]> git.ipfire.org Git - thirdparty/openssl.git/blob - CHANGES
Check for potentially exploitable overflows in asn1_d2i_read_bio
[thirdparty/openssl.git] / CHANGES
1
2 OpenSSL CHANGES
3 _______________
4
5 Changes between 0.9.8u and 0.9.8v [xx XXX xxxx]
6
7 *) Check for potentially exploitable overflows in asn1_d2i_read_bio
8 BUF_mem_grow and BUF_mem_grow_clean. Refuse attempts to shrink buffer
9 in CRYPTO_realloc_clean.
10
11 Thanks to Tavis Ormandy, Google Security Team, for discovering this
12 issue and to Adam Langley <agl@chromium.org> for fixing it.
13 (CVE-2012-2110)
14 [Adam Langley (Google), Tavis Ormandy, Google Security Team]
15
16 Changes between 0.9.8t and 0.9.8u [12 Mar 2012]
17
18 *) Fix MMA (Bleichenbacher's attack on PKCS #1 v1.5 RSA padding) weakness
19 in CMS and PKCS7 code. When RSA decryption fails use a random key for
20 content decryption and always return the same error. Note: this attack
21 needs on average 2^20 messages so it only affects automated senders. The
22 old behaviour can be reenabled in the CMS code by setting the
23 CMS_DEBUG_DECRYPT flag: this is useful for debugging and testing where
24 an MMA defence is not necessary.
25 Thanks to Ivan Nestlerode <inestlerode@us.ibm.com> for discovering
26 this issue. (CVE-2012-0884)
27 [Steve Henson]
28
29 *) Fix CVE-2011-4619: make sure we really are receiving a
30 client hello before rejecting multiple SGC restarts. Thanks to
31 Ivan Nestlerode <inestlerode@us.ibm.com> for discovering this bug.
32 [Steve Henson]
33
34 Changes between 0.9.8s and 0.9.8t [18 Jan 2012]
35
36 *) Fix for DTLS DoS issue introduced by fix for CVE-2011-4109.
37 Thanks to Antonio Martin, Enterprise Secure Access Research and
38 Development, Cisco Systems, Inc. for discovering this bug and
39 preparing a fix. (CVE-2012-0050)
40 [Antonio Martin]
41
42 Changes between 0.9.8r and 0.9.8s [4 Jan 2012]
43
44 *) Nadhem Alfardan and Kenny Paterson have discovered an extension
45 of the Vaudenay padding oracle attack on CBC mode encryption
46 which enables an efficient plaintext recovery attack against
47 the OpenSSL implementation of DTLS. Their attack exploits timing
48 differences arising during decryption processing. A research
49 paper describing this attack can be found at:
50 http://www.isg.rhul.ac.uk/~kp/dtls.pdf
51 Thanks go to Nadhem Alfardan and Kenny Paterson of the Information
52 Security Group at Royal Holloway, University of London
53 (www.isg.rhul.ac.uk) for discovering this flaw and to Robin Seggelmann
54 <seggelmann@fh-muenster.de> and Michael Tuexen <tuexen@fh-muenster.de>
55 for preparing the fix. (CVE-2011-4108)
56 [Robin Seggelmann, Michael Tuexen]
57
58 *) Stop policy check failure freeing same buffer twice. (CVE-2011-4109)
59 [Ben Laurie, Kasper <ekasper@google.com>]
60
61 *) Clear bytes used for block padding of SSL 3.0 records.
62 (CVE-2011-4576)
63 [Adam Langley (Google)]
64
65 *) Only allow one SGC handshake restart for SSL/TLS. Thanks to George
66 Kadianakis <desnacked@gmail.com> for discovering this issue and
67 Adam Langley for preparing the fix. (CVE-2011-4619)
68 [Adam Langley (Google)]
69
70 *) Prevent malformed RFC3779 data triggering an assertion failure.
71 Thanks to Andrew Chi, BBN Technologies, for discovering the flaw
72 and Rob Austein <sra@hactrn.net> for fixing it. (CVE-2011-4577)
73 [Rob Austein <sra@hactrn.net>]
74
75 *) Fix ssl_ciph.c set-up race.
76 [Adam Langley (Google)]
77
78 *) Fix spurious failures in ecdsatest.c.
79 [Emilia Käsper (Google)]
80
81 *) Fix the BIO_f_buffer() implementation (which was mixing different
82 interpretations of the '..._len' fields).
83 [Adam Langley (Google)]
84
85 *) Fix handling of BN_BLINDING: now BN_BLINDING_invert_ex (rather than
86 BN_BLINDING_invert_ex) calls BN_BLINDING_update, ensuring that concurrent
87 threads won't reuse the same blinding coefficients.
88
89 This also avoids the need to obtain the CRYPTO_LOCK_RSA_BLINDING
90 lock to call BN_BLINDING_invert_ex, and avoids one use of
91 BN_BLINDING_update for each BN_BLINDING structure (previously,
92 the last update always remained unused).
93 [Emilia Käsper (Google)]
94
95 *) Fix SSL memory handling for (EC)DH ciphersuites, in particular
96 for multi-threaded use of ECDH.
97 [Adam Langley (Google)]
98
99 *) Fix x509_name_ex_d2i memory leak on bad inputs.
100 [Bodo Moeller]
101
102 *) Add protection against ECDSA timing attacks as mentioned in the paper
103 by Billy Bob Brumley and Nicola Tuveri, see:
104
105 http://eprint.iacr.org/2011/232.pdf
106
107 [Billy Bob Brumley and Nicola Tuveri]
108
109 Changes between 0.9.8q and 0.9.8r [8 Feb 2011]
110
111 *) Fix parsing of OCSP stapling ClientHello extension. CVE-2011-0014
112 [Neel Mehta, Adam Langley, Bodo Moeller (Google)]
113
114 *) Fix bug in string printing code: if *any* escaping is enabled we must
115 escape the escape character (backslash) or the resulting string is
116 ambiguous.
117 [Steve Henson]
118
119 Changes between 0.9.8p and 0.9.8q [2 Dec 2010]
120
121 *) Disable code workaround for ancient and obsolete Netscape browsers
122 and servers: an attacker can use it in a ciphersuite downgrade attack.
123 Thanks to Martin Rex for discovering this bug. CVE-2010-4180
124 [Steve Henson]
125
126 *) Fixed J-PAKE implementation error, originally discovered by
127 Sebastien Martini, further info and confirmation from Stefan
128 Arentz and Feng Hao. Note that this fix is a security fix. CVE-2010-4252
129 [Ben Laurie]
130
131 Changes between 0.9.8o and 0.9.8p [16 Nov 2010]
132
133 *) Fix extension code to avoid race conditions which can result in a buffer
134 overrun vulnerability: resumed sessions must not be modified as they can
135 be shared by multiple threads. CVE-2010-3864
136 [Steve Henson]
137
138 *) Fix for double free bug in ssl/s3_clnt.c CVE-2010-2939
139 [Steve Henson]
140
141 *) Don't reencode certificate when calculating signature: cache and use
142 the original encoding instead. This makes signature verification of
143 some broken encodings work correctly.
144 [Steve Henson]
145
146 *) ec2_GF2m_simple_mul bugfix: compute correct result if the output EC_POINT
147 is also one of the inputs.
148 [Emilia Käsper <emilia.kasper@esat.kuleuven.be> (Google)]
149
150 *) Don't repeatedly append PBE algorithms to table if they already exist.
151 Sort table on each new add. This effectively makes the table read only
152 after all algorithms are added and subsequent calls to PKCS12_pbe_add
153 etc are non-op.
154 [Steve Henson]
155
156 Changes between 0.9.8n and 0.9.8o [01 Jun 2010]
157
158 [NB: OpenSSL 0.9.8o and later 0.9.8 patch levels were released after
159 OpenSSL 1.0.0.]
160
161 *) Correct a typo in the CMS ASN1 module which can result in invalid memory
162 access or freeing data twice (CVE-2010-0742)
163 [Steve Henson, Ronald Moesbergen <intercommit@gmail.com>]
164
165 *) Add SHA2 algorithms to SSL_library_init(). SHA2 is becoming far more
166 common in certificates and some applications which only call
167 SSL_library_init and not OpenSSL_add_all_algorithms() will fail.
168 [Steve Henson]
169
170 *) VMS fixes:
171 Reduce copying into .apps and .test in makevms.com
172 Don't try to use blank CA certificate in CA.com
173 Allow use of C files from original directories in maketests.com
174 [Steven M. Schweda" <sms@antinode.info>]
175
176 Changes between 0.9.8m and 0.9.8n [24 Mar 2010]
177
178 *) When rejecting SSL/TLS records due to an incorrect version number, never
179 update s->server with a new major version number. As of
180 - OpenSSL 0.9.8m if 'short' is a 16-bit type,
181 - OpenSSL 0.9.8f if 'short' is longer than 16 bits,
182 the previous behavior could result in a read attempt at NULL when
183 receiving specific incorrect SSL/TLS records once record payload
184 protection is active. (CVE-2010-0740)
185 [Bodo Moeller, Adam Langley <agl@chromium.org>]
186
187 *) Fix for CVE-2010-0433 where some kerberos enabled versions of OpenSSL
188 could be crashed if the relevant tables were not present (e.g. chrooted).
189 [Tomas Hoger <thoger@redhat.com>]
190
191 Changes between 0.9.8l and 0.9.8m [25 Feb 2010]
192
193 *) Always check bn_wexpend() return values for failure. (CVE-2009-3245)
194 [Martin Olsson, Neel Mehta]
195
196 *) Fix X509_STORE locking: Every 'objs' access requires a lock (to
197 accommodate for stack sorting, always a write lock!).
198 [Bodo Moeller]
199
200 *) On some versions of WIN32 Heap32Next is very slow. This can cause
201 excessive delays in the RAND_poll(): over a minute. As a workaround
202 include a time check in the inner Heap32Next loop too.
203 [Steve Henson]
204
205 *) The code that handled flushing of data in SSL/TLS originally used the
206 BIO_CTRL_INFO ctrl to see if any data was pending first. This caused
207 the problem outlined in PR#1949. The fix suggested there however can
208 trigger problems with buggy BIO_CTRL_WPENDING (e.g. some versions
209 of Apache). So instead simplify the code to flush unconditionally.
210 This should be fine since flushing with no data to flush is a no op.
211 [Steve Henson]
212
213 *) Handle TLS versions 2.0 and later properly and correctly use the
214 highest version of TLS/SSL supported. Although TLS >= 2.0 is some way
215 off ancient servers have a habit of sticking around for a while...
216 [Steve Henson]
217
218 *) Modify compression code so it frees up structures without using the
219 ex_data callbacks. This works around a problem where some applications
220 call CRYPTO_cleanup_all_ex_data() before application exit (e.g. when
221 restarting) then use compression (e.g. SSL with compression) later.
222 This results in significant per-connection memory leaks and
223 has caused some security issues including CVE-2008-1678 and
224 CVE-2009-4355.
225 [Steve Henson]
226
227 *) Constify crypto/cast (i.e., <openssl/cast.h>): a CAST_KEY doesn't
228 change when encrypting or decrypting.
229 [Bodo Moeller]
230
231 *) Add option SSL_OP_LEGACY_SERVER_CONNECT which will allow clients to
232 connect and renegotiate with servers which do not support RI.
233 Until RI is more widely deployed this option is enabled by default.
234 [Steve Henson]
235
236 *) Add "missing" ssl ctrls to clear options and mode.
237 [Steve Henson]
238
239 *) If client attempts to renegotiate and doesn't support RI respond with
240 a no_renegotiation alert as required by RFC5746. Some renegotiating
241 TLS clients will continue a connection gracefully when they receive
242 the alert. Unfortunately OpenSSL mishandled this alert and would hang
243 waiting for a server hello which it will never receive. Now we treat a
244 received no_renegotiation alert as a fatal error. This is because
245 applications requesting a renegotiation might well expect it to succeed
246 and would have no code in place to handle the server denying it so the
247 only safe thing to do is to terminate the connection.
248 [Steve Henson]
249
250 *) Add ctrl macro SSL_get_secure_renegotiation_support() which returns 1 if
251 peer supports secure renegotiation and 0 otherwise. Print out peer
252 renegotiation support in s_client/s_server.
253 [Steve Henson]
254
255 *) Replace the highly broken and deprecated SPKAC certification method with
256 the updated NID creation version. This should correctly handle UTF8.
257 [Steve Henson]
258
259 *) Implement RFC5746. Re-enable renegotiation but require the extension
260 as needed. Unfortunately, SSL3_FLAGS_ALLOW_UNSAFE_LEGACY_RENEGOTIATION
261 turns out to be a bad idea. It has been replaced by
262 SSL_OP_ALLOW_UNSAFE_LEGACY_RENEGOTIATION which can be set with
263 SSL_CTX_set_options(). This is really not recommended unless you
264 know what you are doing.
265 [Eric Rescorla <ekr@networkresonance.com>, Ben Laurie, Steve Henson]
266
267 *) Fixes to stateless session resumption handling. Use initial_ctx when
268 issuing and attempting to decrypt tickets in case it has changed during
269 servername handling. Use a non-zero length session ID when attempting
270 stateless session resumption: this makes it possible to determine if
271 a resumption has occurred immediately after receiving server hello
272 (several places in OpenSSL subtly assume this) instead of later in
273 the handshake.
274 [Steve Henson]
275
276 *) The functions ENGINE_ctrl(), OPENSSL_isservice(),
277 CMS_get1_RecipientRequest() and RAND_bytes() can return <=0 on error
278 fixes for a few places where the return code is not checked
279 correctly.
280 [Julia Lawall <julia@diku.dk>]
281
282 *) Add --strict-warnings option to Configure script to include devteam
283 warnings in other configurations.
284 [Steve Henson]
285
286 *) Add support for --libdir option and LIBDIR variable in makefiles. This
287 makes it possible to install openssl libraries in locations which
288 have names other than "lib", for example "/usr/lib64" which some
289 systems need.
290 [Steve Henson, based on patch from Jeremy Utley]
291
292 *) Don't allow the use of leading 0x80 in OIDs. This is a violation of
293 X690 8.9.12 and can produce some misleading textual output of OIDs.
294 [Steve Henson, reported by Dan Kaminsky]
295
296 *) Delete MD2 from algorithm tables. This follows the recommendation in
297 several standards that it is not used in new applications due to
298 several cryptographic weaknesses. For binary compatibility reasons
299 the MD2 API is still compiled in by default.
300 [Steve Henson]
301
302 *) Add compression id to {d2i,i2d}_SSL_SESSION so it is correctly saved
303 and restored.
304 [Steve Henson]
305
306 *) Rename uni2asc and asc2uni functions to OPENSSL_uni2asc and
307 OPENSSL_asc2uni conditionally on Netware platforms to avoid a name
308 clash.
309 [Guenter <lists@gknw.net>]
310
311 *) Fix the server certificate chain building code to use X509_verify_cert(),
312 it used to have an ad-hoc builder which was unable to cope with anything
313 other than a simple chain.
314 [David Woodhouse <dwmw2@infradead.org>, Steve Henson]
315
316 *) Don't check self signed certificate signatures in X509_verify_cert()
317 by default (a flag can override this): it just wastes time without
318 adding any security. As a useful side effect self signed root CAs
319 with non-FIPS digests are now usable in FIPS mode.
320 [Steve Henson]
321
322 *) In dtls1_process_out_of_seq_message() the check if the current message
323 is already buffered was missing. For every new message was memory
324 allocated, allowing an attacker to perform an denial of service attack
325 with sending out of seq handshake messages until there is no memory
326 left. Additionally every future messege was buffered, even if the
327 sequence number made no sense and would be part of another handshake.
328 So only messages with sequence numbers less than 10 in advance will be
329 buffered. (CVE-2009-1378)
330 [Robin Seggelmann, discovered by Daniel Mentz]
331
332 *) Records are buffered if they arrive with a future epoch to be
333 processed after finishing the corresponding handshake. There is
334 currently no limitation to this buffer allowing an attacker to perform
335 a DOS attack with sending records with future epochs until there is no
336 memory left. This patch adds the pqueue_size() function to detemine
337 the size of a buffer and limits the record buffer to 100 entries.
338 (CVE-2009-1377)
339 [Robin Seggelmann, discovered by Daniel Mentz]
340
341 *) Keep a copy of frag->msg_header.frag_len so it can be used after the
342 parent structure is freed. (CVE-2009-1379)
343 [Daniel Mentz]
344
345 *) Handle non-blocking I/O properly in SSL_shutdown() call.
346 [Darryl Miles <darryl-mailinglists@netbauds.net>]
347
348 *) Add 2.5.4.* OIDs
349 [Ilya O. <vrghost@gmail.com>]
350
351 Changes between 0.9.8k and 0.9.8l [5 Nov 2009]
352
353 *) Disable renegotiation completely - this fixes a severe security
354 problem (CVE-2009-3555) at the cost of breaking all
355 renegotiation. Renegotiation can be re-enabled by setting
356 SSL3_FLAGS_ALLOW_UNSAFE_LEGACY_RENEGOTIATION in s3->flags at
357 run-time. This is really not recommended unless you know what
358 you're doing.
359 [Ben Laurie]
360
361 Changes between 0.9.8j and 0.9.8k [25 Mar 2009]
362
363 *) Don't set val to NULL when freeing up structures, it is freed up by
364 underlying code. If sizeof(void *) > sizeof(long) this can result in
365 zeroing past the valid field. (CVE-2009-0789)
366 [Paolo Ganci <Paolo.Ganci@AdNovum.CH>]
367
368 *) Fix bug where return value of CMS_SignerInfo_verify_content() was not
369 checked correctly. This would allow some invalid signed attributes to
370 appear to verify correctly. (CVE-2009-0591)
371 [Ivan Nestlerode <inestlerode@us.ibm.com>]
372
373 *) Reject UniversalString and BMPString types with invalid lengths. This
374 prevents a crash in ASN1_STRING_print_ex() which assumes the strings have
375 a legal length. (CVE-2009-0590)
376 [Steve Henson]
377
378 *) Set S/MIME signing as the default purpose rather than setting it
379 unconditionally. This allows applications to override it at the store
380 level.
381 [Steve Henson]
382
383 *) Permit restricted recursion of ASN1 strings. This is needed in practice
384 to handle some structures.
385 [Steve Henson]
386
387 *) Improve efficiency of mem_gets: don't search whole buffer each time
388 for a '\n'
389 [Jeremy Shapiro <jnshapir@us.ibm.com>]
390
391 *) New -hex option for openssl rand.
392 [Matthieu Herrb]
393
394 *) Print out UTF8String and NumericString when parsing ASN1.
395 [Steve Henson]
396
397 *) Support NumericString type for name components.
398 [Steve Henson]
399
400 *) Allow CC in the environment to override the automatically chosen
401 compiler. Note that nothing is done to ensure flags work with the
402 chosen compiler.
403 [Ben Laurie]
404
405 Changes between 0.9.8i and 0.9.8j [07 Jan 2009]
406
407 *) Properly check EVP_VerifyFinal() and similar return values
408 (CVE-2008-5077).
409 [Ben Laurie, Bodo Moeller, Google Security Team]
410
411 *) Enable TLS extensions by default.
412 [Ben Laurie]
413
414 *) Allow the CHIL engine to be loaded, whether the application is
415 multithreaded or not. (This does not release the developer from the
416 obligation to set up the dynamic locking callbacks.)
417 [Sander Temme <sander@temme.net>]
418
419 *) Use correct exit code if there is an error in dgst command.
420 [Steve Henson; problem pointed out by Roland Dirlewanger]
421
422 *) Tweak Configure so that you need to say "experimental-jpake" to enable
423 JPAKE, and need to use -DOPENSSL_EXPERIMENTAL_JPAKE in applications.
424 [Bodo Moeller]
425
426 *) Add experimental JPAKE support, including demo authentication in
427 s_client and s_server.
428 [Ben Laurie]
429
430 *) Set the comparison function in v3_addr_canonize().
431 [Rob Austein <sra@hactrn.net>]
432
433 *) Add support for XMPP STARTTLS in s_client.
434 [Philip Paeps <philip@freebsd.org>]
435
436 *) Change the server-side SSL_OP_NETSCAPE_REUSE_CIPHER_CHANGE_BUG behavior
437 to ensure that even with this option, only ciphersuites in the
438 server's preference list will be accepted. (Note that the option
439 applies only when resuming a session, so the earlier behavior was
440 just about the algorithm choice for symmetric cryptography.)
441 [Bodo Moeller]
442
443 Changes between 0.9.8h and 0.9.8i [15 Sep 2008]
444
445 *) Fix NULL pointer dereference if a DTLS server received
446 ChangeCipherSpec as first record (CVE-2009-1386).
447 [PR #1679]
448
449 *) Fix a state transitition in s3_srvr.c and d1_srvr.c
450 (was using SSL3_ST_CW_CLNT_HELLO_B, should be ..._ST_SW_SRVR_...).
451 [Nagendra Modadugu]
452
453 *) The fix in 0.9.8c that supposedly got rid of unsafe
454 double-checked locking was incomplete for RSA blinding,
455 addressing just one layer of what turns out to have been
456 doubly unsafe triple-checked locking.
457
458 So now fix this for real by retiring the MONT_HELPER macro
459 in crypto/rsa/rsa_eay.c.
460
461 [Bodo Moeller; problem pointed out by Marius Schilder]
462
463 *) Various precautionary measures:
464
465 - Avoid size_t integer overflow in HASH_UPDATE (md32_common.h).
466
467 - Avoid a buffer overflow in d2i_SSL_SESSION() (ssl_asn1.c).
468 (NB: This would require knowledge of the secret session ticket key
469 to exploit, in which case you'd be SOL either way.)
470
471 - Change bn_nist.c so that it will properly handle input BIGNUMs
472 outside the expected range.
473
474 - Enforce the 'num' check in BN_div() (bn_div.c) for non-BN_DEBUG
475 builds.
476
477 [Neel Mehta, Bodo Moeller]
478
479 *) Allow engines to be "soft loaded" - i.e. optionally don't die if
480 the load fails. Useful for distros.
481 [Ben Laurie and the FreeBSD team]
482
483 *) Add support for Local Machine Keyset attribute in PKCS#12 files.
484 [Steve Henson]
485
486 *) Fix BN_GF2m_mod_arr() top-bit cleanup code.
487 [Huang Ying]
488
489 *) Expand ENGINE to support engine supplied SSL client certificate functions.
490
491 This work was sponsored by Logica.
492 [Steve Henson]
493
494 *) Add CryptoAPI ENGINE to support use of RSA and DSA keys held in Windows
495 keystores. Support for SSL/TLS client authentication too.
496 Not compiled unless enable-capieng specified to Configure.
497
498 This work was sponsored by Logica.
499 [Steve Henson]
500
501 *) Fix bug in X509_ATTRIBUTE creation: dont set attribute using
502 ASN1_TYPE_set1 if MBSTRING flag set. This bug would crash certain
503 attribute creation routines such as certifcate requests and PKCS#12
504 files.
505 [Steve Henson]
506
507 Changes between 0.9.8g and 0.9.8h [28 May 2008]
508
509 *) Fix flaw if 'Server Key exchange message' is omitted from a TLS
510 handshake which could lead to a cilent crash as found using the
511 Codenomicon TLS test suite (CVE-2008-1672)
512 [Steve Henson, Mark Cox]
513
514 *) Fix double free in TLS server name extensions which could lead to
515 a remote crash found by Codenomicon TLS test suite (CVE-2008-0891)
516 [Joe Orton]
517
518 *) Clear error queue in SSL_CTX_use_certificate_chain_file()
519
520 Clear the error queue to ensure that error entries left from
521 older function calls do not interfere with the correct operation.
522 [Lutz Jaenicke, Erik de Castro Lopo]
523
524 *) Remove root CA certificates of commercial CAs:
525
526 The OpenSSL project does not recommend any specific CA and does not
527 have any policy with respect to including or excluding any CA.
528 Therefore it does not make any sense to ship an arbitrary selection
529 of root CA certificates with the OpenSSL software.
530 [Lutz Jaenicke]
531
532 *) RSA OAEP patches to fix two separate invalid memory reads.
533 The first one involves inputs when 'lzero' is greater than
534 'SHA_DIGEST_LENGTH' (it would read about SHA_DIGEST_LENGTH bytes
535 before the beginning of from). The second one involves inputs where
536 the 'db' section contains nothing but zeroes (there is a one-byte
537 invalid read after the end of 'db').
538 [Ivan Nestlerode <inestlerode@us.ibm.com>]
539
540 *) Partial backport from 0.9.9-dev:
541
542 Introduce bn_mul_mont (dedicated Montgomery multiplication
543 procedure) as a candidate for BIGNUM assembler implementation.
544 While 0.9.9-dev uses assembler for various architectures, only
545 x86_64 is available by default here in the 0.9.8 branch, and
546 32-bit x86 is available through a compile-time setting.
547
548 To try the 32-bit x86 assembler implementation, use Configure
549 option "enable-montasm" (which exists only for this backport).
550
551 As "enable-montasm" for 32-bit x86 disclaims code stability
552 anyway, in this constellation we activate additional code
553 backported from 0.9.9-dev for further performance improvements,
554 namely BN_from_montgomery_word. (To enable this otherwise,
555 e.g. x86_64, try "-DMONT_FROM_WORD___NON_DEFAULT_0_9_8_BUILD".)
556
557 [Andy Polyakov (backport partially by Bodo Moeller)]
558
559 *) Add TLS session ticket callback. This allows an application to set
560 TLS ticket cipher and HMAC keys rather than relying on hardcoded fixed
561 values. This is useful for key rollover for example where several key
562 sets may exist with different names.
563 [Steve Henson]
564
565 *) Reverse ENGINE-internal logic for caching default ENGINE handles.
566 This was broken until now in 0.9.8 releases, such that the only way
567 a registered ENGINE could be used (assuming it initialises
568 successfully on the host) was to explicitly set it as the default
569 for the relevant algorithms. This is in contradiction with 0.9.7
570 behaviour and the documentation. With this fix, when an ENGINE is
571 registered into a given algorithm's table of implementations, the
572 'uptodate' flag is reset so that auto-discovery will be used next
573 time a new context for that algorithm attempts to select an
574 implementation.
575 [Ian Lister (tweaked by Geoff Thorpe)]
576
577 *) Backport of CMS code to OpenSSL 0.9.8. This differs from the 0.9.9
578 implemention in the following ways:
579
580 Lack of EVP_PKEY_ASN1_METHOD means algorithm parameters have to be
581 hard coded.
582
583 Lack of BER streaming support means one pass streaming processing is
584 only supported if data is detached: setting the streaming flag is
585 ignored for embedded content.
586
587 CMS support is disabled by default and must be explicitly enabled
588 with the enable-cms configuration option.
589 [Steve Henson]
590
591 *) Update the GMP engine glue to do direct copies between BIGNUM and
592 mpz_t when openssl and GMP use the same limb size. Otherwise the
593 existing "conversion via a text string export" trick is still used.
594 [Paul Sheer <paulsheer@gmail.com>]
595
596 *) Zlib compression BIO. This is a filter BIO which compressed and
597 uncompresses any data passed through it.
598 [Steve Henson]
599
600 *) Add AES_wrap_key() and AES_unwrap_key() functions to implement
601 RFC3394 compatible AES key wrapping.
602 [Steve Henson]
603
604 *) Add utility functions to handle ASN1 structures. ASN1_STRING_set0():
605 sets string data without copying. X509_ALGOR_set0() and
606 X509_ALGOR_get0(): set and retrieve X509_ALGOR (AlgorithmIdentifier)
607 data. Attribute function X509at_get0_data_by_OBJ(): retrieves data
608 from an X509_ATTRIBUTE structure optionally checking it occurs only
609 once. ASN1_TYPE_set1(): set and ASN1_TYPE structure copying supplied
610 data.
611 [Steve Henson]
612
613 *) Fix BN flag handling in RSA_eay_mod_exp() and BN_MONT_CTX_set()
614 to get the expected BN_FLG_CONSTTIME behavior.
615 [Bodo Moeller (Google)]
616
617 *) Netware support:
618
619 - fixed wrong usage of ioctlsocket() when build for LIBC BSD sockets
620 - fixed do_tests.pl to run the test suite with CLIB builds too (CLIB_OPT)
621 - added some more tests to do_tests.pl
622 - fixed RunningProcess usage so that it works with newer LIBC NDKs too
623 - removed usage of BN_LLONG for CLIB builds to avoid runtime dependency
624 - added new Configure targets netware-clib-bsdsock, netware-clib-gcc,
625 netware-clib-bsdsock-gcc, netware-libc-bsdsock-gcc
626 - various changes to netware.pl to enable gcc-cross builds on Win32
627 platform
628 - changed crypto/bio/b_sock.c to work with macro functions (CLIB BSD)
629 - various changes to fix missing prototype warnings
630 - fixed x86nasm.pl to create correct asm files for NASM COFF output
631 - added AES, WHIRLPOOL and CPUID assembler code to build files
632 - added missing AES assembler make rules to mk1mf.pl
633 - fixed order of includes in apps/ocsp.c so that e_os.h settings apply
634 [Guenter Knauf <eflash@gmx.net>]
635
636 *) Implement certificate status request TLS extension defined in RFC3546.
637 A client can set the appropriate parameters and receive the encoded
638 OCSP response via a callback. A server can query the supplied parameters
639 and set the encoded OCSP response in the callback. Add simplified examples
640 to s_client and s_server.
641 [Steve Henson]
642
643 Changes between 0.9.8f and 0.9.8g [19 Oct 2007]
644
645 *) Fix various bugs:
646 + Binary incompatibility of ssl_ctx_st structure
647 + DTLS interoperation with non-compliant servers
648 + Don't call get_session_cb() without proposed session
649 + Fix ia64 assembler code
650 [Andy Polyakov, Steve Henson]
651
652 Changes between 0.9.8e and 0.9.8f [11 Oct 2007]
653
654 *) DTLS Handshake overhaul. There were longstanding issues with
655 OpenSSL DTLS implementation, which were making it impossible for
656 RFC 4347 compliant client to communicate with OpenSSL server.
657 Unfortunately just fixing these incompatibilities would "cut off"
658 pre-0.9.8f clients. To allow for hassle free upgrade post-0.9.8e
659 server keeps tolerating non RFC compliant syntax. The opposite is
660 not true, 0.9.8f client can not communicate with earlier server.
661 This update even addresses CVE-2007-4995.
662 [Andy Polyakov]
663
664 *) Changes to avoid need for function casts in OpenSSL: some compilers
665 (gcc 4.2 and later) reject their use.
666 [Kurt Roeckx <kurt@roeckx.be>, Peter Hartley <pdh@utter.chaos.org.uk>,
667 Steve Henson]
668
669 *) Add RFC4507 support to OpenSSL. This includes the corrections in
670 RFC4507bis. The encrypted ticket format is an encrypted encoded
671 SSL_SESSION structure, that way new session features are automatically
672 supported.
673
674 If a client application caches session in an SSL_SESSION structure
675 support is transparent because tickets are now stored in the encoded
676 SSL_SESSION.
677
678 The SSL_CTX structure automatically generates keys for ticket
679 protection in servers so again support should be possible
680 with no application modification.
681
682 If a client or server wishes to disable RFC4507 support then the option
683 SSL_OP_NO_TICKET can be set.
684
685 Add a TLS extension debugging callback to allow the contents of any client
686 or server extensions to be examined.
687
688 This work was sponsored by Google.
689 [Steve Henson]
690
691 *) Add initial support for TLS extensions, specifically for the server_name
692 extension so far. The SSL_SESSION, SSL_CTX, and SSL data structures now
693 have new members for a host name. The SSL data structure has an
694 additional member SSL_CTX *initial_ctx so that new sessions can be
695 stored in that context to allow for session resumption, even after the
696 SSL has been switched to a new SSL_CTX in reaction to a client's
697 server_name extension.
698
699 New functions (subject to change):
700
701 SSL_get_servername()
702 SSL_get_servername_type()
703 SSL_set_SSL_CTX()
704
705 New CTRL codes and macros (subject to change):
706
707 SSL_CTRL_SET_TLSEXT_SERVERNAME_CB
708 - SSL_CTX_set_tlsext_servername_callback()
709 SSL_CTRL_SET_TLSEXT_SERVERNAME_ARG
710 - SSL_CTX_set_tlsext_servername_arg()
711 SSL_CTRL_SET_TLSEXT_HOSTNAME - SSL_set_tlsext_host_name()
712
713 openssl s_client has a new '-servername ...' option.
714
715 openssl s_server has new options '-servername_host ...', '-cert2 ...',
716 '-key2 ...', '-servername_fatal' (subject to change). This allows
717 testing the HostName extension for a specific single host name ('-cert'
718 and '-key' remain fallbacks for handshakes without HostName
719 negotiation). If the unrecogninzed_name alert has to be sent, this by
720 default is a warning; it becomes fatal with the '-servername_fatal'
721 option.
722
723 [Peter Sylvester, Remy Allais, Christophe Renou, Steve Henson]
724
725 *) Add AES and SSE2 assembly language support to VC++ build.
726 [Steve Henson]
727
728 *) Mitigate attack on final subtraction in Montgomery reduction.
729 [Andy Polyakov]
730
731 *) Fix crypto/ec/ec_mult.c to work properly with scalars of value 0
732 (which previously caused an internal error).
733 [Bodo Moeller]
734
735 *) Squeeze another 10% out of IGE mode when in != out.
736 [Ben Laurie]
737
738 *) AES IGE mode speedup.
739 [Dean Gaudet (Google)]
740
741 *) Add the Korean symmetric 128-bit cipher SEED (see
742 http://www.kisa.or.kr/kisa/seed/jsp/seed_eng.jsp) and
743 add SEED ciphersuites from RFC 4162:
744
745 TLS_RSA_WITH_SEED_CBC_SHA = "SEED-SHA"
746 TLS_DHE_DSS_WITH_SEED_CBC_SHA = "DHE-DSS-SEED-SHA"
747 TLS_DHE_RSA_WITH_SEED_CBC_SHA = "DHE-RSA-SEED-SHA"
748 TLS_DH_anon_WITH_SEED_CBC_SHA = "ADH-SEED-SHA"
749
750 To minimize changes between patchlevels in the OpenSSL 0.9.8
751 series, SEED remains excluded from compilation unless OpenSSL
752 is configured with 'enable-seed'.
753 [KISA, Bodo Moeller]
754
755 *) Mitigate branch prediction attacks, which can be practical if a
756 single processor is shared, allowing a spy process to extract
757 information. For detailed background information, see
758 http://eprint.iacr.org/2007/039 (O. Aciicmez, S. Gueron,
759 J.-P. Seifert, "New Branch Prediction Vulnerabilities in OpenSSL
760 and Necessary Software Countermeasures"). The core of the change
761 are new versions BN_div_no_branch() and
762 BN_mod_inverse_no_branch() of BN_div() and BN_mod_inverse(),
763 respectively, which are slower, but avoid the security-relevant
764 conditional branches. These are automatically called by BN_div()
765 and BN_mod_inverse() if the flag BN_FLG_CONSTTIME is set for one
766 of the input BIGNUMs. Also, BN_is_bit_set() has been changed to
767 remove a conditional branch.
768
769 BN_FLG_CONSTTIME is the new name for the previous
770 BN_FLG_EXP_CONSTTIME flag, since it now affects more than just
771 modular exponentiation. (Since OpenSSL 0.9.7h, setting this flag
772 in the exponent causes BN_mod_exp_mont() to use the alternative
773 implementation in BN_mod_exp_mont_consttime().) The old name
774 remains as a deprecated alias.
775
776 Similary, RSA_FLAG_NO_EXP_CONSTTIME is replaced by a more general
777 RSA_FLAG_NO_CONSTTIME flag since the RSA implementation now uses
778 constant-time implementations for more than just exponentiation.
779 Here too the old name is kept as a deprecated alias.
780
781 BN_BLINDING_new() will now use BN_dup() for the modulus so that
782 the BN_BLINDING structure gets an independent copy of the
783 modulus. This means that the previous "BIGNUM *m" argument to
784 BN_BLINDING_new() and to BN_BLINDING_create_param() now
785 essentially becomes "const BIGNUM *m", although we can't actually
786 change this in the header file before 0.9.9. It allows
787 RSA_setup_blinding() to use BN_with_flags() on the modulus to
788 enable BN_FLG_CONSTTIME.
789
790 [Matthew D Wood (Intel Corp)]
791
792 *) In the SSL/TLS server implementation, be strict about session ID
793 context matching (which matters if an application uses a single
794 external cache for different purposes). Previously,
795 out-of-context reuse was forbidden only if SSL_VERIFY_PEER was
796 set. This did ensure strict client verification, but meant that,
797 with applications using a single external cache for quite
798 different requirements, clients could circumvent ciphersuite
799 restrictions for a given session ID context by starting a session
800 in a different context.
801 [Bodo Moeller]
802
803 *) Include "!eNULL" in SSL_DEFAULT_CIPHER_LIST to make sure that
804 a ciphersuite string such as "DEFAULT:RSA" cannot enable
805 authentication-only ciphersuites.
806 [Bodo Moeller]
807
808 *) Update the SSL_get_shared_ciphers() fix CVE-2006-3738 which was
809 not complete and could lead to a possible single byte overflow
810 (CVE-2007-5135) [Ben Laurie]
811
812 Changes between 0.9.8d and 0.9.8e [23 Feb 2007]
813
814 *) Since AES128 and AES256 (and similarly Camellia128 and
815 Camellia256) share a single mask bit in the logic of
816 ssl/ssl_ciph.c, the code for masking out disabled ciphers needs a
817 kludge to work properly if AES128 is available and AES256 isn't
818 (or if Camellia128 is available and Camellia256 isn't).
819 [Victor Duchovni]
820
821 *) Fix the BIT STRING encoding generated by crypto/ec/ec_asn1.c
822 (within i2d_ECPrivateKey, i2d_ECPKParameters, i2d_ECParameters):
823 When a point or a seed is encoded in a BIT STRING, we need to
824 prevent the removal of trailing zero bits to get the proper DER
825 encoding. (By default, crypto/asn1/a_bitstr.c assumes the case
826 of a NamedBitList, for which trailing 0 bits need to be removed.)
827 [Bodo Moeller]
828
829 *) Have SSL/TLS server implementation tolerate "mismatched" record
830 protocol version while receiving ClientHello even if the
831 ClientHello is fragmented. (The server can't insist on the
832 particular protocol version it has chosen before the ServerHello
833 message has informed the client about his choice.)
834 [Bodo Moeller]
835
836 *) Add RFC 3779 support.
837 [Rob Austein for ARIN, Ben Laurie]
838
839 *) Load error codes if they are not already present instead of using a
840 static variable. This allows them to be cleanly unloaded and reloaded.
841 Improve header file function name parsing.
842 [Steve Henson]
843
844 *) extend SMTP and IMAP protocol emulation in s_client to use EHLO
845 or CAPABILITY handshake as required by RFCs.
846 [Goetz Babin-Ebell]
847
848 Changes between 0.9.8c and 0.9.8d [28 Sep 2006]
849
850 *) Introduce limits to prevent malicious keys being able to
851 cause a denial of service. (CVE-2006-2940)
852 [Steve Henson, Bodo Moeller]
853
854 *) Fix ASN.1 parsing of certain invalid structures that can result
855 in a denial of service. (CVE-2006-2937) [Steve Henson]
856
857 *) Fix buffer overflow in SSL_get_shared_ciphers() function.
858 (CVE-2006-3738) [Tavis Ormandy and Will Drewry, Google Security Team]
859
860 *) Fix SSL client code which could crash if connecting to a
861 malicious SSLv2 server. (CVE-2006-4343)
862 [Tavis Ormandy and Will Drewry, Google Security Team]
863
864 *) Since 0.9.8b, ciphersuite strings naming explicit ciphersuites
865 match only those. Before that, "AES256-SHA" would be interpreted
866 as a pattern and match "AES128-SHA" too (since AES128-SHA got
867 the same strength classification in 0.9.7h) as we currently only
868 have a single AES bit in the ciphersuite description bitmap.
869 That change, however, also applied to ciphersuite strings such as
870 "RC4-MD5" that intentionally matched multiple ciphersuites --
871 namely, SSL 2.0 ciphersuites in addition to the more common ones
872 from SSL 3.0/TLS 1.0.
873
874 So we change the selection algorithm again: Naming an explicit
875 ciphersuite selects this one ciphersuite, and any other similar
876 ciphersuite (same bitmap) from *other* protocol versions.
877 Thus, "RC4-MD5" again will properly select both the SSL 2.0
878 ciphersuite and the SSL 3.0/TLS 1.0 ciphersuite.
879
880 Since SSL 2.0 does not have any ciphersuites for which the
881 128/256 bit distinction would be relevant, this works for now.
882 The proper fix will be to use different bits for AES128 and
883 AES256, which would have avoided the problems from the beginning;
884 however, bits are scarce, so we can only do this in a new release
885 (not just a patchlevel) when we can change the SSL_CIPHER
886 definition to split the single 'unsigned long mask' bitmap into
887 multiple values to extend the available space.
888
889 [Bodo Moeller]
890
891 Changes between 0.9.8b and 0.9.8c [05 Sep 2006]
892
893 *) Avoid PKCS #1 v1.5 signature attack discovered by Daniel Bleichenbacher
894 (CVE-2006-4339) [Ben Laurie and Google Security Team]
895
896 *) Add AES IGE and biIGE modes.
897 [Ben Laurie]
898
899 *) Change the Unix randomness entropy gathering to use poll() when
900 possible instead of select(), since the latter has some
901 undesirable limitations.
902 [Darryl Miles via Richard Levitte and Bodo Moeller]
903
904 *) Disable "ECCdraft" ciphersuites more thoroughly. Now special
905 treatment in ssl/ssl_ciph.s makes sure that these ciphersuites
906 cannot be implicitly activated as part of, e.g., the "AES" alias.
907 However, please upgrade to OpenSSL 0.9.9[-dev] for
908 non-experimental use of the ECC ciphersuites to get TLS extension
909 support, which is required for curve and point format negotiation
910 to avoid potential handshake problems.
911 [Bodo Moeller]
912
913 *) Disable rogue ciphersuites:
914
915 - SSLv2 0x08 0x00 0x80 ("RC4-64-MD5")
916 - SSLv3/TLSv1 0x00 0x61 ("EXP1024-RC2-CBC-MD5")
917 - SSLv3/TLSv1 0x00 0x60 ("EXP1024-RC4-MD5")
918
919 The latter two were purportedly from
920 draft-ietf-tls-56-bit-ciphersuites-0[01].txt, but do not really
921 appear there.
922
923 Also deactivate the remaining ciphersuites from
924 draft-ietf-tls-56-bit-ciphersuites-01.txt. These are just as
925 unofficial, and the ID has long expired.
926 [Bodo Moeller]
927
928 *) Fix RSA blinding Heisenbug (problems sometimes occured on
929 dual-core machines) and other potential thread-safety issues.
930 [Bodo Moeller]
931
932 *) Add the symmetric cipher Camellia (128-bit, 192-bit, 256-bit key
933 versions), which is now available for royalty-free use
934 (see http://info.isl.ntt.co.jp/crypt/eng/info/chiteki.html).
935 Also, add Camellia TLS ciphersuites from RFC 4132.
936
937 To minimize changes between patchlevels in the OpenSSL 0.9.8
938 series, Camellia remains excluded from compilation unless OpenSSL
939 is configured with 'enable-camellia'.
940 [NTT]
941
942 *) Disable the padding bug check when compression is in use. The padding
943 bug check assumes the first packet is of even length, this is not
944 necessarily true if compresssion is enabled and can result in false
945 positives causing handshake failure. The actual bug test is ancient
946 code so it is hoped that implementations will either have fixed it by
947 now or any which still have the bug do not support compression.
948 [Steve Henson]
949
950 Changes between 0.9.8a and 0.9.8b [04 May 2006]
951
952 *) When applying a cipher rule check to see if string match is an explicit
953 cipher suite and only match that one cipher suite if it is.
954 [Steve Henson]
955
956 *) Link in manifests for VC++ if needed.
957 [Austin Ziegler <halostatue@gmail.com>]
958
959 *) Update support for ECC-based TLS ciphersuites according to
960 draft-ietf-tls-ecc-12.txt with proposed changes (but without
961 TLS extensions, which are supported starting with the 0.9.9
962 branch, not in the OpenSSL 0.9.8 branch).
963 [Douglas Stebila]
964
965 *) New functions EVP_CIPHER_CTX_new() and EVP_CIPHER_CTX_free() to support
966 opaque EVP_CIPHER_CTX handling.
967 [Steve Henson]
968
969 *) Fixes and enhancements to zlib compression code. We now only use
970 "zlib1.dll" and use the default __cdecl calling convention on Win32
971 to conform with the standards mentioned here:
972 http://www.zlib.net/DLL_FAQ.txt
973 Static zlib linking now works on Windows and the new --with-zlib-include
974 --with-zlib-lib options to Configure can be used to supply the location
975 of the headers and library. Gracefully handle case where zlib library
976 can't be loaded.
977 [Steve Henson]
978
979 *) Several fixes and enhancements to the OID generation code. The old code
980 sometimes allowed invalid OIDs (1.X for X >= 40 for example), couldn't
981 handle numbers larger than ULONG_MAX, truncated printing and had a
982 non standard OBJ_obj2txt() behaviour.
983 [Steve Henson]
984
985 *) Add support for building of engines under engine/ as shared libraries
986 under VC++ build system.
987 [Steve Henson]
988
989 *) Corrected the numerous bugs in the Win32 path splitter in DSO.
990 Hopefully, we will not see any false combination of paths any more.
991 [Richard Levitte]
992
993 Changes between 0.9.8 and 0.9.8a [11 Oct 2005]
994
995 *) Remove the functionality of SSL_OP_MSIE_SSLV2_RSA_PADDING
996 (part of SSL_OP_ALL). This option used to disable the
997 countermeasure against man-in-the-middle protocol-version
998 rollback in the SSL 2.0 server implementation, which is a bad
999 idea. (CVE-2005-2969)
1000
1001 [Bodo Moeller; problem pointed out by Yutaka Oiwa (Research Center
1002 for Information Security, National Institute of Advanced Industrial
1003 Science and Technology [AIST], Japan)]
1004
1005 *) Add two function to clear and return the verify parameter flags.
1006 [Steve Henson]
1007
1008 *) Keep cipherlists sorted in the source instead of sorting them at
1009 runtime, thus removing the need for a lock.
1010 [Nils Larsch]
1011
1012 *) Avoid some small subgroup attacks in Diffie-Hellman.
1013 [Nick Mathewson and Ben Laurie]
1014
1015 *) Add functions for well-known primes.
1016 [Nick Mathewson]
1017
1018 *) Extended Windows CE support.
1019 [Satoshi Nakamura and Andy Polyakov]
1020
1021 *) Initialize SSL_METHOD structures at compile time instead of during
1022 runtime, thus removing the need for a lock.
1023 [Steve Henson]
1024
1025 *) Make PKCS7_decrypt() work even if no certificate is supplied by
1026 attempting to decrypt each encrypted key in turn. Add support to
1027 smime utility.
1028 [Steve Henson]
1029
1030 Changes between 0.9.7h and 0.9.8 [05 Jul 2005]
1031
1032 [NB: OpenSSL 0.9.7i and later 0.9.7 patch levels were released after
1033 OpenSSL 0.9.8.]
1034
1035 *) Add libcrypto.pc and libssl.pc for those who feel they need them.
1036 [Richard Levitte]
1037
1038 *) Change CA.sh and CA.pl so they don't bundle the CSR and the private
1039 key into the same file any more.
1040 [Richard Levitte]
1041
1042 *) Add initial support for Win64, both IA64 and AMD64/x64 flavors.
1043 [Andy Polyakov]
1044
1045 *) Add -utf8 command line and config file option to 'ca'.
1046 [Stefan <stf@udoma.org]
1047
1048 *) Removed the macro des_crypt(), as it seems to conflict with some
1049 libraries. Use DES_crypt().
1050 [Richard Levitte]
1051
1052 *) Correct naming of the 'chil' and '4758cca' ENGINEs. This
1053 involves renaming the source and generated shared-libs for
1054 both. The engines will accept the corrected or legacy ids
1055 ('ncipher' and '4758_cca' respectively) when binding. NB,
1056 this only applies when building 'shared'.
1057 [Corinna Vinschen <vinschen@redhat.com> and Geoff Thorpe]
1058
1059 *) Add attribute functions to EVP_PKEY structure. Modify
1060 PKCS12_create() to recognize a CSP name attribute and
1061 use it. Make -CSP option work again in pkcs12 utility.
1062 [Steve Henson]
1063
1064 *) Add new functionality to the bn blinding code:
1065 - automatic re-creation of the BN_BLINDING parameters after
1066 a fixed number of uses (currently 32)
1067 - add new function for parameter creation
1068 - introduce flags to control the update behaviour of the
1069 BN_BLINDING parameters
1070 - hide BN_BLINDING structure
1071 Add a second BN_BLINDING slot to the RSA structure to improve
1072 performance when a single RSA object is shared among several
1073 threads.
1074 [Nils Larsch]
1075
1076 *) Add support for DTLS.
1077 [Nagendra Modadugu <nagendra@cs.stanford.edu> and Ben Laurie]
1078
1079 *) Add support for DER encoded private keys (SSL_FILETYPE_ASN1)
1080 to SSL_CTX_use_PrivateKey_file() and SSL_use_PrivateKey_file()
1081 [Walter Goulet]
1082
1083 *) Remove buggy and incompletet DH cert support from
1084 ssl/ssl_rsa.c and ssl/s3_both.c
1085 [Nils Larsch]
1086
1087 *) Use SHA-1 instead of MD5 as the default digest algorithm for
1088 the apps/openssl applications.
1089 [Nils Larsch]
1090
1091 *) Compile clean with "-Wall -Wmissing-prototypes
1092 -Wstrict-prototypes -Wmissing-declarations -Werror". Currently
1093 DEBUG_SAFESTACK must also be set.
1094 [Ben Laurie]
1095
1096 *) Change ./Configure so that certain algorithms can be disabled by default.
1097 The new counterpiece to "no-xxx" is "enable-xxx".
1098
1099 The patented RC5 and MDC2 algorithms will now be disabled unless
1100 "enable-rc5" and "enable-mdc2", respectively, are specified.
1101
1102 (IDEA remains enabled despite being patented. This is because IDEA
1103 is frequently required for interoperability, and there is no license
1104 fee for non-commercial use. As before, "no-idea" can be used to
1105 avoid this algorithm.)
