]> git.ipfire.org Git - thirdparty/openssl.git/blob - CHANGES
Update default compiler options for default tls extension config.
[thirdparty/openssl.git] / CHANGES
1
2 OpenSSL CHANGES
3 _______________
4
5 Changes between 0.9.8i and 0.9.8j [xx XXX xxxx]
6
7 *) Enable TLS extensions by default.
8 [Ben Laurie]
9
10 *) Allow the CHIL engine to be loaded, whether the application is
11 multithreaded or not. (This does not release the developer from the
12 obligation to set up the dynamic locking callbacks.)
13 [Sander Temme <sander@temme.net>]
14
15 *) Use correct exit code if there is an error in dgst command.
16 [Steve Henson; problem pointed out by Roland Dirlewanger]
17
18 *) Tweak Configure so that you need to say "experimental-jpake" to enable
19 JPAKE, and need to use -DOPENSSL_EXPERIMENTAL_JPAKE in applications.
20 [Bodo Moeller]
21
22 *) Add experimental JPAKE support, including demo authentication in
23 s_client and s_server.
24 [Ben Laurie]
25
26 *) Set the comparison function in v3_addr_canonize().
27 [Rob Austein <sra@hactrn.net>]
28
29 *) Add support for XMPP STARTTLS in s_client.
30 [Philip Paeps <philip@freebsd.org>]
31
32 *) Change the server-side SSL_OP_NETSCAPE_REUSE_CIPHER_CHANGE_BUG behavior
33 to ensure that even with this option, only ciphersuites in the
34 server's preference list will be accepted. (Note that the option
35 applies only when resuming a session, so the earlier behavior was
36 just about the algorithm choice for symmetric cryptography.)
37 [Bodo Moeller]
38
39 Changes between 0.9.8h and 0.9.8i [15 Sep 2008]
40
41 *) Fix a state transitition in s3_srvr.c and d1_srvr.c
42 (was using SSL3_ST_CW_CLNT_HELLO_B, should be ..._ST_SW_SRVR_...).
43 [Nagendra Modadugu]
44
45 *) The fix in 0.9.8c that supposedly got rid of unsafe
46 double-checked locking was incomplete for RSA blinding,
47 addressing just one layer of what turns out to have been
48 doubly unsafe triple-checked locking.
49
50 So now fix this for real by retiring the MONT_HELPER macro
51 in crypto/rsa/rsa_eay.c.
52
53 [Bodo Moeller; problem pointed out by Marius Schilder]
54
55 *) Various precautionary measures:
56
57 - Avoid size_t integer overflow in HASH_UPDATE (md32_common.h).
58
59 - Avoid a buffer overflow in d2i_SSL_SESSION() (ssl_asn1.c).
60 (NB: This would require knowledge of the secret session ticket key
61 to exploit, in which case you'd be SOL either way.)
62
63 - Change bn_nist.c so that it will properly handle input BIGNUMs
64 outside the expected range.
65
66 - Enforce the 'num' check in BN_div() (bn_div.c) for non-BN_DEBUG
67 builds.
68
69 [Neel Mehta, Bodo Moeller]
70
71 *) Allow engines to be "soft loaded" - i.e. optionally don't die if
72 the load fails. Useful for distros.
73 [Ben Laurie and the FreeBSD team]
74
75 *) Add support for Local Machine Keyset attribute in PKCS#12 files.
76 [Steve Henson]
77
78 *) Fix BN_GF2m_mod_arr() top-bit cleanup code.
79 [Huang Ying]
80
81 *) Expand ENGINE to support engine supplied SSL client certificate functions.
82
83 This work was sponsored by Logica.
84 [Steve Henson]
85
86 *) Add CryptoAPI ENGINE to support use of RSA and DSA keys held in Windows
87 keystores. Support for SSL/TLS client authentication too.
88 Not compiled unless enable-capieng specified to Configure.
89
90 This work was sponsored by Logica.
91 [Steve Henson]
92
93 *) Fix bug in X509_ATTRIBUTE creation: dont set attribute using
94 ASN1_TYPE_set1 if MBSTRING flag set. This bug would crash certain
95 attribute creation routines such as certifcate requests and PKCS#12
96 files.
97 [Steve Henson]
98
99 Changes between 0.9.8g and 0.9.8h [28 May 2008]
100
101 *) Fix flaw if 'Server Key exchange message' is omitted from a TLS
102 handshake which could lead to a cilent crash as found using the
103 Codenomicon TLS test suite (CVE-2008-1672)
104 [Steve Henson, Mark Cox]
105
106 *) Fix double free in TLS server name extensions which could lead to
107 a remote crash found by Codenomicon TLS test suite (CVE-2008-0891)
108 [Joe Orton]
109
110 *) Clear error queue in SSL_CTX_use_certificate_chain_file()
111
112 Clear the error queue to ensure that error entries left from
113 older function calls do not interfere with the correct operation.
114 [Lutz Jaenicke, Erik de Castro Lopo]
115
116 *) Remove root CA certificates of commercial CAs:
117
118 The OpenSSL project does not recommend any specific CA and does not
119 have any policy with respect to including or excluding any CA.
120 Therefore it does not make any sense to ship an arbitrary selection
121 of root CA certificates with the OpenSSL software.
122 [Lutz Jaenicke]
123
124 *) RSA OAEP patches to fix two separate invalid memory reads.
125 The first one involves inputs when 'lzero' is greater than
126 'SHA_DIGEST_LENGTH' (it would read about SHA_DIGEST_LENGTH bytes
127 before the beginning of from). The second one involves inputs where
128 the 'db' section contains nothing but zeroes (there is a one-byte
129 invalid read after the end of 'db').
130 [Ivan Nestlerode <inestlerode@us.ibm.com>]
131
132 *) Partial backport from 0.9.9-dev:
133
134 Introduce bn_mul_mont (dedicated Montgomery multiplication
135 procedure) as a candidate for BIGNUM assembler implementation.
136 While 0.9.9-dev uses assembler for various architectures, only
137 x86_64 is available by default here in the 0.9.8 branch, and
138 32-bit x86 is available through a compile-time setting.
139
140 To try the 32-bit x86 assembler implementation, use Configure
141 option "enable-montasm" (which exists only for this backport).
142
143 As "enable-montasm" for 32-bit x86 disclaims code stability
144 anyway, in this constellation we activate additional code
145 backported from 0.9.9-dev for further performance improvements,
146 namely BN_from_montgomery_word. (To enable this otherwise,
147 e.g. x86_64, try "-DMONT_FROM_WORD___NON_DEFAULT_0_9_8_BUILD".)
148
149 [Andy Polyakov (backport partially by Bodo Moeller)]
150
151 *) Add TLS session ticket callback. This allows an application to set
152 TLS ticket cipher and HMAC keys rather than relying on hardcoded fixed
153 values. This is useful for key rollover for example where several key
154 sets may exist with different names.
155 [Steve Henson]
156
157 *) Reverse ENGINE-internal logic for caching default ENGINE handles.
158 This was broken until now in 0.9.8 releases, such that the only way
159 a registered ENGINE could be used (assuming it initialises
160 successfully on the host) was to explicitly set it as the default
161 for the relevant algorithms. This is in contradiction with 0.9.7
162 behaviour and the documentation. With this fix, when an ENGINE is
163 registered into a given algorithm's table of implementations, the
164 'uptodate' flag is reset so that auto-discovery will be used next
165 time a new context for that algorithm attempts to select an
166 implementation.
167 [Ian Lister (tweaked by Geoff Thorpe)]
168
169 *) Backport of CMS code to OpenSSL 0.9.8. This differs from the 0.9.9
170 implemention in the following ways:
171
172 Lack of EVP_PKEY_ASN1_METHOD means algorithm parameters have to be
173 hard coded.
174
175 Lack of BER streaming support means one pass streaming processing is
176 only supported if data is detached: setting the streaming flag is
177 ignored for embedded content.
178
179 CMS support is disabled by default and must be explicitly enabled
180 with the enable-cms configuration option.
181 [Steve Henson]
182
183 *) Update the GMP engine glue to do direct copies between BIGNUM and
184 mpz_t when openssl and GMP use the same limb size. Otherwise the
185 existing "conversion via a text string export" trick is still used.
186 [Paul Sheer <paulsheer@gmail.com>]
187
188 *) Zlib compression BIO. This is a filter BIO which compressed and
189 uncompresses any data passed through it.
190 [Steve Henson]
191
192 *) Add AES_wrap_key() and AES_unwrap_key() functions to implement
193 RFC3394 compatible AES key wrapping.
194 [Steve Henson]
195
196 *) Add utility functions to handle ASN1 structures. ASN1_STRING_set0():
197 sets string data without copying. X509_ALGOR_set0() and
198 X509_ALGOR_get0(): set and retrieve X509_ALGOR (AlgorithmIdentifier)
199 data. Attribute function X509at_get0_data_by_OBJ(): retrieves data
200 from an X509_ATTRIBUTE structure optionally checking it occurs only
201 once. ASN1_TYPE_set1(): set and ASN1_TYPE structure copying supplied
202 data.
203 [Steve Henson]
204
205 *) Fix BN flag handling in RSA_eay_mod_exp() and BN_MONT_CTX_set()
206 to get the expected BN_FLG_CONSTTIME behavior.
207 [Bodo Moeller (Google)]
208
209 *) Netware support:
210
211 - fixed wrong usage of ioctlsocket() when build for LIBC BSD sockets
212 - fixed do_tests.pl to run the test suite with CLIB builds too (CLIB_OPT)
213 - added some more tests to do_tests.pl
214 - fixed RunningProcess usage so that it works with newer LIBC NDKs too
215 - removed usage of BN_LLONG for CLIB builds to avoid runtime dependency
216 - added new Configure targets netware-clib-bsdsock, netware-clib-gcc,
217 netware-clib-bsdsock-gcc, netware-libc-bsdsock-gcc
218 - various changes to netware.pl to enable gcc-cross builds on Win32
219 platform
220 - changed crypto/bio/b_sock.c to work with macro functions (CLIB BSD)
221 - various changes to fix missing prototype warnings
222 - fixed x86nasm.pl to create correct asm files for NASM COFF output
223 - added AES, WHIRLPOOL and CPUID assembler code to build files
224 - added missing AES assembler make rules to mk1mf.pl
225 - fixed order of includes in apps/ocsp.c so that e_os.h settings apply
226 [Guenter Knauf <eflash@gmx.net>]
227
228 *) Implement certificate status request TLS extension defined in RFC3546.
229 A client can set the appropriate parameters and receive the encoded
230 OCSP response via a callback. A server can query the supplied parameters
231 and set the encoded OCSP response in the callback. Add simplified examples
232 to s_client and s_server.
233 [Steve Henson]
234
235 Changes between 0.9.8f and 0.9.8g [19 Oct 2007]
236
237 *) Fix various bugs:
238 + Binary incompatibility of ssl_ctx_st structure
239 + DTLS interoperation with non-compliant servers
240 + Don't call get_session_cb() without proposed session
241 + Fix ia64 assembler code
242 [Andy Polyakov, Steve Henson]
243
244 Changes between 0.9.8e and 0.9.8f [11 Oct 2007]
245
246 *) DTLS Handshake overhaul. There were longstanding issues with
247 OpenSSL DTLS implementation, which were making it impossible for
248 RFC 4347 compliant client to communicate with OpenSSL server.
249 Unfortunately just fixing these incompatibilities would "cut off"
250 pre-0.9.8f clients. To allow for hassle free upgrade post-0.9.8e
251 server keeps tolerating non RFC compliant syntax. The opposite is
252 not true, 0.9.8f client can not communicate with earlier server.
253 This update even addresses CVE-2007-4995.
254 [Andy Polyakov]
255
256 *) Changes to avoid need for function casts in OpenSSL: some compilers
257 (gcc 4.2 and later) reject their use.
258 [Kurt Roeckx <kurt@roeckx.be>, Peter Hartley <pdh@utter.chaos.org.uk>,
259 Steve Henson]
260
261 *) Add RFC4507 support to OpenSSL. This includes the corrections in
262 RFC4507bis. The encrypted ticket format is an encrypted encoded
263 SSL_SESSION structure, that way new session features are automatically
264 supported.
265
266 If a client application caches session in an SSL_SESSION structure
267 support is transparent because tickets are now stored in the encoded
268 SSL_SESSION.
269
270 The SSL_CTX structure automatically generates keys for ticket
271 protection in servers so again support should be possible
272 with no application modification.
273
274 If a client or server wishes to disable RFC4507 support then the option
275 SSL_OP_NO_TICKET can be set.
276
277 Add a TLS extension debugging callback to allow the contents of any client
278 or server extensions to be examined.
279
280 This work was sponsored by Google.
281 [Steve Henson]
282
283 *) Add initial support for TLS extensions, specifically for the server_name
284 extension so far. The SSL_SESSION, SSL_CTX, and SSL data structures now
285 have new members for a host name. The SSL data structure has an
286 additional member SSL_CTX *initial_ctx so that new sessions can be
287 stored in that context to allow for session resumption, even after the
288 SSL has been switched to a new SSL_CTX in reaction to a client's
289 server_name extension.
290
291 New functions (subject to change):
292
293 SSL_get_servername()
294 SSL_get_servername_type()
295 SSL_set_SSL_CTX()
296
297 New CTRL codes and macros (subject to change):
298
299 SSL_CTRL_SET_TLSEXT_SERVERNAME_CB
300 - SSL_CTX_set_tlsext_servername_callback()
301 SSL_CTRL_SET_TLSEXT_SERVERNAME_ARG
302 - SSL_CTX_set_tlsext_servername_arg()
303 SSL_CTRL_SET_TLSEXT_HOSTNAME - SSL_set_tlsext_host_name()
304
305 openssl s_client has a new '-servername ...' option.
306
307 openssl s_server has new options '-servername_host ...', '-cert2 ...',
308 '-key2 ...', '-servername_fatal' (subject to change). This allows
309 testing the HostName extension for a specific single host name ('-cert'
310 and '-key' remain fallbacks for handshakes without HostName
311 negotiation). If the unrecogninzed_name alert has to be sent, this by
312 default is a warning; it becomes fatal with the '-servername_fatal'
313 option.
314
315 [Peter Sylvester, Remy Allais, Christophe Renou, Steve Henson]
316
317 *) Add AES and SSE2 assembly language support to VC++ build.
318 [Steve Henson]
319
320 *) Mitigate attack on final subtraction in Montgomery reduction.
321 [Andy Polyakov]
322
323 *) Fix crypto/ec/ec_mult.c to work properly with scalars of value 0
324 (which previously caused an internal error).
325 [Bodo Moeller]
326
327 *) Squeeze another 10% out of IGE mode when in != out.
328 [Ben Laurie]
329
330 *) AES IGE mode speedup.
331 [Dean Gaudet (Google)]
332
333 *) Add the Korean symmetric 128-bit cipher SEED (see
334 http://www.kisa.or.kr/kisa/seed/jsp/seed_eng.jsp) and
335 add SEED ciphersuites from RFC 4162:
336
337 TLS_RSA_WITH_SEED_CBC_SHA = "SEED-SHA"
338 TLS_DHE_DSS_WITH_SEED_CBC_SHA = "DHE-DSS-SEED-SHA"
339 TLS_DHE_RSA_WITH_SEED_CBC_SHA = "DHE-RSA-SEED-SHA"
340 TLS_DH_anon_WITH_SEED_CBC_SHA = "ADH-SEED-SHA"
341
342 To minimize changes between patchlevels in the OpenSSL 0.9.8
343 series, SEED remains excluded from compilation unless OpenSSL
344 is configured with 'enable-seed'.
345 [KISA, Bodo Moeller]
346
347 *) Mitigate branch prediction attacks, which can be practical if a
348 single processor is shared, allowing a spy process to extract
349 information. For detailed background information, see
350 http://eprint.iacr.org/2007/039 (O. Aciicmez, S. Gueron,
351 J.-P. Seifert, "New Branch Prediction Vulnerabilities in OpenSSL
352 and Necessary Software Countermeasures"). The core of the change
353 are new versions BN_div_no_branch() and
354 BN_mod_inverse_no_branch() of BN_div() and BN_mod_inverse(),
355 respectively, which are slower, but avoid the security-relevant
356 conditional branches. These are automatically called by BN_div()
357 and BN_mod_inverse() if the flag BN_FLG_CONSTTIME is set for one
358 of the input BIGNUMs. Also, BN_is_bit_set() has been changed to
359 remove a conditional branch.
360
361 BN_FLG_CONSTTIME is the new name for the previous
362 BN_FLG_EXP_CONSTTIME flag, since it now affects more than just
363 modular exponentiation. (Since OpenSSL 0.9.7h, setting this flag
364 in the exponent causes BN_mod_exp_mont() to use the alternative
365 implementation in BN_mod_exp_mont_consttime().) The old name
366 remains as a deprecated alias.
367
368 Similary, RSA_FLAG_NO_EXP_CONSTTIME is replaced by a more general
369 RSA_FLAG_NO_CONSTTIME flag since the RSA implementation now uses
370 constant-time implementations for more than just exponentiation.
371 Here too the old name is kept as a deprecated alias.
372
373 BN_BLINDING_new() will now use BN_dup() for the modulus so that
374 the BN_BLINDING structure gets an independent copy of the
375 modulus. This means that the previous "BIGNUM *m" argument to
376 BN_BLINDING_new() and to BN_BLINDING_create_param() now
377 essentially becomes "const BIGNUM *m", although we can't actually
378 change this in the header file before 0.9.9. It allows
379 RSA_setup_blinding() to use BN_with_flags() on the modulus to
380 enable BN_FLG_CONSTTIME.
381
382 [Matthew D Wood (Intel Corp)]
383
384 *) In the SSL/TLS server implementation, be strict about session ID
385 context matching (which matters if an application uses a single
386 external cache for different purposes). Previously,
387 out-of-context reuse was forbidden only if SSL_VERIFY_PEER was
388 set. This did ensure strict client verification, but meant that,
389 with applications using a single external cache for quite
390 different requirements, clients could circumvent ciphersuite
391 restrictions for a given session ID context by starting a session
392 in a different context.
393 [Bodo Moeller]
394
395 *) Include "!eNULL" in SSL_DEFAULT_CIPHER_LIST to make sure that
396 a ciphersuite string such as "DEFAULT:RSA" cannot enable
397 authentication-only ciphersuites.
398 [Bodo Moeller]
399
400 *) Update the SSL_get_shared_ciphers() fix CVE-2006-3738 which was
401 not complete and could lead to a possible single byte overflow
402 (CVE-2007-5135) [Ben Laurie]
403
404 Changes between 0.9.8d and 0.9.8e [23 Feb 2007]
405
406 *) Since AES128 and AES256 (and similarly Camellia128 and
407 Camellia256) share a single mask bit in the logic of
408 ssl/ssl_ciph.c, the code for masking out disabled ciphers needs a
409 kludge to work properly if AES128 is available and AES256 isn't
410 (or if Camellia128 is available and Camellia256 isn't).
411 [Victor Duchovni]
412
413 *) Fix the BIT STRING encoding generated by crypto/ec/ec_asn1.c
414 (within i2d_ECPrivateKey, i2d_ECPKParameters, i2d_ECParameters):
415 When a point or a seed is encoded in a BIT STRING, we need to
416 prevent the removal of trailing zero bits to get the proper DER
417 encoding. (By default, crypto/asn1/a_bitstr.c assumes the case
418 of a NamedBitList, for which trailing 0 bits need to be removed.)
419 [Bodo Moeller]
420
421 *) Have SSL/TLS server implementation tolerate "mismatched" record
422 protocol version while receiving ClientHello even if the
423 ClientHello is fragmented. (The server can't insist on the
424 particular protocol version it has chosen before the ServerHello
425 message has informed the client about his choice.)
426 [Bodo Moeller]
427
428 *) Add RFC 3779 support.
429 [Rob Austein for ARIN, Ben Laurie]
430
431 *) Load error codes if they are not already present instead of using a
432 static variable. This allows them to be cleanly unloaded and reloaded.
433 Improve header file function name parsing.
434 [Steve Henson]
435
436 *) extend SMTP and IMAP protocol emulation in s_client to use EHLO
437 or CAPABILITY handshake as required by RFCs.
438 [Goetz Babin-Ebell]
439
440 Changes between 0.9.8c and 0.9.8d [28 Sep 2006]
441
442 *) Introduce limits to prevent malicious keys being able to
443 cause a denial of service. (CVE-2006-2940)
444 [Steve Henson, Bodo Moeller]
445
446 *) Fix ASN.1 parsing of certain invalid structures that can result
447 in a denial of service. (CVE-2006-2937) [Steve Henson]
448
449 *) Fix buffer overflow in SSL_get_shared_ciphers() function.
450 (CVE-2006-3738) [Tavis Ormandy and Will Drewry, Google Security Team]
451
452 *) Fix SSL client code which could crash if connecting to a
453 malicious SSLv2 server. (CVE-2006-4343)
454 [Tavis Ormandy and Will Drewry, Google Security Team]
455
456 *) Since 0.9.8b, ciphersuite strings naming explicit ciphersuites
457 match only those. Before that, "AES256-SHA" would be interpreted
458 as a pattern and match "AES128-SHA" too (since AES128-SHA got
459 the same strength classification in 0.9.7h) as we currently only
460 have a single AES bit in the ciphersuite description bitmap.
461 That change, however, also applied to ciphersuite strings such as
462 "RC4-MD5" that intentionally matched multiple ciphersuites --
463 namely, SSL 2.0 ciphersuites in addition to the more common ones
464 from SSL 3.0/TLS 1.0.
465
466 So we change the selection algorithm again: Naming an explicit
467 ciphersuite selects this one ciphersuite, and any other similar
468 ciphersuite (same bitmap) from *other* protocol versions.
469 Thus, "RC4-MD5" again will properly select both the SSL 2.0
470 ciphersuite and the SSL 3.0/TLS 1.0 ciphersuite.
471
472 Since SSL 2.0 does not have any ciphersuites for which the
473 128/256 bit distinction would be relevant, this works for now.
474 The proper fix will be to use different bits for AES128 and
475 AES256, which would have avoided the problems from the beginning;
476 however, bits are scarce, so we can only do this in a new release
477 (not just a patchlevel) when we can change the SSL_CIPHER
478 definition to split the single 'unsigned long mask' bitmap into
479 multiple values to extend the available space.
480
481 [Bodo Moeller]
482
483 Changes between 0.9.8b and 0.9.8c [05 Sep 2006]
484
485 *) Avoid PKCS #1 v1.5 signature attack discovered by Daniel Bleichenbacher
486 (CVE-2006-4339) [Ben Laurie and Google Security Team]
487
488 *) Add AES IGE and biIGE modes.
489 [Ben Laurie]
490
491 *) Change the Unix randomness entropy gathering to use poll() when
492 possible instead of select(), since the latter has some
493 undesirable limitations.
494 [Darryl Miles via Richard Levitte and Bodo Moeller]
495
496 *) Disable "ECCdraft" ciphersuites more thoroughly. Now special
497 treatment in ssl/ssl_ciph.s makes sure that these ciphersuites
498 cannot be implicitly activated as part of, e.g., the "AES" alias.
499 However, please upgrade to OpenSSL 0.9.9[-dev] for
500 non-experimental use of the ECC ciphersuites to get TLS extension
501 support, which is required for curve and point format negotiation
502 to avoid potential handshake problems.
503 [Bodo Moeller]
504
505 *) Disable rogue ciphersuites:
506
507 - SSLv2 0x08 0x00 0x80 ("RC4-64-MD5")
508 - SSLv3/TLSv1 0x00 0x61 ("EXP1024-RC2-CBC-MD5")
509 - SSLv3/TLSv1 0x00 0x60 ("EXP1024-RC4-MD5")
510
511 The latter two were purportedly from
512 draft-ietf-tls-56-bit-ciphersuites-0[01].txt, but do not really
513 appear there.
514
515 Also deactivate the remaining ciphersuites from
516 draft-ietf-tls-56-bit-ciphersuites-01.txt. These are just as
517 unofficial, and the ID has long expired.
518 [Bodo Moeller]
519
520 *) Fix RSA blinding Heisenbug (problems sometimes occured on
521 dual-core machines) and other potential thread-safety issues.
522 [Bodo Moeller]
523
524 *) Add the symmetric cipher Camellia (128-bit, 192-bit, 256-bit key
525 versions), which is now available for royalty-free use
526 (see http://info.isl.ntt.co.jp/crypt/eng/info/chiteki.html).
527 Also, add Camellia TLS ciphersuites from RFC 4132.
528
529 To minimize changes between patchlevels in the OpenSSL 0.9.8
530 series, Camellia remains excluded from compilation unless OpenSSL
531 is configured with 'enable-camellia'.
532 [NTT]
533
534 *) Disable the padding bug check when compression is in use. The padding
535 bug check assumes the first packet is of even length, this is not
536 necessarily true if compresssion is enabled and can result in false
537 positives causing handshake failure. The actual bug test is ancient
538 code so it is hoped that implementations will either have fixed it by
539 now or any which still have the bug do not support compression.
540 [Steve Henson]
541
542 Changes between 0.9.8a and 0.9.8b [04 May 2006]
543
544 *) When applying a cipher rule check to see if string match is an explicit
545 cipher suite and only match that one cipher suite if it is.
546 [Steve Henson]
547
548 *) Link in manifests for VC++ if needed.
549 [Austin Ziegler <halostatue@gmail.com>]
550
551 *) Update support for ECC-based TLS ciphersuites according to
552 draft-ietf-tls-ecc-12.txt with proposed changes (but without
553 TLS extensions, which are supported starting with the 0.9.9
554 branch, not in the OpenSSL 0.9.8 branch).
555 [Douglas Stebila]
556
557 *) New functions EVP_CIPHER_CTX_new() and EVP_CIPHER_CTX_free() to support
558 opaque EVP_CIPHER_CTX handling.
559 [Steve Henson]
560
561 *) Fixes and enhancements to zlib compression code. We now only use
562 "zlib1.dll" and use the default __cdecl calling convention on Win32
563 to conform with the standards mentioned here:
564 http://www.zlib.net/DLL_FAQ.txt
565 Static zlib linking now works on Windows and the new --with-zlib-include
566 --with-zlib-lib options to Configure can be used to supply the location
567 of the headers and library. Gracefully handle case where zlib library
568 can't be loaded.
569 [Steve Henson]
570
571 *) Several fixes and enhancements to the OID generation code. The old code
572 sometimes allowed invalid OIDs (1.X for X >= 40 for example), couldn't
573 handle numbers larger than ULONG_MAX, truncated printing and had a
574 non standard OBJ_obj2txt() behaviour.
575 [Steve Henson]
576
577 *) Add support for building of engines under engine/ as shared libraries
578 under VC++ build system.
579 [Steve Henson]
580
581 *) Corrected the numerous bugs in the Win32 path splitter in DSO.
582 Hopefully, we will not see any false combination of paths any more.
583 [Richard Levitte]
584
585 Changes between 0.9.8 and 0.9.8a [11 Oct 2005]
586
587 *) Remove the functionality of SSL_OP_MSIE_SSLV2_RSA_PADDING
588 (part of SSL_OP_ALL). This option used to disable the
589 countermeasure against man-in-the-middle protocol-version
590 rollback in the SSL 2.0 server implementation, which is a bad
591 idea. (CVE-2005-2969)
592
593 [Bodo Moeller; problem pointed out by Yutaka Oiwa (Research Center
594 for Information Security, National Institute of Advanced Industrial
595 Science and Technology [AIST], Japan)]
596
597 *) Add two function to clear and return the verify parameter flags.
598 [Steve Henson]
599
600 *) Keep cipherlists sorted in the source instead of sorting them at
601 runtime, thus removing the need for a lock.
602 [Nils Larsch]
603
604 *) Avoid some small subgroup attacks in Diffie-Hellman.
605 [Nick Mathewson and Ben Laurie]
606
607 *) Add functions for well-known primes.
608 [Nick Mathewson]
609
610 *) Extended Windows CE support.
611 [Satoshi Nakamura and Andy Polyakov]
612
613 *) Initialize SSL_METHOD structures at compile time instead of during
614 runtime, thus removing the need for a lock.
615 [Steve Henson]
616
617 *) Make PKCS7_decrypt() work even if no certificate is supplied by
618 attempting to decrypt each encrypted key in turn. Add support to
619 smime utility.
620 [Steve Henson]
621
622 Changes between 0.9.7h and 0.9.8 [05 Jul 2005]
623
624 [NB: OpenSSL 0.9.7i and later 0.9.7 patch levels were released after
625 OpenSSL 0.9.8.]
626
627 *) Add libcrypto.pc and libssl.pc for those who feel they need them.
628 [Richard Levitte]
629
630 *) Change CA.sh and CA.pl so they don't bundle the CSR and the private
631 key into the same file any more.
632 [Richard Levitte]
633
634 *) Add initial support for Win64, both IA64 and AMD64/x64 flavors.
635 [Andy Polyakov]
636
637 *) Add -utf8 command line and config file option to 'ca'.
638 [Stefan <stf@udoma.org]
639
640 *) Removed the macro des_crypt(), as it seems to conflict with some
641 libraries. Use DES_crypt().
642 [Richard Levitte]
643
644 *) Correct naming of the 'chil' and '4758cca' ENGINEs. This
645 involves renaming the source and generated shared-libs for
646 both. The engines will accept the corrected or legacy ids
647 ('ncipher' and '4758_cca' respectively) when binding. NB,
648 this only applies when building 'shared'.
649 [Corinna Vinschen <vinschen@redhat.com> and Geoff Thorpe]
650
651 *) Add attribute functions to EVP_PKEY structure. Modify
652 PKCS12_create() to recognize a CSP name attribute and
653 use it. Make -CSP option work again in pkcs12 utility.
654 [Steve Henson]
655
656 *) Add new functionality to the bn blinding code:
657 - automatic re-creation of the BN_BLINDING parameters after
658 a fixed number of uses (currently 32)
659 - add new function for parameter creation
660 - introduce flags to control the update behaviour of the
661 BN_BLINDING parameters
662 - hide BN_BLINDING structure
663 Add a second BN_BLINDING slot to the RSA structure to improve
664 performance when a single RSA object is shared among several
665 threads.
666 [Nils Larsch]
667
668 *) Add support for DTLS.
669 [Nagendra Modadugu <nagendra@cs.stanford.edu> and Ben Laurie]
670
671 *) Add support for DER encoded private keys (SSL_FILETYPE_ASN1)
672 to SSL_CTX_use_PrivateKey_file() and SSL_use_PrivateKey_file()
673 [Walter Goulet]
674
675 *) Remove buggy and incompletet DH cert support from
676 ssl/ssl_rsa.c and ssl/s3_both.c
677 [Nils Larsch]
678
679 *) Use SHA-1 instead of MD5 as the default digest algorithm for
680 the apps/openssl applications.
681 [Nils Larsch]
682
683 *) Compile clean with "-Wall -Wmissing-prototypes
684 -Wstrict-prototypes -Wmissing-declarations -Werror". Currently
685 DEBUG_SAFESTACK must also be set.
686 [Ben Laurie]
687
688 *) Change ./Configure so that certain algorithms can be disabled by default.
689 The new counterpiece to "no-xxx" is "enable-xxx".
690
691 The patented RC5 and MDC2 algorithms will now be disabled unless
692 "enable-rc5" and "enable-mdc2", respectively, are specified.
693
694 (IDEA remains enabled despite being patented. This is because IDEA
695 is frequently required for interoperability, and there is no license
696 fee for non-commercial use. As before, "no-idea" can be used to
697 avoid this algorithm.)
698
699 [Bodo Moeller]
700
701 *) Add processing of proxy certificates (see RFC 3820). This work was
702 sponsored by KTH (The Royal Institute of Technology in Stockholm) and
703 EGEE (Enabling Grids for E-science in Europe).
704 [Richard Levitte]
705
706 *) RC4 performance overhaul on modern architectures/implementations, such
707 as Intel P4, IA-64 and AMD64.
708 [Andy Polyakov]
709
710 *) New utility extract-section.pl. This can be used specify an alternative
711 section number in a pod file instead of having to treat each file as
712 a separate case in Makefile. This can be done by adding two lines to the
713 pod file:
714
715 =for comment openssl_section:XXX
716
717 The blank line is mandatory.
718
719 [Steve Henson]
720
721 *) New arguments -certform, -keyform and -pass for s_client and s_server
722 to allow alternative format key and certificate files and passphrase
723 sources.
724 [Steve Henson]
725
726 *) New structure X509_VERIFY_PARAM which combines current verify parameters,
727 update associated structures and add various utility functions.
728
729 Add new policy related verify parameters, include policy checking in
730 standard verify code. Enhance 'smime' application with extra parameters
731 to support policy checking and print out.
732 [Steve Henson]
733
734 *) Add a new engine to support VIA PadLock ACE extensions in the VIA C3
735 Nehemiah processors. These extensions support AES encryption in hardware
736 as well as RNG (though RNG support is currently disabled).
737 [Michal Ludvig <michal@logix.cz>, with help from Andy Polyakov]
738
739 *) Deprecate BN_[get|set]_params() functions (they were ignored internally).
740 [Geoff Thorpe]
741
742 *) New FIPS 180-2 algorithms, SHA-224/-256/-384/-512 are implemented.
743 [Andy Polyakov and a number of other people]
744
745 *) Improved PowerPC platform support. Most notably BIGNUM assembler
746 implementation contributed by IBM.
747 [Suresh Chari, Peter Waltenberg, Andy Polyakov]
748
749 *) The new 'RSA_generate_key_ex' function now takes a BIGNUM for the public
750 exponent rather than 'unsigned long'. There is a corresponding change to
751 the new 'rsa_keygen' element of the RSA_METHOD structure.
752 [Jelte Jansen, Geoff Thorpe]
753
754 *) Functionality for creating the initial serial number file is now
755 moved from CA.pl to the 'ca' utility with a new option -create_serial.
756
757 (Before OpenSSL 0.9.7e, CA.pl used to initialize the serial
758 number file to 1, which is bound to cause problems. To avoid
759 the problems while respecting compatibility between different 0.9.7
760 patchlevels, 0.9.7e employed 'openssl x509 -next_serial' in
761 CA.pl for serial number initialization. With the new release 0.9.8,
762 we can fix the problem directly in the 'ca' utility.)
763 [Steve Henson]
764
765 *) Reduced header interdepencies by declaring more opaque objects in
766 ossl_typ.h. As a consequence, including some headers (eg. engine.h) will
767 give fewer recursive includes, which could break lazy source code - so
768 this change is covered by the OPENSSL_NO_DEPRECATED symbol. As always,
769 developers should define this symbol when building and using openssl to
770 ensure they track the recommended behaviour, interfaces, [etc], but
771 backwards-compatible behaviour prevails when this isn't defined.
772 [Geoff Thorpe]
773
774 *) New function X509_POLICY_NODE_print() which prints out policy nodes.
775 [Steve Henson]
776
777 *) Add new EVP function EVP_CIPHER_CTX_rand_key and associated functionality.
778 This will generate a random key of the appropriate length based on the
779 cipher context. The EVP_CIPHER can provide its own random key generation
780 routine to support keys of a specific form. This is used in the des and
781 3des routines to generate a key of the correct parity. Update S/MIME
782 code to use new functions and hence generate correct parity DES keys.
783 Add EVP_CHECK_DES_KEY #define to return an error if the key is not
784 valid (weak or incorrect parity).
785 [Steve Henson]
786
787 *) Add a local set of CRLs that can be used by X509_verify_cert() as well
788 as looking them up. This is useful when the verified structure may contain
789 CRLs, for example PKCS#7 signedData. Modify PKCS7_verify() to use any CRLs
790 present unless the new PKCS7_NO_CRL flag is asserted.
791 [Steve Henson]
792
793 *) Extend ASN1 oid configuration module. It now additionally accepts the
794 syntax:
795
796 shortName = some long name, 1.2.3.4
797 [Steve Henson]
798
799 *) Reimplemented the BN_CTX implementation. There is now no more static
800 limitation on the number of variables it can handle nor the depth of the
801 "stack" handling for BN_CTX_start()/BN_CTX_end() pairs. The stack
802 information can now expand as required, and rather than having a single
803 static array of bignums, BN_CTX now uses a linked-list of such arrays
804 allowing it to expand on demand whilst maintaining the usefulness of
805 BN_CTX's "bundling".
806 [Geoff Thorpe]
807
808 *) Add a missing BN_CTX parameter to the 'rsa_mod_exp' callback in RSA_METHOD
809 to allow all RSA operations to function using a single BN_CTX.
810 [Geoff Thorpe]
811
812 *) Preliminary support for certificate policy evaluation and checking. This
813 is initially intended to pass the tests outlined in "Conformance Testing
814 of Relying Party Client Certificate Path Processing Logic" v1.07.
815 [Steve Henson]
816
817 *) bn_dup_expand() has been deprecated, it was introduced in 0.9.7 and
818 remained unused and not that useful. A variety of other little bignum
819 tweaks and fixes have also been made continuing on from the audit (see
820 below).
821 [Geoff Thorpe]
822
823 *) Constify all or almost all d2i, c2i, s2i and r2i functions, along with
824 associated ASN1, EVP and SSL functions and old ASN1 macros.
825 [Richard Levitte]
826
827 *) BN_zero() only needs to set 'top' and 'neg' to zero for correct results,
828 and this should never fail. So the return value from the use of
829 BN_set_word() (which can fail due to needless expansion) is now deprecated;
830 if OPENSSL_NO_DEPRECATED is defined, BN_zero() is a void macro.
831 [Geoff Thorpe]
832
833 *) BN_CTX_get() should return zero-valued bignums, providing the same
834 initialised value as BN_new().
835 [Geoff Thorpe, suggested by Ulf Möller]
836
837 *) Support for inhibitAnyPolicy certificate extension.
838 [Steve Henson]
839
840 *) An audit of the BIGNUM code is underway, for which debugging code is
841 enabled when BN_DEBUG is defined. This makes stricter enforcements on what
842 is considered valid when processing BIGNUMs, and causes execution to
843 assert() when a problem is discovered. If BN_DEBUG_RAND is defined,
844 further steps are taken to deliberately pollute unused data in BIGNUM
845 structures to try and expose faulty code further on. For now, openssl will
846 (in its default mode of operation) continue to tolerate the inconsistent
847 forms that it has tolerated in the past, but authors and packagers should
848 consider trying openssl and their own applications when compiled with
849 these debugging symbols defined. It will help highlight potential bugs in
850 their own code, and will improve the test coverage for OpenSSL itself. At
851 some point, these tighter rules will become openssl's default to improve
852 maintainability, though the assert()s and other overheads will remain only
853 in debugging configurations. See bn.h for more details.
854 [Geoff Thorpe, Nils Larsch, Ulf Möller]
855
856 *) BN_CTX_init() has been deprecated, as BN_CTX is an opaque structure
857 that can only be obtained through BN_CTX_new() (which implicitly
858 initialises it). The presence of this function only made it possible
859 to overwrite an existing structure (and cause memory leaks).
860 [Geoff Thorpe]
861
862 *) Because of the callback-based approach for implementing LHASH as a
863 template type, lh_insert() adds opaque objects to hash-tables and
864 lh_doall() or lh_doall_arg() are typically used with a destructor callback
865 to clean up those corresponding objects before destroying the hash table
866 (and losing the object pointers). So some over-zealous constifications in
867 LHASH have been relaxed so that lh_insert() does not take (nor store) the
868 objects as "const" and the lh_doall[_arg] callback wrappers are not
869 prototyped to have "const" restrictions on the object pointers they are
870 given (and so aren't required to cast them away any more).
871 [Geoff Thorpe]
872
873 *) The tmdiff.h API was so ugly and minimal that our own timing utility
874 (speed) prefers to use its own implementation. The two implementations
875 haven't been consolidated as yet (volunteers?) but the tmdiff API has had
876 its object type properly exposed (MS_TM) instead of casting to/from "char
877 *". This may still change yet if someone realises MS_TM and "ms_time_***"
878 aren't necessarily the greatest nomenclatures - but this is what was used
879 internally to the implementation so I've used that for now.
880 [Geoff Thorpe]
881
882 *) Ensure that deprecated functions do not get compiled when
883 OPENSSL_NO_DEPRECATED is defined. Some "openssl" subcommands and a few of
884 the self-tests were still using deprecated key-generation functions so
885 these have been updated also.
886 [Geoff Thorpe]
887
888 *) Reorganise PKCS#7 code to separate the digest location functionality
889 into PKCS7_find_digest(), digest addtion into PKCS7_bio_add_digest().
890 New function PKCS7_set_digest() to set the digest type for PKCS#7
891 digestedData type. Add additional code to correctly generate the
892 digestedData type and add support for this type in PKCS7 initialization
893 functions.
894 [Steve Henson]
895
896 *) New function PKCS7_set0_type_other() this initializes a PKCS7
897 structure of type "other".
898 [Steve Henson]
899
900 *) Fix prime generation loop in crypto/bn/bn_prime.pl by making
901 sure the loop does correctly stop and breaking ("division by zero")
902 modulus operations are not performed. The (pre-generated) prime
903 table crypto/bn/bn_prime.h was already correct, but it could not be
904 re-generated on some platforms because of the "division by zero"
905 situation in the script.
906 [Ralf S. Engelschall]
907
908 *) Update support for ECC-based TLS ciphersuites according to
909 draft-ietf-tls-ecc-03.txt: the KDF1 key derivation function with
910 SHA-1 now is only used for "small" curves (where the
911 representation of a field element takes up to 24 bytes); for
912 larger curves, the field element resulting from ECDH is directly
913 used as premaster secret.
914 [Douglas Stebila (Sun Microsystems Laboratories)]
915
916 *) Add code for kP+lQ timings to crypto/ec/ectest.c, and add SEC2
917 curve secp160r1 to the tests.
918 [Douglas Stebila (Sun Microsystems Laboratories)]
919
920 *) Add the possibility to load symbols globally with DSO.
921 [Götz Babin-Ebell <babin-ebell@trustcenter.de> via Richard Levitte]
922
923 *) Add the functions ERR_set_mark() and ERR_pop_to_mark() for better
924 control of the error stack.
925 [Richard Levitte]
926
927 *) Add support for STORE in ENGINE.
928 [Richard Levitte]
929
930 *) Add the STORE type. The intention is to provide a common interface
931 to certificate and key stores, be they simple file-based stores, or
932 HSM-type store, or LDAP stores, or...
933 NOTE: The code is currently UNTESTED and isn't really used anywhere.
934 [Richard Levitte]
935
936 *) Add a generic structure called OPENSSL_ITEM. This can be used to
937 pass a list of arguments to any function as well as provide a way
938 for a function to pass data back to the caller.
939 [Richard Levitte]
940
941 *) Add the functions BUF_strndup() and BUF_memdup(). BUF_strndup()
942 works like BUF_strdup() but can be used to duplicate a portion of
943 a string. The copy gets NUL-terminated. BUF_memdup() duplicates
944 a memory area.
945 [Richard Levitte]
946
947 *) Add the function sk_find_ex() which works like sk_find(), but will
948 return an index to an element even if an exact match couldn't be
949 found. The index is guaranteed to point at the element where the
950 searched-for key would be inserted to preserve sorting order.
951 [Richard Levitte]
952
953 *) Add the function OBJ_bsearch_ex() which works like OBJ_bsearch() but
954 takes an extra flags argument for optional functionality. Currently,
955 the following flags are defined:
956
957 OBJ_BSEARCH_VALUE_ON_NOMATCH
958 This one gets OBJ_bsearch_ex() to return a pointer to the first
959 element where the comparing function returns a negative or zero
960 number.
961
962 OBJ_BSEARCH_FIRST_VALUE_ON_MATCH
963 This one gets OBJ_bsearch_ex() to return a pointer to the first
964 element where the comparing function returns zero. This is useful
965 if there are more than one element where the comparing function
966 returns zero.
967 [Richard Levitte]
968
969 *) Make it possible to create self-signed certificates with 'openssl ca'
970 in such a way that the self-signed certificate becomes part of the
971 CA database and uses the same mechanisms for serial number generation
972 as all other certificate signing. The new flag '-selfsign' enables
973 this functionality. Adapt CA.sh and CA.pl.in.
974 [Richard Levitte]
975
976 *) Add functionality to check the public key of a certificate request
977 against a given private. This is useful to check that a certificate
978 request can be signed by that key (self-signing).
979 [Richard Levitte]
980
981 *) Make it possible to have multiple active certificates with the same
982 subject in the CA index file. This is done only if the keyword
983 'unique_subject' is set to 'no' in the main CA section (default
984 if 'CA_default') of the configuration file. The value is saved
985 with the database itself in a separate index attribute file,
986 named like the index file with '.attr' appended to the name.
987 [Richard Levitte]
988
989 *) Generate muti valued AVAs using '+' notation in config files for
990 req and dirName.
991 [Steve Henson]
992
993 *) Support for nameConstraints certificate extension.
994 [Steve Henson]
995
996 *) Support for policyConstraints certificate extension.
997 [Steve Henson]
998
999 *) Support for policyMappings certificate extension.
1000 [Steve Henson]
1001
1002 *) Make sure the default DSA_METHOD implementation only uses its
1003 dsa_mod_exp() and/or bn_mod_exp() handlers if they are non-NULL,
1004 and change its own handlers to be NULL so as to remove unnecessary
1005 indirection. This lets alternative implementations fallback to the
1006 default implementation more easily.
1007 [Geoff Thorpe]
1008
1009 *) Support for directoryName in GeneralName related extensions
1010 in config files.
1011 [Steve Henson]
1012
1013 *) Make it possible to link applications using Makefile.shared.
1014 Make that possible even when linking against static libraries!
1015 [Richard Levitte]
1016
1017 *) Support for single pass processing for S/MIME signing. This now
1018 means that S/MIME signing can be done from a pipe, in addition
1019 cleartext signing (multipart/signed type) is effectively streaming
1020 and the signed data does not need to be all held in memory.
1021
1022 This is done with a new flag PKCS7_STREAM. When this flag is set
1023 PKCS7_sign() only initializes the PKCS7 structure and the actual signing
1024 is done after the data is output (and digests calculated) in
1025 SMIME_write_PKCS7().
1026 [Steve Henson]
1027
1028 *) Add full support for -rpath/-R, both in shared libraries and
1029 applications, at least on the platforms where it's known how
1030 to do it.
1031 [Richard Levitte]
1032
1033 *) In crypto/ec/ec_mult.c, implement fast point multiplication with
1034 precomputation, based on wNAF splitting: EC_GROUP_precompute_mult()
1035 will now compute a table of multiples of the generator that
1036 makes subsequent invocations of EC_POINTs_mul() or EC_POINT_mul()
1037 faster (notably in the case of a single point multiplication,
1038 scalar * generator).
1039 [Nils Larsch, Bodo Moeller]
1040
1041 *) IPv6 support for certificate extensions. The various extensions
1042 which use the IP:a.b.c.d can now take IPv6 addresses using the
1043 formats of RFC1884 2.2 . IPv6 addresses are now also displayed
1044 correctly.
1045 [Steve Henson]
1046
1047 *) Added an ENGINE that implements RSA by performing private key
1048 exponentiations with the GMP library. The conversions to and from
1049 GMP's mpz_t format aren't optimised nor are any montgomery forms
1050 cached, and on x86 it appears OpenSSL's own performance has caught up.
1051 However there are likely to be other architectures where GMP could
1052 provide a boost. This ENGINE is not built in by default, but it can be
1053 specified at Configure time and should be accompanied by the necessary
1054 linker additions, eg;
1055 ./config -DOPENSSL_USE_GMP -lgmp
1056 [Geoff Thorpe]
1057
1058 *) "openssl engine" will not display ENGINE/DSO load failure errors when
1059 testing availability of engines with "-t" - the old behaviour is
1060 produced by increasing the feature's verbosity with "-tt".
