]> git.ipfire.org Git - thirdparty/openssl.git/blob - CHANGES
142dc80a51cb6d861f739e920f4b993ae15d2599
[thirdparty/openssl.git] / CHANGES
1
2 OpenSSL CHANGES
3 _______________
4
5 Changes between 1.0.1 and 1.1.0 [xx XXX xxxx]
6
7 *) Add DH parameters from RFC5114 including test data to dhtest.
8 [Steve Henson]
9
10 *) Update fips_test_suite to support multiple command line options. New
11 test to induce all self test errors in sequence and check expected
12 failures.
13 [Steve Henson]
14
15 *) Add FIPS_{rsa,dsa,ecdsa}_{sign,verify} functions which digest and
16 sign or verify all in one operation.
17 [Steve Henson]
18
19 *) Add fips_algvs: a multicall fips utility incorporaing all the algorithm
20 test programs and fips_test_suite. Includes functionality to parse
21 the minimal script output of fipsalgest.pl directly.
22 [Steve Henson]
23
24 *) Add authorisation parameter to FIPS_module_mode_set().
25 [Steve Henson]
26
27 *) Add FIPS selftest for ECDH algorithm using P-224 and B-233 curves.
28 [Steve Henson]
29
30 *) Use separate DRBG fields for internal and external flags. New function
31 FIPS_drbg_health_check() to perform on demand health checking. Add
32 generation tests to fips_test_suite with reduced health check interval to
33 demonstrate periodic health checking. Add "nodh" option to
34 fips_test_suite to skip very slow DH test.
35 [Steve Henson]
36
37 *) New function FIPS_get_cipherbynid() to lookup FIPS supported ciphers
38 based on NID.
39 [Steve Henson]
40
41 *) More extensive health check for DRBG checking many more failure modes.
42 New function FIPS_selftest_drbg_all() to handle every possible DRBG
43 combination: call this in fips_test_suite.
44 [Steve Henson]
45
46 *) Add support for Dual EC DRBG from SP800-90. Update DRBG algorithm test
47 and POST to handle Dual EC cases.
48 [Steve Henson]
49
50 *) Add support for canonical generation of DSA parameter 'g'. See
51 FIPS 186-3 A.2.3.
52
53 *) Add support for HMAC DRBG from SP800-90. Update DRBG algorithm test and
54 POST to handle HMAC cases.
55 [Steve Henson]
56
57 *) Add functions FIPS_module_version() and FIPS_module_version_text()
58 to return numberical and string versions of the FIPS module number.
59 [Steve Henson]
60
61 *) Rename FIPS_mode_set and FIPS_mode to FIPS_module_mode_set and
62 FIPS_module_mode. FIPS_mode and FIPS_mode_set will be implmeneted
63 outside the validated module in the FIPS capable OpenSSL.
64 [Steve Henson]
65
66 *) Minor change to DRBG entropy callback semantics. In some cases
67 there is no mutiple of the block length between min_len and
68 max_len. Allow the callback to return more than max_len bytes
69 of entropy but discard any extra: it is the callback's responsibility
70 to ensure that the extra data discarded does not impact the
71 requested amount of entropy.
72 [Steve Henson]
73
74 *) Add PRNG security strength checks to RSA, DSA and ECDSA using
75 information in FIPS186-3, SP800-57 and SP800-131A.
76 [Steve Henson]
77
78 *) CCM support via EVP. Interface is very similar to GCM case except we
79 must supply all data in one chunk (i.e. no update, final) and the
80 message length must be supplied if AAD is used. Add algorithm test
81 support.
82 [Steve Henson]
83
84 *) Initial version of POST overhaul. Add POST callback to allow the status
85 of POST to be monitored and/or failures induced. Modify fips_test_suite
86 to use callback. Always run all selftests even if one fails.
87 [Steve Henson]
88
89 *) XTS support including algorithm test driver in the fips_gcmtest program.
90 Note: this does increase the maximum key length from 32 to 64 bytes but
91 there should be no binary compatibility issues as existing applications
92 will never use XTS mode.
93 [Steve Henson]
94
95 *) Extensive reorganisation of FIPS PRNG behaviour. Remove all dependencies
96 to OpenSSL RAND code and replace with a tiny FIPS RAND API which also
97 performs algorithm blocking for unapproved PRNG types. Also do not
98 set PRNG type in FIPS_mode_set(): leave this to the application.
99 Add default OpenSSL DRBG handling: sets up FIPS PRNG and seeds with
100 the standard OpenSSL PRNG: set additional data to a date time vector.
101 [Steve Henson]
102
103 *) Rename old X9.31 PRNG functions of the form FIPS_rand* to FIPS_x931*.
104 This shouldn't present any incompatibility problems because applications
105 shouldn't be using these directly and any that are will need to rethink
106 anyway as the X9.31 PRNG is now deprecated by FIPS 140-2
107 [Steve Henson]
108
109 *) Extensive self tests and health checking required by SP800-90 DRBG.
110 Remove strength parameter from FIPS_drbg_instantiate and always
111 instantiate at maximum supported strength.
112 [Steve Henson]
113
114 *) Add ECDH code to fips module and fips_ecdhvs for primitives only testing.
115 [Steve Henson]
116
117 *) New algorithm test program fips_dhvs to handle DH primitives only testing.
118 [Steve Henson]
119
120 *) New function DH_compute_key_padded() to compute a DH key and pad with
121 leading zeroes if needed: this complies with SP800-56A et al.
122 [Steve Henson]
123
124 *) Initial implementation of SP800-90 DRBGs for Hash and CTR. Not used by
125 anything, incomplete, subject to change and largely untested at present.
126 [Steve Henson]
127
128 *) Modify fipscanisteronly build option to only build the necessary object
129 files by filtering FIPS_EX_OBJ through a perl script in crypto/Makefile.
130 [Steve Henson]
131
132 *) Add experimental option FIPSSYMS to give all symbols in
133 fipscanister.o and FIPS or fips prefix. This will avoid
134 conflicts with future versions of OpenSSL. Add perl script
135 util/fipsas.pl to preprocess assembly language source files
136 and rename any affected symbols.
137 [Steve Henson]
138
139 *) Add selftest checks and algorithm block of non-fips algorithms in
140 FIPS mode. Remove DES2 from selftests.
141 [Steve Henson]
142
143 *) Add ECDSA code to fips module. Add tiny fips_ecdsa_check to just
144 return internal method without any ENGINE dependencies. Add new
145 tiny fips sign and verify functions.
146 [Steve Henson]
147
148 *) New build option no-ec2m to disable characteristic 2 code.
149 [Steve Henson]
150
151 *) New build option "fipscanisteronly". This only builds fipscanister.o
152 and (currently) associated fips utilities. Uses the file Makefile.fips
153 instead of Makefile.org as the prototype.
154 [Steve Henson]
155
156 *) Add some FIPS mode restrictions to GCM. Add internal IV generator.
157 Update fips_gcmtest to use IV generator.
158 [Steve Henson]
159
160 *) Initial, experimental EVP support for AES-GCM. AAD can be input by
161 setting output buffer to NULL. The *Final function must be
162 called although it will not retrieve any additional data. The tag
163 can be set or retrieved with a ctrl. The IV length is by default 12
164 bytes (96 bits) but can be set to an alternative value. If the IV
165 length exceeds the maximum IV length (currently 16 bytes) it cannot be
166 set before the key.
167 [Steve Henson]
168
169 *) New flag in ciphers: EVP_CIPH_FLAG_CUSTOM_CIPHER. This means the
170 underlying do_cipher function handles all cipher semantics itself
171 including padding and finalisation. This is useful if (for example)
172 an ENGINE cipher handles block padding itself. The behaviour of
173 do_cipher is subtly changed if this flag is set: the return value
174 is the number of characters written to the output buffer (zero is
175 no longer an error code) or a negative error code. Also if the
176 input buffer is NULL and length 0 finalisation should be performed.
177 [Steve Henson]
178
179 *) If a candidate issuer certificate is already part of the constructed
180 path ignore it: new debug notification X509_V_ERR_PATH_LOOP for this case.
181 [Steve Henson]
182
183 *) Improve forward-security support: add functions
184
185 void SSL_CTX_set_not_resumable_session_callback(SSL_CTX *ctx, int (*cb)(SSL *ssl, int is_forward_secure))
186 void SSL_set_not_resumable_session_callback(SSL *ssl, int (*cb)(SSL *ssl, int is_forward_secure))
187
188 for use by SSL/TLS servers; the callback function will be called whenever a
189 new session is created, and gets to decide whether the session may be
190 cached to make it resumable (return 0) or not (return 1). (As by the
191 SSL/TLS protocol specifications, the session_id sent by the server will be
192 empty to indicate that the session is not resumable; also, the server will
193 not generate RFC 4507 (RFC 5077) session tickets.)
194
195 A simple reasonable callback implementation is to return is_forward_secure.
196 This parameter will be set to 1 or 0 depending on the ciphersuite selected
197 by the SSL/TLS server library, indicating whether it can provide forward
198 security.
199 [Emilia Käsper <emilia.kasper@esat.kuleuven.be> (Google)]
200
201 *) Add TLS key material exporter from RFC 5705.
202 [Eric Rescorla]
203
204 *) Add DTLS-SRTP negotiation from RFC 5764.
205 [Eric Rescorla]
206
207 *) Add DTLS-SRTP negotiation from RFC 5764.
208 [Eric Rescorla]
209
210 *) Add Next Protocol Negotiation,
211 http://tools.ietf.org/html/draft-agl-tls-nextprotoneg-00. Can be
212 disabled with a no-npn flag to config or Configure. Code donated
213 by Google.
214 [Adam Langley <agl@google.com> and Ben Laurie]
215
216 *) New function OPENSSL_gmtime_diff to find the difference in days
217 and seconds between two tm structures. This will be used to provide
218 additional functionality for ASN1_TIME.
219 [Steve Henson]
220
221 *) Add -trusted_first option which attempts to find certificates in the
222 trusted store even if an untrusted chain is also supplied.
223 [Steve Henson]
224
225 *) Initial experimental support for explicitly trusted non-root CAs.
226 OpenSSL still tries to build a complete chain to a root but if an
227 intermediate CA has a trust setting included that is used. The first
228 setting is used: whether to trust or reject.
229 [Steve Henson]
230
231 *) New -verify_name option in command line utilities to set verification
232 parameters by name.
233 [Steve Henson]
234
235 *) Initial CMAC implementation. WARNING: EXPERIMENTAL, API MAY CHANGE.
236 Add CMAC pkey methods.
237 [Steve Henson]
238
239 *) Experimental regnegotiation in s_server -www mode. If the client
240 browses /reneg connection is renegotiated. If /renegcert it is
241 renegotiated requesting a certificate.
242 [Steve Henson]
243
244 *) Add an "external" session cache for debugging purposes to s_server. This
245 should help trace issues which normally are only apparent in deployed
246 multi-process servers.
247 [Steve Henson]
248
249 *) Extensive audit of libcrypto with DEBUG_UNUSED. Fix many cases where
250 return value is ignored. NB. The functions RAND_add(), RAND_seed(),
251 BIO_set_cipher() and some obscure PEM functions were changed so they
252 can now return an error. The RAND changes required a change to the
253 RAND_METHOD structure.
254 [Steve Henson]
255
256 *) New macro __owur for "OpenSSL Warn Unused Result". This makes use of
257 a gcc attribute to warn if the result of a function is ignored. This
258 is enable if DEBUG_UNUSED is set. Add to several functions in evp.h
259 whose return value is often ignored.
260 [Steve Henson]
261
262 Changes between 1.0.0f and 1.0.1 [xx XXX xxxx]
263
264 *) Add optional 64-bit optimized implementations of elliptic curves NIST-P224,
265 NIST-P256, NIST-P521, with constant-time single point multiplication on
266 typical inputs. Compiler support for the nonstandard type __uint128_t is
267 required to use this (present in gcc 4.4 and later, for 64-bit builds).
268 Code made available under Apache License version 2.0.
269
270 Specify "enable-ec_nistp_64_gcc_128" on the Configure (or config) command
271 line to include this in your build of OpenSSL, and run "make depend" (or
272 "make update"). This enables the following EC_METHODs:
273
274 EC_GFp_nistp224_method()
275 EC_GFp_nistp256_method()
276 EC_GFp_nistp521_method()
277
278 EC_GROUP_new_by_curve_name() will automatically use these (while
279 EC_GROUP_new_curve_GFp() currently prefers the more flexible
280 implementations).
281 [Emilia Käsper, Adam Langley, Bodo Moeller (Google)]
282
283 *) Use type ossl_ssize_t instad of ssize_t which isn't available on
284 all platforms. Move ssize_t definition from e_os.h to the public
285 header file e_os2.h as it now appears in public header file cms.h
286 [Steve Henson]
287
288 *) New -sigopt option to the ca, req and x509 utilities. Additional
289 signature parameters can be passed using this option and in
290 particular PSS.
291 [Steve Henson]
292
293 *) Add RSA PSS signing function. This will generate and set the
294 appropriate AlgorithmIdentifiers for PSS based on those in the
295 corresponding EVP_MD_CTX structure. No application support yet.
296 [Steve Henson]
297
298 *) Support for companion algorithm specific ASN1 signing routines.
299 New function ASN1_item_sign_ctx() signs a pre-initialised
300 EVP_MD_CTX structure and sets AlgorithmIdentifiers based on
301 the appropriate parameters.
302 [Steve Henson]
303
304 *) Add new algorithm specific ASN1 verification initialisation function
305 to EVP_PKEY_ASN1_METHOD: this is not in EVP_PKEY_METHOD since the ASN1
306 handling will be the same no matter what EVP_PKEY_METHOD is used.
307 Add a PSS handler to support verification of PSS signatures: checked
308 against a number of sample certificates.
309 [Steve Henson]
310
311 *) Add signature printing for PSS. Add PSS OIDs.
312 [Steve Henson, Martin Kaiser <lists@kaiser.cx>]
313
314 *) Add algorithm specific signature printing. An individual ASN1 method
315 can now print out signatures instead of the standard hex dump.
316
317 More complex signatures (e.g. PSS) can print out more meaningful
318 information. Include DSA version that prints out the signature
319 parameters r, s.
320 [Steve Henson]
321
322 *) Password based recipient info support for CMS library: implementing
323 RFC3211.
324 [Steve Henson]
325
326 *) Split password based encryption into PBES2 and PBKDF2 functions. This
327 neatly separates the code into cipher and PBE sections and is required
328 for some algorithms that split PBES2 into separate pieces (such as
329 password based CMS).
330 [Steve Henson]
331
332 *) Session-handling fixes:
333 - Fix handling of connections that are resuming with a session ID,
334 but also support Session Tickets.
335 - Fix a bug that suppressed issuing of a new ticket if the client
336 presented a ticket with an expired session.
337 - Try to set the ticket lifetime hint to something reasonable.
338 - Make tickets shorter by excluding irrelevant information.
339 - On the client side, don't ignore renewed tickets.
340 [Adam Langley, Bodo Moeller (Google)]
341
342 *) Fix PSK session representation.
343 [Bodo Moeller]
344
345 *) Add RC4-MD5 and AESNI-SHA1 "stitched" implementations.
346
347 This work was sponsored by Intel.
348 [Andy Polyakov]
349
350 *) Add GCM support to TLS library. Some custom code is needed to split
351 the IV between the fixed (from PRF) and explicit (from TLS record)
352 portions. This adds all GCM ciphersuites supported by RFC5288 and
353 RFC5289. Generalise some AES* cipherstrings to inlclude GCM and
354 add a special AESGCM string for GCM only.
355 [Steve Henson]
356
357 *) Expand range of ctrls for AES GCM. Permit setting invocation
358 field on decrypt and retrieval of invocation field only on encrypt.
359 [Steve Henson]
360
361 *) Add HMAC ECC ciphersuites from RFC5289. Include SHA384 PRF support.
362 As required by RFC5289 these ciphersuites cannot be used if for
363 versions of TLS earlier than 1.2.
364 [Steve Henson]
365
366 *) For FIPS capable OpenSSL interpret a NULL default public key method
367 as unset and return the appopriate default but do *not* set the default.
368 This means we can return the appopriate method in applications that
369 swicth between FIPS and non-FIPS modes.
370 [Steve Henson]
371
372 *) Redirect HMAC and CMAC operations to FIPS module in FIPS mode. If an
373 ENGINE is used then we cannot handle that in the FIPS module so we
374 keep original code iff non-FIPS operations are allowed.
375 [Steve Henson]
376
377 *) Add -attime option to openssl verify.
378 [Peter Eckersley <pde@eff.org> and Ben Laurie]
379
380 *) Redirect DSA and DH operations to FIPS module in FIPS mode.
381 [Steve Henson]
382
383 *) Redirect ECDSA and ECDH operations to FIPS module in FIPS mode. Also use
384 FIPS EC methods unconditionally for now.
385 [Steve Henson]
386
387 *) New build option no-ec2m to disable characteristic 2 code.
388 [Steve Henson]
389
390 *) Backport libcrypto audit of return value checking from 1.1.0-dev; not
391 all cases can be covered as some introduce binary incompatibilities.
392 [Steve Henson]
393
394 *) Redirect RSA operations to FIPS module including keygen,
395 encrypt, decrypt, sign and verify. Block use of non FIPS RSA methods.
396 [Steve Henson]
397
398 *) Add similar low level API blocking to ciphers.
399 [Steve Henson]
400
401 *) Low level digest APIs are not approved in FIPS mode: any attempt
402 to use these will cause a fatal error. Applications that *really* want
403 to use them can use the private_* version instead.
404 [Steve Henson]
405
406 *) Redirect cipher operations to FIPS module for FIPS builds.
407 [Steve Henson]
408
409 *) Redirect digest operations to FIPS module for FIPS builds.
410 [Steve Henson]
411
412 *) Update build system to add "fips" flag which will link in fipscanister.o
413 for static and shared library builds embedding a signature if needed.
414 [Steve Henson]
415
416 *) Output TLS supported curves in preference order instead of numerical
417 order. This is currently hardcoded for the highest order curves first.
418 This should be configurable so applications can judge speed vs strength.
419 [Steve Henson]
420
421 *) Add TLS v1.2 server support for client authentication.
422 [Steve Henson]
423
424 *) Add support for FIPS mode in ssl library: disable SSLv3, non-FIPS ciphers
425 and enable MD5.
426 [Steve Henson]
427
428 *) Functions FIPS_mode_set() and FIPS_mode() which call the underlying
429 FIPS modules versions.
430 [Steve Henson]
431
432 *) Add TLS v1.2 client side support for client authentication. Keep cache
433 of handshake records longer as we don't know the hash algorithm to use
434 until after the certificate request message is received.
435 [Steve Henson]
436
437 *) Initial TLS v1.2 client support. Add a default signature algorithms
438 extension including all the algorithms we support. Parse new signature
439 format in client key exchange. Relax some ECC signing restrictions for
440 TLS v1.2 as indicated in RFC5246.
441 [Steve Henson]
442
443 *) Add server support for TLS v1.2 signature algorithms extension. Switch
444 to new signature format when needed using client digest preference.
445 All server ciphersuites should now work correctly in TLS v1.2. No client
446 support yet and no support for client certificates.
447 [Steve Henson]
448
449 *) Initial TLS v1.2 support. Add new SHA256 digest to ssl code, switch
450 to SHA256 for PRF when using TLS v1.2 and later. Add new SHA256 based
451 ciphersuites. At present only RSA key exchange ciphersuites work with
452 TLS v1.2. Add new option for TLS v1.2 replacing the old and obsolete
453 SSL_OP_PKCS1_CHECK flags with SSL_OP_NO_TLSv1_2. New TLSv1.2 methods
454 and version checking.
455 [Steve Henson]
456
457 *) New option OPENSSL_NO_SSL_INTERN. If an application can be compiled
458 with this defined it will not be affected by any changes to ssl internal
459 structures. Add several utility functions to allow openssl application
460 to work with OPENSSL_NO_SSL_INTERN defined.
461 [Steve Henson]
462
463 *) Add SRP support.
464 [Tom Wu <tjw@cs.stanford.edu> and Ben Laurie]
465
466 *) Add functions to copy EVP_PKEY_METHOD and retrieve flags and id.
467 [Steve Henson]
468
469 *) Permit abbreviated handshakes when renegotiating using the function
470 SSL_renegotiate_abbreviated().
471 [Robin Seggelmann <seggelmann@fh-muenster.de>]
472
473 *) Add call to ENGINE_register_all_complete() to
474 ENGINE_load_builtin_engines(), so some implementations get used
475 automatically instead of needing explicit application support.
476 [Steve Henson]
477
478 *) Add support for TLS key exporter as described in RFC5705.
479 [Robin Seggelmann <seggelmann@fh-muenster.de>, Steve Henson]
480
481 *) Initial TLSv1.1 support. Since TLSv1.1 is very similar to TLS v1.0 only
482 a few changes are required:
483
484 Add SSL_OP_NO_TLSv1_1 flag.
485 Add TLSv1_1 methods.
486 Update version checking logic to handle version 1.1.
487 Add explicit IV handling (ported from DTLS code).
488 Add command line options to s_client/s_server.
489 [Steve Henson]
490
491 Changes between 1.0.0e and 1.0.0f [xx XXX xxxx]
492
493 *) Fix handling of BN_BLINDING: now BN_BLINDING_invert_ex (rather than
494 BN_BLINDING_invert_ex) calls BN_BLINDING_update, ensuring that concurrent
495 threads won't reuse the same blinding coefficients.
496
497 This also avoids the need to obtain the CRYPTO_LOCK_RSA_BLINDING
498 lock to call BN_BLINDING_invert_ex, and avoids one use of
499 BN_BLINDING_update for each BN_BLINDING structure (previously,
500 the last update always remained unused).
501 [Emilia Käsper (Google)]
502
503 *) In ssl3_clear, preserve s3->init_extra along with s3->rbuf.
504 [Bob Buckholz (Google)]
505
506 Changes between 1.0.0d and 1.0.0e [6 Sep 2011]
507
508 *) Fix bug where CRLs with nextUpdate in the past are sometimes accepted
509 by initialising X509_STORE_CTX properly. (CVE-2011-3207)
510 [Kaspar Brand <ossl@velox.ch>]
511
512 *) Fix SSL memory handling for (EC)DH ciphersuites, in particular
513 for multi-threaded use of ECDH. (CVE-2011-3210)
514 [Adam Langley (Google)]
515
516 *) Fix x509_name_ex_d2i memory leak on bad inputs.
517 [Bodo Moeller]
518
519 *) Remove hard coded ecdsaWithSHA1 signature tests in ssl code and check
520 signature public key algorithm by using OID xref utilities instead.
521 Before this you could only use some ECC ciphersuites with SHA1 only.
522 [Steve Henson]
523
524 *) Add protection against ECDSA timing attacks as mentioned in the paper
525 by Billy Bob Brumley and Nicola Tuveri, see:
526
527 http://eprint.iacr.org/2011/232.pdf
528
529 [Billy Bob Brumley and Nicola Tuveri]
530
531 Changes between 1.0.0c and 1.0.0d [8 Feb 2011]
532
533 *) Fix parsing of OCSP stapling ClientHello extension. CVE-2011-0014
534 [Neel Mehta, Adam Langley, Bodo Moeller (Google)]
535
536 *) Fix bug in string printing code: if *any* escaping is enabled we must
537 escape the escape character (backslash) or the resulting string is
538 ambiguous.
539 [Steve Henson]
540
541 Changes between 1.0.0b and 1.0.0c [2 Dec 2010]
542
543 *) Disable code workaround for ancient and obsolete Netscape browsers
544 and servers: an attacker can use it in a ciphersuite downgrade attack.
545 Thanks to Martin Rex for discovering this bug. CVE-2010-4180
546 [Steve Henson]
547
548 *) Fixed J-PAKE implementation error, originally discovered by
549 Sebastien Martini, further info and confirmation from Stefan
550 Arentz and Feng Hao. Note that this fix is a security fix. CVE-2010-4252
551 [Ben Laurie]
552
553 Changes between 1.0.0a and 1.0.0b [16 Nov 2010]
554
555 *) Fix extension code to avoid race conditions which can result in a buffer
556 overrun vulnerability: resumed sessions must not be modified as they can
557 be shared by multiple threads. CVE-2010-3864
558 [Steve Henson]
559
560 *) Fix WIN32 build system to correctly link an ENGINE directory into
561 a DLL.
562 [Steve Henson]
563
564 Changes between 1.0.0 and 1.0.0a [01 Jun 2010]
565
566 *) Check return value of int_rsa_verify in pkey_rsa_verifyrecover
567 (CVE-2010-1633)
568 [Steve Henson, Peter-Michael Hager <hager@dortmund.net>]
569
570 Changes between 0.9.8n and 1.0.0 [29 Mar 2010]
571
572 *) Add "missing" function EVP_CIPHER_CTX_copy(). This copies a cipher
573 context. The operation can be customised via the ctrl mechanism in
574 case ENGINEs want to include additional functionality.
575 [Steve Henson]
576
577 *) Tolerate yet another broken PKCS#8 key format: private key value negative.
578 [Steve Henson]
579
580 *) Add new -subject_hash_old and -issuer_hash_old options to x509 utility to
581 output hashes compatible with older versions of OpenSSL.
582 [Willy Weisz <weisz@vcpc.univie.ac.at>]
583
584 *) Fix compression algorithm handling: if resuming a session use the
585 compression algorithm of the resumed session instead of determining
586 it from client hello again. Don't allow server to change algorithm.
587 [Steve Henson]
588
589 *) Add load_crls() function to apps tidying load_certs() too. Add option
590 to verify utility to allow additional CRLs to be included.
591 [Steve Henson]
592
593 *) Update OCSP request code to permit adding custom headers to the request:
594 some responders need this.
595 [Steve Henson]
596
597 *) The function EVP_PKEY_sign() returns <=0 on error: check return code
598 correctly.
599 [Julia Lawall <julia@diku.dk>]
600
601 *) Update verify callback code in apps/s_cb.c and apps/verify.c, it
602 needlessly dereferenced structures, used obsolete functions and
603 didn't handle all updated verify codes correctly.
604 [Steve Henson]
605
606 *) Disable MD2 in the default configuration.
607 [Steve Henson]
608
609 *) In BIO_pop() and BIO_push() use the ctrl argument (which was NULL) to
610 indicate the initial BIO being pushed or popped. This makes it possible
611 to determine whether the BIO is the one explicitly called or as a result
612 of the ctrl being passed down the chain. Fix BIO_pop() and SSL BIOs so
613 it handles reference counts correctly and doesn't zero out the I/O bio
614 when it is not being explicitly popped. WARNING: applications which
615 included workarounds for the old buggy behaviour will need to be modified
616 or they could free up already freed BIOs.
617 [Steve Henson]
618
619 *) Extend the uni2asc/asc2uni => OPENSSL_uni2asc/OPENSSL_asc2uni
620 renaming to all platforms (within the 0.9.8 branch, this was
621 done conditionally on Netware platforms to avoid a name clash).
622 [Guenter <lists@gknw.net>]
623
624 *) Add ECDHE and PSK support to DTLS.
625 [Michael Tuexen <tuexen@fh-muenster.de>]
626
627 *) Add CHECKED_STACK_OF macro to safestack.h, otherwise safestack can't
628 be used on C++.
629 [Steve Henson]
630
631 *) Add "missing" function EVP_MD_flags() (without this the only way to
632 retrieve a digest flags is by accessing the structure directly. Update
633 EVP_MD_do_all*() and EVP_CIPHER_do_all*() to include the name a digest
634 or cipher is registered as in the "from" argument. Print out all
635 registered digests in the dgst usage message instead of manually
636 attempting to work them out.
637 [Steve Henson]
638
639 *) If no SSLv2 ciphers are used don't use an SSLv2 compatible client hello:
640 this allows the use of compression and extensions. Change default cipher
641 string to remove SSLv2 ciphersuites. This effectively avoids ancient SSLv2
642 by default unless an application cipher string requests it.
643 [Steve Henson]
644
645 *) Alter match criteria in PKCS12_parse(). It used to try to use local
646 key ids to find matching certificates and keys but some PKCS#12 files
647 don't follow the (somewhat unwritten) rules and this strategy fails.
648 Now just gather all certificates together and the first private key
649 then look for the first certificate that matches the key.
650 [Steve Henson]
651
652 *) Support use of registered digest and cipher names for dgst and cipher
653 commands instead of having to add each one as a special case. So now
654 you can do:
655
656 openssl sha256 foo
657
658 as well as:
659
660 openssl dgst -sha256 foo
661
662 and this works for ENGINE based algorithms too.
663
664 [Steve Henson]
665
666 *) Update Gost ENGINE to support parameter files.
667 [Victor B. Wagner <vitus@cryptocom.ru>]
668
669 *) Support GeneralizedTime in ca utility.
670 [Oliver Martin <oliver@volatilevoid.net>, Steve Henson]
671
672 *) Enhance the hash format used for certificate directory links. The new
673 form uses the canonical encoding (meaning equivalent names will work
674 even if they aren't identical) and uses SHA1 instead of MD5. This form
675 is incompatible with the older format and as a result c_rehash should
676 be used to rebuild symbolic links.
677 [Steve Henson]
678
679 *) Make PKCS#8 the default write format for private keys, replacing the
680 traditional format. This form is standardised, more secure and doesn't
681 include an implicit MD5 dependency.
682 [Steve Henson]
683
684 *) Add a $gcc_devteam_warn option to Configure. The idea is that any code
685 committed to OpenSSL should pass this lot as a minimum.
686 [Steve Henson]
687
688 *) Add session ticket override functionality for use by EAP-FAST.
689 [Jouni Malinen <j@w1.fi>]
690
691 *) Modify HMAC functions to return a value. Since these can be implemented
692 in an ENGINE errors can occur.
693 [Steve Henson]
694
695 *) Type-checked OBJ_bsearch_ex.
696 [Ben Laurie]
697
698 *) Type-checked OBJ_bsearch. Also some constification necessitated
699 by type-checking. Still to come: TXT_DB, bsearch(?),
700 OBJ_bsearch_ex, qsort, CRYPTO_EX_DATA, ASN1_VALUE, ASN1_STRING,
701 CONF_VALUE.
702 [Ben Laurie]
703
704 *) New function OPENSSL_gmtime_adj() to add a specific number of days and
705 seconds to a tm structure directly, instead of going through OS
706 specific date routines. This avoids any issues with OS routines such
707 as the year 2038 bug. New *_adj() functions for ASN1 time structures
708 and X509_time_adj_ex() to cover the extended range. The existing
709 X509_time_adj() is still usable and will no longer have any date issues.
710 [Steve Henson]
711
712 *) Delta CRL support. New use deltas option which will attempt to locate
713 and search any appropriate delta CRLs available.
714
715 This work was sponsored by Google.
716 [Steve Henson]
717
718 *) Support for CRLs partitioned by reason code. Reorganise CRL processing
719 code and add additional score elements. Validate alternate CRL paths
720 as part of the CRL checking and indicate a new error "CRL path validation
721 error" in this case. Applications wanting additional details can use
722 the verify callback and check the new "parent" field. If this is not
723 NULL CRL path validation is taking place. Existing applications wont
724 see this because it requires extended CRL support which is off by
725 default.
726
727 This work was sponsored by Google.
728 [Steve Henson]
729
730 *) Support for freshest CRL extension.
731
732 This work was sponsored by Google.
733 [Steve Henson]
734
735 *) Initial indirect CRL support. Currently only supported in the CRLs
736 passed directly and not via lookup. Process certificate issuer
737 CRL entry extension and lookup CRL entries by bother issuer name
738 and serial number. Check and process CRL issuer entry in IDP extension.
739
740 This work was sponsored by Google.
741 [Steve Henson]
742
743 *) Add support for distinct certificate and CRL paths. The CRL issuer
744 certificate is validated separately in this case. Only enabled if
745 an extended CRL support flag is set: this flag will enable additional
746 CRL functionality in future.
747
748 This work was sponsored by Google.
749 [Steve Henson]
750
751 *) Add support for policy mappings extension.
752
753 This work was sponsored by Google.
754 [Steve Henson]
755
756 *) Fixes to pathlength constraint, self issued certificate handling,
757 policy processing to align with RFC3280 and PKITS tests.
758
759 This work was sponsored by Google.
760 [Steve Henson]
761
762 *) Support for name constraints certificate extension. DN, email, DNS
763 and URI types are currently supported.
764
765 This work was sponsored by Google.
766 [Steve Henson]
767
768 *) To cater for systems that provide a pointer-based thread ID rather
769 than numeric, deprecate the current numeric thread ID mechanism and
770 replace it with a structure and associated callback type. This
771 mechanism allows a numeric "hash" to be extracted from a thread ID in
772 either case, and on platforms where pointers are larger than 'long',
773 mixing is done to help ensure the numeric 'hash' is usable even if it
774 can't be guaranteed unique. The default mechanism is to use "&errno"
775 as a pointer-based thread ID to distinguish between threads.
776
777 Applications that want to provide their own thread IDs should now use
778 CRYPTO_THREADID_set_callback() to register a callback that will call
779 either CRYPTO_THREADID_set_numeric() or CRYPTO_THREADID_set_pointer().
780
781 Note that ERR_remove_state() is now deprecated, because it is tied
782 to the assumption that thread IDs are numeric. ERR_remove_state(0)
783 to free the current thread's error state should be replaced by
784 ERR_remove_thread_state(NULL).
785
786 (This new approach replaces the functions CRYPTO_set_idptr_callback(),
787 CRYPTO_get_idptr_callback(), and CRYPTO_thread_idptr() that existed in
788 OpenSSL 0.9.9-dev between June 2006 and August 2008. Also, if an
789 application was previously providing a numeric thread callback that
790 was inappropriate for distinguishing threads, then uniqueness might
791 have been obtained with &errno that happened immediately in the
792 intermediate development versions of OpenSSL; this is no longer the
793 case, the numeric thread callback will now override the automatic use
794 of &errno.)
795 [Geoff Thorpe, with help from Bodo Moeller]
796
797 *) Initial support for different CRL issuing certificates. This covers a
798 simple case where the self issued certificates in the chain exist and
799 the real CRL issuer is higher in the existing chain.
800
801 This work was sponsored by Google.
802 [Steve Henson]
803
804 *) Removed effectively defunct crypto/store from the build.
805 [Ben Laurie]
806
807 *) Revamp of STACK to provide stronger type-checking. Still to come:
808 TXT_DB, bsearch(?), OBJ_bsearch, qsort, CRYPTO_EX_DATA, ASN1_VALUE,
809 ASN1_STRING, CONF_VALUE.
810 [Ben Laurie]
811
812 *) Add a new SSL_MODE_RELEASE_BUFFERS mode flag to release unused buffer
813 RAM on SSL connections. This option can save about 34k per idle SSL.
814 [Nick Mathewson]
815
816 *) Revamp of LHASH to provide stronger type-checking. Still to come:
817 STACK, TXT_DB, bsearch, qsort.
818 [Ben Laurie]
819
820 *) Initial support for Cryptographic Message Syntax (aka CMS) based
821 on RFC3850, RFC3851 and RFC3852. New cms directory and cms utility,
822 support for data, signedData, compressedData, digestedData and
823 encryptedData, envelopedData types included. Scripts to check against
824 RFC4134 examples draft and interop and consistency checks of many
825 content types and variants.
826 [Steve Henson]
827
828 *) Add options to enc utility to support use of zlib compression BIO.
829 [Steve Henson]
830
831 *) Extend mk1mf to support importing of options and assembly language
832 files from Configure script, currently only included in VC-WIN32.
833 The assembly language rules can now optionally generate the source
834 files from the associated perl scripts.
835 [Steve Henson]
836
837 *) Implement remaining functionality needed to support GOST ciphersuites.
838 Interop testing has been performed using CryptoPro implementations.
839 [Victor B. Wagner <vitus@cryptocom.ru>]
840
841 *) s390x assembler pack.
842 [Andy Polyakov]
843
844 *) ARMv4 assembler pack. ARMv4 refers to v4 and later ISA, not CPU
845 "family."
846 [Andy Polyakov]
847
848 *) Implement Opaque PRF Input TLS extension as specified in
849 draft-rescorla-tls-opaque-prf-input-00.txt. Since this is not an
850 official specification yet and no extension type assignment by
851 IANA exists, this extension (for now) will have to be explicitly
852 enabled when building OpenSSL by providing the extension number
853 to use. For example, specify an option
854
855 -DTLSEXT_TYPE_opaque_prf_input=0x9527
856
857 to the "config" or "Configure" script to enable the extension,
858 assuming extension number 0x9527 (which is a completely arbitrary
859 and unofficial assignment based on the MD5 hash of the Internet
860 Draft). Note that by doing so, you potentially lose
861 interoperability with other TLS implementations since these might
862 be using the same extension number for other purposes.
863
864 SSL_set_tlsext_opaque_prf_input(ssl, src, len) is used to set the
865 opaque PRF input value to use in the handshake. This will create
866 an interal copy of the length-'len' string at 'src', and will
867 return non-zero for success.
868
869 To get more control and flexibility, provide a callback function
870 by using
871
872 SSL_CTX_set_tlsext_opaque_prf_input_callback(ctx, cb)
873 SSL_CTX_set_tlsext_opaque_prf_input_callback_arg(ctx, arg)
874
875 where
876
877 int (*cb)(SSL *, void *peerinput, size_t len, void *arg);
878 void *arg;
879
880 Callback function 'cb' will be called in handshakes, and is
881 expected to use SSL_set_tlsext_opaque_prf_input() as appropriate.
882 Argument 'arg' is for application purposes (the value as given to
883 SSL_CTX_set_tlsext_opaque_prf_input_callback_arg() will directly
884 be provided to the callback function). The callback function
885 has to return non-zero to report success: usually 1 to use opaque
886 PRF input just if possible, or 2 to enforce use of the opaque PRF
887 input. In the latter case, the library will abort the handshake
888 if opaque PRF input is not successfully negotiated.
889
890 Arguments 'peerinput' and 'len' given to the callback function
891 will always be NULL and 0 in the case of a client. A server will
892 see the client's opaque PRF input through these variables if
893 available (NULL and 0 otherwise). Note that if the server
894 provides an opaque PRF input, the length must be the same as the
895 length of the client's opaque PRF input.
896
897 Note that the callback function will only be called when creating
898 a new session (session resumption can resume whatever was
899 previously negotiated), and will not be called in SSL 2.0
900 handshakes; thus, SSL_CTX_set_options(ctx, SSL_OP_NO_SSLv2) or
901 SSL_set_options(ssl, SSL_OP_NO_SSLv2) is especially recommended
902 for applications that need to enforce opaque PRF input.
903
904 [Bodo Moeller]
905
906 *) Update ssl code to support digests other than SHA1+MD5 for handshake
907 MAC.
908
909 [Victor B. Wagner <vitus@cryptocom.ru>]
910
911 *) Add RFC4507 support to OpenSSL. This includes the corrections in
912 RFC4507bis. The encrypted ticket format is an encrypted encoded
913 SSL_SESSION structure, that way new session features are automatically
914 supported.
915
916 If a client application caches session in an SSL_SESSION structure
917 support is transparent because tickets are now stored in the encoded
918 SSL_SESSION.
919
920 The SSL_CTX structure automatically generates keys for ticket
921 protection in servers so again support should be possible
922 with no application modification.
923
924 If a client or server wishes to disable RFC4507 support then the option
925 SSL_OP_NO_TICKET can be set.
926
927 Add a TLS extension debugging callback to allow the contents of any client
928 or server extensions to be examined.
929
930 This work was sponsored by Google.
931 [Steve Henson]
932
933 *) Final changes to avoid use of pointer pointer casts in OpenSSL.
934 OpenSSL should now compile cleanly on gcc 4.2
935 [Peter Hartley <pdh@utter.chaos.org.uk>, Steve Henson]
936
937 *) Update SSL library to use new EVP_PKEY MAC API. Include generic MAC
938 support including streaming MAC support: this is required for GOST
939 ciphersuite support.
940 [Victor B. Wagner <vitus@cryptocom.ru>, Steve Henson]
941
942 *) Add option -stream to use PKCS#7 streaming in smime utility. New
943 function i2d_PKCS7_bio_stream() and PEM_write_PKCS7_bio_stream()
944 to output in BER and PEM format.
945 [Steve Henson]
946
947 *) Experimental support for use of HMAC via EVP_PKEY interface. This
948 allows HMAC to be handled via the EVP_DigestSign*() interface. The
949 EVP_PKEY "key" in this case is the HMAC key, potentially allowing
950 ENGINE support for HMAC keys which are unextractable. New -mac and
951 -macopt options to dgst utility.
952 [Steve Henson]
953
954 *) New option -sigopt to dgst utility. Update dgst to use
955 EVP_Digest{Sign,Verify}*. These two changes make it possible to use
956 alternative signing paramaters such as X9.31 or PSS in the dgst
957 utility.
958 [Steve Henson]
959
960 *) Change ssl_cipher_apply_rule(), the internal function that does
961 the work each time a ciphersuite string requests enabling
962 ("foo+bar"), moving ("+foo+bar"), disabling ("-foo+bar", or
963 removing ("!foo+bar") a class of ciphersuites: Now it maintains
964 the order of disabled ciphersuites such that those ciphersuites
965 that most recently went from enabled to disabled not only stay
966 in order with respect to each other, but also have higher priority
967 than other disabled ciphersuites the next time ciphersuites are
968 enabled again.
969
970 This means that you can now say, e.g., "PSK:-PSK:HIGH" to enable
971 the same ciphersuites as with "HIGH" alone, but in a specific
972 order where the PSK ciphersuites come first (since they are the
973 most recently disabled ciphersuites when "HIGH" is parsed).
974
975 Also, change ssl_create_cipher_list() (using this new
976 funcionality) such that between otherwise identical
977 cihpersuites, ephemeral ECDH is preferred over ephemeral DH in
978 the default order.
979 [Bodo Moeller]
980
981 *) Change ssl_create_cipher_list() so that it automatically
982 arranges the ciphersuites in reasonable order before starting
983 to process the rule string. Thus, the definition for "DEFAULT"
984 (SSL_DEFAULT_CIPHER_LIST) now is just "ALL:!aNULL:!eNULL", but
985 remains equivalent to "AES:ALL:!aNULL:!eNULL:+aECDH:+kRSA:+RC4:@STRENGTH".
986 This makes it much easier to arrive at a reasonable default order
987 in applications for which anonymous ciphers are OK (meaning
988 that you can't actually use DEFAULT).
989 [Bodo Moeller; suggested by Victor Duchovni]
990
991 *) Split the SSL/TLS algorithm mask (as used for ciphersuite string
992 processing) into multiple integers instead of setting
993 "SSL_MKEY_MASK" bits, "SSL_AUTH_MASK" bits, "SSL_ENC_MASK",
994 "SSL_MAC_MASK", and "SSL_SSL_MASK" bits all in a single integer.
995 (These masks as well as the individual bit definitions are hidden
996 away into the non-exported interface ssl/ssl_locl.h, so this
997 change to the definition of the SSL_CIPHER structure shouldn't
998 affect applications.) This give us more bits for each of these
999 categories, so there is no longer a need to coagulate AES128 and
1000 AES256 into a single algorithm bit, and to coagulate Camellia128
1001 and Camellia256 into a single algorithm bit, which has led to all
1002 kinds of kludges.
1003
1004 Thus, among other things, the kludge introduced in 0.9.7m and
1005 0.9.8e for masking out AES256 independently of AES128 or masking
1006 out Camellia256 independently of AES256 is not needed here in 0.9.9.
1007
1008 With the change, we also introduce new ciphersuite aliases that
1009 so far were missing: "AES128", "AES256", "CAMELLIA128", and
1010 "CAMELLIA256".
1011 [Bodo Moeller]
1012
1013 *) Add support for dsa-with-SHA224 and dsa-with-SHA256.
1014 Use the leftmost N bytes of the signature input if the input is
1015 larger than the prime q (with N being the size in bytes of q).
1016 [Nils Larsch]
1017
1018 *) Very *very* experimental PKCS#7 streaming encoder support. Nothing uses
1019 it yet and it is largely untested.
1020 [Steve Henson]
1021
1022 *) Add support for the ecdsa-with-SHA224/256/384/512 signature types.
1023 [Nils Larsch]
1024
1025 *) Initial incomplete changes to avoid need for function casts in OpenSSL
1026 some compilers (gcc 4.2 and later) reject their use. Safestack is
1027 reimplemented. Update ASN1 to avoid use of legacy functions.
1028 [Steve Henson]
1029
1030 *) Win32/64 targets are linked with Winsock2.
1031 [Andy Polyakov]
1032
1033 *) Add an X509_CRL_METHOD structure to allow CRL processing to be redirected
1034 to external functions. This can be used to increase CRL handling
1035 efficiency especially when CRLs are very large by (for example) storing
1036 the CRL revoked certificates in a database.
1037 [Steve Henson]
1038
1039 *) Overhaul of by_dir code. Add support for dynamic loading of CRLs so
1040 new CRLs added to a directory can be used. New command line option
1041 -verify_return_error to s_client and s_server. This causes real errors
1042 to be returned by the verify callback instead of carrying on no matter
1043 what. This reflects the way a "real world" verify callback would behave.
1044 [Steve Henson]
1045
1046 *) GOST engine, supporting several GOST algorithms and public key formats.
1047 Kindly donated by Cryptocom.
1048 [Cryptocom]
1049
1050 *) Partial support for Issuing Distribution Point CRL extension. CRLs
1051 partitioned by DP are handled but no indirect CRL or reason partitioning
1052 (yet). Complete overhaul of CRL handling: now the most suitable CRL is
1053 selected via a scoring technique which handles IDP and AKID in CRLs.
1054 [Steve Henson]
1055
1056 *) New X509_STORE_CTX callbacks lookup_crls() and lookup_certs() which
1057 will ultimately be used for all verify operations: this will remove the
1058 X509_STORE dependency on certificate verification and allow alternative
1059 lookup methods. X509_STORE based implementations of these two callbacks.
1060 [Steve Henson]
1061
1062 *) Allow multiple CRLs to exist in an X509_STORE with matching issuer names.
1063 Modify get_crl() to find a valid (unexpired) CRL if possible.
1064 [Steve Henson]
1065
1066 *) New function X509_CRL_match() to check if two CRLs are identical. Normally
1067 this would be called X509_CRL_cmp() but that name is already used by
1068 a function that just compares CRL issuer names. Cache several CRL
1069 extensions in X509_CRL structure and cache CRLDP in X509.
1070 [Steve Henson]
1071
1072 *) Store a "canonical" representation of X509_NAME structure (ASN1 Name)
1073 this maps equivalent X509_NAME structures into a consistent structure.
1074 Name comparison can then be performed rapidly using memcmp().
1075 [Steve Henson]
1076
1077 *) Non-blocking OCSP request processing. Add -timeout option to ocsp
1078 utility.
1079 [Steve Henson]
1080
1081 *) Allow digests to supply their own micalg string for S/MIME type using
1082 the ctrl EVP_MD_CTRL_MICALG.
1083 [Steve Henson]
1084
1085 *) During PKCS7 signing pass the PKCS7 SignerInfo structure to the
1086 EVP_PKEY_METHOD before and after signing via the EVP_PKEY_CTRL_PKCS7_SIGN
1087 ctrl. It can then customise the structure before and/or after signing
1088 if necessary.
1089 [Steve Henson]
1090
1091 *) New function OBJ_add_sigid() to allow application defined signature OIDs
1092 to be added to OpenSSLs internal tables. New function OBJ_sigid_free()
1093 to free up any added signature OIDs.
1094 [Steve Henson]
1095
1096 *) New functions EVP_CIPHER_do_all(), EVP_CIPHER_do_all_sorted(),
1097 EVP_MD_do_all() and EVP_MD_do_all_sorted() to enumerate internal
1098 digest and cipher tables. New options added to openssl utility:
1099 list-message-digest-algorithms and list-cipher-algorithms.
1100 [Steve Henson]
1101
1102 *) Change the array representation of binary polynomials: the list
1103 of degrees of non-zero coefficients is now terminated with -1.
1104 Previously it was terminated with 0, which was also part of the
1105 value; thus, the array representation was not applicable to
1106 polynomials where t^0 has coefficient zero. This change makes
1107 the array representation useful in a more general context.
1108 [Douglas Stebila]
1109
1110 *) Various modifications and fixes to SSL/TLS cipher string
1111 handling. For ECC, the code now distinguishes between fixed ECDH
1112 with RSA certificates on the one hand and with ECDSA certificates
1113 on the other hand, since these are separate ciphersuites. The
1114 unused code for Fortezza ciphersuites has been removed.
