]> git.ipfire.org Git - thirdparty/openssl.git/blob - CHANGES
Add ENGINE support for EVP_PKEY_METHOD including lookups of ENGINE
[thirdparty/openssl.git] / CHANGES
1
2 OpenSSL CHANGES
3 _______________
4
5 Changes between 0.9.8b and 0.9.9 [xx XXX xxxx]
6
7 *) Initial engine support for EVP_PKEY_METHOD. New functions to permit
8 an engine to register a method. Add ENGINE lookups for methods and
9 functional reference processing.
10 [Steve Henson]
11
12 *) New functions EVP_Digest{Sign,Verify)*. These are enchance versions of
13 EVP_{Sign,Verify}* which allow an application to customise the signature
14 process.
15 [Steve Henson]
16
17 *) New -resign option to smime utility. This adds one or more signers
18 to an existing PKCS#7 signedData structure. Also -md option to use an
19 alternative message digest algorithm for signing.
20 [Steve Henson]
21
22 *) Tidy up PKCS#7 routines and add new functions to make it easier to
23 create PKCS7 structures containing multiple signers. Update smime
24 application to support multiple signers.
25 [Steve Henson]
26
27 *) New -macalg option to pkcs12 utility to allow setting of an alternative
28 digest MAC.
29 [Steve Henson]
30
31 *) Initial support for PKCS#5 v2.0 PRFs other than default SHA1 HMAC.
32 Reorganize PBE internals to lookup from a static table using NIDs,
33 add support for HMAC PBE OID translation. Add a EVP_CIPHER ctrl:
34 EVP_CTRL_PBE_PRF_NID this allows a cipher to specify an alternative
35 PRF which will be automatically used with PBES2.
36 [Steve Henson]
37
38 *) Replace the algorithm specific calls to generate keys in "req" with the
39 new API.
40 [Steve Henson]
41
42 *) Update PKCS#7 enveloped data routines to use new API. This is now
43 supported by any public key method supporting the encrypt operation. A
44 ctrl is added to allow the public key algorithm to examine or modify
45 the PKCS#7 RecipientInfo structure if it needs to: for RSA this is
46 a no op.
47 [Steve Henson]
48
49 *) Add a ctrl to asn1 method to allow a public key algorithm to express
50 a default digest type to use. In most cases this will be SHA1 but some
51 algorithms (such as GOST) need to specify an alternative digest. The
52 return value indicates how strong the prefernce is 1 means optional and
53 2 is mandatory (that is it is the only supported type). Modify
54 ASN1_item_sign() to accept a NULL digest argument to indicate it should
55 use the default md. Update openssl utilities to use the default digest
56 type for signing if it is not explicitly indicated.
57 [Steve Henson]
58
59 *) Use OID cross reference table in ASN1_sign() and ASN1_verify(). New
60 EVP_MD flag EVP_MD_FLAG_PKEY_METHOD_SIGNATURE. This uses the relevant
61 signing method from the key type. This effectively removes the link
62 between digests and public key types.
63 [Steve Henson]
64
65 *) Add an OID cross reference table and utility functions. Its purpose is to
66 translate between signature OIDs such as SHA1WithrsaEncryption and SHA1,
67 rsaEncryption. This will allow some of the algorithm specific hackery
68 needed to use the correct OID to be removed.
69 [Steve Henson]
70
71 *) Remove algorithm specific dependencies when setting PKCS7_SIGNER_INFO
72 structures for PKCS7_sign(). They are now set up by the relevant public
73 key ASN1 method.
74 [Steve Henson]
75
76 *) Add provisional EC pkey method with support for ECDSA and ECDH.
77 [Steve Henson]
78
79 *) Add support for key derivation (agreement) in the API, DH method and
80 pkeyutl.
81 [Steve Henson]
82
83 *) Add DSA pkey method and DH pkey methods, extend DH ASN1 method to support
84 public and private key formats. As a side effect these add additional
85 command line functionality not previously available: DSA signatures can be
86 generated and verified using pkeyutl and DH key support and generation in
87 pkey, genpkey.
88 [Steve Henson]
89
90 *) BeOS support.
91 [Oliver Tappe <zooey@hirschkaefer.de>]
92
93 *) New make target "install_html_docs" installs HTML renditions of the
94 manual pages.
95 [Oliver Tappe <zooey@hirschkaefer.de>]
96
97 *) New utility "genpkey" this is analagous to "genrsa" etc except it can
98 generate keys for any algorithm. Extend and update EVP_PKEY_METHOD to
99 support key and parameter generation and add initial key generation
100 functionality for RSA.
101 [Steve Henson]
102
103 *) Add functions for main EVP_PKEY_method operations. The undocumented
104 functions EVP_PKEY_{encrypt,decrypt} have been renamed to
105 EVP_PKEY_{encrypt,decrypt}_old.
106 [Steve Henson]
107
108 *) Initial definitions for EVP_PKEY_METHOD. This will be a high level public
109 key API, doesn't do much yet.
110 [Steve Henson]
111
112 *) New function EVP_PKEY_asn1_get0_info() to retrieve information about
113 public key algorithms. New option to openssl utility:
114 "list-public-key-algorithms" to print out info.
115 [Steve Henson]
116
117 *) Implement the Supported Elliptic Curves Extension for
118 ECC ciphersuites from draft-ietf-tls-ecc-12.txt.
119 [Douglas Stebila]
120
121 *) Don't free up OIDs in OBJ_cleanup() if they are in use by EVP_MD or
122 EVP_CIPHER structures to avoid later problems in EVP_cleanup().
123 [Steve Henson]
124
125 *) New utilities pkey and pkeyparam. These are similar to algorithm specific
126 utilities such as rsa, dsa, dsaparam etc except they process any key
127 type.
128 [Steve Henson]
129
130 *) Transfer public key printing routines to EVP_PKEY_ASN1_METHOD. New
131 functions EVP_PKEY_print_public(), EVP_PKEY_print_private(),
132 EVP_PKEY_print_param() to print public key data from an EVP_PKEY
133 structure.
134 [Steve Henson]
135
136 *) Initial support for pluggable public key ASN1.
137 De-spaghettify the public key ASN1 handling. Move public and private
138 key ASN1 handling to a new EVP_PKEY_ASN1_METHOD structure. Relocate
139 algorithm specific handling to a single module within the relevant
140 algorithm directory. Add functions to allow (near) opaque processing
141 of public and private key structures.
142 [Steve Henson]
143
144 *) Implement the Supported Point Formats Extension for
145 ECC ciphersuites from draft-ietf-tls-ecc-12.txt.
146 [Douglas Stebila]
147
148 *) Add initial support for RFC 4279 PSK TLS ciphersuites. Add members
149 for the psk identity [hint] and the psk callback functions to the
150 SSL_SESSION, SSL and SSL_CTX structure.
151
152 New ciphersuites:
153 PSK-RC4-SHA, PSK-3DES-EDE-CBC-SHA, PSK-AES128-CBC-SHA,
154 PSK-AES256-CBC-SHA
155
156 New functions:
157 SSL_CTX_use_psk_identity_hint
158 SSL_get_psk_identity_hint
159 SSL_get_psk_identity
160 SSL_use_psk_identity_hint
161
162 [Mika Kousa and Pasi Eronen of Nokia Corporation]
163
164 *) Add RFC 3161 compliant time stamp request creation, response generation
165 and response verification functionality.
166 [Zoltán Glózik <zglozik@opentsa.org>, The OpenTSA Project]
167
168 *) Add initial support for TLS extensions, specifically for the server_name
169 extension so far. The SSL_SESSION, SSL_CTX, and SSL data structures now
170 have new members for a host name. The SSL data structure has an
171 additional member SSL_CTX *initial_ctx so that new sessions can be
172 stored in that context to allow for session resumption, even after the
173 SSL has been switched to a new SSL_CTX in reaction to a client's
174 server_name extension.
175
176 New functions (subject to change):
177
178 SSL_get_servername()
179 SSL_get_servername_type()
180 SSL_set_SSL_CTX()
181
182 New CTRL codes and macros (subject to change):
183
184 SSL_CTRL_SET_TLSEXT_SERVERNAME_CB
185 - SSL_CTX_set_tlsext_servername_callback()
186 SSL_CTRL_SET_TLSEXT_SERVERNAME_ARG
187 - SSL_CTX_set_tlsext_servername_arg()
188 SSL_CTRL_SET_TLSEXT_HOSTNAME - SSL_set_tlsext_hostname()
189
190 openssl s_client has a new '-servername ...' option.
191
192 openssl s_server has new options '-servername_host ...', '-cert2 ...',
193 '-key2 ...', '-servername_fatal' (subject to change). This allows
194 testing the HostName extension for a specific single host name ('-cert'
195 and '-key' remain fallbacks for handshakes without HostName
196 negotiation). If the unrecogninzed_name alert has to be sent, this by
197 default is a warning; it becomes fatal with the '-servername_fatal'
198 option.
199
200 [Peter Sylvester, Remy Allais, Christophe Renou]
201
202 *) Whirlpool hash implementation is added.
203 [Andy Polyakov]
204
205 *) BIGNUM code on 64-bit SPARCv9 targets is switched from bn(64,64) to
206 bn(64,32). Because of instruction set limitations it doesn't have
207 any negative impact on performance. This was done mostly in order
208 to make it possible to share assembler modules, such as bn_mul_mont
209 implementations, between 32- and 64-bit builds without hassle.
210 [Andy Polyakov]
211
212 *) Disable rogue ciphersuites:
213
214 - SSLv2 0x08 0x00 0x80 ("RC4-64-MD5")
215 - SSLv3/TLSv1 0x00 0x61 ("EXP1024-RC2-CBC-MD5")
216 - SSLv3/TLSv1 0x00 0x60 ("EXP1024-RC4-MD5")
217
218 The latter two were purportedly from
219 draft-ietf-tls-56-bit-ciphersuites-0[01].txt, but do not really
220 appear there.
221
222 Other ciphersuites from draft-ietf-tls-56-bit-ciphersuites-01.txt
223 remain enabled for now, but are just as unofficial, and the ID
224 has long expired; these will probably disappear soon.
225 [Bodo Moeller]
226
227 *) Move code previously exiled into file crypto/ec/ec2_smpt.c
228 to ec2_smpl.c, and no longer require the OPENSSL_EC_BIN_PT_COMP
229 macro.
230 [Bodo Moeller]
231
232 *) New candidate for BIGNUM assembler implementation, bn_mul_mont,
233 dedicated Montgomery multiplication procedure, is introduced.
234 BN_MONT_CTX is modified to allow bn_mul_mont to reach for higher
235 "64-bit" performance on certain 32-bit targets.
236 [Andy Polyakov]
237
238 *) New option SSL_OP_NO_COMP to disable use of compression selectively
239 in SSL structures. New SSL ctrl to set maximum send fragment size.
240 Save memory by seeting the I/O buffer sizes dynamically instead of
241 using the maximum available value.
242 [Steve Henson]
243
244 *) New option -V for 'openssl ciphers'. This prints the ciphersuite code
245 in addition to the text details.
246 [Bodo Moeller]
247
248 *) Very, very preliminary EXPERIMENTAL support for printing of general
249 ASN1 structures. This currently produces rather ugly output and doesn't
250 handle several customised structures at all.
251 [Steve Henson]
252
253 *) Integrated support for PVK file format and some related formats such
254 as MS PUBLICKEYBLOB and PRIVATEKEYBLOB. Command line switches to support
255 these in the 'rsa' and 'dsa' utilities.
256 [Steve Henson]
257
258 *) Support for PKCS#1 RSAPublicKey format on rsa utility command line.
259 [Steve Henson]
260
261 *) Remove the ancient ASN1_METHOD code. This was only ever used in one
262 place for the (very old) "NETSCAPE" format certificates which are now
263 handled using new ASN1 code equivalents.
264 [Steve Henson]
265
266 *) Let the TLSv1_method() etc. functions return a 'const' SSL_METHOD
267 pointer and make the SSL_METHOD parameter in SSL_CTX_new,
268 SSL_CTX_set_ssl_version and SSL_set_ssl_method 'const'.
269 [Nils Larsch]
270
271 *) Modify CRL distribution points extension code to print out previously
272 unsupported fields. Enhance extension setting code to allow setting of
273 all fields.
274 [Steve Henson]
275
276 *) Add print and set support for Issuing Distribution Point CRL extension.
277 [Steve Henson]
278
279 Changes between 0.9.8b and 0.9.8c [xx XXX xxxx]
280
281 *) Disable the padding bug check when compression is in use. The padding
282 bug check assumes the first packet is of even length, this is not
283 necessarily true if compresssion is enabled and can result in false
284 positives causing handshake failure. The actual bug test is ancient
285 code so it is hoped that implementations will either have fixed it by
286 now or any which still have the bug do not support compression.
287 [Steve Henson]
288
289 Changes between 0.9.8a and 0.9.8b [04 May 2006]
290
291 *) When applying a cipher rule check to see if string match is an explicit
292 cipher suite and only match that one cipher suite if it is.
293 [Steve Henson]
294
295 *) Link in manifests for VC++ if needed.
296 [Austin Ziegler <halostatue@gmail.com>]
297
298 *) Update support for ECC-based TLS ciphersuites according to
299 draft-ietf-tls-ecc-12.txt with proposed changes (but without
300 TLS extensions, which are supported starting with the 0.9.9
301 branch, not in the OpenSSL 0.9.8 branch).
302 [Douglas Stebila]
303
304 *) New functions EVP_CIPHER_CTX_new() and EVP_CIPHER_CTX_free() to support
305 opaque EVP_CIPHER_CTX handling.
306 [Steve Henson]
307
308 *) Fixes and enhancements to zlib compression code. We now only use
309 "zlib1.dll" and use the default __cdecl calling convention on Win32
310 to conform with the standards mentioned here:
311 http://www.zlib.net/DLL_FAQ.txt
312 Static zlib linking now works on Windows and the new --with-zlib-include
313 --with-zlib-lib options to Configure can be used to supply the location
314 of the headers and library. Gracefully handle case where zlib library
315 can't be loaded.
316 [Steve Henson]
317
318 *) Several fixes and enhancements to the OID generation code. The old code
319 sometimes allowed invalid OIDs (1.X for X >= 40 for example), couldn't
320 handle numbers larger than ULONG_MAX, truncated printing and had a
321 non standard OBJ_obj2txt() behaviour.
322 [Steve Henson]
323
324 *) Add support for building of engines under engine/ as shared libraries
325 under VC++ build system.
326 [Steve Henson]
327
328 *) Corrected the numerous bugs in the Win32 path splitter in DSO.
329 Hopefully, we will not see any false combination of paths any more.
330 [Richard Levitte]
331
332 Changes between 0.9.8 and 0.9.8a [11 Oct 2005]
333
334 *) Remove the functionality of SSL_OP_MSIE_SSLV2_RSA_PADDING
335 (part of SSL_OP_ALL). This option used to disable the
336 countermeasure against man-in-the-middle protocol-version
337 rollback in the SSL 2.0 server implementation, which is a bad
338 idea. (CVE-2005-2969)
339
340 [Bodo Moeller; problem pointed out by Yutaka Oiwa (Research Center
341 for Information Security, National Institute of Advanced Industrial
342 Science and Technology [AIST], Japan)]
343
344 *) Add two function to clear and return the verify parameter flags.
345 [Steve Henson]
346
347 *) Keep cipherlists sorted in the source instead of sorting them at
348 runtime, thus removing the need for a lock.
349 [Nils Larsch]
350
351 *) Avoid some small subgroup attacks in Diffie-Hellman.
352 [Nick Mathewson and Ben Laurie]
353
354 *) Add functions for well-known primes.
355 [Nick Mathewson]
356
357 *) Extended Windows CE support.
358 [Satoshi Nakamura and Andy Polyakov]
359
360 *) Initialize SSL_METHOD structures at compile time instead of during
361 runtime, thus removing the need for a lock.
362 [Steve Henson]
363
364 *) Make PKCS7_decrypt() work even if no certificate is supplied by
365 attempting to decrypt each encrypted key in turn. Add support to
366 smime utility.
367 [Steve Henson]
368
369 Changes between 0.9.7h and 0.9.8 [05 Jul 2005]
370
371 *) Add libcrypto.pc and libssl.pc for those who feel they need them.
372 [Richard Levitte]
373
374 *) Change CA.sh and CA.pl so they don't bundle the CSR and the private
375 key into the same file any more.
376 [Richard Levitte]
377
378 *) Add initial support for Win64, both IA64 and AMD64/x64 flavors.
379 [Andy Polyakov]
380
381 *) Add -utf8 command line and config file option to 'ca'.
382 [Stefan <stf@udoma.org]
383
384 *) Removed the macro des_crypt(), as it seems to conflict with some
385 libraries. Use DES_crypt().
386 [Richard Levitte]
387
388 *) Correct naming of the 'chil' and '4758cca' ENGINEs. This
389 involves renaming the source and generated shared-libs for
390 both. The engines will accept the corrected or legacy ids
391 ('ncipher' and '4758_cca' respectively) when binding. NB,
392 this only applies when building 'shared'.
393 [Corinna Vinschen <vinschen@redhat.com> and Geoff Thorpe]
394
395 *) Add attribute functions to EVP_PKEY structure. Modify
396 PKCS12_create() to recognize a CSP name attribute and
397 use it. Make -CSP option work again in pkcs12 utility.
398 [Steve Henson]
399
400 *) Add new functionality to the bn blinding code:
401 - automatic re-creation of the BN_BLINDING parameters after
402 a fixed number of uses (currently 32)
403 - add new function for parameter creation
404 - introduce flags to control the update behaviour of the
405 BN_BLINDING parameters
406 - hide BN_BLINDING structure
407 Add a second BN_BLINDING slot to the RSA structure to improve
408 performance when a single RSA object is shared among several
409 threads.
410 [Nils Larsch]
411
412 *) Add support for DTLS.
413 [Nagendra Modadugu <nagendra@cs.stanford.edu> and Ben Laurie]
414
415 *) Add support for DER encoded private keys (SSL_FILETYPE_ASN1)
416 to SSL_CTX_use_PrivateKey_file() and SSL_use_PrivateKey_file()
417 [Walter Goulet]
418
419 *) Remove buggy and incompletet DH cert support from
420 ssl/ssl_rsa.c and ssl/s3_both.c
421 [Nils Larsch]
422
423 *) Use SHA-1 instead of MD5 as the default digest algorithm for
424 the apps/openssl applications.
425 [Nils Larsch]
426
427 *) Compile clean with "-Wall -Wmissing-prototypes
428 -Wstrict-prototypes -Wmissing-declarations -Werror". Currently
429 DEBUG_SAFESTACK must also be set.
430 [Ben Laurie]
431
432 *) Change ./Configure so that certain algorithms can be disabled by default.
433 The new counterpiece to "no-xxx" is "enable-xxx".
434
435 The patented RC5 and MDC2 algorithms will now be disabled unless
436 "enable-rc5" and "enable-mdc2", respectively, are specified.
437
438 (IDEA remains enabled despite being patented. This is because IDEA
439 is frequently required for interoperability, and there is no license
440 fee for non-commercial use. As before, "no-idea" can be used to
441 avoid this algorithm.)
442
443 [Bodo Moeller]
444
445 *) Add processing of proxy certificates (see RFC 3820). This work was
446 sponsored by KTH (The Royal Institute of Technology in Stockholm) and
447 EGEE (Enabling Grids for E-science in Europe).
448 [Richard Levitte]
449
450 *) RC4 performance overhaul on modern architectures/implementations, such
451 as Intel P4, IA-64 and AMD64.
452 [Andy Polyakov]
453
454 *) New utility extract-section.pl. This can be used specify an alternative
455 section number in a pod file instead of having to treat each file as
456 a separate case in Makefile. This can be done by adding two lines to the
457 pod file:
458
459 =for comment openssl_section:XXX
460
461 The blank line is mandatory.
462
463 [Steve Henson]
464
465 *) New arguments -certform, -keyform and -pass for s_client and s_server
466 to allow alternative format key and certificate files and passphrase
467 sources.
468 [Steve Henson]
469
470 *) New structure X509_VERIFY_PARAM which combines current verify parameters,
471 update associated structures and add various utility functions.
472
473 Add new policy related verify parameters, include policy checking in
474 standard verify code. Enhance 'smime' application with extra parameters
475 to support policy checking and print out.
476 [Steve Henson]
477
478 *) Add a new engine to support VIA PadLock ACE extensions in the VIA C3
479 Nehemiah processors. These extensions support AES encryption in hardware
480 as well as RNG (though RNG support is currently disabled).
481 [Michal Ludvig <michal@logix.cz>, with help from Andy Polyakov]
482
483 *) Deprecate BN_[get|set]_params() functions (they were ignored internally).
484 [Geoff Thorpe]
485
486 *) New FIPS 180-2 algorithms, SHA-224/-256/-384/-512 are implemented.
487 [Andy Polyakov and a number of other people]
488
489 *) Improved PowerPC platform support. Most notably BIGNUM assembler
490 implementation contributed by IBM.
491 [Suresh Chari, Peter Waltenberg, Andy Polyakov]
492
493 *) The new 'RSA_generate_key_ex' function now takes a BIGNUM for the public
494 exponent rather than 'unsigned long'. There is a corresponding change to
495 the new 'rsa_keygen' element of the RSA_METHOD structure.
496 [Jelte Jansen, Geoff Thorpe]
497
498 *) Functionality for creating the initial serial number file is now
499 moved from CA.pl to the 'ca' utility with a new option -create_serial.
500
501 (Before OpenSSL 0.9.7e, CA.pl used to initialize the serial
502 number file to 1, which is bound to cause problems. To avoid
503 the problems while respecting compatibility between different 0.9.7
504 patchlevels, 0.9.7e employed 'openssl x509 -next_serial' in
505 CA.pl for serial number initialization. With the new release 0.9.8,
506 we can fix the problem directly in the 'ca' utility.)
507 [Steve Henson]
508
509 *) Reduced header interdepencies by declaring more opaque objects in
510 ossl_typ.h. As a consequence, including some headers (eg. engine.h) will
511 give fewer recursive includes, which could break lazy source code - so
512 this change is covered by the OPENSSL_NO_DEPRECATED symbol. As always,
513 developers should define this symbol when building and using openssl to
514 ensure they track the recommended behaviour, interfaces, [etc], but
515 backwards-compatible behaviour prevails when this isn't defined.
516 [Geoff Thorpe]
517
518 *) New function X509_POLICY_NODE_print() which prints out policy nodes.
519 [Steve Henson]
520
521 *) Add new EVP function EVP_CIPHER_CTX_rand_key and associated functionality.
522 This will generate a random key of the appropriate length based on the
523 cipher context. The EVP_CIPHER can provide its own random key generation
524 routine to support keys of a specific form. This is used in the des and
525 3des routines to generate a key of the correct parity. Update S/MIME
526 code to use new functions and hence generate correct parity DES keys.
527 Add EVP_CHECK_DES_KEY #define to return an error if the key is not
528 valid (weak or incorrect parity).
529 [Steve Henson]
530
531 *) Add a local set of CRLs that can be used by X509_verify_cert() as well
532 as looking them up. This is useful when the verified structure may contain
533 CRLs, for example PKCS#7 signedData. Modify PKCS7_verify() to use any CRLs
534 present unless the new PKCS7_NO_CRL flag is asserted.
535 [Steve Henson]
536
537 *) Extend ASN1 oid configuration module. It now additionally accepts the
538 syntax:
539
540 shortName = some long name, 1.2.3.4
541 [Steve Henson]
542
543 *) Reimplemented the BN_CTX implementation. There is now no more static
544 limitation on the number of variables it can handle nor the depth of the
545 "stack" handling for BN_CTX_start()/BN_CTX_end() pairs. The stack
546 information can now expand as required, and rather than having a single
547 static array of bignums, BN_CTX now uses a linked-list of such arrays
548 allowing it to expand on demand whilst maintaining the usefulness of
549 BN_CTX's "bundling".
550 [Geoff Thorpe]
551
552 *) Add a missing BN_CTX parameter to the 'rsa_mod_exp' callback in RSA_METHOD
553 to allow all RSA operations to function using a single BN_CTX.
554 [Geoff Thorpe]
555
556 *) Preliminary support for certificate policy evaluation and checking. This
557 is initially intended to pass the tests outlined in "Conformance Testing
558 of Relying Party Client Certificate Path Processing Logic" v1.07.
559 [Steve Henson]
560
561 *) bn_dup_expand() has been deprecated, it was introduced in 0.9.7 and
562 remained unused and not that useful. A variety of other little bignum
563 tweaks and fixes have also been made continuing on from the audit (see
564 below).
565 [Geoff Thorpe]
566
567 *) Constify all or almost all d2i, c2i, s2i and r2i functions, along with
568 associated ASN1, EVP and SSL functions and old ASN1 macros.
569 [Richard Levitte]
570
571 *) BN_zero() only needs to set 'top' and 'neg' to zero for correct results,
572 and this should never fail. So the return value from the use of
573 BN_set_word() (which can fail due to needless expansion) is now deprecated;
574 if OPENSSL_NO_DEPRECATED is defined, BN_zero() is a void macro.
575 [Geoff Thorpe]
576
577 *) BN_CTX_get() should return zero-valued bignums, providing the same
578 initialised value as BN_new().
579 [Geoff Thorpe, suggested by Ulf Möller]
580
581 *) Support for inhibitAnyPolicy certificate extension.
582 [Steve Henson]
583
584 *) An audit of the BIGNUM code is underway, for which debugging code is
585 enabled when BN_DEBUG is defined. This makes stricter enforcements on what
586 is considered valid when processing BIGNUMs, and causes execution to
587 assert() when a problem is discovered. If BN_DEBUG_RAND is defined,
588 further steps are taken to deliberately pollute unused data in BIGNUM
589 structures to try and expose faulty code further on. For now, openssl will
590 (in its default mode of operation) continue to tolerate the inconsistent
591 forms that it has tolerated in the past, but authors and packagers should
592 consider trying openssl and their own applications when compiled with
593 these debugging symbols defined. It will help highlight potential bugs in
594 their own code, and will improve the test coverage for OpenSSL itself. At
595 some point, these tighter rules will become openssl's default to improve
596 maintainability, though the assert()s and other overheads will remain only
597 in debugging configurations. See bn.h for more details.
598 [Geoff Thorpe, Nils Larsch, Ulf Möller]
599
600 *) BN_CTX_init() has been deprecated, as BN_CTX is an opaque structure
601 that can only be obtained through BN_CTX_new() (which implicitly
602 initialises it). The presence of this function only made it possible
603 to overwrite an existing structure (and cause memory leaks).
604 [Geoff Thorpe]
605
606 *) Because of the callback-based approach for implementing LHASH as a
607 template type, lh_insert() adds opaque objects to hash-tables and
608 lh_doall() or lh_doall_arg() are typically used with a destructor callback
609 to clean up those corresponding objects before destroying the hash table
610 (and losing the object pointers). So some over-zealous constifications in
611 LHASH have been relaxed so that lh_insert() does not take (nor store) the
612 objects as "const" and the lh_doall[_arg] callback wrappers are not
613 prototyped to have "const" restrictions on the object pointers they are
614 given (and so aren't required to cast them away any more).
615 [Geoff Thorpe]
616
617 *) The tmdiff.h API was so ugly and minimal that our own timing utility
618 (speed) prefers to use its own implementation. The two implementations
619 haven't been consolidated as yet (volunteers?) but the tmdiff API has had
620 its object type properly exposed (MS_TM) instead of casting to/from "char
621 *". This may still change yet if someone realises MS_TM and "ms_time_***"
622 aren't necessarily the greatest nomenclatures - but this is what was used
623 internally to the implementation so I've used that for now.
624 [Geoff Thorpe]
625
626 *) Ensure that deprecated functions do not get compiled when
627 OPENSSL_NO_DEPRECATED is defined. Some "openssl" subcommands and a few of
628 the self-tests were still using deprecated key-generation functions so
629 these have been updated also.
630 [Geoff Thorpe]
631
632 *) Reorganise PKCS#7 code to separate the digest location functionality
633 into PKCS7_find_digest(), digest addtion into PKCS7_bio_add_digest().
634 New function PKCS7_set_digest() to set the digest type for PKCS#7
635 digestedData type. Add additional code to correctly generate the
636 digestedData type and add support for this type in PKCS7 initialization
637 functions.
638 [Steve Henson]
639
640 *) New function PKCS7_set0_type_other() this initializes a PKCS7
641 structure of type "other".
642 [Steve Henson]
643
644 *) Fix prime generation loop in crypto/bn/bn_prime.pl by making
645 sure the loop does correctly stop and breaking ("division by zero")
646 modulus operations are not performed. The (pre-generated) prime
647 table crypto/bn/bn_prime.h was already correct, but it could not be
648 re-generated on some platforms because of the "division by zero"
649 situation in the script.
650 [Ralf S. Engelschall]
651
652 *) Update support for ECC-based TLS ciphersuites according to
653 draft-ietf-tls-ecc-03.txt: the KDF1 key derivation function with
654 SHA-1 now is only used for "small" curves (where the
655 representation of a field element takes up to 24 bytes); for
656 larger curves, the field element resulting from ECDH is directly
657 used as premaster secret.
658 [Douglas Stebila (Sun Microsystems Laboratories)]
659
660 *) Add code for kP+lQ timings to crypto/ec/ectest.c, and add SEC2
661 curve secp160r1 to the tests.
662 [Douglas Stebila (Sun Microsystems Laboratories)]
663
664 *) Add the possibility to load symbols globally with DSO.
665 [Götz Babin-Ebell <babin-ebell@trustcenter.de> via Richard Levitte]
666
667 *) Add the functions ERR_set_mark() and ERR_pop_to_mark() for better
668 control of the error stack.
669 [Richard Levitte]
670
671 *) Add support for STORE in ENGINE.
672 [Richard Levitte]
673
674 *) Add the STORE type. The intention is to provide a common interface
675 to certificate and key stores, be they simple file-based stores, or
676 HSM-type store, or LDAP stores, or...
677 NOTE: The code is currently UNTESTED and isn't really used anywhere.
678 [Richard Levitte]
679
680 *) Add a generic structure called OPENSSL_ITEM. This can be used to
681 pass a list of arguments to any function as well as provide a way
682 for a function to pass data back to the caller.
683 [Richard Levitte]
684
685 *) Add the functions BUF_strndup() and BUF_memdup(). BUF_strndup()
686 works like BUF_strdup() but can be used to duplicate a portion of
687 a string. The copy gets NUL-terminated. BUF_memdup() duplicates
688 a memory area.
689 [Richard Levitte]
690
691 *) Add the function sk_find_ex() which works like sk_find(), but will
692 return an index to an element even if an exact match couldn't be
693 found. The index is guaranteed to point at the element where the
694 searched-for key would be inserted to preserve sorting order.
695 [Richard Levitte]
696
697 *) Add the function OBJ_bsearch_ex() which works like OBJ_bsearch() but
698 takes an extra flags argument for optional functionality. Currently,
699 the following flags are defined:
700
701 OBJ_BSEARCH_VALUE_ON_NOMATCH
702 This one gets OBJ_bsearch_ex() to return a pointer to the first
703 element where the comparing function returns a negative or zero
704 number.
705
706 OBJ_BSEARCH_FIRST_VALUE_ON_MATCH
707 This one gets OBJ_bsearch_ex() to return a pointer to the first
708 element where the comparing function returns zero. This is useful
709 if there are more than one element where the comparing function
710 returns zero.
711 [Richard Levitte]
712
713 *) Make it possible to create self-signed certificates with 'openssl ca'
714 in such a way that the self-signed certificate becomes part of the
715 CA database and uses the same mechanisms for serial number generation
716 as all other certificate signing. The new flag '-selfsign' enables
717 this functionality. Adapt CA.sh and CA.pl.in.
718 [Richard Levitte]
719
720 *) Add functionality to check the public key of a certificate request
721 against a given private. This is useful to check that a certificate
722 request can be signed by that key (self-signing).
723 [Richard Levitte]
724
725 *) Make it possible to have multiple active certificates with the same
726 subject in the CA index file. This is done only if the keyword
727 'unique_subject' is set to 'no' in the main CA section (default
728 if 'CA_default') of the configuration file. The value is saved
729 with the database itself in a separate index attribute file,
730 named like the index file with '.attr' appended to the name.
731 [Richard Levitte]
732
733 *) Generate muti valued AVAs using '+' notation in config files for
734 req and dirName.
735 [Steve Henson]
736
737 *) Support for nameConstraints certificate extension.
738 [Steve Henson]
739
740 *) Support for policyConstraints certificate extension.
741 [Steve Henson]
742
743 *) Support for policyMappings certificate extension.
744 [Steve Henson]
745
746 *) Make sure the default DSA_METHOD implementation only uses its
747 dsa_mod_exp() and/or bn_mod_exp() handlers if they are non-NULL,
748 and change its own handlers to be NULL so as to remove unnecessary
749 indirection. This lets alternative implementations fallback to the
750 default implementation more easily.
751 [Geoff Thorpe]
752
753 *) Support for directoryName in GeneralName related extensions
754 in config files.
755 [Steve Henson]
756
757 *) Make it possible to link applications using Makefile.shared.
758 Make that possible even when linking against static libraries!
759 [Richard Levitte]
760
761 *) Support for single pass processing for S/MIME signing. This now
762 means that S/MIME signing can be done from a pipe, in addition
763 cleartext signing (multipart/signed type) is effectively streaming
764 and the signed data does not need to be all held in memory.
765
766 This is done with a new flag PKCS7_STREAM. When this flag is set
767 PKCS7_sign() only initializes the PKCS7 structure and the actual signing
768 is done after the data is output (and digests calculated) in
769 SMIME_write_PKCS7().
770 [Steve Henson]
771
772 *) Add full support for -rpath/-R, both in shared libraries and
773 applications, at least on the platforms where it's known how
774 to do it.
775 [Richard Levitte]
776
777 *) In crypto/ec/ec_mult.c, implement fast point multiplication with
778 precomputation, based on wNAF splitting: EC_GROUP_precompute_mult()
779 will now compute a table of multiples of the generator that
780 makes subsequent invocations of EC_POINTs_mul() or EC_POINT_mul()
781 faster (notably in the case of a single point multiplication,
782 scalar * generator).
783 [Nils Larsch, Bodo Moeller]
784
785 *) IPv6 support for certificate extensions. The various extensions
786 which use the IP:a.b.c.d can now take IPv6 addresses using the
787 formats of RFC1884 2.2 . IPv6 addresses are now also displayed
788 correctly.
789 [Steve Henson]
790
791 *) Added an ENGINE that implements RSA by performing private key
792 exponentiations with the GMP library. The conversions to and from
793 GMP's mpz_t format aren't optimised nor are any montgomery forms
794 cached, and on x86 it appears OpenSSL's own performance has caught up.
795 However there are likely to be other architectures where GMP could
796 provide a boost. This ENGINE is not built in by default, but it can be
797 specified at Configure time and should be accompanied by the necessary
798 linker additions, eg;
799 ./config -DOPENSSL_USE_GMP -lgmp
800 [Geoff Thorpe]
801
802 *) "openssl engine" will not display ENGINE/DSO load failure errors when
803 testing availability of engines with "-t" - the old behaviour is
804 produced by increasing the feature's verbosity with "-tt".
805 [Geoff Thorpe]
806
807 *) ECDSA routines: under certain error conditions uninitialized BN objects
808 could be freed. Solution: make sure initialization is performed early
809 enough. (Reported and fix supplied by Nils Larsch <nla@trustcenter.de>
810 via PR#459)
811 [Lutz Jaenicke]
812
813 *) Key-generation can now be implemented in RSA_METHOD, DSA_METHOD
814 and DH_METHOD (eg. by ENGINE implementations) to override the normal
815 software implementations. For DSA and DH, parameter generation can
816 also be overriden by providing the appropriate method callbacks.
817 [Geoff Thorpe]
818
819 *) Change the "progress" mechanism used in key-generation and
820 primality testing to functions that take a new BN_GENCB pointer in
821 place of callback/argument pairs. The new API functions have "_ex"
822 postfixes and the older functions are reimplemented as wrappers for
823 the new ones. The OPENSSL_NO_DEPRECATED symbol can be used to hide
824 declarations of the old functions to help (graceful) attempts to
825 migrate to the new functions. Also, the new key-generation API
826 functions operate on a caller-supplied key-structure and return
827 success/failure rather than returning a key or NULL - this is to
828 help make "keygen" another member function of RSA_METHOD etc.
829
830 Example for using the new callback interface:
831
832 int (*my_callback)(int a, int b, BN_GENCB *cb) = ...;
833 void *my_arg = ...;
834 BN_GENCB my_cb;
835
836 BN_GENCB_set(&my_cb, my_callback, my_arg);
837
838 return BN_is_prime_ex(some_bignum, BN_prime_checks, NULL, &cb);
839 /* For the meaning of a, b in calls to my_callback(), see the
840 * documentation of the function that calls the callback.
841 * cb will point to my_cb; my_arg can be retrieved as cb->arg.
842 * my_callback should return 1 if it wants BN_is_prime_ex()
843 * to continue, or 0 to stop.
844 */
845
846 [Geoff Thorpe]
847
848 *) Change the ZLIB compression method to be stateful, and make it
849 available to TLS with the number defined in
850 draft-ietf-tls-compression-04.txt.
851 [Richard Levitte]
852
853 *) Add the ASN.1 structures and functions for CertificatePair, which
854 is defined as follows (according to X.509_4thEditionDraftV6.pdf):
855
856 CertificatePair ::= SEQUENCE {
857 forward [0] Certificate OPTIONAL,
858 reverse [1] Certificate OPTIONAL,
859 -- at least one of the pair shall be present -- }
860
861 Also implement the PEM functions to read and write certificate
862 pairs, and defined the PEM tag as "CERTIFICATE PAIR".
863
864 This needed to be defined, mostly for the sake of the LDAP
865 attribute crossCertificatePair, but may prove useful elsewhere as
866 well.
867 [Richard Levitte]
868
869 *) Make it possible to inhibit symlinking of shared libraries in
870 Makefile.shared, for Cygwin's sake.
871 [Richard Levitte]
872
873 *) Extend the BIGNUM API by creating a function
874 void BN_set_negative(BIGNUM *a, int neg);
875 and a macro that behave like
876 int BN_is_negative(const BIGNUM *a);
877
878 to avoid the need to access 'a->neg' directly in applications.
879 [Nils Larsch]
880
881 *) Implement fast modular reduction for pseudo-Mersenne primes
882 used in NIST curves (crypto/bn/bn_nist.c, crypto/ec/ecp_nist.c).
883 EC_GROUP_new_curve_GFp() will now automatically use this
884 if applicable.
885 [Nils Larsch <nla@trustcenter.de>]
886
887 *) Add new lock type (CRYPTO_LOCK_BN).
888 [Bodo Moeller]
889
890 *) Change the ENGINE framework to automatically load engines
891 dynamically from specific directories unless they could be
892 found to already be built in or loaded. Move all the
893 current engines except for the cryptodev one to a new
894 directory engines/.
895 The engines in engines/ are built as shared libraries if
896 the "shared" options was given to ./Configure or ./config.
897 Otherwise, they are inserted in libcrypto.a.
898 /usr/local/ssl/engines is the default directory for dynamic
899 engines, but that can be overriden at configure time through
900 the usual use of --prefix and/or --openssldir, and at run
901 time with the environment variable OPENSSL_ENGINES.
902 [Geoff Thorpe and Richard Levitte]
903
904 *) Add Makefile.shared, a helper makefile to build shared
905 libraries. Addapt Makefile.org.
906 [Richard Levitte]
907
908 *) Add version info to Win32 DLLs.
909 [Peter 'Luna' Runestig" <peter@runestig.com>]
910
911 *) Add new 'medium level' PKCS#12 API. Certificates and keys
912 can be added using this API to created arbitrary PKCS#12
913 files while avoiding the low level API.
914
915 New options to PKCS12_create(), key or cert can be NULL and
916 will then be omitted from the output file. The encryption
917 algorithm NIDs can be set to -1 for no encryption, the mac
918 iteration count can be set to 0 to omit the mac.
919
920 Enhance pkcs12 utility by making the -nokeys and -nocerts
921 options work when creating a PKCS#12 file. New option -nomac
922 to omit the mac, NONE can be set for an encryption algorithm.
923 New code is modified to use the enhanced PKCS12_create()
924 instead of the low level API.
925 [Steve Henson]
926
927 *) Extend ASN1 encoder to support indefinite length constructed
928 encoding. This can output sequences tags and octet strings in
929 this form. Modify pk7_asn1.c to support indefinite length
930 encoding. This is experimental and needs additional code to
931 be useful, such as an ASN1 bio and some enhanced streaming
932 PKCS#7 code.
933
934 Extend template encode functionality so that tagging is passed
935 down to the template encoder.
936 [Steve Henson]
937
938 *) Let 'openssl req' fail if an argument to '-newkey' is not
939 recognized instead of using RSA as a default.
940 [Bodo Moeller]
941
942 *) Add support for ECC-based ciphersuites from draft-ietf-tls-ecc-01.txt.
943 As these are not official, they are not included in "ALL";
944 the "ECCdraft" ciphersuite group alias can be used to select them.
945 [Vipul Gupta and Sumit Gupta (Sun Microsystems Laboratories)]
946
947 *) Add ECDH engine support.
948 [Nils Gura and Douglas Stebila (Sun Microsystems Laboratories)]
949
950 *) Add ECDH in new directory crypto/ecdh/.
951 [Douglas Stebila (Sun Microsystems Laboratories)]
952
953 *) Let BN_rand_range() abort with an error after 100 iterations
954 without success (which indicates a broken PRNG).
955 [Bodo Moeller]
956
957 *) Change BN_mod_sqrt() so that it verifies that the input value
958 is really the square of the return value. (Previously,
959 BN_mod_sqrt would show GIGO behaviour.)
960 [Bodo Moeller]
961
962 *) Add named elliptic curves over binary fields from X9.62, SECG,
963 and WAP/WTLS; add OIDs that were still missing.
964
965 [Sheueling Chang Shantz and Douglas Stebila
966 (Sun Microsystems Laboratories)]
967
968 *) Extend the EC library for elliptic curves over binary fields
969 (new files ec2_smpl.c, ec2_smpt.c, ec2_mult.c in crypto/ec/).
970 New EC_METHOD:
971
972 EC_GF2m_simple_method
973
974 New API functions:
975
976 EC_GROUP_new_curve_GF2m
977 EC_GROUP_set_curve_GF2m
978 EC_GROUP_get_curve_GF2m
979 EC_POINT_set_affine_coordinates_GF2m
980 EC_POINT_get_affine_coordinates_GF2m
981 EC_POINT_set_compressed_coordinates_GF2m
982
983 Point compression for binary fields is disabled by default for
984 patent reasons (compile with OPENSSL_EC_BIN_PT_COMP defined to
985 enable it).
986
987 As binary polynomials are represented as BIGNUMs, various members
988 of the EC_GROUP and EC_POINT data structures can be shared
989 between the implementations for prime fields and binary fields;
990 the above ..._GF2m functions (except for EX_GROUP_new_curve_GF2m)
991 are essentially identical to their ..._GFp counterparts.
992 (For simplicity, the '..._GFp' prefix has been dropped from
993 various internal method names.)
994
995 An internal 'field_div' method (similar to 'field_mul' and
996 'field_sqr') has been added; this is used only for binary fields.
997
998 [Sheueling Chang Shantz and Douglas Stebila
999 (Sun Microsystems Laboratories)]
1000
1001 *) Optionally dispatch EC_POINT_mul(), EC_POINT_precompute_mult()
1002 through methods ('mul', 'precompute_mult').
1003
1004 The generic implementations (now internally called 'ec_wNAF_mul'
1005 and 'ec_wNAF_precomputed_mult') remain the default if these
1006 methods are undefined.
1007
1008 [Sheueling Chang Shantz and Douglas Stebila
1009 (Sun Microsystems Laboratories)]
1010
1011 *) New function EC_GROUP_get_degree, which is defined through
1012 EC_METHOD. For curves over prime fields, this returns the bit
1013 length of the modulus.
1014
1015 [Sheueling Chang Shantz and Douglas Stebila
1016 (Sun Microsystems Laboratories)]
1017
1018 *) New functions EC_GROUP_dup, EC_POINT_dup.
1019 (These simply call ..._new and ..._copy).
1020
1021 [Sheueling Chang Shantz and Douglas Stebila
1022 (Sun Microsystems Laboratories)]
1023
1024 *) Add binary polynomial arithmetic software in crypto/bn/bn_gf2m.c.
1025 Polynomials are represented as BIGNUMs (where the sign bit is not
1026 used) in the following functions [macros]:
1027
1028 BN_GF2m_add
1029 BN_GF2m_sub [= BN_GF2m_add]
1030 BN_GF2m_mod [wrapper for BN_GF2m_mod_arr]
1031 BN_GF2m_mod_mul [wrapper for BN_GF2m_mod_mul_arr]
1032 BN_GF2m_mod_sqr [wrapper for BN_GF2m_mod_sqr_arr]
1033 BN_GF2m_mod_inv
1034 BN_GF2m_mod_exp [wrapper for BN_GF2m_mod_exp_arr]
1035 BN_GF2m_mod_sqrt [wrapper for BN_GF2m_mod_sqrt_arr]
1036 BN_GF2m_mod_solve_quad [wrapper for BN_GF2m_mod_solve_quad_arr]
1037 BN_GF2m_cmp [= BN_ucmp]
1038
1039 (Note that only the 'mod' functions are actually for fields GF(2^m).