1061 [Geoff Thorpe]
1062
1063 *) ECDSA routines: under certain error conditions uninitialized BN objects
1064 could be freed. Solution: make sure initialization is performed early
1065 enough. (Reported and fix supplied by Nils Larsch <nla@trustcenter.de>
1066 via PR#459)
1067 [Lutz Jaenicke]
1068
1069 *) Key-generation can now be implemented in RSA_METHOD, DSA_METHOD
1070 and DH_METHOD (eg. by ENGINE implementations) to override the normal
1071 software implementations. For DSA and DH, parameter generation can
1072 also be overriden by providing the appropriate method callbacks.
1073 [Geoff Thorpe]
1074
1075 *) Change the "progress" mechanism used in key-generation and
1076 primality testing to functions that take a new BN_GENCB pointer in
1077 place of callback/argument pairs. The new API functions have "_ex"
1078 postfixes and the older functions are reimplemented as wrappers for
1079 the new ones. The OPENSSL_NO_DEPRECATED symbol can be used to hide
1080 declarations of the old functions to help (graceful) attempts to
1081 migrate to the new functions. Also, the new key-generation API
1082 functions operate on a caller-supplied key-structure and return
1083 success/failure rather than returning a key or NULL - this is to
1084 help make "keygen" another member function of RSA_METHOD etc.
1085
1086 Example for using the new callback interface:
1087
1088 int (*my_callback)(int a, int b, BN_GENCB *cb) = ...;
1089 void *my_arg = ...;
1090 BN_GENCB my_cb;
1091
1092 BN_GENCB_set(&my_cb, my_callback, my_arg);
1093
1094 return BN_is_prime_ex(some_bignum, BN_prime_checks, NULL, &cb);
1095 /* For the meaning of a, b in calls to my_callback(), see the
1096 * documentation of the function that calls the callback.
1097 * cb will point to my_cb; my_arg can be retrieved as cb->arg.
1098 * my_callback should return 1 if it wants BN_is_prime_ex()
1099 * to continue, or 0 to stop.
1100 */
1101
1102 [Geoff Thorpe]
1103
1104 *) Change the ZLIB compression method to be stateful, and make it
1105 available to TLS with the number defined in
1106 draft-ietf-tls-compression-04.txt.
1107 [Richard Levitte]
1108
1109 *) Add the ASN.1 structures and functions for CertificatePair, which
1110 is defined as follows (according to X.509_4thEditionDraftV6.pdf):
1111
1112 CertificatePair ::= SEQUENCE {
1113 forward [0] Certificate OPTIONAL,
1114 reverse [1] Certificate OPTIONAL,
1115 -- at least one of the pair shall be present -- }
1116
1117 Also implement the PEM functions to read and write certificate
1118 pairs, and defined the PEM tag as "CERTIFICATE PAIR".
1119
1120 This needed to be defined, mostly for the sake of the LDAP
1121 attribute crossCertificatePair, but may prove useful elsewhere as
1122 well.
1123 [Richard Levitte]
1124
1125 *) Make it possible to inhibit symlinking of shared libraries in
1126 Makefile.shared, for Cygwin's sake.
1127 [Richard Levitte]
1128
1129 *) Extend the BIGNUM API by creating a function
1130 void BN_set_negative(BIGNUM *a, int neg);
1131 and a macro that behave like
1132 int BN_is_negative(const BIGNUM *a);
1133
1134 to avoid the need to access 'a->neg' directly in applications.
1135 [Nils Larsch]
1136
1137 *) Implement fast modular reduction for pseudo-Mersenne primes
1138 used in NIST curves (crypto/bn/bn_nist.c, crypto/ec/ecp_nist.c).
1139 EC_GROUP_new_curve_GFp() will now automatically use this
1140 if applicable.
1141 [Nils Larsch <nla@trustcenter.de>]
1142
1143 *) Add new lock type (CRYPTO_LOCK_BN).
1144 [Bodo Moeller]
1145
1146 *) Change the ENGINE framework to automatically load engines
1147 dynamically from specific directories unless they could be
1148 found to already be built in or loaded. Move all the
1149 current engines except for the cryptodev one to a new
1150 directory engines/.
1151 The engines in engines/ are built as shared libraries if
1152 the "shared" options was given to ./Configure or ./config.
1153 Otherwise, they are inserted in libcrypto.a.
1154 /usr/local/ssl/engines is the default directory for dynamic
1155 engines, but that can be overriden at configure time through
1156 the usual use of --prefix and/or --openssldir, and at run
1157 time with the environment variable OPENSSL_ENGINES.
1158 [Geoff Thorpe and Richard Levitte]
1159
1160 *) Add Makefile.shared, a helper makefile to build shared
1161 libraries. Addapt Makefile.org.
1162 [Richard Levitte]
1163
1164 *) Add version info to Win32 DLLs.
1165 [Peter 'Luna' Runestig" <peter@runestig.com>]
1166
1167 *) Add new 'medium level' PKCS#12 API. Certificates and keys
1168 can be added using this API to created arbitrary PKCS#12
1169 files while avoiding the low level API.
1170
1171 New options to PKCS12_create(), key or cert can be NULL and
1172 will then be omitted from the output file. The encryption
1173 algorithm NIDs can be set to -1 for no encryption, the mac
1174 iteration count can be set to 0 to omit the mac.
1175
1176 Enhance pkcs12 utility by making the -nokeys and -nocerts
1177 options work when creating a PKCS#12 file. New option -nomac
1178 to omit the mac, NONE can be set for an encryption algorithm.
1179 New code is modified to use the enhanced PKCS12_create()
1180 instead of the low level API.
1181 [Steve Henson]
1182
1183 *) Extend ASN1 encoder to support indefinite length constructed
1184 encoding. This can output sequences tags and octet strings in
1185 this form. Modify pk7_asn1.c to support indefinite length
1186 encoding. This is experimental and needs additional code to
1187 be useful, such as an ASN1 bio and some enhanced streaming
1188 PKCS#7 code.
1189
1190 Extend template encode functionality so that tagging is passed
1191 down to the template encoder.
1192 [Steve Henson]
1193
1194 *) Let 'openssl req' fail if an argument to '-newkey' is not
1195 recognized instead of using RSA as a default.
1196 [Bodo Moeller]
1197
1198 *) Add support for ECC-based ciphersuites from draft-ietf-tls-ecc-01.txt.
1199 As these are not official, they are not included in "ALL";
1200 the "ECCdraft" ciphersuite group alias can be used to select them.
1201 [Vipul Gupta and Sumit Gupta (Sun Microsystems Laboratories)]
1202
1203 *) Add ECDH engine support.
1204 [Nils Gura and Douglas Stebila (Sun Microsystems Laboratories)]
1205
1206 *) Add ECDH in new directory crypto/ecdh/.
1207 [Douglas Stebila (Sun Microsystems Laboratories)]
1208
1209 *) Let BN_rand_range() abort with an error after 100 iterations
1210 without success (which indicates a broken PRNG).
1211 [Bodo Moeller]
1212
1213 *) Change BN_mod_sqrt() so that it verifies that the input value
1214 is really the square of the return value. (Previously,
1215 BN_mod_sqrt would show GIGO behaviour.)
1216 [Bodo Moeller]
1217
1218 *) Add named elliptic curves over binary fields from X9.62, SECG,
1219 and WAP/WTLS; add OIDs that were still missing.
1220
1221 [Sheueling Chang Shantz and Douglas Stebila
1222 (Sun Microsystems Laboratories)]
1223
1224 *) Extend the EC library for elliptic curves over binary fields
1225 (new files ec2_smpl.c, ec2_smpt.c, ec2_mult.c in crypto/ec/).
1226 New EC_METHOD:
1227
1228 EC_GF2m_simple_method
1229
1230 New API functions:
1231
1232 EC_GROUP_new_curve_GF2m
1233 EC_GROUP_set_curve_GF2m
1234 EC_GROUP_get_curve_GF2m
1235 EC_POINT_set_affine_coordinates_GF2m
1236 EC_POINT_get_affine_coordinates_GF2m
1237 EC_POINT_set_compressed_coordinates_GF2m
1238
1239 Point compression for binary fields is disabled by default for
1240 patent reasons (compile with OPENSSL_EC_BIN_PT_COMP defined to
1241 enable it).
1242
1243 As binary polynomials are represented as BIGNUMs, various members
1244 of the EC_GROUP and EC_POINT data structures can be shared
1245 between the implementations for prime fields and binary fields;
1246 the above ..._GF2m functions (except for EX_GROUP_new_curve_GF2m)
1247 are essentially identical to their ..._GFp counterparts.
1248 (For simplicity, the '..._GFp' prefix has been dropped from
1249 various internal method names.)
1250
1251 An internal 'field_div' method (similar to 'field_mul' and
1252 'field_sqr') has been added; this is used only for binary fields.
1253
1254 [Sheueling Chang Shantz and Douglas Stebila
1255 (Sun Microsystems Laboratories)]
1256
1257 *) Optionally dispatch EC_POINT_mul(), EC_POINT_precompute_mult()
1258 through methods ('mul', 'precompute_mult').
1259
1260 The generic implementations (now internally called 'ec_wNAF_mul'
1261 and 'ec_wNAF_precomputed_mult') remain the default if these
1262 methods are undefined.
1263
1264 [Sheueling Chang Shantz and Douglas Stebila
1265 (Sun Microsystems Laboratories)]
1266
1267 *) New function EC_GROUP_get_degree, which is defined through
1268 EC_METHOD. For curves over prime fields, this returns the bit
1269 length of the modulus.
1270
1271 [Sheueling Chang Shantz and Douglas Stebila
1272 (Sun Microsystems Laboratories)]
1273
1274 *) New functions EC_GROUP_dup, EC_POINT_dup.
1275 (These simply call ..._new and ..._copy).
1276
1277 [Sheueling Chang Shantz and Douglas Stebila
1278 (Sun Microsystems Laboratories)]
1279
1280 *) Add binary polynomial arithmetic software in crypto/bn/bn_gf2m.c.
1281 Polynomials are represented as BIGNUMs (where the sign bit is not
1282 used) in the following functions [macros]:
1283
1284 BN_GF2m_add
1285 BN_GF2m_sub [= BN_GF2m_add]
1286 BN_GF2m_mod [wrapper for BN_GF2m_mod_arr]
1287 BN_GF2m_mod_mul [wrapper for BN_GF2m_mod_mul_arr]
1288 BN_GF2m_mod_sqr [wrapper for BN_GF2m_mod_sqr_arr]
1289 BN_GF2m_mod_inv
1290 BN_GF2m_mod_exp [wrapper for BN_GF2m_mod_exp_arr]
1291 BN_GF2m_mod_sqrt [wrapper for BN_GF2m_mod_sqrt_arr]
1292 BN_GF2m_mod_solve_quad [wrapper for BN_GF2m_mod_solve_quad_arr]
1293 BN_GF2m_cmp [= BN_ucmp]
1294
1295 (Note that only the 'mod' functions are actually for fields GF(2^m).
1296 BN_GF2m_add() is misnomer, but this is for the sake of consistency.)
1297
1298 For some functions, an the irreducible polynomial defining a
1299 field can be given as an 'unsigned int[]' with strictly
1300 decreasing elements giving the indices of those bits that are set;
1301 i.e., p[] represents the polynomial
1302 f(t) = t^p[0] + t^p[1] + ... + t^p[k]
1303 where
1304 p[0] > p[1] > ... > p[k] = 0.
1305 This applies to the following functions:
1306
1307 BN_GF2m_mod_arr
1308 BN_GF2m_mod_mul_arr
1309 BN_GF2m_mod_sqr_arr
1310 BN_GF2m_mod_inv_arr [wrapper for BN_GF2m_mod_inv]
1311 BN_GF2m_mod_div_arr [wrapper for BN_GF2m_mod_div]
1312 BN_GF2m_mod_exp_arr
1313 BN_GF2m_mod_sqrt_arr
1314 BN_GF2m_mod_solve_quad_arr
1315 BN_GF2m_poly2arr
1316 BN_GF2m_arr2poly
1317
1318 Conversion can be performed by the following functions:
1319
1320 BN_GF2m_poly2arr
1321 BN_GF2m_arr2poly
1322
1323 bntest.c has additional tests for binary polynomial arithmetic.
1324
1325 Two implementations for BN_GF2m_mod_div() are available.
1326 The default algorithm simply uses BN_GF2m_mod_inv() and
1327 BN_GF2m_mod_mul(). The alternative algorithm is compiled in only
1328 if OPENSSL_SUN_GF2M_DIV is defined (patent pending; read the
1329 copyright notice in crypto/bn/bn_gf2m.c before enabling it).
1330
1331 [Sheueling Chang Shantz and Douglas Stebila
1332 (Sun Microsystems Laboratories)]
1333
1334 *) Add new error code 'ERR_R_DISABLED' that can be used when some
1335 functionality is disabled at compile-time.
1336 [Douglas Stebila <douglas.stebila@sun.com>]
1337
1338 *) Change default behaviour of 'openssl asn1parse' so that more
1339 information is visible when viewing, e.g., a certificate:
1340
1341 Modify asn1_parse2 (crypto/asn1/asn1_par.c) so that in non-'dump'
1342 mode the content of non-printable OCTET STRINGs is output in a
1343 style similar to INTEGERs, but with '[HEX DUMP]' prepended to
1344 avoid the appearance of a printable string.
1345 [Nils Larsch <nla@trustcenter.de>]
1346
1347 *) Add 'asn1_flag' and 'asn1_form' member to EC_GROUP with access
1348 functions
1349 EC_GROUP_set_asn1_flag()
1350 EC_GROUP_get_asn1_flag()
1351 EC_GROUP_set_point_conversion_form()
1352 EC_GROUP_get_point_conversion_form()
1353 These control ASN1 encoding details:
1354 - Curves (i.e., groups) are encoded explicitly unless asn1_flag
1355 has been set to OPENSSL_EC_NAMED_CURVE.
1356 - Points are encoded in uncompressed form by default; options for
1357 asn1_for are as for point2oct, namely
1358 POINT_CONVERSION_COMPRESSED
1359 POINT_CONVERSION_UNCOMPRESSED
1360 POINT_CONVERSION_HYBRID
1361
1362 Also add 'seed' and 'seed_len' members to EC_GROUP with access
1363 functions
1364 EC_GROUP_set_seed()
1365 EC_GROUP_get0_seed()
1366 EC_GROUP_get_seed_len()
1367 This is used only for ASN1 purposes (so far).
1368 [Nils Larsch <nla@trustcenter.de>]
1369
1370 *) Add 'field_type' member to EC_METHOD, which holds the NID
1371 of the appropriate field type OID. The new function
1372 EC_METHOD_get_field_type() returns this value.
1373 [Nils Larsch <nla@trustcenter.de>]
1374
1375 *) Add functions
1376 EC_POINT_point2bn()
1377 EC_POINT_bn2point()
1378 EC_POINT_point2hex()
1379 EC_POINT_hex2point()
1380 providing useful interfaces to EC_POINT_point2oct() and
1381 EC_POINT_oct2point().
1382 [Nils Larsch <nla@trustcenter.de>]
1383
1384 *) Change internals of the EC library so that the functions
1385 EC_GROUP_set_generator()
1386 EC_GROUP_get_generator()
1387 EC_GROUP_get_order()
1388 EC_GROUP_get_cofactor()
1389 are implemented directly in crypto/ec/ec_lib.c and not dispatched
1390 to methods, which would lead to unnecessary code duplication when
1391 adding different types of curves.
1392 [Nils Larsch <nla@trustcenter.de> with input by Bodo Moeller]
1393
1394 *) Implement compute_wNAF (crypto/ec/ec_mult.c) without BIGNUM
1395 arithmetic, and such that modified wNAFs are generated
1396 (which avoid length expansion in many cases).
1397 [Bodo Moeller]
1398
1399 *) Add a function EC_GROUP_check_discriminant() (defined via
1400 EC_METHOD) that verifies that the curve discriminant is non-zero.
1401
1402 Add a function EC_GROUP_check() that makes some sanity tests
1403 on a EC_GROUP, its generator and order. This includes
1404 EC_GROUP_check_discriminant().
1405 [Nils Larsch <nla@trustcenter.de>]
1406
1407 *) Add ECDSA in new directory crypto/ecdsa/.
1408
1409 Add applications 'openssl ecparam' and 'openssl ecdsa'
1410 (these are based on 'openssl dsaparam' and 'openssl dsa').
1411
1412 ECDSA support is also included in various other files across the
1413 library. Most notably,
1414 - 'openssl req' now has a '-newkey ecdsa:file' option;
1415 - EVP_PKCS82PKEY (crypto/evp/evp_pkey.c) now can handle ECDSA;
1416 - X509_PUBKEY_get (crypto/asn1/x_pubkey.c) and
1417 d2i_PublicKey (crypto/asn1/d2i_pu.c) have been modified to make
1418 them suitable for ECDSA where domain parameters must be
1419 extracted before the specific public key;
1420 - ECDSA engine support has been added.
1421 [Nils Larsch <nla@trustcenter.de>]
1422
1423 *) Include some named elliptic curves, and add OIDs from X9.62,
1424 SECG, and WAP/WTLS. Each curve can be obtained from the new
1425 function
1426 EC_GROUP_new_by_curve_name(),
1427 and the list of available named curves can be obtained with
1428 EC_get_builtin_curves().
1429 Also add a 'curve_name' member to EC_GROUP objects, which can be
1430 accessed via
1431 EC_GROUP_set_curve_name()
1432 EC_GROUP_get_curve_name()
1433 [Nils Larsch <larsch@trustcenter.de, Bodo Moeller]
1434
1435 *) Remove a few calls to bn_wexpand() in BN_sqr() (the one in there
1436 was actually never needed) and in BN_mul(). The removal in BN_mul()
1437 required a small change in bn_mul_part_recursive() and the addition
1438 of the functions bn_cmp_part_words(), bn_sub_part_words() and
1439 bn_add_part_words(), which do the same thing as bn_cmp_words(),
1440 bn_sub_words() and bn_add_words() except they take arrays with
1441 differing sizes.
1442 [Richard Levitte]
1443
1444 Changes between 0.9.7m and 0.9.7n [xx XXX xxxx]
1445
1446 *) In the SSL/TLS server implementation, be strict about session ID
1447 context matching (which matters if an application uses a single
1448 external cache for different purposes). Previously,
1449 out-of-context reuse was forbidden only if SSL_VERIFY_PEER was
1450 set. This did ensure strict client verification, but meant that,
1451 with applications using a single external cache for quite
1452 different requirements, clients could circumvent ciphersuite
1453 restrictions for a given session ID context by starting a session
1454 in a different context.
1455 [Bodo Moeller]
1456
1457 Changes between 0.9.7l and 0.9.7m [23 Feb 2007]
1458
1459 *) Cleanse PEM buffers before freeing them since they may contain
1460 sensitive data.
1461 [Benjamin Bennett <ben@psc.edu>]
1462
1463 *) Include "!eNULL" in SSL_DEFAULT_CIPHER_LIST to make sure that
1464 a ciphersuite string such as "DEFAULT:RSA" cannot enable
1465 authentication-only ciphersuites.
1466 [Bodo Moeller]
1467
1468 *) Since AES128 and AES256 share a single mask bit in the logic of
1469 ssl/ssl_ciph.c, the code for masking out disabled ciphers needs a
1470 kludge to work properly if AES128 is available and AES256 isn't.
1471 [Victor Duchovni]
1472
1473 *) Expand security boundary to match 1.1.1 module.
1474 [Steve Henson]
1475
1476 *) Remove redundant features: hash file source, editing of test vectors
1477 modify fipsld to use external fips_premain.c signature.
1478 [Steve Henson]
1479
1480 *) New perl script mkfipsscr.pl to create shell scripts or batch files to
1481 run algorithm test programs.
1482 [Steve Henson]
1483
1484 *) Make algorithm test programs more tolerant of whitespace.
1485 [Steve Henson]
1486
1487 *) Have SSL/TLS server implementation tolerate "mismatched" record
1488 protocol version while receiving ClientHello even if the
1489 ClientHello is fragmented. (The server can't insist on the
1490 particular protocol version it has chosen before the ServerHello
1491 message has informed the client about his choice.)
1492 [Bodo Moeller]
1493
1494 *) Load error codes if they are not already present instead of using a
1495 static variable. This allows them to be cleanly unloaded and reloaded.
1496 [Steve Henson]
1497
1498 Changes between 0.9.7k and 0.9.7l [28 Sep 2006]
1499
1500 *) Introduce limits to prevent malicious keys being able to
1501 cause a denial of service. (CVE-2006-2940)
1502 [Steve Henson, Bodo Moeller]
1503
1504 *) Fix ASN.1 parsing of certain invalid structures that can result
1505 in a denial of service. (CVE-2006-2937) [Steve Henson]
1506
1507 *) Fix buffer overflow in SSL_get_shared_ciphers() function.
1508 (CVE-2006-3738) [Tavis Ormandy and Will Drewry, Google Security Team]
1509
1510 *) Fix SSL client code which could crash if connecting to a
1511 malicious SSLv2 server. (CVE-2006-4343)
1512 [Tavis Ormandy and Will Drewry, Google Security Team]
1513
1514 *) Change ciphersuite string processing so that an explicit
1515 ciphersuite selects this one ciphersuite (so that "AES256-SHA"
1516 will no longer include "AES128-SHA"), and any other similar
1517 ciphersuite (same bitmap) from *other* protocol versions (so that
1518 "RC4-MD5" will still include both the SSL 2.0 ciphersuite and the
1519 SSL 3.0/TLS 1.0 ciphersuite). This is a backport combining
1520 changes from 0.9.8b and 0.9.8d.
1521 [Bodo Moeller]
1522
1523 Changes between 0.9.7j and 0.9.7k [05 Sep 2006]
1524
1525 *) Avoid PKCS #1 v1.5 signature attack discovered by Daniel Bleichenbacher
1526 (CVE-2006-4339) [Ben Laurie and Google Security Team]
1527
1528 *) Change the Unix randomness entropy gathering to use poll() when
1529 possible instead of select(), since the latter has some
1530 undesirable limitations.
1531 [Darryl Miles via Richard Levitte and Bodo Moeller]
1532
1533 *) Disable rogue ciphersuites:
1534
1535 - SSLv2 0x08 0x00 0x80 ("RC4-64-MD5")
1536 - SSLv3/TLSv1 0x00 0x61 ("EXP1024-RC2-CBC-MD5")
1537 - SSLv3/TLSv1 0x00 0x60 ("EXP1024-RC4-MD5")
1538
1539 The latter two were purportedly from
1540 draft-ietf-tls-56-bit-ciphersuites-0[01].txt, but do not really
1541 appear there.
1542
1543 Also deactive the remaining ciphersuites from
1544 draft-ietf-tls-56-bit-ciphersuites-01.txt. These are just as
1545 unofficial, and the ID has long expired.
1546 [Bodo Moeller]
1547
1548 *) Fix RSA blinding Heisenbug (problems sometimes occured on
1549 dual-core machines) and other potential thread-safety issues.
1550 [Bodo Moeller]
1551
1552 Changes between 0.9.7i and 0.9.7j [04 May 2006]
1553
1554 *) Adapt fipsld and the build system to link against the validated FIPS
1555 module in FIPS mode.
1556 [Steve Henson]
1557
1558 *) Fixes for VC++ 2005 build under Windows.
1559 [Steve Henson]
1560
1561 *) Add new Windows build target VC-32-GMAKE for VC++. This uses GNU make
1562 from a Windows bash shell such as MSYS. It is autodetected from the
1563 "config" script when run from a VC++ environment. Modify standard VC++
1564 build to use fipscanister.o from the GNU make build.
1565 [Steve Henson]
1566
1567 Changes between 0.9.7h and 0.9.7i [14 Oct 2005]
1568
1569 *) Wrapped the definition of EVP_MAX_MD_SIZE in a #ifdef OPENSSL_FIPS.
1570 The value now differs depending on if you build for FIPS or not.
1571 BEWARE! A program linked with a shared FIPSed libcrypto can't be
1572 safely run with a non-FIPSed libcrypto, as it may crash because of
1573 the difference induced by this change.
1574 [Andy Polyakov]
1575
1576 Changes between 0.9.7g and 0.9.7h [11 Oct 2005]
1577
1578 *) Remove the functionality of SSL_OP_MSIE_SSLV2_RSA_PADDING
1579 (part of SSL_OP_ALL). This option used to disable the
1580 countermeasure against man-in-the-middle protocol-version
1581 rollback in the SSL 2.0 server implementation, which is a bad
1582 idea. (CVE-2005-2969)
1583
1584 [Bodo Moeller; problem pointed out by Yutaka Oiwa (Research Center
1585 for Information Security, National Institute of Advanced Industrial
1586 Science and Technology [AIST], Japan)]
1587
1588 *) Minimal support for X9.31 signatures and PSS padding modes. This is
1589 mainly for FIPS compliance and not fully integrated at this stage.
1590 [Steve Henson]
1591
1592 *) For DSA signing, unless DSA_FLAG_NO_EXP_CONSTTIME is set, perform
1593 the exponentiation using a fixed-length exponent. (Otherwise,
1594 the information leaked through timing could expose the secret key
1595 after many signatures; cf. Bleichenbacher's attack on DSA with
1596 biased k.)
1597 [Bodo Moeller]
1598
1599 *) Make a new fixed-window mod_exp implementation the default for
1600 RSA, DSA, and DH private-key operations so that the sequence of
1601 squares and multiplies and the memory access pattern are
1602 independent of the particular secret key. This will mitigate
1603 cache-timing and potential related attacks.
1604
1605 BN_mod_exp_mont_consttime() is the new exponentiation implementation,
1606 and this is automatically used by BN_mod_exp_mont() if the new flag
1607 BN_FLG_EXP_CONSTTIME is set for the exponent. RSA, DSA, and DH
1608 will use this BN flag for private exponents unless the flag
1609 RSA_FLAG_NO_EXP_CONSTTIME, DSA_FLAG_NO_EXP_CONSTTIME, or
1610 DH_FLAG_NO_EXP_CONSTTIME, respectively, is set.
1611
1612 [Matthew D Wood (Intel Corp), with some changes by Bodo Moeller]
1613
1614 *) Change the client implementation for SSLv23_method() and
1615 SSLv23_client_method() so that is uses the SSL 3.0/TLS 1.0
1616 Client Hello message format if the SSL_OP_NO_SSLv2 option is set.
1617 (Previously, the SSL 2.0 backwards compatible Client Hello
1618 message format would be used even with SSL_OP_NO_SSLv2.)
1619 [Bodo Moeller]
1620
1621 *) Add support for smime-type MIME parameter in S/MIME messages which some
1622 clients need.
1623 [Steve Henson]
1624
1625 *) New function BN_MONT_CTX_set_locked() to set montgomery parameters in
1626 a threadsafe manner. Modify rsa code to use new function and add calls
1627 to dsa and dh code (which had race conditions before).
1628 [Steve Henson]
1629
1630 *) Include the fixed error library code in the C error file definitions
1631 instead of fixing them up at runtime. This keeps the error code
1632 structures constant.
1633 [Steve Henson]
1634
1635 Changes between 0.9.7f and 0.9.7g [11 Apr 2005]
1636
1637 [NB: OpenSSL 0.9.7h and later 0.9.7 patch levels were released after
1638 OpenSSL 0.9.8.]
1639
1640 *) Fixes for newer kerberos headers. NB: the casts are needed because
1641 the 'length' field is signed on one version and unsigned on another
1642 with no (?) obvious way to tell the difference, without these VC++
1643 complains. Also the "definition" of FAR (blank) is no longer included
1644 nor is the error ENOMEM. KRB5_PRIVATE has to be set to 1 to pick up
1645 some needed definitions.
1646 [Steve Henson]
1647
1648 *) Undo Cygwin change.
1649 [Ulf Möller]
1650
1651 *) Added support for proxy certificates according to RFC 3820.
1652 Because they may be a security thread to unaware applications,
1653 they must be explicitely allowed in run-time. See
1654 docs/HOWTO/proxy_certificates.txt for further information.
1655 [Richard Levitte]
1656
1657 Changes between 0.9.7e and 0.9.7f [22 Mar 2005]
1658
1659 *) Use (SSL_RANDOM_VALUE - 4) bytes of pseudo random data when generating
1660 server and client random values. Previously
1661 (SSL_RANDOM_VALUE - sizeof(time_t)) would be used which would result in
1662 less random data when sizeof(time_t) > 4 (some 64 bit platforms).
1663
1664 This change has negligible security impact because:
1665
1666 1. Server and client random values still have 24 bytes of pseudo random
1667 data.
1668
1669 2. Server and client random values are sent in the clear in the initial
1670 handshake.
1671
1672 3. The master secret is derived using the premaster secret (48 bytes in
1673 size for static RSA ciphersuites) as well as client server and random
1674 values.
1675
1676 The OpenSSL team would like to thank the UK NISCC for bringing this issue
1677 to our attention.
1678
1679 [Stephen Henson, reported by UK NISCC]
1680
1681 *) Use Windows randomness collection on Cygwin.
1682 [Ulf Möller]
1683
1684 *) Fix hang in EGD/PRNGD query when communication socket is closed
1685 prematurely by EGD/PRNGD.
1686 [Darren Tucker <dtucker@zip.com.au> via Lutz Jänicke, resolves #1014]
1687
1688 *) Prompt for pass phrases when appropriate for PKCS12 input format.
1689 [Steve Henson]
1690
1691 *) Back-port of selected performance improvements from development
1692 branch, as well as improved support for PowerPC platforms.
1693 [Andy Polyakov]
1694
1695 *) Add lots of checks for memory allocation failure, error codes to indicate
1696 failure and freeing up memory if a failure occurs.
1697 [Nauticus Networks SSL Team <openssl@nauticusnet.com>, Steve Henson]
1698
1699 *) Add new -passin argument to dgst.
1700 [Steve Henson]
1701
1702 *) Perform some character comparisons of different types in X509_NAME_cmp:
1703 this is needed for some certificates that reencode DNs into UTF8Strings
1704 (in violation of RFC3280) and can't or wont issue name rollover
1705 certificates.
1706 [Steve Henson]
1707
1708 *) Make an explicit check during certificate validation to see that
1709 the CA setting in each certificate on the chain is correct. As a
1710 side effect always do the following basic checks on extensions,
1711 not just when there's an associated purpose to the check:
1712
1713 - if there is an unhandled critical extension (unless the user
1714 has chosen to ignore this fault)
1715 - if the path length has been exceeded (if one is set at all)
1716 - that certain extensions fit the associated purpose (if one has
1717 been given)
1718 [Richard Levitte]
1719
1720 Changes between 0.9.7d and 0.9.7e [25 Oct 2004]
1721
1722 *) Avoid a race condition when CRLs are checked in a multi threaded
1723 environment. This would happen due to the reordering of the revoked
1724 entries during signature checking and serial number lookup. Now the
1725 encoding is cached and the serial number sort performed under a lock.
1726 Add new STACK function sk_is_sorted().
1727 [Steve Henson]
1728
1729 *) Add Delta CRL to the extension code.
1730 [Steve Henson]
1731
1732 *) Various fixes to s3_pkt.c so alerts are sent properly.
1733 [David Holmes <d.holmes@f5.com>]
1734
1735 *) Reduce the chances of duplicate issuer name and serial numbers (in
1736 violation of RFC3280) using the OpenSSL certificate creation utilities.
1737 This is done by creating a random 64 bit value for the initial serial
1738 number when a serial number file is created or when a self signed
1739 certificate is created using 'openssl req -x509'. The initial serial
1740 number file is created using 'openssl x509 -next_serial' in CA.pl
1741 rather than being initialized to 1.
1742 [Steve Henson]
1743
1744 Changes between 0.9.7c and 0.9.7d [17 Mar 2004]
1745
1746 *) Fix null-pointer assignment in do_change_cipher_spec() revealed
1747 by using the Codenomicon TLS Test Tool (CVE-2004-0079)
1748 [Joe Orton, Steve Henson]
1749
1750 *) Fix flaw in SSL/TLS handshaking when using Kerberos ciphersuites
1751 (CVE-2004-0112)
1752 [Joe Orton, Steve Henson]
1753
1754 *) Make it possible to have multiple active certificates with the same
1755 subject in the CA index file. This is done only if the keyword
1756 'unique_subject' is set to 'no' in the main CA section (default
1757 if 'CA_default') of the configuration file. The value is saved
1758 with the database itself in a separate index attribute file,
1759 named like the index file with '.attr' appended to the name.
1760 [Richard Levitte]
1761
1762 *) X509 verify fixes. Disable broken certificate workarounds when
1763 X509_V_FLAGS_X509_STRICT is set. Check CRL issuer has cRLSign set if
1764 keyUsage extension present. Don't accept CRLs with unhandled critical
1765 extensions: since verify currently doesn't process CRL extensions this
1766 rejects a CRL with *any* critical extensions. Add new verify error codes
1767 for these cases.
1768 [Steve Henson]
1769
1770 *) When creating an OCSP nonce use an OCTET STRING inside the extnValue.
1771 A clarification of RFC2560 will require the use of OCTET STRINGs and
1772 some implementations cannot handle the current raw format. Since OpenSSL
1773 copies and compares OCSP nonces as opaque blobs without any attempt at
1774 parsing them this should not create any compatibility issues.
1775 [Steve Henson]
1776
1777 *) New md flag EVP_MD_CTX_FLAG_REUSE this allows md_data to be reused when
1778 calling EVP_MD_CTX_copy_ex() to avoid calling OPENSSL_malloc(). Without
1779 this HMAC (and other) operations are several times slower than OpenSSL
1780 < 0.9.7.
1781 [Steve Henson]
1782
1783 *) Print out GeneralizedTime and UTCTime in ASN1_STRING_print_ex().
1784 [Peter Sylvester <Peter.Sylvester@EdelWeb.fr>]
1785
1786 *) Use the correct content when signing type "other".
1787 [Steve Henson]
1788
1789 Changes between 0.9.7b and 0.9.7c [30 Sep 2003]
1790
1791 *) Fix various bugs revealed by running the NISCC test suite:
1792
1793 Stop out of bounds reads in the ASN1 code when presented with
1794 invalid tags (CVE-2003-0543 and CVE-2003-0544).
1795
1796 Free up ASN1_TYPE correctly if ANY type is invalid (CVE-2003-0545).
1797
1798 If verify callback ignores invalid public key errors don't try to check
1799 certificate signature with the NULL public key.
1800
1801 [Steve Henson]
1802
1803 *) New -ignore_err option in ocsp application to stop the server
1804 exiting on the first error in a request.
1805 [Steve Henson]
1806
1807 *) In ssl3_accept() (ssl/s3_srvr.c) only accept a client certificate
1808 if the server requested one: as stated in TLS 1.0 and SSL 3.0
1809 specifications.
1810 [Steve Henson]
1811
1812 *) In ssl3_get_client_hello() (ssl/s3_srvr.c), tolerate additional
1813 extra data after the compression methods not only for TLS 1.0
1814 but also for SSL 3.0 (as required by the specification).
1815 [Bodo Moeller; problem pointed out by Matthias Loepfe]
1816
1817 *) Change X509_certificate_type() to mark the key as exported/exportable
1818 when it's 512 *bits* long, not 512 bytes.
1819 [Richard Levitte]
1820
1821 *) Change AES_cbc_encrypt() so it outputs exact multiple of
1822 blocks during encryption.
1823 [Richard Levitte]
1824
1825 *) Various fixes to base64 BIO and non blocking I/O. On write
1826 flushes were not handled properly if the BIO retried. On read
1827 data was not being buffered properly and had various logic bugs.
1828 This also affects blocking I/O when the data being decoded is a
1829 certain size.
1830 [Steve Henson]
1831
1832 *) Various S/MIME bugfixes and compatibility changes:
1833 output correct application/pkcs7 MIME type if
1834 PKCS7_NOOLDMIMETYPE is set. Tolerate some broken signatures.
1835 Output CR+LF for EOL if PKCS7_CRLFEOL is set (this makes opening
1836 of files as .eml work). Correctly handle very long lines in MIME
1837 parser.
1838 [Steve Henson]
1839
1840 Changes between 0.9.7a and 0.9.7b [10 Apr 2003]
1841
1842 *) Countermeasure against the Klima-Pokorny-Rosa extension of
1843 Bleichbacher's attack on PKCS #1 v1.5 padding: treat
1844 a protocol version number mismatch like a decryption error
1845 in ssl3_get_client_key_exchange (ssl/s3_srvr.c).
1846 [Bodo Moeller]
1847
1848 *) Turn on RSA blinding by default in the default implementation
1849 to avoid a timing attack. Applications that don't want it can call
1850 RSA_blinding_off() or use the new flag RSA_FLAG_NO_BLINDING.
1851 They would be ill-advised to do so in most cases.
1852 [Ben Laurie, Steve Henson, Geoff Thorpe, Bodo Moeller]
1853
1854 *) Change RSA blinding code so that it works when the PRNG is not
1855 seeded (in this case, the secret RSA exponent is abused as
1856 an unpredictable seed -- if it is not unpredictable, there
1857 is no point in blinding anyway). Make RSA blinding thread-safe
1858 by remembering the creator's thread ID in rsa->blinding and
1859 having all other threads use local one-time blinding factors
1860 (this requires more computation than sharing rsa->blinding, but
1861 avoids excessive locking; and if an RSA object is not shared
1862 between threads, blinding will still be very fast).
1863 [Bodo Moeller]
1864
1865 *) Fixed a typo bug that would cause ENGINE_set_default() to set an
1866 ENGINE as defaults for all supported algorithms irrespective of
1867 the 'flags' parameter. 'flags' is now honoured, so applications
1868 should make sure they are passing it correctly.
1869 [Geoff Thorpe]
1870
1871 *) Target "mingw" now allows native Windows code to be generated in
1872 the Cygwin environment as well as with the MinGW compiler.
1873 [Ulf Moeller]
1874
1875 Changes between 0.9.7 and 0.9.7a [19 Feb 2003]
1876
1877 *) In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked
1878 via timing by performing a MAC computation even if incorrrect
1879 block cipher padding has been found. This is a countermeasure
1880 against active attacks where the attacker has to distinguish
1881 between bad padding and a MAC verification error. (CVE-2003-0078)
1882
1883 [Bodo Moeller; problem pointed out by Brice Canvel (EPFL),
1884 Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and
1885 Martin Vuagnoux (EPFL, Ilion)]
1886
1887 *) Make the no-err option work as intended. The intention with no-err
1888 is not to have the whole error stack handling routines removed from
1889 libcrypto, it's only intended to remove all the function name and
1890 reason texts, thereby removing some of the footprint that may not
1891 be interesting if those errors aren't displayed anyway.
1892
1893 NOTE: it's still possible for any application or module to have it's
1894 own set of error texts inserted. The routines are there, just not
1895 used by default when no-err is given.
1896 [Richard Levitte]
1897
1898 *) Add support for FreeBSD on IA64.
1899 [dirk.meyer@dinoex.sub.org via Richard Levitte, resolves #454]
1900
1901 *) Adjust DES_cbc_cksum() so it returns the same value as the MIT
1902 Kerberos function mit_des_cbc_cksum(). Before this change,
1903 the value returned by DES_cbc_cksum() was like the one from
1904 mit_des_cbc_cksum(), except the bytes were swapped.
1905 [Kevin Greaney <Kevin.Greaney@hp.com> and Richard Levitte]
1906
1907 *) Allow an application to disable the automatic SSL chain building.
1908 Before this a rather primitive chain build was always performed in
1909 ssl3_output_cert_chain(): an application had no way to send the
1910 correct chain if the automatic operation produced an incorrect result.
1911
1912 Now the chain builder is disabled if either:
1913
1914 1. Extra certificates are added via SSL_CTX_add_extra_chain_cert().
1915
1916 2. The mode flag SSL_MODE_NO_AUTO_CHAIN is set.
1917
1918 The reasoning behind this is that an application would not want the
1919 auto chain building to take place if extra chain certificates are
1920 present and it might also want a means of sending no additional
1921 certificates (for example the chain has two certificates and the
1922 root is omitted).
1923 [Steve Henson]
1924
1925 *) Add the possibility to build without the ENGINE framework.
1926 [Steven Reddie <smr@essemer.com.au> via Richard Levitte]
1927
1928 *) Under Win32 gmtime() can return NULL: check return value in
1929 OPENSSL_gmtime(). Add error code for case where gmtime() fails.
1930 [Steve Henson]
1931
1932 *) DSA routines: under certain error conditions uninitialized BN objects
1933 could be freed. Solution: make sure initialization is performed early
1934 enough. (Reported and fix supplied by Ivan D Nestlerode <nestler@MIT.EDU>,
1935 Nils Larsch <nla@trustcenter.de> via PR#459)
1936 [Lutz Jaenicke]
1937
1938 *) Another fix for SSLv2 session ID handling: the session ID was incorrectly
1939 checked on reconnect on the client side, therefore session resumption
1940 could still fail with a "ssl session id is different" error. This
1941 behaviour is masked when SSL_OP_ALL is used due to
1942 SSL_OP_MICROSOFT_SESS_ID_BUG being set.
1943 Behaviour observed by Crispin Flowerday <crispin@flowerday.cx> as
1944 followup to PR #377.
1945 [Lutz Jaenicke]
1946
1947 *) IA-32 assembler support enhancements: unified ELF targets, support
1948 for SCO/Caldera platforms, fix for Cygwin shared build.
1949 [Andy Polyakov]
1950
1951 *) Add support for FreeBSD on sparc64. As a consequence, support for
1952 FreeBSD on non-x86 processors is separate from x86 processors on
1953 the config script, much like the NetBSD support.
1954 [Richard Levitte & Kris Kennaway <kris@obsecurity.org>]
1955
1956 Changes between 0.9.6h and 0.9.7 [31 Dec 2002]
1957
1958 [NB: OpenSSL 0.9.6i and later 0.9.6 patch levels were released after
1959 OpenSSL 0.9.7.]
1960
1961 *) Fix session ID handling in SSLv2 client code: the SERVER FINISHED
1962 code (06) was taken as the first octet of the session ID and the last
1963 octet was ignored consequently. As a result SSLv2 client side session
1964 caching could not have worked due to the session ID mismatch between
1965 client and server.
1966 Behaviour observed by Crispin Flowerday <crispin@flowerday.cx> as
1967 PR #377.
1968 [Lutz Jaenicke]
1969
1970 *) Change the declaration of needed Kerberos libraries to use EX_LIBS
1971 instead of the special (and badly supported) LIBKRB5. LIBKRB5 is
1972 removed entirely.
1973 [Richard Levitte]
1974
1975 *) The hw_ncipher.c engine requires dynamic locks. Unfortunately, it
1976 seems that in spite of existing for more than a year, many application
1977 author have done nothing to provide the necessary callbacks, which
1978 means that this particular engine will not work properly anywhere.
1979 This is a very unfortunate situation which forces us, in the name
1980 of usability, to give the hw_ncipher.c a static lock, which is part
1981 of libcrypto.
1982 NOTE: This is for the 0.9.7 series ONLY. This hack will never
1983 appear in 0.9.8 or later. We EXPECT application authors to have
1984 dealt properly with this when 0.9.8 is released (unless we actually
1985 make such changes in the libcrypto locking code that changes will
1986 have to be made anyway).
1987 [Richard Levitte]
1988
1989 *) In asn1_d2i_read_bio() repeatedly call BIO_read() until all content
1990 octets have been read, EOF or an error occurs. Without this change
1991 some truncated ASN1 structures will not produce an error.
1992 [Steve Henson]
1993
1994 *) Disable Heimdal support, since it hasn't been fully implemented.
1995 Still give the possibility to force the use of Heimdal, but with
1996 warnings and a request that patches get sent to openssl-dev.
1997 [Richard Levitte]
1998
1999 *) Add the VC-CE target, introduce the WINCE sysname, and add
2000 INSTALL.WCE and appropriate conditionals to make it build.
2001 [Steven Reddie <smr@essemer.com.au> via Richard Levitte]
2002
2003 *) Change the DLL names for Cygwin to cygcrypto-x.y.z.dll and
2004 cygssl-x.y.z.dll, where x, y and z are the major, minor and
2005 edit numbers of the version.
2006 [Corinna Vinschen <vinschen@redhat.com> and Richard Levitte]
2007
2008 *) Introduce safe string copy and catenation functions
2009 (BUF_strlcpy() and BUF_strlcat()).
2010 [Ben Laurie (CHATS) and Richard Levitte]
2011
2012 *) Avoid using fixed-size buffers for one-line DNs.
2013 [Ben Laurie (CHATS)]
2014
2015 *) Add BUF_MEM_grow_clean() to avoid information leakage when
2016 resizing buffers containing secrets, and use where appropriate.
2017 [Ben Laurie (CHATS)]
2018
2019 *) Avoid using fixed size buffers for configuration file location.
2020 [Ben Laurie (CHATS)]
2021
2022 *) Avoid filename truncation for various CA files.
2023 [Ben Laurie (CHATS)]
2024
2025 *) Use sizeof in preference to magic numbers.
2026 [Ben Laurie (CHATS)]
2027
2028 *) Avoid filename truncation in cert requests.
2029 [Ben Laurie (CHATS)]
2030
2031 *) Add assertions to check for (supposedly impossible) buffer
2032 overflows.
2033 [Ben Laurie (CHATS)]
2034
2035 *) Don't cache truncated DNS entries in the local cache (this could
2036 potentially lead to a spoofing attack).
2037 [Ben Laurie (CHATS)]
2038
2039 *) Fix various buffers to be large enough for hex/decimal
2040 representations in a platform independent manner.
2041 [Ben Laurie (CHATS)]
2042
2043 *) Add CRYPTO_realloc_clean() to avoid information leakage when
2044 resizing buffers containing secrets, and use where appropriate.
2045 [Ben Laurie (CHATS)]
2046
2047 *) Add BIO_indent() to avoid much slightly worrying code to do
2048 indents.
2049 [Ben Laurie (CHATS)]
2050
2051 *) Convert sprintf()/BIO_puts() to BIO_printf().
2052 [Ben Laurie (CHATS)]
2053
2054 *) buffer_gets() could terminate with the buffer only half
2055 full. Fixed.
2056 [Ben Laurie (CHATS)]
2057
2058 *) Add assertions to prevent user-supplied crypto functions from
2059 overflowing internal buffers by having large block sizes, etc.
2060 [Ben Laurie (CHATS)]
2061
2062 *) New OPENSSL_assert() macro (similar to assert(), but enabled
2063 unconditionally).
2064 [Ben Laurie (CHATS)]
2065
2066 *) Eliminate unused copy of key in RC4.
2067 [Ben Laurie (CHATS)]
2068
2069 *) Eliminate unused and incorrectly sized buffers for IV in pem.h.
2070 [Ben Laurie (CHATS)]
2071
2072 *) Fix off-by-one error in EGD path.
2073 [Ben Laurie (CHATS)]
2074
2075 *) If RANDFILE path is too long, ignore instead of truncating.
2076 [Ben Laurie (CHATS)]
2077
2078 *) Eliminate unused and incorrectly sized X.509 structure
2079 CBCParameter.
2080 [Ben Laurie (CHATS)]
2081
2082 *) Eliminate unused and dangerous function knumber().
2083 [Ben Laurie (CHATS)]
2084
2085 *) Eliminate unused and dangerous structure, KSSL_ERR.
2086 [Ben Laurie (CHATS)]
2087
2088 *) Protect against overlong session ID context length in an encoded
2089 session object. Since these are local, this does not appear to be
2090 exploitable.
2091 [Ben Laurie (CHATS)]
2092
2093 *) Change from security patch (see 0.9.6e below) that did not affect
2094 the 0.9.6 release series:
2095
2096 Remote buffer overflow in SSL3 protocol - an attacker could
2097 supply an oversized master key in Kerberos-enabled versions.
2098 (CVE-2002-0657)
2099 [Ben Laurie (CHATS)]
2100
2101 *) Change the SSL kerb5 codes to match RFC 2712.
2102 [Richard Levitte]
2103
2104 *) Make -nameopt work fully for req and add -reqopt switch.
2105 [Michael Bell <michael.bell@rz.hu-berlin.de>, Steve Henson]
2106
2107 *) The "block size" for block ciphers in CFB and OFB mode should be 1.