1115
1116 For consistency with EDH, ephemeral ECDH is now called "EECDH"
1117 (not "ECDHE"). For consistency with the code for DH
1118 certificates, use of ECDH certificates is now considered ECDH
1119 authentication, not RSA or ECDSA authentication (the latter is
1120 merely the CA's signing algorithm and not actively used in the
1121 protocol).
1122
1123 The temporary ciphersuite alias "ECCdraft" is no longer
1124 available, and ECC ciphersuites are no longer excluded from "ALL"
1125 and "DEFAULT". The following aliases now exist for RFC 4492
1126 ciphersuites, most of these by analogy with the DH case:
1127
1128 kECDHr - ECDH cert, signed with RSA
1129 kECDHe - ECDH cert, signed with ECDSA
1130 kECDH - ECDH cert (signed with either RSA or ECDSA)
1131 kEECDH - ephemeral ECDH
1132 ECDH - ECDH cert or ephemeral ECDH
1133
1134 aECDH - ECDH cert
1135 aECDSA - ECDSA cert
1136 ECDSA - ECDSA cert
1137
1138 AECDH - anonymous ECDH
1139 EECDH - non-anonymous ephemeral ECDH (equivalent to "kEECDH:-AECDH")
1140
1141 [Bodo Moeller]
1142
1143 *) Add additional S/MIME capabilities for AES and GOST ciphers if supported.
1144 Use correct micalg parameters depending on digest(s) in signed message.
1145 [Steve Henson]
1146
1147 *) Add engine support for EVP_PKEY_ASN1_METHOD. Add functions to process
1148 an ENGINE asn1 method. Support ENGINE lookups in the ASN1 code.
1149 [Steve Henson]
1150
1151 *) Initial engine support for EVP_PKEY_METHOD. New functions to permit
1152 an engine to register a method. Add ENGINE lookups for methods and
1153 functional reference processing.
1154 [Steve Henson]
1155
1156 *) New functions EVP_Digest{Sign,Verify)*. These are enchance versions of
1157 EVP_{Sign,Verify}* which allow an application to customise the signature
1158 process.
1159 [Steve Henson]
1160
1161 *) New -resign option to smime utility. This adds one or more signers
1162 to an existing PKCS#7 signedData structure. Also -md option to use an
1163 alternative message digest algorithm for signing.
1164 [Steve Henson]
1165
1166 *) Tidy up PKCS#7 routines and add new functions to make it easier to
1167 create PKCS7 structures containing multiple signers. Update smime
1168 application to support multiple signers.
1169 [Steve Henson]
1170
1171 *) New -macalg option to pkcs12 utility to allow setting of an alternative
1172 digest MAC.
1173 [Steve Henson]
1174
1175 *) Initial support for PKCS#5 v2.0 PRFs other than default SHA1 HMAC.
1176 Reorganize PBE internals to lookup from a static table using NIDs,
1177 add support for HMAC PBE OID translation. Add a EVP_CIPHER ctrl:
1178 EVP_CTRL_PBE_PRF_NID this allows a cipher to specify an alternative
1179 PRF which will be automatically used with PBES2.
1180 [Steve Henson]
1181
1182 *) Replace the algorithm specific calls to generate keys in "req" with the
1183 new API.
1184 [Steve Henson]
1185
1186 *) Update PKCS#7 enveloped data routines to use new API. This is now
1187 supported by any public key method supporting the encrypt operation. A
1188 ctrl is added to allow the public key algorithm to examine or modify
1189 the PKCS#7 RecipientInfo structure if it needs to: for RSA this is
1190 a no op.
1191 [Steve Henson]
1192
1193 *) Add a ctrl to asn1 method to allow a public key algorithm to express
1194 a default digest type to use. In most cases this will be SHA1 but some
1195 algorithms (such as GOST) need to specify an alternative digest. The
1196 return value indicates how strong the prefernce is 1 means optional and
1197 2 is mandatory (that is it is the only supported type). Modify
1198 ASN1_item_sign() to accept a NULL digest argument to indicate it should
1199 use the default md. Update openssl utilities to use the default digest
1200 type for signing if it is not explicitly indicated.
1201 [Steve Henson]
1202
1203 *) Use OID cross reference table in ASN1_sign() and ASN1_verify(). New
1204 EVP_MD flag EVP_MD_FLAG_PKEY_METHOD_SIGNATURE. This uses the relevant
1205 signing method from the key type. This effectively removes the link
1206 between digests and public key types.
1207 [Steve Henson]
1208
1209 *) Add an OID cross reference table and utility functions. Its purpose is to
1210 translate between signature OIDs such as SHA1WithrsaEncryption and SHA1,
1211 rsaEncryption. This will allow some of the algorithm specific hackery
1212 needed to use the correct OID to be removed.
1213 [Steve Henson]
1214
1215 *) Remove algorithm specific dependencies when setting PKCS7_SIGNER_INFO
1216 structures for PKCS7_sign(). They are now set up by the relevant public
1217 key ASN1 method.
1218 [Steve Henson]
1219
1220 *) Add provisional EC pkey method with support for ECDSA and ECDH.
1221 [Steve Henson]
1222
1223 *) Add support for key derivation (agreement) in the API, DH method and
1224 pkeyutl.
1225 [Steve Henson]
1226
1227 *) Add DSA pkey method and DH pkey methods, extend DH ASN1 method to support
1228 public and private key formats. As a side effect these add additional
1229 command line functionality not previously available: DSA signatures can be
1230 generated and verified using pkeyutl and DH key support and generation in
1231 pkey, genpkey.
1232 [Steve Henson]
1233
1234 *) BeOS support.
1235 [Oliver Tappe <zooey@hirschkaefer.de>]
1236
1237 *) New make target "install_html_docs" installs HTML renditions of the
1238 manual pages.
1239 [Oliver Tappe <zooey@hirschkaefer.de>]
1240
1241 *) New utility "genpkey" this is analagous to "genrsa" etc except it can
1242 generate keys for any algorithm. Extend and update EVP_PKEY_METHOD to
1243 support key and parameter generation and add initial key generation
1244 functionality for RSA.
1245 [Steve Henson]
1246
1247 *) Add functions for main EVP_PKEY_method operations. The undocumented
1248 functions EVP_PKEY_{encrypt,decrypt} have been renamed to
1249 EVP_PKEY_{encrypt,decrypt}_old.
1250 [Steve Henson]
1251
1252 *) Initial definitions for EVP_PKEY_METHOD. This will be a high level public
1253 key API, doesn't do much yet.
1254 [Steve Henson]
1255
1256 *) New function EVP_PKEY_asn1_get0_info() to retrieve information about
1257 public key algorithms. New option to openssl utility:
1258 "list-public-key-algorithms" to print out info.
1259 [Steve Henson]
1260
1261 *) Implement the Supported Elliptic Curves Extension for
1262 ECC ciphersuites from draft-ietf-tls-ecc-12.txt.
1263 [Douglas Stebila]
1264
1265 *) Don't free up OIDs in OBJ_cleanup() if they are in use by EVP_MD or
1266 EVP_CIPHER structures to avoid later problems in EVP_cleanup().
1267 [Steve Henson]
1268
1269 *) New utilities pkey and pkeyparam. These are similar to algorithm specific
1270 utilities such as rsa, dsa, dsaparam etc except they process any key
1271 type.
1272 [Steve Henson]
1273
1274 *) Transfer public key printing routines to EVP_PKEY_ASN1_METHOD. New
1275 functions EVP_PKEY_print_public(), EVP_PKEY_print_private(),
1276 EVP_PKEY_print_param() to print public key data from an EVP_PKEY
1277 structure.
1278 [Steve Henson]
1279
1280 *) Initial support for pluggable public key ASN1.
1281 De-spaghettify the public key ASN1 handling. Move public and private
1282 key ASN1 handling to a new EVP_PKEY_ASN1_METHOD structure. Relocate
1283 algorithm specific handling to a single module within the relevant
1284 algorithm directory. Add functions to allow (near) opaque processing
1285 of public and private key structures.
1286 [Steve Henson]
1287
1288 *) Implement the Supported Point Formats Extension for
1289 ECC ciphersuites from draft-ietf-tls-ecc-12.txt.
1290 [Douglas Stebila]
1291
1292 *) Add initial support for RFC 4279 PSK TLS ciphersuites. Add members
1293 for the psk identity [hint] and the psk callback functions to the
1294 SSL_SESSION, SSL and SSL_CTX structure.
1295
1296 New ciphersuites:
1297 PSK-RC4-SHA, PSK-3DES-EDE-CBC-SHA, PSK-AES128-CBC-SHA,
1298 PSK-AES256-CBC-SHA
1299
1300 New functions:
1301 SSL_CTX_use_psk_identity_hint
1302 SSL_get_psk_identity_hint
1303 SSL_get_psk_identity
1304 SSL_use_psk_identity_hint
1305
1306 [Mika Kousa and Pasi Eronen of Nokia Corporation]
1307
1308 *) Add RFC 3161 compliant time stamp request creation, response generation
1309 and response verification functionality.
1310 [Zoltán Glózik <zglozik@opentsa.org>, The OpenTSA Project]
1311
1312 *) Add initial support for TLS extensions, specifically for the server_name
1313 extension so far. The SSL_SESSION, SSL_CTX, and SSL data structures now
1314 have new members for a host name. The SSL data structure has an
1315 additional member SSL_CTX *initial_ctx so that new sessions can be
1316 stored in that context to allow for session resumption, even after the
1317 SSL has been switched to a new SSL_CTX in reaction to a client's
1318 server_name extension.
1319
1320 New functions (subject to change):
1321
1322 SSL_get_servername()
1323 SSL_get_servername_type()
1324 SSL_set_SSL_CTX()
1325
1326 New CTRL codes and macros (subject to change):
1327
1328 SSL_CTRL_SET_TLSEXT_SERVERNAME_CB
1329 - SSL_CTX_set_tlsext_servername_callback()
1330 SSL_CTRL_SET_TLSEXT_SERVERNAME_ARG
1331 - SSL_CTX_set_tlsext_servername_arg()
1332 SSL_CTRL_SET_TLSEXT_HOSTNAME - SSL_set_tlsext_host_name()
1333
1334 openssl s_client has a new '-servername ...' option.
1335
1336 openssl s_server has new options '-servername_host ...', '-cert2 ...',
1337 '-key2 ...', '-servername_fatal' (subject to change). This allows
1338 testing the HostName extension for a specific single host name ('-cert'
1339 and '-key' remain fallbacks for handshakes without HostName
1340 negotiation). If the unrecogninzed_name alert has to be sent, this by
1341 default is a warning; it becomes fatal with the '-servername_fatal'
1342 option.
1343
1344 [Peter Sylvester, Remy Allais, Christophe Renou]
1345
1346 *) Whirlpool hash implementation is added.
1347 [Andy Polyakov]
1348
1349 *) BIGNUM code on 64-bit SPARCv9 targets is switched from bn(64,64) to
1350 bn(64,32). Because of instruction set limitations it doesn't have
1351 any negative impact on performance. This was done mostly in order
1352 to make it possible to share assembler modules, such as bn_mul_mont
1353 implementations, between 32- and 64-bit builds without hassle.
1354 [Andy Polyakov]
1355
1356 *) Move code previously exiled into file crypto/ec/ec2_smpt.c
1357 to ec2_smpl.c, and no longer require the OPENSSL_EC_BIN_PT_COMP
1358 macro.
1359 [Bodo Moeller]
1360
1361 *) New candidate for BIGNUM assembler implementation, bn_mul_mont,
1362 dedicated Montgomery multiplication procedure, is introduced.
1363 BN_MONT_CTX is modified to allow bn_mul_mont to reach for higher
1364 "64-bit" performance on certain 32-bit targets.
1365 [Andy Polyakov]
1366
1367 *) New option SSL_OP_NO_COMP to disable use of compression selectively
1368 in SSL structures. New SSL ctrl to set maximum send fragment size.
1369 Save memory by seeting the I/O buffer sizes dynamically instead of
1370 using the maximum available value.
1371 [Steve Henson]
1372
1373 *) New option -V for 'openssl ciphers'. This prints the ciphersuite code
1374 in addition to the text details.
1375 [Bodo Moeller]
1376
1377 *) Very, very preliminary EXPERIMENTAL support for printing of general
1378 ASN1 structures. This currently produces rather ugly output and doesn't
1379 handle several customised structures at all.
1380 [Steve Henson]
1381
1382 *) Integrated support for PVK file format and some related formats such
1383 as MS PUBLICKEYBLOB and PRIVATEKEYBLOB. Command line switches to support
1384 these in the 'rsa' and 'dsa' utilities.
1385 [Steve Henson]
1386
1387 *) Support for PKCS#1 RSAPublicKey format on rsa utility command line.
1388 [Steve Henson]
1389
1390 *) Remove the ancient ASN1_METHOD code. This was only ever used in one
1391 place for the (very old) "NETSCAPE" format certificates which are now
1392 handled using new ASN1 code equivalents.
1393 [Steve Henson]
1394
1395 *) Let the TLSv1_method() etc. functions return a 'const' SSL_METHOD
1396 pointer and make the SSL_METHOD parameter in SSL_CTX_new,
1397 SSL_CTX_set_ssl_version and SSL_set_ssl_method 'const'.
1398 [Nils Larsch]
1399
1400 *) Modify CRL distribution points extension code to print out previously
1401 unsupported fields. Enhance extension setting code to allow setting of
1402 all fields.
1403 [Steve Henson]
1404
1405 *) Add print and set support for Issuing Distribution Point CRL extension.
1406 [Steve Henson]
1407
1408 *) Change 'Configure' script to enable Camellia by default.
1409 [NTT]
1410
1411 Changes between 0.9.8r and 0.9.8s [xx XXX xxxx]
1412
1413 *) Fix handling of BN_BLINDING: now BN_BLINDING_invert_ex (rather than
1414 BN_BLINDING_invert_ex) calls BN_BLINDING_update, ensuring that concurrent
1415 threads won't reuse the same blinding coefficients.
1416
1417 This also avoids the need to obtain the CRYPTO_LOCK_RSA_BLINDING
1418 lock to call BN_BLINDING_invert_ex, and avoids one use of
1419 BN_BLINDING_update for each BN_BLINDING structure (previously,
1420 the last update always remained unused).
1421 [Emilia Käsper (Google)]
1422
1423 *) Fix SSL memory handling for (EC)DH ciphersuites, in particular
1424 for multi-threaded use of ECDH.
1425 [Adam Langley (Google)]
1426
1427 *) Fix x509_name_ex_d2i memory leak on bad inputs.
1428 [Bodo Moeller]
1429
1430 *) Add protection against ECDSA timing attacks as mentioned in the paper
1431 by Billy Bob Brumley and Nicola Tuveri, see:
1432
1433 http://eprint.iacr.org/2011/232.pdf
1434
1435 [Billy Bob Brumley and Nicola Tuveri]
1436
1437 Changes between 0.9.8q and 0.9.8r [8 Feb 2011]
1438
1439 *) Fix parsing of OCSP stapling ClientHello extension. CVE-2011-0014
1440 [Neel Mehta, Adam Langley, Bodo Moeller (Google)]
1441
1442 *) Fix bug in string printing code: if *any* escaping is enabled we must
1443 escape the escape character (backslash) or the resulting string is
1444 ambiguous.
1445 [Steve Henson]
1446
1447 Changes between 0.9.8p and 0.9.8q [2 Dec 2010]
1448
1449 *) Disable code workaround for ancient and obsolete Netscape browsers
1450 and servers: an attacker can use it in a ciphersuite downgrade attack.
1451 Thanks to Martin Rex for discovering this bug. CVE-2010-4180
1452 [Steve Henson]
1453
1454 *) Fixed J-PAKE implementation error, originally discovered by
1455 Sebastien Martini, further info and confirmation from Stefan
1456 Arentz and Feng Hao. Note that this fix is a security fix. CVE-2010-4252
1457 [Ben Laurie]
1458
1459 Changes between 0.9.8o and 0.9.8p [16 Nov 2010]
1460
1461 *) Fix extension code to avoid race conditions which can result in a buffer
1462 overrun vulnerability: resumed sessions must not be modified as they can
1463 be shared by multiple threads. CVE-2010-3864
1464 [Steve Henson]
1465
1466 *) Fix for double free bug in ssl/s3_clnt.c CVE-2010-2939
1467 [Steve Henson]
1468
1469 *) Don't reencode certificate when calculating signature: cache and use
1470 the original encoding instead. This makes signature verification of
1471 some broken encodings work correctly.
1472 [Steve Henson]
1473
1474 *) ec2_GF2m_simple_mul bugfix: compute correct result if the output EC_POINT
1475 is also one of the inputs.
1476 [Emilia Käsper <emilia.kasper@esat.kuleuven.be> (Google)]
1477
1478 *) Don't repeatedly append PBE algorithms to table if they already exist.
1479 Sort table on each new add. This effectively makes the table read only
1480 after all algorithms are added and subsequent calls to PKCS12_pbe_add
1481 etc are non-op.
1482 [Steve Henson]
1483
1484 Changes between 0.9.8n and 0.9.8o [01 Jun 2010]
1485
1486 [NB: OpenSSL 0.9.8o and later 0.9.8 patch levels were released after
1487 OpenSSL 1.0.0.]
1488
1489 *) Correct a typo in the CMS ASN1 module which can result in invalid memory
1490 access or freeing data twice (CVE-2010-0742)
1491 [Steve Henson, Ronald Moesbergen <intercommit@gmail.com>]
1492
1493 *) Add SHA2 algorithms to SSL_library_init(). SHA2 is becoming far more
1494 common in certificates and some applications which only call
1495 SSL_library_init and not OpenSSL_add_all_algorithms() will fail.
1496 [Steve Henson]
1497
1498 *) VMS fixes:
1499 Reduce copying into .apps and .test in makevms.com
1500 Don't try to use blank CA certificate in CA.com
1501 Allow use of C files from original directories in maketests.com
1502 [Steven M. Schweda" <sms@antinode.info>]
1503
1504 Changes between 0.9.8m and 0.9.8n [24 Mar 2010]
1505
1506 *) When rejecting SSL/TLS records due to an incorrect version number, never
1507 update s->server with a new major version number. As of
1508 - OpenSSL 0.9.8m if 'short' is a 16-bit type,
1509 - OpenSSL 0.9.8f if 'short' is longer than 16 bits,
1510 the previous behavior could result in a read attempt at NULL when
1511 receiving specific incorrect SSL/TLS records once record payload
1512 protection is active. (CVE-2010-0740)
1513 [Bodo Moeller, Adam Langley <agl@chromium.org>]
1514
1515 *) Fix for CVE-2010-0433 where some kerberos enabled versions of OpenSSL
1516 could be crashed if the relevant tables were not present (e.g. chrooted).
1517 [Tomas Hoger <thoger@redhat.com>]
1518
1519 Changes between 0.9.8l and 0.9.8m [25 Feb 2010]
1520
1521 *) Always check bn_wexpend() return values for failure. (CVE-2009-3245)
1522 [Martin Olsson, Neel Mehta]
1523
1524 *) Fix X509_STORE locking: Every 'objs' access requires a lock (to
1525 accommodate for stack sorting, always a write lock!).
1526 [Bodo Moeller]
1527
1528 *) On some versions of WIN32 Heap32Next is very slow. This can cause
1529 excessive delays in the RAND_poll(): over a minute. As a workaround
1530 include a time check in the inner Heap32Next loop too.
1531 [Steve Henson]
1532
1533 *) The code that handled flushing of data in SSL/TLS originally used the
1534 BIO_CTRL_INFO ctrl to see if any data was pending first. This caused
1535 the problem outlined in PR#1949. The fix suggested there however can
1536 trigger problems with buggy BIO_CTRL_WPENDING (e.g. some versions
1537 of Apache). So instead simplify the code to flush unconditionally.
1538 This should be fine since flushing with no data to flush is a no op.
1539 [Steve Henson]
1540
1541 *) Handle TLS versions 2.0 and later properly and correctly use the
1542 highest version of TLS/SSL supported. Although TLS >= 2.0 is some way
1543 off ancient servers have a habit of sticking around for a while...
1544 [Steve Henson]
1545
1546 *) Modify compression code so it frees up structures without using the
1547 ex_data callbacks. This works around a problem where some applications
1548 call CRYPTO_cleanup_all_ex_data() before application exit (e.g. when
1549 restarting) then use compression (e.g. SSL with compression) later.
1550 This results in significant per-connection memory leaks and
1551 has caused some security issues including CVE-2008-1678 and
1552 CVE-2009-4355.
1553 [Steve Henson]
1554
1555 *) Constify crypto/cast (i.e., <openssl/cast.h>): a CAST_KEY doesn't
1556 change when encrypting or decrypting.
1557 [Bodo Moeller]
1558
1559 *) Add option SSL_OP_LEGACY_SERVER_CONNECT which will allow clients to
1560 connect and renegotiate with servers which do not support RI.
1561 Until RI is more widely deployed this option is enabled by default.
1562 [Steve Henson]
1563
1564 *) Add "missing" ssl ctrls to clear options and mode.
1565 [Steve Henson]
1566
1567 *) If client attempts to renegotiate and doesn't support RI respond with
1568 a no_renegotiation alert as required by RFC5746. Some renegotiating
1569 TLS clients will continue a connection gracefully when they receive
1570 the alert. Unfortunately OpenSSL mishandled this alert and would hang
1571 waiting for a server hello which it will never receive. Now we treat a
1572 received no_renegotiation alert as a fatal error. This is because
1573 applications requesting a renegotiation might well expect it to succeed
1574 and would have no code in place to handle the server denying it so the
1575 only safe thing to do is to terminate the connection.
1576 [Steve Henson]
1577
1578 *) Add ctrl macro SSL_get_secure_renegotiation_support() which returns 1 if
1579 peer supports secure renegotiation and 0 otherwise. Print out peer
1580 renegotiation support in s_client/s_server.
1581 [Steve Henson]
1582
1583 *) Replace the highly broken and deprecated SPKAC certification method with
1584 the updated NID creation version. This should correctly handle UTF8.
1585 [Steve Henson]
1586
1587 *) Implement RFC5746. Re-enable renegotiation but require the extension
1588 as needed. Unfortunately, SSL3_FLAGS_ALLOW_UNSAFE_LEGACY_RENEGOTIATION
1589 turns out to be a bad idea. It has been replaced by
1590 SSL_OP_ALLOW_UNSAFE_LEGACY_RENEGOTIATION which can be set with
1591 SSL_CTX_set_options(). This is really not recommended unless you
1592 know what you are doing.
1593 [Eric Rescorla <ekr@networkresonance.com>, Ben Laurie, Steve Henson]
1594
1595 *) Fixes to stateless session resumption handling. Use initial_ctx when
1596 issuing and attempting to decrypt tickets in case it has changed during
1597 servername handling. Use a non-zero length session ID when attempting
1598 stateless session resumption: this makes it possible to determine if
1599 a resumption has occurred immediately after receiving server hello
1600 (several places in OpenSSL subtly assume this) instead of later in
1601 the handshake.
1602 [Steve Henson]
1603
1604 *) The functions ENGINE_ctrl(), OPENSSL_isservice(),
1605 CMS_get1_RecipientRequest() and RAND_bytes() can return <=0 on error
1606 fixes for a few places where the return code is not checked
1607 correctly.
1608 [Julia Lawall <julia@diku.dk>]
1609
1610 *) Add --strict-warnings option to Configure script to include devteam
1611 warnings in other configurations.
1612 [Steve Henson]
1613
1614 *) Add support for --libdir option and LIBDIR variable in makefiles. This
1615 makes it possible to install openssl libraries in locations which
1616 have names other than "lib", for example "/usr/lib64" which some
1617 systems need.
1618 [Steve Henson, based on patch from Jeremy Utley]
1619
1620 *) Don't allow the use of leading 0x80 in OIDs. This is a violation of
1621 X690 8.9.12 and can produce some misleading textual output of OIDs.
1622 [Steve Henson, reported by Dan Kaminsky]
1623
1624 *) Delete MD2 from algorithm tables. This follows the recommendation in
1625 several standards that it is not used in new applications due to
1626 several cryptographic weaknesses. For binary compatibility reasons
1627 the MD2 API is still compiled in by default.
1628 [Steve Henson]
1629
1630 *) Add compression id to {d2i,i2d}_SSL_SESSION so it is correctly saved
1631 and restored.
1632 [Steve Henson]
1633
1634 *) Rename uni2asc and asc2uni functions to OPENSSL_uni2asc and
1635 OPENSSL_asc2uni conditionally on Netware platforms to avoid a name
1636 clash.
1637 [Guenter <lists@gknw.net>]
1638
1639 *) Fix the server certificate chain building code to use X509_verify_cert(),
1640 it used to have an ad-hoc builder which was unable to cope with anything
1641 other than a simple chain.
1642 [David Woodhouse <dwmw2@infradead.org>, Steve Henson]
1643
1644 *) Don't check self signed certificate signatures in X509_verify_cert()
1645 by default (a flag can override this): it just wastes time without
1646 adding any security. As a useful side effect self signed root CAs
1647 with non-FIPS digests are now usable in FIPS mode.
1648 [Steve Henson]
1649
1650 *) In dtls1_process_out_of_seq_message() the check if the current message
1651 is already buffered was missing. For every new message was memory
1652 allocated, allowing an attacker to perform an denial of service attack
1653 with sending out of seq handshake messages until there is no memory
1654 left. Additionally every future messege was buffered, even if the
1655 sequence number made no sense and would be part of another handshake.
1656 So only messages with sequence numbers less than 10 in advance will be
1657 buffered. (CVE-2009-1378)
1658 [Robin Seggelmann, discovered by Daniel Mentz]
1659
1660 *) Records are buffered if they arrive with a future epoch to be
1661 processed after finishing the corresponding handshake. There is
1662 currently no limitation to this buffer allowing an attacker to perform
1663 a DOS attack with sending records with future epochs until there is no
1664 memory left. This patch adds the pqueue_size() function to detemine
1665 the size of a buffer and limits the record buffer to 100 entries.
1666 (CVE-2009-1377)
1667 [Robin Seggelmann, discovered by Daniel Mentz]
1668
1669 *) Keep a copy of frag->msg_header.frag_len so it can be used after the
1670 parent structure is freed. (CVE-2009-1379)
1671 [Daniel Mentz]
1672
1673 *) Handle non-blocking I/O properly in SSL_shutdown() call.
1674 [Darryl Miles <darryl-mailinglists@netbauds.net>]
1675
1676 *) Add 2.5.4.* OIDs
1677 [Ilya O. <vrghost@gmail.com>]
1678
1679 Changes between 0.9.8k and 0.9.8l [5 Nov 2009]
1680
1681 *) Disable renegotiation completely - this fixes a severe security
1682 problem (CVE-2009-3555) at the cost of breaking all
1683 renegotiation. Renegotiation can be re-enabled by setting
1684 SSL3_FLAGS_ALLOW_UNSAFE_LEGACY_RENEGOTIATION in s3->flags at
1685 run-time. This is really not recommended unless you know what
1686 you're doing.
1687 [Ben Laurie]
1688
1689 Changes between 0.9.8j and 0.9.8k [25 Mar 2009]
1690
1691 *) Don't set val to NULL when freeing up structures, it is freed up by
1692 underlying code. If sizeof(void *) > sizeof(long) this can result in
1693 zeroing past the valid field. (CVE-2009-0789)
1694 [Paolo Ganci <Paolo.Ganci@AdNovum.CH>]
1695
1696 *) Fix bug where return value of CMS_SignerInfo_verify_content() was not
1697 checked correctly. This would allow some invalid signed attributes to
1698 appear to verify correctly. (CVE-2009-0591)
1699 [Ivan Nestlerode <inestlerode@us.ibm.com>]
1700
1701 *) Reject UniversalString and BMPString types with invalid lengths. This
1702 prevents a crash in ASN1_STRING_print_ex() which assumes the strings have
1703 a legal length. (CVE-2009-0590)
1704 [Steve Henson]
1705
1706 *) Set S/MIME signing as the default purpose rather than setting it
1707 unconditionally. This allows applications to override it at the store
1708 level.
1709 [Steve Henson]
1710
1711 *) Permit restricted recursion of ASN1 strings. This is needed in practice
1712 to handle some structures.
1713 [Steve Henson]
1714
1715 *) Improve efficiency of mem_gets: don't search whole buffer each time
1716 for a '\n'
1717 [Jeremy Shapiro <jnshapir@us.ibm.com>]
1718
1719 *) New -hex option for openssl rand.
1720 [Matthieu Herrb]
1721
1722 *) Print out UTF8String and NumericString when parsing ASN1.
1723 [Steve Henson]
1724
1725 *) Support NumericString type for name components.
1726 [Steve Henson]
1727
1728 *) Allow CC in the environment to override the automatically chosen
1729 compiler. Note that nothing is done to ensure flags work with the
1730 chosen compiler.
1731 [Ben Laurie]
1732
1733 Changes between 0.9.8i and 0.9.8j [07 Jan 2009]
1734
1735 *) Properly check EVP_VerifyFinal() and similar return values
1736 (CVE-2008-5077).
1737 [Ben Laurie, Bodo Moeller, Google Security Team]
1738
1739 *) Enable TLS extensions by default.
1740 [Ben Laurie]
1741
1742 *) Allow the CHIL engine to be loaded, whether the application is
1743 multithreaded or not. (This does not release the developer from the
1744 obligation to set up the dynamic locking callbacks.)
1745 [Sander Temme <sander@temme.net>]
1746
1747 *) Use correct exit code if there is an error in dgst command.
1748 [Steve Henson; problem pointed out by Roland Dirlewanger]
1749
1750 *) Tweak Configure so that you need to say "experimental-jpake" to enable
1751 JPAKE, and need to use -DOPENSSL_EXPERIMENTAL_JPAKE in applications.
1752 [Bodo Moeller]
1753
1754 *) Add experimental JPAKE support, including demo authentication in
1755 s_client and s_server.
1756 [Ben Laurie]
1757
1758 *) Set the comparison function in v3_addr_canonize().
1759 [Rob Austein <sra@hactrn.net>]
1760
1761 *) Add support for XMPP STARTTLS in s_client.
1762 [Philip Paeps <philip@freebsd.org>]
1763
1764 *) Change the server-side SSL_OP_NETSCAPE_REUSE_CIPHER_CHANGE_BUG behavior
1765 to ensure that even with this option, only ciphersuites in the
1766 server's preference list will be accepted. (Note that the option
1767 applies only when resuming a session, so the earlier behavior was
1768 just about the algorithm choice for symmetric cryptography.)
1769 [Bodo Moeller]
1770
1771 Changes between 0.9.8h and 0.9.8i [15 Sep 2008]
1772
1773 *) Fix NULL pointer dereference if a DTLS server received
1774 ChangeCipherSpec as first record (CVE-2009-1386).
1775 [PR #1679]
1776
1777 *) Fix a state transitition in s3_srvr.c and d1_srvr.c
1778 (was using SSL3_ST_CW_CLNT_HELLO_B, should be ..._ST_SW_SRVR_...).
1779 [Nagendra Modadugu]
1780
1781 *) The fix in 0.9.8c that supposedly got rid of unsafe
1782 double-checked locking was incomplete for RSA blinding,
1783 addressing just one layer of what turns out to have been
1784 doubly unsafe triple-checked locking.
1785
1786 So now fix this for real by retiring the MONT_HELPER macro
1787 in crypto/rsa/rsa_eay.c.
1788
1789 [Bodo Moeller; problem pointed out by Marius Schilder]
1790
1791 *) Various precautionary measures:
1792
1793 - Avoid size_t integer overflow in HASH_UPDATE (md32_common.h).
1794
1795 - Avoid a buffer overflow in d2i_SSL_SESSION() (ssl_asn1.c).
1796 (NB: This would require knowledge of the secret session ticket key
1797 to exploit, in which case you'd be SOL either way.)
1798
1799 - Change bn_nist.c so that it will properly handle input BIGNUMs
1800 outside the expected range.
1801
1802 - Enforce the 'num' check in BN_div() (bn_div.c) for non-BN_DEBUG
1803 builds.
1804
1805 [Neel Mehta, Bodo Moeller]
1806
1807 *) Allow engines to be "soft loaded" - i.e. optionally don't die if
1808 the load fails. Useful for distros.
1809 [Ben Laurie and the FreeBSD team]
1810
1811 *) Add support for Local Machine Keyset attribute in PKCS#12 files.
1812 [Steve Henson]
1813
1814 *) Fix BN_GF2m_mod_arr() top-bit cleanup code.
1815 [Huang Ying]
1816
1817 *) Expand ENGINE to support engine supplied SSL client certificate functions.
1818
1819 This work was sponsored by Logica.
1820 [Steve Henson]
1821
1822 *) Add CryptoAPI ENGINE to support use of RSA and DSA keys held in Windows
1823 keystores. Support for SSL/TLS client authentication too.
1824 Not compiled unless enable-capieng specified to Configure.
1825
1826 This work was sponsored by Logica.
1827 [Steve Henson]
1828
1829 *) Fix bug in X509_ATTRIBUTE creation: dont set attribute using
1830 ASN1_TYPE_set1 if MBSTRING flag set. This bug would crash certain
1831 attribute creation routines such as certifcate requests and PKCS#12
1832 files.
1833 [Steve Henson]
1834
1835 Changes between 0.9.8g and 0.9.8h [28 May 2008]
1836
1837 *) Fix flaw if 'Server Key exchange message' is omitted from a TLS
1838 handshake which could lead to a cilent crash as found using the
1839 Codenomicon TLS test suite (CVE-2008-1672)
1840 [Steve Henson, Mark Cox]
1841
1842 *) Fix double free in TLS server name extensions which could lead to
1843 a remote crash found by Codenomicon TLS test suite (CVE-2008-0891)
1844 [Joe Orton]
1845
1846 *) Clear error queue in SSL_CTX_use_certificate_chain_file()
1847
1848 Clear the error queue to ensure that error entries left from
1849 older function calls do not interfere with the correct operation.
1850 [Lutz Jaenicke, Erik de Castro Lopo]
1851
1852 *) Remove root CA certificates of commercial CAs:
1853
1854 The OpenSSL project does not recommend any specific CA and does not
1855 have any policy with respect to including or excluding any CA.
1856 Therefore it does not make any sense to ship an arbitrary selection
1857 of root CA certificates with the OpenSSL software.
1858 [Lutz Jaenicke]
1859
1860 *) RSA OAEP patches to fix two separate invalid memory reads.
1861 The first one involves inputs when 'lzero' is greater than
1862 'SHA_DIGEST_LENGTH' (it would read about SHA_DIGEST_LENGTH bytes
1863 before the beginning of from). The second one involves inputs where
1864 the 'db' section contains nothing but zeroes (there is a one-byte
1865 invalid read after the end of 'db').
1866 [Ivan Nestlerode <inestlerode@us.ibm.com>]
1867
1868 *) Partial backport from 0.9.9-dev:
1869
1870 Introduce bn_mul_mont (dedicated Montgomery multiplication
1871 procedure) as a candidate for BIGNUM assembler implementation.
1872 While 0.9.9-dev uses assembler for various architectures, only
1873 x86_64 is available by default here in the 0.9.8 branch, and
1874 32-bit x86 is available through a compile-time setting.
1875
1876 To try the 32-bit x86 assembler implementation, use Configure
1877 option "enable-montasm" (which exists only for this backport).
1878
1879 As "enable-montasm" for 32-bit x86 disclaims code stability
1880 anyway, in this constellation we activate additional code
1881 backported from 0.9.9-dev for further performance improvements,
1882 namely BN_from_montgomery_word. (To enable this otherwise,
1883 e.g. x86_64, try "-DMONT_FROM_WORD___NON_DEFAULT_0_9_8_BUILD".)
1884
1885 [Andy Polyakov (backport partially by Bodo Moeller)]
1886
1887 *) Add TLS session ticket callback. This allows an application to set
1888 TLS ticket cipher and HMAC keys rather than relying on hardcoded fixed
1889 values. This is useful for key rollover for example where several key
1890 sets may exist with different names.
1891 [Steve Henson]
1892
1893 *) Reverse ENGINE-internal logic for caching default ENGINE handles.
1894 This was broken until now in 0.9.8 releases, such that the only way
1895 a registered ENGINE could be used (assuming it initialises
1896 successfully on the host) was to explicitly set it as the default
1897 for the relevant algorithms. This is in contradiction with 0.9.7
1898 behaviour and the documentation. With this fix, when an ENGINE is
1899 registered into a given algorithm's table of implementations, the
1900 'uptodate' flag is reset so that auto-discovery will be used next
1901 time a new context for that algorithm attempts to select an
1902 implementation.
1903 [Ian Lister (tweaked by Geoff Thorpe)]
1904
1905 *) Backport of CMS code to OpenSSL 0.9.8. This differs from the 0.9.9
1906 implemention in the following ways:
1907
1908 Lack of EVP_PKEY_ASN1_METHOD means algorithm parameters have to be
1909 hard coded.
1910
1911 Lack of BER streaming support means one pass streaming processing is
1912 only supported if data is detached: setting the streaming flag is
1913 ignored for embedded content.
1914
1915 CMS support is disabled by default and must be explicitly enabled
1916 with the enable-cms configuration option.
1917 [Steve Henson]
1918
1919 *) Update the GMP engine glue to do direct copies between BIGNUM and
1920 mpz_t when openssl and GMP use the same limb size. Otherwise the
1921 existing "conversion via a text string export" trick is still used.
1922 [Paul Sheer <paulsheer@gmail.com>]
1923
1924 *) Zlib compression BIO. This is a filter BIO which compressed and
1925 uncompresses any data passed through it.
1926 [Steve Henson]
1927
1928 *) Add AES_wrap_key() and AES_unwrap_key() functions to implement
1929 RFC3394 compatible AES key wrapping.
1930 [Steve Henson]
1931
1932 *) Add utility functions to handle ASN1 structures. ASN1_STRING_set0():
1933 sets string data without copying. X509_ALGOR_set0() and
1934 X509_ALGOR_get0(): set and retrieve X509_ALGOR (AlgorithmIdentifier)
1935 data. Attribute function X509at_get0_data_by_OBJ(): retrieves data
1936 from an X509_ATTRIBUTE structure optionally checking it occurs only
1937 once. ASN1_TYPE_set1(): set and ASN1_TYPE structure copying supplied
1938 data.
1939 [Steve Henson]
1940
1941 *) Fix BN flag handling in RSA_eay_mod_exp() and BN_MONT_CTX_set()
1942 to get the expected BN_FLG_CONSTTIME behavior.
1943 [Bodo Moeller (Google)]
1944
1945 *) Netware support:
1946
1947 - fixed wrong usage of ioctlsocket() when build for LIBC BSD sockets
1948 - fixed do_tests.pl to run the test suite with CLIB builds too (CLIB_OPT)
1949 - added some more tests to do_tests.pl
1950 - fixed RunningProcess usage so that it works with newer LIBC NDKs too
1951 - removed usage of BN_LLONG for CLIB builds to avoid runtime dependency
1952 - added new Configure targets netware-clib-bsdsock, netware-clib-gcc,
1953 netware-clib-bsdsock-gcc, netware-libc-bsdsock-gcc
1954 - various changes to netware.pl to enable gcc-cross builds on Win32
1955 platform
1956 - changed crypto/bio/b_sock.c to work with macro functions (CLIB BSD)
1957 - various changes to fix missing prototype warnings
1958 - fixed x86nasm.pl to create correct asm files for NASM COFF output
1959 - added AES, WHIRLPOOL and CPUID assembler code to build files
1960 - added missing AES assembler make rules to mk1mf.pl
1961 - fixed order of includes in apps/ocsp.c so that e_os.h settings apply
1962 [Guenter Knauf <eflash@gmx.net>]
1963
1964 *) Implement certificate status request TLS extension defined in RFC3546.
1965 A client can set the appropriate parameters and receive the encoded
1966 OCSP response via a callback. A server can query the supplied parameters
1967 and set the encoded OCSP response in the callback. Add simplified examples
1968 to s_client and s_server.
1969 [Steve Henson]
1970
1971 Changes between 0.9.8f and 0.9.8g [19 Oct 2007]
1972
1973 *) Fix various bugs:
1974 + Binary incompatibility of ssl_ctx_st structure
1975 + DTLS interoperation with non-compliant servers
1976 + Don't call get_session_cb() without proposed session
1977 + Fix ia64 assembler code
1978 [Andy Polyakov, Steve Henson]
1979
1980 Changes between 0.9.8e and 0.9.8f [11 Oct 2007]
1981
1982 *) DTLS Handshake overhaul. There were longstanding issues with
1983 OpenSSL DTLS implementation, which were making it impossible for
1984 RFC 4347 compliant client to communicate with OpenSSL server.
1985 Unfortunately just fixing these incompatibilities would "cut off"
1986 pre-0.9.8f clients. To allow for hassle free upgrade post-0.9.8e
1987 server keeps tolerating non RFC compliant syntax. The opposite is
1988 not true, 0.9.8f client can not communicate with earlier server.
1989 This update even addresses CVE-2007-4995.
1990 [Andy Polyakov]
1991
1992 *) Changes to avoid need for function casts in OpenSSL: some compilers
1993 (gcc 4.2 and later) reject their use.
1994 [Kurt Roeckx <kurt@roeckx.be>, Peter Hartley <pdh@utter.chaos.org.uk>,
1995 Steve Henson]
1996
1997 *) Add RFC4507 support to OpenSSL. This includes the corrections in
1998 RFC4507bis. The encrypted ticket format is an encrypted encoded
1999 SSL_SESSION structure, that way new session features are automatically
2000 supported.
2001
2002 If a client application caches session in an SSL_SESSION structure
2003 support is transparent because tickets are now stored in the encoded
2004 SSL_SESSION.
2005
2006 The SSL_CTX structure automatically generates keys for ticket
2007 protection in servers so again support should be possible
2008 with no application modification.
2009
2010 If a client or server wishes to disable RFC4507 support then the option
2011 SSL_OP_NO_TICKET can be set.
2012
2013 Add a TLS extension debugging callback to allow the contents of any client
2014 or server extensions to be examined.
2015
2016 This work was sponsored by Google.
2017 [Steve Henson]
2018
2019 *) Add initial support for TLS extensions, specifically for the server_name
2020 extension so far. The SSL_SESSION, SSL_CTX, and SSL data structures now
2021 have new members for a host name. The SSL data structure has an
2022 additional member SSL_CTX *initial_ctx so that new sessions can be
2023 stored in that context to allow for session resumption, even after the
2024 SSL has been switched to a new SSL_CTX in reaction to a client's
2025 server_name extension.
2026
2027 New functions (subject to change):
2028
2029 SSL_get_servername()
2030 SSL_get_servername_type()
2031 SSL_set_SSL_CTX()
2032
2033 New CTRL codes and macros (subject to change):
2034
2035 SSL_CTRL_SET_TLSEXT_SERVERNAME_CB
2036 - SSL_CTX_set_tlsext_servername_callback()
2037 SSL_CTRL_SET_TLSEXT_SERVERNAME_ARG
2038 - SSL_CTX_set_tlsext_servername_arg()
2039 SSL_CTRL_SET_TLSEXT_HOSTNAME - SSL_set_tlsext_host_name()
2040
2041 openssl s_client has a new '-servername ...' option.
2042
2043 openssl s_server has new options '-servername_host ...', '-cert2 ...',
2044 '-key2 ...', '-servername_fatal' (subject to change). This allows
2045 testing the HostName extension for a specific single host name ('-cert'
2046 and '-key' remain fallbacks for handshakes without HostName
2047 negotiation). If the unrecogninzed_name alert has to be sent, this by
2048 default is a warning; it becomes fatal with the '-servername_fatal'
2049 option.
2050
2051 [Peter Sylvester, Remy Allais, Christophe Renou, Steve Henson]
2052
2053 *) Add AES and SSE2 assembly language support to VC++ build.
2054 [Steve Henson]
2055
2056 *) Mitigate attack on final subtraction in Montgomery reduction.
2057 [Andy Polyakov]
2058
2059 *) Fix crypto/ec/ec_mult.c to work properly with scalars of value 0
2060 (which previously caused an internal error).
2061 [Bodo Moeller]
2062
2063 *) Squeeze another 10% out of IGE mode when in != out.
2064 [Ben Laurie]
2065
2066 *) AES IGE mode speedup.
2067 [Dean Gaudet (Google)]
2068
2069 *) Add the Korean symmetric 128-bit cipher SEED (see
2070 http://www.kisa.or.kr/kisa/seed/jsp/seed_eng.jsp) and
2071 add SEED ciphersuites from RFC 4162:
2072
2073 TLS_RSA_WITH_SEED_CBC_SHA = "SEED-SHA"
2074 TLS_DHE_DSS_WITH_SEED_CBC_SHA = "DHE-DSS-SEED-SHA"
2075 TLS_DHE_RSA_WITH_SEED_CBC_SHA = "DHE-RSA-SEED-SHA"
2076 TLS_DH_anon_WITH_SEED_CBC_SHA = "ADH-SEED-SHA"
2077
2078 To minimize changes between patchlevels in the OpenSSL 0.9.8
2079 series, SEED remains excluded from compilation unless OpenSSL
2080 is configured with 'enable-seed'.
2081 [KISA, Bodo Moeller]
2082
2083 *) Mitigate branch prediction attacks, which can be practical if a
2084 single processor is shared, allowing a spy process to extract
2085 information. For detailed background information, see
2086 http://eprint.iacr.org/2007/039 (O. Aciicmez, S. Gueron,
2087 J.-P. Seifert, "New Branch Prediction Vulnerabilities in OpenSSL
2088 and Necessary Software Countermeasures"). The core of the change
2089 are new versions BN_div_no_branch() and
2090 BN_mod_inverse_no_branch() of BN_div() and BN_mod_inverse(),
2091 respectively, which are slower, but avoid the security-relevant
2092 conditional branches. These are automatically called by BN_div()
2093 and BN_mod_inverse() if the flag BN_FLG_CONSTTIME is set for one
2094 of the input BIGNUMs. Also, BN_is_bit_set() has been changed to
2095 remove a conditional branch.
2096
2097 BN_FLG_CONSTTIME is the new name for the previous
2098 BN_FLG_EXP_CONSTTIME flag, since it now affects more than just
2099 modular exponentiation. (Since OpenSSL 0.9.7h, setting this flag
2100 in the exponent causes BN_mod_exp_mont() to use the alternative
2101 implementation in BN_mod_exp_mont_consttime().) The old name
2102 remains as a deprecated alias.
2103
2104 Similary, RSA_FLAG_NO_EXP_CONSTTIME is replaced by a more general
2105 RSA_FLAG_NO_CONSTTIME flag since the RSA implementation now uses
2106 constant-time implementations for more than just exponentiation.
2107 Here too the old name is kept as a deprecated alias.
2108
2109 BN_BLINDING_new() will now use BN_dup() for the modulus so that
2110 the BN_BLINDING structure gets an independent copy of the
2111 modulus. This means that the previous "BIGNUM *m" argument to
2112 BN_BLINDING_new() and to BN_BLINDING_create_param() now
2113 essentially becomes "const BIGNUM *m", although we can't actually
2114 change this in the header file before 0.9.9. It allows
2115 RSA_setup_blinding() to use BN_with_flags() on the modulus to
2116 enable BN_FLG_CONSTTIME.
2117
2118 [Matthew D Wood (Intel Corp)]
2119
2120 *) In the SSL/TLS server implementation, be strict about session ID
2121 context matching (which matters if an application uses a single
2122 external cache for different purposes). Previously,
2123 out-of-context reuse was forbidden only if SSL_VERIFY_PEER was
2124 set. This did ensure strict client verification, but meant that,
2125 with applications using a single external cache for quite
2126 different requirements, clients could circumvent ciphersuite
2127 restrictions for a given session ID context by starting a session
2128 in a different context.
2129 [Bodo Moeller]
2130
2131 *) Include "!eNULL" in SSL_DEFAULT_CIPHER_LIST to make sure that
2132 a ciphersuite string such as "DEFAULT:RSA" cannot enable
2133 authentication-only ciphersuites.
2134 [Bodo Moeller]
2135
2136 *) Update the SSL_get_shared_ciphers() fix CVE-2006-3738 which was
2137 not complete and could lead to a possible single byte overflow
2138 (CVE-2007-5135) [Ben Laurie]
2139
2140 Changes between 0.9.8d and 0.9.8e [23 Feb 2007]
2141
2142 *) Since AES128 and AES256 (and similarly Camellia128 and
2143 Camellia256) share a single mask bit in the logic of
2144 ssl/ssl_ciph.c, the code for masking out disabled ciphers needs a
2145 kludge to work properly if AES128 is available and AES256 isn't
2146 (or if Camellia128 is available and Camellia256 isn't).