1040 BN_GF2m_add() is misnomer, but this is for the sake of consistency.)
1041
1042 For some functions, an the irreducible polynomial defining a
1043 field can be given as an 'unsigned int[]' with strictly
1044 decreasing elements giving the indices of those bits that are set;
1045 i.e., p[] represents the polynomial
1046 f(t) = t^p[0] + t^p[1] + ... + t^p[k]
1047 where
1048 p[0] > p[1] > ... > p[k] = 0.
1049 This applies to the following functions:
1050
1051 BN_GF2m_mod_arr
1052 BN_GF2m_mod_mul_arr
1053 BN_GF2m_mod_sqr_arr
1054 BN_GF2m_mod_inv_arr [wrapper for BN_GF2m_mod_inv]
1055 BN_GF2m_mod_div_arr [wrapper for BN_GF2m_mod_div]
1056 BN_GF2m_mod_exp_arr
1057 BN_GF2m_mod_sqrt_arr
1058 BN_GF2m_mod_solve_quad_arr
1059 BN_GF2m_poly2arr
1060 BN_GF2m_arr2poly
1061
1062 Conversion can be performed by the following functions:
1063
1064 BN_GF2m_poly2arr
1065 BN_GF2m_arr2poly
1066
1067 bntest.c has additional tests for binary polynomial arithmetic.
1068
1069 Two implementations for BN_GF2m_mod_div() are available.
1070 The default algorithm simply uses BN_GF2m_mod_inv() and
1071 BN_GF2m_mod_mul(). The alternative algorithm is compiled in only
1072 if OPENSSL_SUN_GF2M_DIV is defined (patent pending; read the
1073 copyright notice in crypto/bn/bn_gf2m.c before enabling it).
1074
1075 [Sheueling Chang Shantz and Douglas Stebila
1076 (Sun Microsystems Laboratories)]
1077
1078 *) Add new error code 'ERR_R_DISABLED' that can be used when some
1079 functionality is disabled at compile-time.
1080 [Douglas Stebila <douglas.stebila@sun.com>]
1081
1082 *) Change default behaviour of 'openssl asn1parse' so that more
1083 information is visible when viewing, e.g., a certificate:
1084
1085 Modify asn1_parse2 (crypto/asn1/asn1_par.c) so that in non-'dump'
1086 mode the content of non-printable OCTET STRINGs is output in a
1087 style similar to INTEGERs, but with '[HEX DUMP]' prepended to
1088 avoid the appearance of a printable string.
1089 [Nils Larsch <nla@trustcenter.de>]
1090
1091 *) Add 'asn1_flag' and 'asn1_form' member to EC_GROUP with access
1092 functions
1093 EC_GROUP_set_asn1_flag()
1094 EC_GROUP_get_asn1_flag()
1095 EC_GROUP_set_point_conversion_form()
1096 EC_GROUP_get_point_conversion_form()
1097 These control ASN1 encoding details:
1098 - Curves (i.e., groups) are encoded explicitly unless asn1_flag
1099 has been set to OPENSSL_EC_NAMED_CURVE.
1100 - Points are encoded in uncompressed form by default; options for
1101 asn1_for are as for point2oct, namely
1102 POINT_CONVERSION_COMPRESSED
1103 POINT_CONVERSION_UNCOMPRESSED
1104 POINT_CONVERSION_HYBRID
1105
1106 Also add 'seed' and 'seed_len' members to EC_GROUP with access
1107 functions
1108 EC_GROUP_set_seed()
1109 EC_GROUP_get0_seed()
1110 EC_GROUP_get_seed_len()
1111 This is used only for ASN1 purposes (so far).
1112 [Nils Larsch <nla@trustcenter.de>]
1113
1114 *) Add 'field_type' member to EC_METHOD, which holds the NID
1115 of the appropriate field type OID. The new function
1116 EC_METHOD_get_field_type() returns this value.
1117 [Nils Larsch <nla@trustcenter.de>]
1118
1119 *) Add functions
1120 EC_POINT_point2bn()
1121 EC_POINT_bn2point()
1122 EC_POINT_point2hex()
1123 EC_POINT_hex2point()
1124 providing useful interfaces to EC_POINT_point2oct() and
1125 EC_POINT_oct2point().
1126 [Nils Larsch <nla@trustcenter.de>]
1127
1128 *) Change internals of the EC library so that the functions
1129 EC_GROUP_set_generator()
1130 EC_GROUP_get_generator()
1131 EC_GROUP_get_order()
1132 EC_GROUP_get_cofactor()
1133 are implemented directly in crypto/ec/ec_lib.c and not dispatched
1134 to methods, which would lead to unnecessary code duplication when
1135 adding different types of curves.
1136 [Nils Larsch <nla@trustcenter.de> with input by Bodo Moeller]
1137
1138 *) Implement compute_wNAF (crypto/ec/ec_mult.c) without BIGNUM
1139 arithmetic, and such that modified wNAFs are generated
1140 (which avoid length expansion in many cases).
1141 [Bodo Moeller]
1142
1143 *) Add a function EC_GROUP_check_discriminant() (defined via
1144 EC_METHOD) that verifies that the curve discriminant is non-zero.
1145
1146 Add a function EC_GROUP_check() that makes some sanity tests
1147 on a EC_GROUP, its generator and order. This includes
1148 EC_GROUP_check_discriminant().
1149 [Nils Larsch <nla@trustcenter.de>]
1150
1151 *) Add ECDSA in new directory crypto/ecdsa/.
1152
1153 Add applications 'openssl ecparam' and 'openssl ecdsa'
1154 (these are based on 'openssl dsaparam' and 'openssl dsa').
1155
1156 ECDSA support is also included in various other files across the
1157 library. Most notably,
1158 - 'openssl req' now has a '-newkey ecdsa:file' option;
1159 - EVP_PKCS82PKEY (crypto/evp/evp_pkey.c) now can handle ECDSA;
1160 - X509_PUBKEY_get (crypto/asn1/x_pubkey.c) and
1161 d2i_PublicKey (crypto/asn1/d2i_pu.c) have been modified to make
1162 them suitable for ECDSA where domain parameters must be
1163 extracted before the specific public key;
1164 - ECDSA engine support has been added.
1165 [Nils Larsch <nla@trustcenter.de>]
1166
1167 *) Include some named elliptic curves, and add OIDs from X9.62,
1168 SECG, and WAP/WTLS. Each curve can be obtained from the new
1169 function
1170 EC_GROUP_new_by_curve_name(),
1171 and the list of available named curves can be obtained with
1172 EC_get_builtin_curves().
1173 Also add a 'curve_name' member to EC_GROUP objects, which can be
1174 accessed via
1175 EC_GROUP_set_curve_name()
1176 EC_GROUP_get_curve_name()
1177 [Nils Larsch <larsch@trustcenter.de, Bodo Moeller]
1178
1179 *) Remove a few calls to bn_wexpand() in BN_sqr() (the one in there
1180 was actually never needed) and in BN_mul(). The removal in BN_mul()
1181 required a small change in bn_mul_part_recursive() and the addition
1182 of the functions bn_cmp_part_words(), bn_sub_part_words() and
1183 bn_add_part_words(), which do the same thing as bn_cmp_words(),
1184 bn_sub_words() and bn_add_words() except they take arrays with
1185 differing sizes.
1186 [Richard Levitte]
1187
1188 Changes between 0.9.7h and 0.9.7i [14 Oct 2005]
1189
1190 *) Wrapped the definition of EVP_MAX_MD_SIZE in a #ifdef OPENSSL_FIPS.
1191 The value now differs depending on if you build for FIPS or not.
1192 BEWARE! A program linked with a shared FIPSed libcrypto can't be
1193 safely run with a non-FIPSed libcrypto, as it may crash because of
1194 the difference induced by this change.
1195 [Andy Polyakov]
1196
1197 Changes between 0.9.7g and 0.9.7h [11 Oct 2005]
1198
1199 *) Remove the functionality of SSL_OP_MSIE_SSLV2_RSA_PADDING
1200 (part of SSL_OP_ALL). This option used to disable the
1201 countermeasure against man-in-the-middle protocol-version
1202 rollback in the SSL 2.0 server implementation, which is a bad
1203 idea. (CVE-2005-2969)
1204
1205 [Bodo Moeller; problem pointed out by Yutaka Oiwa (Research Center
1206 for Information Security, National Institute of Advanced Industrial
1207 Science and Technology [AIST], Japan)]
1208
1209 *) Minimal support for X9.31 signatures and PSS padding modes. This is
1210 mainly for FIPS compliance and not fully integrated at this stage.
1211 [Steve Henson]
1212
1213 *) For DSA signing, unless DSA_FLAG_NO_EXP_CONSTTIME is set, perform
1214 the exponentiation using a fixed-length exponent. (Otherwise,
1215 the information leaked through timing could expose the secret key
1216 after many signatures; cf. Bleichenbacher's attack on DSA with
1217 biased k.)
1218 [Bodo Moeller]
1219
1220 *) Make a new fixed-window mod_exp implementation the default for
1221 RSA, DSA, and DH private-key operations so that the sequence of
1222 squares and multiplies and the memory access pattern are
1223 independent of the particular secret key. This will mitigate
1224 cache-timing and potential related attacks.
1225
1226 BN_mod_exp_mont_consttime() is the new exponentiation implementation,
1227 and this is automatically used by BN_mod_exp_mont() if the new flag
1228 BN_FLG_EXP_CONSTTIME is set for the exponent. RSA, DSA, and DH
1229 will use this BN flag for private exponents unless the flag
1230 RSA_FLAG_NO_EXP_CONSTTIME, DSA_FLAG_NO_EXP_CONSTTIME, or
1231 DH_FLAG_NO_EXP_CONSTTIME, respectively, is set.
1232
1233 [Matthew D Wood (Intel Corp), with some changes by Bodo Moeller]
1234
1235 *) Change the client implementation for SSLv23_method() and
1236 SSLv23_client_method() so that is uses the SSL 3.0/TLS 1.0
1237 Client Hello message format if the SSL_OP_NO_SSLv2 option is set.
1238 (Previously, the SSL 2.0 backwards compatible Client Hello
1239 message format would be used even with SSL_OP_NO_SSLv2.)
1240 [Bodo Moeller]
1241
1242 *) Add support for smime-type MIME parameter in S/MIME messages which some
1243 clients need.
1244 [Steve Henson]
1245
1246 *) New function BN_MONT_CTX_set_locked() to set montgomery parameters in
1247 a threadsafe manner. Modify rsa code to use new function and add calls
1248 to dsa and dh code (which had race conditions before).
1249 [Steve Henson]
1250
1251 *) Include the fixed error library code in the C error file definitions
1252 instead of fixing them up at runtime. This keeps the error code
1253 structures constant.
1254 [Steve Henson]
1255
1256 Changes between 0.9.7f and 0.9.7g [11 Apr 2005]
1257
1258 [NB: OpenSSL 0.9.7h and later 0.9.7 patch levels were released after
1259 OpenSSL 0.9.8.]
1260
1261 *) Fixes for newer kerberos headers. NB: the casts are needed because
1262 the 'length' field is signed on one version and unsigned on another
1263 with no (?) obvious way to tell the difference, without these VC++
1264 complains. Also the "definition" of FAR (blank) is no longer included
1265 nor is the error ENOMEM. KRB5_PRIVATE has to be set to 1 to pick up
1266 some needed definitions.
1267 [Steve Henson]
1268
1269 *) Undo Cygwin change.
1270 [Ulf Möller]
1271
1272 *) Added support for proxy certificates according to RFC 3820.
1273 Because they may be a security thread to unaware applications,
1274 they must be explicitely allowed in run-time. See
1275 docs/HOWTO/proxy_certificates.txt for further information.
1276 [Richard Levitte]
1277
1278 Changes between 0.9.7e and 0.9.7f [22 Mar 2005]
1279
1280 *) Use (SSL_RANDOM_VALUE - 4) bytes of pseudo random data when generating
1281 server and client random values. Previously
1282 (SSL_RANDOM_VALUE - sizeof(time_t)) would be used which would result in
1283 less random data when sizeof(time_t) > 4 (some 64 bit platforms).
1284
1285 This change has negligible security impact because:
1286
1287 1. Server and client random values still have 24 bytes of pseudo random
1288 data.
1289
1290 2. Server and client random values are sent in the clear in the initial
1291 handshake.
1292
1293 3. The master secret is derived using the premaster secret (48 bytes in
1294 size for static RSA ciphersuites) as well as client server and random
1295 values.
1296
1297 The OpenSSL team would like to thank the UK NISCC for bringing this issue
1298 to our attention.
1299
1300 [Stephen Henson, reported by UK NISCC]
1301
1302 *) Use Windows randomness collection on Cygwin.
1303 [Ulf Möller]
1304
1305 *) Fix hang in EGD/PRNGD query when communication socket is closed
1306 prematurely by EGD/PRNGD.
1307 [Darren Tucker <dtucker@zip.com.au> via Lutz Jänicke, resolves #1014]
1308
1309 *) Prompt for pass phrases when appropriate for PKCS12 input format.
1310 [Steve Henson]
1311
1312 *) Back-port of selected performance improvements from development
1313 branch, as well as improved support for PowerPC platforms.
1314 [Andy Polyakov]
1315
1316 *) Add lots of checks for memory allocation failure, error codes to indicate
1317 failure and freeing up memory if a failure occurs.
1318 [Nauticus Networks SSL Team <openssl@nauticusnet.com>, Steve Henson]
1319
1320 *) Add new -passin argument to dgst.
1321 [Steve Henson]
1322
1323 *) Perform some character comparisons of different types in X509_NAME_cmp:
1324 this is needed for some certificates that reencode DNs into UTF8Strings
1325 (in violation of RFC3280) and can't or wont issue name rollover
1326 certificates.
1327 [Steve Henson]
1328
1329 *) Make an explicit check during certificate validation to see that
1330 the CA setting in each certificate on the chain is correct. As a
1331 side effect always do the following basic checks on extensions,
1332 not just when there's an associated purpose to the check:
1333
1334 - if there is an unhandled critical extension (unless the user
1335 has chosen to ignore this fault)
1336 - if the path length has been exceeded (if one is set at all)
1337 - that certain extensions fit the associated purpose (if one has
1338 been given)
1339 [Richard Levitte]
1340
1341 Changes between 0.9.7d and 0.9.7e [25 Oct 2004]
1342
1343 *) Avoid a race condition when CRLs are checked in a multi threaded
1344 environment. This would happen due to the reordering of the revoked
1345 entries during signature checking and serial number lookup. Now the
1346 encoding is cached and the serial number sort performed under a lock.
1347 Add new STACK function sk_is_sorted().
1348 [Steve Henson]
1349
1350 *) Add Delta CRL to the extension code.
1351 [Steve Henson]
1352
1353 *) Various fixes to s3_pkt.c so alerts are sent properly.
1354 [David Holmes <d.holmes@f5.com>]
1355
1356 *) Reduce the chances of duplicate issuer name and serial numbers (in
1357 violation of RFC3280) using the OpenSSL certificate creation utilities.
1358 This is done by creating a random 64 bit value for the initial serial
1359 number when a serial number file is created or when a self signed
1360 certificate is created using 'openssl req -x509'. The initial serial
1361 number file is created using 'openssl x509 -next_serial' in CA.pl
1362 rather than being initialized to 1.
1363 [Steve Henson]
1364
1365 Changes between 0.9.7c and 0.9.7d [17 Mar 2004]
1366
1367 *) Fix null-pointer assignment in do_change_cipher_spec() revealed
1368 by using the Codenomicon TLS Test Tool (CVE-2004-0079)
1369 [Joe Orton, Steve Henson]
1370
1371 *) Fix flaw in SSL/TLS handshaking when using Kerberos ciphersuites
1372 (CVE-2004-0112)
1373 [Joe Orton, Steve Henson]
1374
1375 *) Make it possible to have multiple active certificates with the same
1376 subject in the CA index file. This is done only if the keyword
1377 'unique_subject' is set to 'no' in the main CA section (default
1378 if 'CA_default') of the configuration file. The value is saved
1379 with the database itself in a separate index attribute file,
1380 named like the index file with '.attr' appended to the name.
1381 [Richard Levitte]
1382
1383 *) X509 verify fixes. Disable broken certificate workarounds when
1384 X509_V_FLAGS_X509_STRICT is set. Check CRL issuer has cRLSign set if
1385 keyUsage extension present. Don't accept CRLs with unhandled critical
1386 extensions: since verify currently doesn't process CRL extensions this
1387 rejects a CRL with *any* critical extensions. Add new verify error codes
1388 for these cases.
1389 [Steve Henson]
1390
1391 *) When creating an OCSP nonce use an OCTET STRING inside the extnValue.
1392 A clarification of RFC2560 will require the use of OCTET STRINGs and
1393 some implementations cannot handle the current raw format. Since OpenSSL
1394 copies and compares OCSP nonces as opaque blobs without any attempt at
1395 parsing them this should not create any compatibility issues.
1396 [Steve Henson]
1397
1398 *) New md flag EVP_MD_CTX_FLAG_REUSE this allows md_data to be reused when
1399 calling EVP_MD_CTX_copy_ex() to avoid calling OPENSSL_malloc(). Without
1400 this HMAC (and other) operations are several times slower than OpenSSL
1401 < 0.9.7.
1402 [Steve Henson]
1403
1404 *) Print out GeneralizedTime and UTCTime in ASN1_STRING_print_ex().
1405 [Peter Sylvester <Peter.Sylvester@EdelWeb.fr>]
1406
1407 *) Use the correct content when signing type "other".
1408 [Steve Henson]
1409
1410 Changes between 0.9.7b and 0.9.7c [30 Sep 2003]
1411
1412 *) Fix various bugs revealed by running the NISCC test suite:
1413
1414 Stop out of bounds reads in the ASN1 code when presented with
1415 invalid tags (CVE-2003-0543 and CVE-2003-0544).
1416
1417 Free up ASN1_TYPE correctly if ANY type is invalid (CVE-2003-0545).
1418
1419 If verify callback ignores invalid public key errors don't try to check
1420 certificate signature with the NULL public key.
1421
1422 [Steve Henson]
1423
1424 *) New -ignore_err option in ocsp application to stop the server
1425 exiting on the first error in a request.
1426 [Steve Henson]
1427
1428 *) In ssl3_accept() (ssl/s3_srvr.c) only accept a client certificate
1429 if the server requested one: as stated in TLS 1.0 and SSL 3.0
1430 specifications.
1431 [Steve Henson]
1432
1433 *) In ssl3_get_client_hello() (ssl/s3_srvr.c), tolerate additional
1434 extra data after the compression methods not only for TLS 1.0
1435 but also for SSL 3.0 (as required by the specification).
1436 [Bodo Moeller; problem pointed out by Matthias Loepfe]
1437
1438 *) Change X509_certificate_type() to mark the key as exported/exportable
1439 when it's 512 *bits* long, not 512 bytes.
1440 [Richard Levitte]
1441
1442 *) Change AES_cbc_encrypt() so it outputs exact multiple of
1443 blocks during encryption.
1444 [Richard Levitte]
1445
1446 *) Various fixes to base64 BIO and non blocking I/O. On write
1447 flushes were not handled properly if the BIO retried. On read
1448 data was not being buffered properly and had various logic bugs.
1449 This also affects blocking I/O when the data being decoded is a
1450 certain size.
1451 [Steve Henson]
1452
1453 *) Various S/MIME bugfixes and compatibility changes:
1454 output correct application/pkcs7 MIME type if
1455 PKCS7_NOOLDMIMETYPE is set. Tolerate some broken signatures.
1456 Output CR+LF for EOL if PKCS7_CRLFEOL is set (this makes opening
1457 of files as .eml work). Correctly handle very long lines in MIME
1458 parser.
1459 [Steve Henson]
1460
1461 Changes between 0.9.7a and 0.9.7b [10 Apr 2003]
1462
1463 *) Countermeasure against the Klima-Pokorny-Rosa extension of
1464 Bleichbacher's attack on PKCS #1 v1.5 padding: treat
1465 a protocol version number mismatch like a decryption error
1466 in ssl3_get_client_key_exchange (ssl/s3_srvr.c).
1467 [Bodo Moeller]
1468
1469 *) Turn on RSA blinding by default in the default implementation
1470 to avoid a timing attack. Applications that don't want it can call
1471 RSA_blinding_off() or use the new flag RSA_FLAG_NO_BLINDING.
1472 They would be ill-advised to do so in most cases.
1473 [Ben Laurie, Steve Henson, Geoff Thorpe, Bodo Moeller]
1474
1475 *) Change RSA blinding code so that it works when the PRNG is not
1476 seeded (in this case, the secret RSA exponent is abused as
1477 an unpredictable seed -- if it is not unpredictable, there
1478 is no point in blinding anyway). Make RSA blinding thread-safe
1479 by remembering the creator's thread ID in rsa->blinding and
1480 having all other threads use local one-time blinding factors
1481 (this requires more computation than sharing rsa->blinding, but
1482 avoids excessive locking; and if an RSA object is not shared
1483 between threads, blinding will still be very fast).
1484 [Bodo Moeller]
1485
1486 *) Fixed a typo bug that would cause ENGINE_set_default() to set an
1487 ENGINE as defaults for all supported algorithms irrespective of
1488 the 'flags' parameter. 'flags' is now honoured, so applications
1489 should make sure they are passing it correctly.
1490 [Geoff Thorpe]
1491
1492 *) Target "mingw" now allows native Windows code to be generated in
1493 the Cygwin environment as well as with the MinGW compiler.
1494 [Ulf Moeller]
1495
1496 Changes between 0.9.7 and 0.9.7a [19 Feb 2003]
1497
1498 *) In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked
1499 via timing by performing a MAC computation even if incorrrect
1500 block cipher padding has been found. This is a countermeasure
1501 against active attacks where the attacker has to distinguish
1502 between bad padding and a MAC verification error. (CVE-2003-0078)
1503
1504 [Bodo Moeller; problem pointed out by Brice Canvel (EPFL),
1505 Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and
1506 Martin Vuagnoux (EPFL, Ilion)]
1507
1508 *) Make the no-err option work as intended. The intention with no-err
1509 is not to have the whole error stack handling routines removed from
1510 libcrypto, it's only intended to remove all the function name and
1511 reason texts, thereby removing some of the footprint that may not
1512 be interesting if those errors aren't displayed anyway.
1513
1514 NOTE: it's still possible for any application or module to have it's
1515 own set of error texts inserted. The routines are there, just not
1516 used by default when no-err is given.
1517 [Richard Levitte]
1518
1519 *) Add support for FreeBSD on IA64.
1520 [dirk.meyer@dinoex.sub.org via Richard Levitte, resolves #454]
1521
1522 *) Adjust DES_cbc_cksum() so it returns the same value as the MIT
1523 Kerberos function mit_des_cbc_cksum(). Before this change,
1524 the value returned by DES_cbc_cksum() was like the one from
1525 mit_des_cbc_cksum(), except the bytes were swapped.
1526 [Kevin Greaney <Kevin.Greaney@hp.com> and Richard Levitte]
1527
1528 *) Allow an application to disable the automatic SSL chain building.
1529 Before this a rather primitive chain build was always performed in
1530 ssl3_output_cert_chain(): an application had no way to send the
1531 correct chain if the automatic operation produced an incorrect result.
1532
1533 Now the chain builder is disabled if either:
1534
1535 1. Extra certificates are added via SSL_CTX_add_extra_chain_cert().
1536
1537 2. The mode flag SSL_MODE_NO_AUTO_CHAIN is set.
1538
1539 The reasoning behind this is that an application would not want the
1540 auto chain building to take place if extra chain certificates are
1541 present and it might also want a means of sending no additional
1542 certificates (for example the chain has two certificates and the
1543 root is omitted).
1544 [Steve Henson]
1545
1546 *) Add the possibility to build without the ENGINE framework.
1547 [Steven Reddie <smr@essemer.com.au> via Richard Levitte]
1548
1549 *) Under Win32 gmtime() can return NULL: check return value in
1550 OPENSSL_gmtime(). Add error code for case where gmtime() fails.
1551 [Steve Henson]
1552
1553 *) DSA routines: under certain error conditions uninitialized BN objects
1554 could be freed. Solution: make sure initialization is performed early
1555 enough. (Reported and fix supplied by Ivan D Nestlerode <nestler@MIT.EDU>,
1556 Nils Larsch <nla@trustcenter.de> via PR#459)
1557 [Lutz Jaenicke]
1558
1559 *) Another fix for SSLv2 session ID handling: the session ID was incorrectly
1560 checked on reconnect on the client side, therefore session resumption
1561 could still fail with a "ssl session id is different" error. This
1562 behaviour is masked when SSL_OP_ALL is used due to
1563 SSL_OP_MICROSOFT_SESS_ID_BUG being set.
1564 Behaviour observed by Crispin Flowerday <crispin@flowerday.cx> as
1565 followup to PR #377.
1566 [Lutz Jaenicke]
1567
1568 *) IA-32 assembler support enhancements: unified ELF targets, support
1569 for SCO/Caldera platforms, fix for Cygwin shared build.
1570 [Andy Polyakov]
1571
1572 *) Add support for FreeBSD on sparc64. As a consequence, support for
1573 FreeBSD on non-x86 processors is separate from x86 processors on
1574 the config script, much like the NetBSD support.
1575 [Richard Levitte & Kris Kennaway <kris@obsecurity.org>]
1576
1577 Changes between 0.9.6h and 0.9.7 [31 Dec 2002]
1578
1579 [NB: OpenSSL 0.9.6i and later 0.9.6 patch levels were released after
1580 OpenSSL 0.9.7.]
1581
1582 *) Fix session ID handling in SSLv2 client code: the SERVER FINISHED
1583 code (06) was taken as the first octet of the session ID and the last
1584 octet was ignored consequently. As a result SSLv2 client side session
1585 caching could not have worked due to the session ID mismatch between
1586 client and server.
1587 Behaviour observed by Crispin Flowerday <crispin@flowerday.cx> as
1588 PR #377.
1589 [Lutz Jaenicke]
1590
1591 *) Change the declaration of needed Kerberos libraries to use EX_LIBS
1592 instead of the special (and badly supported) LIBKRB5. LIBKRB5 is
1593 removed entirely.
1594 [Richard Levitte]
1595
1596 *) The hw_ncipher.c engine requires dynamic locks. Unfortunately, it
1597 seems that in spite of existing for more than a year, many application
1598 author have done nothing to provide the necessary callbacks, which
1599 means that this particular engine will not work properly anywhere.
1600 This is a very unfortunate situation which forces us, in the name
1601 of usability, to give the hw_ncipher.c a static lock, which is part
1602 of libcrypto.
1603 NOTE: This is for the 0.9.7 series ONLY. This hack will never
1604 appear in 0.9.8 or later. We EXPECT application authors to have
1605 dealt properly with this when 0.9.8 is released (unless we actually
1606 make such changes in the libcrypto locking code that changes will
1607 have to be made anyway).
1608 [Richard Levitte]
1609
1610 *) In asn1_d2i_read_bio() repeatedly call BIO_read() until all content
1611 octets have been read, EOF or an error occurs. Without this change
1612 some truncated ASN1 structures will not produce an error.
1613 [Steve Henson]
1614
1615 *) Disable Heimdal support, since it hasn't been fully implemented.
1616 Still give the possibility to force the use of Heimdal, but with
1617 warnings and a request that patches get sent to openssl-dev.
1618 [Richard Levitte]
1619
1620 *) Add the VC-CE target, introduce the WINCE sysname, and add
1621 INSTALL.WCE and appropriate conditionals to make it build.
1622 [Steven Reddie <smr@essemer.com.au> via Richard Levitte]
1623
1624 *) Change the DLL names for Cygwin to cygcrypto-x.y.z.dll and
1625 cygssl-x.y.z.dll, where x, y and z are the major, minor and
1626 edit numbers of the version.
1627 [Corinna Vinschen <vinschen@redhat.com> and Richard Levitte]
1628
1629 *) Introduce safe string copy and catenation functions
1630 (BUF_strlcpy() and BUF_strlcat()).
1631 [Ben Laurie (CHATS) and Richard Levitte]
1632
1633 *) Avoid using fixed-size buffers for one-line DNs.
1634 [Ben Laurie (CHATS)]
1635
1636 *) Add BUF_MEM_grow_clean() to avoid information leakage when
1637 resizing buffers containing secrets, and use where appropriate.
1638 [Ben Laurie (CHATS)]
1639
1640 *) Avoid using fixed size buffers for configuration file location.
1641 [Ben Laurie (CHATS)]
1642
1643 *) Avoid filename truncation for various CA files.
1644 [Ben Laurie (CHATS)]
1645
1646 *) Use sizeof in preference to magic numbers.
1647 [Ben Laurie (CHATS)]
1648
1649 *) Avoid filename truncation in cert requests.
1650 [Ben Laurie (CHATS)]
1651
1652 *) Add assertions to check for (supposedly impossible) buffer
1653 overflows.
1654 [Ben Laurie (CHATS)]
1655
1656 *) Don't cache truncated DNS entries in the local cache (this could
1657 potentially lead to a spoofing attack).
1658 [Ben Laurie (CHATS)]
1659
1660 *) Fix various buffers to be large enough for hex/decimal
1661 representations in a platform independent manner.
1662 [Ben Laurie (CHATS)]
1663
1664 *) Add CRYPTO_realloc_clean() to avoid information leakage when
1665 resizing buffers containing secrets, and use where appropriate.
1666 [Ben Laurie (CHATS)]
1667
1668 *) Add BIO_indent() to avoid much slightly worrying code to do
1669 indents.
1670 [Ben Laurie (CHATS)]
1671
1672 *) Convert sprintf()/BIO_puts() to BIO_printf().
1673 [Ben Laurie (CHATS)]
1674
1675 *) buffer_gets() could terminate with the buffer only half
1676 full. Fixed.
1677 [Ben Laurie (CHATS)]
1678
1679 *) Add assertions to prevent user-supplied crypto functions from
1680 overflowing internal buffers by having large block sizes, etc.
1681 [Ben Laurie (CHATS)]
1682
1683 *) New OPENSSL_assert() macro (similar to assert(), but enabled
1684 unconditionally).
1685 [Ben Laurie (CHATS)]
1686
1687 *) Eliminate unused copy of key in RC4.
1688 [Ben Laurie (CHATS)]
1689
1690 *) Eliminate unused and incorrectly sized buffers for IV in pem.h.
1691 [Ben Laurie (CHATS)]
1692
1693 *) Fix off-by-one error in EGD path.
1694 [Ben Laurie (CHATS)]
1695
1696 *) If RANDFILE path is too long, ignore instead of truncating.
1697 [Ben Laurie (CHATS)]
1698
1699 *) Eliminate unused and incorrectly sized X.509 structure
1700 CBCParameter.
1701 [Ben Laurie (CHATS)]
1702
1703 *) Eliminate unused and dangerous function knumber().
1704 [Ben Laurie (CHATS)]
1705
1706 *) Eliminate unused and dangerous structure, KSSL_ERR.
1707 [Ben Laurie (CHATS)]
1708
1709 *) Protect against overlong session ID context length in an encoded
1710 session object. Since these are local, this does not appear to be
1711 exploitable.
1712 [Ben Laurie (CHATS)]
1713
1714 *) Change from security patch (see 0.9.6e below) that did not affect
1715 the 0.9.6 release series:
1716
1717 Remote buffer overflow in SSL3 protocol - an attacker could
1718 supply an oversized master key in Kerberos-enabled versions.
1719 (CVE-2002-0657)
1720 [Ben Laurie (CHATS)]
1721
1722 *) Change the SSL kerb5 codes to match RFC 2712.
1723 [Richard Levitte]
1724
1725 *) Make -nameopt work fully for req and add -reqopt switch.
1726 [Michael Bell <michael.bell@rz.hu-berlin.de>, Steve Henson]
1727
1728 *) The "block size" for block ciphers in CFB and OFB mode should be 1.
1729 [Steve Henson, reported by Yngve Nysaeter Pettersen <yngve@opera.com>]
1730
1731 *) Make sure tests can be performed even if the corresponding algorithms
1732 have been removed entirely. This was also the last step to make
1733 OpenSSL compilable with DJGPP under all reasonable conditions.
1734 [Richard Levitte, Doug Kaufman <dkaufman@rahul.net>]
1735
1736 *) Add cipher selection rules COMPLEMENTOFALL and COMPLEMENTOFDEFAULT
1737 to allow version independent disabling of normally unselected ciphers,
1738 which may be activated as a side-effect of selecting a single cipher.
1739
1740 (E.g., cipher list string "RSA" enables ciphersuites that are left
1741 out of "ALL" because they do not provide symmetric encryption.
1742 "RSA:!COMPLEMEMENTOFALL" avoids these unsafe ciphersuites.)
1743 [Lutz Jaenicke, Bodo Moeller]
1744
1745 *) Add appropriate support for separate platform-dependent build
1746 directories. The recommended way to make a platform-dependent
1747 build directory is the following (tested on Linux), maybe with
1748 some local tweaks:
1749
1750 # Place yourself outside of the OpenSSL source tree. In
1751 # this example, the environment variable OPENSSL_SOURCE
1752 # is assumed to contain the absolute OpenSSL source directory.
1753 mkdir -p objtree/"`uname -s`-`uname -r`-`uname -m`"
1754 cd objtree/"`uname -s`-`uname -r`-`uname -m`"
1755 (cd $OPENSSL_SOURCE; find . -type f) | while read F; do
1756 mkdir -p `dirname $F`
1757 ln -s $OPENSSL_SOURCE/$F $F
1758 done
1759
1760 To be absolutely sure not to disturb the source tree, a "make clean"
1761 is a good thing. If it isn't successfull, don't worry about it,
1762 it probably means the source directory is very clean.
1763 [Richard Levitte]
1764
1765 *) Make sure any ENGINE control commands make local copies of string
1766 pointers passed to them whenever necessary. Otherwise it is possible
1767 the caller may have overwritten (or deallocated) the original string
1768 data when a later ENGINE operation tries to use the stored values.
1769 [Götz Babin-Ebell <babinebell@trustcenter.de>]
1770
1771 *) Improve diagnostics in file reading and command-line digests.
1772 [Ben Laurie aided and abetted by Solar Designer <solar@openwall.com>]
1773
1774 *) Add AES modes CFB and OFB to the object database. Correct an
1775 error in AES-CFB decryption.
1776 [Richard Levitte]
1777
1778 *) Remove most calls to EVP_CIPHER_CTX_cleanup() in evp_enc.c, this
1779 allows existing EVP_CIPHER_CTX structures to be reused after
1780 calling EVP_*Final(). This behaviour is used by encryption
1781 BIOs and some applications. This has the side effect that
1782 applications must explicitly clean up cipher contexts with
1783 EVP_CIPHER_CTX_cleanup() or they will leak memory.
1784 [Steve Henson]
1785
1786 *) Check the values of dna and dnb in bn_mul_recursive before calling
1787 bn_mul_comba (a non zero value means the a or b arrays do not contain
1788 n2 elements) and fallback to bn_mul_normal if either is not zero.
1789 [Steve Henson]
1790
1791 *) Fix escaping of non-ASCII characters when using the -subj option
1792 of the "openssl req" command line tool. (Robert Joop <joop@fokus.gmd.de>)
1793 [Lutz Jaenicke]
1794
1795 *) Make object definitions compliant to LDAP (RFC2256): SN is the short
1796 form for "surname", serialNumber has no short form.
1797 Use "mail" as the short name for "rfc822Mailbox" according to RFC2798;
1798 therefore remove "mail" short name for "internet 7".
1799 The OID for unique identifiers in X509 certificates is
1800 x500UniqueIdentifier, not uniqueIdentifier.
1801 Some more OID additions. (Michael Bell <michael.bell@rz.hu-berlin.de>)
1802 [Lutz Jaenicke]
1803
1804 *) Add an "init" command to the ENGINE config module and auto initialize
1805 ENGINEs. Without any "init" command the ENGINE will be initialized
1806 after all ctrl commands have been executed on it. If init=1 the
1807 ENGINE is initailized at that point (ctrls before that point are run
1808 on the uninitialized ENGINE and after on the initialized one). If
1809 init=0 then the ENGINE will not be iniatialized at all.
1810 [Steve Henson]
1811
1812 *) Fix the 'app_verify_callback' interface so that the user-defined
1813 argument is actually passed to the callback: In the
1814 SSL_CTX_set_cert_verify_callback() prototype, the callback
1815 declaration has been changed from
1816 int (*cb)()
1817 into
1818 int (*cb)(X509_STORE_CTX *,void *);
1819 in ssl_verify_cert_chain (ssl/ssl_cert.c), the call
1820 i=s->ctx->app_verify_callback(&ctx)
1821 has been changed into
1822 i=s->ctx->app_verify_callback(&ctx, s->ctx->app_verify_arg).
1823
1824 To update applications using SSL_CTX_set_cert_verify_callback(),
1825 a dummy argument can be added to their callback functions.
1826 [D. K. Smetters <smetters@parc.xerox.com>]
1827
1828 *) Added the '4758cca' ENGINE to support IBM 4758 cards.
1829 [Maurice Gittens <maurice@gittens.nl>, touchups by Geoff Thorpe]
1830
1831 *) Add and OPENSSL_LOAD_CONF define which will cause
1832 OpenSSL_add_all_algorithms() to load the openssl.cnf config file.
1833 This allows older applications to transparently support certain
1834 OpenSSL features: such as crypto acceleration and dynamic ENGINE loading.
1835 Two new functions OPENSSL_add_all_algorithms_noconf() which will never
1836 load the config file and OPENSSL_add_all_algorithms_conf() which will
1837 always load it have also been added.
1838 [Steve Henson]
1839
1840 *) Add the OFB, CFB and CTR (all with 128 bit feedback) to AES.
1841 Adjust NIDs and EVP layer.
1842 [Stephen Sprunk <stephen@sprunk.org> and Richard Levitte]
1843
1844 *) Config modules support in openssl utility.
1845
1846 Most commands now load modules from the config file,
1847 though in a few (such as version) this isn't done
1848 because it couldn't be used for anything.
1849
1850 In the case of ca and req the config file used is
1851 the same as the utility itself: that is the -config
1852 command line option can be used to specify an
1853 alternative file.
1854 [Steve Henson]
1855
1856 *) Move default behaviour from OPENSSL_config(). If appname is NULL
1857 use "openssl_conf" if filename is NULL use default openssl config file.
1858 [Steve Henson]
1859
1860 *) Add an argument to OPENSSL_config() to allow the use of an alternative
1861 config section name. Add a new flag to tolerate a missing config file
1862 and move code to CONF_modules_load_file().
1863 [Steve Henson]
1864
1865 *) Support for crypto accelerator cards from Accelerated Encryption
1866 Processing, www.aep.ie. (Use engine 'aep')
1867 The support was copied from 0.9.6c [engine] and adapted/corrected
1868 to work with the new engine framework.
1869 [AEP Inc. and Richard Levitte]
1870
1871 *) Support for SureWare crypto accelerator cards from Baltimore
1872 Technologies. (Use engine 'sureware')
1873 The support was copied from 0.9.6c [engine] and adapted
1874 to work with the new engine framework.
1875 [Richard Levitte]
1876
1877 *) Have the CHIL engine fork-safe (as defined by nCipher) and actually
1878 make the newer ENGINE framework commands for the CHIL engine work.
1879 [Toomas Kiisk <vix@cyber.ee> and Richard Levitte]
1880
1881 *) Make it possible to produce shared libraries on ReliantUNIX.
1882 [Robert Dahlem <Robert.Dahlem@ffm2.siemens.de> via Richard Levitte]
1883
1884 *) Add the configuration target debug-linux-ppro.
1885 Make 'openssl rsa' use the general key loading routines
1886 implemented in apps.c, and make those routines able to
1887 handle the key format FORMAT_NETSCAPE and the variant
1888 FORMAT_IISSGC.
1889 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte]
1890
1891 *) Fix a crashbug and a logic bug in hwcrhk_load_pubkey().
1892 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte]
1893
1894 *) Add -keyform to rsautl, and document -engine.
1895 [Richard Levitte, inspired by Toomas Kiisk <vix@cyber.ee>]
1896
1897 *) Change BIO_new_file (crypto/bio/bss_file.c) to use new
1898 BIO_R_NO_SUCH_FILE error code rather than the generic
1899 ERR_R_SYS_LIB error code if fopen() fails with ENOENT.
1900 [Ben Laurie]
1901
1902 *) Add new functions
1903 ERR_peek_last_error
1904 ERR_peek_last_error_line
1905 ERR_peek_last_error_line_data.
1906 These are similar to
1907 ERR_peek_error
1908 ERR_peek_error_line
1909 ERR_peek_error_line_data,
1910 but report on the latest error recorded rather than the first one
1911 still in the error queue.
1912 [Ben Laurie, Bodo Moeller]
1913
1914 *) default_algorithms option in ENGINE config module. This allows things
1915 like:
1916 default_algorithms = ALL
1917 default_algorithms = RSA, DSA, RAND, CIPHERS, DIGESTS
1918 [Steve Henson]
1919
1920 *) Prelminary ENGINE config module.
1921 [Steve Henson]
1922
1923 *) New experimental application configuration code.
1924 [Steve Henson]
1925
1926 *) Change the AES code to follow the same name structure as all other
1927 symmetric ciphers, and behave the same way. Move everything to
1928 the directory crypto/aes, thereby obsoleting crypto/rijndael.
1929 [Stephen Sprunk <stephen@sprunk.org> and Richard Levitte]
1930
1931 *) SECURITY: remove unsafe setjmp/signal interaction from ui_openssl.c.
1932 [Ben Laurie and Theo de Raadt]
1933
1934 *) Add option to output public keys in req command.
1935 [Massimiliano Pala madwolf@openca.org]
1936
1937 *) Use wNAFs in EC_POINTs_mul() for improved efficiency
1938 (up to about 10% better than before for P-192 and P-224).
1939 [Bodo Moeller]
1940
1941 *) New functions/macros
1942
1943 SSL_CTX_set_msg_callback(ctx, cb)
1944 SSL_CTX_set_msg_callback_arg(ctx, arg)
1945 SSL_set_msg_callback(ssl, cb)
1946 SSL_set_msg_callback_arg(ssl, arg)
1947
1948 to request calling a callback function
1949
1950 void cb(int write_p, int version, int content_type,
1951 const void *buf, size_t len, SSL *ssl, void *arg)
1952
1953 whenever a protocol message has been completely received
1954 (write_p == 0) or sent (write_p == 1). Here 'version' is the
1955 protocol version according to which the SSL library interprets
1956 the current protocol message (SSL2_VERSION, SSL3_VERSION, or
1957 TLS1_VERSION). 'content_type' is 0 in the case of SSL 2.0, or
1958 the content type as defined in the SSL 3.0/TLS 1.0 protocol
1959 specification (change_cipher_spec(20), alert(21), handshake(22)).
1960 'buf' and 'len' point to the actual message, 'ssl' to the
1961 SSL object, and 'arg' is the application-defined value set by
1962 SSL[_CTX]_set_msg_callback_arg().
1963
1964 'openssl s_client' and 'openssl s_server' have new '-msg' options
1965 to enable a callback that displays all protocol messages.
1966 [Bodo Moeller]
1967
1968 *) Change the shared library support so shared libraries are built as
1969 soon as the corresponding static library is finished, and thereby get
1970 openssl and the test programs linked against the shared library.
1971 This still only happens when the keyword "shard" has been given to
1972 the configuration scripts.
1973
1974 NOTE: shared library support is still an experimental thing, and
1975 backward binary compatibility is still not guaranteed.
1976 ["Maciej W. Rozycki" <macro@ds2.pg.gda.pl> and Richard Levitte]
1977
1978 *) Add support for Subject Information Access extension.
1979 [Peter Sylvester <Peter.Sylvester@EdelWeb.fr>]
1980
1981 *) Make BUF_MEM_grow() behaviour more consistent: Initialise to zero
1982 additional bytes when new memory had to be allocated, not just
1983 when reusing an existing buffer.
1984 [Bodo Moeller]
1985
1986 *) New command line and configuration option 'utf8' for the req command.
1987 This allows field values to be specified as UTF8 strings.
1988 [Steve Henson]
1989
1990 *) Add -multi and -mr options to "openssl speed" - giving multiple parallel
1991 runs for the former and machine-readable output for the latter.
1992 [Ben Laurie]
1993
1994 *) Add '-noemailDN' option to 'openssl ca'. This prevents inclusion
1995 of the e-mail address in the DN (i.e., it will go into a certificate
1996 extension only). The new configuration file option 'email_in_dn = no'
1997 has the same effect.
1998 [Massimiliano Pala madwolf@openca.org]
1999
2000 *) Change all functions with names starting with des_ to be starting
2001 with DES_ instead. Add wrappers that are compatible with libdes,
2002 but are named _ossl_old_des_*. Finally, add macros that map the
2003 des_* symbols to the corresponding _ossl_old_des_* if libdes
2004 compatibility is desired. If OpenSSL 0.9.6c compatibility is
2005 desired, the des_* symbols will be mapped to DES_*, with one
2006 exception.
2007
2008 Since we provide two compatibility mappings, the user needs to
2009 define the macro OPENSSL_DES_LIBDES_COMPATIBILITY if libdes
2010 compatibility is desired. The default (i.e., when that macro
2011 isn't defined) is OpenSSL 0.9.6c compatibility.
2012
2013 There are also macros that enable and disable the support of old
2014 des functions altogether. Those are OPENSSL_ENABLE_OLD_DES_SUPPORT
2015 and OPENSSL_DISABLE_OLD_DES_SUPPORT. If none or both of those
2016 are defined, the default will apply: to support the old des routines.
2017
2018 In either case, one must include openssl/des.h to get the correct
2019 definitions. Do not try to just include openssl/des_old.h, that
2020 won't work.
2021
2022 NOTE: This is a major break of an old API into a new one. Software
2023 authors are encouraged to switch to the DES_ style functions. Some
2024 time in the future, des_old.h and the libdes compatibility functions
2025 will be disable (i.e. OPENSSL_DISABLE_OLD_DES_SUPPORT will be the
2026 default), and then completely removed.