2108 [Steve Henson, reported by Yngve Nysaeter Pettersen <yngve@opera.com>]
2109
2110 *) Make sure tests can be performed even if the corresponding algorithms
2111 have been removed entirely. This was also the last step to make
2112 OpenSSL compilable with DJGPP under all reasonable conditions.
2113 [Richard Levitte, Doug Kaufman <dkaufman@rahul.net>]
2114
2115 *) Add cipher selection rules COMPLEMENTOFALL and COMPLEMENTOFDEFAULT
2116 to allow version independent disabling of normally unselected ciphers,
2117 which may be activated as a side-effect of selecting a single cipher.
2118
2119 (E.g., cipher list string "RSA" enables ciphersuites that are left
2120 out of "ALL" because they do not provide symmetric encryption.
2121 "RSA:!COMPLEMEMENTOFALL" avoids these unsafe ciphersuites.)
2122 [Lutz Jaenicke, Bodo Moeller]
2123
2124 *) Add appropriate support for separate platform-dependent build
2125 directories. The recommended way to make a platform-dependent
2126 build directory is the following (tested on Linux), maybe with
2127 some local tweaks:
2128
2129 # Place yourself outside of the OpenSSL source tree. In
2130 # this example, the environment variable OPENSSL_SOURCE
2131 # is assumed to contain the absolute OpenSSL source directory.
2132 mkdir -p objtree/"`uname -s`-`uname -r`-`uname -m`"
2133 cd objtree/"`uname -s`-`uname -r`-`uname -m`"
2134 (cd $OPENSSL_SOURCE; find . -type f) | while read F; do
2135 mkdir -p `dirname $F`
2136 ln -s $OPENSSL_SOURCE/$F $F
2137 done
2138
2139 To be absolutely sure not to disturb the source tree, a "make clean"
2140 is a good thing. If it isn't successfull, don't worry about it,
2141 it probably means the source directory is very clean.
2142 [Richard Levitte]
2143
2144 *) Make sure any ENGINE control commands make local copies of string
2145 pointers passed to them whenever necessary. Otherwise it is possible
2146 the caller may have overwritten (or deallocated) the original string
2147 data when a later ENGINE operation tries to use the stored values.
2148 [Götz Babin-Ebell <babinebell@trustcenter.de>]
2149
2150 *) Improve diagnostics in file reading and command-line digests.
2151 [Ben Laurie aided and abetted by Solar Designer <solar@openwall.com>]
2152
2153 *) Add AES modes CFB and OFB to the object database. Correct an
2154 error in AES-CFB decryption.
2155 [Richard Levitte]
2156
2157 *) Remove most calls to EVP_CIPHER_CTX_cleanup() in evp_enc.c, this
2158 allows existing EVP_CIPHER_CTX structures to be reused after
2159 calling EVP_*Final(). This behaviour is used by encryption
2160 BIOs and some applications. This has the side effect that
2161 applications must explicitly clean up cipher contexts with
2162 EVP_CIPHER_CTX_cleanup() or they will leak memory.
2163 [Steve Henson]
2164
2165 *) Check the values of dna and dnb in bn_mul_recursive before calling
2166 bn_mul_comba (a non zero value means the a or b arrays do not contain
2167 n2 elements) and fallback to bn_mul_normal if either is not zero.
2168 [Steve Henson]
2169
2170 *) Fix escaping of non-ASCII characters when using the -subj option
2171 of the "openssl req" command line tool. (Robert Joop <joop@fokus.gmd.de>)
2172 [Lutz Jaenicke]
2173
2174 *) Make object definitions compliant to LDAP (RFC2256): SN is the short
2175 form for "surname", serialNumber has no short form.
2176 Use "mail" as the short name for "rfc822Mailbox" according to RFC2798;
2177 therefore remove "mail" short name for "internet 7".
2178 The OID for unique identifiers in X509 certificates is
2179 x500UniqueIdentifier, not uniqueIdentifier.
2180 Some more OID additions. (Michael Bell <michael.bell@rz.hu-berlin.de>)
2181 [Lutz Jaenicke]
2182
2183 *) Add an "init" command to the ENGINE config module and auto initialize
2184 ENGINEs. Without any "init" command the ENGINE will be initialized
2185 after all ctrl commands have been executed on it. If init=1 the
2186 ENGINE is initailized at that point (ctrls before that point are run
2187 on the uninitialized ENGINE and after on the initialized one). If
2188 init=0 then the ENGINE will not be iniatialized at all.
2189 [Steve Henson]
2190
2191 *) Fix the 'app_verify_callback' interface so that the user-defined
2192 argument is actually passed to the callback: In the
2193 SSL_CTX_set_cert_verify_callback() prototype, the callback
2194 declaration has been changed from
2195 int (*cb)()
2196 into
2197 int (*cb)(X509_STORE_CTX *,void *);
2198 in ssl_verify_cert_chain (ssl/ssl_cert.c), the call
2199 i=s->ctx->app_verify_callback(&ctx)
2200 has been changed into
2201 i=s->ctx->app_verify_callback(&ctx, s->ctx->app_verify_arg).
2202
2203 To update applications using SSL_CTX_set_cert_verify_callback(),
2204 a dummy argument can be added to their callback functions.
2205 [D. K. Smetters <smetters@parc.xerox.com>]
2206
2207 *) Added the '4758cca' ENGINE to support IBM 4758 cards.
2208 [Maurice Gittens <maurice@gittens.nl>, touchups by Geoff Thorpe]
2209
2210 *) Add and OPENSSL_LOAD_CONF define which will cause
2211 OpenSSL_add_all_algorithms() to load the openssl.cnf config file.
2212 This allows older applications to transparently support certain
2213 OpenSSL features: such as crypto acceleration and dynamic ENGINE loading.
2214 Two new functions OPENSSL_add_all_algorithms_noconf() which will never
2215 load the config file and OPENSSL_add_all_algorithms_conf() which will
2216 always load it have also been added.
2217 [Steve Henson]
2218
2219 *) Add the OFB, CFB and CTR (all with 128 bit feedback) to AES.
2220 Adjust NIDs and EVP layer.
2221 [Stephen Sprunk <stephen@sprunk.org> and Richard Levitte]
2222
2223 *) Config modules support in openssl utility.
2224
2225 Most commands now load modules from the config file,
2226 though in a few (such as version) this isn't done
2227 because it couldn't be used for anything.
2228
2229 In the case of ca and req the config file used is
2230 the same as the utility itself: that is the -config
2231 command line option can be used to specify an
2232 alternative file.
2233 [Steve Henson]
2234
2235 *) Move default behaviour from OPENSSL_config(). If appname is NULL
2236 use "openssl_conf" if filename is NULL use default openssl config file.
2237 [Steve Henson]
2238
2239 *) Add an argument to OPENSSL_config() to allow the use of an alternative
2240 config section name. Add a new flag to tolerate a missing config file
2241 and move code to CONF_modules_load_file().
2242 [Steve Henson]
2243
2244 *) Support for crypto accelerator cards from Accelerated Encryption
2245 Processing, www.aep.ie. (Use engine 'aep')
2246 The support was copied from 0.9.6c [engine] and adapted/corrected
2247 to work with the new engine framework.
2248 [AEP Inc. and Richard Levitte]
2249
2250 *) Support for SureWare crypto accelerator cards from Baltimore
2251 Technologies. (Use engine 'sureware')
2252 The support was copied from 0.9.6c [engine] and adapted
2253 to work with the new engine framework.
2254 [Richard Levitte]
2255
2256 *) Have the CHIL engine fork-safe (as defined by nCipher) and actually
2257 make the newer ENGINE framework commands for the CHIL engine work.
2258 [Toomas Kiisk <vix@cyber.ee> and Richard Levitte]
2259
2260 *) Make it possible to produce shared libraries on ReliantUNIX.
2261 [Robert Dahlem <Robert.Dahlem@ffm2.siemens.de> via Richard Levitte]
2262
2263 *) Add the configuration target debug-linux-ppro.
2264 Make 'openssl rsa' use the general key loading routines
2265 implemented in apps.c, and make those routines able to
2266 handle the key format FORMAT_NETSCAPE and the variant
2267 FORMAT_IISSGC.
2268 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte]
2269
2270 *) Fix a crashbug and a logic bug in hwcrhk_load_pubkey().
2271 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte]
2272
2273 *) Add -keyform to rsautl, and document -engine.
2274 [Richard Levitte, inspired by Toomas Kiisk <vix@cyber.ee>]
2275
2276 *) Change BIO_new_file (crypto/bio/bss_file.c) to use new
2277 BIO_R_NO_SUCH_FILE error code rather than the generic
2278 ERR_R_SYS_LIB error code if fopen() fails with ENOENT.
2279 [Ben Laurie]
2280
2281 *) Add new functions
2282 ERR_peek_last_error
2283 ERR_peek_last_error_line
2284 ERR_peek_last_error_line_data.
2285 These are similar to
2286 ERR_peek_error
2287 ERR_peek_error_line
2288 ERR_peek_error_line_data,
2289 but report on the latest error recorded rather than the first one
2290 still in the error queue.
2291 [Ben Laurie, Bodo Moeller]
2292
2293 *) default_algorithms option in ENGINE config module. This allows things
2294 like:
2295 default_algorithms = ALL
2296 default_algorithms = RSA, DSA, RAND, CIPHERS, DIGESTS
2297 [Steve Henson]
2298
2299 *) Prelminary ENGINE config module.
2300 [Steve Henson]
2301
2302 *) New experimental application configuration code.
2303 [Steve Henson]
2304
2305 *) Change the AES code to follow the same name structure as all other
2306 symmetric ciphers, and behave the same way. Move everything to
2307 the directory crypto/aes, thereby obsoleting crypto/rijndael.
2308 [Stephen Sprunk <stephen@sprunk.org> and Richard Levitte]
2309
2310 *) SECURITY: remove unsafe setjmp/signal interaction from ui_openssl.c.
2311 [Ben Laurie and Theo de Raadt]
2312
2313 *) Add option to output public keys in req command.
2314 [Massimiliano Pala madwolf@openca.org]
2315
2316 *) Use wNAFs in EC_POINTs_mul() for improved efficiency
2317 (up to about 10% better than before for P-192 and P-224).
2318 [Bodo Moeller]
2319
2320 *) New functions/macros
2321
2322 SSL_CTX_set_msg_callback(ctx, cb)
2323 SSL_CTX_set_msg_callback_arg(ctx, arg)
2324 SSL_set_msg_callback(ssl, cb)
2325 SSL_set_msg_callback_arg(ssl, arg)
2326
2327 to request calling a callback function
2328
2329 void cb(int write_p, int version, int content_type,
2330 const void *buf, size_t len, SSL *ssl, void *arg)
2331
2332 whenever a protocol message has been completely received
2333 (write_p == 0) or sent (write_p == 1). Here 'version' is the
2334 protocol version according to which the SSL library interprets
2335 the current protocol message (SSL2_VERSION, SSL3_VERSION, or
2336 TLS1_VERSION). 'content_type' is 0 in the case of SSL 2.0, or
2337 the content type as defined in the SSL 3.0/TLS 1.0 protocol
2338 specification (change_cipher_spec(20), alert(21), handshake(22)).
2339 'buf' and 'len' point to the actual message, 'ssl' to the
2340 SSL object, and 'arg' is the application-defined value set by
2341 SSL[_CTX]_set_msg_callback_arg().
2342
2343 'openssl s_client' and 'openssl s_server' have new '-msg' options
2344 to enable a callback that displays all protocol messages.
2345 [Bodo Moeller]
2346
2347 *) Change the shared library support so shared libraries are built as
2348 soon as the corresponding static library is finished, and thereby get
2349 openssl and the test programs linked against the shared library.
2350 This still only happens when the keyword "shard" has been given to
2351 the configuration scripts.
2352
2353 NOTE: shared library support is still an experimental thing, and
2354 backward binary compatibility is still not guaranteed.
2355 ["Maciej W. Rozycki" <macro@ds2.pg.gda.pl> and Richard Levitte]
2356
2357 *) Add support for Subject Information Access extension.
2358 [Peter Sylvester <Peter.Sylvester@EdelWeb.fr>]
2359
2360 *) Make BUF_MEM_grow() behaviour more consistent: Initialise to zero
2361 additional bytes when new memory had to be allocated, not just
2362 when reusing an existing buffer.
2363 [Bodo Moeller]
2364
2365 *) New command line and configuration option 'utf8' for the req command.
2366 This allows field values to be specified as UTF8 strings.
2367 [Steve Henson]
2368
2369 *) Add -multi and -mr options to "openssl speed" - giving multiple parallel
2370 runs for the former and machine-readable output for the latter.
2371 [Ben Laurie]
2372
2373 *) Add '-noemailDN' option to 'openssl ca'. This prevents inclusion
2374 of the e-mail address in the DN (i.e., it will go into a certificate
2375 extension only). The new configuration file option 'email_in_dn = no'
2376 has the same effect.
2377 [Massimiliano Pala madwolf@openca.org]
2378
2379 *) Change all functions with names starting with des_ to be starting
2380 with DES_ instead. Add wrappers that are compatible with libdes,
2381 but are named _ossl_old_des_*. Finally, add macros that map the
2382 des_* symbols to the corresponding _ossl_old_des_* if libdes
2383 compatibility is desired. If OpenSSL 0.9.6c compatibility is
2384 desired, the des_* symbols will be mapped to DES_*, with one
2385 exception.
2386
2387 Since we provide two compatibility mappings, the user needs to
2388 define the macro OPENSSL_DES_LIBDES_COMPATIBILITY if libdes
2389 compatibility is desired. The default (i.e., when that macro
2390 isn't defined) is OpenSSL 0.9.6c compatibility.
2391
2392 There are also macros that enable and disable the support of old
2393 des functions altogether. Those are OPENSSL_ENABLE_OLD_DES_SUPPORT
2394 and OPENSSL_DISABLE_OLD_DES_SUPPORT. If none or both of those
2395 are defined, the default will apply: to support the old des routines.
2396
2397 In either case, one must include openssl/des.h to get the correct
2398 definitions. Do not try to just include openssl/des_old.h, that
2399 won't work.
2400
2401 NOTE: This is a major break of an old API into a new one. Software
2402 authors are encouraged to switch to the DES_ style functions. Some
2403 time in the future, des_old.h and the libdes compatibility functions
2404 will be disable (i.e. OPENSSL_DISABLE_OLD_DES_SUPPORT will be the
2405 default), and then completely removed.
2406 [Richard Levitte]
2407
2408 *) Test for certificates which contain unsupported critical extensions.
2409 If such a certificate is found during a verify operation it is
2410 rejected by default: this behaviour can be overridden by either
2411 handling the new error X509_V_ERR_UNHANDLED_CRITICAL_EXTENSION or
2412 by setting the verify flag X509_V_FLAG_IGNORE_CRITICAL. A new function
2413 X509_supported_extension() has also been added which returns 1 if a
2414 particular extension is supported.
2415 [Steve Henson]
2416
2417 *) Modify the behaviour of EVP cipher functions in similar way to digests
2418 to retain compatibility with existing code.
2419 [Steve Henson]
2420
2421 *) Modify the behaviour of EVP_DigestInit() and EVP_DigestFinal() to retain
2422 compatibility with existing code. In particular the 'ctx' parameter does
2423 not have to be to be initialized before the call to EVP_DigestInit() and
2424 it is tidied up after a call to EVP_DigestFinal(). New function
2425 EVP_DigestFinal_ex() which does not tidy up the ctx. Similarly function
2426 EVP_MD_CTX_copy() changed to not require the destination to be
2427 initialized valid and new function EVP_MD_CTX_copy_ex() added which
2428 requires the destination to be valid.
2429
2430 Modify all the OpenSSL digest calls to use EVP_DigestInit_ex(),
2431 EVP_DigestFinal_ex() and EVP_MD_CTX_copy_ex().
2432 [Steve Henson]
2433
2434 *) Change ssl3_get_message (ssl/s3_both.c) and the functions using it
2435 so that complete 'Handshake' protocol structures are kept in memory
2436 instead of overwriting 'msg_type' and 'length' with 'body' data.
2437 [Bodo Moeller]
2438
2439 *) Add an implementation of SSL_add_dir_cert_subjects_to_stack for Win32.
2440 [Massimo Santin via Richard Levitte]
2441
2442 *) Major restructuring to the underlying ENGINE code. This includes
2443 reduction of linker bloat, separation of pure "ENGINE" manipulation
2444 (initialisation, etc) from functionality dealing with implementations
2445 of specific crypto iterfaces. This change also introduces integrated
2446 support for symmetric ciphers and digest implementations - so ENGINEs
2447 can now accelerate these by providing EVP_CIPHER and EVP_MD
2448 implementations of their own. This is detailed in crypto/engine/README
2449 as it couldn't be adequately described here. However, there are a few
2450 API changes worth noting - some RSA, DSA, DH, and RAND functions that
2451 were changed in the original introduction of ENGINE code have now
2452 reverted back - the hooking from this code to ENGINE is now a good
2453 deal more passive and at run-time, operations deal directly with
2454 RSA_METHODs, DSA_METHODs (etc) as they did before, rather than
2455 dereferencing through an ENGINE pointer any more. Also, the ENGINE
2456 functions dealing with BN_MOD_EXP[_CRT] handlers have been removed -
2457 they were not being used by the framework as there is no concept of a
2458 BIGNUM_METHOD and they could not be generalised to the new
2459 'ENGINE_TABLE' mechanism that underlies the new code. Similarly,
2460 ENGINE_cpy() has been removed as it cannot be consistently defined in
2461 the new code.
2462 [Geoff Thorpe]
2463
2464 *) Change ASN1_GENERALIZEDTIME_check() to allow fractional seconds.
2465 [Steve Henson]
2466
2467 *) Change mkdef.pl to sort symbols that get the same entry number,
2468 and make sure the automatically generated functions ERR_load_*
2469 become part of libeay.num as well.
2470 [Richard Levitte]
2471
2472 *) New function SSL_renegotiate_pending(). This returns true once
2473 renegotiation has been requested (either SSL_renegotiate() call
2474 or HelloRequest/ClientHello receveived from the peer) and becomes
2475 false once a handshake has been completed.
2476 (For servers, SSL_renegotiate() followed by SSL_do_handshake()
2477 sends a HelloRequest, but does not ensure that a handshake takes
2478 place. SSL_renegotiate_pending() is useful for checking if the
2479 client has followed the request.)
2480 [Bodo Moeller]
2481
2482 *) New SSL option SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION.
2483 By default, clients may request session resumption even during
2484 renegotiation (if session ID contexts permit); with this option,
2485 session resumption is possible only in the first handshake.
2486
2487 SSL_OP_ALL is now 0x00000FFFL instead of 0x000FFFFFL. This makes
2488 more bits available for options that should not be part of
2489 SSL_OP_ALL (such as SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION).
2490 [Bodo Moeller]
2491
2492 *) Add some demos for certificate and certificate request creation.
2493 [Steve Henson]
2494
2495 *) Make maximum certificate chain size accepted from the peer application
2496 settable (SSL*_get/set_max_cert_list()), as proposed by
2497 "Douglas E. Engert" <deengert@anl.gov>.
2498 [Lutz Jaenicke]
2499
2500 *) Add support for shared libraries for Unixware-7
2501 (Boyd Lynn Gerber <gerberb@zenez.com>).
2502 [Lutz Jaenicke]
2503
2504 *) Add a "destroy" handler to ENGINEs that allows structural cleanup to
2505 be done prior to destruction. Use this to unload error strings from
2506 ENGINEs that load their own error strings. NB: This adds two new API
2507 functions to "get" and "set" this destroy handler in an ENGINE.
2508 [Geoff Thorpe]
2509
2510 *) Alter all existing ENGINE implementations (except "openssl" and
2511 "openbsd") to dynamically instantiate their own error strings. This
2512 makes them more flexible to be built both as statically-linked ENGINEs
2513 and self-contained shared-libraries loadable via the "dynamic" ENGINE.
2514 Also, add stub code to each that makes building them as self-contained
2515 shared-libraries easier (see README.ENGINE).
2516 [Geoff Thorpe]
2517
2518 *) Add a "dynamic" ENGINE that provides a mechanism for binding ENGINE
2519 implementations into applications that are completely implemented in
2520 self-contained shared-libraries. The "dynamic" ENGINE exposes control
2521 commands that can be used to configure what shared-library to load and
2522 to control aspects of the way it is handled. Also, made an update to
2523 the README.ENGINE file that brings its information up-to-date and
2524 provides some information and instructions on the "dynamic" ENGINE
2525 (ie. how to use it, how to build "dynamic"-loadable ENGINEs, etc).
2526 [Geoff Thorpe]
2527
2528 *) Make it possible to unload ranges of ERR strings with a new
2529 "ERR_unload_strings" function.
2530 [Geoff Thorpe]
2531
2532 *) Add a copy() function to EVP_MD.
2533 [Ben Laurie]
2534
2535 *) Make EVP_MD routines take a context pointer instead of just the
2536 md_data void pointer.
2537 [Ben Laurie]
2538
2539 *) Add flags to EVP_MD and EVP_MD_CTX. EVP_MD_FLAG_ONESHOT indicates
2540 that the digest can only process a single chunk of data
2541 (typically because it is provided by a piece of
2542 hardware). EVP_MD_CTX_FLAG_ONESHOT indicates that the application
2543 is only going to provide a single chunk of data, and hence the
2544 framework needn't accumulate the data for oneshot drivers.
2545 [Ben Laurie]
2546
2547 *) As with "ERR", make it possible to replace the underlying "ex_data"
2548 functions. This change also alters the storage and management of global
2549 ex_data state - it's now all inside ex_data.c and all "class" code (eg.
2550 RSA, BIO, SSL_CTX, etc) no longer stores its own STACKS and per-class
2551 index counters. The API functions that use this state have been changed
2552 to take a "class_index" rather than pointers to the class's local STACK
2553 and counter, and there is now an API function to dynamically create new
2554 classes. This centralisation allows us to (a) plug a lot of the
2555 thread-safety problems that existed, and (b) makes it possible to clean
2556 up all allocated state using "CRYPTO_cleanup_all_ex_data()". W.r.t. (b)
2557 such data would previously have always leaked in application code and
2558 workarounds were in place to make the memory debugging turn a blind eye
2559 to it. Application code that doesn't use this new function will still
2560 leak as before, but their memory debugging output will announce it now
2561 rather than letting it slide.
2562
2563 Besides the addition of CRYPTO_cleanup_all_ex_data(), another API change
2564 induced by the "ex_data" overhaul is that X509_STORE_CTX_init() now
2565 has a return value to indicate success or failure.
2566 [Geoff Thorpe]
2567
2568 *) Make it possible to replace the underlying "ERR" functions such that the
2569 global state (2 LHASH tables and 2 locks) is only used by the "default"
2570 implementation. This change also adds two functions to "get" and "set"
2571 the implementation prior to it being automatically set the first time
2572 any other ERR function takes place. Ie. an application can call "get",
2573 pass the return value to a module it has just loaded, and that module
2574 can call its own "set" function using that value. This means the
2575 module's "ERR" operations will use (and modify) the error state in the
2576 application and not in its own statically linked copy of OpenSSL code.
2577 [Geoff Thorpe]
2578
2579 *) Give DH, DSA, and RSA types their own "**_up_ref()" function to increment
2580 reference counts. This performs normal REF_PRINT/REF_CHECK macros on
2581 the operation, and provides a more encapsulated way for external code
2582 (crypto/evp/ and ssl/) to do this. Also changed the evp and ssl code
2583 to use these functions rather than manually incrementing the counts.
2584
2585 Also rename "DSO_up()" function to more descriptive "DSO_up_ref()".
2586 [Geoff Thorpe]
2587
2588 *) Add EVP test program.
2589 [Ben Laurie]
2590
2591 *) Add symmetric cipher support to ENGINE. Expect the API to change!
2592 [Ben Laurie]
2593
2594 *) New CRL functions: X509_CRL_set_version(), X509_CRL_set_issuer_name()
2595 X509_CRL_set_lastUpdate(), X509_CRL_set_nextUpdate(), X509_CRL_sort(),
2596 X509_REVOKED_set_serialNumber(), and X509_REVOKED_set_revocationDate().
2597 These allow a CRL to be built without having to access X509_CRL fields
2598 directly. Modify 'ca' application to use new functions.
2599 [Steve Henson]
2600
2601 *) Move SSL_OP_TLS_ROLLBACK_BUG out of the SSL_OP_ALL list of recommended
2602 bug workarounds. Rollback attack detection is a security feature.
2603 The problem will only arise on OpenSSL servers when TLSv1 is not
2604 available (sslv3_server_method() or SSL_OP_NO_TLSv1).
2605 Software authors not wanting to support TLSv1 will have special reasons
2606 for their choice and can explicitly enable this option.
2607 [Bodo Moeller, Lutz Jaenicke]
2608
2609 *) Rationalise EVP so it can be extended: don't include a union of
2610 cipher/digest structures, add init/cleanup functions for EVP_MD_CTX
2611 (similar to those existing for EVP_CIPHER_CTX).
2612 Usage example:
2613
2614 EVP_MD_CTX md;
2615
2616 EVP_MD_CTX_init(&md); /* new function call */
2617 EVP_DigestInit(&md, EVP_sha1());
2618 EVP_DigestUpdate(&md, in, len);
2619 EVP_DigestFinal(&md, out, NULL);
2620 EVP_MD_CTX_cleanup(&md); /* new function call */
2621
2622 [Ben Laurie]
2623
2624 *) Make DES key schedule conform to the usual scheme, as well as
2625 correcting its structure. This means that calls to DES functions
2626 now have to pass a pointer to a des_key_schedule instead of a
2627 plain des_key_schedule (which was actually always a pointer
2628 anyway): E.g.,
2629
2630 des_key_schedule ks;
2631
2632 des_set_key_checked(..., &ks);
2633 des_ncbc_encrypt(..., &ks, ...);
2634
2635 (Note that a later change renames 'des_...' into 'DES_...'.)
2636 [Ben Laurie]
2637
2638 *) Initial reduction of linker bloat: the use of some functions, such as
2639 PEM causes large amounts of unused functions to be linked in due to
2640 poor organisation. For example pem_all.c contains every PEM function
2641 which has a knock on effect of linking in large amounts of (unused)
2642 ASN1 code. Grouping together similar functions and splitting unrelated
2643 functions prevents this.
2644 [Steve Henson]
2645
2646 *) Cleanup of EVP macros.
2647 [Ben Laurie]
2648
2649 *) Change historical references to {NID,SN,LN}_des_ede and ede3 to add the
2650 correct _ecb suffix.
2651 [Ben Laurie]
2652
2653 *) Add initial OCSP responder support to ocsp application. The
2654 revocation information is handled using the text based index
2655 use by the ca application. The responder can either handle
2656 requests generated internally, supplied in files (for example
2657 via a CGI script) or using an internal minimal server.
2658 [Steve Henson]
2659
2660 *) Add configuration choices to get zlib compression for TLS.
2661 [Richard Levitte]
2662
2663 *) Changes to Kerberos SSL for RFC 2712 compliance:
2664 1. Implemented real KerberosWrapper, instead of just using
2665 KRB5 AP_REQ message. [Thanks to Simon Wilkinson <sxw@sxw.org.uk>]
2666 2. Implemented optional authenticator field of KerberosWrapper.
2667
2668 Added openssl-style ASN.1 macros for Kerberos ticket, ap_req,
2669 and authenticator structs; see crypto/krb5/.
2670
2671 Generalized Kerberos calls to support multiple Kerberos libraries.
2672 [Vern Staats <staatsvr@asc.hpc.mil>,
2673 Jeffrey Altman <jaltman@columbia.edu>
2674 via Richard Levitte]
2675
2676 *) Cause 'openssl speed' to use fully hard-coded DSA keys as it
2677 already does with RSA. testdsa.h now has 'priv_key/pub_key'
2678 values for each of the key sizes rather than having just
2679 parameters (and 'speed' generating keys each time).
2680 [Geoff Thorpe]
2681
2682 *) Speed up EVP routines.
2683 Before:
2684 encrypt
2685 type 8 bytes 64 bytes 256 bytes 1024 bytes 8192 bytes
2686 des-cbc 4408.85k 5560.51k 5778.46k 5862.20k 5825.16k
2687 des-cbc 4389.55k 5571.17k 5792.23k 5846.91k 5832.11k
2688 des-cbc 4394.32k 5575.92k 5807.44k 5848.37k 5841.30k
2689 decrypt
2690 des-cbc 3482.66k 5069.49k 5496.39k 5614.16k 5639.28k
2691 des-cbc 3480.74k 5068.76k 5510.34k 5609.87k 5635.52k
2692 des-cbc 3483.72k 5067.62k 5504.60k 5708.01k 5724.80k
2693 After:
2694 encrypt
2695 des-cbc 4660.16k 5650.19k 5807.19k 5827.13k 5783.32k
2696 decrypt
2697 des-cbc 3624.96k 5258.21k 5530.91k 5624.30k 5628.26k
2698 [Ben Laurie]
2699
2700 *) Added the OS2-EMX target.
2701 ["Brian Havard" <brianh@kheldar.apana.org.au> and Richard Levitte]
2702
2703 *) Rewrite apps to use NCONF routines instead of the old CONF. New functions
2704 to support NCONF routines in extension code. New function CONF_set_nconf()
2705 to allow functions which take an NCONF to also handle the old LHASH
2706 structure: this means that the old CONF compatible routines can be
2707 retained (in particular wrt extensions) without having to duplicate the
2708 code. New function X509V3_add_ext_nconf_sk to add extensions to a stack.
2709 [Steve Henson]
2710
2711 *) Enhance the general user interface with mechanisms for inner control
2712 and with possibilities to have yes/no kind of prompts.
2713 [Richard Levitte]
2714
2715 *) Change all calls to low level digest routines in the library and
2716 applications to use EVP. Add missing calls to HMAC_cleanup() and
2717 don't assume HMAC_CTX can be copied using memcpy().
2718 [Verdon Walker <VWalker@novell.com>, Steve Henson]
2719
2720 *) Add the possibility to control engines through control names but with
2721 arbitrary arguments instead of just a string.
2722 Change the key loaders to take a UI_METHOD instead of a callback
2723 function pointer. NOTE: this breaks binary compatibility with earlier
2724 versions of OpenSSL [engine].
2725 Adapt the nCipher code for these new conditions and add a card insertion
2726 callback.
2727 [Richard Levitte]
2728
2729 *) Enhance the general user interface with mechanisms to better support
2730 dialog box interfaces, application-defined prompts, the possibility
2731 to use defaults (for example default passwords from somewhere else)
2732 and interrupts/cancellations.
2733 [Richard Levitte]
2734
2735 *) Tidy up PKCS#12 attribute handling. Add support for the CSP name
2736 attribute in PKCS#12 files, add new -CSP option to pkcs12 utility.
2737 [Steve Henson]
2738
2739 *) Fix a memory leak in 'sk_dup()' in the case reallocation fails. (Also
2740 tidy up some unnecessarily weird code in 'sk_new()').
2741 [Geoff, reported by Diego Tartara <dtartara@novamens.com>]
2742
2743 *) Change the key loading routines for ENGINEs to use the same kind
2744 callback (pem_password_cb) as all other routines that need this
2745 kind of callback.
2746 [Richard Levitte]
2747
2748 *) Increase ENTROPY_NEEDED to 32 bytes, as Rijndael can operate with
2749 256 bit (=32 byte) keys. Of course seeding with more entropy bytes
2750 than this minimum value is recommended.
2751 [Lutz Jaenicke]
2752
2753 *) New random seeder for OpenVMS, using the system process statistics
2754 that are easily reachable.
2755 [Richard Levitte]
2756
2757 *) Windows apparently can't transparently handle global
2758 variables defined in DLLs. Initialisations such as:
2759
2760 const ASN1_ITEM *it = &ASN1_INTEGER_it;
2761
2762 wont compile. This is used by the any applications that need to
2763 declare their own ASN1 modules. This was fixed by adding the option
2764 EXPORT_VAR_AS_FN to all Win32 platforms, although this isn't strictly
2765 needed for static libraries under Win32.
2766 [Steve Henson]
2767
2768 *) New functions X509_PURPOSE_set() and X509_TRUST_set() to handle
2769 setting of purpose and trust fields. New X509_STORE trust and
2770 purpose functions and tidy up setting in other SSL functions.
2771 [Steve Henson]
2772
2773 *) Add copies of X509_STORE_CTX fields and callbacks to X509_STORE
2774 structure. These are inherited by X509_STORE_CTX when it is
2775 initialised. This allows various defaults to be set in the
2776 X509_STORE structure (such as flags for CRL checking and custom
2777 purpose or trust settings) for functions which only use X509_STORE_CTX
2778 internally such as S/MIME.
2779
2780 Modify X509_STORE_CTX_purpose_inherit() so it only sets purposes and
2781 trust settings if they are not set in X509_STORE. This allows X509_STORE
2782 purposes and trust (in S/MIME for example) to override any set by default.
2783
2784 Add command line options for CRL checking to smime, s_client and s_server
2785 applications.
2786 [Steve Henson]
2787
2788 *) Initial CRL based revocation checking. If the CRL checking flag(s)
2789 are set then the CRL is looked up in the X509_STORE structure and
2790 its validity and signature checked, then if the certificate is found
2791 in the CRL the verify fails with a revoked error.
2792
2793 Various new CRL related callbacks added to X509_STORE_CTX structure.
2794
2795 Command line options added to 'verify' application to support this.
2796
2797 This needs some additional work, such as being able to handle multiple
2798 CRLs with different times, extension based lookup (rather than just
2799 by subject name) and ultimately more complete V2 CRL extension
2800 handling.
2801 [Steve Henson]
2802
2803 *) Add a general user interface API (crypto/ui/). This is designed
2804 to replace things like des_read_password and friends (backward
2805 compatibility functions using this new API are provided).
2806 The purpose is to remove prompting functions from the DES code
2807 section as well as provide for prompting through dialog boxes in
2808 a window system and the like.
2809 [Richard Levitte]
2810
2811 *) Add "ex_data" support to ENGINE so implementations can add state at a
2812 per-structure level rather than having to store it globally.
2813 [Geoff]
2814
2815 *) Make it possible for ENGINE structures to be copied when retrieved by
2816 ENGINE_by_id() if the ENGINE specifies a new flag: ENGINE_FLAGS_BY_ID_COPY.
2817 This causes the "original" ENGINE structure to act like a template,
2818 analogous to the RSA vs. RSA_METHOD type of separation. Because of this
2819 operational state can be localised to each ENGINE structure, despite the
2820 fact they all share the same "methods". New ENGINE structures returned in
2821 this case have no functional references and the return value is the single
2822 structural reference. This matches the single structural reference returned
2823 by ENGINE_by_id() normally, when it is incremented on the pre-existing
2824 ENGINE structure.
2825 [Geoff]
2826
2827 *) Fix ASN1 decoder when decoding type ANY and V_ASN1_OTHER: since this
2828 needs to match any other type at all we need to manually clear the
2829 tag cache.
2830 [Steve Henson]
2831
2832 *) Changes to the "openssl engine" utility to include;
2833 - verbosity levels ('-v', '-vv', and '-vvv') that provide information
2834 about an ENGINE's available control commands.
2835 - executing control commands from command line arguments using the
2836 '-pre' and '-post' switches. '-post' is only used if '-t' is
2837 specified and the ENGINE is successfully initialised. The syntax for
2838 the individual commands are colon-separated, for example;
2839 openssl engine chil -pre FORK_CHECK:0 -pre SO_PATH:/lib/test.so
2840 [Geoff]
2841
2842 *) New dynamic control command support for ENGINEs. ENGINEs can now
2843 declare their own commands (numbers), names (strings), descriptions,
2844 and input types for run-time discovery by calling applications. A
2845 subset of these commands are implicitly classed as "executable"
2846 depending on their input type, and only these can be invoked through
2847 the new string-based API function ENGINE_ctrl_cmd_string(). (Eg. this
2848 can be based on user input, config files, etc). The distinction is
2849 that "executable" commands cannot return anything other than a boolean
2850 result and can only support numeric or string input, whereas some
2851 discoverable commands may only be for direct use through
2852 ENGINE_ctrl(), eg. supporting the exchange of binary data, function
2853 pointers, or other custom uses. The "executable" commands are to
2854 support parameterisations of ENGINE behaviour that can be
2855 unambiguously defined by ENGINEs and used consistently across any
2856 OpenSSL-based application. Commands have been added to all the
2857 existing hardware-supporting ENGINEs, noticeably "SO_PATH" to allow
2858 control over shared-library paths without source code alterations.
2859 [Geoff]
2860
2861 *) Changed all ENGINE implementations to dynamically allocate their
2862 ENGINEs rather than declaring them statically. Apart from this being
2863 necessary with the removal of the ENGINE_FLAGS_MALLOCED distinction,
2864 this also allows the implementations to compile without using the
2865 internal engine_int.h header.
2866 [Geoff]
2867
2868 *) Minor adjustment to "rand" code. RAND_get_rand_method() now returns a
2869 'const' value. Any code that should be able to modify a RAND_METHOD
2870 should already have non-const pointers to it (ie. they should only
2871 modify their own ones).
2872 [Geoff]
2873
2874 *) Made a variety of little tweaks to the ENGINE code.
2875 - "atalla" and "ubsec" string definitions were moved from header files
2876 to C code. "nuron" string definitions were placed in variables
2877 rather than hard-coded - allowing parameterisation of these values
2878 later on via ctrl() commands.
2879 - Removed unused "#if 0"'d code.
2880 - Fixed engine list iteration code so it uses ENGINE_free() to release
2881 structural references.
2882 - Constified the RAND_METHOD element of ENGINE structures.
2883 - Constified various get/set functions as appropriate and added
2884 missing functions (including a catch-all ENGINE_cpy that duplicates
2885 all ENGINE values onto a new ENGINE except reference counts/state).
2886 - Removed NULL parameter checks in get/set functions. Setting a method
2887 or function to NULL is a way of cancelling out a previously set
2888 value. Passing a NULL ENGINE parameter is just plain stupid anyway
2889 and doesn't justify the extra error symbols and code.
2890 - Deprecate the ENGINE_FLAGS_MALLOCED define and move the area for
2891 flags from engine_int.h to engine.h.
2892 - Changed prototypes for ENGINE handler functions (init(), finish(),
2893 ctrl(), key-load functions, etc) to take an (ENGINE*) parameter.
2894 [Geoff]
2895
2896 *) Implement binary inversion algorithm for BN_mod_inverse in addition
2897 to the algorithm using long division. The binary algorithm can be
2898 used only if the modulus is odd. On 32-bit systems, it is faster
2899 only for relatively small moduli (roughly 20-30% for 128-bit moduli,
2900 roughly 5-15% for 256-bit moduli), so we use it only for moduli
2901 up to 450 bits. In 64-bit environments, the binary algorithm
2902 appears to be advantageous for much longer moduli; here we use it
2903 for moduli up to 2048 bits.
2904 [Bodo Moeller]
2905
2906 *) Rewrite CHOICE field setting in ASN1_item_ex_d2i(). The old code
2907 could not support the combine flag in choice fields.
2908 [Steve Henson]
2909
2910 *) Add a 'copy_extensions' option to the 'ca' utility. This copies
2911 extensions from a certificate request to the certificate.
2912 [Steve Henson]
2913
2914 *) Allow multiple 'certopt' and 'nameopt' options to be separated
2915 by commas. Add 'namopt' and 'certopt' options to the 'ca' config
2916 file: this allows the display of the certificate about to be
2917 signed to be customised, to allow certain fields to be included
2918 or excluded and extension details. The old system didn't display
2919 multicharacter strings properly, omitted fields not in the policy
2920 and couldn't display additional details such as extensions.
2921 [Steve Henson]
2922
2923 *) Function EC_POINTs_mul for multiple scalar multiplication
2924 of an arbitrary number of elliptic curve points
2925 \sum scalars[i]*points[i],
2926 optionally including the generator defined for the EC_GROUP:
2927 scalar*generator + \sum scalars[i]*points[i].
2928
2929 EC_POINT_mul is a simple wrapper function for the typical case
2930 that the point list has just one item (besides the optional
2931 generator).
2932 [Bodo Moeller]
2933
2934 *) First EC_METHODs for curves over GF(p):
2935
2936 EC_GFp_simple_method() uses the basic BN_mod_mul and BN_mod_sqr
2937 operations and provides various method functions that can also
2938 operate with faster implementations of modular arithmetic.
2939
2940 EC_GFp_mont_method() reuses most functions that are part of
2941 EC_GFp_simple_method, but uses Montgomery arithmetic.
2942
2943 [Bodo Moeller; point addition and point doubling
2944 implementation directly derived from source code provided by
2945 Lenka Fibikova <fibikova@exp-math.uni-essen.de>]
2946
2947 *) Framework for elliptic curves (crypto/ec/ec.h, crypto/ec/ec_lcl.h,
2948 crypto/ec/ec_lib.c):
2949
2950 Curves are EC_GROUP objects (with an optional group generator)
2951 based on EC_METHODs that are built into the library.
2952
2953 Points are EC_POINT objects based on EC_GROUP objects.
2954
2955 Most of the framework would be able to handle curves over arbitrary
2956 finite fields, but as there are no obvious types for fields other
2957 than GF(p), some functions are limited to that for now.
2958 [Bodo Moeller]
2959
2960 *) Add the -HTTP option to s_server. It is similar to -WWW, but requires
2961 that the file contains a complete HTTP response.
2962 [Richard Levitte]
2963
2964 *) Add the ec directory to mkdef.pl and mkfiles.pl. In mkdef.pl
2965 change the def and num file printf format specifier from "%-40sXXX"
2966 to "%-39s XXX". The latter will always guarantee a space after the
2967 field while the former will cause them to run together if the field
2968 is 40 of more characters long.
2969 [Steve Henson]
2970
2971 *) Constify the cipher and digest 'method' functions and structures
2972 and modify related functions to take constant EVP_MD and EVP_CIPHER
2973 pointers.
2974 [Steve Henson]
2975
2976 *) Hide BN_CTX structure details in bn_lcl.h instead of publishing them
2977 in <openssl/bn.h>. Also further increase BN_CTX_NUM to 32.
2978 [Bodo Moeller]
2979
2980 *) Modify EVP_Digest*() routines so they now return values. Although the
2981 internal software routines can never fail additional hardware versions
2982 might.
2983 [Steve Henson]
2984
2985 *) Clean up crypto/err/err.h and change some error codes to avoid conflicts:
2986
2987 Previously ERR_R_FATAL was too small and coincided with ERR_LIB_PKCS7
2988 (= ERR_R_PKCS7_LIB); it is now 64 instead of 32.
2989
2990 ASN1 error codes
2991 ERR_R_NESTED_ASN1_ERROR
2992 ...
2993 ERR_R_MISSING_ASN1_EOS
2994 were 4 .. 9, conflicting with
2995 ERR_LIB_RSA (= ERR_R_RSA_LIB)
2996 ...
2997 ERR_LIB_PEM (= ERR_R_PEM_LIB).
2998 They are now 58 .. 63 (i.e., just below ERR_R_FATAL).
2999
3000 Add new error code 'ERR_R_INTERNAL_ERROR'.
3001 [Bodo Moeller]
3002
3003 *) Don't overuse locks in crypto/err/err.c: For data retrieval, CRYPTO_r_lock
3004 suffices.
3005 [Bodo Moeller]
3006
3007 *) New option '-subj arg' for 'openssl req' and 'openssl ca'. This
3008 sets the subject name for a new request or supersedes the
3009 subject name in a given request. Formats that can be parsed are
3010 'CN=Some Name, OU=myOU, C=IT'
3011 and
3012 'CN=Some Name/OU=myOU/C=IT'.
3013
3014 Add options '-batch' and '-verbose' to 'openssl req'.
3015 [Massimiliano Pala <madwolf@hackmasters.net>]
3016
3017 *) Introduce the possibility to access global variables through
3018 functions on platform were that's the best way to handle exporting
3019 global variables in shared libraries. To enable this functionality,
3020 one must configure with "EXPORT_VAR_AS_FN" or defined the C macro
3021 "OPENSSL_EXPORT_VAR_AS_FUNCTION" in crypto/opensslconf.h (the latter
3022 is normally done by Configure or something similar).
3023
3024 To implement a global variable, use the macro OPENSSL_IMPLEMENT_GLOBAL
3025 in the source file (foo.c) like this:
3026
3027 OPENSSL_IMPLEMENT_GLOBAL(int,foo)=1;
3028 OPENSSL_IMPLEMENT_GLOBAL(double,bar);
3029
3030 To declare a global variable, use the macros OPENSSL_DECLARE_GLOBAL
3031 and OPENSSL_GLOBAL_REF in the header file (foo.h) like this:
3032
3033 OPENSSL_DECLARE_GLOBAL(int,foo);
3034 #define foo OPENSSL_GLOBAL_REF(foo)
3035 OPENSSL_DECLARE_GLOBAL(double,bar);
3036 #define bar OPENSSL_GLOBAL_REF(bar)
3037
3038 The #defines are very important, and therefore so is including the
3039 header file everywhere where the defined globals are used.
3040
3041 The macro OPENSSL_EXPORT_VAR_AS_FUNCTION also affects the definition
3042 of ASN.1 items, but that structure is a bit different.
3043
3044 The largest change is in util/mkdef.pl which has been enhanced with
3045 better and easier to understand logic to choose which symbols should
3046 go into the Windows .def files as well as a number of fixes and code
3047 cleanup (among others, algorithm keywords are now sorted
3048 lexicographically to avoid constant rewrites).
3049 [Richard Levitte]
3050
3051 *) In BN_div() keep a copy of the sign of 'num' before writing the
3052 result to 'rm' because if rm==num the value will be overwritten
3053 and produce the wrong result if 'num' is negative: this caused
3054 problems with BN_mod() and BN_nnmod().
3055 [Steve Henson]
3056
3057 *) Function OCSP_request_verify(). This checks the signature on an
3058 OCSP request and verifies the signer certificate. The signer
3059 certificate is just checked for a generic purpose and OCSP request
3060 trust settings.
3061 [Steve Henson]
3062
3063 *) Add OCSP_check_validity() function to check the validity of OCSP
3064 responses. OCSP responses are prepared in real time and may only
3065 be a few seconds old. Simply checking that the current time lies
3066 between thisUpdate and nextUpdate max reject otherwise valid responses
3067 caused by either OCSP responder or client clock inaccuracy. Instead
3068 we allow thisUpdate and nextUpdate to fall within a certain period of
3069 the current time. The age of the response can also optionally be
3070 checked. Two new options -validity_period and -status_age added to
3071 ocsp utility.
3072 [Steve Henson]
3073
3074 *) If signature or public key algorithm is unrecognized print out its
3075 OID rather that just UNKNOWN.
3076 [Steve Henson]
3077
3078 *) Change OCSP_cert_to_id() to tolerate a NULL subject certificate and
3079 OCSP_cert_id_new() a NULL serialNumber. This allows a partial certificate
3080 ID to be generated from the issuer certificate alone which can then be
3081 passed to OCSP_id_issuer_cmp().
3082 [Steve Henson]
3083
3084 *) New compilation option ASN1_ITEM_FUNCTIONS. This causes the new
3085 ASN1 modules to export functions returning ASN1_ITEM pointers
3086 instead of the ASN1_ITEM structures themselves. This adds several
3087 new macros which allow the underlying ASN1 function/structure to
3088 be accessed transparently. As a result code should not use ASN1_ITEM
3089 references directly (such as &X509_it) but instead use the relevant
3090 macros (such as ASN1_ITEM_rptr(X509)). This option is to allow
3091 use of the new ASN1 code on platforms where exporting structures
3092 is problematical (for example in shared libraries) but exporting
3093 functions returning pointers to structures is not.
3094 [Steve Henson]
3095
3096 *) Add support for overriding the generation of SSL/TLS session IDs.
3097 These callbacks can be registered either in an SSL_CTX or per SSL.
3098 The purpose of this is to allow applications to control, if they wish,
3099 the arbitrary values chosen for use as session IDs, particularly as it
3100 can be useful for session caching in multiple-server environments. A
3101 command-line switch for testing this (and any client code that wishes
3102 to use such a feature) has been added to "s_server".