1106
1107 [Bodo Moeller]
1108
1109 *) Add processing of proxy certificates (see RFC 3820). This work was
1110 sponsored by KTH (The Royal Institute of Technology in Stockholm) and
1111 EGEE (Enabling Grids for E-science in Europe).
1112 [Richard Levitte]
1113
1114 *) RC4 performance overhaul on modern architectures/implementations, such
1115 as Intel P4, IA-64 and AMD64.
1116 [Andy Polyakov]
1117
1118 *) New utility extract-section.pl. This can be used specify an alternative
1119 section number in a pod file instead of having to treat each file as
1120 a separate case in Makefile. This can be done by adding two lines to the
1121 pod file:
1122
1123 =for comment openssl_section:XXX
1124
1125 The blank line is mandatory.
1126
1127 [Steve Henson]
1128
1129 *) New arguments -certform, -keyform and -pass for s_client and s_server
1130 to allow alternative format key and certificate files and passphrase
1131 sources.
1132 [Steve Henson]
1133
1134 *) New structure X509_VERIFY_PARAM which combines current verify parameters,
1135 update associated structures and add various utility functions.
1136
1137 Add new policy related verify parameters, include policy checking in
1138 standard verify code. Enhance 'smime' application with extra parameters
1139 to support policy checking and print out.
1140 [Steve Henson]
1141
1142 *) Add a new engine to support VIA PadLock ACE extensions in the VIA C3
1143 Nehemiah processors. These extensions support AES encryption in hardware
1144 as well as RNG (though RNG support is currently disabled).
1145 [Michal Ludvig <michal@logix.cz>, with help from Andy Polyakov]
1146
1147 *) Deprecate BN_[get|set]_params() functions (they were ignored internally).
1148 [Geoff Thorpe]
1149
1150 *) New FIPS 180-2 algorithms, SHA-224/-256/-384/-512 are implemented.
1151 [Andy Polyakov and a number of other people]
1152
1153 *) Improved PowerPC platform support. Most notably BIGNUM assembler
1154 implementation contributed by IBM.
1155 [Suresh Chari, Peter Waltenberg, Andy Polyakov]
1156
1157 *) The new 'RSA_generate_key_ex' function now takes a BIGNUM for the public
1158 exponent rather than 'unsigned long'. There is a corresponding change to
1159 the new 'rsa_keygen' element of the RSA_METHOD structure.
1160 [Jelte Jansen, Geoff Thorpe]
1161
1162 *) Functionality for creating the initial serial number file is now
1163 moved from CA.pl to the 'ca' utility with a new option -create_serial.
1164
1165 (Before OpenSSL 0.9.7e, CA.pl used to initialize the serial
1166 number file to 1, which is bound to cause problems. To avoid
1167 the problems while respecting compatibility between different 0.9.7
1168 patchlevels, 0.9.7e employed 'openssl x509 -next_serial' in
1169 CA.pl for serial number initialization. With the new release 0.9.8,
1170 we can fix the problem directly in the 'ca' utility.)
1171 [Steve Henson]
1172
1173 *) Reduced header interdepencies by declaring more opaque objects in
1174 ossl_typ.h. As a consequence, including some headers (eg. engine.h) will
1175 give fewer recursive includes, which could break lazy source code - so
1176 this change is covered by the OPENSSL_NO_DEPRECATED symbol. As always,
1177 developers should define this symbol when building and using openssl to
1178 ensure they track the recommended behaviour, interfaces, [etc], but
1179 backwards-compatible behaviour prevails when this isn't defined.
1180 [Geoff Thorpe]
1181
1182 *) New function X509_POLICY_NODE_print() which prints out policy nodes.
1183 [Steve Henson]
1184
1185 *) Add new EVP function EVP_CIPHER_CTX_rand_key and associated functionality.
1186 This will generate a random key of the appropriate length based on the
1187 cipher context. The EVP_CIPHER can provide its own random key generation
1188 routine to support keys of a specific form. This is used in the des and
1189 3des routines to generate a key of the correct parity. Update S/MIME
1190 code to use new functions and hence generate correct parity DES keys.
1191 Add EVP_CHECK_DES_KEY #define to return an error if the key is not
1192 valid (weak or incorrect parity).
1193 [Steve Henson]
1194
1195 *) Add a local set of CRLs that can be used by X509_verify_cert() as well
1196 as looking them up. This is useful when the verified structure may contain
1197 CRLs, for example PKCS#7 signedData. Modify PKCS7_verify() to use any CRLs
1198 present unless the new PKCS7_NO_CRL flag is asserted.
1199 [Steve Henson]
1200
1201 *) Extend ASN1 oid configuration module. It now additionally accepts the
1202 syntax:
1203
1204 shortName = some long name, 1.2.3.4
1205 [Steve Henson]
1206
1207 *) Reimplemented the BN_CTX implementation. There is now no more static
1208 limitation on the number of variables it can handle nor the depth of the
1209 "stack" handling for BN_CTX_start()/BN_CTX_end() pairs. The stack
1210 information can now expand as required, and rather than having a single
1211 static array of bignums, BN_CTX now uses a linked-list of such arrays
1212 allowing it to expand on demand whilst maintaining the usefulness of
1213 BN_CTX's "bundling".
1214 [Geoff Thorpe]
1215
1216 *) Add a missing BN_CTX parameter to the 'rsa_mod_exp' callback in RSA_METHOD
1217 to allow all RSA operations to function using a single BN_CTX.
1218 [Geoff Thorpe]
1219
1220 *) Preliminary support for certificate policy evaluation and checking. This
1221 is initially intended to pass the tests outlined in "Conformance Testing
1222 of Relying Party Client Certificate Path Processing Logic" v1.07.
1223 [Steve Henson]
1224
1225 *) bn_dup_expand() has been deprecated, it was introduced in 0.9.7 and
1226 remained unused and not that useful. A variety of other little bignum
1227 tweaks and fixes have also been made continuing on from the audit (see
1228 below).
1229 [Geoff Thorpe]
1230
1231 *) Constify all or almost all d2i, c2i, s2i and r2i functions, along with
1232 associated ASN1, EVP and SSL functions and old ASN1 macros.
1233 [Richard Levitte]
1234
1235 *) BN_zero() only needs to set 'top' and 'neg' to zero for correct results,
1236 and this should never fail. So the return value from the use of
1237 BN_set_word() (which can fail due to needless expansion) is now deprecated;
1238 if OPENSSL_NO_DEPRECATED is defined, BN_zero() is a void macro.
1239 [Geoff Thorpe]
1240
1241 *) BN_CTX_get() should return zero-valued bignums, providing the same
1242 initialised value as BN_new().
1243 [Geoff Thorpe, suggested by Ulf Möller]
1244
1245 *) Support for inhibitAnyPolicy certificate extension.
1246 [Steve Henson]
1247
1248 *) An audit of the BIGNUM code is underway, for which debugging code is
1249 enabled when BN_DEBUG is defined. This makes stricter enforcements on what
1250 is considered valid when processing BIGNUMs, and causes execution to
1251 assert() when a problem is discovered. If BN_DEBUG_RAND is defined,
1252 further steps are taken to deliberately pollute unused data in BIGNUM
1253 structures to try and expose faulty code further on. For now, openssl will
1254 (in its default mode of operation) continue to tolerate the inconsistent
1255 forms that it has tolerated in the past, but authors and packagers should
1256 consider trying openssl and their own applications when compiled with
1257 these debugging symbols defined. It will help highlight potential bugs in
1258 their own code, and will improve the test coverage for OpenSSL itself. At
1259 some point, these tighter rules will become openssl's default to improve
1260 maintainability, though the assert()s and other overheads will remain only
1261 in debugging configurations. See bn.h for more details.
1262 [Geoff Thorpe, Nils Larsch, Ulf Möller]
1263
1264 *) BN_CTX_init() has been deprecated, as BN_CTX is an opaque structure
1265 that can only be obtained through BN_CTX_new() (which implicitly
1266 initialises it). The presence of this function only made it possible
1267 to overwrite an existing structure (and cause memory leaks).
1268 [Geoff Thorpe]
1269
1270 *) Because of the callback-based approach for implementing LHASH as a
1271 template type, lh_insert() adds opaque objects to hash-tables and
1272 lh_doall() or lh_doall_arg() are typically used with a destructor callback
1273 to clean up those corresponding objects before destroying the hash table
1274 (and losing the object pointers). So some over-zealous constifications in
1275 LHASH have been relaxed so that lh_insert() does not take (nor store) the
1276 objects as "const" and the lh_doall[_arg] callback wrappers are not
1277 prototyped to have "const" restrictions on the object pointers they are
1278 given (and so aren't required to cast them away any more).
1279 [Geoff Thorpe]
1280
1281 *) The tmdiff.h API was so ugly and minimal that our own timing utility
1282 (speed) prefers to use its own implementation. The two implementations
1283 haven't been consolidated as yet (volunteers?) but the tmdiff API has had
1284 its object type properly exposed (MS_TM) instead of casting to/from "char
1285 *". This may still change yet if someone realises MS_TM and "ms_time_***"
1286 aren't necessarily the greatest nomenclatures - but this is what was used
1287 internally to the implementation so I've used that for now.
1288 [Geoff Thorpe]
1289
1290 *) Ensure that deprecated functions do not get compiled when
1291 OPENSSL_NO_DEPRECATED is defined. Some "openssl" subcommands and a few of
1292 the self-tests were still using deprecated key-generation functions so
1293 these have been updated also.
1294 [Geoff Thorpe]
1295
1296 *) Reorganise PKCS#7 code to separate the digest location functionality
1297 into PKCS7_find_digest(), digest addtion into PKCS7_bio_add_digest().
1298 New function PKCS7_set_digest() to set the digest type for PKCS#7
1299 digestedData type. Add additional code to correctly generate the
1300 digestedData type and add support for this type in PKCS7 initialization
1301 functions.
1302 [Steve Henson]
1303
1304 *) New function PKCS7_set0_type_other() this initializes a PKCS7
1305 structure of type "other".
1306 [Steve Henson]
1307
1308 *) Fix prime generation loop in crypto/bn/bn_prime.pl by making
1309 sure the loop does correctly stop and breaking ("division by zero")
1310 modulus operations are not performed. The (pre-generated) prime
1311 table crypto/bn/bn_prime.h was already correct, but it could not be
1312 re-generated on some platforms because of the "division by zero"
1313 situation in the script.
1314 [Ralf S. Engelschall]
1315
1316 *) Update support for ECC-based TLS ciphersuites according to
1317 draft-ietf-tls-ecc-03.txt: the KDF1 key derivation function with
1318 SHA-1 now is only used for "small" curves (where the
1319 representation of a field element takes up to 24 bytes); for
1320 larger curves, the field element resulting from ECDH is directly
1321 used as premaster secret.
1322 [Douglas Stebila (Sun Microsystems Laboratories)]
1323
1324 *) Add code for kP+lQ timings to crypto/ec/ectest.c, and add SEC2
1325 curve secp160r1 to the tests.
1326 [Douglas Stebila (Sun Microsystems Laboratories)]
1327
1328 *) Add the possibility to load symbols globally with DSO.
1329 [Götz Babin-Ebell <babin-ebell@trustcenter.de> via Richard Levitte]
1330
1331 *) Add the functions ERR_set_mark() and ERR_pop_to_mark() for better
1332 control of the error stack.
1333 [Richard Levitte]
1334
1335 *) Add support for STORE in ENGINE.
1336 [Richard Levitte]
1337
1338 *) Add the STORE type. The intention is to provide a common interface
1339 to certificate and key stores, be they simple file-based stores, or
1340 HSM-type store, or LDAP stores, or...
1341 NOTE: The code is currently UNTESTED and isn't really used anywhere.
1342 [Richard Levitte]
1343
1344 *) Add a generic structure called OPENSSL_ITEM. This can be used to
1345 pass a list of arguments to any function as well as provide a way
1346 for a function to pass data back to the caller.
1347 [Richard Levitte]
1348
1349 *) Add the functions BUF_strndup() and BUF_memdup(). BUF_strndup()
1350 works like BUF_strdup() but can be used to duplicate a portion of
1351 a string. The copy gets NUL-terminated. BUF_memdup() duplicates
1352 a memory area.
1353 [Richard Levitte]
1354
1355 *) Add the function sk_find_ex() which works like sk_find(), but will
1356 return an index to an element even if an exact match couldn't be
1357 found. The index is guaranteed to point at the element where the
1358 searched-for key would be inserted to preserve sorting order.
1359 [Richard Levitte]
1360
1361 *) Add the function OBJ_bsearch_ex() which works like OBJ_bsearch() but
1362 takes an extra flags argument for optional functionality. Currently,
1363 the following flags are defined:
1364
1365 OBJ_BSEARCH_VALUE_ON_NOMATCH
1366 This one gets OBJ_bsearch_ex() to return a pointer to the first
1367 element where the comparing function returns a negative or zero
1368 number.
1369
1370 OBJ_BSEARCH_FIRST_VALUE_ON_MATCH
1371 This one gets OBJ_bsearch_ex() to return a pointer to the first
1372 element where the comparing function returns zero. This is useful
1373 if there are more than one element where the comparing function
1374 returns zero.
1375 [Richard Levitte]
1376
1377 *) Make it possible to create self-signed certificates with 'openssl ca'
1378 in such a way that the self-signed certificate becomes part of the
1379 CA database and uses the same mechanisms for serial number generation
1380 as all other certificate signing. The new flag '-selfsign' enables
1381 this functionality. Adapt CA.sh and CA.pl.in.
1382 [Richard Levitte]
1383
1384 *) Add functionality to check the public key of a certificate request
1385 against a given private. This is useful to check that a certificate
1386 request can be signed by that key (self-signing).
1387 [Richard Levitte]
1388
1389 *) Make it possible to have multiple active certificates with the same
1390 subject in the CA index file. This is done only if the keyword
1391 'unique_subject' is set to 'no' in the main CA section (default
1392 if 'CA_default') of the configuration file. The value is saved
1393 with the database itself in a separate index attribute file,
1394 named like the index file with '.attr' appended to the name.
1395 [Richard Levitte]
1396
1397 *) Generate muti valued AVAs using '+' notation in config files for
1398 req and dirName.
1399 [Steve Henson]
1400
1401 *) Support for nameConstraints certificate extension.
1402 [Steve Henson]
1403
1404 *) Support for policyConstraints certificate extension.
1405 [Steve Henson]
1406
1407 *) Support for policyMappings certificate extension.
1408 [Steve Henson]
1409
1410 *) Make sure the default DSA_METHOD implementation only uses its
1411 dsa_mod_exp() and/or bn_mod_exp() handlers if they are non-NULL,
1412 and change its own handlers to be NULL so as to remove unnecessary
1413 indirection. This lets alternative implementations fallback to the
1414 default implementation more easily.
1415 [Geoff Thorpe]
1416
1417 *) Support for directoryName in GeneralName related extensions
1418 in config files.
1419 [Steve Henson]
1420
1421 *) Make it possible to link applications using Makefile.shared.
1422 Make that possible even when linking against static libraries!
1423 [Richard Levitte]
1424
1425 *) Support for single pass processing for S/MIME signing. This now
1426 means that S/MIME signing can be done from a pipe, in addition
1427 cleartext signing (multipart/signed type) is effectively streaming
1428 and the signed data does not need to be all held in memory.
1429
1430 This is done with a new flag PKCS7_STREAM. When this flag is set
1431 PKCS7_sign() only initializes the PKCS7 structure and the actual signing
1432 is done after the data is output (and digests calculated) in
1433 SMIME_write_PKCS7().
1434 [Steve Henson]
1435
1436 *) Add full support for -rpath/-R, both in shared libraries and
1437 applications, at least on the platforms where it's known how
1438 to do it.
1439 [Richard Levitte]
1440
1441 *) In crypto/ec/ec_mult.c, implement fast point multiplication with
1442 precomputation, based on wNAF splitting: EC_GROUP_precompute_mult()
1443 will now compute a table of multiples of the generator that
1444 makes subsequent invocations of EC_POINTs_mul() or EC_POINT_mul()
1445 faster (notably in the case of a single point multiplication,
1446 scalar * generator).
1447 [Nils Larsch, Bodo Moeller]
1448
1449 *) IPv6 support for certificate extensions. The various extensions
1450 which use the IP:a.b.c.d can now take IPv6 addresses using the
1451 formats of RFC1884 2.2 . IPv6 addresses are now also displayed
1452 correctly.
1453 [Steve Henson]
1454
1455 *) Added an ENGINE that implements RSA by performing private key
1456 exponentiations with the GMP library. The conversions to and from
1457 GMP's mpz_t format aren't optimised nor are any montgomery forms
1458 cached, and on x86 it appears OpenSSL's own performance has caught up.
1459 However there are likely to be other architectures where GMP could
1460 provide a boost. This ENGINE is not built in by default, but it can be
1461 specified at Configure time and should be accompanied by the necessary
1462 linker additions, eg;
1463 ./config -DOPENSSL_USE_GMP -lgmp
1464 [Geoff Thorpe]
1465
1466 *) "openssl engine" will not display ENGINE/DSO load failure errors when
1467 testing availability of engines with "-t" - the old behaviour is
1468 produced by increasing the feature's verbosity with "-tt".
1469 [Geoff Thorpe]
1470
1471 *) ECDSA routines: under certain error conditions uninitialized BN objects
1472 could be freed. Solution: make sure initialization is performed early
1473 enough. (Reported and fix supplied by Nils Larsch <nla@trustcenter.de>
1474 via PR#459)
1475 [Lutz Jaenicke]
1476
1477 *) Key-generation can now be implemented in RSA_METHOD, DSA_METHOD
1478 and DH_METHOD (eg. by ENGINE implementations) to override the normal
1479 software implementations. For DSA and DH, parameter generation can
1480 also be overriden by providing the appropriate method callbacks.
1481 [Geoff Thorpe]
1482
1483 *) Change the "progress" mechanism used in key-generation and
1484 primality testing to functions that take a new BN_GENCB pointer in
1485 place of callback/argument pairs. The new API functions have "_ex"
1486 postfixes and the older functions are reimplemented as wrappers for
1487 the new ones. The OPENSSL_NO_DEPRECATED symbol can be used to hide
1488 declarations of the old functions to help (graceful) attempts to
1489 migrate to the new functions. Also, the new key-generation API
1490 functions operate on a caller-supplied key-structure and return
1491 success/failure rather than returning a key or NULL - this is to
1492 help make "keygen" another member function of RSA_METHOD etc.
1493
1494 Example for using the new callback interface:
1495
1496 int (*my_callback)(int a, int b, BN_GENCB *cb) = ...;
1497 void *my_arg = ...;
1498 BN_GENCB my_cb;
1499
1500 BN_GENCB_set(&my_cb, my_callback, my_arg);
1501
1502 return BN_is_prime_ex(some_bignum, BN_prime_checks, NULL, &cb);
1503 /* For the meaning of a, b in calls to my_callback(), see the
1504 * documentation of the function that calls the callback.
1505 * cb will point to my_cb; my_arg can be retrieved as cb->arg.
1506 * my_callback should return 1 if it wants BN_is_prime_ex()
1507 * to continue, or 0 to stop.
1508 */
1509
1510 [Geoff Thorpe]
1511
1512 *) Change the ZLIB compression method to be stateful, and make it
1513 available to TLS with the number defined in
1514 draft-ietf-tls-compression-04.txt.
1515 [Richard Levitte]
1516
1517 *) Add the ASN.1 structures and functions for CertificatePair, which
1518 is defined as follows (according to X.509_4thEditionDraftV6.pdf):
1519
1520 CertificatePair ::= SEQUENCE {
1521 forward [0] Certificate OPTIONAL,
1522 reverse [1] Certificate OPTIONAL,
1523 -- at least one of the pair shall be present -- }
1524
1525 Also implement the PEM functions to read and write certificate
1526 pairs, and defined the PEM tag as "CERTIFICATE PAIR".
1527
1528 This needed to be defined, mostly for the sake of the LDAP
1529 attribute crossCertificatePair, but may prove useful elsewhere as
1530 well.
1531 [Richard Levitte]
1532
1533 *) Make it possible to inhibit symlinking of shared libraries in
1534 Makefile.shared, for Cygwin's sake.
1535 [Richard Levitte]
1536
1537 *) Extend the BIGNUM API by creating a function
1538 void BN_set_negative(BIGNUM *a, int neg);
1539 and a macro that behave like
1540 int BN_is_negative(const BIGNUM *a);
1541
1542 to avoid the need to access 'a->neg' directly in applications.
1543 [Nils Larsch]
1544
1545 *) Implement fast modular reduction for pseudo-Mersenne primes
1546 used in NIST curves (crypto/bn/bn_nist.c, crypto/ec/ecp_nist.c).
1547 EC_GROUP_new_curve_GFp() will now automatically use this
1548 if applicable.
1549 [Nils Larsch <nla@trustcenter.de>]
1550
1551 *) Add new lock type (CRYPTO_LOCK_BN).
1552 [Bodo Moeller]
1553
1554 *) Change the ENGINE framework to automatically load engines
1555 dynamically from specific directories unless they could be
1556 found to already be built in or loaded. Move all the
1557 current engines except for the cryptodev one to a new
1558 directory engines/.
1559 The engines in engines/ are built as shared libraries if
1560 the "shared" options was given to ./Configure or ./config.
1561 Otherwise, they are inserted in libcrypto.a.
1562 /usr/local/ssl/engines is the default directory for dynamic
1563 engines, but that can be overriden at configure time through
1564 the usual use of --prefix and/or --openssldir, and at run
1565 time with the environment variable OPENSSL_ENGINES.
1566 [Geoff Thorpe and Richard Levitte]
1567
1568 *) Add Makefile.shared, a helper makefile to build shared
1569 libraries. Addapt Makefile.org.
1570 [Richard Levitte]
1571
1572 *) Add version info to Win32 DLLs.
1573 [Peter 'Luna' Runestig" <peter@runestig.com>]
1574
1575 *) Add new 'medium level' PKCS#12 API. Certificates and keys
1576 can be added using this API to created arbitrary PKCS#12
1577 files while avoiding the low level API.
1578
1579 New options to PKCS12_create(), key or cert can be NULL and
1580 will then be omitted from the output file. The encryption
1581 algorithm NIDs can be set to -1 for no encryption, the mac
1582 iteration count can be set to 0 to omit the mac.
1583
1584 Enhance pkcs12 utility by making the -nokeys and -nocerts
1585 options work when creating a PKCS#12 file. New option -nomac
1586 to omit the mac, NONE can be set for an encryption algorithm.
1587 New code is modified to use the enhanced PKCS12_create()
1588 instead of the low level API.
1589 [Steve Henson]
1590
1591 *) Extend ASN1 encoder to support indefinite length constructed
1592 encoding. This can output sequences tags and octet strings in
1593 this form. Modify pk7_asn1.c to support indefinite length
1594 encoding. This is experimental and needs additional code to
1595 be useful, such as an ASN1 bio and some enhanced streaming
1596 PKCS#7 code.
1597
1598 Extend template encode functionality so that tagging is passed
1599 down to the template encoder.
1600 [Steve Henson]
1601
1602 *) Let 'openssl req' fail if an argument to '-newkey' is not
1603 recognized instead of using RSA as a default.
1604 [Bodo Moeller]
1605
1606 *) Add support for ECC-based ciphersuites from draft-ietf-tls-ecc-01.txt.
1607 As these are not official, they are not included in "ALL";
1608 the "ECCdraft" ciphersuite group alias can be used to select them.
1609 [Vipul Gupta and Sumit Gupta (Sun Microsystems Laboratories)]
1610
1611 *) Add ECDH engine support.
1612 [Nils Gura and Douglas Stebila (Sun Microsystems Laboratories)]
1613
1614 *) Add ECDH in new directory crypto/ecdh/.
1615 [Douglas Stebila (Sun Microsystems Laboratories)]
1616
1617 *) Let BN_rand_range() abort with an error after 100 iterations
1618 without success (which indicates a broken PRNG).
1619 [Bodo Moeller]
1620
1621 *) Change BN_mod_sqrt() so that it verifies that the input value
1622 is really the square of the return value. (Previously,
1623 BN_mod_sqrt would show GIGO behaviour.)
1624 [Bodo Moeller]
1625
1626 *) Add named elliptic curves over binary fields from X9.62, SECG,
1627 and WAP/WTLS; add OIDs that were still missing.
1628
1629 [Sheueling Chang Shantz and Douglas Stebila
1630 (Sun Microsystems Laboratories)]
1631
1632 *) Extend the EC library for elliptic curves over binary fields
1633 (new files ec2_smpl.c, ec2_smpt.c, ec2_mult.c in crypto/ec/).
1634 New EC_METHOD:
1635
1636 EC_GF2m_simple_method
1637
1638 New API functions:
1639
1640 EC_GROUP_new_curve_GF2m
1641 EC_GROUP_set_curve_GF2m
1642 EC_GROUP_get_curve_GF2m
1643 EC_POINT_set_affine_coordinates_GF2m
1644 EC_POINT_get_affine_coordinates_GF2m
1645 EC_POINT_set_compressed_coordinates_GF2m
1646
1647 Point compression for binary fields is disabled by default for
1648 patent reasons (compile with OPENSSL_EC_BIN_PT_COMP defined to
1649 enable it).
1650
1651 As binary polynomials are represented as BIGNUMs, various members
1652 of the EC_GROUP and EC_POINT data structures can be shared
1653 between the implementations for prime fields and binary fields;
1654 the above ..._GF2m functions (except for EX_GROUP_new_curve_GF2m)
1655 are essentially identical to their ..._GFp counterparts.
1656 (For simplicity, the '..._GFp' prefix has been dropped from
1657 various internal method names.)
1658
1659 An internal 'field_div' method (similar to 'field_mul' and
1660 'field_sqr') has been added; this is used only for binary fields.
1661
1662 [Sheueling Chang Shantz and Douglas Stebila
1663 (Sun Microsystems Laboratories)]
1664
1665 *) Optionally dispatch EC_POINT_mul(), EC_POINT_precompute_mult()
1666 through methods ('mul', 'precompute_mult').
1667
1668 The generic implementations (now internally called 'ec_wNAF_mul'
1669 and 'ec_wNAF_precomputed_mult') remain the default if these
1670 methods are undefined.
1671
1672 [Sheueling Chang Shantz and Douglas Stebila
1673 (Sun Microsystems Laboratories)]
1674
1675 *) New function EC_GROUP_get_degree, which is defined through
1676 EC_METHOD. For curves over prime fields, this returns the bit
1677 length of the modulus.
1678
1679 [Sheueling Chang Shantz and Douglas Stebila
1680 (Sun Microsystems Laboratories)]
1681
1682 *) New functions EC_GROUP_dup, EC_POINT_dup.
1683 (These simply call ..._new and ..._copy).
1684
1685 [Sheueling Chang Shantz and Douglas Stebila
1686 (Sun Microsystems Laboratories)]
1687
1688 *) Add binary polynomial arithmetic software in crypto/bn/bn_gf2m.c.
1689 Polynomials are represented as BIGNUMs (where the sign bit is not
1690 used) in the following functions [macros]:
1691
1692 BN_GF2m_add
1693 BN_GF2m_sub [= BN_GF2m_add]
1694 BN_GF2m_mod [wrapper for BN_GF2m_mod_arr]
1695 BN_GF2m_mod_mul [wrapper for BN_GF2m_mod_mul_arr]
1696 BN_GF2m_mod_sqr [wrapper for BN_GF2m_mod_sqr_arr]
1697 BN_GF2m_mod_inv
1698 BN_GF2m_mod_exp [wrapper for BN_GF2m_mod_exp_arr]
1699 BN_GF2m_mod_sqrt [wrapper for BN_GF2m_mod_sqrt_arr]
1700 BN_GF2m_mod_solve_quad [wrapper for BN_GF2m_mod_solve_quad_arr]
1701 BN_GF2m_cmp [= BN_ucmp]
1702
1703 (Note that only the 'mod' functions are actually for fields GF(2^m).
1704 BN_GF2m_add() is misnomer, but this is for the sake of consistency.)
1705
1706 For some functions, an the irreducible polynomial defining a
1707 field can be given as an 'unsigned int[]' with strictly
1708 decreasing elements giving the indices of those bits that are set;
1709 i.e., p[] represents the polynomial
1710 f(t) = t^p[0] + t^p[1] + ... + t^p[k]
1711 where
1712 p[0] > p[1] > ... > p[k] = 0.
1713 This applies to the following functions:
1714
1715 BN_GF2m_mod_arr
1716 BN_GF2m_mod_mul_arr
1717 BN_GF2m_mod_sqr_arr
1718 BN_GF2m_mod_inv_arr [wrapper for BN_GF2m_mod_inv]
1719 BN_GF2m_mod_div_arr [wrapper for BN_GF2m_mod_div]
1720 BN_GF2m_mod_exp_arr
1721 BN_GF2m_mod_sqrt_arr
1722 BN_GF2m_mod_solve_quad_arr
1723 BN_GF2m_poly2arr
1724 BN_GF2m_arr2poly
1725
1726 Conversion can be performed by the following functions:
1727
1728 BN_GF2m_poly2arr
1729 BN_GF2m_arr2poly
1730
1731 bntest.c has additional tests for binary polynomial arithmetic.
1732
1733 Two implementations for BN_GF2m_mod_div() are available.
1734 The default algorithm simply uses BN_GF2m_mod_inv() and
1735 BN_GF2m_mod_mul(). The alternative algorithm is compiled in only
1736 if OPENSSL_SUN_GF2M_DIV is defined (patent pending; read the
1737 copyright notice in crypto/bn/bn_gf2m.c before enabling it).
1738
1739 [Sheueling Chang Shantz and Douglas Stebila
1740 (Sun Microsystems Laboratories)]
1741
1742 *) Add new error code 'ERR_R_DISABLED' that can be used when some
1743 functionality is disabled at compile-time.
1744 [Douglas Stebila <douglas.stebila@sun.com>]
1745
1746 *) Change default behaviour of 'openssl asn1parse' so that more
1747 information is visible when viewing, e.g., a certificate:
1748
1749 Modify asn1_parse2 (crypto/asn1/asn1_par.c) so that in non-'dump'
1750 mode the content of non-printable OCTET STRINGs is output in a
1751 style similar to INTEGERs, but with '[HEX DUMP]' prepended to
1752 avoid the appearance of a printable string.
1753 [Nils Larsch <nla@trustcenter.de>]
1754
1755 *) Add 'asn1_flag' and 'asn1_form' member to EC_GROUP with access
1756 functions
1757 EC_GROUP_set_asn1_flag()
1758 EC_GROUP_get_asn1_flag()
1759 EC_GROUP_set_point_conversion_form()
1760 EC_GROUP_get_point_conversion_form()
1761 These control ASN1 encoding details:
1762 - Curves (i.e., groups) are encoded explicitly unless asn1_flag
1763 has been set to OPENSSL_EC_NAMED_CURVE.
1764 - Points are encoded in uncompressed form by default; options for
1765 asn1_for are as for point2oct, namely
1766 POINT_CONVERSION_COMPRESSED
1767 POINT_CONVERSION_UNCOMPRESSED
1768 POINT_CONVERSION_HYBRID
1769
1770 Also add 'seed' and 'seed_len' members to EC_GROUP with access
1771 functions
1772 EC_GROUP_set_seed()
1773 EC_GROUP_get0_seed()
1774 EC_GROUP_get_seed_len()
1775 This is used only for ASN1 purposes (so far).
1776 [Nils Larsch <nla@trustcenter.de>]
1777
1778 *) Add 'field_type' member to EC_METHOD, which holds the NID
1779 of the appropriate field type OID. The new function
1780 EC_METHOD_get_field_type() returns this value.
1781 [Nils Larsch <nla@trustcenter.de>]
1782
1783 *) Add functions
1784 EC_POINT_point2bn()
1785 EC_POINT_bn2point()
1786 EC_POINT_point2hex()
1787 EC_POINT_hex2point()
1788 providing useful interfaces to EC_POINT_point2oct() and
1789 EC_POINT_oct2point().
1790 [Nils Larsch <nla@trustcenter.de>]
1791
1792 *) Change internals of the EC library so that the functions
1793 EC_GROUP_set_generator()
1794 EC_GROUP_get_generator()
1795 EC_GROUP_get_order()
1796 EC_GROUP_get_cofactor()
1797 are implemented directly in crypto/ec/ec_lib.c and not dispatched
1798 to methods, which would lead to unnecessary code duplication when
1799 adding different types of curves.
1800 [Nils Larsch <nla@trustcenter.de> with input by Bodo Moeller]
1801
1802 *) Implement compute_wNAF (crypto/ec/ec_mult.c) without BIGNUM
1803 arithmetic, and such that modified wNAFs are generated
1804 (which avoid length expansion in many cases).
1805 [Bodo Moeller]
1806
1807 *) Add a function EC_GROUP_check_discriminant() (defined via
1808 EC_METHOD) that verifies that the curve discriminant is non-zero.
1809
1810 Add a function EC_GROUP_check() that makes some sanity tests
1811 on a EC_GROUP, its generator and order. This includes
1812 EC_GROUP_check_discriminant().
1813 [Nils Larsch <nla@trustcenter.de>]
1814
1815 *) Add ECDSA in new directory crypto/ecdsa/.
1816
1817 Add applications 'openssl ecparam' and 'openssl ecdsa'
1818 (these are based on 'openssl dsaparam' and 'openssl dsa').
1819
1820 ECDSA support is also included in various other files across the
1821 library. Most notably,
1822 - 'openssl req' now has a '-newkey ecdsa:file' option;
1823 - EVP_PKCS82PKEY (crypto/evp/evp_pkey.c) now can handle ECDSA;
1824 - X509_PUBKEY_get (crypto/asn1/x_pubkey.c) and
1825 d2i_PublicKey (crypto/asn1/d2i_pu.c) have been modified to make
1826 them suitable for ECDSA where domain parameters must be
1827 extracted before the specific public key;
1828 - ECDSA engine support has been added.
1829 [Nils Larsch <nla@trustcenter.de>]
1830
1831 *) Include some named elliptic curves, and add OIDs from X9.62,
1832 SECG, and WAP/WTLS. Each curve can be obtained from the new
1833 function
1834 EC_GROUP_new_by_curve_name(),
1835 and the list of available named curves can be obtained with
1836 EC_get_builtin_curves().
1837 Also add a 'curve_name' member to EC_GROUP objects, which can be
1838 accessed via
1839 EC_GROUP_set_curve_name()
1840 EC_GROUP_get_curve_name()
1841 [Nils Larsch <larsch@trustcenter.de, Bodo Moeller]
1842
1843 *) Remove a few calls to bn_wexpand() in BN_sqr() (the one in there
1844 was actually never needed) and in BN_mul(). The removal in BN_mul()
1845 required a small change in bn_mul_part_recursive() and the addition
1846 of the functions bn_cmp_part_words(), bn_sub_part_words() and
1847 bn_add_part_words(), which do the same thing as bn_cmp_words(),
1848 bn_sub_words() and bn_add_words() except they take arrays with
1849 differing sizes.
1850 [Richard Levitte]
1851
1852 Changes between 0.9.7l and 0.9.7m [23 Feb 2007]
1853
1854 *) Cleanse PEM buffers before freeing them since they may contain
1855 sensitive data.
1856 [Benjamin Bennett <ben@psc.edu>]
1857
1858 *) Include "!eNULL" in SSL_DEFAULT_CIPHER_LIST to make sure that
1859 a ciphersuite string such as "DEFAULT:RSA" cannot enable
1860 authentication-only ciphersuites.
1861 [Bodo Moeller]
1862
1863 *) Since AES128 and AES256 share a single mask bit in the logic of
1864 ssl/ssl_ciph.c, the code for masking out disabled ciphers needs a
1865 kludge to work properly if AES128 is available and AES256 isn't.
1866 [Victor Duchovni]
1867
1868 *) Expand security boundary to match 1.1.1 module.
1869 [Steve Henson]
1870
1871 *) Remove redundant features: hash file source, editing of test vectors
1872 modify fipsld to use external fips_premain.c signature.
1873 [Steve Henson]
1874
1875 *) New perl script mkfipsscr.pl to create shell scripts or batch files to
1876 run algorithm test programs.
1877 [Steve Henson]
1878
1879 *) Make algorithm test programs more tolerant of whitespace.
1880 [Steve Henson]
1881
1882 *) Have SSL/TLS server implementation tolerate "mismatched" record
1883 protocol version while receiving ClientHello even if the
1884 ClientHello is fragmented. (The server can't insist on the
1885 particular protocol version it has chosen before the ServerHello
1886 message has informed the client about his choice.)
1887 [Bodo Moeller]
1888
1889 *) Load error codes if they are not already present instead of using a
1890 static variable. This allows them to be cleanly unloaded and reloaded.
1891 [Steve Henson]
1892
1893 Changes between 0.9.7k and 0.9.7l [28 Sep 2006]
1894
1895 *) Introduce limits to prevent malicious keys being able to
1896 cause a denial of service. (CVE-2006-2940)
1897 [Steve Henson, Bodo Moeller]
1898
1899 *) Fix ASN.1 parsing of certain invalid structures that can result
1900 in a denial of service. (CVE-2006-2937) [Steve Henson]
1901
1902 *) Fix buffer overflow in SSL_get_shared_ciphers() function.
1903 (CVE-2006-3738) [Tavis Ormandy and Will Drewry, Google Security Team]
1904
1905 *) Fix SSL client code which could crash if connecting to a
1906 malicious SSLv2 server. (CVE-2006-4343)
1907 [Tavis Ormandy and Will Drewry, Google Security Team]
1908
1909 *) Change ciphersuite string processing so that an explicit
1910 ciphersuite selects this one ciphersuite (so that "AES256-SHA"
1911 will no longer include "AES128-SHA"), and any other similar
1912 ciphersuite (same bitmap) from *other* protocol versions (so that
1913 "RC4-MD5" will still include both the SSL 2.0 ciphersuite and the
1914 SSL 3.0/TLS 1.0 ciphersuite). This is a backport combining
1915 changes from 0.9.8b and 0.9.8d.
1916 [Bodo Moeller]
1917
1918 Changes between 0.9.7j and 0.9.7k [05 Sep 2006]
1919
1920 *) Avoid PKCS #1 v1.5 signature attack discovered by Daniel Bleichenbacher
1921 (CVE-2006-4339) [Ben Laurie and Google Security Team]
1922
1923 *) Change the Unix randomness entropy gathering to use poll() when
1924 possible instead of select(), since the latter has some
1925 undesirable limitations.
1926 [Darryl Miles via Richard Levitte and Bodo Moeller]
1927
1928 *) Disable rogue ciphersuites:
1929
1930 - SSLv2 0x08 0x00 0x80 ("RC4-64-MD5")
1931 - SSLv3/TLSv1 0x00 0x61 ("EXP1024-RC2-CBC-MD5")
1932 - SSLv3/TLSv1 0x00 0x60 ("EXP1024-RC4-MD5")
1933
1934 The latter two were purportedly from
1935 draft-ietf-tls-56-bit-ciphersuites-0[01].txt, but do not really
1936 appear there.
1937
1938 Also deactive the remaining ciphersuites from
1939 draft-ietf-tls-56-bit-ciphersuites-01.txt. These are just as
1940 unofficial, and the ID has long expired.
1941 [Bodo Moeller]
1942
1943 *) Fix RSA blinding Heisenbug (problems sometimes occured on
1944 dual-core machines) and other potential thread-safety issues.
1945 [Bodo Moeller]
1946
1947 Changes between 0.9.7i and 0.9.7j [04 May 2006]
1948
1949 *) Adapt fipsld and the build system to link against the validated FIPS
1950 module in FIPS mode.
1951 [Steve Henson]
1952
1953 *) Fixes for VC++ 2005 build under Windows.
1954 [Steve Henson]
1955
1956 *) Add new Windows build target VC-32-GMAKE for VC++. This uses GNU make
1957 from a Windows bash shell such as MSYS. It is autodetected from the
1958 "config" script when run from a VC++ environment. Modify standard VC++
1959 build to use fipscanister.o from the GNU make build.
1960 [Steve Henson]
1961
1962 Changes between 0.9.7h and 0.9.7i [14 Oct 2005]
1963
1964 *) Wrapped the definition of EVP_MAX_MD_SIZE in a #ifdef OPENSSL_FIPS.
1965 The value now differs depending on if you build for FIPS or not.
1966 BEWARE! A program linked with a shared FIPSed libcrypto can't be
1967 safely run with a non-FIPSed libcrypto, as it may crash because of
1968 the difference induced by this change.
1969 [Andy Polyakov]
1970
1971 Changes between 0.9.7g and 0.9.7h [11 Oct 2005]
1972
1973 *) Remove the functionality of SSL_OP_MSIE_SSLV2_RSA_PADDING
1974 (part of SSL_OP_ALL). This option used to disable the
1975 countermeasure against man-in-the-middle protocol-version
1976 rollback in the SSL 2.0 server implementation, which is a bad
1977 idea. (CVE-2005-2969)
1978
1979 [Bodo Moeller; problem pointed out by Yutaka Oiwa (Research Center
1980 for Information Security, National Institute of Advanced Industrial
1981 Science and Technology [AIST], Japan)]
1982
1983 *) Minimal support for X9.31 signatures and PSS padding modes. This is
1984 mainly for FIPS compliance and not fully integrated at this stage.
1985 [Steve Henson]
1986
1987 *) For DSA signing, unless DSA_FLAG_NO_EXP_CONSTTIME is set, perform
1988 the exponentiation using a fixed-length exponent. (Otherwise,
1989 the information leaked through timing could expose the secret key
1990 after many signatures; cf. Bleichenbacher's attack on DSA with
1991 biased k.)
1992 [Bodo Moeller]
1993
1994 *) Make a new fixed-window mod_exp implementation the default for
1995 RSA, DSA, and DH private-key operations so that the sequence of
1996 squares and multiplies and the memory access pattern are
1997 independent of the particular secret key. This will mitigate
1998 cache-timing and potential related attacks.
1999
2000 BN_mod_exp_mont_consttime() is the new exponentiation implementation,
2001 and this is automatically used by BN_mod_exp_mont() if the new flag
2002 BN_FLG_EXP_CONSTTIME is set for the exponent. RSA, DSA, and DH
2003 will use this BN flag for private exponents unless the flag
2004 RSA_FLAG_NO_EXP_CONSTTIME, DSA_FLAG_NO_EXP_CONSTTIME, or
2005 DH_FLAG_NO_EXP_CONSTTIME, respectively, is set.
2006
2007 [Matthew D Wood (Intel Corp), with some changes by Bodo Moeller]
2008
2009 *) Change the client implementation for SSLv23_method() and
2010 SSLv23_client_method() so that is uses the SSL 3.0/TLS 1.0
2011 Client Hello message format if the SSL_OP_NO_SSLv2 option is set.
2012 (Previously, the SSL 2.0 backwards compatible Client Hello
2013 message format would be used even with SSL_OP_NO_SSLv2.)
2014 [Bodo Moeller]
2015
2016 *) Add support for smime-type MIME parameter in S/MIME messages which some
2017 clients need.
2018 [Steve Henson]
2019
2020 *) New function BN_MONT_CTX_set_locked() to set montgomery parameters in
2021 a threadsafe manner. Modify rsa code to use new function and add calls
2022 to dsa and dh code (which had race conditions before).
2023 [Steve Henson]
2024
2025 *) Include the fixed error library code in the C error file definitions
2026 instead of fixing them up at runtime. This keeps the error code
2027 structures constant.
2028 [Steve Henson]
2029
2030 Changes between 0.9.7f and 0.9.7g [11 Apr 2005]
2031
2032 [NB: OpenSSL 0.9.7h and later 0.9.7 patch levels were released after
2033 OpenSSL 0.9.8.]
2034
2035 *) Fixes for newer kerberos headers. NB: the casts are needed because
2036 the 'length' field is signed on one version and unsigned on another
2037 with no (?) obvious way to tell the difference, without these VC++
2038 complains. Also the "definition" of FAR (blank) is no longer included
2039 nor is the error ENOMEM. KRB5_PRIVATE has to be set to 1 to pick up
2040 some needed definitions.
2041 [Steve Henson]
2042
2043 *) Undo Cygwin change.
2044 [Ulf Möller]
2045
2046 *) Added support for proxy certificates according to RFC 3820.
2047 Because they may be a security thread to unaware applications,
2048 they must be explicitely allowed in run-time. See
2049 docs/HOWTO/proxy_certificates.txt for further information.
2050 [Richard Levitte]
2051
2052 Changes between 0.9.7e and 0.9.7f [22 Mar 2005]
2053
2054 *) Use (SSL_RANDOM_VALUE - 4) bytes of pseudo random data when generating
2055 server and client random values. Previously
2056 (SSL_RANDOM_VALUE - sizeof(time_t)) would be used which would result in
2057 less random data when sizeof(time_t) > 4 (some 64 bit platforms).
2058
2059 This change has negligible security impact because:
2060
2061 1. Server and client random values still have 24 bytes of pseudo random
2062 data.
2063
2064 2. Server and client random values are sent in the clear in the initial
2065 handshake.
2066
2067 3. The master secret is derived using the premaster secret (48 bytes in
2068 size for static RSA ciphersuites) as well as client server and random
2069 values.
2070
2071 The OpenSSL team would like to thank the UK NISCC for bringing this issue
2072 to our attention.
2073
2074 [Stephen Henson, reported by UK NISCC]
2075
2076 *) Use Windows randomness collection on Cygwin.
2077 [Ulf Möller]
2078
2079 *) Fix hang in EGD/PRNGD query when communication socket is closed
2080 prematurely by EGD/PRNGD.
2081 [Darren Tucker <dtucker@zip.com.au> via Lutz Jänicke, resolves #1014]
2082
2083 *) Prompt for pass phrases when appropriate for PKCS12 input format.
2084 [Steve Henson]
2085
2086 *) Back-port of selected performance improvements from development
2087 branch, as well as improved support for PowerPC platforms.
2088 [Andy Polyakov]
2089
2090 *) Add lots of checks for memory allocation failure, error codes to indicate
2091 failure and freeing up memory if a failure occurs.
2092 [Nauticus Networks SSL Team <openssl@nauticusnet.com>, Steve Henson]
2093
2094 *) Add new -passin argument to dgst.
2095 [Steve Henson]
2096
2097 *) Perform some character comparisons of different types in X509_NAME_cmp:
2098 this is needed for some certificates that reencode DNs into UTF8Strings
2099 (in violation of RFC3280) and can't or wont issue name rollover
2100 certificates.
2101 [Steve Henson]
2102
2103 *) Make an explicit check during certificate validation to see that
2104 the CA setting in each certificate on the chain is correct. As a
2105 side effect always do the following basic checks on extensions,
2106 not just when there's an associated purpose to the check:
2107
2108 - if there is an unhandled critical extension (unless the user
2109 has chosen to ignore this fault)
2110 - if the path length has been exceeded (if one is set at all)
2111 - that certain extensions fit the associated purpose (if one has
2112 been given)
2113 [Richard Levitte]
2114
2115 Changes between 0.9.7d and 0.9.7e [25 Oct 2004]
2116
2117 *) Avoid a race condition when CRLs are checked in a multi threaded
2118 environment. This would happen due to the reordering of the revoked
2119 entries during signature checking and serial number lookup. Now the
2120 encoding is cached and the serial number sort performed under a lock.
2121 Add new STACK function sk_is_sorted().
2122 [Steve Henson]
2123
2124 *) Add Delta CRL to the extension code.
2125 [Steve Henson]
2126
2127 *) Various fixes to s3_pkt.c so alerts are sent properly.
2128 [David Holmes <d.holmes@f5.com>]
2129
2130 *) Reduce the chances of duplicate issuer name and serial numbers (in
2131 violation of RFC3280) using the OpenSSL certificate creation utilities.
2132 This is done by creating a random 64 bit value for the initial serial
2133 number when a serial number file is created or when a self signed
2134 certificate is created using 'openssl req -x509'. The initial serial
2135 number file is created using 'openssl x509 -next_serial' in CA.pl
2136 rather than being initialized to 1.
2137 [Steve Henson]
2138
2139 Changes between 0.9.7c and 0.9.7d [17 Mar 2004]
2140
2141 *) Fix null-pointer assignment in do_change_cipher_spec() revealed
2142 by using the Codenomicon TLS Test Tool (CVE-2004-0079)
2143 [Joe Orton, Steve Henson]
2144
2145 *) Fix flaw in SSL/TLS handshaking when using Kerberos ciphersuites
2146 (CVE-2004-0112)
2147 [Joe Orton, Steve Henson]
2148
2149 *) Make it possible to have multiple active certificates with the same
2150 subject in the CA index file. This is done only if the keyword
2151 'unique_subject' is set to 'no' in the main CA section (default
2152 if 'CA_default') of the configuration file. The value is saved
2153 with the database itself in a separate index attribute file,
2154 named like the index file with '.attr' appended to the name.
2155 [Richard Levitte]
2156
2157 *) X509 verify fixes. Disable broken certificate workarounds when
2158 X509_V_FLAGS_X509_STRICT is set. Check CRL issuer has cRLSign set if
2159 keyUsage extension present. Don't accept CRLs with unhandled critical
2160 extensions: since verify currently doesn't process CRL extensions this
2161 rejects a CRL with *any* critical extensions. Add new verify error codes
2162 for these cases.
2163 [Steve Henson]
2164
2165 *) When creating an OCSP nonce use an OCTET STRING inside the extnValue.
2166 A clarification of RFC2560 will require the use of OCTET STRINGs and
2167 some implementations cannot handle the current raw format. Since OpenSSL
2168 copies and compares OCSP nonces as opaque blobs without any attempt at
2169 parsing them this should not create any compatibility issues.
2170 [Steve Henson]
2171
2172 *) New md flag EVP_MD_CTX_FLAG_REUSE this allows md_data to be reused when
2173 calling EVP_MD_CTX_copy_ex() to avoid calling OPENSSL_malloc(). Without
2174 this HMAC (and other) operations are several times slower than OpenSSL
2175 < 0.9.7.
2176 [Steve Henson]
2177
2178 *) Print out GeneralizedTime and UTCTime in ASN1_STRING_print_ex().
2179 [Peter Sylvester <Peter.Sylvester@EdelWeb.fr>]
2180
2181 *) Use the correct content when signing type "other".