2147 [Victor Duchovni]
2148
2149 *) Fix the BIT STRING encoding generated by crypto/ec/ec_asn1.c
2150 (within i2d_ECPrivateKey, i2d_ECPKParameters, i2d_ECParameters):
2151 When a point or a seed is encoded in a BIT STRING, we need to
2152 prevent the removal of trailing zero bits to get the proper DER
2153 encoding. (By default, crypto/asn1/a_bitstr.c assumes the case
2154 of a NamedBitList, for which trailing 0 bits need to be removed.)
2155 [Bodo Moeller]
2156
2157 *) Have SSL/TLS server implementation tolerate "mismatched" record
2158 protocol version while receiving ClientHello even if the
2159 ClientHello is fragmented. (The server can't insist on the
2160 particular protocol version it has chosen before the ServerHello
2161 message has informed the client about his choice.)
2162 [Bodo Moeller]
2163
2164 *) Add RFC 3779 support.
2165 [Rob Austein for ARIN, Ben Laurie]
2166
2167 *) Load error codes if they are not already present instead of using a
2168 static variable. This allows them to be cleanly unloaded and reloaded.
2169 Improve header file function name parsing.
2170 [Steve Henson]
2171
2172 *) extend SMTP and IMAP protocol emulation in s_client to use EHLO
2173 or CAPABILITY handshake as required by RFCs.
2174 [Goetz Babin-Ebell]
2175
2176 Changes between 0.9.8c and 0.9.8d [28 Sep 2006]
2177
2178 *) Introduce limits to prevent malicious keys being able to
2179 cause a denial of service. (CVE-2006-2940)
2180 [Steve Henson, Bodo Moeller]
2181
2182 *) Fix ASN.1 parsing of certain invalid structures that can result
2183 in a denial of service. (CVE-2006-2937) [Steve Henson]
2184
2185 *) Fix buffer overflow in SSL_get_shared_ciphers() function.
2186 (CVE-2006-3738) [Tavis Ormandy and Will Drewry, Google Security Team]
2187
2188 *) Fix SSL client code which could crash if connecting to a
2189 malicious SSLv2 server. (CVE-2006-4343)
2190 [Tavis Ormandy and Will Drewry, Google Security Team]
2191
2192 *) Since 0.9.8b, ciphersuite strings naming explicit ciphersuites
2193 match only those. Before that, "AES256-SHA" would be interpreted
2194 as a pattern and match "AES128-SHA" too (since AES128-SHA got
2195 the same strength classification in 0.9.7h) as we currently only
2196 have a single AES bit in the ciphersuite description bitmap.
2197 That change, however, also applied to ciphersuite strings such as
2198 "RC4-MD5" that intentionally matched multiple ciphersuites --
2199 namely, SSL 2.0 ciphersuites in addition to the more common ones
2200 from SSL 3.0/TLS 1.0.
2201
2202 So we change the selection algorithm again: Naming an explicit
2203 ciphersuite selects this one ciphersuite, and any other similar
2204 ciphersuite (same bitmap) from *other* protocol versions.
2205 Thus, "RC4-MD5" again will properly select both the SSL 2.0
2206 ciphersuite and the SSL 3.0/TLS 1.0 ciphersuite.
2207
2208 Since SSL 2.0 does not have any ciphersuites for which the
2209 128/256 bit distinction would be relevant, this works for now.
2210 The proper fix will be to use different bits for AES128 and
2211 AES256, which would have avoided the problems from the beginning;
2212 however, bits are scarce, so we can only do this in a new release
2213 (not just a patchlevel) when we can change the SSL_CIPHER
2214 definition to split the single 'unsigned long mask' bitmap into
2215 multiple values to extend the available space.
2216
2217 [Bodo Moeller]
2218
2219 Changes between 0.9.8b and 0.9.8c [05 Sep 2006]
2220
2221 *) Avoid PKCS #1 v1.5 signature attack discovered by Daniel Bleichenbacher
2222 (CVE-2006-4339) [Ben Laurie and Google Security Team]
2223
2224 *) Add AES IGE and biIGE modes.
2225 [Ben Laurie]
2226
2227 *) Change the Unix randomness entropy gathering to use poll() when
2228 possible instead of select(), since the latter has some
2229 undesirable limitations.
2230 [Darryl Miles via Richard Levitte and Bodo Moeller]
2231
2232 *) Disable "ECCdraft" ciphersuites more thoroughly. Now special
2233 treatment in ssl/ssl_ciph.s makes sure that these ciphersuites
2234 cannot be implicitly activated as part of, e.g., the "AES" alias.
2235 However, please upgrade to OpenSSL 0.9.9[-dev] for
2236 non-experimental use of the ECC ciphersuites to get TLS extension
2237 support, which is required for curve and point format negotiation
2238 to avoid potential handshake problems.
2239 [Bodo Moeller]
2240
2241 *) Disable rogue ciphersuites:
2242
2243 - SSLv2 0x08 0x00 0x80 ("RC4-64-MD5")
2244 - SSLv3/TLSv1 0x00 0x61 ("EXP1024-RC2-CBC-MD5")
2245 - SSLv3/TLSv1 0x00 0x60 ("EXP1024-RC4-MD5")
2246
2247 The latter two were purportedly from
2248 draft-ietf-tls-56-bit-ciphersuites-0[01].txt, but do not really
2249 appear there.
2250
2251 Also deactivate the remaining ciphersuites from
2252 draft-ietf-tls-56-bit-ciphersuites-01.txt. These are just as
2253 unofficial, and the ID has long expired.
2254 [Bodo Moeller]
2255
2256 *) Fix RSA blinding Heisenbug (problems sometimes occured on
2257 dual-core machines) and other potential thread-safety issues.
2258 [Bodo Moeller]
2259
2260 *) Add the symmetric cipher Camellia (128-bit, 192-bit, 256-bit key
2261 versions), which is now available for royalty-free use
2262 (see http://info.isl.ntt.co.jp/crypt/eng/info/chiteki.html).
2263 Also, add Camellia TLS ciphersuites from RFC 4132.
2264
2265 To minimize changes between patchlevels in the OpenSSL 0.9.8
2266 series, Camellia remains excluded from compilation unless OpenSSL
2267 is configured with 'enable-camellia'.
2268 [NTT]
2269
2270 *) Disable the padding bug check when compression is in use. The padding
2271 bug check assumes the first packet is of even length, this is not
2272 necessarily true if compresssion is enabled and can result in false
2273 positives causing handshake failure. The actual bug test is ancient
2274 code so it is hoped that implementations will either have fixed it by
2275 now or any which still have the bug do not support compression.
2276 [Steve Henson]
2277
2278 Changes between 0.9.8a and 0.9.8b [04 May 2006]
2279
2280 *) When applying a cipher rule check to see if string match is an explicit
2281 cipher suite and only match that one cipher suite if it is.
2282 [Steve Henson]
2283
2284 *) Link in manifests for VC++ if needed.
2285 [Austin Ziegler <halostatue@gmail.com>]
2286
2287 *) Update support for ECC-based TLS ciphersuites according to
2288 draft-ietf-tls-ecc-12.txt with proposed changes (but without
2289 TLS extensions, which are supported starting with the 0.9.9
2290 branch, not in the OpenSSL 0.9.8 branch).
2291 [Douglas Stebila]
2292
2293 *) New functions EVP_CIPHER_CTX_new() and EVP_CIPHER_CTX_free() to support
2294 opaque EVP_CIPHER_CTX handling.
2295 [Steve Henson]
2296
2297 *) Fixes and enhancements to zlib compression code. We now only use
2298 "zlib1.dll" and use the default __cdecl calling convention on Win32
2299 to conform with the standards mentioned here:
2300 http://www.zlib.net/DLL_FAQ.txt
2301 Static zlib linking now works on Windows and the new --with-zlib-include
2302 --with-zlib-lib options to Configure can be used to supply the location
2303 of the headers and library. Gracefully handle case where zlib library
2304 can't be loaded.
2305 [Steve Henson]
2306
2307 *) Several fixes and enhancements to the OID generation code. The old code
2308 sometimes allowed invalid OIDs (1.X for X >= 40 for example), couldn't
2309 handle numbers larger than ULONG_MAX, truncated printing and had a
2310 non standard OBJ_obj2txt() behaviour.
2311 [Steve Henson]
2312
2313 *) Add support for building of engines under engine/ as shared libraries
2314 under VC++ build system.
2315 [Steve Henson]
2316
2317 *) Corrected the numerous bugs in the Win32 path splitter in DSO.
2318 Hopefully, we will not see any false combination of paths any more.
2319 [Richard Levitte]
2320
2321 Changes between 0.9.8 and 0.9.8a [11 Oct 2005]
2322
2323 *) Remove the functionality of SSL_OP_MSIE_SSLV2_RSA_PADDING
2324 (part of SSL_OP_ALL). This option used to disable the
2325 countermeasure against man-in-the-middle protocol-version
2326 rollback in the SSL 2.0 server implementation, which is a bad
2327 idea. (CVE-2005-2969)
2328
2329 [Bodo Moeller; problem pointed out by Yutaka Oiwa (Research Center
2330 for Information Security, National Institute of Advanced Industrial
2331 Science and Technology [AIST], Japan)]
2332
2333 *) Add two function to clear and return the verify parameter flags.
2334 [Steve Henson]
2335
2336 *) Keep cipherlists sorted in the source instead of sorting them at
2337 runtime, thus removing the need for a lock.
2338 [Nils Larsch]
2339
2340 *) Avoid some small subgroup attacks in Diffie-Hellman.
2341 [Nick Mathewson and Ben Laurie]
2342
2343 *) Add functions for well-known primes.
2344 [Nick Mathewson]
2345
2346 *) Extended Windows CE support.
2347 [Satoshi Nakamura and Andy Polyakov]
2348
2349 *) Initialize SSL_METHOD structures at compile time instead of during
2350 runtime, thus removing the need for a lock.
2351 [Steve Henson]
2352
2353 *) Make PKCS7_decrypt() work even if no certificate is supplied by
2354 attempting to decrypt each encrypted key in turn. Add support to
2355 smime utility.
2356 [Steve Henson]
2357
2358 Changes between 0.9.7h and 0.9.8 [05 Jul 2005]
2359
2360 [NB: OpenSSL 0.9.7i and later 0.9.7 patch levels were released after
2361 OpenSSL 0.9.8.]
2362
2363 *) Add libcrypto.pc and libssl.pc for those who feel they need them.
2364 [Richard Levitte]
2365
2366 *) Change CA.sh and CA.pl so they don't bundle the CSR and the private
2367 key into the same file any more.
2368 [Richard Levitte]
2369
2370 *) Add initial support for Win64, both IA64 and AMD64/x64 flavors.
2371 [Andy Polyakov]
2372
2373 *) Add -utf8 command line and config file option to 'ca'.
2374 [Stefan <stf@udoma.org]
2375
2376 *) Removed the macro des_crypt(), as it seems to conflict with some
2377 libraries. Use DES_crypt().
2378 [Richard Levitte]
2379
2380 *) Correct naming of the 'chil' and '4758cca' ENGINEs. This
2381 involves renaming the source and generated shared-libs for
2382 both. The engines will accept the corrected or legacy ids
2383 ('ncipher' and '4758_cca' respectively) when binding. NB,
2384 this only applies when building 'shared'.
2385 [Corinna Vinschen <vinschen@redhat.com> and Geoff Thorpe]
2386
2387 *) Add attribute functions to EVP_PKEY structure. Modify
2388 PKCS12_create() to recognize a CSP name attribute and
2389 use it. Make -CSP option work again in pkcs12 utility.
2390 [Steve Henson]
2391
2392 *) Add new functionality to the bn blinding code:
2393 - automatic re-creation of the BN_BLINDING parameters after
2394 a fixed number of uses (currently 32)
2395 - add new function for parameter creation
2396 - introduce flags to control the update behaviour of the
2397 BN_BLINDING parameters
2398 - hide BN_BLINDING structure
2399 Add a second BN_BLINDING slot to the RSA structure to improve
2400 performance when a single RSA object is shared among several
2401 threads.
2402 [Nils Larsch]
2403
2404 *) Add support for DTLS.
2405 [Nagendra Modadugu <nagendra@cs.stanford.edu> and Ben Laurie]
2406
2407 *) Add support for DER encoded private keys (SSL_FILETYPE_ASN1)
2408 to SSL_CTX_use_PrivateKey_file() and SSL_use_PrivateKey_file()
2409 [Walter Goulet]
2410
2411 *) Remove buggy and incompletet DH cert support from
2412 ssl/ssl_rsa.c and ssl/s3_both.c
2413 [Nils Larsch]
2414
2415 *) Use SHA-1 instead of MD5 as the default digest algorithm for
2416 the apps/openssl applications.
2417 [Nils Larsch]
2418
2419 *) Compile clean with "-Wall -Wmissing-prototypes
2420 -Wstrict-prototypes -Wmissing-declarations -Werror". Currently
2421 DEBUG_SAFESTACK must also be set.
2422 [Ben Laurie]
2423
2424 *) Change ./Configure so that certain algorithms can be disabled by default.
2425 The new counterpiece to "no-xxx" is "enable-xxx".
2426
2427 The patented RC5 and MDC2 algorithms will now be disabled unless
2428 "enable-rc5" and "enable-mdc2", respectively, are specified.
2429
2430 (IDEA remains enabled despite being patented. This is because IDEA
2431 is frequently required for interoperability, and there is no license
2432 fee for non-commercial use. As before, "no-idea" can be used to
2433 avoid this algorithm.)
2434
2435 [Bodo Moeller]
2436
2437 *) Add processing of proxy certificates (see RFC 3820). This work was
2438 sponsored by KTH (The Royal Institute of Technology in Stockholm) and
2439 EGEE (Enabling Grids for E-science in Europe).
2440 [Richard Levitte]
2441
2442 *) RC4 performance overhaul on modern architectures/implementations, such
2443 as Intel P4, IA-64 and AMD64.
2444 [Andy Polyakov]
2445
2446 *) New utility extract-section.pl. This can be used specify an alternative
2447 section number in a pod file instead of having to treat each file as
2448 a separate case in Makefile. This can be done by adding two lines to the
2449 pod file:
2450
2451 =for comment openssl_section:XXX
2452
2453 The blank line is mandatory.
2454
2455 [Steve Henson]
2456
2457 *) New arguments -certform, -keyform and -pass for s_client and s_server
2458 to allow alternative format key and certificate files and passphrase
2459 sources.
2460 [Steve Henson]
2461
2462 *) New structure X509_VERIFY_PARAM which combines current verify parameters,
2463 update associated structures and add various utility functions.
2464
2465 Add new policy related verify parameters, include policy checking in
2466 standard verify code. Enhance 'smime' application with extra parameters
2467 to support policy checking and print out.
2468 [Steve Henson]
2469
2470 *) Add a new engine to support VIA PadLock ACE extensions in the VIA C3
2471 Nehemiah processors. These extensions support AES encryption in hardware
2472 as well as RNG (though RNG support is currently disabled).
2473 [Michal Ludvig <michal@logix.cz>, with help from Andy Polyakov]
2474
2475 *) Deprecate BN_[get|set]_params() functions (they were ignored internally).
2476 [Geoff Thorpe]
2477
2478 *) New FIPS 180-2 algorithms, SHA-224/-256/-384/-512 are implemented.
2479 [Andy Polyakov and a number of other people]
2480
2481 *) Improved PowerPC platform support. Most notably BIGNUM assembler
2482 implementation contributed by IBM.
2483 [Suresh Chari, Peter Waltenberg, Andy Polyakov]
2484
2485 *) The new 'RSA_generate_key_ex' function now takes a BIGNUM for the public
2486 exponent rather than 'unsigned long'. There is a corresponding change to
2487 the new 'rsa_keygen' element of the RSA_METHOD structure.
2488 [Jelte Jansen, Geoff Thorpe]
2489
2490 *) Functionality for creating the initial serial number file is now
2491 moved from CA.pl to the 'ca' utility with a new option -create_serial.
2492
2493 (Before OpenSSL 0.9.7e, CA.pl used to initialize the serial
2494 number file to 1, which is bound to cause problems. To avoid
2495 the problems while respecting compatibility between different 0.9.7
2496 patchlevels, 0.9.7e employed 'openssl x509 -next_serial' in
2497 CA.pl for serial number initialization. With the new release 0.9.8,
2498 we can fix the problem directly in the 'ca' utility.)
2499 [Steve Henson]
2500
2501 *) Reduced header interdepencies by declaring more opaque objects in
2502 ossl_typ.h. As a consequence, including some headers (eg. engine.h) will
2503 give fewer recursive includes, which could break lazy source code - so
2504 this change is covered by the OPENSSL_NO_DEPRECATED symbol. As always,
2505 developers should define this symbol when building and using openssl to
2506 ensure they track the recommended behaviour, interfaces, [etc], but
2507 backwards-compatible behaviour prevails when this isn't defined.
2508 [Geoff Thorpe]
2509
2510 *) New function X509_POLICY_NODE_print() which prints out policy nodes.
2511 [Steve Henson]
2512
2513 *) Add new EVP function EVP_CIPHER_CTX_rand_key and associated functionality.
2514 This will generate a random key of the appropriate length based on the
2515 cipher context. The EVP_CIPHER can provide its own random key generation
2516 routine to support keys of a specific form. This is used in the des and
2517 3des routines to generate a key of the correct parity. Update S/MIME
2518 code to use new functions and hence generate correct parity DES keys.
2519 Add EVP_CHECK_DES_KEY #define to return an error if the key is not
2520 valid (weak or incorrect parity).
2521 [Steve Henson]
2522
2523 *) Add a local set of CRLs that can be used by X509_verify_cert() as well
2524 as looking them up. This is useful when the verified structure may contain
2525 CRLs, for example PKCS#7 signedData. Modify PKCS7_verify() to use any CRLs
2526 present unless the new PKCS7_NO_CRL flag is asserted.
2527 [Steve Henson]
2528
2529 *) Extend ASN1 oid configuration module. It now additionally accepts the
2530 syntax:
2531
2532 shortName = some long name, 1.2.3.4
2533 [Steve Henson]
2534
2535 *) Reimplemented the BN_CTX implementation. There is now no more static
2536 limitation on the number of variables it can handle nor the depth of the
2537 "stack" handling for BN_CTX_start()/BN_CTX_end() pairs. The stack
2538 information can now expand as required, and rather than having a single
2539 static array of bignums, BN_CTX now uses a linked-list of such arrays
2540 allowing it to expand on demand whilst maintaining the usefulness of
2541 BN_CTX's "bundling".
2542 [Geoff Thorpe]
2543
2544 *) Add a missing BN_CTX parameter to the 'rsa_mod_exp' callback in RSA_METHOD
2545 to allow all RSA operations to function using a single BN_CTX.
2546 [Geoff Thorpe]
2547
2548 *) Preliminary support for certificate policy evaluation and checking. This
2549 is initially intended to pass the tests outlined in "Conformance Testing
2550 of Relying Party Client Certificate Path Processing Logic" v1.07.
2551 [Steve Henson]
2552
2553 *) bn_dup_expand() has been deprecated, it was introduced in 0.9.7 and
2554 remained unused and not that useful. A variety of other little bignum
2555 tweaks and fixes have also been made continuing on from the audit (see
2556 below).
2557 [Geoff Thorpe]
2558
2559 *) Constify all or almost all d2i, c2i, s2i and r2i functions, along with
2560 associated ASN1, EVP and SSL functions and old ASN1 macros.
2561 [Richard Levitte]
2562
2563 *) BN_zero() only needs to set 'top' and 'neg' to zero for correct results,
2564 and this should never fail. So the return value from the use of
2565 BN_set_word() (which can fail due to needless expansion) is now deprecated;
2566 if OPENSSL_NO_DEPRECATED is defined, BN_zero() is a void macro.
2567 [Geoff Thorpe]
2568
2569 *) BN_CTX_get() should return zero-valued bignums, providing the same
2570 initialised value as BN_new().
2571 [Geoff Thorpe, suggested by Ulf Möller]
2572
2573 *) Support for inhibitAnyPolicy certificate extension.
2574 [Steve Henson]
2575
2576 *) An audit of the BIGNUM code is underway, for which debugging code is
2577 enabled when BN_DEBUG is defined. This makes stricter enforcements on what
2578 is considered valid when processing BIGNUMs, and causes execution to
2579 assert() when a problem is discovered. If BN_DEBUG_RAND is defined,
2580 further steps are taken to deliberately pollute unused data in BIGNUM
2581 structures to try and expose faulty code further on. For now, openssl will
2582 (in its default mode of operation) continue to tolerate the inconsistent
2583 forms that it has tolerated in the past, but authors and packagers should
2584 consider trying openssl and their own applications when compiled with
2585 these debugging symbols defined. It will help highlight potential bugs in
2586 their own code, and will improve the test coverage for OpenSSL itself. At
2587 some point, these tighter rules will become openssl's default to improve
2588 maintainability, though the assert()s and other overheads will remain only
2589 in debugging configurations. See bn.h for more details.
2590 [Geoff Thorpe, Nils Larsch, Ulf Möller]
2591
2592 *) BN_CTX_init() has been deprecated, as BN_CTX is an opaque structure
2593 that can only be obtained through BN_CTX_new() (which implicitly
2594 initialises it). The presence of this function only made it possible
2595 to overwrite an existing structure (and cause memory leaks).
2596 [Geoff Thorpe]
2597
2598 *) Because of the callback-based approach for implementing LHASH as a
2599 template type, lh_insert() adds opaque objects to hash-tables and
2600 lh_doall() or lh_doall_arg() are typically used with a destructor callback
2601 to clean up those corresponding objects before destroying the hash table
2602 (and losing the object pointers). So some over-zealous constifications in
2603 LHASH have been relaxed so that lh_insert() does not take (nor store) the
2604 objects as "const" and the lh_doall[_arg] callback wrappers are not
2605 prototyped to have "const" restrictions on the object pointers they are
2606 given (and so aren't required to cast them away any more).
2607 [Geoff Thorpe]
2608
2609 *) The tmdiff.h API was so ugly and minimal that our own timing utility
2610 (speed) prefers to use its own implementation. The two implementations
2611 haven't been consolidated as yet (volunteers?) but the tmdiff API has had
2612 its object type properly exposed (MS_TM) instead of casting to/from "char
2613 *". This may still change yet if someone realises MS_TM and "ms_time_***"
2614 aren't necessarily the greatest nomenclatures - but this is what was used
2615 internally to the implementation so I've used that for now.
2616 [Geoff Thorpe]
2617
2618 *) Ensure that deprecated functions do not get compiled when
2619 OPENSSL_NO_DEPRECATED is defined. Some "openssl" subcommands and a few of
2620 the self-tests were still using deprecated key-generation functions so
2621 these have been updated also.
2622 [Geoff Thorpe]
2623
2624 *) Reorganise PKCS#7 code to separate the digest location functionality
2625 into PKCS7_find_digest(), digest addtion into PKCS7_bio_add_digest().
2626 New function PKCS7_set_digest() to set the digest type for PKCS#7
2627 digestedData type. Add additional code to correctly generate the
2628 digestedData type and add support for this type in PKCS7 initialization
2629 functions.
2630 [Steve Henson]
2631
2632 *) New function PKCS7_set0_type_other() this initializes a PKCS7
2633 structure of type "other".
2634 [Steve Henson]
2635
2636 *) Fix prime generation loop in crypto/bn/bn_prime.pl by making
2637 sure the loop does correctly stop and breaking ("division by zero")
2638 modulus operations are not performed. The (pre-generated) prime
2639 table crypto/bn/bn_prime.h was already correct, but it could not be
2640 re-generated on some platforms because of the "division by zero"
2641 situation in the script.
2642 [Ralf S. Engelschall]
2643
2644 *) Update support for ECC-based TLS ciphersuites according to
2645 draft-ietf-tls-ecc-03.txt: the KDF1 key derivation function with
2646 SHA-1 now is only used for "small" curves (where the
2647 representation of a field element takes up to 24 bytes); for
2648 larger curves, the field element resulting from ECDH is directly
2649 used as premaster secret.
2650 [Douglas Stebila (Sun Microsystems Laboratories)]
2651
2652 *) Add code for kP+lQ timings to crypto/ec/ectest.c, and add SEC2
2653 curve secp160r1 to the tests.
2654 [Douglas Stebila (Sun Microsystems Laboratories)]
2655
2656 *) Add the possibility to load symbols globally with DSO.
2657 [Götz Babin-Ebell <babin-ebell@trustcenter.de> via Richard Levitte]
2658
2659 *) Add the functions ERR_set_mark() and ERR_pop_to_mark() for better
2660 control of the error stack.
2661 [Richard Levitte]
2662
2663 *) Add support for STORE in ENGINE.
2664 [Richard Levitte]
2665
2666 *) Add the STORE type. The intention is to provide a common interface
2667 to certificate and key stores, be they simple file-based stores, or
2668 HSM-type store, or LDAP stores, or...
2669 NOTE: The code is currently UNTESTED and isn't really used anywhere.
2670 [Richard Levitte]
2671
2672 *) Add a generic structure called OPENSSL_ITEM. This can be used to
2673 pass a list of arguments to any function as well as provide a way
2674 for a function to pass data back to the caller.
2675 [Richard Levitte]
2676
2677 *) Add the functions BUF_strndup() and BUF_memdup(). BUF_strndup()
2678 works like BUF_strdup() but can be used to duplicate a portion of
2679 a string. The copy gets NUL-terminated. BUF_memdup() duplicates
2680 a memory area.
2681 [Richard Levitte]
2682
2683 *) Add the function sk_find_ex() which works like sk_find(), but will
2684 return an index to an element even if an exact match couldn't be
2685 found. The index is guaranteed to point at the element where the
2686 searched-for key would be inserted to preserve sorting order.
2687 [Richard Levitte]
2688
2689 *) Add the function OBJ_bsearch_ex() which works like OBJ_bsearch() but
2690 takes an extra flags argument for optional functionality. Currently,
2691 the following flags are defined:
2692
2693 OBJ_BSEARCH_VALUE_ON_NOMATCH
2694 This one gets OBJ_bsearch_ex() to return a pointer to the first
2695 element where the comparing function returns a negative or zero
2696 number.
2697
2698 OBJ_BSEARCH_FIRST_VALUE_ON_MATCH
2699 This one gets OBJ_bsearch_ex() to return a pointer to the first
2700 element where the comparing function returns zero. This is useful
2701 if there are more than one element where the comparing function
2702 returns zero.
2703 [Richard Levitte]
2704
2705 *) Make it possible to create self-signed certificates with 'openssl ca'
2706 in such a way that the self-signed certificate becomes part of the
2707 CA database and uses the same mechanisms for serial number generation
2708 as all other certificate signing. The new flag '-selfsign' enables
2709 this functionality. Adapt CA.sh and CA.pl.in.
2710 [Richard Levitte]
2711
2712 *) Add functionality to check the public key of a certificate request
2713 against a given private. This is useful to check that a certificate
2714 request can be signed by that key (self-signing).
2715 [Richard Levitte]
2716
2717 *) Make it possible to have multiple active certificates with the same
2718 subject in the CA index file. This is done only if the keyword
2719 'unique_subject' is set to 'no' in the main CA section (default
2720 if 'CA_default') of the configuration file. The value is saved
2721 with the database itself in a separate index attribute file,
2722 named like the index file with '.attr' appended to the name.
2723 [Richard Levitte]
2724
2725 *) Generate muti valued AVAs using '+' notation in config files for
2726 req and dirName.
2727 [Steve Henson]
2728
2729 *) Support for nameConstraints certificate extension.
2730 [Steve Henson]
2731
2732 *) Support for policyConstraints certificate extension.
2733 [Steve Henson]
2734
2735 *) Support for policyMappings certificate extension.
2736 [Steve Henson]
2737
2738 *) Make sure the default DSA_METHOD implementation only uses its
2739 dsa_mod_exp() and/or bn_mod_exp() handlers if they are non-NULL,
2740 and change its own handlers to be NULL so as to remove unnecessary
2741 indirection. This lets alternative implementations fallback to the
2742 default implementation more easily.
2743 [Geoff Thorpe]
2744
2745 *) Support for directoryName in GeneralName related extensions
2746 in config files.
2747 [Steve Henson]
2748
2749 *) Make it possible to link applications using Makefile.shared.
2750 Make that possible even when linking against static libraries!
2751 [Richard Levitte]
2752
2753 *) Support for single pass processing for S/MIME signing. This now
2754 means that S/MIME signing can be done from a pipe, in addition
2755 cleartext signing (multipart/signed type) is effectively streaming
2756 and the signed data does not need to be all held in memory.
2757
2758 This is done with a new flag PKCS7_STREAM. When this flag is set
2759 PKCS7_sign() only initializes the PKCS7 structure and the actual signing
2760 is done after the data is output (and digests calculated) in
2761 SMIME_write_PKCS7().
2762 [Steve Henson]
2763
2764 *) Add full support for -rpath/-R, both in shared libraries and
2765 applications, at least on the platforms where it's known how
2766 to do it.
2767 [Richard Levitte]
2768
2769 *) In crypto/ec/ec_mult.c, implement fast point multiplication with
2770 precomputation, based on wNAF splitting: EC_GROUP_precompute_mult()
2771 will now compute a table of multiples of the generator that
2772 makes subsequent invocations of EC_POINTs_mul() or EC_POINT_mul()
2773 faster (notably in the case of a single point multiplication,
2774 scalar * generator).
2775 [Nils Larsch, Bodo Moeller]
2776
2777 *) IPv6 support for certificate extensions. The various extensions
2778 which use the IP:a.b.c.d can now take IPv6 addresses using the
2779 formats of RFC1884 2.2 . IPv6 addresses are now also displayed
2780 correctly.
2781 [Steve Henson]
2782
2783 *) Added an ENGINE that implements RSA by performing private key
2784 exponentiations with the GMP library. The conversions to and from
2785 GMP's mpz_t format aren't optimised nor are any montgomery forms
2786 cached, and on x86 it appears OpenSSL's own performance has caught up.
2787 However there are likely to be other architectures where GMP could
2788 provide a boost. This ENGINE is not built in by default, but it can be
2789 specified at Configure time and should be accompanied by the necessary
2790 linker additions, eg;
2791 ./config -DOPENSSL_USE_GMP -lgmp
2792 [Geoff Thorpe]
2793
2794 *) "openssl engine" will not display ENGINE/DSO load failure errors when
2795 testing availability of engines with "-t" - the old behaviour is
2796 produced by increasing the feature's verbosity with "-tt".
2797 [Geoff Thorpe]
2798
2799 *) ECDSA routines: under certain error conditions uninitialized BN objects
2800 could be freed. Solution: make sure initialization is performed early
2801 enough. (Reported and fix supplied by Nils Larsch <nla@trustcenter.de>
2802 via PR#459)
2803 [Lutz Jaenicke]
2804
2805 *) Key-generation can now be implemented in RSA_METHOD, DSA_METHOD
2806 and DH_METHOD (eg. by ENGINE implementations) to override the normal
2807 software implementations. For DSA and DH, parameter generation can
2808 also be overriden by providing the appropriate method callbacks.
2809 [Geoff Thorpe]
2810
2811 *) Change the "progress" mechanism used in key-generation and
2812 primality testing to functions that take a new BN_GENCB pointer in
2813 place of callback/argument pairs. The new API functions have "_ex"
2814 postfixes and the older functions are reimplemented as wrappers for
2815 the new ones. The OPENSSL_NO_DEPRECATED symbol can be used to hide
2816 declarations of the old functions to help (graceful) attempts to
2817 migrate to the new functions. Also, the new key-generation API
2818 functions operate on a caller-supplied key-structure and return
2819 success/failure rather than returning a key or NULL - this is to
2820 help make "keygen" another member function of RSA_METHOD etc.
2821
2822 Example for using the new callback interface:
2823
2824 int (*my_callback)(int a, int b, BN_GENCB *cb) = ...;
2825 void *my_arg = ...;
2826 BN_GENCB my_cb;
2827
2828 BN_GENCB_set(&my_cb, my_callback, my_arg);
2829
2830 return BN_is_prime_ex(some_bignum, BN_prime_checks, NULL, &cb);
2831 /* For the meaning of a, b in calls to my_callback(), see the
2832 * documentation of the function that calls the callback.
2833 * cb will point to my_cb; my_arg can be retrieved as cb->arg.
2834 * my_callback should return 1 if it wants BN_is_prime_ex()
2835 * to continue, or 0 to stop.
2836 */
2837
2838 [Geoff Thorpe]
2839
2840 *) Change the ZLIB compression method to be stateful, and make it
2841 available to TLS with the number defined in
2842 draft-ietf-tls-compression-04.txt.
2843 [Richard Levitte]
2844
2845 *) Add the ASN.1 structures and functions for CertificatePair, which
2846 is defined as follows (according to X.509_4thEditionDraftV6.pdf):
2847
2848 CertificatePair ::= SEQUENCE {
2849 forward [0] Certificate OPTIONAL,
2850 reverse [1] Certificate OPTIONAL,
2851 -- at least one of the pair shall be present -- }
2852
2853 Also implement the PEM functions to read and write certificate
2854 pairs, and defined the PEM tag as "CERTIFICATE PAIR".
2855
2856 This needed to be defined, mostly for the sake of the LDAP
2857 attribute crossCertificatePair, but may prove useful elsewhere as
2858 well.
2859 [Richard Levitte]
2860
2861 *) Make it possible to inhibit symlinking of shared libraries in
2862 Makefile.shared, for Cygwin's sake.
2863 [Richard Levitte]
2864
2865 *) Extend the BIGNUM API by creating a function
2866 void BN_set_negative(BIGNUM *a, int neg);
2867 and a macro that behave like
2868 int BN_is_negative(const BIGNUM *a);
2869
2870 to avoid the need to access 'a->neg' directly in applications.
2871 [Nils Larsch]
2872
2873 *) Implement fast modular reduction for pseudo-Mersenne primes
2874 used in NIST curves (crypto/bn/bn_nist.c, crypto/ec/ecp_nist.c).
2875 EC_GROUP_new_curve_GFp() will now automatically use this
2876 if applicable.
2877 [Nils Larsch <nla@trustcenter.de>]
2878
2879 *) Add new lock type (CRYPTO_LOCK_BN).
2880 [Bodo Moeller]
2881
2882 *) Change the ENGINE framework to automatically load engines
2883 dynamically from specific directories unless they could be
2884 found to already be built in or loaded. Move all the
2885 current engines except for the cryptodev one to a new
2886 directory engines/.
2887 The engines in engines/ are built as shared libraries if
2888 the "shared" options was given to ./Configure or ./config.
2889 Otherwise, they are inserted in libcrypto.a.
2890 /usr/local/ssl/engines is the default directory for dynamic
2891 engines, but that can be overriden at configure time through
2892 the usual use of --prefix and/or --openssldir, and at run
2893 time with the environment variable OPENSSL_ENGINES.
2894 [Geoff Thorpe and Richard Levitte]
2895
2896 *) Add Makefile.shared, a helper makefile to build shared
2897 libraries. Addapt Makefile.org.
2898 [Richard Levitte]
2899
2900 *) Add version info to Win32 DLLs.
2901 [Peter 'Luna' Runestig" <peter@runestig.com>]
2902
2903 *) Add new 'medium level' PKCS#12 API. Certificates and keys
2904 can be added using this API to created arbitrary PKCS#12
2905 files while avoiding the low level API.
2906
2907 New options to PKCS12_create(), key or cert can be NULL and
2908 will then be omitted from the output file. The encryption
2909 algorithm NIDs can be set to -1 for no encryption, the mac
2910 iteration count can be set to 0 to omit the mac.
2911
2912 Enhance pkcs12 utility by making the -nokeys and -nocerts
2913 options work when creating a PKCS#12 file. New option -nomac
2914 to omit the mac, NONE can be set for an encryption algorithm.
2915 New code is modified to use the enhanced PKCS12_create()
2916 instead of the low level API.
2917 [Steve Henson]
2918
2919 *) Extend ASN1 encoder to support indefinite length constructed
2920 encoding. This can output sequences tags and octet strings in
2921 this form. Modify pk7_asn1.c to support indefinite length
2922 encoding. This is experimental and needs additional code to
2923 be useful, such as an ASN1 bio and some enhanced streaming
2924 PKCS#7 code.
2925
2926 Extend template encode functionality so that tagging is passed
2927 down to the template encoder.
2928 [Steve Henson]
2929
2930 *) Let 'openssl req' fail if an argument to '-newkey' is not
2931 recognized instead of using RSA as a default.
2932 [Bodo Moeller]
2933
2934 *) Add support for ECC-based ciphersuites from draft-ietf-tls-ecc-01.txt.
2935 As these are not official, they are not included in "ALL";
2936 the "ECCdraft" ciphersuite group alias can be used to select them.
2937 [Vipul Gupta and Sumit Gupta (Sun Microsystems Laboratories)]
2938
2939 *) Add ECDH engine support.
2940 [Nils Gura and Douglas Stebila (Sun Microsystems Laboratories)]
2941
2942 *) Add ECDH in new directory crypto/ecdh/.
2943 [Douglas Stebila (Sun Microsystems Laboratories)]
2944
2945 *) Let BN_rand_range() abort with an error after 100 iterations
2946 without success (which indicates a broken PRNG).
2947 [Bodo Moeller]
2948
2949 *) Change BN_mod_sqrt() so that it verifies that the input value
2950 is really the square of the return value. (Previously,
2951 BN_mod_sqrt would show GIGO behaviour.)
2952 [Bodo Moeller]
2953
2954 *) Add named elliptic curves over binary fields from X9.62, SECG,
2955 and WAP/WTLS; add OIDs that were still missing.
2956
2957 [Sheueling Chang Shantz and Douglas Stebila
2958 (Sun Microsystems Laboratories)]
2959
2960 *) Extend the EC library for elliptic curves over binary fields
2961 (new files ec2_smpl.c, ec2_smpt.c, ec2_mult.c in crypto/ec/).
2962 New EC_METHOD:
2963
2964 EC_GF2m_simple_method
2965
2966 New API functions:
2967
2968 EC_GROUP_new_curve_GF2m
2969 EC_GROUP_set_curve_GF2m
2970 EC_GROUP_get_curve_GF2m
2971 EC_POINT_set_affine_coordinates_GF2m
2972 EC_POINT_get_affine_coordinates_GF2m
2973 EC_POINT_set_compressed_coordinates_GF2m
2974
2975 Point compression for binary fields is disabled by default for
2976 patent reasons (compile with OPENSSL_EC_BIN_PT_COMP defined to
2977 enable it).
2978
2979 As binary polynomials are represented as BIGNUMs, various members
2980 of the EC_GROUP and EC_POINT data structures can be shared
2981 between the implementations for prime fields and binary fields;
2982 the above ..._GF2m functions (except for EX_GROUP_new_curve_GF2m)
2983 are essentially identical to their ..._GFp counterparts.
2984 (For simplicity, the '..._GFp' prefix has been dropped from
2985 various internal method names.)
2986
2987 An internal 'field_div' method (similar to 'field_mul' and
2988 'field_sqr') has been added; this is used only for binary fields.
2989
2990 [Sheueling Chang Shantz and Douglas Stebila
2991 (Sun Microsystems Laboratories)]
2992
2993 *) Optionally dispatch EC_POINT_mul(), EC_POINT_precompute_mult()
2994 through methods ('mul', 'precompute_mult').
2995
2996 The generic implementations (now internally called 'ec_wNAF_mul'
2997 and 'ec_wNAF_precomputed_mult') remain the default if these
2998 methods are undefined.
2999
3000 [Sheueling Chang Shantz and Douglas Stebila
3001 (Sun Microsystems Laboratories)]
3002
3003 *) New function EC_GROUP_get_degree, which is defined through
3004 EC_METHOD. For curves over prime fields, this returns the bit
3005 length of the modulus.
3006
3007 [Sheueling Chang Shantz and Douglas Stebila
3008 (Sun Microsystems Laboratories)]
3009
3010 *) New functions EC_GROUP_dup, EC_POINT_dup.
3011 (These simply call ..._new and ..._copy).
3012
3013 [Sheueling Chang Shantz and Douglas Stebila
3014 (Sun Microsystems Laboratories)]
3015
3016 *) Add binary polynomial arithmetic software in crypto/bn/bn_gf2m.c.
3017 Polynomials are represented as BIGNUMs (where the sign bit is not
3018 used) in the following functions [macros]:
3019
3020 BN_GF2m_add
3021 BN_GF2m_sub [= BN_GF2m_add]
3022 BN_GF2m_mod [wrapper for BN_GF2m_mod_arr]
3023 BN_GF2m_mod_mul [wrapper for BN_GF2m_mod_mul_arr]
3024 BN_GF2m_mod_sqr [wrapper for BN_GF2m_mod_sqr_arr]
3025 BN_GF2m_mod_inv
3026 BN_GF2m_mod_exp [wrapper for BN_GF2m_mod_exp_arr]
3027 BN_GF2m_mod_sqrt [wrapper for BN_GF2m_mod_sqrt_arr]
3028 BN_GF2m_mod_solve_quad [wrapper for BN_GF2m_mod_solve_quad_arr]
3029 BN_GF2m_cmp [= BN_ucmp]
3030
3031 (Note that only the 'mod' functions are actually for fields GF(2^m).
3032 BN_GF2m_add() is misnomer, but this is for the sake of consistency.)
3033
3034 For some functions, an the irreducible polynomial defining a
3035 field can be given as an 'unsigned int[]' with strictly
3036 decreasing elements giving the indices of those bits that are set;
3037 i.e., p[] represents the polynomial
3038 f(t) = t^p[0] + t^p[1] + ... + t^p[k]
3039 where
3040 p[0] > p[1] > ... > p[k] = 0.
3041 This applies to the following functions:
3042
3043 BN_GF2m_mod_arr
3044 BN_GF2m_mod_mul_arr
3045 BN_GF2m_mod_sqr_arr
3046 BN_GF2m_mod_inv_arr [wrapper for BN_GF2m_mod_inv]
3047 BN_GF2m_mod_div_arr [wrapper for BN_GF2m_mod_div]
3048 BN_GF2m_mod_exp_arr
3049 BN_GF2m_mod_sqrt_arr
3050 BN_GF2m_mod_solve_quad_arr
3051 BN_GF2m_poly2arr
3052 BN_GF2m_arr2poly
3053
3054 Conversion can be performed by the following functions:
3055
3056 BN_GF2m_poly2arr
3057 BN_GF2m_arr2poly
3058
3059 bntest.c has additional tests for binary polynomial arithmetic.
3060
3061 Two implementations for BN_GF2m_mod_div() are available.
3062 The default algorithm simply uses BN_GF2m_mod_inv() and
3063 BN_GF2m_mod_mul(). The alternative algorithm is compiled in only
3064 if OPENSSL_SUN_GF2M_DIV is defined (patent pending; read the
3065 copyright notice in crypto/bn/bn_gf2m.c before enabling it).
3066
3067 [Sheueling Chang Shantz and Douglas Stebila
3068 (Sun Microsystems Laboratories)]
3069
3070 *) Add new error code 'ERR_R_DISABLED' that can be used when some
3071 functionality is disabled at compile-time.
3072 [Douglas Stebila <douglas.stebila@sun.com>]
3073
3074 *) Change default behaviour of 'openssl asn1parse' so that more
3075 information is visible when viewing, e.g., a certificate:
3076
3077 Modify asn1_parse2 (crypto/asn1/asn1_par.c) so that in non-'dump'
3078 mode the content of non-printable OCTET STRINGs is output in a
3079 style similar to INTEGERs, but with '[HEX DUMP]' prepended to
3080 avoid the appearance of a printable string.
3081 [Nils Larsch <nla@trustcenter.de>]
3082
3083 *) Add 'asn1_flag' and 'asn1_form' member to EC_GROUP with access
3084 functions
3085 EC_GROUP_set_asn1_flag()
3086 EC_GROUP_get_asn1_flag()
3087 EC_GROUP_set_point_conversion_form()
3088 EC_GROUP_get_point_conversion_form()
3089 These control ASN1 encoding details:
3090 - Curves (i.e., groups) are encoded explicitly unless asn1_flag
3091 has been set to OPENSSL_EC_NAMED_CURVE.
3092 - Points are encoded in uncompressed form by default; options for
3093 asn1_for are as for point2oct, namely
3094 POINT_CONVERSION_COMPRESSED
3095 POINT_CONVERSION_UNCOMPRESSED
3096 POINT_CONVERSION_HYBRID
3097
3098 Also add 'seed' and 'seed_len' members to EC_GROUP with access
3099 functions
3100 EC_GROUP_set_seed()
3101 EC_GROUP_get0_seed()
3102 EC_GROUP_get_seed_len()
3103 This is used only for ASN1 purposes (so far).
3104 [Nils Larsch <nla@trustcenter.de>]
3105
3106 *) Add 'field_type' member to EC_METHOD, which holds the NID
3107 of the appropriate field type OID. The new function
3108 EC_METHOD_get_field_type() returns this value.
3109 [Nils Larsch <nla@trustcenter.de>]
3110
3111 *) Add functions
3112 EC_POINT_point2bn()
3113 EC_POINT_bn2point()
3114 EC_POINT_point2hex()
3115 EC_POINT_hex2point()
3116 providing useful interfaces to EC_POINT_point2oct() and
3117 EC_POINT_oct2point().
3118 [Nils Larsch <nla@trustcenter.de>]
3119
3120 *) Change internals of the EC library so that the functions
3121 EC_GROUP_set_generator()
3122 EC_GROUP_get_generator()
3123 EC_GROUP_get_order()
3124 EC_GROUP_get_cofactor()
3125 are implemented directly in crypto/ec/ec_lib.c and not dispatched
3126 to methods, which would lead to unnecessary code duplication when
3127 adding different types of curves.
3128 [Nils Larsch <nla@trustcenter.de> with input by Bodo Moeller]
3129
3130 *) Implement compute_wNAF (crypto/ec/ec_mult.c) without BIGNUM
3131 arithmetic, and such that modified wNAFs are generated
3132 (which avoid length expansion in many cases).
3133 [Bodo Moeller]
3134
3135 *) Add a function EC_GROUP_check_discriminant() (defined via
3136 EC_METHOD) that verifies that the curve discriminant is non-zero.
3137
3138 Add a function EC_GROUP_check() that makes some sanity tests
3139 on a EC_GROUP, its generator and order. This includes
3140 EC_GROUP_check_discriminant().
3141 [Nils Larsch <nla@trustcenter.de>]
3142
3143 *) Add ECDSA in new directory crypto/ecdsa/.
3144
3145 Add applications 'openssl ecparam' and 'openssl ecdsa'
3146 (these are based on 'openssl dsaparam' and 'openssl dsa').
3147
3148 ECDSA support is also included in various other files across the
3149 library. Most notably,
3150 - 'openssl req' now has a '-newkey ecdsa:file' option;
3151 - EVP_PKCS82PKEY (crypto/evp/evp_pkey.c) now can handle ECDSA;
3152 - X509_PUBKEY_get (crypto/asn1/x_pubkey.c) and
3153 d2i_PublicKey (crypto/asn1/d2i_pu.c) have been modified to make
3154 them suitable for ECDSA where domain parameters must be
3155 extracted before the specific public key;
3156 - ECDSA engine support has been added.
3157 [Nils Larsch <nla@trustcenter.de>]
3158
3159 *) Include some named elliptic curves, and add OIDs from X9.62,
3160 SECG, and WAP/WTLS. Each curve can be obtained from the new
3161 function
3162 EC_GROUP_new_by_curve_name(),
3163 and the list of available named curves can be obtained with
3164 EC_get_builtin_curves().
3165 Also add a 'curve_name' member to EC_GROUP objects, which can be
3166 accessed via
3167 EC_GROUP_set_curve_name()
3168 EC_GROUP_get_curve_name()
3169 [Nils Larsch <larsch@trustcenter.de, Bodo Moeller]
3170
3171 *) Remove a few calls to bn_wexpand() in BN_sqr() (the one in there
3172 was actually never needed) and in BN_mul(). The removal in BN_mul()
3173 required a small change in bn_mul_part_recursive() and the addition
3174 of the functions bn_cmp_part_words(), bn_sub_part_words() and
3175 bn_add_part_words(), which do the same thing as bn_cmp_words(),
3176 bn_sub_words() and bn_add_words() except they take arrays with
3177 differing sizes.
3178 [Richard Levitte]
3179
3180 Changes between 0.9.7l and 0.9.7m [23 Feb 2007]
3181
3182 *) Cleanse PEM buffers before freeing them since they may contain
3183 sensitive data.
3184 [Benjamin Bennett <ben@psc.edu>]
3185
3186 *) Include "!eNULL" in SSL_DEFAULT_CIPHER_LIST to make sure that
3187 a ciphersuite string such as "DEFAULT:RSA" cannot enable
3188 authentication-only ciphersuites.
3189 [Bodo Moeller]
3190
3191 *) Since AES128 and AES256 share a single mask bit in the logic of
3192 ssl/ssl_ciph.c, the code for masking out disabled ciphers needs a
3193 kludge to work properly if AES128 is available and AES256 isn't.