2027 [Richard Levitte]
2028
2029 *) Test for certificates which contain unsupported critical extensions.
2030 If such a certificate is found during a verify operation it is
2031 rejected by default: this behaviour can be overridden by either
2032 handling the new error X509_V_ERR_UNHANDLED_CRITICAL_EXTENSION or
2033 by setting the verify flag X509_V_FLAG_IGNORE_CRITICAL. A new function
2034 X509_supported_extension() has also been added which returns 1 if a
2035 particular extension is supported.
2036 [Steve Henson]
2037
2038 *) Modify the behaviour of EVP cipher functions in similar way to digests
2039 to retain compatibility with existing code.
2040 [Steve Henson]
2041
2042 *) Modify the behaviour of EVP_DigestInit() and EVP_DigestFinal() to retain
2043 compatibility with existing code. In particular the 'ctx' parameter does
2044 not have to be to be initialized before the call to EVP_DigestInit() and
2045 it is tidied up after a call to EVP_DigestFinal(). New function
2046 EVP_DigestFinal_ex() which does not tidy up the ctx. Similarly function
2047 EVP_MD_CTX_copy() changed to not require the destination to be
2048 initialized valid and new function EVP_MD_CTX_copy_ex() added which
2049 requires the destination to be valid.
2050
2051 Modify all the OpenSSL digest calls to use EVP_DigestInit_ex(),
2052 EVP_DigestFinal_ex() and EVP_MD_CTX_copy_ex().
2053 [Steve Henson]
2054
2055 *) Change ssl3_get_message (ssl/s3_both.c) and the functions using it
2056 so that complete 'Handshake' protocol structures are kept in memory
2057 instead of overwriting 'msg_type' and 'length' with 'body' data.
2058 [Bodo Moeller]
2059
2060 *) Add an implementation of SSL_add_dir_cert_subjects_to_stack for Win32.
2061 [Massimo Santin via Richard Levitte]
2062
2063 *) Major restructuring to the underlying ENGINE code. This includes
2064 reduction of linker bloat, separation of pure "ENGINE" manipulation
2065 (initialisation, etc) from functionality dealing with implementations
2066 of specific crypto iterfaces. This change also introduces integrated
2067 support for symmetric ciphers and digest implementations - so ENGINEs
2068 can now accelerate these by providing EVP_CIPHER and EVP_MD
2069 implementations of their own. This is detailed in crypto/engine/README
2070 as it couldn't be adequately described here. However, there are a few
2071 API changes worth noting - some RSA, DSA, DH, and RAND functions that
2072 were changed in the original introduction of ENGINE code have now
2073 reverted back - the hooking from this code to ENGINE is now a good
2074 deal more passive and at run-time, operations deal directly with
2075 RSA_METHODs, DSA_METHODs (etc) as they did before, rather than
2076 dereferencing through an ENGINE pointer any more. Also, the ENGINE
2077 functions dealing with BN_MOD_EXP[_CRT] handlers have been removed -
2078 they were not being used by the framework as there is no concept of a
2079 BIGNUM_METHOD and they could not be generalised to the new
2080 'ENGINE_TABLE' mechanism that underlies the new code. Similarly,
2081 ENGINE_cpy() has been removed as it cannot be consistently defined in
2082 the new code.
2083 [Geoff Thorpe]
2084
2085 *) Change ASN1_GENERALIZEDTIME_check() to allow fractional seconds.
2086 [Steve Henson]
2087
2088 *) Change mkdef.pl to sort symbols that get the same entry number,
2089 and make sure the automatically generated functions ERR_load_*
2090 become part of libeay.num as well.
2091 [Richard Levitte]
2092
2093 *) New function SSL_renegotiate_pending(). This returns true once
2094 renegotiation has been requested (either SSL_renegotiate() call
2095 or HelloRequest/ClientHello receveived from the peer) and becomes
2096 false once a handshake has been completed.
2097 (For servers, SSL_renegotiate() followed by SSL_do_handshake()
2098 sends a HelloRequest, but does not ensure that a handshake takes
2099 place. SSL_renegotiate_pending() is useful for checking if the
2100 client has followed the request.)
2101 [Bodo Moeller]
2102
2103 *) New SSL option SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION.
2104 By default, clients may request session resumption even during
2105 renegotiation (if session ID contexts permit); with this option,
2106 session resumption is possible only in the first handshake.
2107
2108 SSL_OP_ALL is now 0x00000FFFL instead of 0x000FFFFFL. This makes
2109 more bits available for options that should not be part of
2110 SSL_OP_ALL (such as SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION).
2111 [Bodo Moeller]
2112
2113 *) Add some demos for certificate and certificate request creation.
2114 [Steve Henson]
2115
2116 *) Make maximum certificate chain size accepted from the peer application
2117 settable (SSL*_get/set_max_cert_list()), as proposed by
2118 "Douglas E. Engert" <deengert@anl.gov>.
2119 [Lutz Jaenicke]
2120
2121 *) Add support for shared libraries for Unixware-7
2122 (Boyd Lynn Gerber <gerberb@zenez.com>).
2123 [Lutz Jaenicke]
2124
2125 *) Add a "destroy" handler to ENGINEs that allows structural cleanup to
2126 be done prior to destruction. Use this to unload error strings from
2127 ENGINEs that load their own error strings. NB: This adds two new API
2128 functions to "get" and "set" this destroy handler in an ENGINE.
2129 [Geoff Thorpe]
2130
2131 *) Alter all existing ENGINE implementations (except "openssl" and
2132 "openbsd") to dynamically instantiate their own error strings. This
2133 makes them more flexible to be built both as statically-linked ENGINEs
2134 and self-contained shared-libraries loadable via the "dynamic" ENGINE.
2135 Also, add stub code to each that makes building them as self-contained
2136 shared-libraries easier (see README.ENGINE).
2137 [Geoff Thorpe]
2138
2139 *) Add a "dynamic" ENGINE that provides a mechanism for binding ENGINE
2140 implementations into applications that are completely implemented in
2141 self-contained shared-libraries. The "dynamic" ENGINE exposes control
2142 commands that can be used to configure what shared-library to load and
2143 to control aspects of the way it is handled. Also, made an update to
2144 the README.ENGINE file that brings its information up-to-date and
2145 provides some information and instructions on the "dynamic" ENGINE
2146 (ie. how to use it, how to build "dynamic"-loadable ENGINEs, etc).
2147 [Geoff Thorpe]
2148
2149 *) Make it possible to unload ranges of ERR strings with a new
2150 "ERR_unload_strings" function.
2151 [Geoff Thorpe]
2152
2153 *) Add a copy() function to EVP_MD.
2154 [Ben Laurie]
2155
2156 *) Make EVP_MD routines take a context pointer instead of just the
2157 md_data void pointer.
2158 [Ben Laurie]
2159
2160 *) Add flags to EVP_MD and EVP_MD_CTX. EVP_MD_FLAG_ONESHOT indicates
2161 that the digest can only process a single chunk of data
2162 (typically because it is provided by a piece of
2163 hardware). EVP_MD_CTX_FLAG_ONESHOT indicates that the application
2164 is only going to provide a single chunk of data, and hence the
2165 framework needn't accumulate the data for oneshot drivers.
2166 [Ben Laurie]
2167
2168 *) As with "ERR", make it possible to replace the underlying "ex_data"
2169 functions. This change also alters the storage and management of global
2170 ex_data state - it's now all inside ex_data.c and all "class" code (eg.
2171 RSA, BIO, SSL_CTX, etc) no longer stores its own STACKS and per-class
2172 index counters. The API functions that use this state have been changed
2173 to take a "class_index" rather than pointers to the class's local STACK
2174 and counter, and there is now an API function to dynamically create new
2175 classes. This centralisation allows us to (a) plug a lot of the
2176 thread-safety problems that existed, and (b) makes it possible to clean
2177 up all allocated state using "CRYPTO_cleanup_all_ex_data()". W.r.t. (b)
2178 such data would previously have always leaked in application code and
2179 workarounds were in place to make the memory debugging turn a blind eye
2180 to it. Application code that doesn't use this new function will still
2181 leak as before, but their memory debugging output will announce it now
2182 rather than letting it slide.
2183
2184 Besides the addition of CRYPTO_cleanup_all_ex_data(), another API change
2185 induced by the "ex_data" overhaul is that X509_STORE_CTX_init() now
2186 has a return value to indicate success or failure.
2187 [Geoff Thorpe]
2188
2189 *) Make it possible to replace the underlying "ERR" functions such that the
2190 global state (2 LHASH tables and 2 locks) is only used by the "default"
2191 implementation. This change also adds two functions to "get" and "set"
2192 the implementation prior to it being automatically set the first time
2193 any other ERR function takes place. Ie. an application can call "get",
2194 pass the return value to a module it has just loaded, and that module
2195 can call its own "set" function using that value. This means the
2196 module's "ERR" operations will use (and modify) the error state in the
2197 application and not in its own statically linked copy of OpenSSL code.
2198 [Geoff Thorpe]
2199
2200 *) Give DH, DSA, and RSA types their own "**_up_ref()" function to increment
2201 reference counts. This performs normal REF_PRINT/REF_CHECK macros on
2202 the operation, and provides a more encapsulated way for external code
2203 (crypto/evp/ and ssl/) to do this. Also changed the evp and ssl code
2204 to use these functions rather than manually incrementing the counts.
2205
2206 Also rename "DSO_up()" function to more descriptive "DSO_up_ref()".
2207 [Geoff Thorpe]
2208
2209 *) Add EVP test program.
2210 [Ben Laurie]
2211
2212 *) Add symmetric cipher support to ENGINE. Expect the API to change!
2213 [Ben Laurie]
2214
2215 *) New CRL functions: X509_CRL_set_version(), X509_CRL_set_issuer_name()
2216 X509_CRL_set_lastUpdate(), X509_CRL_set_nextUpdate(), X509_CRL_sort(),
2217 X509_REVOKED_set_serialNumber(), and X509_REVOKED_set_revocationDate().
2218 These allow a CRL to be built without having to access X509_CRL fields
2219 directly. Modify 'ca' application to use new functions.
2220 [Steve Henson]
2221
2222 *) Move SSL_OP_TLS_ROLLBACK_BUG out of the SSL_OP_ALL list of recommended
2223 bug workarounds. Rollback attack detection is a security feature.
2224 The problem will only arise on OpenSSL servers when TLSv1 is not
2225 available (sslv3_server_method() or SSL_OP_NO_TLSv1).
2226 Software authors not wanting to support TLSv1 will have special reasons
2227 for their choice and can explicitly enable this option.
2228 [Bodo Moeller, Lutz Jaenicke]
2229
2230 *) Rationalise EVP so it can be extended: don't include a union of
2231 cipher/digest structures, add init/cleanup functions for EVP_MD_CTX
2232 (similar to those existing for EVP_CIPHER_CTX).
2233 Usage example:
2234
2235 EVP_MD_CTX md;
2236
2237 EVP_MD_CTX_init(&md); /* new function call */
2238 EVP_DigestInit(&md, EVP_sha1());
2239 EVP_DigestUpdate(&md, in, len);
2240 EVP_DigestFinal(&md, out, NULL);
2241 EVP_MD_CTX_cleanup(&md); /* new function call */
2242
2243 [Ben Laurie]
2244
2245 *) Make DES key schedule conform to the usual scheme, as well as
2246 correcting its structure. This means that calls to DES functions
2247 now have to pass a pointer to a des_key_schedule instead of a
2248 plain des_key_schedule (which was actually always a pointer
2249 anyway): E.g.,
2250
2251 des_key_schedule ks;
2252
2253 des_set_key_checked(..., &ks);
2254 des_ncbc_encrypt(..., &ks, ...);
2255
2256 (Note that a later change renames 'des_...' into 'DES_...'.)
2257 [Ben Laurie]
2258
2259 *) Initial reduction of linker bloat: the use of some functions, such as
2260 PEM causes large amounts of unused functions to be linked in due to
2261 poor organisation. For example pem_all.c contains every PEM function
2262 which has a knock on effect of linking in large amounts of (unused)
2263 ASN1 code. Grouping together similar functions and splitting unrelated
2264 functions prevents this.
2265 [Steve Henson]
2266
2267 *) Cleanup of EVP macros.
2268 [Ben Laurie]
2269
2270 *) Change historical references to {NID,SN,LN}_des_ede and ede3 to add the
2271 correct _ecb suffix.
2272 [Ben Laurie]
2273
2274 *) Add initial OCSP responder support to ocsp application. The
2275 revocation information is handled using the text based index
2276 use by the ca application. The responder can either handle
2277 requests generated internally, supplied in files (for example
2278 via a CGI script) or using an internal minimal server.
2279 [Steve Henson]
2280
2281 *) Add configuration choices to get zlib compression for TLS.
2282 [Richard Levitte]
2283
2284 *) Changes to Kerberos SSL for RFC 2712 compliance:
2285 1. Implemented real KerberosWrapper, instead of just using
2286 KRB5 AP_REQ message. [Thanks to Simon Wilkinson <sxw@sxw.org.uk>]
2287 2. Implemented optional authenticator field of KerberosWrapper.
2288
2289 Added openssl-style ASN.1 macros for Kerberos ticket, ap_req,
2290 and authenticator structs; see crypto/krb5/.
2291
2292 Generalized Kerberos calls to support multiple Kerberos libraries.
2293 [Vern Staats <staatsvr@asc.hpc.mil>,
2294 Jeffrey Altman <jaltman@columbia.edu>
2295 via Richard Levitte]
2296
2297 *) Cause 'openssl speed' to use fully hard-coded DSA keys as it
2298 already does with RSA. testdsa.h now has 'priv_key/pub_key'
2299 values for each of the key sizes rather than having just
2300 parameters (and 'speed' generating keys each time).
2301 [Geoff Thorpe]
2302
2303 *) Speed up EVP routines.
2304 Before:
2305 encrypt
2306 type 8 bytes 64 bytes 256 bytes 1024 bytes 8192 bytes
2307 des-cbc 4408.85k 5560.51k 5778.46k 5862.20k 5825.16k
2308 des-cbc 4389.55k 5571.17k 5792.23k 5846.91k 5832.11k
2309 des-cbc 4394.32k 5575.92k 5807.44k 5848.37k 5841.30k
2310 decrypt
2311 des-cbc 3482.66k 5069.49k 5496.39k 5614.16k 5639.28k
2312 des-cbc 3480.74k 5068.76k 5510.34k 5609.87k 5635.52k
2313 des-cbc 3483.72k 5067.62k 5504.60k 5708.01k 5724.80k
2314 After:
2315 encrypt
2316 des-cbc 4660.16k 5650.19k 5807.19k 5827.13k 5783.32k
2317 decrypt
2318 des-cbc 3624.96k 5258.21k 5530.91k 5624.30k 5628.26k
2319 [Ben Laurie]
2320
2321 *) Added the OS2-EMX target.
2322 ["Brian Havard" <brianh@kheldar.apana.org.au> and Richard Levitte]
2323
2324 *) Rewrite apps to use NCONF routines instead of the old CONF. New functions
2325 to support NCONF routines in extension code. New function CONF_set_nconf()
2326 to allow functions which take an NCONF to also handle the old LHASH
2327 structure: this means that the old CONF compatible routines can be
2328 retained (in particular wrt extensions) without having to duplicate the
2329 code. New function X509V3_add_ext_nconf_sk to add extensions to a stack.
2330 [Steve Henson]
2331
2332 *) Enhance the general user interface with mechanisms for inner control
2333 and with possibilities to have yes/no kind of prompts.
2334 [Richard Levitte]
2335
2336 *) Change all calls to low level digest routines in the library and
2337 applications to use EVP. Add missing calls to HMAC_cleanup() and
2338 don't assume HMAC_CTX can be copied using memcpy().
2339 [Verdon Walker <VWalker@novell.com>, Steve Henson]
2340
2341 *) Add the possibility to control engines through control names but with
2342 arbitrary arguments instead of just a string.
2343 Change the key loaders to take a UI_METHOD instead of a callback
2344 function pointer. NOTE: this breaks binary compatibility with earlier
2345 versions of OpenSSL [engine].
2346 Adapt the nCipher code for these new conditions and add a card insertion
2347 callback.
2348 [Richard Levitte]
2349
2350 *) Enhance the general user interface with mechanisms to better support
2351 dialog box interfaces, application-defined prompts, the possibility
2352 to use defaults (for example default passwords from somewhere else)
2353 and interrupts/cancellations.
2354 [Richard Levitte]
2355
2356 *) Tidy up PKCS#12 attribute handling. Add support for the CSP name
2357 attribute in PKCS#12 files, add new -CSP option to pkcs12 utility.
2358 [Steve Henson]
2359
2360 *) Fix a memory leak in 'sk_dup()' in the case reallocation fails. (Also
2361 tidy up some unnecessarily weird code in 'sk_new()').
2362 [Geoff, reported by Diego Tartara <dtartara@novamens.com>]
2363
2364 *) Change the key loading routines for ENGINEs to use the same kind
2365 callback (pem_password_cb) as all other routines that need this
2366 kind of callback.
2367 [Richard Levitte]
2368
2369 *) Increase ENTROPY_NEEDED to 32 bytes, as Rijndael can operate with
2370 256 bit (=32 byte) keys. Of course seeding with more entropy bytes
2371 than this minimum value is recommended.
2372 [Lutz Jaenicke]
2373
2374 *) New random seeder for OpenVMS, using the system process statistics
2375 that are easily reachable.
2376 [Richard Levitte]
2377
2378 *) Windows apparently can't transparently handle global
2379 variables defined in DLLs. Initialisations such as:
2380
2381 const ASN1_ITEM *it = &ASN1_INTEGER_it;
2382
2383 wont compile. This is used by the any applications that need to
2384 declare their own ASN1 modules. This was fixed by adding the option
2385 EXPORT_VAR_AS_FN to all Win32 platforms, although this isn't strictly
2386 needed for static libraries under Win32.
2387 [Steve Henson]
2388
2389 *) New functions X509_PURPOSE_set() and X509_TRUST_set() to handle
2390 setting of purpose and trust fields. New X509_STORE trust and
2391 purpose functions and tidy up setting in other SSL functions.
2392 [Steve Henson]
2393
2394 *) Add copies of X509_STORE_CTX fields and callbacks to X509_STORE
2395 structure. These are inherited by X509_STORE_CTX when it is
2396 initialised. This allows various defaults to be set in the
2397 X509_STORE structure (such as flags for CRL checking and custom
2398 purpose or trust settings) for functions which only use X509_STORE_CTX
2399 internally such as S/MIME.
2400
2401 Modify X509_STORE_CTX_purpose_inherit() so it only sets purposes and
2402 trust settings if they are not set in X509_STORE. This allows X509_STORE
2403 purposes and trust (in S/MIME for example) to override any set by default.
2404
2405 Add command line options for CRL checking to smime, s_client and s_server
2406 applications.
2407 [Steve Henson]
2408
2409 *) Initial CRL based revocation checking. If the CRL checking flag(s)
2410 are set then the CRL is looked up in the X509_STORE structure and
2411 its validity and signature checked, then if the certificate is found
2412 in the CRL the verify fails with a revoked error.
2413
2414 Various new CRL related callbacks added to X509_STORE_CTX structure.
2415
2416 Command line options added to 'verify' application to support this.
2417
2418 This needs some additional work, such as being able to handle multiple
2419 CRLs with different times, extension based lookup (rather than just
2420 by subject name) and ultimately more complete V2 CRL extension
2421 handling.
2422 [Steve Henson]
2423
2424 *) Add a general user interface API (crypto/ui/). This is designed
2425 to replace things like des_read_password and friends (backward
2426 compatibility functions using this new API are provided).
2427 The purpose is to remove prompting functions from the DES code
2428 section as well as provide for prompting through dialog boxes in
2429 a window system and the like.
2430 [Richard Levitte]
2431
2432 *) Add "ex_data" support to ENGINE so implementations can add state at a
2433 per-structure level rather than having to store it globally.
2434 [Geoff]
2435
2436 *) Make it possible for ENGINE structures to be copied when retrieved by
2437 ENGINE_by_id() if the ENGINE specifies a new flag: ENGINE_FLAGS_BY_ID_COPY.
2438 This causes the "original" ENGINE structure to act like a template,
2439 analogous to the RSA vs. RSA_METHOD type of separation. Because of this
2440 operational state can be localised to each ENGINE structure, despite the
2441 fact they all share the same "methods". New ENGINE structures returned in
2442 this case have no functional references and the return value is the single
2443 structural reference. This matches the single structural reference returned
2444 by ENGINE_by_id() normally, when it is incremented on the pre-existing
2445 ENGINE structure.
2446 [Geoff]
2447
2448 *) Fix ASN1 decoder when decoding type ANY and V_ASN1_OTHER: since this
2449 needs to match any other type at all we need to manually clear the
2450 tag cache.
2451 [Steve Henson]
2452
2453 *) Changes to the "openssl engine" utility to include;
2454 - verbosity levels ('-v', '-vv', and '-vvv') that provide information
2455 about an ENGINE's available control commands.
2456 - executing control commands from command line arguments using the
2457 '-pre' and '-post' switches. '-post' is only used if '-t' is
2458 specified and the ENGINE is successfully initialised. The syntax for
2459 the individual commands are colon-separated, for example;
2460 openssl engine chil -pre FORK_CHECK:0 -pre SO_PATH:/lib/test.so
2461 [Geoff]
2462
2463 *) New dynamic control command support for ENGINEs. ENGINEs can now
2464 declare their own commands (numbers), names (strings), descriptions,
2465 and input types for run-time discovery by calling applications. A
2466 subset of these commands are implicitly classed as "executable"
2467 depending on their input type, and only these can be invoked through
2468 the new string-based API function ENGINE_ctrl_cmd_string(). (Eg. this
2469 can be based on user input, config files, etc). The distinction is
2470 that "executable" commands cannot return anything other than a boolean
2471 result and can only support numeric or string input, whereas some
2472 discoverable commands may only be for direct use through
2473 ENGINE_ctrl(), eg. supporting the exchange of binary data, function
2474 pointers, or other custom uses. The "executable" commands are to
2475 support parameterisations of ENGINE behaviour that can be
2476 unambiguously defined by ENGINEs and used consistently across any
2477 OpenSSL-based application. Commands have been added to all the
2478 existing hardware-supporting ENGINEs, noticeably "SO_PATH" to allow
2479 control over shared-library paths without source code alterations.
2480 [Geoff]
2481
2482 *) Changed all ENGINE implementations to dynamically allocate their
2483 ENGINEs rather than declaring them statically. Apart from this being
2484 necessary with the removal of the ENGINE_FLAGS_MALLOCED distinction,
2485 this also allows the implementations to compile without using the
2486 internal engine_int.h header.
2487 [Geoff]
2488
2489 *) Minor adjustment to "rand" code. RAND_get_rand_method() now returns a
2490 'const' value. Any code that should be able to modify a RAND_METHOD
2491 should already have non-const pointers to it (ie. they should only
2492 modify their own ones).
2493 [Geoff]
2494
2495 *) Made a variety of little tweaks to the ENGINE code.
2496 - "atalla" and "ubsec" string definitions were moved from header files
2497 to C code. "nuron" string definitions were placed in variables
2498 rather than hard-coded - allowing parameterisation of these values
2499 later on via ctrl() commands.
2500 - Removed unused "#if 0"'d code.
2501 - Fixed engine list iteration code so it uses ENGINE_free() to release
2502 structural references.
2503 - Constified the RAND_METHOD element of ENGINE structures.
2504 - Constified various get/set functions as appropriate and added
2505 missing functions (including a catch-all ENGINE_cpy that duplicates
2506 all ENGINE values onto a new ENGINE except reference counts/state).
2507 - Removed NULL parameter checks in get/set functions. Setting a method
2508 or function to NULL is a way of cancelling out a previously set
2509 value. Passing a NULL ENGINE parameter is just plain stupid anyway
2510 and doesn't justify the extra error symbols and code.
2511 - Deprecate the ENGINE_FLAGS_MALLOCED define and move the area for
2512 flags from engine_int.h to engine.h.
2513 - Changed prototypes for ENGINE handler functions (init(), finish(),
2514 ctrl(), key-load functions, etc) to take an (ENGINE*) parameter.
2515 [Geoff]
2516
2517 *) Implement binary inversion algorithm for BN_mod_inverse in addition
2518 to the algorithm using long division. The binary algorithm can be
2519 used only if the modulus is odd. On 32-bit systems, it is faster
2520 only for relatively small moduli (roughly 20-30% for 128-bit moduli,
2521 roughly 5-15% for 256-bit moduli), so we use it only for moduli
2522 up to 450 bits. In 64-bit environments, the binary algorithm
2523 appears to be advantageous for much longer moduli; here we use it
2524 for moduli up to 2048 bits.
2525 [Bodo Moeller]
2526
2527 *) Rewrite CHOICE field setting in ASN1_item_ex_d2i(). The old code
2528 could not support the combine flag in choice fields.
2529 [Steve Henson]
2530
2531 *) Add a 'copy_extensions' option to the 'ca' utility. This copies
2532 extensions from a certificate request to the certificate.
2533 [Steve Henson]
2534
2535 *) Allow multiple 'certopt' and 'nameopt' options to be separated
2536 by commas. Add 'namopt' and 'certopt' options to the 'ca' config
2537 file: this allows the display of the certificate about to be
2538 signed to be customised, to allow certain fields to be included
2539 or excluded and extension details. The old system didn't display
2540 multicharacter strings properly, omitted fields not in the policy
2541 and couldn't display additional details such as extensions.
2542 [Steve Henson]
2543
2544 *) Function EC_POINTs_mul for multiple scalar multiplication
2545 of an arbitrary number of elliptic curve points
2546 \sum scalars[i]*points[i],
2547 optionally including the generator defined for the EC_GROUP:
2548 scalar*generator + \sum scalars[i]*points[i].
2549
2550 EC_POINT_mul is a simple wrapper function for the typical case
2551 that the point list has just one item (besides the optional
2552 generator).
2553 [Bodo Moeller]
2554
2555 *) First EC_METHODs for curves over GF(p):
2556
2557 EC_GFp_simple_method() uses the basic BN_mod_mul and BN_mod_sqr
2558 operations and provides various method functions that can also
2559 operate with faster implementations of modular arithmetic.
2560
2561 EC_GFp_mont_method() reuses most functions that are part of
2562 EC_GFp_simple_method, but uses Montgomery arithmetic.
2563
2564 [Bodo Moeller; point addition and point doubling
2565 implementation directly derived from source code provided by
2566 Lenka Fibikova <fibikova@exp-math.uni-essen.de>]
2567
2568 *) Framework for elliptic curves (crypto/ec/ec.h, crypto/ec/ec_lcl.h,
2569 crypto/ec/ec_lib.c):
2570
2571 Curves are EC_GROUP objects (with an optional group generator)
2572 based on EC_METHODs that are built into the library.
2573
2574 Points are EC_POINT objects based on EC_GROUP objects.
2575
2576 Most of the framework would be able to handle curves over arbitrary
2577 finite fields, but as there are no obvious types for fields other
2578 than GF(p), some functions are limited to that for now.
2579 [Bodo Moeller]
2580
2581 *) Add the -HTTP option to s_server. It is similar to -WWW, but requires
2582 that the file contains a complete HTTP response.
2583 [Richard Levitte]
2584
2585 *) Add the ec directory to mkdef.pl and mkfiles.pl. In mkdef.pl
2586 change the def and num file printf format specifier from "%-40sXXX"
2587 to "%-39s XXX". The latter will always guarantee a space after the
2588 field while the former will cause them to run together if the field
2589 is 40 of more characters long.
2590 [Steve Henson]
2591
2592 *) Constify the cipher and digest 'method' functions and structures
2593 and modify related functions to take constant EVP_MD and EVP_CIPHER
2594 pointers.
2595 [Steve Henson]
2596
2597 *) Hide BN_CTX structure details in bn_lcl.h instead of publishing them
2598 in <openssl/bn.h>. Also further increase BN_CTX_NUM to 32.
2599 [Bodo Moeller]
2600
2601 *) Modify EVP_Digest*() routines so they now return values. Although the
2602 internal software routines can never fail additional hardware versions
2603 might.
2604 [Steve Henson]
2605
2606 *) Clean up crypto/err/err.h and change some error codes to avoid conflicts:
2607
2608 Previously ERR_R_FATAL was too small and coincided with ERR_LIB_PKCS7
2609 (= ERR_R_PKCS7_LIB); it is now 64 instead of 32.
2610
2611 ASN1 error codes
2612 ERR_R_NESTED_ASN1_ERROR
2613 ...
2614 ERR_R_MISSING_ASN1_EOS
2615 were 4 .. 9, conflicting with
2616 ERR_LIB_RSA (= ERR_R_RSA_LIB)
2617 ...
2618 ERR_LIB_PEM (= ERR_R_PEM_LIB).
2619 They are now 58 .. 63 (i.e., just below ERR_R_FATAL).
2620
2621 Add new error code 'ERR_R_INTERNAL_ERROR'.
2622 [Bodo Moeller]
2623
2624 *) Don't overuse locks in crypto/err/err.c: For data retrieval, CRYPTO_r_lock
2625 suffices.
2626 [Bodo Moeller]
2627
2628 *) New option '-subj arg' for 'openssl req' and 'openssl ca'. This
2629 sets the subject name for a new request or supersedes the
2630 subject name in a given request. Formats that can be parsed are
2631 'CN=Some Name, OU=myOU, C=IT'
2632 and
2633 'CN=Some Name/OU=myOU/C=IT'.
2634
2635 Add options '-batch' and '-verbose' to 'openssl req'.
2636 [Massimiliano Pala <madwolf@hackmasters.net>]
2637
2638 *) Introduce the possibility to access global variables through
2639 functions on platform were that's the best way to handle exporting
2640 global variables in shared libraries. To enable this functionality,
2641 one must configure with "EXPORT_VAR_AS_FN" or defined the C macro
2642 "OPENSSL_EXPORT_VAR_AS_FUNCTION" in crypto/opensslconf.h (the latter
2643 is normally done by Configure or something similar).
2644
2645 To implement a global variable, use the macro OPENSSL_IMPLEMENT_GLOBAL
2646 in the source file (foo.c) like this:
2647
2648 OPENSSL_IMPLEMENT_GLOBAL(int,foo)=1;
2649 OPENSSL_IMPLEMENT_GLOBAL(double,bar);
2650
2651 To declare a global variable, use the macros OPENSSL_DECLARE_GLOBAL
2652 and OPENSSL_GLOBAL_REF in the header file (foo.h) like this:
2653
2654 OPENSSL_DECLARE_GLOBAL(int,foo);
2655 #define foo OPENSSL_GLOBAL_REF(foo)
2656 OPENSSL_DECLARE_GLOBAL(double,bar);
2657 #define bar OPENSSL_GLOBAL_REF(bar)
2658
2659 The #defines are very important, and therefore so is including the
2660 header file everywhere where the defined globals are used.
2661
2662 The macro OPENSSL_EXPORT_VAR_AS_FUNCTION also affects the definition
2663 of ASN.1 items, but that structure is a bit different.
2664
2665 The largest change is in util/mkdef.pl which has been enhanced with
2666 better and easier to understand logic to choose which symbols should
2667 go into the Windows .def files as well as a number of fixes and code
2668 cleanup (among others, algorithm keywords are now sorted
2669 lexicographically to avoid constant rewrites).
2670 [Richard Levitte]
2671
2672 *) In BN_div() keep a copy of the sign of 'num' before writing the
2673 result to 'rm' because if rm==num the value will be overwritten
2674 and produce the wrong result if 'num' is negative: this caused
2675 problems with BN_mod() and BN_nnmod().
2676 [Steve Henson]
2677
2678 *) Function OCSP_request_verify(). This checks the signature on an
2679 OCSP request and verifies the signer certificate. The signer
2680 certificate is just checked for a generic purpose and OCSP request
2681 trust settings.
2682 [Steve Henson]
2683
2684 *) Add OCSP_check_validity() function to check the validity of OCSP
2685 responses. OCSP responses are prepared in real time and may only
2686 be a few seconds old. Simply checking that the current time lies
2687 between thisUpdate and nextUpdate max reject otherwise valid responses
2688 caused by either OCSP responder or client clock inaccuracy. Instead
2689 we allow thisUpdate and nextUpdate to fall within a certain period of
2690 the current time. The age of the response can also optionally be
2691 checked. Two new options -validity_period and -status_age added to
2692 ocsp utility.
2693 [Steve Henson]
2694
2695 *) If signature or public key algorithm is unrecognized print out its
2696 OID rather that just UNKNOWN.
2697 [Steve Henson]
2698
2699 *) Change OCSP_cert_to_id() to tolerate a NULL subject certificate and
2700 OCSP_cert_id_new() a NULL serialNumber. This allows a partial certificate
2701 ID to be generated from the issuer certificate alone which can then be
2702 passed to OCSP_id_issuer_cmp().
2703 [Steve Henson]
2704
2705 *) New compilation option ASN1_ITEM_FUNCTIONS. This causes the new
2706 ASN1 modules to export functions returning ASN1_ITEM pointers
2707 instead of the ASN1_ITEM structures themselves. This adds several
2708 new macros which allow the underlying ASN1 function/structure to
2709 be accessed transparently. As a result code should not use ASN1_ITEM
2710 references directly (such as &X509_it) but instead use the relevant
2711 macros (such as ASN1_ITEM_rptr(X509)). This option is to allow
2712 use of the new ASN1 code on platforms where exporting structures
2713 is problematical (for example in shared libraries) but exporting
2714 functions returning pointers to structures is not.
2715 [Steve Henson]
2716
2717 *) Add support for overriding the generation of SSL/TLS session IDs.
2718 These callbacks can be registered either in an SSL_CTX or per SSL.
2719 The purpose of this is to allow applications to control, if they wish,
2720 the arbitrary values chosen for use as session IDs, particularly as it
2721 can be useful for session caching in multiple-server environments. A
2722 command-line switch for testing this (and any client code that wishes
2723 to use such a feature) has been added to "s_server".
2724 [Geoff Thorpe, Lutz Jaenicke]
2725
2726 *) Modify mkdef.pl to recognise and parse preprocessor conditionals
2727 of the form '#if defined(...) || defined(...) || ...' and
2728 '#if !defined(...) && !defined(...) && ...'. This also avoids
2729 the growing number of special cases it was previously handling.
2730 [Richard Levitte]
2731
2732 *) Make all configuration macros available for application by making
2733 sure they are available in opensslconf.h, by giving them names starting
2734 with "OPENSSL_" to avoid conflicts with other packages and by making
2735 sure e_os2.h will cover all platform-specific cases together with
2736 opensslconf.h.
2737 Additionally, it is now possible to define configuration/platform-
2738 specific names (called "system identities"). In the C code, these
2739 are prefixed with "OPENSSL_SYSNAME_". e_os2.h will create another
2740 macro with the name beginning with "OPENSSL_SYS_", which is determined
2741 from "OPENSSL_SYSNAME_*" or compiler-specific macros depending on
2742 what is available.
2743 [Richard Levitte]
2744
2745 *) New option -set_serial to 'req' and 'x509' this allows the serial
2746 number to use to be specified on the command line. Previously self
2747 signed certificates were hard coded with serial number 0 and the
2748 CA options of 'x509' had to use a serial number in a file which was
2749 auto incremented.
2750 [Steve Henson]
2751
2752 *) New options to 'ca' utility to support V2 CRL entry extensions.
2753 Currently CRL reason, invalidity date and hold instruction are
2754 supported. Add new CRL extensions to V3 code and some new objects.
2755 [Steve Henson]
2756
2757 *) New function EVP_CIPHER_CTX_set_padding() this is used to
2758 disable standard block padding (aka PKCS#5 padding) in the EVP
2759 API, which was previously mandatory. This means that the data is
2760 not padded in any way and so the total length much be a multiple
2761 of the block size, otherwise an error occurs.
2762 [Steve Henson]
2763
2764 *) Initial (incomplete) OCSP SSL support.
2765 [Steve Henson]
2766
2767 *) New function OCSP_parse_url(). This splits up a URL into its host,
2768 port and path components: primarily to parse OCSP URLs. New -url
2769 option to ocsp utility.
2770 [Steve Henson]
2771
2772 *) New nonce behavior. The return value of OCSP_check_nonce() now
2773 reflects the various checks performed. Applications can decide
2774 whether to tolerate certain situations such as an absent nonce
2775 in a response when one was present in a request: the ocsp application
2776 just prints out a warning. New function OCSP_add1_basic_nonce()
2777 this is to allow responders to include a nonce in a response even if
2778 the request is nonce-less.
2779 [Steve Henson]
2780
2781 *) Disable stdin buffering in load_cert (apps/apps.c) so that no certs are
2782 skipped when using openssl x509 multiple times on a single input file,
2783 e.g. "(openssl x509 -out cert1; openssl x509 -out cert2) <certs".
2784 [Bodo Moeller]
2785
2786 *) Make ASN1_UTCTIME_set_string() and ASN1_GENERALIZEDTIME_set_string()
2787 set string type: to handle setting ASN1_TIME structures. Fix ca
2788 utility to correctly initialize revocation date of CRLs.
2789 [Steve Henson]
2790
2791 *) New option SSL_OP_CIPHER_SERVER_PREFERENCE allows the server to override
2792 the clients preferred ciphersuites and rather use its own preferences.
2793 Should help to work around M$ SGC (Server Gated Cryptography) bug in
2794 Internet Explorer by ensuring unchanged hash method during stepup.
2795 (Also replaces the broken/deactivated SSL_OP_NON_EXPORT_FIRST option.)
2796 [Lutz Jaenicke]
2797
2798 *) Make mkdef.pl recognise all DECLARE_ASN1 macros, change rijndael
2799 to aes and add a new 'exist' option to print out symbols that don't
2800 appear to exist.
2801 [Steve Henson]
2802
2803 *) Additional options to ocsp utility to allow flags to be set and
2804 additional certificates supplied.
2805 [Steve Henson]
2806
2807 *) Add the option -VAfile to 'openssl ocsp', so the user can give the
2808 OCSP client a number of certificate to only verify the response
2809 signature against.
2810 [Richard Levitte]
2811
2812 *) Update Rijndael code to version 3.0 and change EVP AES ciphers to
2813 handle the new API. Currently only ECB, CBC modes supported. Add new
2814 AES OIDs.
2815
2816 Add TLS AES ciphersuites as described in RFC3268, "Advanced
2817 Encryption Standard (AES) Ciphersuites for Transport Layer
2818 Security (TLS)". (In beta versions of OpenSSL 0.9.7, these were
2819 not enabled by default and were not part of the "ALL" ciphersuite
2820 alias because they were not yet official; they could be
2821 explicitly requested by specifying the "AESdraft" ciphersuite
2822 group alias. In the final release of OpenSSL 0.9.7, the group
2823 alias is called "AES" and is part of "ALL".)
2824 [Ben Laurie, Steve Henson, Bodo Moeller]
2825
2826 *) New function OCSP_copy_nonce() to copy nonce value (if present) from
2827 request to response.
2828 [Steve Henson]
2829
2830 *) Functions for OCSP responders. OCSP_request_onereq_count(),
2831 OCSP_request_onereq_get0(), OCSP_onereq_get0_id() and OCSP_id_get0_info()
2832 extract information from a certificate request. OCSP_response_create()
2833 creates a response and optionally adds a basic response structure.
2834 OCSP_basic_add1_status() adds a complete single response to a basic
2835 response and returns the OCSP_SINGLERESP structure just added (to allow
2836 extensions to be included for example). OCSP_basic_add1_cert() adds a
2837 certificate to a basic response and OCSP_basic_sign() signs a basic
2838 response with various flags. New helper functions ASN1_TIME_check()
2839 (checks validity of ASN1_TIME structure) and ASN1_TIME_to_generalizedtime()
2840 (converts ASN1_TIME to GeneralizedTime).
2841 [Steve Henson]
2842
2843 *) Various new functions. EVP_Digest() combines EVP_Digest{Init,Update,Final}()
2844 in a single operation. X509_get0_pubkey_bitstr() extracts the public_key
2845 structure from a certificate. X509_pubkey_digest() digests the public_key
2846 contents: this is used in various key identifiers.
2847 [Steve Henson]
2848
2849 *) Make sk_sort() tolerate a NULL argument.
2850 [Steve Henson reported by Massimiliano Pala <madwolf@comune.modena.it>]
2851
2852 *) New OCSP verify flag OCSP_TRUSTOTHER. When set the "other" certificates
2853 passed by the function are trusted implicitly. If any of them signed the
2854 response then it is assumed to be valid and is not verified.
2855 [Steve Henson]
2856
2857 *) In PKCS7_set_type() initialise content_type in PKCS7_ENC_CONTENT
2858 to data. This was previously part of the PKCS7 ASN1 code. This
2859 was causing problems with OpenSSL created PKCS#12 and PKCS#7 structures.
2860 [Steve Henson, reported by Kenneth R. Robinette
2861 <support@securenetterm.com>]
2862
2863 *) Add CRYPTO_push_info() and CRYPTO_pop_info() calls to new ASN1
2864 routines: without these tracing memory leaks is very painful.
2865 Fix leaks in PKCS12 and PKCS7 routines.
2866 [Steve Henson]
2867
2868 *) Make X509_time_adj() cope with the new behaviour of ASN1_TIME_new().
2869 Previously it initialised the 'type' argument to V_ASN1_UTCTIME which
2870 effectively meant GeneralizedTime would never be used. Now it
2871 is initialised to -1 but X509_time_adj() now has to check the value
2872 and use ASN1_TIME_set() if the value is not V_ASN1_UTCTIME or
2873 V_ASN1_GENERALIZEDTIME, without this it always uses GeneralizedTime.
2874 [Steve Henson, reported by Kenneth R. Robinette
2875 <support@securenetterm.com>]
2876
2877 *) Fixes to BN_to_ASN1_INTEGER when bn is zero. This would previously
2878 result in a zero length in the ASN1_INTEGER structure which was
2879 not consistent with the structure when d2i_ASN1_INTEGER() was used
2880 and would cause ASN1_INTEGER_cmp() to fail. Enhance s2i_ASN1_INTEGER()
2881 to cope with hex and negative integers. Fix bug in i2a_ASN1_INTEGER()
2882 where it did not print out a minus for negative ASN1_INTEGER.
2883 [Steve Henson]
2884
2885 *) Add summary printout to ocsp utility. The various functions which
2886 convert status values to strings have been renamed to:
2887 OCSP_response_status_str(), OCSP_cert_status_str() and
2888 OCSP_crl_reason_str() and are no longer static. New options
2889 to verify nonce values and to disable verification. OCSP response
2890 printout format cleaned up.
2891 [Steve Henson]
2892
2893 *) Add additional OCSP certificate checks. These are those specified
2894 in RFC2560. This consists of two separate checks: the CA of the
2895 certificate being checked must either be the OCSP signer certificate
2896 or the issuer of the OCSP signer certificate. In the latter case the
2897 OCSP signer certificate must contain the OCSP signing extended key
2898 usage. This check is performed by attempting to match the OCSP
2899 signer or the OCSP signer CA to the issuerNameHash and issuerKeyHash
2900 in the OCSP_CERTID structures of the response.
2901 [Steve Henson]
2902
2903 *) Initial OCSP certificate verification added to OCSP_basic_verify()
2904 and related routines. This uses the standard OpenSSL certificate
2905 verify routines to perform initial checks (just CA validity) and
2906 to obtain the certificate chain. Then additional checks will be
2907 performed on the chain. Currently the root CA is checked to see
2908 if it is explicitly trusted for OCSP signing. This is used to set
2909 a root CA as a global signing root: that is any certificate that
2910 chains to that CA is an acceptable OCSP signing certificate.
2911 [Steve Henson]
2912
2913 *) New '-extfile ...' option to 'openssl ca' for reading X.509v3
2914 extensions from a separate configuration file.
2915 As when reading extensions from the main configuration file,
2916 the '-extensions ...' option may be used for specifying the
2917 section to use.
2918 [Massimiliano Pala <madwolf@comune.modena.it>]
2919
2920 *) New OCSP utility. Allows OCSP requests to be generated or
2921 read. The request can be sent to a responder and the output
2922 parsed, outputed or printed in text form. Not complete yet:
2923 still needs to check the OCSP response validity.
2924 [Steve Henson]
2925
2926 *) New subcommands for 'openssl ca':
2927 'openssl ca -status <serial>' prints the status of the cert with
2928 the given serial number (according to the index file).
2929 'openssl ca -updatedb' updates the expiry status of certificates
2930 in the index file.
2931 [Massimiliano Pala <madwolf@comune.modena.it>]
2932
2933 *) New '-newreq-nodes' command option to CA.pl. This is like
2934 '-newreq', but calls 'openssl req' with the '-nodes' option
2935 so that the resulting key is not encrypted.
2936 [Damien Miller <djm@mindrot.org>]
2937
2938 *) New configuration for the GNU Hurd.
2939 [Jonathan Bartlett <johnnyb@wolfram.com> via Richard Levitte]
2940
2941 *) Initial code to implement OCSP basic response verify. This
2942 is currently incomplete. Currently just finds the signer's
2943 certificate and verifies the signature on the response.
2944 [Steve Henson]
2945
2946 *) New SSLeay_version code SSLEAY_DIR to determine the compiled-in
2947 value of OPENSSLDIR. This is available via the new '-d' option
2948 to 'openssl version', and is also included in 'openssl version -a'.