3103 [Geoff Thorpe, Lutz Jaenicke]
3104
3105 *) Modify mkdef.pl to recognise and parse preprocessor conditionals
3106 of the form '#if defined(...) || defined(...) || ...' and
3107 '#if !defined(...) && !defined(...) && ...'. This also avoids
3108 the growing number of special cases it was previously handling.
3109 [Richard Levitte]
3110
3111 *) Make all configuration macros available for application by making
3112 sure they are available in opensslconf.h, by giving them names starting
3113 with "OPENSSL_" to avoid conflicts with other packages and by making
3114 sure e_os2.h will cover all platform-specific cases together with
3115 opensslconf.h.
3116 Additionally, it is now possible to define configuration/platform-
3117 specific names (called "system identities"). In the C code, these
3118 are prefixed with "OPENSSL_SYSNAME_". e_os2.h will create another
3119 macro with the name beginning with "OPENSSL_SYS_", which is determined
3120 from "OPENSSL_SYSNAME_*" or compiler-specific macros depending on
3121 what is available.
3122 [Richard Levitte]
3123
3124 *) New option -set_serial to 'req' and 'x509' this allows the serial
3125 number to use to be specified on the command line. Previously self
3126 signed certificates were hard coded with serial number 0 and the
3127 CA options of 'x509' had to use a serial number in a file which was
3128 auto incremented.
3129 [Steve Henson]
3130
3131 *) New options to 'ca' utility to support V2 CRL entry extensions.
3132 Currently CRL reason, invalidity date and hold instruction are
3133 supported. Add new CRL extensions to V3 code and some new objects.
3134 [Steve Henson]
3135
3136 *) New function EVP_CIPHER_CTX_set_padding() this is used to
3137 disable standard block padding (aka PKCS#5 padding) in the EVP
3138 API, which was previously mandatory. This means that the data is
3139 not padded in any way and so the total length much be a multiple
3140 of the block size, otherwise an error occurs.
3141 [Steve Henson]
3142
3143 *) Initial (incomplete) OCSP SSL support.
3144 [Steve Henson]
3145
3146 *) New function OCSP_parse_url(). This splits up a URL into its host,
3147 port and path components: primarily to parse OCSP URLs. New -url
3148 option to ocsp utility.
3149 [Steve Henson]
3150
3151 *) New nonce behavior. The return value of OCSP_check_nonce() now
3152 reflects the various checks performed. Applications can decide
3153 whether to tolerate certain situations such as an absent nonce
3154 in a response when one was present in a request: the ocsp application
3155 just prints out a warning. New function OCSP_add1_basic_nonce()
3156 this is to allow responders to include a nonce in a response even if
3157 the request is nonce-less.
3158 [Steve Henson]
3159
3160 *) Disable stdin buffering in load_cert (apps/apps.c) so that no certs are
3161 skipped when using openssl x509 multiple times on a single input file,
3162 e.g. "(openssl x509 -out cert1; openssl x509 -out cert2) <certs".
3163 [Bodo Moeller]
3164
3165 *) Make ASN1_UTCTIME_set_string() and ASN1_GENERALIZEDTIME_set_string()
3166 set string type: to handle setting ASN1_TIME structures. Fix ca
3167 utility to correctly initialize revocation date of CRLs.
3168 [Steve Henson]
3169
3170 *) New option SSL_OP_CIPHER_SERVER_PREFERENCE allows the server to override
3171 the clients preferred ciphersuites and rather use its own preferences.
3172 Should help to work around M$ SGC (Server Gated Cryptography) bug in
3173 Internet Explorer by ensuring unchanged hash method during stepup.
3174 (Also replaces the broken/deactivated SSL_OP_NON_EXPORT_FIRST option.)
3175 [Lutz Jaenicke]
3176
3177 *) Make mkdef.pl recognise all DECLARE_ASN1 macros, change rijndael
3178 to aes and add a new 'exist' option to print out symbols that don't
3179 appear to exist.
3180 [Steve Henson]
3181
3182 *) Additional options to ocsp utility to allow flags to be set and
3183 additional certificates supplied.
3184 [Steve Henson]
3185
3186 *) Add the option -VAfile to 'openssl ocsp', so the user can give the
3187 OCSP client a number of certificate to only verify the response
3188 signature against.
3189 [Richard Levitte]
3190
3191 *) Update Rijndael code to version 3.0 and change EVP AES ciphers to
3192 handle the new API. Currently only ECB, CBC modes supported. Add new
3193 AES OIDs.
3194
3195 Add TLS AES ciphersuites as described in RFC3268, "Advanced
3196 Encryption Standard (AES) Ciphersuites for Transport Layer
3197 Security (TLS)". (In beta versions of OpenSSL 0.9.7, these were
3198 not enabled by default and were not part of the "ALL" ciphersuite
3199 alias because they were not yet official; they could be
3200 explicitly requested by specifying the "AESdraft" ciphersuite
3201 group alias. In the final release of OpenSSL 0.9.7, the group
3202 alias is called "AES" and is part of "ALL".)
3203 [Ben Laurie, Steve Henson, Bodo Moeller]
3204
3205 *) New function OCSP_copy_nonce() to copy nonce value (if present) from
3206 request to response.
3207 [Steve Henson]
3208
3209 *) Functions for OCSP responders. OCSP_request_onereq_count(),
3210 OCSP_request_onereq_get0(), OCSP_onereq_get0_id() and OCSP_id_get0_info()
3211 extract information from a certificate request. OCSP_response_create()
3212 creates a response and optionally adds a basic response structure.
3213 OCSP_basic_add1_status() adds a complete single response to a basic
3214 response and returns the OCSP_SINGLERESP structure just added (to allow
3215 extensions to be included for example). OCSP_basic_add1_cert() adds a
3216 certificate to a basic response and OCSP_basic_sign() signs a basic
3217 response with various flags. New helper functions ASN1_TIME_check()
3218 (checks validity of ASN1_TIME structure) and ASN1_TIME_to_generalizedtime()
3219 (converts ASN1_TIME to GeneralizedTime).
3220 [Steve Henson]
3221
3222 *) Various new functions. EVP_Digest() combines EVP_Digest{Init,Update,Final}()
3223 in a single operation. X509_get0_pubkey_bitstr() extracts the public_key
3224 structure from a certificate. X509_pubkey_digest() digests the public_key
3225 contents: this is used in various key identifiers.
3226 [Steve Henson]
3227
3228 *) Make sk_sort() tolerate a NULL argument.
3229 [Steve Henson reported by Massimiliano Pala <madwolf@comune.modena.it>]
3230
3231 *) New OCSP verify flag OCSP_TRUSTOTHER. When set the "other" certificates
3232 passed by the function are trusted implicitly. If any of them signed the
3233 response then it is assumed to be valid and is not verified.
3234 [Steve Henson]
3235
3236 *) In PKCS7_set_type() initialise content_type in PKCS7_ENC_CONTENT
3237 to data. This was previously part of the PKCS7 ASN1 code. This
3238 was causing problems with OpenSSL created PKCS#12 and PKCS#7 structures.
3239 [Steve Henson, reported by Kenneth R. Robinette
3240 <support@securenetterm.com>]
3241
3242 *) Add CRYPTO_push_info() and CRYPTO_pop_info() calls to new ASN1
3243 routines: without these tracing memory leaks is very painful.
3244 Fix leaks in PKCS12 and PKCS7 routines.
3245 [Steve Henson]
3246
3247 *) Make X509_time_adj() cope with the new behaviour of ASN1_TIME_new().
3248 Previously it initialised the 'type' argument to V_ASN1_UTCTIME which
3249 effectively meant GeneralizedTime would never be used. Now it
3250 is initialised to -1 but X509_time_adj() now has to check the value
3251 and use ASN1_TIME_set() if the value is not V_ASN1_UTCTIME or
3252 V_ASN1_GENERALIZEDTIME, without this it always uses GeneralizedTime.
3253 [Steve Henson, reported by Kenneth R. Robinette
3254 <support@securenetterm.com>]
3255
3256 *) Fixes to BN_to_ASN1_INTEGER when bn is zero. This would previously
3257 result in a zero length in the ASN1_INTEGER structure which was
3258 not consistent with the structure when d2i_ASN1_INTEGER() was used
3259 and would cause ASN1_INTEGER_cmp() to fail. Enhance s2i_ASN1_INTEGER()
3260 to cope with hex and negative integers. Fix bug in i2a_ASN1_INTEGER()
3261 where it did not print out a minus for negative ASN1_INTEGER.
3262 [Steve Henson]
3263
3264 *) Add summary printout to ocsp utility. The various functions which
3265 convert status values to strings have been renamed to:
3266 OCSP_response_status_str(), OCSP_cert_status_str() and
3267 OCSP_crl_reason_str() and are no longer static. New options
3268 to verify nonce values and to disable verification. OCSP response
3269 printout format cleaned up.
3270 [Steve Henson]
3271
3272 *) Add additional OCSP certificate checks. These are those specified
3273 in RFC2560. This consists of two separate checks: the CA of the
3274 certificate being checked must either be the OCSP signer certificate
3275 or the issuer of the OCSP signer certificate. In the latter case the
3276 OCSP signer certificate must contain the OCSP signing extended key
3277 usage. This check is performed by attempting to match the OCSP
3278 signer or the OCSP signer CA to the issuerNameHash and issuerKeyHash
3279 in the OCSP_CERTID structures of the response.
3280 [Steve Henson]
3281
3282 *) Initial OCSP certificate verification added to OCSP_basic_verify()
3283 and related routines. This uses the standard OpenSSL certificate
3284 verify routines to perform initial checks (just CA validity) and
3285 to obtain the certificate chain. Then additional checks will be
3286 performed on the chain. Currently the root CA is checked to see
3287 if it is explicitly trusted for OCSP signing. This is used to set
3288 a root CA as a global signing root: that is any certificate that
3289 chains to that CA is an acceptable OCSP signing certificate.
3290 [Steve Henson]
3291
3292 *) New '-extfile ...' option to 'openssl ca' for reading X.509v3
3293 extensions from a separate configuration file.
3294 As when reading extensions from the main configuration file,
3295 the '-extensions ...' option may be used for specifying the
3296 section to use.
3297 [Massimiliano Pala <madwolf@comune.modena.it>]
3298
3299 *) New OCSP utility. Allows OCSP requests to be generated or
3300 read. The request can be sent to a responder and the output
3301 parsed, outputed or printed in text form. Not complete yet:
3302 still needs to check the OCSP response validity.
3303 [Steve Henson]
3304
3305 *) New subcommands for 'openssl ca':
3306 'openssl ca -status <serial>' prints the status of the cert with
3307 the given serial number (according to the index file).
3308 'openssl ca -updatedb' updates the expiry status of certificates
3309 in the index file.
3310 [Massimiliano Pala <madwolf@comune.modena.it>]
3311
3312 *) New '-newreq-nodes' command option to CA.pl. This is like
3313 '-newreq', but calls 'openssl req' with the '-nodes' option
3314 so that the resulting key is not encrypted.
3315 [Damien Miller <djm@mindrot.org>]
3316
3317 *) New configuration for the GNU Hurd.
3318 [Jonathan Bartlett <johnnyb@wolfram.com> via Richard Levitte]
3319
3320 *) Initial code to implement OCSP basic response verify. This
3321 is currently incomplete. Currently just finds the signer's
3322 certificate and verifies the signature on the response.
3323 [Steve Henson]
3324
3325 *) New SSLeay_version code SSLEAY_DIR to determine the compiled-in
3326 value of OPENSSLDIR. This is available via the new '-d' option
3327 to 'openssl version', and is also included in 'openssl version -a'.
3328 [Bodo Moeller]
3329
3330 *) Allowing defining memory allocation callbacks that will be given
3331 file name and line number information in additional arguments
3332 (a const char* and an int). The basic functionality remains, as
3333 well as the original possibility to just replace malloc(),
3334 realloc() and free() by functions that do not know about these
3335 additional arguments. To register and find out the current
3336 settings for extended allocation functions, the following
3337 functions are provided:
3338
3339 CRYPTO_set_mem_ex_functions
3340 CRYPTO_set_locked_mem_ex_functions
3341 CRYPTO_get_mem_ex_functions
3342 CRYPTO_get_locked_mem_ex_functions
3343
3344 These work the same way as CRYPTO_set_mem_functions and friends.
3345 CRYPTO_get_[locked_]mem_functions now writes 0 where such an
3346 extended allocation function is enabled.
3347 Similarly, CRYPTO_get_[locked_]mem_ex_functions writes 0 where
3348 a conventional allocation function is enabled.
3349 [Richard Levitte, Bodo Moeller]
3350
3351 *) Finish off removing the remaining LHASH function pointer casts.
3352 There should no longer be any prototype-casting required when using
3353 the LHASH abstraction, and any casts that remain are "bugs". See
3354 the callback types and macros at the head of lhash.h for details
3355 (and "OBJ_cleanup" in crypto/objects/obj_dat.c as an example).
3356 [Geoff Thorpe]
3357
3358 *) Add automatic query of EGD sockets in RAND_poll() for the unix variant.
3359 If /dev/[u]random devices are not available or do not return enough
3360 entropy, EGD style sockets (served by EGD or PRNGD) will automatically
3361 be queried.
3362 The locations /var/run/egd-pool, /dev/egd-pool, /etc/egd-pool, and
3363 /etc/entropy will be queried once each in this sequence, quering stops
3364 when enough entropy was collected without querying more sockets.
3365 [Lutz Jaenicke]
3366
3367 *) Change the Unix RAND_poll() variant to be able to poll several
3368 random devices, as specified by DEVRANDOM, until a sufficient amount
3369 of data has been collected. We spend at most 10 ms on each file
3370 (select timeout) and read in non-blocking mode. DEVRANDOM now
3371 defaults to the list "/dev/urandom", "/dev/random", "/dev/srandom"
3372 (previously it was just the string "/dev/urandom"), so on typical
3373 platforms the 10 ms delay will never occur.
3374 Also separate out the Unix variant to its own file, rand_unix.c.
3375 For VMS, there's a currently-empty rand_vms.c.
3376 [Richard Levitte]
3377
3378 *) Move OCSP client related routines to ocsp_cl.c. These
3379 provide utility functions which an application needing
3380 to issue a request to an OCSP responder and analyse the
3381 response will typically need: as opposed to those which an
3382 OCSP responder itself would need which will be added later.
3383
3384 OCSP_request_sign() signs an OCSP request with an API similar
3385 to PKCS7_sign(). OCSP_response_status() returns status of OCSP
3386 response. OCSP_response_get1_basic() extracts basic response
3387 from response. OCSP_resp_find_status(): finds and extracts status
3388 information from an OCSP_CERTID structure (which will be created
3389 when the request structure is built). These are built from lower
3390 level functions which work on OCSP_SINGLERESP structures but
3391 wont normally be used unless the application wishes to examine
3392 extensions in the OCSP response for example.
3393
3394 Replace nonce routines with a pair of functions.
3395 OCSP_request_add1_nonce() adds a nonce value and optionally
3396 generates a random value. OCSP_check_nonce() checks the
3397 validity of the nonce in an OCSP response.
3398 [Steve Henson]
3399
3400 *) Change function OCSP_request_add() to OCSP_request_add0_id().
3401 This doesn't copy the supplied OCSP_CERTID and avoids the
3402 need to free up the newly created id. Change return type
3403 to OCSP_ONEREQ to return the internal OCSP_ONEREQ structure.
3404 This can then be used to add extensions to the request.
3405 Deleted OCSP_request_new(), since most of its functionality
3406 is now in OCSP_REQUEST_new() (and the case insensitive name
3407 clash) apart from the ability to set the request name which
3408 will be added elsewhere.
3409 [Steve Henson]
3410
3411 *) Update OCSP API. Remove obsolete extensions argument from
3412 various functions. Extensions are now handled using the new
3413 OCSP extension code. New simple OCSP HTTP function which
3414 can be used to send requests and parse the response.
3415 [Steve Henson]
3416
3417 *) Fix the PKCS#7 (S/MIME) code to work with new ASN1. Two new
3418 ASN1_ITEM structures help with sign and verify. PKCS7_ATTR_SIGN
3419 uses the special reorder version of SET OF to sort the attributes
3420 and reorder them to match the encoded order. This resolves a long
3421 standing problem: a verify on a PKCS7 structure just after signing
3422 it used to fail because the attribute order did not match the
3423 encoded order. PKCS7_ATTR_VERIFY does not reorder the attributes:
3424 it uses the received order. This is necessary to tolerate some broken
3425 software that does not order SET OF. This is handled by encoding
3426 as a SEQUENCE OF but using implicit tagging (with UNIVERSAL class)
3427 to produce the required SET OF.
3428 [Steve Henson]
3429
3430 *) Have mk1mf.pl generate the macros OPENSSL_BUILD_SHLIBCRYPTO and
3431 OPENSSL_BUILD_SHLIBSSL and use them appropriately in the header
3432 files to get correct declarations of the ASN.1 item variables.
3433 [Richard Levitte]
3434
3435 *) Rewrite of PKCS#12 code to use new ASN1 functionality. Replace many
3436 PKCS#12 macros with real functions. Fix two unrelated ASN1 bugs:
3437 asn1_check_tlen() would sometimes attempt to use 'ctx' when it was
3438 NULL and ASN1_TYPE was not dereferenced properly in asn1_ex_c2i().
3439 New ASN1 macro: DECLARE_ASN1_ITEM() which just declares the relevant
3440 ASN1_ITEM and no wrapper functions.
3441 [Steve Henson]
3442
3443 *) New functions or ASN1_item_d2i_fp() and ASN1_item_d2i_bio(). These
3444 replace the old function pointer based I/O routines. Change most of
3445 the *_d2i_bio() and *_d2i_fp() functions to use these.
3446 [Steve Henson]
3447
3448 *) Enhance mkdef.pl to be more accepting about spacing in C preprocessor
3449 lines, recognice more "algorithms" that can be deselected, and make
3450 it complain about algorithm deselection that isn't recognised.
3451 [Richard Levitte]
3452
3453 *) New ASN1 functions to handle dup, sign, verify, digest, pack and
3454 unpack operations in terms of ASN1_ITEM. Modify existing wrappers
3455 to use new functions. Add NO_ASN1_OLD which can be set to remove
3456 some old style ASN1 functions: this can be used to determine if old
3457 code will still work when these eventually go away.
3458 [Steve Henson]
3459
3460 *) New extension functions for OCSP structures, these follow the
3461 same conventions as certificates and CRLs.
3462 [Steve Henson]
3463
3464 *) New function X509V3_add1_i2d(). This automatically encodes and
3465 adds an extension. Its behaviour can be customised with various
3466 flags to append, replace or delete. Various wrappers added for
3467 certifcates and CRLs.
3468 [Steve Henson]
3469
3470 *) Fix to avoid calling the underlying ASN1 print routine when
3471 an extension cannot be parsed. Correct a typo in the
3472 OCSP_SERVICELOC extension. Tidy up print OCSP format.
3473 [Steve Henson]
3474
3475 *) Make mkdef.pl parse some of the ASN1 macros and add apropriate
3476 entries for variables.
3477 [Steve Henson]
3478
3479 *) Add functionality to apps/openssl.c for detecting locking
3480 problems: As the program is single-threaded, all we have
3481 to do is register a locking callback using an array for
3482 storing which locks are currently held by the program.
3483 [Bodo Moeller]
3484
3485 *) Use a lock around the call to CRYPTO_get_ex_new_index() in
3486 SSL_get_ex_data_X509_STORE_idx(), which is used in
3487 ssl_verify_cert_chain() and thus can be called at any time
3488 during TLS/SSL handshakes so that thread-safety is essential.
3489 Unfortunately, the ex_data design is not at all suited
3490 for multi-threaded use, so it probably should be abolished.
3491 [Bodo Moeller]
3492
3493 *) Added Broadcom "ubsec" ENGINE to OpenSSL.
3494 [Broadcom, tweaked and integrated by Geoff Thorpe]
3495
3496 *) Move common extension printing code to new function
3497 X509V3_print_extensions(). Reorganise OCSP print routines and
3498 implement some needed OCSP ASN1 functions. Add OCSP extensions.
3499 [Steve Henson]
3500
3501 *) New function X509_signature_print() to remove duplication in some
3502 print routines.
3503 [Steve Henson]
3504
3505 *) Add a special meaning when SET OF and SEQUENCE OF flags are both
3506 set (this was treated exactly the same as SET OF previously). This
3507 is used to reorder the STACK representing the structure to match the
3508 encoding. This will be used to get round a problem where a PKCS7
3509 structure which was signed could not be verified because the STACK
3510 order did not reflect the encoded order.
3511 [Steve Henson]
3512
3513 *) Reimplement the OCSP ASN1 module using the new code.
3514 [Steve Henson]
3515
3516 *) Update the X509V3 code to permit the use of an ASN1_ITEM structure
3517 for its ASN1 operations. The old style function pointers still exist
3518 for now but they will eventually go away.
3519 [Steve Henson]
3520
3521 *) Merge in replacement ASN1 code from the ASN1 branch. This almost
3522 completely replaces the old ASN1 functionality with a table driven
3523 encoder and decoder which interprets an ASN1_ITEM structure describing
3524 the ASN1 module. Compatibility with the existing ASN1 API (i2d,d2i) is
3525 largely maintained. Almost all of the old asn1_mac.h macro based ASN1
3526 has also been converted to the new form.
3527 [Steve Henson]
3528
3529 *) Change BN_mod_exp_recp so that negative moduli are tolerated
3530 (the sign is ignored). Similarly, ignore the sign in BN_MONT_CTX_set
3531 so that BN_mod_exp_mont and BN_mod_exp_mont_word work
3532 for negative moduli.
3533 [Bodo Moeller]
3534
3535 *) Fix BN_uadd and BN_usub: Always return non-negative results instead
3536 of not touching the result's sign bit.
3537 [Bodo Moeller]
3538
3539 *) BN_div bugfix: If the result is 0, the sign (res->neg) must not be
3540 set.
3541 [Bodo Moeller]
3542
3543 *) Changed the LHASH code to use prototypes for callbacks, and created
3544 macros to declare and implement thin (optionally static) functions
3545 that provide type-safety and avoid function pointer casting for the
3546 type-specific callbacks.
3547 [Geoff Thorpe]
3548
3549 *) Added Kerberos Cipher Suites to be used with TLS, as written in
3550 RFC 2712.
3551 [Veers Staats <staatsvr@asc.hpc.mil>,
3552 Jeffrey Altman <jaltman@columbia.edu>, via Richard Levitte]
3553
3554 *) Reformat the FAQ so the different questions and answers can be divided
3555 in sections depending on the subject.
3556 [Richard Levitte]
3557
3558 *) Have the zlib compression code load ZLIB.DLL dynamically under
3559 Windows.
3560 [Richard Levitte]
3561
3562 *) New function BN_mod_sqrt for computing square roots modulo a prime
3563 (using the probabilistic Tonelli-Shanks algorithm unless
3564 p == 3 (mod 4) or p == 5 (mod 8), which are cases that can
3565 be handled deterministically).
3566 [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller]
3567
3568 *) Make BN_mod_inverse faster by explicitly handling small quotients
3569 in the Euclid loop. (Speed gain about 20% for small moduli [256 or
3570 512 bits], about 30% for larger ones [1024 or 2048 bits].)
3571 [Bodo Moeller]
3572
3573 *) New function BN_kronecker.
3574 [Bodo Moeller]
3575
3576 *) Fix BN_gcd so that it works on negative inputs; the result is
3577 positive unless both parameters are zero.
3578 Previously something reasonably close to an infinite loop was
3579 possible because numbers could be growing instead of shrinking
3580 in the implementation of Euclid's algorithm.
3581 [Bodo Moeller]
3582
3583 *) Fix BN_is_word() and BN_is_one() macros to take into account the
3584 sign of the number in question.
3585
3586 Fix BN_is_word(a,w) to work correctly for w == 0.
3587
3588 The old BN_is_word(a,w) macro is now called BN_abs_is_word(a,w)
3589 because its test if the absolute value of 'a' equals 'w'.
3590 Note that BN_abs_is_word does *not* handle w == 0 reliably;
3591 it exists mostly for use in the implementations of BN_is_zero(),
3592 BN_is_one(), and BN_is_word().
3593 [Bodo Moeller]
3594
3595 *) New function BN_swap.
3596 [Bodo Moeller]
3597
3598 *) Use BN_nnmod instead of BN_mod in crypto/bn/bn_exp.c so that
3599 the exponentiation functions are more likely to produce reasonable
3600 results on negative inputs.
3601 [Bodo Moeller]
3602
3603 *) Change BN_mod_mul so that the result is always non-negative.
3604 Previously, it could be negative if one of the factors was negative;
3605 I don't think anyone really wanted that behaviour.
3606 [Bodo Moeller]
3607
3608 *) Move BN_mod_... functions into new file crypto/bn/bn_mod.c
3609 (except for exponentiation, which stays in crypto/bn/bn_exp.c,
3610 and BN_mod_mul_reciprocal, which stays in crypto/bn/bn_recp.c)
3611 and add new functions:
3612
3613 BN_nnmod
3614 BN_mod_sqr
3615 BN_mod_add
3616 BN_mod_add_quick
3617 BN_mod_sub
3618 BN_mod_sub_quick
3619 BN_mod_lshift1
3620 BN_mod_lshift1_quick
3621 BN_mod_lshift
3622 BN_mod_lshift_quick
3623
3624 These functions always generate non-negative results.
3625
3626 BN_nnmod otherwise is like BN_mod (if BN_mod computes a remainder r
3627 such that |m| < r < 0, BN_nnmod will output rem + |m| instead).
3628
3629 BN_mod_XXX_quick(r, a, [b,] m) generates the same result as
3630 BN_mod_XXX(r, a, [b,] m, ctx), but requires that a [and b]
3631 be reduced modulo m.
3632 [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller]
3633
3634 #if 0
3635 The following entry accidentily appeared in the CHANGES file
3636 distributed with OpenSSL 0.9.7. The modifications described in
3637 it do *not* apply to OpenSSL 0.9.7.
3638
3639 *) Remove a few calls to bn_wexpand() in BN_sqr() (the one in there
3640 was actually never needed) and in BN_mul(). The removal in BN_mul()
3641 required a small change in bn_mul_part_recursive() and the addition
3642 of the functions bn_cmp_part_words(), bn_sub_part_words() and
3643 bn_add_part_words(), which do the same thing as bn_cmp_words(),
3644 bn_sub_words() and bn_add_words() except they take arrays with
3645 differing sizes.
3646 [Richard Levitte]
3647 #endif
3648
3649 *) In 'openssl passwd', verify passwords read from the terminal
3650 unless the '-salt' option is used (which usually means that
3651 verification would just waste user's time since the resulting
3652 hash is going to be compared with some given password hash)
3653 or the new '-noverify' option is used.
3654
3655 This is an incompatible change, but it does not affect
3656 non-interactive use of 'openssl passwd' (passwords on the command
3657 line, '-stdin' option, '-in ...' option) and thus should not
3658 cause any problems.
3659 [Bodo Moeller]
3660
3661 *) Remove all references to RSAref, since there's no more need for it.
3662 [Richard Levitte]
3663
3664 *) Make DSO load along a path given through an environment variable
3665 (SHLIB_PATH) with shl_load().
3666 [Richard Levitte]
3667
3668 *) Constify the ENGINE code as a result of BIGNUM constification.
3669 Also constify the RSA code and most things related to it. In a
3670 few places, most notable in the depth of the ASN.1 code, ugly
3671 casts back to non-const were required (to be solved at a later
3672 time)
3673 [Richard Levitte]
3674
3675 *) Make it so the openssl application has all engines loaded by default.
3676 [Richard Levitte]
3677
3678 *) Constify the BIGNUM routines a little more.
3679 [Richard Levitte]
3680
3681 *) Add the following functions:
3682
3683 ENGINE_load_cswift()
3684 ENGINE_load_chil()
3685 ENGINE_load_atalla()
3686 ENGINE_load_nuron()
3687 ENGINE_load_builtin_engines()
3688
3689 That way, an application can itself choose if external engines that
3690 are built-in in OpenSSL shall ever be used or not. The benefit is
3691 that applications won't have to be linked with libdl or other dso
3692 libraries unless it's really needed.
3693
3694 Changed 'openssl engine' to load all engines on demand.
3695 Changed the engine header files to avoid the duplication of some
3696 declarations (they differed!).
3697 [Richard Levitte]
3698
3699 *) 'openssl engine' can now list capabilities.
3700 [Richard Levitte]
3701
3702 *) Better error reporting in 'openssl engine'.
3703 [Richard Levitte]
3704
3705 *) Never call load_dh_param(NULL) in s_server.
3706 [Bodo Moeller]
3707
3708 *) Add engine application. It can currently list engines by name and
3709 identity, and test if they are actually available.
3710 [Richard Levitte]
3711
3712 *) Improve RPM specification file by forcing symbolic linking and making
3713 sure the installed documentation is also owned by root.root.
3714 [Damien Miller <djm@mindrot.org>]
3715
3716 *) Give the OpenSSL applications more possibilities to make use of
3717 keys (public as well as private) handled by engines.
3718 [Richard Levitte]
3719
3720 *) Add OCSP code that comes from CertCo.
3721 [Richard Levitte]
3722
3723 *) Add VMS support for the Rijndael code.
3724 [Richard Levitte]
3725
3726 *) Added untested support for Nuron crypto accelerator.
3727 [Ben Laurie]
3728
3729 *) Add support for external cryptographic devices. This code was
3730 previously distributed separately as the "engine" branch.
3731 [Geoff Thorpe, Richard Levitte]
3732
3733 *) Rework the filename-translation in the DSO code. It is now possible to
3734 have far greater control over how a "name" is turned into a filename
3735 depending on the operating environment and any oddities about the
3736 different shared library filenames on each system.
3737 [Geoff Thorpe]
3738
3739 *) Support threads on FreeBSD-elf in Configure.
3740 [Richard Levitte]
3741
3742 *) Fix for SHA1 assembly problem with MASM: it produces
3743 warnings about corrupt line number information when assembling
3744 with debugging information. This is caused by the overlapping
3745 of two sections.
3746 [Bernd Matthes <mainbug@celocom.de>, Steve Henson]
3747
3748 *) NCONF changes.
3749 NCONF_get_number() has no error checking at all. As a replacement,
3750 NCONF_get_number_e() is defined (_e for "error checking") and is
3751 promoted strongly. The old NCONF_get_number is kept around for
3752 binary backward compatibility.
3753 Make it possible for methods to load from something other than a BIO,
3754 by providing a function pointer that is given a name instead of a BIO.
3755 For example, this could be used to load configuration data from an
3756 LDAP server.
3757 [Richard Levitte]
3758
3759 *) Fix for non blocking accept BIOs. Added new I/O special reason
3760 BIO_RR_ACCEPT to cover this case. Previously use of accept BIOs
3761 with non blocking I/O was not possible because no retry code was
3762 implemented. Also added new SSL code SSL_WANT_ACCEPT to cover
3763 this case.
3764 [Steve Henson]
3765
3766 *) Added the beginnings of Rijndael support.
3767 [Ben Laurie]
3768
3769 *) Fix for bug in DirectoryString mask setting. Add support for
3770 X509_NAME_print_ex() in 'req' and X509_print_ex() function
3771 to allow certificate printing to more controllable, additional
3772 'certopt' option to 'x509' to allow new printing options to be
3773 set.
3774 [Steve Henson]
3775
3776 *) Clean old EAY MD5 hack from e_os.h.
3777 [Richard Levitte]
3778
3779 Changes between 0.9.6l and 0.9.6m [17 Mar 2004]
3780
3781 *) Fix null-pointer assignment in do_change_cipher_spec() revealed
3782 by using the Codenomicon TLS Test Tool (CVE-2004-0079)
3783 [Joe Orton, Steve Henson]
3784
3785 Changes between 0.9.6k and 0.9.6l [04 Nov 2003]
3786
3787 *) Fix additional bug revealed by the NISCC test suite:
3788
3789 Stop bug triggering large recursion when presented with
3790 certain ASN.1 tags (CVE-2003-0851)
3791 [Steve Henson]
3792
3793 Changes between 0.9.6j and 0.9.6k [30 Sep 2003]
3794
3795 *) Fix various bugs revealed by running the NISCC test suite:
3796
3797 Stop out of bounds reads in the ASN1 code when presented with
3798 invalid tags (CVE-2003-0543 and CVE-2003-0544).
3799
3800 If verify callback ignores invalid public key errors don't try to check
3801 certificate signature with the NULL public key.
3802
3803 [Steve Henson]
3804
3805 *) In ssl3_accept() (ssl/s3_srvr.c) only accept a client certificate
3806 if the server requested one: as stated in TLS 1.0 and SSL 3.0
3807 specifications.
3808 [Steve Henson]
3809
3810 *) In ssl3_get_client_hello() (ssl/s3_srvr.c), tolerate additional
3811 extra data after the compression methods not only for TLS 1.0
3812 but also for SSL 3.0 (as required by the specification).
3813 [Bodo Moeller; problem pointed out by Matthias Loepfe]
3814
3815 *) Change X509_certificate_type() to mark the key as exported/exportable
3816 when it's 512 *bits* long, not 512 bytes.
3817 [Richard Levitte]
3818
3819 Changes between 0.9.6i and 0.9.6j [10 Apr 2003]
3820
3821 *) Countermeasure against the Klima-Pokorny-Rosa extension of
3822 Bleichbacher's attack on PKCS #1 v1.5 padding: treat
3823 a protocol version number mismatch like a decryption error
3824 in ssl3_get_client_key_exchange (ssl/s3_srvr.c).
3825 [Bodo Moeller]
3826
3827 *) Turn on RSA blinding by default in the default implementation
3828 to avoid a timing attack. Applications that don't want it can call
3829 RSA_blinding_off() or use the new flag RSA_FLAG_NO_BLINDING.
3830 They would be ill-advised to do so in most cases.
3831 [Ben Laurie, Steve Henson, Geoff Thorpe, Bodo Moeller]
3832
3833 *) Change RSA blinding code so that it works when the PRNG is not
3834 seeded (in this case, the secret RSA exponent is abused as
3835 an unpredictable seed -- if it is not unpredictable, there
3836 is no point in blinding anyway). Make RSA blinding thread-safe
3837 by remembering the creator's thread ID in rsa->blinding and
3838 having all other threads use local one-time blinding factors
3839 (this requires more computation than sharing rsa->blinding, but
3840 avoids excessive locking; and if an RSA object is not shared
3841 between threads, blinding will still be very fast).
3842 [Bodo Moeller]
3843
3844 Changes between 0.9.6h and 0.9.6i [19 Feb 2003]
3845
3846 *) In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked
3847 via timing by performing a MAC computation even if incorrrect
3848 block cipher padding has been found. This is a countermeasure
3849 against active attacks where the attacker has to distinguish
3850 between bad padding and a MAC verification error. (CVE-2003-0078)
3851
3852 [Bodo Moeller; problem pointed out by Brice Canvel (EPFL),
3853 Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and
3854 Martin Vuagnoux (EPFL, Ilion)]
3855
3856 Changes between 0.9.6g and 0.9.6h [5 Dec 2002]
3857
3858 *) New function OPENSSL_cleanse(), which is used to cleanse a section of
3859 memory from it's contents. This is done with a counter that will
3860 place alternating values in each byte. This can be used to solve
3861 two issues: 1) the removal of calls to memset() by highly optimizing
3862 compilers, and 2) cleansing with other values than 0, since those can
3863 be read through on certain media, for example a swap space on disk.
3864 [Geoff Thorpe]
3865
3866 *) Bugfix: client side session caching did not work with external caching,
3867 because the session->cipher setting was not restored when reloading
3868 from the external cache. This problem was masked, when
3869 SSL_OP_NETSCAPE_REUSE_CIPHER_CHANGE_BUG (part of SSL_OP_ALL) was set.
3870 (Found by Steve Haslam <steve@araqnid.ddts.net>.)
3871 [Lutz Jaenicke]
3872
3873 *) Fix client_certificate (ssl/s2_clnt.c): The permissible total
3874 length of the REQUEST-CERTIFICATE message is 18 .. 34, not 17 .. 33.
3875 [Zeev Lieber <zeev-l@yahoo.com>]
3876
3877 *) Undo an undocumented change introduced in 0.9.6e which caused
3878 repeated calls to OpenSSL_add_all_ciphers() and
3879 OpenSSL_add_all_digests() to be ignored, even after calling
3880 EVP_cleanup().
3881 [Richard Levitte]
3882
3883 *) Change the default configuration reader to deal with last line not
3884 being properly terminated.
3885 [Richard Levitte]
3886
3887 *) Change X509_NAME_cmp() so it applies the special rules on handling
3888 DN values that are of type PrintableString, as well as RDNs of type
3889 emailAddress where the value has the type ia5String.
3890 [stefank@valicert.com via Richard Levitte]
3891
3892 *) Add a SSL_SESS_CACHE_NO_INTERNAL_STORE flag to take over half
3893 the job SSL_SESS_CACHE_NO_INTERNAL_LOOKUP was inconsistently
3894 doing, define a new flag (SSL_SESS_CACHE_NO_INTERNAL) to be
3895 the bitwise-OR of the two for use by the majority of applications
3896 wanting this behaviour, and update the docs. The documented
3897 behaviour and actual behaviour were inconsistent and had been
3898 changing anyway, so this is more a bug-fix than a behavioural
3899 change.
3900 [Geoff Thorpe, diagnosed by Nadav Har'El]
3901
3902 *) Don't impose a 16-byte length minimum on session IDs in ssl/s3_clnt.c
3903 (the SSL 3.0 and TLS 1.0 specifications allow any length up to 32 bytes).
3904 [Bodo Moeller]
3905
3906 *) Fix initialization code race conditions in
3907 SSLv23_method(), SSLv23_client_method(), SSLv23_server_method(),
3908 SSLv2_method(), SSLv2_client_method(), SSLv2_server_method(),
3909 SSLv3_method(), SSLv3_client_method(), SSLv3_server_method(),
3910 TLSv1_method(), TLSv1_client_method(), TLSv1_server_method(),
3911 ssl2_get_cipher_by_char(),
3912 ssl3_get_cipher_by_char().
3913 [Patrick McCormick <patrick@tellme.com>, Bodo Moeller]
3914
3915 *) Reorder cleanup sequence in SSL_CTX_free(): only remove the ex_data after
3916 the cached sessions are flushed, as the remove_cb() might use ex_data
3917 contents. Bug found by Sam Varshavchik <mrsam@courier-mta.com>
3918 (see [openssl.org #212]).
3919 [Geoff Thorpe, Lutz Jaenicke]
3920
3921 *) Fix typo in OBJ_txt2obj which incorrectly passed the content
3922 length, instead of the encoding length to d2i_ASN1_OBJECT.
3923 [Steve Henson]
3924
3925 Changes between 0.9.6f and 0.9.6g [9 Aug 2002]
3926
3927 *) [In 0.9.6g-engine release:]
3928 Fix crypto/engine/vendor_defns/cswift.h for WIN32 (use '_stdcall').
3929 [Lynn Gazis <lgazis@rainbow.com>]
3930
3931 Changes between 0.9.6e and 0.9.6f [8 Aug 2002]
3932
3933 *) Fix ASN1 checks. Check for overflow by comparing with LONG_MAX
3934 and get fix the header length calculation.
3935 [Florian Weimer <Weimer@CERT.Uni-Stuttgart.DE>,
3936 Alon Kantor <alonk@checkpoint.com> (and others),
3937 Steve Henson]
3938
3939 *) Use proper error handling instead of 'assertions' in buffer
3940 overflow checks added in 0.9.6e. This prevents DoS (the
3941 assertions could call abort()).
3942 [Arne Ansper <arne@ats.cyber.ee>, Bodo Moeller]
3943
3944 Changes between 0.9.6d and 0.9.6e [30 Jul 2002]
3945
3946 *) Add various sanity checks to asn1_get_length() to reject
3947 the ASN1 length bytes if they exceed sizeof(long), will appear
3948 negative or the content length exceeds the length of the
3949 supplied buffer.
3950 [Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>]
3951
3952 *) Fix cipher selection routines: ciphers without encryption had no flags
3953 for the cipher strength set and where therefore not handled correctly
3954 by the selection routines (PR #130).
3955 [Lutz Jaenicke]
3956
3957 *) Fix EVP_dsa_sha macro.
3958 [Nils Larsch]
3959
3960 *) New option
3961 SSL_OP_DONT_INSERT_EMPTY_FRAGMENTS
3962 for disabling the SSL 3.0/TLS 1.0 CBC vulnerability countermeasure
3963 that was added in OpenSSL 0.9.6d.
3964
3965 As the countermeasure turned out to be incompatible with some
3966 broken SSL implementations, the new option is part of SSL_OP_ALL.
3967 SSL_OP_ALL is usually employed when compatibility with weird SSL
3968 implementations is desired (e.g. '-bugs' option to 's_client' and
3969 's_server'), so the new option is automatically set in many
3970 applications.
3971 [Bodo Moeller]
3972
3973 *) Changes in security patch:
3974
3975 Changes marked "(CHATS)" were sponsored by the Defense Advanced
3976 Research Projects Agency (DARPA) and Air Force Research Laboratory,
3977 Air Force Materiel Command, USAF, under agreement number
3978 F30602-01-2-0537.
3979
3980 *) Add various sanity checks to asn1_get_length() to reject
3981 the ASN1 length bytes if they exceed sizeof(long), will appear
3982 negative or the content length exceeds the length of the
3983 supplied buffer. (CVE-2002-0659)
3984 [Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>]
3985
3986 *) Assertions for various potential buffer overflows, not known to
3987 happen in practice.
3988 [Ben Laurie (CHATS)]
3989
3990 *) Various temporary buffers to hold ASCII versions of integers were
3991 too small for 64 bit platforms. (CVE-2002-0655)
3992 [Matthew Byng-Maddick <mbm@aldigital.co.uk> and Ben Laurie (CHATS)>
3993
3994 *) Remote buffer overflow in SSL3 protocol - an attacker could
3995 supply an oversized session ID to a client. (CVE-2002-0656)
3996 [Ben Laurie (CHATS)]
3997
3998 *) Remote buffer overflow in SSL2 protocol - an attacker could
3999 supply an oversized client master key. (CVE-2002-0656)
4000 [Ben Laurie (CHATS)]
4001
4002 Changes between 0.9.6c and 0.9.6d [9 May 2002]
4003
4004 *) Fix crypto/asn1/a_sign.c so that 'parameters' is omitted (not
4005 encoded as NULL) with id-dsa-with-sha1.
4006 [Nils Larsch <nla@trustcenter.de>; problem pointed out by Bodo Moeller]
4007
4008 *) Check various X509_...() return values in apps/req.c.
4009 [Nils Larsch <nla@trustcenter.de>]
4010
4011 *) Fix BASE64 decode (EVP_DecodeUpdate) for data with CR/LF ended lines:
4012 an end-of-file condition would erronously be flagged, when the CRLF
4013 was just at the end of a processed block. The bug was discovered when
4014 processing data through a buffering memory BIO handing the data to a
4015 BASE64-decoding BIO. Bug fund and patch submitted by Pavel Tsekov
4016 <ptsekov@syntrex.com> and Nedelcho Stanev.
4017 [Lutz Jaenicke]
4018
4019 *) Implement a countermeasure against a vulnerability recently found
4020 in CBC ciphersuites in SSL 3.0/TLS 1.0: Send an empty fragment
4021 before application data chunks to avoid the use of known IVs
4022 with data potentially chosen by the attacker.
4023 [Bodo Moeller]
4024
4025 *) Fix length checks in ssl3_get_client_hello().
4026 [Bodo Moeller]
4027
4028 *) TLS/SSL library bugfix: use s->s3->in_read_app_data differently
4029 to prevent ssl3_read_internal() from incorrectly assuming that
4030 ssl3_read_bytes() found application data while handshake
4031 processing was enabled when in fact s->s3->in_read_app_data was
4032 merely automatically cleared during the initial handshake.
4033 [Bodo Moeller; problem pointed out by Arne Ansper <arne@ats.cyber.ee>]
4034
4035 *) Fix object definitions for Private and Enterprise: they were not
4036 recognized in their shortname (=lowercase) representation. Extend
4037 obj_dat.pl to issue an error when using undefined keywords instead
4038 of silently ignoring the problem (Svenning Sorensen
4039 <sss@sss.dnsalias.net>).
4040 [Lutz Jaenicke]
4041
4042 *) Fix DH_generate_parameters() so that it works for 'non-standard'
4043 generators, i.e. generators other than 2 and 5. (Previously, the
4044 code did not properly initialise the 'add' and 'rem' values to
4045 BN_generate_prime().)
4046
4047 In the new general case, we do not insist that 'generator' is
4048 actually a primitive root: This requirement is rather pointless;
4049 a generator of the order-q subgroup is just as good, if not
4050 better.
4051 [Bodo Moeller]
4052
4053 *) Map new X509 verification errors to alerts. Discovered and submitted by
4054 Tom Wu <tom@arcot.com>.
4055 [Lutz Jaenicke]
4056
4057 *) Fix ssl3_pending() (ssl/s3_lib.c) to prevent SSL_pending() from
4058 returning non-zero before the data has been completely received
4059 when using non-blocking I/O.
4060 [Bodo Moeller; problem pointed out by John Hughes]
4061
4062 *) Some of the ciphers missed the strength entry (SSL_LOW etc).
4063 [Ben Laurie, Lutz Jaenicke]
4064
4065 *) Fix bug in SSL_clear(): bad sessions were not removed (found by
4066 Yoram Zahavi <YoramZ@gilian.com>).
4067 [Lutz Jaenicke]
4068
4069 *) Add information about CygWin 1.3 and on, and preserve proper
4070 configuration for the versions before that.
4071 [Corinna Vinschen <vinschen@redhat.com> and Richard Levitte]
4072
4073 *) Make removal from session cache (SSL_CTX_remove_session()) more robust:
4074 check whether we deal with a copy of a session and do not delete from
4075 the cache in this case. Problem reported by "Izhar Shoshani Levi"
4076 <izhar@checkpoint.com>.
4077 [Lutz Jaenicke]
4078
4079 *) Do not store session data into the internal session cache, if it
4080 is never intended to be looked up (SSL_SESS_CACHE_NO_INTERNAL_LOOKUP
4081 flag is set). Proposed by Aslam <aslam@funk.com>.
4082 [Lutz Jaenicke]
4083
4084 *) Have ASN1_BIT_STRING_set_bit() really clear a bit when the requested
4085 value is 0.
4086 [Richard Levitte]
4087
4088 *) [In 0.9.6d-engine release:]
4089 Fix a crashbug and a logic bug in hwcrhk_load_pubkey().
4090 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte]
4091
4092 *) Add the configuration target linux-s390x.
4093 [Neale Ferguson <Neale.Ferguson@SoftwareAG-USA.com> via Richard Levitte]
4094
4095 *) The earlier bugfix for the SSL3_ST_SW_HELLO_REQ_C case of
4096 ssl3_accept (ssl/s3_srvr.c) incorrectly used a local flag
4097 variable as an indication that a ClientHello message has been
4098 received. As the flag value will be lost between multiple
4099 invocations of ssl3_accept when using non-blocking I/O, the
4100 function may not be aware that a handshake has actually taken
4101 place, thus preventing a new session from being added to the
4102 session cache.
4103
4104 To avoid this problem, we now set s->new_session to 2 instead of
4105 using a local variable.
4106 [Lutz Jaenicke, Bodo Moeller]
4107
4108 *) Bugfix: Return -1 from ssl3_get_server_done (ssl3/s3_clnt.c)
4109 if the SSL_R_LENGTH_MISMATCH error is detected.
4110 [Geoff Thorpe, Bodo Moeller]
4111
4112 *) New 'shared_ldflag' column in Configure platform table.
4113 [Richard Levitte]
4114
4115 *) Fix EVP_CIPHER_mode macro.
4116 ["Dan S. Camper" <dan@bti.net>]
4117
4118 *) Fix ssl3_read_bytes (ssl/s3_pkt.c): To ignore messages of unknown
4119 type, we must throw them away by setting rr->length to 0.