3194 [Victor Duchovni]
3195
3196 *) Expand security boundary to match 1.1.1 module.
3197 [Steve Henson]
3198
3199 *) Remove redundant features: hash file source, editing of test vectors
3200 modify fipsld to use external fips_premain.c signature.
3201 [Steve Henson]
3202
3203 *) New perl script mkfipsscr.pl to create shell scripts or batch files to
3204 run algorithm test programs.
3205 [Steve Henson]
3206
3207 *) Make algorithm test programs more tolerant of whitespace.
3208 [Steve Henson]
3209
3210 *) Have SSL/TLS server implementation tolerate "mismatched" record
3211 protocol version while receiving ClientHello even if the
3212 ClientHello is fragmented. (The server can't insist on the
3213 particular protocol version it has chosen before the ServerHello
3214 message has informed the client about his choice.)
3215 [Bodo Moeller]
3216
3217 *) Load error codes if they are not already present instead of using a
3218 static variable. This allows them to be cleanly unloaded and reloaded.
3219 [Steve Henson]
3220
3221 Changes between 0.9.7k and 0.9.7l [28 Sep 2006]
3222
3223 *) Introduce limits to prevent malicious keys being able to
3224 cause a denial of service. (CVE-2006-2940)
3225 [Steve Henson, Bodo Moeller]
3226
3227 *) Fix ASN.1 parsing of certain invalid structures that can result
3228 in a denial of service. (CVE-2006-2937) [Steve Henson]
3229
3230 *) Fix buffer overflow in SSL_get_shared_ciphers() function.
3231 (CVE-2006-3738) [Tavis Ormandy and Will Drewry, Google Security Team]
3232
3233 *) Fix SSL client code which could crash if connecting to a
3234 malicious SSLv2 server. (CVE-2006-4343)
3235 [Tavis Ormandy and Will Drewry, Google Security Team]
3236
3237 *) Change ciphersuite string processing so that an explicit
3238 ciphersuite selects this one ciphersuite (so that "AES256-SHA"
3239 will no longer include "AES128-SHA"), and any other similar
3240 ciphersuite (same bitmap) from *other* protocol versions (so that
3241 "RC4-MD5" will still include both the SSL 2.0 ciphersuite and the
3242 SSL 3.0/TLS 1.0 ciphersuite). This is a backport combining
3243 changes from 0.9.8b and 0.9.8d.
3244 [Bodo Moeller]
3245
3246 Changes between 0.9.7j and 0.9.7k [05 Sep 2006]
3247
3248 *) Avoid PKCS #1 v1.5 signature attack discovered by Daniel Bleichenbacher
3249 (CVE-2006-4339) [Ben Laurie and Google Security Team]
3250
3251 *) Change the Unix randomness entropy gathering to use poll() when
3252 possible instead of select(), since the latter has some
3253 undesirable limitations.
3254 [Darryl Miles via Richard Levitte and Bodo Moeller]
3255
3256 *) Disable rogue ciphersuites:
3257
3258 - SSLv2 0x08 0x00 0x80 ("RC4-64-MD5")
3259 - SSLv3/TLSv1 0x00 0x61 ("EXP1024-RC2-CBC-MD5")
3260 - SSLv3/TLSv1 0x00 0x60 ("EXP1024-RC4-MD5")
3261
3262 The latter two were purportedly from
3263 draft-ietf-tls-56-bit-ciphersuites-0[01].txt, but do not really
3264 appear there.
3265
3266 Also deactive the remaining ciphersuites from
3267 draft-ietf-tls-56-bit-ciphersuites-01.txt. These are just as
3268 unofficial, and the ID has long expired.
3269 [Bodo Moeller]
3270
3271 *) Fix RSA blinding Heisenbug (problems sometimes occured on
3272 dual-core machines) and other potential thread-safety issues.
3273 [Bodo Moeller]
3274
3275 Changes between 0.9.7i and 0.9.7j [04 May 2006]
3276
3277 *) Adapt fipsld and the build system to link against the validated FIPS
3278 module in FIPS mode.
3279 [Steve Henson]
3280
3281 *) Fixes for VC++ 2005 build under Windows.
3282 [Steve Henson]
3283
3284 *) Add new Windows build target VC-32-GMAKE for VC++. This uses GNU make
3285 from a Windows bash shell such as MSYS. It is autodetected from the
3286 "config" script when run from a VC++ environment. Modify standard VC++
3287 build to use fipscanister.o from the GNU make build.
3288 [Steve Henson]
3289
3290 Changes between 0.9.7h and 0.9.7i [14 Oct 2005]
3291
3292 *) Wrapped the definition of EVP_MAX_MD_SIZE in a #ifdef OPENSSL_FIPS.
3293 The value now differs depending on if you build for FIPS or not.
3294 BEWARE! A program linked with a shared FIPSed libcrypto can't be
3295 safely run with a non-FIPSed libcrypto, as it may crash because of
3296 the difference induced by this change.
3297 [Andy Polyakov]
3298
3299 Changes between 0.9.7g and 0.9.7h [11 Oct 2005]
3300
3301 *) Remove the functionality of SSL_OP_MSIE_SSLV2_RSA_PADDING
3302 (part of SSL_OP_ALL). This option used to disable the
3303 countermeasure against man-in-the-middle protocol-version
3304 rollback in the SSL 2.0 server implementation, which is a bad
3305 idea. (CVE-2005-2969)
3306
3307 [Bodo Moeller; problem pointed out by Yutaka Oiwa (Research Center
3308 for Information Security, National Institute of Advanced Industrial
3309 Science and Technology [AIST], Japan)]
3310
3311 *) Minimal support for X9.31 signatures and PSS padding modes. This is
3312 mainly for FIPS compliance and not fully integrated at this stage.
3313 [Steve Henson]
3314
3315 *) For DSA signing, unless DSA_FLAG_NO_EXP_CONSTTIME is set, perform
3316 the exponentiation using a fixed-length exponent. (Otherwise,
3317 the information leaked through timing could expose the secret key
3318 after many signatures; cf. Bleichenbacher's attack on DSA with
3319 biased k.)
3320 [Bodo Moeller]
3321
3322 *) Make a new fixed-window mod_exp implementation the default for
3323 RSA, DSA, and DH private-key operations so that the sequence of
3324 squares and multiplies and the memory access pattern are
3325 independent of the particular secret key. This will mitigate
3326 cache-timing and potential related attacks.
3327
3328 BN_mod_exp_mont_consttime() is the new exponentiation implementation,
3329 and this is automatically used by BN_mod_exp_mont() if the new flag
3330 BN_FLG_EXP_CONSTTIME is set for the exponent. RSA, DSA, and DH
3331 will use this BN flag for private exponents unless the flag
3332 RSA_FLAG_NO_EXP_CONSTTIME, DSA_FLAG_NO_EXP_CONSTTIME, or
3333 DH_FLAG_NO_EXP_CONSTTIME, respectively, is set.
3334
3335 [Matthew D Wood (Intel Corp), with some changes by Bodo Moeller]
3336
3337 *) Change the client implementation for SSLv23_method() and
3338 SSLv23_client_method() so that is uses the SSL 3.0/TLS 1.0
3339 Client Hello message format if the SSL_OP_NO_SSLv2 option is set.
3340 (Previously, the SSL 2.0 backwards compatible Client Hello
3341 message format would be used even with SSL_OP_NO_SSLv2.)
3342 [Bodo Moeller]
3343
3344 *) Add support for smime-type MIME parameter in S/MIME messages which some
3345 clients need.
3346 [Steve Henson]
3347
3348 *) New function BN_MONT_CTX_set_locked() to set montgomery parameters in
3349 a threadsafe manner. Modify rsa code to use new function and add calls
3350 to dsa and dh code (which had race conditions before).
3351 [Steve Henson]
3352
3353 *) Include the fixed error library code in the C error file definitions
3354 instead of fixing them up at runtime. This keeps the error code
3355 structures constant.
3356 [Steve Henson]
3357
3358 Changes between 0.9.7f and 0.9.7g [11 Apr 2005]
3359
3360 [NB: OpenSSL 0.9.7h and later 0.9.7 patch levels were released after
3361 OpenSSL 0.9.8.]
3362
3363 *) Fixes for newer kerberos headers. NB: the casts are needed because
3364 the 'length' field is signed on one version and unsigned on another
3365 with no (?) obvious way to tell the difference, without these VC++
3366 complains. Also the "definition" of FAR (blank) is no longer included
3367 nor is the error ENOMEM. KRB5_PRIVATE has to be set to 1 to pick up
3368 some needed definitions.
3369 [Steve Henson]
3370
3371 *) Undo Cygwin change.
3372 [Ulf Möller]
3373
3374 *) Added support for proxy certificates according to RFC 3820.
3375 Because they may be a security thread to unaware applications,
3376 they must be explicitely allowed in run-time. See
3377 docs/HOWTO/proxy_certificates.txt for further information.
3378 [Richard Levitte]
3379
3380 Changes between 0.9.7e and 0.9.7f [22 Mar 2005]
3381
3382 *) Use (SSL_RANDOM_VALUE - 4) bytes of pseudo random data when generating
3383 server and client random values. Previously
3384 (SSL_RANDOM_VALUE - sizeof(time_t)) would be used which would result in
3385 less random data when sizeof(time_t) > 4 (some 64 bit platforms).
3386
3387 This change has negligible security impact because:
3388
3389 1. Server and client random values still have 24 bytes of pseudo random
3390 data.
3391
3392 2. Server and client random values are sent in the clear in the initial
3393 handshake.
3394
3395 3. The master secret is derived using the premaster secret (48 bytes in
3396 size for static RSA ciphersuites) as well as client server and random
3397 values.
3398
3399 The OpenSSL team would like to thank the UK NISCC for bringing this issue
3400 to our attention.
3401
3402 [Stephen Henson, reported by UK NISCC]
3403
3404 *) Use Windows randomness collection on Cygwin.
3405 [Ulf Möller]
3406
3407 *) Fix hang in EGD/PRNGD query when communication socket is closed
3408 prematurely by EGD/PRNGD.
3409 [Darren Tucker <dtucker@zip.com.au> via Lutz Jänicke, resolves #1014]
3410
3411 *) Prompt for pass phrases when appropriate for PKCS12 input format.
3412 [Steve Henson]
3413
3414 *) Back-port of selected performance improvements from development
3415 branch, as well as improved support for PowerPC platforms.
3416 [Andy Polyakov]
3417
3418 *) Add lots of checks for memory allocation failure, error codes to indicate
3419 failure and freeing up memory if a failure occurs.
3420 [Nauticus Networks SSL Team <openssl@nauticusnet.com>, Steve Henson]
3421
3422 *) Add new -passin argument to dgst.
3423 [Steve Henson]
3424
3425 *) Perform some character comparisons of different types in X509_NAME_cmp:
3426 this is needed for some certificates that reencode DNs into UTF8Strings
3427 (in violation of RFC3280) and can't or wont issue name rollover
3428 certificates.
3429 [Steve Henson]
3430
3431 *) Make an explicit check during certificate validation to see that
3432 the CA setting in each certificate on the chain is correct. As a
3433 side effect always do the following basic checks on extensions,
3434 not just when there's an associated purpose to the check:
3435
3436 - if there is an unhandled critical extension (unless the user
3437 has chosen to ignore this fault)
3438 - if the path length has been exceeded (if one is set at all)
3439 - that certain extensions fit the associated purpose (if one has
3440 been given)
3441 [Richard Levitte]
3442
3443 Changes between 0.9.7d and 0.9.7e [25 Oct 2004]
3444
3445 *) Avoid a race condition when CRLs are checked in a multi threaded
3446 environment. This would happen due to the reordering of the revoked
3447 entries during signature checking and serial number lookup. Now the
3448 encoding is cached and the serial number sort performed under a lock.
3449 Add new STACK function sk_is_sorted().
3450 [Steve Henson]
3451
3452 *) Add Delta CRL to the extension code.
3453 [Steve Henson]
3454
3455 *) Various fixes to s3_pkt.c so alerts are sent properly.
3456 [David Holmes <d.holmes@f5.com>]
3457
3458 *) Reduce the chances of duplicate issuer name and serial numbers (in
3459 violation of RFC3280) using the OpenSSL certificate creation utilities.
3460 This is done by creating a random 64 bit value for the initial serial
3461 number when a serial number file is created or when a self signed
3462 certificate is created using 'openssl req -x509'. The initial serial
3463 number file is created using 'openssl x509 -next_serial' in CA.pl
3464 rather than being initialized to 1.
3465 [Steve Henson]
3466
3467 Changes between 0.9.7c and 0.9.7d [17 Mar 2004]
3468
3469 *) Fix null-pointer assignment in do_change_cipher_spec() revealed
3470 by using the Codenomicon TLS Test Tool (CVE-2004-0079)
3471 [Joe Orton, Steve Henson]
3472
3473 *) Fix flaw in SSL/TLS handshaking when using Kerberos ciphersuites
3474 (CVE-2004-0112)
3475 [Joe Orton, Steve Henson]
3476
3477 *) Make it possible to have multiple active certificates with the same
3478 subject in the CA index file. This is done only if the keyword
3479 'unique_subject' is set to 'no' in the main CA section (default
3480 if 'CA_default') of the configuration file. The value is saved
3481 with the database itself in a separate index attribute file,
3482 named like the index file with '.attr' appended to the name.
3483 [Richard Levitte]
3484
3485 *) X509 verify fixes. Disable broken certificate workarounds when
3486 X509_V_FLAGS_X509_STRICT is set. Check CRL issuer has cRLSign set if
3487 keyUsage extension present. Don't accept CRLs with unhandled critical
3488 extensions: since verify currently doesn't process CRL extensions this
3489 rejects a CRL with *any* critical extensions. Add new verify error codes
3490 for these cases.
3491 [Steve Henson]
3492
3493 *) When creating an OCSP nonce use an OCTET STRING inside the extnValue.
3494 A clarification of RFC2560 will require the use of OCTET STRINGs and
3495 some implementations cannot handle the current raw format. Since OpenSSL
3496 copies and compares OCSP nonces as opaque blobs without any attempt at
3497 parsing them this should not create any compatibility issues.
3498 [Steve Henson]
3499
3500 *) New md flag EVP_MD_CTX_FLAG_REUSE this allows md_data to be reused when
3501 calling EVP_MD_CTX_copy_ex() to avoid calling OPENSSL_malloc(). Without
3502 this HMAC (and other) operations are several times slower than OpenSSL
3503 < 0.9.7.
3504 [Steve Henson]
3505
3506 *) Print out GeneralizedTime and UTCTime in ASN1_STRING_print_ex().
3507 [Peter Sylvester <Peter.Sylvester@EdelWeb.fr>]
3508
3509 *) Use the correct content when signing type "other".
3510 [Steve Henson]
3511
3512 Changes between 0.9.7b and 0.9.7c [30 Sep 2003]
3513
3514 *) Fix various bugs revealed by running the NISCC test suite:
3515
3516 Stop out of bounds reads in the ASN1 code when presented with
3517 invalid tags (CVE-2003-0543 and CVE-2003-0544).
3518
3519 Free up ASN1_TYPE correctly if ANY type is invalid (CVE-2003-0545).
3520
3521 If verify callback ignores invalid public key errors don't try to check
3522 certificate signature with the NULL public key.
3523
3524 [Steve Henson]
3525
3526 *) New -ignore_err option in ocsp application to stop the server
3527 exiting on the first error in a request.
3528 [Steve Henson]
3529
3530 *) In ssl3_accept() (ssl/s3_srvr.c) only accept a client certificate
3531 if the server requested one: as stated in TLS 1.0 and SSL 3.0
3532 specifications.
3533 [Steve Henson]
3534
3535 *) In ssl3_get_client_hello() (ssl/s3_srvr.c), tolerate additional
3536 extra data after the compression methods not only for TLS 1.0
3537 but also for SSL 3.0 (as required by the specification).
3538 [Bodo Moeller; problem pointed out by Matthias Loepfe]
3539
3540 *) Change X509_certificate_type() to mark the key as exported/exportable
3541 when it's 512 *bits* long, not 512 bytes.
3542 [Richard Levitte]
3543
3544 *) Change AES_cbc_encrypt() so it outputs exact multiple of
3545 blocks during encryption.
3546 [Richard Levitte]
3547
3548 *) Various fixes to base64 BIO and non blocking I/O. On write
3549 flushes were not handled properly if the BIO retried. On read
3550 data was not being buffered properly and had various logic bugs.
3551 This also affects blocking I/O when the data being decoded is a
3552 certain size.
3553 [Steve Henson]
3554
3555 *) Various S/MIME bugfixes and compatibility changes:
3556 output correct application/pkcs7 MIME type if
3557 PKCS7_NOOLDMIMETYPE is set. Tolerate some broken signatures.
3558 Output CR+LF for EOL if PKCS7_CRLFEOL is set (this makes opening
3559 of files as .eml work). Correctly handle very long lines in MIME
3560 parser.
3561 [Steve Henson]
3562
3563 Changes between 0.9.7a and 0.9.7b [10 Apr 2003]
3564
3565 *) Countermeasure against the Klima-Pokorny-Rosa extension of
3566 Bleichbacher's attack on PKCS #1 v1.5 padding: treat
3567 a protocol version number mismatch like a decryption error
3568 in ssl3_get_client_key_exchange (ssl/s3_srvr.c).
3569 [Bodo Moeller]
3570
3571 *) Turn on RSA blinding by default in the default implementation
3572 to avoid a timing attack. Applications that don't want it can call
3573 RSA_blinding_off() or use the new flag RSA_FLAG_NO_BLINDING.
3574 They would be ill-advised to do so in most cases.
3575 [Ben Laurie, Steve Henson, Geoff Thorpe, Bodo Moeller]
3576
3577 *) Change RSA blinding code so that it works when the PRNG is not
3578 seeded (in this case, the secret RSA exponent is abused as
3579 an unpredictable seed -- if it is not unpredictable, there
3580 is no point in blinding anyway). Make RSA blinding thread-safe
3581 by remembering the creator's thread ID in rsa->blinding and
3582 having all other threads use local one-time blinding factors
3583 (this requires more computation than sharing rsa->blinding, but
3584 avoids excessive locking; and if an RSA object is not shared
3585 between threads, blinding will still be very fast).
3586 [Bodo Moeller]
3587
3588 *) Fixed a typo bug that would cause ENGINE_set_default() to set an
3589 ENGINE as defaults for all supported algorithms irrespective of
3590 the 'flags' parameter. 'flags' is now honoured, so applications
3591 should make sure they are passing it correctly.
3592 [Geoff Thorpe]
3593
3594 *) Target "mingw" now allows native Windows code to be generated in
3595 the Cygwin environment as well as with the MinGW compiler.
3596 [Ulf Moeller]
3597
3598 Changes between 0.9.7 and 0.9.7a [19 Feb 2003]
3599
3600 *) In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked
3601 via timing by performing a MAC computation even if incorrrect
3602 block cipher padding has been found. This is a countermeasure
3603 against active attacks where the attacker has to distinguish
3604 between bad padding and a MAC verification error. (CVE-2003-0078)
3605
3606 [Bodo Moeller; problem pointed out by Brice Canvel (EPFL),
3607 Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and
3608 Martin Vuagnoux (EPFL, Ilion)]
3609
3610 *) Make the no-err option work as intended. The intention with no-err
3611 is not to have the whole error stack handling routines removed from
3612 libcrypto, it's only intended to remove all the function name and
3613 reason texts, thereby removing some of the footprint that may not
3614 be interesting if those errors aren't displayed anyway.
3615
3616 NOTE: it's still possible for any application or module to have it's
3617 own set of error texts inserted. The routines are there, just not
3618 used by default when no-err is given.
3619 [Richard Levitte]
3620
3621 *) Add support for FreeBSD on IA64.
3622 [dirk.meyer@dinoex.sub.org via Richard Levitte, resolves #454]
3623
3624 *) Adjust DES_cbc_cksum() so it returns the same value as the MIT
3625 Kerberos function mit_des_cbc_cksum(). Before this change,
3626 the value returned by DES_cbc_cksum() was like the one from
3627 mit_des_cbc_cksum(), except the bytes were swapped.
3628 [Kevin Greaney <Kevin.Greaney@hp.com> and Richard Levitte]
3629
3630 *) Allow an application to disable the automatic SSL chain building.
3631 Before this a rather primitive chain build was always performed in
3632 ssl3_output_cert_chain(): an application had no way to send the
3633 correct chain if the automatic operation produced an incorrect result.
3634
3635 Now the chain builder is disabled if either:
3636
3637 1. Extra certificates are added via SSL_CTX_add_extra_chain_cert().
3638
3639 2. The mode flag SSL_MODE_NO_AUTO_CHAIN is set.
3640
3641 The reasoning behind this is that an application would not want the
3642 auto chain building to take place if extra chain certificates are
3643 present and it might also want a means of sending no additional
3644 certificates (for example the chain has two certificates and the
3645 root is omitted).
3646 [Steve Henson]
3647
3648 *) Add the possibility to build without the ENGINE framework.
3649 [Steven Reddie <smr@essemer.com.au> via Richard Levitte]
3650
3651 *) Under Win32 gmtime() can return NULL: check return value in
3652 OPENSSL_gmtime(). Add error code for case where gmtime() fails.
3653 [Steve Henson]
3654
3655 *) DSA routines: under certain error conditions uninitialized BN objects
3656 could be freed. Solution: make sure initialization is performed early
3657 enough. (Reported and fix supplied by Ivan D Nestlerode <nestler@MIT.EDU>,
3658 Nils Larsch <nla@trustcenter.de> via PR#459)
3659 [Lutz Jaenicke]
3660
3661 *) Another fix for SSLv2 session ID handling: the session ID was incorrectly
3662 checked on reconnect on the client side, therefore session resumption
3663 could still fail with a "ssl session id is different" error. This
3664 behaviour is masked when SSL_OP_ALL is used due to
3665 SSL_OP_MICROSOFT_SESS_ID_BUG being set.
3666 Behaviour observed by Crispin Flowerday <crispin@flowerday.cx> as
3667 followup to PR #377.
3668 [Lutz Jaenicke]
3669
3670 *) IA-32 assembler support enhancements: unified ELF targets, support
3671 for SCO/Caldera platforms, fix for Cygwin shared build.
3672 [Andy Polyakov]
3673
3674 *) Add support for FreeBSD on sparc64. As a consequence, support for
3675 FreeBSD on non-x86 processors is separate from x86 processors on
3676 the config script, much like the NetBSD support.
3677 [Richard Levitte & Kris Kennaway <kris@obsecurity.org>]
3678
3679 Changes between 0.9.6h and 0.9.7 [31 Dec 2002]
3680
3681 [NB: OpenSSL 0.9.6i and later 0.9.6 patch levels were released after
3682 OpenSSL 0.9.7.]
3683
3684 *) Fix session ID handling in SSLv2 client code: the SERVER FINISHED
3685 code (06) was taken as the first octet of the session ID and the last
3686 octet was ignored consequently. As a result SSLv2 client side session
3687 caching could not have worked due to the session ID mismatch between
3688 client and server.
3689 Behaviour observed by Crispin Flowerday <crispin@flowerday.cx> as
3690 PR #377.
3691 [Lutz Jaenicke]
3692
3693 *) Change the declaration of needed Kerberos libraries to use EX_LIBS
3694 instead of the special (and badly supported) LIBKRB5. LIBKRB5 is
3695 removed entirely.
3696 [Richard Levitte]
3697
3698 *) The hw_ncipher.c engine requires dynamic locks. Unfortunately, it
3699 seems that in spite of existing for more than a year, many application
3700 author have done nothing to provide the necessary callbacks, which
3701 means that this particular engine will not work properly anywhere.
3702 This is a very unfortunate situation which forces us, in the name
3703 of usability, to give the hw_ncipher.c a static lock, which is part
3704 of libcrypto.
3705 NOTE: This is for the 0.9.7 series ONLY. This hack will never
3706 appear in 0.9.8 or later. We EXPECT application authors to have
3707 dealt properly with this when 0.9.8 is released (unless we actually
3708 make such changes in the libcrypto locking code that changes will
3709 have to be made anyway).
3710 [Richard Levitte]
3711
3712 *) In asn1_d2i_read_bio() repeatedly call BIO_read() until all content
3713 octets have been read, EOF or an error occurs. Without this change
3714 some truncated ASN1 structures will not produce an error.
3715 [Steve Henson]
3716
3717 *) Disable Heimdal support, since it hasn't been fully implemented.
3718 Still give the possibility to force the use of Heimdal, but with
3719 warnings and a request that patches get sent to openssl-dev.
3720 [Richard Levitte]
3721
3722 *) Add the VC-CE target, introduce the WINCE sysname, and add
3723 INSTALL.WCE and appropriate conditionals to make it build.
3724 [Steven Reddie <smr@essemer.com.au> via Richard Levitte]
3725
3726 *) Change the DLL names for Cygwin to cygcrypto-x.y.z.dll and
3727 cygssl-x.y.z.dll, where x, y and z are the major, minor and
3728 edit numbers of the version.
3729 [Corinna Vinschen <vinschen@redhat.com> and Richard Levitte]
3730
3731 *) Introduce safe string copy and catenation functions
3732 (BUF_strlcpy() and BUF_strlcat()).
3733 [Ben Laurie (CHATS) and Richard Levitte]
3734
3735 *) Avoid using fixed-size buffers for one-line DNs.
3736 [Ben Laurie (CHATS)]
3737
3738 *) Add BUF_MEM_grow_clean() to avoid information leakage when
3739 resizing buffers containing secrets, and use where appropriate.
3740 [Ben Laurie (CHATS)]
3741
3742 *) Avoid using fixed size buffers for configuration file location.
3743 [Ben Laurie (CHATS)]
3744
3745 *) Avoid filename truncation for various CA files.
3746 [Ben Laurie (CHATS)]
3747
3748 *) Use sizeof in preference to magic numbers.
3749 [Ben Laurie (CHATS)]
3750
3751 *) Avoid filename truncation in cert requests.
3752 [Ben Laurie (CHATS)]
3753
3754 *) Add assertions to check for (supposedly impossible) buffer
3755 overflows.
3756 [Ben Laurie (CHATS)]
3757
3758 *) Don't cache truncated DNS entries in the local cache (this could
3759 potentially lead to a spoofing attack).
3760 [Ben Laurie (CHATS)]
3761
3762 *) Fix various buffers to be large enough for hex/decimal
3763 representations in a platform independent manner.
3764 [Ben Laurie (CHATS)]
3765
3766 *) Add CRYPTO_realloc_clean() to avoid information leakage when
3767 resizing buffers containing secrets, and use where appropriate.
3768 [Ben Laurie (CHATS)]
3769
3770 *) Add BIO_indent() to avoid much slightly worrying code to do
3771 indents.
3772 [Ben Laurie (CHATS)]
3773
3774 *) Convert sprintf()/BIO_puts() to BIO_printf().
3775 [Ben Laurie (CHATS)]
3776
3777 *) buffer_gets() could terminate with the buffer only half
3778 full. Fixed.
3779 [Ben Laurie (CHATS)]
3780
3781 *) Add assertions to prevent user-supplied crypto functions from
3782 overflowing internal buffers by having large block sizes, etc.
3783 [Ben Laurie (CHATS)]
3784
3785 *) New OPENSSL_assert() macro (similar to assert(), but enabled
3786 unconditionally).
3787 [Ben Laurie (CHATS)]
3788
3789 *) Eliminate unused copy of key in RC4.
3790 [Ben Laurie (CHATS)]
3791
3792 *) Eliminate unused and incorrectly sized buffers for IV in pem.h.
3793 [Ben Laurie (CHATS)]
3794
3795 *) Fix off-by-one error in EGD path.
3796 [Ben Laurie (CHATS)]
3797
3798 *) If RANDFILE path is too long, ignore instead of truncating.
3799 [Ben Laurie (CHATS)]
3800
3801 *) Eliminate unused and incorrectly sized X.509 structure
3802 CBCParameter.
3803 [Ben Laurie (CHATS)]
3804
3805 *) Eliminate unused and dangerous function knumber().
3806 [Ben Laurie (CHATS)]
3807
3808 *) Eliminate unused and dangerous structure, KSSL_ERR.
3809 [Ben Laurie (CHATS)]
3810
3811 *) Protect against overlong session ID context length in an encoded
3812 session object. Since these are local, this does not appear to be
3813 exploitable.
3814 [Ben Laurie (CHATS)]
3815
3816 *) Change from security patch (see 0.9.6e below) that did not affect
3817 the 0.9.6 release series:
3818
3819 Remote buffer overflow in SSL3 protocol - an attacker could
3820 supply an oversized master key in Kerberos-enabled versions.
3821 (CVE-2002-0657)
3822 [Ben Laurie (CHATS)]
3823
3824 *) Change the SSL kerb5 codes to match RFC 2712.
3825 [Richard Levitte]
3826
3827 *) Make -nameopt work fully for req and add -reqopt switch.
3828 [Michael Bell <michael.bell@rz.hu-berlin.de>, Steve Henson]
3829
3830 *) The "block size" for block ciphers in CFB and OFB mode should be 1.
3831 [Steve Henson, reported by Yngve Nysaeter Pettersen <yngve@opera.com>]
3832
3833 *) Make sure tests can be performed even if the corresponding algorithms
3834 have been removed entirely. This was also the last step to make
3835 OpenSSL compilable with DJGPP under all reasonable conditions.
3836 [Richard Levitte, Doug Kaufman <dkaufman@rahul.net>]
3837
3838 *) Add cipher selection rules COMPLEMENTOFALL and COMPLEMENTOFDEFAULT
3839 to allow version independent disabling of normally unselected ciphers,
3840 which may be activated as a side-effect of selecting a single cipher.
3841
3842 (E.g., cipher list string "RSA" enables ciphersuites that are left
3843 out of "ALL" because they do not provide symmetric encryption.
3844 "RSA:!COMPLEMEMENTOFALL" avoids these unsafe ciphersuites.)
3845 [Lutz Jaenicke, Bodo Moeller]
3846
3847 *) Add appropriate support for separate platform-dependent build
3848 directories. The recommended way to make a platform-dependent
3849 build directory is the following (tested on Linux), maybe with
3850 some local tweaks:
3851
3852 # Place yourself outside of the OpenSSL source tree. In
3853 # this example, the environment variable OPENSSL_SOURCE
3854 # is assumed to contain the absolute OpenSSL source directory.
3855 mkdir -p objtree/"`uname -s`-`uname -r`-`uname -m`"
3856 cd objtree/"`uname -s`-`uname -r`-`uname -m`"
3857 (cd $OPENSSL_SOURCE; find . -type f) | while read F; do
3858 mkdir -p `dirname $F`
3859 ln -s $OPENSSL_SOURCE/$F $F
3860 done
3861
3862 To be absolutely sure not to disturb the source tree, a "make clean"
3863 is a good thing. If it isn't successfull, don't worry about it,
3864 it probably means the source directory is very clean.
3865 [Richard Levitte]
3866
3867 *) Make sure any ENGINE control commands make local copies of string
3868 pointers passed to them whenever necessary. Otherwise it is possible
3869 the caller may have overwritten (or deallocated) the original string
3870 data when a later ENGINE operation tries to use the stored values.
3871 [Götz Babin-Ebell <babinebell@trustcenter.de>]
3872
3873 *) Improve diagnostics in file reading and command-line digests.
3874 [Ben Laurie aided and abetted by Solar Designer <solar@openwall.com>]
3875
3876 *) Add AES modes CFB and OFB to the object database. Correct an
3877 error in AES-CFB decryption.
3878 [Richard Levitte]
3879
3880 *) Remove most calls to EVP_CIPHER_CTX_cleanup() in evp_enc.c, this
3881 allows existing EVP_CIPHER_CTX structures to be reused after
3882 calling EVP_*Final(). This behaviour is used by encryption
3883 BIOs and some applications. This has the side effect that
3884 applications must explicitly clean up cipher contexts with
3885 EVP_CIPHER_CTX_cleanup() or they will leak memory.
3886 [Steve Henson]
3887
3888 *) Check the values of dna and dnb in bn_mul_recursive before calling
3889 bn_mul_comba (a non zero value means the a or b arrays do not contain
3890 n2 elements) and fallback to bn_mul_normal if either is not zero.
3891 [Steve Henson]
3892
3893 *) Fix escaping of non-ASCII characters when using the -subj option
3894 of the "openssl req" command line tool. (Robert Joop <joop@fokus.gmd.de>)
3895 [Lutz Jaenicke]
3896
3897 *) Make object definitions compliant to LDAP (RFC2256): SN is the short
3898 form for "surname", serialNumber has no short form.
3899 Use "mail" as the short name for "rfc822Mailbox" according to RFC2798;
3900 therefore remove "mail" short name for "internet 7".
3901 The OID for unique identifiers in X509 certificates is
3902 x500UniqueIdentifier, not uniqueIdentifier.
3903 Some more OID additions. (Michael Bell <michael.bell@rz.hu-berlin.de>)
3904 [Lutz Jaenicke]
3905
3906 *) Add an "init" command to the ENGINE config module and auto initialize
3907 ENGINEs. Without any "init" command the ENGINE will be initialized
3908 after all ctrl commands have been executed on it. If init=1 the
3909 ENGINE is initailized at that point (ctrls before that point are run
3910 on the uninitialized ENGINE and after on the initialized one). If
3911 init=0 then the ENGINE will not be iniatialized at all.
3912 [Steve Henson]
3913
3914 *) Fix the 'app_verify_callback' interface so that the user-defined
3915 argument is actually passed to the callback: In the
3916 SSL_CTX_set_cert_verify_callback() prototype, the callback
3917 declaration has been changed from
3918 int (*cb)()
3919 into
3920 int (*cb)(X509_STORE_CTX *,void *);
3921 in ssl_verify_cert_chain (ssl/ssl_cert.c), the call
3922 i=s->ctx->app_verify_callback(&ctx)
3923 has been changed into
3924 i=s->ctx->app_verify_callback(&ctx, s->ctx->app_verify_arg).
3925
3926 To update applications using SSL_CTX_set_cert_verify_callback(),
3927 a dummy argument can be added to their callback functions.
3928 [D. K. Smetters <smetters@parc.xerox.com>]
3929
3930 *) Added the '4758cca' ENGINE to support IBM 4758 cards.
3931 [Maurice Gittens <maurice@gittens.nl>, touchups by Geoff Thorpe]
3932
3933 *) Add and OPENSSL_LOAD_CONF define which will cause
3934 OpenSSL_add_all_algorithms() to load the openssl.cnf config file.
3935 This allows older applications to transparently support certain
3936 OpenSSL features: such as crypto acceleration and dynamic ENGINE loading.
3937 Two new functions OPENSSL_add_all_algorithms_noconf() which will never
3938 load the config file and OPENSSL_add_all_algorithms_conf() which will
3939 always load it have also been added.
3940 [Steve Henson]
3941
3942 *) Add the OFB, CFB and CTR (all with 128 bit feedback) to AES.
3943 Adjust NIDs and EVP layer.
3944 [Stephen Sprunk <stephen@sprunk.org> and Richard Levitte]
3945
3946 *) Config modules support in openssl utility.
3947
3948 Most commands now load modules from the config file,
3949 though in a few (such as version) this isn't done
3950 because it couldn't be used for anything.
3951
3952 In the case of ca and req the config file used is
3953 the same as the utility itself: that is the -config
3954 command line option can be used to specify an
3955 alternative file.
3956 [Steve Henson]
3957
3958 *) Move default behaviour from OPENSSL_config(). If appname is NULL
3959 use "openssl_conf" if filename is NULL use default openssl config file.
3960 [Steve Henson]
3961
3962 *) Add an argument to OPENSSL_config() to allow the use of an alternative
3963 config section name. Add a new flag to tolerate a missing config file
3964 and move code to CONF_modules_load_file().
3965 [Steve Henson]
3966
3967 *) Support for crypto accelerator cards from Accelerated Encryption
3968 Processing, www.aep.ie. (Use engine 'aep')
3969 The support was copied from 0.9.6c [engine] and adapted/corrected
3970 to work with the new engine framework.
3971 [AEP Inc. and Richard Levitte]
3972
3973 *) Support for SureWare crypto accelerator cards from Baltimore
3974 Technologies. (Use engine 'sureware')
3975 The support was copied from 0.9.6c [engine] and adapted
3976 to work with the new engine framework.
3977 [Richard Levitte]
3978
3979 *) Have the CHIL engine fork-safe (as defined by nCipher) and actually
3980 make the newer ENGINE framework commands for the CHIL engine work.
3981 [Toomas Kiisk <vix@cyber.ee> and Richard Levitte]
3982
3983 *) Make it possible to produce shared libraries on ReliantUNIX.
3984 [Robert Dahlem <Robert.Dahlem@ffm2.siemens.de> via Richard Levitte]
3985
3986 *) Add the configuration target debug-linux-ppro.
3987 Make 'openssl rsa' use the general key loading routines
3988 implemented in apps.c, and make those routines able to
3989 handle the key format FORMAT_NETSCAPE and the variant
3990 FORMAT_IISSGC.
3991 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte]
3992
3993 *) Fix a crashbug and a logic bug in hwcrhk_load_pubkey().
3994 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte]
3995
3996 *) Add -keyform to rsautl, and document -engine.
3997 [Richard Levitte, inspired by Toomas Kiisk <vix@cyber.ee>]
3998
3999 *) Change BIO_new_file (crypto/bio/bss_file.c) to use new
4000 BIO_R_NO_SUCH_FILE error code rather than the generic
4001 ERR_R_SYS_LIB error code if fopen() fails with ENOENT.
4002 [Ben Laurie]
4003
4004 *) Add new functions
4005 ERR_peek_last_error
4006 ERR_peek_last_error_line
4007 ERR_peek_last_error_line_data.
4008 These are similar to
4009 ERR_peek_error
4010 ERR_peek_error_line
4011 ERR_peek_error_line_data,
4012 but report on the latest error recorded rather than the first one
4013 still in the error queue.
4014 [Ben Laurie, Bodo Moeller]
4015
4016 *) default_algorithms option in ENGINE config module. This allows things
4017 like:
4018 default_algorithms = ALL
4019 default_algorithms = RSA, DSA, RAND, CIPHERS, DIGESTS
4020 [Steve Henson]
4021
4022 *) Prelminary ENGINE config module.
4023 [Steve Henson]
4024
4025 *) New experimental application configuration code.
4026 [Steve Henson]
4027
4028 *) Change the AES code to follow the same name structure as all other
4029 symmetric ciphers, and behave the same way. Move everything to
4030 the directory crypto/aes, thereby obsoleting crypto/rijndael.
4031 [Stephen Sprunk <stephen@sprunk.org> and Richard Levitte]
4032
4033 *) SECURITY: remove unsafe setjmp/signal interaction from ui_openssl.c.
4034 [Ben Laurie and Theo de Raadt]
4035
4036 *) Add option to output public keys in req command.
4037 [Massimiliano Pala madwolf@openca.org]
4038
4039 *) Use wNAFs in EC_POINTs_mul() for improved efficiency
4040 (up to about 10% better than before for P-192 and P-224).
4041 [Bodo Moeller]
4042
4043 *) New functions/macros
4044
4045 SSL_CTX_set_msg_callback(ctx, cb)
4046 SSL_CTX_set_msg_callback_arg(ctx, arg)
4047 SSL_set_msg_callback(ssl, cb)
4048 SSL_set_msg_callback_arg(ssl, arg)
4049
4050 to request calling a callback function
4051
4052 void cb(int write_p, int version, int content_type,
4053 const void *buf, size_t len, SSL *ssl, void *arg)
4054
4055 whenever a protocol message has been completely received
4056 (write_p == 0) or sent (write_p == 1). Here 'version' is the
4057 protocol version according to which the SSL library interprets
4058 the current protocol message (SSL2_VERSION, SSL3_VERSION, or
4059 TLS1_VERSION). 'content_type' is 0 in the case of SSL 2.0, or
4060 the content type as defined in the SSL 3.0/TLS 1.0 protocol
4061 specification (change_cipher_spec(20), alert(21), handshake(22)).
4062 'buf' and 'len' point to the actual message, 'ssl' to the
4063 SSL object, and 'arg' is the application-defined value set by
4064 SSL[_CTX]_set_msg_callback_arg().
4065
4066 'openssl s_client' and 'openssl s_server' have new '-msg' options
4067 to enable a callback that displays all protocol messages.
4068 [Bodo Moeller]
4069
4070 *) Change the shared library support so shared libraries are built as
4071 soon as the corresponding static library is finished, and thereby get
4072 openssl and the test programs linked against the shared library.
4073 This still only happens when the keyword "shard" has been given to
4074 the configuration scripts.
4075
4076 NOTE: shared library support is still an experimental thing, and
4077 backward binary compatibility is still not guaranteed.
4078 ["Maciej W. Rozycki" <macro@ds2.pg.gda.pl> and Richard Levitte]
4079
4080 *) Add support for Subject Information Access extension.
4081 [Peter Sylvester <Peter.Sylvester@EdelWeb.fr>]
4082
4083 *) Make BUF_MEM_grow() behaviour more consistent: Initialise to zero
4084 additional bytes when new memory had to be allocated, not just
4085 when reusing an existing buffer.
4086 [Bodo Moeller]
4087
4088 *) New command line and configuration option 'utf8' for the req command.
4089 This allows field values to be specified as UTF8 strings.
4090 [Steve Henson]
4091
4092 *) Add -multi and -mr options to "openssl speed" - giving multiple parallel
4093 runs for the former and machine-readable output for the latter.
4094 [Ben Laurie]
4095
4096 *) Add '-noemailDN' option to 'openssl ca'. This prevents inclusion
4097 of the e-mail address in the DN (i.e., it will go into a certificate
4098 extension only). The new configuration file option 'email_in_dn = no'
4099 has the same effect.
4100 [Massimiliano Pala madwolf@openca.org]
4101
4102 *) Change all functions with names starting with des_ to be starting
4103 with DES_ instead. Add wrappers that are compatible with libdes,
4104 but are named _ossl_old_des_*. Finally, add macros that map the
4105 des_* symbols to the corresponding _ossl_old_des_* if libdes
4106 compatibility is desired. If OpenSSL 0.9.6c compatibility is
4107 desired, the des_* symbols will be mapped to DES_*, with one
4108 exception.
4109
4110 Since we provide two compatibility mappings, the user needs to
4111 define the macro OPENSSL_DES_LIBDES_COMPATIBILITY if libdes
4112 compatibility is desired. The default (i.e., when that macro
4113 isn't defined) is OpenSSL 0.9.6c compatibility.
4114
4115 There are also macros that enable and disable the support of old
4116 des functions altogether. Those are OPENSSL_ENABLE_OLD_DES_SUPPORT
4117 and OPENSSL_DISABLE_OLD_DES_SUPPORT. If none or both of those
4118 are defined, the default will apply: to support the old des routines.
4119
4120 In either case, one must include openssl/des.h to get the correct
4121 definitions. Do not try to just include openssl/des_old.h, that
4122 won't work.
4123
4124 NOTE: This is a major break of an old API into a new one. Software
4125 authors are encouraged to switch to the DES_ style functions. Some
4126 time in the future, des_old.h and the libdes compatibility functions
4127 will be disable (i.e. OPENSSL_DISABLE_OLD_DES_SUPPORT will be the
4128 default), and then completely removed.
4129 [Richard Levitte]
4130
4131 *) Test for certificates which contain unsupported critical extensions.
4132 If such a certificate is found during a verify operation it is
4133 rejected by default: this behaviour can be overridden by either
4134 handling the new error X509_V_ERR_UNHANDLED_CRITICAL_EXTENSION or
4135 by setting the verify flag X509_V_FLAG_IGNORE_CRITICAL. A new function
4136 X509_supported_extension() has also been added which returns 1 if a
4137 particular extension is supported.
4138 [Steve Henson]
4139
4140 *) Modify the behaviour of EVP cipher functions in similar way to digests
4141 to retain compatibility with existing code.
4142 [Steve Henson]
4143
4144 *) Modify the behaviour of EVP_DigestInit() and EVP_DigestFinal() to retain
4145 compatibility with existing code. In particular the 'ctx' parameter does
4146 not have to be to be initialized before the call to EVP_DigestInit() and
4147 it is tidied up after a call to EVP_DigestFinal(). New function
4148 EVP_DigestFinal_ex() which does not tidy up the ctx. Similarly function
4149 EVP_MD_CTX_copy() changed to not require the destination to be
4150 initialized valid and new function EVP_MD_CTX_copy_ex() added which
4151 requires the destination to be valid.
4152
4153 Modify all the OpenSSL digest calls to use EVP_DigestInit_ex(),
4154 EVP_DigestFinal_ex() and EVP_MD_CTX_copy_ex().
4155 [Steve Henson]
4156
4157 *) Change ssl3_get_message (ssl/s3_both.c) and the functions using it
4158 so that complete 'Handshake' protocol structures are kept in memory
4159 instead of overwriting 'msg_type' and 'length' with 'body' data.
4160 [Bodo Moeller]
4161
4162 *) Add an implementation of SSL_add_dir_cert_subjects_to_stack for Win32.
4163 [Massimo Santin via Richard Levitte]
4164
4165 *) Major restructuring to the underlying ENGINE code. This includes
4166 reduction of linker bloat, separation of pure "ENGINE" manipulation
4167 (initialisation, etc) from functionality dealing with implementations
4168 of specific crypto iterfaces. This change also introduces integrated
4169 support for symmetric ciphers and digest implementations - so ENGINEs
4170 can now accelerate these by providing EVP_CIPHER and EVP_MD
4171 implementations of their own. This is detailed in crypto/engine/README
4172 as it couldn't be adequately described here. However, there are a few
4173 API changes worth noting - some RSA, DSA, DH, and RAND functions that
4174 were changed in the original introduction of ENGINE code have now
4175 reverted back - the hooking from this code to ENGINE is now a good
4176 deal more passive and at run-time, operations deal directly with
4177 RSA_METHODs, DSA_METHODs (etc) as they did before, rather than
4178 dereferencing through an ENGINE pointer any more. Also, the ENGINE
4179 functions dealing with BN_MOD_EXP[_CRT] handlers have been removed -
4180 they were not being used by the framework as there is no concept of a
4181 BIGNUM_METHOD and they could not be generalised to the new
4182 'ENGINE_TABLE' mechanism that underlies the new code. Similarly,
4183 ENGINE_cpy() has been removed as it cannot be consistently defined in
4184 the new code.
4185 [Geoff Thorpe]
4186
4187 *) Change ASN1_GENERALIZEDTIME_check() to allow fractional seconds.
4188 [Steve Henson]
4189
4190 *) Change mkdef.pl to sort symbols that get the same entry number,
4191 and make sure the automatically generated functions ERR_load_*
4192 become part of libeay.num as well.
4193 [Richard Levitte]
4194
4195 *) New function SSL_renegotiate_pending(). This returns true once
4196 renegotiation has been requested (either SSL_renegotiate() call
4197 or HelloRequest/ClientHello receveived from the peer) and becomes
4198 false once a handshake has been completed.
4199 (For servers, SSL_renegotiate() followed by SSL_do_handshake()
4200 sends a HelloRequest, but does not ensure that a handshake takes
4201 place. SSL_renegotiate_pending() is useful for checking if the
4202 client has followed the request.)
4203 [Bodo Moeller]
4204
4205 *) New SSL option SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION.
4206 By default, clients may request session resumption even during
4207 renegotiation (if session ID contexts permit); with this option,
4208 session resumption is possible only in the first handshake.
4209
4210 SSL_OP_ALL is now 0x00000FFFL instead of 0x000FFFFFL. This makes
4211 more bits available for options that should not be part of
4212 SSL_OP_ALL (such as SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION).
4213 [Bodo Moeller]
4214
4215 *) Add some demos for certificate and certificate request creation.
4216 [Steve Henson]
4217
4218 *) Make maximum certificate chain size accepted from the peer application
4219 settable (SSL*_get/set_max_cert_list()), as proposed by
4220 "Douglas E. Engert" <deengert@anl.gov>.
4221 [Lutz Jaenicke]
4222
4223 *) Add support for shared libraries for Unixware-7
4224 (Boyd Lynn Gerber <gerberb@zenez.com>).
4225 [Lutz Jaenicke]
4226
4227 *) Add a "destroy" handler to ENGINEs that allows structural cleanup to
4228 be done prior to destruction. Use this to unload error strings from
4229 ENGINEs that load their own error strings. NB: This adds two new API
4230 functions to "get" and "set" this destroy handler in an ENGINE.
4231 [Geoff Thorpe]
4232
4233 *) Alter all existing ENGINE implementations (except "openssl" and
4234 "openbsd") to dynamically instantiate their own error strings. This
4235 makes them more flexible to be built both as statically-linked ENGINEs
4236 and self-contained shared-libraries loadable via the "dynamic" ENGINE.