2949 [Bodo Moeller]
2950
2951 *) Allowing defining memory allocation callbacks that will be given
2952 file name and line number information in additional arguments
2953 (a const char* and an int). The basic functionality remains, as
2954 well as the original possibility to just replace malloc(),
2955 realloc() and free() by functions that do not know about these
2956 additional arguments. To register and find out the current
2957 settings for extended allocation functions, the following
2958 functions are provided:
2959
2960 CRYPTO_set_mem_ex_functions
2961 CRYPTO_set_locked_mem_ex_functions
2962 CRYPTO_get_mem_ex_functions
2963 CRYPTO_get_locked_mem_ex_functions
2964
2965 These work the same way as CRYPTO_set_mem_functions and friends.
2966 CRYPTO_get_[locked_]mem_functions now writes 0 where such an
2967 extended allocation function is enabled.
2968 Similarly, CRYPTO_get_[locked_]mem_ex_functions writes 0 where
2969 a conventional allocation function is enabled.
2970 [Richard Levitte, Bodo Moeller]
2971
2972 *) Finish off removing the remaining LHASH function pointer casts.
2973 There should no longer be any prototype-casting required when using
2974 the LHASH abstraction, and any casts that remain are "bugs". See
2975 the callback types and macros at the head of lhash.h for details
2976 (and "OBJ_cleanup" in crypto/objects/obj_dat.c as an example).
2977 [Geoff Thorpe]
2978
2979 *) Add automatic query of EGD sockets in RAND_poll() for the unix variant.
2980 If /dev/[u]random devices are not available or do not return enough
2981 entropy, EGD style sockets (served by EGD or PRNGD) will automatically
2982 be queried.
2983 The locations /var/run/egd-pool, /dev/egd-pool, /etc/egd-pool, and
2984 /etc/entropy will be queried once each in this sequence, quering stops
2985 when enough entropy was collected without querying more sockets.
2986 [Lutz Jaenicke]
2987
2988 *) Change the Unix RAND_poll() variant to be able to poll several
2989 random devices, as specified by DEVRANDOM, until a sufficient amount
2990 of data has been collected. We spend at most 10 ms on each file
2991 (select timeout) and read in non-blocking mode. DEVRANDOM now
2992 defaults to the list "/dev/urandom", "/dev/random", "/dev/srandom"
2993 (previously it was just the string "/dev/urandom"), so on typical
2994 platforms the 10 ms delay will never occur.
2995 Also separate out the Unix variant to its own file, rand_unix.c.
2996 For VMS, there's a currently-empty rand_vms.c.
2997 [Richard Levitte]
2998
2999 *) Move OCSP client related routines to ocsp_cl.c. These
3000 provide utility functions which an application needing
3001 to issue a request to an OCSP responder and analyse the
3002 response will typically need: as opposed to those which an
3003 OCSP responder itself would need which will be added later.
3004
3005 OCSP_request_sign() signs an OCSP request with an API similar
3006 to PKCS7_sign(). OCSP_response_status() returns status of OCSP
3007 response. OCSP_response_get1_basic() extracts basic response
3008 from response. OCSP_resp_find_status(): finds and extracts status
3009 information from an OCSP_CERTID structure (which will be created
3010 when the request structure is built). These are built from lower
3011 level functions which work on OCSP_SINGLERESP structures but
3012 wont normally be used unless the application wishes to examine
3013 extensions in the OCSP response for example.
3014
3015 Replace nonce routines with a pair of functions.
3016 OCSP_request_add1_nonce() adds a nonce value and optionally
3017 generates a random value. OCSP_check_nonce() checks the
3018 validity of the nonce in an OCSP response.
3019 [Steve Henson]
3020
3021 *) Change function OCSP_request_add() to OCSP_request_add0_id().
3022 This doesn't copy the supplied OCSP_CERTID and avoids the
3023 need to free up the newly created id. Change return type
3024 to OCSP_ONEREQ to return the internal OCSP_ONEREQ structure.
3025 This can then be used to add extensions to the request.
3026 Deleted OCSP_request_new(), since most of its functionality
3027 is now in OCSP_REQUEST_new() (and the case insensitive name
3028 clash) apart from the ability to set the request name which
3029 will be added elsewhere.
3030 [Steve Henson]
3031
3032 *) Update OCSP API. Remove obsolete extensions argument from
3033 various functions. Extensions are now handled using the new
3034 OCSP extension code. New simple OCSP HTTP function which
3035 can be used to send requests and parse the response.
3036 [Steve Henson]
3037
3038 *) Fix the PKCS#7 (S/MIME) code to work with new ASN1. Two new
3039 ASN1_ITEM structures help with sign and verify. PKCS7_ATTR_SIGN
3040 uses the special reorder version of SET OF to sort the attributes
3041 and reorder them to match the encoded order. This resolves a long
3042 standing problem: a verify on a PKCS7 structure just after signing
3043 it used to fail because the attribute order did not match the
3044 encoded order. PKCS7_ATTR_VERIFY does not reorder the attributes:
3045 it uses the received order. This is necessary to tolerate some broken
3046 software that does not order SET OF. This is handled by encoding
3047 as a SEQUENCE OF but using implicit tagging (with UNIVERSAL class)
3048 to produce the required SET OF.
3049 [Steve Henson]
3050
3051 *) Have mk1mf.pl generate the macros OPENSSL_BUILD_SHLIBCRYPTO and
3052 OPENSSL_BUILD_SHLIBSSL and use them appropriately in the header
3053 files to get correct declarations of the ASN.1 item variables.
3054 [Richard Levitte]
3055
3056 *) Rewrite of PKCS#12 code to use new ASN1 functionality. Replace many
3057 PKCS#12 macros with real functions. Fix two unrelated ASN1 bugs:
3058 asn1_check_tlen() would sometimes attempt to use 'ctx' when it was
3059 NULL and ASN1_TYPE was not dereferenced properly in asn1_ex_c2i().
3060 New ASN1 macro: DECLARE_ASN1_ITEM() which just declares the relevant
3061 ASN1_ITEM and no wrapper functions.
3062 [Steve Henson]
3063
3064 *) New functions or ASN1_item_d2i_fp() and ASN1_item_d2i_bio(). These
3065 replace the old function pointer based I/O routines. Change most of
3066 the *_d2i_bio() and *_d2i_fp() functions to use these.
3067 [Steve Henson]
3068
3069 *) Enhance mkdef.pl to be more accepting about spacing in C preprocessor
3070 lines, recognice more "algorithms" that can be deselected, and make
3071 it complain about algorithm deselection that isn't recognised.
3072 [Richard Levitte]
3073
3074 *) New ASN1 functions to handle dup, sign, verify, digest, pack and
3075 unpack operations in terms of ASN1_ITEM. Modify existing wrappers
3076 to use new functions. Add NO_ASN1_OLD which can be set to remove
3077 some old style ASN1 functions: this can be used to determine if old
3078 code will still work when these eventually go away.
3079 [Steve Henson]
3080
3081 *) New extension functions for OCSP structures, these follow the
3082 same conventions as certificates and CRLs.
3083 [Steve Henson]
3084
3085 *) New function X509V3_add1_i2d(). This automatically encodes and
3086 adds an extension. Its behaviour can be customised with various
3087 flags to append, replace or delete. Various wrappers added for
3088 certifcates and CRLs.
3089 [Steve Henson]
3090
3091 *) Fix to avoid calling the underlying ASN1 print routine when
3092 an extension cannot be parsed. Correct a typo in the
3093 OCSP_SERVICELOC extension. Tidy up print OCSP format.
3094 [Steve Henson]
3095
3096 *) Make mkdef.pl parse some of the ASN1 macros and add apropriate
3097 entries for variables.
3098 [Steve Henson]
3099
3100 *) Add functionality to apps/openssl.c for detecting locking
3101 problems: As the program is single-threaded, all we have
3102 to do is register a locking callback using an array for
3103 storing which locks are currently held by the program.
3104 [Bodo Moeller]
3105
3106 *) Use a lock around the call to CRYPTO_get_ex_new_index() in
3107 SSL_get_ex_data_X509_STORE_idx(), which is used in
3108 ssl_verify_cert_chain() and thus can be called at any time
3109 during TLS/SSL handshakes so that thread-safety is essential.
3110 Unfortunately, the ex_data design is not at all suited
3111 for multi-threaded use, so it probably should be abolished.
3112 [Bodo Moeller]
3113
3114 *) Added Broadcom "ubsec" ENGINE to OpenSSL.
3115 [Broadcom, tweaked and integrated by Geoff Thorpe]
3116
3117 *) Move common extension printing code to new function
3118 X509V3_print_extensions(). Reorganise OCSP print routines and
3119 implement some needed OCSP ASN1 functions. Add OCSP extensions.
3120 [Steve Henson]
3121
3122 *) New function X509_signature_print() to remove duplication in some
3123 print routines.
3124 [Steve Henson]
3125
3126 *) Add a special meaning when SET OF and SEQUENCE OF flags are both
3127 set (this was treated exactly the same as SET OF previously). This
3128 is used to reorder the STACK representing the structure to match the
3129 encoding. This will be used to get round a problem where a PKCS7
3130 structure which was signed could not be verified because the STACK
3131 order did not reflect the encoded order.
3132 [Steve Henson]
3133
3134 *) Reimplement the OCSP ASN1 module using the new code.
3135 [Steve Henson]
3136
3137 *) Update the X509V3 code to permit the use of an ASN1_ITEM structure
3138 for its ASN1 operations. The old style function pointers still exist
3139 for now but they will eventually go away.
3140 [Steve Henson]
3141
3142 *) Merge in replacement ASN1 code from the ASN1 branch. This almost
3143 completely replaces the old ASN1 functionality with a table driven
3144 encoder and decoder which interprets an ASN1_ITEM structure describing
3145 the ASN1 module. Compatibility with the existing ASN1 API (i2d,d2i) is
3146 largely maintained. Almost all of the old asn1_mac.h macro based ASN1
3147 has also been converted to the new form.
3148 [Steve Henson]
3149
3150 *) Change BN_mod_exp_recp so that negative moduli are tolerated
3151 (the sign is ignored). Similarly, ignore the sign in BN_MONT_CTX_set
3152 so that BN_mod_exp_mont and BN_mod_exp_mont_word work
3153 for negative moduli.
3154 [Bodo Moeller]
3155
3156 *) Fix BN_uadd and BN_usub: Always return non-negative results instead
3157 of not touching the result's sign bit.
3158 [Bodo Moeller]
3159
3160 *) BN_div bugfix: If the result is 0, the sign (res->neg) must not be
3161 set.
3162 [Bodo Moeller]
3163
3164 *) Changed the LHASH code to use prototypes for callbacks, and created
3165 macros to declare and implement thin (optionally static) functions
3166 that provide type-safety and avoid function pointer casting for the
3167 type-specific callbacks.
3168 [Geoff Thorpe]
3169
3170 *) Added Kerberos Cipher Suites to be used with TLS, as written in
3171 RFC 2712.
3172 [Veers Staats <staatsvr@asc.hpc.mil>,
3173 Jeffrey Altman <jaltman@columbia.edu>, via Richard Levitte]
3174
3175 *) Reformat the FAQ so the different questions and answers can be divided
3176 in sections depending on the subject.
3177 [Richard Levitte]
3178
3179 *) Have the zlib compression code load ZLIB.DLL dynamically under
3180 Windows.
3181 [Richard Levitte]
3182
3183 *) New function BN_mod_sqrt for computing square roots modulo a prime
3184 (using the probabilistic Tonelli-Shanks algorithm unless
3185 p == 3 (mod 4) or p == 5 (mod 8), which are cases that can
3186 be handled deterministically).
3187 [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller]
3188
3189 *) Make BN_mod_inverse faster by explicitly handling small quotients
3190 in the Euclid loop. (Speed gain about 20% for small moduli [256 or
3191 512 bits], about 30% for larger ones [1024 or 2048 bits].)
3192 [Bodo Moeller]
3193
3194 *) New function BN_kronecker.
3195 [Bodo Moeller]
3196
3197 *) Fix BN_gcd so that it works on negative inputs; the result is
3198 positive unless both parameters are zero.
3199 Previously something reasonably close to an infinite loop was
3200 possible because numbers could be growing instead of shrinking
3201 in the implementation of Euclid's algorithm.
3202 [Bodo Moeller]
3203
3204 *) Fix BN_is_word() and BN_is_one() macros to take into account the
3205 sign of the number in question.
3206
3207 Fix BN_is_word(a,w) to work correctly for w == 0.
3208
3209 The old BN_is_word(a,w) macro is now called BN_abs_is_word(a,w)
3210 because its test if the absolute value of 'a' equals 'w'.
3211 Note that BN_abs_is_word does *not* handle w == 0 reliably;
3212 it exists mostly for use in the implementations of BN_is_zero(),
3213 BN_is_one(), and BN_is_word().
3214 [Bodo Moeller]
3215
3216 *) New function BN_swap.
3217 [Bodo Moeller]
3218
3219 *) Use BN_nnmod instead of BN_mod in crypto/bn/bn_exp.c so that
3220 the exponentiation functions are more likely to produce reasonable
3221 results on negative inputs.
3222 [Bodo Moeller]
3223
3224 *) Change BN_mod_mul so that the result is always non-negative.
3225 Previously, it could be negative if one of the factors was negative;
3226 I don't think anyone really wanted that behaviour.
3227 [Bodo Moeller]
3228
3229 *) Move BN_mod_... functions into new file crypto/bn/bn_mod.c
3230 (except for exponentiation, which stays in crypto/bn/bn_exp.c,
3231 and BN_mod_mul_reciprocal, which stays in crypto/bn/bn_recp.c)
3232 and add new functions:
3233
3234 BN_nnmod
3235 BN_mod_sqr
3236 BN_mod_add
3237 BN_mod_add_quick
3238 BN_mod_sub
3239 BN_mod_sub_quick
3240 BN_mod_lshift1
3241 BN_mod_lshift1_quick
3242 BN_mod_lshift
3243 BN_mod_lshift_quick
3244
3245 These functions always generate non-negative results.
3246
3247 BN_nnmod otherwise is like BN_mod (if BN_mod computes a remainder r
3248 such that |m| < r < 0, BN_nnmod will output rem + |m| instead).
3249
3250 BN_mod_XXX_quick(r, a, [b,] m) generates the same result as
3251 BN_mod_XXX(r, a, [b,] m, ctx), but requires that a [and b]
3252 be reduced modulo m.
3253 [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller]
3254
3255 #if 0
3256 The following entry accidentily appeared in the CHANGES file
3257 distributed with OpenSSL 0.9.7. The modifications described in
3258 it do *not* apply to OpenSSL 0.9.7.
3259
3260 *) Remove a few calls to bn_wexpand() in BN_sqr() (the one in there
3261 was actually never needed) and in BN_mul(). The removal in BN_mul()
3262 required a small change in bn_mul_part_recursive() and the addition
3263 of the functions bn_cmp_part_words(), bn_sub_part_words() and
3264 bn_add_part_words(), which do the same thing as bn_cmp_words(),
3265 bn_sub_words() and bn_add_words() except they take arrays with
3266 differing sizes.
3267 [Richard Levitte]
3268 #endif
3269
3270 *) In 'openssl passwd', verify passwords read from the terminal
3271 unless the '-salt' option is used (which usually means that
3272 verification would just waste user's time since the resulting
3273 hash is going to be compared with some given password hash)
3274 or the new '-noverify' option is used.
3275
3276 This is an incompatible change, but it does not affect
3277 non-interactive use of 'openssl passwd' (passwords on the command
3278 line, '-stdin' option, '-in ...' option) and thus should not
3279 cause any problems.
3280 [Bodo Moeller]
3281
3282 *) Remove all references to RSAref, since there's no more need for it.
3283 [Richard Levitte]
3284
3285 *) Make DSO load along a path given through an environment variable
3286 (SHLIB_PATH) with shl_load().
3287 [Richard Levitte]
3288
3289 *) Constify the ENGINE code as a result of BIGNUM constification.
3290 Also constify the RSA code and most things related to it. In a
3291 few places, most notable in the depth of the ASN.1 code, ugly
3292 casts back to non-const were required (to be solved at a later
3293 time)
3294 [Richard Levitte]
3295
3296 *) Make it so the openssl application has all engines loaded by default.
3297 [Richard Levitte]
3298
3299 *) Constify the BIGNUM routines a little more.
3300 [Richard Levitte]
3301
3302 *) Add the following functions:
3303
3304 ENGINE_load_cswift()
3305 ENGINE_load_chil()
3306 ENGINE_load_atalla()
3307 ENGINE_load_nuron()
3308 ENGINE_load_builtin_engines()
3309
3310 That way, an application can itself choose if external engines that
3311 are built-in in OpenSSL shall ever be used or not. The benefit is
3312 that applications won't have to be linked with libdl or other dso
3313 libraries unless it's really needed.
3314
3315 Changed 'openssl engine' to load all engines on demand.
3316 Changed the engine header files to avoid the duplication of some
3317 declarations (they differed!).
3318 [Richard Levitte]
3319
3320 *) 'openssl engine' can now list capabilities.
3321 [Richard Levitte]
3322
3323 *) Better error reporting in 'openssl engine'.
3324 [Richard Levitte]
3325
3326 *) Never call load_dh_param(NULL) in s_server.
3327 [Bodo Moeller]
3328
3329 *) Add engine application. It can currently list engines by name and
3330 identity, and test if they are actually available.
3331 [Richard Levitte]
3332
3333 *) Improve RPM specification file by forcing symbolic linking and making
3334 sure the installed documentation is also owned by root.root.
3335 [Damien Miller <djm@mindrot.org>]
3336
3337 *) Give the OpenSSL applications more possibilities to make use of
3338 keys (public as well as private) handled by engines.
3339 [Richard Levitte]
3340
3341 *) Add OCSP code that comes from CertCo.
3342 [Richard Levitte]
3343
3344 *) Add VMS support for the Rijndael code.
3345 [Richard Levitte]
3346
3347 *) Added untested support for Nuron crypto accelerator.
3348 [Ben Laurie]
3349
3350 *) Add support for external cryptographic devices. This code was
3351 previously distributed separately as the "engine" branch.
3352 [Geoff Thorpe, Richard Levitte]
3353
3354 *) Rework the filename-translation in the DSO code. It is now possible to
3355 have far greater control over how a "name" is turned into a filename
3356 depending on the operating environment and any oddities about the
3357 different shared library filenames on each system.
3358 [Geoff Thorpe]
3359
3360 *) Support threads on FreeBSD-elf in Configure.
3361 [Richard Levitte]
3362
3363 *) Fix for SHA1 assembly problem with MASM: it produces
3364 warnings about corrupt line number information when assembling
3365 with debugging information. This is caused by the overlapping
3366 of two sections.
3367 [Bernd Matthes <mainbug@celocom.de>, Steve Henson]
3368
3369 *) NCONF changes.
3370 NCONF_get_number() has no error checking at all. As a replacement,
3371 NCONF_get_number_e() is defined (_e for "error checking") and is
3372 promoted strongly. The old NCONF_get_number is kept around for
3373 binary backward compatibility.
3374 Make it possible for methods to load from something other than a BIO,
3375 by providing a function pointer that is given a name instead of a BIO.
3376 For example, this could be used to load configuration data from an
3377 LDAP server.
3378 [Richard Levitte]
3379
3380 *) Fix for non blocking accept BIOs. Added new I/O special reason
3381 BIO_RR_ACCEPT to cover this case. Previously use of accept BIOs
3382 with non blocking I/O was not possible because no retry code was
3383 implemented. Also added new SSL code SSL_WANT_ACCEPT to cover
3384 this case.
3385 [Steve Henson]
3386
3387 *) Added the beginnings of Rijndael support.
3388 [Ben Laurie]
3389
3390 *) Fix for bug in DirectoryString mask setting. Add support for
3391 X509_NAME_print_ex() in 'req' and X509_print_ex() function
3392 to allow certificate printing to more controllable, additional
3393 'certopt' option to 'x509' to allow new printing options to be
3394 set.
3395 [Steve Henson]
3396
3397 *) Clean old EAY MD5 hack from e_os.h.
3398 [Richard Levitte]
3399
3400 Changes between 0.9.6l and 0.9.6m [17 Mar 2004]
3401
3402 *) Fix null-pointer assignment in do_change_cipher_spec() revealed
3403 by using the Codenomicon TLS Test Tool (CVE-2004-0079)
3404 [Joe Orton, Steve Henson]
3405
3406 Changes between 0.9.6k and 0.9.6l [04 Nov 2003]
3407
3408 *) Fix additional bug revealed by the NISCC test suite:
3409
3410 Stop bug triggering large recursion when presented with
3411 certain ASN.1 tags (CVE-2003-0851)
3412 [Steve Henson]
3413
3414 Changes between 0.9.6j and 0.9.6k [30 Sep 2003]
3415
3416 *) Fix various bugs revealed by running the NISCC test suite:
3417
3418 Stop out of bounds reads in the ASN1 code when presented with
3419 invalid tags (CVE-2003-0543 and CVE-2003-0544).
3420
3421 If verify callback ignores invalid public key errors don't try to check
3422 certificate signature with the NULL public key.
3423
3424 [Steve Henson]
3425
3426 *) In ssl3_accept() (ssl/s3_srvr.c) only accept a client certificate
3427 if the server requested one: as stated in TLS 1.0 and SSL 3.0
3428 specifications.
3429 [Steve Henson]
3430
3431 *) In ssl3_get_client_hello() (ssl/s3_srvr.c), tolerate additional
3432 extra data after the compression methods not only for TLS 1.0
3433 but also for SSL 3.0 (as required by the specification).
3434 [Bodo Moeller; problem pointed out by Matthias Loepfe]
3435
3436 *) Change X509_certificate_type() to mark the key as exported/exportable
3437 when it's 512 *bits* long, not 512 bytes.
3438 [Richard Levitte]
3439
3440 Changes between 0.9.6i and 0.9.6j [10 Apr 2003]
3441
3442 *) Countermeasure against the Klima-Pokorny-Rosa extension of
3443 Bleichbacher's attack on PKCS #1 v1.5 padding: treat
3444 a protocol version number mismatch like a decryption error
3445 in ssl3_get_client_key_exchange (ssl/s3_srvr.c).
3446 [Bodo Moeller]
3447
3448 *) Turn on RSA blinding by default in the default implementation
3449 to avoid a timing attack. Applications that don't want it can call
3450 RSA_blinding_off() or use the new flag RSA_FLAG_NO_BLINDING.
3451 They would be ill-advised to do so in most cases.
3452 [Ben Laurie, Steve Henson, Geoff Thorpe, Bodo Moeller]
3453
3454 *) Change RSA blinding code so that it works when the PRNG is not
3455 seeded (in this case, the secret RSA exponent is abused as
3456 an unpredictable seed -- if it is not unpredictable, there
3457 is no point in blinding anyway). Make RSA blinding thread-safe
3458 by remembering the creator's thread ID in rsa->blinding and
3459 having all other threads use local one-time blinding factors
3460 (this requires more computation than sharing rsa->blinding, but
3461 avoids excessive locking; and if an RSA object is not shared
3462 between threads, blinding will still be very fast).
3463 [Bodo Moeller]
3464
3465 Changes between 0.9.6h and 0.9.6i [19 Feb 2003]
3466
3467 *) In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked
3468 via timing by performing a MAC computation even if incorrrect
3469 block cipher padding has been found. This is a countermeasure
3470 against active attacks where the attacker has to distinguish
3471 between bad padding and a MAC verification error. (CVE-2003-0078)
3472
3473 [Bodo Moeller; problem pointed out by Brice Canvel (EPFL),
3474 Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and
3475 Martin Vuagnoux (EPFL, Ilion)]
3476
3477 Changes between 0.9.6g and 0.9.6h [5 Dec 2002]
3478
3479 *) New function OPENSSL_cleanse(), which is used to cleanse a section of
3480 memory from it's contents. This is done with a counter that will
3481 place alternating values in each byte. This can be used to solve
3482 two issues: 1) the removal of calls to memset() by highly optimizing
3483 compilers, and 2) cleansing with other values than 0, since those can
3484 be read through on certain media, for example a swap space on disk.
3485 [Geoff Thorpe]
3486
3487 *) Bugfix: client side session caching did not work with external caching,
3488 because the session->cipher setting was not restored when reloading
3489 from the external cache. This problem was masked, when
3490 SSL_OP_NETSCAPE_REUSE_CIPHER_CHANGE_BUG (part of SSL_OP_ALL) was set.
3491 (Found by Steve Haslam <steve@araqnid.ddts.net>.)
3492 [Lutz Jaenicke]
3493
3494 *) Fix client_certificate (ssl/s2_clnt.c): The permissible total
3495 length of the REQUEST-CERTIFICATE message is 18 .. 34, not 17 .. 33.
3496 [Zeev Lieber <zeev-l@yahoo.com>]
3497
3498 *) Undo an undocumented change introduced in 0.9.6e which caused
3499 repeated calls to OpenSSL_add_all_ciphers() and
3500 OpenSSL_add_all_digests() to be ignored, even after calling
3501 EVP_cleanup().
3502 [Richard Levitte]
3503
3504 *) Change the default configuration reader to deal with last line not
3505 being properly terminated.
3506 [Richard Levitte]
3507
3508 *) Change X509_NAME_cmp() so it applies the special rules on handling
3509 DN values that are of type PrintableString, as well as RDNs of type
3510 emailAddress where the value has the type ia5String.
3511 [stefank@valicert.com via Richard Levitte]
3512
3513 *) Add a SSL_SESS_CACHE_NO_INTERNAL_STORE flag to take over half
3514 the job SSL_SESS_CACHE_NO_INTERNAL_LOOKUP was inconsistently
3515 doing, define a new flag (SSL_SESS_CACHE_NO_INTERNAL) to be
3516 the bitwise-OR of the two for use by the majority of applications
3517 wanting this behaviour, and update the docs. The documented
3518 behaviour and actual behaviour were inconsistent and had been
3519 changing anyway, so this is more a bug-fix than a behavioural
3520 change.
3521 [Geoff Thorpe, diagnosed by Nadav Har'El]
3522
3523 *) Don't impose a 16-byte length minimum on session IDs in ssl/s3_clnt.c
3524 (the SSL 3.0 and TLS 1.0 specifications allow any length up to 32 bytes).
3525 [Bodo Moeller]
3526
3527 *) Fix initialization code race conditions in
3528 SSLv23_method(), SSLv23_client_method(), SSLv23_server_method(),
3529 SSLv2_method(), SSLv2_client_method(), SSLv2_server_method(),
3530 SSLv3_method(), SSLv3_client_method(), SSLv3_server_method(),
3531 TLSv1_method(), TLSv1_client_method(), TLSv1_server_method(),
3532 ssl2_get_cipher_by_char(),
3533 ssl3_get_cipher_by_char().
3534 [Patrick McCormick <patrick@tellme.com>, Bodo Moeller]
3535
3536 *) Reorder cleanup sequence in SSL_CTX_free(): only remove the ex_data after
3537 the cached sessions are flushed, as the remove_cb() might use ex_data
3538 contents. Bug found by Sam Varshavchik <mrsam@courier-mta.com>
3539 (see [openssl.org #212]).
3540 [Geoff Thorpe, Lutz Jaenicke]
3541
3542 *) Fix typo in OBJ_txt2obj which incorrectly passed the content
3543 length, instead of the encoding length to d2i_ASN1_OBJECT.
3544 [Steve Henson]
3545
3546 Changes between 0.9.6f and 0.9.6g [9 Aug 2002]
3547
3548 *) [In 0.9.6g-engine release:]
3549 Fix crypto/engine/vendor_defns/cswift.h for WIN32 (use '_stdcall').
3550 [Lynn Gazis <lgazis@rainbow.com>]
3551
3552 Changes between 0.9.6e and 0.9.6f [8 Aug 2002]
3553
3554 *) Fix ASN1 checks. Check for overflow by comparing with LONG_MAX
3555 and get fix the header length calculation.
3556 [Florian Weimer <Weimer@CERT.Uni-Stuttgart.DE>,
3557 Alon Kantor <alonk@checkpoint.com> (and others),
3558 Steve Henson]
3559
3560 *) Use proper error handling instead of 'assertions' in buffer
3561 overflow checks added in 0.9.6e. This prevents DoS (the
3562 assertions could call abort()).
3563 [Arne Ansper <arne@ats.cyber.ee>, Bodo Moeller]
3564
3565 Changes between 0.9.6d and 0.9.6e [30 Jul 2002]
3566
3567 *) Add various sanity checks to asn1_get_length() to reject
3568 the ASN1 length bytes if they exceed sizeof(long), will appear
3569 negative or the content length exceeds the length of the
3570 supplied buffer.
3571 [Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>]
3572
3573 *) Fix cipher selection routines: ciphers without encryption had no flags
3574 for the cipher strength set and where therefore not handled correctly
3575 by the selection routines (PR #130).
3576 [Lutz Jaenicke]
3577
3578 *) Fix EVP_dsa_sha macro.
3579 [Nils Larsch]
3580
3581 *) New option
3582 SSL_OP_DONT_INSERT_EMPTY_FRAGMENTS
3583 for disabling the SSL 3.0/TLS 1.0 CBC vulnerability countermeasure
3584 that was added in OpenSSL 0.9.6d.
3585
3586 As the countermeasure turned out to be incompatible with some
3587 broken SSL implementations, the new option is part of SSL_OP_ALL.
3588 SSL_OP_ALL is usually employed when compatibility with weird SSL
3589 implementations is desired (e.g. '-bugs' option to 's_client' and
3590 's_server'), so the new option is automatically set in many
3591 applications.
3592 [Bodo Moeller]
3593
3594 *) Changes in security patch:
3595
3596 Changes marked "(CHATS)" were sponsored by the Defense Advanced
3597 Research Projects Agency (DARPA) and Air Force Research Laboratory,
3598 Air Force Materiel Command, USAF, under agreement number
3599 F30602-01-2-0537.
3600
3601 *) Add various sanity checks to asn1_get_length() to reject
3602 the ASN1 length bytes if they exceed sizeof(long), will appear
3603 negative or the content length exceeds the length of the
3604 supplied buffer. (CVE-2002-0659)
3605 [Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>]
3606
3607 *) Assertions for various potential buffer overflows, not known to
3608 happen in practice.
3609 [Ben Laurie (CHATS)]
3610
3611 *) Various temporary buffers to hold ASCII versions of integers were
3612 too small for 64 bit platforms. (CVE-2002-0655)
3613 [Matthew Byng-Maddick <mbm@aldigital.co.uk> and Ben Laurie (CHATS)>
3614
3615 *) Remote buffer overflow in SSL3 protocol - an attacker could
3616 supply an oversized session ID to a client. (CVE-2002-0656)
3617 [Ben Laurie (CHATS)]
3618
3619 *) Remote buffer overflow in SSL2 protocol - an attacker could
3620 supply an oversized client master key. (CVE-2002-0656)
3621 [Ben Laurie (CHATS)]
3622
3623 Changes between 0.9.6c and 0.9.6d [9 May 2002]
3624
3625 *) Fix crypto/asn1/a_sign.c so that 'parameters' is omitted (not
3626 encoded as NULL) with id-dsa-with-sha1.
3627 [Nils Larsch <nla@trustcenter.de>; problem pointed out by Bodo Moeller]
3628
3629 *) Check various X509_...() return values in apps/req.c.
3630 [Nils Larsch <nla@trustcenter.de>]
3631
3632 *) Fix BASE64 decode (EVP_DecodeUpdate) for data with CR/LF ended lines:
3633 an end-of-file condition would erronously be flagged, when the CRLF
3634 was just at the end of a processed block. The bug was discovered when
3635 processing data through a buffering memory BIO handing the data to a
3636 BASE64-decoding BIO. Bug fund and patch submitted by Pavel Tsekov
3637 <ptsekov@syntrex.com> and Nedelcho Stanev.
3638 [Lutz Jaenicke]
3639
3640 *) Implement a countermeasure against a vulnerability recently found
3641 in CBC ciphersuites in SSL 3.0/TLS 1.0: Send an empty fragment
3642 before application data chunks to avoid the use of known IVs
3643 with data potentially chosen by the attacker.
3644 [Bodo Moeller]
3645
3646 *) Fix length checks in ssl3_get_client_hello().
3647 [Bodo Moeller]
3648
3649 *) TLS/SSL library bugfix: use s->s3->in_read_app_data differently
3650 to prevent ssl3_read_internal() from incorrectly assuming that
3651 ssl3_read_bytes() found application data while handshake
3652 processing was enabled when in fact s->s3->in_read_app_data was
3653 merely automatically cleared during the initial handshake.
3654 [Bodo Moeller; problem pointed out by Arne Ansper <arne@ats.cyber.ee>]
3655
3656 *) Fix object definitions for Private and Enterprise: they were not
3657 recognized in their shortname (=lowercase) representation. Extend
3658 obj_dat.pl to issue an error when using undefined keywords instead
3659 of silently ignoring the problem (Svenning Sorensen
3660 <sss@sss.dnsalias.net>).
3661 [Lutz Jaenicke]
3662
3663 *) Fix DH_generate_parameters() so that it works for 'non-standard'
3664 generators, i.e. generators other than 2 and 5. (Previously, the
3665 code did not properly initialise the 'add' and 'rem' values to
3666 BN_generate_prime().)
3667
3668 In the new general case, we do not insist that 'generator' is
3669 actually a primitive root: This requirement is rather pointless;
3670 a generator of the order-q subgroup is just as good, if not
3671 better.
3672 [Bodo Moeller]
3673
3674 *) Map new X509 verification errors to alerts. Discovered and submitted by
3675 Tom Wu <tom@arcot.com>.
3676 [Lutz Jaenicke]
3677
3678 *) Fix ssl3_pending() (ssl/s3_lib.c) to prevent SSL_pending() from
3679 returning non-zero before the data has been completely received
3680 when using non-blocking I/O.
3681 [Bodo Moeller; problem pointed out by John Hughes]
3682
3683 *) Some of the ciphers missed the strength entry (SSL_LOW etc).
3684 [Ben Laurie, Lutz Jaenicke]
3685
3686 *) Fix bug in SSL_clear(): bad sessions were not removed (found by
3687 Yoram Zahavi <YoramZ@gilian.com>).
3688 [Lutz Jaenicke]
3689
3690 *) Add information about CygWin 1.3 and on, and preserve proper
3691 configuration for the versions before that.
3692 [Corinna Vinschen <vinschen@redhat.com> and Richard Levitte]
3693
3694 *) Make removal from session cache (SSL_CTX_remove_session()) more robust:
3695 check whether we deal with a copy of a session and do not delete from
3696 the cache in this case. Problem reported by "Izhar Shoshani Levi"
3697 <izhar@checkpoint.com>.
3698 [Lutz Jaenicke]
3699
3700 *) Do not store session data into the internal session cache, if it
3701 is never intended to be looked up (SSL_SESS_CACHE_NO_INTERNAL_LOOKUP
3702 flag is set). Proposed by Aslam <aslam@funk.com>.
3703 [Lutz Jaenicke]
3704
3705 *) Have ASN1_BIT_STRING_set_bit() really clear a bit when the requested
3706 value is 0.
3707 [Richard Levitte]
3708
3709 *) [In 0.9.6d-engine release:]
3710 Fix a crashbug and a logic bug in hwcrhk_load_pubkey().
3711 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte]
3712
3713 *) Add the configuration target linux-s390x.
3714 [Neale Ferguson <Neale.Ferguson@SoftwareAG-USA.com> via Richard Levitte]
3715
3716 *) The earlier bugfix for the SSL3_ST_SW_HELLO_REQ_C case of
3717 ssl3_accept (ssl/s3_srvr.c) incorrectly used a local flag
3718 variable as an indication that a ClientHello message has been
3719 received. As the flag value will be lost between multiple
3720 invocations of ssl3_accept when using non-blocking I/O, the
3721 function may not be aware that a handshake has actually taken
3722 place, thus preventing a new session from being added to the
3723 session cache.
3724
3725 To avoid this problem, we now set s->new_session to 2 instead of
3726 using a local variable.
3727 [Lutz Jaenicke, Bodo Moeller]
3728
3729 *) Bugfix: Return -1 from ssl3_get_server_done (ssl3/s3_clnt.c)
3730 if the SSL_R_LENGTH_MISMATCH error is detected.
3731 [Geoff Thorpe, Bodo Moeller]
3732
3733 *) New 'shared_ldflag' column in Configure platform table.
3734 [Richard Levitte]
3735
3736 *) Fix EVP_CIPHER_mode macro.
3737 ["Dan S. Camper" <dan@bti.net>]
3738
3739 *) Fix ssl3_read_bytes (ssl/s3_pkt.c): To ignore messages of unknown
3740 type, we must throw them away by setting rr->length to 0.
3741 [D P Chang <dpc@qualys.com>]
3742
3743 Changes between 0.9.6b and 0.9.6c [21 dec 2001]
3744
3745 *) Fix BN_rand_range bug pointed out by Dominikus Scherkl
3746 <Dominikus.Scherkl@biodata.com>. (The previous implementation
3747 worked incorrectly for those cases where range = 10..._2 and
3748 3*range is two bits longer than range.)
3749 [Bodo Moeller]
3750
3751 *) Only add signing time to PKCS7 structures if it is not already
3752 present.
3753 [Steve Henson]
3754
3755 *) Fix crypto/objects/objects.h: "ld-ce" should be "id-ce",
3756 OBJ_ld_ce should be OBJ_id_ce.
3757 Also some ip-pda OIDs in crypto/objects/objects.txt were
3758 incorrect (cf. RFC 3039).
3759 [Matt Cooper, Frederic Giudicelli, Bodo Moeller]
3760
3761 *) Release CRYPTO_LOCK_DYNLOCK when CRYPTO_destroy_dynlockid()
3762 returns early because it has nothing to do.
3763 [Andy Schneider <andy.schneider@bjss.co.uk>]
3764
3765 *) [In 0.9.6c-engine release:]
3766 Fix mutex callback return values in crypto/engine/hw_ncipher.c.
3767 [Andy Schneider <andy.schneider@bjss.co.uk>]
3768
3769 *) [In 0.9.6c-engine release:]
3770 Add support for Cryptographic Appliance's keyserver technology.
3771 (Use engine 'keyclient')
3772 [Cryptographic Appliances and Geoff Thorpe]
3773
3774 *) Add a configuration entry for OS/390 Unix. The C compiler 'c89'
3775 is called via tools/c89.sh because arguments have to be
3776 rearranged (all '-L' options must appear before the first object
3777 modules).
3778 [Richard Shapiro <rshapiro@abinitio.com>]
3779
3780 *) [In 0.9.6c-engine release:]
3781 Add support for Broadcom crypto accelerator cards, backported
3782 from 0.9.7.
3783 [Broadcom, Nalin Dahyabhai <nalin@redhat.com>, Mark Cox]
3784
3785 *) [In 0.9.6c-engine release:]
3786 Add support for SureWare crypto accelerator cards from
3787 Baltimore Technologies. (Use engine 'sureware')
3788 [Baltimore Technologies and Mark Cox]
3789
3790 *) [In 0.9.6c-engine release:]
3791 Add support for crypto accelerator cards from Accelerated
3792 Encryption Processing, www.aep.ie. (Use engine 'aep')
3793 [AEP Inc. and Mark Cox]
3794
3795 *) Add a configuration entry for gcc on UnixWare.
3796 [Gary Benson <gbenson@redhat.com>]
3797
3798 *) Change ssl/s2_clnt.c and ssl/s2_srvr.c so that received handshake
3799 messages are stored in a single piece (fixed-length part and
3800 variable-length part combined) and fix various bugs found on the way.
3801 [Bodo Moeller]
3802
3803 *) Disable caching in BIO_gethostbyname(), directly use gethostbyname()
3804 instead. BIO_gethostbyname() does not know what timeouts are
3805 appropriate, so entries would stay in cache even when they have
3806 become invalid.
3807 [Bodo Moeller; problem pointed out by Rich Salz <rsalz@zolera.com>
3808
3809 *) Change ssl23_get_client_hello (ssl/s23_srvr.c) behaviour when
3810 faced with a pathologically small ClientHello fragment that does
3811 not contain client_version: Instead of aborting with an error,
3812 simply choose the highest available protocol version (i.e.,
3813 TLS 1.0 unless it is disabled). In practice, ClientHello
3814 messages are never sent like this, but this change gives us
3815 strictly correct behaviour at least for TLS.
3816 [Bodo Moeller]
3817
3818 *) Fix SSL handshake functions and SSL_clear() such that SSL_clear()
3819 never resets s->method to s->ctx->method when called from within
3820 one of the SSL handshake functions.
3821 [Bodo Moeller; problem pointed out by Niko Baric]
3822
3823 *) In ssl3_get_client_hello (ssl/s3_srvr.c), generate a fatal alert
3824 (sent using the client's version number) if client_version is
3825 smaller than the protocol version in use. Also change
3826 ssl23_get_client_hello (ssl/s23_srvr.c) to select TLS 1.0 if
3827 the client demanded SSL 3.0 but only TLS 1.0 is enabled; then
3828 the client will at least see that alert.
3829 [Bodo Moeller]
3830
3831 *) Fix ssl3_get_message (ssl/s3_both.c) to handle message fragmentation
3832 correctly.
3833 [Bodo Moeller]
3834
3835 *) Avoid infinite loop in ssl3_get_message (ssl/s3_both.c) if a
3836 client receives HelloRequest while in a handshake.
3837 [Bodo Moeller; bug noticed by Andy Schneider <andy.schneider@bjss.co.uk>]
3838
3839 *) Bugfix in ssl3_accept (ssl/s3_srvr.c): Case SSL3_ST_SW_HELLO_REQ_C
3840 should end in 'break', not 'goto end' which circuments various
3841 cleanups done in state SSL_ST_OK. But session related stuff
3842 must be disabled for SSL_ST_OK in the case that we just sent a
3843 HelloRequest.
3844
3845 Also avoid some overhead by not calling ssl_init_wbio_buffer()
3846 before just sending a HelloRequest.
3847 [Bodo Moeller, Eric Rescorla <ekr@rtfm.com>]
3848
3849 *) Fix ssl/s3_enc.c, ssl/t1_enc.c and ssl/s3_pkt.c so that we don't
3850 reveal whether illegal block cipher padding was found or a MAC
3851 verification error occured. (Neither SSLerr() codes nor alerts
3852 are directly visible to potential attackers, but the information
3853 may leak via logfiles.)
3854
3855 Similar changes are not required for the SSL 2.0 implementation
3856 because the number of padding bytes is sent in clear for SSL 2.0,
3857 and the extra bytes are just ignored. However ssl/s2_pkt.c
3858 failed to verify that the purported number of padding bytes is in
3859 the legal range.
3860 [Bodo Moeller]
3861
3862 *) Add OpenUNIX-8 support including shared libraries
3863 (Boyd Lynn Gerber <gerberb@zenez.com>).
3864 [Lutz Jaenicke]
3865
3866 *) Improve RSA_padding_check_PKCS1_OAEP() check again to avoid
3867 'wristwatch attack' using huge encoding parameters (cf.
3868 James H. Manger's CRYPTO 2001 paper). Note that the
3869 RSA_PKCS1_OAEP_PADDING case of RSA_private_decrypt() does not use
3870 encoding parameters and hence was not vulnerable.
3871 [Bodo Moeller]
3872
3873 *) BN_sqr() bug fix.
3874 [Ulf Möller, reported by Jim Ellis <jim.ellis@cavium.com>]
3875
3876 *) Rabin-Miller test analyses assume uniformly distributed witnesses,
3877 so use BN_pseudo_rand_range() instead of using BN_pseudo_rand()
3878 followed by modular reduction.
3879 [Bodo Moeller; pointed out by Adam Young <AYoung1@NCSUS.JNJ.COM>]
3880
3881 *) Add BN_pseudo_rand_range() with obvious functionality: BN_rand_range()
3882 equivalent based on BN_pseudo_rand() instead of BN_rand().
3883 [Bodo Moeller]
3884
3885 *) s3_srvr.c: allow sending of large client certificate lists (> 16 kB).
3886 This function was broken, as the check for a new client hello message
3887 to handle SGC did not allow these large messages.
3888 (Tracked down by "Douglas E. Engert" <deengert@anl.gov>.)
3889 [Lutz Jaenicke]
3890
3891 *) Add alert descriptions for TLSv1 to SSL_alert_desc_string[_long]().
3892 [Lutz Jaenicke]
3893
3894 *) Fix buggy behaviour of BIO_get_num_renegotiates() and BIO_ctrl()
3895 for BIO_C_GET_WRITE_BUF_SIZE ("Stephen Hinton" <shinton@netopia.com>).
3896 [Lutz Jaenicke]
3897
3898 *) Rework the configuration and shared library support for Tru64 Unix.
3899 The configuration part makes use of modern compiler features and
3900 still retains old compiler behavior for those that run older versions
3901 of the OS. The shared library support part includes a variant that
3902 uses the RPATH feature, and is available through the special
3903 configuration target "alpha-cc-rpath", which will never be selected
3904 automatically.
3905 [Tim Mooney <mooney@dogbert.cc.ndsu.NoDak.edu> via Richard Levitte]
3906
3907 *) In ssl3_get_key_exchange (ssl/s3_clnt.c), call ssl3_get_message()
3908 with the same message size as in ssl3_get_certificate_request().
3909 Otherwise, if no ServerKeyExchange message occurs, CertificateRequest
3910 messages might inadvertently be reject as too long.
3911 [Petr Lampa <lampa@fee.vutbr.cz>]
3912
3913 *) Enhanced support for IA-64 Unix platforms (well, Linux and HP-UX).
3914 [Andy Polyakov]
3915
3916 *) Modified SSL library such that the verify_callback that has been set
3917 specificly for an SSL object with SSL_set_verify() is actually being
3918 used. Before the change, a verify_callback set with this function was
3919 ignored and the verify_callback() set in the SSL_CTX at the time of
3920 the call was used. New function X509_STORE_CTX_set_verify_cb() introduced
3921 to allow the necessary settings.