4120 [D P Chang <dpc@qualys.com>]
4121
4122 Changes between 0.9.6b and 0.9.6c [21 dec 2001]
4123
4124 *) Fix BN_rand_range bug pointed out by Dominikus Scherkl
4125 <Dominikus.Scherkl@biodata.com>. (The previous implementation
4126 worked incorrectly for those cases where range = 10..._2 and
4127 3*range is two bits longer than range.)
4128 [Bodo Moeller]
4129
4130 *) Only add signing time to PKCS7 structures if it is not already
4131 present.
4132 [Steve Henson]
4133
4134 *) Fix crypto/objects/objects.h: "ld-ce" should be "id-ce",
4135 OBJ_ld_ce should be OBJ_id_ce.
4136 Also some ip-pda OIDs in crypto/objects/objects.txt were
4137 incorrect (cf. RFC 3039).
4138 [Matt Cooper, Frederic Giudicelli, Bodo Moeller]
4139
4140 *) Release CRYPTO_LOCK_DYNLOCK when CRYPTO_destroy_dynlockid()
4141 returns early because it has nothing to do.
4142 [Andy Schneider <andy.schneider@bjss.co.uk>]
4143
4144 *) [In 0.9.6c-engine release:]
4145 Fix mutex callback return values in crypto/engine/hw_ncipher.c.
4146 [Andy Schneider <andy.schneider@bjss.co.uk>]
4147
4148 *) [In 0.9.6c-engine release:]
4149 Add support for Cryptographic Appliance's keyserver technology.
4150 (Use engine 'keyclient')
4151 [Cryptographic Appliances and Geoff Thorpe]
4152
4153 *) Add a configuration entry for OS/390 Unix. The C compiler 'c89'
4154 is called via tools/c89.sh because arguments have to be
4155 rearranged (all '-L' options must appear before the first object
4156 modules).
4157 [Richard Shapiro <rshapiro@abinitio.com>]
4158
4159 *) [In 0.9.6c-engine release:]
4160 Add support for Broadcom crypto accelerator cards, backported
4161 from 0.9.7.
4162 [Broadcom, Nalin Dahyabhai <nalin@redhat.com>, Mark Cox]
4163
4164 *) [In 0.9.6c-engine release:]
4165 Add support for SureWare crypto accelerator cards from
4166 Baltimore Technologies. (Use engine 'sureware')
4167 [Baltimore Technologies and Mark Cox]
4168
4169 *) [In 0.9.6c-engine release:]
4170 Add support for crypto accelerator cards from Accelerated
4171 Encryption Processing, www.aep.ie. (Use engine 'aep')
4172 [AEP Inc. and Mark Cox]
4173
4174 *) Add a configuration entry for gcc on UnixWare.
4175 [Gary Benson <gbenson@redhat.com>]
4176
4177 *) Change ssl/s2_clnt.c and ssl/s2_srvr.c so that received handshake
4178 messages are stored in a single piece (fixed-length part and
4179 variable-length part combined) and fix various bugs found on the way.
4180 [Bodo Moeller]
4181
4182 *) Disable caching in BIO_gethostbyname(), directly use gethostbyname()
4183 instead. BIO_gethostbyname() does not know what timeouts are
4184 appropriate, so entries would stay in cache even when they have
4185 become invalid.
4186 [Bodo Moeller; problem pointed out by Rich Salz <rsalz@zolera.com>
4187
4188 *) Change ssl23_get_client_hello (ssl/s23_srvr.c) behaviour when
4189 faced with a pathologically small ClientHello fragment that does
4190 not contain client_version: Instead of aborting with an error,
4191 simply choose the highest available protocol version (i.e.,
4192 TLS 1.0 unless it is disabled). In practice, ClientHello
4193 messages are never sent like this, but this change gives us
4194 strictly correct behaviour at least for TLS.
4195 [Bodo Moeller]
4196
4197 *) Fix SSL handshake functions and SSL_clear() such that SSL_clear()
4198 never resets s->method to s->ctx->method when called from within
4199 one of the SSL handshake functions.
4200 [Bodo Moeller; problem pointed out by Niko Baric]
4201
4202 *) In ssl3_get_client_hello (ssl/s3_srvr.c), generate a fatal alert
4203 (sent using the client's version number) if client_version is
4204 smaller than the protocol version in use. Also change
4205 ssl23_get_client_hello (ssl/s23_srvr.c) to select TLS 1.0 if
4206 the client demanded SSL 3.0 but only TLS 1.0 is enabled; then
4207 the client will at least see that alert.
4208 [Bodo Moeller]
4209
4210 *) Fix ssl3_get_message (ssl/s3_both.c) to handle message fragmentation
4211 correctly.
4212 [Bodo Moeller]
4213
4214 *) Avoid infinite loop in ssl3_get_message (ssl/s3_both.c) if a
4215 client receives HelloRequest while in a handshake.
4216 [Bodo Moeller; bug noticed by Andy Schneider <andy.schneider@bjss.co.uk>]
4217
4218 *) Bugfix in ssl3_accept (ssl/s3_srvr.c): Case SSL3_ST_SW_HELLO_REQ_C
4219 should end in 'break', not 'goto end' which circuments various
4220 cleanups done in state SSL_ST_OK. But session related stuff
4221 must be disabled for SSL_ST_OK in the case that we just sent a
4222 HelloRequest.
4223
4224 Also avoid some overhead by not calling ssl_init_wbio_buffer()
4225 before just sending a HelloRequest.
4226 [Bodo Moeller, Eric Rescorla <ekr@rtfm.com>]
4227
4228 *) Fix ssl/s3_enc.c, ssl/t1_enc.c and ssl/s3_pkt.c so that we don't
4229 reveal whether illegal block cipher padding was found or a MAC
4230 verification error occured. (Neither SSLerr() codes nor alerts
4231 are directly visible to potential attackers, but the information
4232 may leak via logfiles.)
4233
4234 Similar changes are not required for the SSL 2.0 implementation
4235 because the number of padding bytes is sent in clear for SSL 2.0,
4236 and the extra bytes are just ignored. However ssl/s2_pkt.c
4237 failed to verify that the purported number of padding bytes is in
4238 the legal range.
4239 [Bodo Moeller]
4240
4241 *) Add OpenUNIX-8 support including shared libraries
4242 (Boyd Lynn Gerber <gerberb@zenez.com>).
4243 [Lutz Jaenicke]
4244
4245 *) Improve RSA_padding_check_PKCS1_OAEP() check again to avoid
4246 'wristwatch attack' using huge encoding parameters (cf.
4247 James H. Manger's CRYPTO 2001 paper). Note that the
4248 RSA_PKCS1_OAEP_PADDING case of RSA_private_decrypt() does not use
4249 encoding parameters and hence was not vulnerable.
4250 [Bodo Moeller]
4251
4252 *) BN_sqr() bug fix.
4253 [Ulf Möller, reported by Jim Ellis <jim.ellis@cavium.com>]
4254
4255 *) Rabin-Miller test analyses assume uniformly distributed witnesses,
4256 so use BN_pseudo_rand_range() instead of using BN_pseudo_rand()
4257 followed by modular reduction.
4258 [Bodo Moeller; pointed out by Adam Young <AYoung1@NCSUS.JNJ.COM>]
4259
4260 *) Add BN_pseudo_rand_range() with obvious functionality: BN_rand_range()
4261 equivalent based on BN_pseudo_rand() instead of BN_rand().
4262 [Bodo Moeller]
4263
4264 *) s3_srvr.c: allow sending of large client certificate lists (> 16 kB).
4265 This function was broken, as the check for a new client hello message
4266 to handle SGC did not allow these large messages.
4267 (Tracked down by "Douglas E. Engert" <deengert@anl.gov>.)
4268 [Lutz Jaenicke]
4269
4270 *) Add alert descriptions for TLSv1 to SSL_alert_desc_string[_long]().
4271 [Lutz Jaenicke]
4272
4273 *) Fix buggy behaviour of BIO_get_num_renegotiates() and BIO_ctrl()
4274 for BIO_C_GET_WRITE_BUF_SIZE ("Stephen Hinton" <shinton@netopia.com>).
4275 [Lutz Jaenicke]
4276
4277 *) Rework the configuration and shared library support for Tru64 Unix.
4278 The configuration part makes use of modern compiler features and
4279 still retains old compiler behavior for those that run older versions
4280 of the OS. The shared library support part includes a variant that
4281 uses the RPATH feature, and is available through the special
4282 configuration target "alpha-cc-rpath", which will never be selected
4283 automatically.
4284 [Tim Mooney <mooney@dogbert.cc.ndsu.NoDak.edu> via Richard Levitte]
4285
4286 *) In ssl3_get_key_exchange (ssl/s3_clnt.c), call ssl3_get_message()
4287 with the same message size as in ssl3_get_certificate_request().
4288 Otherwise, if no ServerKeyExchange message occurs, CertificateRequest
4289 messages might inadvertently be reject as too long.
4290 [Petr Lampa <lampa@fee.vutbr.cz>]
4291
4292 *) Enhanced support for IA-64 Unix platforms (well, Linux and HP-UX).
4293 [Andy Polyakov]
4294
4295 *) Modified SSL library such that the verify_callback that has been set
4296 specificly for an SSL object with SSL_set_verify() is actually being
4297 used. Before the change, a verify_callback set with this function was
4298 ignored and the verify_callback() set in the SSL_CTX at the time of
4299 the call was used. New function X509_STORE_CTX_set_verify_cb() introduced
4300 to allow the necessary settings.
4301 [Lutz Jaenicke]
4302
4303 *) Initialize static variable in crypto/dsa/dsa_lib.c and crypto/dh/dh_lib.c
4304 explicitly to NULL, as at least on Solaris 8 this seems not always to be
4305 done automatically (in contradiction to the requirements of the C
4306 standard). This made problems when used from OpenSSH.
4307 [Lutz Jaenicke]
4308
4309 *) In OpenSSL 0.9.6a and 0.9.6b, crypto/dh/dh_key.c ignored
4310 dh->length and always used
4311
4312 BN_rand_range(priv_key, dh->p).
4313
4314 BN_rand_range() is not necessary for Diffie-Hellman, and this
4315 specific range makes Diffie-Hellman unnecessarily inefficient if
4316 dh->length (recommended exponent length) is much smaller than the
4317 length of dh->p. We could use BN_rand_range() if the order of
4318 the subgroup was stored in the DH structure, but we only have
4319 dh->length.
4320
4321 So switch back to
4322
4323 BN_rand(priv_key, l, ...)
4324
4325 where 'l' is dh->length if this is defined, or BN_num_bits(dh->p)-1
4326 otherwise.
4327 [Bodo Moeller]
4328
4329 *) In
4330
4331 RSA_eay_public_encrypt
4332 RSA_eay_private_decrypt
4333 RSA_eay_private_encrypt (signing)
4334 RSA_eay_public_decrypt (signature verification)
4335
4336 (default implementations for RSA_public_encrypt,
4337 RSA_private_decrypt, RSA_private_encrypt, RSA_public_decrypt),
4338 always reject numbers >= n.
4339 [Bodo Moeller]
4340
4341 *) In crypto/rand/md_rand.c, use a new short-time lock CRYPTO_LOCK_RAND2
4342 to synchronize access to 'locking_thread'. This is necessary on
4343 systems where access to 'locking_thread' (an 'unsigned long'
4344 variable) is not atomic.
4345 [Bodo Moeller]
4346
4347 *) In crypto/rand/md_rand.c, set 'locking_thread' to current thread's ID
4348 *before* setting the 'crypto_lock_rand' flag. The previous code had
4349 a race condition if 0 is a valid thread ID.
4350 [Travis Vitek <vitek@roguewave.com>]
4351
4352 *) Add support for shared libraries under Irix.
4353 [Albert Chin-A-Young <china@thewrittenword.com>]
4354
4355 *) Add configuration option to build on Linux on both big-endian and
4356 little-endian MIPS.
4357 [Ralf Baechle <ralf@uni-koblenz.de>]
4358
4359 *) Add the possibility to create shared libraries on HP-UX.
4360 [Richard Levitte]
4361
4362 Changes between 0.9.6a and 0.9.6b [9 Jul 2001]
4363
4364 *) Change ssleay_rand_bytes (crypto/rand/md_rand.c)
4365 to avoid a SSLeay/OpenSSL PRNG weakness pointed out by
4366 Markku-Juhani O. Saarinen <markku-juhani.saarinen@nokia.com>:
4367 PRNG state recovery was possible based on the output of
4368 one PRNG request appropriately sized to gain knowledge on
4369 'md' followed by enough consecutive 1-byte PRNG requests
4370 to traverse all of 'state'.
4371
4372 1. When updating 'md_local' (the current thread's copy of 'md')
4373 during PRNG output generation, hash all of the previous
4374 'md_local' value, not just the half used for PRNG output.
4375
4376 2. Make the number of bytes from 'state' included into the hash
4377 independent from the number of PRNG bytes requested.
4378
4379 The first measure alone would be sufficient to avoid
4380 Markku-Juhani's attack. (Actually it had never occurred
4381 to me that the half of 'md_local' used for chaining was the
4382 half from which PRNG output bytes were taken -- I had always
4383 assumed that the secret half would be used.) The second
4384 measure makes sure that additional data from 'state' is never
4385 mixed into 'md_local' in small portions; this heuristically
4386 further strengthens the PRNG.
4387 [Bodo Moeller]
4388
4389 *) Fix crypto/bn/asm/mips3.s.
4390 [Andy Polyakov]
4391
4392 *) When only the key is given to "enc", the IV is undefined. Print out
4393 an error message in this case.
4394 [Lutz Jaenicke]
4395
4396 *) Handle special case when X509_NAME is empty in X509 printing routines.
4397 [Steve Henson]
4398
4399 *) In dsa_do_verify (crypto/dsa/dsa_ossl.c), verify that r and s are
4400 positive and less than q.
4401 [Bodo Moeller]
4402
4403 *) Don't change *pointer in CRYPTO_add_lock() is add_lock_callback is
4404 used: it isn't thread safe and the add_lock_callback should handle
4405 that itself.
4406 [Paul Rose <Paul.Rose@bridge.com>]
4407
4408 *) Verify that incoming data obeys the block size in
4409 ssl3_enc (ssl/s3_enc.c) and tls1_enc (ssl/t1_enc.c).
4410 [Bodo Moeller]
4411
4412 *) Fix OAEP check.
4413 [Ulf Möller, Bodo Möller]
4414
4415 *) The countermeasure against Bleichbacher's attack on PKCS #1 v1.5
4416 RSA encryption was accidentally removed in s3_srvr.c in OpenSSL 0.9.5
4417 when fixing the server behaviour for backwards-compatible 'client
4418 hello' messages. (Note that the attack is impractical against
4419 SSL 3.0 and TLS 1.0 anyway because length and version checking
4420 means that the probability of guessing a valid ciphertext is
4421 around 2^-40; see section 5 in Bleichenbacher's CRYPTO '98
4422 paper.)
4423
4424 Before 0.9.5, the countermeasure (hide the error by generating a
4425 random 'decryption result') did not work properly because
4426 ERR_clear_error() was missing, meaning that SSL_get_error() would
4427 detect the supposedly ignored error.
4428
4429 Both problems are now fixed.
4430 [Bodo Moeller]
4431
4432 *) In crypto/bio/bf_buff.c, increase DEFAULT_BUFFER_SIZE to 4096
4433 (previously it was 1024).
4434 [Bodo Moeller]
4435
4436 *) Fix for compatibility mode trust settings: ignore trust settings
4437 unless some valid trust or reject settings are present.
4438 [Steve Henson]
4439
4440 *) Fix for blowfish EVP: its a variable length cipher.
4441 [Steve Henson]
4442
4443 *) Fix various bugs related to DSA S/MIME verification. Handle missing
4444 parameters in DSA public key structures and return an error in the
4445 DSA routines if parameters are absent.
4446 [Steve Henson]
4447
4448 *) In versions up to 0.9.6, RAND_file_name() resorted to file ".rnd"
4449 in the current directory if neither $RANDFILE nor $HOME was set.
4450 RAND_file_name() in 0.9.6a returned NULL in this case. This has
4451 caused some confusion to Windows users who haven't defined $HOME.
4452 Thus RAND_file_name() is changed again: e_os.h can define a
4453 DEFAULT_HOME, which will be used if $HOME is not set.
4454 For Windows, we use "C:"; on other platforms, we still require
4455 environment variables.
4456
4457 *) Move 'if (!initialized) RAND_poll()' into regions protected by
4458 CRYPTO_LOCK_RAND. This is not strictly necessary, but avoids
4459 having multiple threads call RAND_poll() concurrently.
4460 [Bodo Moeller]
4461
4462 *) In crypto/rand/md_rand.c, replace 'add_do_not_lock' flag by a
4463 combination of a flag and a thread ID variable.
4464 Otherwise while one thread is in ssleay_rand_bytes (which sets the
4465 flag), *other* threads can enter ssleay_add_bytes without obeying
4466 the CRYPTO_LOCK_RAND lock (and may even illegally release the lock
4467 that they do not hold after the first thread unsets add_do_not_lock).
4468 [Bodo Moeller]
4469
4470 *) Change bctest again: '-x' expressions are not available in all
4471 versions of 'test'.
4472 [Bodo Moeller]
4473
4474 Changes between 0.9.6 and 0.9.6a [5 Apr 2001]
4475
4476 *) Fix a couple of memory leaks in PKCS7_dataDecode()
4477 [Steve Henson, reported by Heyun Zheng <hzheng@atdsprint.com>]
4478
4479 *) Change Configure and Makefiles to provide EXE_EXT, which will contain
4480 the default extension for executables, if any. Also, make the perl
4481 scripts that use symlink() to test if it really exists and use "cp"
4482 if it doesn't. All this made OpenSSL compilable and installable in
4483 CygWin.
4484 [Richard Levitte]
4485
4486 *) Fix for asn1_GetSequence() for indefinite length constructed data.
4487 If SEQUENCE is length is indefinite just set c->slen to the total
4488 amount of data available.
4489 [Steve Henson, reported by shige@FreeBSD.org]
4490 [This change does not apply to 0.9.7.]
4491
4492 *) Change bctest to avoid here-documents inside command substitution
4493 (workaround for FreeBSD /bin/sh bug).
4494 For compatibility with Ultrix, avoid shell functions (introduced
4495 in the bctest version that searches along $PATH).
4496 [Bodo Moeller]
4497
4498 *) Rename 'des_encrypt' to 'des_encrypt1'. This avoids the clashes
4499 with des_encrypt() defined on some operating systems, like Solaris
4500 and UnixWare.
4501 [Richard Levitte]
4502
4503 *) Check the result of RSA-CRT (see D. Boneh, R. DeMillo, R. Lipton:
4504 On the Importance of Eliminating Errors in Cryptographic
4505 Computations, J. Cryptology 14 (2001) 2, 101-119,
4506 http://theory.stanford.edu/~dabo/papers/faults.ps.gz).
4507 [Ulf Moeller]
4508
4509 *) MIPS assembler BIGNUM division bug fix.
4510 [Andy Polyakov]
4511
4512 *) Disabled incorrect Alpha assembler code.
4513 [Richard Levitte]
4514
4515 *) Fix PKCS#7 decode routines so they correctly update the length
4516 after reading an EOC for the EXPLICIT tag.
4517 [Steve Henson]
4518 [This change does not apply to 0.9.7.]
4519
4520 *) Fix bug in PKCS#12 key generation routines. This was triggered
4521 if a 3DES key was generated with a 0 initial byte. Include
4522 PKCS12_BROKEN_KEYGEN compilation option to retain the old
4523 (but broken) behaviour.
4524 [Steve Henson]
4525
4526 *) Enhance bctest to search for a working bc along $PATH and print
4527 it when found.
4528 [Tim Rice <tim@multitalents.net> via Richard Levitte]
4529
4530 *) Fix memory leaks in err.c: free err_data string if necessary;
4531 don't write to the wrong index in ERR_set_error_data.
4532 [Bodo Moeller]
4533
4534 *) Implement ssl23_peek (analogous to ssl23_read), which previously
4535 did not exist.
4536 [Bodo Moeller]
4537
4538 *) Replace rdtsc with _emit statements for VC++ version 5.
4539 [Jeremy Cooper <jeremy@baymoo.org>]
4540
4541 *) Make it possible to reuse SSLv2 sessions.
4542 [Richard Levitte]
4543
4544 *) In copy_email() check for >= 0 as a return value for
4545 X509_NAME_get_index_by_NID() since 0 is a valid index.
4546 [Steve Henson reported by Massimiliano Pala <madwolf@opensca.org>]
4547
4548 *) Avoid coredump with unsupported or invalid public keys by checking if
4549 X509_get_pubkey() fails in PKCS7_verify(). Fix memory leak when
4550 PKCS7_verify() fails with non detached data.
4551 [Steve Henson]
4552
4553 *) Don't use getenv in library functions when run as setuid/setgid.
4554 New function OPENSSL_issetugid().
4555 [Ulf Moeller]
4556
4557 *) Avoid false positives in memory leak detection code (crypto/mem_dbg.c)
4558 due to incorrect handling of multi-threading:
4559
4560 1. Fix timing glitch in the MemCheck_off() portion of CRYPTO_mem_ctrl().
4561
4562 2. Fix logical glitch in is_MemCheck_on() aka CRYPTO_is_mem_check_on().
4563
4564 3. Count how many times MemCheck_off() has been called so that
4565 nested use can be treated correctly. This also avoids
4566 inband-signalling in the previous code (which relied on the
4567 assumption that thread ID 0 is impossible).
4568 [Bodo Moeller]
4569
4570 *) Add "-rand" option also to s_client and s_server.
4571 [Lutz Jaenicke]
4572
4573 *) Fix CPU detection on Irix 6.x.
4574 [Kurt Hockenbury <khockenb@stevens-tech.edu> and
4575 "Bruce W. Forsberg" <bruce.forsberg@baesystems.com>]
4576
4577 *) Fix X509_NAME bug which produced incorrect encoding if X509_NAME
4578 was empty.
4579 [Steve Henson]
4580 [This change does not apply to 0.9.7.]
4581
4582 *) Use the cached encoding of an X509_NAME structure rather than
4583 copying it. This is apparently the reason for the libsafe "errors"
4584 but the code is actually correct.
4585 [Steve Henson]
4586
4587 *) Add new function BN_rand_range(), and fix DSA_sign_setup() to prevent
4588 Bleichenbacher's DSA attack.
4589 Extend BN_[pseudo_]rand: As before, top=1 forces the highest two bits
4590 to be set and top=0 forces the highest bit to be set; top=-1 is new
4591 and leaves the highest bit random.
4592 [Ulf Moeller, Bodo Moeller]
4593
4594 *) In the NCONF_...-based implementations for CONF_... queries
4595 (crypto/conf/conf_lib.c), if the input LHASH is NULL, avoid using
4596 a temporary CONF structure with the data component set to NULL
4597 (which gives segmentation faults in lh_retrieve).
4598 Instead, use NULL for the CONF pointer in CONF_get_string and
4599 CONF_get_number (which may use environment variables) and directly
4600 return NULL from CONF_get_section.
4601 [Bodo Moeller]
4602
4603 *) Fix potential buffer overrun for EBCDIC.
4604 [Ulf Moeller]
4605
4606 *) Tolerate nonRepudiation as being valid for S/MIME signing and certSign
4607 keyUsage if basicConstraints absent for a CA.
4608 [Steve Henson]
4609
4610 *) Make SMIME_write_PKCS7() write mail header values with a format that
4611 is more generally accepted (no spaces before the semicolon), since
4612 some programs can't parse those values properly otherwise. Also make
4613 sure BIO's that break lines after each write do not create invalid
4614 headers.
4615 [Richard Levitte]
4616
4617 *) Make the CRL encoding routines work with empty SEQUENCE OF. The
4618 macros previously used would not encode an empty SEQUENCE OF
4619 and break the signature.
4620 [Steve Henson]
4621 [This change does not apply to 0.9.7.]
4622
4623 *) Zero the premaster secret after deriving the master secret in
4624 DH ciphersuites.
4625 [Steve Henson]
4626
4627 *) Add some EVP_add_digest_alias registrations (as found in
4628 OpenSSL_add_all_digests()) to SSL_library_init()
4629 aka OpenSSL_add_ssl_algorithms(). This provides improved
4630 compatibility with peers using X.509 certificates
4631 with unconventional AlgorithmIdentifier OIDs.
4632 [Bodo Moeller]
4633
4634 *) Fix for Irix with NO_ASM.
4635 ["Bruce W. Forsberg" <bruce.forsberg@baesystems.com>]
4636
4637 *) ./config script fixes.
4638 [Ulf Moeller, Richard Levitte]
4639
4640 *) Fix 'openssl passwd -1'.
4641 [Bodo Moeller]
4642
4643 *) Change PKCS12_key_gen_asc() so it can cope with non null
4644 terminated strings whose length is passed in the passlen
4645 parameter, for example from PEM callbacks. This was done
4646 by adding an extra length parameter to asc2uni().
4647 [Steve Henson, reported by <oddissey@samsung.co.kr>]
4648
4649 *) Fix C code generated by 'openssl dsaparam -C': If a BN_bin2bn
4650 call failed, free the DSA structure.
4651 [Bodo Moeller]
4652
4653 *) Fix to uni2asc() to cope with zero length Unicode strings.
4654 These are present in some PKCS#12 files.
4655 [Steve Henson]
4656
4657 *) Increase s2->wbuf allocation by one byte in ssl2_new (ssl/s2_lib.c).
4658 Otherwise do_ssl_write (ssl/s2_pkt.c) will write beyond buffer limits
4659 when writing a 32767 byte record.
4660 [Bodo Moeller; problem reported by Eric Day <eday@concentric.net>]
4661
4662 *) In RSA_eay_public_{en,ed}crypt and RSA_eay_mod_exp (rsa_eay.c),
4663 obtain lock CRYPTO_LOCK_RSA before setting rsa->_method_mod_{n,p,q}.
4664
4665 (RSA objects have a reference count access to which is protected
4666 by CRYPTO_LOCK_RSA [see rsa_lib.c, s3_srvr.c, ssl_cert.c, ssl_rsa.c],
4667 so they are meant to be shared between threads.)
4668 [Bodo Moeller, Geoff Thorpe; original patch submitted by
4669 "Reddie, Steven" <Steven.Reddie@ca.com>]
4670
4671 *) Fix a deadlock in CRYPTO_mem_leaks().
4672 [Bodo Moeller]
4673
4674 *) Use better test patterns in bntest.
4675 [Ulf Möller]
4676
4677 *) rand_win.c fix for Borland C.
4678 [Ulf Möller]
4679
4680 *) BN_rshift bugfix for n == 0.
4681 [Bodo Moeller]
4682
4683 *) Add a 'bctest' script that checks for some known 'bc' bugs
4684 so that 'make test' does not abort just because 'bc' is broken.
4685 [Bodo Moeller]
4686
4687 *) Store verify_result within SSL_SESSION also for client side to
4688 avoid potential security hole. (Re-used sessions on the client side
4689 always resulted in verify_result==X509_V_OK, not using the original
4690 result of the server certificate verification.)
4691 [Lutz Jaenicke]
4692
4693 *) Fix ssl3_pending: If the record in s->s3->rrec is not of type
4694 SSL3_RT_APPLICATION_DATA, return 0.
4695 Similarly, change ssl2_pending to return 0 if SSL_in_init(s) is true.
4696 [Bodo Moeller]
4697
4698 *) Fix SSL_peek:
4699 Both ssl2_peek and ssl3_peek, which were totally broken in earlier
4700 releases, have been re-implemented by renaming the previous
4701 implementations of ssl2_read and ssl3_read to ssl2_read_internal
4702 and ssl3_read_internal, respectively, and adding 'peek' parameters
4703 to them. The new ssl[23]_{read,peek} functions are calls to
4704 ssl[23]_read_internal with the 'peek' flag set appropriately.
4705 A 'peek' parameter has also been added to ssl3_read_bytes, which
4706 does the actual work for ssl3_read_internal.
4707 [Bodo Moeller]
4708
4709 *) Initialise "ex_data" member of RSA/DSA/DH structures prior to calling
4710 the method-specific "init()" handler. Also clean up ex_data after
4711 calling the method-specific "finish()" handler. Previously, this was
4712 happening the other way round.
4713 [Geoff Thorpe]
4714
4715 *) Increase BN_CTX_NUM (the number of BIGNUMs in a BN_CTX) to 16.
4716 The previous value, 12, was not always sufficient for BN_mod_exp().
4717 [Bodo Moeller]
4718
4719 *) Make sure that shared libraries get the internal name engine with
4720 the full version number and not just 0. This should mark the
4721 shared libraries as not backward compatible. Of course, this should
4722 be changed again when we can guarantee backward binary compatibility.
4723 [Richard Levitte]
4724
4725 *) Fix typo in get_cert_by_subject() in by_dir.c
4726 [Jean-Marc Desperrier <jean-marc.desperrier@certplus.com>]
4727
4728 *) Rework the system to generate shared libraries:
4729
4730 - Make note of the expected extension for the shared libraries and
4731 if there is a need for symbolic links from for example libcrypto.so.0
4732 to libcrypto.so.0.9.7. There is extended info in Configure for
4733 that.
4734
4735 - Make as few rebuilds of the shared libraries as possible.
4736
4737 - Still avoid linking the OpenSSL programs with the shared libraries.
4738
4739 - When installing, install the shared libraries separately from the
4740 static ones.
4741 [Richard Levitte]
4742
4743 *) Fix SSL_CTX_set_read_ahead macro to actually use its argument.
4744
4745 Copy SSL_CTX's read_ahead flag to SSL object directly in SSL_new
4746 and not in SSL_clear because the latter is also used by the
4747 accept/connect functions; previously, the settings made by
4748 SSL_set_read_ahead would be lost during the handshake.
4749 [Bodo Moeller; problems reported by Anders Gertz <gertz@epact.se>]
4750
4751 *) Correct util/mkdef.pl to be selective about disabled algorithms.
4752 Previously, it would create entries for disableed algorithms no
4753 matter what.
4754 [Richard Levitte]
4755
4756 *) Added several new manual pages for SSL_* function.
4757 [Lutz Jaenicke]
4758
4759 Changes between 0.9.5a and 0.9.6 [24 Sep 2000]
4760
4761 *) In ssl23_get_client_hello, generate an error message when faced
4762 with an initial SSL 3.0/TLS record that is too small to contain the
4763 first two bytes of the ClientHello message, i.e. client_version.
4764 (Note that this is a pathologic case that probably has never happened
4765 in real life.) The previous approach was to use the version number
4766 from the record header as a substitute; but our protocol choice
4767 should not depend on that one because it is not authenticated
4768 by the Finished messages.
4769 [Bodo Moeller]
4770
4771 *) More robust randomness gathering functions for Windows.
4772 [Jeffrey Altman <jaltman@columbia.edu>]
4773
4774 *) For compatibility reasons if the flag X509_V_FLAG_ISSUER_CHECK is
4775 not set then we don't setup the error code for issuer check errors
4776 to avoid possibly overwriting other errors which the callback does
4777 handle. If an application does set the flag then we assume it knows
4778 what it is doing and can handle the new informational codes
4779 appropriately.
4780 [Steve Henson]
4781
4782 *) Fix for a nasty bug in ASN1_TYPE handling. ASN1_TYPE is used for
4783 a general "ANY" type, as such it should be able to decode anything
4784 including tagged types. However it didn't check the class so it would
4785 wrongly interpret tagged types in the same way as their universal
4786 counterpart and unknown types were just rejected. Changed so that the
4787 tagged and unknown types are handled in the same way as a SEQUENCE:
4788 that is the encoding is stored intact. There is also a new type
4789 "V_ASN1_OTHER" which is used when the class is not universal, in this
4790 case we have no idea what the actual type is so we just lump them all
4791 together.
4792 [Steve Henson]
4793
4794 *) On VMS, stdout may very well lead to a file that is written to
4795 in a record-oriented fashion. That means that every write() will
4796 write a separate record, which will be read separately by the
4797 programs trying to read from it. This can be very confusing.
4798
4799 The solution is to put a BIO filter in the way that will buffer
4800 text until a linefeed is reached, and then write everything a
4801 line at a time, so every record written will be an actual line,
4802 not chunks of lines and not (usually doesn't happen, but I've
4803 seen it once) several lines in one record. BIO_f_linebuffer() is
4804 the answer.
4805
4806 Currently, it's a VMS-only method, because that's where it has
4807 been tested well enough.
4808 [Richard Levitte]
4809
4810 *) Remove 'optimized' squaring variant in BN_mod_mul_montgomery,
4811 it can return incorrect results.
4812 (Note: The buggy variant was not enabled in OpenSSL 0.9.5a,
4813 but it was in 0.9.6-beta[12].)
4814 [Bodo Moeller]
4815
4816 *) Disable the check for content being present when verifying detached
4817 signatures in pk7_smime.c. Some versions of Netscape (wrongly)
4818 include zero length content when signing messages.
4819 [Steve Henson]
4820
4821 *) New BIO_shutdown_wr macro, which invokes the BIO_C_SHUTDOWN_WR
4822 BIO_ctrl (for BIO pairs).
4823 [Bodo Möller]
4824
4825 *) Add DSO method for VMS.
4826 [Richard Levitte]
4827
4828 *) Bug fix: Montgomery multiplication could produce results with the
4829 wrong sign.
4830 [Ulf Möller]
4831
4832 *) Add RPM specification openssl.spec and modify it to build three
4833 packages. The default package contains applications, application
4834 documentation and run-time libraries. The devel package contains
4835 include files, static libraries and function documentation. The
4836 doc package contains the contents of the doc directory. The original
4837 openssl.spec was provided by Damien Miller <djm@mindrot.org>.
4838 [Richard Levitte]
4839
4840 *) Add a large number of documentation files for many SSL routines.
4841 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>]
4842
4843 *) Add a configuration entry for Sony News 4.
4844 [NAKAJI Hiroyuki <nakaji@tutrp.tut.ac.jp>]
4845
4846 *) Don't set the two most significant bits to one when generating a
4847 random number < q in the DSA library.
4848 [Ulf Möller]
4849
4850 *) New SSL API mode 'SSL_MODE_AUTO_RETRY'. This disables the default
4851 behaviour that SSL_read may result in SSL_ERROR_WANT_READ (even if
4852 the underlying transport is blocking) if a handshake took place.
4853 (The default behaviour is needed by applications such as s_client
4854 and s_server that use select() to determine when to use SSL_read;
4855 but for applications that know in advance when to expect data, it
4856 just makes things more complicated.)
4857 [Bodo Moeller]
4858
4859 *) Add RAND_egd_bytes(), which gives control over the number of bytes read
4860 from EGD.
4861 [Ben Laurie]
4862
4863 *) Add a few more EBCDIC conditionals that make `req' and `x509'
4864 work better on such systems.
4865 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>]
4866
4867 *) Add two demo programs for PKCS12_parse() and PKCS12_create().
4868 Update PKCS12_parse() so it copies the friendlyName and the
4869 keyid to the certificates aux info.
4870 [Steve Henson]
4871
4872 *) Fix bug in PKCS7_verify() which caused an infinite loop
4873 if there was more than one signature.
4874 [Sven Uszpelkat <su@celocom.de>]
4875
4876 *) Major change in util/mkdef.pl to include extra information
4877 about each symbol, as well as presentig variables as well
4878 as functions. This change means that there's n more need
4879 to rebuild the .num files when some algorithms are excluded.
4880 [Richard Levitte]
4881
4882 *) Allow the verify time to be set by an application,
4883 rather than always using the current time.
4884 [Steve Henson]
4885
4886 *) Phase 2 verify code reorganisation. The certificate
4887 verify code now looks up an issuer certificate by a
4888 number of criteria: subject name, authority key id
4889 and key usage. It also verifies self signed certificates
4890 by the same criteria. The main comparison function is
4891 X509_check_issued() which performs these checks.
4892
4893 Lot of changes were necessary in order to support this
4894 without completely rewriting the lookup code.
4895
4896 Authority and subject key identifier are now cached.
4897
4898 The LHASH 'certs' is X509_STORE has now been replaced
4899 by a STACK_OF(X509_OBJECT). This is mainly because an
4900 LHASH can't store or retrieve multiple objects with
4901 the same hash value.
4902
4903 As a result various functions (which were all internal
4904 use only) have changed to handle the new X509_STORE
4905 structure. This will break anything that messed round
4906 with X509_STORE internally.
4907
4908 The functions X509_STORE_add_cert() now checks for an
4909 exact match, rather than just subject name.
4910
4911 The X509_STORE API doesn't directly support the retrieval
4912 of multiple certificates matching a given criteria, however
4913 this can be worked round by performing a lookup first
4914 (which will fill the cache with candidate certificates)
4915 and then examining the cache for matches. This is probably
4916 the best we can do without throwing out X509_LOOKUP
4917 entirely (maybe later...).
4918
4919 The X509_VERIFY_CTX structure has been enhanced considerably.
4920
4921 All certificate lookup operations now go via a get_issuer()
4922 callback. Although this currently uses an X509_STORE it
4923 can be replaced by custom lookups. This is a simple way
4924 to bypass the X509_STORE hackery necessary to make this
4925 work and makes it possible to use more efficient techniques
4926 in future. A very simple version which uses a simple
4927 STACK for its trusted certificate store is also provided
4928 using X509_STORE_CTX_trusted_stack().
4929
4930 The verify_cb() and verify() callbacks now have equivalents
4931 in the X509_STORE_CTX structure.
4932
4933 X509_STORE_CTX also has a 'flags' field which can be used
4934 to customise the verify behaviour.
4935 [Steve Henson]
4936
4937 *) Add new PKCS#7 signing option PKCS7_NOSMIMECAP which
4938 excludes S/MIME capabilities.
4939 [Steve Henson]
4940
4941 *) When a certificate request is read in keep a copy of the
4942 original encoding of the signed data and use it when outputing
4943 again. Signatures then use the original encoding rather than
4944 a decoded, encoded version which may cause problems if the
4945 request is improperly encoded.
4946 [Steve Henson]
4947
4948 *) For consistency with other BIO_puts implementations, call
4949 buffer_write(b, ...) directly in buffer_puts instead of calling
4950 BIO_write(b, ...).
4951
4952 In BIO_puts, increment b->num_write as in BIO_write.
4953 [Peter.Sylvester@EdelWeb.fr]
4954
4955 *) Fix BN_mul_word for the case where the word is 0. (We have to use
4956 BN_zero, we may not return a BIGNUM with an array consisting of
4957 words set to zero.)
4958 [Bodo Moeller]
4959
4960 *) Avoid calling abort() from within the library when problems are
4961 detected, except if preprocessor symbols have been defined
4962 (such as REF_CHECK, BN_DEBUG etc.).
4963 [Bodo Moeller]
4964
4965 *) New openssl application 'rsautl'. This utility can be
4966 used for low level RSA operations. DER public key
4967 BIO/fp routines also added.
4968 [Steve Henson]
4969
4970 *) New Configure entry and patches for compiling on QNX 4.
4971 [Andreas Schneider <andreas@ds3.etech.fh-hamburg.de>]
4972
4973 *) A demo state-machine implementation was sponsored by
4974 Nuron (http://www.nuron.com/) and is now available in
4975 demos/state_machine.
4976 [Ben Laurie]
4977
4978 *) New options added to the 'dgst' utility for signature
4979 generation and verification.
4980 [Steve Henson]
4981
4982 *) Unrecognized PKCS#7 content types are now handled via a
4983 catch all ASN1_TYPE structure. This allows unsupported
4984 types to be stored as a "blob" and an application can
4985 encode and decode it manually.
4986 [Steve Henson]
4987
4988 *) Fix various signed/unsigned issues to make a_strex.c
4989 compile under VC++.
4990 [Oscar Jacobsson <oscar.jacobsson@celocom.com>]
4991
4992 *) ASN1 fixes. i2d_ASN1_OBJECT was not returning the correct
4993 length if passed a buffer. ASN1_INTEGER_to_BN failed
4994 if passed a NULL BN and its argument was negative.
4995 [Steve Henson, pointed out by Sven Heiberg <sven@tartu.cyber.ee>]
4996
4997 *) Modification to PKCS#7 encoding routines to output definite
4998 length encoding. Since currently the whole structures are in
4999 memory there's not real point in using indefinite length
5000 constructed encoding. However if OpenSSL is compiled with
5001 the flag PKCS7_INDEFINITE_ENCODING the old form is used.
5002 [Steve Henson]
5003
5004 *) Added BIO_vprintf() and BIO_vsnprintf().
5005 [Richard Levitte]
5006
5007 *) Added more prefixes to parse for in the the strings written
5008 through a logging bio, to cover all the levels that are available
5009 through syslog. The prefixes are now:
5010
5011 PANIC, EMERG, EMR => LOG_EMERG
5012 ALERT, ALR => LOG_ALERT
5013 CRIT, CRI => LOG_CRIT
5014 ERROR, ERR => LOG_ERR
5015 WARNING, WARN, WAR => LOG_WARNING
5016 NOTICE, NOTE, NOT => LOG_NOTICE
5017 INFO, INF => LOG_INFO
5018 DEBUG, DBG => LOG_DEBUG
5019
5020 and as before, if none of those prefixes are present at the
5021 beginning of the string, LOG_ERR is chosen.
5022
5023 On Win32, the LOG_* levels are mapped according to this:
5024
5025 LOG_EMERG, LOG_ALERT, LOG_CRIT, LOG_ERR => EVENTLOG_ERROR_TYPE
5026 LOG_WARNING => EVENTLOG_WARNING_TYPE
5027 LOG_NOTICE, LOG_INFO, LOG_DEBUG => EVENTLOG_INFORMATION_TYPE
5028
5029 [Richard Levitte]
5030
5031 *) Made it possible to reconfigure with just the configuration
5032 argument "reconf" or "reconfigure". The command line arguments
5033 are stored in Makefile.ssl in the variable CONFIGURE_ARGS,
5034 and are retrieved from there when reconfiguring.
5035 [Richard Levitte]
5036
5037 *) MD4 implemented.
5038 [Assar Westerlund <assar@sics.se>, Richard Levitte]
5039
5040 *) Add the arguments -CAfile and -CApath to the pkcs12 utility.
5041 [Richard Levitte]
5042
5043 *) The obj_dat.pl script was messing up the sorting of object
5044 names. The reason was that it compared the quoted version
5045 of strings as a result "OCSP" > "OCSP Signing" because
5046 " > SPACE. Changed script to store unquoted versions of
5047 names and add quotes on output. It was also omitting some
5048 names from the lookup table if they were given a default
5049 value (that is if SN is missing it is given the same
5050 value as LN and vice versa), these are now added on the
5051 grounds that if an object has a name we should be able to
5052 look it up. Finally added warning output when duplicate
5053 short or long names are found.
5054 [Steve Henson]
5055
5056 *) Changes needed for Tandem NSK.
5057 [Scott Uroff <scott@xypro.com>]
5058
5059 *) Fix SSL 2.0 rollback checking: Due to an off-by-one error in
5060 RSA_padding_check_SSLv23(), special padding was never detected
5061 and thus the SSL 3.0/TLS 1.0 countermeasure against protocol
5062 version rollback attacks was not effective.
5063
5064 In s23_clnt.c, don't use special rollback-attack detection padding
5065 (RSA_SSLV23_PADDING) if SSL 2.0 is the only protocol enabled in the
5066 client; similarly, in s23_srvr.c, don't do the rollback check if
5067 SSL 2.0 is the only protocol enabled in the server.
5068 [Bodo Moeller]
5069
5070 *) Make it possible to get hexdumps of unprintable data with 'openssl
5071 asn1parse'. By implication, the functions ASN1_parse_dump() and
5072 BIO_dump_indent() are added.
5073 [Richard Levitte]
5074
5075 *) New functions ASN1_STRING_print_ex() and X509_NAME_print_ex()
5076 these print out strings and name structures based on various
5077 flags including RFC2253 support and proper handling of
5078 multibyte characters. Added options to the 'x509' utility
5079 to allow the various flags to be set.
5080 [Steve Henson]
5081
5082 *) Various fixes to use ASN1_TIME instead of ASN1_UTCTIME.
5083 Also change the functions X509_cmp_current_time() and
5084 X509_gmtime_adj() work with an ASN1_TIME structure,
5085 this will enable certificates using GeneralizedTime in validity
5086 dates to be checked.
5087 [Steve Henson]
5088
5089 *) Make the NEG_PUBKEY_BUG code (which tolerates invalid
5090 negative public key encodings) on by default,
5091 NO_NEG_PUBKEY_BUG can be set to disable it.
5092 [Steve Henson]
5093
5094 *) New function c2i_ASN1_OBJECT() which acts on ASN1_OBJECT
5095 content octets. An i2c_ASN1_OBJECT is unnecessary because
5096 the encoding can be trivially obtained from the structure.
5097 [Steve Henson]
5098
5099 *) crypto/err.c locking bugfix: Use write locks (CRYPTO_w_[un]lock),
5100 not read locks (CRYPTO_r_[un]lock).
5101 [Bodo Moeller]
5102
5103 *) A first attempt at creating official support for shared
5104 libraries through configuration. I've kept it so the
5105 default is static libraries only, and the OpenSSL programs
5106 are always statically linked for now, but there are
5107 preparations for dynamic linking in place.
5108 This has been tested on Linux and Tru64.
5109 [Richard Levitte]
5110
5111 *) Randomness polling function for Win9x, as described in:
5112 Peter Gutmann, Software Generation of Practically Strong
5113 Random Numbers.
5114 [Ulf Möller]
5115
5116 *) Fix so PRNG is seeded in req if using an already existing
5117 DSA key.
5118 [Steve Henson]
5119
5120 *) New options to smime application. -inform and -outform
5121 allow alternative formats for the S/MIME message including
5122 PEM and DER. The -content option allows the content to be
5123 specified separately. This should allow things like Netscape
5124 form signing output easier to verify.
5125 [Steve Henson]
5126
5127 *) Fix the ASN1 encoding of tags using the 'long form'.
5128 [Steve Henson]
5129
5130 *) New ASN1 functions, i2c_* and c2i_* for INTEGER and BIT
5131 STRING types. These convert content octets to and from the
5132 underlying type. The actual tag and length octets are
5133 already assumed to have been read in and checked. These
5134 are needed because all other string types have virtually
5135 identical handling apart from the tag. By having versions
5136 of the ASN1 functions that just operate on content octets
5137 IMPLICIT tagging can be handled properly. It also allows
5138 the ASN1_ENUMERATED code to be cut down because ASN1_ENUMERATED
5139 and ASN1_INTEGER are identical apart from the tag.
5140 [Steve Henson]
5141
5142 *) Change the handling of OID objects as follows:
5143
5144 - New object identifiers are inserted in objects.txt, following
5145 the syntax given in objects.README.
5146 - objects.pl is used to process obj_mac.num and create a new
5147 obj_mac.h.
5148 - obj_dat.pl is used to create a new obj_dat.h, using the data in
5149 obj_mac.h.
5150
5151 This is currently kind of a hack, and the perl code in objects.pl
5152 isn't very elegant, but it works as I intended. The simplest way
5153 to check that it worked correctly is to look in obj_dat.h and
5154 check the array nid_objs and make sure the objects haven't moved
5155 around (this is important!). Additions are OK, as well as
5156 consistent name changes.
5157 [Richard Levitte]
5158
5159 *) Add BSD-style MD5-based passwords to 'openssl passwd' (option '-1').
5160 [Bodo Moeller]
5161
5162 *) Addition of the command line parameter '-rand file' to 'openssl req'.
5163 The given file adds to whatever has already been seeded into the
5164 random pool through the RANDFILE configuration file option or
5165 environment variable, or the default random state file.
5166 [Richard Levitte]
5167
5168 *) mkstack.pl now sorts each macro group into lexical order.
5169 Previously the output order depended on the order the files
5170 appeared in the directory, resulting in needless rewriting
5171 of safestack.h .