4237 Also, add stub code to each that makes building them as self-contained
4238 shared-libraries easier (see README.ENGINE).
4239 [Geoff Thorpe]
4240
4241 *) Add a "dynamic" ENGINE that provides a mechanism for binding ENGINE
4242 implementations into applications that are completely implemented in
4243 self-contained shared-libraries. The "dynamic" ENGINE exposes control
4244 commands that can be used to configure what shared-library to load and
4245 to control aspects of the way it is handled. Also, made an update to
4246 the README.ENGINE file that brings its information up-to-date and
4247 provides some information and instructions on the "dynamic" ENGINE
4248 (ie. how to use it, how to build "dynamic"-loadable ENGINEs, etc).
4249 [Geoff Thorpe]
4250
4251 *) Make it possible to unload ranges of ERR strings with a new
4252 "ERR_unload_strings" function.
4253 [Geoff Thorpe]
4254
4255 *) Add a copy() function to EVP_MD.
4256 [Ben Laurie]
4257
4258 *) Make EVP_MD routines take a context pointer instead of just the
4259 md_data void pointer.
4260 [Ben Laurie]
4261
4262 *) Add flags to EVP_MD and EVP_MD_CTX. EVP_MD_FLAG_ONESHOT indicates
4263 that the digest can only process a single chunk of data
4264 (typically because it is provided by a piece of
4265 hardware). EVP_MD_CTX_FLAG_ONESHOT indicates that the application
4266 is only going to provide a single chunk of data, and hence the
4267 framework needn't accumulate the data for oneshot drivers.
4268 [Ben Laurie]
4269
4270 *) As with "ERR", make it possible to replace the underlying "ex_data"
4271 functions. This change also alters the storage and management of global
4272 ex_data state - it's now all inside ex_data.c and all "class" code (eg.
4273 RSA, BIO, SSL_CTX, etc) no longer stores its own STACKS and per-class
4274 index counters. The API functions that use this state have been changed
4275 to take a "class_index" rather than pointers to the class's local STACK
4276 and counter, and there is now an API function to dynamically create new
4277 classes. This centralisation allows us to (a) plug a lot of the
4278 thread-safety problems that existed, and (b) makes it possible to clean
4279 up all allocated state using "CRYPTO_cleanup_all_ex_data()". W.r.t. (b)
4280 such data would previously have always leaked in application code and
4281 workarounds were in place to make the memory debugging turn a blind eye
4282 to it. Application code that doesn't use this new function will still
4283 leak as before, but their memory debugging output will announce it now
4284 rather than letting it slide.
4285
4286 Besides the addition of CRYPTO_cleanup_all_ex_data(), another API change
4287 induced by the "ex_data" overhaul is that X509_STORE_CTX_init() now
4288 has a return value to indicate success or failure.
4289 [Geoff Thorpe]
4290
4291 *) Make it possible to replace the underlying "ERR" functions such that the
4292 global state (2 LHASH tables and 2 locks) is only used by the "default"
4293 implementation. This change also adds two functions to "get" and "set"
4294 the implementation prior to it being automatically set the first time
4295 any other ERR function takes place. Ie. an application can call "get",
4296 pass the return value to a module it has just loaded, and that module
4297 can call its own "set" function using that value. This means the
4298 module's "ERR" operations will use (and modify) the error state in the
4299 application and not in its own statically linked copy of OpenSSL code.
4300 [Geoff Thorpe]
4301
4302 *) Give DH, DSA, and RSA types their own "**_up_ref()" function to increment
4303 reference counts. This performs normal REF_PRINT/REF_CHECK macros on
4304 the operation, and provides a more encapsulated way for external code
4305 (crypto/evp/ and ssl/) to do this. Also changed the evp and ssl code
4306 to use these functions rather than manually incrementing the counts.
4307
4308 Also rename "DSO_up()" function to more descriptive "DSO_up_ref()".
4309 [Geoff Thorpe]
4310
4311 *) Add EVP test program.
4312 [Ben Laurie]
4313
4314 *) Add symmetric cipher support to ENGINE. Expect the API to change!
4315 [Ben Laurie]
4316
4317 *) New CRL functions: X509_CRL_set_version(), X509_CRL_set_issuer_name()
4318 X509_CRL_set_lastUpdate(), X509_CRL_set_nextUpdate(), X509_CRL_sort(),
4319 X509_REVOKED_set_serialNumber(), and X509_REVOKED_set_revocationDate().
4320 These allow a CRL to be built without having to access X509_CRL fields
4321 directly. Modify 'ca' application to use new functions.
4322 [Steve Henson]
4323
4324 *) Move SSL_OP_TLS_ROLLBACK_BUG out of the SSL_OP_ALL list of recommended
4325 bug workarounds. Rollback attack detection is a security feature.
4326 The problem will only arise on OpenSSL servers when TLSv1 is not
4327 available (sslv3_server_method() or SSL_OP_NO_TLSv1).
4328 Software authors not wanting to support TLSv1 will have special reasons
4329 for their choice and can explicitly enable this option.
4330 [Bodo Moeller, Lutz Jaenicke]
4331
4332 *) Rationalise EVP so it can be extended: don't include a union of
4333 cipher/digest structures, add init/cleanup functions for EVP_MD_CTX
4334 (similar to those existing for EVP_CIPHER_CTX).
4335 Usage example:
4336
4337 EVP_MD_CTX md;
4338
4339 EVP_MD_CTX_init(&md); /* new function call */
4340 EVP_DigestInit(&md, EVP_sha1());
4341 EVP_DigestUpdate(&md, in, len);
4342 EVP_DigestFinal(&md, out, NULL);
4343 EVP_MD_CTX_cleanup(&md); /* new function call */
4344
4345 [Ben Laurie]
4346
4347 *) Make DES key schedule conform to the usual scheme, as well as
4348 correcting its structure. This means that calls to DES functions
4349 now have to pass a pointer to a des_key_schedule instead of a
4350 plain des_key_schedule (which was actually always a pointer
4351 anyway): E.g.,
4352
4353 des_key_schedule ks;
4354
4355 des_set_key_checked(..., &ks);
4356 des_ncbc_encrypt(..., &ks, ...);
4357
4358 (Note that a later change renames 'des_...' into 'DES_...'.)
4359 [Ben Laurie]
4360
4361 *) Initial reduction of linker bloat: the use of some functions, such as
4362 PEM causes large amounts of unused functions to be linked in due to
4363 poor organisation. For example pem_all.c contains every PEM function
4364 which has a knock on effect of linking in large amounts of (unused)
4365 ASN1 code. Grouping together similar functions and splitting unrelated
4366 functions prevents this.
4367 [Steve Henson]
4368
4369 *) Cleanup of EVP macros.
4370 [Ben Laurie]
4371
4372 *) Change historical references to {NID,SN,LN}_des_ede and ede3 to add the
4373 correct _ecb suffix.
4374 [Ben Laurie]
4375
4376 *) Add initial OCSP responder support to ocsp application. The
4377 revocation information is handled using the text based index
4378 use by the ca application. The responder can either handle
4379 requests generated internally, supplied in files (for example
4380 via a CGI script) or using an internal minimal server.
4381 [Steve Henson]
4382
4383 *) Add configuration choices to get zlib compression for TLS.
4384 [Richard Levitte]
4385
4386 *) Changes to Kerberos SSL for RFC 2712 compliance:
4387 1. Implemented real KerberosWrapper, instead of just using
4388 KRB5 AP_REQ message. [Thanks to Simon Wilkinson <sxw@sxw.org.uk>]
4389 2. Implemented optional authenticator field of KerberosWrapper.
4390
4391 Added openssl-style ASN.1 macros for Kerberos ticket, ap_req,
4392 and authenticator structs; see crypto/krb5/.
4393
4394 Generalized Kerberos calls to support multiple Kerberos libraries.
4395 [Vern Staats <staatsvr@asc.hpc.mil>,
4396 Jeffrey Altman <jaltman@columbia.edu>
4397 via Richard Levitte]
4398
4399 *) Cause 'openssl speed' to use fully hard-coded DSA keys as it
4400 already does with RSA. testdsa.h now has 'priv_key/pub_key'
4401 values for each of the key sizes rather than having just
4402 parameters (and 'speed' generating keys each time).
4403 [Geoff Thorpe]
4404
4405 *) Speed up EVP routines.
4406 Before:
4407 encrypt
4408 type 8 bytes 64 bytes 256 bytes 1024 bytes 8192 bytes
4409 des-cbc 4408.85k 5560.51k 5778.46k 5862.20k 5825.16k
4410 des-cbc 4389.55k 5571.17k 5792.23k 5846.91k 5832.11k
4411 des-cbc 4394.32k 5575.92k 5807.44k 5848.37k 5841.30k
4412 decrypt
4413 des-cbc 3482.66k 5069.49k 5496.39k 5614.16k 5639.28k
4414 des-cbc 3480.74k 5068.76k 5510.34k 5609.87k 5635.52k
4415 des-cbc 3483.72k 5067.62k 5504.60k 5708.01k 5724.80k
4416 After:
4417 encrypt
4418 des-cbc 4660.16k 5650.19k 5807.19k 5827.13k 5783.32k
4419 decrypt
4420 des-cbc 3624.96k 5258.21k 5530.91k 5624.30k 5628.26k
4421 [Ben Laurie]
4422
4423 *) Added the OS2-EMX target.
4424 ["Brian Havard" <brianh@kheldar.apana.org.au> and Richard Levitte]
4425
4426 *) Rewrite apps to use NCONF routines instead of the old CONF. New functions
4427 to support NCONF routines in extension code. New function CONF_set_nconf()
4428 to allow functions which take an NCONF to also handle the old LHASH
4429 structure: this means that the old CONF compatible routines can be
4430 retained (in particular wrt extensions) without having to duplicate the
4431 code. New function X509V3_add_ext_nconf_sk to add extensions to a stack.
4432 [Steve Henson]
4433
4434 *) Enhance the general user interface with mechanisms for inner control
4435 and with possibilities to have yes/no kind of prompts.
4436 [Richard Levitte]
4437
4438 *) Change all calls to low level digest routines in the library and
4439 applications to use EVP. Add missing calls to HMAC_cleanup() and
4440 don't assume HMAC_CTX can be copied using memcpy().
4441 [Verdon Walker <VWalker@novell.com>, Steve Henson]
4442
4443 *) Add the possibility to control engines through control names but with
4444 arbitrary arguments instead of just a string.
4445 Change the key loaders to take a UI_METHOD instead of a callback
4446 function pointer. NOTE: this breaks binary compatibility with earlier
4447 versions of OpenSSL [engine].
4448 Adapt the nCipher code for these new conditions and add a card insertion
4449 callback.
4450 [Richard Levitte]
4451
4452 *) Enhance the general user interface with mechanisms to better support
4453 dialog box interfaces, application-defined prompts, the possibility
4454 to use defaults (for example default passwords from somewhere else)
4455 and interrupts/cancellations.
4456 [Richard Levitte]
4457
4458 *) Tidy up PKCS#12 attribute handling. Add support for the CSP name
4459 attribute in PKCS#12 files, add new -CSP option to pkcs12 utility.
4460 [Steve Henson]
4461
4462 *) Fix a memory leak in 'sk_dup()' in the case reallocation fails. (Also
4463 tidy up some unnecessarily weird code in 'sk_new()').
4464 [Geoff, reported by Diego Tartara <dtartara@novamens.com>]
4465
4466 *) Change the key loading routines for ENGINEs to use the same kind
4467 callback (pem_password_cb) as all other routines that need this
4468 kind of callback.
4469 [Richard Levitte]
4470
4471 *) Increase ENTROPY_NEEDED to 32 bytes, as Rijndael can operate with
4472 256 bit (=32 byte) keys. Of course seeding with more entropy bytes
4473 than this minimum value is recommended.
4474 [Lutz Jaenicke]
4475
4476 *) New random seeder for OpenVMS, using the system process statistics
4477 that are easily reachable.
4478 [Richard Levitte]
4479
4480 *) Windows apparently can't transparently handle global
4481 variables defined in DLLs. Initialisations such as:
4482
4483 const ASN1_ITEM *it = &ASN1_INTEGER_it;
4484
4485 wont compile. This is used by the any applications that need to
4486 declare their own ASN1 modules. This was fixed by adding the option
4487 EXPORT_VAR_AS_FN to all Win32 platforms, although this isn't strictly
4488 needed for static libraries under Win32.
4489 [Steve Henson]
4490
4491 *) New functions X509_PURPOSE_set() and X509_TRUST_set() to handle
4492 setting of purpose and trust fields. New X509_STORE trust and
4493 purpose functions and tidy up setting in other SSL functions.
4494 [Steve Henson]
4495
4496 *) Add copies of X509_STORE_CTX fields and callbacks to X509_STORE
4497 structure. These are inherited by X509_STORE_CTX when it is
4498 initialised. This allows various defaults to be set in the
4499 X509_STORE structure (such as flags for CRL checking and custom
4500 purpose or trust settings) for functions which only use X509_STORE_CTX
4501 internally such as S/MIME.
4502
4503 Modify X509_STORE_CTX_purpose_inherit() so it only sets purposes and
4504 trust settings if they are not set in X509_STORE. This allows X509_STORE
4505 purposes and trust (in S/MIME for example) to override any set by default.
4506
4507 Add command line options for CRL checking to smime, s_client and s_server
4508 applications.
4509 [Steve Henson]
4510
4511 *) Initial CRL based revocation checking. If the CRL checking flag(s)
4512 are set then the CRL is looked up in the X509_STORE structure and
4513 its validity and signature checked, then if the certificate is found
4514 in the CRL the verify fails with a revoked error.
4515
4516 Various new CRL related callbacks added to X509_STORE_CTX structure.
4517
4518 Command line options added to 'verify' application to support this.
4519
4520 This needs some additional work, such as being able to handle multiple
4521 CRLs with different times, extension based lookup (rather than just
4522 by subject name) and ultimately more complete V2 CRL extension
4523 handling.
4524 [Steve Henson]
4525
4526 *) Add a general user interface API (crypto/ui/). This is designed
4527 to replace things like des_read_password and friends (backward
4528 compatibility functions using this new API are provided).
4529 The purpose is to remove prompting functions from the DES code
4530 section as well as provide for prompting through dialog boxes in
4531 a window system and the like.
4532 [Richard Levitte]
4533
4534 *) Add "ex_data" support to ENGINE so implementations can add state at a
4535 per-structure level rather than having to store it globally.
4536 [Geoff]
4537
4538 *) Make it possible for ENGINE structures to be copied when retrieved by
4539 ENGINE_by_id() if the ENGINE specifies a new flag: ENGINE_FLAGS_BY_ID_COPY.
4540 This causes the "original" ENGINE structure to act like a template,
4541 analogous to the RSA vs. RSA_METHOD type of separation. Because of this
4542 operational state can be localised to each ENGINE structure, despite the
4543 fact they all share the same "methods". New ENGINE structures returned in
4544 this case have no functional references and the return value is the single
4545 structural reference. This matches the single structural reference returned
4546 by ENGINE_by_id() normally, when it is incremented on the pre-existing
4547 ENGINE structure.
4548 [Geoff]
4549
4550 *) Fix ASN1 decoder when decoding type ANY and V_ASN1_OTHER: since this
4551 needs to match any other type at all we need to manually clear the
4552 tag cache.
4553 [Steve Henson]
4554
4555 *) Changes to the "openssl engine" utility to include;
4556 - verbosity levels ('-v', '-vv', and '-vvv') that provide information
4557 about an ENGINE's available control commands.
4558 - executing control commands from command line arguments using the
4559 '-pre' and '-post' switches. '-post' is only used if '-t' is
4560 specified and the ENGINE is successfully initialised. The syntax for
4561 the individual commands are colon-separated, for example;
4562 openssl engine chil -pre FORK_CHECK:0 -pre SO_PATH:/lib/test.so
4563 [Geoff]
4564
4565 *) New dynamic control command support for ENGINEs. ENGINEs can now
4566 declare their own commands (numbers), names (strings), descriptions,
4567 and input types for run-time discovery by calling applications. A
4568 subset of these commands are implicitly classed as "executable"
4569 depending on their input type, and only these can be invoked through
4570 the new string-based API function ENGINE_ctrl_cmd_string(). (Eg. this
4571 can be based on user input, config files, etc). The distinction is
4572 that "executable" commands cannot return anything other than a boolean
4573 result and can only support numeric or string input, whereas some
4574 discoverable commands may only be for direct use through
4575 ENGINE_ctrl(), eg. supporting the exchange of binary data, function
4576 pointers, or other custom uses. The "executable" commands are to
4577 support parameterisations of ENGINE behaviour that can be
4578 unambiguously defined by ENGINEs and used consistently across any
4579 OpenSSL-based application. Commands have been added to all the
4580 existing hardware-supporting ENGINEs, noticeably "SO_PATH" to allow
4581 control over shared-library paths without source code alterations.
4582 [Geoff]
4583
4584 *) Changed all ENGINE implementations to dynamically allocate their
4585 ENGINEs rather than declaring them statically. Apart from this being
4586 necessary with the removal of the ENGINE_FLAGS_MALLOCED distinction,
4587 this also allows the implementations to compile without using the
4588 internal engine_int.h header.
4589 [Geoff]
4590
4591 *) Minor adjustment to "rand" code. RAND_get_rand_method() now returns a
4592 'const' value. Any code that should be able to modify a RAND_METHOD
4593 should already have non-const pointers to it (ie. they should only
4594 modify their own ones).
4595 [Geoff]
4596
4597 *) Made a variety of little tweaks to the ENGINE code.
4598 - "atalla" and "ubsec" string definitions were moved from header files
4599 to C code. "nuron" string definitions were placed in variables
4600 rather than hard-coded - allowing parameterisation of these values
4601 later on via ctrl() commands.
4602 - Removed unused "#if 0"'d code.
4603 - Fixed engine list iteration code so it uses ENGINE_free() to release
4604 structural references.
4605 - Constified the RAND_METHOD element of ENGINE structures.
4606 - Constified various get/set functions as appropriate and added
4607 missing functions (including a catch-all ENGINE_cpy that duplicates
4608 all ENGINE values onto a new ENGINE except reference counts/state).
4609 - Removed NULL parameter checks in get/set functions. Setting a method
4610 or function to NULL is a way of cancelling out a previously set
4611 value. Passing a NULL ENGINE parameter is just plain stupid anyway
4612 and doesn't justify the extra error symbols and code.
4613 - Deprecate the ENGINE_FLAGS_MALLOCED define and move the area for
4614 flags from engine_int.h to engine.h.
4615 - Changed prototypes for ENGINE handler functions (init(), finish(),
4616 ctrl(), key-load functions, etc) to take an (ENGINE*) parameter.
4617 [Geoff]
4618
4619 *) Implement binary inversion algorithm for BN_mod_inverse in addition
4620 to the algorithm using long division. The binary algorithm can be
4621 used only if the modulus is odd. On 32-bit systems, it is faster
4622 only for relatively small moduli (roughly 20-30% for 128-bit moduli,
4623 roughly 5-15% for 256-bit moduli), so we use it only for moduli
4624 up to 450 bits. In 64-bit environments, the binary algorithm
4625 appears to be advantageous for much longer moduli; here we use it
4626 for moduli up to 2048 bits.
4627 [Bodo Moeller]
4628
4629 *) Rewrite CHOICE field setting in ASN1_item_ex_d2i(). The old code
4630 could not support the combine flag in choice fields.
4631 [Steve Henson]
4632
4633 *) Add a 'copy_extensions' option to the 'ca' utility. This copies
4634 extensions from a certificate request to the certificate.
4635 [Steve Henson]
4636
4637 *) Allow multiple 'certopt' and 'nameopt' options to be separated
4638 by commas. Add 'namopt' and 'certopt' options to the 'ca' config
4639 file: this allows the display of the certificate about to be
4640 signed to be customised, to allow certain fields to be included
4641 or excluded and extension details. The old system didn't display
4642 multicharacter strings properly, omitted fields not in the policy
4643 and couldn't display additional details such as extensions.
4644 [Steve Henson]
4645
4646 *) Function EC_POINTs_mul for multiple scalar multiplication
4647 of an arbitrary number of elliptic curve points
4648 \sum scalars[i]*points[i],
4649 optionally including the generator defined for the EC_GROUP:
4650 scalar*generator + \sum scalars[i]*points[i].
4651
4652 EC_POINT_mul is a simple wrapper function for the typical case
4653 that the point list has just one item (besides the optional
4654 generator).
4655 [Bodo Moeller]
4656
4657 *) First EC_METHODs for curves over GF(p):
4658
4659 EC_GFp_simple_method() uses the basic BN_mod_mul and BN_mod_sqr
4660 operations and provides various method functions that can also
4661 operate with faster implementations of modular arithmetic.
4662
4663 EC_GFp_mont_method() reuses most functions that are part of
4664 EC_GFp_simple_method, but uses Montgomery arithmetic.
4665
4666 [Bodo Moeller; point addition and point doubling
4667 implementation directly derived from source code provided by
4668 Lenka Fibikova <fibikova@exp-math.uni-essen.de>]
4669
4670 *) Framework for elliptic curves (crypto/ec/ec.h, crypto/ec/ec_lcl.h,
4671 crypto/ec/ec_lib.c):
4672
4673 Curves are EC_GROUP objects (with an optional group generator)
4674 based on EC_METHODs that are built into the library.
4675
4676 Points are EC_POINT objects based on EC_GROUP objects.
4677
4678 Most of the framework would be able to handle curves over arbitrary
4679 finite fields, but as there are no obvious types for fields other
4680 than GF(p), some functions are limited to that for now.
4681 [Bodo Moeller]
4682
4683 *) Add the -HTTP option to s_server. It is similar to -WWW, but requires
4684 that the file contains a complete HTTP response.
4685 [Richard Levitte]
4686
4687 *) Add the ec directory to mkdef.pl and mkfiles.pl. In mkdef.pl
4688 change the def and num file printf format specifier from "%-40sXXX"
4689 to "%-39s XXX". The latter will always guarantee a space after the
4690 field while the former will cause them to run together if the field
4691 is 40 of more characters long.
4692 [Steve Henson]
4693
4694 *) Constify the cipher and digest 'method' functions and structures
4695 and modify related functions to take constant EVP_MD and EVP_CIPHER
4696 pointers.
4697 [Steve Henson]
4698
4699 *) Hide BN_CTX structure details in bn_lcl.h instead of publishing them
4700 in <openssl/bn.h>. Also further increase BN_CTX_NUM to 32.
4701 [Bodo Moeller]
4702
4703 *) Modify EVP_Digest*() routines so they now return values. Although the
4704 internal software routines can never fail additional hardware versions
4705 might.
4706 [Steve Henson]
4707
4708 *) Clean up crypto/err/err.h and change some error codes to avoid conflicts:
4709
4710 Previously ERR_R_FATAL was too small and coincided with ERR_LIB_PKCS7
4711 (= ERR_R_PKCS7_LIB); it is now 64 instead of 32.
4712
4713 ASN1 error codes
4714 ERR_R_NESTED_ASN1_ERROR
4715 ...
4716 ERR_R_MISSING_ASN1_EOS
4717 were 4 .. 9, conflicting with
4718 ERR_LIB_RSA (= ERR_R_RSA_LIB)
4719 ...
4720 ERR_LIB_PEM (= ERR_R_PEM_LIB).
4721 They are now 58 .. 63 (i.e., just below ERR_R_FATAL).
4722
4723 Add new error code 'ERR_R_INTERNAL_ERROR'.
4724 [Bodo Moeller]
4725
4726 *) Don't overuse locks in crypto/err/err.c: For data retrieval, CRYPTO_r_lock
4727 suffices.
4728 [Bodo Moeller]
4729
4730 *) New option '-subj arg' for 'openssl req' and 'openssl ca'. This
4731 sets the subject name for a new request or supersedes the
4732 subject name in a given request. Formats that can be parsed are
4733 'CN=Some Name, OU=myOU, C=IT'
4734 and
4735 'CN=Some Name/OU=myOU/C=IT'.
4736
4737 Add options '-batch' and '-verbose' to 'openssl req'.
4738 [Massimiliano Pala <madwolf@hackmasters.net>]
4739
4740 *) Introduce the possibility to access global variables through
4741 functions on platform were that's the best way to handle exporting
4742 global variables in shared libraries. To enable this functionality,
4743 one must configure with "EXPORT_VAR_AS_FN" or defined the C macro
4744 "OPENSSL_EXPORT_VAR_AS_FUNCTION" in crypto/opensslconf.h (the latter
4745 is normally done by Configure or something similar).
4746
4747 To implement a global variable, use the macro OPENSSL_IMPLEMENT_GLOBAL
4748 in the source file (foo.c) like this:
4749
4750 OPENSSL_IMPLEMENT_GLOBAL(int,foo)=1;
4751 OPENSSL_IMPLEMENT_GLOBAL(double,bar);
4752
4753 To declare a global variable, use the macros OPENSSL_DECLARE_GLOBAL
4754 and OPENSSL_GLOBAL_REF in the header file (foo.h) like this:
4755
4756 OPENSSL_DECLARE_GLOBAL(int,foo);
4757 #define foo OPENSSL_GLOBAL_REF(foo)
4758 OPENSSL_DECLARE_GLOBAL(double,bar);
4759 #define bar OPENSSL_GLOBAL_REF(bar)
4760
4761 The #defines are very important, and therefore so is including the
4762 header file everywhere where the defined globals are used.
4763
4764 The macro OPENSSL_EXPORT_VAR_AS_FUNCTION also affects the definition
4765 of ASN.1 items, but that structure is a bit different.
4766
4767 The largest change is in util/mkdef.pl which has been enhanced with
4768 better and easier to understand logic to choose which symbols should
4769 go into the Windows .def files as well as a number of fixes and code
4770 cleanup (among others, algorithm keywords are now sorted
4771 lexicographically to avoid constant rewrites).
4772 [Richard Levitte]
4773
4774 *) In BN_div() keep a copy of the sign of 'num' before writing the
4775 result to 'rm' because if rm==num the value will be overwritten
4776 and produce the wrong result if 'num' is negative: this caused
4777 problems with BN_mod() and BN_nnmod().
4778 [Steve Henson]
4779
4780 *) Function OCSP_request_verify(). This checks the signature on an
4781 OCSP request and verifies the signer certificate. The signer
4782 certificate is just checked for a generic purpose and OCSP request
4783 trust settings.
4784 [Steve Henson]
4785
4786 *) Add OCSP_check_validity() function to check the validity of OCSP
4787 responses. OCSP responses are prepared in real time and may only
4788 be a few seconds old. Simply checking that the current time lies
4789 between thisUpdate and nextUpdate max reject otherwise valid responses
4790 caused by either OCSP responder or client clock inaccuracy. Instead
4791 we allow thisUpdate and nextUpdate to fall within a certain period of
4792 the current time. The age of the response can also optionally be
4793 checked. Two new options -validity_period and -status_age added to
4794 ocsp utility.
4795 [Steve Henson]
4796
4797 *) If signature or public key algorithm is unrecognized print out its
4798 OID rather that just UNKNOWN.
4799 [Steve Henson]
4800
4801 *) Change OCSP_cert_to_id() to tolerate a NULL subject certificate and
4802 OCSP_cert_id_new() a NULL serialNumber. This allows a partial certificate
4803 ID to be generated from the issuer certificate alone which can then be
4804 passed to OCSP_id_issuer_cmp().
4805 [Steve Henson]
4806
4807 *) New compilation option ASN1_ITEM_FUNCTIONS. This causes the new
4808 ASN1 modules to export functions returning ASN1_ITEM pointers
4809 instead of the ASN1_ITEM structures themselves. This adds several
4810 new macros which allow the underlying ASN1 function/structure to
4811 be accessed transparently. As a result code should not use ASN1_ITEM
4812 references directly (such as &X509_it) but instead use the relevant
4813 macros (such as ASN1_ITEM_rptr(X509)). This option is to allow
4814 use of the new ASN1 code on platforms where exporting structures
4815 is problematical (for example in shared libraries) but exporting
4816 functions returning pointers to structures is not.
4817 [Steve Henson]
4818
4819 *) Add support for overriding the generation of SSL/TLS session IDs.
4820 These callbacks can be registered either in an SSL_CTX or per SSL.
4821 The purpose of this is to allow applications to control, if they wish,
4822 the arbitrary values chosen for use as session IDs, particularly as it
4823 can be useful for session caching in multiple-server environments. A
4824 command-line switch for testing this (and any client code that wishes
4825 to use such a feature) has been added to "s_server".
4826 [Geoff Thorpe, Lutz Jaenicke]
4827
4828 *) Modify mkdef.pl to recognise and parse preprocessor conditionals
4829 of the form '#if defined(...) || defined(...) || ...' and
4830 '#if !defined(...) && !defined(...) && ...'. This also avoids
4831 the growing number of special cases it was previously handling.
4832 [Richard Levitte]
4833
4834 *) Make all configuration macros available for application by making
4835 sure they are available in opensslconf.h, by giving them names starting
4836 with "OPENSSL_" to avoid conflicts with other packages and by making
4837 sure e_os2.h will cover all platform-specific cases together with
4838 opensslconf.h.
4839 Additionally, it is now possible to define configuration/platform-
4840 specific names (called "system identities"). In the C code, these
4841 are prefixed with "OPENSSL_SYSNAME_". e_os2.h will create another
4842 macro with the name beginning with "OPENSSL_SYS_", which is determined
4843 from "OPENSSL_SYSNAME_*" or compiler-specific macros depending on
4844 what is available.
4845 [Richard Levitte]
4846
4847 *) New option -set_serial to 'req' and 'x509' this allows the serial
4848 number to use to be specified on the command line. Previously self
4849 signed certificates were hard coded with serial number 0 and the
4850 CA options of 'x509' had to use a serial number in a file which was
4851 auto incremented.
4852 [Steve Henson]
4853
4854 *) New options to 'ca' utility to support V2 CRL entry extensions.
4855 Currently CRL reason, invalidity date and hold instruction are
4856 supported. Add new CRL extensions to V3 code and some new objects.
4857 [Steve Henson]
4858
4859 *) New function EVP_CIPHER_CTX_set_padding() this is used to
4860 disable standard block padding (aka PKCS#5 padding) in the EVP
4861 API, which was previously mandatory. This means that the data is
4862 not padded in any way and so the total length much be a multiple
4863 of the block size, otherwise an error occurs.
4864 [Steve Henson]
4865
4866 *) Initial (incomplete) OCSP SSL support.
4867 [Steve Henson]
4868
4869 *) New function OCSP_parse_url(). This splits up a URL into its host,
4870 port and path components: primarily to parse OCSP URLs. New -url
4871 option to ocsp utility.
4872 [Steve Henson]
4873
4874 *) New nonce behavior. The return value of OCSP_check_nonce() now
4875 reflects the various checks performed. Applications can decide
4876 whether to tolerate certain situations such as an absent nonce
4877 in a response when one was present in a request: the ocsp application
4878 just prints out a warning. New function OCSP_add1_basic_nonce()
4879 this is to allow responders to include a nonce in a response even if
4880 the request is nonce-less.
4881 [Steve Henson]
4882
4883 *) Disable stdin buffering in load_cert (apps/apps.c) so that no certs are
4884 skipped when using openssl x509 multiple times on a single input file,
4885 e.g. "(openssl x509 -out cert1; openssl x509 -out cert2) <certs".
4886 [Bodo Moeller]
4887
4888 *) Make ASN1_UTCTIME_set_string() and ASN1_GENERALIZEDTIME_set_string()
4889 set string type: to handle setting ASN1_TIME structures. Fix ca
4890 utility to correctly initialize revocation date of CRLs.
4891 [Steve Henson]
4892
4893 *) New option SSL_OP_CIPHER_SERVER_PREFERENCE allows the server to override
4894 the clients preferred ciphersuites and rather use its own preferences.
4895 Should help to work around M$ SGC (Server Gated Cryptography) bug in
4896 Internet Explorer by ensuring unchanged hash method during stepup.
4897 (Also replaces the broken/deactivated SSL_OP_NON_EXPORT_FIRST option.)
4898 [Lutz Jaenicke]
4899
4900 *) Make mkdef.pl recognise all DECLARE_ASN1 macros, change rijndael
4901 to aes and add a new 'exist' option to print out symbols that don't
4902 appear to exist.
4903 [Steve Henson]
4904
4905 *) Additional options to ocsp utility to allow flags to be set and
4906 additional certificates supplied.
4907 [Steve Henson]
4908
4909 *) Add the option -VAfile to 'openssl ocsp', so the user can give the
4910 OCSP client a number of certificate to only verify the response
4911 signature against.
4912 [Richard Levitte]
4913
4914 *) Update Rijndael code to version 3.0 and change EVP AES ciphers to
4915 handle the new API. Currently only ECB, CBC modes supported. Add new
4916 AES OIDs.
4917
4918 Add TLS AES ciphersuites as described in RFC3268, "Advanced
4919 Encryption Standard (AES) Ciphersuites for Transport Layer
4920 Security (TLS)". (In beta versions of OpenSSL 0.9.7, these were
4921 not enabled by default and were not part of the "ALL" ciphersuite
4922 alias because they were not yet official; they could be
4923 explicitly requested by specifying the "AESdraft" ciphersuite
4924 group alias. In the final release of OpenSSL 0.9.7, the group
4925 alias is called "AES" and is part of "ALL".)
4926 [Ben Laurie, Steve Henson, Bodo Moeller]
4927
4928 *) New function OCSP_copy_nonce() to copy nonce value (if present) from
4929 request to response.
4930 [Steve Henson]
4931
4932 *) Functions for OCSP responders. OCSP_request_onereq_count(),
4933 OCSP_request_onereq_get0(), OCSP_onereq_get0_id() and OCSP_id_get0_info()
4934 extract information from a certificate request. OCSP_response_create()
4935 creates a response and optionally adds a basic response structure.
4936 OCSP_basic_add1_status() adds a complete single response to a basic
4937 response and returns the OCSP_SINGLERESP structure just added (to allow
4938 extensions to be included for example). OCSP_basic_add1_cert() adds a
4939 certificate to a basic response and OCSP_basic_sign() signs a basic
4940 response with various flags. New helper functions ASN1_TIME_check()
4941 (checks validity of ASN1_TIME structure) and ASN1_TIME_to_generalizedtime()
4942 (converts ASN1_TIME to GeneralizedTime).
4943 [Steve Henson]
4944
4945 *) Various new functions. EVP_Digest() combines EVP_Digest{Init,Update,Final}()
4946 in a single operation. X509_get0_pubkey_bitstr() extracts the public_key
4947 structure from a certificate. X509_pubkey_digest() digests the public_key
4948 contents: this is used in various key identifiers.
4949 [Steve Henson]
4950
4951 *) Make sk_sort() tolerate a NULL argument.
4952 [Steve Henson reported by Massimiliano Pala <madwolf@comune.modena.it>]
4953
4954 *) New OCSP verify flag OCSP_TRUSTOTHER. When set the "other" certificates
4955 passed by the function are trusted implicitly. If any of them signed the
4956 response then it is assumed to be valid and is not verified.
4957 [Steve Henson]
4958
4959 *) In PKCS7_set_type() initialise content_type in PKCS7_ENC_CONTENT
4960 to data. This was previously part of the PKCS7 ASN1 code. This
4961 was causing problems with OpenSSL created PKCS#12 and PKCS#7 structures.
4962 [Steve Henson, reported by Kenneth R. Robinette
4963 <support@securenetterm.com>]
4964
4965 *) Add CRYPTO_push_info() and CRYPTO_pop_info() calls to new ASN1
4966 routines: without these tracing memory leaks is very painful.
4967 Fix leaks in PKCS12 and PKCS7 routines.
4968 [Steve Henson]
4969
4970 *) Make X509_time_adj() cope with the new behaviour of ASN1_TIME_new().
4971 Previously it initialised the 'type' argument to V_ASN1_UTCTIME which
4972 effectively meant GeneralizedTime would never be used. Now it
4973 is initialised to -1 but X509_time_adj() now has to check the value
4974 and use ASN1_TIME_set() if the value is not V_ASN1_UTCTIME or
4975 V_ASN1_GENERALIZEDTIME, without this it always uses GeneralizedTime.
4976 [Steve Henson, reported by Kenneth R. Robinette
4977 <support@securenetterm.com>]
4978
4979 *) Fixes to BN_to_ASN1_INTEGER when bn is zero. This would previously
4980 result in a zero length in the ASN1_INTEGER structure which was
4981 not consistent with the structure when d2i_ASN1_INTEGER() was used
4982 and would cause ASN1_INTEGER_cmp() to fail. Enhance s2i_ASN1_INTEGER()
4983 to cope with hex and negative integers. Fix bug in i2a_ASN1_INTEGER()
4984 where it did not print out a minus for negative ASN1_INTEGER.
4985 [Steve Henson]
4986
4987 *) Add summary printout to ocsp utility. The various functions which
4988 convert status values to strings have been renamed to:
4989 OCSP_response_status_str(), OCSP_cert_status_str() and
4990 OCSP_crl_reason_str() and are no longer static. New options
4991 to verify nonce values and to disable verification. OCSP response
4992 printout format cleaned up.
4993 [Steve Henson]
4994
4995 *) Add additional OCSP certificate checks. These are those specified
4996 in RFC2560. This consists of two separate checks: the CA of the
4997 certificate being checked must either be the OCSP signer certificate
4998 or the issuer of the OCSP signer certificate. In the latter case the
4999 OCSP signer certificate must contain the OCSP signing extended key
5000 usage. This check is performed by attempting to match the OCSP
5001 signer or the OCSP signer CA to the issuerNameHash and issuerKeyHash
5002 in the OCSP_CERTID structures of the response.
5003 [Steve Henson]
5004
5005 *) Initial OCSP certificate verification added to OCSP_basic_verify()
5006 and related routines. This uses the standard OpenSSL certificate
5007 verify routines to perform initial checks (just CA validity) and
5008 to obtain the certificate chain. Then additional checks will be
5009 performed on the chain. Currently the root CA is checked to see
5010 if it is explicitly trusted for OCSP signing. This is used to set
5011 a root CA as a global signing root: that is any certificate that
5012 chains to that CA is an acceptable OCSP signing certificate.
5013 [Steve Henson]
5014
5015 *) New '-extfile ...' option to 'openssl ca' for reading X.509v3
5016 extensions from a separate configuration file.
5017 As when reading extensions from the main configuration file,
5018 the '-extensions ...' option may be used for specifying the
5019 section to use.
5020 [Massimiliano Pala <madwolf@comune.modena.it>]
5021
5022 *) New OCSP utility. Allows OCSP requests to be generated or
5023 read. The request can be sent to a responder and the output
5024 parsed, outputed or printed in text form. Not complete yet:
5025 still needs to check the OCSP response validity.
5026 [Steve Henson]
5027
5028 *) New subcommands for 'openssl ca':
5029 'openssl ca -status <serial>' prints the status of the cert with
5030 the given serial number (according to the index file).
5031 'openssl ca -updatedb' updates the expiry status of certificates
5032 in the index file.
5033 [Massimiliano Pala <madwolf@comune.modena.it>]
5034
5035 *) New '-newreq-nodes' command option to CA.pl. This is like
5036 '-newreq', but calls 'openssl req' with the '-nodes' option
5037 so that the resulting key is not encrypted.
5038 [Damien Miller <djm@mindrot.org>]
5039
5040 *) New configuration for the GNU Hurd.
5041 [Jonathan Bartlett <johnnyb@wolfram.com> via Richard Levitte]
5042
5043 *) Initial code to implement OCSP basic response verify. This
5044 is currently incomplete. Currently just finds the signer's
5045 certificate and verifies the signature on the response.
5046 [Steve Henson]
5047
5048 *) New SSLeay_version code SSLEAY_DIR to determine the compiled-in
5049 value of OPENSSLDIR. This is available via the new '-d' option
5050 to 'openssl version', and is also included in 'openssl version -a'.
5051 [Bodo Moeller]
5052
5053 *) Allowing defining memory allocation callbacks that will be given
5054 file name and line number information in additional arguments
5055 (a const char* and an int). The basic functionality remains, as
5056 well as the original possibility to just replace malloc(),
5057 realloc() and free() by functions that do not know about these
5058 additional arguments. To register and find out the current
5059 settings for extended allocation functions, the following
5060 functions are provided:
5061
5062 CRYPTO_set_mem_ex_functions
5063 CRYPTO_set_locked_mem_ex_functions
5064 CRYPTO_get_mem_ex_functions
5065 CRYPTO_get_locked_mem_ex_functions
5066
5067 These work the same way as CRYPTO_set_mem_functions and friends.
5068 CRYPTO_get_[locked_]mem_functions now writes 0 where such an
5069 extended allocation function is enabled.
5070 Similarly, CRYPTO_get_[locked_]mem_ex_functions writes 0 where
5071 a conventional allocation function is enabled.
5072 [Richard Levitte, Bodo Moeller]
5073
5074 *) Finish off removing the remaining LHASH function pointer casts.
5075 There should no longer be any prototype-casting required when using
5076 the LHASH abstraction, and any casts that remain are "bugs". See
5077 the callback types and macros at the head of lhash.h for details
5078 (and "OBJ_cleanup" in crypto/objects/obj_dat.c as an example).
5079 [Geoff Thorpe]
5080
5081 *) Add automatic query of EGD sockets in RAND_poll() for the unix variant.
5082 If /dev/[u]random devices are not available or do not return enough
5083 entropy, EGD style sockets (served by EGD or PRNGD) will automatically
5084 be queried.
5085 The locations /var/run/egd-pool, /dev/egd-pool, /etc/egd-pool, and
5086 /etc/entropy will be queried once each in this sequence, quering stops
5087 when enough entropy was collected without querying more sockets.
5088 [Lutz Jaenicke]
5089
5090 *) Change the Unix RAND_poll() variant to be able to poll several
5091 random devices, as specified by DEVRANDOM, until a sufficient amount
5092 of data has been collected. We spend at most 10 ms on each file
5093 (select timeout) and read in non-blocking mode. DEVRANDOM now
5094 defaults to the list "/dev/urandom", "/dev/random", "/dev/srandom"
5095 (previously it was just the string "/dev/urandom"), so on typical
5096 platforms the 10 ms delay will never occur.
5097 Also separate out the Unix variant to its own file, rand_unix.c.
5098 For VMS, there's a currently-empty rand_vms.c.
5099 [Richard Levitte]
5100
5101 *) Move OCSP client related routines to ocsp_cl.c. These
5102 provide utility functions which an application needing
5103 to issue a request to an OCSP responder and analyse the
5104 response will typically need: as opposed to those which an
5105 OCSP responder itself would need which will be added later.
5106
5107 OCSP_request_sign() signs an OCSP request with an API similar
5108 to PKCS7_sign(). OCSP_response_status() returns status of OCSP
5109 response. OCSP_response_get1_basic() extracts basic response
5110 from response. OCSP_resp_find_status(): finds and extracts status
5111 information from an OCSP_CERTID structure (which will be created
5112 when the request structure is built). These are built from lower
5113 level functions which work on OCSP_SINGLERESP structures but
5114 wont normally be used unless the application wishes to examine
5115 extensions in the OCSP response for example.
5116
5117 Replace nonce routines with a pair of functions.
5118 OCSP_request_add1_nonce() adds a nonce value and optionally
5119 generates a random value. OCSP_check_nonce() checks the
5120 validity of the nonce in an OCSP response.
5121 [Steve Henson]
5122
5123 *) Change function OCSP_request_add() to OCSP_request_add0_id().
5124 This doesn't copy the supplied OCSP_CERTID and avoids the
5125 need to free up the newly created id. Change return type
5126 to OCSP_ONEREQ to return the internal OCSP_ONEREQ structure.
5127 This can then be used to add extensions to the request.
5128 Deleted OCSP_request_new(), since most of its functionality
5129 is now in OCSP_REQUEST_new() (and the case insensitive name
5130 clash) apart from the ability to set the request name which
5131 will be added elsewhere.
5132 [Steve Henson]
5133
5134 *) Update OCSP API. Remove obsolete extensions argument from
5135 various functions. Extensions are now handled using the new
5136 OCSP extension code. New simple OCSP HTTP function which
5137 can be used to send requests and parse the response.
5138 [Steve Henson]
5139
5140 *) Fix the PKCS#7 (S/MIME) code to work with new ASN1. Two new
5141 ASN1_ITEM structures help with sign and verify. PKCS7_ATTR_SIGN
5142 uses the special reorder version of SET OF to sort the attributes
5143 and reorder them to match the encoded order. This resolves a long
5144 standing problem: a verify on a PKCS7 structure just after signing
5145 it used to fail because the attribute order did not match the
5146 encoded order. PKCS7_ATTR_VERIFY does not reorder the attributes:
5147 it uses the received order. This is necessary to tolerate some broken
5148 software that does not order SET OF. This is handled by encoding
5149 as a SEQUENCE OF but using implicit tagging (with UNIVERSAL class)
5150 to produce the required SET OF.
5151 [Steve Henson]
5152
5153 *) Have mk1mf.pl generate the macros OPENSSL_BUILD_SHLIBCRYPTO and
5154 OPENSSL_BUILD_SHLIBSSL and use them appropriately in the header
5155 files to get correct declarations of the ASN.1 item variables.
5156 [Richard Levitte]
5157
5158 *) Rewrite of PKCS#12 code to use new ASN1 functionality. Replace many
5159 PKCS#12 macros with real functions. Fix two unrelated ASN1 bugs:
5160 asn1_check_tlen() would sometimes attempt to use 'ctx' when it was
5161 NULL and ASN1_TYPE was not dereferenced properly in asn1_ex_c2i().
5162 New ASN1 macro: DECLARE_ASN1_ITEM() which just declares the relevant
5163 ASN1_ITEM and no wrapper functions.
5164 [Steve Henson]
5165
5166 *) New functions or ASN1_item_d2i_fp() and ASN1_item_d2i_bio(). These
5167 replace the old function pointer based I/O routines. Change most of
5168 the *_d2i_bio() and *_d2i_fp() functions to use these.
5169 [Steve Henson]
5170
5171 *) Enhance mkdef.pl to be more accepting about spacing in C preprocessor
5172 lines, recognice more "algorithms" that can be deselected, and make
5173 it complain about algorithm deselection that isn't recognised.
5174 [Richard Levitte]
5175
5176 *) New ASN1 functions to handle dup, sign, verify, digest, pack and
5177 unpack operations in terms of ASN1_ITEM. Modify existing wrappers
5178 to use new functions. Add NO_ASN1_OLD which can be set to remove
5179 some old style ASN1 functions: this can be used to determine if old
5180 code will still work when these eventually go away.
5181 [Steve Henson]
5182
5183 *) New extension functions for OCSP structures, these follow the
5184 same conventions as certificates and CRLs.
5185 [Steve Henson]
5186
5187 *) New function X509V3_add1_i2d(). This automatically encodes and
5188 adds an extension. Its behaviour can be customised with various
5189 flags to append, replace or delete. Various wrappers added for
5190 certifcates and CRLs.
5191 [Steve Henson]
5192
5193 *) Fix to avoid calling the underlying ASN1 print routine when
5194 an extension cannot be parsed. Correct a typo in the
5195 OCSP_SERVICELOC extension. Tidy up print OCSP format.
5196 [Steve Henson]
5197
5198 *) Make mkdef.pl parse some of the ASN1 macros and add apropriate
5199 entries for variables.
5200 [Steve Henson]
5201
5202 *) Add functionality to apps/openssl.c for detecting locking
5203 problems: As the program is single-threaded, all we have
5204 to do is register a locking callback using an array for
5205 storing which locks are currently held by the program.
5206 [Bodo Moeller]
5207
5208 *) Use a lock around the call to CRYPTO_get_ex_new_index() in
5209 SSL_get_ex_data_X509_STORE_idx(), which is used in
5210 ssl_verify_cert_chain() and thus can be called at any time
5211 during TLS/SSL handshakes so that thread-safety is essential.
5212 Unfortunately, the ex_data design is not at all suited
5213 for multi-threaded use, so it probably should be abolished.
5214 [Bodo Moeller]
5215
5216 *) Added Broadcom "ubsec" ENGINE to OpenSSL.
5217 [Broadcom, tweaked and integrated by Geoff Thorpe]
5218
5219 *) Move common extension printing code to new function
5220 X509V3_print_extensions(). Reorganise OCSP print routines and
5221 implement some needed OCSP ASN1 functions. Add OCSP extensions.
5222 [Steve Henson]
5223
5224 *) New function X509_signature_print() to remove duplication in some
5225 print routines.