2182 [Steve Henson]
2183
2184 Changes between 0.9.7b and 0.9.7c [30 Sep 2003]
2185
2186 *) Fix various bugs revealed by running the NISCC test suite:
2187
2188 Stop out of bounds reads in the ASN1 code when presented with
2189 invalid tags (CVE-2003-0543 and CVE-2003-0544).
2190
2191 Free up ASN1_TYPE correctly if ANY type is invalid (CVE-2003-0545).
2192
2193 If verify callback ignores invalid public key errors don't try to check
2194 certificate signature with the NULL public key.
2195
2196 [Steve Henson]
2197
2198 *) New -ignore_err option in ocsp application to stop the server
2199 exiting on the first error in a request.
2200 [Steve Henson]
2201
2202 *) In ssl3_accept() (ssl/s3_srvr.c) only accept a client certificate
2203 if the server requested one: as stated in TLS 1.0 and SSL 3.0
2204 specifications.
2205 [Steve Henson]
2206
2207 *) In ssl3_get_client_hello() (ssl/s3_srvr.c), tolerate additional
2208 extra data after the compression methods not only for TLS 1.0
2209 but also for SSL 3.0 (as required by the specification).
2210 [Bodo Moeller; problem pointed out by Matthias Loepfe]
2211
2212 *) Change X509_certificate_type() to mark the key as exported/exportable
2213 when it's 512 *bits* long, not 512 bytes.
2214 [Richard Levitte]
2215
2216 *) Change AES_cbc_encrypt() so it outputs exact multiple of
2217 blocks during encryption.
2218 [Richard Levitte]
2219
2220 *) Various fixes to base64 BIO and non blocking I/O. On write
2221 flushes were not handled properly if the BIO retried. On read
2222 data was not being buffered properly and had various logic bugs.
2223 This also affects blocking I/O when the data being decoded is a
2224 certain size.
2225 [Steve Henson]
2226
2227 *) Various S/MIME bugfixes and compatibility changes:
2228 output correct application/pkcs7 MIME type if
2229 PKCS7_NOOLDMIMETYPE is set. Tolerate some broken signatures.
2230 Output CR+LF for EOL if PKCS7_CRLFEOL is set (this makes opening
2231 of files as .eml work). Correctly handle very long lines in MIME
2232 parser.
2233 [Steve Henson]
2234
2235 Changes between 0.9.7a and 0.9.7b [10 Apr 2003]
2236
2237 *) Countermeasure against the Klima-Pokorny-Rosa extension of
2238 Bleichbacher's attack on PKCS #1 v1.5 padding: treat
2239 a protocol version number mismatch like a decryption error
2240 in ssl3_get_client_key_exchange (ssl/s3_srvr.c).
2241 [Bodo Moeller]
2242
2243 *) Turn on RSA blinding by default in the default implementation
2244 to avoid a timing attack. Applications that don't want it can call
2245 RSA_blinding_off() or use the new flag RSA_FLAG_NO_BLINDING.
2246 They would be ill-advised to do so in most cases.
2247 [Ben Laurie, Steve Henson, Geoff Thorpe, Bodo Moeller]
2248
2249 *) Change RSA blinding code so that it works when the PRNG is not
2250 seeded (in this case, the secret RSA exponent is abused as
2251 an unpredictable seed -- if it is not unpredictable, there
2252 is no point in blinding anyway). Make RSA blinding thread-safe
2253 by remembering the creator's thread ID in rsa->blinding and
2254 having all other threads use local one-time blinding factors
2255 (this requires more computation than sharing rsa->blinding, but
2256 avoids excessive locking; and if an RSA object is not shared
2257 between threads, blinding will still be very fast).
2258 [Bodo Moeller]
2259
2260 *) Fixed a typo bug that would cause ENGINE_set_default() to set an
2261 ENGINE as defaults for all supported algorithms irrespective of
2262 the 'flags' parameter. 'flags' is now honoured, so applications
2263 should make sure they are passing it correctly.
2264 [Geoff Thorpe]
2265
2266 *) Target "mingw" now allows native Windows code to be generated in
2267 the Cygwin environment as well as with the MinGW compiler.
2268 [Ulf Moeller]
2269
2270 Changes between 0.9.7 and 0.9.7a [19 Feb 2003]
2271
2272 *) In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked
2273 via timing by performing a MAC computation even if incorrrect
2274 block cipher padding has been found. This is a countermeasure
2275 against active attacks where the attacker has to distinguish
2276 between bad padding and a MAC verification error. (CVE-2003-0078)
2277
2278 [Bodo Moeller; problem pointed out by Brice Canvel (EPFL),
2279 Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and
2280 Martin Vuagnoux (EPFL, Ilion)]
2281
2282 *) Make the no-err option work as intended. The intention with no-err
2283 is not to have the whole error stack handling routines removed from
2284 libcrypto, it's only intended to remove all the function name and
2285 reason texts, thereby removing some of the footprint that may not
2286 be interesting if those errors aren't displayed anyway.
2287
2288 NOTE: it's still possible for any application or module to have it's
2289 own set of error texts inserted. The routines are there, just not
2290 used by default when no-err is given.
2291 [Richard Levitte]
2292
2293 *) Add support for FreeBSD on IA64.
2294 [dirk.meyer@dinoex.sub.org via Richard Levitte, resolves #454]
2295
2296 *) Adjust DES_cbc_cksum() so it returns the same value as the MIT
2297 Kerberos function mit_des_cbc_cksum(). Before this change,
2298 the value returned by DES_cbc_cksum() was like the one from
2299 mit_des_cbc_cksum(), except the bytes were swapped.
2300 [Kevin Greaney <Kevin.Greaney@hp.com> and Richard Levitte]
2301
2302 *) Allow an application to disable the automatic SSL chain building.
2303 Before this a rather primitive chain build was always performed in
2304 ssl3_output_cert_chain(): an application had no way to send the
2305 correct chain if the automatic operation produced an incorrect result.
2306
2307 Now the chain builder is disabled if either:
2308
2309 1. Extra certificates are added via SSL_CTX_add_extra_chain_cert().
2310
2311 2. The mode flag SSL_MODE_NO_AUTO_CHAIN is set.
2312
2313 The reasoning behind this is that an application would not want the
2314 auto chain building to take place if extra chain certificates are
2315 present and it might also want a means of sending no additional
2316 certificates (for example the chain has two certificates and the
2317 root is omitted).
2318 [Steve Henson]
2319
2320 *) Add the possibility to build without the ENGINE framework.
2321 [Steven Reddie <smr@essemer.com.au> via Richard Levitte]
2322
2323 *) Under Win32 gmtime() can return NULL: check return value in
2324 OPENSSL_gmtime(). Add error code for case where gmtime() fails.
2325 [Steve Henson]
2326
2327 *) DSA routines: under certain error conditions uninitialized BN objects
2328 could be freed. Solution: make sure initialization is performed early
2329 enough. (Reported and fix supplied by Ivan D Nestlerode <nestler@MIT.EDU>,
2330 Nils Larsch <nla@trustcenter.de> via PR#459)
2331 [Lutz Jaenicke]
2332
2333 *) Another fix for SSLv2 session ID handling: the session ID was incorrectly
2334 checked on reconnect on the client side, therefore session resumption
2335 could still fail with a "ssl session id is different" error. This
2336 behaviour is masked when SSL_OP_ALL is used due to
2337 SSL_OP_MICROSOFT_SESS_ID_BUG being set.
2338 Behaviour observed by Crispin Flowerday <crispin@flowerday.cx> as
2339 followup to PR #377.
2340 [Lutz Jaenicke]
2341
2342 *) IA-32 assembler support enhancements: unified ELF targets, support
2343 for SCO/Caldera platforms, fix for Cygwin shared build.
2344 [Andy Polyakov]
2345
2346 *) Add support for FreeBSD on sparc64. As a consequence, support for
2347 FreeBSD on non-x86 processors is separate from x86 processors on
2348 the config script, much like the NetBSD support.
2349 [Richard Levitte & Kris Kennaway <kris@obsecurity.org>]
2350
2351 Changes between 0.9.6h and 0.9.7 [31 Dec 2002]
2352
2353 [NB: OpenSSL 0.9.6i and later 0.9.6 patch levels were released after
2354 OpenSSL 0.9.7.]
2355
2356 *) Fix session ID handling in SSLv2 client code: the SERVER FINISHED
2357 code (06) was taken as the first octet of the session ID and the last
2358 octet was ignored consequently. As a result SSLv2 client side session
2359 caching could not have worked due to the session ID mismatch between
2360 client and server.
2361 Behaviour observed by Crispin Flowerday <crispin@flowerday.cx> as
2362 PR #377.
2363 [Lutz Jaenicke]
2364
2365 *) Change the declaration of needed Kerberos libraries to use EX_LIBS
2366 instead of the special (and badly supported) LIBKRB5. LIBKRB5 is
2367 removed entirely.
2368 [Richard Levitte]
2369
2370 *) The hw_ncipher.c engine requires dynamic locks. Unfortunately, it
2371 seems that in spite of existing for more than a year, many application
2372 author have done nothing to provide the necessary callbacks, which
2373 means that this particular engine will not work properly anywhere.
2374 This is a very unfortunate situation which forces us, in the name
2375 of usability, to give the hw_ncipher.c a static lock, which is part
2376 of libcrypto.
2377 NOTE: This is for the 0.9.7 series ONLY. This hack will never
2378 appear in 0.9.8 or later. We EXPECT application authors to have
2379 dealt properly with this when 0.9.8 is released (unless we actually
2380 make such changes in the libcrypto locking code that changes will
2381 have to be made anyway).
2382 [Richard Levitte]
2383
2384 *) In asn1_d2i_read_bio() repeatedly call BIO_read() until all content
2385 octets have been read, EOF or an error occurs. Without this change
2386 some truncated ASN1 structures will not produce an error.
2387 [Steve Henson]
2388
2389 *) Disable Heimdal support, since it hasn't been fully implemented.
2390 Still give the possibility to force the use of Heimdal, but with
2391 warnings and a request that patches get sent to openssl-dev.
2392 [Richard Levitte]
2393
2394 *) Add the VC-CE target, introduce the WINCE sysname, and add
2395 INSTALL.WCE and appropriate conditionals to make it build.
2396 [Steven Reddie <smr@essemer.com.au> via Richard Levitte]
2397
2398 *) Change the DLL names for Cygwin to cygcrypto-x.y.z.dll and
2399 cygssl-x.y.z.dll, where x, y and z are the major, minor and
2400 edit numbers of the version.
2401 [Corinna Vinschen <vinschen@redhat.com> and Richard Levitte]
2402
2403 *) Introduce safe string copy and catenation functions
2404 (BUF_strlcpy() and BUF_strlcat()).
2405 [Ben Laurie (CHATS) and Richard Levitte]
2406
2407 *) Avoid using fixed-size buffers for one-line DNs.
2408 [Ben Laurie (CHATS)]
2409
2410 *) Add BUF_MEM_grow_clean() to avoid information leakage when
2411 resizing buffers containing secrets, and use where appropriate.
2412 [Ben Laurie (CHATS)]
2413
2414 *) Avoid using fixed size buffers for configuration file location.
2415 [Ben Laurie (CHATS)]
2416
2417 *) Avoid filename truncation for various CA files.
2418 [Ben Laurie (CHATS)]
2419
2420 *) Use sizeof in preference to magic numbers.
2421 [Ben Laurie (CHATS)]
2422
2423 *) Avoid filename truncation in cert requests.
2424 [Ben Laurie (CHATS)]
2425
2426 *) Add assertions to check for (supposedly impossible) buffer
2427 overflows.
2428 [Ben Laurie (CHATS)]
2429
2430 *) Don't cache truncated DNS entries in the local cache (this could
2431 potentially lead to a spoofing attack).
2432 [Ben Laurie (CHATS)]
2433
2434 *) Fix various buffers to be large enough for hex/decimal
2435 representations in a platform independent manner.
2436 [Ben Laurie (CHATS)]
2437
2438 *) Add CRYPTO_realloc_clean() to avoid information leakage when
2439 resizing buffers containing secrets, and use where appropriate.
2440 [Ben Laurie (CHATS)]
2441
2442 *) Add BIO_indent() to avoid much slightly worrying code to do
2443 indents.
2444 [Ben Laurie (CHATS)]
2445
2446 *) Convert sprintf()/BIO_puts() to BIO_printf().
2447 [Ben Laurie (CHATS)]
2448
2449 *) buffer_gets() could terminate with the buffer only half
2450 full. Fixed.
2451 [Ben Laurie (CHATS)]
2452
2453 *) Add assertions to prevent user-supplied crypto functions from
2454 overflowing internal buffers by having large block sizes, etc.
2455 [Ben Laurie (CHATS)]
2456
2457 *) New OPENSSL_assert() macro (similar to assert(), but enabled
2458 unconditionally).
2459 [Ben Laurie (CHATS)]
2460
2461 *) Eliminate unused copy of key in RC4.
2462 [Ben Laurie (CHATS)]
2463
2464 *) Eliminate unused and incorrectly sized buffers for IV in pem.h.
2465 [Ben Laurie (CHATS)]
2466
2467 *) Fix off-by-one error in EGD path.
2468 [Ben Laurie (CHATS)]
2469
2470 *) If RANDFILE path is too long, ignore instead of truncating.
2471 [Ben Laurie (CHATS)]
2472
2473 *) Eliminate unused and incorrectly sized X.509 structure
2474 CBCParameter.
2475 [Ben Laurie (CHATS)]
2476
2477 *) Eliminate unused and dangerous function knumber().
2478 [Ben Laurie (CHATS)]
2479
2480 *) Eliminate unused and dangerous structure, KSSL_ERR.
2481 [Ben Laurie (CHATS)]
2482
2483 *) Protect against overlong session ID context length in an encoded
2484 session object. Since these are local, this does not appear to be
2485 exploitable.
2486 [Ben Laurie (CHATS)]
2487
2488 *) Change from security patch (see 0.9.6e below) that did not affect
2489 the 0.9.6 release series:
2490
2491 Remote buffer overflow in SSL3 protocol - an attacker could
2492 supply an oversized master key in Kerberos-enabled versions.
2493 (CVE-2002-0657)
2494 [Ben Laurie (CHATS)]
2495
2496 *) Change the SSL kerb5 codes to match RFC 2712.
2497 [Richard Levitte]
2498
2499 *) Make -nameopt work fully for req and add -reqopt switch.
2500 [Michael Bell <michael.bell@rz.hu-berlin.de>, Steve Henson]
2501
2502 *) The "block size" for block ciphers in CFB and OFB mode should be 1.
2503 [Steve Henson, reported by Yngve Nysaeter Pettersen <yngve@opera.com>]
2504
2505 *) Make sure tests can be performed even if the corresponding algorithms
2506 have been removed entirely. This was also the last step to make
2507 OpenSSL compilable with DJGPP under all reasonable conditions.
2508 [Richard Levitte, Doug Kaufman <dkaufman@rahul.net>]
2509
2510 *) Add cipher selection rules COMPLEMENTOFALL and COMPLEMENTOFDEFAULT
2511 to allow version independent disabling of normally unselected ciphers,
2512 which may be activated as a side-effect of selecting a single cipher.
2513
2514 (E.g., cipher list string "RSA" enables ciphersuites that are left
2515 out of "ALL" because they do not provide symmetric encryption.
2516 "RSA:!COMPLEMEMENTOFALL" avoids these unsafe ciphersuites.)
2517 [Lutz Jaenicke, Bodo Moeller]
2518
2519 *) Add appropriate support for separate platform-dependent build
2520 directories. The recommended way to make a platform-dependent
2521 build directory is the following (tested on Linux), maybe with
2522 some local tweaks:
2523
2524 # Place yourself outside of the OpenSSL source tree. In
2525 # this example, the environment variable OPENSSL_SOURCE
2526 # is assumed to contain the absolute OpenSSL source directory.
2527 mkdir -p objtree/"`uname -s`-`uname -r`-`uname -m`"
2528 cd objtree/"`uname -s`-`uname -r`-`uname -m`"
2529 (cd $OPENSSL_SOURCE; find . -type f) | while read F; do
2530 mkdir -p `dirname $F`
2531 ln -s $OPENSSL_SOURCE/$F $F
2532 done
2533
2534 To be absolutely sure not to disturb the source tree, a "make clean"
2535 is a good thing. If it isn't successfull, don't worry about it,
2536 it probably means the source directory is very clean.
2537 [Richard Levitte]
2538
2539 *) Make sure any ENGINE control commands make local copies of string
2540 pointers passed to them whenever necessary. Otherwise it is possible
2541 the caller may have overwritten (or deallocated) the original string
2542 data when a later ENGINE operation tries to use the stored values.
2543 [Götz Babin-Ebell <babinebell@trustcenter.de>]
2544
2545 *) Improve diagnostics in file reading and command-line digests.
2546 [Ben Laurie aided and abetted by Solar Designer <solar@openwall.com>]
2547
2548 *) Add AES modes CFB and OFB to the object database. Correct an
2549 error in AES-CFB decryption.
2550 [Richard Levitte]
2551
2552 *) Remove most calls to EVP_CIPHER_CTX_cleanup() in evp_enc.c, this
2553 allows existing EVP_CIPHER_CTX structures to be reused after
2554 calling EVP_*Final(). This behaviour is used by encryption
2555 BIOs and some applications. This has the side effect that
2556 applications must explicitly clean up cipher contexts with
2557 EVP_CIPHER_CTX_cleanup() or they will leak memory.
2558 [Steve Henson]
2559
2560 *) Check the values of dna and dnb in bn_mul_recursive before calling
2561 bn_mul_comba (a non zero value means the a or b arrays do not contain
2562 n2 elements) and fallback to bn_mul_normal if either is not zero.
2563 [Steve Henson]
2564
2565 *) Fix escaping of non-ASCII characters when using the -subj option
2566 of the "openssl req" command line tool. (Robert Joop <joop@fokus.gmd.de>)
2567 [Lutz Jaenicke]
2568
2569 *) Make object definitions compliant to LDAP (RFC2256): SN is the short
2570 form for "surname", serialNumber has no short form.
2571 Use "mail" as the short name for "rfc822Mailbox" according to RFC2798;
2572 therefore remove "mail" short name for "internet 7".
2573 The OID for unique identifiers in X509 certificates is
2574 x500UniqueIdentifier, not uniqueIdentifier.
2575 Some more OID additions. (Michael Bell <michael.bell@rz.hu-berlin.de>)
2576 [Lutz Jaenicke]
2577
2578 *) Add an "init" command to the ENGINE config module and auto initialize
2579 ENGINEs. Without any "init" command the ENGINE will be initialized
2580 after all ctrl commands have been executed on it. If init=1 the
2581 ENGINE is initailized at that point (ctrls before that point are run
2582 on the uninitialized ENGINE and after on the initialized one). If
2583 init=0 then the ENGINE will not be iniatialized at all.
2584 [Steve Henson]
2585
2586 *) Fix the 'app_verify_callback' interface so that the user-defined
2587 argument is actually passed to the callback: In the
2588 SSL_CTX_set_cert_verify_callback() prototype, the callback
2589 declaration has been changed from
2590 int (*cb)()
2591 into
2592 int (*cb)(X509_STORE_CTX *,void *);
2593 in ssl_verify_cert_chain (ssl/ssl_cert.c), the call
2594 i=s->ctx->app_verify_callback(&ctx)
2595 has been changed into
2596 i=s->ctx->app_verify_callback(&ctx, s->ctx->app_verify_arg).
2597
2598 To update applications using SSL_CTX_set_cert_verify_callback(),
2599 a dummy argument can be added to their callback functions.
2600 [D. K. Smetters <smetters@parc.xerox.com>]
2601
2602 *) Added the '4758cca' ENGINE to support IBM 4758 cards.
2603 [Maurice Gittens <maurice@gittens.nl>, touchups by Geoff Thorpe]
2604
2605 *) Add and OPENSSL_LOAD_CONF define which will cause
2606 OpenSSL_add_all_algorithms() to load the openssl.cnf config file.
2607 This allows older applications to transparently support certain
2608 OpenSSL features: such as crypto acceleration and dynamic ENGINE loading.
2609 Two new functions OPENSSL_add_all_algorithms_noconf() which will never
2610 load the config file and OPENSSL_add_all_algorithms_conf() which will
2611 always load it have also been added.
2612 [Steve Henson]
2613
2614 *) Add the OFB, CFB and CTR (all with 128 bit feedback) to AES.
2615 Adjust NIDs and EVP layer.
2616 [Stephen Sprunk <stephen@sprunk.org> and Richard Levitte]
2617
2618 *) Config modules support in openssl utility.
2619
2620 Most commands now load modules from the config file,
2621 though in a few (such as version) this isn't done
2622 because it couldn't be used for anything.
2623
2624 In the case of ca and req the config file used is
2625 the same as the utility itself: that is the -config
2626 command line option can be used to specify an
2627 alternative file.
2628 [Steve Henson]
2629
2630 *) Move default behaviour from OPENSSL_config(). If appname is NULL
2631 use "openssl_conf" if filename is NULL use default openssl config file.
2632 [Steve Henson]
2633
2634 *) Add an argument to OPENSSL_config() to allow the use of an alternative
2635 config section name. Add a new flag to tolerate a missing config file
2636 and move code to CONF_modules_load_file().
2637 [Steve Henson]
2638
2639 *) Support for crypto accelerator cards from Accelerated Encryption
2640 Processing, www.aep.ie. (Use engine 'aep')
2641 The support was copied from 0.9.6c [engine] and adapted/corrected
2642 to work with the new engine framework.
2643 [AEP Inc. and Richard Levitte]
2644
2645 *) Support for SureWare crypto accelerator cards from Baltimore
2646 Technologies. (Use engine 'sureware')
2647 The support was copied from 0.9.6c [engine] and adapted
2648 to work with the new engine framework.
2649 [Richard Levitte]
2650
2651 *) Have the CHIL engine fork-safe (as defined by nCipher) and actually
2652 make the newer ENGINE framework commands for the CHIL engine work.
2653 [Toomas Kiisk <vix@cyber.ee> and Richard Levitte]
2654
2655 *) Make it possible to produce shared libraries on ReliantUNIX.
2656 [Robert Dahlem <Robert.Dahlem@ffm2.siemens.de> via Richard Levitte]
2657
2658 *) Add the configuration target debug-linux-ppro.
2659 Make 'openssl rsa' use the general key loading routines
2660 implemented in apps.c, and make those routines able to
2661 handle the key format FORMAT_NETSCAPE and the variant
2662 FORMAT_IISSGC.
2663 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte]
2664
2665 *) Fix a crashbug and a logic bug in hwcrhk_load_pubkey().
2666 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte]
2667
2668 *) Add -keyform to rsautl, and document -engine.
2669 [Richard Levitte, inspired by Toomas Kiisk <vix@cyber.ee>]
2670
2671 *) Change BIO_new_file (crypto/bio/bss_file.c) to use new
2672 BIO_R_NO_SUCH_FILE error code rather than the generic
2673 ERR_R_SYS_LIB error code if fopen() fails with ENOENT.
2674 [Ben Laurie]
2675
2676 *) Add new functions
2677 ERR_peek_last_error
2678 ERR_peek_last_error_line
2679 ERR_peek_last_error_line_data.
2680 These are similar to
2681 ERR_peek_error
2682 ERR_peek_error_line
2683 ERR_peek_error_line_data,
2684 but report on the latest error recorded rather than the first one
2685 still in the error queue.
2686 [Ben Laurie, Bodo Moeller]
2687
2688 *) default_algorithms option in ENGINE config module. This allows things
2689 like:
2690 default_algorithms = ALL
2691 default_algorithms = RSA, DSA, RAND, CIPHERS, DIGESTS
2692 [Steve Henson]
2693
2694 *) Prelminary ENGINE config module.
2695 [Steve Henson]
2696
2697 *) New experimental application configuration code.
2698 [Steve Henson]
2699
2700 *) Change the AES code to follow the same name structure as all other
2701 symmetric ciphers, and behave the same way. Move everything to
2702 the directory crypto/aes, thereby obsoleting crypto/rijndael.
2703 [Stephen Sprunk <stephen@sprunk.org> and Richard Levitte]
2704
2705 *) SECURITY: remove unsafe setjmp/signal interaction from ui_openssl.c.
2706 [Ben Laurie and Theo de Raadt]
2707
2708 *) Add option to output public keys in req command.
2709 [Massimiliano Pala madwolf@openca.org]
2710
2711 *) Use wNAFs in EC_POINTs_mul() for improved efficiency
2712 (up to about 10% better than before for P-192 and P-224).
2713 [Bodo Moeller]
2714
2715 *) New functions/macros
2716
2717 SSL_CTX_set_msg_callback(ctx, cb)
2718 SSL_CTX_set_msg_callback_arg(ctx, arg)
2719 SSL_set_msg_callback(ssl, cb)
2720 SSL_set_msg_callback_arg(ssl, arg)
2721
2722 to request calling a callback function
2723
2724 void cb(int write_p, int version, int content_type,
2725 const void *buf, size_t len, SSL *ssl, void *arg)
2726
2727 whenever a protocol message has been completely received
2728 (write_p == 0) or sent (write_p == 1). Here 'version' is the
2729 protocol version according to which the SSL library interprets
2730 the current protocol message (SSL2_VERSION, SSL3_VERSION, or
2731 TLS1_VERSION). 'content_type' is 0 in the case of SSL 2.0, or
2732 the content type as defined in the SSL 3.0/TLS 1.0 protocol
2733 specification (change_cipher_spec(20), alert(21), handshake(22)).
2734 'buf' and 'len' point to the actual message, 'ssl' to the
2735 SSL object, and 'arg' is the application-defined value set by
2736 SSL[_CTX]_set_msg_callback_arg().
2737
2738 'openssl s_client' and 'openssl s_server' have new '-msg' options
2739 to enable a callback that displays all protocol messages.
2740 [Bodo Moeller]
2741
2742 *) Change the shared library support so shared libraries are built as
2743 soon as the corresponding static library is finished, and thereby get
2744 openssl and the test programs linked against the shared library.
2745 This still only happens when the keyword "shard" has been given to
2746 the configuration scripts.
2747
2748 NOTE: shared library support is still an experimental thing, and
2749 backward binary compatibility is still not guaranteed.
2750 ["Maciej W. Rozycki" <macro@ds2.pg.gda.pl> and Richard Levitte]
2751
2752 *) Add support for Subject Information Access extension.
2753 [Peter Sylvester <Peter.Sylvester@EdelWeb.fr>]
2754
2755 *) Make BUF_MEM_grow() behaviour more consistent: Initialise to zero
2756 additional bytes when new memory had to be allocated, not just
2757 when reusing an existing buffer.
2758 [Bodo Moeller]
2759
2760 *) New command line and configuration option 'utf8' for the req command.
2761 This allows field values to be specified as UTF8 strings.
2762 [Steve Henson]
2763
2764 *) Add -multi and -mr options to "openssl speed" - giving multiple parallel
2765 runs for the former and machine-readable output for the latter.
2766 [Ben Laurie]
2767
2768 *) Add '-noemailDN' option to 'openssl ca'. This prevents inclusion
2769 of the e-mail address in the DN (i.e., it will go into a certificate
2770 extension only). The new configuration file option 'email_in_dn = no'
2771 has the same effect.
2772 [Massimiliano Pala madwolf@openca.org]
2773
2774 *) Change all functions with names starting with des_ to be starting
2775 with DES_ instead. Add wrappers that are compatible with libdes,
2776 but are named _ossl_old_des_*. Finally, add macros that map the
2777 des_* symbols to the corresponding _ossl_old_des_* if libdes
2778 compatibility is desired. If OpenSSL 0.9.6c compatibility is
2779 desired, the des_* symbols will be mapped to DES_*, with one
2780 exception.
2781
2782 Since we provide two compatibility mappings, the user needs to
2783 define the macro OPENSSL_DES_LIBDES_COMPATIBILITY if libdes
2784 compatibility is desired. The default (i.e., when that macro
2785 isn't defined) is OpenSSL 0.9.6c compatibility.
2786
2787 There are also macros that enable and disable the support of old
2788 des functions altogether. Those are OPENSSL_ENABLE_OLD_DES_SUPPORT
2789 and OPENSSL_DISABLE_OLD_DES_SUPPORT. If none or both of those
2790 are defined, the default will apply: to support the old des routines.
2791
2792 In either case, one must include openssl/des.h to get the correct
2793 definitions. Do not try to just include openssl/des_old.h, that
2794 won't work.
2795
2796 NOTE: This is a major break of an old API into a new one. Software
2797 authors are encouraged to switch to the DES_ style functions. Some
2798 time in the future, des_old.h and the libdes compatibility functions
2799 will be disable (i.e. OPENSSL_DISABLE_OLD_DES_SUPPORT will be the
2800 default), and then completely removed.
2801 [Richard Levitte]
2802
2803 *) Test for certificates which contain unsupported critical extensions.
2804 If such a certificate is found during a verify operation it is
2805 rejected by default: this behaviour can be overridden by either
2806 handling the new error X509_V_ERR_UNHANDLED_CRITICAL_EXTENSION or
2807 by setting the verify flag X509_V_FLAG_IGNORE_CRITICAL. A new function
2808 X509_supported_extension() has also been added which returns 1 if a
2809 particular extension is supported.
2810 [Steve Henson]
2811
2812 *) Modify the behaviour of EVP cipher functions in similar way to digests
2813 to retain compatibility with existing code.
2814 [Steve Henson]
2815
2816 *) Modify the behaviour of EVP_DigestInit() and EVP_DigestFinal() to retain
2817 compatibility with existing code. In particular the 'ctx' parameter does
2818 not have to be to be initialized before the call to EVP_DigestInit() and
2819 it is tidied up after a call to EVP_DigestFinal(). New function
2820 EVP_DigestFinal_ex() which does not tidy up the ctx. Similarly function
2821 EVP_MD_CTX_copy() changed to not require the destination to be
2822 initialized valid and new function EVP_MD_CTX_copy_ex() added which
2823 requires the destination to be valid.
2824
2825 Modify all the OpenSSL digest calls to use EVP_DigestInit_ex(),
2826 EVP_DigestFinal_ex() and EVP_MD_CTX_copy_ex().
2827 [Steve Henson]
2828
2829 *) Change ssl3_get_message (ssl/s3_both.c) and the functions using it
2830 so that complete 'Handshake' protocol structures are kept in memory
2831 instead of overwriting 'msg_type' and 'length' with 'body' data.
2832 [Bodo Moeller]
2833
2834 *) Add an implementation of SSL_add_dir_cert_subjects_to_stack for Win32.
2835 [Massimo Santin via Richard Levitte]
2836
2837 *) Major restructuring to the underlying ENGINE code. This includes
2838 reduction of linker bloat, separation of pure "ENGINE" manipulation
2839 (initialisation, etc) from functionality dealing with implementations
2840 of specific crypto iterfaces. This change also introduces integrated
2841 support for symmetric ciphers and digest implementations - so ENGINEs
2842 can now accelerate these by providing EVP_CIPHER and EVP_MD
2843 implementations of their own. This is detailed in crypto/engine/README
2844 as it couldn't be adequately described here. However, there are a few
2845 API changes worth noting - some RSA, DSA, DH, and RAND functions that
2846 were changed in the original introduction of ENGINE code have now
2847 reverted back - the hooking from this code to ENGINE is now a good
2848 deal more passive and at run-time, operations deal directly with
2849 RSA_METHODs, DSA_METHODs (etc) as they did before, rather than
2850 dereferencing through an ENGINE pointer any more. Also, the ENGINE
2851 functions dealing with BN_MOD_EXP[_CRT] handlers have been removed -
2852 they were not being used by the framework as there is no concept of a
2853 BIGNUM_METHOD and they could not be generalised to the new
2854 'ENGINE_TABLE' mechanism that underlies the new code. Similarly,
2855 ENGINE_cpy() has been removed as it cannot be consistently defined in
2856 the new code.
2857 [Geoff Thorpe]
2858
2859 *) Change ASN1_GENERALIZEDTIME_check() to allow fractional seconds.
2860 [Steve Henson]
2861
2862 *) Change mkdef.pl to sort symbols that get the same entry number,
2863 and make sure the automatically generated functions ERR_load_*
2864 become part of libeay.num as well.
2865 [Richard Levitte]
2866
2867 *) New function SSL_renegotiate_pending(). This returns true once
2868 renegotiation has been requested (either SSL_renegotiate() call
2869 or HelloRequest/ClientHello receveived from the peer) and becomes
2870 false once a handshake has been completed.
2871 (For servers, SSL_renegotiate() followed by SSL_do_handshake()
2872 sends a HelloRequest, but does not ensure that a handshake takes
2873 place. SSL_renegotiate_pending() is useful for checking if the
2874 client has followed the request.)
2875 [Bodo Moeller]
2876
2877 *) New SSL option SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION.
2878 By default, clients may request session resumption even during
2879 renegotiation (if session ID contexts permit); with this option,
2880 session resumption is possible only in the first handshake.
2881
2882 SSL_OP_ALL is now 0x00000FFFL instead of 0x000FFFFFL. This makes
2883 more bits available for options that should not be part of
2884 SSL_OP_ALL (such as SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION).
2885 [Bodo Moeller]
2886
2887 *) Add some demos for certificate and certificate request creation.
2888 [Steve Henson]
2889
2890 *) Make maximum certificate chain size accepted from the peer application
2891 settable (SSL*_get/set_max_cert_list()), as proposed by
2892 "Douglas E. Engert" <deengert@anl.gov>.
2893 [Lutz Jaenicke]
2894
2895 *) Add support for shared libraries for Unixware-7
2896 (Boyd Lynn Gerber <gerberb@zenez.com>).
2897 [Lutz Jaenicke]
2898
2899 *) Add a "destroy" handler to ENGINEs that allows structural cleanup to
2900 be done prior to destruction. Use this to unload error strings from
2901 ENGINEs that load their own error strings. NB: This adds two new API
2902 functions to "get" and "set" this destroy handler in an ENGINE.
2903 [Geoff Thorpe]
2904
2905 *) Alter all existing ENGINE implementations (except "openssl" and
2906 "openbsd") to dynamically instantiate their own error strings. This
2907 makes them more flexible to be built both as statically-linked ENGINEs
2908 and self-contained shared-libraries loadable via the "dynamic" ENGINE.
2909 Also, add stub code to each that makes building them as self-contained
2910 shared-libraries easier (see README.ENGINE).
2911 [Geoff Thorpe]
2912
2913 *) Add a "dynamic" ENGINE that provides a mechanism for binding ENGINE
2914 implementations into applications that are completely implemented in
2915 self-contained shared-libraries. The "dynamic" ENGINE exposes control
2916 commands that can be used to configure what shared-library to load and
2917 to control aspects of the way it is handled. Also, made an update to
2918 the README.ENGINE file that brings its information up-to-date and
2919 provides some information and instructions on the "dynamic" ENGINE
2920 (ie. how to use it, how to build "dynamic"-loadable ENGINEs, etc).
2921 [Geoff Thorpe]
2922
2923 *) Make it possible to unload ranges of ERR strings with a new
2924 "ERR_unload_strings" function.
2925 [Geoff Thorpe]
2926
2927 *) Add a copy() function to EVP_MD.
2928 [Ben Laurie]
2929
2930 *) Make EVP_MD routines take a context pointer instead of just the
2931 md_data void pointer.
2932 [Ben Laurie]
2933
2934 *) Add flags to EVP_MD and EVP_MD_CTX. EVP_MD_FLAG_ONESHOT indicates
2935 that the digest can only process a single chunk of data
2936 (typically because it is provided by a piece of
2937 hardware). EVP_MD_CTX_FLAG_ONESHOT indicates that the application
2938 is only going to provide a single chunk of data, and hence the
2939 framework needn't accumulate the data for oneshot drivers.
2940 [Ben Laurie]
2941
2942 *) As with "ERR", make it possible to replace the underlying "ex_data"
2943 functions. This change also alters the storage and management of global
2944 ex_data state - it's now all inside ex_data.c and all "class" code (eg.
2945 RSA, BIO, SSL_CTX, etc) no longer stores its own STACKS and per-class
2946 index counters. The API functions that use this state have been changed
2947 to take a "class_index" rather than pointers to the class's local STACK
2948 and counter, and there is now an API function to dynamically create new
2949 classes. This centralisation allows us to (a) plug a lot of the
2950 thread-safety problems that existed, and (b) makes it possible to clean
2951 up all allocated state using "CRYPTO_cleanup_all_ex_data()". W.r.t. (b)
2952 such data would previously have always leaked in application code and
2953 workarounds were in place to make the memory debugging turn a blind eye
2954 to it. Application code that doesn't use this new function will still
2955 leak as before, but their memory debugging output will announce it now
2956 rather than letting it slide.
2957
2958 Besides the addition of CRYPTO_cleanup_all_ex_data(), another API change
2959 induced by the "ex_data" overhaul is that X509_STORE_CTX_init() now
2960 has a return value to indicate success or failure.
2961 [Geoff Thorpe]
2962
2963 *) Make it possible to replace the underlying "ERR" functions such that the
2964 global state (2 LHASH tables and 2 locks) is only used by the "default"
2965 implementation. This change also adds two functions to "get" and "set"
2966 the implementation prior to it being automatically set the first time
2967 any other ERR function takes place. Ie. an application can call "get",
2968 pass the return value to a module it has just loaded, and that module
2969 can call its own "set" function using that value. This means the
2970 module's "ERR" operations will use (and modify) the error state in the
2971 application and not in its own statically linked copy of OpenSSL code.
2972 [Geoff Thorpe]
2973
2974 *) Give DH, DSA, and RSA types their own "**_up_ref()" function to increment
2975 reference counts. This performs normal REF_PRINT/REF_CHECK macros on
2976 the operation, and provides a more encapsulated way for external code
2977 (crypto/evp/ and ssl/) to do this. Also changed the evp and ssl code
2978 to use these functions rather than manually incrementing the counts.
2979
2980 Also rename "DSO_up()" function to more descriptive "DSO_up_ref()".
2981 [Geoff Thorpe]
2982
2983 *) Add EVP test program.
2984 [Ben Laurie]
2985
2986 *) Add symmetric cipher support to ENGINE. Expect the API to change!
2987 [Ben Laurie]
2988
2989 *) New CRL functions: X509_CRL_set_version(), X509_CRL_set_issuer_name()
2990 X509_CRL_set_lastUpdate(), X509_CRL_set_nextUpdate(), X509_CRL_sort(),
2991 X509_REVOKED_set_serialNumber(), and X509_REVOKED_set_revocationDate().
2992 These allow a CRL to be built without having to access X509_CRL fields
2993 directly. Modify 'ca' application to use new functions.
2994 [Steve Henson]
2995
2996 *) Move SSL_OP_TLS_ROLLBACK_BUG out of the SSL_OP_ALL list of recommended
2997 bug workarounds. Rollback attack detection is a security feature.
2998 The problem will only arise on OpenSSL servers when TLSv1 is not
2999 available (sslv3_server_method() or SSL_OP_NO_TLSv1).
3000 Software authors not wanting to support TLSv1 will have special reasons
3001 for their choice and can explicitly enable this option.
3002 [Bodo Moeller, Lutz Jaenicke]
3003
3004 *) Rationalise EVP so it can be extended: don't include a union of
3005 cipher/digest structures, add init/cleanup functions for EVP_MD_CTX
3006 (similar to those existing for EVP_CIPHER_CTX).
3007 Usage example:
3008
3009 EVP_MD_CTX md;
3010
3011 EVP_MD_CTX_init(&md); /* new function call */
3012 EVP_DigestInit(&md, EVP_sha1());
3013 EVP_DigestUpdate(&md, in, len);
3014 EVP_DigestFinal(&md, out, NULL);
3015 EVP_MD_CTX_cleanup(&md); /* new function call */
3016
3017 [Ben Laurie]
3018
3019 *) Make DES key schedule conform to the usual scheme, as well as
3020 correcting its structure. This means that calls to DES functions
3021 now have to pass a pointer to a des_key_schedule instead of a
3022 plain des_key_schedule (which was actually always a pointer
3023 anyway): E.g.,
3024
3025 des_key_schedule ks;
3026
3027 des_set_key_checked(..., &ks);
3028 des_ncbc_encrypt(..., &ks, ...);
3029
3030 (Note that a later change renames 'des_...' into 'DES_...'.)
3031 [Ben Laurie]
3032
3033 *) Initial reduction of linker bloat: the use of some functions, such as
3034 PEM causes large amounts of unused functions to be linked in due to
3035 poor organisation. For example pem_all.c contains every PEM function
3036 which has a knock on effect of linking in large amounts of (unused)
3037 ASN1 code. Grouping together similar functions and splitting unrelated
3038 functions prevents this.
3039 [Steve Henson]
3040
3041 *) Cleanup of EVP macros.
3042 [Ben Laurie]
3043
3044 *) Change historical references to {NID,SN,LN}_des_ede and ede3 to add the
3045 correct _ecb suffix.
3046 [Ben Laurie]
3047
3048 *) Add initial OCSP responder support to ocsp application. The
3049 revocation information is handled using the text based index
3050 use by the ca application. The responder can either handle
3051 requests generated internally, supplied in files (for example
3052 via a CGI script) or using an internal minimal server.
3053 [Steve Henson]
3054
3055 *) Add configuration choices to get zlib compression for TLS.
3056 [Richard Levitte]
3057
3058 *) Changes to Kerberos SSL for RFC 2712 compliance:
3059 1. Implemented real KerberosWrapper, instead of just using
3060 KRB5 AP_REQ message. [Thanks to Simon Wilkinson <sxw@sxw.org.uk>]
3061 2. Implemented optional authenticator field of KerberosWrapper.
3062
3063 Added openssl-style ASN.1 macros for Kerberos ticket, ap_req,
3064 and authenticator structs; see crypto/krb5/.
3065
3066 Generalized Kerberos calls to support multiple Kerberos libraries.
3067 [Vern Staats <staatsvr@asc.hpc.mil>,
3068 Jeffrey Altman <jaltman@columbia.edu>
3069 via Richard Levitte]
3070
3071 *) Cause 'openssl speed' to use fully hard-coded DSA keys as it
3072 already does with RSA. testdsa.h now has 'priv_key/pub_key'
3073 values for each of the key sizes rather than having just
3074 parameters (and 'speed' generating keys each time).
3075 [Geoff Thorpe]
3076
3077 *) Speed up EVP routines.
3078 Before:
3079 encrypt
3080 type 8 bytes 64 bytes 256 bytes 1024 bytes 8192 bytes
3081 des-cbc 4408.85k 5560.51k 5778.46k 5862.20k 5825.16k
3082 des-cbc 4389.55k 5571.17k 5792.23k 5846.91k 5832.11k
3083 des-cbc 4394.32k 5575.92k 5807.44k 5848.37k 5841.30k
3084 decrypt
3085 des-cbc 3482.66k 5069.49k 5496.39k 5614.16k 5639.28k
3086 des-cbc 3480.74k 5068.76k 5510.34k 5609.87k 5635.52k
3087 des-cbc 3483.72k 5067.62k 5504.60k 5708.01k 5724.80k
3088 After:
3089 encrypt
3090 des-cbc 4660.16k 5650.19k 5807.19k 5827.13k 5783.32k
3091 decrypt
3092 des-cbc 3624.96k 5258.21k 5530.91k 5624.30k 5628.26k
3093 [Ben Laurie]
3094
3095 *) Added the OS2-EMX target.
3096 ["Brian Havard" <brianh@kheldar.apana.org.au> and Richard Levitte]
3097
3098 *) Rewrite apps to use NCONF routines instead of the old CONF. New functions
3099 to support NCONF routines in extension code. New function CONF_set_nconf()
3100 to allow functions which take an NCONF to also handle the old LHASH
3101 structure: this means that the old CONF compatible routines can be
3102 retained (in particular wrt extensions) without having to duplicate the
3103 code. New function X509V3_add_ext_nconf_sk to add extensions to a stack.
3104 [Steve Henson]
3105
3106 *) Enhance the general user interface with mechanisms for inner control
3107 and with possibilities to have yes/no kind of prompts.
3108 [Richard Levitte]
3109
3110 *) Change all calls to low level digest routines in the library and
3111 applications to use EVP. Add missing calls to HMAC_cleanup() and
3112 don't assume HMAC_CTX can be copied using memcpy().
3113 [Verdon Walker <VWalker@novell.com>, Steve Henson]
3114
3115 *) Add the possibility to control engines through control names but with
3116 arbitrary arguments instead of just a string.
3117 Change the key loaders to take a UI_METHOD instead of a callback
3118 function pointer. NOTE: this breaks binary compatibility with earlier
3119 versions of OpenSSL [engine].
3120 Adapt the nCipher code for these new conditions and add a card insertion
3121 callback.
3122 [Richard Levitte]
3123
3124 *) Enhance the general user interface with mechanisms to better support
3125 dialog box interfaces, application-defined prompts, the possibility
3126 to use defaults (for example default passwords from somewhere else)
3127 and interrupts/cancellations.
3128 [Richard Levitte]
3129
3130 *) Tidy up PKCS#12 attribute handling. Add support for the CSP name
3131 attribute in PKCS#12 files, add new -CSP option to pkcs12 utility.
3132 [Steve Henson]
3133
3134 *) Fix a memory leak in 'sk_dup()' in the case reallocation fails. (Also
3135 tidy up some unnecessarily weird code in 'sk_new()').
3136 [Geoff, reported by Diego Tartara <dtartara@novamens.com>]
3137
3138 *) Change the key loading routines for ENGINEs to use the same kind
3139 callback (pem_password_cb) as all other routines that need this
3140 kind of callback.
3141 [Richard Levitte]
3142
3143 *) Increase ENTROPY_NEEDED to 32 bytes, as Rijndael can operate with
3144 256 bit (=32 byte) keys. Of course seeding with more entropy bytes
3145 than this minimum value is recommended.
3146 [Lutz Jaenicke]
3147
3148 *) New random seeder for OpenVMS, using the system process statistics
3149 that are easily reachable.
3150 [Richard Levitte]
3151
3152 *) Windows apparently can't transparently handle global
3153 variables defined in DLLs. Initialisations such as:
3154
3155 const ASN1_ITEM *it = &ASN1_INTEGER_it;
3156
3157 wont compile. This is used by the any applications that need to
3158 declare their own ASN1 modules. This was fixed by adding the option
3159 EXPORT_VAR_AS_FN to all Win32 platforms, although this isn't strictly
3160 needed for static libraries under Win32.
3161 [Steve Henson]
3162
3163 *) New functions X509_PURPOSE_set() and X509_TRUST_set() to handle
3164 setting of purpose and trust fields. New X509_STORE trust and
3165 purpose functions and tidy up setting in other SSL functions.
3166 [Steve Henson]
3167
3168 *) Add copies of X509_STORE_CTX fields and callbacks to X509_STORE
3169 structure. These are inherited by X509_STORE_CTX when it is
3170 initialised. This allows various defaults to be set in the
3171 X509_STORE structure (such as flags for CRL checking and custom
3172 purpose or trust settings) for functions which only use X509_STORE_CTX
3173 internally such as S/MIME.
3174
3175 Modify X509_STORE_CTX_purpose_inherit() so it only sets purposes and
3176 trust settings if they are not set in X509_STORE. This allows X509_STORE
3177 purposes and trust (in S/MIME for example) to override any set by default.
3178
3179 Add command line options for CRL checking to smime, s_client and s_server
3180 applications.
3181 [Steve Henson]
3182
3183 *) Initial CRL based revocation checking. If the CRL checking flag(s)
3184 are set then the CRL is looked up in the X509_STORE structure and
3185 its validity and signature checked, then if the certificate is found
3186 in the CRL the verify fails with a revoked error.
3187
3188 Various new CRL related callbacks added to X509_STORE_CTX structure.
3189
3190 Command line options added to 'verify' application to support this.
3191
3192 This needs some additional work, such as being able to handle multiple
3193 CRLs with different times, extension based lookup (rather than just
3194 by subject name) and ultimately more complete V2 CRL extension
3195 handling.
3196 [Steve Henson]
3197
3198 *) Add a general user interface API (crypto/ui/). This is designed
3199 to replace things like des_read_password and friends (backward
3200 compatibility functions using this new API are provided).
3201 The purpose is to remove prompting functions from the DES code
3202 section as well as provide for prompting through dialog boxes in
3203 a window system and the like.
3204 [Richard Levitte]
3205
3206 *) Add "ex_data" support to ENGINE so implementations can add state at a
3207 per-structure level rather than having to store it globally.
3208 [Geoff]
3209
3210 *) Make it possible for ENGINE structures to be copied when retrieved by
3211 ENGINE_by_id() if the ENGINE specifies a new flag: ENGINE_FLAGS_BY_ID_COPY.
3212 This causes the "original" ENGINE structure to act like a template,
3213 analogous to the RSA vs. RSA_METHOD type of separation. Because of this
3214 operational state can be localised to each ENGINE structure, despite the
3215 fact they all share the same "methods". New ENGINE structures returned in
3216 this case have no functional references and the return value is the single
3217 structural reference. This matches the single structural reference returned
3218 by ENGINE_by_id() normally, when it is incremented on the pre-existing
3219 ENGINE structure.
3220 [Geoff]
3221
3222 *) Fix ASN1 decoder when decoding type ANY and V_ASN1_OTHER: since this
3223 needs to match any other type at all we need to manually clear the
3224 tag cache.
3225 [Steve Henson]
3226
3227 *) Changes to the "openssl engine" utility to include;
3228 - verbosity levels ('-v', '-vv', and '-vvv') that provide information
3229 about an ENGINE's available control commands.
3230 - executing control commands from command line arguments using the
3231 '-pre' and '-post' switches. '-post' is only used if '-t' is
3232 specified and the ENGINE is successfully initialised. The syntax for
3233 the individual commands are colon-separated, for example;
3234 openssl engine chil -pre FORK_CHECK:0 -pre SO_PATH:/lib/test.so
3235 [Geoff]
3236
3237 *) New dynamic control command support for ENGINEs. ENGINEs can now
3238 declare their own commands (numbers), names (strings), descriptions,
3239 and input types for run-time discovery by calling applications. A
3240 subset of these commands are implicitly classed as "executable"
3241 depending on their input type, and only these can be invoked through
3242 the new string-based API function ENGINE_ctrl_cmd_string(). (Eg. this
3243 can be based on user input, config files, etc). The distinction is
3244 that "executable" commands cannot return anything other than a boolean
3245 result and can only support numeric or string input, whereas some
3246 discoverable commands may only be for direct use through
3247 ENGINE_ctrl(), eg. supporting the exchange of binary data, function
3248 pointers, or other custom uses. The "executable" commands are to
3249 support parameterisations of ENGINE behaviour that can be
3250 unambiguously defined by ENGINEs and used consistently across any
3251 OpenSSL-based application. Commands have been added to all the
3252 existing hardware-supporting ENGINEs, noticeably "SO_PATH" to allow
3253 control over shared-library paths without source code alterations.