5172 [Steve Henson]
5173
5174 *) Patches to make OpenSSL compile under Win32 again. Mostly
5175 work arounds for the VC++ problem that it treats func() as
5176 func(void). Also stripped out the parts of mkdef.pl that
5177 added extra typesafe functions: these no longer exist.
5178 [Steve Henson]
5179
5180 *) Reorganisation of the stack code. The macros are now all
5181 collected in safestack.h . Each macro is defined in terms of
5182 a "stack macro" of the form SKM_<name>(type, a, b). The
5183 DEBUG_SAFESTACK is now handled in terms of function casts,
5184 this has the advantage of retaining type safety without the
5185 use of additional functions. If DEBUG_SAFESTACK is not defined
5186 then the non typesafe macros are used instead. Also modified the
5187 mkstack.pl script to handle the new form. Needs testing to see
5188 if which (if any) compilers it chokes and maybe make DEBUG_SAFESTACK
5189 the default if no major problems. Similar behaviour for ASN1_SET_OF
5190 and PKCS12_STACK_OF.
5191 [Steve Henson]
5192
5193 *) When some versions of IIS use the 'NET' form of private key the
5194 key derivation algorithm is different. Normally MD5(password) is
5195 used as a 128 bit RC4 key. In the modified case
5196 MD5(MD5(password) + "SGCKEYSALT") is used insted. Added some
5197 new functions i2d_RSA_NET(), d2i_RSA_NET() etc which are the same
5198 as the old Netscape_RSA functions except they have an additional
5199 'sgckey' parameter which uses the modified algorithm. Also added
5200 an -sgckey command line option to the rsa utility. Thanks to
5201 Adrian Peck <bertie@ncipher.com> for posting details of the modified
5202 algorithm to openssl-dev.
5203 [Steve Henson]
5204
5205 *) The evp_local.h macros were using 'c.##kname' which resulted in
5206 invalid expansion on some systems (SCO 5.0.5 for example).
5207 Corrected to 'c.kname'.
5208 [Phillip Porch <root@theporch.com>]
5209
5210 *) New X509_get1_email() and X509_REQ_get1_email() functions that return
5211 a STACK of email addresses from a certificate or request, these look
5212 in the subject name and the subject alternative name extensions and
5213 omit any duplicate addresses.
5214 [Steve Henson]
5215
5216 *) Re-implement BN_mod_exp2_mont using independent (and larger) windows.
5217 This makes DSA verification about 2 % faster.
5218 [Bodo Moeller]
5219
5220 *) Increase maximum window size in BN_mod_exp_... to 6 bits instead of 5
5221 (meaning that now 2^5 values will be precomputed, which is only 4 KB
5222 plus overhead for 1024 bit moduli).
5223 This makes exponentiations about 0.5 % faster for 1024 bit
5224 exponents (as measured by "openssl speed rsa2048").
5225 [Bodo Moeller]
5226
5227 *) Rename memory handling macros to avoid conflicts with other
5228 software:
5229 Malloc => OPENSSL_malloc
5230 Malloc_locked => OPENSSL_malloc_locked
5231 Realloc => OPENSSL_realloc
5232 Free => OPENSSL_free
5233 [Richard Levitte]
5234
5235 *) New function BN_mod_exp_mont_word for small bases (roughly 15%
5236 faster than BN_mod_exp_mont, i.e. 7% for a full DH exchange).
5237 [Bodo Moeller]
5238
5239 *) CygWin32 support.
5240 [John Jarvie <jjarvie@newsguy.com>]
5241
5242 *) The type-safe stack code has been rejigged. It is now only compiled
5243 in when OpenSSL is configured with the DEBUG_SAFESTACK option and
5244 by default all type-specific stack functions are "#define"d back to
5245 standard stack functions. This results in more streamlined output
5246 but retains the type-safety checking possibilities of the original
5247 approach.
5248 [Geoff Thorpe]
5249
5250 *) The STACK code has been cleaned up, and certain type declarations
5251 that didn't make a lot of sense have been brought in line. This has
5252 also involved a cleanup of sorts in safestack.h to more correctly
5253 map type-safe stack functions onto their plain stack counterparts.
5254 This work has also resulted in a variety of "const"ifications of
5255 lots of the code, especially "_cmp" operations which should normally
5256 be prototyped with "const" parameters anyway.
5257 [Geoff Thorpe]
5258
5259 *) When generating bytes for the first time in md_rand.c, 'stir the pool'
5260 by seeding with STATE_SIZE dummy bytes (with zero entropy count).
5261 (The PRNG state consists of two parts, the large pool 'state' and 'md',
5262 where all of 'md' is used each time the PRNG is used, but 'state'
5263 is used only indexed by a cyclic counter. As entropy may not be
5264 well distributed from the beginning, 'md' is important as a
5265 chaining variable. However, the output function chains only half
5266 of 'md', i.e. 80 bits. ssleay_rand_add, on the other hand, chains
5267 all of 'md', and seeding with STATE_SIZE dummy bytes will result
5268 in all of 'state' being rewritten, with the new values depending
5269 on virtually all of 'md'. This overcomes the 80 bit limitation.)
5270 [Bodo Moeller]
5271
5272 *) In ssl/s2_clnt.c and ssl/s3_clnt.c, call ERR_clear_error() when
5273 the handshake is continued after ssl_verify_cert_chain();
5274 otherwise, if SSL_VERIFY_NONE is set, remaining error codes
5275 can lead to 'unexplainable' connection aborts later.
5276 [Bodo Moeller; problem tracked down by Lutz Jaenicke]
5277
5278 *) Major EVP API cipher revision.
5279 Add hooks for extra EVP features. This allows various cipher
5280 parameters to be set in the EVP interface. Support added for variable
5281 key length ciphers via the EVP_CIPHER_CTX_set_key_length() function and
5282 setting of RC2 and RC5 parameters.
5283
5284 Modify EVP_OpenInit() and EVP_SealInit() to cope with variable key length
5285 ciphers.
5286
5287 Remove lots of duplicated code from the EVP library. For example *every*
5288 cipher init() function handles the 'iv' in the same way according to the
5289 cipher mode. They also all do nothing if the 'key' parameter is NULL and
5290 for CFB and OFB modes they zero ctx->num.
5291
5292 New functionality allows removal of S/MIME code RC2 hack.
5293
5294 Most of the routines have the same form and so can be declared in terms
5295 of macros.
5296
5297 By shifting this to the top level EVP_CipherInit() it can be removed from
5298 all individual ciphers. If the cipher wants to handle IVs or keys
5299 differently it can set the EVP_CIPH_CUSTOM_IV or EVP_CIPH_ALWAYS_CALL_INIT
5300 flags.
5301
5302 Change lots of functions like EVP_EncryptUpdate() to now return a
5303 value: although software versions of the algorithms cannot fail
5304 any installed hardware versions can.
5305 [Steve Henson]
5306
5307 *) Implement SSL_OP_TLS_ROLLBACK_BUG: In ssl3_get_client_key_exchange, if
5308 this option is set, tolerate broken clients that send the negotiated
5309 protocol version number instead of the requested protocol version
5310 number.
5311 [Bodo Moeller]
5312
5313 *) Call dh_tmp_cb (set by ..._TMP_DH_CB) with correct 'is_export' flag;
5314 i.e. non-zero for export ciphersuites, zero otherwise.
5315 Previous versions had this flag inverted, inconsistent with
5316 rsa_tmp_cb (..._TMP_RSA_CB).
5317 [Bodo Moeller; problem reported by Amit Chopra]
5318
5319 *) Add missing DSA library text string. Work around for some IIS
5320 key files with invalid SEQUENCE encoding.
5321 [Steve Henson]
5322
5323 *) Add a document (doc/standards.txt) that list all kinds of standards
5324 and so on that are implemented in OpenSSL.
5325 [Richard Levitte]
5326
5327 *) Enhance c_rehash script. Old version would mishandle certificates
5328 with the same subject name hash and wouldn't handle CRLs at all.
5329 Added -fingerprint option to crl utility, to support new c_rehash
5330 features.
5331 [Steve Henson]
5332
5333 *) Eliminate non-ANSI declarations in crypto.h and stack.h.
5334 [Ulf Möller]
5335
5336 *) Fix for SSL server purpose checking. Server checking was
5337 rejecting certificates which had extended key usage present
5338 but no ssl client purpose.
5339 [Steve Henson, reported by Rene Grosser <grosser@hisolutions.com>]
5340
5341 *) Make PKCS#12 code work with no password. The PKCS#12 spec
5342 is a little unclear about how a blank password is handled.
5343 Since the password in encoded as a BMPString with terminating
5344 double NULL a zero length password would end up as just the
5345 double NULL. However no password at all is different and is
5346 handled differently in the PKCS#12 key generation code. NS
5347 treats a blank password as zero length. MSIE treats it as no
5348 password on export: but it will try both on import. We now do
5349 the same: PKCS12_parse() tries zero length and no password if
5350 the password is set to "" or NULL (NULL is now a valid password:
5351 it wasn't before) as does the pkcs12 application.
5352 [Steve Henson]
5353
5354 *) Bugfixes in apps/x509.c: Avoid a memory leak; and don't use
5355 perror when PEM_read_bio_X509_REQ fails, the error message must
5356 be obtained from the error queue.
5357 [Bodo Moeller]
5358
5359 *) Avoid 'thread_hash' memory leak in crypto/err/err.c by freeing
5360 it in ERR_remove_state if appropriate, and change ERR_get_state
5361 accordingly to avoid race conditions (this is necessary because
5362 thread_hash is no longer constant once set).
5363 [Bodo Moeller]
5364
5365 *) Bugfix for linux-elf makefile.one.
5366 [Ulf Möller]
5367
5368 *) RSA_get_default_method() will now cause a default
5369 RSA_METHOD to be chosen if one doesn't exist already.
5370 Previously this was only set during a call to RSA_new()
5371 or RSA_new_method(NULL) meaning it was possible for
5372 RSA_get_default_method() to return NULL.
5373 [Geoff Thorpe]
5374
5375 *) Added native name translation to the existing DSO code
5376 that will convert (if the flag to do so is set) filenames
5377 that are sufficiently small and have no path information
5378 into a canonical native form. Eg. "blah" converted to
5379 "libblah.so" or "blah.dll" etc.
5380 [Geoff Thorpe]
5381
5382 *) New function ERR_error_string_n(e, buf, len) which is like
5383 ERR_error_string(e, buf), but writes at most 'len' bytes
5384 including the 0 terminator. For ERR_error_string_n, 'buf'
5385 may not be NULL.
5386 [Damien Miller <djm@mindrot.org>, Bodo Moeller]
5387
5388 *) CONF library reworked to become more general. A new CONF
5389 configuration file reader "class" is implemented as well as a
5390 new functions (NCONF_*, for "New CONF") to handle it. The now
5391 old CONF_* functions are still there, but are reimplemented to
5392 work in terms of the new functions. Also, a set of functions
5393 to handle the internal storage of the configuration data is
5394 provided to make it easier to write new configuration file
5395 reader "classes" (I can definitely see something reading a
5396 configuration file in XML format, for example), called _CONF_*,
5397 or "the configuration storage API"...
5398
5399 The new configuration file reading functions are:
5400
5401 NCONF_new, NCONF_free, NCONF_load, NCONF_load_fp, NCONF_load_bio,
5402 NCONF_get_section, NCONF_get_string, NCONF_get_numbre
5403
5404 NCONF_default, NCONF_WIN32
5405
5406 NCONF_dump_fp, NCONF_dump_bio
5407
5408 NCONF_default and NCONF_WIN32 are method (or "class") choosers,
5409 NCONF_new creates a new CONF object. This works in the same way
5410 as other interfaces in OpenSSL, like the BIO interface.
5411 NCONF_dump_* dump the internal storage of the configuration file,
5412 which is useful for debugging. All other functions take the same
5413 arguments as the old CONF_* functions wth the exception of the
5414 first that must be a `CONF *' instead of a `LHASH *'.
5415
5416 To make it easer to use the new classes with the old CONF_* functions,
5417 the function CONF_set_default_method is provided.
5418 [Richard Levitte]
5419
5420 *) Add '-tls1' option to 'openssl ciphers', which was already
5421 mentioned in the documentation but had not been implemented.
5422 (This option is not yet really useful because even the additional
5423 experimental TLS 1.0 ciphers are currently treated as SSL 3.0 ciphers.)
5424 [Bodo Moeller]
5425
5426 *) Initial DSO code added into libcrypto for letting OpenSSL (and
5427 OpenSSL-based applications) load shared libraries and bind to
5428 them in a portable way.
5429 [Geoff Thorpe, with contributions from Richard Levitte]
5430
5431 Changes between 0.9.5 and 0.9.5a [1 Apr 2000]
5432
5433 *) Make sure _lrotl and _lrotr are only used with MSVC.
5434
5435 *) Use lock CRYPTO_LOCK_RAND correctly in ssleay_rand_status
5436 (the default implementation of RAND_status).
5437
5438 *) Rename openssl x509 option '-crlext', which was added in 0.9.5,
5439 to '-clrext' (= clear extensions), as intended and documented.
5440 [Bodo Moeller; inconsistency pointed out by Michael Attili
5441 <attili@amaxo.com>]
5442
5443 *) Fix for HMAC. It wasn't zeroing the rest of the block if the key length
5444 was larger than the MD block size.
5445 [Steve Henson, pointed out by Yost William <YostW@tce.com>]
5446
5447 *) Modernise PKCS12_parse() so it uses STACK_OF(X509) for its ca argument
5448 fix a leak when the ca argument was passed as NULL. Stop X509_PUBKEY_set()
5449 using the passed key: if the passed key was a private key the result
5450 of X509_print(), for example, would be to print out all the private key
5451 components.
5452 [Steve Henson]
5453
5454 *) des_quad_cksum() byte order bug fix.
5455 [Ulf Möller, using the problem description in krb4-0.9.7, where
5456 the solution is attributed to Derrick J Brashear <shadow@DEMENTIA.ORG>]
5457
5458 *) Fix so V_ASN1_APP_CHOOSE works again: however its use is strongly
5459 discouraged.
5460 [Steve Henson, pointed out by Brian Korver <briank@cs.stanford.edu>]
5461
5462 *) For easily testing in shell scripts whether some command
5463 'openssl XXX' exists, the new pseudo-command 'openssl no-XXX'
5464 returns with exit code 0 iff no command of the given name is available.
5465 'no-XXX' is printed in this case, 'XXX' otherwise. In both cases,
5466 the output goes to stdout and nothing is printed to stderr.
5467 Additional arguments are always ignored.
5468
5469 Since for each cipher there is a command of the same name,
5470 the 'no-cipher' compilation switches can be tested this way.
5471
5472 ('openssl no-XXX' is not able to detect pseudo-commands such
5473 as 'quit', 'list-XXX-commands', or 'no-XXX' itself.)
5474 [Bodo Moeller]
5475
5476 *) Update test suite so that 'make test' succeeds in 'no-rsa' configuration.
5477 [Bodo Moeller]
5478
5479 *) For SSL_[CTX_]set_tmp_dh, don't create a DH key if SSL_OP_SINGLE_DH_USE
5480 is set; it will be thrown away anyway because each handshake creates
5481 its own key.
5482 ssl_cert_dup, which is used by SSL_new, now copies DH keys in addition
5483 to parameters -- in previous versions (since OpenSSL 0.9.3) the
5484 'default key' from SSL_CTX_set_tmp_dh would always be lost, meanining
5485 you effectivly got SSL_OP_SINGLE_DH_USE when using this macro.
5486 [Bodo Moeller]
5487
5488 *) New s_client option -ign_eof: EOF at stdin is ignored, and
5489 'Q' and 'R' lose their special meanings (quit/renegotiate).
5490 This is part of what -quiet does; unlike -quiet, -ign_eof
5491 does not suppress any output.
5492 [Richard Levitte]
5493
5494 *) Add compatibility options to the purpose and trust code. The
5495 purpose X509_PURPOSE_ANY is "any purpose" which automatically
5496 accepts a certificate or CA, this was the previous behaviour,
5497 with all the associated security issues.
5498
5499 X509_TRUST_COMPAT is the old trust behaviour: only and
5500 automatically trust self signed roots in certificate store. A
5501 new trust setting X509_TRUST_DEFAULT is used to specify that
5502 a purpose has no associated trust setting and it should instead
5503 use the value in the default purpose.
5504 [Steve Henson]
5505
5506 *) Fix the PKCS#8 DSA private key code so it decodes keys again
5507 and fix a memory leak.
5508 [Steve Henson]
5509
5510 *) In util/mkerr.pl (which implements 'make errors'), preserve
5511 reason strings from the previous version of the .c file, as
5512 the default to have only downcase letters (and digits) in
5513 automatically generated reasons codes is not always appropriate.
5514 [Bodo Moeller]
5515
5516 *) In ERR_load_ERR_strings(), build an ERR_LIB_SYS error reason table
5517 using strerror. Previously, ERR_reason_error_string() returned
5518 library names as reason strings for SYSerr; but SYSerr is a special
5519 case where small numbers are errno values, not library numbers.
5520 [Bodo Moeller]
5521
5522 *) Add '-dsaparam' option to 'openssl dhparam' application. This
5523 converts DSA parameters into DH parameters. (When creating parameters,
5524 DSA_generate_parameters is used.)
5525 [Bodo Moeller]
5526
5527 *) Include 'length' (recommended exponent length) in C code generated
5528 by 'openssl dhparam -C'.
5529 [Bodo Moeller]
5530
5531 *) The second argument to set_label in perlasm was already being used
5532 so couldn't be used as a "file scope" flag. Moved to third argument
5533 which was free.
5534 [Steve Henson]
5535
5536 *) In PEM_ASN1_write_bio and some other functions, use RAND_pseudo_bytes
5537 instead of RAND_bytes for encryption IVs and salts.
5538 [Bodo Moeller]
5539
5540 *) Include RAND_status() into RAND_METHOD instead of implementing
5541 it only for md_rand.c Otherwise replacing the PRNG by calling
5542 RAND_set_rand_method would be impossible.
5543 [Bodo Moeller]
5544
5545 *) Don't let DSA_generate_key() enter an infinite loop if the random
5546 number generation fails.
5547 [Bodo Moeller]
5548
5549 *) New 'rand' application for creating pseudo-random output.
5550 [Bodo Moeller]
5551
5552 *) Added configuration support for Linux/IA64
5553 [Rolf Haberrecker <rolf@suse.de>]
5554
5555 *) Assembler module support for Mingw32.
5556 [Ulf Möller]
5557
5558 *) Shared library support for HPUX (in shlib/).
5559 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE> and Anonymous]
5560
5561 *) Shared library support for Solaris gcc.
5562 [Lutz Behnke <behnke@trustcenter.de>]
5563
5564 Changes between 0.9.4 and 0.9.5 [28 Feb 2000]
5565
5566 *) PKCS7_encrypt() was adding text MIME headers twice because they
5567 were added manually and by SMIME_crlf_copy().
5568 [Steve Henson]
5569
5570 *) In bntest.c don't call BN_rand with zero bits argument.
5571 [Steve Henson, pointed out by Andrew W. Gray <agray@iconsinc.com>]
5572
5573 *) BN_mul bugfix: In bn_mul_part_recursion() only the a>a[n] && b>b[n]
5574 case was implemented. This caused BN_div_recp() to fail occasionally.
5575 [Ulf Möller]
5576
5577 *) Add an optional second argument to the set_label() in the perl
5578 assembly language builder. If this argument exists and is set
5579 to 1 it signals that the assembler should use a symbol whose
5580 scope is the entire file, not just the current function. This
5581 is needed with MASM which uses the format label:: for this scope.
5582 [Steve Henson, pointed out by Peter Runestig <peter@runestig.com>]
5583
5584 *) Change the ASN1 types so they are typedefs by default. Before
5585 almost all types were #define'd to ASN1_STRING which was causing
5586 STACK_OF() problems: you couldn't declare STACK_OF(ASN1_UTF8STRING)
5587 for example.
5588 [Steve Henson]
5589
5590 *) Change names of new functions to the new get1/get0 naming
5591 convention: After 'get1', the caller owns a reference count
5592 and has to call ..._free; 'get0' returns a pointer to some
5593 data structure without incrementing reference counters.
5594 (Some of the existing 'get' functions increment a reference
5595 counter, some don't.)
5596 Similarly, 'set1' and 'add1' functions increase reference
5597 counters or duplicate objects.
5598 [Steve Henson]
5599
5600 *) Allow for the possibility of temp RSA key generation failure:
5601 the code used to assume it always worked and crashed on failure.
5602 [Steve Henson]
5603
5604 *) Fix potential buffer overrun problem in BIO_printf().
5605 [Ulf Möller, using public domain code by Patrick Powell; problem
5606 pointed out by David Sacerdote <das33@cornell.edu>]
5607
5608 *) Support EGD <http://www.lothar.com/tech/crypto/>. New functions
5609 RAND_egd() and RAND_status(). In the command line application,
5610 the EGD socket can be specified like a seed file using RANDFILE
5611 or -rand.
5612 [Ulf Möller]
5613
5614 *) Allow the string CERTIFICATE to be tolerated in PKCS#7 structures.
5615 Some CAs (e.g. Verisign) distribute certificates in this form.
5616 [Steve Henson]
5617
5618 *) Remove the SSL_ALLOW_ADH compile option and set the default cipher
5619 list to exclude them. This means that no special compilation option
5620 is needed to use anonymous DH: it just needs to be included in the
5621 cipher list.
5622 [Steve Henson]
5623
5624 *) Change the EVP_MD_CTX_type macro so its meaning consistent with
5625 EVP_MD_type. The old functionality is available in a new macro called
5626 EVP_MD_md(). Change code that uses it and update docs.
5627 [Steve Henson]
5628
5629 *) ..._ctrl functions now have corresponding ..._callback_ctrl functions
5630 where the 'void *' argument is replaced by a function pointer argument.
5631 Previously 'void *' was abused to point to functions, which works on
5632 many platforms, but is not correct. As these functions are usually
5633 called by macros defined in OpenSSL header files, most source code
5634 should work without changes.
5635 [Richard Levitte]
5636
5637 *) <openssl/opensslconf.h> (which is created by Configure) now contains
5638 sections with information on -D... compiler switches used for
5639 compiling the library so that applications can see them. To enable
5640 one of these sections, a pre-processor symbol OPENSSL_..._DEFINES
5641 must be defined. E.g.,
5642 #define OPENSSL_ALGORITHM_DEFINES
5643 #include <openssl/opensslconf.h>
5644 defines all pertinent NO_<algo> symbols, such as NO_IDEA, NO_RSA, etc.
5645 [Richard Levitte, Ulf and Bodo Möller]
5646
5647 *) Bugfix: Tolerate fragmentation and interleaving in the SSL 3/TLS
5648 record layer.
5649 [Bodo Moeller]
5650
5651 *) Change the 'other' type in certificate aux info to a STACK_OF
5652 X509_ALGOR. Although not an AlgorithmIdentifier as such it has
5653 the required ASN1 format: arbitrary types determined by an OID.
5654 [Steve Henson]
5655
5656 *) Add some PEM_write_X509_REQ_NEW() functions and a command line
5657 argument to 'req'. This is not because the function is newer or
5658 better than others it just uses the work 'NEW' in the certificate
5659 request header lines. Some software needs this.
5660 [Steve Henson]
5661
5662 *) Reorganise password command line arguments: now passwords can be
5663 obtained from various sources. Delete the PEM_cb function and make
5664 it the default behaviour: i.e. if the callback is NULL and the
5665 usrdata argument is not NULL interpret it as a null terminated pass
5666 phrase. If usrdata and the callback are NULL then the pass phrase
5667 is prompted for as usual.
5668 [Steve Henson]
5669
5670 *) Add support for the Compaq Atalla crypto accelerator. If it is installed,
5671 the support is automatically enabled. The resulting binaries will
5672 autodetect the card and use it if present.
5673 [Ben Laurie and Compaq Inc.]
5674
5675 *) Work around for Netscape hang bug. This sends certificate request
5676 and server done in one record. Since this is perfectly legal in the
5677 SSL/TLS protocol it isn't a "bug" option and is on by default. See
5678 the bugs/SSLv3 entry for more info.
5679 [Steve Henson]
5680
5681 *) HP-UX tune-up: new unified configs, HP C compiler bug workaround.
5682 [Andy Polyakov]
5683
5684 *) Add -rand argument to smime and pkcs12 applications and read/write
5685 of seed file.
5686 [Steve Henson]
5687
5688 *) New 'passwd' tool for crypt(3) and apr1 password hashes.
5689 [Bodo Moeller]
5690
5691 *) Add command line password options to the remaining applications.
5692 [Steve Henson]
5693
5694 *) Bug fix for BN_div_recp() for numerators with an even number of
5695 bits.
5696 [Ulf Möller]
5697
5698 *) More tests in bntest.c, and changed test_bn output.
5699 [Ulf Möller]
5700
5701 *) ./config recognizes MacOS X now.
5702 [Andy Polyakov]
5703
5704 *) Bug fix for BN_div() when the first words of num and divsor are
5705 equal (it gave wrong results if (rem=(n1-q*d0)&BN_MASK2) < d0).
5706 [Ulf Möller]
5707
5708 *) Add support for various broken PKCS#8 formats, and command line
5709 options to produce them.
5710 [Steve Henson]
5711
5712 *) New functions BN_CTX_start(), BN_CTX_get() and BT_CTX_end() to
5713 get temporary BIGNUMs from a BN_CTX.
5714 [Ulf Möller]
5715
5716 *) Correct return values in BN_mod_exp_mont() and BN_mod_exp2_mont()
5717 for p == 0.
5718 [Ulf Möller]
5719
5720 *) Change the SSLeay_add_all_*() functions to OpenSSL_add_all_*() and
5721 include a #define from the old name to the new. The original intent
5722 was that statically linked binaries could for example just call
5723 SSLeay_add_all_ciphers() to just add ciphers to the table and not
5724 link with digests. This never worked becayse SSLeay_add_all_digests()
5725 and SSLeay_add_all_ciphers() were in the same source file so calling
5726 one would link with the other. They are now in separate source files.
5727 [Steve Henson]
5728
5729 *) Add a new -notext option to 'ca' and a -pubkey option to 'spkac'.
5730 [Steve Henson]
5731
5732 *) Use a less unusual form of the Miller-Rabin primality test (it used
5733 a binary algorithm for exponentiation integrated into the Miller-Rabin
5734 loop, our standard modexp algorithms are faster).
5735 [Bodo Moeller]
5736
5737 *) Support for the EBCDIC character set completed.
5738 [Martin Kraemer <Martin.Kraemer@Mch.SNI.De>]
5739
5740 *) Source code cleanups: use const where appropriate, eliminate casts,
5741 use void * instead of char * in lhash.
5742 [Ulf Möller]
5743
5744 *) Bugfix: ssl3_send_server_key_exchange was not restartable
5745 (the state was not changed to SSL3_ST_SW_KEY_EXCH_B, and because of
5746 this the server could overwrite ephemeral keys that the client
5747 has already seen).
5748 [Bodo Moeller]
5749
5750 *) Turn DSA_is_prime into a macro that calls BN_is_prime,
5751 using 50 iterations of the Rabin-Miller test.
5752
5753 DSA_generate_parameters now uses BN_is_prime_fasttest (with 50
5754 iterations of the Rabin-Miller test as required by the appendix
5755 to FIPS PUB 186[-1]) instead of DSA_is_prime.
5756 As BN_is_prime_fasttest includes trial division, DSA parameter
5757 generation becomes much faster.
5758
5759 This implies a change for the callback functions in DSA_is_prime
5760 and DSA_generate_parameters: The callback function is called once
5761 for each positive witness in the Rabin-Miller test, not just
5762 occasionally in the inner loop; and the parameters to the
5763 callback function now provide an iteration count for the outer
5764 loop rather than for the current invocation of the inner loop.
5765 DSA_generate_parameters additionally can call the callback
5766 function with an 'iteration count' of -1, meaning that a
5767 candidate has passed the trial division test (when q is generated
5768 from an application-provided seed, trial division is skipped).
5769 [Bodo Moeller]
5770
5771 *) New function BN_is_prime_fasttest that optionally does trial
5772 division before starting the Rabin-Miller test and has
5773 an additional BN_CTX * argument (whereas BN_is_prime always
5774 has to allocate at least one BN_CTX).
5775 'callback(1, -1, cb_arg)' is called when a number has passed the
5776 trial division stage.
5777 [Bodo Moeller]
5778
5779 *) Fix for bug in CRL encoding. The validity dates weren't being handled
5780 as ASN1_TIME.
5781 [Steve Henson]
5782
5783 *) New -pkcs12 option to CA.pl script to write out a PKCS#12 file.
5784 [Steve Henson]
5785
5786 *) New function BN_pseudo_rand().
5787 [Ulf Möller]
5788
5789 *) Clean up BN_mod_mul_montgomery(): replace the broken (and unreadable)
5790 bignum version of BN_from_montgomery() with the working code from
5791 SSLeay 0.9.0 (the word based version is faster anyway), and clean up
5792 the comments.
5793 [Ulf Möller]
5794
5795 *) Avoid a race condition in s2_clnt.c (function get_server_hello) that
5796 made it impossible to use the same SSL_SESSION data structure in
5797 SSL2 clients in multiple threads.
5798 [Bodo Moeller]
5799
5800 *) The return value of RAND_load_file() no longer counts bytes obtained
5801 by stat(). RAND_load_file(..., -1) is new and uses the complete file
5802 to seed the PRNG (previously an explicit byte count was required).
5803 [Ulf Möller, Bodo Möller]
5804
5805 *) Clean up CRYPTO_EX_DATA functions, some of these didn't have prototypes
5806 used (char *) instead of (void *) and had casts all over the place.
5807 [Steve Henson]
5808
5809 *) Make BN_generate_prime() return NULL on error if ret!=NULL.
5810 [Ulf Möller]
5811
5812 *) Retain source code compatibility for BN_prime_checks macro:
5813 BN_is_prime(..., BN_prime_checks, ...) now uses
5814 BN_prime_checks_for_size to determine the appropriate number of
5815 Rabin-Miller iterations.
5816 [Ulf Möller]
5817
5818 *) Diffie-Hellman uses "safe" primes: DH_check() return code renamed to
5819 DH_CHECK_P_NOT_SAFE_PRIME.
5820 (Check if this is true? OpenPGP calls them "strong".)
5821 [Ulf Möller]
5822
5823 *) Merge the functionality of "dh" and "gendh" programs into a new program
5824 "dhparam". The old programs are retained for now but will handle DH keys
5825 (instead of parameters) in future.
5826 [Steve Henson]
5827
5828 *) Make the ciphers, s_server and s_client programs check the return values
5829 when a new cipher list is set.
5830 [Steve Henson]
5831
5832 *) Enhance the SSL/TLS cipher mechanism to correctly handle the TLS 56bit
5833 ciphers. Before when the 56bit ciphers were enabled the sorting was
5834 wrong.
5835
5836 The syntax for the cipher sorting has been extended to support sorting by
5837 cipher-strength (using the strength_bits hard coded in the tables).
5838 The new command is "@STRENGTH" (see also doc/apps/ciphers.pod).
5839
5840 Fix a bug in the cipher-command parser: when supplying a cipher command
5841 string with an "undefined" symbol (neither command nor alphanumeric
5842 [A-Za-z0-9], ssl_set_cipher_list used to hang in an endless loop. Now
5843 an error is flagged.
5844
5845 Due to the strength-sorting extension, the code of the
5846 ssl_create_cipher_list() function was completely rearranged. I hope that
5847 the readability was also increased :-)
5848 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>]
5849
5850 *) Minor change to 'x509' utility. The -CAcreateserial option now uses 1
5851 for the first serial number and places 2 in the serial number file. This
5852 avoids problems when the root CA is created with serial number zero and
5853 the first user certificate has the same issuer name and serial number
5854 as the root CA.
5855 [Steve Henson]
5856
5857 *) Fixes to X509_ATTRIBUTE utilities, change the 'req' program so it uses
5858 the new code. Add documentation for this stuff.
5859 [Steve Henson]
5860
5861 *) Changes to X509_ATTRIBUTE utilities. These have been renamed from
5862 X509_*() to X509at_*() on the grounds that they don't handle X509
5863 structures and behave in an analagous way to the X509v3 functions:
5864 they shouldn't be called directly but wrapper functions should be used
5865 instead.
5866
5867 So we also now have some wrapper functions that call the X509at functions
5868 when passed certificate requests. (TO DO: similar things can be done with
5869 PKCS#7 signed and unsigned attributes, PKCS#12 attributes and a few other
5870 things. Some of these need some d2i or i2d and print functionality
5871 because they handle more complex structures.)
5872 [Steve Henson]
5873
5874 *) Add missing #ifndefs that caused missing symbols when building libssl
5875 as a shared library without RSA. Use #ifndef NO_SSL2 instead of
5876 NO_RSA in ssl/s2*.c.
5877 [Kris Kennaway <kris@hub.freebsd.org>, modified by Ulf Möller]
5878
5879 *) Precautions against using the PRNG uninitialized: RAND_bytes() now
5880 has a return value which indicates the quality of the random data
5881 (1 = ok, 0 = not seeded). Also an error is recorded on the thread's
5882 error queue. New function RAND_pseudo_bytes() generates output that is
5883 guaranteed to be unique but not unpredictable. RAND_add is like
5884 RAND_seed, but takes an extra argument for an entropy estimate
5885 (RAND_seed always assumes full entropy).
5886 [Ulf Möller]
5887
5888 *) Do more iterations of Rabin-Miller probable prime test (specifically,
5889 3 for 1024-bit primes, 6 for 512-bit primes, 12 for 256-bit primes
5890 instead of only 2 for all lengths; see BN_prime_checks_for_size definition
5891 in crypto/bn/bn_prime.c for the complete table). This guarantees a
5892 false-positive rate of at most 2^-80 for random input.
5893 [Bodo Moeller]
5894
5895 *) Rewrite ssl3_read_n (ssl/s3_pkt.c) avoiding a couple of bugs.
5896 [Bodo Moeller]
5897
5898 *) New function X509_CTX_rget_chain() (renamed to X509_CTX_get1_chain
5899 in the 0.9.5 release), this returns the chain
5900 from an X509_CTX structure with a dup of the stack and all
5901 the X509 reference counts upped: so the stack will exist
5902 after X509_CTX_cleanup() has been called. Modify pkcs12.c
5903 to use this.
5904
5905 Also make SSL_SESSION_print() print out the verify return
5906 code.
5907 [Steve Henson]
5908
5909 *) Add manpage for the pkcs12 command. Also change the default
5910 behaviour so MAC iteration counts are used unless the new
5911 -nomaciter option is used. This improves file security and
5912 only older versions of MSIE (4.0 for example) need it.
5913 [Steve Henson]
5914
5915 *) Honor the no-xxx Configure options when creating .DEF files.
5916 [Ulf Möller]
5917
5918 *) Add PKCS#10 attributes to field table: challengePassword,
5919 unstructuredName and unstructuredAddress. These are taken from
5920 draft PKCS#9 v2.0 but are compatible with v1.2 provided no
5921 international characters are used.
5922
5923 More changes to X509_ATTRIBUTE code: allow the setting of types
5924 based on strings. Remove the 'loc' parameter when adding
5925 attributes because these will be a SET OF encoding which is sorted
5926 in ASN1 order.
5927 [Steve Henson]
5928
5929 *) Initial changes to the 'req' utility to allow request generation
5930 automation. This will allow an application to just generate a template
5931 file containing all the field values and have req construct the
5932 request.
5933
5934 Initial support for X509_ATTRIBUTE handling. Stacks of these are
5935 used all over the place including certificate requests and PKCS#7
5936 structures. They are currently handled manually where necessary with
5937 some primitive wrappers for PKCS#7. The new functions behave in a
5938 manner analogous to the X509 extension functions: they allow
5939 attributes to be looked up by NID and added.
5940
5941 Later something similar to the X509V3 code would be desirable to
5942 automatically handle the encoding, decoding and printing of the
5943 more complex types. The string types like challengePassword can
5944 be handled by the string table functions.
5945
5946 Also modified the multi byte string table handling. Now there is
5947 a 'global mask' which masks out certain types. The table itself
5948 can use the flag STABLE_NO_MASK to ignore the mask setting: this
5949 is useful when for example there is only one permissible type
5950 (as in countryName) and using the mask might result in no valid
5951 types at all.
5952 [Steve Henson]
5953
5954 *) Clean up 'Finished' handling, and add functions SSL_get_finished and
5955 SSL_get_peer_finished to allow applications to obtain the latest
5956 Finished messages sent to the peer or expected from the peer,
5957 respectively. (SSL_get_peer_finished is usually the Finished message
5958 actually received from the peer, otherwise the protocol will be aborted.)
5959
5960 As the Finished message are message digests of the complete handshake
5961 (with a total of 192 bits for TLS 1.0 and more for SSL 3.0), they can
5962 be used for external authentication procedures when the authentication
5963 provided by SSL/TLS is not desired or is not enough.
5964 [Bodo Moeller]
5965
5966 *) Enhanced support for Alpha Linux is added. Now ./config checks if
5967 the host supports BWX extension and if Compaq C is present on the
5968 $PATH. Just exploiting of the BWX extension results in 20-30%
5969 performance kick for some algorithms, e.g. DES and RC4 to mention
5970 a couple. Compaq C in turn generates ~20% faster code for MD5 and
5971 SHA1.
5972 [Andy Polyakov]
5973
5974 *) Add support for MS "fast SGC". This is arguably a violation of the
5975 SSL3/TLS protocol. Netscape SGC does two handshakes: the first with
5976 weak crypto and after checking the certificate is SGC a second one
5977 with strong crypto. MS SGC stops the first handshake after receiving
5978 the server certificate message and sends a second client hello. Since
5979 a server will typically do all the time consuming operations before
5980 expecting any further messages from the client (server key exchange
5981 is the most expensive) there is little difference between the two.
5982
5983 To get OpenSSL to support MS SGC we have to permit a second client
5984 hello message after we have sent server done. In addition we have to
5985 reset the MAC if we do get this second client hello.
5986 [Steve Henson]
5987
5988 *) Add a function 'd2i_AutoPrivateKey()' this will automatically decide
5989 if a DER encoded private key is RSA or DSA traditional format. Changed
5990 d2i_PrivateKey_bio() to use it. This is only needed for the "traditional"
5991 format DER encoded private key. Newer code should use PKCS#8 format which
5992 has the key type encoded in the ASN1 structure. Added DER private key
5993 support to pkcs8 application.
5994 [Steve Henson]
5995
5996 *) SSL 3/TLS 1 servers now don't request certificates when an anonymous
5997 ciphersuites has been selected (as required by the SSL 3/TLS 1
5998 specifications). Exception: When SSL_VERIFY_FAIL_IF_NO_PEER_CERT
5999 is set, we interpret this as a request to violate the specification
6000 (the worst that can happen is a handshake failure, and 'correct'
6001 behaviour would result in a handshake failure anyway).
6002 [Bodo Moeller]
6003
6004 *) In SSL_CTX_add_session, take into account that there might be multiple
6005 SSL_SESSION structures with the same session ID (e.g. when two threads
6006 concurrently obtain them from an external cache).
6007 The internal cache can handle only one SSL_SESSION with a given ID,
6008 so if there's a conflict, we now throw out the old one to achieve
6009 consistency.
6010 [Bodo Moeller]
6011
6012 *) Add OIDs for idea and blowfish in CBC mode. This will allow both
6013 to be used in PKCS#5 v2.0 and S/MIME. Also add checking to
6014 some routines that use cipher OIDs: some ciphers do not have OIDs
6015 defined and so they cannot be used for S/MIME and PKCS#5 v2.0 for
6016 example.
6017 [Steve Henson]
6018
6019 *) Simplify the trust setting structure and code. Now we just have
6020 two sequences of OIDs for trusted and rejected settings. These will
6021 typically have values the same as the extended key usage extension
6022 and any application specific purposes.
6023
6024 The trust checking code now has a default behaviour: it will just
6025 check for an object with the same NID as the passed id. Functions can
6026 be provided to override either the default behaviour or the behaviour
6027 for a given id. SSL client, server and email already have functions
6028 in place for compatibility: they check the NID and also return "trusted"
6029 if the certificate is self signed.
6030 [Steve Henson]
6031
6032 *) Add d2i,i2d bio/fp functions for PrivateKey: these convert the
6033 traditional format into an EVP_PKEY structure.
6034 [Steve Henson]
6035
6036 *) Add a password callback function PEM_cb() which either prompts for
6037 a password if usr_data is NULL or otherwise assumes it is a null
6038 terminated password. Allow passwords to be passed on command line
6039 environment or config files in a few more utilities.
6040 [Steve Henson]
6041
6042 *) Add a bunch of DER and PEM functions to handle PKCS#8 format private
6043 keys. Add some short names for PKCS#8 PBE algorithms and allow them
6044 to be specified on the command line for the pkcs8 and pkcs12 utilities.
6045 Update documentation.
6046 [Steve Henson]
6047
6048 *) Support for ASN1 "NULL" type. This could be handled before by using
6049 ASN1_TYPE but there wasn't any function that would try to read a NULL
6050 and produce an error if it couldn't. For compatibility we also have
6051 ASN1_NULL_new() and ASN1_NULL_free() functions but these are faked and
6052 don't allocate anything because they don't need to.
6053 [Steve Henson]
6054
6055 *) Initial support for MacOS is now provided. Examine INSTALL.MacOS
6056 for details.
6057 [Andy Polyakov, Roy Woods <roy@centicsystems.ca>]
6058
6059 *) Rebuild of the memory allocation routines used by OpenSSL code and
6060 possibly others as well. The purpose is to make an interface that
6061 provide hooks so anyone can build a separate set of allocation and
6062 deallocation routines to be used by OpenSSL, for example memory
6063 pool implementations, or something else, which was previously hard
6064 since Malloc(), Realloc() and Free() were defined as macros having
6065 the values malloc, realloc and free, respectively (except for Win32
6066 compilations). The same is provided for memory debugging code.
6067 OpenSSL already comes with functionality to find memory leaks, but
6068 this gives people a chance to debug other memory problems.
6069
6070 With these changes, a new set of functions and macros have appeared:
6071
6072 CRYPTO_set_mem_debug_functions() [F]
6073 CRYPTO_get_mem_debug_functions() [F]
6074 CRYPTO_dbg_set_options() [F]
6075 CRYPTO_dbg_get_options() [F]
6076 CRYPTO_malloc_debug_init() [M]
6077
6078 The memory debug functions are NULL by default, unless the library
6079 is compiled with CRYPTO_MDEBUG or friends is defined. If someone
6080 wants to debug memory anyway, CRYPTO_malloc_debug_init() (which
6081 gives the standard debugging functions that come with OpenSSL) or
6082 CRYPTO_set_mem_debug_functions() (tells OpenSSL to use functions
6083 provided by the library user) must be used. When the standard
6084 debugging functions are used, CRYPTO_dbg_set_options can be used to
6085 request additional information:
6086 CRYPTO_dbg_set_options(V_CYRPTO_MDEBUG_xxx) corresponds to setting
6087 the CRYPTO_MDEBUG_xxx macro when compiling the library.
6088
6089 Also, things like CRYPTO_set_mem_functions will always give the
6090 expected result (the new set of functions is used for allocation
6091 and deallocation) at all times, regardless of platform and compiler
6092 options.
6093
6094 To finish it up, some functions that were never use in any other
6095 way than through macros have a new API and new semantic:
6096
6097 CRYPTO_dbg_malloc()
6098 CRYPTO_dbg_realloc()
6099 CRYPTO_dbg_free()
6100
6101 All macros of value have retained their old syntax.
6102 [Richard Levitte and Bodo Moeller]
6103
6104 *) Some S/MIME fixes. The OID for SMIMECapabilities was wrong, the
6105 ordering of SMIMECapabilities wasn't in "strength order" and there
6106 was a missing NULL in the AlgorithmIdentifier for the SHA1 signature
6107 algorithm.
6108 [Steve Henson]
6109
6110 *) Some ASN1 types with illegal zero length encoding (INTEGER,
6111 ENUMERATED and OBJECT IDENTIFIER) choked the ASN1 routines.
6112 [Frans Heymans <fheymans@isaserver.be>, modified by Steve Henson]
6113
6114 *) Merge in my S/MIME library for OpenSSL. This provides a simple
6115 S/MIME API on top of the PKCS#7 code, a MIME parser (with enough
6116 functionality to handle multipart/signed properly) and a utility
6117 called 'smime' to call all this stuff. This is based on code I
6118 originally wrote for Celo who have kindly allowed it to be
6119 included in OpenSSL.
6120 [Steve Henson]
6121
6122 *) Add variants des_set_key_checked and des_set_key_unchecked of
6123 des_set_key (aka des_key_sched). Global variable des_check_key
6124 decides which of these is called by des_set_key; this way
6125 des_check_key behaves as it always did, but applications and
6126 the library itself, which was buggy for des_check_key == 1,
6127 have a cleaner way to pick the version they need.
6128 [Bodo Moeller]
6129
6130 *) New function PKCS12_newpass() which changes the password of a
6131 PKCS12 structure.
6132 [Steve Henson]
6133
6134 *) Modify X509_TRUST and X509_PURPOSE so it also uses a static and
6135 dynamic mix. In both cases the ids can be used as an index into the
6136 table. Also modified the X509_TRUST_add() and X509_PURPOSE_add()
6137 functions so they accept a list of the field values and the
6138 application doesn't need to directly manipulate the X509_TRUST
6139 structure.
6140 [Steve Henson]
6141
6142 *) Modify the ASN1_STRING_TABLE stuff so it also uses bsearch and doesn't
6143 need initialising.
6144 [Steve Henson]
6145
6146 *) Modify the way the V3 extension code looks up extensions. This now
6147 works in a similar way to the object code: we have some "standard"
6148 extensions in a static table which is searched with OBJ_bsearch()
6149 and the application can add dynamic ones if needed. The file
6150 crypto/x509v3/ext_dat.h now has the info: this file needs to be
6151 updated whenever a new extension is added to the core code and kept
6152 in ext_nid order. There is a simple program 'tabtest.c' which checks
6153 this. New extensions are not added too often so this file can readily
6154 be maintained manually.
6155
6156 There are two big advantages in doing things this way. The extensions
6157 can be looked up immediately and no longer need to be "added" using
6158 X509V3_add_standard_extensions(): this function now does nothing.
6159 [Side note: I get *lots* of email saying the extension code doesn't
6160 work because people forget to call this function]
6161 Also no dynamic allocation is done unless new extensions are added:
6162 so if we don't add custom extensions there is no need to call
6163 X509V3_EXT_cleanup().
6164 [Steve Henson]
6165
6166 *) Modify enc utility's salting as follows: make salting the default. Add a
6167 magic header, so unsalted files fail gracefully instead of just decrypting
6168 to garbage. This is because not salting is a big security hole, so people
6169 should be discouraged from doing it.
6170 [Ben Laurie]
6171
6172 *) Fixes and enhancements to the 'x509' utility. It allowed a message
6173 digest to be passed on the command line but it only used this
6174 parameter when signing a certificate. Modified so all relevant
6175 operations are affected by the digest parameter including the
6176 -fingerprint and -x509toreq options. Also -x509toreq choked if a
6177 DSA key was used because it didn't fix the digest.
6178 [Steve Henson]
6179
6180 *) Initial certificate chain verify code. Currently tests the untrusted
6181 certificates for consistency with the verify purpose (which is set
6182 when the X509_STORE_CTX structure is set up) and checks the pathlength.
6183
6184 There is a NO_CHAIN_VERIFY compilation option to keep the old behaviour:
6185 this is because it will reject chains with invalid extensions whereas
6186 every previous version of OpenSSL and SSLeay made no checks at all.
6187
6188 Trust code: checks the root CA for the relevant trust settings. Trust
6189 settings have an initial value consistent with the verify purpose: e.g.
6190 if the verify purpose is for SSL client use it expects the CA to be
6191 trusted for SSL client use. However the default value can be changed to
6192 permit custom trust settings: one example of this would be to only trust
6193 certificates from a specific "secure" set of CAs.