5226 [Steve Henson]
5227
5228 *) Add a special meaning when SET OF and SEQUENCE OF flags are both
5229 set (this was treated exactly the same as SET OF previously). This
5230 is used to reorder the STACK representing the structure to match the
5231 encoding. This will be used to get round a problem where a PKCS7
5232 structure which was signed could not be verified because the STACK
5233 order did not reflect the encoded order.
5234 [Steve Henson]
5235
5236 *) Reimplement the OCSP ASN1 module using the new code.
5237 [Steve Henson]
5238
5239 *) Update the X509V3 code to permit the use of an ASN1_ITEM structure
5240 for its ASN1 operations. The old style function pointers still exist
5241 for now but they will eventually go away.
5242 [Steve Henson]
5243
5244 *) Merge in replacement ASN1 code from the ASN1 branch. This almost
5245 completely replaces the old ASN1 functionality with a table driven
5246 encoder and decoder which interprets an ASN1_ITEM structure describing
5247 the ASN1 module. Compatibility with the existing ASN1 API (i2d,d2i) is
5248 largely maintained. Almost all of the old asn1_mac.h macro based ASN1
5249 has also been converted to the new form.
5250 [Steve Henson]
5251
5252 *) Change BN_mod_exp_recp so that negative moduli are tolerated
5253 (the sign is ignored). Similarly, ignore the sign in BN_MONT_CTX_set
5254 so that BN_mod_exp_mont and BN_mod_exp_mont_word work
5255 for negative moduli.
5256 [Bodo Moeller]
5257
5258 *) Fix BN_uadd and BN_usub: Always return non-negative results instead
5259 of not touching the result's sign bit.
5260 [Bodo Moeller]
5261
5262 *) BN_div bugfix: If the result is 0, the sign (res->neg) must not be
5263 set.
5264 [Bodo Moeller]
5265
5266 *) Changed the LHASH code to use prototypes for callbacks, and created
5267 macros to declare and implement thin (optionally static) functions
5268 that provide type-safety and avoid function pointer casting for the
5269 type-specific callbacks.
5270 [Geoff Thorpe]
5271
5272 *) Added Kerberos Cipher Suites to be used with TLS, as written in
5273 RFC 2712.
5274 [Veers Staats <staatsvr@asc.hpc.mil>,
5275 Jeffrey Altman <jaltman@columbia.edu>, via Richard Levitte]
5276
5277 *) Reformat the FAQ so the different questions and answers can be divided
5278 in sections depending on the subject.
5279 [Richard Levitte]
5280
5281 *) Have the zlib compression code load ZLIB.DLL dynamically under
5282 Windows.
5283 [Richard Levitte]
5284
5285 *) New function BN_mod_sqrt for computing square roots modulo a prime
5286 (using the probabilistic Tonelli-Shanks algorithm unless
5287 p == 3 (mod 4) or p == 5 (mod 8), which are cases that can
5288 be handled deterministically).
5289 [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller]
5290
5291 *) Make BN_mod_inverse faster by explicitly handling small quotients
5292 in the Euclid loop. (Speed gain about 20% for small moduli [256 or
5293 512 bits], about 30% for larger ones [1024 or 2048 bits].)
5294 [Bodo Moeller]
5295
5296 *) New function BN_kronecker.
5297 [Bodo Moeller]
5298
5299 *) Fix BN_gcd so that it works on negative inputs; the result is
5300 positive unless both parameters are zero.
5301 Previously something reasonably close to an infinite loop was
5302 possible because numbers could be growing instead of shrinking
5303 in the implementation of Euclid's algorithm.
5304 [Bodo Moeller]
5305
5306 *) Fix BN_is_word() and BN_is_one() macros to take into account the
5307 sign of the number in question.
5308
5309 Fix BN_is_word(a,w) to work correctly for w == 0.
5310
5311 The old BN_is_word(a,w) macro is now called BN_abs_is_word(a,w)
5312 because its test if the absolute value of 'a' equals 'w'.
5313 Note that BN_abs_is_word does *not* handle w == 0 reliably;
5314 it exists mostly for use in the implementations of BN_is_zero(),
5315 BN_is_one(), and BN_is_word().
5316 [Bodo Moeller]
5317
5318 *) New function BN_swap.
5319 [Bodo Moeller]
5320
5321 *) Use BN_nnmod instead of BN_mod in crypto/bn/bn_exp.c so that
5322 the exponentiation functions are more likely to produce reasonable
5323 results on negative inputs.
5324 [Bodo Moeller]
5325
5326 *) Change BN_mod_mul so that the result is always non-negative.
5327 Previously, it could be negative if one of the factors was negative;
5328 I don't think anyone really wanted that behaviour.
5329 [Bodo Moeller]
5330
5331 *) Move BN_mod_... functions into new file crypto/bn/bn_mod.c
5332 (except for exponentiation, which stays in crypto/bn/bn_exp.c,
5333 and BN_mod_mul_reciprocal, which stays in crypto/bn/bn_recp.c)
5334 and add new functions:
5335
5336 BN_nnmod
5337 BN_mod_sqr
5338 BN_mod_add
5339 BN_mod_add_quick
5340 BN_mod_sub
5341 BN_mod_sub_quick
5342 BN_mod_lshift1
5343 BN_mod_lshift1_quick
5344 BN_mod_lshift
5345 BN_mod_lshift_quick
5346
5347 These functions always generate non-negative results.
5348
5349 BN_nnmod otherwise is like BN_mod (if BN_mod computes a remainder r
5350 such that |m| < r < 0, BN_nnmod will output rem + |m| instead).
5351
5352 BN_mod_XXX_quick(r, a, [b,] m) generates the same result as
5353 BN_mod_XXX(r, a, [b,] m, ctx), but requires that a [and b]
5354 be reduced modulo m.
5355 [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller]
5356
5357 #if 0
5358 The following entry accidentily appeared in the CHANGES file
5359 distributed with OpenSSL 0.9.7. The modifications described in
5360 it do *not* apply to OpenSSL 0.9.7.
5361
5362 *) Remove a few calls to bn_wexpand() in BN_sqr() (the one in there
5363 was actually never needed) and in BN_mul(). The removal in BN_mul()
5364 required a small change in bn_mul_part_recursive() and the addition
5365 of the functions bn_cmp_part_words(), bn_sub_part_words() and
5366 bn_add_part_words(), which do the same thing as bn_cmp_words(),
5367 bn_sub_words() and bn_add_words() except they take arrays with
5368 differing sizes.
5369 [Richard Levitte]
5370 #endif
5371
5372 *) In 'openssl passwd', verify passwords read from the terminal
5373 unless the '-salt' option is used (which usually means that
5374 verification would just waste user's time since the resulting
5375 hash is going to be compared with some given password hash)
5376 or the new '-noverify' option is used.
5377
5378 This is an incompatible change, but it does not affect
5379 non-interactive use of 'openssl passwd' (passwords on the command
5380 line, '-stdin' option, '-in ...' option) and thus should not
5381 cause any problems.
5382 [Bodo Moeller]
5383
5384 *) Remove all references to RSAref, since there's no more need for it.
5385 [Richard Levitte]
5386
5387 *) Make DSO load along a path given through an environment variable
5388 (SHLIB_PATH) with shl_load().
5389 [Richard Levitte]
5390
5391 *) Constify the ENGINE code as a result of BIGNUM constification.
5392 Also constify the RSA code and most things related to it. In a
5393 few places, most notable in the depth of the ASN.1 code, ugly
5394 casts back to non-const were required (to be solved at a later
5395 time)
5396 [Richard Levitte]
5397
5398 *) Make it so the openssl application has all engines loaded by default.
5399 [Richard Levitte]
5400
5401 *) Constify the BIGNUM routines a little more.
5402 [Richard Levitte]
5403
5404 *) Add the following functions:
5405
5406 ENGINE_load_cswift()
5407 ENGINE_load_chil()
5408 ENGINE_load_atalla()
5409 ENGINE_load_nuron()
5410 ENGINE_load_builtin_engines()
5411
5412 That way, an application can itself choose if external engines that
5413 are built-in in OpenSSL shall ever be used or not. The benefit is
5414 that applications won't have to be linked with libdl or other dso
5415 libraries unless it's really needed.
5416
5417 Changed 'openssl engine' to load all engines on demand.
5418 Changed the engine header files to avoid the duplication of some
5419 declarations (they differed!).
5420 [Richard Levitte]
5421
5422 *) 'openssl engine' can now list capabilities.
5423 [Richard Levitte]
5424
5425 *) Better error reporting in 'openssl engine'.
5426 [Richard Levitte]
5427
5428 *) Never call load_dh_param(NULL) in s_server.
5429 [Bodo Moeller]
5430
5431 *) Add engine application. It can currently list engines by name and
5432 identity, and test if they are actually available.
5433 [Richard Levitte]
5434
5435 *) Improve RPM specification file by forcing symbolic linking and making
5436 sure the installed documentation is also owned by root.root.
5437 [Damien Miller <djm@mindrot.org>]
5438
5439 *) Give the OpenSSL applications more possibilities to make use of
5440 keys (public as well as private) handled by engines.
5441 [Richard Levitte]
5442
5443 *) Add OCSP code that comes from CertCo.
5444 [Richard Levitte]
5445
5446 *) Add VMS support for the Rijndael code.
5447 [Richard Levitte]
5448
5449 *) Added untested support for Nuron crypto accelerator.
5450 [Ben Laurie]
5451
5452 *) Add support for external cryptographic devices. This code was
5453 previously distributed separately as the "engine" branch.
5454 [Geoff Thorpe, Richard Levitte]
5455
5456 *) Rework the filename-translation in the DSO code. It is now possible to
5457 have far greater control over how a "name" is turned into a filename
5458 depending on the operating environment and any oddities about the
5459 different shared library filenames on each system.
5460 [Geoff Thorpe]
5461
5462 *) Support threads on FreeBSD-elf in Configure.
5463 [Richard Levitte]
5464
5465 *) Fix for SHA1 assembly problem with MASM: it produces
5466 warnings about corrupt line number information when assembling
5467 with debugging information. This is caused by the overlapping
5468 of two sections.
5469 [Bernd Matthes <mainbug@celocom.de>, Steve Henson]
5470
5471 *) NCONF changes.
5472 NCONF_get_number() has no error checking at all. As a replacement,
5473 NCONF_get_number_e() is defined (_e for "error checking") and is
5474 promoted strongly. The old NCONF_get_number is kept around for
5475 binary backward compatibility.
5476 Make it possible for methods to load from something other than a BIO,
5477 by providing a function pointer that is given a name instead of a BIO.
5478 For example, this could be used to load configuration data from an
5479 LDAP server.
5480 [Richard Levitte]
5481
5482 *) Fix for non blocking accept BIOs. Added new I/O special reason
5483 BIO_RR_ACCEPT to cover this case. Previously use of accept BIOs
5484 with non blocking I/O was not possible because no retry code was
5485 implemented. Also added new SSL code SSL_WANT_ACCEPT to cover
5486 this case.
5487 [Steve Henson]
5488
5489 *) Added the beginnings of Rijndael support.
5490 [Ben Laurie]
5491
5492 *) Fix for bug in DirectoryString mask setting. Add support for
5493 X509_NAME_print_ex() in 'req' and X509_print_ex() function
5494 to allow certificate printing to more controllable, additional
5495 'certopt' option to 'x509' to allow new printing options to be
5496 set.
5497 [Steve Henson]
5498
5499 *) Clean old EAY MD5 hack from e_os.h.
5500 [Richard Levitte]
5501
5502 Changes between 0.9.6l and 0.9.6m [17 Mar 2004]
5503
5504 *) Fix null-pointer assignment in do_change_cipher_spec() revealed
5505 by using the Codenomicon TLS Test Tool (CVE-2004-0079)
5506 [Joe Orton, Steve Henson]
5507
5508 Changes between 0.9.6k and 0.9.6l [04 Nov 2003]
5509
5510 *) Fix additional bug revealed by the NISCC test suite:
5511
5512 Stop bug triggering large recursion when presented with
5513 certain ASN.1 tags (CVE-2003-0851)
5514 [Steve Henson]
5515
5516 Changes between 0.9.6j and 0.9.6k [30 Sep 2003]
5517
5518 *) Fix various bugs revealed by running the NISCC test suite:
5519
5520 Stop out of bounds reads in the ASN1 code when presented with
5521 invalid tags (CVE-2003-0543 and CVE-2003-0544).
5522
5523 If verify callback ignores invalid public key errors don't try to check
5524 certificate signature with the NULL public key.
5525
5526 [Steve Henson]
5527
5528 *) In ssl3_accept() (ssl/s3_srvr.c) only accept a client certificate
5529 if the server requested one: as stated in TLS 1.0 and SSL 3.0
5530 specifications.
5531 [Steve Henson]
5532
5533 *) In ssl3_get_client_hello() (ssl/s3_srvr.c), tolerate additional
5534 extra data after the compression methods not only for TLS 1.0
5535 but also for SSL 3.0 (as required by the specification).
5536 [Bodo Moeller; problem pointed out by Matthias Loepfe]
5537
5538 *) Change X509_certificate_type() to mark the key as exported/exportable
5539 when it's 512 *bits* long, not 512 bytes.
5540 [Richard Levitte]
5541
5542 Changes between 0.9.6i and 0.9.6j [10 Apr 2003]
5543
5544 *) Countermeasure against the Klima-Pokorny-Rosa extension of
5545 Bleichbacher's attack on PKCS #1 v1.5 padding: treat
5546 a protocol version number mismatch like a decryption error
5547 in ssl3_get_client_key_exchange (ssl/s3_srvr.c).
5548 [Bodo Moeller]
5549
5550 *) Turn on RSA blinding by default in the default implementation
5551 to avoid a timing attack. Applications that don't want it can call
5552 RSA_blinding_off() or use the new flag RSA_FLAG_NO_BLINDING.
5553 They would be ill-advised to do so in most cases.
5554 [Ben Laurie, Steve Henson, Geoff Thorpe, Bodo Moeller]
5555
5556 *) Change RSA blinding code so that it works when the PRNG is not
5557 seeded (in this case, the secret RSA exponent is abused as
5558 an unpredictable seed -- if it is not unpredictable, there
5559 is no point in blinding anyway). Make RSA blinding thread-safe
5560 by remembering the creator's thread ID in rsa->blinding and
5561 having all other threads use local one-time blinding factors
5562 (this requires more computation than sharing rsa->blinding, but
5563 avoids excessive locking; and if an RSA object is not shared
5564 between threads, blinding will still be very fast).
5565 [Bodo Moeller]
5566
5567 Changes between 0.9.6h and 0.9.6i [19 Feb 2003]
5568
5569 *) In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked
5570 via timing by performing a MAC computation even if incorrrect
5571 block cipher padding has been found. This is a countermeasure
5572 against active attacks where the attacker has to distinguish
5573 between bad padding and a MAC verification error. (CVE-2003-0078)
5574
5575 [Bodo Moeller; problem pointed out by Brice Canvel (EPFL),
5576 Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and
5577 Martin Vuagnoux (EPFL, Ilion)]
5578
5579 Changes between 0.9.6g and 0.9.6h [5 Dec 2002]
5580
5581 *) New function OPENSSL_cleanse(), which is used to cleanse a section of
5582 memory from it's contents. This is done with a counter that will
5583 place alternating values in each byte. This can be used to solve
5584 two issues: 1) the removal of calls to memset() by highly optimizing
5585 compilers, and 2) cleansing with other values than 0, since those can
5586 be read through on certain media, for example a swap space on disk.
5587 [Geoff Thorpe]
5588
5589 *) Bugfix: client side session caching did not work with external caching,
5590 because the session->cipher setting was not restored when reloading
5591 from the external cache. This problem was masked, when
5592 SSL_OP_NETSCAPE_REUSE_CIPHER_CHANGE_BUG (part of SSL_OP_ALL) was set.
5593 (Found by Steve Haslam <steve@araqnid.ddts.net>.)
5594 [Lutz Jaenicke]
5595
5596 *) Fix client_certificate (ssl/s2_clnt.c): The permissible total
5597 length of the REQUEST-CERTIFICATE message is 18 .. 34, not 17 .. 33.
5598 [Zeev Lieber <zeev-l@yahoo.com>]
5599
5600 *) Undo an undocumented change introduced in 0.9.6e which caused
5601 repeated calls to OpenSSL_add_all_ciphers() and
5602 OpenSSL_add_all_digests() to be ignored, even after calling
5603 EVP_cleanup().
5604 [Richard Levitte]
5605
5606 *) Change the default configuration reader to deal with last line not
5607 being properly terminated.
5608 [Richard Levitte]
5609
5610 *) Change X509_NAME_cmp() so it applies the special rules on handling
5611 DN values that are of type PrintableString, as well as RDNs of type
5612 emailAddress where the value has the type ia5String.
5613 [stefank@valicert.com via Richard Levitte]
5614
5615 *) Add a SSL_SESS_CACHE_NO_INTERNAL_STORE flag to take over half
5616 the job SSL_SESS_CACHE_NO_INTERNAL_LOOKUP was inconsistently
5617 doing, define a new flag (SSL_SESS_CACHE_NO_INTERNAL) to be
5618 the bitwise-OR of the two for use by the majority of applications
5619 wanting this behaviour, and update the docs. The documented
5620 behaviour and actual behaviour were inconsistent and had been
5621 changing anyway, so this is more a bug-fix than a behavioural
5622 change.
5623 [Geoff Thorpe, diagnosed by Nadav Har'El]
5624
5625 *) Don't impose a 16-byte length minimum on session IDs in ssl/s3_clnt.c
5626 (the SSL 3.0 and TLS 1.0 specifications allow any length up to 32 bytes).
5627 [Bodo Moeller]
5628
5629 *) Fix initialization code race conditions in
5630 SSLv23_method(), SSLv23_client_method(), SSLv23_server_method(),
5631 SSLv2_method(), SSLv2_client_method(), SSLv2_server_method(),
5632 SSLv3_method(), SSLv3_client_method(), SSLv3_server_method(),
5633 TLSv1_method(), TLSv1_client_method(), TLSv1_server_method(),
5634 ssl2_get_cipher_by_char(),
5635 ssl3_get_cipher_by_char().
5636 [Patrick McCormick <patrick@tellme.com>, Bodo Moeller]
5637
5638 *) Reorder cleanup sequence in SSL_CTX_free(): only remove the ex_data after
5639 the cached sessions are flushed, as the remove_cb() might use ex_data
5640 contents. Bug found by Sam Varshavchik <mrsam@courier-mta.com>
5641 (see [openssl.org #212]).
5642 [Geoff Thorpe, Lutz Jaenicke]
5643
5644 *) Fix typo in OBJ_txt2obj which incorrectly passed the content
5645 length, instead of the encoding length to d2i_ASN1_OBJECT.
5646 [Steve Henson]
5647
5648 Changes between 0.9.6f and 0.9.6g [9 Aug 2002]
5649
5650 *) [In 0.9.6g-engine release:]
5651 Fix crypto/engine/vendor_defns/cswift.h for WIN32 (use '_stdcall').
5652 [Lynn Gazis <lgazis@rainbow.com>]
5653
5654 Changes between 0.9.6e and 0.9.6f [8 Aug 2002]
5655
5656 *) Fix ASN1 checks. Check for overflow by comparing with LONG_MAX
5657 and get fix the header length calculation.
5658 [Florian Weimer <Weimer@CERT.Uni-Stuttgart.DE>,
5659 Alon Kantor <alonk@checkpoint.com> (and others),
5660 Steve Henson]
5661
5662 *) Use proper error handling instead of 'assertions' in buffer
5663 overflow checks added in 0.9.6e. This prevents DoS (the
5664 assertions could call abort()).
5665 [Arne Ansper <arne@ats.cyber.ee>, Bodo Moeller]
5666
5667 Changes between 0.9.6d and 0.9.6e [30 Jul 2002]
5668
5669 *) Add various sanity checks to asn1_get_length() to reject
5670 the ASN1 length bytes if they exceed sizeof(long), will appear
5671 negative or the content length exceeds the length of the
5672 supplied buffer.
5673 [Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>]
5674
5675 *) Fix cipher selection routines: ciphers without encryption had no flags
5676 for the cipher strength set and where therefore not handled correctly
5677 by the selection routines (PR #130).
5678 [Lutz Jaenicke]
5679
5680 *) Fix EVP_dsa_sha macro.
5681 [Nils Larsch]
5682
5683 *) New option
5684 SSL_OP_DONT_INSERT_EMPTY_FRAGMENTS
5685 for disabling the SSL 3.0/TLS 1.0 CBC vulnerability countermeasure
5686 that was added in OpenSSL 0.9.6d.
5687
5688 As the countermeasure turned out to be incompatible with some
5689 broken SSL implementations, the new option is part of SSL_OP_ALL.
5690 SSL_OP_ALL is usually employed when compatibility with weird SSL
5691 implementations is desired (e.g. '-bugs' option to 's_client' and
5692 's_server'), so the new option is automatically set in many
5693 applications.
5694 [Bodo Moeller]
5695
5696 *) Changes in security patch:
5697
5698 Changes marked "(CHATS)" were sponsored by the Defense Advanced
5699 Research Projects Agency (DARPA) and Air Force Research Laboratory,
5700 Air Force Materiel Command, USAF, under agreement number
5701 F30602-01-2-0537.
5702
5703 *) Add various sanity checks to asn1_get_length() to reject
5704 the ASN1 length bytes if they exceed sizeof(long), will appear
5705 negative or the content length exceeds the length of the
5706 supplied buffer. (CVE-2002-0659)
5707 [Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>]
5708
5709 *) Assertions for various potential buffer overflows, not known to
5710 happen in practice.
5711 [Ben Laurie (CHATS)]
5712
5713 *) Various temporary buffers to hold ASCII versions of integers were
5714 too small for 64 bit platforms. (CVE-2002-0655)
5715 [Matthew Byng-Maddick <mbm@aldigital.co.uk> and Ben Laurie (CHATS)>
5716
5717 *) Remote buffer overflow in SSL3 protocol - an attacker could
5718 supply an oversized session ID to a client. (CVE-2002-0656)
5719 [Ben Laurie (CHATS)]
5720
5721 *) Remote buffer overflow in SSL2 protocol - an attacker could
5722 supply an oversized client master key. (CVE-2002-0656)
5723 [Ben Laurie (CHATS)]
5724
5725 Changes between 0.9.6c and 0.9.6d [9 May 2002]
5726
5727 *) Fix crypto/asn1/a_sign.c so that 'parameters' is omitted (not
5728 encoded as NULL) with id-dsa-with-sha1.
5729 [Nils Larsch <nla@trustcenter.de>; problem pointed out by Bodo Moeller]
5730
5731 *) Check various X509_...() return values in apps/req.c.
5732 [Nils Larsch <nla@trustcenter.de>]
5733
5734 *) Fix BASE64 decode (EVP_DecodeUpdate) for data with CR/LF ended lines:
5735 an end-of-file condition would erronously be flagged, when the CRLF
5736 was just at the end of a processed block. The bug was discovered when
5737 processing data through a buffering memory BIO handing the data to a
5738 BASE64-decoding BIO. Bug fund and patch submitted by Pavel Tsekov
5739 <ptsekov@syntrex.com> and Nedelcho Stanev.
5740 [Lutz Jaenicke]
5741
5742 *) Implement a countermeasure against a vulnerability recently found
5743 in CBC ciphersuites in SSL 3.0/TLS 1.0: Send an empty fragment
5744 before application data chunks to avoid the use of known IVs
5745 with data potentially chosen by the attacker.
5746 [Bodo Moeller]
5747
5748 *) Fix length checks in ssl3_get_client_hello().
5749 [Bodo Moeller]
5750
5751 *) TLS/SSL library bugfix: use s->s3->in_read_app_data differently
5752 to prevent ssl3_read_internal() from incorrectly assuming that
5753 ssl3_read_bytes() found application data while handshake
5754 processing was enabled when in fact s->s3->in_read_app_data was
5755 merely automatically cleared during the initial handshake.
5756 [Bodo Moeller; problem pointed out by Arne Ansper <arne@ats.cyber.ee>]
5757
5758 *) Fix object definitions for Private and Enterprise: they were not
5759 recognized in their shortname (=lowercase) representation. Extend
5760 obj_dat.pl to issue an error when using undefined keywords instead
5761 of silently ignoring the problem (Svenning Sorensen
5762 <sss@sss.dnsalias.net>).
5763 [Lutz Jaenicke]
5764
5765 *) Fix DH_generate_parameters() so that it works for 'non-standard'
5766 generators, i.e. generators other than 2 and 5. (Previously, the
5767 code did not properly initialise the 'add' and 'rem' values to
5768 BN_generate_prime().)
5769
5770 In the new general case, we do not insist that 'generator' is
5771 actually a primitive root: This requirement is rather pointless;
5772 a generator of the order-q subgroup is just as good, if not
5773 better.
5774 [Bodo Moeller]
5775
5776 *) Map new X509 verification errors to alerts. Discovered and submitted by
5777 Tom Wu <tom@arcot.com>.
5778 [Lutz Jaenicke]
5779
5780 *) Fix ssl3_pending() (ssl/s3_lib.c) to prevent SSL_pending() from
5781 returning non-zero before the data has been completely received
5782 when using non-blocking I/O.
5783 [Bodo Moeller; problem pointed out by John Hughes]
5784
5785 *) Some of the ciphers missed the strength entry (SSL_LOW etc).
5786 [Ben Laurie, Lutz Jaenicke]
5787
5788 *) Fix bug in SSL_clear(): bad sessions were not removed (found by
5789 Yoram Zahavi <YoramZ@gilian.com>).
5790 [Lutz Jaenicke]
5791
5792 *) Add information about CygWin 1.3 and on, and preserve proper
5793 configuration for the versions before that.
5794 [Corinna Vinschen <vinschen@redhat.com> and Richard Levitte]
5795
5796 *) Make removal from session cache (SSL_CTX_remove_session()) more robust:
5797 check whether we deal with a copy of a session and do not delete from
5798 the cache in this case. Problem reported by "Izhar Shoshani Levi"
5799 <izhar@checkpoint.com>.
5800 [Lutz Jaenicke]
5801
5802 *) Do not store session data into the internal session cache, if it
5803 is never intended to be looked up (SSL_SESS_CACHE_NO_INTERNAL_LOOKUP
5804 flag is set). Proposed by Aslam <aslam@funk.com>.
5805 [Lutz Jaenicke]
5806
5807 *) Have ASN1_BIT_STRING_set_bit() really clear a bit when the requested
5808 value is 0.
5809 [Richard Levitte]
5810
5811 *) [In 0.9.6d-engine release:]
5812 Fix a crashbug and a logic bug in hwcrhk_load_pubkey().
5813 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte]
5814
5815 *) Add the configuration target linux-s390x.
5816 [Neale Ferguson <Neale.Ferguson@SoftwareAG-USA.com> via Richard Levitte]
5817
5818 *) The earlier bugfix for the SSL3_ST_SW_HELLO_REQ_C case of
5819 ssl3_accept (ssl/s3_srvr.c) incorrectly used a local flag
5820 variable as an indication that a ClientHello message has been
5821 received. As the flag value will be lost between multiple
5822 invocations of ssl3_accept when using non-blocking I/O, the
5823 function may not be aware that a handshake has actually taken
5824 place, thus preventing a new session from being added to the
5825 session cache.
5826
5827 To avoid this problem, we now set s->new_session to 2 instead of
5828 using a local variable.
5829 [Lutz Jaenicke, Bodo Moeller]
5830
5831 *) Bugfix: Return -1 from ssl3_get_server_done (ssl3/s3_clnt.c)
5832 if the SSL_R_LENGTH_MISMATCH error is detected.
5833 [Geoff Thorpe, Bodo Moeller]
5834
5835 *) New 'shared_ldflag' column in Configure platform table.
5836 [Richard Levitte]
5837
5838 *) Fix EVP_CIPHER_mode macro.
5839 ["Dan S. Camper" <dan@bti.net>]
5840
5841 *) Fix ssl3_read_bytes (ssl/s3_pkt.c): To ignore messages of unknown
5842 type, we must throw them away by setting rr->length to 0.
5843 [D P Chang <dpc@qualys.com>]
5844
5845 Changes between 0.9.6b and 0.9.6c [21 dec 2001]
5846
5847 *) Fix BN_rand_range bug pointed out by Dominikus Scherkl
5848 <Dominikus.Scherkl@biodata.com>. (The previous implementation
5849 worked incorrectly for those cases where range = 10..._2 and
5850 3*range is two bits longer than range.)
5851 [Bodo Moeller]
5852
5853 *) Only add signing time to PKCS7 structures if it is not already
5854 present.
5855 [Steve Henson]
5856
5857 *) Fix crypto/objects/objects.h: "ld-ce" should be "id-ce",
5858 OBJ_ld_ce should be OBJ_id_ce.
5859 Also some ip-pda OIDs in crypto/objects/objects.txt were
5860 incorrect (cf. RFC 3039).
5861 [Matt Cooper, Frederic Giudicelli, Bodo Moeller]
5862
5863 *) Release CRYPTO_LOCK_DYNLOCK when CRYPTO_destroy_dynlockid()
5864 returns early because it has nothing to do.
5865 [Andy Schneider <andy.schneider@bjss.co.uk>]
5866
5867 *) [In 0.9.6c-engine release:]
5868 Fix mutex callback return values in crypto/engine/hw_ncipher.c.
5869 [Andy Schneider <andy.schneider@bjss.co.uk>]
5870
5871 *) [In 0.9.6c-engine release:]
5872 Add support for Cryptographic Appliance's keyserver technology.
5873 (Use engine 'keyclient')
5874 [Cryptographic Appliances and Geoff Thorpe]
5875
5876 *) Add a configuration entry for OS/390 Unix. The C compiler 'c89'
5877 is called via tools/c89.sh because arguments have to be
5878 rearranged (all '-L' options must appear before the first object
5879 modules).
5880 [Richard Shapiro <rshapiro@abinitio.com>]
5881
5882 *) [In 0.9.6c-engine release:]
5883 Add support for Broadcom crypto accelerator cards, backported
5884 from 0.9.7.
5885 [Broadcom, Nalin Dahyabhai <nalin@redhat.com>, Mark Cox]
5886
5887 *) [In 0.9.6c-engine release:]
5888 Add support for SureWare crypto accelerator cards from
5889 Baltimore Technologies. (Use engine 'sureware')
5890 [Baltimore Technologies and Mark Cox]
5891
5892 *) [In 0.9.6c-engine release:]
5893 Add support for crypto accelerator cards from Accelerated
5894 Encryption Processing, www.aep.ie. (Use engine 'aep')
5895 [AEP Inc. and Mark Cox]
5896
5897 *) Add a configuration entry for gcc on UnixWare.
5898 [Gary Benson <gbenson@redhat.com>]
5899
5900 *) Change ssl/s2_clnt.c and ssl/s2_srvr.c so that received handshake
5901 messages are stored in a single piece (fixed-length part and
5902 variable-length part combined) and fix various bugs found on the way.
5903 [Bodo Moeller]
5904
5905 *) Disable caching in BIO_gethostbyname(), directly use gethostbyname()
5906 instead. BIO_gethostbyname() does not know what timeouts are
5907 appropriate, so entries would stay in cache even when they have
5908 become invalid.
5909 [Bodo Moeller; problem pointed out by Rich Salz <rsalz@zolera.com>
5910
5911 *) Change ssl23_get_client_hello (ssl/s23_srvr.c) behaviour when
5912 faced with a pathologically small ClientHello fragment that does
5913 not contain client_version: Instead of aborting with an error,
5914 simply choose the highest available protocol version (i.e.,
5915 TLS 1.0 unless it is disabled). In practice, ClientHello
5916 messages are never sent like this, but this change gives us
5917 strictly correct behaviour at least for TLS.
5918 [Bodo Moeller]
5919
5920 *) Fix SSL handshake functions and SSL_clear() such that SSL_clear()
5921 never resets s->method to s->ctx->method when called from within
5922 one of the SSL handshake functions.
5923 [Bodo Moeller; problem pointed out by Niko Baric]
5924
5925 *) In ssl3_get_client_hello (ssl/s3_srvr.c), generate a fatal alert
5926 (sent using the client's version number) if client_version is
5927 smaller than the protocol version in use. Also change
5928 ssl23_get_client_hello (ssl/s23_srvr.c) to select TLS 1.0 if
5929 the client demanded SSL 3.0 but only TLS 1.0 is enabled; then
5930 the client will at least see that alert.
5931 [Bodo Moeller]
5932
5933 *) Fix ssl3_get_message (ssl/s3_both.c) to handle message fragmentation
5934 correctly.
5935 [Bodo Moeller]
5936
5937 *) Avoid infinite loop in ssl3_get_message (ssl/s3_both.c) if a
5938 client receives HelloRequest while in a handshake.
5939 [Bodo Moeller; bug noticed by Andy Schneider <andy.schneider@bjss.co.uk>]
5940
5941 *) Bugfix in ssl3_accept (ssl/s3_srvr.c): Case SSL3_ST_SW_HELLO_REQ_C
5942 should end in 'break', not 'goto end' which circuments various
5943 cleanups done in state SSL_ST_OK. But session related stuff
5944 must be disabled for SSL_ST_OK in the case that we just sent a
5945 HelloRequest.
5946
5947 Also avoid some overhead by not calling ssl_init_wbio_buffer()
5948 before just sending a HelloRequest.
5949 [Bodo Moeller, Eric Rescorla <ekr@rtfm.com>]
5950
5951 *) Fix ssl/s3_enc.c, ssl/t1_enc.c and ssl/s3_pkt.c so that we don't
5952 reveal whether illegal block cipher padding was found or a MAC
5953 verification error occured. (Neither SSLerr() codes nor alerts
5954 are directly visible to potential attackers, but the information
5955 may leak via logfiles.)
5956
5957 Similar changes are not required for the SSL 2.0 implementation
5958 because the number of padding bytes is sent in clear for SSL 2.0,
5959 and the extra bytes are just ignored. However ssl/s2_pkt.c
5960 failed to verify that the purported number of padding bytes is in
5961 the legal range.
5962 [Bodo Moeller]
5963
5964 *) Add OpenUNIX-8 support including shared libraries
5965 (Boyd Lynn Gerber <gerberb@zenez.com>).
5966 [Lutz Jaenicke]
5967
5968 *) Improve RSA_padding_check_PKCS1_OAEP() check again to avoid
5969 'wristwatch attack' using huge encoding parameters (cf.
5970 James H. Manger's CRYPTO 2001 paper). Note that the
5971 RSA_PKCS1_OAEP_PADDING case of RSA_private_decrypt() does not use
5972 encoding parameters and hence was not vulnerable.
5973 [Bodo Moeller]
5974
5975 *) BN_sqr() bug fix.
5976 [Ulf Möller, reported by Jim Ellis <jim.ellis@cavium.com>]
5977
5978 *) Rabin-Miller test analyses assume uniformly distributed witnesses,
5979 so use BN_pseudo_rand_range() instead of using BN_pseudo_rand()
5980 followed by modular reduction.
5981 [Bodo Moeller; pointed out by Adam Young <AYoung1@NCSUS.JNJ.COM>]
5982
5983 *) Add BN_pseudo_rand_range() with obvious functionality: BN_rand_range()
5984 equivalent based on BN_pseudo_rand() instead of BN_rand().
5985 [Bodo Moeller]
5986
5987 *) s3_srvr.c: allow sending of large client certificate lists (> 16 kB).
5988 This function was broken, as the check for a new client hello message
5989 to handle SGC did not allow these large messages.
5990 (Tracked down by "Douglas E. Engert" <deengert@anl.gov>.)
5991 [Lutz Jaenicke]
5992
5993 *) Add alert descriptions for TLSv1 to SSL_alert_desc_string[_long]().
5994 [Lutz Jaenicke]
5995
5996 *) Fix buggy behaviour of BIO_get_num_renegotiates() and BIO_ctrl()
5997 for BIO_C_GET_WRITE_BUF_SIZE ("Stephen Hinton" <shinton@netopia.com>).
5998 [Lutz Jaenicke]
5999
6000 *) Rework the configuration and shared library support for Tru64 Unix.
6001 The configuration part makes use of modern compiler features and
6002 still retains old compiler behavior for those that run older versions
6003 of the OS. The shared library support part includes a variant that
6004 uses the RPATH feature, and is available through the special
6005 configuration target "alpha-cc-rpath", which will never be selected
6006 automatically.
6007 [Tim Mooney <mooney@dogbert.cc.ndsu.NoDak.edu> via Richard Levitte]
6008
6009 *) In ssl3_get_key_exchange (ssl/s3_clnt.c), call ssl3_get_message()
6010 with the same message size as in ssl3_get_certificate_request().
6011 Otherwise, if no ServerKeyExchange message occurs, CertificateRequest
6012 messages might inadvertently be reject as too long.
6013 [Petr Lampa <lampa@fee.vutbr.cz>]
6014
6015 *) Enhanced support for IA-64 Unix platforms (well, Linux and HP-UX).
6016 [Andy Polyakov]
6017
6018 *) Modified SSL library such that the verify_callback that has been set
6019 specificly for an SSL object with SSL_set_verify() is actually being
6020 used. Before the change, a verify_callback set with this function was
6021 ignored and the verify_callback() set in the SSL_CTX at the time of
6022 the call was used. New function X509_STORE_CTX_set_verify_cb() introduced
6023 to allow the necessary settings.
6024 [Lutz Jaenicke]
6025
6026 *) Initialize static variable in crypto/dsa/dsa_lib.c and crypto/dh/dh_lib.c
6027 explicitly to NULL, as at least on Solaris 8 this seems not always to be
6028 done automatically (in contradiction to the requirements of the C
6029 standard). This made problems when used from OpenSSH.
6030 [Lutz Jaenicke]
6031
6032 *) In OpenSSL 0.9.6a and 0.9.6b, crypto/dh/dh_key.c ignored
6033 dh->length and always used
6034
6035 BN_rand_range(priv_key, dh->p).
6036
6037 BN_rand_range() is not necessary for Diffie-Hellman, and this
6038 specific range makes Diffie-Hellman unnecessarily inefficient if
6039 dh->length (recommended exponent length) is much smaller than the
6040 length of dh->p. We could use BN_rand_range() if the order of
6041 the subgroup was stored in the DH structure, but we only have
6042 dh->length.
6043
6044 So switch back to
6045
6046 BN_rand(priv_key, l, ...)
6047
6048 where 'l' is dh->length if this is defined, or BN_num_bits(dh->p)-1
6049 otherwise.
6050 [Bodo Moeller]
6051
6052 *) In
6053
6054 RSA_eay_public_encrypt
6055 RSA_eay_private_decrypt
6056 RSA_eay_private_encrypt (signing)
6057 RSA_eay_public_decrypt (signature verification)
6058
6059 (default implementations for RSA_public_encrypt,
6060 RSA_private_decrypt, RSA_private_encrypt, RSA_public_decrypt),
6061 always reject numbers >= n.
6062 [Bodo Moeller]
6063
6064 *) In crypto/rand/md_rand.c, use a new short-time lock CRYPTO_LOCK_RAND2
6065 to synchronize access to 'locking_thread'. This is necessary on
6066 systems where access to 'locking_thread' (an 'unsigned long'
6067 variable) is not atomic.
6068 [Bodo Moeller]
6069
6070 *) In crypto/rand/md_rand.c, set 'locking_thread' to current thread's ID
6071 *before* setting the 'crypto_lock_rand' flag. The previous code had
6072 a race condition if 0 is a valid thread ID.
6073 [Travis Vitek <vitek@roguewave.com>]
6074
6075 *) Add support for shared libraries under Irix.
6076 [Albert Chin-A-Young <china@thewrittenword.com>]
6077
6078 *) Add configuration option to build on Linux on both big-endian and
6079 little-endian MIPS.
6080 [Ralf Baechle <ralf@uni-koblenz.de>]
6081
6082 *) Add the possibility to create shared libraries on HP-UX.
6083 [Richard Levitte]
6084
6085 Changes between 0.9.6a and 0.9.6b [9 Jul 2001]
6086
6087 *) Change ssleay_rand_bytes (crypto/rand/md_rand.c)
6088 to avoid a SSLeay/OpenSSL PRNG weakness pointed out by
6089 Markku-Juhani O. Saarinen <markku-juhani.saarinen@nokia.com>:
6090 PRNG state recovery was possible based on the output of
6091 one PRNG request appropriately sized to gain knowledge on
6092 'md' followed by enough consecutive 1-byte PRNG requests
6093 to traverse all of 'state'.
6094
6095 1. When updating 'md_local' (the current thread's copy of 'md')
6096 during PRNG output generation, hash all of the previous
6097 'md_local' value, not just the half used for PRNG output.
6098
6099 2. Make the number of bytes from 'state' included into the hash
6100 independent from the number of PRNG bytes requested.
6101
6102 The first measure alone would be sufficient to avoid
6103 Markku-Juhani's attack. (Actually it had never occurred
6104 to me that the half of 'md_local' used for chaining was the
6105 half from which PRNG output bytes were taken -- I had always
6106 assumed that the secret half would be used.) The second
6107 measure makes sure that additional data from 'state' is never
6108 mixed into 'md_local' in small portions; this heuristically
6109 further strengthens the PRNG.
6110 [Bodo Moeller]
6111
6112 *) Fix crypto/bn/asm/mips3.s.
6113 [Andy Polyakov]
6114
6115 *) When only the key is given to "enc", the IV is undefined. Print out
6116 an error message in this case.
6117 [Lutz Jaenicke]
6118
6119 *) Handle special case when X509_NAME is empty in X509 printing routines.
6120 [Steve Henson]
6121
6122 *) In dsa_do_verify (crypto/dsa/dsa_ossl.c), verify that r and s are
6123 positive and less than q.
6124 [Bodo Moeller]
6125
6126 *) Don't change *pointer in CRYPTO_add_lock() is add_lock_callback is
6127 used: it isn't thread safe and the add_lock_callback should handle
6128 that itself.
6129 [Paul Rose <Paul.Rose@bridge.com>]
6130
6131 *) Verify that incoming data obeys the block size in
6132 ssl3_enc (ssl/s3_enc.c) and tls1_enc (ssl/t1_enc.c).
6133 [Bodo Moeller]
6134
6135 *) Fix OAEP check.
6136 [Ulf Möller, Bodo Möller]
6137
6138 *) The countermeasure against Bleichbacher's attack on PKCS #1 v1.5
6139 RSA encryption was accidentally removed in s3_srvr.c in OpenSSL 0.9.5
6140 when fixing the server behaviour for backwards-compatible 'client
6141 hello' messages. (Note that the attack is impractical against
6142 SSL 3.0 and TLS 1.0 anyway because length and version checking
6143 means that the probability of guessing a valid ciphertext is
6144 around 2^-40; see section 5 in Bleichenbacher's CRYPTO '98
6145 paper.)
6146
6147 Before 0.9.5, the countermeasure (hide the error by generating a
6148 random 'decryption result') did not work properly because
6149 ERR_clear_error() was missing, meaning that SSL_get_error() would
6150 detect the supposedly ignored error.
6151
6152 Both problems are now fixed.
6153 [Bodo Moeller]
6154
6155 *) In crypto/bio/bf_buff.c, increase DEFAULT_BUFFER_SIZE to 4096
6156 (previously it was 1024).
6157 [Bodo Moeller]
6158
6159 *) Fix for compatibility mode trust settings: ignore trust settings
6160 unless some valid trust or reject settings are present.
6161 [Steve Henson]
6162
6163 *) Fix for blowfish EVP: its a variable length cipher.
6164 [Steve Henson]
6165
6166 *) Fix various bugs related to DSA S/MIME verification. Handle missing
6167 parameters in DSA public key structures and return an error in the
6168 DSA routines if parameters are absent.
6169 [Steve Henson]
6170
6171 *) In versions up to 0.9.6, RAND_file_name() resorted to file ".rnd"
6172 in the current directory if neither $RANDFILE nor $HOME was set.
6173 RAND_file_name() in 0.9.6a returned NULL in this case. This has
6174 caused some confusion to Windows users who haven't defined $HOME.
6175 Thus RAND_file_name() is changed again: e_os.h can define a
6176 DEFAULT_HOME, which will be used if $HOME is not set.
6177 For Windows, we use "C:"; on other platforms, we still require
6178 environment variables.
6179
6180 *) Move 'if (!initialized) RAND_poll()' into regions protected by
6181 CRYPTO_LOCK_RAND. This is not strictly necessary, but avoids
6182 having multiple threads call RAND_poll() concurrently.
6183 [Bodo Moeller]
6184
6185 *) In crypto/rand/md_rand.c, replace 'add_do_not_lock' flag by a
6186 combination of a flag and a thread ID variable.
6187 Otherwise while one thread is in ssleay_rand_bytes (which sets the
6188 flag), *other* threads can enter ssleay_add_bytes without obeying
6189 the CRYPTO_LOCK_RAND lock (and may even illegally release the lock
6190 that they do not hold after the first thread unsets add_do_not_lock).
6191 [Bodo Moeller]
6192
6193 *) Change bctest again: '-x' expressions are not available in all
6194 versions of 'test'.
6195 [Bodo Moeller]
6196
6197 Changes between 0.9.6 and 0.9.6a [5 Apr 2001]
6198
6199 *) Fix a couple of memory leaks in PKCS7_dataDecode()
6200 [Steve Henson, reported by Heyun Zheng <hzheng@atdsprint.com>]
6201
6202 *) Change Configure and Makefiles to provide EXE_EXT, which will contain
6203 the default extension for executables, if any. Also, make the perl
6204 scripts that use symlink() to test if it really exists and use "cp"
6205 if it doesn't. All this made OpenSSL compilable and installable in
6206 CygWin.
6207 [Richard Levitte]
6208
6209 *) Fix for asn1_GetSequence() for indefinite length constructed data.
6210 If SEQUENCE is length is indefinite just set c->slen to the total
6211 amount of data available.
6212 [Steve Henson, reported by shige@FreeBSD.org]
6213 [This change does not apply to 0.9.7.]
6214
6215 *) Change bctest to avoid here-documents inside command substitution
6216 (workaround for FreeBSD /bin/sh bug).
6217 For compatibility with Ultrix, avoid shell functions (introduced
6218 in the bctest version that searches along $PATH).
6219 [Bodo Moeller]
6220
6221 *) Rename 'des_encrypt' to 'des_encrypt1'. This avoids the clashes
6222 with des_encrypt() defined on some operating systems, like Solaris
6223 and UnixWare.
6224 [Richard Levitte]
6225
6226 *) Check the result of RSA-CRT (see D. Boneh, R. DeMillo, R. Lipton:
6227 On the Importance of Eliminating Errors in Cryptographic
6228 Computations, J. Cryptology 14 (2001) 2, 101-119,
6229 http://theory.stanford.edu/~dabo/papers/faults.ps.gz).
6230 [Ulf Moeller]
6231
6232 *) MIPS assembler BIGNUM division bug fix.
6233 [Andy Polyakov]
6234
6235 *) Disabled incorrect Alpha assembler code.
6236 [Richard Levitte]
6237
6238 *) Fix PKCS#7 decode routines so they correctly update the length
6239 after reading an EOC for the EXPLICIT tag.
6240 [Steve Henson]
6241 [This change does not apply to 0.9.7.]
6242
6243 *) Fix bug in PKCS#12 key generation routines. This was triggered
6244 if a 3DES key was generated with a 0 initial byte. Include
6245 PKCS12_BROKEN_KEYGEN compilation option to retain the old
6246 (but broken) behaviour.
6247 [Steve Henson]
6248
6249 *) Enhance bctest to search for a working bc along $PATH and print
6250 it when found.
6251 [Tim Rice <tim@multitalents.net> via Richard Levitte]
6252
6253 *) Fix memory leaks in err.c: free err_data string if necessary;
6254 don't write to the wrong index in ERR_set_error_data.
6255 [Bodo Moeller]
6256
6257 *) Implement ssl23_peek (analogous to ssl23_read), which previously
6258 did not exist.
6259 [Bodo Moeller]
6260
6261 *) Replace rdtsc with _emit statements for VC++ version 5.
6262 [Jeremy Cooper <jeremy@baymoo.org>]
6263
6264 *) Make it possible to reuse SSLv2 sessions.
6265 [Richard Levitte]
6266
6267 *) In copy_email() check for >= 0 as a return value for
6268 X509_NAME_get_index_by_NID() since 0 is a valid index.
6269 [Steve Henson reported by Massimiliano Pala <madwolf@opensca.org>]
6270
6271 *) Avoid coredump with unsupported or invalid public keys by checking if
6272 X509_get_pubkey() fails in PKCS7_verify(). Fix memory leak when
6273 PKCS7_verify() fails with non detached data.
6274 [Steve Henson]
6275
6276 *) Don't use getenv in library functions when run as setuid/setgid.
