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