6194
6195 Also added X509_STORE_CTX_new() and X509_STORE_CTX_free() functions
6196 which should be used for version portability: especially since the
6197 verify structure is likely to change more often now.
6198
6199 SSL integration. Add purpose and trust to SSL_CTX and SSL and functions
6200 to set them. If not set then assume SSL clients will verify SSL servers
6201 and vice versa.
6202
6203 Two new options to the verify program: -untrusted allows a set of
6204 untrusted certificates to be passed in and -purpose which sets the
6205 intended purpose of the certificate. If a purpose is set then the
6206 new chain verify code is used to check extension consistency.
6207 [Steve Henson]
6208
6209 *) Support for the authority information access extension.
6210 [Steve Henson]
6211
6212 *) Modify RSA and DSA PEM read routines to transparently handle
6213 PKCS#8 format private keys. New *_PUBKEY_* functions that handle
6214 public keys in a format compatible with certificate
6215 SubjectPublicKeyInfo structures. Unfortunately there were already
6216 functions called *_PublicKey_* which used various odd formats so
6217 these are retained for compatibility: however the DSA variants were
6218 never in a public release so they have been deleted. Changed dsa/rsa
6219 utilities to handle the new format: note no releases ever handled public
6220 keys so we should be OK.
6221
6222 The primary motivation for this change is to avoid the same fiasco
6223 that dogs private keys: there are several incompatible private key
6224 formats some of which are standard and some OpenSSL specific and
6225 require various evil hacks to allow partial transparent handling and
6226 even then it doesn't work with DER formats. Given the option anything
6227 other than PKCS#8 should be dumped: but the other formats have to
6228 stay in the name of compatibility.
6229
6230 With public keys and the benefit of hindsight one standard format
6231 is used which works with EVP_PKEY, RSA or DSA structures: though
6232 it clearly returns an error if you try to read the wrong kind of key.
6233
6234 Added a -pubkey option to the 'x509' utility to output the public key.
6235 Also rename the EVP_PKEY_get_*() to EVP_PKEY_rget_*()
6236 (renamed to EVP_PKEY_get1_*() in the OpenSSL 0.9.5 release) and add
6237 EVP_PKEY_rset_*() functions (renamed to EVP_PKEY_set1_*())
6238 that do the same as the EVP_PKEY_assign_*() except they up the
6239 reference count of the added key (they don't "swallow" the
6240 supplied key).
6241 [Steve Henson]
6242
6243 *) Fixes to crypto/x509/by_file.c the code to read in certificates and
6244 CRLs would fail if the file contained no certificates or no CRLs:
6245 added a new function to read in both types and return the number
6246 read: this means that if none are read it will be an error. The
6247 DER versions of the certificate and CRL reader would always fail
6248 because it isn't possible to mix certificates and CRLs in DER format
6249 without choking one or the other routine. Changed this to just read
6250 a certificate: this is the best we can do. Also modified the code
6251 in apps/verify.c to take notice of return codes: it was previously
6252 attempting to read in certificates from NULL pointers and ignoring
6253 any errors: this is one reason why the cert and CRL reader seemed
6254 to work. It doesn't check return codes from the default certificate
6255 routines: these may well fail if the certificates aren't installed.
6256 [Steve Henson]
6257
6258 *) Code to support otherName option in GeneralName.
6259 [Steve Henson]
6260
6261 *) First update to verify code. Change the verify utility
6262 so it warns if it is passed a self signed certificate:
6263 for consistency with the normal behaviour. X509_verify
6264 has been modified to it will now verify a self signed
6265 certificate if *exactly* the same certificate appears
6266 in the store: it was previously impossible to trust a
6267 single self signed certificate. This means that:
6268 openssl verify ss.pem
6269 now gives a warning about a self signed certificate but
6270 openssl verify -CAfile ss.pem ss.pem
6271 is OK.
6272 [Steve Henson]
6273
6274 *) For servers, store verify_result in SSL_SESSION data structure
6275 (and add it to external session representation).
6276 This is needed when client certificate verifications fails,
6277 but an application-provided verification callback (set by
6278 SSL_CTX_set_cert_verify_callback) allows accepting the session
6279 anyway (i.e. leaves x509_store_ctx->error != X509_V_OK
6280 but returns 1): When the session is reused, we have to set
6281 ssl->verify_result to the appropriate error code to avoid
6282 security holes.
6283 [Bodo Moeller, problem pointed out by Lutz Jaenicke]
6284
6285 *) Fix a bug in the new PKCS#7 code: it didn't consider the
6286 case in PKCS7_dataInit() where the signed PKCS7 structure
6287 didn't contain any existing data because it was being created.
6288 [Po-Cheng Chen <pocheng@nst.com.tw>, slightly modified by Steve Henson]
6289
6290 *) Add a salt to the key derivation routines in enc.c. This
6291 forms the first 8 bytes of the encrypted file. Also add a
6292 -S option to allow a salt to be input on the command line.
6293 [Steve Henson]
6294
6295 *) New function X509_cmp(). Oddly enough there wasn't a function
6296 to compare two certificates. We do this by working out the SHA1
6297 hash and comparing that. X509_cmp() will be needed by the trust
6298 code.
6299 [Steve Henson]
6300
6301 *) SSL_get1_session() is like SSL_get_session(), but increments
6302 the reference count in the SSL_SESSION returned.
6303 [Geoff Thorpe <geoff@eu.c2.net>]
6304
6305 *) Fix for 'req': it was adding a null to request attributes.
6306 Also change the X509_LOOKUP and X509_INFO code to handle
6307 certificate auxiliary information.
6308 [Steve Henson]
6309
6310 *) Add support for 40 and 64 bit RC2 and RC4 algorithms: document
6311 the 'enc' command.
6312 [Steve Henson]
6313
6314 *) Add the possibility to add extra information to the memory leak
6315 detecting output, to form tracebacks, showing from where each
6316 allocation was originated: CRYPTO_push_info("constant string") adds
6317 the string plus current file name and line number to a per-thread
6318 stack, CRYPTO_pop_info() does the obvious, CRYPTO_remove_all_info()
6319 is like calling CYRPTO_pop_info() until the stack is empty.
6320 Also updated memory leak detection code to be multi-thread-safe.
6321 [Richard Levitte]
6322
6323 *) Add options -text and -noout to pkcs7 utility and delete the
6324 encryption options which never did anything. Update docs.
6325 [Steve Henson]
6326
6327 *) Add options to some of the utilities to allow the pass phrase
6328 to be included on either the command line (not recommended on
6329 OSes like Unix) or read from the environment. Update the
6330 manpages and fix a few bugs.
6331 [Steve Henson]
6332
6333 *) Add a few manpages for some of the openssl commands.
6334 [Steve Henson]
6335
6336 *) Fix the -revoke option in ca. It was freeing up memory twice,
6337 leaking and not finding already revoked certificates.
6338 [Steve Henson]
6339
6340 *) Extensive changes to support certificate auxiliary information.
6341 This involves the use of X509_CERT_AUX structure and X509_AUX
6342 functions. An X509_AUX function such as PEM_read_X509_AUX()
6343 can still read in a certificate file in the usual way but it
6344 will also read in any additional "auxiliary information". By
6345 doing things this way a fair degree of compatibility can be
6346 retained: existing certificates can have this information added
6347 using the new 'x509' options.
6348
6349 Current auxiliary information includes an "alias" and some trust
6350 settings. The trust settings will ultimately be used in enhanced
6351 certificate chain verification routines: currently a certificate
6352 can only be trusted if it is self signed and then it is trusted
6353 for all purposes.
6354 [Steve Henson]
6355
6356 *) Fix assembler for Alpha (tested only on DEC OSF not Linux or *BSD).
6357 The problem was that one of the replacement routines had not been working
6358 since SSLeay releases. For now the offending routine has been replaced
6359 with non-optimised assembler. Even so, this now gives around 95%
6360 performance improvement for 1024 bit RSA signs.
6361 [Mark Cox]
6362
6363 *) Hack to fix PKCS#7 decryption when used with some unorthodox RC2
6364 handling. Most clients have the effective key size in bits equal to
6365 the key length in bits: so a 40 bit RC2 key uses a 40 bit (5 byte) key.
6366 A few however don't do this and instead use the size of the decrypted key
6367 to determine the RC2 key length and the AlgorithmIdentifier to determine
6368 the effective key length. In this case the effective key length can still
6369 be 40 bits but the key length can be 168 bits for example. This is fixed
6370 by manually forcing an RC2 key into the EVP_PKEY structure because the
6371 EVP code can't currently handle unusual RC2 key sizes: it always assumes
6372 the key length and effective key length are equal.
6373 [Steve Henson]
6374
6375 *) Add a bunch of functions that should simplify the creation of
6376 X509_NAME structures. Now you should be able to do:
6377 X509_NAME_add_entry_by_txt(nm, "CN", MBSTRING_ASC, "Steve", -1, -1, 0);
6378 and have it automatically work out the correct field type and fill in
6379 the structures. The more adventurous can try:
6380 X509_NAME_add_entry_by_txt(nm, field, MBSTRING_UTF8, str, -1, -1, 0);
6381 and it will (hopefully) work out the correct multibyte encoding.
6382 [Steve Henson]
6383
6384 *) Change the 'req' utility to use the new field handling and multibyte
6385 copy routines. Before the DN field creation was handled in an ad hoc
6386 way in req, ca, and x509 which was rather broken and didn't support
6387 BMPStrings or UTF8Strings. Since some software doesn't implement
6388 BMPStrings or UTF8Strings yet, they can be enabled using the config file
6389 using the dirstring_type option. See the new comment in the default
6390 openssl.cnf for more info.
6391 [Steve Henson]
6392
6393 *) Make crypto/rand/md_rand.c more robust:
6394 - Assure unique random numbers after fork().
6395 - Make sure that concurrent threads access the global counter and
6396 md serializably so that we never lose entropy in them
6397 or use exactly the same state in multiple threads.
6398 Access to the large state is not always serializable because
6399 the additional locking could be a performance killer, and
6400 md should be large enough anyway.
6401 [Bodo Moeller]
6402
6403 *) New file apps/app_rand.c with commonly needed functionality
6404 for handling the random seed file.
6405
6406 Use the random seed file in some applications that previously did not:
6407 ca,
6408 dsaparam -genkey (which also ignored its '-rand' option),
6409 s_client,
6410 s_server,
6411 x509 (when signing).
6412 Except on systems with /dev/urandom, it is crucial to have a random
6413 seed file at least for key creation, DSA signing, and for DH exchanges;
6414 for RSA signatures we could do without one.
6415
6416 gendh and gendsa (unlike genrsa) used to read only the first byte
6417 of each file listed in the '-rand' option. The function as previously
6418 found in genrsa is now in app_rand.c and is used by all programs
6419 that support '-rand'.
6420 [Bodo Moeller]
6421
6422 *) In RAND_write_file, use mode 0600 for creating files;
6423 don't just chmod when it may be too late.
6424 [Bodo Moeller]
6425
6426 *) Report an error from X509_STORE_load_locations
6427 when X509_LOOKUP_load_file or X509_LOOKUP_add_dir failed.
6428 [Bill Perry]
6429
6430 *) New function ASN1_mbstring_copy() this copies a string in either
6431 ASCII, Unicode, Universal (4 bytes per character) or UTF8 format
6432 into an ASN1_STRING type. A mask of permissible types is passed
6433 and it chooses the "minimal" type to use or an error if not type
6434 is suitable.
6435 [Steve Henson]
6436
6437 *) Add function equivalents to the various macros in asn1.h. The old
6438 macros are retained with an M_ prefix. Code inside the library can
6439 use the M_ macros. External code (including the openssl utility)
6440 should *NOT* in order to be "shared library friendly".
6441 [Steve Henson]
6442
6443 *) Add various functions that can check a certificate's extensions
6444 to see if it usable for various purposes such as SSL client,
6445 server or S/MIME and CAs of these types. This is currently
6446 VERY EXPERIMENTAL but will ultimately be used for certificate chain
6447 verification. Also added a -purpose flag to x509 utility to
6448 print out all the purposes.
6449 [Steve Henson]
6450
6451 *) Add a CRYPTO_EX_DATA to X509 certificate structure and associated
6452 functions.
6453 [Steve Henson]
6454
6455 *) New X509V3_{X509,CRL,REVOKED}_get_d2i() functions. These will search
6456 for, obtain and decode and extension and obtain its critical flag.
6457 This allows all the necessary extension code to be handled in a
6458 single function call.
6459 [Steve Henson]
6460
6461 *) RC4 tune-up featuring 30-40% performance improvement on most RISC
6462 platforms. See crypto/rc4/rc4_enc.c for further details.
6463 [Andy Polyakov]
6464
6465 *) New -noout option to asn1parse. This causes no output to be produced
6466 its main use is when combined with -strparse and -out to extract data
6467 from a file (which may not be in ASN.1 format).
6468 [Steve Henson]
6469
6470 *) Fix for pkcs12 program. It was hashing an invalid certificate pointer
6471 when producing the local key id.
6472 [Richard Levitte <levitte@stacken.kth.se>]
6473
6474 *) New option -dhparam in s_server. This allows a DH parameter file to be
6475 stated explicitly. If it is not stated then it tries the first server
6476 certificate file. The previous behaviour hard coded the filename
6477 "server.pem".
6478 [Steve Henson]
6479
6480 *) Add -pubin and -pubout options to the rsa and dsa commands. These allow
6481 a public key to be input or output. For example:
6482 openssl rsa -in key.pem -pubout -out pubkey.pem
6483 Also added necessary DSA public key functions to handle this.
6484 [Steve Henson]
6485
6486 *) Fix so PKCS7_dataVerify() doesn't crash if no certificates are contained
6487 in the message. This was handled by allowing
6488 X509_find_by_issuer_and_serial() to tolerate a NULL passed to it.
6489 [Steve Henson, reported by Sampo Kellomaki <sampo@mail.neuronio.pt>]
6490
6491 *) Fix for bug in d2i_ASN1_bytes(): other ASN1 functions add an extra null
6492 to the end of the strings whereas this didn't. This would cause problems
6493 if strings read with d2i_ASN1_bytes() were later modified.
6494 [Steve Henson, reported by Arne Ansper <arne@ats.cyber.ee>]
6495
6496 *) Fix for base64 decode bug. When a base64 bio reads only one line of
6497 data and it contains EOF it will end up returning an error. This is
6498 caused by input 46 bytes long. The cause is due to the way base64
6499 BIOs find the start of base64 encoded data. They do this by trying a
6500 trial decode on each line until they find one that works. When they
6501 do a flag is set and it starts again knowing it can pass all the
6502 data directly through the decoder. Unfortunately it doesn't reset
6503 the context it uses. This means that if EOF is reached an attempt
6504 is made to pass two EOFs through the context and this causes the
6505 resulting error. This can also cause other problems as well. As is
6506 usual with these problems it takes *ages* to find and the fix is
6507 trivial: move one line.
6508 [Steve Henson, reported by ian@uns.ns.ac.yu (Ivan Nejgebauer) ]
6509
6510 *) Ugly workaround to get s_client and s_server working under Windows. The
6511 old code wouldn't work because it needed to select() on sockets and the
6512 tty (for keypresses and to see if data could be written). Win32 only
6513 supports select() on sockets so we select() with a 1s timeout on the
6514 sockets and then see if any characters are waiting to be read, if none
6515 are present then we retry, we also assume we can always write data to
6516 the tty. This isn't nice because the code then blocks until we've
6517 received a complete line of data and it is effectively polling the
6518 keyboard at 1s intervals: however it's quite a bit better than not
6519 working at all :-) A dedicated Windows application might handle this
6520 with an event loop for example.
6521 [Steve Henson]
6522
6523 *) Enhance RSA_METHOD structure. Now there are two extra methods, rsa_sign
6524 and rsa_verify. When the RSA_FLAGS_SIGN_VER option is set these functions
6525 will be called when RSA_sign() and RSA_verify() are used. This is useful
6526 if rsa_pub_dec() and rsa_priv_enc() equivalents are not available.
6527 For this to work properly RSA_public_decrypt() and RSA_private_encrypt()
6528 should *not* be used: RSA_sign() and RSA_verify() must be used instead.
6529 This necessitated the support of an extra signature type NID_md5_sha1
6530 for SSL signatures and modifications to the SSL library to use it instead
6531 of calling RSA_public_decrypt() and RSA_private_encrypt().
6532 [Steve Henson]
6533
6534 *) Add new -verify -CAfile and -CApath options to the crl program, these
6535 will lookup a CRL issuers certificate and verify the signature in a
6536 similar way to the verify program. Tidy up the crl program so it
6537 no longer accesses structures directly. Make the ASN1 CRL parsing a bit
6538 less strict. It will now permit CRL extensions even if it is not
6539 a V2 CRL: this will allow it to tolerate some broken CRLs.
6540 [Steve Henson]
6541
6542 *) Initialize all non-automatic variables each time one of the openssl
6543 sub-programs is started (this is necessary as they may be started
6544 multiple times from the "OpenSSL>" prompt).
6545 [Lennart Bang, Bodo Moeller]
6546
6547 *) Preliminary compilation option RSA_NULL which disables RSA crypto without
6548 removing all other RSA functionality (this is what NO_RSA does). This
6549 is so (for example) those in the US can disable those operations covered
6550 by the RSA patent while allowing storage and parsing of RSA keys and RSA
6551 key generation.
6552 [Steve Henson]
6553
6554 *) Non-copying interface to BIO pairs.
6555 (still largely untested)
6556 [Bodo Moeller]
6557
6558 *) New function ANS1_tag2str() to convert an ASN1 tag to a descriptive
6559 ASCII string. This was handled independently in various places before.
6560 [Steve Henson]
6561
6562 *) New functions UTF8_getc() and UTF8_putc() that parse and generate
6563 UTF8 strings a character at a time.
6564 [Steve Henson]
6565
6566 *) Use client_version from client hello to select the protocol
6567 (s23_srvr.c) and for RSA client key exchange verification
6568 (s3_srvr.c), as required by the SSL 3.0/TLS 1.0 specifications.
6569 [Bodo Moeller]
6570
6571 *) Add various utility functions to handle SPKACs, these were previously
6572 handled by poking round in the structure internals. Added new function
6573 NETSCAPE_SPKI_print() to print out SPKAC and a new utility 'spkac' to
6574 print, verify and generate SPKACs. Based on an original idea from
6575 Massimiliano Pala <madwolf@comune.modena.it> but extensively modified.
6576 [Steve Henson]
6577
6578 *) RIPEMD160 is operational on all platforms and is back in 'make test'.
6579 [Andy Polyakov]
6580
6581 *) Allow the config file extension section to be overwritten on the
6582 command line. Based on an original idea from Massimiliano Pala
6583 <madwolf@comune.modena.it>. The new option is called -extensions
6584 and can be applied to ca, req and x509. Also -reqexts to override
6585 the request extensions in req and -crlexts to override the crl extensions
6586 in ca.
6587 [Steve Henson]
6588
6589 *) Add new feature to the SPKAC handling in ca. Now you can include
6590 the same field multiple times by preceding it by "XXXX." for example:
6591 1.OU="Unit name 1"
6592 2.OU="Unit name 2"
6593 this is the same syntax as used in the req config file.
6594 [Steve Henson]
6595
6596 *) Allow certificate extensions to be added to certificate requests. These
6597 are specified in a 'req_extensions' option of the req section of the
6598 config file. They can be printed out with the -text option to req but
6599 are otherwise ignored at present.
6600 [Steve Henson]
6601
6602 *) Fix a horrible bug in enc_read() in crypto/evp/bio_enc.c: if the first
6603 data read consists of only the final block it would not decrypted because
6604 EVP_CipherUpdate() would correctly report zero bytes had been decrypted.
6605 A misplaced 'break' also meant the decrypted final block might not be
6606 copied until the next read.
6607 [Steve Henson]
6608
6609 *) Initial support for DH_METHOD. Again based on RSA_METHOD. Also added
6610 a few extra parameters to the DH structure: these will be useful if
6611 for example we want the value of 'q' or implement X9.42 DH.
6612 [Steve Henson]
6613
6614 *) Initial support for DSA_METHOD. This is based on the RSA_METHOD and
6615 provides hooks that allow the default DSA functions or functions on a
6616 "per key" basis to be replaced. This allows hardware acceleration and
6617 hardware key storage to be handled without major modification to the
6618 library. Also added low level modexp hooks and CRYPTO_EX structure and
6619 associated functions.
6620 [Steve Henson]
6621
6622 *) Add a new flag to memory BIOs, BIO_FLAG_MEM_RDONLY. This marks the BIO
6623 as "read only": it can't be written to and the buffer it points to will
6624 not be freed. Reading from a read only BIO is much more efficient than
6625 a normal memory BIO. This was added because there are several times when
6626 an area of memory needs to be read from a BIO. The previous method was
6627 to create a memory BIO and write the data to it, this results in two
6628 copies of the data and an O(n^2) reading algorithm. There is a new
6629 function BIO_new_mem_buf() which creates a read only memory BIO from
6630 an area of memory. Also modified the PKCS#7 routines to use read only
6631 memory BIOs.
6632 [Steve Henson]
6633
6634 *) Bugfix: ssl23_get_client_hello did not work properly when called in
6635 state SSL23_ST_SR_CLNT_HELLO_B, i.e. when the first 7 bytes of
6636 a SSLv2-compatible client hello for SSLv3 or TLSv1 could be read,
6637 but a retry condition occured while trying to read the rest.
6638 [Bodo Moeller]
6639
6640 *) The PKCS7_ENC_CONTENT_new() function was setting the content type as
6641 NID_pkcs7_encrypted by default: this was wrong since this should almost
6642 always be NID_pkcs7_data. Also modified the PKCS7_set_type() to handle
6643 the encrypted data type: this is a more sensible place to put it and it
6644 allows the PKCS#12 code to be tidied up that duplicated this
6645 functionality.
6646 [Steve Henson]
6647
6648 *) Changed obj_dat.pl script so it takes its input and output files on
6649 the command line. This should avoid shell escape redirection problems
6650 under Win32.
6651 [Steve Henson]
6652
6653 *) Initial support for certificate extension requests, these are included
6654 in things like Xenroll certificate requests. Included functions to allow
6655 extensions to be obtained and added.
6656 [Steve Henson]
6657
6658 *) -crlf option to s_client and s_server for sending newlines as
6659 CRLF (as required by many protocols).
6660 [Bodo Moeller]
6661
6662 Changes between 0.9.3a and 0.9.4 [09 Aug 1999]
6663
6664 *) Install libRSAglue.a when OpenSSL is built with RSAref.
6665 [Ralf S. Engelschall]
6666
6667 *) A few more ``#ifndef NO_FP_API / #endif'' pairs for consistency.
6668 [Andrija Antonijevic <TheAntony2@bigfoot.com>]
6669
6670 *) Fix -startdate and -enddate (which was missing) arguments to 'ca'
6671 program.
6672 [Steve Henson]
6673
6674 *) New function DSA_dup_DH, which duplicates DSA parameters/keys as
6675 DH parameters/keys (q is lost during that conversion, but the resulting
6676 DH parameters contain its length).
6677
6678 For 1024-bit p, DSA_generate_parameters followed by DSA_dup_DH is
6679 much faster than DH_generate_parameters (which creates parameters
6680 where p = 2*q + 1), and also the smaller q makes DH computations
6681 much more efficient (160-bit exponentiation instead of 1024-bit
6682 exponentiation); so this provides a convenient way to support DHE
6683 ciphersuites in SSL/TLS servers (see ssl/ssltest.c). It is of
6684 utter importance to use
6685 SSL_CTX_set_options(s_ctx, SSL_OP_SINGLE_DH_USE);
6686 or
6687 SSL_set_options(s_ctx, SSL_OP_SINGLE_DH_USE);
6688 when such DH parameters are used, because otherwise small subgroup
6689 attacks may become possible!
6690 [Bodo Moeller]
6691
6692 *) Avoid memory leak in i2d_DHparams.
6693 [Bodo Moeller]
6694
6695 *) Allow the -k option to be used more than once in the enc program:
6696 this allows the same encrypted message to be read by multiple recipients.
6697 [Steve Henson]
6698
6699 *) New function OBJ_obj2txt(buf, buf_len, a, no_name), this converts
6700 an ASN1_OBJECT to a text string. If the "no_name" parameter is set then
6701 it will always use the numerical form of the OID, even if it has a short
6702 or long name.
6703 [Steve Henson]
6704
6705 *) Added an extra RSA flag: RSA_FLAG_EXT_PKEY. Previously the rsa_mod_exp
6706 method only got called if p,q,dmp1,dmq1,iqmp components were present,
6707 otherwise bn_mod_exp was called. In the case of hardware keys for example
6708 no private key components need be present and it might store extra data
6709 in the RSA structure, which cannot be accessed from bn_mod_exp.
6710 By setting RSA_FLAG_EXT_PKEY rsa_mod_exp will always be called for
6711 private key operations.
6712 [Steve Henson]
6713
6714 *) Added support for SPARC Linux.
6715 [Andy Polyakov]
6716
6717 *) pem_password_cb function type incompatibly changed from
6718 typedef int pem_password_cb(char *buf, int size, int rwflag);
6719 to
6720 ....(char *buf, int size, int rwflag, void *userdata);
6721 so that applications can pass data to their callbacks:
6722 The PEM[_ASN1]_{read,write}... functions and macros now take an
6723 additional void * argument, which is just handed through whenever
6724 the password callback is called.
6725 [Damien Miller <dmiller@ilogic.com.au>; tiny changes by Bodo Moeller]
6726
6727 New function SSL_CTX_set_default_passwd_cb_userdata.
6728
6729 Compatibility note: As many C implementations push function arguments
6730 onto the stack in reverse order, the new library version is likely to
6731 interoperate with programs that have been compiled with the old
6732 pem_password_cb definition (PEM_whatever takes some data that
6733 happens to be on the stack as its last argument, and the callback
6734 just ignores this garbage); but there is no guarantee whatsoever that
6735 this will work.
6736
6737 *) The -DPLATFORM="\"$(PLATFORM)\"" definition and the similar -DCFLAGS=...
6738 (both in crypto/Makefile.ssl for use by crypto/cversion.c) caused
6739 problems not only on Windows, but also on some Unix platforms.
6740 To avoid problematic command lines, these definitions are now in an
6741 auto-generated file crypto/buildinf.h (created by crypto/Makefile.ssl
6742 for standard "make" builds, by util/mk1mf.pl for "mk1mf" builds).
6743 [Bodo Moeller]
6744
6745 *) MIPS III/IV assembler module is reimplemented.
6746 [Andy Polyakov]
6747
6748 *) More DES library cleanups: remove references to srand/rand and
6749 delete an unused file.
6750 [Ulf Möller]
6751
6752 *) Add support for the the free Netwide assembler (NASM) under Win32,
6753 since not many people have MASM (ml) and it can be hard to obtain.
6754 This is currently experimental but it seems to work OK and pass all
6755 the tests. Check out INSTALL.W32 for info.
6756 [Steve Henson]
6757
6758 *) Fix memory leaks in s3_clnt.c: All non-anonymous SSL3/TLS1 connections
6759 without temporary keys kept an extra copy of the server key,
6760 and connections with temporary keys did not free everything in case
6761 of an error.
6762 [Bodo Moeller]
6763
6764 *) New function RSA_check_key and new openssl rsa option -check
6765 for verifying the consistency of RSA keys.
6766 [Ulf Moeller, Bodo Moeller]
6767
6768 *) Various changes to make Win32 compile work:
6769 1. Casts to avoid "loss of data" warnings in p5_crpt2.c
6770 2. Change unsigned int to int in b_dump.c to avoid "signed/unsigned
6771 comparison" warnings.
6772 3. Add sk_<TYPE>_sort to DEF file generator and do make update.
6773 [Steve Henson]
6774
6775 *) Add a debugging option to PKCS#5 v2 key generation function: when
6776 you #define DEBUG_PKCS5V2 passwords, salts, iteration counts and
6777 derived keys are printed to stderr.
6778 [Steve Henson]
6779
6780 *) Copy the flags in ASN1_STRING_dup().
6781 [Roman E. Pavlov <pre@mo.msk.ru>]
6782
6783 *) The x509 application mishandled signing requests containing DSA
6784 keys when the signing key was also DSA and the parameters didn't match.
6785
6786 It was supposed to omit the parameters when they matched the signing key:
6787 the verifying software was then supposed to automatically use the CA's
6788 parameters if they were absent from the end user certificate.
6789
6790 Omitting parameters is no longer recommended. The test was also
6791 the wrong way round! This was probably due to unusual behaviour in
6792 EVP_cmp_parameters() which returns 1 if the parameters match.
6793 This meant that parameters were omitted when they *didn't* match and
6794 the certificate was useless. Certificates signed with 'ca' didn't have
6795 this bug.
6796 [Steve Henson, reported by Doug Erickson <Doug.Erickson@Part.NET>]
6797
6798 *) Memory leak checking (-DCRYPTO_MDEBUG) had some problems.
6799 The interface is as follows:
6800 Applications can use
6801 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_ON) aka MemCheck_start(),
6802 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_OFF) aka MemCheck_stop();
6803 "off" is now the default.
6804 The library internally uses
6805 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_DISABLE) aka MemCheck_off(),
6806 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_ENABLE) aka MemCheck_on()
6807 to disable memory-checking temporarily.
6808
6809 Some inconsistent states that previously were possible (and were
6810 even the default) are now avoided.
6811
6812 -DCRYPTO_MDEBUG_TIME is new and additionally stores the current time
6813 with each memory chunk allocated; this is occasionally more helpful
6814 than just having a counter.
6815
6816 -DCRYPTO_MDEBUG_THREAD is also new and adds the thread ID.
6817
6818 -DCRYPTO_MDEBUG_ALL enables all of the above, plus any future
6819 extensions.
6820 [Bodo Moeller]
6821
6822 *) Introduce "mode" for SSL structures (with defaults in SSL_CTX),
6823 which largely parallels "options", but is for changing API behaviour,
6824 whereas "options" are about protocol behaviour.
6825 Initial "mode" flags are:
6826
6827 SSL_MODE_ENABLE_PARTIAL_WRITE Allow SSL_write to report success when
6828 a single record has been written.
6829 SSL_MODE_ACCEPT_MOVING_WRITE_BUFFER Don't insist that SSL_write
6830 retries use the same buffer location.
6831 (But all of the contents must be
6832 copied!)
6833 [Bodo Moeller]
6834
6835 *) Bugfix: SSL_set_options ignored its parameter, only SSL_CTX_set_options
6836 worked.
6837
6838 *) Fix problems with no-hmac etc.
6839 [Ulf Möller, pointed out by Brian Wellington <bwelling@tislabs.com>]
6840
6841 *) New functions RSA_get_default_method(), RSA_set_method() and
6842 RSA_get_method(). These allows replacement of RSA_METHODs without having
6843 to mess around with the internals of an RSA structure.
6844 [Steve Henson]
6845
6846 *) Fix memory leaks in DSA_do_sign and DSA_is_prime.
6847 Also really enable memory leak checks in openssl.c and in some
6848 test programs.
6849 [Chad C. Mulligan, Bodo Moeller]
6850
6851 *) Fix a bug in d2i_ASN1_INTEGER() and i2d_ASN1_INTEGER() which can mess
6852 up the length of negative integers. This has now been simplified to just
6853 store the length when it is first determined and use it later, rather
6854 than trying to keep track of where data is copied and updating it to
6855 point to the end.
6856 [Steve Henson, reported by Brien Wheeler
6857 <bwheeler@authentica-security.com>]
6858
6859 *) Add a new function PKCS7_signatureVerify. This allows the verification
6860 of a PKCS#7 signature but with the signing certificate passed to the
6861 function itself. This contrasts with PKCS7_dataVerify which assumes the
6862 certificate is present in the PKCS#7 structure. This isn't always the
6863 case: certificates can be omitted from a PKCS#7 structure and be
6864 distributed by "out of band" means (such as a certificate database).
6865 [Steve Henson]
6866
6867 *) Complete the PEM_* macros with DECLARE_PEM versions to replace the
6868 function prototypes in pem.h, also change util/mkdef.pl to add the
6869 necessary function names.
6870 [Steve Henson]
6871
6872 *) mk1mf.pl (used by Windows builds) did not properly read the
6873 options set by Configure in the top level Makefile, and Configure
6874 was not even able to write more than one option correctly.
6875 Fixed, now "no-idea no-rc5 -DCRYPTO_MDEBUG" etc. works as intended.
6876 [Bodo Moeller]
6877
6878 *) New functions CONF_load_bio() and CONF_load_fp() to allow a config
6879 file to be loaded from a BIO or FILE pointer. The BIO version will
6880 for example allow memory BIOs to contain config info.
6881 [Steve Henson]
6882
6883 *) New function "CRYPTO_num_locks" that returns CRYPTO_NUM_LOCKS.
6884 Whoever hopes to achieve shared-library compatibility across versions
6885 must use this, not the compile-time macro.
6886 (Exercise 0.9.4: Which is the minimum library version required by
6887 such programs?)
6888 Note: All this applies only to multi-threaded programs, others don't
6889 need locks.
6890 [Bodo Moeller]
6891
6892 *) Add missing case to s3_clnt.c state machine -- one of the new SSL tests
6893 through a BIO pair triggered the default case, i.e.
6894 SSLerr(...,SSL_R_UNKNOWN_STATE).
6895 [Bodo Moeller]
6896
6897 *) New "BIO pair" concept (crypto/bio/bss_bio.c) so that applications
6898 can use the SSL library even if none of the specific BIOs is
6899 appropriate.
6900 [Bodo Moeller]
6901
6902 *) Fix a bug in i2d_DSAPublicKey() which meant it returned the wrong value
6903 for the encoded length.
6904 [Jeon KyoungHo <khjeon@sds.samsung.co.kr>]
6905
6906 *) Add initial documentation of the X509V3 functions.
6907 [Steve Henson]
6908
6909 *) Add a new pair of functions PEM_write_PKCS8PrivateKey() and
6910 PEM_write_bio_PKCS8PrivateKey() that are equivalent to
6911 PEM_write_PrivateKey() and PEM_write_bio_PrivateKey() but use the more
6912 secure PKCS#8 private key format with a high iteration count.
6913 [Steve Henson]
6914
6915 *) Fix determination of Perl interpreter: A perl or perl5
6916 _directory_ in $PATH was also accepted as the interpreter.
6917 [Ralf S. Engelschall]
6918
6919 *) Fix demos/sign/sign.c: well there wasn't anything strictly speaking
6920 wrong with it but it was very old and did things like calling
6921 PEM_ASN1_read() directly and used MD5 for the hash not to mention some
6922 unusual formatting.
6923 [Steve Henson]
6924
6925 *) Fix demos/selfsign.c: it used obsolete and deleted functions, changed
6926 to use the new extension code.
6927 [Steve Henson]
6928
6929 *) Implement the PEM_read/PEM_write functions in crypto/pem/pem_all.c
6930 with macros. This should make it easier to change their form, add extra
6931 arguments etc. Fix a few PEM prototypes which didn't have cipher as a
6932 constant.
6933 [Steve Henson]
6934
6935 *) Add to configuration table a new entry that can specify an alternative
6936 name for unistd.h (for pre-POSIX systems); we need this for NeXTstep,
6937 according to Mark Crispin <MRC@Panda.COM>.
6938 [Bodo Moeller]
6939
6940 #if 0
6941 *) DES CBC did not update the IV. Weird.
6942 [Ben Laurie]
6943 #else
6944 des_cbc_encrypt does not update the IV, but des_ncbc_encrypt does.
6945 Changing the behaviour of the former might break existing programs --
6946 where IV updating is needed, des_ncbc_encrypt can be used.
6947 #endif
6948
6949 *) When bntest is run from "make test" it drives bc to check its
6950 calculations, as well as internally checking them. If an internal check
6951 fails, it needs to cause bc to give a non-zero result or make test carries
6952 on without noticing the failure. Fixed.
6953 [Ben Laurie]
6954
6955 *) DES library cleanups.
6956 [Ulf Möller]
6957
6958 *) Add support for PKCS#5 v2.0 PBE algorithms. This will permit PKCS#8 to be
6959 used with any cipher unlike PKCS#5 v1.5 which can at most handle 64 bit
6960 ciphers. NOTE: although the key derivation function has been verified
6961 against some published test vectors it has not been extensively tested
6962 yet. Added a -v2 "cipher" option to pkcs8 application to allow the use
6963 of v2.0.
6964 [Steve Henson]
6965
6966 *) Instead of "mkdir -p", which is not fully portable, use new
6967 Perl script "util/mkdir-p.pl".
6968 [Bodo Moeller]
6969
6970 *) Rewrite the way password based encryption (PBE) is handled. It used to
6971 assume that the ASN1 AlgorithmIdentifier parameter was a PBEParameter
6972 structure. This was true for the PKCS#5 v1.5 and PKCS#12 PBE algorithms
6973 but doesn't apply to PKCS#5 v2.0 where it can be something else. Now
6974 the 'parameter' field of the AlgorithmIdentifier is passed to the
6975 underlying key generation function so it must do its own ASN1 parsing.
6976 This has also changed the EVP_PBE_CipherInit() function which now has a
6977 'parameter' argument instead of literal salt and iteration count values
6978 and the function EVP_PBE_ALGOR_CipherInit() has been deleted.
6979 [Steve Henson]
6980
6981 *) Support for PKCS#5 v1.5 compatible password based encryption algorithms
6982 and PKCS#8 functionality. New 'pkcs8' application linked to openssl.
6983 Needed to change the PEM_STRING_EVP_PKEY value which was just "PRIVATE
6984 KEY" because this clashed with PKCS#8 unencrypted string. Since this
6985 value was just used as a "magic string" and not used directly its
6986 value doesn't matter.
6987 [Steve Henson]
6988
6989 *) Introduce some semblance of const correctness to BN. Shame C doesn't
6990 support mutable.
6991 [Ben Laurie]
6992
6993 *) "linux-sparc64" configuration (ultrapenguin).
6994 [Ray Miller <ray.miller@oucs.ox.ac.uk>]
6995 "linux-sparc" configuration.
6996 [Christian Forster <fo@hawo.stw.uni-erlangen.de>]
6997
6998 *) config now generates no-xxx options for missing ciphers.
6999 [Ulf Möller]
7000
7001 *) Support the EBCDIC character set (work in progress).
7002 File ebcdic.c not yet included because it has a different license.
7003 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>]
7004
7005 *) Support BS2000/OSD-POSIX.
7006 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>]
7007
7008 *) Make callbacks for key generation use void * instead of char *.
7009 [Ben Laurie]
7010
7011 *) Make S/MIME samples compile (not yet tested).
7012 [Ben Laurie]
7013
7014 *) Additional typesafe stacks.
7015 [Ben Laurie]
7016
7017 *) New configuration variants "bsdi-elf-gcc" (BSD/OS 4.x).
7018 [Bodo Moeller]
7019
7020
7021 Changes between 0.9.3 and 0.9.3a [29 May 1999]
7022
7023 *) New configuration variant "sco5-gcc".
7024
7025 *) Updated some demos.
7026 [Sean O Riordain, Wade Scholine]
7027
7028 *) Add missing BIO_free at exit of pkcs12 application.
7029 [Wu Zhigang]
7030
7031 *) Fix memory leak in conf.c.
7032 [Steve Henson]
7033
7034 *) Updates for Win32 to assembler version of MD5.
7035 [Steve Henson]
7036
7037 *) Set #! path to perl in apps/der_chop to where we found it
7038 instead of using a fixed path.
7039 [Bodo Moeller]
7040
7041 *) SHA library changes for irix64-mips4-cc.
7042 [Andy Polyakov]
7043
7044 *) Improvements for VMS support.
7045 [Richard Levitte]
7046
7047
7048 Changes between 0.9.2b and 0.9.3 [24 May 1999]
7049
7050 *) Bignum library bug fix. IRIX 6 passes "make test" now!
7051 This also avoids the problems with SC4.2 and unpatched SC5.
7052 [Andy Polyakov <appro@fy.chalmers.se>]
7053
7054 *) New functions sk_num, sk_value and sk_set to replace the previous macros.
7055 These are required because of the typesafe stack would otherwise break
7056 existing code. If old code used a structure member which used to be STACK
7057 and is now STACK_OF (for example cert in a PKCS7_SIGNED structure) with
7058 sk_num or sk_value it would produce an error because the num, data members
7059 are not present in STACK_OF. Now it just produces a warning. sk_set
7060 replaces the old method of assigning a value to sk_value
7061 (e.g. sk_value(x, i) = y) which the library used in a few cases. Any code
7062 that does this will no longer work (and should use sk_set instead) but
7063 this could be regarded as a "questionable" behaviour anyway.
7064 [Steve Henson]
7065
7066 *) Fix most of the other PKCS#7 bugs. The "experimental" code can now
7067 correctly handle encrypted S/MIME data.
7068 [Steve Henson]
7069
7070 *) Change type of various DES function arguments from des_cblock
7071 (which means, in function argument declarations, pointer to char)
7072 to des_cblock * (meaning pointer to array with 8 char elements),
7073 which allows the compiler to do more typechecking; it was like
7074 that back in SSLeay, but with lots of ugly casts.
7075
7076 Introduce new type const_des_cblock.
7077 [Bodo Moeller]
7078
7079 *) Reorganise the PKCS#7 library and get rid of some of the more obvious
7080 problems: find RecipientInfo structure that matches recipient certificate
7081 and initialise the ASN1 structures properly based on passed cipher.
7082 [Steve Henson]
7083
7084 *) Belatedly make the BN tests actually check the results.
7085 [Ben Laurie]
7086
7087 *) Fix the encoding and decoding of negative ASN1 INTEGERS and conversion
7088 to and from BNs: it was completely broken. New compilation option
7089 NEG_PUBKEY_BUG to allow for some broken certificates that encode public
7090 key elements as negative integers.
7091 [Steve Henson]
7092
7093 *) Reorganize and speed up MD5.
7094 [Andy Polyakov <appro@fy.chalmers.se>]
7095
7096 *) VMS support.
7097 [Richard Levitte <richard@levitte.org>]
7098
7099 *) New option -out to asn1parse to allow the parsed structure to be
7100 output to a file. This is most useful when combined with the -strparse
7101 option to examine the output of things like OCTET STRINGS.
7102 [Steve Henson]
7103
7104 *) Make SSL library a little more fool-proof by not requiring any longer
7105 that SSL_set_{accept,connect}_state be called before
7106 SSL_{accept,connect} may be used (SSL_set_..._state is omitted
7107 in many applications because usually everything *appeared* to work as
7108 intended anyway -- now it really works as intended).
7109 [Bodo Moeller]
7110
7111 *) Move openssl.cnf out of lib/.
7112 [Ulf Möller]
7113
7114 *) Fix various things to let OpenSSL even pass ``egcc -pipe -O2 -Wall
7115 -Wshadow -Wpointer-arith -Wcast-align -Wmissing-prototypes
7116 -Wmissing-declarations -Wnested-externs -Winline'' with EGCS 1.1.2+
7117 [Ralf S. Engelschall]
7118
7119 *) Various fixes to the EVP and PKCS#7 code. It may now be able to
7120 handle PKCS#7 enveloped data properly.
7121 [Sebastian Akerman <sak@parallelconsulting.com>, modified by Steve]
7122
7123 *) Create a duplicate of the SSL_CTX's CERT in SSL_new instead of
7124 copying pointers. The cert_st handling is changed by this in
7125 various ways (and thus what used to be known as ctx->default_cert
7126 is now called ctx->cert, since we don't resort to s->ctx->[default_]cert
7127 any longer when s->cert does not give us what we need).
7128 ssl_cert_instantiate becomes obsolete by this change.
7129 As soon as we've got the new code right (possibly it already is?),
7130 we have solved a couple of bugs of the earlier code where s->cert
7131 was used as if it could not have been shared with other SSL structures.
7132
7133 Note that using the SSL API in certain dirty ways now will result
7134 in different behaviour than observed with earlier library versions:
7135 Changing settings for an SSL_CTX *ctx after having done s = SSL_new(ctx)
7136 does not influence s as it used to.
7137
7138 In order to clean up things more thoroughly, inside SSL_SESSION
7139 we don't use CERT any longer, but a new structure SESS_CERT
7140 that holds per-session data (if available); currently, this is
7141 the peer's certificate chain and, for clients, the server's certificate
7142 and temporary key. CERT holds only those values that can have
7143 meaningful defaults in an SSL_CTX.
7144 [Bodo Moeller]
7145
7146 *) New function X509V3_EXT_i2d() to create an X509_EXTENSION structure
7147 from the internal representation. Various PKCS#7 fixes: remove some
7148 evil casts and set the enc_dig_alg field properly based on the signing
7149 key type.
7150 [Steve Henson]
7151
7152 *) Allow PKCS#12 password to be set from the command line or the
7153 environment. Let 'ca' get its config file name from the environment
7154 variables "OPENSSL_CONF" or "SSLEAY_CONF" (for consistency with 'req'
7155 and 'x509').
7156 [Steve Henson]
7157
7158 *) Allow certificate policies extension to use an IA5STRING for the
7159 organization field. This is contrary to the PKIX definition but
7160 VeriSign uses it and IE5 only recognises this form. Document 'x509'
7161 extension option.
7162 [Steve Henson]
7163
7164 *) Add PEDANTIC compiler flag to allow compilation with gcc -pedantic,
7165 without disallowing inline assembler and the like for non-pedantic builds.
7166 [Ben Laurie]
7167
7168 *) Support Borland C++ builder.
7169 [Janez Jere <jj@void.si>, modified by Ulf Möller]
7170
7171 *) Support Mingw32.
7172 [Ulf Möller]
7173
7174 *) SHA-1 cleanups and performance enhancements.
7175 [Andy Polyakov <appro@fy.chalmers.se>]
7176
7177 *) Sparc v8plus assembler for the bignum library.
7178 [Andy Polyakov <appro@fy.chalmers.se>]
7179
7180 *) Accept any -xxx and +xxx compiler options in Configure.
7181 [Ulf Möller]
7182
7183 *) Update HPUX configuration.
7184 [Anonymous]
7185
7186 *) Add missing sk_<type>_unshift() function to safestack.h
7187 [Ralf S. Engelschall]
7188
7189 *) New function SSL_CTX_use_certificate_chain_file that sets the
7190 "extra_cert"s in addition to the certificate. (This makes sense
7191 only for "PEM" format files, as chains as a whole are not
7192 DER-encoded.)
7193 [Bodo Moeller]
7194
7195 *) Support verify_depth from the SSL API.
7196 x509_vfy.c had what can be considered an off-by-one-error:
7197 Its depth (which was not part of the external interface)
7198 was actually counting the number of certificates in a chain;
7199 now it really counts the depth.
7200 [Bodo Moeller]
7201
7202 *) Bugfix in crypto/x509/x509_cmp.c: The SSLerr macro was used
7203 instead of X509err, which often resulted in confusing error
7204 messages since the error codes are not globally unique
7205 (e.g. an alleged error in ssl3_accept when a certificate
7206 didn't match the private key).
7207
7208 *) New function SSL_CTX_set_session_id_context that allows to set a default
7209 value (so that you don't need SSL_set_session_id_context for each
7210 connection using the SSL_CTX).
7211 [Bodo Moeller]
7212
7213 *) OAEP decoding bug fix.
7214 [Ulf Möller]
7215
7216 *) Support INSTALL_PREFIX for package builders, as proposed by
7217 David Harris.
7218 [Bodo Moeller]
7219
7220 *) New Configure options "threads" and "no-threads". For systems
7221 where the proper compiler options are known (currently Solaris
7222 and Linux), "threads" is the default.
7223 [Bodo Moeller]
7224
7225 *) New script util/mklink.pl as a faster substitute for util/mklink.sh.
7226 [Bodo Moeller]
7227
7228 *) Install various scripts to $(OPENSSLDIR)/misc, not to
7229 $(INSTALLTOP)/bin -- they shouldn't clutter directories
7230 such as /usr/local/bin.
7231 [Bodo Moeller]
7232
7233 *) "make linux-shared" to build shared libraries.
7234 [Niels Poppe <niels@netbox.org>]
7235
7236 *) New Configure option no-<cipher> (rsa, idea, rc5, ...).