3254 [Geoff]
3255
3256 *) Changed all ENGINE implementations to dynamically allocate their
3257 ENGINEs rather than declaring them statically. Apart from this being
3258 necessary with the removal of the ENGINE_FLAGS_MALLOCED distinction,
3259 this also allows the implementations to compile without using the
3260 internal engine_int.h header.
3261 [Geoff]
3262
3263 *) Minor adjustment to "rand" code. RAND_get_rand_method() now returns a
3264 'const' value. Any code that should be able to modify a RAND_METHOD
3265 should already have non-const pointers to it (ie. they should only
3266 modify their own ones).
3267 [Geoff]
3268
3269 *) Made a variety of little tweaks to the ENGINE code.
3270 - "atalla" and "ubsec" string definitions were moved from header files
3271 to C code. "nuron" string definitions were placed in variables
3272 rather than hard-coded - allowing parameterisation of these values
3273 later on via ctrl() commands.
3274 - Removed unused "#if 0"'d code.
3275 - Fixed engine list iteration code so it uses ENGINE_free() to release
3276 structural references.
3277 - Constified the RAND_METHOD element of ENGINE structures.
3278 - Constified various get/set functions as appropriate and added
3279 missing functions (including a catch-all ENGINE_cpy that duplicates
3280 all ENGINE values onto a new ENGINE except reference counts/state).
3281 - Removed NULL parameter checks in get/set functions. Setting a method
3282 or function to NULL is a way of cancelling out a previously set
3283 value. Passing a NULL ENGINE parameter is just plain stupid anyway
3284 and doesn't justify the extra error symbols and code.
3285 - Deprecate the ENGINE_FLAGS_MALLOCED define and move the area for
3286 flags from engine_int.h to engine.h.
3287 - Changed prototypes for ENGINE handler functions (init(), finish(),
3288 ctrl(), key-load functions, etc) to take an (ENGINE*) parameter.
3289 [Geoff]
3290
3291 *) Implement binary inversion algorithm for BN_mod_inverse in addition
3292 to the algorithm using long division. The binary algorithm can be
3293 used only if the modulus is odd. On 32-bit systems, it is faster
3294 only for relatively small moduli (roughly 20-30% for 128-bit moduli,
3295 roughly 5-15% for 256-bit moduli), so we use it only for moduli
3296 up to 450 bits. In 64-bit environments, the binary algorithm
3297 appears to be advantageous for much longer moduli; here we use it
3298 for moduli up to 2048 bits.
3299 [Bodo Moeller]
3300
3301 *) Rewrite CHOICE field setting in ASN1_item_ex_d2i(). The old code
3302 could not support the combine flag in choice fields.
3303 [Steve Henson]
3304
3305 *) Add a 'copy_extensions' option to the 'ca' utility. This copies
3306 extensions from a certificate request to the certificate.
3307 [Steve Henson]
3308
3309 *) Allow multiple 'certopt' and 'nameopt' options to be separated
3310 by commas. Add 'namopt' and 'certopt' options to the 'ca' config
3311 file: this allows the display of the certificate about to be
3312 signed to be customised, to allow certain fields to be included
3313 or excluded and extension details. The old system didn't display
3314 multicharacter strings properly, omitted fields not in the policy
3315 and couldn't display additional details such as extensions.
3316 [Steve Henson]
3317
3318 *) Function EC_POINTs_mul for multiple scalar multiplication
3319 of an arbitrary number of elliptic curve points
3320 \sum scalars[i]*points[i],
3321 optionally including the generator defined for the EC_GROUP:
3322 scalar*generator + \sum scalars[i]*points[i].
3323
3324 EC_POINT_mul is a simple wrapper function for the typical case
3325 that the point list has just one item (besides the optional
3326 generator).
3327 [Bodo Moeller]
3328
3329 *) First EC_METHODs for curves over GF(p):
3330
3331 EC_GFp_simple_method() uses the basic BN_mod_mul and BN_mod_sqr
3332 operations and provides various method functions that can also
3333 operate with faster implementations of modular arithmetic.
3334
3335 EC_GFp_mont_method() reuses most functions that are part of
3336 EC_GFp_simple_method, but uses Montgomery arithmetic.
3337
3338 [Bodo Moeller; point addition and point doubling
3339 implementation directly derived from source code provided by
3340 Lenka Fibikova <fibikova@exp-math.uni-essen.de>]
3341
3342 *) Framework for elliptic curves (crypto/ec/ec.h, crypto/ec/ec_lcl.h,
3343 crypto/ec/ec_lib.c):
3344
3345 Curves are EC_GROUP objects (with an optional group generator)
3346 based on EC_METHODs that are built into the library.
3347
3348 Points are EC_POINT objects based on EC_GROUP objects.
3349
3350 Most of the framework would be able to handle curves over arbitrary
3351 finite fields, but as there are no obvious types for fields other
3352 than GF(p), some functions are limited to that for now.
3353 [Bodo Moeller]
3354
3355 *) Add the -HTTP option to s_server. It is similar to -WWW, but requires
3356 that the file contains a complete HTTP response.
3357 [Richard Levitte]
3358
3359 *) Add the ec directory to mkdef.pl and mkfiles.pl. In mkdef.pl
3360 change the def and num file printf format specifier from "%-40sXXX"
3361 to "%-39s XXX". The latter will always guarantee a space after the
3362 field while the former will cause them to run together if the field
3363 is 40 of more characters long.
3364 [Steve Henson]
3365
3366 *) Constify the cipher and digest 'method' functions and structures
3367 and modify related functions to take constant EVP_MD and EVP_CIPHER
3368 pointers.
3369 [Steve Henson]
3370
3371 *) Hide BN_CTX structure details in bn_lcl.h instead of publishing them
3372 in <openssl/bn.h>. Also further increase BN_CTX_NUM to 32.
3373 [Bodo Moeller]
3374
3375 *) Modify EVP_Digest*() routines so they now return values. Although the
3376 internal software routines can never fail additional hardware versions
3377 might.
3378 [Steve Henson]
3379
3380 *) Clean up crypto/err/err.h and change some error codes to avoid conflicts:
3381
3382 Previously ERR_R_FATAL was too small and coincided with ERR_LIB_PKCS7
3383 (= ERR_R_PKCS7_LIB); it is now 64 instead of 32.
3384
3385 ASN1 error codes
3386 ERR_R_NESTED_ASN1_ERROR
3387 ...
3388 ERR_R_MISSING_ASN1_EOS
3389 were 4 .. 9, conflicting with
3390 ERR_LIB_RSA (= ERR_R_RSA_LIB)
3391 ...
3392 ERR_LIB_PEM (= ERR_R_PEM_LIB).
3393 They are now 58 .. 63 (i.e., just below ERR_R_FATAL).
3394
3395 Add new error code 'ERR_R_INTERNAL_ERROR'.
3396 [Bodo Moeller]
3397
3398 *) Don't overuse locks in crypto/err/err.c: For data retrieval, CRYPTO_r_lock
3399 suffices.
3400 [Bodo Moeller]
3401
3402 *) New option '-subj arg' for 'openssl req' and 'openssl ca'. This
3403 sets the subject name for a new request or supersedes the
3404 subject name in a given request. Formats that can be parsed are
3405 'CN=Some Name, OU=myOU, C=IT'
3406 and
3407 'CN=Some Name/OU=myOU/C=IT'.
3408
3409 Add options '-batch' and '-verbose' to 'openssl req'.
3410 [Massimiliano Pala <madwolf@hackmasters.net>]
3411
3412 *) Introduce the possibility to access global variables through
3413 functions on platform were that's the best way to handle exporting
3414 global variables in shared libraries. To enable this functionality,
3415 one must configure with "EXPORT_VAR_AS_FN" or defined the C macro
3416 "OPENSSL_EXPORT_VAR_AS_FUNCTION" in crypto/opensslconf.h (the latter
3417 is normally done by Configure or something similar).
3418
3419 To implement a global variable, use the macro OPENSSL_IMPLEMENT_GLOBAL
3420 in the source file (foo.c) like this:
3421
3422 OPENSSL_IMPLEMENT_GLOBAL(int,foo)=1;
3423 OPENSSL_IMPLEMENT_GLOBAL(double,bar);
3424
3425 To declare a global variable, use the macros OPENSSL_DECLARE_GLOBAL
3426 and OPENSSL_GLOBAL_REF in the header file (foo.h) like this:
3427
3428 OPENSSL_DECLARE_GLOBAL(int,foo);
3429 #define foo OPENSSL_GLOBAL_REF(foo)
3430 OPENSSL_DECLARE_GLOBAL(double,bar);
3431 #define bar OPENSSL_GLOBAL_REF(bar)
3432
3433 The #defines are very important, and therefore so is including the
3434 header file everywhere where the defined globals are used.
3435
3436 The macro OPENSSL_EXPORT_VAR_AS_FUNCTION also affects the definition
3437 of ASN.1 items, but that structure is a bit different.
3438
3439 The largest change is in util/mkdef.pl which has been enhanced with
3440 better and easier to understand logic to choose which symbols should
3441 go into the Windows .def files as well as a number of fixes and code
3442 cleanup (among others, algorithm keywords are now sorted
3443 lexicographically to avoid constant rewrites).
3444 [Richard Levitte]
3445
3446 *) In BN_div() keep a copy of the sign of 'num' before writing the
3447 result to 'rm' because if rm==num the value will be overwritten
3448 and produce the wrong result if 'num' is negative: this caused
3449 problems with BN_mod() and BN_nnmod().
3450 [Steve Henson]
3451
3452 *) Function OCSP_request_verify(). This checks the signature on an
3453 OCSP request and verifies the signer certificate. The signer
3454 certificate is just checked for a generic purpose and OCSP request
3455 trust settings.
3456 [Steve Henson]
3457
3458 *) Add OCSP_check_validity() function to check the validity of OCSP
3459 responses. OCSP responses are prepared in real time and may only
3460 be a few seconds old. Simply checking that the current time lies
3461 between thisUpdate and nextUpdate max reject otherwise valid responses
3462 caused by either OCSP responder or client clock inaccuracy. Instead
3463 we allow thisUpdate and nextUpdate to fall within a certain period of
3464 the current time. The age of the response can also optionally be
3465 checked. Two new options -validity_period and -status_age added to
3466 ocsp utility.
3467 [Steve Henson]
3468
3469 *) If signature or public key algorithm is unrecognized print out its
3470 OID rather that just UNKNOWN.
3471 [Steve Henson]
3472
3473 *) Change OCSP_cert_to_id() to tolerate a NULL subject certificate and
3474 OCSP_cert_id_new() a NULL serialNumber. This allows a partial certificate
3475 ID to be generated from the issuer certificate alone which can then be
3476 passed to OCSP_id_issuer_cmp().
3477 [Steve Henson]
3478
3479 *) New compilation option ASN1_ITEM_FUNCTIONS. This causes the new
3480 ASN1 modules to export functions returning ASN1_ITEM pointers
3481 instead of the ASN1_ITEM structures themselves. This adds several
3482 new macros which allow the underlying ASN1 function/structure to
3483 be accessed transparently. As a result code should not use ASN1_ITEM
3484 references directly (such as &X509_it) but instead use the relevant
3485 macros (such as ASN1_ITEM_rptr(X509)). This option is to allow
3486 use of the new ASN1 code on platforms where exporting structures
3487 is problematical (for example in shared libraries) but exporting
3488 functions returning pointers to structures is not.
3489 [Steve Henson]
3490
3491 *) Add support for overriding the generation of SSL/TLS session IDs.
3492 These callbacks can be registered either in an SSL_CTX or per SSL.
3493 The purpose of this is to allow applications to control, if they wish,
3494 the arbitrary values chosen for use as session IDs, particularly as it
3495 can be useful for session caching in multiple-server environments. A
3496 command-line switch for testing this (and any client code that wishes
3497 to use such a feature) has been added to "s_server".
3498 [Geoff Thorpe, Lutz Jaenicke]
3499
3500 *) Modify mkdef.pl to recognise and parse preprocessor conditionals
3501 of the form '#if defined(...) || defined(...) || ...' and
3502 '#if !defined(...) && !defined(...) && ...'. This also avoids
3503 the growing number of special cases it was previously handling.
3504 [Richard Levitte]
3505
3506 *) Make all configuration macros available for application by making
3507 sure they are available in opensslconf.h, by giving them names starting
3508 with "OPENSSL_" to avoid conflicts with other packages and by making
3509 sure e_os2.h will cover all platform-specific cases together with
3510 opensslconf.h.
3511 Additionally, it is now possible to define configuration/platform-
3512 specific names (called "system identities"). In the C code, these
3513 are prefixed with "OPENSSL_SYSNAME_". e_os2.h will create another
3514 macro with the name beginning with "OPENSSL_SYS_", which is determined
3515 from "OPENSSL_SYSNAME_*" or compiler-specific macros depending on
3516 what is available.
3517 [Richard Levitte]
3518
3519 *) New option -set_serial to 'req' and 'x509' this allows the serial
3520 number to use to be specified on the command line. Previously self
3521 signed certificates were hard coded with serial number 0 and the
3522 CA options of 'x509' had to use a serial number in a file which was
3523 auto incremented.
3524 [Steve Henson]
3525
3526 *) New options to 'ca' utility to support V2 CRL entry extensions.
3527 Currently CRL reason, invalidity date and hold instruction are
3528 supported. Add new CRL extensions to V3 code and some new objects.
3529 [Steve Henson]
3530
3531 *) New function EVP_CIPHER_CTX_set_padding() this is used to
3532 disable standard block padding (aka PKCS#5 padding) in the EVP
3533 API, which was previously mandatory. This means that the data is
3534 not padded in any way and so the total length much be a multiple
3535 of the block size, otherwise an error occurs.
3536 [Steve Henson]
3537
3538 *) Initial (incomplete) OCSP SSL support.
3539 [Steve Henson]
3540
3541 *) New function OCSP_parse_url(). This splits up a URL into its host,
3542 port and path components: primarily to parse OCSP URLs. New -url
3543 option to ocsp utility.
3544 [Steve Henson]
3545
3546 *) New nonce behavior. The return value of OCSP_check_nonce() now
3547 reflects the various checks performed. Applications can decide
3548 whether to tolerate certain situations such as an absent nonce
3549 in a response when one was present in a request: the ocsp application
3550 just prints out a warning. New function OCSP_add1_basic_nonce()
3551 this is to allow responders to include a nonce in a response even if
3552 the request is nonce-less.
3553 [Steve Henson]
3554
3555 *) Disable stdin buffering in load_cert (apps/apps.c) so that no certs are
3556 skipped when using openssl x509 multiple times on a single input file,
3557 e.g. "(openssl x509 -out cert1; openssl x509 -out cert2) <certs".
3558 [Bodo Moeller]
3559
3560 *) Make ASN1_UTCTIME_set_string() and ASN1_GENERALIZEDTIME_set_string()
3561 set string type: to handle setting ASN1_TIME structures. Fix ca
3562 utility to correctly initialize revocation date of CRLs.
3563 [Steve Henson]
3564
3565 *) New option SSL_OP_CIPHER_SERVER_PREFERENCE allows the server to override
3566 the clients preferred ciphersuites and rather use its own preferences.
3567 Should help to work around M$ SGC (Server Gated Cryptography) bug in
3568 Internet Explorer by ensuring unchanged hash method during stepup.
3569 (Also replaces the broken/deactivated SSL_OP_NON_EXPORT_FIRST option.)
3570 [Lutz Jaenicke]
3571
3572 *) Make mkdef.pl recognise all DECLARE_ASN1 macros, change rijndael
3573 to aes and add a new 'exist' option to print out symbols that don't
3574 appear to exist.
3575 [Steve Henson]
3576
3577 *) Additional options to ocsp utility to allow flags to be set and
3578 additional certificates supplied.
3579 [Steve Henson]
3580
3581 *) Add the option -VAfile to 'openssl ocsp', so the user can give the
3582 OCSP client a number of certificate to only verify the response
3583 signature against.
3584 [Richard Levitte]
3585
3586 *) Update Rijndael code to version 3.0 and change EVP AES ciphers to
3587 handle the new API. Currently only ECB, CBC modes supported. Add new
3588 AES OIDs.
3589
3590 Add TLS AES ciphersuites as described in RFC3268, "Advanced
3591 Encryption Standard (AES) Ciphersuites for Transport Layer
3592 Security (TLS)". (In beta versions of OpenSSL 0.9.7, these were
3593 not enabled by default and were not part of the "ALL" ciphersuite
3594 alias because they were not yet official; they could be
3595 explicitly requested by specifying the "AESdraft" ciphersuite
3596 group alias. In the final release of OpenSSL 0.9.7, the group
3597 alias is called "AES" and is part of "ALL".)
3598 [Ben Laurie, Steve Henson, Bodo Moeller]
3599
3600 *) New function OCSP_copy_nonce() to copy nonce value (if present) from
3601 request to response.
3602 [Steve Henson]
3603
3604 *) Functions for OCSP responders. OCSP_request_onereq_count(),
3605 OCSP_request_onereq_get0(), OCSP_onereq_get0_id() and OCSP_id_get0_info()
3606 extract information from a certificate request. OCSP_response_create()
3607 creates a response and optionally adds a basic response structure.
3608 OCSP_basic_add1_status() adds a complete single response to a basic
3609 response and returns the OCSP_SINGLERESP structure just added (to allow
3610 extensions to be included for example). OCSP_basic_add1_cert() adds a
3611 certificate to a basic response and OCSP_basic_sign() signs a basic
3612 response with various flags. New helper functions ASN1_TIME_check()
3613 (checks validity of ASN1_TIME structure) and ASN1_TIME_to_generalizedtime()
3614 (converts ASN1_TIME to GeneralizedTime).
3615 [Steve Henson]
3616
3617 *) Various new functions. EVP_Digest() combines EVP_Digest{Init,Update,Final}()
3618 in a single operation. X509_get0_pubkey_bitstr() extracts the public_key
3619 structure from a certificate. X509_pubkey_digest() digests the public_key
3620 contents: this is used in various key identifiers.
3621 [Steve Henson]
3622
3623 *) Make sk_sort() tolerate a NULL argument.
3624 [Steve Henson reported by Massimiliano Pala <madwolf@comune.modena.it>]
3625
3626 *) New OCSP verify flag OCSP_TRUSTOTHER. When set the "other" certificates
3627 passed by the function are trusted implicitly. If any of them signed the
3628 response then it is assumed to be valid and is not verified.
3629 [Steve Henson]
3630
3631 *) In PKCS7_set_type() initialise content_type in PKCS7_ENC_CONTENT
3632 to data. This was previously part of the PKCS7 ASN1 code. This
3633 was causing problems with OpenSSL created PKCS#12 and PKCS#7 structures.
3634 [Steve Henson, reported by Kenneth R. Robinette
3635 <support@securenetterm.com>]
3636
3637 *) Add CRYPTO_push_info() and CRYPTO_pop_info() calls to new ASN1
3638 routines: without these tracing memory leaks is very painful.
3639 Fix leaks in PKCS12 and PKCS7 routines.
3640 [Steve Henson]
3641
3642 *) Make X509_time_adj() cope with the new behaviour of ASN1_TIME_new().
3643 Previously it initialised the 'type' argument to V_ASN1_UTCTIME which
3644 effectively meant GeneralizedTime would never be used. Now it
3645 is initialised to -1 but X509_time_adj() now has to check the value
3646 and use ASN1_TIME_set() if the value is not V_ASN1_UTCTIME or
3647 V_ASN1_GENERALIZEDTIME, without this it always uses GeneralizedTime.
3648 [Steve Henson, reported by Kenneth R. Robinette
3649 <support@securenetterm.com>]
3650
3651 *) Fixes to BN_to_ASN1_INTEGER when bn is zero. This would previously
3652 result in a zero length in the ASN1_INTEGER structure which was
3653 not consistent with the structure when d2i_ASN1_INTEGER() was used
3654 and would cause ASN1_INTEGER_cmp() to fail. Enhance s2i_ASN1_INTEGER()
3655 to cope with hex and negative integers. Fix bug in i2a_ASN1_INTEGER()
3656 where it did not print out a minus for negative ASN1_INTEGER.
3657 [Steve Henson]
3658
3659 *) Add summary printout to ocsp utility. The various functions which
3660 convert status values to strings have been renamed to:
3661 OCSP_response_status_str(), OCSP_cert_status_str() and
3662 OCSP_crl_reason_str() and are no longer static. New options
3663 to verify nonce values and to disable verification. OCSP response
3664 printout format cleaned up.
3665 [Steve Henson]
3666
3667 *) Add additional OCSP certificate checks. These are those specified
3668 in RFC2560. This consists of two separate checks: the CA of the
3669 certificate being checked must either be the OCSP signer certificate
3670 or the issuer of the OCSP signer certificate. In the latter case the
3671 OCSP signer certificate must contain the OCSP signing extended key
3672 usage. This check is performed by attempting to match the OCSP
3673 signer or the OCSP signer CA to the issuerNameHash and issuerKeyHash
3674 in the OCSP_CERTID structures of the response.
3675 [Steve Henson]
3676
3677 *) Initial OCSP certificate verification added to OCSP_basic_verify()
3678 and related routines. This uses the standard OpenSSL certificate
3679 verify routines to perform initial checks (just CA validity) and
3680 to obtain the certificate chain. Then additional checks will be
3681 performed on the chain. Currently the root CA is checked to see
3682 if it is explicitly trusted for OCSP signing. This is used to set
3683 a root CA as a global signing root: that is any certificate that
3684 chains to that CA is an acceptable OCSP signing certificate.
3685 [Steve Henson]
3686
3687 *) New '-extfile ...' option to 'openssl ca' for reading X.509v3
3688 extensions from a separate configuration file.
3689 As when reading extensions from the main configuration file,
3690 the '-extensions ...' option may be used for specifying the
3691 section to use.
3692 [Massimiliano Pala <madwolf@comune.modena.it>]
3693
3694 *) New OCSP utility. Allows OCSP requests to be generated or
3695 read. The request can be sent to a responder and the output
3696 parsed, outputed or printed in text form. Not complete yet:
3697 still needs to check the OCSP response validity.
3698 [Steve Henson]
3699
3700 *) New subcommands for 'openssl ca':
3701 'openssl ca -status <serial>' prints the status of the cert with
3702 the given serial number (according to the index file).
3703 'openssl ca -updatedb' updates the expiry status of certificates
3704 in the index file.
3705 [Massimiliano Pala <madwolf@comune.modena.it>]
3706
3707 *) New '-newreq-nodes' command option to CA.pl. This is like
3708 '-newreq', but calls 'openssl req' with the '-nodes' option
3709 so that the resulting key is not encrypted.
3710 [Damien Miller <djm@mindrot.org>]
3711
3712 *) New configuration for the GNU Hurd.
3713 [Jonathan Bartlett <johnnyb@wolfram.com> via Richard Levitte]
3714
3715 *) Initial code to implement OCSP basic response verify. This
3716 is currently incomplete. Currently just finds the signer's
3717 certificate and verifies the signature on the response.
3718 [Steve Henson]
3719
3720 *) New SSLeay_version code SSLEAY_DIR to determine the compiled-in
3721 value of OPENSSLDIR. This is available via the new '-d' option
3722 to 'openssl version', and is also included in 'openssl version -a'.
3723 [Bodo Moeller]
3724
3725 *) Allowing defining memory allocation callbacks that will be given
3726 file name and line number information in additional arguments
3727 (a const char* and an int). The basic functionality remains, as
3728 well as the original possibility to just replace malloc(),
3729 realloc() and free() by functions that do not know about these
3730 additional arguments. To register and find out the current
3731 settings for extended allocation functions, the following
3732 functions are provided:
3733
3734 CRYPTO_set_mem_ex_functions
3735 CRYPTO_set_locked_mem_ex_functions
3736 CRYPTO_get_mem_ex_functions
3737 CRYPTO_get_locked_mem_ex_functions
3738
3739 These work the same way as CRYPTO_set_mem_functions and friends.
3740 CRYPTO_get_[locked_]mem_functions now writes 0 where such an
3741 extended allocation function is enabled.
3742 Similarly, CRYPTO_get_[locked_]mem_ex_functions writes 0 where
3743 a conventional allocation function is enabled.
3744 [Richard Levitte, Bodo Moeller]
3745
3746 *) Finish off removing the remaining LHASH function pointer casts.
3747 There should no longer be any prototype-casting required when using
3748 the LHASH abstraction, and any casts that remain are "bugs". See
3749 the callback types and macros at the head of lhash.h for details
3750 (and "OBJ_cleanup" in crypto/objects/obj_dat.c as an example).
3751 [Geoff Thorpe]
3752
3753 *) Add automatic query of EGD sockets in RAND_poll() for the unix variant.
3754 If /dev/[u]random devices are not available or do not return enough
3755 entropy, EGD style sockets (served by EGD or PRNGD) will automatically
3756 be queried.
3757 The locations /var/run/egd-pool, /dev/egd-pool, /etc/egd-pool, and
3758 /etc/entropy will be queried once each in this sequence, quering stops
3759 when enough entropy was collected without querying more sockets.
3760 [Lutz Jaenicke]
3761
3762 *) Change the Unix RAND_poll() variant to be able to poll several
3763 random devices, as specified by DEVRANDOM, until a sufficient amount
3764 of data has been collected. We spend at most 10 ms on each file
3765 (select timeout) and read in non-blocking mode. DEVRANDOM now
3766 defaults to the list "/dev/urandom", "/dev/random", "/dev/srandom"
3767 (previously it was just the string "/dev/urandom"), so on typical
3768 platforms the 10 ms delay will never occur.
3769 Also separate out the Unix variant to its own file, rand_unix.c.
3770 For VMS, there's a currently-empty rand_vms.c.
3771 [Richard Levitte]
3772
3773 *) Move OCSP client related routines to ocsp_cl.c. These
3774 provide utility functions which an application needing
3775 to issue a request to an OCSP responder and analyse the
3776 response will typically need: as opposed to those which an
3777 OCSP responder itself would need which will be added later.
3778
3779 OCSP_request_sign() signs an OCSP request with an API similar
3780 to PKCS7_sign(). OCSP_response_status() returns status of OCSP
3781 response. OCSP_response_get1_basic() extracts basic response
3782 from response. OCSP_resp_find_status(): finds and extracts status
3783 information from an OCSP_CERTID structure (which will be created
3784 when the request structure is built). These are built from lower
3785 level functions which work on OCSP_SINGLERESP structures but
3786 wont normally be used unless the application wishes to examine
3787 extensions in the OCSP response for example.
3788
3789 Replace nonce routines with a pair of functions.
3790 OCSP_request_add1_nonce() adds a nonce value and optionally
3791 generates a random value. OCSP_check_nonce() checks the
3792 validity of the nonce in an OCSP response.
3793 [Steve Henson]
3794
3795 *) Change function OCSP_request_add() to OCSP_request_add0_id().
3796 This doesn't copy the supplied OCSP_CERTID and avoids the
3797 need to free up the newly created id. Change return type
3798 to OCSP_ONEREQ to return the internal OCSP_ONEREQ structure.
3799 This can then be used to add extensions to the request.
3800 Deleted OCSP_request_new(), since most of its functionality
3801 is now in OCSP_REQUEST_new() (and the case insensitive name
3802 clash) apart from the ability to set the request name which
3803 will be added elsewhere.
3804 [Steve Henson]
3805
3806 *) Update OCSP API. Remove obsolete extensions argument from
3807 various functions. Extensions are now handled using the new
3808 OCSP extension code. New simple OCSP HTTP function which
3809 can be used to send requests and parse the response.
3810 [Steve Henson]
3811
3812 *) Fix the PKCS#7 (S/MIME) code to work with new ASN1. Two new
3813 ASN1_ITEM structures help with sign and verify. PKCS7_ATTR_SIGN
3814 uses the special reorder version of SET OF to sort the attributes
3815 and reorder them to match the encoded order. This resolves a long
3816 standing problem: a verify on a PKCS7 structure just after signing
3817 it used to fail because the attribute order did not match the
3818 encoded order. PKCS7_ATTR_VERIFY does not reorder the attributes:
3819 it uses the received order. This is necessary to tolerate some broken
3820 software that does not order SET OF. This is handled by encoding
3821 as a SEQUENCE OF but using implicit tagging (with UNIVERSAL class)
3822 to produce the required SET OF.
3823 [Steve Henson]
3824
3825 *) Have mk1mf.pl generate the macros OPENSSL_BUILD_SHLIBCRYPTO and
3826 OPENSSL_BUILD_SHLIBSSL and use them appropriately in the header
3827 files to get correct declarations of the ASN.1 item variables.
3828 [Richard Levitte]
3829
3830 *) Rewrite of PKCS#12 code to use new ASN1 functionality. Replace many
3831 PKCS#12 macros with real functions. Fix two unrelated ASN1 bugs:
3832 asn1_check_tlen() would sometimes attempt to use 'ctx' when it was
3833 NULL and ASN1_TYPE was not dereferenced properly in asn1_ex_c2i().
3834 New ASN1 macro: DECLARE_ASN1_ITEM() which just declares the relevant
3835 ASN1_ITEM and no wrapper functions.
3836 [Steve Henson]
3837
3838 *) New functions or ASN1_item_d2i_fp() and ASN1_item_d2i_bio(). These
3839 replace the old function pointer based I/O routines. Change most of
3840 the *_d2i_bio() and *_d2i_fp() functions to use these.
3841 [Steve Henson]
3842
3843 *) Enhance mkdef.pl to be more accepting about spacing in C preprocessor
3844 lines, recognice more "algorithms" that can be deselected, and make
3845 it complain about algorithm deselection that isn't recognised.
3846 [Richard Levitte]
3847
3848 *) New ASN1 functions to handle dup, sign, verify, digest, pack and
3849 unpack operations in terms of ASN1_ITEM. Modify existing wrappers
3850 to use new functions. Add NO_ASN1_OLD which can be set to remove
3851 some old style ASN1 functions: this can be used to determine if old
3852 code will still work when these eventually go away.
3853 [Steve Henson]
3854
3855 *) New extension functions for OCSP structures, these follow the
3856 same conventions as certificates and CRLs.
3857 [Steve Henson]
3858
3859 *) New function X509V3_add1_i2d(). This automatically encodes and
3860 adds an extension. Its behaviour can be customised with various
3861 flags to append, replace or delete. Various wrappers added for
3862 certifcates and CRLs.
3863 [Steve Henson]
3864
3865 *) Fix to avoid calling the underlying ASN1 print routine when
3866 an extension cannot be parsed. Correct a typo in the
3867 OCSP_SERVICELOC extension. Tidy up print OCSP format.
3868 [Steve Henson]
3869
3870 *) Make mkdef.pl parse some of the ASN1 macros and add apropriate
3871 entries for variables.
3872 [Steve Henson]
3873
3874 *) Add functionality to apps/openssl.c for detecting locking
3875 problems: As the program is single-threaded, all we have
3876 to do is register a locking callback using an array for
3877 storing which locks are currently held by the program.
3878 [Bodo Moeller]
3879
3880 *) Use a lock around the call to CRYPTO_get_ex_new_index() in
3881 SSL_get_ex_data_X509_STORE_idx(), which is used in
3882 ssl_verify_cert_chain() and thus can be called at any time
3883 during TLS/SSL handshakes so that thread-safety is essential.
3884 Unfortunately, the ex_data design is not at all suited
3885 for multi-threaded use, so it probably should be abolished.
3886 [Bodo Moeller]
3887
3888 *) Added Broadcom "ubsec" ENGINE to OpenSSL.
3889 [Broadcom, tweaked and integrated by Geoff Thorpe]
3890
3891 *) Move common extension printing code to new function
3892 X509V3_print_extensions(). Reorganise OCSP print routines and
3893 implement some needed OCSP ASN1 functions. Add OCSP extensions.
3894 [Steve Henson]
3895
3896 *) New function X509_signature_print() to remove duplication in some
3897 print routines.
3898 [Steve Henson]
3899
3900 *) Add a special meaning when SET OF and SEQUENCE OF flags are both
3901 set (this was treated exactly the same as SET OF previously). This
3902 is used to reorder the STACK representing the structure to match the
3903 encoding. This will be used to get round a problem where a PKCS7
3904 structure which was signed could not be verified because the STACK
3905 order did not reflect the encoded order.
3906 [Steve Henson]
3907
3908 *) Reimplement the OCSP ASN1 module using the new code.
3909 [Steve Henson]
3910
3911 *) Update the X509V3 code to permit the use of an ASN1_ITEM structure
3912 for its ASN1 operations. The old style function pointers still exist
3913 for now but they will eventually go away.
3914 [Steve Henson]
3915
3916 *) Merge in replacement ASN1 code from the ASN1 branch. This almost
3917 completely replaces the old ASN1 functionality with a table driven
3918 encoder and decoder which interprets an ASN1_ITEM structure describing
3919 the ASN1 module. Compatibility with the existing ASN1 API (i2d,d2i) is
3920 largely maintained. Almost all of the old asn1_mac.h macro based ASN1
3921 has also been converted to the new form.
3922 [Steve Henson]
3923
3924 *) Change BN_mod_exp_recp so that negative moduli are tolerated
3925 (the sign is ignored). Similarly, ignore the sign in BN_MONT_CTX_set
3926 so that BN_mod_exp_mont and BN_mod_exp_mont_word work
3927 for negative moduli.
3928 [Bodo Moeller]
3929
3930 *) Fix BN_uadd and BN_usub: Always return non-negative results instead
3931 of not touching the result's sign bit.
3932 [Bodo Moeller]
3933
3934 *) BN_div bugfix: If the result is 0, the sign (res->neg) must not be
3935 set.
3936 [Bodo Moeller]
3937
3938 *) Changed the LHASH code to use prototypes for callbacks, and created
3939 macros to declare and implement thin (optionally static) functions
3940 that provide type-safety and avoid function pointer casting for the
3941 type-specific callbacks.
3942 [Geoff Thorpe]
3943
3944 *) Added Kerberos Cipher Suites to be used with TLS, as written in
3945 RFC 2712.
3946 [Veers Staats <staatsvr@asc.hpc.mil>,
3947 Jeffrey Altman <jaltman@columbia.edu>, via Richard Levitte]
3948
3949 *) Reformat the FAQ so the different questions and answers can be divided
3950 in sections depending on the subject.
3951 [Richard Levitte]
3952
3953 *) Have the zlib compression code load ZLIB.DLL dynamically under
3954 Windows.
3955 [Richard Levitte]
3956
3957 *) New function BN_mod_sqrt for computing square roots modulo a prime
3958 (using the probabilistic Tonelli-Shanks algorithm unless
3959 p == 3 (mod 4) or p == 5 (mod 8), which are cases that can
3960 be handled deterministically).
3961 [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller]
3962
3963 *) Make BN_mod_inverse faster by explicitly handling small quotients
3964 in the Euclid loop. (Speed gain about 20% for small moduli [256 or
3965 512 bits], about 30% for larger ones [1024 or 2048 bits].)
3966 [Bodo Moeller]
3967
3968 *) New function BN_kronecker.
3969 [Bodo Moeller]
3970
3971 *) Fix BN_gcd so that it works on negative inputs; the result is
3972 positive unless both parameters are zero.
3973 Previously something reasonably close to an infinite loop was
3974 possible because numbers could be growing instead of shrinking
3975 in the implementation of Euclid's algorithm.
3976 [Bodo Moeller]
3977
3978 *) Fix BN_is_word() and BN_is_one() macros to take into account the
3979 sign of the number in question.
3980
3981 Fix BN_is_word(a,w) to work correctly for w == 0.
3982
3983 The old BN_is_word(a,w) macro is now called BN_abs_is_word(a,w)
3984 because its test if the absolute value of 'a' equals 'w'.
3985 Note that BN_abs_is_word does *not* handle w == 0 reliably;
3986 it exists mostly for use in the implementations of BN_is_zero(),
3987 BN_is_one(), and BN_is_word().
3988 [Bodo Moeller]
3989
3990 *) New function BN_swap.
3991 [Bodo Moeller]
3992
3993 *) Use BN_nnmod instead of BN_mod in crypto/bn/bn_exp.c so that
3994 the exponentiation functions are more likely to produce reasonable
3995 results on negative inputs.
3996 [Bodo Moeller]
3997
3998 *) Change BN_mod_mul so that the result is always non-negative.
3999 Previously, it could be negative if one of the factors was negative;
4000 I don't think anyone really wanted that behaviour.
4001 [Bodo Moeller]
4002
4003 *) Move BN_mod_... functions into new file crypto/bn/bn_mod.c
4004 (except for exponentiation, which stays in crypto/bn/bn_exp.c,
4005 and BN_mod_mul_reciprocal, which stays in crypto/bn/bn_recp.c)
4006 and add new functions:
4007
4008 BN_nnmod
4009 BN_mod_sqr
4010 BN_mod_add
4011 BN_mod_add_quick
4012 BN_mod_sub
4013 BN_mod_sub_quick
4014 BN_mod_lshift1
4015 BN_mod_lshift1_quick
4016 BN_mod_lshift
4017 BN_mod_lshift_quick
4018
4019 These functions always generate non-negative results.
4020
4021 BN_nnmod otherwise is like BN_mod (if BN_mod computes a remainder r
4022 such that |m| < r < 0, BN_nnmod will output rem + |m| instead).
4023
4024 BN_mod_XXX_quick(r, a, [b,] m) generates the same result as
4025 BN_mod_XXX(r, a, [b,] m, ctx), but requires that a [and b]
4026 be reduced modulo m.
4027 [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller]
4028
4029 #if 0
4030 The following entry accidentily appeared in the CHANGES file
4031 distributed with OpenSSL 0.9.7. The modifications described in
4032 it do *not* apply to OpenSSL 0.9.7.
4033
4034 *) Remove a few calls to bn_wexpand() in BN_sqr() (the one in there
4035 was actually never needed) and in BN_mul(). The removal in BN_mul()
4036 required a small change in bn_mul_part_recursive() and the addition
4037 of the functions bn_cmp_part_words(), bn_sub_part_words() and
4038 bn_add_part_words(), which do the same thing as bn_cmp_words(),
4039 bn_sub_words() and bn_add_words() except they take arrays with
4040 differing sizes.
4041 [Richard Levitte]
4042 #endif
4043
4044 *) In 'openssl passwd', verify passwords read from the terminal
4045 unless the '-salt' option is used (which usually means that
4046 verification would just waste user's time since the resulting
4047 hash is going to be compared with some given password hash)
4048 or the new '-noverify' option is used.
4049
4050 This is an incompatible change, but it does not affect
4051 non-interactive use of 'openssl passwd' (passwords on the command
4052 line, '-stdin' option, '-in ...' option) and thus should not
4053 cause any problems.
4054 [Bodo Moeller]
4055
4056 *) Remove all references to RSAref, since there's no more need for it.
4057 [Richard Levitte]
4058
4059 *) Make DSO load along a path given through an environment variable
4060 (SHLIB_PATH) with shl_load().
4061 [Richard Levitte]
4062
4063 *) Constify the ENGINE code as a result of BIGNUM constification.
4064 Also constify the RSA code and most things related to it. In a
4065 few places, most notable in the depth of the ASN.1 code, ugly
4066 casts back to non-const were required (to be solved at a later
4067 time)
4068 [Richard Levitte]
4069
4070 *) Make it so the openssl application has all engines loaded by default.
4071 [Richard Levitte]
4072
4073 *) Constify the BIGNUM routines a little more.
4074 [Richard Levitte]
4075
4076 *) Add the following functions:
4077
4078 ENGINE_load_cswift()
4079 ENGINE_load_chil()
4080 ENGINE_load_atalla()
4081 ENGINE_load_nuron()
4082 ENGINE_load_builtin_engines()
4083
4084 That way, an application can itself choose if external engines that
4085 are built-in in OpenSSL shall ever be used or not. The benefit is
4086 that applications won't have to be linked with libdl or other dso
4087 libraries unless it's really needed.
4088
4089 Changed 'openssl engine' to load all engines on demand.
4090 Changed the engine header files to avoid the duplication of some
4091 declarations (they differed!).
4092 [Richard Levitte]
4093
4094 *) 'openssl engine' can now list capabilities.
4095 [Richard Levitte]
4096
4097 *) Better error reporting in 'openssl engine'.
4098 [Richard Levitte]
4099
4100 *) Never call load_dh_param(NULL) in s_server.
4101 [Bodo Moeller]
4102
4103 *) Add engine application. It can currently list engines by name and
4104 identity, and test if they are actually available.
4105 [Richard Levitte]
4106
4107 *) Improve RPM specification file by forcing symbolic linking and making
4108 sure the installed documentation is also owned by root.root.
4109 [Damien Miller <djm@mindrot.org>]
4110
4111 *) Give the OpenSSL applications more possibilities to make use of
4112 keys (public as well as private) handled by engines.
4113 [Richard Levitte]
4114
4115 *) Add OCSP code that comes from CertCo.
4116 [Richard Levitte]
4117
4118 *) Add VMS support for the Rijndael code.
4119 [Richard Levitte]
4120
4121 *) Added untested support for Nuron crypto accelerator.
4122 [Ben Laurie]
4123
4124 *) Add support for external cryptographic devices. This code was
4125 previously distributed separately as the "engine" branch.
4126 [Geoff Thorpe, Richard Levitte]
4127
4128 *) Rework the filename-translation in the DSO code. It is now possible to
4129 have far greater control over how a "name" is turned into a filename
4130 depending on the operating environment and any oddities about the
4131 different shared library filenames on each system.
4132 [Geoff Thorpe]
4133
4134 *) Support threads on FreeBSD-elf in Configure.
4135 [Richard Levitte]
4136
4137 *) Fix for SHA1 assembly problem with MASM: it produces
4138 warnings about corrupt line number information when assembling
4139 with debugging information. This is caused by the overlapping
4140 of two sections.
4141 [Bernd Matthes <mainbug@celocom.de>, Steve Henson]
4142
4143 *) NCONF changes.
4144 NCONF_get_number() has no error checking at all. As a replacement,
4145 NCONF_get_number_e() is defined (_e for "error checking") and is
4146 promoted strongly. The old NCONF_get_number is kept around for
4147 binary backward compatibility.
4148 Make it possible for methods to load from something other than a BIO,
4149 by providing a function pointer that is given a name instead of a BIO.
4150 For example, this could be used to load configuration data from an
4151 LDAP server.
4152 [Richard Levitte]
4153
4154 *) Fix for non blocking accept BIOs. Added new I/O special reason
4155 BIO_RR_ACCEPT to cover this case. Previously use of accept BIOs
4156 with non blocking I/O was not possible because no retry code was
4157 implemented. Also added new SSL code SSL_WANT_ACCEPT to cover
4158 this case.
4159 [Steve Henson]
4160
4161 *) Added the beginnings of Rijndael support.
4162 [Ben Laurie]
4163
4164 *) Fix for bug in DirectoryString mask setting. Add support for
4165 X509_NAME_print_ex() in 'req' and X509_print_ex() function
4166 to allow certificate printing to more controllable, additional
4167 'certopt' option to 'x509' to allow new printing options to be
4168 set.
4169 [Steve Henson]
4170
4171 *) Clean old EAY MD5 hack from e_os.h.
4172 [Richard Levitte]
4173
4174 Changes between 0.9.6l and 0.9.6m [17 Mar 2004]
4175
4176 *) Fix null-pointer assignment in do_change_cipher_spec() revealed
4177 by using the Codenomicon TLS Test Tool (CVE-2004-0079)
4178 [Joe Orton, Steve Henson]
4179
4180 Changes between 0.9.6k and 0.9.6l [04 Nov 2003]
4181
4182 *) Fix additional bug revealed by the NISCC test suite:
4183
4184 Stop bug triggering large recursion when presented with
4185 certain ASN.1 tags (CVE-2003-0851)
4186 [Steve Henson]
4187
4188 Changes between 0.9.6j and 0.9.6k [30 Sep 2003]
4189
4190 *) Fix various bugs revealed by running the NISCC test suite:
4191
4192 Stop out of bounds reads in the ASN1 code when presented with
4193 invalid tags (CVE-2003-0543 and CVE-2003-0544).
4194
4195 If verify callback ignores invalid public key errors don't try to check
4196 certificate signature with the NULL public key.
4197
4198 [Steve Henson]
4199
4200 *) In ssl3_accept() (ssl/s3_srvr.c) only accept a client certificate
4201 if the server requested one: as stated in TLS 1.0 and SSL 3.0
4202 specifications.
4203 [Steve Henson]
4204
4205 *) In ssl3_get_client_hello() (ssl/s3_srvr.c), tolerate additional
4206 extra data after the compression methods not only for TLS 1.0
4207 but also for SSL 3.0 (as required by the specification).
4208 [Bodo Moeller; problem pointed out by Matthias Loepfe]
4209
4210 *) Change X509_certificate_type() to mark the key as exported/exportable
4211 when it's 512 *bits* long, not 512 bytes.
4212 [Richard Levitte]
4213
4214 Changes between 0.9.6i and 0.9.6j [10 Apr 2003]
4215
4216 *) Countermeasure against the Klima-Pokorny-Rosa extension of
4217 Bleichbacher's attack on PKCS #1 v1.5 padding: treat
4218 a protocol version number mismatch like a decryption error
4219 in ssl3_get_client_key_exchange (ssl/s3_srvr.c).
4220 [Bodo Moeller]
4221
4222 *) Turn on RSA blinding by default in the default implementation
4223 to avoid a timing attack. Applications that don't want it can call
4224 RSA_blinding_off() or use the new flag RSA_FLAG_NO_BLINDING.
4225 They would be ill-advised to do so in most cases.
4226 [Ben Laurie, Steve Henson, Geoff Thorpe, Bodo Moeller]
4227
4228 *) Change RSA blinding code so that it works when the PRNG is not
4229 seeded (in this case, the secret RSA exponent is abused as
4230 an unpredictable seed -- if it is not unpredictable, there
4231 is no point in blinding anyway). Make RSA blinding thread-safe
4232 by remembering the creator's thread ID in rsa->blinding and
4233 having all other threads use local one-time blinding factors
4234 (this requires more computation than sharing rsa->blinding, but
4235 avoids excessive locking; and if an RSA object is not shared
4236 between threads, blinding will still be very fast).
4237 [Bodo Moeller]
4238
4239 Changes between 0.9.6h and 0.9.6i [19 Feb 2003]
4240
4241 *) In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked
4242 via timing by performing a MAC computation even if incorrrect
4243 block cipher padding has been found. This is a countermeasure
4244 against active attacks where the attacker has to distinguish
4245 between bad padding and a MAC verification error. (CVE-2003-0078)
4246
4247 [Bodo Moeller; problem pointed out by Brice Canvel (EPFL),
4248 Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and
4249 Martin Vuagnoux (EPFL, Ilion)]
4250
4251 Changes between 0.9.6g and 0.9.6h [5 Dec 2002]
4252
4253 *) New function OPENSSL_cleanse(), which is used to cleanse a section of
4254 memory from it's contents. This is done with a counter that will
4255 place alternating values in each byte. This can be used to solve
4256 two issues: 1) the removal of calls to memset() by highly optimizing
4257 compilers, and 2) cleansing with other values than 0, since those can
4258 be read through on certain media, for example a swap space on disk.
4259 [Geoff Thorpe]
4260
4261 *) Bugfix: client side session caching did not work with external caching,
4262 because the session->cipher setting was not restored when reloading
4263 from the external cache. This problem was masked, when
4264 SSL_OP_NETSCAPE_REUSE_CIPHER_CHANGE_BUG (part of SSL_OP_ALL) was set.
4265 (Found by Steve Haslam <steve@araqnid.ddts.net>.)
4266 [Lutz Jaenicke]
4267
4268 *) Fix client_certificate (ssl/s2_clnt.c): The permissible total
4269 length of the REQUEST-CERTIFICATE message is 18 .. 34, not 17 .. 33.
4270 [Zeev Lieber <zeev-l@yahoo.com>]
4271
4272 *) Undo an undocumented change introduced in 0.9.6e which caused
4273 repeated calls to OpenSSL_add_all_ciphers() and
4274 OpenSSL_add_all_digests() to be ignored, even after calling
4275 EVP_cleanup().
4276 [Richard Levitte]
4277
4278 *) Change the default configuration reader to deal with last line not
4279 being properly terminated.
4280 [Richard Levitte]
4281
4282 *) Change X509_NAME_cmp() so it applies the special rules on handling
4283 DN values that are of type PrintableString, as well as RDNs of type
4284 emailAddress where the value has the type ia5String.
4285 [stefank@valicert.com via Richard Levitte]
4286
4287 *) Add a SSL_SESS_CACHE_NO_INTERNAL_STORE flag to take over half
4288 the job SSL_SESS_CACHE_NO_INTERNAL_LOOKUP was inconsistently
4289 doing, define a new flag (SSL_SESS_CACHE_NO_INTERNAL) to be
4290 the bitwise-OR of the two for use by the majority of applications
4291 wanting this behaviour, and update the docs. The documented
4292 behaviour and actual behaviour were inconsistent and had been
4293 changing anyway, so this is more a bug-fix than a behavioural
4294 change.
4295 [Geoff Thorpe, diagnosed by Nadav Har'El]
4296
4297 *) Don't impose a 16-byte length minimum on session IDs in ssl/s3_clnt.c
4298 (the SSL 3.0 and TLS 1.0 specifications allow any length up to 32 bytes).
4299 [Bodo Moeller]
4300
4301 *) Fix initialization code race conditions in
4302 SSLv23_method(), SSLv23_client_method(), SSLv23_server_method(),
4303 SSLv2_method(), SSLv2_client_method(), SSLv2_server_method(),
4304 SSLv3_method(), SSLv3_client_method(), SSLv3_server_method(),
4305 TLSv1_method(), TLSv1_client_method(), TLSv1_server_method(),
4306 ssl2_get_cipher_by_char(),
4307 ssl3_get_cipher_by_char().
4308 [Patrick McCormick <patrick@tellme.com>, Bodo Moeller]
4309
4310 *) Reorder cleanup sequence in SSL_CTX_free(): only remove the ex_data after
4311 the cached sessions are flushed, as the remove_cb() might use ex_data
4312 contents. Bug found by Sam Varshavchik <mrsam@courier-mta.com>
4313 (see [openssl.org #212]).