6277 New function OPENSSL_issetugid().
6278 [Ulf Moeller]
6279
6280 *) Avoid false positives in memory leak detection code (crypto/mem_dbg.c)
6281 due to incorrect handling of multi-threading:
6282
6283 1. Fix timing glitch in the MemCheck_off() portion of CRYPTO_mem_ctrl().
6284
6285 2. Fix logical glitch in is_MemCheck_on() aka CRYPTO_is_mem_check_on().
6286
6287 3. Count how many times MemCheck_off() has been called so that
6288 nested use can be treated correctly. This also avoids
6289 inband-signalling in the previous code (which relied on the
6290 assumption that thread ID 0 is impossible).
6291 [Bodo Moeller]
6292
6293 *) Add "-rand" option also to s_client and s_server.
6294 [Lutz Jaenicke]
6295
6296 *) Fix CPU detection on Irix 6.x.
6297 [Kurt Hockenbury <khockenb@stevens-tech.edu> and
6298 "Bruce W. Forsberg" <bruce.forsberg@baesystems.com>]
6299
6300 *) Fix X509_NAME bug which produced incorrect encoding if X509_NAME
6301 was empty.
6302 [Steve Henson]
6303 [This change does not apply to 0.9.7.]
6304
6305 *) Use the cached encoding of an X509_NAME structure rather than
6306 copying it. This is apparently the reason for the libsafe "errors"
6307 but the code is actually correct.
6308 [Steve Henson]
6309
6310 *) Add new function BN_rand_range(), and fix DSA_sign_setup() to prevent
6311 Bleichenbacher's DSA attack.
6312 Extend BN_[pseudo_]rand: As before, top=1 forces the highest two bits
6313 to be set and top=0 forces the highest bit to be set; top=-1 is new
6314 and leaves the highest bit random.
6315 [Ulf Moeller, Bodo Moeller]
6316
6317 *) In the NCONF_...-based implementations for CONF_... queries
6318 (crypto/conf/conf_lib.c), if the input LHASH is NULL, avoid using
6319 a temporary CONF structure with the data component set to NULL
6320 (which gives segmentation faults in lh_retrieve).
6321 Instead, use NULL for the CONF pointer in CONF_get_string and
6322 CONF_get_number (which may use environment variables) and directly
6323 return NULL from CONF_get_section.
6324 [Bodo Moeller]
6325
6326 *) Fix potential buffer overrun for EBCDIC.
6327 [Ulf Moeller]
6328
6329 *) Tolerate nonRepudiation as being valid for S/MIME signing and certSign
6330 keyUsage if basicConstraints absent for a CA.
6331 [Steve Henson]
6332
6333 *) Make SMIME_write_PKCS7() write mail header values with a format that
6334 is more generally accepted (no spaces before the semicolon), since
6335 some programs can't parse those values properly otherwise. Also make
6336 sure BIO's that break lines after each write do not create invalid
6337 headers.
6338 [Richard Levitte]
6339
6340 *) Make the CRL encoding routines work with empty SEQUENCE OF. The
6341 macros previously used would not encode an empty SEQUENCE OF
6342 and break the signature.
6343 [Steve Henson]
6344 [This change does not apply to 0.9.7.]
6345
6346 *) Zero the premaster secret after deriving the master secret in
6347 DH ciphersuites.
6348 [Steve Henson]
6349
6350 *) Add some EVP_add_digest_alias registrations (as found in
6351 OpenSSL_add_all_digests()) to SSL_library_init()
6352 aka OpenSSL_add_ssl_algorithms(). This provides improved
6353 compatibility with peers using X.509 certificates
6354 with unconventional AlgorithmIdentifier OIDs.
6355 [Bodo Moeller]
6356
6357 *) Fix for Irix with NO_ASM.
6358 ["Bruce W. Forsberg" <bruce.forsberg@baesystems.com>]
6359
6360 *) ./config script fixes.
6361 [Ulf Moeller, Richard Levitte]
6362
6363 *) Fix 'openssl passwd -1'.
6364 [Bodo Moeller]
6365
6366 *) Change PKCS12_key_gen_asc() so it can cope with non null
6367 terminated strings whose length is passed in the passlen
6368 parameter, for example from PEM callbacks. This was done
6369 by adding an extra length parameter to asc2uni().
6370 [Steve Henson, reported by <oddissey@samsung.co.kr>]
6371
6372 *) Fix C code generated by 'openssl dsaparam -C': If a BN_bin2bn
6373 call failed, free the DSA structure.
6374 [Bodo Moeller]
6375
6376 *) Fix to uni2asc() to cope with zero length Unicode strings.
6377 These are present in some PKCS#12 files.
6378 [Steve Henson]
6379
6380 *) Increase s2->wbuf allocation by one byte in ssl2_new (ssl/s2_lib.c).
6381 Otherwise do_ssl_write (ssl/s2_pkt.c) will write beyond buffer limits
6382 when writing a 32767 byte record.
6383 [Bodo Moeller; problem reported by Eric Day <eday@concentric.net>]
6384
6385 *) In RSA_eay_public_{en,ed}crypt and RSA_eay_mod_exp (rsa_eay.c),
6386 obtain lock CRYPTO_LOCK_RSA before setting rsa->_method_mod_{n,p,q}.
6387
6388 (RSA objects have a reference count access to which is protected
6389 by CRYPTO_LOCK_RSA [see rsa_lib.c, s3_srvr.c, ssl_cert.c, ssl_rsa.c],
6390 so they are meant to be shared between threads.)
6391 [Bodo Moeller, Geoff Thorpe; original patch submitted by
6392 "Reddie, Steven" <Steven.Reddie@ca.com>]
6393
6394 *) Fix a deadlock in CRYPTO_mem_leaks().
6395 [Bodo Moeller]
6396
6397 *) Use better test patterns in bntest.
6398 [Ulf Möller]
6399
6400 *) rand_win.c fix for Borland C.
6401 [Ulf Möller]
6402
6403 *) BN_rshift bugfix for n == 0.
6404 [Bodo Moeller]
6405
6406 *) Add a 'bctest' script that checks for some known 'bc' bugs
6407 so that 'make test' does not abort just because 'bc' is broken.
6408 [Bodo Moeller]
6409
6410 *) Store verify_result within SSL_SESSION also for client side to
6411 avoid potential security hole. (Re-used sessions on the client side
6412 always resulted in verify_result==X509_V_OK, not using the original
6413 result of the server certificate verification.)
6414 [Lutz Jaenicke]
6415
6416 *) Fix ssl3_pending: If the record in s->s3->rrec is not of type
6417 SSL3_RT_APPLICATION_DATA, return 0.
6418 Similarly, change ssl2_pending to return 0 if SSL_in_init(s) is true.
6419 [Bodo Moeller]
6420
6421 *) Fix SSL_peek:
6422 Both ssl2_peek and ssl3_peek, which were totally broken in earlier
6423 releases, have been re-implemented by renaming the previous
6424 implementations of ssl2_read and ssl3_read to ssl2_read_internal
6425 and ssl3_read_internal, respectively, and adding 'peek' parameters
6426 to them. The new ssl[23]_{read,peek} functions are calls to
6427 ssl[23]_read_internal with the 'peek' flag set appropriately.
6428 A 'peek' parameter has also been added to ssl3_read_bytes, which
6429 does the actual work for ssl3_read_internal.
6430 [Bodo Moeller]
6431
6432 *) Initialise "ex_data" member of RSA/DSA/DH structures prior to calling
6433 the method-specific "init()" handler. Also clean up ex_data after
6434 calling the method-specific "finish()" handler. Previously, this was
6435 happening the other way round.
6436 [Geoff Thorpe]
6437
6438 *) Increase BN_CTX_NUM (the number of BIGNUMs in a BN_CTX) to 16.
6439 The previous value, 12, was not always sufficient for BN_mod_exp().
6440 [Bodo Moeller]
6441
6442 *) Make sure that shared libraries get the internal name engine with
6443 the full version number and not just 0. This should mark the
6444 shared libraries as not backward compatible. Of course, this should
6445 be changed again when we can guarantee backward binary compatibility.
6446 [Richard Levitte]
6447
6448 *) Fix typo in get_cert_by_subject() in by_dir.c
6449 [Jean-Marc Desperrier <jean-marc.desperrier@certplus.com>]
6450
6451 *) Rework the system to generate shared libraries:
6452
6453 - Make note of the expected extension for the shared libraries and
6454 if there is a need for symbolic links from for example libcrypto.so.0
6455 to libcrypto.so.0.9.7. There is extended info in Configure for
6456 that.
6457
6458 - Make as few rebuilds of the shared libraries as possible.
6459
6460 - Still avoid linking the OpenSSL programs with the shared libraries.
6461
6462 - When installing, install the shared libraries separately from the
6463 static ones.
6464 [Richard Levitte]
6465
6466 *) Fix SSL_CTX_set_read_ahead macro to actually use its argument.
6467
6468 Copy SSL_CTX's read_ahead flag to SSL object directly in SSL_new
6469 and not in SSL_clear because the latter is also used by the
6470 accept/connect functions; previously, the settings made by
6471 SSL_set_read_ahead would be lost during the handshake.
6472 [Bodo Moeller; problems reported by Anders Gertz <gertz@epact.se>]
6473
6474 *) Correct util/mkdef.pl to be selective about disabled algorithms.
6475 Previously, it would create entries for disableed algorithms no
6476 matter what.
6477 [Richard Levitte]
6478
6479 *) Added several new manual pages for SSL_* function.
6480 [Lutz Jaenicke]
6481
6482 Changes between 0.9.5a and 0.9.6 [24 Sep 2000]
6483
6484 *) In ssl23_get_client_hello, generate an error message when faced
6485 with an initial SSL 3.0/TLS record that is too small to contain the
6486 first two bytes of the ClientHello message, i.e. client_version.
6487 (Note that this is a pathologic case that probably has never happened
6488 in real life.) The previous approach was to use the version number
6489 from the record header as a substitute; but our protocol choice
6490 should not depend on that one because it is not authenticated
6491 by the Finished messages.
6492 [Bodo Moeller]
6493
6494 *) More robust randomness gathering functions for Windows.
6495 [Jeffrey Altman <jaltman@columbia.edu>]
6496
6497 *) For compatibility reasons if the flag X509_V_FLAG_ISSUER_CHECK is
6498 not set then we don't setup the error code for issuer check errors
6499 to avoid possibly overwriting other errors which the callback does
6500 handle. If an application does set the flag then we assume it knows
6501 what it is doing and can handle the new informational codes
6502 appropriately.
6503 [Steve Henson]
6504
6505 *) Fix for a nasty bug in ASN1_TYPE handling. ASN1_TYPE is used for
6506 a general "ANY" type, as such it should be able to decode anything
6507 including tagged types. However it didn't check the class so it would
6508 wrongly interpret tagged types in the same way as their universal
6509 counterpart and unknown types were just rejected. Changed so that the
6510 tagged and unknown types are handled in the same way as a SEQUENCE:
6511 that is the encoding is stored intact. There is also a new type
6512 "V_ASN1_OTHER" which is used when the class is not universal, in this
6513 case we have no idea what the actual type is so we just lump them all
6514 together.
6515 [Steve Henson]
6516
6517 *) On VMS, stdout may very well lead to a file that is written to
6518 in a record-oriented fashion. That means that every write() will
6519 write a separate record, which will be read separately by the
6520 programs trying to read from it. This can be very confusing.
6521
6522 The solution is to put a BIO filter in the way that will buffer
6523 text until a linefeed is reached, and then write everything a
6524 line at a time, so every record written will be an actual line,
6525 not chunks of lines and not (usually doesn't happen, but I've
6526 seen it once) several lines in one record. BIO_f_linebuffer() is
6527 the answer.
6528
6529 Currently, it's a VMS-only method, because that's where it has
6530 been tested well enough.
6531 [Richard Levitte]
6532
6533 *) Remove 'optimized' squaring variant in BN_mod_mul_montgomery,
6534 it can return incorrect results.
6535 (Note: The buggy variant was not enabled in OpenSSL 0.9.5a,
6536 but it was in 0.9.6-beta[12].)
6537 [Bodo Moeller]
6538
6539 *) Disable the check for content being present when verifying detached
6540 signatures in pk7_smime.c. Some versions of Netscape (wrongly)
6541 include zero length content when signing messages.
6542 [Steve Henson]
6543
6544 *) New BIO_shutdown_wr macro, which invokes the BIO_C_SHUTDOWN_WR
6545 BIO_ctrl (for BIO pairs).
6546 [Bodo Möller]
6547
6548 *) Add DSO method for VMS.
6549 [Richard Levitte]
6550
6551 *) Bug fix: Montgomery multiplication could produce results with the
6552 wrong sign.
6553 [Ulf Möller]
6554
6555 *) Add RPM specification openssl.spec and modify it to build three
6556 packages. The default package contains applications, application
6557 documentation and run-time libraries. The devel package contains
6558 include files, static libraries and function documentation. The
6559 doc package contains the contents of the doc directory. The original
6560 openssl.spec was provided by Damien Miller <djm@mindrot.org>.
6561 [Richard Levitte]
6562
6563 *) Add a large number of documentation files for many SSL routines.
6564 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>]
6565
6566 *) Add a configuration entry for Sony News 4.
6567 [NAKAJI Hiroyuki <nakaji@tutrp.tut.ac.jp>]
6568
6569 *) Don't set the two most significant bits to one when generating a
6570 random number < q in the DSA library.
6571 [Ulf Möller]
6572
6573 *) New SSL API mode 'SSL_MODE_AUTO_RETRY'. This disables the default
6574 behaviour that SSL_read may result in SSL_ERROR_WANT_READ (even if
6575 the underlying transport is blocking) if a handshake took place.
6576 (The default behaviour is needed by applications such as s_client
6577 and s_server that use select() to determine when to use SSL_read;
6578 but for applications that know in advance when to expect data, it
6579 just makes things more complicated.)
6580 [Bodo Moeller]
6581
6582 *) Add RAND_egd_bytes(), which gives control over the number of bytes read
6583 from EGD.
6584 [Ben Laurie]
6585
6586 *) Add a few more EBCDIC conditionals that make `req' and `x509'
6587 work better on such systems.
6588 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>]
6589
6590 *) Add two demo programs for PKCS12_parse() and PKCS12_create().
6591 Update PKCS12_parse() so it copies the friendlyName and the
6592 keyid to the certificates aux info.
6593 [Steve Henson]
6594
6595 *) Fix bug in PKCS7_verify() which caused an infinite loop
6596 if there was more than one signature.
6597 [Sven Uszpelkat <su@celocom.de>]
6598
6599 *) Major change in util/mkdef.pl to include extra information
6600 about each symbol, as well as presentig variables as well
6601 as functions. This change means that there's n more need
6602 to rebuild the .num files when some algorithms are excluded.
6603 [Richard Levitte]
6604
6605 *) Allow the verify time to be set by an application,
6606 rather than always using the current time.
6607 [Steve Henson]
6608
6609 *) Phase 2 verify code reorganisation. The certificate
6610 verify code now looks up an issuer certificate by a
6611 number of criteria: subject name, authority key id
6612 and key usage. It also verifies self signed certificates
6613 by the same criteria. The main comparison function is
6614 X509_check_issued() which performs these checks.
6615
6616 Lot of changes were necessary in order to support this
6617 without completely rewriting the lookup code.
6618
6619 Authority and subject key identifier are now cached.
6620
6621 The LHASH 'certs' is X509_STORE has now been replaced
6622 by a STACK_OF(X509_OBJECT). This is mainly because an
6623 LHASH can't store or retrieve multiple objects with
6624 the same hash value.
6625
6626 As a result various functions (which were all internal
6627 use only) have changed to handle the new X509_STORE
6628 structure. This will break anything that messed round
6629 with X509_STORE internally.
6630
6631 The functions X509_STORE_add_cert() now checks for an
6632 exact match, rather than just subject name.
6633
6634 The X509_STORE API doesn't directly support the retrieval
6635 of multiple certificates matching a given criteria, however
6636 this can be worked round by performing a lookup first
6637 (which will fill the cache with candidate certificates)
6638 and then examining the cache for matches. This is probably
6639 the best we can do without throwing out X509_LOOKUP
6640 entirely (maybe later...).
6641
6642 The X509_VERIFY_CTX structure has been enhanced considerably.
6643
6644 All certificate lookup operations now go via a get_issuer()
6645 callback. Although this currently uses an X509_STORE it
6646 can be replaced by custom lookups. This is a simple way
6647 to bypass the X509_STORE hackery necessary to make this
6648 work and makes it possible to use more efficient techniques
6649 in future. A very simple version which uses a simple
6650 STACK for its trusted certificate store is also provided
6651 using X509_STORE_CTX_trusted_stack().
6652
6653 The verify_cb() and verify() callbacks now have equivalents
6654 in the X509_STORE_CTX structure.
6655
6656 X509_STORE_CTX also has a 'flags' field which can be used
6657 to customise the verify behaviour.
6658 [Steve Henson]
6659
6660 *) Add new PKCS#7 signing option PKCS7_NOSMIMECAP which
6661 excludes S/MIME capabilities.
6662 [Steve Henson]
6663
6664 *) When a certificate request is read in keep a copy of the
6665 original encoding of the signed data and use it when outputing
6666 again. Signatures then use the original encoding rather than
6667 a decoded, encoded version which may cause problems if the
6668 request is improperly encoded.
6669 [Steve Henson]
6670
6671 *) For consistency with other BIO_puts implementations, call
6672 buffer_write(b, ...) directly in buffer_puts instead of calling
6673 BIO_write(b, ...).
6674
6675 In BIO_puts, increment b->num_write as in BIO_write.
6676 [Peter.Sylvester@EdelWeb.fr]
6677
6678 *) Fix BN_mul_word for the case where the word is 0. (We have to use
6679 BN_zero, we may not return a BIGNUM with an array consisting of
6680 words set to zero.)
6681 [Bodo Moeller]
6682
6683 *) Avoid calling abort() from within the library when problems are
6684 detected, except if preprocessor symbols have been defined
6685 (such as REF_CHECK, BN_DEBUG etc.).
6686 [Bodo Moeller]
6687
6688 *) New openssl application 'rsautl'. This utility can be
6689 used for low level RSA operations. DER public key
6690 BIO/fp routines also added.
6691 [Steve Henson]
6692
6693 *) New Configure entry and patches for compiling on QNX 4.
6694 [Andreas Schneider <andreas@ds3.etech.fh-hamburg.de>]
6695
6696 *) A demo state-machine implementation was sponsored by
6697 Nuron (http://www.nuron.com/) and is now available in
6698 demos/state_machine.
6699 [Ben Laurie]
6700
6701 *) New options added to the 'dgst' utility for signature
6702 generation and verification.
6703 [Steve Henson]
6704
6705 *) Unrecognized PKCS#7 content types are now handled via a
6706 catch all ASN1_TYPE structure. This allows unsupported
6707 types to be stored as a "blob" and an application can
6708 encode and decode it manually.
6709 [Steve Henson]
6710
6711 *) Fix various signed/unsigned issues to make a_strex.c
6712 compile under VC++.
6713 [Oscar Jacobsson <oscar.jacobsson@celocom.com>]
6714
6715 *) ASN1 fixes. i2d_ASN1_OBJECT was not returning the correct
6716 length if passed a buffer. ASN1_INTEGER_to_BN failed
6717 if passed a NULL BN and its argument was negative.
6718 [Steve Henson, pointed out by Sven Heiberg <sven@tartu.cyber.ee>]
6719
6720 *) Modification to PKCS#7 encoding routines to output definite
6721 length encoding. Since currently the whole structures are in
6722 memory there's not real point in using indefinite length
6723 constructed encoding. However if OpenSSL is compiled with
6724 the flag PKCS7_INDEFINITE_ENCODING the old form is used.
6725 [Steve Henson]
6726
6727 *) Added BIO_vprintf() and BIO_vsnprintf().
6728 [Richard Levitte]
6729
6730 *) Added more prefixes to parse for in the the strings written
6731 through a logging bio, to cover all the levels that are available
6732 through syslog. The prefixes are now:
6733
6734 PANIC, EMERG, EMR => LOG_EMERG
6735 ALERT, ALR => LOG_ALERT
6736 CRIT, CRI => LOG_CRIT
6737 ERROR, ERR => LOG_ERR
6738 WARNING, WARN, WAR => LOG_WARNING
6739 NOTICE, NOTE, NOT => LOG_NOTICE
6740 INFO, INF => LOG_INFO
6741 DEBUG, DBG => LOG_DEBUG
6742
6743 and as before, if none of those prefixes are present at the
6744 beginning of the string, LOG_ERR is chosen.
6745
6746 On Win32, the LOG_* levels are mapped according to this:
6747
6748 LOG_EMERG, LOG_ALERT, LOG_CRIT, LOG_ERR => EVENTLOG_ERROR_TYPE
6749 LOG_WARNING => EVENTLOG_WARNING_TYPE
6750 LOG_NOTICE, LOG_INFO, LOG_DEBUG => EVENTLOG_INFORMATION_TYPE
6751
6752 [Richard Levitte]
6753
6754 *) Made it possible to reconfigure with just the configuration
6755 argument "reconf" or "reconfigure". The command line arguments
6756 are stored in Makefile.ssl in the variable CONFIGURE_ARGS,
6757 and are retrieved from there when reconfiguring.
6758 [Richard Levitte]
6759
6760 *) MD4 implemented.
6761 [Assar Westerlund <assar@sics.se>, Richard Levitte]
6762
6763 *) Add the arguments -CAfile and -CApath to the pkcs12 utility.
6764 [Richard Levitte]
6765
6766 *) The obj_dat.pl script was messing up the sorting of object
6767 names. The reason was that it compared the quoted version
6768 of strings as a result "OCSP" > "OCSP Signing" because
6769 " > SPACE. Changed script to store unquoted versions of
6770 names and add quotes on output. It was also omitting some
6771 names from the lookup table if they were given a default
6772 value (that is if SN is missing it is given the same
6773 value as LN and vice versa), these are now added on the
6774 grounds that if an object has a name we should be able to
6775 look it up. Finally added warning output when duplicate
6776 short or long names are found.
6777 [Steve Henson]
6778
6779 *) Changes needed for Tandem NSK.
6780 [Scott Uroff <scott@xypro.com>]
6781
6782 *) Fix SSL 2.0 rollback checking: Due to an off-by-one error in
6783 RSA_padding_check_SSLv23(), special padding was never detected
6784 and thus the SSL 3.0/TLS 1.0 countermeasure against protocol
6785 version rollback attacks was not effective.
6786
6787 In s23_clnt.c, don't use special rollback-attack detection padding
6788 (RSA_SSLV23_PADDING) if SSL 2.0 is the only protocol enabled in the
6789 client; similarly, in s23_srvr.c, don't do the rollback check if
6790 SSL 2.0 is the only protocol enabled in the server.
6791 [Bodo Moeller]
6792
6793 *) Make it possible to get hexdumps of unprintable data with 'openssl
6794 asn1parse'. By implication, the functions ASN1_parse_dump() and
6795 BIO_dump_indent() are added.
6796 [Richard Levitte]
6797
6798 *) New functions ASN1_STRING_print_ex() and X509_NAME_print_ex()
6799 these print out strings and name structures based on various
6800 flags including RFC2253 support and proper handling of
6801 multibyte characters. Added options to the 'x509' utility
6802 to allow the various flags to be set.
6803 [Steve Henson]
6804
6805 *) Various fixes to use ASN1_TIME instead of ASN1_UTCTIME.
6806 Also change the functions X509_cmp_current_time() and
6807 X509_gmtime_adj() work with an ASN1_TIME structure,
6808 this will enable certificates using GeneralizedTime in validity
6809 dates to be checked.
6810 [Steve Henson]
6811
6812 *) Make the NEG_PUBKEY_BUG code (which tolerates invalid
6813 negative public key encodings) on by default,
6814 NO_NEG_PUBKEY_BUG can be set to disable it.
6815 [Steve Henson]
6816
6817 *) New function c2i_ASN1_OBJECT() which acts on ASN1_OBJECT
6818 content octets. An i2c_ASN1_OBJECT is unnecessary because
6819 the encoding can be trivially obtained from the structure.
6820 [Steve Henson]
6821
6822 *) crypto/err.c locking bugfix: Use write locks (CRYPTO_w_[un]lock),
6823 not read locks (CRYPTO_r_[un]lock).
6824 [Bodo Moeller]
6825
6826 *) A first attempt at creating official support for shared
6827 libraries through configuration. I've kept it so the
6828 default is static libraries only, and the OpenSSL programs
6829 are always statically linked for now, but there are
6830 preparations for dynamic linking in place.
6831 This has been tested on Linux and Tru64.
6832 [Richard Levitte]
6833
6834 *) Randomness polling function for Win9x, as described in:
6835 Peter Gutmann, Software Generation of Practically Strong
6836 Random Numbers.
6837 [Ulf Möller]
6838
6839 *) Fix so PRNG is seeded in req if using an already existing
6840 DSA key.
6841 [Steve Henson]
6842
6843 *) New options to smime application. -inform and -outform
6844 allow alternative formats for the S/MIME message including
6845 PEM and DER. The -content option allows the content to be
6846 specified separately. This should allow things like Netscape
6847 form signing output easier to verify.
6848 [Steve Henson]
6849
6850 *) Fix the ASN1 encoding of tags using the 'long form'.
6851 [Steve Henson]
6852
6853 *) New ASN1 functions, i2c_* and c2i_* for INTEGER and BIT
6854 STRING types. These convert content octets to and from the
6855 underlying type. The actual tag and length octets are
6856 already assumed to have been read in and checked. These
6857 are needed because all other string types have virtually
6858 identical handling apart from the tag. By having versions
6859 of the ASN1 functions that just operate on content octets
6860 IMPLICIT tagging can be handled properly. It also allows
6861 the ASN1_ENUMERATED code to be cut down because ASN1_ENUMERATED
6862 and ASN1_INTEGER are identical apart from the tag.
6863 [Steve Henson]
6864
6865 *) Change the handling of OID objects as follows:
6866
6867 - New object identifiers are inserted in objects.txt, following
6868 the syntax given in objects.README.
6869 - objects.pl is used to process obj_mac.num and create a new
6870 obj_mac.h.
6871 - obj_dat.pl is used to create a new obj_dat.h, using the data in
6872 obj_mac.h.
6873
6874 This is currently kind of a hack, and the perl code in objects.pl
6875 isn't very elegant, but it works as I intended. The simplest way
6876 to check that it worked correctly is to look in obj_dat.h and
6877 check the array nid_objs and make sure the objects haven't moved
6878 around (this is important!). Additions are OK, as well as
6879 consistent name changes.
6880 [Richard Levitte]
6881
6882 *) Add BSD-style MD5-based passwords to 'openssl passwd' (option '-1').
6883 [Bodo Moeller]
6884
6885 *) Addition of the command line parameter '-rand file' to 'openssl req'.
6886 The given file adds to whatever has already been seeded into the
6887 random pool through the RANDFILE configuration file option or
6888 environment variable, or the default random state file.
6889 [Richard Levitte]
6890
6891 *) mkstack.pl now sorts each macro group into lexical order.
6892 Previously the output order depended on the order the files
6893 appeared in the directory, resulting in needless rewriting
6894 of safestack.h .
6895 [Steve Henson]
6896
6897 *) Patches to make OpenSSL compile under Win32 again. Mostly
6898 work arounds for the VC++ problem that it treats func() as
6899 func(void). Also stripped out the parts of mkdef.pl that
6900 added extra typesafe functions: these no longer exist.
6901 [Steve Henson]
6902
6903 *) Reorganisation of the stack code. The macros are now all
6904 collected in safestack.h . Each macro is defined in terms of
6905 a "stack macro" of the form SKM_<name>(type, a, b). The
6906 DEBUG_SAFESTACK is now handled in terms of function casts,
6907 this has the advantage of retaining type safety without the
6908 use of additional functions. If DEBUG_SAFESTACK is not defined
6909 then the non typesafe macros are used instead. Also modified the
6910 mkstack.pl script to handle the new form. Needs testing to see
6911 if which (if any) compilers it chokes and maybe make DEBUG_SAFESTACK
6912 the default if no major problems. Similar behaviour for ASN1_SET_OF
6913 and PKCS12_STACK_OF.
6914 [Steve Henson]
6915
6916 *) When some versions of IIS use the 'NET' form of private key the
6917 key derivation algorithm is different. Normally MD5(password) is
6918 used as a 128 bit RC4 key. In the modified case
6919 MD5(MD5(password) + "SGCKEYSALT") is used insted. Added some
6920 new functions i2d_RSA_NET(), d2i_RSA_NET() etc which are the same
6921 as the old Netscape_RSA functions except they have an additional
6922 'sgckey' parameter which uses the modified algorithm. Also added
6923 an -sgckey command line option to the rsa utility. Thanks to
6924 Adrian Peck <bertie@ncipher.com> for posting details of the modified
6925 algorithm to openssl-dev.
6926 [Steve Henson]
6927
6928 *) The evp_local.h macros were using 'c.##kname' which resulted in
6929 invalid expansion on some systems (SCO 5.0.5 for example).
6930 Corrected to 'c.kname'.
6931 [Phillip Porch <root@theporch.com>]
6932
6933 *) New X509_get1_email() and X509_REQ_get1_email() functions that return
6934 a STACK of email addresses from a certificate or request, these look
6935 in the subject name and the subject alternative name extensions and
6936 omit any duplicate addresses.
6937 [Steve Henson]
6938
6939 *) Re-implement BN_mod_exp2_mont using independent (and larger) windows.
6940 This makes DSA verification about 2 % faster.
6941 [Bodo Moeller]
6942
6943 *) Increase maximum window size in BN_mod_exp_... to 6 bits instead of 5
6944 (meaning that now 2^5 values will be precomputed, which is only 4 KB
6945 plus overhead for 1024 bit moduli).
6946 This makes exponentiations about 0.5 % faster for 1024 bit
6947 exponents (as measured by "openssl speed rsa2048").
6948 [Bodo Moeller]
6949
6950 *) Rename memory handling macros to avoid conflicts with other
6951 software:
6952 Malloc => OPENSSL_malloc
6953 Malloc_locked => OPENSSL_malloc_locked
6954 Realloc => OPENSSL_realloc
6955 Free => OPENSSL_free
6956 [Richard Levitte]
6957
6958 *) New function BN_mod_exp_mont_word for small bases (roughly 15%
6959 faster than BN_mod_exp_mont, i.e. 7% for a full DH exchange).
6960 [Bodo Moeller]
6961
6962 *) CygWin32 support.
6963 [John Jarvie <jjarvie@newsguy.com>]
6964
6965 *) The type-safe stack code has been rejigged. It is now only compiled
6966 in when OpenSSL is configured with the DEBUG_SAFESTACK option and
6967 by default all type-specific stack functions are "#define"d back to
6968 standard stack functions. This results in more streamlined output
6969 but retains the type-safety checking possibilities of the original
6970 approach.
6971 [Geoff Thorpe]
6972
6973 *) The STACK code has been cleaned up, and certain type declarations
6974 that didn't make a lot of sense have been brought in line. This has
6975 also involved a cleanup of sorts in safestack.h to more correctly
6976 map type-safe stack functions onto their plain stack counterparts.
6977 This work has also resulted in a variety of "const"ifications of
6978 lots of the code, especially "_cmp" operations which should normally
6979 be prototyped with "const" parameters anyway.
6980 [Geoff Thorpe]
6981
6982 *) When generating bytes for the first time in md_rand.c, 'stir the pool'
6983 by seeding with STATE_SIZE dummy bytes (with zero entropy count).
6984 (The PRNG state consists of two parts, the large pool 'state' and 'md',
6985 where all of 'md' is used each time the PRNG is used, but 'state'
6986 is used only indexed by a cyclic counter. As entropy may not be
6987 well distributed from the beginning, 'md' is important as a
6988 chaining variable. However, the output function chains only half
6989 of 'md', i.e. 80 bits. ssleay_rand_add, on the other hand, chains
6990 all of 'md', and seeding with STATE_SIZE dummy bytes will result
6991 in all of 'state' being rewritten, with the new values depending
6992 on virtually all of 'md'. This overcomes the 80 bit limitation.)
6993 [Bodo Moeller]
6994
6995 *) In ssl/s2_clnt.c and ssl/s3_clnt.c, call ERR_clear_error() when
6996 the handshake is continued after ssl_verify_cert_chain();
6997 otherwise, if SSL_VERIFY_NONE is set, remaining error codes
6998 can lead to 'unexplainable' connection aborts later.
6999 [Bodo Moeller; problem tracked down by Lutz Jaenicke]
7000
7001 *) Major EVP API cipher revision.
7002 Add hooks for extra EVP features. This allows various cipher
7003 parameters to be set in the EVP interface. Support added for variable
7004 key length ciphers via the EVP_CIPHER_CTX_set_key_length() function and
7005 setting of RC2 and RC5 parameters.
7006
7007 Modify EVP_OpenInit() and EVP_SealInit() to cope with variable key length
7008 ciphers.
7009
7010 Remove lots of duplicated code from the EVP library. For example *every*
7011 cipher init() function handles the 'iv' in the same way according to the
7012 cipher mode. They also all do nothing if the 'key' parameter is NULL and
7013 for CFB and OFB modes they zero ctx->num.
7014
7015 New functionality allows removal of S/MIME code RC2 hack.
7016
7017 Most of the routines have the same form and so can be declared in terms
7018 of macros.
7019
7020 By shifting this to the top level EVP_CipherInit() it can be removed from
7021 all individual ciphers. If the cipher wants to handle IVs or keys
7022 differently it can set the EVP_CIPH_CUSTOM_IV or EVP_CIPH_ALWAYS_CALL_INIT
7023 flags.
7024
7025 Change lots of functions like EVP_EncryptUpdate() to now return a
7026 value: although software versions of the algorithms cannot fail
7027 any installed hardware versions can.
7028 [Steve Henson]
7029
7030 *) Implement SSL_OP_TLS_ROLLBACK_BUG: In ssl3_get_client_key_exchange, if
7031 this option is set, tolerate broken clients that send the negotiated
7032 protocol version number instead of the requested protocol version
7033 number.
7034 [Bodo Moeller]
7035
7036 *) Call dh_tmp_cb (set by ..._TMP_DH_CB) with correct 'is_export' flag;
7037 i.e. non-zero for export ciphersuites, zero otherwise.
7038 Previous versions had this flag inverted, inconsistent with
7039 rsa_tmp_cb (..._TMP_RSA_CB).
7040 [Bodo Moeller; problem reported by Amit Chopra]
7041
7042 *) Add missing DSA library text string. Work around for some IIS
7043 key files with invalid SEQUENCE encoding.
7044 [Steve Henson]
7045
7046 *) Add a document (doc/standards.txt) that list all kinds of standards
7047 and so on that are implemented in OpenSSL.
7048 [Richard Levitte]
7049
7050 *) Enhance c_rehash script. Old version would mishandle certificates
7051 with the same subject name hash and wouldn't handle CRLs at all.
7052 Added -fingerprint option to crl utility, to support new c_rehash
7053 features.
7054 [Steve Henson]
7055
7056 *) Eliminate non-ANSI declarations in crypto.h and stack.h.
7057 [Ulf Möller]
7058
7059 *) Fix for SSL server purpose checking. Server checking was
7060 rejecting certificates which had extended key usage present
7061 but no ssl client purpose.
7062 [Steve Henson, reported by Rene Grosser <grosser@hisolutions.com>]
7063
7064 *) Make PKCS#12 code work with no password. The PKCS#12 spec
7065 is a little unclear about how a blank password is handled.
7066 Since the password in encoded as a BMPString with terminating
7067 double NULL a zero length password would end up as just the
7068 double NULL. However no password at all is different and is
7069 handled differently in the PKCS#12 key generation code. NS
7070 treats a blank password as zero length. MSIE treats it as no
7071 password on export: but it will try both on import. We now do
7072 the same: PKCS12_parse() tries zero length and no password if
7073 the password is set to "" or NULL (NULL is now a valid password:
7074 it wasn't before) as does the pkcs12 application.
7075 [Steve Henson]
7076
7077 *) Bugfixes in apps/x509.c: Avoid a memory leak; and don't use
7078 perror when PEM_read_bio_X509_REQ fails, the error message must
7079 be obtained from the error queue.
7080 [Bodo Moeller]
7081
7082 *) Avoid 'thread_hash' memory leak in crypto/err/err.c by freeing
7083 it in ERR_remove_state if appropriate, and change ERR_get_state
7084 accordingly to avoid race conditions (this is necessary because
7085 thread_hash is no longer constant once set).
7086 [Bodo Moeller]
7087
7088 *) Bugfix for linux-elf makefile.one.
7089 [Ulf Möller]
7090
7091 *) RSA_get_default_method() will now cause a default
7092 RSA_METHOD to be chosen if one doesn't exist already.
7093 Previously this was only set during a call to RSA_new()
7094 or RSA_new_method(NULL) meaning it was possible for
7095 RSA_get_default_method() to return NULL.
7096 [Geoff Thorpe]
7097
7098 *) Added native name translation to the existing DSO code
7099 that will convert (if the flag to do so is set) filenames
7100 that are sufficiently small and have no path information
7101 into a canonical native form. Eg. "blah" converted to
7102 "libblah.so" or "blah.dll" etc.
7103 [Geoff Thorpe]
7104
7105 *) New function ERR_error_string_n(e, buf, len) which is like
7106 ERR_error_string(e, buf), but writes at most 'len' bytes
7107 including the 0 terminator. For ERR_error_string_n, 'buf'
7108 may not be NULL.
7109 [Damien Miller <djm@mindrot.org>, Bodo Moeller]
7110
7111 *) CONF library reworked to become more general. A new CONF
7112 configuration file reader "class" is implemented as well as a
7113 new functions (NCONF_*, for "New CONF") to handle it. The now
7114 old CONF_* functions are still there, but are reimplemented to
7115 work in terms of the new functions. Also, a set of functions
7116 to handle the internal storage of the configuration data is
7117 provided to make it easier to write new configuration file
7118 reader "classes" (I can definitely see something reading a
7119 configuration file in XML format, for example), called _CONF_*,
7120 or "the configuration storage API"...
7121
7122 The new configuration file reading functions are:
7123
7124 NCONF_new, NCONF_free, NCONF_load, NCONF_load_fp, NCONF_load_bio,
7125 NCONF_get_section, NCONF_get_string, NCONF_get_numbre
7126
7127 NCONF_default, NCONF_WIN32
7128
7129 NCONF_dump_fp, NCONF_dump_bio
7130
7131 NCONF_default and NCONF_WIN32 are method (or "class") choosers,
7132 NCONF_new creates a new CONF object. This works in the same way
7133 as other interfaces in OpenSSL, like the BIO interface.
7134 NCONF_dump_* dump the internal storage of the configuration file,
7135 which is useful for debugging. All other functions take the same
7136 arguments as the old CONF_* functions wth the exception of the
7137 first that must be a `CONF *' instead of a `LHASH *'.
7138
7139 To make it easer to use the new classes with the old CONF_* functions,
7140 the function CONF_set_default_method is provided.
7141 [Richard Levitte]
7142
7143 *) Add '-tls1' option to 'openssl ciphers', which was already
7144 mentioned in the documentation but had not been implemented.
7145 (This option is not yet really useful because even the additional
7146 experimental TLS 1.0 ciphers are currently treated as SSL 3.0 ciphers.)
7147 [Bodo Moeller]
7148
7149 *) Initial DSO code added into libcrypto for letting OpenSSL (and
7150 OpenSSL-based applications) load shared libraries and bind to
7151 them in a portable way.
7152 [Geoff Thorpe, with contributions from Richard Levitte]
7153
7154 Changes between 0.9.5 and 0.9.5a [1 Apr 2000]
7155
7156 *) Make sure _lrotl and _lrotr are only used with MSVC.
7157
7158 *) Use lock CRYPTO_LOCK_RAND correctly in ssleay_rand_status
7159 (the default implementation of RAND_status).
7160
7161 *) Rename openssl x509 option '-crlext', which was added in 0.9.5,
7162 to '-clrext' (= clear extensions), as intended and documented.
7163 [Bodo Moeller; inconsistency pointed out by Michael Attili
7164 <attili@amaxo.com>]
7165
7166 *) Fix for HMAC. It wasn't zeroing the rest of the block if the key length
7167 was larger than the MD block size.
7168 [Steve Henson, pointed out by Yost William <YostW@tce.com>]
7169
7170 *) Modernise PKCS12_parse() so it uses STACK_OF(X509) for its ca argument
7171 fix a leak when the ca argument was passed as NULL. Stop X509_PUBKEY_set()
7172 using the passed key: if the passed key was a private key the result
7173 of X509_print(), for example, would be to print out all the private key
7174 components.
7175 [Steve Henson]
7176
7177 *) des_quad_cksum() byte order bug fix.
7178 [Ulf Möller, using the problem description in krb4-0.9.7, where
7179 the solution is attributed to Derrick J Brashear <shadow@DEMENTIA.ORG>]
7180
7181 *) Fix so V_ASN1_APP_CHOOSE works again: however its use is strongly
7182 discouraged.
7183 [Steve Henson, pointed out by Brian Korver <briank@cs.stanford.edu>]
7184
7185 *) For easily testing in shell scripts whether some command
7186 'openssl XXX' exists, the new pseudo-command 'openssl no-XXX'
7187 returns with exit code 0 iff no command of the given name is available.
7188 'no-XXX' is printed in this case, 'XXX' otherwise. In both cases,
7189 the output goes to stdout and nothing is printed to stderr.
7190 Additional arguments are always ignored.
7191
7192 Since for each cipher there is a command of the same name,
7193 the 'no-cipher' compilation switches can be tested this way.
7194
7195 ('openssl no-XXX' is not able to detect pseudo-commands such
7196 as 'quit', 'list-XXX-commands', or 'no-XXX' itself.)
7197 [Bodo Moeller]
7198
7199 *) Update test suite so that 'make test' succeeds in 'no-rsa' configuration.
7200 [Bodo Moeller]
7201
7202 *) For SSL_[CTX_]set_tmp_dh, don't create a DH key if SSL_OP_SINGLE_DH_USE
7203 is set; it will be thrown away anyway because each handshake creates
7204 its own key.
7205 ssl_cert_dup, which is used by SSL_new, now copies DH keys in addition
7206 to parameters -- in previous versions (since OpenSSL 0.9.3) the
7207 'default key' from SSL_CTX_set_tmp_dh would always be lost, meanining
7208 you effectivly got SSL_OP_SINGLE_DH_USE when using this macro.
7209 [Bodo Moeller]
7210
7211 *) New s_client option -ign_eof: EOF at stdin is ignored, and
7212 'Q' and 'R' lose their special meanings (quit/renegotiate).
7213 This is part of what -quiet does; unlike -quiet, -ign_eof
7214 does not suppress any output.
7215 [Richard Levitte]
7216
7217 *) Add compatibility options to the purpose and trust code. The
7218 purpose X509_PURPOSE_ANY is "any purpose" which automatically
7219 accepts a certificate or CA, this was the previous behaviour,
7220 with all the associated security issues.
7221
7222 X509_TRUST_COMPAT is the old trust behaviour: only and
7223 automatically trust self signed roots in certificate store. A
7224 new trust setting X509_TRUST_DEFAULT is used to specify that
7225 a purpose has no associated trust setting and it should instead
7226 use the value in the default purpose.
7227 [Steve Henson]
7228
7229 *) Fix the PKCS#8 DSA private key code so it decodes keys again
7230 and fix a memory leak.
7231 [Steve Henson]
7232
7233 *) In util/mkerr.pl (which implements 'make errors'), preserve
7234 reason strings from the previous version of the .c file, as
7235 the default to have only downcase letters (and digits) in
7236 automatically generated reasons codes is not always appropriate.
7237 [Bodo Moeller]
7238
7239 *) In ERR_load_ERR_strings(), build an ERR_LIB_SYS error reason table
7240 using strerror. Previously, ERR_reason_error_string() returned
7241 library names as reason strings for SYSerr; but SYSerr is a special
7242 case where small numbers are errno values, not library numbers.
7243 [Bodo Moeller]
7244
7245 *) Add '-dsaparam' option to 'openssl dhparam' application. This
7246 converts DSA parameters into DH parameters. (When creating parameters,
7247 DSA_generate_parameters is used.)
7248 [Bodo Moeller]
7249
7250 *) Include 'length' (recommended exponent length) in C code generated
7251 by 'openssl dhparam -C'.
7252 [Bodo Moeller]
7253
7254 *) The second argument to set_label in perlasm was already being used
7255 so couldn't be used as a "file scope" flag. Moved to third argument
7256 which was free.
7257 [Steve Henson]
7258
7259 *) In PEM_ASN1_write_bio and some other functions, use RAND_pseudo_bytes
7260 instead of RAND_bytes for encryption IVs and salts.
7261 [Bodo Moeller]
7262
7263 *) Include RAND_status() into RAND_METHOD instead of implementing
7264 it only for md_rand.c Otherwise replacing the PRNG by calling
7265 RAND_set_rand_method would be impossible.
7266 [Bodo Moeller]
7267
7268 *) Don't let DSA_generate_key() enter an infinite loop if the random
7269 number generation fails.
7270 [Bodo Moeller]
7271
7272 *) New 'rand' application for creating pseudo-random output.
7273 [Bodo Moeller]
7274
7275 *) Added configuration support for Linux/IA64
7276 [Rolf Haberrecker <rolf@suse.de>]
7277
7278 *) Assembler module support for Mingw32.
7279 [Ulf Möller]
7280
7281 *) Shared library support for HPUX (in shlib/).
7282 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE> and Anonymous]
7283
7284 *) Shared library support for Solaris gcc.
7285 [Lutz Behnke <behnke@trustcenter.de>]
7286
7287 Changes between 0.9.4 and 0.9.5 [28 Feb 2000]
7288
7289 *) PKCS7_encrypt() was adding text MIME headers twice because they
7290 were added manually and by SMIME_crlf_copy().
7291 [Steve Henson]
7292
7293 *) In bntest.c don't call BN_rand with zero bits argument.
7294 [Steve Henson, pointed out by Andrew W. Gray <agray@iconsinc.com>]
7295
7296 *) BN_mul bugfix: In bn_mul_part_recursion() only the a>a[n] && b>b[n]
7297 case was implemented. This caused BN_div_recp() to fail occasionally.
7298 [Ulf Möller]
7299
7300 *) Add an optional second argument to the set_label() in the perl
7301 assembly language builder. If this argument exists and is set
7302 to 1 it signals that the assembler should use a symbol whose
7303 scope is the entire file, not just the current function. This
7304 is needed with MASM which uses the format label:: for this scope.
7305 [Steve Henson, pointed out by Peter Runestig <peter@runestig.com>]
7306
7307 *) Change the ASN1 types so they are typedefs by default. Before
7308 almost all types were #define'd to ASN1_STRING which was causing
7309 STACK_OF() problems: you couldn't declare STACK_OF(ASN1_UTF8STRING)
7310 for example.
7311 [Steve Henson]
7312
7313 *) Change names of new functions to the new get1/get0 naming
7314 convention: After 'get1', the caller owns a reference count
7315 and has to call ..._free; 'get0' returns a pointer to some
7316 data structure without incrementing reference counters.
7317 (Some of the existing 'get' functions increment a reference
7318 counter, some don't.)
7319 Similarly, 'set1' and 'add1' functions increase reference
7320 counters or duplicate objects.
7321 [Steve Henson]
7322
7323 *) Allow for the possibility of temp RSA key generation failure:
7324 the code used to assume it always worked and crashed on failure.
7325 [Steve Henson]
7326
7327 *) Fix potential buffer overrun problem in BIO_printf().
7328 [Ulf Möller, using public domain code by Patrick Powell; problem
7329 pointed out by David Sacerdote <das33@cornell.edu>]
7330
7331 *) Support EGD <http://www.lothar.com/tech/crypto/>. New functions
7332 RAND_egd() and RAND_status(). In the command line application,
7333 the EGD socket can be specified like a seed file using RANDFILE
7334 or -rand.
7335 [Ulf Möller]
7336
7337 *) Allow the string CERTIFICATE to be tolerated in PKCS#7 structures.
7338 Some CAs (e.g. Verisign) distribute certificates in this form.
7339 [Steve Henson]
7340
7341 *) Remove the SSL_ALLOW_ADH compile option and set the default cipher
7342 list to exclude them. This means that no special compilation option
7343 is needed to use anonymous DH: it just needs to be included in the
7344 cipher list.
7345 [Steve Henson]
7346
7347 *) Change the EVP_MD_CTX_type macro so its meaning consistent with
7348 EVP_MD_type. The old functionality is available in a new macro called
7349 EVP_MD_md(). Change code that uses it and update docs.
7350 [Steve Henson]
7351
7352 *) ..._ctrl functions now have corresponding ..._callback_ctrl functions
7353 where the 'void *' argument is replaced by a function pointer argument.