3922 [Lutz Jaenicke]
3923
3924 *) Initialize static variable in crypto/dsa/dsa_lib.c and crypto/dh/dh_lib.c
3925 explicitly to NULL, as at least on Solaris 8 this seems not always to be
3926 done automatically (in contradiction to the requirements of the C
3927 standard). This made problems when used from OpenSSH.
3928 [Lutz Jaenicke]
3929
3930 *) In OpenSSL 0.9.6a and 0.9.6b, crypto/dh/dh_key.c ignored
3931 dh->length and always used
3932
3933 BN_rand_range(priv_key, dh->p).
3934
3935 BN_rand_range() is not necessary for Diffie-Hellman, and this
3936 specific range makes Diffie-Hellman unnecessarily inefficient if
3937 dh->length (recommended exponent length) is much smaller than the
3938 length of dh->p. We could use BN_rand_range() if the order of
3939 the subgroup was stored in the DH structure, but we only have
3940 dh->length.
3941
3942 So switch back to
3943
3944 BN_rand(priv_key, l, ...)
3945
3946 where 'l' is dh->length if this is defined, or BN_num_bits(dh->p)-1
3947 otherwise.
3948 [Bodo Moeller]
3949
3950 *) In
3951
3952 RSA_eay_public_encrypt
3953 RSA_eay_private_decrypt
3954 RSA_eay_private_encrypt (signing)
3955 RSA_eay_public_decrypt (signature verification)
3956
3957 (default implementations for RSA_public_encrypt,
3958 RSA_private_decrypt, RSA_private_encrypt, RSA_public_decrypt),
3959 always reject numbers >= n.
3960 [Bodo Moeller]
3961
3962 *) In crypto/rand/md_rand.c, use a new short-time lock CRYPTO_LOCK_RAND2
3963 to synchronize access to 'locking_thread'. This is necessary on
3964 systems where access to 'locking_thread' (an 'unsigned long'
3965 variable) is not atomic.
3966 [Bodo Moeller]
3967
3968 *) In crypto/rand/md_rand.c, set 'locking_thread' to current thread's ID
3969 *before* setting the 'crypto_lock_rand' flag. The previous code had
3970 a race condition if 0 is a valid thread ID.
3971 [Travis Vitek <vitek@roguewave.com>]
3972
3973 *) Add support for shared libraries under Irix.
3974 [Albert Chin-A-Young <china@thewrittenword.com>]
3975
3976 *) Add configuration option to build on Linux on both big-endian and
3977 little-endian MIPS.
3978 [Ralf Baechle <ralf@uni-koblenz.de>]
3979
3980 *) Add the possibility to create shared libraries on HP-UX.
3981 [Richard Levitte]
3982
3983 Changes between 0.9.6a and 0.9.6b [9 Jul 2001]
3984
3985 *) Change ssleay_rand_bytes (crypto/rand/md_rand.c)
3986 to avoid a SSLeay/OpenSSL PRNG weakness pointed out by
3987 Markku-Juhani O. Saarinen <markku-juhani.saarinen@nokia.com>:
3988 PRNG state recovery was possible based on the output of
3989 one PRNG request appropriately sized to gain knowledge on
3990 'md' followed by enough consecutive 1-byte PRNG requests
3991 to traverse all of 'state'.
3992
3993 1. When updating 'md_local' (the current thread's copy of 'md')
3994 during PRNG output generation, hash all of the previous
3995 'md_local' value, not just the half used for PRNG output.
3996
3997 2. Make the number of bytes from 'state' included into the hash
3998 independent from the number of PRNG bytes requested.
3999
4000 The first measure alone would be sufficient to avoid
4001 Markku-Juhani's attack. (Actually it had never occurred
4002 to me that the half of 'md_local' used for chaining was the
4003 half from which PRNG output bytes were taken -- I had always
4004 assumed that the secret half would be used.) The second
4005 measure makes sure that additional data from 'state' is never
4006 mixed into 'md_local' in small portions; this heuristically
4007 further strengthens the PRNG.
4008 [Bodo Moeller]
4009
4010 *) Fix crypto/bn/asm/mips3.s.
4011 [Andy Polyakov]
4012
4013 *) When only the key is given to "enc", the IV is undefined. Print out
4014 an error message in this case.
4015 [Lutz Jaenicke]
4016
4017 *) Handle special case when X509_NAME is empty in X509 printing routines.
4018 [Steve Henson]
4019
4020 *) In dsa_do_verify (crypto/dsa/dsa_ossl.c), verify that r and s are
4021 positive and less than q.
4022 [Bodo Moeller]
4023
4024 *) Don't change *pointer in CRYPTO_add_lock() is add_lock_callback is
4025 used: it isn't thread safe and the add_lock_callback should handle
4026 that itself.
4027 [Paul Rose <Paul.Rose@bridge.com>]
4028
4029 *) Verify that incoming data obeys the block size in
4030 ssl3_enc (ssl/s3_enc.c) and tls1_enc (ssl/t1_enc.c).
4031 [Bodo Moeller]
4032
4033 *) Fix OAEP check.
4034 [Ulf Möller, Bodo Möller]
4035
4036 *) The countermeasure against Bleichbacher's attack on PKCS #1 v1.5
4037 RSA encryption was accidentally removed in s3_srvr.c in OpenSSL 0.9.5
4038 when fixing the server behaviour for backwards-compatible 'client
4039 hello' messages. (Note that the attack is impractical against
4040 SSL 3.0 and TLS 1.0 anyway because length and version checking
4041 means that the probability of guessing a valid ciphertext is
4042 around 2^-40; see section 5 in Bleichenbacher's CRYPTO '98
4043 paper.)
4044
4045 Before 0.9.5, the countermeasure (hide the error by generating a
4046 random 'decryption result') did not work properly because
4047 ERR_clear_error() was missing, meaning that SSL_get_error() would
4048 detect the supposedly ignored error.
4049
4050 Both problems are now fixed.
4051 [Bodo Moeller]
4052
4053 *) In crypto/bio/bf_buff.c, increase DEFAULT_BUFFER_SIZE to 4096
4054 (previously it was 1024).
4055 [Bodo Moeller]
4056
4057 *) Fix for compatibility mode trust settings: ignore trust settings
4058 unless some valid trust or reject settings are present.
4059 [Steve Henson]
4060
4061 *) Fix for blowfish EVP: its a variable length cipher.
4062 [Steve Henson]
4063
4064 *) Fix various bugs related to DSA S/MIME verification. Handle missing
4065 parameters in DSA public key structures and return an error in the
4066 DSA routines if parameters are absent.
4067 [Steve Henson]
4068
4069 *) In versions up to 0.9.6, RAND_file_name() resorted to file ".rnd"
4070 in the current directory if neither $RANDFILE nor $HOME was set.
4071 RAND_file_name() in 0.9.6a returned NULL in this case. This has
4072 caused some confusion to Windows users who haven't defined $HOME.
4073 Thus RAND_file_name() is changed again: e_os.h can define a
4074 DEFAULT_HOME, which will be used if $HOME is not set.
4075 For Windows, we use "C:"; on other platforms, we still require
4076 environment variables.
4077
4078 *) Move 'if (!initialized) RAND_poll()' into regions protected by
4079 CRYPTO_LOCK_RAND. This is not strictly necessary, but avoids
4080 having multiple threads call RAND_poll() concurrently.
4081 [Bodo Moeller]
4082
4083 *) In crypto/rand/md_rand.c, replace 'add_do_not_lock' flag by a
4084 combination of a flag and a thread ID variable.
4085 Otherwise while one thread is in ssleay_rand_bytes (which sets the
4086 flag), *other* threads can enter ssleay_add_bytes without obeying
4087 the CRYPTO_LOCK_RAND lock (and may even illegally release the lock
4088 that they do not hold after the first thread unsets add_do_not_lock).
4089 [Bodo Moeller]
4090
4091 *) Change bctest again: '-x' expressions are not available in all
4092 versions of 'test'.
4093 [Bodo Moeller]
4094
4095 Changes between 0.9.6 and 0.9.6a [5 Apr 2001]
4096
4097 *) Fix a couple of memory leaks in PKCS7_dataDecode()
4098 [Steve Henson, reported by Heyun Zheng <hzheng@atdsprint.com>]
4099
4100 *) Change Configure and Makefiles to provide EXE_EXT, which will contain
4101 the default extension for executables, if any. Also, make the perl
4102 scripts that use symlink() to test if it really exists and use "cp"
4103 if it doesn't. All this made OpenSSL compilable and installable in
4104 CygWin.
4105 [Richard Levitte]
4106
4107 *) Fix for asn1_GetSequence() for indefinite length constructed data.
4108 If SEQUENCE is length is indefinite just set c->slen to the total
4109 amount of data available.
4110 [Steve Henson, reported by shige@FreeBSD.org]
4111 [This change does not apply to 0.9.7.]
4112
4113 *) Change bctest to avoid here-documents inside command substitution
4114 (workaround for FreeBSD /bin/sh bug).
4115 For compatibility with Ultrix, avoid shell functions (introduced
4116 in the bctest version that searches along $PATH).
4117 [Bodo Moeller]
4118
4119 *) Rename 'des_encrypt' to 'des_encrypt1'. This avoids the clashes
4120 with des_encrypt() defined on some operating systems, like Solaris
4121 and UnixWare.
4122 [Richard Levitte]
4123
4124 *) Check the result of RSA-CRT (see D. Boneh, R. DeMillo, R. Lipton:
4125 On the Importance of Eliminating Errors in Cryptographic
4126 Computations, J. Cryptology 14 (2001) 2, 101-119,
4127 http://theory.stanford.edu/~dabo/papers/faults.ps.gz).
4128 [Ulf Moeller]
4129
4130 *) MIPS assembler BIGNUM division bug fix.
4131 [Andy Polyakov]
4132
4133 *) Disabled incorrect Alpha assembler code.
4134 [Richard Levitte]
4135
4136 *) Fix PKCS#7 decode routines so they correctly update the length
4137 after reading an EOC for the EXPLICIT tag.
4138 [Steve Henson]
4139 [This change does not apply to 0.9.7.]
4140
4141 *) Fix bug in PKCS#12 key generation routines. This was triggered
4142 if a 3DES key was generated with a 0 initial byte. Include
4143 PKCS12_BROKEN_KEYGEN compilation option to retain the old
4144 (but broken) behaviour.
4145 [Steve Henson]
4146
4147 *) Enhance bctest to search for a working bc along $PATH and print
4148 it when found.
4149 [Tim Rice <tim@multitalents.net> via Richard Levitte]
4150
4151 *) Fix memory leaks in err.c: free err_data string if necessary;
4152 don't write to the wrong index in ERR_set_error_data.
4153 [Bodo Moeller]
4154
4155 *) Implement ssl23_peek (analogous to ssl23_read), which previously
4156 did not exist.
4157 [Bodo Moeller]
4158
4159 *) Replace rdtsc with _emit statements for VC++ version 5.
4160 [Jeremy Cooper <jeremy@baymoo.org>]
4161
4162 *) Make it possible to reuse SSLv2 sessions.
4163 [Richard Levitte]
4164
4165 *) In copy_email() check for >= 0 as a return value for
4166 X509_NAME_get_index_by_NID() since 0 is a valid index.
4167 [Steve Henson reported by Massimiliano Pala <madwolf@opensca.org>]
4168
4169 *) Avoid coredump with unsupported or invalid public keys by checking if
4170 X509_get_pubkey() fails in PKCS7_verify(). Fix memory leak when
4171 PKCS7_verify() fails with non detached data.
4172 [Steve Henson]
4173
4174 *) Don't use getenv in library functions when run as setuid/setgid.
4175 New function OPENSSL_issetugid().
4176 [Ulf Moeller]
4177
4178 *) Avoid false positives in memory leak detection code (crypto/mem_dbg.c)
4179 due to incorrect handling of multi-threading:
4180
4181 1. Fix timing glitch in the MemCheck_off() portion of CRYPTO_mem_ctrl().
4182
4183 2. Fix logical glitch in is_MemCheck_on() aka CRYPTO_is_mem_check_on().
4184
4185 3. Count how many times MemCheck_off() has been called so that
4186 nested use can be treated correctly. This also avoids
4187 inband-signalling in the previous code (which relied on the
4188 assumption that thread ID 0 is impossible).
4189 [Bodo Moeller]
4190
4191 *) Add "-rand" option also to s_client and s_server.
4192 [Lutz Jaenicke]
4193
4194 *) Fix CPU detection on Irix 6.x.
4195 [Kurt Hockenbury <khockenb@stevens-tech.edu> and
4196 "Bruce W. Forsberg" <bruce.forsberg@baesystems.com>]
4197
4198 *) Fix X509_NAME bug which produced incorrect encoding if X509_NAME
4199 was empty.
4200 [Steve Henson]
4201 [This change does not apply to 0.9.7.]
4202
4203 *) Use the cached encoding of an X509_NAME structure rather than
4204 copying it. This is apparently the reason for the libsafe "errors"
4205 but the code is actually correct.
4206 [Steve Henson]
4207
4208 *) Add new function BN_rand_range(), and fix DSA_sign_setup() to prevent
4209 Bleichenbacher's DSA attack.
4210 Extend BN_[pseudo_]rand: As before, top=1 forces the highest two bits
4211 to be set and top=0 forces the highest bit to be set; top=-1 is new
4212 and leaves the highest bit random.
4213 [Ulf Moeller, Bodo Moeller]
4214
4215 *) In the NCONF_...-based implementations for CONF_... queries
4216 (crypto/conf/conf_lib.c), if the input LHASH is NULL, avoid using
4217 a temporary CONF structure with the data component set to NULL
4218 (which gives segmentation faults in lh_retrieve).
4219 Instead, use NULL for the CONF pointer in CONF_get_string and
4220 CONF_get_number (which may use environment variables) and directly
4221 return NULL from CONF_get_section.
4222 [Bodo Moeller]
4223
4224 *) Fix potential buffer overrun for EBCDIC.
4225 [Ulf Moeller]
4226
4227 *) Tolerate nonRepudiation as being valid for S/MIME signing and certSign
4228 keyUsage if basicConstraints absent for a CA.
4229 [Steve Henson]
4230
4231 *) Make SMIME_write_PKCS7() write mail header values with a format that
4232 is more generally accepted (no spaces before the semicolon), since
4233 some programs can't parse those values properly otherwise. Also make
4234 sure BIO's that break lines after each write do not create invalid
4235 headers.
4236 [Richard Levitte]
4237
4238 *) Make the CRL encoding routines work with empty SEQUENCE OF. The
4239 macros previously used would not encode an empty SEQUENCE OF
4240 and break the signature.
4241 [Steve Henson]
4242 [This change does not apply to 0.9.7.]
4243
4244 *) Zero the premaster secret after deriving the master secret in
4245 DH ciphersuites.
4246 [Steve Henson]
4247
4248 *) Add some EVP_add_digest_alias registrations (as found in
4249 OpenSSL_add_all_digests()) to SSL_library_init()
4250 aka OpenSSL_add_ssl_algorithms(). This provides improved
4251 compatibility with peers using X.509 certificates
4252 with unconventional AlgorithmIdentifier OIDs.
4253 [Bodo Moeller]
4254
4255 *) Fix for Irix with NO_ASM.
4256 ["Bruce W. Forsberg" <bruce.forsberg@baesystems.com>]
4257
4258 *) ./config script fixes.
4259 [Ulf Moeller, Richard Levitte]
4260
4261 *) Fix 'openssl passwd -1'.
4262 [Bodo Moeller]
4263
4264 *) Change PKCS12_key_gen_asc() so it can cope with non null
4265 terminated strings whose length is passed in the passlen
4266 parameter, for example from PEM callbacks. This was done
4267 by adding an extra length parameter to asc2uni().
4268 [Steve Henson, reported by <oddissey@samsung.co.kr>]
4269
4270 *) Fix C code generated by 'openssl dsaparam -C': If a BN_bin2bn
4271 call failed, free the DSA structure.
4272 [Bodo Moeller]
4273
4274 *) Fix to uni2asc() to cope with zero length Unicode strings.
4275 These are present in some PKCS#12 files.
4276 [Steve Henson]
4277
4278 *) Increase s2->wbuf allocation by one byte in ssl2_new (ssl/s2_lib.c).
4279 Otherwise do_ssl_write (ssl/s2_pkt.c) will write beyond buffer limits
4280 when writing a 32767 byte record.
4281 [Bodo Moeller; problem reported by Eric Day <eday@concentric.net>]
4282
4283 *) In RSA_eay_public_{en,ed}crypt and RSA_eay_mod_exp (rsa_eay.c),
4284 obtain lock CRYPTO_LOCK_RSA before setting rsa->_method_mod_{n,p,q}.
4285
4286 (RSA objects have a reference count access to which is protected
4287 by CRYPTO_LOCK_RSA [see rsa_lib.c, s3_srvr.c, ssl_cert.c, ssl_rsa.c],
4288 so they are meant to be shared between threads.)
4289 [Bodo Moeller, Geoff Thorpe; original patch submitted by
4290 "Reddie, Steven" <Steven.Reddie@ca.com>]
4291
4292 *) Fix a deadlock in CRYPTO_mem_leaks().
4293 [Bodo Moeller]
4294
4295 *) Use better test patterns in bntest.
4296 [Ulf Möller]
4297
4298 *) rand_win.c fix for Borland C.
4299 [Ulf Möller]
4300
4301 *) BN_rshift bugfix for n == 0.
4302 [Bodo Moeller]
4303
4304 *) Add a 'bctest' script that checks for some known 'bc' bugs
4305 so that 'make test' does not abort just because 'bc' is broken.
4306 [Bodo Moeller]
4307
4308 *) Store verify_result within SSL_SESSION also for client side to
4309 avoid potential security hole. (Re-used sessions on the client side
4310 always resulted in verify_result==X509_V_OK, not using the original
4311 result of the server certificate verification.)
4312 [Lutz Jaenicke]
4313
4314 *) Fix ssl3_pending: If the record in s->s3->rrec is not of type
4315 SSL3_RT_APPLICATION_DATA, return 0.
4316 Similarly, change ssl2_pending to return 0 if SSL_in_init(s) is true.
4317 [Bodo Moeller]
4318
4319 *) Fix SSL_peek:
4320 Both ssl2_peek and ssl3_peek, which were totally broken in earlier
4321 releases, have been re-implemented by renaming the previous
4322 implementations of ssl2_read and ssl3_read to ssl2_read_internal
4323 and ssl3_read_internal, respectively, and adding 'peek' parameters
4324 to them. The new ssl[23]_{read,peek} functions are calls to
4325 ssl[23]_read_internal with the 'peek' flag set appropriately.
4326 A 'peek' parameter has also been added to ssl3_read_bytes, which
4327 does the actual work for ssl3_read_internal.
4328 [Bodo Moeller]
4329
4330 *) Initialise "ex_data" member of RSA/DSA/DH structures prior to calling
4331 the method-specific "init()" handler. Also clean up ex_data after
4332 calling the method-specific "finish()" handler. Previously, this was
4333 happening the other way round.
4334 [Geoff Thorpe]
4335
4336 *) Increase BN_CTX_NUM (the number of BIGNUMs in a BN_CTX) to 16.
4337 The previous value, 12, was not always sufficient for BN_mod_exp().
4338 [Bodo Moeller]
4339
4340 *) Make sure that shared libraries get the internal name engine with
4341 the full version number and not just 0. This should mark the
4342 shared libraries as not backward compatible. Of course, this should
4343 be changed again when we can guarantee backward binary compatibility.
4344 [Richard Levitte]
4345
4346 *) Fix typo in get_cert_by_subject() in by_dir.c
4347 [Jean-Marc Desperrier <jean-marc.desperrier@certplus.com>]
4348
4349 *) Rework the system to generate shared libraries:
4350
4351 - Make note of the expected extension for the shared libraries and
4352 if there is a need for symbolic links from for example libcrypto.so.0
4353 to libcrypto.so.0.9.7. There is extended info in Configure for
4354 that.
4355
4356 - Make as few rebuilds of the shared libraries as possible.
4357
4358 - Still avoid linking the OpenSSL programs with the shared libraries.
4359
4360 - When installing, install the shared libraries separately from the
4361 static ones.
4362 [Richard Levitte]
4363
4364 *) Fix SSL_CTX_set_read_ahead macro to actually use its argument.
4365
4366 Copy SSL_CTX's read_ahead flag to SSL object directly in SSL_new
4367 and not in SSL_clear because the latter is also used by the
4368 accept/connect functions; previously, the settings made by
4369 SSL_set_read_ahead would be lost during the handshake.
4370 [Bodo Moeller; problems reported by Anders Gertz <gertz@epact.se>]
4371
4372 *) Correct util/mkdef.pl to be selective about disabled algorithms.
4373 Previously, it would create entries for disableed algorithms no
4374 matter what.
4375 [Richard Levitte]
4376
4377 *) Added several new manual pages for SSL_* function.
4378 [Lutz Jaenicke]
4379
4380 Changes between 0.9.5a and 0.9.6 [24 Sep 2000]
4381
4382 *) In ssl23_get_client_hello, generate an error message when faced
4383 with an initial SSL 3.0/TLS record that is too small to contain the
4384 first two bytes of the ClientHello message, i.e. client_version.
4385 (Note that this is a pathologic case that probably has never happened
4386 in real life.) The previous approach was to use the version number
4387 from the record header as a substitute; but our protocol choice
4388 should not depend on that one because it is not authenticated
4389 by the Finished messages.
4390 [Bodo Moeller]
4391
4392 *) More robust randomness gathering functions for Windows.
4393 [Jeffrey Altman <jaltman@columbia.edu>]
4394
4395 *) For compatibility reasons if the flag X509_V_FLAG_ISSUER_CHECK is
4396 not set then we don't setup the error code for issuer check errors
4397 to avoid possibly overwriting other errors which the callback does
4398 handle. If an application does set the flag then we assume it knows
4399 what it is doing and can handle the new informational codes
4400 appropriately.
4401 [Steve Henson]
4402
4403 *) Fix for a nasty bug in ASN1_TYPE handling. ASN1_TYPE is used for
4404 a general "ANY" type, as such it should be able to decode anything
4405 including tagged types. However it didn't check the class so it would
4406 wrongly interpret tagged types in the same way as their universal
4407 counterpart and unknown types were just rejected. Changed so that the
4408 tagged and unknown types are handled in the same way as a SEQUENCE:
4409 that is the encoding is stored intact. There is also a new type
4410 "V_ASN1_OTHER" which is used when the class is not universal, in this
4411 case we have no idea what the actual type is so we just lump them all
4412 together.
4413 [Steve Henson]
4414
4415 *) On VMS, stdout may very well lead to a file that is written to
4416 in a record-oriented fashion. That means that every write() will
4417 write a separate record, which will be read separately by the
4418 programs trying to read from it. This can be very confusing.
4419
4420 The solution is to put a BIO filter in the way that will buffer
4421 text until a linefeed is reached, and then write everything a
4422 line at a time, so every record written will be an actual line,
4423 not chunks of lines and not (usually doesn't happen, but I've
4424 seen it once) several lines in one record. BIO_f_linebuffer() is
4425 the answer.
4426
4427 Currently, it's a VMS-only method, because that's where it has
4428 been tested well enough.
4429 [Richard Levitte]
4430
4431 *) Remove 'optimized' squaring variant in BN_mod_mul_montgomery,
4432 it can return incorrect results.
4433 (Note: The buggy variant was not enabled in OpenSSL 0.9.5a,
4434 but it was in 0.9.6-beta[12].)
4435 [Bodo Moeller]
4436
4437 *) Disable the check for content being present when verifying detached
4438 signatures in pk7_smime.c. Some versions of Netscape (wrongly)
4439 include zero length content when signing messages.
4440 [Steve Henson]
4441
4442 *) New BIO_shutdown_wr macro, which invokes the BIO_C_SHUTDOWN_WR
4443 BIO_ctrl (for BIO pairs).
4444 [Bodo Möller]
4445
4446 *) Add DSO method for VMS.
4447 [Richard Levitte]
4448
4449 *) Bug fix: Montgomery multiplication could produce results with the
4450 wrong sign.
4451 [Ulf Möller]
4452
4453 *) Add RPM specification openssl.spec and modify it to build three
4454 packages. The default package contains applications, application
4455 documentation and run-time libraries. The devel package contains
4456 include files, static libraries and function documentation. The
4457 doc package contains the contents of the doc directory. The original
4458 openssl.spec was provided by Damien Miller <djm@mindrot.org>.
4459 [Richard Levitte]
4460
4461 *) Add a large number of documentation files for many SSL routines.
4462 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>]
4463
4464 *) Add a configuration entry for Sony News 4.
4465 [NAKAJI Hiroyuki <nakaji@tutrp.tut.ac.jp>]
4466
4467 *) Don't set the two most significant bits to one when generating a
4468 random number < q in the DSA library.
4469 [Ulf Möller]
4470
4471 *) New SSL API mode 'SSL_MODE_AUTO_RETRY'. This disables the default
4472 behaviour that SSL_read may result in SSL_ERROR_WANT_READ (even if
4473 the underlying transport is blocking) if a handshake took place.
4474 (The default behaviour is needed by applications such as s_client
4475 and s_server that use select() to determine when to use SSL_read;
4476 but for applications that know in advance when to expect data, it
4477 just makes things more complicated.)
4478 [Bodo Moeller]
4479
4480 *) Add RAND_egd_bytes(), which gives control over the number of bytes read
4481 from EGD.
4482 [Ben Laurie]
4483
4484 *) Add a few more EBCDIC conditionals that make `req' and `x509'
4485 work better on such systems.
4486 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>]
4487
4488 *) Add two demo programs for PKCS12_parse() and PKCS12_create().
4489 Update PKCS12_parse() so it copies the friendlyName and the
4490 keyid to the certificates aux info.
4491 [Steve Henson]
4492
4493 *) Fix bug in PKCS7_verify() which caused an infinite loop
4494 if there was more than one signature.
4495 [Sven Uszpelkat <su@celocom.de>]
4496
4497 *) Major change in util/mkdef.pl to include extra information
4498 about each symbol, as well as presentig variables as well
4499 as functions. This change means that there's n more need
4500 to rebuild the .num files when some algorithms are excluded.
4501 [Richard Levitte]
4502
4503 *) Allow the verify time to be set by an application,
4504 rather than always using the current time.
4505 [Steve Henson]
4506
4507 *) Phase 2 verify code reorganisation. The certificate
4508 verify code now looks up an issuer certificate by a
4509 number of criteria: subject name, authority key id
4510 and key usage. It also verifies self signed certificates
4511 by the same criteria. The main comparison function is
4512 X509_check_issued() which performs these checks.
4513
4514 Lot of changes were necessary in order to support this
4515 without completely rewriting the lookup code.
4516
4517 Authority and subject key identifier are now cached.
4518
4519 The LHASH 'certs' is X509_STORE has now been replaced
4520 by a STACK_OF(X509_OBJECT). This is mainly because an
4521 LHASH can't store or retrieve multiple objects with
4522 the same hash value.
4523
4524 As a result various functions (which were all internal
4525 use only) have changed to handle the new X509_STORE
4526 structure. This will break anything that messed round
4527 with X509_STORE internally.
4528
4529 The functions X509_STORE_add_cert() now checks for an
4530 exact match, rather than just subject name.
4531
4532 The X509_STORE API doesn't directly support the retrieval
4533 of multiple certificates matching a given criteria, however
4534 this can be worked round by performing a lookup first
4535 (which will fill the cache with candidate certificates)
4536 and then examining the cache for matches. This is probably
4537 the best we can do without throwing out X509_LOOKUP
4538 entirely (maybe later...).
4539
4540 The X509_VERIFY_CTX structure has been enhanced considerably.
4541
4542 All certificate lookup operations now go via a get_issuer()
4543 callback. Although this currently uses an X509_STORE it
4544 can be replaced by custom lookups. This is a simple way
4545 to bypass the X509_STORE hackery necessary to make this
4546 work and makes it possible to use more efficient techniques
4547 in future. A very simple version which uses a simple
4548 STACK for its trusted certificate store is also provided
4549 using X509_STORE_CTX_trusted_stack().
4550
4551 The verify_cb() and verify() callbacks now have equivalents
4552 in the X509_STORE_CTX structure.
4553
4554 X509_STORE_CTX also has a 'flags' field which can be used
4555 to customise the verify behaviour.
4556 [Steve Henson]
4557
4558 *) Add new PKCS#7 signing option PKCS7_NOSMIMECAP which
4559 excludes S/MIME capabilities.
4560 [Steve Henson]
4561
4562 *) When a certificate request is read in keep a copy of the
4563 original encoding of the signed data and use it when outputing
4564 again. Signatures then use the original encoding rather than
4565 a decoded, encoded version which may cause problems if the
4566 request is improperly encoded.
4567 [Steve Henson]
4568
4569 *) For consistency with other BIO_puts implementations, call
4570 buffer_write(b, ...) directly in buffer_puts instead of calling
4571 BIO_write(b, ...).
4572
4573 In BIO_puts, increment b->num_write as in BIO_write.
4574 [Peter.Sylvester@EdelWeb.fr]
4575
4576 *) Fix BN_mul_word for the case where the word is 0. (We have to use
4577 BN_zero, we may not return a BIGNUM with an array consisting of
4578 words set to zero.)
4579 [Bodo Moeller]
4580
4581 *) Avoid calling abort() from within the library when problems are
4582 detected, except if preprocessor symbols have been defined
4583 (such as REF_CHECK, BN_DEBUG etc.).
4584 [Bodo Moeller]
4585
4586 *) New openssl application 'rsautl'. This utility can be
4587 used for low level RSA operations. DER public key
4588 BIO/fp routines also added.
4589 [Steve Henson]
4590
4591 *) New Configure entry and patches for compiling on QNX 4.
4592 [Andreas Schneider <andreas@ds3.etech.fh-hamburg.de>]
4593
4594 *) A demo state-machine implementation was sponsored by
4595 Nuron (http://www.nuron.com/) and is now available in
4596 demos/state_machine.
4597 [Ben Laurie]
4598
4599 *) New options added to the 'dgst' utility for signature
4600 generation and verification.
4601 [Steve Henson]
4602
4603 *) Unrecognized PKCS#7 content types are now handled via a
4604 catch all ASN1_TYPE structure. This allows unsupported
4605 types to be stored as a "blob" and an application can
4606 encode and decode it manually.
4607 [Steve Henson]
4608
4609 *) Fix various signed/unsigned issues to make a_strex.c
4610 compile under VC++.
4611 [Oscar Jacobsson <oscar.jacobsson@celocom.com>]
4612
4613 *) ASN1 fixes. i2d_ASN1_OBJECT was not returning the correct
4614 length if passed a buffer. ASN1_INTEGER_to_BN failed
4615 if passed a NULL BN and its argument was negative.
4616 [Steve Henson, pointed out by Sven Heiberg <sven@tartu.cyber.ee>]
4617
4618 *) Modification to PKCS#7 encoding routines to output definite
4619 length encoding. Since currently the whole structures are in
4620 memory there's not real point in using indefinite length
4621 constructed encoding. However if OpenSSL is compiled with
4622 the flag PKCS7_INDEFINITE_ENCODING the old form is used.
4623 [Steve Henson]
4624
4625 *) Added BIO_vprintf() and BIO_vsnprintf().
4626 [Richard Levitte]
4627
4628 *) Added more prefixes to parse for in the the strings written
4629 through a logging bio, to cover all the levels that are available
4630 through syslog. The prefixes are now:
4631
4632 PANIC, EMERG, EMR => LOG_EMERG
4633 ALERT, ALR => LOG_ALERT
4634 CRIT, CRI => LOG_CRIT
4635 ERROR, ERR => LOG_ERR
4636 WARNING, WARN, WAR => LOG_WARNING
4637 NOTICE, NOTE, NOT => LOG_NOTICE
4638 INFO, INF => LOG_INFO
4639 DEBUG, DBG => LOG_DEBUG
4640
4641 and as before, if none of those prefixes are present at the
4642 beginning of the string, LOG_ERR is chosen.
4643
4644 On Win32, the LOG_* levels are mapped according to this:
4645
4646 LOG_EMERG, LOG_ALERT, LOG_CRIT, LOG_ERR => EVENTLOG_ERROR_TYPE
4647 LOG_WARNING => EVENTLOG_WARNING_TYPE
4648 LOG_NOTICE, LOG_INFO, LOG_DEBUG => EVENTLOG_INFORMATION_TYPE
4649
4650 [Richard Levitte]
4651
4652 *) Made it possible to reconfigure with just the configuration
4653 argument "reconf" or "reconfigure". The command line arguments
4654 are stored in Makefile.ssl in the variable CONFIGURE_ARGS,
4655 and are retrieved from there when reconfiguring.
4656 [Richard Levitte]
4657
4658 *) MD4 implemented.
4659 [Assar Westerlund <assar@sics.se>, Richard Levitte]
4660
4661 *) Add the arguments -CAfile and -CApath to the pkcs12 utility.
4662 [Richard Levitte]
4663
4664 *) The obj_dat.pl script was messing up the sorting of object
4665 names. The reason was that it compared the quoted version
4666 of strings as a result "OCSP" > "OCSP Signing" because
4667 " > SPACE. Changed script to store unquoted versions of
4668 names and add quotes on output. It was also omitting some
4669 names from the lookup table if they were given a default
4670 value (that is if SN is missing it is given the same
4671 value as LN and vice versa), these are now added on the
4672 grounds that if an object has a name we should be able to
4673 look it up. Finally added warning output when duplicate
4674 short or long names are found.
4675 [Steve Henson]
4676
4677 *) Changes needed for Tandem NSK.
4678 [Scott Uroff <scott@xypro.com>]
4679
4680 *) Fix SSL 2.0 rollback checking: Due to an off-by-one error in
4681 RSA_padding_check_SSLv23(), special padding was never detected
4682 and thus the SSL 3.0/TLS 1.0 countermeasure against protocol
4683 version rollback attacks was not effective.
4684
4685 In s23_clnt.c, don't use special rollback-attack detection padding
4686 (RSA_SSLV23_PADDING) if SSL 2.0 is the only protocol enabled in the
4687 client; similarly, in s23_srvr.c, don't do the rollback check if
4688 SSL 2.0 is the only protocol enabled in the server.
4689 [Bodo Moeller]
4690
4691 *) Make it possible to get hexdumps of unprintable data with 'openssl
4692 asn1parse'. By implication, the functions ASN1_parse_dump() and
4693 BIO_dump_indent() are added.
4694 [Richard Levitte]
4695
4696 *) New functions ASN1_STRING_print_ex() and X509_NAME_print_ex()
4697 these print out strings and name structures based on various
4698 flags including RFC2253 support and proper handling of
4699 multibyte characters. Added options to the 'x509' utility
4700 to allow the various flags to be set.
4701 [Steve Henson]
4702
4703 *) Various fixes to use ASN1_TIME instead of ASN1_UTCTIME.
4704 Also change the functions X509_cmp_current_time() and
4705 X509_gmtime_adj() work with an ASN1_TIME structure,
4706 this will enable certificates using GeneralizedTime in validity
4707 dates to be checked.
4708 [Steve Henson]
4709
4710 *) Make the NEG_PUBKEY_BUG code (which tolerates invalid
4711 negative public key encodings) on by default,
4712 NO_NEG_PUBKEY_BUG can be set to disable it.
4713 [Steve Henson]
4714
4715 *) New function c2i_ASN1_OBJECT() which acts on ASN1_OBJECT
4716 content octets. An i2c_ASN1_OBJECT is unnecessary because
4717 the encoding can be trivially obtained from the structure.
4718 [Steve Henson]
4719
4720 *) crypto/err.c locking bugfix: Use write locks (CRYPTO_w_[un]lock),
4721 not read locks (CRYPTO_r_[un]lock).
4722 [Bodo Moeller]
4723
4724 *) A first attempt at creating official support for shared
4725 libraries through configuration. I've kept it so the
4726 default is static libraries only, and the OpenSSL programs
4727 are always statically linked for now, but there are
4728 preparations for dynamic linking in place.
4729 This has been tested on Linux and Tru64.
4730 [Richard Levitte]
4731
4732 *) Randomness polling function for Win9x, as described in:
4733 Peter Gutmann, Software Generation of Practically Strong
4734 Random Numbers.
4735 [Ulf Möller]
4736
4737 *) Fix so PRNG is seeded in req if using an already existing
4738 DSA key.
4739 [Steve Henson]
4740
4741 *) New options to smime application. -inform and -outform
4742 allow alternative formats for the S/MIME message including
4743 PEM and DER. The -content option allows the content to be
4744 specified separately. This should allow things like Netscape
4745 form signing output easier to verify.
4746 [Steve Henson]
4747
4748 *) Fix the ASN1 encoding of tags using the 'long form'.
4749 [Steve Henson]
4750
4751 *) New ASN1 functions, i2c_* and c2i_* for INTEGER and BIT
4752 STRING types. These convert content octets to and from the
4753 underlying type. The actual tag and length octets are
4754 already assumed to have been read in and checked. These
4755 are needed because all other string types have virtually
4756 identical handling apart from the tag. By having versions
4757 of the ASN1 functions that just operate on content octets
4758 IMPLICIT tagging can be handled properly. It also allows
4759 the ASN1_ENUMERATED code to be cut down because ASN1_ENUMERATED
4760 and ASN1_INTEGER are identical apart from the tag.
4761 [Steve Henson]
4762
4763 *) Change the handling of OID objects as follows:
4764
4765 - New object identifiers are inserted in objects.txt, following
4766 the syntax given in objects.README.
4767 - objects.pl is used to process obj_mac.num and create a new
4768 obj_mac.h.
4769 - obj_dat.pl is used to create a new obj_dat.h, using the data in
4770 obj_mac.h.
4771
4772 This is currently kind of a hack, and the perl code in objects.pl
4773 isn't very elegant, but it works as I intended. The simplest way
4774 to check that it worked correctly is to look in obj_dat.h and
4775 check the array nid_objs and make sure the objects haven't moved
4776 around (this is important!). Additions are OK, as well as
4777 consistent name changes.
4778 [Richard Levitte]
4779
4780 *) Add BSD-style MD5-based passwords to 'openssl passwd' (option '-1').
4781 [Bodo Moeller]
4782
4783 *) Addition of the command line parameter '-rand file' to 'openssl req'.
4784 The given file adds to whatever has already been seeded into the
4785 random pool through the RANDFILE configuration file option or
4786 environment variable, or the default random state file.
4787 [Richard Levitte]
4788
4789 *) mkstack.pl now sorts each macro group into lexical order.
4790 Previously the output order depended on the order the files
4791 appeared in the directory, resulting in needless rewriting
4792 of safestack.h .
4793 [Steve Henson]
4794
4795 *) Patches to make OpenSSL compile under Win32 again. Mostly
4796 work arounds for the VC++ problem that it treats func() as
4797 func(void). Also stripped out the parts of mkdef.pl that
4798 added extra typesafe functions: these no longer exist.
4799 [Steve Henson]
4800
4801 *) Reorganisation of the stack code. The macros are now all
4802 collected in safestack.h . Each macro is defined in terms of
4803 a "stack macro" of the form SKM_<name>(type, a, b). The
4804 DEBUG_SAFESTACK is now handled in terms of function casts,
4805 this has the advantage of retaining type safety without the
4806 use of additional functions. If DEBUG_SAFESTACK is not defined
4807 then the non typesafe macros are used instead. Also modified the
4808 mkstack.pl script to handle the new form. Needs testing to see
4809 if which (if any) compilers it chokes and maybe make DEBUG_SAFESTACK
4810 the default if no major problems. Similar behaviour for ASN1_SET_OF
4811 and PKCS12_STACK_OF.
4812 [Steve Henson]
4813
4814 *) When some versions of IIS use the 'NET' form of private key the
4815 key derivation algorithm is different. Normally MD5(password) is
4816 used as a 128 bit RC4 key. In the modified case
4817 MD5(MD5(password) + "SGCKEYSALT") is used insted. Added some
4818 new functions i2d_RSA_NET(), d2i_RSA_NET() etc which are the same
4819 as the old Netscape_RSA functions except they have an additional
4820 'sgckey' parameter which uses the modified algorithm. Also added
4821 an -sgckey command line option to the rsa utility. Thanks to
4822 Adrian Peck <bertie@ncipher.com> for posting details of the modified
4823 algorithm to openssl-dev.
4824 [Steve Henson]
4825
4826 *) The evp_local.h macros were using 'c.##kname' which resulted in
4827 invalid expansion on some systems (SCO 5.0.5 for example).
4828 Corrected to 'c.kname'.
4829 [Phillip Porch <root@theporch.com>]
4830
4831 *) New X509_get1_email() and X509_REQ_get1_email() functions that return
4832 a STACK of email addresses from a certificate or request, these look
4833 in the subject name and the subject alternative name extensions and
4834 omit any duplicate addresses.
4835 [Steve Henson]
4836
4837 *) Re-implement BN_mod_exp2_mont using independent (and larger) windows.
4838 This makes DSA verification about 2 % faster.
4839 [Bodo Moeller]
4840
4841 *) Increase maximum window size in BN_mod_exp_... to 6 bits instead of 5
4842 (meaning that now 2^5 values will be precomputed, which is only 4 KB
4843 plus overhead for 1024 bit moduli).
4844 This makes exponentiations about 0.5 % faster for 1024 bit
4845 exponents (as measured by "openssl speed rsa2048").
4846 [Bodo Moeller]
4847
4848 *) Rename memory handling macros to avoid conflicts with other
4849 software:
4850 Malloc => OPENSSL_malloc
4851 Malloc_locked => OPENSSL_malloc_locked
4852 Realloc => OPENSSL_realloc
4853 Free => OPENSSL_free
4854 [Richard Levitte]
4855
4856 *) New function BN_mod_exp_mont_word for small bases (roughly 15%
4857 faster than BN_mod_exp_mont, i.e. 7% for a full DH exchange).
4858 [Bodo Moeller]
4859
4860 *) CygWin32 support.
4861 [John Jarvie <jjarvie@newsguy.com>]
4862
4863 *) The type-safe stack code has been rejigged. It is now only compiled
4864 in when OpenSSL is configured with the DEBUG_SAFESTACK option and
4865 by default all type-specific stack functions are "#define"d back to
4866 standard stack functions. This results in more streamlined output
4867 but retains the type-safety checking possibilities of the original
4868 approach.
4869 [Geoff Thorpe]
4870
4871 *) The STACK code has been cleaned up, and certain type declarations
4872 that didn't make a lot of sense have been brought in line. This has
4873 also involved a cleanup of sorts in safestack.h to more correctly
4874 map type-safe stack functions onto their plain stack counterparts.
4875 This work has also resulted in a variety of "const"ifications of
4876 lots of the code, especially "_cmp" operations which should normally
4877 be prototyped with "const" parameters anyway.
4878 [Geoff Thorpe]
4879
4880 *) When generating bytes for the first time in md_rand.c, 'stir the pool'
4881 by seeding with STATE_SIZE dummy bytes (with zero entropy count).
4882 (The PRNG state consists of two parts, the large pool 'state' and 'md',
4883 where all of 'md' is used each time the PRNG is used, but 'state'
4884 is used only indexed by a cyclic counter. As entropy may not be
4885 well distributed from the beginning, 'md' is important as a
4886 chaining variable. However, the output function chains only half
4887 of 'md', i.e. 80 bits. ssleay_rand_add, on the other hand, chains
4888 all of 'md', and seeding with STATE_SIZE dummy bytes will result
4889 in all of 'state' being rewritten, with the new values depending
4890 on virtually all of 'md'. This overcomes the 80 bit limitation.)
4891 [Bodo Moeller]
4892
4893 *) In ssl/s2_clnt.c and ssl/s3_clnt.c, call ERR_clear_error() when
4894 the handshake is continued after ssl_verify_cert_chain();
4895 otherwise, if SSL_VERIFY_NONE is set, remaining error codes
4896 can lead to 'unexplainable' connection aborts later.
4897 [Bodo Moeller; problem tracked down by Lutz Jaenicke]
4898
4899 *) Major EVP API cipher revision.
4900 Add hooks for extra EVP features. This allows various cipher
4901 parameters to be set in the EVP interface. Support added for variable
4902 key length ciphers via the EVP_CIPHER_CTX_set_key_length() function and
4903 setting of RC2 and RC5 parameters.
4904
4905 Modify EVP_OpenInit() and EVP_SealInit() to cope with variable key length
4906 ciphers.
4907
4908 Remove lots of duplicated code from the EVP library. For example *every*
4909 cipher init() function handles the 'iv' in the same way according to the
4910 cipher mode. They also all do nothing if the 'key' parameter is NULL and
4911 for CFB and OFB modes they zero ctx->num.
4912
4913 New functionality allows removal of S/MIME code RC2 hack.
4914
4915 Most of the routines have the same form and so can be declared in terms
4916 of macros.
4917
4918 By shifting this to the top level EVP_CipherInit() it can be removed from
4919 all individual ciphers. If the cipher wants to handle IVs or keys
4920 differently it can set the EVP_CIPH_CUSTOM_IV or EVP_CIPH_ALWAYS_CALL_INIT
4921 flags.
4922
4923 Change lots of functions like EVP_EncryptUpdate() to now return a
4924 value: although software versions of the algorithms cannot fail
4925 any installed hardware versions can.
4926 [Steve Henson]
4927
4928 *) Implement SSL_OP_TLS_ROLLBACK_BUG: In ssl3_get_client_key_exchange, if
4929 this option is set, tolerate broken clients that send the negotiated
4930 protocol version number instead of the requested protocol version
4931 number.
4932 [Bodo Moeller]
4933
4934 *) Call dh_tmp_cb (set by ..._TMP_DH_CB) with correct 'is_export' flag;
4935 i.e. non-zero for export ciphersuites, zero otherwise.