4314 [Geoff Thorpe, Lutz Jaenicke]
4315
4316 *) Fix typo in OBJ_txt2obj which incorrectly passed the content
4317 length, instead of the encoding length to d2i_ASN1_OBJECT.
4318 [Steve Henson]
4319
4320 Changes between 0.9.6f and 0.9.6g [9 Aug 2002]
4321
4322 *) [In 0.9.6g-engine release:]
4323 Fix crypto/engine/vendor_defns/cswift.h for WIN32 (use '_stdcall').
4324 [Lynn Gazis <lgazis@rainbow.com>]
4325
4326 Changes between 0.9.6e and 0.9.6f [8 Aug 2002]
4327
4328 *) Fix ASN1 checks. Check for overflow by comparing with LONG_MAX
4329 and get fix the header length calculation.
4330 [Florian Weimer <Weimer@CERT.Uni-Stuttgart.DE>,
4331 Alon Kantor <alonk@checkpoint.com> (and others),
4332 Steve Henson]
4333
4334 *) Use proper error handling instead of 'assertions' in buffer
4335 overflow checks added in 0.9.6e. This prevents DoS (the
4336 assertions could call abort()).
4337 [Arne Ansper <arne@ats.cyber.ee>, Bodo Moeller]
4338
4339 Changes between 0.9.6d and 0.9.6e [30 Jul 2002]
4340
4341 *) Add various sanity checks to asn1_get_length() to reject
4342 the ASN1 length bytes if they exceed sizeof(long), will appear
4343 negative or the content length exceeds the length of the
4344 supplied buffer.
4345 [Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>]
4346
4347 *) Fix cipher selection routines: ciphers without encryption had no flags
4348 for the cipher strength set and where therefore not handled correctly
4349 by the selection routines (PR #130).
4350 [Lutz Jaenicke]
4351
4352 *) Fix EVP_dsa_sha macro.
4353 [Nils Larsch]
4354
4355 *) New option
4356 SSL_OP_DONT_INSERT_EMPTY_FRAGMENTS
4357 for disabling the SSL 3.0/TLS 1.0 CBC vulnerability countermeasure
4358 that was added in OpenSSL 0.9.6d.
4359
4360 As the countermeasure turned out to be incompatible with some
4361 broken SSL implementations, the new option is part of SSL_OP_ALL.
4362 SSL_OP_ALL is usually employed when compatibility with weird SSL
4363 implementations is desired (e.g. '-bugs' option to 's_client' and
4364 's_server'), so the new option is automatically set in many
4365 applications.
4366 [Bodo Moeller]
4367
4368 *) Changes in security patch:
4369
4370 Changes marked "(CHATS)" were sponsored by the Defense Advanced
4371 Research Projects Agency (DARPA) and Air Force Research Laboratory,
4372 Air Force Materiel Command, USAF, under agreement number
4373 F30602-01-2-0537.
4374
4375 *) Add various sanity checks to asn1_get_length() to reject
4376 the ASN1 length bytes if they exceed sizeof(long), will appear
4377 negative or the content length exceeds the length of the
4378 supplied buffer. (CVE-2002-0659)
4379 [Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>]
4380
4381 *) Assertions for various potential buffer overflows, not known to
4382 happen in practice.
4383 [Ben Laurie (CHATS)]
4384
4385 *) Various temporary buffers to hold ASCII versions of integers were
4386 too small for 64 bit platforms. (CVE-2002-0655)
4387 [Matthew Byng-Maddick <mbm@aldigital.co.uk> and Ben Laurie (CHATS)>
4388
4389 *) Remote buffer overflow in SSL3 protocol - an attacker could
4390 supply an oversized session ID to a client. (CVE-2002-0656)
4391 [Ben Laurie (CHATS)]
4392
4393 *) Remote buffer overflow in SSL2 protocol - an attacker could
4394 supply an oversized client master key. (CVE-2002-0656)
4395 [Ben Laurie (CHATS)]
4396
4397 Changes between 0.9.6c and 0.9.6d [9 May 2002]
4398
4399 *) Fix crypto/asn1/a_sign.c so that 'parameters' is omitted (not
4400 encoded as NULL) with id-dsa-with-sha1.
4401 [Nils Larsch <nla@trustcenter.de>; problem pointed out by Bodo Moeller]
4402
4403 *) Check various X509_...() return values in apps/req.c.
4404 [Nils Larsch <nla@trustcenter.de>]
4405
4406 *) Fix BASE64 decode (EVP_DecodeUpdate) for data with CR/LF ended lines:
4407 an end-of-file condition would erronously be flagged, when the CRLF
4408 was just at the end of a processed block. The bug was discovered when
4409 processing data through a buffering memory BIO handing the data to a
4410 BASE64-decoding BIO. Bug fund and patch submitted by Pavel Tsekov
4411 <ptsekov@syntrex.com> and Nedelcho Stanev.
4412 [Lutz Jaenicke]
4413
4414 *) Implement a countermeasure against a vulnerability recently found
4415 in CBC ciphersuites in SSL 3.0/TLS 1.0: Send an empty fragment
4416 before application data chunks to avoid the use of known IVs
4417 with data potentially chosen by the attacker.
4418 [Bodo Moeller]
4419
4420 *) Fix length checks in ssl3_get_client_hello().
4421 [Bodo Moeller]
4422
4423 *) TLS/SSL library bugfix: use s->s3->in_read_app_data differently
4424 to prevent ssl3_read_internal() from incorrectly assuming that
4425 ssl3_read_bytes() found application data while handshake
4426 processing was enabled when in fact s->s3->in_read_app_data was
4427 merely automatically cleared during the initial handshake.
4428 [Bodo Moeller; problem pointed out by Arne Ansper <arne@ats.cyber.ee>]
4429
4430 *) Fix object definitions for Private and Enterprise: they were not
4431 recognized in their shortname (=lowercase) representation. Extend
4432 obj_dat.pl to issue an error when using undefined keywords instead
4433 of silently ignoring the problem (Svenning Sorensen
4434 <sss@sss.dnsalias.net>).
4435 [Lutz Jaenicke]
4436
4437 *) Fix DH_generate_parameters() so that it works for 'non-standard'
4438 generators, i.e. generators other than 2 and 5. (Previously, the
4439 code did not properly initialise the 'add' and 'rem' values to
4440 BN_generate_prime().)
4441
4442 In the new general case, we do not insist that 'generator' is
4443 actually a primitive root: This requirement is rather pointless;
4444 a generator of the order-q subgroup is just as good, if not
4445 better.
4446 [Bodo Moeller]
4447
4448 *) Map new X509 verification errors to alerts. Discovered and submitted by
4449 Tom Wu <tom@arcot.com>.
4450 [Lutz Jaenicke]
4451
4452 *) Fix ssl3_pending() (ssl/s3_lib.c) to prevent SSL_pending() from
4453 returning non-zero before the data has been completely received
4454 when using non-blocking I/O.
4455 [Bodo Moeller; problem pointed out by John Hughes]
4456
4457 *) Some of the ciphers missed the strength entry (SSL_LOW etc).
4458 [Ben Laurie, Lutz Jaenicke]
4459
4460 *) Fix bug in SSL_clear(): bad sessions were not removed (found by
4461 Yoram Zahavi <YoramZ@gilian.com>).
4462 [Lutz Jaenicke]
4463
4464 *) Add information about CygWin 1.3 and on, and preserve proper
4465 configuration for the versions before that.
4466 [Corinna Vinschen <vinschen@redhat.com> and Richard Levitte]
4467
4468 *) Make removal from session cache (SSL_CTX_remove_session()) more robust:
4469 check whether we deal with a copy of a session and do not delete from
4470 the cache in this case. Problem reported by "Izhar Shoshani Levi"
4471 <izhar@checkpoint.com>.
4472 [Lutz Jaenicke]
4473
4474 *) Do not store session data into the internal session cache, if it
4475 is never intended to be looked up (SSL_SESS_CACHE_NO_INTERNAL_LOOKUP
4476 flag is set). Proposed by Aslam <aslam@funk.com>.
4477 [Lutz Jaenicke]
4478
4479 *) Have ASN1_BIT_STRING_set_bit() really clear a bit when the requested
4480 value is 0.
4481 [Richard Levitte]
4482
4483 *) [In 0.9.6d-engine release:]
4484 Fix a crashbug and a logic bug in hwcrhk_load_pubkey().
4485 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte]
4486
4487 *) Add the configuration target linux-s390x.
4488 [Neale Ferguson <Neale.Ferguson@SoftwareAG-USA.com> via Richard Levitte]
4489
4490 *) The earlier bugfix for the SSL3_ST_SW_HELLO_REQ_C case of
4491 ssl3_accept (ssl/s3_srvr.c) incorrectly used a local flag
4492 variable as an indication that a ClientHello message has been
4493 received. As the flag value will be lost between multiple
4494 invocations of ssl3_accept when using non-blocking I/O, the
4495 function may not be aware that a handshake has actually taken
4496 place, thus preventing a new session from being added to the
4497 session cache.
4498
4499 To avoid this problem, we now set s->new_session to 2 instead of
4500 using a local variable.
4501 [Lutz Jaenicke, Bodo Moeller]
4502
4503 *) Bugfix: Return -1 from ssl3_get_server_done (ssl3/s3_clnt.c)
4504 if the SSL_R_LENGTH_MISMATCH error is detected.
4505 [Geoff Thorpe, Bodo Moeller]
4506
4507 *) New 'shared_ldflag' column in Configure platform table.
4508 [Richard Levitte]
4509
4510 *) Fix EVP_CIPHER_mode macro.
4511 ["Dan S. Camper" <dan@bti.net>]
4512
4513 *) Fix ssl3_read_bytes (ssl/s3_pkt.c): To ignore messages of unknown
4514 type, we must throw them away by setting rr->length to 0.
4515 [D P Chang <dpc@qualys.com>]
4516
4517 Changes between 0.9.6b and 0.9.6c [21 dec 2001]
4518
4519 *) Fix BN_rand_range bug pointed out by Dominikus Scherkl
4520 <Dominikus.Scherkl@biodata.com>. (The previous implementation
4521 worked incorrectly for those cases where range = 10..._2 and
4522 3*range is two bits longer than range.)
4523 [Bodo Moeller]
4524
4525 *) Only add signing time to PKCS7 structures if it is not already
4526 present.
4527 [Steve Henson]
4528
4529 *) Fix crypto/objects/objects.h: "ld-ce" should be "id-ce",
4530 OBJ_ld_ce should be OBJ_id_ce.
4531 Also some ip-pda OIDs in crypto/objects/objects.txt were
4532 incorrect (cf. RFC 3039).
4533 [Matt Cooper, Frederic Giudicelli, Bodo Moeller]
4534
4535 *) Release CRYPTO_LOCK_DYNLOCK when CRYPTO_destroy_dynlockid()
4536 returns early because it has nothing to do.
4537 [Andy Schneider <andy.schneider@bjss.co.uk>]
4538
4539 *) [In 0.9.6c-engine release:]
4540 Fix mutex callback return values in crypto/engine/hw_ncipher.c.
4541 [Andy Schneider <andy.schneider@bjss.co.uk>]
4542
4543 *) [In 0.9.6c-engine release:]
4544 Add support for Cryptographic Appliance's keyserver technology.
4545 (Use engine 'keyclient')
4546 [Cryptographic Appliances and Geoff Thorpe]
4547
4548 *) Add a configuration entry for OS/390 Unix. The C compiler 'c89'
4549 is called via tools/c89.sh because arguments have to be
4550 rearranged (all '-L' options must appear before the first object
4551 modules).
4552 [Richard Shapiro <rshapiro@abinitio.com>]
4553
4554 *) [In 0.9.6c-engine release:]
4555 Add support for Broadcom crypto accelerator cards, backported
4556 from 0.9.7.
4557 [Broadcom, Nalin Dahyabhai <nalin@redhat.com>, Mark Cox]
4558
4559 *) [In 0.9.6c-engine release:]
4560 Add support for SureWare crypto accelerator cards from
4561 Baltimore Technologies. (Use engine 'sureware')
4562 [Baltimore Technologies and Mark Cox]
4563
4564 *) [In 0.9.6c-engine release:]
4565 Add support for crypto accelerator cards from Accelerated
4566 Encryption Processing, www.aep.ie. (Use engine 'aep')
4567 [AEP Inc. and Mark Cox]
4568
4569 *) Add a configuration entry for gcc on UnixWare.
4570 [Gary Benson <gbenson@redhat.com>]
4571
4572 *) Change ssl/s2_clnt.c and ssl/s2_srvr.c so that received handshake
4573 messages are stored in a single piece (fixed-length part and
4574 variable-length part combined) and fix various bugs found on the way.
4575 [Bodo Moeller]
4576
4577 *) Disable caching in BIO_gethostbyname(), directly use gethostbyname()
4578 instead. BIO_gethostbyname() does not know what timeouts are
4579 appropriate, so entries would stay in cache even when they have
4580 become invalid.
4581 [Bodo Moeller; problem pointed out by Rich Salz <rsalz@zolera.com>
4582
4583 *) Change ssl23_get_client_hello (ssl/s23_srvr.c) behaviour when
4584 faced with a pathologically small ClientHello fragment that does
4585 not contain client_version: Instead of aborting with an error,
4586 simply choose the highest available protocol version (i.e.,
4587 TLS 1.0 unless it is disabled). In practice, ClientHello
4588 messages are never sent like this, but this change gives us
4589 strictly correct behaviour at least for TLS.
4590 [Bodo Moeller]
4591
4592 *) Fix SSL handshake functions and SSL_clear() such that SSL_clear()
4593 never resets s->method to s->ctx->method when called from within
4594 one of the SSL handshake functions.
4595 [Bodo Moeller; problem pointed out by Niko Baric]
4596
4597 *) In ssl3_get_client_hello (ssl/s3_srvr.c), generate a fatal alert
4598 (sent using the client's version number) if client_version is
4599 smaller than the protocol version in use. Also change
4600 ssl23_get_client_hello (ssl/s23_srvr.c) to select TLS 1.0 if
4601 the client demanded SSL 3.0 but only TLS 1.0 is enabled; then
4602 the client will at least see that alert.
4603 [Bodo Moeller]
4604
4605 *) Fix ssl3_get_message (ssl/s3_both.c) to handle message fragmentation
4606 correctly.
4607 [Bodo Moeller]
4608
4609 *) Avoid infinite loop in ssl3_get_message (ssl/s3_both.c) if a
4610 client receives HelloRequest while in a handshake.
4611 [Bodo Moeller; bug noticed by Andy Schneider <andy.schneider@bjss.co.uk>]
4612
4613 *) Bugfix in ssl3_accept (ssl/s3_srvr.c): Case SSL3_ST_SW_HELLO_REQ_C
4614 should end in 'break', not 'goto end' which circuments various
4615 cleanups done in state SSL_ST_OK. But session related stuff
4616 must be disabled for SSL_ST_OK in the case that we just sent a
4617 HelloRequest.
4618
4619 Also avoid some overhead by not calling ssl_init_wbio_buffer()
4620 before just sending a HelloRequest.
4621 [Bodo Moeller, Eric Rescorla <ekr@rtfm.com>]
4622
4623 *) Fix ssl/s3_enc.c, ssl/t1_enc.c and ssl/s3_pkt.c so that we don't
4624 reveal whether illegal block cipher padding was found or a MAC
4625 verification error occured. (Neither SSLerr() codes nor alerts
4626 are directly visible to potential attackers, but the information
4627 may leak via logfiles.)
4628
4629 Similar changes are not required for the SSL 2.0 implementation
4630 because the number of padding bytes is sent in clear for SSL 2.0,
4631 and the extra bytes are just ignored. However ssl/s2_pkt.c
4632 failed to verify that the purported number of padding bytes is in
4633 the legal range.
4634 [Bodo Moeller]
4635
4636 *) Add OpenUNIX-8 support including shared libraries
4637 (Boyd Lynn Gerber <gerberb@zenez.com>).
4638 [Lutz Jaenicke]
4639
4640 *) Improve RSA_padding_check_PKCS1_OAEP() check again to avoid
4641 'wristwatch attack' using huge encoding parameters (cf.
4642 James H. Manger's CRYPTO 2001 paper). Note that the
4643 RSA_PKCS1_OAEP_PADDING case of RSA_private_decrypt() does not use
4644 encoding parameters and hence was not vulnerable.
4645 [Bodo Moeller]
4646
4647 *) BN_sqr() bug fix.
4648 [Ulf Möller, reported by Jim Ellis <jim.ellis@cavium.com>]
4649
4650 *) Rabin-Miller test analyses assume uniformly distributed witnesses,
4651 so use BN_pseudo_rand_range() instead of using BN_pseudo_rand()
4652 followed by modular reduction.
4653 [Bodo Moeller; pointed out by Adam Young <AYoung1@NCSUS.JNJ.COM>]
4654
4655 *) Add BN_pseudo_rand_range() with obvious functionality: BN_rand_range()
4656 equivalent based on BN_pseudo_rand() instead of BN_rand().
4657 [Bodo Moeller]
4658
4659 *) s3_srvr.c: allow sending of large client certificate lists (> 16 kB).
4660 This function was broken, as the check for a new client hello message
4661 to handle SGC did not allow these large messages.
4662 (Tracked down by "Douglas E. Engert" <deengert@anl.gov>.)
4663 [Lutz Jaenicke]
4664
4665 *) Add alert descriptions for TLSv1 to SSL_alert_desc_string[_long]().
4666 [Lutz Jaenicke]
4667
4668 *) Fix buggy behaviour of BIO_get_num_renegotiates() and BIO_ctrl()
4669 for BIO_C_GET_WRITE_BUF_SIZE ("Stephen Hinton" <shinton@netopia.com>).
4670 [Lutz Jaenicke]
4671
4672 *) Rework the configuration and shared library support for Tru64 Unix.
4673 The configuration part makes use of modern compiler features and
4674 still retains old compiler behavior for those that run older versions
4675 of the OS. The shared library support part includes a variant that
4676 uses the RPATH feature, and is available through the special
4677 configuration target "alpha-cc-rpath", which will never be selected
4678 automatically.
4679 [Tim Mooney <mooney@dogbert.cc.ndsu.NoDak.edu> via Richard Levitte]
4680
4681 *) In ssl3_get_key_exchange (ssl/s3_clnt.c), call ssl3_get_message()
4682 with the same message size as in ssl3_get_certificate_request().
4683 Otherwise, if no ServerKeyExchange message occurs, CertificateRequest
4684 messages might inadvertently be reject as too long.
4685 [Petr Lampa <lampa@fee.vutbr.cz>]
4686
4687 *) Enhanced support for IA-64 Unix platforms (well, Linux and HP-UX).
4688 [Andy Polyakov]
4689
4690 *) Modified SSL library such that the verify_callback that has been set
4691 specificly for an SSL object with SSL_set_verify() is actually being
4692 used. Before the change, a verify_callback set with this function was
4693 ignored and the verify_callback() set in the SSL_CTX at the time of
4694 the call was used. New function X509_STORE_CTX_set_verify_cb() introduced
4695 to allow the necessary settings.
4696 [Lutz Jaenicke]
4697
4698 *) Initialize static variable in crypto/dsa/dsa_lib.c and crypto/dh/dh_lib.c
4699 explicitly to NULL, as at least on Solaris 8 this seems not always to be
4700 done automatically (in contradiction to the requirements of the C
4701 standard). This made problems when used from OpenSSH.
4702 [Lutz Jaenicke]
4703
4704 *) In OpenSSL 0.9.6a and 0.9.6b, crypto/dh/dh_key.c ignored
4705 dh->length and always used
4706
4707 BN_rand_range(priv_key, dh->p).
4708
4709 BN_rand_range() is not necessary for Diffie-Hellman, and this
4710 specific range makes Diffie-Hellman unnecessarily inefficient if
4711 dh->length (recommended exponent length) is much smaller than the
4712 length of dh->p. We could use BN_rand_range() if the order of
4713 the subgroup was stored in the DH structure, but we only have
4714 dh->length.
4715
4716 So switch back to
4717
4718 BN_rand(priv_key, l, ...)
4719
4720 where 'l' is dh->length if this is defined, or BN_num_bits(dh->p)-1
4721 otherwise.
4722 [Bodo Moeller]
4723
4724 *) In
4725
4726 RSA_eay_public_encrypt
4727 RSA_eay_private_decrypt
4728 RSA_eay_private_encrypt (signing)
4729 RSA_eay_public_decrypt (signature verification)
4730
4731 (default implementations for RSA_public_encrypt,
4732 RSA_private_decrypt, RSA_private_encrypt, RSA_public_decrypt),
4733 always reject numbers >= n.
4734 [Bodo Moeller]
4735
4736 *) In crypto/rand/md_rand.c, use a new short-time lock CRYPTO_LOCK_RAND2
4737 to synchronize access to 'locking_thread'. This is necessary on
4738 systems where access to 'locking_thread' (an 'unsigned long'
4739 variable) is not atomic.
4740 [Bodo Moeller]
4741
4742 *) In crypto/rand/md_rand.c, set 'locking_thread' to current thread's ID
4743 *before* setting the 'crypto_lock_rand' flag. The previous code had
4744 a race condition if 0 is a valid thread ID.
4745 [Travis Vitek <vitek@roguewave.com>]
4746
4747 *) Add support for shared libraries under Irix.
4748 [Albert Chin-A-Young <china@thewrittenword.com>]
4749
4750 *) Add configuration option to build on Linux on both big-endian and
4751 little-endian MIPS.
4752 [Ralf Baechle <ralf@uni-koblenz.de>]
4753
4754 *) Add the possibility to create shared libraries on HP-UX.
4755 [Richard Levitte]
4756
4757 Changes between 0.9.6a and 0.9.6b [9 Jul 2001]
4758
4759 *) Change ssleay_rand_bytes (crypto/rand/md_rand.c)
4760 to avoid a SSLeay/OpenSSL PRNG weakness pointed out by
4761 Markku-Juhani O. Saarinen <markku-juhani.saarinen@nokia.com>:
4762 PRNG state recovery was possible based on the output of
4763 one PRNG request appropriately sized to gain knowledge on
4764 'md' followed by enough consecutive 1-byte PRNG requests
4765 to traverse all of 'state'.
4766
4767 1. When updating 'md_local' (the current thread's copy of 'md')
4768 during PRNG output generation, hash all of the previous
4769 'md_local' value, not just the half used for PRNG output.
4770
4771 2. Make the number of bytes from 'state' included into the hash
4772 independent from the number of PRNG bytes requested.
4773
4774 The first measure alone would be sufficient to avoid
4775 Markku-Juhani's attack. (Actually it had never occurred
4776 to me that the half of 'md_local' used for chaining was the
4777 half from which PRNG output bytes were taken -- I had always
4778 assumed that the secret half would be used.) The second
4779 measure makes sure that additional data from 'state' is never
4780 mixed into 'md_local' in small portions; this heuristically
4781 further strengthens the PRNG.
4782 [Bodo Moeller]
4783
4784 *) Fix crypto/bn/asm/mips3.s.
4785 [Andy Polyakov]
4786
4787 *) When only the key is given to "enc", the IV is undefined. Print out
4788 an error message in this case.
4789 [Lutz Jaenicke]
4790
4791 *) Handle special case when X509_NAME is empty in X509 printing routines.
4792 [Steve Henson]
4793
4794 *) In dsa_do_verify (crypto/dsa/dsa_ossl.c), verify that r and s are
4795 positive and less than q.
4796 [Bodo Moeller]
4797
4798 *) Don't change *pointer in CRYPTO_add_lock() is add_lock_callback is
4799 used: it isn't thread safe and the add_lock_callback should handle
4800 that itself.
4801 [Paul Rose <Paul.Rose@bridge.com>]
4802
4803 *) Verify that incoming data obeys the block size in
4804 ssl3_enc (ssl/s3_enc.c) and tls1_enc (ssl/t1_enc.c).
4805 [Bodo Moeller]
4806
4807 *) Fix OAEP check.
4808 [Ulf Möller, Bodo Möller]
4809
4810 *) The countermeasure against Bleichbacher's attack on PKCS #1 v1.5
4811 RSA encryption was accidentally removed in s3_srvr.c in OpenSSL 0.9.5
4812 when fixing the server behaviour for backwards-compatible 'client
4813 hello' messages. (Note that the attack is impractical against
4814 SSL 3.0 and TLS 1.0 anyway because length and version checking
4815 means that the probability of guessing a valid ciphertext is
4816 around 2^-40; see section 5 in Bleichenbacher's CRYPTO '98
4817 paper.)
4818
4819 Before 0.9.5, the countermeasure (hide the error by generating a
4820 random 'decryption result') did not work properly because
4821 ERR_clear_error() was missing, meaning that SSL_get_error() would
4822 detect the supposedly ignored error.
4823
4824 Both problems are now fixed.
4825 [Bodo Moeller]
4826
4827 *) In crypto/bio/bf_buff.c, increase DEFAULT_BUFFER_SIZE to 4096
4828 (previously it was 1024).
4829 [Bodo Moeller]
4830
4831 *) Fix for compatibility mode trust settings: ignore trust settings
4832 unless some valid trust or reject settings are present.
4833 [Steve Henson]
4834
4835 *) Fix for blowfish EVP: its a variable length cipher.
4836 [Steve Henson]
4837
4838 *) Fix various bugs related to DSA S/MIME verification. Handle missing
4839 parameters in DSA public key structures and return an error in the
4840 DSA routines if parameters are absent.
4841 [Steve Henson]
4842
4843 *) In versions up to 0.9.6, RAND_file_name() resorted to file ".rnd"
4844 in the current directory if neither $RANDFILE nor $HOME was set.
4845 RAND_file_name() in 0.9.6a returned NULL in this case. This has
4846 caused some confusion to Windows users who haven't defined $HOME.
4847 Thus RAND_file_name() is changed again: e_os.h can define a
4848 DEFAULT_HOME, which will be used if $HOME is not set.
4849 For Windows, we use "C:"; on other platforms, we still require
4850 environment variables.
4851
4852 *) Move 'if (!initialized) RAND_poll()' into regions protected by
4853 CRYPTO_LOCK_RAND. This is not strictly necessary, but avoids
4854 having multiple threads call RAND_poll() concurrently.
4855 [Bodo Moeller]
4856
4857 *) In crypto/rand/md_rand.c, replace 'add_do_not_lock' flag by a
4858 combination of a flag and a thread ID variable.
4859 Otherwise while one thread is in ssleay_rand_bytes (which sets the
4860 flag), *other* threads can enter ssleay_add_bytes without obeying
4861 the CRYPTO_LOCK_RAND lock (and may even illegally release the lock
4862 that they do not hold after the first thread unsets add_do_not_lock).
4863 [Bodo Moeller]
4864
4865 *) Change bctest again: '-x' expressions are not available in all
4866 versions of 'test'.
4867 [Bodo Moeller]
4868
4869 Changes between 0.9.6 and 0.9.6a [5 Apr 2001]
4870
4871 *) Fix a couple of memory leaks in PKCS7_dataDecode()
4872 [Steve Henson, reported by Heyun Zheng <hzheng@atdsprint.com>]
4873
4874 *) Change Configure and Makefiles to provide EXE_EXT, which will contain
4875 the default extension for executables, if any. Also, make the perl
4876 scripts that use symlink() to test if it really exists and use "cp"
4877 if it doesn't. All this made OpenSSL compilable and installable in
4878 CygWin.
4879 [Richard Levitte]
4880
4881 *) Fix for asn1_GetSequence() for indefinite length constructed data.
4882 If SEQUENCE is length is indefinite just set c->slen to the total
4883 amount of data available.
4884 [Steve Henson, reported by shige@FreeBSD.org]
4885 [This change does not apply to 0.9.7.]
4886
4887 *) Change bctest to avoid here-documents inside command substitution
4888 (workaround for FreeBSD /bin/sh bug).
4889 For compatibility with Ultrix, avoid shell functions (introduced
4890 in the bctest version that searches along $PATH).
4891 [Bodo Moeller]
4892
4893 *) Rename 'des_encrypt' to 'des_encrypt1'. This avoids the clashes
4894 with des_encrypt() defined on some operating systems, like Solaris
4895 and UnixWare.
4896 [Richard Levitte]
4897
4898 *) Check the result of RSA-CRT (see D. Boneh, R. DeMillo, R. Lipton:
4899 On the Importance of Eliminating Errors in Cryptographic
4900 Computations, J. Cryptology 14 (2001) 2, 101-119,
4901 http://theory.stanford.edu/~dabo/papers/faults.ps.gz).
4902 [Ulf Moeller]
4903
4904 *) MIPS assembler BIGNUM division bug fix.
4905 [Andy Polyakov]
4906
4907 *) Disabled incorrect Alpha assembler code.
4908 [Richard Levitte]
4909
4910 *) Fix PKCS#7 decode routines so they correctly update the length
4911 after reading an EOC for the EXPLICIT tag.
4912 [Steve Henson]
4913 [This change does not apply to 0.9.7.]
4914
4915 *) Fix bug in PKCS#12 key generation routines. This was triggered
4916 if a 3DES key was generated with a 0 initial byte. Include
4917 PKCS12_BROKEN_KEYGEN compilation option to retain the old
4918 (but broken) behaviour.
4919 [Steve Henson]
4920
4921 *) Enhance bctest to search for a working bc along $PATH and print
4922 it when found.
4923 [Tim Rice <tim@multitalents.net> via Richard Levitte]
4924
4925 *) Fix memory leaks in err.c: free err_data string if necessary;
4926 don't write to the wrong index in ERR_set_error_data.
4927 [Bodo Moeller]
4928
4929 *) Implement ssl23_peek (analogous to ssl23_read), which previously
4930 did not exist.
4931 [Bodo Moeller]
4932
4933 *) Replace rdtsc with _emit statements for VC++ version 5.
4934 [Jeremy Cooper <jeremy@baymoo.org>]
4935
4936 *) Make it possible to reuse SSLv2 sessions.
4937 [Richard Levitte]
4938
4939 *) In copy_email() check for >= 0 as a return value for
4940 X509_NAME_get_index_by_NID() since 0 is a valid index.
4941 [Steve Henson reported by Massimiliano Pala <madwolf@opensca.org>]
4942
4943 *) Avoid coredump with unsupported or invalid public keys by checking if
4944 X509_get_pubkey() fails in PKCS7_verify(). Fix memory leak when
4945 PKCS7_verify() fails with non detached data.
4946 [Steve Henson]
4947
4948 *) Don't use getenv in library functions when run as setuid/setgid.
4949 New function OPENSSL_issetugid().
4950 [Ulf Moeller]
4951
4952 *) Avoid false positives in memory leak detection code (crypto/mem_dbg.c)
4953 due to incorrect handling of multi-threading:
4954
4955 1. Fix timing glitch in the MemCheck_off() portion of CRYPTO_mem_ctrl().
4956
4957 2. Fix logical glitch in is_MemCheck_on() aka CRYPTO_is_mem_check_on().
4958
4959 3. Count how many times MemCheck_off() has been called so that
4960 nested use can be treated correctly. This also avoids
4961 inband-signalling in the previous code (which relied on the
4962 assumption that thread ID 0 is impossible).
4963 [Bodo Moeller]
4964
4965 *) Add "-rand" option also to s_client and s_server.
4966 [Lutz Jaenicke]
4967
4968 *) Fix CPU detection on Irix 6.x.
4969 [Kurt Hockenbury <khockenb@stevens-tech.edu> and
4970 "Bruce W. Forsberg" <bruce.forsberg@baesystems.com>]
4971
4972 *) Fix X509_NAME bug which produced incorrect encoding if X509_NAME
4973 was empty.
4974 [Steve Henson]
4975 [This change does not apply to 0.9.7.]
4976
4977 *) Use the cached encoding of an X509_NAME structure rather than
4978 copying it. This is apparently the reason for the libsafe "errors"
4979 but the code is actually correct.
4980 [Steve Henson]
4981
4982 *) Add new function BN_rand_range(), and fix DSA_sign_setup() to prevent
4983 Bleichenbacher's DSA attack.
4984 Extend BN_[pseudo_]rand: As before, top=1 forces the highest two bits
4985 to be set and top=0 forces the highest bit to be set; top=-1 is new
4986 and leaves the highest bit random.
4987 [Ulf Moeller, Bodo Moeller]
4988
4989 *) In the NCONF_...-based implementations for CONF_... queries
4990 (crypto/conf/conf_lib.c), if the input LHASH is NULL, avoid using
4991 a temporary CONF structure with the data component set to NULL
4992 (which gives segmentation faults in lh_retrieve).
4993 Instead, use NULL for the CONF pointer in CONF_get_string and
4994 CONF_get_number (which may use environment variables) and directly
4995 return NULL from CONF_get_section.
4996 [Bodo Moeller]
4997
4998 *) Fix potential buffer overrun for EBCDIC.
4999 [Ulf Moeller]
5000
5001 *) Tolerate nonRepudiation as being valid for S/MIME signing and certSign
5002 keyUsage if basicConstraints absent for a CA.
5003 [Steve Henson]
5004
5005 *) Make SMIME_write_PKCS7() write mail header values with a format that
5006 is more generally accepted (no spaces before the semicolon), since
5007 some programs can't parse those values properly otherwise. Also make
5008 sure BIO's that break lines after each write do not create invalid
5009 headers.
5010 [Richard Levitte]
5011
5012 *) Make the CRL encoding routines work with empty SEQUENCE OF. The
5013 macros previously used would not encode an empty SEQUENCE OF
5014 and break the signature.
5015 [Steve Henson]
5016 [This change does not apply to 0.9.7.]
5017
5018 *) Zero the premaster secret after deriving the master secret in
5019 DH ciphersuites.
5020 [Steve Henson]
5021
5022 *) Add some EVP_add_digest_alias registrations (as found in
5023 OpenSSL_add_all_digests()) to SSL_library_init()
5024 aka OpenSSL_add_ssl_algorithms(). This provides improved
5025 compatibility with peers using X.509 certificates
5026 with unconventional AlgorithmIdentifier OIDs.
5027 [Bodo Moeller]
5028
5029 *) Fix for Irix with NO_ASM.
5030 ["Bruce W. Forsberg" <bruce.forsberg@baesystems.com>]
5031
5032 *) ./config script fixes.
5033 [Ulf Moeller, Richard Levitte]
5034
5035 *) Fix 'openssl passwd -1'.
5036 [Bodo Moeller]
5037
5038 *) Change PKCS12_key_gen_asc() so it can cope with non null
5039 terminated strings whose length is passed in the passlen
5040 parameter, for example from PEM callbacks. This was done
5041 by adding an extra length parameter to asc2uni().
5042 [Steve Henson, reported by <oddissey@samsung.co.kr>]
5043
5044 *) Fix C code generated by 'openssl dsaparam -C': If a BN_bin2bn
5045 call failed, free the DSA structure.
5046 [Bodo Moeller]
5047
5048 *) Fix to uni2asc() to cope with zero length Unicode strings.
5049 These are present in some PKCS#12 files.
5050 [Steve Henson]
5051
5052 *) Increase s2->wbuf allocation by one byte in ssl2_new (ssl/s2_lib.c).
5053 Otherwise do_ssl_write (ssl/s2_pkt.c) will write beyond buffer limits
5054 when writing a 32767 byte record.
5055 [Bodo Moeller; problem reported by Eric Day <eday@concentric.net>]
5056
5057 *) In RSA_eay_public_{en,ed}crypt and RSA_eay_mod_exp (rsa_eay.c),
5058 obtain lock CRYPTO_LOCK_RSA before setting rsa->_method_mod_{n,p,q}.
5059
5060 (RSA objects have a reference count access to which is protected
5061 by CRYPTO_LOCK_RSA [see rsa_lib.c, s3_srvr.c, ssl_cert.c, ssl_rsa.c],
5062 so they are meant to be shared between threads.)
5063 [Bodo Moeller, Geoff Thorpe; original patch submitted by
5064 "Reddie, Steven" <Steven.Reddie@ca.com>]
5065
5066 *) Fix a deadlock in CRYPTO_mem_leaks().
5067 [Bodo Moeller]
5068
5069 *) Use better test patterns in bntest.
5070 [Ulf Möller]
5071
5072 *) rand_win.c fix for Borland C.
5073 [Ulf Möller]
5074
5075 *) BN_rshift bugfix for n == 0.
5076 [Bodo Moeller]
5077
5078 *) Add a 'bctest' script that checks for some known 'bc' bugs
5079 so that 'make test' does not abort just because 'bc' is broken.
5080 [Bodo Moeller]
5081
5082 *) Store verify_result within SSL_SESSION also for client side to
5083 avoid potential security hole. (Re-used sessions on the client side
5084 always resulted in verify_result==X509_V_OK, not using the original
5085 result of the server certificate verification.)
5086 [Lutz Jaenicke]
5087
5088 *) Fix ssl3_pending: If the record in s->s3->rrec is not of type
5089 SSL3_RT_APPLICATION_DATA, return 0.
5090 Similarly, change ssl2_pending to return 0 if SSL_in_init(s) is true.
5091 [Bodo Moeller]
5092
5093 *) Fix SSL_peek:
5094 Both ssl2_peek and ssl3_peek, which were totally broken in earlier
5095 releases, have been re-implemented by renaming the previous
5096 implementations of ssl2_read and ssl3_read to ssl2_read_internal
5097 and ssl3_read_internal, respectively, and adding 'peek' parameters
5098 to them. The new ssl[23]_{read,peek} functions are calls to
5099 ssl[23]_read_internal with the 'peek' flag set appropriately.
5100 A 'peek' parameter has also been added to ssl3_read_bytes, which
5101 does the actual work for ssl3_read_internal.
5102 [Bodo Moeller]
5103
5104 *) Initialise "ex_data" member of RSA/DSA/DH structures prior to calling
5105 the method-specific "init()" handler. Also clean up ex_data after
5106 calling the method-specific "finish()" handler. Previously, this was
5107 happening the other way round.
5108 [Geoff Thorpe]
5109
5110 *) Increase BN_CTX_NUM (the number of BIGNUMs in a BN_CTX) to 16.
5111 The previous value, 12, was not always sufficient for BN_mod_exp().
5112 [Bodo Moeller]
5113
5114 *) Make sure that shared libraries get the internal name engine with
5115 the full version number and not just 0. This should mark the
5116 shared libraries as not backward compatible. Of course, this should
5117 be changed again when we can guarantee backward binary compatibility.
5118 [Richard Levitte]
5119
5120 *) Fix typo in get_cert_by_subject() in by_dir.c
5121 [Jean-Marc Desperrier <jean-marc.desperrier@certplus.com>]
5122
5123 *) Rework the system to generate shared libraries:
5124
5125 - Make note of the expected extension for the shared libraries and
5126 if there is a need for symbolic links from for example libcrypto.so.0
5127 to libcrypto.so.0.9.7. There is extended info in Configure for
5128 that.
5129
5130 - Make as few rebuilds of the shared libraries as possible.
5131
5132 - Still avoid linking the OpenSSL programs with the shared libraries.
5133
5134 - When installing, install the shared libraries separately from the
5135 static ones.
5136 [Richard Levitte]
5137
5138 *) Fix SSL_CTX_set_read_ahead macro to actually use its argument.
5139
5140 Copy SSL_CTX's read_ahead flag to SSL object directly in SSL_new
5141 and not in SSL_clear because the latter is also used by the
5142 accept/connect functions; previously, the settings made by
5143 SSL_set_read_ahead would be lost during the handshake.
5144 [Bodo Moeller; problems reported by Anders Gertz <gertz@epact.se>]
5145
5146 *) Correct util/mkdef.pl to be selective about disabled algorithms.
5147 Previously, it would create entries for disableed algorithms no
5148 matter what.
5149 [Richard Levitte]
5150
5151 *) Added several new manual pages for SSL_* function.
5152 [Lutz Jaenicke]
5153
5154 Changes between 0.9.5a and 0.9.6 [24 Sep 2000]
5155
5156 *) In ssl23_get_client_hello, generate an error message when faced
5157 with an initial SSL 3.0/TLS record that is too small to contain the
5158 first two bytes of the ClientHello message, i.e. client_version.
5159 (Note that this is a pathologic case that probably has never happened
5160 in real life.) The previous approach was to use the version number
5161 from the record header as a substitute; but our protocol choice
5162 should not depend on that one because it is not authenticated
5163 by the Finished messages.
5164 [Bodo Moeller]
5165
5166 *) More robust randomness gathering functions for Windows.
5167 [Jeffrey Altman <jaltman@columbia.edu>]
5168
5169 *) For compatibility reasons if the flag X509_V_FLAG_ISSUER_CHECK is
5170 not set then we don't setup the error code for issuer check errors
5171 to avoid possibly overwriting other errors which the callback does
5172 handle. If an application does set the flag then we assume it knows
5173 what it is doing and can handle the new informational codes
5174 appropriately.
5175 [Steve Henson]
5176
5177 *) Fix for a nasty bug in ASN1_TYPE handling. ASN1_TYPE is used for
5178 a general "ANY" type, as such it should be able to decode anything
5179 including tagged types. However it didn't check the class so it would
5180 wrongly interpret tagged types in the same way as their universal
5181 counterpart and unknown types were just rejected. Changed so that the
5182 tagged and unknown types are handled in the same way as a SEQUENCE:
5183 that is the encoding is stored intact. There is also a new type
5184 "V_ASN1_OTHER" which is used when the class is not universal, in this
5185 case we have no idea what the actual type is so we just lump them all
5186 together.
5187 [Steve Henson]
5188
5189 *) On VMS, stdout may very well lead to a file that is written to
5190 in a record-oriented fashion. That means that every write() will
5191 write a separate record, which will be read separately by the
5192 programs trying to read from it. This can be very confusing.
5193
5194 The solution is to put a BIO filter in the way that will buffer
5195 text until a linefeed is reached, and then write everything a
5196 line at a time, so every record written will be an actual line,
5197 not chunks of lines and not (usually doesn't happen, but I've
5198 seen it once) several lines in one record. BIO_f_linebuffer() is
5199 the answer.
5200
5201 Currently, it's a VMS-only method, because that's where it has
5202 been tested well enough.
5203 [Richard Levitte]
5204
5205 *) Remove 'optimized' squaring variant in BN_mod_mul_montgomery,
5206 it can return incorrect results.
5207 (Note: The buggy variant was not enabled in OpenSSL 0.9.5a,
5208 but it was in 0.9.6-beta[12].)
5209 [Bodo Moeller]
5210
5211 *) Disable the check for content being present when verifying detached
5212 signatures in pk7_smime.c. Some versions of Netscape (wrongly)
5213 include zero length content when signing messages.
5214 [Steve Henson]
5215
5216 *) New BIO_shutdown_wr macro, which invokes the BIO_C_SHUTDOWN_WR
5217 BIO_ctrl (for BIO pairs).
5218 [Bodo Möller]
5219
5220 *) Add DSO method for VMS.
5221 [Richard Levitte]
5222
5223 *) Bug fix: Montgomery multiplication could produce results with the
5224 wrong sign.
5225 [Ulf Möller]
5226
5227 *) Add RPM specification openssl.spec and modify it to build three
5228 packages. The default package contains applications, application
5229 documentation and run-time libraries. The devel package contains
5230 include files, static libraries and function documentation. The
5231 doc package contains the contents of the doc directory. The original
5232 openssl.spec was provided by Damien Miller <djm@mindrot.org>.
5233 [Richard Levitte]
5234
5235 *) Add a large number of documentation files for many SSL routines.
5236 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>]
5237
5238 *) Add a configuration entry for Sony News 4.
5239 [NAKAJI Hiroyuki <nakaji@tutrp.tut.ac.jp>]
5240
5241 *) Don't set the two most significant bits to one when generating a
5242 random number < q in the DSA library.
5243 [Ulf Möller]
5244
5245 *) New SSL API mode 'SSL_MODE_AUTO_RETRY'. This disables the default
5246 behaviour that SSL_read may result in SSL_ERROR_WANT_READ (even if
5247 the underlying transport is blocking) if a handshake took place.
5248 (The default behaviour is needed by applications such as s_client
5249 and s_server that use select() to determine when to use SSL_read;
5250 but for applications that know in advance when to expect data, it
5251 just makes things more complicated.)
5252 [Bodo Moeller]
5253
5254 *) Add RAND_egd_bytes(), which gives control over the number of bytes read
5255 from EGD.
5256 [Ben Laurie]
5257
5258 *) Add a few more EBCDIC conditionals that make `req' and `x509'
5259 work better on such systems.
5260 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>]
5261
5262 *) Add two demo programs for PKCS12_parse() and PKCS12_create().
5263 Update PKCS12_parse() so it copies the friendlyName and the
5264 keyid to the certificates aux info.
5265 [Steve Henson]
5266
5267 *) Fix bug in PKCS7_verify() which caused an infinite loop
5268 if there was more than one signature.
5269 [Sven Uszpelkat <su@celocom.de>]
5270
5271 *) Major change in util/mkdef.pl to include extra information
5272 about each symbol, as well as presentig variables as well
5273 as functions. This change means that there's n more need
5274 to rebuild the .num files when some algorithms are excluded.
5275 [Richard Levitte]
5276
5277 *) Allow the verify time to be set by an application,
5278 rather than always using the current time.
5279 [Steve Henson]
5280
5281 *) Phase 2 verify code reorganisation. The certificate
5282 verify code now looks up an issuer certificate by a
5283 number of criteria: subject name, authority key id
5284 and key usage. It also verifies self signed certificates
5285 by the same criteria. The main comparison function is
5286 X509_check_issued() which performs these checks.
5287
5288 Lot of changes were necessary in order to support this
5289 without completely rewriting the lookup code.
5290
5291 Authority and subject key identifier are now cached.
5292
5293 The LHASH 'certs' is X509_STORE has now been replaced
5294 by a STACK_OF(X509_OBJECT). This is mainly because an
5295 LHASH can't store or retrieve multiple objects with
5296 the same hash value.
5297
5298 As a result various functions (which were all internal
5299 use only) have changed to handle the new X509_STORE
5300 structure. This will break anything that messed round
5301 with X509_STORE internally.
5302
5303 The functions X509_STORE_add_cert() now checks for an
5304 exact match, rather than just subject name.
5305
5306 The X509_STORE API doesn't directly support the retrieval
5307 of multiple certificates matching a given criteria, however
5308 this can be worked round by performing a lookup first
5309 (which will fill the cache with candidate certificates)
5310 and then examining the cache for matches. This is probably
5311 the best we can do without throwing out X509_LOOKUP
5312 entirely (maybe later...).
5313
5314 The X509_VERIFY_CTX structure has been enhanced considerably.
5315
5316 All certificate lookup operations now go via a get_issuer()
5317 callback. Although this currently uses an X509_STORE it
5318 can be replaced by custom lookups. This is a simple way
5319 to bypass the X509_STORE hackery necessary to make this
5320 work and makes it possible to use more efficient techniques
5321 in future. A very simple version which uses a simple
5322 STACK for its trusted certificate store is also provided
5323 using X509_STORE_CTX_trusted_stack().
5324
5325 The verify_cb() and verify() callbacks now have equivalents
5326 in the X509_STORE_CTX structure.
5327
5328 X509_STORE_CTX also has a 'flags' field which can be used
5329 to customise the verify behaviour.
5330 [Steve Henson]
5331
5332 *) Add new PKCS#7 signing option PKCS7_NOSMIMECAP which
5333 excludes S/MIME capabilities.
5334 [Steve Henson]
5335
5336 *) When a certificate request is read in keep a copy of the
5337 original encoding of the signed data and use it when outputing
5338 again. Signatures then use the original encoding rather than
5339 a decoded, encoded version which may cause problems if the
5340 request is improperly encoded.
5341 [Steve Henson]
5342
5343 *) For consistency with other BIO_puts implementations, call
5344 buffer_write(b, ...) directly in buffer_puts instead of calling
5345 BIO_write(b, ...).
5346
5347 In BIO_puts, increment b->num_write as in BIO_write.
5348 [Peter.Sylvester@EdelWeb.fr]
5349
5350 *) Fix BN_mul_word for the case where the word is 0. (We have to use
5351 BN_zero, we may not return a BIGNUM with an array consisting of
5352 words set to zero.)
5353 [Bodo Moeller]
5354
5355 *) Avoid calling abort() from within the library when problems are
5356 detected, except if preprocessor symbols have been defined
5357 (such as REF_CHECK, BN_DEBUG etc.).
5358 [Bodo Moeller]
5359
5360 *) New openssl application 'rsautl'. This utility can be
5361 used for low level RSA operations. DER public key
5362 BIO/fp routines also added.
5363 [Steve Henson]
5364
5365 *) New Configure entry and patches for compiling on QNX 4.
5366 [Andreas Schneider <andreas@ds3.etech.fh-hamburg.de>]
5367
5368 *) A demo state-machine implementation was sponsored by
5369 Nuron (http://www.nuron.com/) and is now available in
5370 demos/state_machine.
5371 [Ben Laurie]
5372
5373 *) New options added to the 'dgst' utility for signature
5374 generation and verification.
5375 [Steve Henson]
5376
5377 *) Unrecognized PKCS#7 content types are now handled via a
5378 catch all ASN1_TYPE structure. This allows unsupported
5379 types to be stored as a "blob" and an application can
5380 encode and decode it manually.
5381 [Steve Henson]
5382
5383 *) Fix various signed/unsigned issues to make a_strex.c
5384 compile under VC++.
5385 [Oscar Jacobsson <oscar.jacobsson@celocom.com>]
5386
5387 *) ASN1 fixes. i2d_ASN1_OBJECT was not returning the correct
5388 length if passed a buffer. ASN1_INTEGER_to_BN failed
5389 if passed a NULL BN and its argument was negative.
5390 [Steve Henson, pointed out by Sven Heiberg <sven@tartu.cyber.ee>]
5391
5392 *) Modification to PKCS#7 encoding routines to output definite
5393 length encoding. Since currently the whole structures are in
5394 memory there's not real point in using indefinite length
5395 constructed encoding. However if OpenSSL is compiled with
5396 the flag PKCS7_INDEFINITE_ENCODING the old form is used.
5397 [Steve Henson]
5398
5399 *) Added BIO_vprintf() and BIO_vsnprintf().