7237 [Ulf Möller]
7238
7239 *) Add the PKCS#12 API documentation to openssl.txt. Preliminary support for
7240 extension adding in x509 utility.
7241 [Steve Henson]
7242
7243 *) Remove NOPROTO sections and error code comments.
7244 [Ulf Möller]
7245
7246 *) Partial rewrite of the DEF file generator to now parse the ANSI
7247 prototypes.
7248 [Steve Henson]
7249
7250 *) New Configure options --prefix=DIR and --openssldir=DIR.
7251 [Ulf Möller]
7252
7253 *) Complete rewrite of the error code script(s). It is all now handled
7254 by one script at the top level which handles error code gathering,
7255 header rewriting and C source file generation. It should be much better
7256 than the old method: it now uses a modified version of Ulf's parser to
7257 read the ANSI prototypes in all header files (thus the old K&R definitions
7258 aren't needed for error creation any more) and do a better job of
7259 translating function codes into names. The old 'ASN1 error code imbedded
7260 in a comment' is no longer necessary and it doesn't use .err files which
7261 have now been deleted. Also the error code call doesn't have to appear all
7262 on one line (which resulted in some large lines...).
7263 [Steve Henson]
7264
7265 *) Change #include filenames from <foo.h> to <openssl/foo.h>.
7266 [Bodo Moeller]
7267
7268 *) Change behaviour of ssl2_read when facing length-0 packets: Don't return
7269 0 (which usually indicates a closed connection), but continue reading.
7270 [Bodo Moeller]
7271
7272 *) Fix some race conditions.
7273 [Bodo Moeller]
7274
7275 *) Add support for CRL distribution points extension. Add Certificate
7276 Policies and CRL distribution points documentation.
7277 [Steve Henson]
7278
7279 *) Move the autogenerated header file parts to crypto/opensslconf.h.
7280 [Ulf Möller]
7281
7282 *) Fix new 56-bit DES export ciphersuites: they were using 7 bytes instead of
7283 8 of keying material. Merlin has also confirmed interop with this fix
7284 between OpenSSL and Baltimore C/SSL 2.0 and J/SSL 2.0.
7285 [Merlin Hughes <merlin@baltimore.ie>]
7286
7287 *) Fix lots of warnings.
7288 [Richard Levitte <levitte@stacken.kth.se>]
7289
7290 *) In add_cert_dir() in crypto/x509/by_dir.c, break out of the loop if
7291 the directory spec didn't end with a LIST_SEPARATOR_CHAR.
7292 [Richard Levitte <levitte@stacken.kth.se>]
7293
7294 *) Fix problems with sizeof(long) == 8.
7295 [Andy Polyakov <appro@fy.chalmers.se>]
7296
7297 *) Change functions to ANSI C.
7298 [Ulf Möller]
7299
7300 *) Fix typos in error codes.
7301 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>, Ulf Möller]
7302
7303 *) Remove defunct assembler files from Configure.
7304 [Ulf Möller]
7305
7306 *) SPARC v8 assembler BIGNUM implementation.
7307 [Andy Polyakov <appro@fy.chalmers.se>]
7308
7309 *) Support for Certificate Policies extension: both print and set.
7310 Various additions to support the r2i method this uses.
7311 [Steve Henson]
7312
7313 *) A lot of constification, and fix a bug in X509_NAME_oneline() that could
7314 return a const string when you are expecting an allocated buffer.
7315 [Ben Laurie]
7316
7317 *) Add support for ASN1 types UTF8String and VISIBLESTRING, also the CHOICE
7318 types DirectoryString and DisplayText.
7319 [Steve Henson]
7320
7321 *) Add code to allow r2i extensions to access the configuration database,
7322 add an LHASH database driver and add several ctx helper functions.
7323 [Steve Henson]
7324
7325 *) Fix an evil bug in bn_expand2() which caused various BN functions to
7326 fail when they extended the size of a BIGNUM.
7327 [Steve Henson]
7328
7329 *) Various utility functions to handle SXNet extension. Modify mkdef.pl to
7330 support typesafe stack.
7331 [Steve Henson]
7332
7333 *) Fix typo in SSL_[gs]et_options().
7334 [Nils Frostberg <nils@medcom.se>]
7335
7336 *) Delete various functions and files that belonged to the (now obsolete)
7337 old X509V3 handling code.
7338 [Steve Henson]
7339
7340 *) New Configure option "rsaref".
7341 [Ulf Möller]
7342
7343 *) Don't auto-generate pem.h.
7344 [Bodo Moeller]
7345
7346 *) Introduce type-safe ASN.1 SETs.
7347 [Ben Laurie]
7348
7349 *) Convert various additional casted stacks to type-safe STACK_OF() variants.
7350 [Ben Laurie, Ralf S. Engelschall, Steve Henson]
7351
7352 *) Introduce type-safe STACKs. This will almost certainly break lots of code
7353 that links with OpenSSL (well at least cause lots of warnings), but fear
7354 not: the conversion is trivial, and it eliminates loads of evil casts. A
7355 few STACKed things have been converted already. Feel free to convert more.
7356 In the fullness of time, I'll do away with the STACK type altogether.
7357 [Ben Laurie]
7358
7359 *) Add `openssl ca -revoke <certfile>' facility which revokes a certificate
7360 specified in <certfile> by updating the entry in the index.txt file.
7361 This way one no longer has to edit the index.txt file manually for
7362 revoking a certificate. The -revoke option does the gory details now.
7363 [Massimiliano Pala <madwolf@openca.org>, Ralf S. Engelschall]
7364
7365 *) Fix `openssl crl -noout -text' combination where `-noout' killed the
7366 `-text' option at all and this way the `-noout -text' combination was
7367 inconsistent in `openssl crl' with the friends in `openssl x509|rsa|dsa'.
7368 [Ralf S. Engelschall]
7369
7370 *) Make sure a corresponding plain text error message exists for the
7371 X509_V_ERR_CERT_REVOKED/23 error number which can occur when a
7372 verify callback function determined that a certificate was revoked.
7373 [Ralf S. Engelschall]
7374
7375 *) Bugfix: In test/testenc, don't test "openssl <cipher>" for
7376 ciphers that were excluded, e.g. by -DNO_IDEA. Also, test
7377 all available cipers including rc5, which was forgotten until now.
7378 In order to let the testing shell script know which algorithms
7379 are available, a new (up to now undocumented) command
7380 "openssl list-cipher-commands" is used.
7381 [Bodo Moeller]
7382
7383 *) Bugfix: s_client occasionally would sleep in select() when
7384 it should have checked SSL_pending() first.
7385 [Bodo Moeller]
7386
7387 *) New functions DSA_do_sign and DSA_do_verify to provide access to
7388 the raw DSA values prior to ASN.1 encoding.
7389 [Ulf Möller]
7390
7391 *) Tweaks to Configure
7392 [Niels Poppe <niels@netbox.org>]
7393
7394 *) Add support for PKCS#5 v2.0 ASN1 PBES2 structures. No other support,
7395 yet...
7396 [Steve Henson]
7397
7398 *) New variables $(RANLIB) and $(PERL) in the Makefiles.
7399 [Ulf Möller]
7400
7401 *) New config option to avoid instructions that are illegal on the 80386.
7402 The default code is faster, but requires at least a 486.
7403 [Ulf Möller]
7404
7405 *) Got rid of old SSL2_CLIENT_VERSION (inconsistently used) and
7406 SSL2_SERVER_VERSION (not used at all) macros, which are now the
7407 same as SSL2_VERSION anyway.
7408 [Bodo Moeller]
7409
7410 *) New "-showcerts" option for s_client.
7411 [Bodo Moeller]
7412
7413 *) Still more PKCS#12 integration. Add pkcs12 application to openssl
7414 application. Various cleanups and fixes.
7415 [Steve Henson]
7416
7417 *) More PKCS#12 integration. Add new pkcs12 directory with Makefile.ssl and
7418 modify error routines to work internally. Add error codes and PBE init
7419 to library startup routines.
7420 [Steve Henson]
7421
7422 *) Further PKCS#12 integration. Added password based encryption, PKCS#8 and
7423 packing functions to asn1 and evp. Changed function names and error
7424 codes along the way.
7425 [Steve Henson]
7426
7427 *) PKCS12 integration: and so it begins... First of several patches to
7428 slowly integrate PKCS#12 functionality into OpenSSL. Add PKCS#12
7429 objects to objects.h
7430 [Steve Henson]
7431
7432 *) Add a new 'indent' option to some X509V3 extension code. Initial ASN1
7433 and display support for Thawte strong extranet extension.
7434 [Steve Henson]
7435
7436 *) Add LinuxPPC support.
7437 [Jeff Dubrule <igor@pobox.org>]
7438
7439 *) Get rid of redundant BN file bn_mulw.c, and rename bn_div64 to
7440 bn_div_words in alpha.s.
7441 [Hannes Reinecke <H.Reinecke@hw.ac.uk> and Ben Laurie]
7442
7443 *) Make sure the RSA OAEP test is skipped under -DRSAref because
7444 OAEP isn't supported when OpenSSL is built with RSAref.
7445 [Ulf Moeller <ulf@fitug.de>]
7446
7447 *) Move definitions of IS_SET/IS_SEQUENCE inside crypto/asn1/asn1.h
7448 so they no longer are missing under -DNOPROTO.
7449 [Soren S. Jorvang <soren@t.dk>]
7450
7451
7452 Changes between 0.9.1c and 0.9.2b [22 Mar 1999]
7453
7454 *) Make SSL_get_peer_cert_chain() work in servers. Unfortunately, it still
7455 doesn't work when the session is reused. Coming soon!
7456 [Ben Laurie]
7457
7458 *) Fix a security hole, that allows sessions to be reused in the wrong
7459 context thus bypassing client cert protection! All software that uses
7460 client certs and session caches in multiple contexts NEEDS PATCHING to
7461 allow session reuse! A fuller solution is in the works.
7462 [Ben Laurie, problem pointed out by Holger Reif, Bodo Moeller (and ???)]
7463
7464 *) Some more source tree cleanups (removed obsolete files
7465 crypto/bf/asm/bf586.pl, test/test.txt and crypto/sha/asm/f.s; changed
7466 permission on "config" script to be executable) and a fix for the INSTALL
7467 document.
7468 [Ulf Moeller <ulf@fitug.de>]
7469
7470 *) Remove some legacy and erroneous uses of malloc, free instead of
7471 Malloc, Free.
7472 [Lennart Bang <lob@netstream.se>, with minor changes by Steve]
7473
7474 *) Make rsa_oaep_test return non-zero on error.
7475 [Ulf Moeller <ulf@fitug.de>]
7476
7477 *) Add support for native Solaris shared libraries. Configure
7478 solaris-sparc-sc4-pic, make, then run shlib/solaris-sc4.sh. It'd be nice
7479 if someone would make that last step automatic.
7480 [Matthias Loepfe <Matthias.Loepfe@AdNovum.CH>]
7481
7482 *) ctx_size was not built with the right compiler during "make links". Fixed.
7483 [Ben Laurie]
7484
7485 *) Change the meaning of 'ALL' in the cipher list. It now means "everything
7486 except NULL ciphers". This means the default cipher list will no longer
7487 enable NULL ciphers. They need to be specifically enabled e.g. with
7488 the string "DEFAULT:eNULL".
7489 [Steve Henson]
7490
7491 *) Fix to RSA private encryption routines: if p < q then it would
7492 occasionally produce an invalid result. This will only happen with
7493 externally generated keys because OpenSSL (and SSLeay) ensure p > q.
7494 [Steve Henson]
7495
7496 *) Be less restrictive and allow also `perl util/perlpath.pl
7497 /path/to/bin/perl' in addition to `perl util/perlpath.pl /path/to/bin',
7498 because this way one can also use an interpreter named `perl5' (which is
7499 usually the name of Perl 5.xxx on platforms where an Perl 4.x is still
7500 installed as `perl').
7501 [Matthias Loepfe <Matthias.Loepfe@adnovum.ch>]
7502
7503 *) Let util/clean-depend.pl work also with older Perl 5.00x versions.
7504 [Matthias Loepfe <Matthias.Loepfe@adnovum.ch>]
7505
7506 *) Fix Makefile.org so CC,CFLAG etc are passed to 'make links' add
7507 advapi32.lib to Win32 build and change the pem test comparision
7508 to fc.exe (thanks to Ulrich Kroener <kroneru@yahoo.com> for the
7509 suggestion). Fix misplaced ASNI prototypes and declarations in evp.h
7510 and crypto/des/ede_cbcm_enc.c.
7511 [Steve Henson]
7512
7513 *) DES quad checksum was broken on big-endian architectures. Fixed.
7514 [Ben Laurie]
7515
7516 *) Comment out two functions in bio.h that aren't implemented. Fix up the
7517 Win32 test batch file so it (might) work again. The Win32 test batch file
7518 is horrible: I feel ill....
7519 [Steve Henson]
7520
7521 *) Move various #ifdefs around so NO_SYSLOG, NO_DIRENT etc are now selected
7522 in e_os.h. Audit of header files to check ANSI and non ANSI
7523 sections: 10 functions were absent from non ANSI section and not exported
7524 from Windows DLLs. Fixed up libeay.num for new functions.
7525 [Steve Henson]
7526
7527 *) Make `openssl version' output lines consistent.
7528 [Ralf S. Engelschall]
7529
7530 *) Fix Win32 symbol export lists for BIO functions: Added
7531 BIO_get_ex_new_index, BIO_get_ex_num, BIO_get_ex_data and BIO_set_ex_data
7532 to ms/libeay{16,32}.def.
7533 [Ralf S. Engelschall]
7534
7535 *) Second round of fixing the OpenSSL perl/ stuff. It now at least compiled
7536 fine under Unix and passes some trivial tests I've now added. But the
7537 whole stuff is horribly incomplete, so a README.1ST with a disclaimer was
7538 added to make sure no one expects that this stuff really works in the
7539 OpenSSL 0.9.2 release. Additionally I've started to clean the XS sources
7540 up and fixed a few little bugs and inconsistencies in OpenSSL.{pm,xs} and
7541 openssl_bio.xs.
7542 [Ralf S. Engelschall]
7543
7544 *) Fix the generation of two part addresses in perl.
7545 [Kenji Miyake <kenji@miyake.org>, integrated by Ben Laurie]
7546
7547 *) Add config entry for Linux on MIPS.
7548 [John Tobey <jtobey@channel1.com>]
7549
7550 *) Make links whenever Configure is run, unless we are on Windoze.
7551 [Ben Laurie]
7552
7553 *) Permit extensions to be added to CRLs using crl_section in openssl.cnf.
7554 Currently only issuerAltName and AuthorityKeyIdentifier make any sense
7555 in CRLs.
7556 [Steve Henson]
7557
7558 *) Add a useful kludge to allow package maintainers to specify compiler and
7559 other platforms details on the command line without having to patch the
7560 Configure script everytime: One now can use ``perl Configure
7561 <id>:<details>'', i.e. platform ids are allowed to have details appended
7562 to them (seperated by colons). This is treated as there would be a static
7563 pre-configured entry in Configure's %table under key <id> with value
7564 <details> and ``perl Configure <id>'' is called. So, when you want to
7565 perform a quick test-compile under FreeBSD 3.1 with pgcc and without
7566 assembler stuff you can use ``perl Configure "FreeBSD-elf:pgcc:-O6:::"''
7567 now, which overrides the FreeBSD-elf entry on-the-fly.
7568 [Ralf S. Engelschall]
7569
7570 *) Disable new TLS1 ciphersuites by default: they aren't official yet.
7571 [Ben Laurie]
7572
7573 *) Allow DSO flags like -fpic, -fPIC, -KPIC etc. to be specified
7574 on the `perl Configure ...' command line. This way one can compile
7575 OpenSSL libraries with Position Independent Code (PIC) which is needed
7576 for linking it into DSOs.
7577 [Ralf S. Engelschall]
7578
7579 *) Remarkably, export ciphers were totally broken and no-one had noticed!
7580 Fixed.
7581 [Ben Laurie]
7582
7583 *) Cleaned up the LICENSE document: The official contact for any license
7584 questions now is the OpenSSL core team under openssl-core@openssl.org.
7585 And add a paragraph about the dual-license situation to make sure people
7586 recognize that _BOTH_ the OpenSSL license _AND_ the SSLeay license apply
7587 to the OpenSSL toolkit.
7588 [Ralf S. Engelschall]
7589
7590 *) General source tree makefile cleanups: Made `making xxx in yyy...'
7591 display consistent in the source tree and replaced `/bin/rm' by `rm'.
7592 Additonally cleaned up the `make links' target: Remove unnecessary
7593 semicolons, subsequent redundant removes, inline point.sh into mklink.sh
7594 to speed processing and no longer clutter the display with confusing
7595 stuff. Instead only the actually done links are displayed.
7596 [Ralf S. Engelschall]
7597
7598 *) Permit null encryption ciphersuites, used for authentication only. It used
7599 to be necessary to set the preprocessor define SSL_ALLOW_ENULL to do this.
7600 It is now necessary to set SSL_FORBID_ENULL to prevent the use of null
7601 encryption.
7602 [Ben Laurie]
7603
7604 *) Add a bunch of fixes to the PKCS#7 stuff. It used to sometimes reorder
7605 signed attributes when verifying signatures (this would break them),
7606 the detached data encoding was wrong and public keys obtained using
7607 X509_get_pubkey() weren't freed.
7608 [Steve Henson]
7609
7610 *) Add text documentation for the BUFFER functions. Also added a work around
7611 to a Win95 console bug. This was triggered by the password read stuff: the
7612 last character typed gets carried over to the next fread(). If you were
7613 generating a new cert request using 'req' for example then the last
7614 character of the passphrase would be CR which would then enter the first
7615 field as blank.
7616 [Steve Henson]
7617
7618 *) Added the new `Includes OpenSSL Cryptography Software' button as
7619 doc/openssl_button.{gif,html} which is similar in style to the old SSLeay
7620 button and can be used by applications based on OpenSSL to show the
7621 relationship to the OpenSSL project.
7622 [Ralf S. Engelschall]
7623
7624 *) Remove confusing variables in function signatures in files
7625 ssl/ssl_lib.c and ssl/ssl.h.
7626 [Lennart Bong <lob@kulthea.stacken.kth.se>]
7627
7628 *) Don't install bss_file.c under PREFIX/include/
7629 [Lennart Bong <lob@kulthea.stacken.kth.se>]
7630
7631 *) Get the Win32 compile working again. Modify mkdef.pl so it can handle
7632 functions that return function pointers and has support for NT specific
7633 stuff. Fix mk1mf.pl and VC-32.pl to support NT differences also. Various
7634 #ifdef WIN32 and WINNTs sprinkled about the place and some changes from
7635 unsigned to signed types: this was killing the Win32 compile.
7636 [Steve Henson]
7637
7638 *) Add new certificate file to stack functions,
7639 SSL_add_dir_cert_subjects_to_stack() and
7640 SSL_add_file_cert_subjects_to_stack(). These largely supplant
7641 SSL_load_client_CA_file(), and can be used to add multiple certs easily
7642 to a stack (usually this is then handed to SSL_CTX_set_client_CA_list()).
7643 This means that Apache-SSL and similar packages don't have to mess around
7644 to add as many CAs as they want to the preferred list.
7645 [Ben Laurie]
7646
7647 *) Experiment with doxygen documentation. Currently only partially applied to
7648 ssl/ssl_lib.c.
7649 See http://www.stack.nl/~dimitri/doxygen/index.html, and run doxygen with
7650 openssl.doxy as the configuration file.
7651 [Ben Laurie]
7652
7653 *) Get rid of remaining C++-style comments which strict C compilers hate.
7654 [Ralf S. Engelschall, pointed out by Carlos Amengual]
7655
7656 *) Changed BN_RECURSION in bn_mont.c to BN_RECURSION_MONT so it is not
7657 compiled in by default: it has problems with large keys.
7658 [Steve Henson]
7659
7660 *) Add a bunch of SSL_xxx() functions for configuring the temporary RSA and
7661 DH private keys and/or callback functions which directly correspond to
7662 their SSL_CTX_xxx() counterparts but work on a per-connection basis. This
7663 is needed for applications which have to configure certificates on a
7664 per-connection basis (e.g. Apache+mod_ssl) instead of a per-context basis
7665 (e.g. s_server).
7666 For the RSA certificate situation is makes no difference, but
7667 for the DSA certificate situation this fixes the "no shared cipher"
7668 problem where the OpenSSL cipher selection procedure failed because the
7669 temporary keys were not overtaken from the context and the API provided
7670 no way to reconfigure them.
7671 The new functions now let applications reconfigure the stuff and they
7672 are in detail: SSL_need_tmp_RSA, SSL_set_tmp_rsa, SSL_set_tmp_dh,
7673 SSL_set_tmp_rsa_callback and SSL_set_tmp_dh_callback. Additionally a new
7674 non-public-API function ssl_cert_instantiate() is used as a helper
7675 function and also to reduce code redundancy inside ssl_rsa.c.
7676 [Ralf S. Engelschall]
7677
7678 *) Move s_server -dcert and -dkey options out of the undocumented feature
7679 area because they are useful for the DSA situation and should be
7680 recognized by the users.
7681 [Ralf S. Engelschall]
7682
7683 *) Fix the cipher decision scheme for export ciphers: the export bits are
7684 *not* within SSL_MKEY_MASK or SSL_AUTH_MASK, they are within
7685 SSL_EXP_MASK. So, the original variable has to be used instead of the
7686 already masked variable.
7687 [Richard Levitte <levitte@stacken.kth.se>]
7688
7689 *) Fix 'port' variable from `int' to `unsigned int' in crypto/bio/b_sock.c
7690 [Richard Levitte <levitte@stacken.kth.se>]
7691
7692 *) Change type of another md_len variable in pk7_doit.c:PKCS7_dataFinal()
7693 from `int' to `unsigned int' because it's a length and initialized by
7694 EVP_DigestFinal() which expects an `unsigned int *'.
7695 [Richard Levitte <levitte@stacken.kth.se>]
7696
7697 *) Don't hard-code path to Perl interpreter on shebang line of Configure
7698 script. Instead use the usual Shell->Perl transition trick.
7699 [Ralf S. Engelschall]
7700
7701 *) Make `openssl x509 -noout -modulus' functional also for DSA certificates
7702 (in addition to RSA certificates) to match the behaviour of `openssl dsa
7703 -noout -modulus' as it's already the case for `openssl rsa -noout
7704 -modulus'. For RSA the -modulus is the real "modulus" while for DSA
7705 currently the public key is printed (a decision which was already done by
7706 `openssl dsa -modulus' in the past) which serves a similar purpose.
7707 Additionally the NO_RSA no longer completely removes the whole -modulus
7708 option; it now only avoids using the RSA stuff. Same applies to NO_DSA
7709 now, too.
7710 [Ralf S. Engelschall]
7711
7712 *) Add Arne Ansper's reliable BIO - this is an encrypted, block-digested
7713 BIO. See the source (crypto/evp/bio_ok.c) for more info.
7714 [Arne Ansper <arne@ats.cyber.ee>]
7715
7716 *) Dump the old yucky req code that tried (and failed) to allow raw OIDs
7717 to be added. Now both 'req' and 'ca' can use new objects defined in the
7718 config file.
7719 [Steve Henson]
7720
7721 *) Add cool BIO that does syslog (or event log on NT).
7722 [Arne Ansper <arne@ats.cyber.ee>, integrated by Ben Laurie]
7723
7724 *) Add support for new TLS ciphersuites, TLS_RSA_EXPORT56_WITH_RC4_56_MD5,
7725 TLS_RSA_EXPORT56_WITH_RC2_CBC_56_MD5 and
7726 TLS_RSA_EXPORT56_WITH_DES_CBC_SHA, as specified in "56-bit Export Cipher
7727 Suites For TLS", draft-ietf-tls-56-bit-ciphersuites-00.txt.
7728 [Ben Laurie]
7729
7730 *) Add preliminary config info for new extension code.
7731 [Steve Henson]
7732
7733 *) Make RSA_NO_PADDING really use no padding.
7734 [Ulf Moeller <ulf@fitug.de>]
7735
7736 *) Generate errors when private/public key check is done.
7737 [Ben Laurie]
7738
7739 *) Overhaul for 'crl' utility. New function X509_CRL_print. Partial support
7740 for some CRL extensions and new objects added.
7741 [Steve Henson]
7742
7743 *) Really fix the ASN1 IMPLICIT bug this time... Partial support for private
7744 key usage extension and fuller support for authority key id.
7745 [Steve Henson]
7746
7747 *) Add OAEP encryption for the OpenSSL crypto library. OAEP is the improved
7748 padding method for RSA, which is recommended for new applications in PKCS
7749 #1 v2.0 (RFC 2437, October 1998).
7750 OAEP (Optimal Asymmetric Encryption Padding) has better theoretical
7751 foundations than the ad-hoc padding used in PKCS #1 v1.5. It is secure
7752 against Bleichbacher's attack on RSA.
7753 [Ulf Moeller <ulf@fitug.de>, reformatted, corrected and integrated by
7754 Ben Laurie]
7755
7756 *) Updates to the new SSL compression code
7757 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)]
7758
7759 *) Fix so that the version number in the master secret, when passed
7760 via RSA, checks that if TLS was proposed, but we roll back to SSLv3
7761 (because the server will not accept higher), that the version number
7762 is 0x03,0x01, not 0x03,0x00
7763 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)]
7764
7765 *) Run extensive memory leak checks on SSL apps. Fixed *lots* of memory
7766 leaks in ssl/ relating to new X509_get_pubkey() behaviour. Also fixes
7767 in apps/ and an unrelated leak in crypto/dsa/dsa_vrf.c
7768 [Steve Henson]
7769
7770 *) Support for RAW extensions where an arbitrary extension can be
7771 created by including its DER encoding. See apps/openssl.cnf for
7772 an example.
7773 [Steve Henson]
7774
7775 *) Make sure latest Perl versions don't interpret some generated C array
7776 code as Perl array code in the crypto/err/err_genc.pl script.
7777 [Lars Weber <3weber@informatik.uni-hamburg.de>]
7778
7779 *) Modify ms/do_ms.bat to not generate assembly language makefiles since
7780 not many people have the assembler. Various Win32 compilation fixes and
7781 update to the INSTALL.W32 file with (hopefully) more accurate Win32
7782 build instructions.
7783 [Steve Henson]
7784
7785 *) Modify configure script 'Configure' to automatically create crypto/date.h
7786 file under Win32 and also build pem.h from pem.org. New script
7787 util/mkfiles.pl to create the MINFO file on environments that can't do a
7788 'make files': perl util/mkfiles.pl >MINFO should work.
7789 [Steve Henson]
7790
7791 *) Major rework of DES function declarations, in the pursuit of correctness
7792 and purity. As a result, many evil casts evaporated, and some weirdness,
7793 too. You may find this causes warnings in your code. Zapping your evil
7794 casts will probably fix them. Mostly.
7795 [Ben Laurie]
7796
7797 *) Fix for a typo in asn1.h. Bug fix to object creation script
7798 obj_dat.pl. It considered a zero in an object definition to mean
7799 "end of object": none of the objects in objects.h have any zeros
7800 so it wasn't spotted.
7801 [Steve Henson, reported by Erwann ABALEA <eabalea@certplus.com>]
7802
7803 *) Add support for Triple DES Cipher Block Chaining with Output Feedback
7804 Masking (CBCM). In the absence of test vectors, the best I have been able
7805 to do is check that the decrypt undoes the encrypt, so far. Send me test
7806 vectors if you have them.
7807 [Ben Laurie]
7808
7809 *) Correct calculation of key length for export ciphers (too much space was
7810 allocated for null ciphers). This has not been tested!
7811 [Ben Laurie]
7812
7813 *) Modifications to the mkdef.pl for Win32 DEF file creation. The usage
7814 message is now correct (it understands "crypto" and "ssl" on its
7815 command line). There is also now an "update" option. This will update
7816 the util/ssleay.num and util/libeay.num files with any new functions.
7817 If you do a:
7818 perl util/mkdef.pl crypto ssl update
7819 it will update them.
7820 [Steve Henson]
7821
7822 *) Overhauled the Perl interface (perl/*):
7823 - ported BN stuff to OpenSSL's different BN library
7824 - made the perl/ source tree CVS-aware
7825 - renamed the package from SSLeay to OpenSSL (the files still contain
7826 their history because I've copied them in the repository)
7827 - removed obsolete files (the test scripts will be replaced
7828 by better Test::Harness variants in the future)
7829 [Ralf S. Engelschall]
7830
7831 *) First cut for a very conservative source tree cleanup:
7832 1. merge various obsolete readme texts into doc/ssleay.txt
7833 where we collect the old documents and readme texts.
7834 2. remove the first part of files where I'm already sure that we no
7835 longer need them because of three reasons: either they are just temporary
7836 files which were left by Eric or they are preserved original files where
7837 I've verified that the diff is also available in the CVS via "cvs diff
7838 -rSSLeay_0_8_1b" or they were renamed (as it was definitely the case for
7839 the crypto/md/ stuff).
7840 [Ralf S. Engelschall]
7841
7842 *) More extension code. Incomplete support for subject and issuer alt
7843 name, issuer and authority key id. Change the i2v function parameters
7844 and add an extra 'crl' parameter in the X509V3_CTX structure: guess
7845 what that's for :-) Fix to ASN1 macro which messed up
7846 IMPLICIT tag and add f_enum.c which adds a2i, i2a for ENUMERATED.
7847 [Steve Henson]
7848
7849 *) Preliminary support for ENUMERATED type. This is largely copied from the
7850 INTEGER code.
7851 [Steve Henson]
7852
7853 *) Add new function, EVP_MD_CTX_copy() to replace frequent use of memcpy.
7854 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)]
7855
7856 *) Make sure `make rehash' target really finds the `openssl' program.
7857 [Ralf S. Engelschall, Matthias Loepfe <Matthias.Loepfe@adnovum.ch>]
7858
7859 *) Squeeze another 7% of speed out of MD5 assembler, at least on a P2. I'd
7860 like to hear about it if this slows down other processors.
7861 [Ben Laurie]
7862
7863 *) Add CygWin32 platform information to Configure script.
7864 [Alan Batie <batie@aahz.jf.intel.com>]
7865
7866 *) Fixed ms/32all.bat script: `no_asm' -> `no-asm'
7867 [Rainer W. Gerling <gerling@mpg-gv.mpg.de>]
7868
7869 *) New program nseq to manipulate netscape certificate sequences
7870 [Steve Henson]
7871
7872 *) Modify crl2pkcs7 so it supports multiple -certfile arguments. Fix a
7873 few typos.
7874 [Steve Henson]
7875
7876 *) Fixes to BN code. Previously the default was to define BN_RECURSION
7877 but the BN code had some problems that would cause failures when
7878 doing certificate verification and some other functions.
7879 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)]
7880
7881 *) Add ASN1 and PEM code to support netscape certificate sequences.
7882 [Steve Henson]
7883
7884 *) Add ASN1 and PEM code to support netscape certificate sequences.
7885 [Steve Henson]
7886
7887 *) Add several PKIX and private extended key usage OIDs.
7888 [Steve Henson]
7889
7890 *) Modify the 'ca' program to handle the new extension code. Modify
7891 openssl.cnf for new extension format, add comments.
7892 [Steve Henson]
7893
7894 *) More X509 V3 changes. Fix typo in v3_bitstr.c. Add support to 'req'
7895 and add a sample to openssl.cnf so req -x509 now adds appropriate
7896 CA extensions.
7897 [Steve Henson]
7898
7899 *) Continued X509 V3 changes. Add to other makefiles, integrate with the
7900 error code, add initial support to X509_print() and x509 application.
7901 [Steve Henson]
7902
7903 *) Takes a deep breath and start addding X509 V3 extension support code. Add
7904 files in crypto/x509v3. Move original stuff to crypto/x509v3/old. All this
7905 stuff is currently isolated and isn't even compiled yet.
7906 [Steve Henson]
7907
7908 *) Continuing patches for GeneralizedTime. Fix up certificate and CRL
7909 ASN1 to use ASN1_TIME and modify print routines to use ASN1_TIME_print.
7910 Removed the versions check from X509 routines when loading extensions:
7911 this allows certain broken certificates that don't set the version
7912 properly to be processed.
7913 [Steve Henson]
7914
7915 *) Deal with irritating shit to do with dependencies, in YAAHW (Yet Another
7916 Ad Hoc Way) - Makefile.ssls now all contain local dependencies, which
7917 can still be regenerated with "make depend".
7918 [Ben Laurie]
7919
7920 *) Spelling mistake in C version of CAST-128.
7921 [Ben Laurie, reported by Jeremy Hylton <jeremy@cnri.reston.va.us>]
7922
7923 *) Changes to the error generation code. The perl script err-code.pl
7924 now reads in the old error codes and retains the old numbers, only
7925 adding new ones if necessary. It also only changes the .err files if new
7926 codes are added. The makefiles have been modified to only insert errors
7927 when needed (to avoid needlessly modifying header files). This is done
7928 by only inserting errors if the .err file is newer than the auto generated
7929 C file. To rebuild all the error codes from scratch (the old behaviour)
7930 either modify crypto/Makefile.ssl to pass the -regen flag to err_code.pl
7931 or delete all the .err files.
7932 [Steve Henson]
7933
7934 *) CAST-128 was incorrectly implemented for short keys. The C version has
7935 been fixed, but is untested. The assembler versions are also fixed, but
7936 new assembler HAS NOT BEEN GENERATED FOR WIN32 - the Makefile needs fixing
7937 to regenerate it if needed.
7938 [Ben Laurie, reported (with fix for C version) by Jun-ichiro itojun
7939 Hagino <itojun@kame.net>]
7940
7941 *) File was opened incorrectly in randfile.c.
7942 [Ulf Möller <ulf@fitug.de>]
7943
7944 *) Beginning of support for GeneralizedTime. d2i, i2d, check and print
7945 functions. Also ASN1_TIME suite which is a CHOICE of UTCTime or
7946 GeneralizedTime. ASN1_TIME is the proper type used in certificates et
7947 al: it's just almost always a UTCTime. Note this patch adds new error
7948 codes so do a "make errors" if there are problems.
7949 [Steve Henson]
7950
7951 *) Correct Linux 1 recognition in config.
7952 [Ulf Möller <ulf@fitug.de>]
7953
7954 *) Remove pointless MD5 hash when using DSA keys in ca.
7955 [Anonymous <nobody@replay.com>]
7956
7957 *) Generate an error if given an empty string as a cert directory. Also
7958 generate an error if handed NULL (previously returned 0 to indicate an
7959 error, but didn't set one).
7960 [Ben Laurie, reported by Anonymous <nobody@replay.com>]
7961
7962 *) Add prototypes to SSL methods. Make SSL_write's buffer const, at last.
7963 [Ben Laurie]
7964
7965 *) Fix the dummy function BN_ref_mod_exp() in rsaref.c to have the correct
7966 parameters. This was causing a warning which killed off the Win32 compile.
7967 [Steve Henson]
7968
7969 *) Remove C++ style comments from crypto/bn/bn_local.h.
7970 [Neil Costigan <neil.costigan@celocom.com>]
7971
7972 *) The function OBJ_txt2nid was broken. It was supposed to return a nid
7973 based on a text string, looking up short and long names and finally
7974 "dot" format. The "dot" format stuff didn't work. Added new function
7975 OBJ_txt2obj to do the same but return an ASN1_OBJECT and rewrote
7976 OBJ_txt2nid to use it. OBJ_txt2obj can also return objects even if the
7977 OID is not part of the table.
7978 [Steve Henson]
7979
7980 *) Add prototypes to X509 lookup/verify methods, fixing a bug in
7981 X509_LOOKUP_by_alias().
7982 [Ben Laurie]
7983
7984 *) Sort openssl functions by name.
7985 [Ben Laurie]
7986
7987 *) Get the gendsa program working (hopefully) and add it to app list. Remove
7988 encryption from sample DSA keys (in case anyone is interested the password
7989 was "1234").
7990 [Steve Henson]
7991
7992 *) Make _all_ *_free functions accept a NULL pointer.
7993 [Frans Heymans <fheymans@isaserver.be>]
7994
7995 *) If a DH key is generated in s3_srvr.c, don't blow it by trying to use
7996 NULL pointers.
7997 [Anonymous <nobody@replay.com>]
7998
7999 *) s_server should send the CAfile as acceptable CAs, not its own cert.
8000 [Bodo Moeller <3moeller@informatik.uni-hamburg.de>]
8001
8002 *) Don't blow it for numeric -newkey arguments to apps/req.
8003 [Bodo Moeller <3moeller@informatik.uni-hamburg.de>]
8004
8005 *) Temp key "for export" tests were wrong in s3_srvr.c.
8006 [Anonymous <nobody@replay.com>]
8007
8008 *) Add prototype for temp key callback functions
8009 SSL_CTX_set_tmp_{rsa,dh}_callback().
8010 [Ben Laurie]
8011
8012 *) Make DH_free() tolerate being passed a NULL pointer (like RSA_free() and
8013 DSA_free()). Make X509_PUBKEY_set() check for errors in d2i_PublicKey().
8014 [Steve Henson]
8015
8016 *) X509_name_add_entry() freed the wrong thing after an error.
8017 [Arne Ansper <arne@ats.cyber.ee>]
8018
8019 *) rsa_eay.c would attempt to free a NULL context.
8020 [Arne Ansper <arne@ats.cyber.ee>]
8021
8022 *) BIO_s_socket() had a broken should_retry() on Windoze.
8023 [Arne Ansper <arne@ats.cyber.ee>]
8024
8025 *) BIO_f_buffer() didn't pass on BIO_CTRL_FLUSH.
8026 [Arne Ansper <arne@ats.cyber.ee>]
8027
8028 *) Make sure the already existing X509_STORE->depth variable is initialized
8029 in X509_STORE_new(), but document the fact that this variable is still
8030 unused in the certificate verification process.
8031 [Ralf S. Engelschall]
8032
8033 *) Fix the various library and apps files to free up pkeys obtained from
8034 X509_PUBKEY_get() et al. Also allow x509.c to handle netscape extensions.
8035 [Steve Henson]
8036
8037 *) Fix reference counting in X509_PUBKEY_get(). This makes
8038 demos/maurice/example2.c work, amongst others, probably.
8039 [Steve Henson and Ben Laurie]
8040
8041 *) First cut of a cleanup for apps/. First the `ssleay' program is now named
8042 `openssl' and second, the shortcut symlinks for the `openssl <command>'
8043 are no longer created. This way we have a single and consistent command
8044 line interface `openssl <command>', similar to `cvs <command>'.
8045 [Ralf S. Engelschall, Paul Sutton and Ben Laurie]
8046
8047 *) ca.c: move test for DSA keys inside #ifndef NO_DSA. Make pubkey
8048 BIT STRING wrapper always have zero unused bits.
8049 [Steve Henson]
8050
8051 *) Add CA.pl, perl version of CA.sh, add extended key usage OID.
8052 [Steve Henson]
8053
8054 *) Make the top-level INSTALL documentation easier to understand.
8055 [Paul Sutton]
8056
8057 *) Makefiles updated to exit if an error occurs in a sub-directory
8058 make (including if user presses ^C) [Paul Sutton]
8059
8060 *) Make Montgomery context stuff explicit in RSA data structure.
8061 [Ben Laurie]
8062
8063 *) Fix build order of pem and err to allow for generated pem.h.
8064 [Ben Laurie]
8065
8066 *) Fix renumbering bug in X509_NAME_delete_entry().
8067 [Ben Laurie]
8068
8069 *) Enhanced the err-ins.pl script so it makes the error library number
8070 global and can add a library name. This is needed for external ASN1 and
8071 other error libraries.
8072 [Steve Henson]
8073
8074 *) Fixed sk_insert which never worked properly.
8075 [Steve Henson]
8076
8077 *) Fix ASN1 macros so they can handle indefinite length construted
8078 EXPLICIT tags. Some non standard certificates use these: they can now
8079 be read in.
8080 [Steve Henson]
8081
8082 *) Merged the various old/obsolete SSLeay documentation files (doc/xxx.doc)
8083 into a single doc/ssleay.txt bundle. This way the information is still
8084 preserved but no longer messes up this directory. Now it's new room for
8085 the new set of documenation files.
8086 [Ralf S. Engelschall]
8087
8088 *) SETs were incorrectly DER encoded. This was a major pain, because they
8089 shared code with SEQUENCEs, which aren't coded the same. This means that
8090 almost everything to do with SETs or SEQUENCEs has either changed name or
8091 number of arguments.
8092 [Ben Laurie, based on a partial fix by GP Jayan <gp@nsj.co.jp>]
8093
8094 *) Fix test data to work with the above.
8095 [Ben Laurie]
8096
8097 *) Fix the RSA header declarations that hid a bug I fixed in 0.9.0b but
8098 was already fixed by Eric for 0.9.1 it seems.
8099 [Ben Laurie - pointed out by Ulf Möller <ulf@fitug.de>]
8100
8101 *) Autodetect FreeBSD3.
8102 [Ben Laurie]
8103
8104 *) Fix various bugs in Configure. This affects the following platforms:
8105 nextstep
8106 ncr-scde
8107 unixware-2.0
8108 unixware-2.0-pentium
8109 sco5-cc.
8110 [Ben Laurie]
8111
8112 *) Eliminate generated files from CVS. Reorder tests to regenerate files
8113 before they are needed.
8114 [Ben Laurie]
8115
8116 *) Generate Makefile.ssl from Makefile.org (to keep CVS happy).
8117 [Ben Laurie]
8118
8119
8120 Changes between 0.9.1b and 0.9.1c [23-Dec-1998]
8121
8122 *) Added OPENSSL_VERSION_NUMBER to crypto/crypto.h and
8123 changed SSLeay to OpenSSL in version strings.
8124 [Ralf S. Engelschall]
8125
8126 *) Some fixups to the top-level documents.
8127 [Paul Sutton]
8128
8129 *) Fixed the nasty bug where rsaref.h was not found under compile-time
8130 because the symlink to include/ was missing.
8131 [Ralf S. Engelschall]
8132
8133 *) Incorporated the popular no-RSA/DSA-only patches
8134 which allow to compile a RSA-free SSLeay.
8135 [Andrew Cooke / Interrader Ldt., Ralf S. Engelschall]
8136
8137 *) Fixed nasty rehash problem under `make -f Makefile.ssl links'
8138 when "ssleay" is still not found.
8139 [Ralf S. Engelschall]
8140
8141 *) Added more platforms to Configure: Cray T3E, HPUX 11,
8142 [Ralf S. Engelschall, Beckmann <beckman@acl.lanl.gov>]
8143
8144 *) Updated the README file.
8145 [Ralf S. Engelschall]
8146
8147 *) Added various .cvsignore files in the CVS repository subdirs
8148 to make a "cvs update" really silent.
8149 [Ralf S. Engelschall]
8150
8151 *) Recompiled the error-definition header files and added
8152 missing symbols to the Win32 linker tables.
8153 [Ralf S. Engelschall]
8154
8155 *) Cleaned up the top-level documents;
8156 o new files: CHANGES and LICENSE
8157 o merged VERSION, HISTORY* and README* files a CHANGES.SSLeay
8158 o merged COPYRIGHT into LICENSE
8159 o removed obsolete TODO file
8160 o renamed MICROSOFT to INSTALL.W32
8161 [Ralf S. Engelschall]
8162
8163 *) Removed dummy files from the 0.9.1b source tree:
8164 crypto/asn1/x crypto/bio/cd crypto/bio/fg crypto/bio/grep crypto/bio/vi
8165 crypto/bn/asm/......add.c crypto/bn/asm/a.out crypto/dsa/f crypto/md5/f
8166 crypto/pem/gmon.out crypto/perlasm/f crypto/pkcs7/build crypto/rsa/f
8167 crypto/sha/asm/f crypto/threads/f ms/zzz ssl/f ssl/f.mak test/f
8168 util/f.mak util/pl/f util/pl/f.mak crypto/bf/bf_locl.old apps/f
8169 [Ralf S. Engelschall]
8170
8171 *) Added various platform portability fixes.
8172 [Mark J. Cox]
8173
8174 *) The Genesis of the OpenSSL rpject:
8175 We start with the latest (unreleased) SSLeay version 0.9.1b which Eric A.
8176 Young and Tim J. Hudson created while they were working for C2Net until
8177 summer 1998.
8178 [The OpenSSL Project]
8179
8180
8181 Changes between 0.9.0b and 0.9.1b [not released]
8182
8183 *) Updated a few CA certificates under certs/
8184 [Eric A. Young]
8185
8186 *) Changed some BIGNUM api stuff.
8187 [Eric A. Young]
8188
8189 *) Various platform ports: OpenBSD, Ultrix, IRIX 64bit, NetBSD,
8190 DGUX x86, Linux Alpha, etc.
8191 [Eric A. Young]
8192
8193 *) New COMP library [crypto/comp/] for SSL Record Layer Compression:
8194 RLE (dummy implemented) and ZLIB (really implemented when ZLIB is
8195 available).
8196 [Eric A. Young]
8197
8198 *) Add -strparse option to asn1pars program which parses nested
8199 binary structures
8200 [Dr Stephen Henson <shenson@bigfoot.com>]
8201
8202 *) Added "oid_file" to ssleay.cnf for "ca" and "req" programs.
8203 [Eric A. Young]
8204
8205 *) DSA fix for "ca" program.
8206 [Eric A. Young]
8207
8208 *) Added "-genkey" option to "dsaparam" program.
8209 [Eric A. Young]
8210
8211 *) Added RIPE MD160 (rmd160) message digest.
8212 [Eric A. Young]
8213
8214 *) Added -a (all) option to "ssleay version" command.
8215 [Eric A. Young]
8216
8217 *) Added PLATFORM define which is the id given to Configure.
8218 [Eric A. Young]
8219
8220 *) Added MemCheck_XXXX functions to crypto/mem.c for memory checking.
8221 [Eric A. Young]
8222
8223 *) Extended the ASN.1 parser routines.
8224 [Eric A. Young]
8225
8226 *) Extended BIO routines to support REUSEADDR, seek, tell, etc.
8227 [Eric A. Young]
8228
8229 *) Added a BN_CTX to the BN library.
8230 [Eric A. Young]
8231
8232 *) Fixed the weak key values in DES library
8233 [Eric A. Young]
8234
8235 *) Changed API in EVP library for cipher aliases.
8236 [Eric A. Young]
8237
8238 *) Added support for RC2/64bit cipher.
8239 [Eric A. Young]
8240
8241 *) Converted the lhash library to the crypto/mem.c functions.
8242 [Eric A. Young]
8243
8244 *) Added more recognized ASN.1 object ids.
8245 [Eric A. Young]
8246
8247 *) Added more RSA padding checks for SSL/TLS.
8248 [Eric A. Young]
8249
8250 *) Added BIO proxy/filter functionality.
8251 [Eric A. Young]
8252
8253 *) Added extra_certs to SSL_CTX which can be used
8254 send extra CA certificates to the client in the CA cert chain sending
8255 process. It can be configured with SSL_CTX_add_extra_chain_cert().
8256 [Eric A. Young]
8257
8258 *) Now Fortezza is denied in the authentication phase because
8259 this is key exchange mechanism is not supported by SSLeay at all.
8260 [Eric A. Young]
8261
8262 *) Additional PKCS1 checks.
8263 [Eric A. Young]
8264
8265 *) Support the string "TLSv1" for all TLS v1 ciphers.
8266 [Eric A. Young]
8267
8268 *) Added function SSL_get_ex_data_X509_STORE_CTX_idx() which gives the
8269 ex_data index of the SSL context in the X509_STORE_CTX ex_data.
8270 [Eric A. Young]
8271
8272 *) Fixed a few memory leaks.
8273 [Eric A. Young]
8274
8275 *) Fixed various code and comment typos.
8276 [Eric A. Young]
8277
8278 *) A minor bug in ssl/s3_clnt.c where there would always be 4 0
8279 bytes sent in the client random.
8280 [Edward Bishop <ebishop@spyglass.com>]
8281