4936 Previous versions had this flag inverted, inconsistent with
4937 rsa_tmp_cb (..._TMP_RSA_CB).
4938 [Bodo Moeller; problem reported by Amit Chopra]
4939
4940 *) Add missing DSA library text string. Work around for some IIS
4941 key files with invalid SEQUENCE encoding.
4942 [Steve Henson]
4943
4944 *) Add a document (doc/standards.txt) that list all kinds of standards
4945 and so on that are implemented in OpenSSL.
4946 [Richard Levitte]
4947
4948 *) Enhance c_rehash script. Old version would mishandle certificates
4949 with the same subject name hash and wouldn't handle CRLs at all.
4950 Added -fingerprint option to crl utility, to support new c_rehash
4951 features.
4952 [Steve Henson]
4953
4954 *) Eliminate non-ANSI declarations in crypto.h and stack.h.
4955 [Ulf Möller]
4956
4957 *) Fix for SSL server purpose checking. Server checking was
4958 rejecting certificates which had extended key usage present
4959 but no ssl client purpose.
4960 [Steve Henson, reported by Rene Grosser <grosser@hisolutions.com>]
4961
4962 *) Make PKCS#12 code work with no password. The PKCS#12 spec
4963 is a little unclear about how a blank password is handled.
4964 Since the password in encoded as a BMPString with terminating
4965 double NULL a zero length password would end up as just the
4966 double NULL. However no password at all is different and is
4967 handled differently in the PKCS#12 key generation code. NS
4968 treats a blank password as zero length. MSIE treats it as no
4969 password on export: but it will try both on import. We now do
4970 the same: PKCS12_parse() tries zero length and no password if
4971 the password is set to "" or NULL (NULL is now a valid password:
4972 it wasn't before) as does the pkcs12 application.
4973 [Steve Henson]
4974
4975 *) Bugfixes in apps/x509.c: Avoid a memory leak; and don't use
4976 perror when PEM_read_bio_X509_REQ fails, the error message must
4977 be obtained from the error queue.
4978 [Bodo Moeller]
4979
4980 *) Avoid 'thread_hash' memory leak in crypto/err/err.c by freeing
4981 it in ERR_remove_state if appropriate, and change ERR_get_state
4982 accordingly to avoid race conditions (this is necessary because
4983 thread_hash is no longer constant once set).
4984 [Bodo Moeller]
4985
4986 *) Bugfix for linux-elf makefile.one.
4987 [Ulf Möller]
4988
4989 *) RSA_get_default_method() will now cause a default
4990 RSA_METHOD to be chosen if one doesn't exist already.
4991 Previously this was only set during a call to RSA_new()
4992 or RSA_new_method(NULL) meaning it was possible for
4993 RSA_get_default_method() to return NULL.
4994 [Geoff Thorpe]
4995
4996 *) Added native name translation to the existing DSO code
4997 that will convert (if the flag to do so is set) filenames
4998 that are sufficiently small and have no path information
4999 into a canonical native form. Eg. "blah" converted to
5000 "libblah.so" or "blah.dll" etc.
5001 [Geoff Thorpe]
5002
5003 *) New function ERR_error_string_n(e, buf, len) which is like
5004 ERR_error_string(e, buf), but writes at most 'len' bytes
5005 including the 0 terminator. For ERR_error_string_n, 'buf'
5006 may not be NULL.
5007 [Damien Miller <djm@mindrot.org>, Bodo Moeller]
5008
5009 *) CONF library reworked to become more general. A new CONF
5010 configuration file reader "class" is implemented as well as a
5011 new functions (NCONF_*, for "New CONF") to handle it. The now
5012 old CONF_* functions are still there, but are reimplemented to
5013 work in terms of the new functions. Also, a set of functions
5014 to handle the internal storage of the configuration data is
5015 provided to make it easier to write new configuration file
5016 reader "classes" (I can definitely see something reading a
5017 configuration file in XML format, for example), called _CONF_*,
5018 or "the configuration storage API"...
5019
5020 The new configuration file reading functions are:
5021
5022 NCONF_new, NCONF_free, NCONF_load, NCONF_load_fp, NCONF_load_bio,
5023 NCONF_get_section, NCONF_get_string, NCONF_get_numbre
5024
5025 NCONF_default, NCONF_WIN32
5026
5027 NCONF_dump_fp, NCONF_dump_bio
5028
5029 NCONF_default and NCONF_WIN32 are method (or "class") choosers,
5030 NCONF_new creates a new CONF object. This works in the same way
5031 as other interfaces in OpenSSL, like the BIO interface.
5032 NCONF_dump_* dump the internal storage of the configuration file,
5033 which is useful for debugging. All other functions take the same
5034 arguments as the old CONF_* functions wth the exception of the
5035 first that must be a `CONF *' instead of a `LHASH *'.
5036
5037 To make it easer to use the new classes with the old CONF_* functions,
5038 the function CONF_set_default_method is provided.
5039 [Richard Levitte]
5040
5041 *) Add '-tls1' option to 'openssl ciphers', which was already
5042 mentioned in the documentation but had not been implemented.
5043 (This option is not yet really useful because even the additional
5044 experimental TLS 1.0 ciphers are currently treated as SSL 3.0 ciphers.)
5045 [Bodo Moeller]
5046
5047 *) Initial DSO code added into libcrypto for letting OpenSSL (and
5048 OpenSSL-based applications) load shared libraries and bind to
5049 them in a portable way.
5050 [Geoff Thorpe, with contributions from Richard Levitte]
5051
5052 Changes between 0.9.5 and 0.9.5a [1 Apr 2000]
5053
5054 *) Make sure _lrotl and _lrotr are only used with MSVC.
5055
5056 *) Use lock CRYPTO_LOCK_RAND correctly in ssleay_rand_status
5057 (the default implementation of RAND_status).
5058
5059 *) Rename openssl x509 option '-crlext', which was added in 0.9.5,
5060 to '-clrext' (= clear extensions), as intended and documented.
5061 [Bodo Moeller; inconsistency pointed out by Michael Attili
5062 <attili@amaxo.com>]
5063
5064 *) Fix for HMAC. It wasn't zeroing the rest of the block if the key length
5065 was larger than the MD block size.
5066 [Steve Henson, pointed out by Yost William <YostW@tce.com>]
5067
5068 *) Modernise PKCS12_parse() so it uses STACK_OF(X509) for its ca argument
5069 fix a leak when the ca argument was passed as NULL. Stop X509_PUBKEY_set()
5070 using the passed key: if the passed key was a private key the result
5071 of X509_print(), for example, would be to print out all the private key
5072 components.
5073 [Steve Henson]
5074
5075 *) des_quad_cksum() byte order bug fix.
5076 [Ulf Möller, using the problem description in krb4-0.9.7, where
5077 the solution is attributed to Derrick J Brashear <shadow@DEMENTIA.ORG>]
5078
5079 *) Fix so V_ASN1_APP_CHOOSE works again: however its use is strongly
5080 discouraged.
5081 [Steve Henson, pointed out by Brian Korver <briank@cs.stanford.edu>]
5082
5083 *) For easily testing in shell scripts whether some command
5084 'openssl XXX' exists, the new pseudo-command 'openssl no-XXX'
5085 returns with exit code 0 iff no command of the given name is available.
5086 'no-XXX' is printed in this case, 'XXX' otherwise. In both cases,
5087 the output goes to stdout and nothing is printed to stderr.
5088 Additional arguments are always ignored.
5089
5090 Since for each cipher there is a command of the same name,
5091 the 'no-cipher' compilation switches can be tested this way.
5092
5093 ('openssl no-XXX' is not able to detect pseudo-commands such
5094 as 'quit', 'list-XXX-commands', or 'no-XXX' itself.)
5095 [Bodo Moeller]
5096
5097 *) Update test suite so that 'make test' succeeds in 'no-rsa' configuration.
5098 [Bodo Moeller]
5099
5100 *) For SSL_[CTX_]set_tmp_dh, don't create a DH key if SSL_OP_SINGLE_DH_USE
5101 is set; it will be thrown away anyway because each handshake creates
5102 its own key.
5103 ssl_cert_dup, which is used by SSL_new, now copies DH keys in addition
5104 to parameters -- in previous versions (since OpenSSL 0.9.3) the
5105 'default key' from SSL_CTX_set_tmp_dh would always be lost, meanining
5106 you effectivly got SSL_OP_SINGLE_DH_USE when using this macro.
5107 [Bodo Moeller]
5108
5109 *) New s_client option -ign_eof: EOF at stdin is ignored, and
5110 'Q' and 'R' lose their special meanings (quit/renegotiate).
5111 This is part of what -quiet does; unlike -quiet, -ign_eof
5112 does not suppress any output.
5113 [Richard Levitte]
5114
5115 *) Add compatibility options to the purpose and trust code. The
5116 purpose X509_PURPOSE_ANY is "any purpose" which automatically
5117 accepts a certificate or CA, this was the previous behaviour,
5118 with all the associated security issues.
5119
5120 X509_TRUST_COMPAT is the old trust behaviour: only and
5121 automatically trust self signed roots in certificate store. A
5122 new trust setting X509_TRUST_DEFAULT is used to specify that
5123 a purpose has no associated trust setting and it should instead
5124 use the value in the default purpose.
5125 [Steve Henson]
5126
5127 *) Fix the PKCS#8 DSA private key code so it decodes keys again
5128 and fix a memory leak.
5129 [Steve Henson]
5130
5131 *) In util/mkerr.pl (which implements 'make errors'), preserve
5132 reason strings from the previous version of the .c file, as
5133 the default to have only downcase letters (and digits) in
5134 automatically generated reasons codes is not always appropriate.
5135 [Bodo Moeller]
5136
5137 *) In ERR_load_ERR_strings(), build an ERR_LIB_SYS error reason table
5138 using strerror. Previously, ERR_reason_error_string() returned
5139 library names as reason strings for SYSerr; but SYSerr is a special
5140 case where small numbers are errno values, not library numbers.
5141 [Bodo Moeller]
5142
5143 *) Add '-dsaparam' option to 'openssl dhparam' application. This
5144 converts DSA parameters into DH parameters. (When creating parameters,
5145 DSA_generate_parameters is used.)
5146 [Bodo Moeller]
5147
5148 *) Include 'length' (recommended exponent length) in C code generated
5149 by 'openssl dhparam -C'.
5150 [Bodo Moeller]
5151
5152 *) The second argument to set_label in perlasm was already being used
5153 so couldn't be used as a "file scope" flag. Moved to third argument
5154 which was free.
5155 [Steve Henson]
5156
5157 *) In PEM_ASN1_write_bio and some other functions, use RAND_pseudo_bytes
5158 instead of RAND_bytes for encryption IVs and salts.
5159 [Bodo Moeller]
5160
5161 *) Include RAND_status() into RAND_METHOD instead of implementing
5162 it only for md_rand.c Otherwise replacing the PRNG by calling
5163 RAND_set_rand_method would be impossible.
5164 [Bodo Moeller]
5165
5166 *) Don't let DSA_generate_key() enter an infinite loop if the random
5167 number generation fails.
5168 [Bodo Moeller]
5169
5170 *) New 'rand' application for creating pseudo-random output.
5171 [Bodo Moeller]
5172
5173 *) Added configuration support for Linux/IA64
5174 [Rolf Haberrecker <rolf@suse.de>]
5175
5176 *) Assembler module support for Mingw32.
5177 [Ulf Möller]
5178
5179 *) Shared library support for HPUX (in shlib/).
5180 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE> and Anonymous]
5181
5182 *) Shared library support for Solaris gcc.
5183 [Lutz Behnke <behnke@trustcenter.de>]
5184
5185 Changes between 0.9.4 and 0.9.5 [28 Feb 2000]
5186
5187 *) PKCS7_encrypt() was adding text MIME headers twice because they
5188 were added manually and by SMIME_crlf_copy().
5189 [Steve Henson]
5190
5191 *) In bntest.c don't call BN_rand with zero bits argument.
5192 [Steve Henson, pointed out by Andrew W. Gray <agray@iconsinc.com>]
5193
5194 *) BN_mul bugfix: In bn_mul_part_recursion() only the a>a[n] && b>b[n]
5195 case was implemented. This caused BN_div_recp() to fail occasionally.
5196 [Ulf Möller]
5197
5198 *) Add an optional second argument to the set_label() in the perl
5199 assembly language builder. If this argument exists and is set
5200 to 1 it signals that the assembler should use a symbol whose
5201 scope is the entire file, not just the current function. This
5202 is needed with MASM which uses the format label:: for this scope.
5203 [Steve Henson, pointed out by Peter Runestig <peter@runestig.com>]
5204
5205 *) Change the ASN1 types so they are typedefs by default. Before
5206 almost all types were #define'd to ASN1_STRING which was causing
5207 STACK_OF() problems: you couldn't declare STACK_OF(ASN1_UTF8STRING)
5208 for example.
5209 [Steve Henson]
5210
5211 *) Change names of new functions to the new get1/get0 naming
5212 convention: After 'get1', the caller owns a reference count
5213 and has to call ..._free; 'get0' returns a pointer to some
5214 data structure without incrementing reference counters.
5215 (Some of the existing 'get' functions increment a reference
5216 counter, some don't.)
5217 Similarly, 'set1' and 'add1' functions increase reference
5218 counters or duplicate objects.
5219 [Steve Henson]
5220
5221 *) Allow for the possibility of temp RSA key generation failure:
5222 the code used to assume it always worked and crashed on failure.
5223 [Steve Henson]
5224
5225 *) Fix potential buffer overrun problem in BIO_printf().
5226 [Ulf Möller, using public domain code by Patrick Powell; problem
5227 pointed out by David Sacerdote <das33@cornell.edu>]
5228
5229 *) Support EGD <http://www.lothar.com/tech/crypto/>. New functions
5230 RAND_egd() and RAND_status(). In the command line application,
5231 the EGD socket can be specified like a seed file using RANDFILE
5232 or -rand.
5233 [Ulf Möller]
5234
5235 *) Allow the string CERTIFICATE to be tolerated in PKCS#7 structures.
5236 Some CAs (e.g. Verisign) distribute certificates in this form.
5237 [Steve Henson]
5238
5239 *) Remove the SSL_ALLOW_ADH compile option and set the default cipher
5240 list to exclude them. This means that no special compilation option
5241 is needed to use anonymous DH: it just needs to be included in the
5242 cipher list.
5243 [Steve Henson]
5244
5245 *) Change the EVP_MD_CTX_type macro so its meaning consistent with
5246 EVP_MD_type. The old functionality is available in a new macro called
5247 EVP_MD_md(). Change code that uses it and update docs.
5248 [Steve Henson]
5249
5250 *) ..._ctrl functions now have corresponding ..._callback_ctrl functions
5251 where the 'void *' argument is replaced by a function pointer argument.
5252 Previously 'void *' was abused to point to functions, which works on
5253 many platforms, but is not correct. As these functions are usually
5254 called by macros defined in OpenSSL header files, most source code
5255 should work without changes.
5256 [Richard Levitte]
5257
5258 *) <openssl/opensslconf.h> (which is created by Configure) now contains
5259 sections with information on -D... compiler switches used for
5260 compiling the library so that applications can see them. To enable
5261 one of these sections, a pre-processor symbol OPENSSL_..._DEFINES
5262 must be defined. E.g.,
5263 #define OPENSSL_ALGORITHM_DEFINES
5264 #include <openssl/opensslconf.h>
5265 defines all pertinent NO_<algo> symbols, such as NO_IDEA, NO_RSA, etc.
5266 [Richard Levitte, Ulf and Bodo Möller]
5267
5268 *) Bugfix: Tolerate fragmentation and interleaving in the SSL 3/TLS
5269 record layer.
5270 [Bodo Moeller]
5271
5272 *) Change the 'other' type in certificate aux info to a STACK_OF
5273 X509_ALGOR. Although not an AlgorithmIdentifier as such it has
5274 the required ASN1 format: arbitrary types determined by an OID.
5275 [Steve Henson]
5276
5277 *) Add some PEM_write_X509_REQ_NEW() functions and a command line
5278 argument to 'req'. This is not because the function is newer or
5279 better than others it just uses the work 'NEW' in the certificate
5280 request header lines. Some software needs this.
5281 [Steve Henson]
5282
5283 *) Reorganise password command line arguments: now passwords can be
5284 obtained from various sources. Delete the PEM_cb function and make
5285 it the default behaviour: i.e. if the callback is NULL and the
5286 usrdata argument is not NULL interpret it as a null terminated pass
5287 phrase. If usrdata and the callback are NULL then the pass phrase
5288 is prompted for as usual.
5289 [Steve Henson]
5290
5291 *) Add support for the Compaq Atalla crypto accelerator. If it is installed,
5292 the support is automatically enabled. The resulting binaries will
5293 autodetect the card and use it if present.
5294 [Ben Laurie and Compaq Inc.]
5295
5296 *) Work around for Netscape hang bug. This sends certificate request
5297 and server done in one record. Since this is perfectly legal in the
5298 SSL/TLS protocol it isn't a "bug" option and is on by default. See
5299 the bugs/SSLv3 entry for more info.
5300 [Steve Henson]
5301
5302 *) HP-UX tune-up: new unified configs, HP C compiler bug workaround.
5303 [Andy Polyakov]
5304
5305 *) Add -rand argument to smime and pkcs12 applications and read/write
5306 of seed file.
5307 [Steve Henson]
5308
5309 *) New 'passwd' tool for crypt(3) and apr1 password hashes.
5310 [Bodo Moeller]
5311
5312 *) Add command line password options to the remaining applications.
5313 [Steve Henson]
5314
5315 *) Bug fix for BN_div_recp() for numerators with an even number of
5316 bits.
5317 [Ulf Möller]
5318
5319 *) More tests in bntest.c, and changed test_bn output.
5320 [Ulf Möller]
5321
5322 *) ./config recognizes MacOS X now.
5323 [Andy Polyakov]
5324
5325 *) Bug fix for BN_div() when the first words of num and divsor are
5326 equal (it gave wrong results if (rem=(n1-q*d0)&BN_MASK2) < d0).
5327 [Ulf Möller]
5328
5329 *) Add support for various broken PKCS#8 formats, and command line
5330 options to produce them.
5331 [Steve Henson]
5332
5333 *) New functions BN_CTX_start(), BN_CTX_get() and BT_CTX_end() to
5334 get temporary BIGNUMs from a BN_CTX.
5335 [Ulf Möller]
5336
5337 *) Correct return values in BN_mod_exp_mont() and BN_mod_exp2_mont()
5338 for p == 0.
5339 [Ulf Möller]
5340
5341 *) Change the SSLeay_add_all_*() functions to OpenSSL_add_all_*() and
5342 include a #define from the old name to the new. The original intent
5343 was that statically linked binaries could for example just call
5344 SSLeay_add_all_ciphers() to just add ciphers to the table and not
5345 link with digests. This never worked becayse SSLeay_add_all_digests()
5346 and SSLeay_add_all_ciphers() were in the same source file so calling
5347 one would link with the other. They are now in separate source files.
5348 [Steve Henson]
5349
5350 *) Add a new -notext option to 'ca' and a -pubkey option to 'spkac'.
5351 [Steve Henson]
5352
5353 *) Use a less unusual form of the Miller-Rabin primality test (it used
5354 a binary algorithm for exponentiation integrated into the Miller-Rabin
5355 loop, our standard modexp algorithms are faster).
5356 [Bodo Moeller]
5357
5358 *) Support for the EBCDIC character set completed.
5359 [Martin Kraemer <Martin.Kraemer@Mch.SNI.De>]
5360
5361 *) Source code cleanups: use const where appropriate, eliminate casts,
5362 use void * instead of char * in lhash.
5363 [Ulf Möller]
5364
5365 *) Bugfix: ssl3_send_server_key_exchange was not restartable
5366 (the state was not changed to SSL3_ST_SW_KEY_EXCH_B, and because of
5367 this the server could overwrite ephemeral keys that the client
5368 has already seen).
5369 [Bodo Moeller]
5370
5371 *) Turn DSA_is_prime into a macro that calls BN_is_prime,
5372 using 50 iterations of the Rabin-Miller test.
5373
5374 DSA_generate_parameters now uses BN_is_prime_fasttest (with 50
5375 iterations of the Rabin-Miller test as required by the appendix
5376 to FIPS PUB 186[-1]) instead of DSA_is_prime.
5377 As BN_is_prime_fasttest includes trial division, DSA parameter
5378 generation becomes much faster.
5379
5380 This implies a change for the callback functions in DSA_is_prime
5381 and DSA_generate_parameters: The callback function is called once
5382 for each positive witness in the Rabin-Miller test, not just
5383 occasionally in the inner loop; and the parameters to the
5384 callback function now provide an iteration count for the outer
5385 loop rather than for the current invocation of the inner loop.
5386 DSA_generate_parameters additionally can call the callback
5387 function with an 'iteration count' of -1, meaning that a
5388 candidate has passed the trial division test (when q is generated
5389 from an application-provided seed, trial division is skipped).
5390 [Bodo Moeller]
5391
5392 *) New function BN_is_prime_fasttest that optionally does trial
5393 division before starting the Rabin-Miller test and has
5394 an additional BN_CTX * argument (whereas BN_is_prime always
5395 has to allocate at least one BN_CTX).
5396 'callback(1, -1, cb_arg)' is called when a number has passed the
5397 trial division stage.
5398 [Bodo Moeller]
5399
5400 *) Fix for bug in CRL encoding. The validity dates weren't being handled
5401 as ASN1_TIME.
5402 [Steve Henson]
5403
5404 *) New -pkcs12 option to CA.pl script to write out a PKCS#12 file.
5405 [Steve Henson]
5406
5407 *) New function BN_pseudo_rand().
5408 [Ulf Möller]
5409
5410 *) Clean up BN_mod_mul_montgomery(): replace the broken (and unreadable)
5411 bignum version of BN_from_montgomery() with the working code from
5412 SSLeay 0.9.0 (the word based version is faster anyway), and clean up
5413 the comments.
5414 [Ulf Möller]
5415
5416 *) Avoid a race condition in s2_clnt.c (function get_server_hello) that
5417 made it impossible to use the same SSL_SESSION data structure in
5418 SSL2 clients in multiple threads.
5419 [Bodo Moeller]
5420
5421 *) The return value of RAND_load_file() no longer counts bytes obtained
5422 by stat(). RAND_load_file(..., -1) is new and uses the complete file
5423 to seed the PRNG (previously an explicit byte count was required).
5424 [Ulf Möller, Bodo Möller]
5425
5426 *) Clean up CRYPTO_EX_DATA functions, some of these didn't have prototypes
5427 used (char *) instead of (void *) and had casts all over the place.
5428 [Steve Henson]
5429
5430 *) Make BN_generate_prime() return NULL on error if ret!=NULL.
5431 [Ulf Möller]
5432
5433 *) Retain source code compatibility for BN_prime_checks macro:
5434 BN_is_prime(..., BN_prime_checks, ...) now uses
5435 BN_prime_checks_for_size to determine the appropriate number of
5436 Rabin-Miller iterations.
5437 [Ulf Möller]
5438
5439 *) Diffie-Hellman uses "safe" primes: DH_check() return code renamed to
5440 DH_CHECK_P_NOT_SAFE_PRIME.
5441 (Check if this is true? OpenPGP calls them "strong".)
5442 [Ulf Möller]
5443
5444 *) Merge the functionality of "dh" and "gendh" programs into a new program
5445 "dhparam". The old programs are retained for now but will handle DH keys
5446 (instead of parameters) in future.
5447 [Steve Henson]
5448
5449 *) Make the ciphers, s_server and s_client programs check the return values
5450 when a new cipher list is set.
5451 [Steve Henson]
5452
5453 *) Enhance the SSL/TLS cipher mechanism to correctly handle the TLS 56bit
5454 ciphers. Before when the 56bit ciphers were enabled the sorting was
5455 wrong.
5456
5457 The syntax for the cipher sorting has been extended to support sorting by
5458 cipher-strength (using the strength_bits hard coded in the tables).
5459 The new command is "@STRENGTH" (see also doc/apps/ciphers.pod).
5460
5461 Fix a bug in the cipher-command parser: when supplying a cipher command
5462 string with an "undefined" symbol (neither command nor alphanumeric
5463 [A-Za-z0-9], ssl_set_cipher_list used to hang in an endless loop. Now
5464 an error is flagged.
5465
5466 Due to the strength-sorting extension, the code of the
5467 ssl_create_cipher_list() function was completely rearranged. I hope that
5468 the readability was also increased :-)
5469 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>]
5470
5471 *) Minor change to 'x509' utility. The -CAcreateserial option now uses 1
5472 for the first serial number and places 2 in the serial number file. This
5473 avoids problems when the root CA is created with serial number zero and
5474 the first user certificate has the same issuer name and serial number
5475 as the root CA.
5476 [Steve Henson]
5477
5478 *) Fixes to X509_ATTRIBUTE utilities, change the 'req' program so it uses
5479 the new code. Add documentation for this stuff.
5480 [Steve Henson]
5481
5482 *) Changes to X509_ATTRIBUTE utilities. These have been renamed from
5483 X509_*() to X509at_*() on the grounds that they don't handle X509
5484 structures and behave in an analagous way to the X509v3 functions:
5485 they shouldn't be called directly but wrapper functions should be used
5486 instead.
5487
5488 So we also now have some wrapper functions that call the X509at functions
5489 when passed certificate requests. (TO DO: similar things can be done with
5490 PKCS#7 signed and unsigned attributes, PKCS#12 attributes and a few other
5491 things. Some of these need some d2i or i2d and print functionality
5492 because they handle more complex structures.)
5493 [Steve Henson]
5494
5495 *) Add missing #ifndefs that caused missing symbols when building libssl
5496 as a shared library without RSA. Use #ifndef NO_SSL2 instead of
5497 NO_RSA in ssl/s2*.c.
5498 [Kris Kennaway <kris@hub.freebsd.org>, modified by Ulf Möller]
5499
5500 *) Precautions against using the PRNG uninitialized: RAND_bytes() now
5501 has a return value which indicates the quality of the random data
5502 (1 = ok, 0 = not seeded). Also an error is recorded on the thread's
5503 error queue. New function RAND_pseudo_bytes() generates output that is
5504 guaranteed to be unique but not unpredictable. RAND_add is like
5505 RAND_seed, but takes an extra argument for an entropy estimate
5506 (RAND_seed always assumes full entropy).
5507 [Ulf Möller]
5508
5509 *) Do more iterations of Rabin-Miller probable prime test (specifically,
5510 3 for 1024-bit primes, 6 for 512-bit primes, 12 for 256-bit primes
5511 instead of only 2 for all lengths; see BN_prime_checks_for_size definition
5512 in crypto/bn/bn_prime.c for the complete table). This guarantees a
5513 false-positive rate of at most 2^-80 for random input.
5514 [Bodo Moeller]
5515
5516 *) Rewrite ssl3_read_n (ssl/s3_pkt.c) avoiding a couple of bugs.
5517 [Bodo Moeller]
5518
5519 *) New function X509_CTX_rget_chain() (renamed to X509_CTX_get1_chain
5520 in the 0.9.5 release), this returns the chain
5521 from an X509_CTX structure with a dup of the stack and all
5522 the X509 reference counts upped: so the stack will exist
5523 after X509_CTX_cleanup() has been called. Modify pkcs12.c
5524 to use this.
5525
5526 Also make SSL_SESSION_print() print out the verify return
5527 code.
5528 [Steve Henson]
5529
5530 *) Add manpage for the pkcs12 command. Also change the default
5531 behaviour so MAC iteration counts are used unless the new
5532 -nomaciter option is used. This improves file security and
5533 only older versions of MSIE (4.0 for example) need it.
5534 [Steve Henson]
5535
5536 *) Honor the no-xxx Configure options when creating .DEF files.
5537 [Ulf Möller]
5538
5539 *) Add PKCS#10 attributes to field table: challengePassword,
5540 unstructuredName and unstructuredAddress. These are taken from
5541 draft PKCS#9 v2.0 but are compatible with v1.2 provided no
5542 international characters are used.
5543
5544 More changes to X509_ATTRIBUTE code: allow the setting of types
5545 based on strings. Remove the 'loc' parameter when adding
5546 attributes because these will be a SET OF encoding which is sorted
5547 in ASN1 order.
5548 [Steve Henson]
5549
5550 *) Initial changes to the 'req' utility to allow request generation
5551 automation. This will allow an application to just generate a template
5552 file containing all the field values and have req construct the
5553 request.
5554
5555 Initial support for X509_ATTRIBUTE handling. Stacks of these are
5556 used all over the place including certificate requests and PKCS#7
5557 structures. They are currently handled manually where necessary with
5558 some primitive wrappers for PKCS#7. The new functions behave in a
5559 manner analogous to the X509 extension functions: they allow
5560 attributes to be looked up by NID and added.
5561
5562 Later something similar to the X509V3 code would be desirable to
5563 automatically handle the encoding, decoding and printing of the
5564 more complex types. The string types like challengePassword can
5565 be handled by the string table functions.
5566
5567 Also modified the multi byte string table handling. Now there is
5568 a 'global mask' which masks out certain types. The table itself
5569 can use the flag STABLE_NO_MASK to ignore the mask setting: this
5570 is useful when for example there is only one permissible type
5571 (as in countryName) and using the mask might result in no valid
5572 types at all.
5573 [Steve Henson]
5574
5575 *) Clean up 'Finished' handling, and add functions SSL_get_finished and
5576 SSL_get_peer_finished to allow applications to obtain the latest
5577 Finished messages sent to the peer or expected from the peer,
5578 respectively. (SSL_get_peer_finished is usually the Finished message
5579 actually received from the peer, otherwise the protocol will be aborted.)
5580
5581 As the Finished message are message digests of the complete handshake
5582 (with a total of 192 bits for TLS 1.0 and more for SSL 3.0), they can
5583 be used for external authentication procedures when the authentication
5584 provided by SSL/TLS is not desired or is not enough.
5585 [Bodo Moeller]
5586
5587 *) Enhanced support for Alpha Linux is added. Now ./config checks if
5588 the host supports BWX extension and if Compaq C is present on the
5589 $PATH. Just exploiting of the BWX extension results in 20-30%
5590 performance kick for some algorithms, e.g. DES and RC4 to mention
5591 a couple. Compaq C in turn generates ~20% faster code for MD5 and
5592 SHA1.
5593 [Andy Polyakov]
5594
5595 *) Add support for MS "fast SGC". This is arguably a violation of the
5596 SSL3/TLS protocol. Netscape SGC does two handshakes: the first with
5597 weak crypto and after checking the certificate is SGC a second one
5598 with strong crypto. MS SGC stops the first handshake after receiving
5599 the server certificate message and sends a second client hello. Since
5600 a server will typically do all the time consuming operations before
5601 expecting any further messages from the client (server key exchange
5602 is the most expensive) there is little difference between the two.
5603
5604 To get OpenSSL to support MS SGC we have to permit a second client
5605 hello message after we have sent server done. In addition we have to
5606 reset the MAC if we do get this second client hello.
5607 [Steve Henson]
5608
5609 *) Add a function 'd2i_AutoPrivateKey()' this will automatically decide
5610 if a DER encoded private key is RSA or DSA traditional format. Changed
5611 d2i_PrivateKey_bio() to use it. This is only needed for the "traditional"
5612 format DER encoded private key. Newer code should use PKCS#8 format which
5613 has the key type encoded in the ASN1 structure. Added DER private key
5614 support to pkcs8 application.
5615 [Steve Henson]
5616
5617 *) SSL 3/TLS 1 servers now don't request certificates when an anonymous
5618 ciphersuites has been selected (as required by the SSL 3/TLS 1
5619 specifications). Exception: When SSL_VERIFY_FAIL_IF_NO_PEER_CERT
5620 is set, we interpret this as a request to violate the specification
5621 (the worst that can happen is a handshake failure, and 'correct'
5622 behaviour would result in a handshake failure anyway).
5623 [Bodo Moeller]
5624
5625 *) In SSL_CTX_add_session, take into account that there might be multiple
5626 SSL_SESSION structures with the same session ID (e.g. when two threads
5627 concurrently obtain them from an external cache).
5628 The internal cache can handle only one SSL_SESSION with a given ID,
5629 so if there's a conflict, we now throw out the old one to achieve
5630 consistency.
5631 [Bodo Moeller]
5632
5633 *) Add OIDs for idea and blowfish in CBC mode. This will allow both
5634 to be used in PKCS#5 v2.0 and S/MIME. Also add checking to
5635 some routines that use cipher OIDs: some ciphers do not have OIDs
5636 defined and so they cannot be used for S/MIME and PKCS#5 v2.0 for
5637 example.
5638 [Steve Henson]
5639
5640 *) Simplify the trust setting structure and code. Now we just have
5641 two sequences of OIDs for trusted and rejected settings. These will
5642 typically have values the same as the extended key usage extension
5643 and any application specific purposes.
5644
5645 The trust checking code now has a default behaviour: it will just
5646 check for an object with the same NID as the passed id. Functions can
5647 be provided to override either the default behaviour or the behaviour
5648 for a given id. SSL client, server and email already have functions
5649 in place for compatibility: they check the NID and also return "trusted"
5650 if the certificate is self signed.
5651 [Steve Henson]
5652
5653 *) Add d2i,i2d bio/fp functions for PrivateKey: these convert the
5654 traditional format into an EVP_PKEY structure.
5655 [Steve Henson]
5656
5657 *) Add a password callback function PEM_cb() which either prompts for
5658 a password if usr_data is NULL or otherwise assumes it is a null
5659 terminated password. Allow passwords to be passed on command line
5660 environment or config files in a few more utilities.
5661 [Steve Henson]
5662
5663 *) Add a bunch of DER and PEM functions to handle PKCS#8 format private
5664 keys. Add some short names for PKCS#8 PBE algorithms and allow them
5665 to be specified on the command line for the pkcs8 and pkcs12 utilities.
5666 Update documentation.
5667 [Steve Henson]
5668
5669 *) Support for ASN1 "NULL" type. This could be handled before by using
5670 ASN1_TYPE but there wasn't any function that would try to read a NULL
5671 and produce an error if it couldn't. For compatibility we also have
5672 ASN1_NULL_new() and ASN1_NULL_free() functions but these are faked and
5673 don't allocate anything because they don't need to.
5674 [Steve Henson]
5675
5676 *) Initial support for MacOS is now provided. Examine INSTALL.MacOS
5677 for details.
5678 [Andy Polyakov, Roy Woods <roy@centicsystems.ca>]
5679
5680 *) Rebuild of the memory allocation routines used by OpenSSL code and
5681 possibly others as well. The purpose is to make an interface that
5682 provide hooks so anyone can build a separate set of allocation and
5683 deallocation routines to be used by OpenSSL, for example memory
5684 pool implementations, or something else, which was previously hard
5685 since Malloc(), Realloc() and Free() were defined as macros having
5686 the values malloc, realloc and free, respectively (except for Win32
5687 compilations). The same is provided for memory debugging code.
5688 OpenSSL already comes with functionality to find memory leaks, but
5689 this gives people a chance to debug other memory problems.
5690
5691 With these changes, a new set of functions and macros have appeared:
5692
5693 CRYPTO_set_mem_debug_functions() [F]
5694 CRYPTO_get_mem_debug_functions() [F]
5695 CRYPTO_dbg_set_options() [F]
5696 CRYPTO_dbg_get_options() [F]
5697 CRYPTO_malloc_debug_init() [M]
5698
5699 The memory debug functions are NULL by default, unless the library
5700 is compiled with CRYPTO_MDEBUG or friends is defined. If someone
5701 wants to debug memory anyway, CRYPTO_malloc_debug_init() (which
5702 gives the standard debugging functions that come with OpenSSL) or
5703 CRYPTO_set_mem_debug_functions() (tells OpenSSL to use functions
5704 provided by the library user) must be used. When the standard
5705 debugging functions are used, CRYPTO_dbg_set_options can be used to
5706 request additional information:
5707 CRYPTO_dbg_set_options(V_CYRPTO_MDEBUG_xxx) corresponds to setting
5708 the CRYPTO_MDEBUG_xxx macro when compiling the library.
5709
5710 Also, things like CRYPTO_set_mem_functions will always give the
5711 expected result (the new set of functions is used for allocation
5712 and deallocation) at all times, regardless of platform and compiler
5713 options.
5714
5715 To finish it up, some functions that were never use in any other
5716 way than through macros have a new API and new semantic:
5717
5718 CRYPTO_dbg_malloc()
5719 CRYPTO_dbg_realloc()
5720 CRYPTO_dbg_free()
5721
5722 All macros of value have retained their old syntax.
5723 [Richard Levitte and Bodo Moeller]
5724
5725 *) Some S/MIME fixes. The OID for SMIMECapabilities was wrong, the
5726 ordering of SMIMECapabilities wasn't in "strength order" and there
5727 was a missing NULL in the AlgorithmIdentifier for the SHA1 signature
5728 algorithm.
5729 [Steve Henson]
5730
5731 *) Some ASN1 types with illegal zero length encoding (INTEGER,
5732 ENUMERATED and OBJECT IDENTIFIER) choked the ASN1 routines.
5733 [Frans Heymans <fheymans@isaserver.be>, modified by Steve Henson]
5734
5735 *) Merge in my S/MIME library for OpenSSL. This provides a simple
5736 S/MIME API on top of the PKCS#7 code, a MIME parser (with enough
5737 functionality to handle multipart/signed properly) and a utility
5738 called 'smime' to call all this stuff. This is based on code I
5739 originally wrote for Celo who have kindly allowed it to be
5740 included in OpenSSL.
5741 [Steve Henson]
5742
5743 *) Add variants des_set_key_checked and des_set_key_unchecked of
5744 des_set_key (aka des_key_sched). Global variable des_check_key
5745 decides which of these is called by des_set_key; this way
5746 des_check_key behaves as it always did, but applications and
5747 the library itself, which was buggy for des_check_key == 1,
5748 have a cleaner way to pick the version they need.
5749 [Bodo Moeller]
5750
5751 *) New function PKCS12_newpass() which changes the password of a
5752 PKCS12 structure.
5753 [Steve Henson]
5754
5755 *) Modify X509_TRUST and X509_PURPOSE so it also uses a static and
5756 dynamic mix. In both cases the ids can be used as an index into the
5757 table. Also modified the X509_TRUST_add() and X509_PURPOSE_add()
5758 functions so they accept a list of the field values and the
5759 application doesn't need to directly manipulate the X509_TRUST
5760 structure.
5761 [Steve Henson]
5762
5763 *) Modify the ASN1_STRING_TABLE stuff so it also uses bsearch and doesn't
5764 need initialising.
5765 [Steve Henson]
5766
5767 *) Modify the way the V3 extension code looks up extensions. This now
5768 works in a similar way to the object code: we have some "standard"
5769 extensions in a static table which is searched with OBJ_bsearch()
5770 and the application can add dynamic ones if needed. The file
5771 crypto/x509v3/ext_dat.h now has the info: this file needs to be
5772 updated whenever a new extension is added to the core code and kept
5773 in ext_nid order. There is a simple program 'tabtest.c' which checks
5774 this. New extensions are not added too often so this file can readily
5775 be maintained manually.
5776
5777 There are two big advantages in doing things this way. The extensions
5778 can be looked up immediately and no longer need to be "added" using
5779 X509V3_add_standard_extensions(): this function now does nothing.
5780 [Side note: I get *lots* of email saying the extension code doesn't
5781 work because people forget to call this function]
5782 Also no dynamic allocation is done unless new extensions are added:
5783 so if we don't add custom extensions there is no need to call
5784 X509V3_EXT_cleanup().
5785 [Steve Henson]
5786
5787 *) Modify enc utility's salting as follows: make salting the default. Add a
5788 magic header, so unsalted files fail gracefully instead of just decrypting
5789 to garbage. This is because not salting is a big security hole, so people
5790 should be discouraged from doing it.
5791 [Ben Laurie]
5792
5793 *) Fixes and enhancements to the 'x509' utility. It allowed a message
5794 digest to be passed on the command line but it only used this
5795 parameter when signing a certificate. Modified so all relevant
5796 operations are affected by the digest parameter including the
5797 -fingerprint and -x509toreq options. Also -x509toreq choked if a
5798 DSA key was used because it didn't fix the digest.
5799 [Steve Henson]
5800
5801 *) Initial certificate chain verify code. Currently tests the untrusted
5802 certificates for consistency with the verify purpose (which is set
5803 when the X509_STORE_CTX structure is set up) and checks the pathlength.
5804
5805 There is a NO_CHAIN_VERIFY compilation option to keep the old behaviour:
5806 this is because it will reject chains with invalid extensions whereas
5807 every previous version of OpenSSL and SSLeay made no checks at all.
5808
5809 Trust code: checks the root CA for the relevant trust settings. Trust
5810 settings have an initial value consistent with the verify purpose: e.g.
5811 if the verify purpose is for SSL client use it expects the CA to be
5812 trusted for SSL client use. However the default value can be changed to
5813 permit custom trust settings: one example of this would be to only trust
5814 certificates from a specific "secure" set of CAs.
5815
5816 Also added X509_STORE_CTX_new() and X509_STORE_CTX_free() functions
5817 which should be used for version portability: especially since the
5818 verify structure is likely to change more often now.
5819
5820 SSL integration. Add purpose and trust to SSL_CTX and SSL and functions
5821 to set them. If not set then assume SSL clients will verify SSL servers
5822 and vice versa.
5823
5824 Two new options to the verify program: -untrusted allows a set of
5825 untrusted certificates to be passed in and -purpose which sets the
5826 intended purpose of the certificate. If a purpose is set then the
5827 new chain verify code is used to check extension consistency.
5828 [Steve Henson]
5829
5830 *) Support for the authority information access extension.
5831 [Steve Henson]
5832
5833 *) Modify RSA and DSA PEM read routines to transparently handle
5834 PKCS#8 format private keys. New *_PUBKEY_* functions that handle
5835 public keys in a format compatible with certificate
5836 SubjectPublicKeyInfo structures. Unfortunately there were already
5837 functions called *_PublicKey_* which used various odd formats so
5838 these are retained for compatibility: however the DSA variants were
5839 never in a public release so they have been deleted. Changed dsa/rsa
5840 utilities to handle the new format: note no releases ever handled public
5841 keys so we should be OK.
5842
5843 The primary motivation for this change is to avoid the same fiasco
5844 that dogs private keys: there are several incompatible private key
5845 formats some of which are standard and some OpenSSL specific and
5846 require various evil hacks to allow partial transparent handling and
5847 even then it doesn't work with DER formats. Given the option anything
5848 other than PKCS#8 should be dumped: but the other formats have to
5849 stay in the name of compatibility.
5850
5851 With public keys and the benefit of hindsight one standard format
5852 is used which works with EVP_PKEY, RSA or DSA structures: though
5853 it clearly returns an error if you try to read the wrong kind of key.
5854
5855 Added a -pubkey option to the 'x509' utility to output the public key.
5856 Also rename the EVP_PKEY_get_*() to EVP_PKEY_rget_*()
5857 (renamed to EVP_PKEY_get1_*() in the OpenSSL 0.9.5 release) and add
5858 EVP_PKEY_rset_*() functions (renamed to EVP_PKEY_set1_*())
5859 that do the same as the EVP_PKEY_assign_*() except they up the
5860 reference count of the added key (they don't "swallow" the
5861 supplied key).
5862 [Steve Henson]
5863
5864 *) Fixes to crypto/x509/by_file.c the code to read in certificates and
5865 CRLs would fail if the file contained no certificates or no CRLs:
5866 added a new function to read in both types and return the number
5867 read: this means that if none are read it will be an error. The
5868 DER versions of the certificate and CRL reader would always fail
5869 because it isn't possible to mix certificates and CRLs in DER format
5870 without choking one or the other routine. Changed this to just read
5871 a certificate: this is the best we can do. Also modified the code
5872 in apps/verify.c to take notice of return codes: it was previously
5873 attempting to read in certificates from NULL pointers and ignoring
5874 any errors: this is one reason why the cert and CRL reader seemed
5875 to work. It doesn't check return codes from the default certificate
5876 routines: these may well fail if the certificates aren't installed.
5877 [Steve Henson]
5878
5879 *) Code to support otherName option in GeneralName.
5880 [Steve Henson]
5881
5882 *) First update to verify code. Change the verify utility
5883 so it warns if it is passed a self signed certificate:
5884 for consistency with the normal behaviour. X509_verify
5885 has been modified to it will now verify a self signed
5886 certificate if *exactly* the same certificate appears
5887 in the store: it was previously impossible to trust a
5888 single self signed certificate. This means that:
5889 openssl verify ss.pem
5890 now gives a warning about a self signed certificate but
5891 openssl verify -CAfile ss.pem ss.pem
5892 is OK.
5893 [Steve Henson]
5894
5895 *) For servers, store verify_result in SSL_SESSION data structure
5896 (and add it to external session representation).
5897 This is needed when client certificate verifications fails,
5898 but an application-provided verification callback (set by
5899 SSL_CTX_set_cert_verify_callback) allows accepting the session
5900 anyway (i.e. leaves x509_store_ctx->error != X509_V_OK
5901 but returns 1): When the session is reused, we have to set
5902 ssl->verify_result to the appropriate error code to avoid
5903 security holes.
5904 [Bodo Moeller, problem pointed out by Lutz Jaenicke]
5905
5906 *) Fix a bug in the new PKCS#7 code: it didn't consider the
5907 case in PKCS7_dataInit() where the signed PKCS7 structure
5908 didn't contain any existing data because it was being created.
5909 [Po-Cheng Chen <pocheng@nst.com.tw>, slightly modified by Steve Henson]
5910
5911 *) Add a salt to the key derivation routines in enc.c. This
5912 forms the first 8 bytes of the encrypted file. Also add a
5913 -S option to allow a salt to be input on the command line.