5400 [Richard Levitte]
5401
5402 *) Added more prefixes to parse for in the the strings written
5403 through a logging bio, to cover all the levels that are available
5404 through syslog. The prefixes are now:
5405
5406 PANIC, EMERG, EMR => LOG_EMERG
5407 ALERT, ALR => LOG_ALERT
5408 CRIT, CRI => LOG_CRIT
5409 ERROR, ERR => LOG_ERR
5410 WARNING, WARN, WAR => LOG_WARNING
5411 NOTICE, NOTE, NOT => LOG_NOTICE
5412 INFO, INF => LOG_INFO
5413 DEBUG, DBG => LOG_DEBUG
5414
5415 and as before, if none of those prefixes are present at the
5416 beginning of the string, LOG_ERR is chosen.
5417
5418 On Win32, the LOG_* levels are mapped according to this:
5419
5420 LOG_EMERG, LOG_ALERT, LOG_CRIT, LOG_ERR => EVENTLOG_ERROR_TYPE
5421 LOG_WARNING => EVENTLOG_WARNING_TYPE
5422 LOG_NOTICE, LOG_INFO, LOG_DEBUG => EVENTLOG_INFORMATION_TYPE
5423
5424 [Richard Levitte]
5425
5426 *) Made it possible to reconfigure with just the configuration
5427 argument "reconf" or "reconfigure". The command line arguments
5428 are stored in Makefile.ssl in the variable CONFIGURE_ARGS,
5429 and are retrieved from there when reconfiguring.
5430 [Richard Levitte]
5431
5432 *) MD4 implemented.
5433 [Assar Westerlund <assar@sics.se>, Richard Levitte]
5434
5435 *) Add the arguments -CAfile and -CApath to the pkcs12 utility.
5436 [Richard Levitte]
5437
5438 *) The obj_dat.pl script was messing up the sorting of object
5439 names. The reason was that it compared the quoted version
5440 of strings as a result "OCSP" > "OCSP Signing" because
5441 " > SPACE. Changed script to store unquoted versions of
5442 names and add quotes on output. It was also omitting some
5443 names from the lookup table if they were given a default
5444 value (that is if SN is missing it is given the same
5445 value as LN and vice versa), these are now added on the
5446 grounds that if an object has a name we should be able to
5447 look it up. Finally added warning output when duplicate
5448 short or long names are found.
5449 [Steve Henson]
5450
5451 *) Changes needed for Tandem NSK.
5452 [Scott Uroff <scott@xypro.com>]
5453
5454 *) Fix SSL 2.0 rollback checking: Due to an off-by-one error in
5455 RSA_padding_check_SSLv23(), special padding was never detected
5456 and thus the SSL 3.0/TLS 1.0 countermeasure against protocol
5457 version rollback attacks was not effective.
5458
5459 In s23_clnt.c, don't use special rollback-attack detection padding
5460 (RSA_SSLV23_PADDING) if SSL 2.0 is the only protocol enabled in the
5461 client; similarly, in s23_srvr.c, don't do the rollback check if
5462 SSL 2.0 is the only protocol enabled in the server.
5463 [Bodo Moeller]
5464
5465 *) Make it possible to get hexdumps of unprintable data with 'openssl
5466 asn1parse'. By implication, the functions ASN1_parse_dump() and
5467 BIO_dump_indent() are added.
5468 [Richard Levitte]
5469
5470 *) New functions ASN1_STRING_print_ex() and X509_NAME_print_ex()
5471 these print out strings and name structures based on various
5472 flags including RFC2253 support and proper handling of
5473 multibyte characters. Added options to the 'x509' utility
5474 to allow the various flags to be set.
5475 [Steve Henson]
5476
5477 *) Various fixes to use ASN1_TIME instead of ASN1_UTCTIME.
5478 Also change the functions X509_cmp_current_time() and
5479 X509_gmtime_adj() work with an ASN1_TIME structure,
5480 this will enable certificates using GeneralizedTime in validity
5481 dates to be checked.
5482 [Steve Henson]
5483
5484 *) Make the NEG_PUBKEY_BUG code (which tolerates invalid
5485 negative public key encodings) on by default,
5486 NO_NEG_PUBKEY_BUG can be set to disable it.
5487 [Steve Henson]
5488
5489 *) New function c2i_ASN1_OBJECT() which acts on ASN1_OBJECT
5490 content octets. An i2c_ASN1_OBJECT is unnecessary because
5491 the encoding can be trivially obtained from the structure.
5492 [Steve Henson]
5493
5494 *) crypto/err.c locking bugfix: Use write locks (CRYPTO_w_[un]lock),
5495 not read locks (CRYPTO_r_[un]lock).
5496 [Bodo Moeller]
5497
5498 *) A first attempt at creating official support for shared
5499 libraries through configuration. I've kept it so the
5500 default is static libraries only, and the OpenSSL programs
5501 are always statically linked for now, but there are
5502 preparations for dynamic linking in place.
5503 This has been tested on Linux and Tru64.
5504 [Richard Levitte]
5505
5506 *) Randomness polling function for Win9x, as described in:
5507 Peter Gutmann, Software Generation of Practically Strong
5508 Random Numbers.
5509 [Ulf Möller]
5510
5511 *) Fix so PRNG is seeded in req if using an already existing
5512 DSA key.
5513 [Steve Henson]
5514
5515 *) New options to smime application. -inform and -outform
5516 allow alternative formats for the S/MIME message including
5517 PEM and DER. The -content option allows the content to be
5518 specified separately. This should allow things like Netscape
5519 form signing output easier to verify.
5520 [Steve Henson]
5521
5522 *) Fix the ASN1 encoding of tags using the 'long form'.
5523 [Steve Henson]
5524
5525 *) New ASN1 functions, i2c_* and c2i_* for INTEGER and BIT
5526 STRING types. These convert content octets to and from the
5527 underlying type. The actual tag and length octets are
5528 already assumed to have been read in and checked. These
5529 are needed because all other string types have virtually
5530 identical handling apart from the tag. By having versions
5531 of the ASN1 functions that just operate on content octets
5532 IMPLICIT tagging can be handled properly. It also allows
5533 the ASN1_ENUMERATED code to be cut down because ASN1_ENUMERATED
5534 and ASN1_INTEGER are identical apart from the tag.
5535 [Steve Henson]
5536
5537 *) Change the handling of OID objects as follows:
5538
5539 - New object identifiers are inserted in objects.txt, following
5540 the syntax given in objects.README.
5541 - objects.pl is used to process obj_mac.num and create a new
5542 obj_mac.h.
5543 - obj_dat.pl is used to create a new obj_dat.h, using the data in
5544 obj_mac.h.
5545
5546 This is currently kind of a hack, and the perl code in objects.pl
5547 isn't very elegant, but it works as I intended. The simplest way
5548 to check that it worked correctly is to look in obj_dat.h and
5549 check the array nid_objs and make sure the objects haven't moved
5550 around (this is important!). Additions are OK, as well as
5551 consistent name changes.
5552 [Richard Levitte]
5553
5554 *) Add BSD-style MD5-based passwords to 'openssl passwd' (option '-1').
5555 [Bodo Moeller]
5556
5557 *) Addition of the command line parameter '-rand file' to 'openssl req'.
5558 The given file adds to whatever has already been seeded into the
5559 random pool through the RANDFILE configuration file option or
5560 environment variable, or the default random state file.
5561 [Richard Levitte]
5562
5563 *) mkstack.pl now sorts each macro group into lexical order.
5564 Previously the output order depended on the order the files
5565 appeared in the directory, resulting in needless rewriting
5566 of safestack.h .
5567 [Steve Henson]
5568
5569 *) Patches to make OpenSSL compile under Win32 again. Mostly
5570 work arounds for the VC++ problem that it treats func() as
5571 func(void). Also stripped out the parts of mkdef.pl that
5572 added extra typesafe functions: these no longer exist.
5573 [Steve Henson]
5574
5575 *) Reorganisation of the stack code. The macros are now all
5576 collected in safestack.h . Each macro is defined in terms of
5577 a "stack macro" of the form SKM_<name>(type, a, b). The
5578 DEBUG_SAFESTACK is now handled in terms of function casts,
5579 this has the advantage of retaining type safety without the
5580 use of additional functions. If DEBUG_SAFESTACK is not defined
5581 then the non typesafe macros are used instead. Also modified the
5582 mkstack.pl script to handle the new form. Needs testing to see
5583 if which (if any) compilers it chokes and maybe make DEBUG_SAFESTACK
5584 the default if no major problems. Similar behaviour for ASN1_SET_OF
5585 and PKCS12_STACK_OF.
5586 [Steve Henson]
5587
5588 *) When some versions of IIS use the 'NET' form of private key the
5589 key derivation algorithm is different. Normally MD5(password) is
5590 used as a 128 bit RC4 key. In the modified case
5591 MD5(MD5(password) + "SGCKEYSALT") is used insted. Added some
5592 new functions i2d_RSA_NET(), d2i_RSA_NET() etc which are the same
5593 as the old Netscape_RSA functions except they have an additional
5594 'sgckey' parameter which uses the modified algorithm. Also added
5595 an -sgckey command line option to the rsa utility. Thanks to
5596 Adrian Peck <bertie@ncipher.com> for posting details of the modified
5597 algorithm to openssl-dev.
5598 [Steve Henson]
5599
5600 *) The evp_local.h macros were using 'c.##kname' which resulted in
5601 invalid expansion on some systems (SCO 5.0.5 for example).
5602 Corrected to 'c.kname'.
5603 [Phillip Porch <root@theporch.com>]
5604
5605 *) New X509_get1_email() and X509_REQ_get1_email() functions that return
5606 a STACK of email addresses from a certificate or request, these look
5607 in the subject name and the subject alternative name extensions and
5608 omit any duplicate addresses.
5609 [Steve Henson]
5610
5611 *) Re-implement BN_mod_exp2_mont using independent (and larger) windows.
5612 This makes DSA verification about 2 % faster.
5613 [Bodo Moeller]
5614
5615 *) Increase maximum window size in BN_mod_exp_... to 6 bits instead of 5
5616 (meaning that now 2^5 values will be precomputed, which is only 4 KB
5617 plus overhead for 1024 bit moduli).
5618 This makes exponentiations about 0.5 % faster for 1024 bit
5619 exponents (as measured by "openssl speed rsa2048").
5620 [Bodo Moeller]
5621
5622 *) Rename memory handling macros to avoid conflicts with other
5623 software:
5624 Malloc => OPENSSL_malloc
5625 Malloc_locked => OPENSSL_malloc_locked
5626 Realloc => OPENSSL_realloc
5627 Free => OPENSSL_free
5628 [Richard Levitte]
5629
5630 *) New function BN_mod_exp_mont_word for small bases (roughly 15%
5631 faster than BN_mod_exp_mont, i.e. 7% for a full DH exchange).
5632 [Bodo Moeller]
5633
5634 *) CygWin32 support.
5635 [John Jarvie <jjarvie@newsguy.com>]
5636
5637 *) The type-safe stack code has been rejigged. It is now only compiled
5638 in when OpenSSL is configured with the DEBUG_SAFESTACK option and
5639 by default all type-specific stack functions are "#define"d back to
5640 standard stack functions. This results in more streamlined output
5641 but retains the type-safety checking possibilities of the original
5642 approach.
5643 [Geoff Thorpe]
5644
5645 *) The STACK code has been cleaned up, and certain type declarations
5646 that didn't make a lot of sense have been brought in line. This has
5647 also involved a cleanup of sorts in safestack.h to more correctly
5648 map type-safe stack functions onto their plain stack counterparts.
5649 This work has also resulted in a variety of "const"ifications of
5650 lots of the code, especially "_cmp" operations which should normally
5651 be prototyped with "const" parameters anyway.
5652 [Geoff Thorpe]
5653
5654 *) When generating bytes for the first time in md_rand.c, 'stir the pool'
5655 by seeding with STATE_SIZE dummy bytes (with zero entropy count).
5656 (The PRNG state consists of two parts, the large pool 'state' and 'md',
5657 where all of 'md' is used each time the PRNG is used, but 'state'
5658 is used only indexed by a cyclic counter. As entropy may not be
5659 well distributed from the beginning, 'md' is important as a
5660 chaining variable. However, the output function chains only half
5661 of 'md', i.e. 80 bits. ssleay_rand_add, on the other hand, chains
5662 all of 'md', and seeding with STATE_SIZE dummy bytes will result
5663 in all of 'state' being rewritten, with the new values depending
5664 on virtually all of 'md'. This overcomes the 80 bit limitation.)
5665 [Bodo Moeller]
5666
5667 *) In ssl/s2_clnt.c and ssl/s3_clnt.c, call ERR_clear_error() when
5668 the handshake is continued after ssl_verify_cert_chain();
5669 otherwise, if SSL_VERIFY_NONE is set, remaining error codes
5670 can lead to 'unexplainable' connection aborts later.
5671 [Bodo Moeller; problem tracked down by Lutz Jaenicke]
5672
5673 *) Major EVP API cipher revision.
5674 Add hooks for extra EVP features. This allows various cipher
5675 parameters to be set in the EVP interface. Support added for variable
5676 key length ciphers via the EVP_CIPHER_CTX_set_key_length() function and
5677 setting of RC2 and RC5 parameters.
5678
5679 Modify EVP_OpenInit() and EVP_SealInit() to cope with variable key length
5680 ciphers.
5681
5682 Remove lots of duplicated code from the EVP library. For example *every*
5683 cipher init() function handles the 'iv' in the same way according to the
5684 cipher mode. They also all do nothing if the 'key' parameter is NULL and
5685 for CFB and OFB modes they zero ctx->num.
5686
5687 New functionality allows removal of S/MIME code RC2 hack.
5688
5689 Most of the routines have the same form and so can be declared in terms
5690 of macros.
5691
5692 By shifting this to the top level EVP_CipherInit() it can be removed from
5693 all individual ciphers. If the cipher wants to handle IVs or keys
5694 differently it can set the EVP_CIPH_CUSTOM_IV or EVP_CIPH_ALWAYS_CALL_INIT
5695 flags.
5696
5697 Change lots of functions like EVP_EncryptUpdate() to now return a
5698 value: although software versions of the algorithms cannot fail
5699 any installed hardware versions can.
5700 [Steve Henson]
5701
5702 *) Implement SSL_OP_TLS_ROLLBACK_BUG: In ssl3_get_client_key_exchange, if
5703 this option is set, tolerate broken clients that send the negotiated
5704 protocol version number instead of the requested protocol version
5705 number.
5706 [Bodo Moeller]
5707
5708 *) Call dh_tmp_cb (set by ..._TMP_DH_CB) with correct 'is_export' flag;
5709 i.e. non-zero for export ciphersuites, zero otherwise.
5710 Previous versions had this flag inverted, inconsistent with
5711 rsa_tmp_cb (..._TMP_RSA_CB).
5712 [Bodo Moeller; problem reported by Amit Chopra]
5713
5714 *) Add missing DSA library text string. Work around for some IIS
5715 key files with invalid SEQUENCE encoding.
5716 [Steve Henson]
5717
5718 *) Add a document (doc/standards.txt) that list all kinds of standards
5719 and so on that are implemented in OpenSSL.
5720 [Richard Levitte]
5721
5722 *) Enhance c_rehash script. Old version would mishandle certificates
5723 with the same subject name hash and wouldn't handle CRLs at all.
5724 Added -fingerprint option to crl utility, to support new c_rehash
5725 features.
5726 [Steve Henson]
5727
5728 *) Eliminate non-ANSI declarations in crypto.h and stack.h.
5729 [Ulf Möller]
5730
5731 *) Fix for SSL server purpose checking. Server checking was
5732 rejecting certificates which had extended key usage present
5733 but no ssl client purpose.
5734 [Steve Henson, reported by Rene Grosser <grosser@hisolutions.com>]
5735
5736 *) Make PKCS#12 code work with no password. The PKCS#12 spec
5737 is a little unclear about how a blank password is handled.
5738 Since the password in encoded as a BMPString with terminating
5739 double NULL a zero length password would end up as just the
5740 double NULL. However no password at all is different and is
5741 handled differently in the PKCS#12 key generation code. NS
5742 treats a blank password as zero length. MSIE treats it as no
5743 password on export: but it will try both on import. We now do
5744 the same: PKCS12_parse() tries zero length and no password if
5745 the password is set to "" or NULL (NULL is now a valid password:
5746 it wasn't before) as does the pkcs12 application.
5747 [Steve Henson]
5748
5749 *) Bugfixes in apps/x509.c: Avoid a memory leak; and don't use
5750 perror when PEM_read_bio_X509_REQ fails, the error message must
5751 be obtained from the error queue.
5752 [Bodo Moeller]
5753
5754 *) Avoid 'thread_hash' memory leak in crypto/err/err.c by freeing
5755 it in ERR_remove_state if appropriate, and change ERR_get_state
5756 accordingly to avoid race conditions (this is necessary because
5757 thread_hash is no longer constant once set).
5758 [Bodo Moeller]
5759
5760 *) Bugfix for linux-elf makefile.one.
5761 [Ulf Möller]
5762
5763 *) RSA_get_default_method() will now cause a default
5764 RSA_METHOD to be chosen if one doesn't exist already.
5765 Previously this was only set during a call to RSA_new()
5766 or RSA_new_method(NULL) meaning it was possible for
5767 RSA_get_default_method() to return NULL.
5768 [Geoff Thorpe]
5769
5770 *) Added native name translation to the existing DSO code
5771 that will convert (if the flag to do so is set) filenames
5772 that are sufficiently small and have no path information
5773 into a canonical native form. Eg. "blah" converted to
5774 "libblah.so" or "blah.dll" etc.
5775 [Geoff Thorpe]
5776
5777 *) New function ERR_error_string_n(e, buf, len) which is like
5778 ERR_error_string(e, buf), but writes at most 'len' bytes
5779 including the 0 terminator. For ERR_error_string_n, 'buf'
5780 may not be NULL.
5781 [Damien Miller <djm@mindrot.org>, Bodo Moeller]
5782
5783 *) CONF library reworked to become more general. A new CONF
5784 configuration file reader "class" is implemented as well as a
5785 new functions (NCONF_*, for "New CONF") to handle it. The now
5786 old CONF_* functions are still there, but are reimplemented to
5787 work in terms of the new functions. Also, a set of functions
5788 to handle the internal storage of the configuration data is
5789 provided to make it easier to write new configuration file
5790 reader "classes" (I can definitely see something reading a
5791 configuration file in XML format, for example), called _CONF_*,
5792 or "the configuration storage API"...
5793
5794 The new configuration file reading functions are:
5795
5796 NCONF_new, NCONF_free, NCONF_load, NCONF_load_fp, NCONF_load_bio,
5797 NCONF_get_section, NCONF_get_string, NCONF_get_numbre
5798
5799 NCONF_default, NCONF_WIN32
5800
5801 NCONF_dump_fp, NCONF_dump_bio
5802
5803 NCONF_default and NCONF_WIN32 are method (or "class") choosers,
5804 NCONF_new creates a new CONF object. This works in the same way
5805 as other interfaces in OpenSSL, like the BIO interface.
5806 NCONF_dump_* dump the internal storage of the configuration file,
5807 which is useful for debugging. All other functions take the same
5808 arguments as the old CONF_* functions wth the exception of the
5809 first that must be a `CONF *' instead of a `LHASH *'.
5810
5811 To make it easer to use the new classes with the old CONF_* functions,
5812 the function CONF_set_default_method is provided.
5813 [Richard Levitte]
5814
5815 *) Add '-tls1' option to 'openssl ciphers', which was already
5816 mentioned in the documentation but had not been implemented.
5817 (This option is not yet really useful because even the additional
5818 experimental TLS 1.0 ciphers are currently treated as SSL 3.0 ciphers.)
5819 [Bodo Moeller]
5820
5821 *) Initial DSO code added into libcrypto for letting OpenSSL (and
5822 OpenSSL-based applications) load shared libraries and bind to
5823 them in a portable way.
5824 [Geoff Thorpe, with contributions from Richard Levitte]
5825
5826 Changes between 0.9.5 and 0.9.5a [1 Apr 2000]
5827
5828 *) Make sure _lrotl and _lrotr are only used with MSVC.
5829
5830 *) Use lock CRYPTO_LOCK_RAND correctly in ssleay_rand_status
5831 (the default implementation of RAND_status).
5832
5833 *) Rename openssl x509 option '-crlext', which was added in 0.9.5,
5834 to '-clrext' (= clear extensions), as intended and documented.
5835 [Bodo Moeller; inconsistency pointed out by Michael Attili
5836 <attili@amaxo.com>]
5837
5838 *) Fix for HMAC. It wasn't zeroing the rest of the block if the key length
5839 was larger than the MD block size.
5840 [Steve Henson, pointed out by Yost William <YostW@tce.com>]
5841
5842 *) Modernise PKCS12_parse() so it uses STACK_OF(X509) for its ca argument
5843 fix a leak when the ca argument was passed as NULL. Stop X509_PUBKEY_set()
5844 using the passed key: if the passed key was a private key the result
5845 of X509_print(), for example, would be to print out all the private key
5846 components.
5847 [Steve Henson]
5848
5849 *) des_quad_cksum() byte order bug fix.
5850 [Ulf Möller, using the problem description in krb4-0.9.7, where
5851 the solution is attributed to Derrick J Brashear <shadow@DEMENTIA.ORG>]
5852
5853 *) Fix so V_ASN1_APP_CHOOSE works again: however its use is strongly
5854 discouraged.
5855 [Steve Henson, pointed out by Brian Korver <briank@cs.stanford.edu>]
5856
5857 *) For easily testing in shell scripts whether some command
5858 'openssl XXX' exists, the new pseudo-command 'openssl no-XXX'
5859 returns with exit code 0 iff no command of the given name is available.
5860 'no-XXX' is printed in this case, 'XXX' otherwise. In both cases,
5861 the output goes to stdout and nothing is printed to stderr.
5862 Additional arguments are always ignored.
5863
5864 Since for each cipher there is a command of the same name,
5865 the 'no-cipher' compilation switches can be tested this way.
5866
5867 ('openssl no-XXX' is not able to detect pseudo-commands such
5868 as 'quit', 'list-XXX-commands', or 'no-XXX' itself.)
5869 [Bodo Moeller]
5870
5871 *) Update test suite so that 'make test' succeeds in 'no-rsa' configuration.
5872 [Bodo Moeller]
5873
5874 *) For SSL_[CTX_]set_tmp_dh, don't create a DH key if SSL_OP_SINGLE_DH_USE
5875 is set; it will be thrown away anyway because each handshake creates
5876 its own key.
5877 ssl_cert_dup, which is used by SSL_new, now copies DH keys in addition
5878 to parameters -- in previous versions (since OpenSSL 0.9.3) the
5879 'default key' from SSL_CTX_set_tmp_dh would always be lost, meanining
5880 you effectivly got SSL_OP_SINGLE_DH_USE when using this macro.
5881 [Bodo Moeller]
5882
5883 *) New s_client option -ign_eof: EOF at stdin is ignored, and
5884 'Q' and 'R' lose their special meanings (quit/renegotiate).
5885 This is part of what -quiet does; unlike -quiet, -ign_eof
5886 does not suppress any output.
5887 [Richard Levitte]
5888
5889 *) Add compatibility options to the purpose and trust code. The
5890 purpose X509_PURPOSE_ANY is "any purpose" which automatically
5891 accepts a certificate or CA, this was the previous behaviour,
5892 with all the associated security issues.
5893
5894 X509_TRUST_COMPAT is the old trust behaviour: only and
5895 automatically trust self signed roots in certificate store. A
5896 new trust setting X509_TRUST_DEFAULT is used to specify that
5897 a purpose has no associated trust setting and it should instead
5898 use the value in the default purpose.
5899 [Steve Henson]
5900
5901 *) Fix the PKCS#8 DSA private key code so it decodes keys again
5902 and fix a memory leak.
5903 [Steve Henson]
5904
5905 *) In util/mkerr.pl (which implements 'make errors'), preserve
5906 reason strings from the previous version of the .c file, as
5907 the default to have only downcase letters (and digits) in
5908 automatically generated reasons codes is not always appropriate.
5909 [Bodo Moeller]
5910
5911 *) In ERR_load_ERR_strings(), build an ERR_LIB_SYS error reason table
5912 using strerror. Previously, ERR_reason_error_string() returned
5913 library names as reason strings for SYSerr; but SYSerr is a special
5914 case where small numbers are errno values, not library numbers.
5915 [Bodo Moeller]
5916
5917 *) Add '-dsaparam' option to 'openssl dhparam' application. This
5918 converts DSA parameters into DH parameters. (When creating parameters,
5919 DSA_generate_parameters is used.)
5920 [Bodo Moeller]
5921
5922 *) Include 'length' (recommended exponent length) in C code generated
5923 by 'openssl dhparam -C'.
5924 [Bodo Moeller]
5925
5926 *) The second argument to set_label in perlasm was already being used
5927 so couldn't be used as a "file scope" flag. Moved to third argument
5928 which was free.
5929 [Steve Henson]
5930
5931 *) In PEM_ASN1_write_bio and some other functions, use RAND_pseudo_bytes
5932 instead of RAND_bytes for encryption IVs and salts.
5933 [Bodo Moeller]
5934
5935 *) Include RAND_status() into RAND_METHOD instead of implementing
5936 it only for md_rand.c Otherwise replacing the PRNG by calling
5937 RAND_set_rand_method would be impossible.
5938 [Bodo Moeller]
5939
5940 *) Don't let DSA_generate_key() enter an infinite loop if the random
5941 number generation fails.
5942 [Bodo Moeller]
5943
5944 *) New 'rand' application for creating pseudo-random output.
5945 [Bodo Moeller]
5946
5947 *) Added configuration support for Linux/IA64
5948 [Rolf Haberrecker <rolf@suse.de>]
5949
5950 *) Assembler module support for Mingw32.
5951 [Ulf Möller]
5952
5953 *) Shared library support for HPUX (in shlib/).
5954 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE> and Anonymous]
5955
5956 *) Shared library support for Solaris gcc.
5957 [Lutz Behnke <behnke@trustcenter.de>]
5958
5959 Changes between 0.9.4 and 0.9.5 [28 Feb 2000]
5960
5961 *) PKCS7_encrypt() was adding text MIME headers twice because they
5962 were added manually and by SMIME_crlf_copy().
5963 [Steve Henson]
5964
5965 *) In bntest.c don't call BN_rand with zero bits argument.
5966 [Steve Henson, pointed out by Andrew W. Gray <agray@iconsinc.com>]
5967
5968 *) BN_mul bugfix: In bn_mul_part_recursion() only the a>a[n] && b>b[n]
5969 case was implemented. This caused BN_div_recp() to fail occasionally.
5970 [Ulf Möller]
5971
5972 *) Add an optional second argument to the set_label() in the perl
5973 assembly language builder. If this argument exists and is set
5974 to 1 it signals that the assembler should use a symbol whose
5975 scope is the entire file, not just the current function. This
5976 is needed with MASM which uses the format label:: for this scope.
5977 [Steve Henson, pointed out by Peter Runestig <peter@runestig.com>]
5978
5979 *) Change the ASN1 types so they are typedefs by default. Before
5980 almost all types were #define'd to ASN1_STRING which was causing
5981 STACK_OF() problems: you couldn't declare STACK_OF(ASN1_UTF8STRING)
5982 for example.
5983 [Steve Henson]
5984
5985 *) Change names of new functions to the new get1/get0 naming
5986 convention: After 'get1', the caller owns a reference count
5987 and has to call ..._free; 'get0' returns a pointer to some
5988 data structure without incrementing reference counters.
5989 (Some of the existing 'get' functions increment a reference
5990 counter, some don't.)
5991 Similarly, 'set1' and 'add1' functions increase reference
5992 counters or duplicate objects.
5993 [Steve Henson]
5994
5995 *) Allow for the possibility of temp RSA key generation failure:
5996 the code used to assume it always worked and crashed on failure.
5997 [Steve Henson]
5998
5999 *) Fix potential buffer overrun problem in BIO_printf().
6000 [Ulf Möller, using public domain code by Patrick Powell; problem
6001 pointed out by David Sacerdote <das33@cornell.edu>]
6002
6003 *) Support EGD <http://www.lothar.com/tech/crypto/>. New functions
6004 RAND_egd() and RAND_status(). In the command line application,
6005 the EGD socket can be specified like a seed file using RANDFILE
6006 or -rand.
6007 [Ulf Möller]
6008
6009 *) Allow the string CERTIFICATE to be tolerated in PKCS#7 structures.
6010 Some CAs (e.g. Verisign) distribute certificates in this form.
6011 [Steve Henson]
6012
6013 *) Remove the SSL_ALLOW_ADH compile option and set the default cipher
6014 list to exclude them. This means that no special compilation option
6015 is needed to use anonymous DH: it just needs to be included in the
6016 cipher list.
6017 [Steve Henson]
6018
6019 *) Change the EVP_MD_CTX_type macro so its meaning consistent with
6020 EVP_MD_type. The old functionality is available in a new macro called
6021 EVP_MD_md(). Change code that uses it and update docs.
6022 [Steve Henson]
6023
6024 *) ..._ctrl functions now have corresponding ..._callback_ctrl functions
6025 where the 'void *' argument is replaced by a function pointer argument.
6026 Previously 'void *' was abused to point to functions, which works on
6027 many platforms, but is not correct. As these functions are usually
6028 called by macros defined in OpenSSL header files, most source code
6029 should work without changes.
6030 [Richard Levitte]
6031
6032 *) <openssl/opensslconf.h> (which is created by Configure) now contains
6033 sections with information on -D... compiler switches used for
6034 compiling the library so that applications can see them. To enable
6035 one of these sections, a pre-processor symbol OPENSSL_..._DEFINES
6036 must be defined. E.g.,
6037 #define OPENSSL_ALGORITHM_DEFINES
6038 #include <openssl/opensslconf.h>
6039 defines all pertinent NO_<algo> symbols, such as NO_IDEA, NO_RSA, etc.
6040 [Richard Levitte, Ulf and Bodo Möller]
6041
6042 *) Bugfix: Tolerate fragmentation and interleaving in the SSL 3/TLS
6043 record layer.
6044 [Bodo Moeller]
6045
6046 *) Change the 'other' type in certificate aux info to a STACK_OF
6047 X509_ALGOR. Although not an AlgorithmIdentifier as such it has
6048 the required ASN1 format: arbitrary types determined by an OID.
6049 [Steve Henson]
6050
6051 *) Add some PEM_write_X509_REQ_NEW() functions and a command line
6052 argument to 'req'. This is not because the function is newer or
6053 better than others it just uses the work 'NEW' in the certificate
6054 request header lines. Some software needs this.
6055 [Steve Henson]
6056
6057 *) Reorganise password command line arguments: now passwords can be
6058 obtained from various sources. Delete the PEM_cb function and make
6059 it the default behaviour: i.e. if the callback is NULL and the
6060 usrdata argument is not NULL interpret it as a null terminated pass
6061 phrase. If usrdata and the callback are NULL then the pass phrase
6062 is prompted for as usual.
6063 [Steve Henson]
6064
6065 *) Add support for the Compaq Atalla crypto accelerator. If it is installed,
6066 the support is automatically enabled. The resulting binaries will
6067 autodetect the card and use it if present.
6068 [Ben Laurie and Compaq Inc.]
6069
6070 *) Work around for Netscape hang bug. This sends certificate request
6071 and server done in one record. Since this is perfectly legal in the
6072 SSL/TLS protocol it isn't a "bug" option and is on by default. See
6073 the bugs/SSLv3 entry for more info.
6074 [Steve Henson]
6075
6076 *) HP-UX tune-up: new unified configs, HP C compiler bug workaround.
6077 [Andy Polyakov]
6078
6079 *) Add -rand argument to smime and pkcs12 applications and read/write
6080 of seed file.
6081 [Steve Henson]
6082
6083 *) New 'passwd' tool for crypt(3) and apr1 password hashes.
6084 [Bodo Moeller]
6085
6086 *) Add command line password options to the remaining applications.
6087 [Steve Henson]
6088
6089 *) Bug fix for BN_div_recp() for numerators with an even number of
6090 bits.
6091 [Ulf Möller]
6092
6093 *) More tests in bntest.c, and changed test_bn output.
6094 [Ulf Möller]
6095
6096 *) ./config recognizes MacOS X now.
6097 [Andy Polyakov]
6098
6099 *) Bug fix for BN_div() when the first words of num and divsor are
6100 equal (it gave wrong results if (rem=(n1-q*d0)&BN_MASK2) < d0).
6101 [Ulf Möller]
6102
6103 *) Add support for various broken PKCS#8 formats, and command line
6104 options to produce them.
6105 [Steve Henson]
6106
6107 *) New functions BN_CTX_start(), BN_CTX_get() and BT_CTX_end() to
6108 get temporary BIGNUMs from a BN_CTX.
6109 [Ulf Möller]
6110
6111 *) Correct return values in BN_mod_exp_mont() and BN_mod_exp2_mont()
6112 for p == 0.
6113 [Ulf Möller]
6114
6115 *) Change the SSLeay_add_all_*() functions to OpenSSL_add_all_*() and
6116 include a #define from the old name to the new. The original intent
6117 was that statically linked binaries could for example just call
6118 SSLeay_add_all_ciphers() to just add ciphers to the table and not
6119 link with digests. This never worked becayse SSLeay_add_all_digests()
6120 and SSLeay_add_all_ciphers() were in the same source file so calling
6121 one would link with the other. They are now in separate source files.
6122 [Steve Henson]
6123
6124 *) Add a new -notext option to 'ca' and a -pubkey option to 'spkac'.
6125 [Steve Henson]
6126
6127 *) Use a less unusual form of the Miller-Rabin primality test (it used
6128 a binary algorithm for exponentiation integrated into the Miller-Rabin
6129 loop, our standard modexp algorithms are faster).
6130 [Bodo Moeller]
6131
6132 *) Support for the EBCDIC character set completed.
6133 [Martin Kraemer <Martin.Kraemer@Mch.SNI.De>]
6134
6135 *) Source code cleanups: use const where appropriate, eliminate casts,
6136 use void * instead of char * in lhash.
6137 [Ulf Möller]
6138
6139 *) Bugfix: ssl3_send_server_key_exchange was not restartable
6140 (the state was not changed to SSL3_ST_SW_KEY_EXCH_B, and because of
6141 this the server could overwrite ephemeral keys that the client
6142 has already seen).
6143 [Bodo Moeller]
6144
6145 *) Turn DSA_is_prime into a macro that calls BN_is_prime,
6146 using 50 iterations of the Rabin-Miller test.
6147
6148 DSA_generate_parameters now uses BN_is_prime_fasttest (with 50
6149 iterations of the Rabin-Miller test as required by the appendix
6150 to FIPS PUB 186[-1]) instead of DSA_is_prime.
6151 As BN_is_prime_fasttest includes trial division, DSA parameter
6152 generation becomes much faster.
6153
6154 This implies a change for the callback functions in DSA_is_prime
6155 and DSA_generate_parameters: The callback function is called once
6156 for each positive witness in the Rabin-Miller test, not just
6157 occasionally in the inner loop; and the parameters to the
6158 callback function now provide an iteration count for the outer
6159 loop rather than for the current invocation of the inner loop.
6160 DSA_generate_parameters additionally can call the callback
6161 function with an 'iteration count' of -1, meaning that a
6162 candidate has passed the trial division test (when q is generated
6163 from an application-provided seed, trial division is skipped).
6164 [Bodo Moeller]
6165
6166 *) New function BN_is_prime_fasttest that optionally does trial
6167 division before starting the Rabin-Miller test and has
6168 an additional BN_CTX * argument (whereas BN_is_prime always
6169 has to allocate at least one BN_CTX).
6170 'callback(1, -1, cb_arg)' is called when a number has passed the
6171 trial division stage.
6172 [Bodo Moeller]
6173
6174 *) Fix for bug in CRL encoding. The validity dates weren't being handled
6175 as ASN1_TIME.
6176 [Steve Henson]
6177
6178 *) New -pkcs12 option to CA.pl script to write out a PKCS#12 file.
6179 [Steve Henson]
6180
6181 *) New function BN_pseudo_rand().
6182 [Ulf Möller]
6183
6184 *) Clean up BN_mod_mul_montgomery(): replace the broken (and unreadable)
6185 bignum version of BN_from_montgomery() with the working code from
6186 SSLeay 0.9.0 (the word based version is faster anyway), and clean up
6187 the comments.
6188 [Ulf Möller]
6189
6190 *) Avoid a race condition in s2_clnt.c (function get_server_hello) that
6191 made it impossible to use the same SSL_SESSION data structure in
6192 SSL2 clients in multiple threads.
6193 [Bodo Moeller]
6194
6195 *) The return value of RAND_load_file() no longer counts bytes obtained
6196 by stat(). RAND_load_file(..., -1) is new and uses the complete file
6197 to seed the PRNG (previously an explicit byte count was required).
6198 [Ulf Möller, Bodo Möller]
6199
6200 *) Clean up CRYPTO_EX_DATA functions, some of these didn't have prototypes
6201 used (char *) instead of (void *) and had casts all over the place.
6202 [Steve Henson]
6203
6204 *) Make BN_generate_prime() return NULL on error if ret!=NULL.
6205 [Ulf Möller]
6206
6207 *) Retain source code compatibility for BN_prime_checks macro:
6208 BN_is_prime(..., BN_prime_checks, ...) now uses
6209 BN_prime_checks_for_size to determine the appropriate number of
6210 Rabin-Miller iterations.
6211 [Ulf Möller]
6212
6213 *) Diffie-Hellman uses "safe" primes: DH_check() return code renamed to
6214 DH_CHECK_P_NOT_SAFE_PRIME.
6215 (Check if this is true? OpenPGP calls them "strong".)
6216 [Ulf Möller]
6217
6218 *) Merge the functionality of "dh" and "gendh" programs into a new program
6219 "dhparam". The old programs are retained for now but will handle DH keys
6220 (instead of parameters) in future.
6221 [Steve Henson]
6222
6223 *) Make the ciphers, s_server and s_client programs check the return values
6224 when a new cipher list is set.
6225 [Steve Henson]
6226
6227 *) Enhance the SSL/TLS cipher mechanism to correctly handle the TLS 56bit
6228 ciphers. Before when the 56bit ciphers were enabled the sorting was
6229 wrong.
6230
6231 The syntax for the cipher sorting has been extended to support sorting by
6232 cipher-strength (using the strength_bits hard coded in the tables).
6233 The new command is "@STRENGTH" (see also doc/apps/ciphers.pod).
6234
6235 Fix a bug in the cipher-command parser: when supplying a cipher command
6236 string with an "undefined" symbol (neither command nor alphanumeric
6237 [A-Za-z0-9], ssl_set_cipher_list used to hang in an endless loop. Now
6238 an error is flagged.
6239
6240 Due to the strength-sorting extension, the code of the
6241 ssl_create_cipher_list() function was completely rearranged. I hope that
6242 the readability was also increased :-)
6243 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>]
6244
6245 *) Minor change to 'x509' utility. The -CAcreateserial option now uses 1
6246 for the first serial number and places 2 in the serial number file. This
6247 avoids problems when the root CA is created with serial number zero and
6248 the first user certificate has the same issuer name and serial number
6249 as the root CA.
6250 [Steve Henson]
6251
6252 *) Fixes to X509_ATTRIBUTE utilities, change the 'req' program so it uses
6253 the new code. Add documentation for this stuff.
6254 [Steve Henson]
6255
6256 *) Changes to X509_ATTRIBUTE utilities. These have been renamed from
6257 X509_*() to X509at_*() on the grounds that they don't handle X509
6258 structures and behave in an analagous way to the X509v3 functions:
6259 they shouldn't be called directly but wrapper functions should be used
6260 instead.
6261
6262 So we also now have some wrapper functions that call the X509at functions
6263 when passed certificate requests. (TO DO: similar things can be done with
6264 PKCS#7 signed and unsigned attributes, PKCS#12 attributes and a few other
6265 things. Some of these need some d2i or i2d and print functionality
6266 because they handle more complex structures.)
6267 [Steve Henson]
6268
6269 *) Add missing #ifndefs that caused missing symbols when building libssl
6270 as a shared library without RSA. Use #ifndef NO_SSL2 instead of
6271 NO_RSA in ssl/s2*.c.
6272 [Kris Kennaway <kris@hub.freebsd.org>, modified by Ulf Möller]
6273
6274 *) Precautions against using the PRNG uninitialized: RAND_bytes() now
6275 has a return value which indicates the quality of the random data
6276 (1 = ok, 0 = not seeded). Also an error is recorded on the thread's
6277 error queue. New function RAND_pseudo_bytes() generates output that is
6278 guaranteed to be unique but not unpredictable. RAND_add is like
6279 RAND_seed, but takes an extra argument for an entropy estimate
6280 (RAND_seed always assumes full entropy).
6281 [Ulf Möller]
6282
6283 *) Do more iterations of Rabin-Miller probable prime test (specifically,
6284 3 for 1024-bit primes, 6 for 512-bit primes, 12 for 256-bit primes
6285 instead of only 2 for all lengths; see BN_prime_checks_for_size definition
6286 in crypto/bn/bn_prime.c for the complete table). This guarantees a
6287 false-positive rate of at most 2^-80 for random input.
6288 [Bodo Moeller]
6289
6290 *) Rewrite ssl3_read_n (ssl/s3_pkt.c) avoiding a couple of bugs.
6291 [Bodo Moeller]
6292
6293 *) New function X509_CTX_rget_chain() (renamed to X509_CTX_get1_chain
6294 in the 0.9.5 release), this returns the chain
6295 from an X509_CTX structure with a dup of the stack and all
6296 the X509 reference counts upped: so the stack will exist
6297 after X509_CTX_cleanup() has been called. Modify pkcs12.c
6298 to use this.
6299
6300 Also make SSL_SESSION_print() print out the verify return
6301 code.
6302 [Steve Henson]
6303
6304 *) Add manpage for the pkcs12 command. Also change the default
6305 behaviour so MAC iteration counts are used unless the new
6306 -nomaciter option is used. This improves file security and
6307 only older versions of MSIE (4.0 for example) need it.
6308 [Steve Henson]
6309
6310 *) Honor the no-xxx Configure options when creating .DEF files.
6311 [Ulf Möller]
6312
6313 *) Add PKCS#10 attributes to field table: challengePassword,
6314 unstructuredName and unstructuredAddress. These are taken from
6315 draft PKCS#9 v2.0 but are compatible with v1.2 provided no
6316 international characters are used.
6317
6318 More changes to X509_ATTRIBUTE code: allow the setting of types
6319 based on strings. Remove the 'loc' parameter when adding
6320 attributes because these will be a SET OF encoding which is sorted
6321 in ASN1 order.
6322 [Steve Henson]
6323
6324 *) Initial changes to the 'req' utility to allow request generation
6325 automation. This will allow an application to just generate a template
6326 file containing all the field values and have req construct the
6327 request.
6328
6329 Initial support for X509_ATTRIBUTE handling. Stacks of these are
6330 used all over the place including certificate requests and PKCS#7
6331 structures. They are currently handled manually where necessary with
6332 some primitive wrappers for PKCS#7. The new functions behave in a
6333 manner analogous to the X509 extension functions: they allow
6334 attributes to be looked up by NID and added.
6335
6336 Later something similar to the X509V3 code would be desirable to
6337 automatically handle the encoding, decoding and printing of the
6338 more complex types. The string types like challengePassword can
6339 be handled by the string table functions.
6340
6341 Also modified the multi byte string table handling. Now there is
6342 a 'global mask' which masks out certain types. The table itself
6343 can use the flag STABLE_NO_MASK to ignore the mask setting: this
6344 is useful when for example there is only one permissible type
6345 (as in countryName) and using the mask might result in no valid
6346 types at all.
6347 [Steve Henson]
6348
6349 *) Clean up 'Finished' handling, and add functions SSL_get_finished and
6350 SSL_get_peer_finished to allow applications to obtain the latest
6351 Finished messages sent to the peer or expected from the peer,
6352 respectively. (SSL_get_peer_finished is usually the Finished message
6353 actually received from the peer, otherwise the protocol will be aborted.)
6354
6355 As the Finished message are message digests of the complete handshake
6356 (with a total of 192 bits for TLS 1.0 and more for SSL 3.0), they can
6357 be used for external authentication procedures when the authentication
6358 provided by SSL/TLS is not desired or is not enough.
6359 [Bodo Moeller]
6360
6361 *) Enhanced support for Alpha Linux is added. Now ./config checks if
6362 the host supports BWX extension and if Compaq C is present on the
6363 $PATH. Just exploiting of the BWX extension results in 20-30%
6364 performance kick for some algorithms, e.g. DES and RC4 to mention
6365 a couple. Compaq C in turn generates ~20% faster code for MD5 and
6366 SHA1.
6367 [Andy Polyakov]
6368
6369 *) Add support for MS "fast SGC". This is arguably a violation of the
6370 SSL3/TLS protocol. Netscape SGC does two handshakes: the first with
6371 weak crypto and after checking the certificate is SGC a second one
6372 with strong crypto. MS SGC stops the first handshake after receiving
6373 the server certificate message and sends a second client hello. Since
6374 a server will typically do all the time consuming operations before
6375 expecting any further messages from the client (server key exchange
6376 is the most expensive) there is little difference between the two.
6377
6378 To get OpenSSL to support MS SGC we have to permit a second client
6379 hello message after we have sent server done. In addition we have to
6380 reset the MAC if we do get this second client hello.
6381 [Steve Henson]
6382
6383 *) Add a function 'd2i_AutoPrivateKey()' this will automatically decide
6384 if a DER encoded private key is RSA or DSA traditional format. Changed
6385 d2i_PrivateKey_bio() to use it. This is only needed for the "traditional"
6386 format DER encoded private key. Newer code should use PKCS#8 format which
6387 has the key type encoded in the ASN1 structure. Added DER private key
6388 support to pkcs8 application.
6389 [Steve Henson]
6390
6391 *) SSL 3/TLS 1 servers now don't request certificates when an anonymous
6392 ciphersuites has been selected (as required by the SSL 3/TLS 1
6393 specifications). Exception: When SSL_VERIFY_FAIL_IF_NO_PEER_CERT
6394 is set, we interpret this as a request to violate the specification
6395 (the worst that can happen is a handshake failure, and 'correct'
6396 behaviour would result in a handshake failure anyway).
6397 [Bodo Moeller]
6398
6399 *) In SSL_CTX_add_session, take into account that there might be multiple
6400 SSL_SESSION structures with the same session ID (e.g. when two threads
6401 concurrently obtain them from an external cache).
6402 The internal cache can handle only one SSL_SESSION with a given ID,
6403 so if there's a conflict, we now throw out the old one to achieve
6404 consistency.
6405 [Bodo Moeller]
6406
6407 *) Add OIDs for idea and blowfish in CBC mode. This will allow both
6408 to be used in PKCS#5 v2.0 and S/MIME. Also add checking to
6409 some routines that use cipher OIDs: some ciphers do not have OIDs
6410 defined and so they cannot be used for S/MIME and PKCS#5 v2.0 for
6411 example.
6412 [Steve Henson]
6413
6414 *) Simplify the trust setting structure and code. Now we just have
6415 two sequences of OIDs for trusted and rejected settings. These will
6416 typically have values the same as the extended key usage extension
6417 and any application specific purposes.
6418
6419 The trust checking code now has a default behaviour: it will just
6420 check for an object with the same NID as the passed id. Functions can
6421 be provided to override either the default behaviour or the behaviour
6422 for a given id. SSL client, server and email already have functions
6423 in place for compatibility: they check the NID and also return "trusted"
6424 if the certificate is self signed.
6425 [Steve Henson]
6426
6427 *) Add d2i,i2d bio/fp functions for PrivateKey: these convert the
6428 traditional format into an EVP_PKEY structure.
6429 [Steve Henson]
6430
6431 *) Add a password callback function PEM_cb() which either prompts for
6432 a password if usr_data is NULL or otherwise assumes it is a null
6433 terminated password. Allow passwords to be passed on command line
6434 environment or config files in a few more utilities.
6435 [Steve Henson]
6436
6437 *) Add a bunch of DER and PEM functions to handle PKCS#8 format private
6438 keys. Add some short names for PKCS#8 PBE algorithms and allow them
6439 to be specified on the command line for the pkcs8 and pkcs12 utilities.
6440 Update documentation.
6441 [Steve Henson]
6442
6443 *) Support for ASN1 "NULL" type. This could be handled before by using
6444 ASN1_TYPE but there wasn't any function that would try to read a NULL
6445 and produce an error if it couldn't. For compatibility we also have
6446 ASN1_NULL_new() and ASN1_NULL_free() functions but these are faked and
6447 don't allocate anything because they don't need to.
6448 [Steve Henson]
6449
6450 *) Initial support for MacOS is now provided. Examine INSTALL.MacOS
6451 for details.
6452 [Andy Polyakov, Roy Woods <roy@centicsystems.ca>]
6453
6454 *) Rebuild of the memory allocation routines used by OpenSSL code and
6455 possibly others as well. The purpose is to make an interface that
6456 provide hooks so anyone can build a separate set of allocation and
6457 deallocation routines to be used by OpenSSL, for example memory
6458 pool implementations, or something else, which was previously hard
6459 since Malloc(), Realloc() and Free() were defined as macros having
6460 the values malloc, realloc and free, respectively (except for Win32
6461 compilations). The same is provided for memory debugging code.
6462 OpenSSL already comes with functionality to find memory leaks, but
6463 this gives people a chance to debug other memory problems.
6464
6465 With these changes, a new set of functions and macros have appeared:
6466
6467 CRYPTO_set_mem_debug_functions() [F]
6468 CRYPTO_get_mem_debug_functions() [F]
6469 CRYPTO_dbg_set_options() [F]
6470 CRYPTO_dbg_get_options() [F]
6471 CRYPTO_malloc_debug_init() [M]
6472
6473 The memory debug functions are NULL by default, unless the library
6474 is compiled with CRYPTO_MDEBUG or friends is defined. If someone
6475 wants to debug memory anyway, CRYPTO_malloc_debug_init() (which
6476 gives the standard debugging functions that come with OpenSSL) or
6477 CRYPTO_set_mem_debug_functions() (tells OpenSSL to use functions
6478 provided by the library user) must be used. When the standard
6479 debugging functions are used, CRYPTO_dbg_set_options can be used to
6480 request additional information:
6481 CRYPTO_dbg_set_options(V_CYRPTO_MDEBUG_xxx) corresponds to setting
6482 the CRYPTO_MDEBUG_xxx macro when compiling the library.
6483
6484 Also, things like CRYPTO_set_mem_functions will always give the
6485 expected result (the new set of functions is used for allocation
6486 and deallocation) at all times, regardless of platform and compiler
6487 options.
6488
6489 To finish it up, some functions that were never use in any other
6490 way than through macros have a new API and new semantic:
6491
6492 CRYPTO_dbg_malloc()
6493 CRYPTO_dbg_realloc()
6494 CRYPTO_dbg_free()
6495
6496 All macros of value have retained their old syntax.
6497 [Richard Levitte and Bodo Moeller]
6498
6499 *) Some S/MIME fixes. The OID for SMIMECapabilities was wrong, the
6500 ordering of SMIMECapabilities wasn't in "strength order" and there
6501 was a missing NULL in the AlgorithmIdentifier for the SHA1 signature
6502 algorithm.
6503 [Steve Henson]
6504
6505 *) Some ASN1 types with illegal zero length encoding (INTEGER,
6506 ENUMERATED and OBJECT IDENTIFIER) choked the ASN1 routines.
6507 [Frans Heymans <fheymans@isaserver.be>, modified by Steve Henson]
6508
6509 *) Merge in my S/MIME library for OpenSSL. This provides a simple
6510 S/MIME API on top of the PKCS#7 code, a MIME parser (with enough
6511 functionality to handle multipart/signed properly) and a utility
6512 called 'smime' to call all this stuff. This is based on code I
6513 originally wrote for Celo who have kindly allowed it to be
6514 included in OpenSSL.
6515 [Steve Henson]
6516
6517 *) Add variants des_set_key_checked and des_set_key_unchecked of
6518 des_set_key (aka des_key_sched). Global variable des_check_key
6519 decides which of these is called by des_set_key; this way
6520 des_check_key behaves as it always did, but applications and
6521 the library itself, which was buggy for des_check_key == 1,
6522 have a cleaner way to pick the version they need.
6523 [Bodo Moeller]
6524
6525 *) New function PKCS12_newpass() which changes the password of a
6526 PKCS12 structure.
6527 [Steve Henson]
6528
6529 *) Modify X509_TRUST and X509_PURPOSE so it also uses a static and
6530 dynamic mix. In both cases the ids can be used as an index into the
6531 table. Also modified the X509_TRUST_add() and X509_PURPOSE_add()
6532 functions so they accept a list of the field values and the
6533 application doesn't need to directly manipulate the X509_TRUST
6534 structure.
6535 [Steve Henson]
6536
6537 *) Modify the ASN1_STRING_TABLE stuff so it also uses bsearch and doesn't
6538 need initialising.
6539 [Steve Henson]
6540
6541 *) Modify the way the V3 extension code looks up extensions. This now
6542 works in a similar way to the object code: we have some "standard"
6543 extensions in a static table which is searched with OBJ_bsearch()
6544 and the application can add dynamic ones if needed. The file
6545 crypto/x509v3/ext_dat.h now has the info: this file needs to be
6546 updated whenever a new extension is added to the core code and kept
6547 in ext_nid order. There is a simple program 'tabtest.c' which checks
6548 this. New extensions are not added too often so this file can readily
6549 be maintained manually.
6550
6551 There are two big advantages in doing things this way. The extensions
6552 can be looked up immediately and no longer need to be "added" using
6553 X509V3_add_standard_extensions(): this function now does nothing.
6554 [Side note: I get *lots* of email saying the extension code doesn't
6555 work because people forget to call this function]
6556 Also no dynamic allocation is done unless new extensions are added:
6557 so if we don't add custom extensions there is no need to call
6558 X509V3_EXT_cleanup().
6559 [Steve Henson]
6560
6561 *) Modify enc utility's salting as follows: make salting the default. Add a
6562 magic header, so unsalted files fail gracefully instead of just decrypting
6563 to garbage. This is because not salting is a big security hole, so people
6564 should be discouraged from doing it.
6565 [Ben Laurie]
6566
6567 *) Fixes and enhancements to the 'x509' utility. It allowed a message
6568 digest to be passed on the command line but it only used this
6569 parameter when signing a certificate. Modified so all relevant
6570 operations are affected by the digest parameter including the
6571 -fingerprint and -x509toreq options. Also -x509toreq choked if a
6572 DSA key was used because it didn't fix the digest.
6573 [Steve Henson]
6574
6575 *) Initial certificate chain verify code. Currently tests the untrusted
6576 certificates for consistency with the verify purpose (which is set
6577 when the X509_STORE_CTX structure is set up) and checks the pathlength.
6578
6579 There is a NO_CHAIN_VERIFY compilation option to keep the old behaviour:
6580 this is because it will reject chains with invalid extensions whereas
6581 every previous version of OpenSSL and SSLeay made no checks at all.
6582
6583 Trust code: checks the root CA for the relevant trust settings. Trust
6584 settings have an initial value consistent with the verify purpose: e.g.
6585 if the verify purpose is for SSL client use it expects the CA to be
6586 trusted for SSL client use. However the default value can be changed to
6587 permit custom trust settings: one example of this would be to only trust
6588 certificates from a specific "secure" set of CAs.
6589
6590 Also added X509_STORE_CTX_new() and X509_STORE_CTX_free() functions
6591 which should be used for version portability: especially since the
6592 verify structure is likely to change more often now.
6593
6594 SSL integration. Add purpose and trust to SSL_CTX and SSL and functions
6595 to set them. If not set then assume SSL clients will verify SSL servers
6596 and vice versa.
6597
6598 Two new options to the verify program: -untrusted allows a set of
6599 untrusted certificates to be passed in and -purpose which sets the
6600 intended purpose of the certificate. If a purpose is set then the
6601 new chain verify code is used to check extension consistency.
6602 [Steve Henson]
6603
6604 *) Support for the authority information access extension.
6605 [Steve Henson]
6606
6607 *) Modify RSA and DSA PEM read routines to transparently handle
6608 PKCS#8 format private keys. New *_PUBKEY_* functions that handle
6609 public keys in a format compatible with certificate
6610 SubjectPublicKeyInfo structures. Unfortunately there were already
6611 functions called *_PublicKey_* which used various odd formats so
6612 these are retained for compatibility: however the DSA variants were
6613 never in a public release so they have been deleted. Changed dsa/rsa
6614 utilities to handle the new format: note no releases ever handled public
6615 keys so we should be OK.
6616
6617 The primary motivation for this change is to avoid the same fiasco
6618 that dogs private keys: there are several incompatible private key
6619 formats some of which are standard and some OpenSSL specific and
6620 require various evil hacks to allow partial transparent handling and
6621 even then it doesn't work with DER formats. Given the option anything
6622 other than PKCS#8 should be dumped: but the other formats have to
6623 stay in the name of compatibility.
6624
6625 With public keys and the benefit of hindsight one standard format
6626 is used which works with EVP_PKEY, RSA or DSA structures: though
6627 it clearly returns an error if you try to read the wrong kind of key.
6628
6629 Added a -pubkey option to the 'x509' utility to output the public key.
6630 Also rename the EVP_PKEY_get_*() to EVP_PKEY_rget_*()
6631 (renamed to EVP_PKEY_get1_*() in the OpenSSL 0.9.5 release) and add
6632 EVP_PKEY_rset_*() functions (renamed to EVP_PKEY_set1_*())
6633 that do the same as the EVP_PKEY_assign_*() except they up the
6634 reference count of the added key (they don't "swallow" the
6635 supplied key).
6636 [Steve Henson]
6637
6638 *) Fixes to crypto/x509/by_file.c the code to read in certificates and
6639 CRLs would fail if the file contained no certificates or no CRLs:
6640 added a new function to read in both types and return the number
6641 read: this means that if none are read it will be an error. The
6642 DER versions of the certificate and CRL reader would always fail
6643 because it isn't possible to mix certificates and CRLs in DER format
6644 without choking one or the other routine. Changed this to just read
6645 a certificate: this is the best we can do. Also modified the code
6646 in apps/verify.c to take notice of return codes: it was previously
6647 attempting to read in certificates from NULL pointers and ignoring
6648 any errors: this is one reason why the cert and CRL reader seemed
6649 to work. It doesn't check return codes from the default certificate
6650 routines: these may well fail if the certificates aren't installed.
6651 [Steve Henson]
6652
6653 *) Code to support otherName option in GeneralName.
6654 [Steve Henson]
6655
6656 *) First update to verify code. Change the verify utility
6657 so it warns if it is passed a self signed certificate:
6658 for consistency with the normal behaviour. X509_verify
6659 has been modified to it will now verify a self signed
6660 certificate if *exactly* the same certificate appears
6661 in the store: it was previously impossible to trust a
6662 single self signed certificate. This means that:
6663 openssl verify ss.pem
6664 now gives a warning about a self signed certificate but
6665 openssl verify -CAfile ss.pem ss.pem
6666 is OK.
6667 [Steve Henson]
6668
6669 *) For servers, store verify_result in SSL_SESSION data structure
6670 (and add it to external session representation).
6671 This is needed when client certificate verifications fails,
6672 but an application-provided verification callback (set by
6673 SSL_CTX_set_cert_verify_callback) allows accepting the session
6674 anyway (i.e. leaves x509_store_ctx->error != X509_V_OK
6675 but returns 1): When the session is reused, we have to set
6676 ssl->verify_result to the appropriate error code to avoid
6677 security holes.
6678 [Bodo Moeller, problem pointed out by Lutz Jaenicke]
6679
6680 *) Fix a bug in the new PKCS#7 code: it didn't consider the
6681 case in PKCS7_dataInit() where the signed PKCS7 structure
6682 didn't contain any existing data because it was being created.
6683 [Po-Cheng Chen <pocheng@nst.com.tw>, slightly modified by Steve Henson]
6684
6685 *) Add a salt to the key derivation routines in enc.c. This
6686 forms the first 8 bytes of the encrypted file. Also add a
6687 -S option to allow a salt to be input on the command line.
6688 [Steve Henson]
6689
6690 *) New function X509_cmp(). Oddly enough there wasn't a function
6691 to compare two certificates. We do this by working out the SHA1
6692 hash and comparing that. X509_cmp() will be needed by the trust
6693 code.
6694 [Steve Henson]
6695
6696 *) SSL_get1_session() is like SSL_get_session(), but increments
6697 the reference count in the SSL_SESSION returned.
6698 [Geoff Thorpe <geoff@eu.c2.net>]
6699
6700 *) Fix for 'req': it was adding a null to request attributes.
6701 Also change the X509_LOOKUP and X509_INFO code to handle
6702 certificate auxiliary information.
6703 [Steve Henson]
6704
6705 *) Add support for 40 and 64 bit RC2 and RC4 algorithms: document
6706 the 'enc' command.
6707 [Steve Henson]
6708
6709 *) Add the possibility to add extra information to the memory leak
6710 detecting output, to form tracebacks, showing from where each
6711 allocation was originated: CRYPTO_push_info("constant string") adds
6712 the string plus current file name and line number to a per-thread
6713 stack, CRYPTO_pop_info() does the obvious, CRYPTO_remove_all_info()
6714 is like calling CYRPTO_pop_info() until the stack is empty.
6715 Also updated memory leak detection code to be multi-thread-safe.
6716 [Richard Levitte]
6717
6718 *) Add options -text and -noout to pkcs7 utility and delete the
6719 encryption options which never did anything. Update docs.
6720 [Steve Henson]
6721
6722 *) Add options to some of the utilities to allow the pass phrase
6723 to be included on either the command line (not recommended on
6724 OSes like Unix) or read from the environment. Update the
6725 manpages and fix a few bugs.
6726 [Steve Henson]
6727
6728 *) Add a few manpages for some of the openssl commands.
6729 [Steve Henson]
6730
6731 *) Fix the -revoke option in ca. It was freeing up memory twice,
6732 leaking and not finding already revoked certificates.
6733 [Steve Henson]
6734
6735 *) Extensive changes to support certificate auxiliary information.
6736 This involves the use of X509_CERT_AUX structure and X509_AUX
6737 functions. An X509_AUX function such as PEM_read_X509_AUX()
6738 can still read in a certificate file in the usual way but it
6739 will also read in any additional "auxiliary information". By
6740 doing things this way a fair degree of compatibility can be
6741 retained: existing certificates can have this information added
6742 using the new 'x509' options.
6743
6744 Current auxiliary information includes an "alias" and some trust
6745 settings. The trust settings will ultimately be used in enhanced
6746 certificate chain verification routines: currently a certificate
6747 can only be trusted if it is self signed and then it is trusted
6748 for all purposes.
6749 [Steve Henson]
6750
6751 *) Fix assembler for Alpha (tested only on DEC OSF not Linux or *BSD).
6752 The problem was that one of the replacement routines had not been working
6753 since SSLeay releases. For now the offending routine has been replaced
6754 with non-optimised assembler. Even so, this now gives around 95%
6755 performance improvement for 1024 bit RSA signs.
6756 [Mark Cox]
6757
6758 *) Hack to fix PKCS#7 decryption when used with some unorthodox RC2
6759 handling. Most clients have the effective key size in bits equal to
6760 the key length in bits: so a 40 bit RC2 key uses a 40 bit (5 byte) key.
6761 A few however don't do this and instead use the size of the decrypted key
6762 to determine the RC2 key length and the AlgorithmIdentifier to determine
6763 the effective key length. In this case the effective key length can still
6764 be 40 bits but the key length can be 168 bits for example. This is fixed
6765 by manually forcing an RC2 key into the EVP_PKEY structure because the
6766 EVP code can't currently handle unusual RC2 key sizes: it always assumes
6767 the key length and effective key length are equal.
6768 [Steve Henson]
6769
6770 *) Add a bunch of functions that should simplify the creation of
6771 X509_NAME structures. Now you should be able to do:
6772 X509_NAME_add_entry_by_txt(nm, "CN", MBSTRING_ASC, "Steve", -1, -1, 0);
6773 and have it automatically work out the correct field type and fill in
6774 the structures. The more adventurous can try:
6775 X509_NAME_add_entry_by_txt(nm, field, MBSTRING_UTF8, str, -1, -1, 0);
6776 and it will (hopefully) work out the correct multibyte encoding.
6777 [Steve Henson]
6778
6779 *) Change the 'req' utility to use the new field handling and multibyte
6780 copy routines. Before the DN field creation was handled in an ad hoc
6781 way in req, ca, and x509 which was rather broken and didn't support
6782 BMPStrings or UTF8Strings. Since some software doesn't implement
6783 BMPStrings or UTF8Strings yet, they can be enabled using the config file
6784 using the dirstring_type option. See the new comment in the default
6785 openssl.cnf for more info.
6786 [Steve Henson]
6787
6788 *) Make crypto/rand/md_rand.c more robust:
6789 - Assure unique random numbers after fork().
6790 - Make sure that concurrent threads access the global counter and
6791 md serializably so that we never lose entropy in them
6792 or use exactly the same state in multiple threads.
6793 Access to the large state is not always serializable because
6794 the additional locking could be a performance killer, and
6795 md should be large enough anyway.
6796 [Bodo Moeller]
6797
6798 *) New file apps/app_rand.c with commonly needed functionality
6799 for handling the random seed file.
6800
6801 Use the random seed file in some applications that previously did not:
6802 ca,
6803 dsaparam -genkey (which also ignored its '-rand' option),
6804 s_client,
6805 s_server,
6806 x509 (when signing).
6807 Except on systems with /dev/urandom, it is crucial to have a random
6808 seed file at least for key creation, DSA signing, and for DH exchanges;
6809 for RSA signatures we could do without one.
6810
6811 gendh and gendsa (unlike genrsa) used to read only the first byte
6812 of each file listed in the '-rand' option. The function as previously
6813 found in genrsa is now in app_rand.c and is used by all programs
6814 that support '-rand'.
6815 [Bodo Moeller]
6816
6817 *) In RAND_write_file, use mode 0600 for creating files;
6818 don't just chmod when it may be too late.
6819 [Bodo Moeller]
6820
6821 *) Report an error from X509_STORE_load_locations
6822 when X509_LOOKUP_load_file or X509_LOOKUP_add_dir failed.
6823 [Bill Perry]
6824
6825 *) New function ASN1_mbstring_copy() this copies a string in either
6826 ASCII, Unicode, Universal (4 bytes per character) or UTF8 format
6827 into an ASN1_STRING type. A mask of permissible types is passed
6828 and it chooses the "minimal" type to use or an error if not type
6829 is suitable.
6830 [Steve Henson]
6831
6832 *) Add function equivalents to the various macros in asn1.h. The old
6833 macros are retained with an M_ prefix. Code inside the library can
6834 use the M_ macros. External code (including the openssl utility)
6835 should *NOT* in order to be "shared library friendly".
6836 [Steve Henson]
6837
6838 *) Add various functions that can check a certificate's extensions
6839 to see if it usable for various purposes such as SSL client,
6840 server or S/MIME and CAs of these types. This is currently
6841 VERY EXPERIMENTAL but will ultimately be used for certificate chain
6842 verification. Also added a -purpose flag to x509 utility to
6843 print out all the purposes.
6844 [Steve Henson]
6845
6846 *) Add a CRYPTO_EX_DATA to X509 certificate structure and associated
6847 functions.
6848 [Steve Henson]
6849
6850 *) New X509V3_{X509,CRL,REVOKED}_get_d2i() functions. These will search
6851 for, obtain and decode and extension and obtain its critical flag.
6852 This allows all the necessary extension code to be handled in a
6853 single function call.
6854 [Steve Henson]
6855
6856 *) RC4 tune-up featuring 30-40% performance improvement on most RISC
6857 platforms. See crypto/rc4/rc4_enc.c for further details.
6858 [Andy Polyakov]
6859
6860 *) New -noout option to asn1parse. This causes no output to be produced
6861 its main use is when combined with -strparse and -out to extract data
6862 from a file (which may not be in ASN.1 format).
6863 [Steve Henson]
6864
6865 *) Fix for pkcs12 program. It was hashing an invalid certificate pointer
6866 when producing the local key id.
6867 [Richard Levitte <levitte@stacken.kth.se>]
6868
6869 *) New option -dhparam in s_server. This allows a DH parameter file to be
6870 stated explicitly. If it is not stated then it tries the first server
6871 certificate file. The previous behaviour hard coded the filename
6872 "server.pem".
6873 [Steve Henson]
6874
6875 *) Add -pubin and -pubout options to the rsa and dsa commands. These allow
6876 a public key to be input or output. For example:
6877 openssl rsa -in key.pem -pubout -out pubkey.pem
6878 Also added necessary DSA public key functions to handle this.
6879 [Steve Henson]
6880
6881 *) Fix so PKCS7_dataVerify() doesn't crash if no certificates are contained
6882 in the message. This was handled by allowing
6883 X509_find_by_issuer_and_serial() to tolerate a NULL passed to it.
6884 [Steve Henson, reported by Sampo Kellomaki <sampo@mail.neuronio.pt>]
6885
6886 *) Fix for bug in d2i_ASN1_bytes(): other ASN1 functions add an extra null
6887 to the end of the strings whereas this didn't. This would cause problems
6888 if strings read with d2i_ASN1_bytes() were later modified.
6889 [Steve Henson, reported by Arne Ansper <arne@ats.cyber.ee>]
6890
6891 *) Fix for base64 decode bug. When a base64 bio reads only one line of
6892 data and it contains EOF it will end up returning an error. This is
6893 caused by input 46 bytes long. The cause is due to the way base64
6894 BIOs find the start of base64 encoded data. They do this by trying a
6895 trial decode on each line until they find one that works. When they
6896 do a flag is set and it starts again knowing it can pass all the
6897 data directly through the decoder. Unfortunately it doesn't reset
6898 the context it uses. This means that if EOF is reached an attempt
6899 is made to pass two EOFs through the context and this causes the
6900 resulting error. This can also cause other problems as well. As is
6901 usual with these problems it takes *ages* to find and the fix is
6902 trivial: move one line.
6903 [Steve Henson, reported by ian@uns.ns.ac.yu (Ivan Nejgebauer) ]
6904
6905 *) Ugly workaround to get s_client and s_server working under Windows. The
6906 old code wouldn't work because it needed to select() on sockets and the
6907 tty (for keypresses and to see if data could be written). Win32 only
6908 supports select() on sockets so we select() with a 1s timeout on the
6909 sockets and then see if any characters are waiting to be read, if none
6910 are present then we retry, we also assume we can always write data to
6911 the tty. This isn't nice because the code then blocks until we've
6912 received a complete line of data and it is effectively polling the
6913 keyboard at 1s intervals: however it's quite a bit better than not
6914 working at all :-) A dedicated Windows application might handle this
6915 with an event loop for example.
6916 [Steve Henson]
6917
6918 *) Enhance RSA_METHOD structure. Now there are two extra methods, rsa_sign
6919 and rsa_verify. When the RSA_FLAGS_SIGN_VER option is set these functions
6920 will be called when RSA_sign() and RSA_verify() are used. This is useful
6921 if rsa_pub_dec() and rsa_priv_enc() equivalents are not available.
6922 For this to work properly RSA_public_decrypt() and RSA_private_encrypt()
6923 should *not* be used: RSA_sign() and RSA_verify() must be used instead.
6924 This necessitated the support of an extra signature type NID_md5_sha1
6925 for SSL signatures and modifications to the SSL library to use it instead
6926 of calling RSA_public_decrypt() and RSA_private_encrypt().
6927 [Steve Henson]
6928
6929 *) Add new -verify -CAfile and -CApath options to the crl program, these
6930 will lookup a CRL issuers certificate and verify the signature in a
6931 similar way to the verify program. Tidy up the crl program so it
6932 no longer accesses structures directly. Make the ASN1 CRL parsing a bit
6933 less strict. It will now permit CRL extensions even if it is not
6934 a V2 CRL: this will allow it to tolerate some broken CRLs.
6935 [Steve Henson]
6936
6937 *) Initialize all non-automatic variables each time one of the openssl
6938 sub-programs is started (this is necessary as they may be started
6939 multiple times from the "OpenSSL>" prompt).
6940 [Lennart Bang, Bodo Moeller]
6941
6942 *) Preliminary compilation option RSA_NULL which disables RSA crypto without
6943 removing all other RSA functionality (this is what NO_RSA does). This
6944 is so (for example) those in the US can disable those operations covered
6945 by the RSA patent while allowing storage and parsing of RSA keys and RSA
6946 key generation.
6947 [Steve Henson]
6948
6949 *) Non-copying interface to BIO pairs.
6950 (still largely untested)
6951 [Bodo Moeller]
6952
6953 *) New function ANS1_tag2str() to convert an ASN1 tag to a descriptive
6954 ASCII string. This was handled independently in various places before.
6955 [Steve Henson]
6956
6957 *) New functions UTF8_getc() and UTF8_putc() that parse and generate
6958 UTF8 strings a character at a time.
6959 [Steve Henson]
6960
6961 *) Use client_version from client hello to select the protocol
6962 (s23_srvr.c) and for RSA client key exchange verification
6963 (s3_srvr.c), as required by the SSL 3.0/TLS 1.0 specifications.
6964 [Bodo Moeller]
6965
6966 *) Add various utility functions to handle SPKACs, these were previously
6967 handled by poking round in the structure internals. Added new function
6968 NETSCAPE_SPKI_print() to print out SPKAC and a new utility 'spkac' to
6969 print, verify and generate SPKACs. Based on an original idea from
6970 Massimiliano Pala <madwolf@comune.modena.it> but extensively modified.
6971 [Steve Henson]
6972
6973 *) RIPEMD160 is operational on all platforms and is back in 'make test'.
6974 [Andy Polyakov]
6975
6976 *) Allow the config file extension section to be overwritten on the
6977 command line. Based on an original idea from Massimiliano Pala
6978 <madwolf@comune.modena.it>. The new option is called -extensions
6979 and can be applied to ca, req and x509. Also -reqexts to override
6980 the request extensions in req and -crlexts to override the crl extensions
6981 in ca.
6982 [Steve Henson]
6983
6984 *) Add new feature to the SPKAC handling in ca. Now you can include
6985 the same field multiple times by preceding it by "XXXX." for example:
6986 1.OU="Unit name 1"
6987 2.OU="Unit name 2"
6988 this is the same syntax as used in the req config file.
6989 [Steve Henson]
6990
6991 *) Allow certificate extensions to be added to certificate requests. These
6992 are specified in a 'req_extensions' option of the req section of the
6993 config file. They can be printed out with the -text option to req but
6994 are otherwise ignored at present.
6995 [Steve Henson]
6996
6997 *) Fix a horrible bug in enc_read() in crypto/evp/bio_enc.c: if the first
6998 data read consists of only the final block it would not decrypted because
6999 EVP_CipherUpdate() would correctly report zero bytes had been decrypted.
7000 A misplaced 'break' also meant the decrypted final block might not be
7001 copied until the next read.
7002 [Steve Henson]
7003
7004 *) Initial support for DH_METHOD. Again based on RSA_METHOD. Also added
7005 a few extra parameters to the DH structure: these will be useful if
7006 for example we want the value of 'q' or implement X9.42 DH.
7007 [Steve Henson]
7008
7009 *) Initial support for DSA_METHOD. This is based on the RSA_METHOD and
7010 provides hooks that allow the default DSA functions or functions on a
7011 "per key" basis to be replaced. This allows hardware acceleration and
7012 hardware key storage to be handled without major modification to the
7013 library. Also added low level modexp hooks and CRYPTO_EX structure and
7014 associated functions.
7015 [Steve Henson]
7016
7017 *) Add a new flag to memory BIOs, BIO_FLAG_MEM_RDONLY. This marks the BIO
7018 as "read only": it can't be written to and the buffer it points to will
7019 not be freed. Reading from a read only BIO is much more efficient than
7020 a normal memory BIO. This was added because there are several times when
7021 an area of memory needs to be read from a BIO. The previous method was
7022 to create a memory BIO and write the data to it, this results in two
7023 copies of the data and an O(n^2) reading algorithm. There is a new
7024 function BIO_new_mem_buf() which creates a read only memory BIO from
7025 an area of memory. Also modified the PKCS#7 routines to use read only
7026 memory BIOs.
7027 [Steve Henson]
7028
7029 *) Bugfix: ssl23_get_client_hello did not work properly when called in
7030 state SSL23_ST_SR_CLNT_HELLO_B, i.e. when the first 7 bytes of
7031 a SSLv2-compatible client hello for SSLv3 or TLSv1 could be read,
7032 but a retry condition occured while trying to read the rest.
7033 [Bodo Moeller]
7034
7035 *) The PKCS7_ENC_CONTENT_new() function was setting the content type as
7036 NID_pkcs7_encrypted by default: this was wrong since this should almost
7037 always be NID_pkcs7_data. Also modified the PKCS7_set_type() to handle
7038 the encrypted data type: this is a more sensible place to put it and it
7039 allows the PKCS#12 code to be tidied up that duplicated this
7040 functionality.
7041 [Steve Henson]
7042
7043 *) Changed obj_dat.pl script so it takes its input and output files on
7044 the command line. This should avoid shell escape redirection problems
7045 under Win32.
7046 [Steve Henson]
7047
7048 *) Initial support for certificate extension requests, these are included
7049 in things like Xenroll certificate requests. Included functions to allow
7050 extensions to be obtained and added.
7051 [Steve Henson]
7052
7053 *) -crlf option to s_client and s_server for sending newlines as
7054 CRLF (as required by many protocols).
7055 [Bodo Moeller]
7056
7057 Changes between 0.9.3a and 0.9.4 [09 Aug 1999]
7058
7059 *) Install libRSAglue.a when OpenSSL is built with RSAref.
7060 [Ralf S. Engelschall]
7061
7062 *) A few more ``#ifndef NO_FP_API / #endif'' pairs for consistency.
7063 [Andrija Antonijevic <TheAntony2@bigfoot.com>]
7064
7065 *) Fix -startdate and -enddate (which was missing) arguments to 'ca'
7066 program.
7067 [Steve Henson]
7068
7069 *) New function DSA_dup_DH, which duplicates DSA parameters/keys as
7070 DH parameters/keys (q is lost during that conversion, but the resulting
7071 DH parameters contain its length).
7072
7073 For 1024-bit p, DSA_generate_parameters followed by DSA_dup_DH is
7074 much faster than DH_generate_parameters (which creates parameters
7075 where p = 2*q + 1), and also the smaller q makes DH computations
7076 much more efficient (160-bit exponentiation instead of 1024-bit
7077 exponentiation); so this provides a convenient way to support DHE
7078 ciphersuites in SSL/TLS servers (see ssl/ssltest.c). It is of
7079 utter importance to use
7080 SSL_CTX_set_options(s_ctx, SSL_OP_SINGLE_DH_USE);
7081 or
7082 SSL_set_options(s_ctx, SSL_OP_SINGLE_DH_USE);
7083 when such DH parameters are used, because otherwise small subgroup
7084 attacks may become possible!
7085 [Bodo Moeller]
7086
7087 *) Avoid memory leak in i2d_DHparams.
7088 [Bodo Moeller]
7089
7090 *) Allow the -k option to be used more than once in the enc program:
7091 this allows the same encrypted message to be read by multiple recipients.
7092 [Steve Henson]
7093
7094 *) New function OBJ_obj2txt(buf, buf_len, a, no_name), this converts
7095 an ASN1_OBJECT to a text string. If the "no_name" parameter is set then
7096 it will always use the numerical form of the OID, even if it has a short
7097 or long name.
7098 [Steve Henson]
7099
7100 *) Added an extra RSA flag: RSA_FLAG_EXT_PKEY. Previously the rsa_mod_exp
7101 method only got called if p,q,dmp1,dmq1,iqmp components were present,
7102 otherwise bn_mod_exp was called. In the case of hardware keys for example
7103 no private key components need be present and it might store extra data
7104 in the RSA structure, which cannot be accessed from bn_mod_exp.
7105 By setting RSA_FLAG_EXT_PKEY rsa_mod_exp will always be called for
7106 private key operations.
7107 [Steve Henson]
7108
7109 *) Added support for SPARC Linux.
7110 [Andy Polyakov]
7111
7112 *) pem_password_cb function type incompatibly changed from
7113 typedef int pem_password_cb(char *buf, int size, int rwflag);
7114 to
7115 ....(char *buf, int size, int rwflag, void *userdata);
7116 so that applications can pass data to their callbacks:
7117 The PEM[_ASN1]_{read,write}... functions and macros now take an
7118 additional void * argument, which is just handed through whenever
7119 the password callback is called.
7120 [Damien Miller <dmiller@ilogic.com.au>; tiny changes by Bodo Moeller]
7121
7122 New function SSL_CTX_set_default_passwd_cb_userdata.
7123
7124 Compatibility note: As many C implementations push function arguments
7125 onto the stack in reverse order, the new library version is likely to
7126 interoperate with programs that have been compiled with the old
7127 pem_password_cb definition (PEM_whatever takes some data that
7128 happens to be on the stack as its last argument, and the callback
7129 just ignores this garbage); but there is no guarantee whatsoever that
7130 this will work.
7131
7132 *) The -DPLATFORM="\"$(PLATFORM)\"" definition and the similar -DCFLAGS=...
7133 (both in crypto/Makefile.ssl for use by crypto/cversion.c) caused
7134 problems not only on Windows, but also on some Unix platforms.
7135 To avoid problematic command lines, these definitions are now in an
7136 auto-generated file crypto/buildinf.h (created by crypto/Makefile.ssl
7137 for standard "make" builds, by util/mk1mf.pl for "mk1mf" builds).
7138 [Bodo Moeller]
7139
7140 *) MIPS III/IV assembler module is reimplemented.
7141 [Andy Polyakov]
7142
7143 *) More DES library cleanups: remove references to srand/rand and
7144 delete an unused file.
7145 [Ulf Möller]
7146
7147 *) Add support for the the free Netwide assembler (NASM) under Win32,
7148 since not many people have MASM (ml) and it can be hard to obtain.
7149 This is currently experimental but it seems to work OK and pass all
7150 the tests. Check out INSTALL.W32 for info.
7151 [Steve Henson]
7152
7153 *) Fix memory leaks in s3_clnt.c: All non-anonymous SSL3/TLS1 connections
7154 without temporary keys kept an extra copy of the server key,
7155 and connections with temporary keys did not free everything in case
7156 of an error.
7157 [Bodo Moeller]
7158
7159 *) New function RSA_check_key and new openssl rsa option -check
7160 for verifying the consistency of RSA keys.
7161 [Ulf Moeller, Bodo Moeller]
7162
7163 *) Various changes to make Win32 compile work:
7164 1. Casts to avoid "loss of data" warnings in p5_crpt2.c
7165 2. Change unsigned int to int in b_dump.c to avoid "signed/unsigned
7166 comparison" warnings.
7167 3. Add sk_<TYPE>_sort to DEF file generator and do make update.
7168 [Steve Henson]
7169
7170 *) Add a debugging option to PKCS#5 v2 key generation function: when
7171 you #define DEBUG_PKCS5V2 passwords, salts, iteration counts and
7172 derived keys are printed to stderr.
7173 [Steve Henson]
7174
7175 *) Copy the flags in ASN1_STRING_dup().
7176 [Roman E. Pavlov <pre@mo.msk.ru>]
7177
7178 *) The x509 application mishandled signing requests containing DSA
7179 keys when the signing key was also DSA and the parameters didn't match.
7180
7181 It was supposed to omit the parameters when they matched the signing key:
7182 the verifying software was then supposed to automatically use the CA's
7183 parameters if they were absent from the end user certificate.
7184
7185 Omitting parameters is no longer recommended. The test was also
7186 the wrong way round! This was probably due to unusual behaviour in
7187 EVP_cmp_parameters() which returns 1 if the parameters match.
7188 This meant that parameters were omitted when they *didn't* match and
7189 the certificate was useless. Certificates signed with 'ca' didn't have
7190 this bug.
7191 [Steve Henson, reported by Doug Erickson <Doug.Erickson@Part.NET>]
7192
7193 *) Memory leak checking (-DCRYPTO_MDEBUG) had some problems.
7194 The interface is as follows:
7195 Applications can use
7196 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_ON) aka MemCheck_start(),
7197 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_OFF) aka MemCheck_stop();
7198 "off" is now the default.
7199 The library internally uses
7200 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_DISABLE) aka MemCheck_off(),
7201 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_ENABLE) aka MemCheck_on()
7202 to disable memory-checking temporarily.
7203
7204 Some inconsistent states that previously were possible (and were
7205 even the default) are now avoided.
7206
7207 -DCRYPTO_MDEBUG_TIME is new and additionally stores the current time
7208 with each memory chunk allocated; this is occasionally more helpful
7209 than just having a counter.
7210
7211 -DCRYPTO_MDEBUG_THREAD is also new and adds the thread ID.
7212
7213 -DCRYPTO_MDEBUG_ALL enables all of the above, plus any future
7214 extensions.
7215 [Bodo Moeller]
7216
7217 *) Introduce "mode" for SSL structures (with defaults in SSL_CTX),
7218 which largely parallels "options", but is for changing API behaviour,
7219 whereas "options" are about protocol behaviour.
7220 Initial "mode" flags are:
7221
7222 SSL_MODE_ENABLE_PARTIAL_WRITE Allow SSL_write to report success when
7223 a single record has been written.
7224 SSL_MODE_ACCEPT_MOVING_WRITE_BUFFER Don't insist that SSL_write
7225 retries use the same buffer location.
7226 (But all of the contents must be
7227 copied!)
7228 [Bodo Moeller]
7229
7230 *) Bugfix: SSL_set_options ignored its parameter, only SSL_CTX_set_options
7231 worked.
7232
7233 *) Fix problems with no-hmac etc.
7234 [Ulf Möller, pointed out by Brian Wellington <bwelling@tislabs.com>]
7235
7236 *) New functions RSA_get_default_method(), RSA_set_method() and
7237 RSA_get_method(). These allows replacement of RSA_METHODs without having
7238 to mess around with the internals of an RSA structure.
7239 [Steve Henson]
7240
7241 *) Fix memory leaks in DSA_do_sign and DSA_is_prime.
7242 Also really enable memory leak checks in openssl.c and in some
7243 test programs.
7244 [Chad C. Mulligan, Bodo Moeller]
7245
7246 *) Fix a bug in d2i_ASN1_INTEGER() and i2d_ASN1_INTEGER() which can mess
7247 up the length of negative integers. This has now been simplified to just
7248 store the length when it is first determined and use it later, rather
7249 than trying to keep track of where data is copied and updating it to
7250 point to the end.
7251 [Steve Henson, reported by Brien Wheeler
7252 <bwheeler@authentica-security.com>]
7253
7254 *) Add a new function PKCS7_signatureVerify. This allows the verification
7255 of a PKCS#7 signature but with the signing certificate passed to the
7256 function itself. This contrasts with PKCS7_dataVerify which assumes the
7257 certificate is present in the PKCS#7 structure. This isn't always the
7258 case: certificates can be omitted from a PKCS#7 structure and be
7259 distributed by "out of band" means (such as a certificate database).
7260 [Steve Henson]
7261
7262 *) Complete the PEM_* macros with DECLARE_PEM versions to replace the
7263 function prototypes in pem.h, also change util/mkdef.pl to add the
7264 necessary function names.
7265 [Steve Henson]
7266
7267 *) mk1mf.pl (used by Windows builds) did not properly read the
7268 options set by Configure in the top level Makefile, and Configure
7269 was not even able to write more than one option correctly.
7270 Fixed, now "no-idea no-rc5 -DCRYPTO_MDEBUG" etc. works as intended.
7271 [Bodo Moeller]
7272
7273 *) New functions CONF_load_bio() and CONF_load_fp() to allow a config
7274 file to be loaded from a BIO or FILE pointer. The BIO version will
7275 for example allow memory BIOs to contain config info.
7276 [Steve Henson]
7277
7278 *) New function "CRYPTO_num_locks" that returns CRYPTO_NUM_LOCKS.
7279 Whoever hopes to achieve shared-library compatibility across versions
7280 must use this, not the compile-time macro.
7281 (Exercise 0.9.4: Which is the minimum library version required by
7282 such programs?)
7283 Note: All this applies only to multi-threaded programs, others don't
7284 need locks.
7285 [Bodo Moeller]
7286
7287 *) Add missing case to s3_clnt.c state machine -- one of the new SSL tests
7288 through a BIO pair triggered the default case, i.e.
7289 SSLerr(...,SSL_R_UNKNOWN_STATE).
7290 [Bodo Moeller]
7291
7292 *) New "BIO pair" concept (crypto/bio/bss_bio.c) so that applications
7293 can use the SSL library even if none of the specific BIOs is
7294 appropriate.
7295 [Bodo Moeller]
7296
7297 *) Fix a bug in i2d_DSAPublicKey() which meant it returned the wrong value
7298 for the encoded length.
7299 [Jeon KyoungHo <khjeon@sds.samsung.co.kr>]
7300
7301 *) Add initial documentation of the X509V3 functions.
7302 [Steve Henson]
7303
7304 *) Add a new pair of functions PEM_write_PKCS8PrivateKey() and
7305 PEM_write_bio_PKCS8PrivateKey() that are equivalent to
7306 PEM_write_PrivateKey() and PEM_write_bio_PrivateKey() but use the more
7307 secure PKCS#8 private key format with a high iteration count.
7308 [Steve Henson]
7309
7310 *) Fix determination of Perl interpreter: A perl or perl5
7311 _directory_ in $PATH was also accepted as the interpreter.
7312 [Ralf S. Engelschall]
7313
7314 *) Fix demos/sign/sign.c: well there wasn't anything strictly speaking
7315 wrong with it but it was very old and did things like calling
7316 PEM_ASN1_read() directly and used MD5 for the hash not to mention some
7317 unusual formatting.
7318 [Steve Henson]
7319
7320 *) Fix demos/selfsign.c: it used obsolete and deleted functions, changed
7321 to use the new extension code.
7322 [Steve Henson]
7323
7324 *) Implement the PEM_read/PEM_write functions in crypto/pem/pem_all.c
7325 with macros. This should make it easier to change their form, add extra
7326 arguments etc. Fix a few PEM prototypes which didn't have cipher as a
7327 constant.
7328 [Steve Henson]
7329
7330 *) Add to configuration table a new entry that can specify an alternative
7331 name for unistd.h (for pre-POSIX systems); we need this for NeXTstep,
7332 according to Mark Crispin <MRC@Panda.COM>.
7333 [Bodo Moeller]
7334
7335 #if 0
7336 *) DES CBC did not update the IV. Weird.
7337 [Ben Laurie]
7338 #else
7339 des_cbc_encrypt does not update the IV, but des_ncbc_encrypt does.
7340 Changing the behaviour of the former might break existing programs --
7341 where IV updating is needed, des_ncbc_encrypt can be used.
7342 #endif
7343
7344 *) When bntest is run from "make test" it drives bc to check its
7345 calculations, as well as internally checking them. If an internal check
7346 fails, it needs to cause bc to give a non-zero result or make test carries
7347 on without noticing the failure. Fixed.
7348 [Ben Laurie]
7349
7350 *) DES library cleanups.
7351 [Ulf Möller]
7352
7353 *) Add support for PKCS#5 v2.0 PBE algorithms. This will permit PKCS#8 to be
7354 used with any cipher unlike PKCS#5 v1.5 which can at most handle 64 bit
7355 ciphers. NOTE: although the key derivation function has been verified
7356 against some published test vectors it has not been extensively tested
7357 yet. Added a -v2 "cipher" option to pkcs8 application to allow the use
7358 of v2.0.
7359 [Steve Henson]
7360
7361 *) Instead of "mkdir -p", which is not fully portable, use new
7362 Perl script "util/mkdir-p.pl".
7363 [Bodo Moeller]
7364
7365 *) Rewrite the way password based encryption (PBE) is handled. It used to
7366 assume that the ASN1 AlgorithmIdentifier parameter was a PBEParameter
7367 structure. This was true for the PKCS#5 v1.5 and PKCS#12 PBE algorithms
7368 but doesn't apply to PKCS#5 v2.0 where it can be something else. Now
7369 the 'parameter' field of the AlgorithmIdentifier is passed to the
7370 underlying key generation function so it must do its own ASN1 parsing.
7371 This has also changed the EVP_PBE_CipherInit() function which now has a
7372 'parameter' argument instead of literal salt and iteration count values
7373 and the function EVP_PBE_ALGOR_CipherInit() has been deleted.
7374 [Steve Henson]
7375
7376 *) Support for PKCS#5 v1.5 compatible password based encryption algorithms
7377 and PKCS#8 functionality. New 'pkcs8' application linked to openssl.
7378 Needed to change the PEM_STRING_EVP_PKEY value which was just "PRIVATE
7379 KEY" because this clashed with PKCS#8 unencrypted string. Since this
7380 value was just used as a "magic string" and not used directly its
7381 value doesn't matter.
7382 [Steve Henson]
7383
7384 *) Introduce some semblance of const correctness to BN. Shame C doesn't
7385 support mutable.
7386 [Ben Laurie]
7387
7388 *) "linux-sparc64" configuration (ultrapenguin).
7389 [Ray Miller <ray.miller@oucs.ox.ac.uk>]
7390 "linux-sparc" configuration.
7391 [Christian Forster <fo@hawo.stw.uni-erlangen.de>]
7392
7393 *) config now generates no-xxx options for missing ciphers.
7394 [Ulf Möller]
7395
7396 *) Support the EBCDIC character set (work in progress).
7397 File ebcdic.c not yet included because it has a different license.
7398 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>]
7399
7400 *) Support BS2000/OSD-POSIX.
7401 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>]
7402
7403 *) Make callbacks for key generation use void * instead of char *.
7404 [Ben Laurie]
7405
7406 *) Make S/MIME samples compile (not yet tested).
7407 [Ben Laurie]
7408
7409 *) Additional typesafe stacks.
7410 [Ben Laurie]
7411
7412 *) New configuration variants "bsdi-elf-gcc" (BSD/OS 4.x).
7413 [Bodo Moeller]
7414
7415
7416 Changes between 0.9.3 and 0.9.3a [29 May 1999]
7417
7418 *) New configuration variant "sco5-gcc".
7419
7420 *) Updated some demos.
7421 [Sean O Riordain, Wade Scholine]
7422
7423 *) Add missing BIO_free at exit of pkcs12 application.
7424 [Wu Zhigang]
7425
7426 *) Fix memory leak in conf.c.
7427 [Steve Henson]
7428
7429 *) Updates for Win32 to assembler version of MD5.
7430 [Steve Henson]
7431
7432 *) Set #! path to perl in apps/der_chop to where we found it
7433 instead of using a fixed path.
7434 [Bodo Moeller]
7435
7436 *) SHA library changes for irix64-mips4-cc.
7437 [Andy Polyakov]
7438
7439 *) Improvements for VMS support.
7440 [Richard Levitte]
7441
7442
7443 Changes between 0.9.2b and 0.9.3 [24 May 1999]
7444
7445 *) Bignum library bug fix. IRIX 6 passes "make test" now!
7446 This also avoids the problems with SC4.2 and unpatched SC5.
7447 [Andy Polyakov <appro@fy.chalmers.se>]
7448
7449 *) New functions sk_num, sk_value and sk_set to replace the previous macros.
7450 These are required because of the typesafe stack would otherwise break
7451 existing code. If old code used a structure member which used to be STACK
7452 and is now STACK_OF (for example cert in a PKCS7_SIGNED structure) with
7453 sk_num or sk_value it would produce an error because the num, data members
7454 are not present in STACK_OF. Now it just produces a warning. sk_set
7455 replaces the old method of assigning a value to sk_value
7456 (e.g. sk_value(x, i) = y) which the library used in a few cases. Any code
7457 that does this will no longer work (and should use sk_set instead) but
7458 this could be regarded as a "questionable" behaviour anyway.
7459 [Steve Henson]
7460
7461 *) Fix most of the other PKCS#7 bugs. The "experimental" code can now
7462 correctly handle encrypted S/MIME data.
7463 [Steve Henson]
7464
7465 *) Change type of various DES function arguments from des_cblock
7466 (which means, in function argument declarations, pointer to char)
7467 to des_cblock * (meaning pointer to array with 8 char elements),
7468 which allows the compiler to do more typechecking; it was like
7469 that back in SSLeay, but with lots of ugly casts.
7470
7471 Introduce new type const_des_cblock.
7472 [Bodo Moeller]
7473
7474 *) Reorganise the PKCS#7 library and get rid of some of the more obvious
7475 problems: find RecipientInfo structure that matches recipient certificate
7476 and initialise the ASN1 structures properly based on passed cipher.
7477 [Steve Henson]
7478
7479 *) Belatedly make the BN tests actually check the results.
7480 [Ben Laurie]
7481
7482 *) Fix the encoding and decoding of negative ASN1 INTEGERS and conversion
7483 to and from BNs: it was completely broken. New compilation option
7484 NEG_PUBKEY_BUG to allow for some broken certificates that encode public
7485 key elements as negative integers.
7486 [Steve Henson]
7487
7488 *) Reorganize and speed up MD5.
7489 [Andy Polyakov <appro@fy.chalmers.se>]
7490
7491 *) VMS support.
7492 [Richard Levitte <richard@levitte.org>]
7493
7494 *) New option -out to asn1parse to allow the parsed structure to be
7495 output to a file. This is most useful when combined with the -strparse
7496 option to examine the output of things like OCTET STRINGS.
7497 [Steve Henson]
7498
7499 *) Make SSL library a little more fool-proof by not requiring any longer
7500 that SSL_set_{accept,connect}_state be called before
7501 SSL_{accept,connect} may be used (SSL_set_..._state is omitted
7502 in many applications because usually everything *appeared* to work as
7503 intended anyway -- now it really works as intended).
7504 [Bodo Moeller]
7505
7506 *) Move openssl.cnf out of lib/.
7507 [Ulf Möller]
7508
7509 *) Fix various things to let OpenSSL even pass ``egcc -pipe -O2 -Wall
7510 -Wshadow -Wpointer-arith -Wcast-align -Wmissing-prototypes
7511 -Wmissing-declarations -Wnested-externs -Winline'' with EGCS 1.1.2+
7512 [Ralf S. Engelschall]
7513
7514 *) Various fixes to the EVP and PKCS#7 code. It may now be able to
7515 handle PKCS#7 enveloped data properly.
7516 [Sebastian Akerman <sak@parallelconsulting.com>, modified by Steve]
7517
7518 *) Create a duplicate of the SSL_CTX's CERT in SSL_new instead of
7519 copying pointers. The cert_st handling is changed by this in
7520 various ways (and thus what used to be known as ctx->default_cert
7521 is now called ctx->cert, since we don't resort to s->ctx->[default_]cert
7522 any longer when s->cert does not give us what we need).
7523 ssl_cert_instantiate becomes obsolete by this change.
7524 As soon as we've got the new code right (possibly it already is?),
7525 we have solved a couple of bugs of the earlier code where s->cert
7526 was used as if it could not have been shared with other SSL structures.
7527
7528 Note that using the SSL API in certain dirty ways now will result
7529 in different behaviour than observed with earlier library versions:
7530 Changing settings for an SSL_CTX *ctx after having done s = SSL_new(ctx)
7531 does not influence s as it used to.
7532
7533 In order to clean up things more thoroughly, inside SSL_SESSION
7534 we don't use CERT any longer, but a new structure SESS_CERT
7535 that holds per-session data (if available); currently, this is
7536 the peer's certificate chain and, for clients, the server's certificate
7537 and temporary key. CERT holds only those values that can have
7538 meaningful defaults in an SSL_CTX.
7539 [Bodo Moeller]
7540
7541 *) New function X509V3_EXT_i2d() to create an X509_EXTENSION structure
7542 from the internal representation. Various PKCS#7 fixes: remove some
7543 evil casts and set the enc_dig_alg field properly based on the signing
7544 key type.
7545 [Steve Henson]
7546
7547 *) Allow PKCS#12 password to be set from the command line or the
7548 environment. Let 'ca' get its config file name from the environment
7549 variables "OPENSSL_CONF" or "SSLEAY_CONF" (for consistency with 'req'
7550 and 'x509').
7551 [Steve Henson]
7552
7553 *) Allow certificate policies extension to use an IA5STRING for the
7554 organization field. This is contrary to the PKIX definition but
7555 VeriSign uses it and IE5 only recognises this form. Document 'x509'
7556 extension option.
7557 [Steve Henson]
7558
7559 *) Add PEDANTIC compiler flag to allow compilation with gcc -pedantic,
7560 without disallowing inline assembler and the like for non-pedantic builds.
7561 [Ben Laurie]
7562
7563 *) Support Borland C++ builder.
7564 [Janez Jere <jj@void.si>, modified by Ulf Möller]
7565
7566 *) Support Mingw32.
7567 [Ulf Möller]
7568
7569 *) SHA-1 cleanups and performance enhancements.