5914 [Steve Henson]
5915
5916 *) New function X509_cmp(). Oddly enough there wasn't a function
5917 to compare two certificates. We do this by working out the SHA1
5918 hash and comparing that. X509_cmp() will be needed by the trust
5919 code.
5920 [Steve Henson]
5921
5922 *) SSL_get1_session() is like SSL_get_session(), but increments
5923 the reference count in the SSL_SESSION returned.
5924 [Geoff Thorpe <geoff@eu.c2.net>]
5925
5926 *) Fix for 'req': it was adding a null to request attributes.
5927 Also change the X509_LOOKUP and X509_INFO code to handle
5928 certificate auxiliary information.
5929 [Steve Henson]
5930
5931 *) Add support for 40 and 64 bit RC2 and RC4 algorithms: document
5932 the 'enc' command.
5933 [Steve Henson]
5934
5935 *) Add the possibility to add extra information to the memory leak
5936 detecting output, to form tracebacks, showing from where each
5937 allocation was originated: CRYPTO_push_info("constant string") adds
5938 the string plus current file name and line number to a per-thread
5939 stack, CRYPTO_pop_info() does the obvious, CRYPTO_remove_all_info()
5940 is like calling CYRPTO_pop_info() until the stack is empty.
5941 Also updated memory leak detection code to be multi-thread-safe.
5942 [Richard Levitte]
5943
5944 *) Add options -text and -noout to pkcs7 utility and delete the
5945 encryption options which never did anything. Update docs.
5946 [Steve Henson]
5947
5948 *) Add options to some of the utilities to allow the pass phrase
5949 to be included on either the command line (not recommended on
5950 OSes like Unix) or read from the environment. Update the
5951 manpages and fix a few bugs.
5952 [Steve Henson]
5953
5954 *) Add a few manpages for some of the openssl commands.
5955 [Steve Henson]
5956
5957 *) Fix the -revoke option in ca. It was freeing up memory twice,
5958 leaking and not finding already revoked certificates.
5959 [Steve Henson]
5960
5961 *) Extensive changes to support certificate auxiliary information.
5962 This involves the use of X509_CERT_AUX structure and X509_AUX
5963 functions. An X509_AUX function such as PEM_read_X509_AUX()
5964 can still read in a certificate file in the usual way but it
5965 will also read in any additional "auxiliary information". By
5966 doing things this way a fair degree of compatibility can be
5967 retained: existing certificates can have this information added
5968 using the new 'x509' options.
5969
5970 Current auxiliary information includes an "alias" and some trust
5971 settings. The trust settings will ultimately be used in enhanced
5972 certificate chain verification routines: currently a certificate
5973 can only be trusted if it is self signed and then it is trusted
5974 for all purposes.
5975 [Steve Henson]
5976
5977 *) Fix assembler for Alpha (tested only on DEC OSF not Linux or *BSD).
5978 The problem was that one of the replacement routines had not been working
5979 since SSLeay releases. For now the offending routine has been replaced
5980 with non-optimised assembler. Even so, this now gives around 95%
5981 performance improvement for 1024 bit RSA signs.
5982 [Mark Cox]
5983
5984 *) Hack to fix PKCS#7 decryption when used with some unorthodox RC2
5985 handling. Most clients have the effective key size in bits equal to
5986 the key length in bits: so a 40 bit RC2 key uses a 40 bit (5 byte) key.
5987 A few however don't do this and instead use the size of the decrypted key
5988 to determine the RC2 key length and the AlgorithmIdentifier to determine
5989 the effective key length. In this case the effective key length can still
5990 be 40 bits but the key length can be 168 bits for example. This is fixed
5991 by manually forcing an RC2 key into the EVP_PKEY structure because the
5992 EVP code can't currently handle unusual RC2 key sizes: it always assumes
5993 the key length and effective key length are equal.
5994 [Steve Henson]
5995
5996 *) Add a bunch of functions that should simplify the creation of
5997 X509_NAME structures. Now you should be able to do:
5998 X509_NAME_add_entry_by_txt(nm, "CN", MBSTRING_ASC, "Steve", -1, -1, 0);
5999 and have it automatically work out the correct field type and fill in
6000 the structures. The more adventurous can try:
6001 X509_NAME_add_entry_by_txt(nm, field, MBSTRING_UTF8, str, -1, -1, 0);
6002 and it will (hopefully) work out the correct multibyte encoding.
6003 [Steve Henson]
6004
6005 *) Change the 'req' utility to use the new field handling and multibyte
6006 copy routines. Before the DN field creation was handled in an ad hoc
6007 way in req, ca, and x509 which was rather broken and didn't support
6008 BMPStrings or UTF8Strings. Since some software doesn't implement
6009 BMPStrings or UTF8Strings yet, they can be enabled using the config file
6010 using the dirstring_type option. See the new comment in the default
6011 openssl.cnf for more info.
6012 [Steve Henson]
6013
6014 *) Make crypto/rand/md_rand.c more robust:
6015 - Assure unique random numbers after fork().
6016 - Make sure that concurrent threads access the global counter and
6017 md serializably so that we never lose entropy in them
6018 or use exactly the same state in multiple threads.
6019 Access to the large state is not always serializable because
6020 the additional locking could be a performance killer, and
6021 md should be large enough anyway.
6022 [Bodo Moeller]
6023
6024 *) New file apps/app_rand.c with commonly needed functionality
6025 for handling the random seed file.
6026
6027 Use the random seed file in some applications that previously did not:
6028 ca,
6029 dsaparam -genkey (which also ignored its '-rand' option),
6030 s_client,
6031 s_server,
6032 x509 (when signing).
6033 Except on systems with /dev/urandom, it is crucial to have a random
6034 seed file at least for key creation, DSA signing, and for DH exchanges;
6035 for RSA signatures we could do without one.
6036
6037 gendh and gendsa (unlike genrsa) used to read only the first byte
6038 of each file listed in the '-rand' option. The function as previously
6039 found in genrsa is now in app_rand.c and is used by all programs
6040 that support '-rand'.
6041 [Bodo Moeller]
6042
6043 *) In RAND_write_file, use mode 0600 for creating files;
6044 don't just chmod when it may be too late.
6045 [Bodo Moeller]
6046
6047 *) Report an error from X509_STORE_load_locations
6048 when X509_LOOKUP_load_file or X509_LOOKUP_add_dir failed.
6049 [Bill Perry]
6050
6051 *) New function ASN1_mbstring_copy() this copies a string in either
6052 ASCII, Unicode, Universal (4 bytes per character) or UTF8 format
6053 into an ASN1_STRING type. A mask of permissible types is passed
6054 and it chooses the "minimal" type to use or an error if not type
6055 is suitable.
6056 [Steve Henson]
6057
6058 *) Add function equivalents to the various macros in asn1.h. The old
6059 macros are retained with an M_ prefix. Code inside the library can
6060 use the M_ macros. External code (including the openssl utility)
6061 should *NOT* in order to be "shared library friendly".
6062 [Steve Henson]
6063
6064 *) Add various functions that can check a certificate's extensions
6065 to see if it usable for various purposes such as SSL client,
6066 server or S/MIME and CAs of these types. This is currently
6067 VERY EXPERIMENTAL but will ultimately be used for certificate chain
6068 verification. Also added a -purpose flag to x509 utility to
6069 print out all the purposes.
6070 [Steve Henson]
6071
6072 *) Add a CRYPTO_EX_DATA to X509 certificate structure and associated
6073 functions.
6074 [Steve Henson]
6075
6076 *) New X509V3_{X509,CRL,REVOKED}_get_d2i() functions. These will search
6077 for, obtain and decode and extension and obtain its critical flag.
6078 This allows all the necessary extension code to be handled in a
6079 single function call.
6080 [Steve Henson]
6081
6082 *) RC4 tune-up featuring 30-40% performance improvement on most RISC
6083 platforms. See crypto/rc4/rc4_enc.c for further details.
6084 [Andy Polyakov]
6085
6086 *) New -noout option to asn1parse. This causes no output to be produced
6087 its main use is when combined with -strparse and -out to extract data
6088 from a file (which may not be in ASN.1 format).
6089 [Steve Henson]
6090
6091 *) Fix for pkcs12 program. It was hashing an invalid certificate pointer
6092 when producing the local key id.
6093 [Richard Levitte <levitte@stacken.kth.se>]
6094
6095 *) New option -dhparam in s_server. This allows a DH parameter file to be
6096 stated explicitly. If it is not stated then it tries the first server
6097 certificate file. The previous behaviour hard coded the filename
6098 "server.pem".
6099 [Steve Henson]
6100
6101 *) Add -pubin and -pubout options to the rsa and dsa commands. These allow
6102 a public key to be input or output. For example:
6103 openssl rsa -in key.pem -pubout -out pubkey.pem
6104 Also added necessary DSA public key functions to handle this.
6105 [Steve Henson]
6106
6107 *) Fix so PKCS7_dataVerify() doesn't crash if no certificates are contained
6108 in the message. This was handled by allowing
6109 X509_find_by_issuer_and_serial() to tolerate a NULL passed to it.
6110 [Steve Henson, reported by Sampo Kellomaki <sampo@mail.neuronio.pt>]
6111
6112 *) Fix for bug in d2i_ASN1_bytes(): other ASN1 functions add an extra null
6113 to the end of the strings whereas this didn't. This would cause problems
6114 if strings read with d2i_ASN1_bytes() were later modified.
6115 [Steve Henson, reported by Arne Ansper <arne@ats.cyber.ee>]
6116
6117 *) Fix for base64 decode bug. When a base64 bio reads only one line of
6118 data and it contains EOF it will end up returning an error. This is
6119 caused by input 46 bytes long. The cause is due to the way base64
6120 BIOs find the start of base64 encoded data. They do this by trying a
6121 trial decode on each line until they find one that works. When they
6122 do a flag is set and it starts again knowing it can pass all the
6123 data directly through the decoder. Unfortunately it doesn't reset
6124 the context it uses. This means that if EOF is reached an attempt
6125 is made to pass two EOFs through the context and this causes the
6126 resulting error. This can also cause other problems as well. As is
6127 usual with these problems it takes *ages* to find and the fix is
6128 trivial: move one line.
6129 [Steve Henson, reported by ian@uns.ns.ac.yu (Ivan Nejgebauer) ]
6130
6131 *) Ugly workaround to get s_client and s_server working under Windows. The
6132 old code wouldn't work because it needed to select() on sockets and the
6133 tty (for keypresses and to see if data could be written). Win32 only
6134 supports select() on sockets so we select() with a 1s timeout on the
6135 sockets and then see if any characters are waiting to be read, if none
6136 are present then we retry, we also assume we can always write data to
6137 the tty. This isn't nice because the code then blocks until we've
6138 received a complete line of data and it is effectively polling the
6139 keyboard at 1s intervals: however it's quite a bit better than not
6140 working at all :-) A dedicated Windows application might handle this
6141 with an event loop for example.
6142 [Steve Henson]
6143
6144 *) Enhance RSA_METHOD structure. Now there are two extra methods, rsa_sign
6145 and rsa_verify. When the RSA_FLAGS_SIGN_VER option is set these functions
6146 will be called when RSA_sign() and RSA_verify() are used. This is useful
6147 if rsa_pub_dec() and rsa_priv_enc() equivalents are not available.
6148 For this to work properly RSA_public_decrypt() and RSA_private_encrypt()
6149 should *not* be used: RSA_sign() and RSA_verify() must be used instead.
6150 This necessitated the support of an extra signature type NID_md5_sha1
6151 for SSL signatures and modifications to the SSL library to use it instead
6152 of calling RSA_public_decrypt() and RSA_private_encrypt().
6153 [Steve Henson]
6154
6155 *) Add new -verify -CAfile and -CApath options to the crl program, these
6156 will lookup a CRL issuers certificate and verify the signature in a
6157 similar way to the verify program. Tidy up the crl program so it
6158 no longer accesses structures directly. Make the ASN1 CRL parsing a bit
6159 less strict. It will now permit CRL extensions even if it is not
6160 a V2 CRL: this will allow it to tolerate some broken CRLs.
6161 [Steve Henson]
6162
6163 *) Initialize all non-automatic variables each time one of the openssl
6164 sub-programs is started (this is necessary as they may be started
6165 multiple times from the "OpenSSL>" prompt).
6166 [Lennart Bang, Bodo Moeller]
6167
6168 *) Preliminary compilation option RSA_NULL which disables RSA crypto without
6169 removing all other RSA functionality (this is what NO_RSA does). This
6170 is so (for example) those in the US can disable those operations covered
6171 by the RSA patent while allowing storage and parsing of RSA keys and RSA
6172 key generation.
6173 [Steve Henson]
6174
6175 *) Non-copying interface to BIO pairs.
6176 (still largely untested)
6177 [Bodo Moeller]
6178
6179 *) New function ANS1_tag2str() to convert an ASN1 tag to a descriptive
6180 ASCII string. This was handled independently in various places before.
6181 [Steve Henson]
6182
6183 *) New functions UTF8_getc() and UTF8_putc() that parse and generate
6184 UTF8 strings a character at a time.
6185 [Steve Henson]
6186
6187 *) Use client_version from client hello to select the protocol
6188 (s23_srvr.c) and for RSA client key exchange verification
6189 (s3_srvr.c), as required by the SSL 3.0/TLS 1.0 specifications.
6190 [Bodo Moeller]
6191
6192 *) Add various utility functions to handle SPKACs, these were previously
6193 handled by poking round in the structure internals. Added new function
6194 NETSCAPE_SPKI_print() to print out SPKAC and a new utility 'spkac' to
6195 print, verify and generate SPKACs. Based on an original idea from
6196 Massimiliano Pala <madwolf@comune.modena.it> but extensively modified.
6197 [Steve Henson]
6198
6199 *) RIPEMD160 is operational on all platforms and is back in 'make test'.
6200 [Andy Polyakov]
6201
6202 *) Allow the config file extension section to be overwritten on the
6203 command line. Based on an original idea from Massimiliano Pala
6204 <madwolf@comune.modena.it>. The new option is called -extensions
6205 and can be applied to ca, req and x509. Also -reqexts to override
6206 the request extensions in req and -crlexts to override the crl extensions
6207 in ca.
6208 [Steve Henson]
6209
6210 *) Add new feature to the SPKAC handling in ca. Now you can include
6211 the same field multiple times by preceding it by "XXXX." for example:
6212 1.OU="Unit name 1"
6213 2.OU="Unit name 2"
6214 this is the same syntax as used in the req config file.
6215 [Steve Henson]
6216
6217 *) Allow certificate extensions to be added to certificate requests. These
6218 are specified in a 'req_extensions' option of the req section of the
6219 config file. They can be printed out with the -text option to req but
6220 are otherwise ignored at present.
6221 [Steve Henson]
6222
6223 *) Fix a horrible bug in enc_read() in crypto/evp/bio_enc.c: if the first
6224 data read consists of only the final block it would not decrypted because
6225 EVP_CipherUpdate() would correctly report zero bytes had been decrypted.
6226 A misplaced 'break' also meant the decrypted final block might not be
6227 copied until the next read.
6228 [Steve Henson]
6229
6230 *) Initial support for DH_METHOD. Again based on RSA_METHOD. Also added
6231 a few extra parameters to the DH structure: these will be useful if
6232 for example we want the value of 'q' or implement X9.42 DH.
6233 [Steve Henson]
6234
6235 *) Initial support for DSA_METHOD. This is based on the RSA_METHOD and
6236 provides hooks that allow the default DSA functions or functions on a
6237 "per key" basis to be replaced. This allows hardware acceleration and
6238 hardware key storage to be handled without major modification to the
6239 library. Also added low level modexp hooks and CRYPTO_EX structure and
6240 associated functions.
6241 [Steve Henson]
6242
6243 *) Add a new flag to memory BIOs, BIO_FLAG_MEM_RDONLY. This marks the BIO
6244 as "read only": it can't be written to and the buffer it points to will
6245 not be freed. Reading from a read only BIO is much more efficient than
6246 a normal memory BIO. This was added because there are several times when
6247 an area of memory needs to be read from a BIO. The previous method was
6248 to create a memory BIO and write the data to it, this results in two
6249 copies of the data and an O(n^2) reading algorithm. There is a new
6250 function BIO_new_mem_buf() which creates a read only memory BIO from
6251 an area of memory. Also modified the PKCS#7 routines to use read only
6252 memory BIOs.
6253 [Steve Henson]
6254
6255 *) Bugfix: ssl23_get_client_hello did not work properly when called in
6256 state SSL23_ST_SR_CLNT_HELLO_B, i.e. when the first 7 bytes of
6257 a SSLv2-compatible client hello for SSLv3 or TLSv1 could be read,
6258 but a retry condition occured while trying to read the rest.
6259 [Bodo Moeller]
6260
6261 *) The PKCS7_ENC_CONTENT_new() function was setting the content type as
6262 NID_pkcs7_encrypted by default: this was wrong since this should almost
6263 always be NID_pkcs7_data. Also modified the PKCS7_set_type() to handle
6264 the encrypted data type: this is a more sensible place to put it and it
6265 allows the PKCS#12 code to be tidied up that duplicated this
6266 functionality.
6267 [Steve Henson]
6268
6269 *) Changed obj_dat.pl script so it takes its input and output files on
6270 the command line. This should avoid shell escape redirection problems
6271 under Win32.
6272 [Steve Henson]
6273
6274 *) Initial support for certificate extension requests, these are included
6275 in things like Xenroll certificate requests. Included functions to allow
6276 extensions to be obtained and added.
6277 [Steve Henson]
6278
6279 *) -crlf option to s_client and s_server for sending newlines as
6280 CRLF (as required by many protocols).
6281 [Bodo Moeller]
6282
6283 Changes between 0.9.3a and 0.9.4 [09 Aug 1999]
6284
6285 *) Install libRSAglue.a when OpenSSL is built with RSAref.
6286 [Ralf S. Engelschall]
6287
6288 *) A few more ``#ifndef NO_FP_API / #endif'' pairs for consistency.
6289 [Andrija Antonijevic <TheAntony2@bigfoot.com>]
6290
6291 *) Fix -startdate and -enddate (which was missing) arguments to 'ca'
6292 program.
6293 [Steve Henson]
6294
6295 *) New function DSA_dup_DH, which duplicates DSA parameters/keys as
6296 DH parameters/keys (q is lost during that conversion, but the resulting
6297 DH parameters contain its length).
6298
6299 For 1024-bit p, DSA_generate_parameters followed by DSA_dup_DH is
6300 much faster than DH_generate_parameters (which creates parameters
6301 where p = 2*q + 1), and also the smaller q makes DH computations
6302 much more efficient (160-bit exponentiation instead of 1024-bit
6303 exponentiation); so this provides a convenient way to support DHE
6304 ciphersuites in SSL/TLS servers (see ssl/ssltest.c). It is of
6305 utter importance to use
6306 SSL_CTX_set_options(s_ctx, SSL_OP_SINGLE_DH_USE);
6307 or
6308 SSL_set_options(s_ctx, SSL_OP_SINGLE_DH_USE);
6309 when such DH parameters are used, because otherwise small subgroup
6310 attacks may become possible!
6311 [Bodo Moeller]
6312
6313 *) Avoid memory leak in i2d_DHparams.
6314 [Bodo Moeller]
6315
6316 *) Allow the -k option to be used more than once in the enc program:
6317 this allows the same encrypted message to be read by multiple recipients.
6318 [Steve Henson]
6319
6320 *) New function OBJ_obj2txt(buf, buf_len, a, no_name), this converts
6321 an ASN1_OBJECT to a text string. If the "no_name" parameter is set then
6322 it will always use the numerical form of the OID, even if it has a short
6323 or long name.
6324 [Steve Henson]
6325
6326 *) Added an extra RSA flag: RSA_FLAG_EXT_PKEY. Previously the rsa_mod_exp
6327 method only got called if p,q,dmp1,dmq1,iqmp components were present,
6328 otherwise bn_mod_exp was called. In the case of hardware keys for example
6329 no private key components need be present and it might store extra data
6330 in the RSA structure, which cannot be accessed from bn_mod_exp.
6331 By setting RSA_FLAG_EXT_PKEY rsa_mod_exp will always be called for
6332 private key operations.
6333 [Steve Henson]
6334
6335 *) Added support for SPARC Linux.
6336 [Andy Polyakov]
6337
6338 *) pem_password_cb function type incompatibly changed from
6339 typedef int pem_password_cb(char *buf, int size, int rwflag);
6340 to
6341 ....(char *buf, int size, int rwflag, void *userdata);
6342 so that applications can pass data to their callbacks:
6343 The PEM[_ASN1]_{read,write}... functions and macros now take an
6344 additional void * argument, which is just handed through whenever
6345 the password callback is called.
6346 [Damien Miller <dmiller@ilogic.com.au>; tiny changes by Bodo Moeller]
6347
6348 New function SSL_CTX_set_default_passwd_cb_userdata.
6349
6350 Compatibility note: As many C implementations push function arguments
6351 onto the stack in reverse order, the new library version is likely to
6352 interoperate with programs that have been compiled with the old
6353 pem_password_cb definition (PEM_whatever takes some data that
6354 happens to be on the stack as its last argument, and the callback
6355 just ignores this garbage); but there is no guarantee whatsoever that
6356 this will work.
6357
6358 *) The -DPLATFORM="\"$(PLATFORM)\"" definition and the similar -DCFLAGS=...
6359 (both in crypto/Makefile.ssl for use by crypto/cversion.c) caused
6360 problems not only on Windows, but also on some Unix platforms.
6361 To avoid problematic command lines, these definitions are now in an
6362 auto-generated file crypto/buildinf.h (created by crypto/Makefile.ssl
6363 for standard "make" builds, by util/mk1mf.pl for "mk1mf" builds).
6364 [Bodo Moeller]
6365
6366 *) MIPS III/IV assembler module is reimplemented.
6367 [Andy Polyakov]
6368
6369 *) More DES library cleanups: remove references to srand/rand and
6370 delete an unused file.
6371 [Ulf Möller]
6372
6373 *) Add support for the the free Netwide assembler (NASM) under Win32,
6374 since not many people have MASM (ml) and it can be hard to obtain.
6375 This is currently experimental but it seems to work OK and pass all
6376 the tests. Check out INSTALL.W32 for info.
6377 [Steve Henson]
6378
6379 *) Fix memory leaks in s3_clnt.c: All non-anonymous SSL3/TLS1 connections
6380 without temporary keys kept an extra copy of the server key,
6381 and connections with temporary keys did not free everything in case
6382 of an error.
6383 [Bodo Moeller]
6384
6385 *) New function RSA_check_key and new openssl rsa option -check
6386 for verifying the consistency of RSA keys.
6387 [Ulf Moeller, Bodo Moeller]
6388
6389 *) Various changes to make Win32 compile work:
6390 1. Casts to avoid "loss of data" warnings in p5_crpt2.c
6391 2. Change unsigned int to int in b_dump.c to avoid "signed/unsigned
6392 comparison" warnings.
6393 3. Add sk_<TYPE>_sort to DEF file generator and do make update.
6394 [Steve Henson]
6395
6396 *) Add a debugging option to PKCS#5 v2 key generation function: when
6397 you #define DEBUG_PKCS5V2 passwords, salts, iteration counts and
6398 derived keys are printed to stderr.
6399 [Steve Henson]
6400
6401 *) Copy the flags in ASN1_STRING_dup().
6402 [Roman E. Pavlov <pre@mo.msk.ru>]
6403
6404 *) The x509 application mishandled signing requests containing DSA
6405 keys when the signing key was also DSA and the parameters didn't match.
6406
6407 It was supposed to omit the parameters when they matched the signing key:
6408 the verifying software was then supposed to automatically use the CA's
6409 parameters if they were absent from the end user certificate.
6410
6411 Omitting parameters is no longer recommended. The test was also
6412 the wrong way round! This was probably due to unusual behaviour in
6413 EVP_cmp_parameters() which returns 1 if the parameters match.
6414 This meant that parameters were omitted when they *didn't* match and
6415 the certificate was useless. Certificates signed with 'ca' didn't have
6416 this bug.
6417 [Steve Henson, reported by Doug Erickson <Doug.Erickson@Part.NET>]
6418
6419 *) Memory leak checking (-DCRYPTO_MDEBUG) had some problems.
6420 The interface is as follows:
6421 Applications can use
6422 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_ON) aka MemCheck_start(),
6423 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_OFF) aka MemCheck_stop();
6424 "off" is now the default.
6425 The library internally uses
6426 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_DISABLE) aka MemCheck_off(),
6427 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_ENABLE) aka MemCheck_on()
6428 to disable memory-checking temporarily.
6429
6430 Some inconsistent states that previously were possible (and were
6431 even the default) are now avoided.
6432
6433 -DCRYPTO_MDEBUG_TIME is new and additionally stores the current time
6434 with each memory chunk allocated; this is occasionally more helpful
6435 than just having a counter.
6436
6437 -DCRYPTO_MDEBUG_THREAD is also new and adds the thread ID.
6438
6439 -DCRYPTO_MDEBUG_ALL enables all of the above, plus any future
6440 extensions.
6441 [Bodo Moeller]
6442
6443 *) Introduce "mode" for SSL structures (with defaults in SSL_CTX),
6444 which largely parallels "options", but is for changing API behaviour,
6445 whereas "options" are about protocol behaviour.
6446 Initial "mode" flags are:
6447
6448 SSL_MODE_ENABLE_PARTIAL_WRITE Allow SSL_write to report success when
6449 a single record has been written.
6450 SSL_MODE_ACCEPT_MOVING_WRITE_BUFFER Don't insist that SSL_write
6451 retries use the same buffer location.
6452 (But all of the contents must be
6453 copied!)
6454 [Bodo Moeller]
6455
6456 *) Bugfix: SSL_set_options ignored its parameter, only SSL_CTX_set_options
6457 worked.
6458
6459 *) Fix problems with no-hmac etc.
6460 [Ulf Möller, pointed out by Brian Wellington <bwelling@tislabs.com>]
6461
6462 *) New functions RSA_get_default_method(), RSA_set_method() and
6463 RSA_get_method(). These allows replacement of RSA_METHODs without having
6464 to mess around with the internals of an RSA structure.
6465 [Steve Henson]
6466
6467 *) Fix memory leaks in DSA_do_sign and DSA_is_prime.
6468 Also really enable memory leak checks in openssl.c and in some
6469 test programs.
6470 [Chad C. Mulligan, Bodo Moeller]
6471
6472 *) Fix a bug in d2i_ASN1_INTEGER() and i2d_ASN1_INTEGER() which can mess
6473 up the length of negative integers. This has now been simplified to just
6474 store the length when it is first determined and use it later, rather
6475 than trying to keep track of where data is copied and updating it to
6476 point to the end.
6477 [Steve Henson, reported by Brien Wheeler
6478 <bwheeler@authentica-security.com>]
6479
6480 *) Add a new function PKCS7_signatureVerify. This allows the verification
6481 of a PKCS#7 signature but with the signing certificate passed to the
6482 function itself. This contrasts with PKCS7_dataVerify which assumes the
6483 certificate is present in the PKCS#7 structure. This isn't always the
6484 case: certificates can be omitted from a PKCS#7 structure and be
6485 distributed by "out of band" means (such as a certificate database).
6486 [Steve Henson]
6487
6488 *) Complete the PEM_* macros with DECLARE_PEM versions to replace the
6489 function prototypes in pem.h, also change util/mkdef.pl to add the
6490 necessary function names.
6491 [Steve Henson]
6492
6493 *) mk1mf.pl (used by Windows builds) did not properly read the
6494 options set by Configure in the top level Makefile, and Configure
6495 was not even able to write more than one option correctly.
6496 Fixed, now "no-idea no-rc5 -DCRYPTO_MDEBUG" etc. works as intended.
6497 [Bodo Moeller]
6498
6499 *) New functions CONF_load_bio() and CONF_load_fp() to allow a config
6500 file to be loaded from a BIO or FILE pointer. The BIO version will
6501 for example allow memory BIOs to contain config info.
6502 [Steve Henson]
6503
6504 *) New function "CRYPTO_num_locks" that returns CRYPTO_NUM_LOCKS.
6505 Whoever hopes to achieve shared-library compatibility across versions
6506 must use this, not the compile-time macro.
6507 (Exercise 0.9.4: Which is the minimum library version required by
6508 such programs?)
6509 Note: All this applies only to multi-threaded programs, others don't
6510 need locks.
6511 [Bodo Moeller]
6512
6513 *) Add missing case to s3_clnt.c state machine -- one of the new SSL tests
6514 through a BIO pair triggered the default case, i.e.
6515 SSLerr(...,SSL_R_UNKNOWN_STATE).
6516 [Bodo Moeller]
6517
6518 *) New "BIO pair" concept (crypto/bio/bss_bio.c) so that applications
6519 can use the SSL library even if none of the specific BIOs is
6520 appropriate.
6521 [Bodo Moeller]
6522
6523 *) Fix a bug in i2d_DSAPublicKey() which meant it returned the wrong value
6524 for the encoded length.
6525 [Jeon KyoungHo <khjeon@sds.samsung.co.kr>]
6526
6527 *) Add initial documentation of the X509V3 functions.
6528 [Steve Henson]
6529
6530 *) Add a new pair of functions PEM_write_PKCS8PrivateKey() and
6531 PEM_write_bio_PKCS8PrivateKey() that are equivalent to
6532 PEM_write_PrivateKey() and PEM_write_bio_PrivateKey() but use the more
6533 secure PKCS#8 private key format with a high iteration count.
6534 [Steve Henson]
6535
6536 *) Fix determination of Perl interpreter: A perl or perl5
6537 _directory_ in $PATH was also accepted as the interpreter.
6538 [Ralf S. Engelschall]
6539
6540 *) Fix demos/sign/sign.c: well there wasn't anything strictly speaking
6541 wrong with it but it was very old and did things like calling
6542 PEM_ASN1_read() directly and used MD5 for the hash not to mention some
6543 unusual formatting.
6544 [Steve Henson]
6545
6546 *) Fix demos/selfsign.c: it used obsolete and deleted functions, changed
6547 to use the new extension code.
6548 [Steve Henson]
6549
6550 *) Implement the PEM_read/PEM_write functions in crypto/pem/pem_all.c
6551 with macros. This should make it easier to change their form, add extra
6552 arguments etc. Fix a few PEM prototypes which didn't have cipher as a
6553 constant.
6554 [Steve Henson]
6555
6556 *) Add to configuration table a new entry that can specify an alternative
6557 name for unistd.h (for pre-POSIX systems); we need this for NeXTstep,
6558 according to Mark Crispin <MRC@Panda.COM>.
6559 [Bodo Moeller]
6560
6561 #if 0
6562 *) DES CBC did not update the IV. Weird.
6563 [Ben Laurie]
6564 #else
6565 des_cbc_encrypt does not update the IV, but des_ncbc_encrypt does.
6566 Changing the behaviour of the former might break existing programs --
6567 where IV updating is needed, des_ncbc_encrypt can be used.
6568 #endif
6569
6570 *) When bntest is run from "make test" it drives bc to check its
6571 calculations, as well as internally checking them. If an internal check
6572 fails, it needs to cause bc to give a non-zero result or make test carries
6573 on without noticing the failure. Fixed.
6574 [Ben Laurie]
6575
6576 *) DES library cleanups.
6577 [Ulf Möller]
6578
6579 *) Add support for PKCS#5 v2.0 PBE algorithms. This will permit PKCS#8 to be
6580 used with any cipher unlike PKCS#5 v1.5 which can at most handle 64 bit
6581 ciphers. NOTE: although the key derivation function has been verified
6582 against some published test vectors it has not been extensively tested
6583 yet. Added a -v2 "cipher" option to pkcs8 application to allow the use
6584 of v2.0.
6585 [Steve Henson]
6586
6587 *) Instead of "mkdir -p", which is not fully portable, use new
6588 Perl script "util/mkdir-p.pl".
6589 [Bodo Moeller]
6590
6591 *) Rewrite the way password based encryption (PBE) is handled. It used to
6592 assume that the ASN1 AlgorithmIdentifier parameter was a PBEParameter
6593 structure. This was true for the PKCS#5 v1.5 and PKCS#12 PBE algorithms
6594 but doesn't apply to PKCS#5 v2.0 where it can be something else. Now
6595 the 'parameter' field of the AlgorithmIdentifier is passed to the
6596 underlying key generation function so it must do its own ASN1 parsing.
6597 This has also changed the EVP_PBE_CipherInit() function which now has a
6598 'parameter' argument instead of literal salt and iteration count values
6599 and the function EVP_PBE_ALGOR_CipherInit() has been deleted.
6600 [Steve Henson]
6601
6602 *) Support for PKCS#5 v1.5 compatible password based encryption algorithms
6603 and PKCS#8 functionality. New 'pkcs8' application linked to openssl.
6604 Needed to change the PEM_STRING_EVP_PKEY value which was just "PRIVATE
6605 KEY" because this clashed with PKCS#8 unencrypted string. Since this
6606 value was just used as a "magic string" and not used directly its
6607 value doesn't matter.
6608 [Steve Henson]
6609
6610 *) Introduce some semblance of const correctness to BN. Shame C doesn't
6611 support mutable.
6612 [Ben Laurie]
6613
6614 *) "linux-sparc64" configuration (ultrapenguin).
6615 [Ray Miller <ray.miller@oucs.ox.ac.uk>]
6616 "linux-sparc" configuration.
6617 [Christian Forster <fo@hawo.stw.uni-erlangen.de>]
6618
6619 *) config now generates no-xxx options for missing ciphers.
6620 [Ulf Möller]
6621
6622 *) Support the EBCDIC character set (work in progress).
6623 File ebcdic.c not yet included because it has a different license.
6624 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>]
6625
6626 *) Support BS2000/OSD-POSIX.
6627 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>]
6628
6629 *) Make callbacks for key generation use void * instead of char *.
6630 [Ben Laurie]
6631
6632 *) Make S/MIME samples compile (not yet tested).
6633 [Ben Laurie]
6634
6635 *) Additional typesafe stacks.
6636 [Ben Laurie]
6637
6638 *) New configuration variants "bsdi-elf-gcc" (BSD/OS 4.x).
6639 [Bodo Moeller]
6640
6641
6642 Changes between 0.9.3 and 0.9.3a [29 May 1999]
6643
6644 *) New configuration variant "sco5-gcc".
6645
6646 *) Updated some demos.
6647 [Sean O Riordain, Wade Scholine]
6648
6649 *) Add missing BIO_free at exit of pkcs12 application.
6650 [Wu Zhigang]
6651
6652 *) Fix memory leak in conf.c.
6653 [Steve Henson]
6654
6655 *) Updates for Win32 to assembler version of MD5.
6656 [Steve Henson]
6657
6658 *) Set #! path to perl in apps/der_chop to where we found it
6659 instead of using a fixed path.
6660 [Bodo Moeller]
6661
6662 *) SHA library changes for irix64-mips4-cc.
6663 [Andy Polyakov]
6664
6665 *) Improvements for VMS support.
6666 [Richard Levitte]
6667
6668
6669 Changes between 0.9.2b and 0.9.3 [24 May 1999]
6670
6671 *) Bignum library bug fix. IRIX 6 passes "make test" now!
6672 This also avoids the problems with SC4.2 and unpatched SC5.
6673 [Andy Polyakov <appro@fy.chalmers.se>]
6674
6675 *) New functions sk_num, sk_value and sk_set to replace the previous macros.
6676 These are required because of the typesafe stack would otherwise break
6677 existing code. If old code used a structure member which used to be STACK
6678 and is now STACK_OF (for example cert in a PKCS7_SIGNED structure) with
6679 sk_num or sk_value it would produce an error because the num, data members
6680 are not present in STACK_OF. Now it just produces a warning. sk_set
6681 replaces the old method of assigning a value to sk_value
6682 (e.g. sk_value(x, i) = y) which the library used in a few cases. Any code
6683 that does this will no longer work (and should use sk_set instead) but
6684 this could be regarded as a "questionable" behaviour anyway.
6685 [Steve Henson]
6686
6687 *) Fix most of the other PKCS#7 bugs. The "experimental" code can now
6688 correctly handle encrypted S/MIME data.
6689 [Steve Henson]
6690
6691 *) Change type of various DES function arguments from des_cblock
6692 (which means, in function argument declarations, pointer to char)
6693 to des_cblock * (meaning pointer to array with 8 char elements),
6694 which allows the compiler to do more typechecking; it was like
6695 that back in SSLeay, but with lots of ugly casts.
6696
6697 Introduce new type const_des_cblock.
6698 [Bodo Moeller]
6699
6700 *) Reorganise the PKCS#7 library and get rid of some of the more obvious
6701 problems: find RecipientInfo structure that matches recipient certificate
6702 and initialise the ASN1 structures properly based on passed cipher.
6703 [Steve Henson]
6704
6705 *) Belatedly make the BN tests actually check the results.
6706 [Ben Laurie]
6707
6708 *) Fix the encoding and decoding of negative ASN1 INTEGERS and conversion
6709 to and from BNs: it was completely broken. New compilation option
6710 NEG_PUBKEY_BUG to allow for some broken certificates that encode public
6711 key elements as negative integers.
6712 [Steve Henson]
6713
6714 *) Reorganize and speed up MD5.
6715 [Andy Polyakov <appro@fy.chalmers.se>]
6716
6717 *) VMS support.
6718 [Richard Levitte <richard@levitte.org>]
6719
6720 *) New option -out to asn1parse to allow the parsed structure to be
6721 output to a file. This is most useful when combined with the -strparse
6722 option to examine the output of things like OCTET STRINGS.
6723 [Steve Henson]
6724
6725 *) Make SSL library a little more fool-proof by not requiring any longer
6726 that SSL_set_{accept,connect}_state be called before
6727 SSL_{accept,connect} may be used (SSL_set_..._state is omitted
6728 in many applications because usually everything *appeared* to work as
6729 intended anyway -- now it really works as intended).
6730 [Bodo Moeller]
6731
6732 *) Move openssl.cnf out of lib/.
6733 [Ulf Möller]
6734
6735 *) Fix various things to let OpenSSL even pass ``egcc -pipe -O2 -Wall
6736 -Wshadow -Wpointer-arith -Wcast-align -Wmissing-prototypes
6737 -Wmissing-declarations -Wnested-externs -Winline'' with EGCS 1.1.2+
6738 [Ralf S. Engelschall]
6739
6740 *) Various fixes to the EVP and PKCS#7 code. It may now be able to
6741 handle PKCS#7 enveloped data properly.
6742 [Sebastian Akerman <sak@parallelconsulting.com>, modified by Steve]
6743
6744 *) Create a duplicate of the SSL_CTX's CERT in SSL_new instead of
6745 copying pointers. The cert_st handling is changed by this in
6746 various ways (and thus what used to be known as ctx->default_cert
6747 is now called ctx->cert, since we don't resort to s->ctx->[default_]cert
6748 any longer when s->cert does not give us what we need).
6749 ssl_cert_instantiate becomes obsolete by this change.
6750 As soon as we've got the new code right (possibly it already is?),
6751 we have solved a couple of bugs of the earlier code where s->cert
6752 was used as if it could not have been shared with other SSL structures.
6753
6754 Note that using the SSL API in certain dirty ways now will result
6755 in different behaviour than observed with earlier library versions:
6756 Changing settings for an SSL_CTX *ctx after having done s = SSL_new(ctx)
6757 does not influence s as it used to.
6758
6759 In order to clean up things more thoroughly, inside SSL_SESSION
6760 we don't use CERT any longer, but a new structure SESS_CERT
6761 that holds per-session data (if available); currently, this is
6762 the peer's certificate chain and, for clients, the server's certificate
6763 and temporary key. CERT holds only those values that can have
6764 meaningful defaults in an SSL_CTX.
6765 [Bodo Moeller]
6766
6767 *) New function X509V3_EXT_i2d() to create an X509_EXTENSION structure
6768 from the internal representation. Various PKCS#7 fixes: remove some
6769 evil casts and set the enc_dig_alg field properly based on the signing
6770 key type.
6771 [Steve Henson]
6772
6773 *) Allow PKCS#12 password to be set from the command line or the
6774 environment. Let 'ca' get its config file name from the environment
6775 variables "OPENSSL_CONF" or "SSLEAY_CONF" (for consistency with 'req'
6776 and 'x509').
6777 [Steve Henson]
6778
6779 *) Allow certificate policies extension to use an IA5STRING for the
6780 organization field. This is contrary to the PKIX definition but
6781 VeriSign uses it and IE5 only recognises this form. Document 'x509'
6782 extension option.
6783 [Steve Henson]
6784
6785 *) Add PEDANTIC compiler flag to allow compilation with gcc -pedantic,
6786 without disallowing inline assembler and the like for non-pedantic builds.
6787 [Ben Laurie]
6788
6789 *) Support Borland C++ builder.
6790 [Janez Jere <jj@void.si>, modified by Ulf Möller]
6791
6792 *) Support Mingw32.
6793 [Ulf Möller]
6794
6795 *) SHA-1 cleanups and performance enhancements.
6796 [Andy Polyakov <appro@fy.chalmers.se>]
6797
6798 *) Sparc v8plus assembler for the bignum library.
6799 [Andy Polyakov <appro@fy.chalmers.se>]
6800
6801 *) Accept any -xxx and +xxx compiler options in Configure.
6802 [Ulf Möller]
6803
6804 *) Update HPUX configuration.
6805 [Anonymous]
6806
6807 *) Add missing sk_<type>_unshift() function to safestack.h
6808 [Ralf S. Engelschall]
6809
6810 *) New function SSL_CTX_use_certificate_chain_file that sets the
6811 "extra_cert"s in addition to the certificate. (This makes sense
6812 only for "PEM" format files, as chains as a whole are not
6813 DER-encoded.)
6814 [Bodo Moeller]
6815
6816 *) Support verify_depth from the SSL API.
6817 x509_vfy.c had what can be considered an off-by-one-error:
6818 Its depth (which was not part of the external interface)
6819 was actually counting the number of certificates in a chain;
6820 now it really counts the depth.
6821 [Bodo Moeller]
6822
6823 *) Bugfix in crypto/x509/x509_cmp.c: The SSLerr macro was used
6824 instead of X509err, which often resulted in confusing error
6825 messages since the error codes are not globally unique
6826 (e.g. an alleged error in ssl3_accept when a certificate
6827 didn't match the private key).
6828
6829 *) New function SSL_CTX_set_session_id_context that allows to set a default
6830 value (so that you don't need SSL_set_session_id_context for each
6831 connection using the SSL_CTX).
6832 [Bodo Moeller]
6833
6834 *) OAEP decoding bug fix.
6835 [Ulf Möller]
6836
6837 *) Support INSTALL_PREFIX for package builders, as proposed by
6838 David Harris.
6839 [Bodo Moeller]
6840
6841 *) New Configure options "threads" and "no-threads". For systems
6842 where the proper compiler options are known (currently Solaris
6843 and Linux), "threads" is the default.
6844 [Bodo Moeller]
6845
6846 *) New script util/mklink.pl as a faster substitute for util/mklink.sh.
6847 [Bodo Moeller]
6848
6849 *) Install various scripts to $(OPENSSLDIR)/misc, not to
6850 $(INSTALLTOP)/bin -- they shouldn't clutter directories
6851 such as /usr/local/bin.
6852 [Bodo Moeller]
6853
6854 *) "make linux-shared" to build shared libraries.
6855 [Niels Poppe <niels@netbox.org>]
6856
6857 *) New Configure option no-<cipher> (rsa, idea, rc5, ...).
6858 [Ulf Möller]
6859
6860 *) Add the PKCS#12 API documentation to openssl.txt. Preliminary support for
6861 extension adding in x509 utility.
6862 [Steve Henson]
6863
6864 *) Remove NOPROTO sections and error code comments.
6865 [Ulf Möller]
6866
6867 *) Partial rewrite of the DEF file generator to now parse the ANSI
6868 prototypes.
6869 [Steve Henson]
6870
6871 *) New Configure options --prefix=DIR and --openssldir=DIR.
6872 [Ulf Möller]
6873
6874 *) Complete rewrite of the error code script(s). It is all now handled
6875 by one script at the top level which handles error code gathering,
6876 header rewriting and C source file generation. It should be much better
6877 than the old method: it now uses a modified version of Ulf's parser to
6878 read the ANSI prototypes in all header files (thus the old K&R definitions
6879 aren't needed for error creation any more) and do a better job of
6880 translating function codes into names. The old 'ASN1 error code imbedded
6881 in a comment' is no longer necessary and it doesn't use .err files which
6882 have now been deleted. Also the error code call doesn't have to appear all
6883 on one line (which resulted in some large lines...).
6884 [Steve Henson]
6885
6886 *) Change #include filenames from <foo.h> to <openssl/foo.h>.
6887 [Bodo Moeller]
6888
6889 *) Change behaviour of ssl2_read when facing length-0 packets: Don't return
6890 0 (which usually indicates a closed connection), but continue reading.
6891 [Bodo Moeller]
6892
6893 *) Fix some race conditions.
6894 [Bodo Moeller]
6895
6896 *) Add support for CRL distribution points extension. Add Certificate
6897 Policies and CRL distribution points documentation.
6898 [Steve Henson]
6899
6900 *) Move the autogenerated header file parts to crypto/opensslconf.h.
6901 [Ulf Möller]
6902
6903 *) Fix new 56-bit DES export ciphersuites: they were using 7 bytes instead of
6904 8 of keying material. Merlin has also confirmed interop with this fix
6905 between OpenSSL and Baltimore C/SSL 2.0 and J/SSL 2.0.
6906 [Merlin Hughes <merlin@baltimore.ie>]
6907
6908 *) Fix lots of warnings.