7570 [Andy Polyakov <appro@fy.chalmers.se>]
7571
7572 *) Sparc v8plus assembler for the bignum library.
7573 [Andy Polyakov <appro@fy.chalmers.se>]
7574
7575 *) Accept any -xxx and +xxx compiler options in Configure.
7576 [Ulf Möller]
7577
7578 *) Update HPUX configuration.
7579 [Anonymous]
7580
7581 *) Add missing sk_<type>_unshift() function to safestack.h
7582 [Ralf S. Engelschall]
7583
7584 *) New function SSL_CTX_use_certificate_chain_file that sets the
7585 "extra_cert"s in addition to the certificate. (This makes sense
7586 only for "PEM" format files, as chains as a whole are not
7587 DER-encoded.)
7588 [Bodo Moeller]
7589
7590 *) Support verify_depth from the SSL API.
7591 x509_vfy.c had what can be considered an off-by-one-error:
7592 Its depth (which was not part of the external interface)
7593 was actually counting the number of certificates in a chain;
7594 now it really counts the depth.
7595 [Bodo Moeller]
7596
7597 *) Bugfix in crypto/x509/x509_cmp.c: The SSLerr macro was used
7598 instead of X509err, which often resulted in confusing error
7599 messages since the error codes are not globally unique
7600 (e.g. an alleged error in ssl3_accept when a certificate
7601 didn't match the private key).
7602
7603 *) New function SSL_CTX_set_session_id_context that allows to set a default
7604 value (so that you don't need SSL_set_session_id_context for each
7605 connection using the SSL_CTX).
7606 [Bodo Moeller]
7607
7608 *) OAEP decoding bug fix.
7609 [Ulf Möller]
7610
7611 *) Support INSTALL_PREFIX for package builders, as proposed by
7612 David Harris.
7613 [Bodo Moeller]
7614
7615 *) New Configure options "threads" and "no-threads". For systems
7616 where the proper compiler options are known (currently Solaris
7617 and Linux), "threads" is the default.
7618 [Bodo Moeller]
7619
7620 *) New script util/mklink.pl as a faster substitute for util/mklink.sh.
7621 [Bodo Moeller]
7622
7623 *) Install various scripts to $(OPENSSLDIR)/misc, not to
7624 $(INSTALLTOP)/bin -- they shouldn't clutter directories
7625 such as /usr/local/bin.
7626 [Bodo Moeller]
7627
7628 *) "make linux-shared" to build shared libraries.
7629 [Niels Poppe <niels@netbox.org>]
7630
7631 *) New Configure option no-<cipher> (rsa, idea, rc5, ...).
7632 [Ulf Möller]
7633
7634 *) Add the PKCS#12 API documentation to openssl.txt. Preliminary support for
7635 extension adding in x509 utility.
7636 [Steve Henson]
7637
7638 *) Remove NOPROTO sections and error code comments.
7639 [Ulf Möller]
7640
7641 *) Partial rewrite of the DEF file generator to now parse the ANSI
7642 prototypes.
7643 [Steve Henson]
7644
7645 *) New Configure options --prefix=DIR and --openssldir=DIR.
7646 [Ulf Möller]
7647
7648 *) Complete rewrite of the error code script(s). It is all now handled
7649 by one script at the top level which handles error code gathering,
7650 header rewriting and C source file generation. It should be much better
7651 than the old method: it now uses a modified version of Ulf's parser to
7652 read the ANSI prototypes in all header files (thus the old K&R definitions
7653 aren't needed for error creation any more) and do a better job of
7654 translating function codes into names. The old 'ASN1 error code imbedded
7655 in a comment' is no longer necessary and it doesn't use .err files which
7656 have now been deleted. Also the error code call doesn't have to appear all
7657 on one line (which resulted in some large lines...).
7658 [Steve Henson]
7659
7660 *) Change #include filenames from <foo.h> to <openssl/foo.h>.
7661 [Bodo Moeller]
7662
7663 *) Change behaviour of ssl2_read when facing length-0 packets: Don't return
7664 0 (which usually indicates a closed connection), but continue reading.
7665 [Bodo Moeller]
7666
7667 *) Fix some race conditions.
7668 [Bodo Moeller]
7669
7670 *) Add support for CRL distribution points extension. Add Certificate
7671 Policies and CRL distribution points documentation.
7672 [Steve Henson]
7673
7674 *) Move the autogenerated header file parts to crypto/opensslconf.h.
7675 [Ulf Möller]
7676
7677 *) Fix new 56-bit DES export ciphersuites: they were using 7 bytes instead of
7678 8 of keying material. Merlin has also confirmed interop with this fix
7679 between OpenSSL and Baltimore C/SSL 2.0 and J/SSL 2.0.
7680 [Merlin Hughes <merlin@baltimore.ie>]
7681
7682 *) Fix lots of warnings.
7683 [Richard Levitte <levitte@stacken.kth.se>]
7684
7685 *) In add_cert_dir() in crypto/x509/by_dir.c, break out of the loop if
7686 the directory spec didn't end with a LIST_SEPARATOR_CHAR.
7687 [Richard Levitte <levitte@stacken.kth.se>]
7688
7689 *) Fix problems with sizeof(long) == 8.
7690 [Andy Polyakov <appro@fy.chalmers.se>]
7691
7692 *) Change functions to ANSI C.
7693 [Ulf Möller]
7694
7695 *) Fix typos in error codes.
7696 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>, Ulf Möller]
7697
7698 *) Remove defunct assembler files from Configure.
7699 [Ulf Möller]
7700
7701 *) SPARC v8 assembler BIGNUM implementation.
7702 [Andy Polyakov <appro@fy.chalmers.se>]
7703
7704 *) Support for Certificate Policies extension: both print and set.
7705 Various additions to support the r2i method this uses.
7706 [Steve Henson]
7707
7708 *) A lot of constification, and fix a bug in X509_NAME_oneline() that could
7709 return a const string when you are expecting an allocated buffer.
7710 [Ben Laurie]
7711
7712 *) Add support for ASN1 types UTF8String and VISIBLESTRING, also the CHOICE
7713 types DirectoryString and DisplayText.
7714 [Steve Henson]
7715
7716 *) Add code to allow r2i extensions to access the configuration database,
7717 add an LHASH database driver and add several ctx helper functions.
7718 [Steve Henson]
7719
7720 *) Fix an evil bug in bn_expand2() which caused various BN functions to
7721 fail when they extended the size of a BIGNUM.
7722 [Steve Henson]
7723
7724 *) Various utility functions to handle SXNet extension. Modify mkdef.pl to
7725 support typesafe stack.
7726 [Steve Henson]
7727
7728 *) Fix typo in SSL_[gs]et_options().
7729 [Nils Frostberg <nils@medcom.se>]
7730
7731 *) Delete various functions and files that belonged to the (now obsolete)
7732 old X509V3 handling code.
7733 [Steve Henson]
7734
7735 *) New Configure option "rsaref".
7736 [Ulf Möller]
7737
7738 *) Don't auto-generate pem.h.
7739 [Bodo Moeller]
7740
7741 *) Introduce type-safe ASN.1 SETs.
7742 [Ben Laurie]
7743
7744 *) Convert various additional casted stacks to type-safe STACK_OF() variants.
7745 [Ben Laurie, Ralf S. Engelschall, Steve Henson]
7746
7747 *) Introduce type-safe STACKs. This will almost certainly break lots of code
7748 that links with OpenSSL (well at least cause lots of warnings), but fear
7749 not: the conversion is trivial, and it eliminates loads of evil casts. A
7750 few STACKed things have been converted already. Feel free to convert more.
7751 In the fullness of time, I'll do away with the STACK type altogether.
7752 [Ben Laurie]
7753
7754 *) Add `openssl ca -revoke <certfile>' facility which revokes a certificate
7755 specified in <certfile> by updating the entry in the index.txt file.
7756 This way one no longer has to edit the index.txt file manually for
7757 revoking a certificate. The -revoke option does the gory details now.
7758 [Massimiliano Pala <madwolf@openca.org>, Ralf S. Engelschall]
7759
7760 *) Fix `openssl crl -noout -text' combination where `-noout' killed the
7761 `-text' option at all and this way the `-noout -text' combination was
7762 inconsistent in `openssl crl' with the friends in `openssl x509|rsa|dsa'.
7763 [Ralf S. Engelschall]
7764
7765 *) Make sure a corresponding plain text error message exists for the
7766 X509_V_ERR_CERT_REVOKED/23 error number which can occur when a
7767 verify callback function determined that a certificate was revoked.
7768 [Ralf S. Engelschall]
7769
7770 *) Bugfix: In test/testenc, don't test "openssl <cipher>" for
7771 ciphers that were excluded, e.g. by -DNO_IDEA. Also, test
7772 all available cipers including rc5, which was forgotten until now.
7773 In order to let the testing shell script know which algorithms
7774 are available, a new (up to now undocumented) command
7775 "openssl list-cipher-commands" is used.
7776 [Bodo Moeller]
7777
7778 *) Bugfix: s_client occasionally would sleep in select() when
7779 it should have checked SSL_pending() first.
7780 [Bodo Moeller]
7781
7782 *) New functions DSA_do_sign and DSA_do_verify to provide access to
7783 the raw DSA values prior to ASN.1 encoding.
7784 [Ulf Möller]
7785
7786 *) Tweaks to Configure
7787 [Niels Poppe <niels@netbox.org>]
7788
7789 *) Add support for PKCS#5 v2.0 ASN1 PBES2 structures. No other support,
7790 yet...
7791 [Steve Henson]
7792
7793 *) New variables $(RANLIB) and $(PERL) in the Makefiles.
7794 [Ulf Möller]
7795
7796 *) New config option to avoid instructions that are illegal on the 80386.
7797 The default code is faster, but requires at least a 486.
7798 [Ulf Möller]
7799
7800 *) Got rid of old SSL2_CLIENT_VERSION (inconsistently used) and
7801 SSL2_SERVER_VERSION (not used at all) macros, which are now the
7802 same as SSL2_VERSION anyway.
7803 [Bodo Moeller]
7804
7805 *) New "-showcerts" option for s_client.
7806 [Bodo Moeller]
7807
7808 *) Still more PKCS#12 integration. Add pkcs12 application to openssl
7809 application. Various cleanups and fixes.
7810 [Steve Henson]
7811
7812 *) More PKCS#12 integration. Add new pkcs12 directory with Makefile.ssl and
7813 modify error routines to work internally. Add error codes and PBE init
7814 to library startup routines.
7815 [Steve Henson]
7816
7817 *) Further PKCS#12 integration. Added password based encryption, PKCS#8 and
7818 packing functions to asn1 and evp. Changed function names and error
7819 codes along the way.
7820 [Steve Henson]
7821
7822 *) PKCS12 integration: and so it begins... First of several patches to
7823 slowly integrate PKCS#12 functionality into OpenSSL. Add PKCS#12
7824 objects to objects.h
7825 [Steve Henson]
7826
7827 *) Add a new 'indent' option to some X509V3 extension code. Initial ASN1
7828 and display support for Thawte strong extranet extension.
7829 [Steve Henson]
7830
7831 *) Add LinuxPPC support.
7832 [Jeff Dubrule <igor@pobox.org>]
7833
7834 *) Get rid of redundant BN file bn_mulw.c, and rename bn_div64 to
7835 bn_div_words in alpha.s.
7836 [Hannes Reinecke <H.Reinecke@hw.ac.uk> and Ben Laurie]
7837
7838 *) Make sure the RSA OAEP test is skipped under -DRSAref because
7839 OAEP isn't supported when OpenSSL is built with RSAref.
7840 [Ulf Moeller <ulf@fitug.de>]
7841
7842 *) Move definitions of IS_SET/IS_SEQUENCE inside crypto/asn1/asn1.h
7843 so they no longer are missing under -DNOPROTO.
7844 [Soren S. Jorvang <soren@t.dk>]
7845
7846
7847 Changes between 0.9.1c and 0.9.2b [22 Mar 1999]
7848
7849 *) Make SSL_get_peer_cert_chain() work in servers. Unfortunately, it still
7850 doesn't work when the session is reused. Coming soon!
7851 [Ben Laurie]
7852
7853 *) Fix a security hole, that allows sessions to be reused in the wrong
7854 context thus bypassing client cert protection! All software that uses
7855 client certs and session caches in multiple contexts NEEDS PATCHING to
7856 allow session reuse! A fuller solution is in the works.
7857 [Ben Laurie, problem pointed out by Holger Reif, Bodo Moeller (and ???)]
7858
7859 *) Some more source tree cleanups (removed obsolete files
7860 crypto/bf/asm/bf586.pl, test/test.txt and crypto/sha/asm/f.s; changed
7861 permission on "config" script to be executable) and a fix for the INSTALL
7862 document.
7863 [Ulf Moeller <ulf@fitug.de>]
7864
7865 *) Remove some legacy and erroneous uses of malloc, free instead of
7866 Malloc, Free.
7867 [Lennart Bang <lob@netstream.se>, with minor changes by Steve]
7868
7869 *) Make rsa_oaep_test return non-zero on error.
7870 [Ulf Moeller <ulf@fitug.de>]
7871
7872 *) Add support for native Solaris shared libraries. Configure
7873 solaris-sparc-sc4-pic, make, then run shlib/solaris-sc4.sh. It'd be nice
7874 if someone would make that last step automatic.
7875 [Matthias Loepfe <Matthias.Loepfe@AdNovum.CH>]
7876
7877 *) ctx_size was not built with the right compiler during "make links". Fixed.
7878 [Ben Laurie]
7879
7880 *) Change the meaning of 'ALL' in the cipher list. It now means "everything
7881 except NULL ciphers". This means the default cipher list will no longer
7882 enable NULL ciphers. They need to be specifically enabled e.g. with
7883 the string "DEFAULT:eNULL".
7884 [Steve Henson]
7885
7886 *) Fix to RSA private encryption routines: if p < q then it would
7887 occasionally produce an invalid result. This will only happen with
7888 externally generated keys because OpenSSL (and SSLeay) ensure p > q.
7889 [Steve Henson]
7890
7891 *) Be less restrictive and allow also `perl util/perlpath.pl
7892 /path/to/bin/perl' in addition to `perl util/perlpath.pl /path/to/bin',
7893 because this way one can also use an interpreter named `perl5' (which is
7894 usually the name of Perl 5.xxx on platforms where an Perl 4.x is still
7895 installed as `perl').
7896 [Matthias Loepfe <Matthias.Loepfe@adnovum.ch>]
7897
7898 *) Let util/clean-depend.pl work also with older Perl 5.00x versions.
7899 [Matthias Loepfe <Matthias.Loepfe@adnovum.ch>]
7900
7901 *) Fix Makefile.org so CC,CFLAG etc are passed to 'make links' add
7902 advapi32.lib to Win32 build and change the pem test comparision
7903 to fc.exe (thanks to Ulrich Kroener <kroneru@yahoo.com> for the
7904 suggestion). Fix misplaced ASNI prototypes and declarations in evp.h
7905 and crypto/des/ede_cbcm_enc.c.
7906 [Steve Henson]
7907
7908 *) DES quad checksum was broken on big-endian architectures. Fixed.
7909 [Ben Laurie]
7910
7911 *) Comment out two functions in bio.h that aren't implemented. Fix up the
7912 Win32 test batch file so it (might) work again. The Win32 test batch file
7913 is horrible: I feel ill....
7914 [Steve Henson]
7915
7916 *) Move various #ifdefs around so NO_SYSLOG, NO_DIRENT etc are now selected
7917 in e_os.h. Audit of header files to check ANSI and non ANSI
7918 sections: 10 functions were absent from non ANSI section and not exported
7919 from Windows DLLs. Fixed up libeay.num for new functions.
7920 [Steve Henson]
7921
7922 *) Make `openssl version' output lines consistent.
7923 [Ralf S. Engelschall]
7924
7925 *) Fix Win32 symbol export lists for BIO functions: Added
7926 BIO_get_ex_new_index, BIO_get_ex_num, BIO_get_ex_data and BIO_set_ex_data
7927 to ms/libeay{16,32}.def.
7928 [Ralf S. Engelschall]
7929
7930 *) Second round of fixing the OpenSSL perl/ stuff. It now at least compiled
7931 fine under Unix and passes some trivial tests I've now added. But the
7932 whole stuff is horribly incomplete, so a README.1ST with a disclaimer was
7933 added to make sure no one expects that this stuff really works in the
7934 OpenSSL 0.9.2 release. Additionally I've started to clean the XS sources
7935 up and fixed a few little bugs and inconsistencies in OpenSSL.{pm,xs} and
7936 openssl_bio.xs.
7937 [Ralf S. Engelschall]
7938
7939 *) Fix the generation of two part addresses in perl.
7940 [Kenji Miyake <kenji@miyake.org>, integrated by Ben Laurie]
7941
7942 *) Add config entry for Linux on MIPS.
7943 [John Tobey <jtobey@channel1.com>]
7944
7945 *) Make links whenever Configure is run, unless we are on Windoze.
7946 [Ben Laurie]
7947
7948 *) Permit extensions to be added to CRLs using crl_section in openssl.cnf.
7949 Currently only issuerAltName and AuthorityKeyIdentifier make any sense
7950 in CRLs.
7951 [Steve Henson]
7952
7953 *) Add a useful kludge to allow package maintainers to specify compiler and
7954 other platforms details on the command line without having to patch the
7955 Configure script everytime: One now can use ``perl Configure
7956 <id>:<details>'', i.e. platform ids are allowed to have details appended
7957 to them (seperated by colons). This is treated as there would be a static
7958 pre-configured entry in Configure's %table under key <id> with value
7959 <details> and ``perl Configure <id>'' is called. So, when you want to
7960 perform a quick test-compile under FreeBSD 3.1 with pgcc and without
7961 assembler stuff you can use ``perl Configure "FreeBSD-elf:pgcc:-O6:::"''
7962 now, which overrides the FreeBSD-elf entry on-the-fly.
7963 [Ralf S. Engelschall]
7964
7965 *) Disable new TLS1 ciphersuites by default: they aren't official yet.
7966 [Ben Laurie]
7967
7968 *) Allow DSO flags like -fpic, -fPIC, -KPIC etc. to be specified
7969 on the `perl Configure ...' command line. This way one can compile
7970 OpenSSL libraries with Position Independent Code (PIC) which is needed
7971 for linking it into DSOs.
7972 [Ralf S. Engelschall]
7973
7974 *) Remarkably, export ciphers were totally broken and no-one had noticed!
7975 Fixed.
7976 [Ben Laurie]
7977
7978 *) Cleaned up the LICENSE document: The official contact for any license
7979 questions now is the OpenSSL core team under openssl-core@openssl.org.
7980 And add a paragraph about the dual-license situation to make sure people
7981 recognize that _BOTH_ the OpenSSL license _AND_ the SSLeay license apply
7982 to the OpenSSL toolkit.
7983 [Ralf S. Engelschall]
7984
7985 *) General source tree makefile cleanups: Made `making xxx in yyy...'
7986 display consistent in the source tree and replaced `/bin/rm' by `rm'.
7987 Additonally cleaned up the `make links' target: Remove unnecessary
7988 semicolons, subsequent redundant removes, inline point.sh into mklink.sh
7989 to speed processing and no longer clutter the display with confusing
7990 stuff. Instead only the actually done links are displayed.
7991 [Ralf S. Engelschall]
7992
7993 *) Permit null encryption ciphersuites, used for authentication only. It used
7994 to be necessary to set the preprocessor define SSL_ALLOW_ENULL to do this.
7995 It is now necessary to set SSL_FORBID_ENULL to prevent the use of null
7996 encryption.
7997 [Ben Laurie]
7998
7999 *) Add a bunch of fixes to the PKCS#7 stuff. It used to sometimes reorder
8000 signed attributes when verifying signatures (this would break them),
8001 the detached data encoding was wrong and public keys obtained using
8002 X509_get_pubkey() weren't freed.
8003 [Steve Henson]
8004
8005 *) Add text documentation for the BUFFER functions. Also added a work around
8006 to a Win95 console bug. This was triggered by the password read stuff: the
8007 last character typed gets carried over to the next fread(). If you were
8008 generating a new cert request using 'req' for example then the last
8009 character of the passphrase would be CR which would then enter the first
8010 field as blank.
8011 [Steve Henson]
8012
8013 *) Added the new `Includes OpenSSL Cryptography Software' button as
8014 doc/openssl_button.{gif,html} which is similar in style to the old SSLeay
8015 button and can be used by applications based on OpenSSL to show the
8016 relationship to the OpenSSL project.
8017 [Ralf S. Engelschall]
8018
8019 *) Remove confusing variables in function signatures in files
8020 ssl/ssl_lib.c and ssl/ssl.h.
8021 [Lennart Bong <lob@kulthea.stacken.kth.se>]
8022
8023 *) Don't install bss_file.c under PREFIX/include/
8024 [Lennart Bong <lob@kulthea.stacken.kth.se>]
8025
8026 *) Get the Win32 compile working again. Modify mkdef.pl so it can handle
8027 functions that return function pointers and has support for NT specific
8028 stuff. Fix mk1mf.pl and VC-32.pl to support NT differences also. Various
8029 #ifdef WIN32 and WINNTs sprinkled about the place and some changes from
8030 unsigned to signed types: this was killing the Win32 compile.
8031 [Steve Henson]
8032
8033 *) Add new certificate file to stack functions,
8034 SSL_add_dir_cert_subjects_to_stack() and
8035 SSL_add_file_cert_subjects_to_stack(). These largely supplant
8036 SSL_load_client_CA_file(), and can be used to add multiple certs easily
8037 to a stack (usually this is then handed to SSL_CTX_set_client_CA_list()).
8038 This means that Apache-SSL and similar packages don't have to mess around
8039 to add as many CAs as they want to the preferred list.
8040 [Ben Laurie]
8041
8042 *) Experiment with doxygen documentation. Currently only partially applied to
8043 ssl/ssl_lib.c.
8044 See http://www.stack.nl/~dimitri/doxygen/index.html, and run doxygen with
8045 openssl.doxy as the configuration file.
8046 [Ben Laurie]
8047
8048 *) Get rid of remaining C++-style comments which strict C compilers hate.
8049 [Ralf S. Engelschall, pointed out by Carlos Amengual]
8050
8051 *) Changed BN_RECURSION in bn_mont.c to BN_RECURSION_MONT so it is not
8052 compiled in by default: it has problems with large keys.
8053 [Steve Henson]
8054
8055 *) Add a bunch of SSL_xxx() functions for configuring the temporary RSA and
8056 DH private keys and/or callback functions which directly correspond to
8057 their SSL_CTX_xxx() counterparts but work on a per-connection basis. This
8058 is needed for applications which have to configure certificates on a
8059 per-connection basis (e.g. Apache+mod_ssl) instead of a per-context basis
8060 (e.g. s_server).
8061 For the RSA certificate situation is makes no difference, but
8062 for the DSA certificate situation this fixes the "no shared cipher"
8063 problem where the OpenSSL cipher selection procedure failed because the
8064 temporary keys were not overtaken from the context and the API provided
8065 no way to reconfigure them.
8066 The new functions now let applications reconfigure the stuff and they
8067 are in detail: SSL_need_tmp_RSA, SSL_set_tmp_rsa, SSL_set_tmp_dh,
8068 SSL_set_tmp_rsa_callback and SSL_set_tmp_dh_callback. Additionally a new
8069 non-public-API function ssl_cert_instantiate() is used as a helper
8070 function and also to reduce code redundancy inside ssl_rsa.c.
8071 [Ralf S. Engelschall]
8072
8073 *) Move s_server -dcert and -dkey options out of the undocumented feature
8074 area because they are useful for the DSA situation and should be
8075 recognized by the users.
8076 [Ralf S. Engelschall]
8077
8078 *) Fix the cipher decision scheme for export ciphers: the export bits are
8079 *not* within SSL_MKEY_MASK or SSL_AUTH_MASK, they are within
8080 SSL_EXP_MASK. So, the original variable has to be used instead of the
8081 already masked variable.
8082 [Richard Levitte <levitte@stacken.kth.se>]
8083
8084 *) Fix 'port' variable from `int' to `unsigned int' in crypto/bio/b_sock.c
8085 [Richard Levitte <levitte@stacken.kth.se>]
8086
8087 *) Change type of another md_len variable in pk7_doit.c:PKCS7_dataFinal()
8088 from `int' to `unsigned int' because it's a length and initialized by
8089 EVP_DigestFinal() which expects an `unsigned int *'.
8090 [Richard Levitte <levitte@stacken.kth.se>]
8091
8092 *) Don't hard-code path to Perl interpreter on shebang line of Configure
8093 script. Instead use the usual Shell->Perl transition trick.
8094 [Ralf S. Engelschall]
8095
8096 *) Make `openssl x509 -noout -modulus' functional also for DSA certificates
8097 (in addition to RSA certificates) to match the behaviour of `openssl dsa
8098 -noout -modulus' as it's already the case for `openssl rsa -noout
8099 -modulus'. For RSA the -modulus is the real "modulus" while for DSA
8100 currently the public key is printed (a decision which was already done by
8101 `openssl dsa -modulus' in the past) which serves a similar purpose.
8102 Additionally the NO_RSA no longer completely removes the whole -modulus
8103 option; it now only avoids using the RSA stuff. Same applies to NO_DSA
8104 now, too.
8105 [Ralf S. Engelschall]
8106
8107 *) Add Arne Ansper's reliable BIO - this is an encrypted, block-digested
8108 BIO. See the source (crypto/evp/bio_ok.c) for more info.
8109 [Arne Ansper <arne@ats.cyber.ee>]
8110
8111 *) Dump the old yucky req code that tried (and failed) to allow raw OIDs
8112 to be added. Now both 'req' and 'ca' can use new objects defined in the
8113 config file.
8114 [Steve Henson]
8115
8116 *) Add cool BIO that does syslog (or event log on NT).
8117 [Arne Ansper <arne@ats.cyber.ee>, integrated by Ben Laurie]
8118
8119 *) Add support for new TLS ciphersuites, TLS_RSA_EXPORT56_WITH_RC4_56_MD5,
8120 TLS_RSA_EXPORT56_WITH_RC2_CBC_56_MD5 and
8121 TLS_RSA_EXPORT56_WITH_DES_CBC_SHA, as specified in "56-bit Export Cipher
8122 Suites For TLS", draft-ietf-tls-56-bit-ciphersuites-00.txt.
8123 [Ben Laurie]
8124
8125 *) Add preliminary config info for new extension code.
8126 [Steve Henson]
8127
8128 *) Make RSA_NO_PADDING really use no padding.
8129 [Ulf Moeller <ulf@fitug.de>]
8130
8131 *) Generate errors when private/public key check is done.
8132 [Ben Laurie]
8133
8134 *) Overhaul for 'crl' utility. New function X509_CRL_print. Partial support
8135 for some CRL extensions and new objects added.
8136 [Steve Henson]
8137
8138 *) Really fix the ASN1 IMPLICIT bug this time... Partial support for private
8139 key usage extension and fuller support for authority key id.
8140 [Steve Henson]
8141
8142 *) Add OAEP encryption for the OpenSSL crypto library. OAEP is the improved
8143 padding method for RSA, which is recommended for new applications in PKCS
8144 #1 v2.0 (RFC 2437, October 1998).
8145 OAEP (Optimal Asymmetric Encryption Padding) has better theoretical
8146 foundations than the ad-hoc padding used in PKCS #1 v1.5. It is secure
8147 against Bleichbacher's attack on RSA.
8148 [Ulf Moeller <ulf@fitug.de>, reformatted, corrected and integrated by
8149 Ben Laurie]
8150
8151 *) Updates to the new SSL compression code
8152 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)]
8153
8154 *) Fix so that the version number in the master secret, when passed
8155 via RSA, checks that if TLS was proposed, but we roll back to SSLv3
8156 (because the server will not accept higher), that the version number
8157 is 0x03,0x01, not 0x03,0x00
8158 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)]
8159
8160 *) Run extensive memory leak checks on SSL apps. Fixed *lots* of memory
8161 leaks in ssl/ relating to new X509_get_pubkey() behaviour. Also fixes
8162 in apps/ and an unrelated leak in crypto/dsa/dsa_vrf.c
8163 [Steve Henson]
8164
8165 *) Support for RAW extensions where an arbitrary extension can be
8166 created by including its DER encoding. See apps/openssl.cnf for
8167 an example.
8168 [Steve Henson]
8169
8170 *) Make sure latest Perl versions don't interpret some generated C array
8171 code as Perl array code in the crypto/err/err_genc.pl script.
8172 [Lars Weber <3weber@informatik.uni-hamburg.de>]
8173
8174 *) Modify ms/do_ms.bat to not generate assembly language makefiles since
8175 not many people have the assembler. Various Win32 compilation fixes and
8176 update to the INSTALL.W32 file with (hopefully) more accurate Win32
8177 build instructions.
8178 [Steve Henson]
8179
8180 *) Modify configure script 'Configure' to automatically create crypto/date.h
8181 file under Win32 and also build pem.h from pem.org. New script
8182 util/mkfiles.pl to create the MINFO file on environments that can't do a
8183 'make files': perl util/mkfiles.pl >MINFO should work.
8184 [Steve Henson]
8185
8186 *) Major rework of DES function declarations, in the pursuit of correctness
8187 and purity. As a result, many evil casts evaporated, and some weirdness,
8188 too. You may find this causes warnings in your code. Zapping your evil
8189 casts will probably fix them. Mostly.
8190 [Ben Laurie]
8191
8192 *) Fix for a typo in asn1.h. Bug fix to object creation script
8193 obj_dat.pl. It considered a zero in an object definition to mean
8194 "end of object": none of the objects in objects.h have any zeros
8195 so it wasn't spotted.
8196 [Steve Henson, reported by Erwann ABALEA <eabalea@certplus.com>]
8197
8198 *) Add support for Triple DES Cipher Block Chaining with Output Feedback
8199 Masking (CBCM). In the absence of test vectors, the best I have been able
8200 to do is check that the decrypt undoes the encrypt, so far. Send me test
8201 vectors if you have them.
8202 [Ben Laurie]
8203
8204 *) Correct calculation of key length for export ciphers (too much space was
8205 allocated for null ciphers). This has not been tested!
8206 [Ben Laurie]
8207
8208 *) Modifications to the mkdef.pl for Win32 DEF file creation. The usage
8209 message is now correct (it understands "crypto" and "ssl" on its
8210 command line). There is also now an "update" option. This will update
8211 the util/ssleay.num and util/libeay.num files with any new functions.
8212 If you do a:
8213 perl util/mkdef.pl crypto ssl update
8214 it will update them.
8215 [Steve Henson]
8216
8217 *) Overhauled the Perl interface (perl/*):
8218 - ported BN stuff to OpenSSL's different BN library
8219 - made the perl/ source tree CVS-aware
8220 - renamed the package from SSLeay to OpenSSL (the files still contain
8221 their history because I've copied them in the repository)
8222 - removed obsolete files (the test scripts will be replaced
8223 by better Test::Harness variants in the future)
8224 [Ralf S. Engelschall]
8225
8226 *) First cut for a very conservative source tree cleanup:
8227 1. merge various obsolete readme texts into doc/ssleay.txt
8228 where we collect the old documents and readme texts.
8229 2. remove the first part of files where I'm already sure that we no
8230 longer need them because of three reasons: either they are just temporary
8231 files which were left by Eric or they are preserved original files where
8232 I've verified that the diff is also available in the CVS via "cvs diff
8233 -rSSLeay_0_8_1b" or they were renamed (as it was definitely the case for
8234 the crypto/md/ stuff).
8235 [Ralf S. Engelschall]
8236
8237 *) More extension code. Incomplete support for subject and issuer alt
8238 name, issuer and authority key id. Change the i2v function parameters
8239 and add an extra 'crl' parameter in the X509V3_CTX structure: guess
8240 what that's for :-) Fix to ASN1 macro which messed up
8241 IMPLICIT tag and add f_enum.c which adds a2i, i2a for ENUMERATED.
8242 [Steve Henson]
8243
8244 *) Preliminary support for ENUMERATED type. This is largely copied from the
8245 INTEGER code.
8246 [Steve Henson]
8247
8248 *) Add new function, EVP_MD_CTX_copy() to replace frequent use of memcpy.
8249 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)]
8250
8251 *) Make sure `make rehash' target really finds the `openssl' program.
8252 [Ralf S. Engelschall, Matthias Loepfe <Matthias.Loepfe@adnovum.ch>]
8253
8254 *) Squeeze another 7% of speed out of MD5 assembler, at least on a P2. I'd
8255 like to hear about it if this slows down other processors.
8256 [Ben Laurie]
8257
8258 *) Add CygWin32 platform information to Configure script.
8259 [Alan Batie <batie@aahz.jf.intel.com>]
8260
8261 *) Fixed ms/32all.bat script: `no_asm' -> `no-asm'
8262 [Rainer W. Gerling <gerling@mpg-gv.mpg.de>]
8263
8264 *) New program nseq to manipulate netscape certificate sequences
8265 [Steve Henson]
8266
8267 *) Modify crl2pkcs7 so it supports multiple -certfile arguments. Fix a
8268 few typos.
8269 [Steve Henson]
8270
8271 *) Fixes to BN code. Previously the default was to define BN_RECURSION
8272 but the BN code had some problems that would cause failures when
8273 doing certificate verification and some other functions.
8274 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)]
8275
8276 *) Add ASN1 and PEM code to support netscape certificate sequences.
8277 [Steve Henson]
8278
8279 *) Add ASN1 and PEM code to support netscape certificate sequences.
8280 [Steve Henson]
8281
8282 *) Add several PKIX and private extended key usage OIDs.
8283 [Steve Henson]
8284
8285 *) Modify the 'ca' program to handle the new extension code. Modify
8286 openssl.cnf for new extension format, add comments.
8287 [Steve Henson]
8288
8289 *) More X509 V3 changes. Fix typo in v3_bitstr.c. Add support to 'req'
8290 and add a sample to openssl.cnf so req -x509 now adds appropriate
8291 CA extensions.
8292 [Steve Henson]
8293
8294 *) Continued X509 V3 changes. Add to other makefiles, integrate with the
8295 error code, add initial support to X509_print() and x509 application.
8296 [Steve Henson]
8297
8298 *) Takes a deep breath and start addding X509 V3 extension support code. Add
8299 files in crypto/x509v3. Move original stuff to crypto/x509v3/old. All this
8300 stuff is currently isolated and isn't even compiled yet.
8301 [Steve Henson]
8302
8303 *) Continuing patches for GeneralizedTime. Fix up certificate and CRL
8304 ASN1 to use ASN1_TIME and modify print routines to use ASN1_TIME_print.
8305 Removed the versions check from X509 routines when loading extensions:
8306 this allows certain broken certificates that don't set the version
8307 properly to be processed.
8308 [Steve Henson]
8309
8310 *) Deal with irritating shit to do with dependencies, in YAAHW (Yet Another
8311 Ad Hoc Way) - Makefile.ssls now all contain local dependencies, which
8312 can still be regenerated with "make depend".
8313 [Ben Laurie]
8314
8315 *) Spelling mistake in C version of CAST-128.
8316 [Ben Laurie, reported by Jeremy Hylton <jeremy@cnri.reston.va.us>]
8317
8318 *) Changes to the error generation code. The perl script err-code.pl
8319 now reads in the old error codes and retains the old numbers, only
8320 adding new ones if necessary. It also only changes the .err files if new
8321 codes are added. The makefiles have been modified to only insert errors
8322 when needed (to avoid needlessly modifying header files). This is done
8323 by only inserting errors if the .err file is newer than the auto generated
8324 C file. To rebuild all the error codes from scratch (the old behaviour)
8325 either modify crypto/Makefile.ssl to pass the -regen flag to err_code.pl
8326 or delete all the .err files.
8327 [Steve Henson]
8328
8329 *) CAST-128 was incorrectly implemented for short keys. The C version has
8330 been fixed, but is untested. The assembler versions are also fixed, but
8331 new assembler HAS NOT BEEN GENERATED FOR WIN32 - the Makefile needs fixing
8332 to regenerate it if needed.
8333 [Ben Laurie, reported (with fix for C version) by Jun-ichiro itojun
8334 Hagino <itojun@kame.net>]
8335
8336 *) File was opened incorrectly in randfile.c.
8337 [Ulf Möller <ulf@fitug.de>]
8338
8339 *) Beginning of support for GeneralizedTime. d2i, i2d, check and print
8340 functions. Also ASN1_TIME suite which is a CHOICE of UTCTime or
8341 GeneralizedTime. ASN1_TIME is the proper type used in certificates et
8342 al: it's just almost always a UTCTime. Note this patch adds new error
8343 codes so do a "make errors" if there are problems.
8344 [Steve Henson]
8345
8346 *) Correct Linux 1 recognition in config.
8347 [Ulf Möller <ulf@fitug.de>]
8348
8349 *) Remove pointless MD5 hash when using DSA keys in ca.
8350 [Anonymous <nobody@replay.com>]
8351
8352 *) Generate an error if given an empty string as a cert directory. Also
8353 generate an error if handed NULL (previously returned 0 to indicate an
8354 error, but didn't set one).
8355 [Ben Laurie, reported by Anonymous <nobody@replay.com>]
8356
8357 *) Add prototypes to SSL methods. Make SSL_write's buffer const, at last.
8358 [Ben Laurie]
8359
8360 *) Fix the dummy function BN_ref_mod_exp() in rsaref.c to have the correct
8361 parameters. This was causing a warning which killed off the Win32 compile.
8362 [Steve Henson]
8363
8364 *) Remove C++ style comments from crypto/bn/bn_local.h.
8365 [Neil Costigan <neil.costigan@celocom.com>]
8366
8367 *) The function OBJ_txt2nid was broken. It was supposed to return a nid
8368 based on a text string, looking up short and long names and finally
8369 "dot" format. The "dot" format stuff didn't work. Added new function
8370 OBJ_txt2obj to do the same but return an ASN1_OBJECT and rewrote
8371 OBJ_txt2nid to use it. OBJ_txt2obj can also return objects even if the
8372 OID is not part of the table.
8373 [Steve Henson]
8374
8375 *) Add prototypes to X509 lookup/verify methods, fixing a bug in
8376 X509_LOOKUP_by_alias().
8377 [Ben Laurie]
8378
8379 *) Sort openssl functions by name.
8380 [Ben Laurie]
8381
8382 *) Get the gendsa program working (hopefully) and add it to app list. Remove
8383 encryption from sample DSA keys (in case anyone is interested the password
8384 was "1234").
8385 [Steve Henson]
8386
8387 *) Make _all_ *_free functions accept a NULL pointer.
8388 [Frans Heymans <fheymans@isaserver.be>]
8389
8390 *) If a DH key is generated in s3_srvr.c, don't blow it by trying to use
8391 NULL pointers.
8392 [Anonymous <nobody@replay.com>]
8393
8394 *) s_server should send the CAfile as acceptable CAs, not its own cert.
8395 [Bodo Moeller <3moeller@informatik.uni-hamburg.de>]
8396
8397 *) Don't blow it for numeric -newkey arguments to apps/req.
8398 [Bodo Moeller <3moeller@informatik.uni-hamburg.de>]
8399
8400 *) Temp key "for export" tests were wrong in s3_srvr.c.
8401 [Anonymous <nobody@replay.com>]
8402
8403 *) Add prototype for temp key callback functions
8404 SSL_CTX_set_tmp_{rsa,dh}_callback().
8405 [Ben Laurie]
8406
8407 *) Make DH_free() tolerate being passed a NULL pointer (like RSA_free() and
8408 DSA_free()). Make X509_PUBKEY_set() check for errors in d2i_PublicKey().
8409 [Steve Henson]
8410
8411 *) X509_name_add_entry() freed the wrong thing after an error.
8412 [Arne Ansper <arne@ats.cyber.ee>]
8413
8414 *) rsa_eay.c would attempt to free a NULL context.
8415 [Arne Ansper <arne@ats.cyber.ee>]
8416
8417 *) BIO_s_socket() had a broken should_retry() on Windoze.
8418 [Arne Ansper <arne@ats.cyber.ee>]
8419
8420 *) BIO_f_buffer() didn't pass on BIO_CTRL_FLUSH.
8421 [Arne Ansper <arne@ats.cyber.ee>]
8422
8423 *) Make sure the already existing X509_STORE->depth variable is initialized
8424 in X509_STORE_new(), but document the fact that this variable is still
8425 unused in the certificate verification process.
8426 [Ralf S. Engelschall]
8427
8428 *) Fix the various library and apps files to free up pkeys obtained from
8429 X509_PUBKEY_get() et al. Also allow x509.c to handle netscape extensions.
8430 [Steve Henson]
8431
8432 *) Fix reference counting in X509_PUBKEY_get(). This makes
8433 demos/maurice/example2.c work, amongst others, probably.
8434 [Steve Henson and Ben Laurie]
8435
8436 *) First cut of a cleanup for apps/. First the `ssleay' program is now named
8437 `openssl' and second, the shortcut symlinks for the `openssl <command>'
8438 are no longer created. This way we have a single and consistent command
8439 line interface `openssl <command>', similar to `cvs <command>'.
8440 [Ralf S. Engelschall, Paul Sutton and Ben Laurie]
8441
8442 *) ca.c: move test for DSA keys inside #ifndef NO_DSA. Make pubkey
8443 BIT STRING wrapper always have zero unused bits.
8444 [Steve Henson]
8445
8446 *) Add CA.pl, perl version of CA.sh, add extended key usage OID.
8447 [Steve Henson]
8448
8449 *) Make the top-level INSTALL documentation easier to understand.
8450 [Paul Sutton]
8451
8452 *) Makefiles updated to exit if an error occurs in a sub-directory
8453 make (including if user presses ^C) [Paul Sutton]
8454
8455 *) Make Montgomery context stuff explicit in RSA data structure.
8456 [Ben Laurie]
8457
8458 *) Fix build order of pem and err to allow for generated pem.h.
8459 [Ben Laurie]
8460
8461 *) Fix renumbering bug in X509_NAME_delete_entry().
8462 [Ben Laurie]
8463
8464 *) Enhanced the err-ins.pl script so it makes the error library number
8465 global and can add a library name. This is needed for external ASN1 and
8466 other error libraries.
8467 [Steve Henson]
8468
8469 *) Fixed sk_insert which never worked properly.
8470 [Steve Henson]
8471
8472 *) Fix ASN1 macros so they can handle indefinite length construted
8473 EXPLICIT tags. Some non standard certificates use these: they can now
8474 be read in.
8475 [Steve Henson]
8476
8477 *) Merged the various old/obsolete SSLeay documentation files (doc/xxx.doc)
8478 into a single doc/ssleay.txt bundle. This way the information is still
8479 preserved but no longer messes up this directory. Now it's new room for
8480 the new set of documenation files.
8481 [Ralf S. Engelschall]
8482
8483 *) SETs were incorrectly DER encoded. This was a major pain, because they
8484 shared code with SEQUENCEs, which aren't coded the same. This means that
8485 almost everything to do with SETs or SEQUENCEs has either changed name or
8486 number of arguments.
8487 [Ben Laurie, based on a partial fix by GP Jayan <gp@nsj.co.jp>]
8488
8489 *) Fix test data to work with the above.
8490 [Ben Laurie]
8491
8492 *) Fix the RSA header declarations that hid a bug I fixed in 0.9.0b but
8493 was already fixed by Eric for 0.9.1 it seems.
8494 [Ben Laurie - pointed out by Ulf Möller <ulf@fitug.de>]
8495
8496 *) Autodetect FreeBSD3.
8497 [Ben Laurie]
8498
8499 *) Fix various bugs in Configure. This affects the following platforms:
8500 nextstep
8501 ncr-scde
8502 unixware-2.0
8503 unixware-2.0-pentium
8504 sco5-cc.
8505 [Ben Laurie]
8506
8507 *) Eliminate generated files from CVS. Reorder tests to regenerate files
8508 before they are needed.
8509 [Ben Laurie]
8510
8511 *) Generate Makefile.ssl from Makefile.org (to keep CVS happy).
8512 [Ben Laurie]
8513
8514
8515 Changes between 0.9.1b and 0.9.1c [23-Dec-1998]
8516
8517 *) Added OPENSSL_VERSION_NUMBER to crypto/crypto.h and
8518 changed SSLeay to OpenSSL in version strings.
8519 [Ralf S. Engelschall]
8520
8521 *) Some fixups to the top-level documents.
8522 [Paul Sutton]
8523
8524 *) Fixed the nasty bug where rsaref.h was not found under compile-time
8525 because the symlink to include/ was missing.
8526 [Ralf S. Engelschall]
8527
8528 *) Incorporated the popular no-RSA/DSA-only patches
8529 which allow to compile a RSA-free SSLeay.
8530 [Andrew Cooke / Interrader Ldt., Ralf S. Engelschall]
8531
8532 *) Fixed nasty rehash problem under `make -f Makefile.ssl links'
8533 when "ssleay" is still not found.
8534 [Ralf S. Engelschall]
8535
8536 *) Added more platforms to Configure: Cray T3E, HPUX 11,
8537 [Ralf S. Engelschall, Beckmann <beckman@acl.lanl.gov>]
8538
8539 *) Updated the README file.
8540 [Ralf S. Engelschall]
8541
8542 *) Added various .cvsignore files in the CVS repository subdirs
8543 to make a "cvs update" really silent.
8544 [Ralf S. Engelschall]
8545
8546 *) Recompiled the error-definition header files and added
8547 missing symbols to the Win32 linker tables.
8548 [Ralf S. Engelschall]
8549
8550 *) Cleaned up the top-level documents;
8551 o new files: CHANGES and LICENSE
8552 o merged VERSION, HISTORY* and README* files a CHANGES.SSLeay
8553 o merged COPYRIGHT into LICENSE
8554 o removed obsolete TODO file
8555 o renamed MICROSOFT to INSTALL.W32
8556 [Ralf S. Engelschall]
8557
8558 *) Removed dummy files from the 0.9.1b source tree:
8559 crypto/asn1/x crypto/bio/cd crypto/bio/fg crypto/bio/grep crypto/bio/vi
8560 crypto/bn/asm/......add.c crypto/bn/asm/a.out crypto/dsa/f crypto/md5/f
8561 crypto/pem/gmon.out crypto/perlasm/f crypto/pkcs7/build crypto/rsa/f
8562 crypto/sha/asm/f crypto/threads/f ms/zzz ssl/f ssl/f.mak test/f
8563 util/f.mak util/pl/f util/pl/f.mak crypto/bf/bf_locl.old apps/f
8564 [Ralf S. Engelschall]
8565
8566 *) Added various platform portability fixes.
8567 [Mark J. Cox]
8568
8569 *) The Genesis of the OpenSSL rpject:
8570 We start with the latest (unreleased) SSLeay version 0.9.1b which Eric A.
8571 Young and Tim J. Hudson created while they were working for C2Net until
8572 summer 1998.
8573 [The OpenSSL Project]
8574
8575
8576 Changes between 0.9.0b and 0.9.1b [not released]
8577
8578 *) Updated a few CA certificates under certs/
8579 [Eric A. Young]
8580
8581 *) Changed some BIGNUM api stuff.
8582 [Eric A. Young]
8583
8584 *) Various platform ports: OpenBSD, Ultrix, IRIX 64bit, NetBSD,
8585 DGUX x86, Linux Alpha, etc.
8586 [Eric A. Young]
8587
8588 *) New COMP library [crypto/comp/] for SSL Record Layer Compression:
8589 RLE (dummy implemented) and ZLIB (really implemented when ZLIB is
8590 available).
8591 [Eric A. Young]
8592
8593 *) Add -strparse option to asn1pars program which parses nested
8594 binary structures
8595 [Dr Stephen Henson <shenson@bigfoot.com>]
8596
8597 *) Added "oid_file" to ssleay.cnf for "ca" and "req" programs.
8598 [Eric A. Young]
8599
8600 *) DSA fix for "ca" program.
8601 [Eric A. Young]
8602
8603 *) Added "-genkey" option to "dsaparam" program.
8604 [Eric A. Young]
8605
8606 *) Added RIPE MD160 (rmd160) message digest.
8607 [Eric A. Young]
8608
8609 *) Added -a (all) option to "ssleay version" command.
8610 [Eric A. Young]
8611
8612 *) Added PLATFORM define which is the id given to Configure.
8613 [Eric A. Young]
8614
8615 *) Added MemCheck_XXXX functions to crypto/mem.c for memory checking.
8616 [Eric A. Young]
8617
8618 *) Extended the ASN.1 parser routines.
8619 [Eric A. Young]
8620
8621 *) Extended BIO routines to support REUSEADDR, seek, tell, etc.
8622 [Eric A. Young]
8623
8624 *) Added a BN_CTX to the BN library.
8625 [Eric A. Young]
8626
8627 *) Fixed the weak key values in DES library
8628 [Eric A. Young]
8629
8630 *) Changed API in EVP library for cipher aliases.
8631 [Eric A. Young]
8632
8633 *) Added support for RC2/64bit cipher.
8634 [Eric A. Young]
8635
8636 *) Converted the lhash library to the crypto/mem.c functions.
8637 [Eric A. Young]
8638
8639 *) Added more recognized ASN.1 object ids.
8640 [Eric A. Young]
8641
8642 *) Added more RSA padding checks for SSL/TLS.
8643 [Eric A. Young]
8644
8645 *) Added BIO proxy/filter functionality.
8646 [Eric A. Young]
8647
8648 *) Added extra_certs to SSL_CTX which can be used
8649 send extra CA certificates to the client in the CA cert chain sending
8650 process. It can be configured with SSL_CTX_add_extra_chain_cert().
8651 [Eric A. Young]
8652
8653 *) Now Fortezza is denied in the authentication phase because
8654 this is key exchange mechanism is not supported by SSLeay at all.
8655 [Eric A. Young]
8656
8657 *) Additional PKCS1 checks.
8658 [Eric A. Young]
8659
8660 *) Support the string "TLSv1" for all TLS v1 ciphers.
8661 [Eric A. Young]
8662
8663 *) Added function SSL_get_ex_data_X509_STORE_CTX_idx() which gives the
8664 ex_data index of the SSL context in the X509_STORE_CTX ex_data.
8665 [Eric A. Young]
8666
8667 *) Fixed a few memory leaks.
8668 [Eric A. Young]
8669
8670 *) Fixed various code and comment typos.
8671 [Eric A. Young]
8672
8673 *) A minor bug in ssl/s3_clnt.c where there would always be 4 0
8674 bytes sent in the client random.
8675 [Edward Bishop <ebishop@spyglass.com>]
8676