6909 [Richard Levitte <levitte@stacken.kth.se>]
6910
6911 *) In add_cert_dir() in crypto/x509/by_dir.c, break out of the loop if
6912 the directory spec didn't end with a LIST_SEPARATOR_CHAR.
6913 [Richard Levitte <levitte@stacken.kth.se>]
6914
6915 *) Fix problems with sizeof(long) == 8.
6916 [Andy Polyakov <appro@fy.chalmers.se>]
6917
6918 *) Change functions to ANSI C.
6919 [Ulf Möller]
6920
6921 *) Fix typos in error codes.
6922 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>, Ulf Möller]
6923
6924 *) Remove defunct assembler files from Configure.
6925 [Ulf Möller]
6926
6927 *) SPARC v8 assembler BIGNUM implementation.
6928 [Andy Polyakov <appro@fy.chalmers.se>]
6929
6930 *) Support for Certificate Policies extension: both print and set.
6931 Various additions to support the r2i method this uses.
6932 [Steve Henson]
6933
6934 *) A lot of constification, and fix a bug in X509_NAME_oneline() that could
6935 return a const string when you are expecting an allocated buffer.
6936 [Ben Laurie]
6937
6938 *) Add support for ASN1 types UTF8String and VISIBLESTRING, also the CHOICE
6939 types DirectoryString and DisplayText.
6940 [Steve Henson]
6941
6942 *) Add code to allow r2i extensions to access the configuration database,
6943 add an LHASH database driver and add several ctx helper functions.
6944 [Steve Henson]
6945
6946 *) Fix an evil bug in bn_expand2() which caused various BN functions to
6947 fail when they extended the size of a BIGNUM.
6948 [Steve Henson]
6949
6950 *) Various utility functions to handle SXNet extension. Modify mkdef.pl to
6951 support typesafe stack.
6952 [Steve Henson]
6953
6954 *) Fix typo in SSL_[gs]et_options().
6955 [Nils Frostberg <nils@medcom.se>]
6956
6957 *) Delete various functions and files that belonged to the (now obsolete)
6958 old X509V3 handling code.
6959 [Steve Henson]
6960
6961 *) New Configure option "rsaref".
6962 [Ulf Möller]
6963
6964 *) Don't auto-generate pem.h.
6965 [Bodo Moeller]
6966
6967 *) Introduce type-safe ASN.1 SETs.
6968 [Ben Laurie]
6969
6970 *) Convert various additional casted stacks to type-safe STACK_OF() variants.
6971 [Ben Laurie, Ralf S. Engelschall, Steve Henson]
6972
6973 *) Introduce type-safe STACKs. This will almost certainly break lots of code
6974 that links with OpenSSL (well at least cause lots of warnings), but fear
6975 not: the conversion is trivial, and it eliminates loads of evil casts. A
6976 few STACKed things have been converted already. Feel free to convert more.
6977 In the fullness of time, I'll do away with the STACK type altogether.
6978 [Ben Laurie]
6979
6980 *) Add `openssl ca -revoke <certfile>' facility which revokes a certificate
6981 specified in <certfile> by updating the entry in the index.txt file.
6982 This way one no longer has to edit the index.txt file manually for
6983 revoking a certificate. The -revoke option does the gory details now.
6984 [Massimiliano Pala <madwolf@openca.org>, Ralf S. Engelschall]
6985
6986 *) Fix `openssl crl -noout -text' combination where `-noout' killed the
6987 `-text' option at all and this way the `-noout -text' combination was
6988 inconsistent in `openssl crl' with the friends in `openssl x509|rsa|dsa'.
6989 [Ralf S. Engelschall]
6990
6991 *) Make sure a corresponding plain text error message exists for the
6992 X509_V_ERR_CERT_REVOKED/23 error number which can occur when a
6993 verify callback function determined that a certificate was revoked.
6994 [Ralf S. Engelschall]
6995
6996 *) Bugfix: In test/testenc, don't test "openssl <cipher>" for
6997 ciphers that were excluded, e.g. by -DNO_IDEA. Also, test
6998 all available cipers including rc5, which was forgotten until now.
6999 In order to let the testing shell script know which algorithms
7000 are available, a new (up to now undocumented) command
7001 "openssl list-cipher-commands" is used.
7002 [Bodo Moeller]
7003
7004 *) Bugfix: s_client occasionally would sleep in select() when
7005 it should have checked SSL_pending() first.
7006 [Bodo Moeller]
7007
7008 *) New functions DSA_do_sign and DSA_do_verify to provide access to
7009 the raw DSA values prior to ASN.1 encoding.
7010 [Ulf Möller]
7011
7012 *) Tweaks to Configure
7013 [Niels Poppe <niels@netbox.org>]
7014
7015 *) Add support for PKCS#5 v2.0 ASN1 PBES2 structures. No other support,
7016 yet...
7017 [Steve Henson]
7018
7019 *) New variables $(RANLIB) and $(PERL) in the Makefiles.
7020 [Ulf Möller]
7021
7022 *) New config option to avoid instructions that are illegal on the 80386.
7023 The default code is faster, but requires at least a 486.
7024 [Ulf Möller]
7025
7026 *) Got rid of old SSL2_CLIENT_VERSION (inconsistently used) and
7027 SSL2_SERVER_VERSION (not used at all) macros, which are now the
7028 same as SSL2_VERSION anyway.
7029 [Bodo Moeller]
7030
7031 *) New "-showcerts" option for s_client.
7032 [Bodo Moeller]
7033
7034 *) Still more PKCS#12 integration. Add pkcs12 application to openssl
7035 application. Various cleanups and fixes.
7036 [Steve Henson]
7037
7038 *) More PKCS#12 integration. Add new pkcs12 directory with Makefile.ssl and
7039 modify error routines to work internally. Add error codes and PBE init
7040 to library startup routines.
7041 [Steve Henson]
7042
7043 *) Further PKCS#12 integration. Added password based encryption, PKCS#8 and
7044 packing functions to asn1 and evp. Changed function names and error
7045 codes along the way.
7046 [Steve Henson]
7047
7048 *) PKCS12 integration: and so it begins... First of several patches to
7049 slowly integrate PKCS#12 functionality into OpenSSL. Add PKCS#12
7050 objects to objects.h
7051 [Steve Henson]
7052
7053 *) Add a new 'indent' option to some X509V3 extension code. Initial ASN1
7054 and display support for Thawte strong extranet extension.
7055 [Steve Henson]
7056
7057 *) Add LinuxPPC support.
7058 [Jeff Dubrule <igor@pobox.org>]
7059
7060 *) Get rid of redundant BN file bn_mulw.c, and rename bn_div64 to
7061 bn_div_words in alpha.s.
7062 [Hannes Reinecke <H.Reinecke@hw.ac.uk> and Ben Laurie]
7063
7064 *) Make sure the RSA OAEP test is skipped under -DRSAref because
7065 OAEP isn't supported when OpenSSL is built with RSAref.
7066 [Ulf Moeller <ulf@fitug.de>]
7067
7068 *) Move definitions of IS_SET/IS_SEQUENCE inside crypto/asn1/asn1.h
7069 so they no longer are missing under -DNOPROTO.
7070 [Soren S. Jorvang <soren@t.dk>]
7071
7072
7073 Changes between 0.9.1c and 0.9.2b [22 Mar 1999]
7074
7075 *) Make SSL_get_peer_cert_chain() work in servers. Unfortunately, it still
7076 doesn't work when the session is reused. Coming soon!
7077 [Ben Laurie]
7078
7079 *) Fix a security hole, that allows sessions to be reused in the wrong
7080 context thus bypassing client cert protection! All software that uses
7081 client certs and session caches in multiple contexts NEEDS PATCHING to
7082 allow session reuse! A fuller solution is in the works.
7083 [Ben Laurie, problem pointed out by Holger Reif, Bodo Moeller (and ???)]
7084
7085 *) Some more source tree cleanups (removed obsolete files
7086 crypto/bf/asm/bf586.pl, test/test.txt and crypto/sha/asm/f.s; changed
7087 permission on "config" script to be executable) and a fix for the INSTALL
7088 document.
7089 [Ulf Moeller <ulf@fitug.de>]
7090
7091 *) Remove some legacy and erroneous uses of malloc, free instead of
7092 Malloc, Free.
7093 [Lennart Bang <lob@netstream.se>, with minor changes by Steve]
7094
7095 *) Make rsa_oaep_test return non-zero on error.
7096 [Ulf Moeller <ulf@fitug.de>]
7097
7098 *) Add support for native Solaris shared libraries. Configure
7099 solaris-sparc-sc4-pic, make, then run shlib/solaris-sc4.sh. It'd be nice
7100 if someone would make that last step automatic.
7101 [Matthias Loepfe <Matthias.Loepfe@AdNovum.CH>]
7102
7103 *) ctx_size was not built with the right compiler during "make links". Fixed.
7104 [Ben Laurie]
7105
7106 *) Change the meaning of 'ALL' in the cipher list. It now means "everything
7107 except NULL ciphers". This means the default cipher list will no longer
7108 enable NULL ciphers. They need to be specifically enabled e.g. with
7109 the string "DEFAULT:eNULL".
7110 [Steve Henson]
7111
7112 *) Fix to RSA private encryption routines: if p < q then it would
7113 occasionally produce an invalid result. This will only happen with
7114 externally generated keys because OpenSSL (and SSLeay) ensure p > q.
7115 [Steve Henson]
7116
7117 *) Be less restrictive and allow also `perl util/perlpath.pl
7118 /path/to/bin/perl' in addition to `perl util/perlpath.pl /path/to/bin',
7119 because this way one can also use an interpreter named `perl5' (which is
7120 usually the name of Perl 5.xxx on platforms where an Perl 4.x is still
7121 installed as `perl').
7122 [Matthias Loepfe <Matthias.Loepfe@adnovum.ch>]
7123
7124 *) Let util/clean-depend.pl work also with older Perl 5.00x versions.
7125 [Matthias Loepfe <Matthias.Loepfe@adnovum.ch>]
7126
7127 *) Fix Makefile.org so CC,CFLAG etc are passed to 'make links' add
7128 advapi32.lib to Win32 build and change the pem test comparision
7129 to fc.exe (thanks to Ulrich Kroener <kroneru@yahoo.com> for the
7130 suggestion). Fix misplaced ASNI prototypes and declarations in evp.h
7131 and crypto/des/ede_cbcm_enc.c.
7132 [Steve Henson]
7133
7134 *) DES quad checksum was broken on big-endian architectures. Fixed.
7135 [Ben Laurie]
7136
7137 *) Comment out two functions in bio.h that aren't implemented. Fix up the
7138 Win32 test batch file so it (might) work again. The Win32 test batch file
7139 is horrible: I feel ill....
7140 [Steve Henson]
7141
7142 *) Move various #ifdefs around so NO_SYSLOG, NO_DIRENT etc are now selected
7143 in e_os.h. Audit of header files to check ANSI and non ANSI
7144 sections: 10 functions were absent from non ANSI section and not exported
7145 from Windows DLLs. Fixed up libeay.num for new functions.
7146 [Steve Henson]
7147
7148 *) Make `openssl version' output lines consistent.
7149 [Ralf S. Engelschall]
7150
7151 *) Fix Win32 symbol export lists for BIO functions: Added
7152 BIO_get_ex_new_index, BIO_get_ex_num, BIO_get_ex_data and BIO_set_ex_data
7153 to ms/libeay{16,32}.def.
7154 [Ralf S. Engelschall]
7155
7156 *) Second round of fixing the OpenSSL perl/ stuff. It now at least compiled
7157 fine under Unix and passes some trivial tests I've now added. But the
7158 whole stuff is horribly incomplete, so a README.1ST with a disclaimer was
7159 added to make sure no one expects that this stuff really works in the
7160 OpenSSL 0.9.2 release. Additionally I've started to clean the XS sources
7161 up and fixed a few little bugs and inconsistencies in OpenSSL.{pm,xs} and
7162 openssl_bio.xs.
7163 [Ralf S. Engelschall]
7164
7165 *) Fix the generation of two part addresses in perl.
7166 [Kenji Miyake <kenji@miyake.org>, integrated by Ben Laurie]
7167
7168 *) Add config entry for Linux on MIPS.
7169 [John Tobey <jtobey@channel1.com>]
7170
7171 *) Make links whenever Configure is run, unless we are on Windoze.
7172 [Ben Laurie]
7173
7174 *) Permit extensions to be added to CRLs using crl_section in openssl.cnf.
7175 Currently only issuerAltName and AuthorityKeyIdentifier make any sense
7176 in CRLs.
7177 [Steve Henson]
7178
7179 *) Add a useful kludge to allow package maintainers to specify compiler and
7180 other platforms details on the command line without having to patch the
7181 Configure script everytime: One now can use ``perl Configure
7182 <id>:<details>'', i.e. platform ids are allowed to have details appended
7183 to them (seperated by colons). This is treated as there would be a static
7184 pre-configured entry in Configure's %table under key <id> with value
7185 <details> and ``perl Configure <id>'' is called. So, when you want to
7186 perform a quick test-compile under FreeBSD 3.1 with pgcc and without
7187 assembler stuff you can use ``perl Configure "FreeBSD-elf:pgcc:-O6:::"''
7188 now, which overrides the FreeBSD-elf entry on-the-fly.
7189 [Ralf S. Engelschall]
7190
7191 *) Disable new TLS1 ciphersuites by default: they aren't official yet.
7192 [Ben Laurie]
7193
7194 *) Allow DSO flags like -fpic, -fPIC, -KPIC etc. to be specified
7195 on the `perl Configure ...' command line. This way one can compile
7196 OpenSSL libraries with Position Independent Code (PIC) which is needed
7197 for linking it into DSOs.
7198 [Ralf S. Engelschall]
7199
7200 *) Remarkably, export ciphers were totally broken and no-one had noticed!
7201 Fixed.
7202 [Ben Laurie]
7203
7204 *) Cleaned up the LICENSE document: The official contact for any license
7205 questions now is the OpenSSL core team under openssl-core@openssl.org.
7206 And add a paragraph about the dual-license situation to make sure people
7207 recognize that _BOTH_ the OpenSSL license _AND_ the SSLeay license apply
7208 to the OpenSSL toolkit.
7209 [Ralf S. Engelschall]
7210
7211 *) General source tree makefile cleanups: Made `making xxx in yyy...'
7212 display consistent in the source tree and replaced `/bin/rm' by `rm'.
7213 Additonally cleaned up the `make links' target: Remove unnecessary
7214 semicolons, subsequent redundant removes, inline point.sh into mklink.sh
7215 to speed processing and no longer clutter the display with confusing
7216 stuff. Instead only the actually done links are displayed.
7217 [Ralf S. Engelschall]
7218
7219 *) Permit null encryption ciphersuites, used for authentication only. It used
7220 to be necessary to set the preprocessor define SSL_ALLOW_ENULL to do this.
7221 It is now necessary to set SSL_FORBID_ENULL to prevent the use of null
7222 encryption.
7223 [Ben Laurie]
7224
7225 *) Add a bunch of fixes to the PKCS#7 stuff. It used to sometimes reorder
7226 signed attributes when verifying signatures (this would break them),
7227 the detached data encoding was wrong and public keys obtained using
7228 X509_get_pubkey() weren't freed.
7229 [Steve Henson]
7230
7231 *) Add text documentation for the BUFFER functions. Also added a work around
7232 to a Win95 console bug. This was triggered by the password read stuff: the
7233 last character typed gets carried over to the next fread(). If you were
7234 generating a new cert request using 'req' for example then the last
7235 character of the passphrase would be CR which would then enter the first
7236 field as blank.
7237 [Steve Henson]
7238
7239 *) Added the new `Includes OpenSSL Cryptography Software' button as
7240 doc/openssl_button.{gif,html} which is similar in style to the old SSLeay
7241 button and can be used by applications based on OpenSSL to show the
7242 relationship to the OpenSSL project.
7243 [Ralf S. Engelschall]
7244
7245 *) Remove confusing variables in function signatures in files
7246 ssl/ssl_lib.c and ssl/ssl.h.
7247 [Lennart Bong <lob@kulthea.stacken.kth.se>]
7248
7249 *) Don't install bss_file.c under PREFIX/include/
7250 [Lennart Bong <lob@kulthea.stacken.kth.se>]
7251
7252 *) Get the Win32 compile working again. Modify mkdef.pl so it can handle
7253 functions that return function pointers and has support for NT specific
7254 stuff. Fix mk1mf.pl and VC-32.pl to support NT differences also. Various
7255 #ifdef WIN32 and WINNTs sprinkled about the place and some changes from
7256 unsigned to signed types: this was killing the Win32 compile.
7257 [Steve Henson]
7258
7259 *) Add new certificate file to stack functions,
7260 SSL_add_dir_cert_subjects_to_stack() and
7261 SSL_add_file_cert_subjects_to_stack(). These largely supplant
7262 SSL_load_client_CA_file(), and can be used to add multiple certs easily
7263 to a stack (usually this is then handed to SSL_CTX_set_client_CA_list()).
7264 This means that Apache-SSL and similar packages don't have to mess around
7265 to add as many CAs as they want to the preferred list.
7266 [Ben Laurie]
7267
7268 *) Experiment with doxygen documentation. Currently only partially applied to
7269 ssl/ssl_lib.c.
7270 See http://www.stack.nl/~dimitri/doxygen/index.html, and run doxygen with
7271 openssl.doxy as the configuration file.
7272 [Ben Laurie]
7273
7274 *) Get rid of remaining C++-style comments which strict C compilers hate.
7275 [Ralf S. Engelschall, pointed out by Carlos Amengual]
7276
7277 *) Changed BN_RECURSION in bn_mont.c to BN_RECURSION_MONT so it is not
7278 compiled in by default: it has problems with large keys.
7279 [Steve Henson]
7280
7281 *) Add a bunch of SSL_xxx() functions for configuring the temporary RSA and
7282 DH private keys and/or callback functions which directly correspond to
7283 their SSL_CTX_xxx() counterparts but work on a per-connection basis. This
7284 is needed for applications which have to configure certificates on a
7285 per-connection basis (e.g. Apache+mod_ssl) instead of a per-context basis
7286 (e.g. s_server).
7287 For the RSA certificate situation is makes no difference, but
7288 for the DSA certificate situation this fixes the "no shared cipher"
7289 problem where the OpenSSL cipher selection procedure failed because the
7290 temporary keys were not overtaken from the context and the API provided
7291 no way to reconfigure them.
7292 The new functions now let applications reconfigure the stuff and they
7293 are in detail: SSL_need_tmp_RSA, SSL_set_tmp_rsa, SSL_set_tmp_dh,
7294 SSL_set_tmp_rsa_callback and SSL_set_tmp_dh_callback. Additionally a new
7295 non-public-API function ssl_cert_instantiate() is used as a helper
7296 function and also to reduce code redundancy inside ssl_rsa.c.
7297 [Ralf S. Engelschall]
7298
7299 *) Move s_server -dcert and -dkey options out of the undocumented feature
7300 area because they are useful for the DSA situation and should be
7301 recognized by the users.
7302 [Ralf S. Engelschall]
7303
7304 *) Fix the cipher decision scheme for export ciphers: the export bits are
7305 *not* within SSL_MKEY_MASK or SSL_AUTH_MASK, they are within
7306 SSL_EXP_MASK. So, the original variable has to be used instead of the
7307 already masked variable.
7308 [Richard Levitte <levitte@stacken.kth.se>]
7309
7310 *) Fix 'port' variable from `int' to `unsigned int' in crypto/bio/b_sock.c
7311 [Richard Levitte <levitte@stacken.kth.se>]
7312
7313 *) Change type of another md_len variable in pk7_doit.c:PKCS7_dataFinal()
7314 from `int' to `unsigned int' because it's a length and initialized by
7315 EVP_DigestFinal() which expects an `unsigned int *'.
7316 [Richard Levitte <levitte@stacken.kth.se>]
7317
7318 *) Don't hard-code path to Perl interpreter on shebang line of Configure
7319 script. Instead use the usual Shell->Perl transition trick.
7320 [Ralf S. Engelschall]
7321
7322 *) Make `openssl x509 -noout -modulus' functional also for DSA certificates
7323 (in addition to RSA certificates) to match the behaviour of `openssl dsa
7324 -noout -modulus' as it's already the case for `openssl rsa -noout
7325 -modulus'. For RSA the -modulus is the real "modulus" while for DSA
7326 currently the public key is printed (a decision which was already done by
7327 `openssl dsa -modulus' in the past) which serves a similar purpose.
7328 Additionally the NO_RSA no longer completely removes the whole -modulus
7329 option; it now only avoids using the RSA stuff. Same applies to NO_DSA
7330 now, too.
7331 [Ralf S. Engelschall]
7332
7333 *) Add Arne Ansper's reliable BIO - this is an encrypted, block-digested
7334 BIO. See the source (crypto/evp/bio_ok.c) for more info.
7335 [Arne Ansper <arne@ats.cyber.ee>]
7336
7337 *) Dump the old yucky req code that tried (and failed) to allow raw OIDs
7338 to be added. Now both 'req' and 'ca' can use new objects defined in the
7339 config file.
7340 [Steve Henson]
7341
7342 *) Add cool BIO that does syslog (or event log on NT).
7343 [Arne Ansper <arne@ats.cyber.ee>, integrated by Ben Laurie]
7344
7345 *) Add support for new TLS ciphersuites, TLS_RSA_EXPORT56_WITH_RC4_56_MD5,
7346 TLS_RSA_EXPORT56_WITH_RC2_CBC_56_MD5 and
7347 TLS_RSA_EXPORT56_WITH_DES_CBC_SHA, as specified in "56-bit Export Cipher
7348 Suites For TLS", draft-ietf-tls-56-bit-ciphersuites-00.txt.
7349 [Ben Laurie]
7350
7351 *) Add preliminary config info for new extension code.
7352 [Steve Henson]
7353
7354 *) Make RSA_NO_PADDING really use no padding.
7355 [Ulf Moeller <ulf@fitug.de>]
7356
7357 *) Generate errors when private/public key check is done.
7358 [Ben Laurie]
7359
7360 *) Overhaul for 'crl' utility. New function X509_CRL_print. Partial support
7361 for some CRL extensions and new objects added.
7362 [Steve Henson]
7363
7364 *) Really fix the ASN1 IMPLICIT bug this time... Partial support for private
7365 key usage extension and fuller support for authority key id.
7366 [Steve Henson]
7367
7368 *) Add OAEP encryption for the OpenSSL crypto library. OAEP is the improved
7369 padding method for RSA, which is recommended for new applications in PKCS
7370 #1 v2.0 (RFC 2437, October 1998).
7371 OAEP (Optimal Asymmetric Encryption Padding) has better theoretical
7372 foundations than the ad-hoc padding used in PKCS #1 v1.5. It is secure
7373 against Bleichbacher's attack on RSA.
7374 [Ulf Moeller <ulf@fitug.de>, reformatted, corrected and integrated by
7375 Ben Laurie]
7376
7377 *) Updates to the new SSL compression code
7378 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)]
7379
7380 *) Fix so that the version number in the master secret, when passed
7381 via RSA, checks that if TLS was proposed, but we roll back to SSLv3
7382 (because the server will not accept higher), that the version number
7383 is 0x03,0x01, not 0x03,0x00
7384 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)]
7385
7386 *) Run extensive memory leak checks on SSL apps. Fixed *lots* of memory
7387 leaks in ssl/ relating to new X509_get_pubkey() behaviour. Also fixes
7388 in apps/ and an unrelated leak in crypto/dsa/dsa_vrf.c
7389 [Steve Henson]
7390
7391 *) Support for RAW extensions where an arbitrary extension can be
7392 created by including its DER encoding. See apps/openssl.cnf for
7393 an example.
7394 [Steve Henson]
7395
7396 *) Make sure latest Perl versions don't interpret some generated C array
7397 code as Perl array code in the crypto/err/err_genc.pl script.
7398 [Lars Weber <3weber@informatik.uni-hamburg.de>]
7399
7400 *) Modify ms/do_ms.bat to not generate assembly language makefiles since
7401 not many people have the assembler. Various Win32 compilation fixes and
7402 update to the INSTALL.W32 file with (hopefully) more accurate Win32
7403 build instructions.
7404 [Steve Henson]
7405
7406 *) Modify configure script 'Configure' to automatically create crypto/date.h
7407 file under Win32 and also build pem.h from pem.org. New script
7408 util/mkfiles.pl to create the MINFO file on environments that can't do a
7409 'make files': perl util/mkfiles.pl >MINFO should work.
7410 [Steve Henson]
7411
7412 *) Major rework of DES function declarations, in the pursuit of correctness
7413 and purity. As a result, many evil casts evaporated, and some weirdness,
7414 too. You may find this causes warnings in your code. Zapping your evil
7415 casts will probably fix them. Mostly.
7416 [Ben Laurie]
7417
7418 *) Fix for a typo in asn1.h. Bug fix to object creation script
7419 obj_dat.pl. It considered a zero in an object definition to mean
7420 "end of object": none of the objects in objects.h have any zeros
7421 so it wasn't spotted.
7422 [Steve Henson, reported by Erwann ABALEA <eabalea@certplus.com>]
7423
7424 *) Add support for Triple DES Cipher Block Chaining with Output Feedback
7425 Masking (CBCM). In the absence of test vectors, the best I have been able
7426 to do is check that the decrypt undoes the encrypt, so far. Send me test
7427 vectors if you have them.
7428 [Ben Laurie]
7429
7430 *) Correct calculation of key length for export ciphers (too much space was
7431 allocated for null ciphers). This has not been tested!
7432 [Ben Laurie]
7433
7434 *) Modifications to the mkdef.pl for Win32 DEF file creation. The usage
7435 message is now correct (it understands "crypto" and "ssl" on its
7436 command line). There is also now an "update" option. This will update
7437 the util/ssleay.num and util/libeay.num files with any new functions.
7438 If you do a:
7439 perl util/mkdef.pl crypto ssl update
7440 it will update them.
7441 [Steve Henson]
7442
7443 *) Overhauled the Perl interface (perl/*):
7444 - ported BN stuff to OpenSSL's different BN library
7445 - made the perl/ source tree CVS-aware
7446 - renamed the package from SSLeay to OpenSSL (the files still contain
7447 their history because I've copied them in the repository)
7448 - removed obsolete files (the test scripts will be replaced
7449 by better Test::Harness variants in the future)
7450 [Ralf S. Engelschall]
7451
7452 *) First cut for a very conservative source tree cleanup:
7453 1. merge various obsolete readme texts into doc/ssleay.txt
7454 where we collect the old documents and readme texts.
7455 2. remove the first part of files where I'm already sure that we no
7456 longer need them because of three reasons: either they are just temporary
7457 files which were left by Eric or they are preserved original files where
7458 I've verified that the diff is also available in the CVS via "cvs diff
7459 -rSSLeay_0_8_1b" or they were renamed (as it was definitely the case for
7460 the crypto/md/ stuff).
7461 [Ralf S. Engelschall]
7462
7463 *) More extension code. Incomplete support for subject and issuer alt
7464 name, issuer and authority key id. Change the i2v function parameters
7465 and add an extra 'crl' parameter in the X509V3_CTX structure: guess
7466 what that's for :-) Fix to ASN1 macro which messed up
7467 IMPLICIT tag and add f_enum.c which adds a2i, i2a for ENUMERATED.
7468 [Steve Henson]
7469
7470 *) Preliminary support for ENUMERATED type. This is largely copied from the
7471 INTEGER code.
7472 [Steve Henson]
7473
7474 *) Add new function, EVP_MD_CTX_copy() to replace frequent use of memcpy.
7475 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)]
7476
7477 *) Make sure `make rehash' target really finds the `openssl' program.
7478 [Ralf S. Engelschall, Matthias Loepfe <Matthias.Loepfe@adnovum.ch>]
7479
7480 *) Squeeze another 7% of speed out of MD5 assembler, at least on a P2. I'd
7481 like to hear about it if this slows down other processors.
7482 [Ben Laurie]
7483
7484 *) Add CygWin32 platform information to Configure script.
7485 [Alan Batie <batie@aahz.jf.intel.com>]
7486
7487 *) Fixed ms/32all.bat script: `no_asm' -> `no-asm'
7488 [Rainer W. Gerling <gerling@mpg-gv.mpg.de>]
7489
7490 *) New program nseq to manipulate netscape certificate sequences
7491 [Steve Henson]
7492
7493 *) Modify crl2pkcs7 so it supports multiple -certfile arguments. Fix a
7494 few typos.
7495 [Steve Henson]
7496
7497 *) Fixes to BN code. Previously the default was to define BN_RECURSION
7498 but the BN code had some problems that would cause failures when
7499 doing certificate verification and some other functions.
7500 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)]
7501
7502 *) Add ASN1 and PEM code to support netscape certificate sequences.
7503 [Steve Henson]
7504
7505 *) Add ASN1 and PEM code to support netscape certificate sequences.
7506 [Steve Henson]
7507
7508 *) Add several PKIX and private extended key usage OIDs.
7509 [Steve Henson]
7510
7511 *) Modify the 'ca' program to handle the new extension code. Modify
7512 openssl.cnf for new extension format, add comments.
7513 [Steve Henson]
7514
7515 *) More X509 V3 changes. Fix typo in v3_bitstr.c. Add support to 'req'
7516 and add a sample to openssl.cnf so req -x509 now adds appropriate
7517 CA extensions.
7518 [Steve Henson]
7519
7520 *) Continued X509 V3 changes. Add to other makefiles, integrate with the
7521 error code, add initial support to X509_print() and x509 application.
7522 [Steve Henson]
7523
7524 *) Takes a deep breath and start addding X509 V3 extension support code. Add
7525 files in crypto/x509v3. Move original stuff to crypto/x509v3/old. All this
7526 stuff is currently isolated and isn't even compiled yet.
7527 [Steve Henson]
7528
7529 *) Continuing patches for GeneralizedTime. Fix up certificate and CRL
7530 ASN1 to use ASN1_TIME and modify print routines to use ASN1_TIME_print.
7531 Removed the versions check from X509 routines when loading extensions:
7532 this allows certain broken certificates that don't set the version
7533 properly to be processed.
7534 [Steve Henson]
7535
7536 *) Deal with irritating shit to do with dependencies, in YAAHW (Yet Another
7537 Ad Hoc Way) - Makefile.ssls now all contain local dependencies, which
7538 can still be regenerated with "make depend".
7539 [Ben Laurie]
7540
7541 *) Spelling mistake in C version of CAST-128.
7542 [Ben Laurie, reported by Jeremy Hylton <jeremy@cnri.reston.va.us>]
7543
7544 *) Changes to the error generation code. The perl script err-code.pl
7545 now reads in the old error codes and retains the old numbers, only
7546 adding new ones if necessary. It also only changes the .err files if new
7547 codes are added. The makefiles have been modified to only insert errors
7548 when needed (to avoid needlessly modifying header files). This is done
7549 by only inserting errors if the .err file is newer than the auto generated
7550 C file. To rebuild all the error codes from scratch (the old behaviour)
7551 either modify crypto/Makefile.ssl to pass the -regen flag to err_code.pl
7552 or delete all the .err files.
7553 [Steve Henson]
7554
7555 *) CAST-128 was incorrectly implemented for short keys. The C version has
7556 been fixed, but is untested. The assembler versions are also fixed, but
7557 new assembler HAS NOT BEEN GENERATED FOR WIN32 - the Makefile needs fixing
7558 to regenerate it if needed.
7559 [Ben Laurie, reported (with fix for C version) by Jun-ichiro itojun
7560 Hagino <itojun@kame.net>]
7561
7562 *) File was opened incorrectly in randfile.c.
7563 [Ulf Möller <ulf@fitug.de>]
7564
7565 *) Beginning of support for GeneralizedTime. d2i, i2d, check and print
7566 functions. Also ASN1_TIME suite which is a CHOICE of UTCTime or
7567 GeneralizedTime. ASN1_TIME is the proper type used in certificates et
7568 al: it's just almost always a UTCTime. Note this patch adds new error
7569 codes so do a "make errors" if there are problems.
7570 [Steve Henson]
7571
7572 *) Correct Linux 1 recognition in config.
7573 [Ulf Möller <ulf@fitug.de>]
7574
7575 *) Remove pointless MD5 hash when using DSA keys in ca.
7576 [Anonymous <nobody@replay.com>]
7577
7578 *) Generate an error if given an empty string as a cert directory. Also
7579 generate an error if handed NULL (previously returned 0 to indicate an
7580 error, but didn't set one).
7581 [Ben Laurie, reported by Anonymous <nobody@replay.com>]
7582
7583 *) Add prototypes to SSL methods. Make SSL_write's buffer const, at last.
7584 [Ben Laurie]
7585
7586 *) Fix the dummy function BN_ref_mod_exp() in rsaref.c to have the correct
7587 parameters. This was causing a warning which killed off the Win32 compile.
7588 [Steve Henson]
7589
7590 *) Remove C++ style comments from crypto/bn/bn_local.h.
7591 [Neil Costigan <neil.costigan@celocom.com>]
7592
7593 *) The function OBJ_txt2nid was broken. It was supposed to return a nid
7594 based on a text string, looking up short and long names and finally
7595 "dot" format. The "dot" format stuff didn't work. Added new function
7596 OBJ_txt2obj to do the same but return an ASN1_OBJECT and rewrote
7597 OBJ_txt2nid to use it. OBJ_txt2obj can also return objects even if the
7598 OID is not part of the table.
7599 [Steve Henson]
7600
7601 *) Add prototypes to X509 lookup/verify methods, fixing a bug in
7602 X509_LOOKUP_by_alias().
7603 [Ben Laurie]
7604
7605 *) Sort openssl functions by name.
7606 [Ben Laurie]
7607
7608 *) Get the gendsa program working (hopefully) and add it to app list. Remove
7609 encryption from sample DSA keys (in case anyone is interested the password
7610 was "1234").
7611 [Steve Henson]
7612
7613 *) Make _all_ *_free functions accept a NULL pointer.
7614 [Frans Heymans <fheymans@isaserver.be>]
7615
7616 *) If a DH key is generated in s3_srvr.c, don't blow it by trying to use
7617 NULL pointers.
7618 [Anonymous <nobody@replay.com>]
7619
7620 *) s_server should send the CAfile as acceptable CAs, not its own cert.
7621 [Bodo Moeller <3moeller@informatik.uni-hamburg.de>]
7622
7623 *) Don't blow it for numeric -newkey arguments to apps/req.
7624 [Bodo Moeller <3moeller@informatik.uni-hamburg.de>]
7625
7626 *) Temp key "for export" tests were wrong in s3_srvr.c.
7627 [Anonymous <nobody@replay.com>]
7628
7629 *) Add prototype for temp key callback functions
7630 SSL_CTX_set_tmp_{rsa,dh}_callback().
7631 [Ben Laurie]
7632
7633 *) Make DH_free() tolerate being passed a NULL pointer (like RSA_free() and
7634 DSA_free()). Make X509_PUBKEY_set() check for errors in d2i_PublicKey().
7635 [Steve Henson]
7636
7637 *) X509_name_add_entry() freed the wrong thing after an error.
7638 [Arne Ansper <arne@ats.cyber.ee>]
7639
7640 *) rsa_eay.c would attempt to free a NULL context.
7641 [Arne Ansper <arne@ats.cyber.ee>]
7642
7643 *) BIO_s_socket() had a broken should_retry() on Windoze.
7644 [Arne Ansper <arne@ats.cyber.ee>]
7645
7646 *) BIO_f_buffer() didn't pass on BIO_CTRL_FLUSH.
7647 [Arne Ansper <arne@ats.cyber.ee>]
7648
7649 *) Make sure the already existing X509_STORE->depth variable is initialized
7650 in X509_STORE_new(), but document the fact that this variable is still
7651 unused in the certificate verification process.
7652 [Ralf S. Engelschall]
7653
7654 *) Fix the various library and apps files to free up pkeys obtained from
7655 X509_PUBKEY_get() et al. Also allow x509.c to handle netscape extensions.
7656 [Steve Henson]
7657
7658 *) Fix reference counting in X509_PUBKEY_get(). This makes
7659 demos/maurice/example2.c work, amongst others, probably.
7660 [Steve Henson and Ben Laurie]
7661
7662 *) First cut of a cleanup for apps/. First the `ssleay' program is now named
7663 `openssl' and second, the shortcut symlinks for the `openssl <command>'
7664 are no longer created. This way we have a single and consistent command
7665 line interface `openssl <command>', similar to `cvs <command>'.
7666 [Ralf S. Engelschall, Paul Sutton and Ben Laurie]
7667
7668 *) ca.c: move test for DSA keys inside #ifndef NO_DSA. Make pubkey
7669 BIT STRING wrapper always have zero unused bits.
7670 [Steve Henson]
7671
7672 *) Add CA.pl, perl version of CA.sh, add extended key usage OID.
7673 [Steve Henson]
7674
7675 *) Make the top-level INSTALL documentation easier to understand.
7676 [Paul Sutton]
7677
7678 *) Makefiles updated to exit if an error occurs in a sub-directory
7679 make (including if user presses ^C) [Paul Sutton]
7680
7681 *) Make Montgomery context stuff explicit in RSA data structure.
7682 [Ben Laurie]
7683
7684 *) Fix build order of pem and err to allow for generated pem.h.
7685 [Ben Laurie]
7686
7687 *) Fix renumbering bug in X509_NAME_delete_entry().
7688 [Ben Laurie]
7689
7690 *) Enhanced the err-ins.pl script so it makes the error library number
7691 global and can add a library name. This is needed for external ASN1 and
7692 other error libraries.
7693 [Steve Henson]
7694
7695 *) Fixed sk_insert which never worked properly.
7696 [Steve Henson]
7697
7698 *) Fix ASN1 macros so they can handle indefinite length construted
7699 EXPLICIT tags. Some non standard certificates use these: they can now
7700 be read in.
7701 [Steve Henson]
7702
7703 *) Merged the various old/obsolete SSLeay documentation files (doc/xxx.doc)
7704 into a single doc/ssleay.txt bundle. This way the information is still
7705 preserved but no longer messes up this directory. Now it's new room for
7706 the new set of documenation files.
7707 [Ralf S. Engelschall]
7708
7709 *) SETs were incorrectly DER encoded. This was a major pain, because they
7710 shared code with SEQUENCEs, which aren't coded the same. This means that
7711 almost everything to do with SETs or SEQUENCEs has either changed name or
7712 number of arguments.
7713 [Ben Laurie, based on a partial fix by GP Jayan <gp@nsj.co.jp>]
7714
7715 *) Fix test data to work with the above.
7716 [Ben Laurie]
7717
7718 *) Fix the RSA header declarations that hid a bug I fixed in 0.9.0b but
7719 was already fixed by Eric for 0.9.1 it seems.
7720 [Ben Laurie - pointed out by Ulf Möller <ulf@fitug.de>]
7721
7722 *) Autodetect FreeBSD3.
7723 [Ben Laurie]
7724
7725 *) Fix various bugs in Configure. This affects the following platforms:
7726 nextstep
7727 ncr-scde
7728 unixware-2.0
7729 unixware-2.0-pentium
7730 sco5-cc.
7731 [Ben Laurie]
7732
7733 *) Eliminate generated files from CVS. Reorder tests to regenerate files
7734 before they are needed.
7735 [Ben Laurie]
7736
7737 *) Generate Makefile.ssl from Makefile.org (to keep CVS happy).
7738 [Ben Laurie]
7739
7740
7741 Changes between 0.9.1b and 0.9.1c [23-Dec-1998]
7742
7743 *) Added OPENSSL_VERSION_NUMBER to crypto/crypto.h and
7744 changed SSLeay to OpenSSL in version strings.
7745 [Ralf S. Engelschall]
7746
7747 *) Some fixups to the top-level documents.
7748 [Paul Sutton]
7749
7750 *) Fixed the nasty bug where rsaref.h was not found under compile-time
7751 because the symlink to include/ was missing.
7752 [Ralf S. Engelschall]
7753
7754 *) Incorporated the popular no-RSA/DSA-only patches
7755 which allow to compile a RSA-free SSLeay.
7756 [Andrew Cooke / Interrader Ldt., Ralf S. Engelschall]
7757
7758 *) Fixed nasty rehash problem under `make -f Makefile.ssl links'
7759 when "ssleay" is still not found.
7760 [Ralf S. Engelschall]
7761
7762 *) Added more platforms to Configure: Cray T3E, HPUX 11,
7763 [Ralf S. Engelschall, Beckmann <beckman@acl.lanl.gov>]
7764
7765 *) Updated the README file.
7766 [Ralf S. Engelschall]
7767
7768 *) Added various .cvsignore files in the CVS repository subdirs
7769 to make a "cvs update" really silent.
7770 [Ralf S. Engelschall]
7771
7772 *) Recompiled the error-definition header files and added
7773 missing symbols to the Win32 linker tables.
7774 [Ralf S. Engelschall]
7775
7776 *) Cleaned up the top-level documents;
7777 o new files: CHANGES and LICENSE
7778 o merged VERSION, HISTORY* and README* files a CHANGES.SSLeay
7779 o merged COPYRIGHT into LICENSE
7780 o removed obsolete TODO file
7781 o renamed MICROSOFT to INSTALL.W32
7782 [Ralf S. Engelschall]
7783
7784 *) Removed dummy files from the 0.9.1b source tree:
7785 crypto/asn1/x crypto/bio/cd crypto/bio/fg crypto/bio/grep crypto/bio/vi
7786 crypto/bn/asm/......add.c crypto/bn/asm/a.out crypto/dsa/f crypto/md5/f
7787 crypto/pem/gmon.out crypto/perlasm/f crypto/pkcs7/build crypto/rsa/f
7788 crypto/sha/asm/f crypto/threads/f ms/zzz ssl/f ssl/f.mak test/f
7789 util/f.mak util/pl/f util/pl/f.mak crypto/bf/bf_locl.old apps/f
7790 [Ralf S. Engelschall]
7791
7792 *) Added various platform portability fixes.
7793 [Mark J. Cox]
7794
7795 *) The Genesis of the OpenSSL rpject:
7796 We start with the latest (unreleased) SSLeay version 0.9.1b which Eric A.
7797 Young and Tim J. Hudson created while they were working for C2Net until
7798 summer 1998.
7799 [The OpenSSL Project]
7800
7801
7802 Changes between 0.9.0b and 0.9.1b [not released]
7803
7804 *) Updated a few CA certificates under certs/
7805 [Eric A. Young]
7806
7807 *) Changed some BIGNUM api stuff.
7808 [Eric A. Young]
7809
7810 *) Various platform ports: OpenBSD, Ultrix, IRIX 64bit, NetBSD,
7811 DGUX x86, Linux Alpha, etc.
7812 [Eric A. Young]
7813
7814 *) New COMP library [crypto/comp/] for SSL Record Layer Compression:
7815 RLE (dummy implemented) and ZLIB (really implemented when ZLIB is
7816 available).
7817 [Eric A. Young]
7818
7819 *) Add -strparse option to asn1pars program which parses nested
7820 binary structures
7821 [Dr Stephen Henson <shenson@bigfoot.com>]
7822
7823 *) Added "oid_file" to ssleay.cnf for "ca" and "req" programs.
7824 [Eric A. Young]
7825
7826 *) DSA fix for "ca" program.
7827 [Eric A. Young]
7828
7829 *) Added "-genkey" option to "dsaparam" program.
7830 [Eric A. Young]
7831
7832 *) Added RIPE MD160 (rmd160) message digest.
7833 [Eric A. Young]
7834
7835 *) Added -a (all) option to "ssleay version" command.
7836 [Eric A. Young]
7837
7838 *) Added PLATFORM define which is the id given to Configure.
7839 [Eric A. Young]
7840
7841 *) Added MemCheck_XXXX functions to crypto/mem.c for memory checking.
7842 [Eric A. Young]
7843
7844 *) Extended the ASN.1 parser routines.
7845 [Eric A. Young]
7846
7847 *) Extended BIO routines to support REUSEADDR, seek, tell, etc.
7848 [Eric A. Young]
7849
7850 *) Added a BN_CTX to the BN library.
7851 [Eric A. Young]
7852
7853 *) Fixed the weak key values in DES library
7854 [Eric A. Young]
7855
7856 *) Changed API in EVP library for cipher aliases.
7857 [Eric A. Young]
7858
7859 *) Added support for RC2/64bit cipher.
7860 [Eric A. Young]
7861
7862 *) Converted the lhash library to the crypto/mem.c functions.
7863 [Eric A. Young]
7864
7865 *) Added more recognized ASN.1 object ids.
7866 [Eric A. Young]
7867
7868 *) Added more RSA padding checks for SSL/TLS.
7869 [Eric A. Young]
7870
7871 *) Added BIO proxy/filter functionality.
7872 [Eric A. Young]
7873
7874 *) Added extra_certs to SSL_CTX which can be used
7875 send extra CA certificates to the client in the CA cert chain sending
7876 process. It can be configured with SSL_CTX_add_extra_chain_cert().
7877 [Eric A. Young]
7878
7879 *) Now Fortezza is denied in the authentication phase because
7880 this is key exchange mechanism is not supported by SSLeay at all.
7881 [Eric A. Young]
7882
7883 *) Additional PKCS1 checks.
7884 [Eric A. Young]
7885
7886 *) Support the string "TLSv1" for all TLS v1 ciphers.
7887 [Eric A. Young]
7888
7889 *) Added function SSL_get_ex_data_X509_STORE_CTX_idx() which gives the
7890 ex_data index of the SSL context in the X509_STORE_CTX ex_data.
7891 [Eric A. Young]
7892
7893 *) Fixed a few memory leaks.
7894 [Eric A. Young]
7895
7896 *) Fixed various code and comment typos.
7897 [Eric A. Young]
7898
7899 *) A minor bug in ssl/s3_clnt.c where there would always be 4 0
7900 bytes sent in the client random.
7901 [Edward Bishop <ebishop@spyglass.com>]
7902