7354 Previously 'void *' was abused to point to functions, which works on
7355 many platforms, but is not correct. As these functions are usually
7356 called by macros defined in OpenSSL header files, most source code
7357 should work without changes.
7358 [Richard Levitte]
7359
7360 *) <openssl/opensslconf.h> (which is created by Configure) now contains
7361 sections with information on -D... compiler switches used for
7362 compiling the library so that applications can see them. To enable
7363 one of these sections, a pre-processor symbol OPENSSL_..._DEFINES
7364 must be defined. E.g.,
7365 #define OPENSSL_ALGORITHM_DEFINES
7366 #include <openssl/opensslconf.h>
7367 defines all pertinent NO_<algo> symbols, such as NO_IDEA, NO_RSA, etc.
7368 [Richard Levitte, Ulf and Bodo Möller]
7369
7370 *) Bugfix: Tolerate fragmentation and interleaving in the SSL 3/TLS
7371 record layer.
7372 [Bodo Moeller]
7373
7374 *) Change the 'other' type in certificate aux info to a STACK_OF
7375 X509_ALGOR. Although not an AlgorithmIdentifier as such it has
7376 the required ASN1 format: arbitrary types determined by an OID.
7377 [Steve Henson]
7378
7379 *) Add some PEM_write_X509_REQ_NEW() functions and a command line
7380 argument to 'req'. This is not because the function is newer or
7381 better than others it just uses the work 'NEW' in the certificate
7382 request header lines. Some software needs this.
7383 [Steve Henson]
7384
7385 *) Reorganise password command line arguments: now passwords can be
7386 obtained from various sources. Delete the PEM_cb function and make
7387 it the default behaviour: i.e. if the callback is NULL and the
7388 usrdata argument is not NULL interpret it as a null terminated pass
7389 phrase. If usrdata and the callback are NULL then the pass phrase
7390 is prompted for as usual.
7391 [Steve Henson]
7392
7393 *) Add support for the Compaq Atalla crypto accelerator. If it is installed,
7394 the support is automatically enabled. The resulting binaries will
7395 autodetect the card and use it if present.
7396 [Ben Laurie and Compaq Inc.]
7397
7398 *) Work around for Netscape hang bug. This sends certificate request
7399 and server done in one record. Since this is perfectly legal in the
7400 SSL/TLS protocol it isn't a "bug" option and is on by default. See
7401 the bugs/SSLv3 entry for more info.
7402 [Steve Henson]
7403
7404 *) HP-UX tune-up: new unified configs, HP C compiler bug workaround.
7405 [Andy Polyakov]
7406
7407 *) Add -rand argument to smime and pkcs12 applications and read/write
7408 of seed file.
7409 [Steve Henson]
7410
7411 *) New 'passwd' tool for crypt(3) and apr1 password hashes.
7412 [Bodo Moeller]
7413
7414 *) Add command line password options to the remaining applications.
7415 [Steve Henson]
7416
7417 *) Bug fix for BN_div_recp() for numerators with an even number of
7418 bits.
7419 [Ulf Möller]
7420
7421 *) More tests in bntest.c, and changed test_bn output.
7422 [Ulf Möller]
7423
7424 *) ./config recognizes MacOS X now.
7425 [Andy Polyakov]
7426
7427 *) Bug fix for BN_div() when the first words of num and divsor are
7428 equal (it gave wrong results if (rem=(n1-q*d0)&BN_MASK2) < d0).
7429 [Ulf Möller]
7430
7431 *) Add support for various broken PKCS#8 formats, and command line
7432 options to produce them.
7433 [Steve Henson]
7434
7435 *) New functions BN_CTX_start(), BN_CTX_get() and BT_CTX_end() to
7436 get temporary BIGNUMs from a BN_CTX.
7437 [Ulf Möller]
7438
7439 *) Correct return values in BN_mod_exp_mont() and BN_mod_exp2_mont()
7440 for p == 0.
7441 [Ulf Möller]
7442
7443 *) Change the SSLeay_add_all_*() functions to OpenSSL_add_all_*() and
7444 include a #define from the old name to the new. The original intent
7445 was that statically linked binaries could for example just call
7446 SSLeay_add_all_ciphers() to just add ciphers to the table and not
7447 link with digests. This never worked becayse SSLeay_add_all_digests()
7448 and SSLeay_add_all_ciphers() were in the same source file so calling
7449 one would link with the other. They are now in separate source files.
7450 [Steve Henson]
7451
7452 *) Add a new -notext option to 'ca' and a -pubkey option to 'spkac'.
7453 [Steve Henson]
7454
7455 *) Use a less unusual form of the Miller-Rabin primality test (it used
7456 a binary algorithm for exponentiation integrated into the Miller-Rabin
7457 loop, our standard modexp algorithms are faster).
7458 [Bodo Moeller]
7459
7460 *) Support for the EBCDIC character set completed.
7461 [Martin Kraemer <Martin.Kraemer@Mch.SNI.De>]
7462
7463 *) Source code cleanups: use const where appropriate, eliminate casts,
7464 use void * instead of char * in lhash.
7465 [Ulf Möller]
7466
7467 *) Bugfix: ssl3_send_server_key_exchange was not restartable
7468 (the state was not changed to SSL3_ST_SW_KEY_EXCH_B, and because of
7469 this the server could overwrite ephemeral keys that the client
7470 has already seen).
7471 [Bodo Moeller]
7472
7473 *) Turn DSA_is_prime into a macro that calls BN_is_prime,
7474 using 50 iterations of the Rabin-Miller test.
7475
7476 DSA_generate_parameters now uses BN_is_prime_fasttest (with 50
7477 iterations of the Rabin-Miller test as required by the appendix
7478 to FIPS PUB 186[-1]) instead of DSA_is_prime.
7479 As BN_is_prime_fasttest includes trial division, DSA parameter
7480 generation becomes much faster.
7481
7482 This implies a change for the callback functions in DSA_is_prime
7483 and DSA_generate_parameters: The callback function is called once
7484 for each positive witness in the Rabin-Miller test, not just
7485 occasionally in the inner loop; and the parameters to the
7486 callback function now provide an iteration count for the outer
7487 loop rather than for the current invocation of the inner loop.
7488 DSA_generate_parameters additionally can call the callback
7489 function with an 'iteration count' of -1, meaning that a
7490 candidate has passed the trial division test (when q is generated
7491 from an application-provided seed, trial division is skipped).
7492 [Bodo Moeller]
7493
7494 *) New function BN_is_prime_fasttest that optionally does trial
7495 division before starting the Rabin-Miller test and has
7496 an additional BN_CTX * argument (whereas BN_is_prime always
7497 has to allocate at least one BN_CTX).
7498 'callback(1, -1, cb_arg)' is called when a number has passed the
7499 trial division stage.
7500 [Bodo Moeller]
7501
7502 *) Fix for bug in CRL encoding. The validity dates weren't being handled
7503 as ASN1_TIME.
7504 [Steve Henson]
7505
7506 *) New -pkcs12 option to CA.pl script to write out a PKCS#12 file.
7507 [Steve Henson]
7508
7509 *) New function BN_pseudo_rand().
7510 [Ulf Möller]
7511
7512 *) Clean up BN_mod_mul_montgomery(): replace the broken (and unreadable)
7513 bignum version of BN_from_montgomery() with the working code from
7514 SSLeay 0.9.0 (the word based version is faster anyway), and clean up
7515 the comments.
7516 [Ulf Möller]
7517
7518 *) Avoid a race condition in s2_clnt.c (function get_server_hello) that
7519 made it impossible to use the same SSL_SESSION data structure in
7520 SSL2 clients in multiple threads.
7521 [Bodo Moeller]
7522
7523 *) The return value of RAND_load_file() no longer counts bytes obtained
7524 by stat(). RAND_load_file(..., -1) is new and uses the complete file
7525 to seed the PRNG (previously an explicit byte count was required).
7526 [Ulf Möller, Bodo Möller]
7527
7528 *) Clean up CRYPTO_EX_DATA functions, some of these didn't have prototypes
7529 used (char *) instead of (void *) and had casts all over the place.
7530 [Steve Henson]
7531
7532 *) Make BN_generate_prime() return NULL on error if ret!=NULL.
7533 [Ulf Möller]
7534
7535 *) Retain source code compatibility for BN_prime_checks macro:
7536 BN_is_prime(..., BN_prime_checks, ...) now uses
7537 BN_prime_checks_for_size to determine the appropriate number of
7538 Rabin-Miller iterations.
7539 [Ulf Möller]
7540
7541 *) Diffie-Hellman uses "safe" primes: DH_check() return code renamed to
7542 DH_CHECK_P_NOT_SAFE_PRIME.
7543 (Check if this is true? OpenPGP calls them "strong".)
7544 [Ulf Möller]
7545
7546 *) Merge the functionality of "dh" and "gendh" programs into a new program
7547 "dhparam". The old programs are retained for now but will handle DH keys
7548 (instead of parameters) in future.
7549 [Steve Henson]
7550
7551 *) Make the ciphers, s_server and s_client programs check the return values
7552 when a new cipher list is set.
7553 [Steve Henson]
7554
7555 *) Enhance the SSL/TLS cipher mechanism to correctly handle the TLS 56bit
7556 ciphers. Before when the 56bit ciphers were enabled the sorting was
7557 wrong.
7558
7559 The syntax for the cipher sorting has been extended to support sorting by
7560 cipher-strength (using the strength_bits hard coded in the tables).
7561 The new command is "@STRENGTH" (see also doc/apps/ciphers.pod).
7562
7563 Fix a bug in the cipher-command parser: when supplying a cipher command
7564 string with an "undefined" symbol (neither command nor alphanumeric
7565 [A-Za-z0-9], ssl_set_cipher_list used to hang in an endless loop. Now
7566 an error is flagged.
7567
7568 Due to the strength-sorting extension, the code of the
7569 ssl_create_cipher_list() function was completely rearranged. I hope that
7570 the readability was also increased :-)
7571 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>]
7572
7573 *) Minor change to 'x509' utility. The -CAcreateserial option now uses 1
7574 for the first serial number and places 2 in the serial number file. This
7575 avoids problems when the root CA is created with serial number zero and
7576 the first user certificate has the same issuer name and serial number
7577 as the root CA.
7578 [Steve Henson]
7579
7580 *) Fixes to X509_ATTRIBUTE utilities, change the 'req' program so it uses
7581 the new code. Add documentation for this stuff.
7582 [Steve Henson]
7583
7584 *) Changes to X509_ATTRIBUTE utilities. These have been renamed from
7585 X509_*() to X509at_*() on the grounds that they don't handle X509
7586 structures and behave in an analagous way to the X509v3 functions:
7587 they shouldn't be called directly but wrapper functions should be used
7588 instead.
7589
7590 So we also now have some wrapper functions that call the X509at functions
7591 when passed certificate requests. (TO DO: similar things can be done with
7592 PKCS#7 signed and unsigned attributes, PKCS#12 attributes and a few other
7593 things. Some of these need some d2i or i2d and print functionality
7594 because they handle more complex structures.)
7595 [Steve Henson]
7596
7597 *) Add missing #ifndefs that caused missing symbols when building libssl
7598 as a shared library without RSA. Use #ifndef NO_SSL2 instead of
7599 NO_RSA in ssl/s2*.c.
7600 [Kris Kennaway <kris@hub.freebsd.org>, modified by Ulf Möller]
7601
7602 *) Precautions against using the PRNG uninitialized: RAND_bytes() now
7603 has a return value which indicates the quality of the random data
7604 (1 = ok, 0 = not seeded). Also an error is recorded on the thread's
7605 error queue. New function RAND_pseudo_bytes() generates output that is
7606 guaranteed to be unique but not unpredictable. RAND_add is like
7607 RAND_seed, but takes an extra argument for an entropy estimate
7608 (RAND_seed always assumes full entropy).
7609 [Ulf Möller]
7610
7611 *) Do more iterations of Rabin-Miller probable prime test (specifically,
7612 3 for 1024-bit primes, 6 for 512-bit primes, 12 for 256-bit primes
7613 instead of only 2 for all lengths; see BN_prime_checks_for_size definition
7614 in crypto/bn/bn_prime.c for the complete table). This guarantees a
7615 false-positive rate of at most 2^-80 for random input.
7616 [Bodo Moeller]
7617
7618 *) Rewrite ssl3_read_n (ssl/s3_pkt.c) avoiding a couple of bugs.
7619 [Bodo Moeller]
7620
7621 *) New function X509_CTX_rget_chain() (renamed to X509_CTX_get1_chain
7622 in the 0.9.5 release), this returns the chain
7623 from an X509_CTX structure with a dup of the stack and all
7624 the X509 reference counts upped: so the stack will exist
7625 after X509_CTX_cleanup() has been called. Modify pkcs12.c
7626 to use this.
7627
7628 Also make SSL_SESSION_print() print out the verify return
7629 code.
7630 [Steve Henson]
7631
7632 *) Add manpage for the pkcs12 command. Also change the default
7633 behaviour so MAC iteration counts are used unless the new
7634 -nomaciter option is used. This improves file security and
7635 only older versions of MSIE (4.0 for example) need it.
7636 [Steve Henson]
7637
7638 *) Honor the no-xxx Configure options when creating .DEF files.
7639 [Ulf Möller]
7640
7641 *) Add PKCS#10 attributes to field table: challengePassword,
7642 unstructuredName and unstructuredAddress. These are taken from
7643 draft PKCS#9 v2.0 but are compatible with v1.2 provided no
7644 international characters are used.
7645
7646 More changes to X509_ATTRIBUTE code: allow the setting of types
7647 based on strings. Remove the 'loc' parameter when adding
7648 attributes because these will be a SET OF encoding which is sorted
7649 in ASN1 order.
7650 [Steve Henson]
7651
7652 *) Initial changes to the 'req' utility to allow request generation
7653 automation. This will allow an application to just generate a template
7654 file containing all the field values and have req construct the
7655 request.
7656
7657 Initial support for X509_ATTRIBUTE handling. Stacks of these are
7658 used all over the place including certificate requests and PKCS#7
7659 structures. They are currently handled manually where necessary with
7660 some primitive wrappers for PKCS#7. The new functions behave in a
7661 manner analogous to the X509 extension functions: they allow
7662 attributes to be looked up by NID and added.
7663
7664 Later something similar to the X509V3 code would be desirable to
7665 automatically handle the encoding, decoding and printing of the
7666 more complex types. The string types like challengePassword can
7667 be handled by the string table functions.
7668
7669 Also modified the multi byte string table handling. Now there is
7670 a 'global mask' which masks out certain types. The table itself
7671 can use the flag STABLE_NO_MASK to ignore the mask setting: this
7672 is useful when for example there is only one permissible type
7673 (as in countryName) and using the mask might result in no valid
7674 types at all.
7675 [Steve Henson]
7676
7677 *) Clean up 'Finished' handling, and add functions SSL_get_finished and
7678 SSL_get_peer_finished to allow applications to obtain the latest
7679 Finished messages sent to the peer or expected from the peer,
7680 respectively. (SSL_get_peer_finished is usually the Finished message
7681 actually received from the peer, otherwise the protocol will be aborted.)
7682
7683 As the Finished message are message digests of the complete handshake
7684 (with a total of 192 bits for TLS 1.0 and more for SSL 3.0), they can
7685 be used for external authentication procedures when the authentication
7686 provided by SSL/TLS is not desired or is not enough.
7687 [Bodo Moeller]
7688
7689 *) Enhanced support for Alpha Linux is added. Now ./config checks if
7690 the host supports BWX extension and if Compaq C is present on the
7691 $PATH. Just exploiting of the BWX extension results in 20-30%
7692 performance kick for some algorithms, e.g. DES and RC4 to mention
7693 a couple. Compaq C in turn generates ~20% faster code for MD5 and
7694 SHA1.
7695 [Andy Polyakov]
7696
7697 *) Add support for MS "fast SGC". This is arguably a violation of the
7698 SSL3/TLS protocol. Netscape SGC does two handshakes: the first with
7699 weak crypto and after checking the certificate is SGC a second one
7700 with strong crypto. MS SGC stops the first handshake after receiving
7701 the server certificate message and sends a second client hello. Since
7702 a server will typically do all the time consuming operations before
7703 expecting any further messages from the client (server key exchange
7704 is the most expensive) there is little difference between the two.
7705
7706 To get OpenSSL to support MS SGC we have to permit a second client
7707 hello message after we have sent server done. In addition we have to
7708 reset the MAC if we do get this second client hello.
7709 [Steve Henson]
7710
7711 *) Add a function 'd2i_AutoPrivateKey()' this will automatically decide
7712 if a DER encoded private key is RSA or DSA traditional format. Changed
7713 d2i_PrivateKey_bio() to use it. This is only needed for the "traditional"
7714 format DER encoded private key. Newer code should use PKCS#8 format which
7715 has the key type encoded in the ASN1 structure. Added DER private key
7716 support to pkcs8 application.
7717 [Steve Henson]
7718
7719 *) SSL 3/TLS 1 servers now don't request certificates when an anonymous
7720 ciphersuites has been selected (as required by the SSL 3/TLS 1
7721 specifications). Exception: When SSL_VERIFY_FAIL_IF_NO_PEER_CERT
7722 is set, we interpret this as a request to violate the specification
7723 (the worst that can happen is a handshake failure, and 'correct'
7724 behaviour would result in a handshake failure anyway).
7725 [Bodo Moeller]
7726
7727 *) In SSL_CTX_add_session, take into account that there might be multiple
7728 SSL_SESSION structures with the same session ID (e.g. when two threads
7729 concurrently obtain them from an external cache).
7730 The internal cache can handle only one SSL_SESSION with a given ID,
7731 so if there's a conflict, we now throw out the old one to achieve
7732 consistency.
7733 [Bodo Moeller]
7734
7735 *) Add OIDs for idea and blowfish in CBC mode. This will allow both
7736 to be used in PKCS#5 v2.0 and S/MIME. Also add checking to
7737 some routines that use cipher OIDs: some ciphers do not have OIDs
7738 defined and so they cannot be used for S/MIME and PKCS#5 v2.0 for
7739 example.
7740 [Steve Henson]
7741
7742 *) Simplify the trust setting structure and code. Now we just have
7743 two sequences of OIDs for trusted and rejected settings. These will
7744 typically have values the same as the extended key usage extension
7745 and any application specific purposes.
7746
7747 The trust checking code now has a default behaviour: it will just
7748 check for an object with the same NID as the passed id. Functions can
7749 be provided to override either the default behaviour or the behaviour
7750 for a given id. SSL client, server and email already have functions
7751 in place for compatibility: they check the NID and also return "trusted"
7752 if the certificate is self signed.
7753 [Steve Henson]
7754
7755 *) Add d2i,i2d bio/fp functions for PrivateKey: these convert the
7756 traditional format into an EVP_PKEY structure.
7757 [Steve Henson]
7758
7759 *) Add a password callback function PEM_cb() which either prompts for
7760 a password if usr_data is NULL or otherwise assumes it is a null
7761 terminated password. Allow passwords to be passed on command line
7762 environment or config files in a few more utilities.
7763 [Steve Henson]
7764
7765 *) Add a bunch of DER and PEM functions to handle PKCS#8 format private
7766 keys. Add some short names for PKCS#8 PBE algorithms and allow them
7767 to be specified on the command line for the pkcs8 and pkcs12 utilities.
7768 Update documentation.
7769 [Steve Henson]
7770
7771 *) Support for ASN1 "NULL" type. This could be handled before by using
7772 ASN1_TYPE but there wasn't any function that would try to read a NULL
7773 and produce an error if it couldn't. For compatibility we also have
7774 ASN1_NULL_new() and ASN1_NULL_free() functions but these are faked and
7775 don't allocate anything because they don't need to.
7776 [Steve Henson]
7777
7778 *) Initial support for MacOS is now provided. Examine INSTALL.MacOS
7779 for details.
7780 [Andy Polyakov, Roy Woods <roy@centicsystems.ca>]
7781
7782 *) Rebuild of the memory allocation routines used by OpenSSL code and
7783 possibly others as well. The purpose is to make an interface that
7784 provide hooks so anyone can build a separate set of allocation and
7785 deallocation routines to be used by OpenSSL, for example memory
7786 pool implementations, or something else, which was previously hard
7787 since Malloc(), Realloc() and Free() were defined as macros having
7788 the values malloc, realloc and free, respectively (except for Win32
7789 compilations). The same is provided for memory debugging code.
7790 OpenSSL already comes with functionality to find memory leaks, but
7791 this gives people a chance to debug other memory problems.
7792
7793 With these changes, a new set of functions and macros have appeared:
7794
7795 CRYPTO_set_mem_debug_functions() [F]
7796 CRYPTO_get_mem_debug_functions() [F]
7797 CRYPTO_dbg_set_options() [F]
7798 CRYPTO_dbg_get_options() [F]
7799 CRYPTO_malloc_debug_init() [M]
7800
7801 The memory debug functions are NULL by default, unless the library
7802 is compiled with CRYPTO_MDEBUG or friends is defined. If someone
7803 wants to debug memory anyway, CRYPTO_malloc_debug_init() (which
7804 gives the standard debugging functions that come with OpenSSL) or
7805 CRYPTO_set_mem_debug_functions() (tells OpenSSL to use functions
7806 provided by the library user) must be used. When the standard
7807 debugging functions are used, CRYPTO_dbg_set_options can be used to
7808 request additional information:
7809 CRYPTO_dbg_set_options(V_CYRPTO_MDEBUG_xxx) corresponds to setting
7810 the CRYPTO_MDEBUG_xxx macro when compiling the library.
7811
7812 Also, things like CRYPTO_set_mem_functions will always give the
7813 expected result (the new set of functions is used for allocation
7814 and deallocation) at all times, regardless of platform and compiler
7815 options.
7816
7817 To finish it up, some functions that were never use in any other
7818 way than through macros have a new API and new semantic:
7819
7820 CRYPTO_dbg_malloc()
7821 CRYPTO_dbg_realloc()
7822 CRYPTO_dbg_free()
7823
7824 All macros of value have retained their old syntax.
7825 [Richard Levitte and Bodo Moeller]
7826
7827 *) Some S/MIME fixes. The OID for SMIMECapabilities was wrong, the
7828 ordering of SMIMECapabilities wasn't in "strength order" and there
7829 was a missing NULL in the AlgorithmIdentifier for the SHA1 signature
7830 algorithm.
7831 [Steve Henson]
7832
7833 *) Some ASN1 types with illegal zero length encoding (INTEGER,
7834 ENUMERATED and OBJECT IDENTIFIER) choked the ASN1 routines.
7835 [Frans Heymans <fheymans@isaserver.be>, modified by Steve Henson]
7836
7837 *) Merge in my S/MIME library for OpenSSL. This provides a simple
7838 S/MIME API on top of the PKCS#7 code, a MIME parser (with enough
7839 functionality to handle multipart/signed properly) and a utility
7840 called 'smime' to call all this stuff. This is based on code I
7841 originally wrote for Celo who have kindly allowed it to be
7842 included in OpenSSL.
7843 [Steve Henson]
7844
7845 *) Add variants des_set_key_checked and des_set_key_unchecked of
7846 des_set_key (aka des_key_sched). Global variable des_check_key
7847 decides which of these is called by des_set_key; this way
7848 des_check_key behaves as it always did, but applications and
7849 the library itself, which was buggy for des_check_key == 1,
7850 have a cleaner way to pick the version they need.
7851 [Bodo Moeller]
7852
7853 *) New function PKCS12_newpass() which changes the password of a
7854 PKCS12 structure.
7855 [Steve Henson]
7856
7857 *) Modify X509_TRUST and X509_PURPOSE so it also uses a static and
7858 dynamic mix. In both cases the ids can be used as an index into the
7859 table. Also modified the X509_TRUST_add() and X509_PURPOSE_add()
7860 functions so they accept a list of the field values and the
7861 application doesn't need to directly manipulate the X509_TRUST
7862 structure.
7863 [Steve Henson]
7864
7865 *) Modify the ASN1_STRING_TABLE stuff so it also uses bsearch and doesn't
7866 need initialising.
7867 [Steve Henson]
7868
7869 *) Modify the way the V3 extension code looks up extensions. This now
7870 works in a similar way to the object code: we have some "standard"
7871 extensions in a static table which is searched with OBJ_bsearch()
7872 and the application can add dynamic ones if needed. The file
7873 crypto/x509v3/ext_dat.h now has the info: this file needs to be
7874 updated whenever a new extension is added to the core code and kept
7875 in ext_nid order. There is a simple program 'tabtest.c' which checks
7876 this. New extensions are not added too often so this file can readily
7877 be maintained manually.
7878
7879 There are two big advantages in doing things this way. The extensions
7880 can be looked up immediately and no longer need to be "added" using
7881 X509V3_add_standard_extensions(): this function now does nothing.
7882 [Side note: I get *lots* of email saying the extension code doesn't
7883 work because people forget to call this function]
7884 Also no dynamic allocation is done unless new extensions are added:
7885 so if we don't add custom extensions there is no need to call
7886 X509V3_EXT_cleanup().
7887 [Steve Henson]
7888
7889 *) Modify enc utility's salting as follows: make salting the default. Add a
7890 magic header, so unsalted files fail gracefully instead of just decrypting
7891 to garbage. This is because not salting is a big security hole, so people
7892 should be discouraged from doing it.
7893 [Ben Laurie]
7894
7895 *) Fixes and enhancements to the 'x509' utility. It allowed a message
7896 digest to be passed on the command line but it only used this
7897 parameter when signing a certificate. Modified so all relevant
7898 operations are affected by the digest parameter including the
7899 -fingerprint and -x509toreq options. Also -x509toreq choked if a
7900 DSA key was used because it didn't fix the digest.
7901 [Steve Henson]
7902
7903 *) Initial certificate chain verify code. Currently tests the untrusted
7904 certificates for consistency with the verify purpose (which is set
7905 when the X509_STORE_CTX structure is set up) and checks the pathlength.
7906
7907 There is a NO_CHAIN_VERIFY compilation option to keep the old behaviour:
7908 this is because it will reject chains with invalid extensions whereas
7909 every previous version of OpenSSL and SSLeay made no checks at all.
7910
7911 Trust code: checks the root CA for the relevant trust settings. Trust
7912 settings have an initial value consistent with the verify purpose: e.g.
7913 if the verify purpose is for SSL client use it expects the CA to be
7914 trusted for SSL client use. However the default value can be changed to
7915 permit custom trust settings: one example of this would be to only trust
7916 certificates from a specific "secure" set of CAs.
7917
7918 Also added X509_STORE_CTX_new() and X509_STORE_CTX_free() functions
7919 which should be used for version portability: especially since the
7920 verify structure is likely to change more often now.
7921
7922 SSL integration. Add purpose and trust to SSL_CTX and SSL and functions
7923 to set them. If not set then assume SSL clients will verify SSL servers
7924 and vice versa.
7925
7926 Two new options to the verify program: -untrusted allows a set of
7927 untrusted certificates to be passed in and -purpose which sets the
7928 intended purpose of the certificate. If a purpose is set then the
7929 new chain verify code is used to check extension consistency.
7930 [Steve Henson]
7931
7932 *) Support for the authority information access extension.
7933 [Steve Henson]
7934
7935 *) Modify RSA and DSA PEM read routines to transparently handle
7936 PKCS#8 format private keys. New *_PUBKEY_* functions that handle
7937 public keys in a format compatible with certificate
7938 SubjectPublicKeyInfo structures. Unfortunately there were already
7939 functions called *_PublicKey_* which used various odd formats so
7940 these are retained for compatibility: however the DSA variants were
7941 never in a public release so they have been deleted. Changed dsa/rsa
7942 utilities to handle the new format: note no releases ever handled public
7943 keys so we should be OK.
7944
7945 The primary motivation for this change is to avoid the same fiasco
7946 that dogs private keys: there are several incompatible private key
7947 formats some of which are standard and some OpenSSL specific and
7948 require various evil hacks to allow partial transparent handling and
7949 even then it doesn't work with DER formats. Given the option anything
7950 other than PKCS#8 should be dumped: but the other formats have to
7951 stay in the name of compatibility.
7952
7953 With public keys and the benefit of hindsight one standard format
7954 is used which works with EVP_PKEY, RSA or DSA structures: though
7955 it clearly returns an error if you try to read the wrong kind of key.
7956
7957 Added a -pubkey option to the 'x509' utility to output the public key.
7958 Also rename the EVP_PKEY_get_*() to EVP_PKEY_rget_*()
7959 (renamed to EVP_PKEY_get1_*() in the OpenSSL 0.9.5 release) and add
7960 EVP_PKEY_rset_*() functions (renamed to EVP_PKEY_set1_*())
7961 that do the same as the EVP_PKEY_assign_*() except they up the
7962 reference count of the added key (they don't "swallow" the
7963 supplied key).
7964 [Steve Henson]
7965
7966 *) Fixes to crypto/x509/by_file.c the code to read in certificates and
7967 CRLs would fail if the file contained no certificates or no CRLs:
7968 added a new function to read in both types and return the number
7969 read: this means that if none are read it will be an error. The
7970 DER versions of the certificate and CRL reader would always fail
7971 because it isn't possible to mix certificates and CRLs in DER format
7972 without choking one or the other routine. Changed this to just read
7973 a certificate: this is the best we can do. Also modified the code
7974 in apps/verify.c to take notice of return codes: it was previously
7975 attempting to read in certificates from NULL pointers and ignoring
7976 any errors: this is one reason why the cert and CRL reader seemed
7977 to work. It doesn't check return codes from the default certificate
7978 routines: these may well fail if the certificates aren't installed.
7979 [Steve Henson]
7980
7981 *) Code to support otherName option in GeneralName.
7982 [Steve Henson]
7983
7984 *) First update to verify code. Change the verify utility
7985 so it warns if it is passed a self signed certificate:
7986 for consistency with the normal behaviour. X509_verify
7987 has been modified to it will now verify a self signed
7988 certificate if *exactly* the same certificate appears
7989 in the store: it was previously impossible to trust a
7990 single self signed certificate. This means that:
7991 openssl verify ss.pem
7992 now gives a warning about a self signed certificate but
7993 openssl verify -CAfile ss.pem ss.pem
7994 is OK.
7995 [Steve Henson]
7996
7997 *) For servers, store verify_result in SSL_SESSION data structure
7998 (and add it to external session representation).
7999 This is needed when client certificate verifications fails,
8000 but an application-provided verification callback (set by
8001 SSL_CTX_set_cert_verify_callback) allows accepting the session
8002 anyway (i.e. leaves x509_store_ctx->error != X509_V_OK
8003 but returns 1): When the session is reused, we have to set
8004 ssl->verify_result to the appropriate error code to avoid
8005 security holes.
8006 [Bodo Moeller, problem pointed out by Lutz Jaenicke]
8007
8008 *) Fix a bug in the new PKCS#7 code: it didn't consider the
8009 case in PKCS7_dataInit() where the signed PKCS7 structure
8010 didn't contain any existing data because it was being created.
8011 [Po-Cheng Chen <pocheng@nst.com.tw>, slightly modified by Steve Henson]
8012
8013 *) Add a salt to the key derivation routines in enc.c. This
8014 forms the first 8 bytes of the encrypted file. Also add a
8015 -S option to allow a salt to be input on the command line.
8016 [Steve Henson]
8017
8018 *) New function X509_cmp(). Oddly enough there wasn't a function
8019 to compare two certificates. We do this by working out the SHA1
8020 hash and comparing that. X509_cmp() will be needed by the trust
8021 code.
8022 [Steve Henson]
8023
8024 *) SSL_get1_session() is like SSL_get_session(), but increments
8025 the reference count in the SSL_SESSION returned.
8026 [Geoff Thorpe <geoff@eu.c2.net>]
8027
8028 *) Fix for 'req': it was adding a null to request attributes.
8029 Also change the X509_LOOKUP and X509_INFO code to handle
8030 certificate auxiliary information.
8031 [Steve Henson]
8032
8033 *) Add support for 40 and 64 bit RC2 and RC4 algorithms: document
8034 the 'enc' command.
8035 [Steve Henson]
8036
8037 *) Add the possibility to add extra information to the memory leak
8038 detecting output, to form tracebacks, showing from where each
8039 allocation was originated: CRYPTO_push_info("constant string") adds
8040 the string plus current file name and line number to a per-thread
8041 stack, CRYPTO_pop_info() does the obvious, CRYPTO_remove_all_info()
8042 is like calling CYRPTO_pop_info() until the stack is empty.
8043 Also updated memory leak detection code to be multi-thread-safe.
8044 [Richard Levitte]
8045
8046 *) Add options -text and -noout to pkcs7 utility and delete the
8047 encryption options which never did anything. Update docs.
8048 [Steve Henson]
8049
8050 *) Add options to some of the utilities to allow the pass phrase
8051 to be included on either the command line (not recommended on
8052 OSes like Unix) or read from the environment. Update the
8053 manpages and fix a few bugs.
8054 [Steve Henson]
8055
8056 *) Add a few manpages for some of the openssl commands.
8057 [Steve Henson]
8058
8059 *) Fix the -revoke option in ca. It was freeing up memory twice,
8060 leaking and not finding already revoked certificates.
8061 [Steve Henson]
8062
8063 *) Extensive changes to support certificate auxiliary information.
8064 This involves the use of X509_CERT_AUX structure and X509_AUX
8065 functions. An X509_AUX function such as PEM_read_X509_AUX()
8066 can still read in a certificate file in the usual way but it
8067 will also read in any additional "auxiliary information". By
8068 doing things this way a fair degree of compatibility can be
8069 retained: existing certificates can have this information added
8070 using the new 'x509' options.
8071
8072 Current auxiliary information includes an "alias" and some trust
8073 settings. The trust settings will ultimately be used in enhanced
8074 certificate chain verification routines: currently a certificate
8075 can only be trusted if it is self signed and then it is trusted
8076 for all purposes.
8077 [Steve Henson]
8078
8079 *) Fix assembler for Alpha (tested only on DEC OSF not Linux or *BSD).
8080 The problem was that one of the replacement routines had not been working
8081 since SSLeay releases. For now the offending routine has been replaced
8082 with non-optimised assembler. Even so, this now gives around 95%
8083 performance improvement for 1024 bit RSA signs.
8084 [Mark Cox]
8085
8086 *) Hack to fix PKCS#7 decryption when used with some unorthodox RC2
8087 handling. Most clients have the effective key size in bits equal to
8088 the key length in bits: so a 40 bit RC2 key uses a 40 bit (5 byte) key.
8089 A few however don't do this and instead use the size of the decrypted key
8090 to determine the RC2 key length and the AlgorithmIdentifier to determine
8091 the effective key length. In this case the effective key length can still
8092 be 40 bits but the key length can be 168 bits for example. This is fixed
8093 by manually forcing an RC2 key into the EVP_PKEY structure because the
8094 EVP code can't currently handle unusual RC2 key sizes: it always assumes
8095 the key length and effective key length are equal.
8096 [Steve Henson]
8097
8098 *) Add a bunch of functions that should simplify the creation of
8099 X509_NAME structures. Now you should be able to do:
8100 X509_NAME_add_entry_by_txt(nm, "CN", MBSTRING_ASC, "Steve", -1, -1, 0);
8101 and have it automatically work out the correct field type and fill in
8102 the structures. The more adventurous can try:
8103 X509_NAME_add_entry_by_txt(nm, field, MBSTRING_UTF8, str, -1, -1, 0);
8104 and it will (hopefully) work out the correct multibyte encoding.
8105 [Steve Henson]
8106
8107 *) Change the 'req' utility to use the new field handling and multibyte
8108 copy routines. Before the DN field creation was handled in an ad hoc
8109 way in req, ca, and x509 which was rather broken and didn't support
8110 BMPStrings or UTF8Strings. Since some software doesn't implement
8111 BMPStrings or UTF8Strings yet, they can be enabled using the config file
8112 using the dirstring_type option. See the new comment in the default
8113 openssl.cnf for more info.
8114 [Steve Henson]
8115
8116 *) Make crypto/rand/md_rand.c more robust:
8117 - Assure unique random numbers after fork().
8118 - Make sure that concurrent threads access the global counter and
8119 md serializably so that we never lose entropy in them
8120 or use exactly the same state in multiple threads.
8121 Access to the large state is not always serializable because
8122 the additional locking could be a performance killer, and
8123 md should be large enough anyway.
8124 [Bodo Moeller]
8125
8126 *) New file apps/app_rand.c with commonly needed functionality
8127 for handling the random seed file.
8128
8129 Use the random seed file in some applications that previously did not:
8130 ca,
8131 dsaparam -genkey (which also ignored its '-rand' option),
8132 s_client,
8133 s_server,
8134 x509 (when signing).
8135 Except on systems with /dev/urandom, it is crucial to have a random
8136 seed file at least for key creation, DSA signing, and for DH exchanges;
8137 for RSA signatures we could do without one.
8138
8139 gendh and gendsa (unlike genrsa) used to read only the first byte
8140 of each file listed in the '-rand' option. The function as previously
8141 found in genrsa is now in app_rand.c and is used by all programs
8142 that support '-rand'.
8143 [Bodo Moeller]
8144
8145 *) In RAND_write_file, use mode 0600 for creating files;
8146 don't just chmod when it may be too late.
8147 [Bodo Moeller]
8148
8149 *) Report an error from X509_STORE_load_locations
8150 when X509_LOOKUP_load_file or X509_LOOKUP_add_dir failed.
8151 [Bill Perry]
8152
8153 *) New function ASN1_mbstring_copy() this copies a string in either
8154 ASCII, Unicode, Universal (4 bytes per character) or UTF8 format
8155 into an ASN1_STRING type. A mask of permissible types is passed
8156 and it chooses the "minimal" type to use or an error if not type
8157 is suitable.
8158 [Steve Henson]
8159
8160 *) Add function equivalents to the various macros in asn1.h. The old
8161 macros are retained with an M_ prefix. Code inside the library can
8162 use the M_ macros. External code (including the openssl utility)
8163 should *NOT* in order to be "shared library friendly".
8164 [Steve Henson]
8165
8166 *) Add various functions that can check a certificate's extensions
8167 to see if it usable for various purposes such as SSL client,
8168 server or S/MIME and CAs of these types. This is currently
8169 VERY EXPERIMENTAL but will ultimately be used for certificate chain
8170 verification. Also added a -purpose flag to x509 utility to
8171 print out all the purposes.
8172 [Steve Henson]
8173
8174 *) Add a CRYPTO_EX_DATA to X509 certificate structure and associated
8175 functions.
8176 [Steve Henson]
8177
8178 *) New X509V3_{X509,CRL,REVOKED}_get_d2i() functions. These will search
8179 for, obtain and decode and extension and obtain its critical flag.
8180 This allows all the necessary extension code to be handled in a
8181 single function call.
8182 [Steve Henson]
8183
8184 *) RC4 tune-up featuring 30-40% performance improvement on most RISC
8185 platforms. See crypto/rc4/rc4_enc.c for further details.
8186 [Andy Polyakov]
8187
8188 *) New -noout option to asn1parse. This causes no output to be produced
8189 its main use is when combined with -strparse and -out to extract data
8190 from a file (which may not be in ASN.1 format).
8191 [Steve Henson]
8192
8193 *) Fix for pkcs12 program. It was hashing an invalid certificate pointer
8194 when producing the local key id.
8195 [Richard Levitte <levitte@stacken.kth.se>]
8196
8197 *) New option -dhparam in s_server. This allows a DH parameter file to be
8198 stated explicitly. If it is not stated then it tries the first server
8199 certificate file. The previous behaviour hard coded the filename
8200 "server.pem".
8201 [Steve Henson]
8202
8203 *) Add -pubin and -pubout options to the rsa and dsa commands. These allow
8204 a public key to be input or output. For example:
8205 openssl rsa -in key.pem -pubout -out pubkey.pem
8206 Also added necessary DSA public key functions to handle this.
8207 [Steve Henson]
8208
8209 *) Fix so PKCS7_dataVerify() doesn't crash if no certificates are contained
8210 in the message. This was handled by allowing
8211 X509_find_by_issuer_and_serial() to tolerate a NULL passed to it.
8212 [Steve Henson, reported by Sampo Kellomaki <sampo@mail.neuronio.pt>]
8213
8214 *) Fix for bug in d2i_ASN1_bytes(): other ASN1 functions add an extra null
8215 to the end of the strings whereas this didn't. This would cause problems
8216 if strings read with d2i_ASN1_bytes() were later modified.
8217 [Steve Henson, reported by Arne Ansper <arne@ats.cyber.ee>]
8218
8219 *) Fix for base64 decode bug. When a base64 bio reads only one line of
8220 data and it contains EOF it will end up returning an error. This is
8221 caused by input 46 bytes long. The cause is due to the way base64
8222 BIOs find the start of base64 encoded data. They do this by trying a
8223 trial decode on each line until they find one that works. When they
8224 do a flag is set and it starts again knowing it can pass all the
8225 data directly through the decoder. Unfortunately it doesn't reset
8226 the context it uses. This means that if EOF is reached an attempt
8227 is made to pass two EOFs through the context and this causes the
8228 resulting error. This can also cause other problems as well. As is
8229 usual with these problems it takes *ages* to find and the fix is
8230 trivial: move one line.
8231 [Steve Henson, reported by ian@uns.ns.ac.yu (Ivan Nejgebauer) ]
8232
8233 *) Ugly workaround to get s_client and s_server working under Windows. The
8234 old code wouldn't work because it needed to select() on sockets and the
8235 tty (for keypresses and to see if data could be written). Win32 only
8236 supports select() on sockets so we select() with a 1s timeout on the
8237 sockets and then see if any characters are waiting to be read, if none
8238 are present then we retry, we also assume we can always write data to
8239 the tty. This isn't nice because the code then blocks until we've
8240 received a complete line of data and it is effectively polling the
8241 keyboard at 1s intervals: however it's quite a bit better than not
8242 working at all :-) A dedicated Windows application might handle this
8243 with an event loop for example.
8244 [Steve Henson]
8245
8246 *) Enhance RSA_METHOD structure. Now there are two extra methods, rsa_sign
8247 and rsa_verify. When the RSA_FLAGS_SIGN_VER option is set these functions
8248 will be called when RSA_sign() and RSA_verify() are used. This is useful
8249 if rsa_pub_dec() and rsa_priv_enc() equivalents are not available.
8250 For this to work properly RSA_public_decrypt() and RSA_private_encrypt()
8251 should *not* be used: RSA_sign() and RSA_verify() must be used instead.
8252 This necessitated the support of an extra signature type NID_md5_sha1
8253 for SSL signatures and modifications to the SSL library to use it instead
8254 of calling RSA_public_decrypt() and RSA_private_encrypt().
8255 [Steve Henson]
8256
8257 *) Add new -verify -CAfile and -CApath options to the crl program, these
8258 will lookup a CRL issuers certificate and verify the signature in a
8259 similar way to the verify program. Tidy up the crl program so it
8260 no longer accesses structures directly. Make the ASN1 CRL parsing a bit
8261 less strict. It will now permit CRL extensions even if it is not
8262 a V2 CRL: this will allow it to tolerate some broken CRLs.
8263 [Steve Henson]
8264
8265 *) Initialize all non-automatic variables each time one of the openssl
8266 sub-programs is started (this is necessary as they may be started
8267 multiple times from the "OpenSSL>" prompt).
8268 [Lennart Bang, Bodo Moeller]
8269
8270 *) Preliminary compilation option RSA_NULL which disables RSA crypto without
8271 removing all other RSA functionality (this is what NO_RSA does). This
8272 is so (for example) those in the US can disable those operations covered
8273 by the RSA patent while allowing storage and parsing of RSA keys and RSA
8274 key generation.
8275 [Steve Henson]
8276
8277 *) Non-copying interface to BIO pairs.
8278 (still largely untested)
8279 [Bodo Moeller]
8280
8281 *) New function ANS1_tag2str() to convert an ASN1 tag to a descriptive
8282 ASCII string. This was handled independently in various places before.
8283 [Steve Henson]
8284
8285 *) New functions UTF8_getc() and UTF8_putc() that parse and generate
8286 UTF8 strings a character at a time.
8287 [Steve Henson]
8288
8289 *) Use client_version from client hello to select the protocol
8290 (s23_srvr.c) and for RSA client key exchange verification
8291 (s3_srvr.c), as required by the SSL 3.0/TLS 1.0 specifications.
8292 [Bodo Moeller]
8293
8294 *) Add various utility functions to handle SPKACs, these were previously
8295 handled by poking round in the structure internals. Added new function
8296 NETSCAPE_SPKI_print() to print out SPKAC and a new utility 'spkac' to
8297 print, verify and generate SPKACs. Based on an original idea from
8298 Massimiliano Pala <madwolf@comune.modena.it> but extensively modified.
8299 [Steve Henson]
8300
8301 *) RIPEMD160 is operational on all platforms and is back in 'make test'.
8302 [Andy Polyakov]
8303
8304 *) Allow the config file extension section to be overwritten on the
8305 command line. Based on an original idea from Massimiliano Pala
8306 <madwolf@comune.modena.it>. The new option is called -extensions
8307 and can be applied to ca, req and x509. Also -reqexts to override
8308 the request extensions in req and -crlexts to override the crl extensions
8309 in ca.
8310 [Steve Henson]
8311
8312 *) Add new feature to the SPKAC handling in ca. Now you can include
8313 the same field multiple times by preceding it by "XXXX." for example:
8314 1.OU="Unit name 1"
8315 2.OU="Unit name 2"
8316 this is the same syntax as used in the req config file.
8317 [Steve Henson]
8318
8319 *) Allow certificate extensions to be added to certificate requests. These
8320 are specified in a 'req_extensions' option of the req section of the
8321 config file. They can be printed out with the -text option to req but
8322 are otherwise ignored at present.
8323 [Steve Henson]
8324
8325 *) Fix a horrible bug in enc_read() in crypto/evp/bio_enc.c: if the first
8326 data read consists of only the final block it would not decrypted because
8327 EVP_CipherUpdate() would correctly report zero bytes had been decrypted.
8328 A misplaced 'break' also meant the decrypted final block might not be
8329 copied until the next read.
8330 [Steve Henson]
8331
8332 *) Initial support for DH_METHOD. Again based on RSA_METHOD. Also added
8333 a few extra parameters to the DH structure: these will be useful if
8334 for example we want the value of 'q' or implement X9.42 DH.
8335 [Steve Henson]
8336
8337 *) Initial support for DSA_METHOD. This is based on the RSA_METHOD and
8338 provides hooks that allow the default DSA functions or functions on a
8339 "per key" basis to be replaced. This allows hardware acceleration and
8340 hardware key storage to be handled without major modification to the
8341 library. Also added low level modexp hooks and CRYPTO_EX structure and
8342 associated functions.
8343 [Steve Henson]
8344
8345 *) Add a new flag to memory BIOs, BIO_FLAG_MEM_RDONLY. This marks the BIO
8346 as "read only": it can't be written to and the buffer it points to will
8347 not be freed. Reading from a read only BIO is much more efficient than
8348 a normal memory BIO. This was added because there are several times when
8349 an area of memory needs to be read from a BIO. The previous method was
8350 to create a memory BIO and write the data to it, this results in two
8351 copies of the data and an O(n^2) reading algorithm. There is a new
8352 function BIO_new_mem_buf() which creates a read only memory BIO from
8353 an area of memory. Also modified the PKCS#7 routines to use read only
8354 memory BIOs.
8355 [Steve Henson]
8356
8357 *) Bugfix: ssl23_get_client_hello did not work properly when called in
8358 state SSL23_ST_SR_CLNT_HELLO_B, i.e. when the first 7 bytes of
8359 a SSLv2-compatible client hello for SSLv3 or TLSv1 could be read,
8360 but a retry condition occured while trying to read the rest.
8361 [Bodo Moeller]
8362
8363 *) The PKCS7_ENC_CONTENT_new() function was setting the content type as
8364 NID_pkcs7_encrypted by default: this was wrong since this should almost
8365 always be NID_pkcs7_data. Also modified the PKCS7_set_type() to handle
8366 the encrypted data type: this is a more sensible place to put it and it
8367 allows the PKCS#12 code to be tidied up that duplicated this
8368 functionality.
8369 [Steve Henson]
8370
8371 *) Changed obj_dat.pl script so it takes its input and output files on
8372 the command line. This should avoid shell escape redirection problems
8373 under Win32.
8374 [Steve Henson]
8375
8376 *) Initial support for certificate extension requests, these are included
8377 in things like Xenroll certificate requests. Included functions to allow
8378 extensions to be obtained and added.
8379 [Steve Henson]
8380
8381 *) -crlf option to s_client and s_server for sending newlines as
8382 CRLF (as required by many protocols).
8383 [Bodo Moeller]
8384
8385 Changes between 0.9.3a and 0.9.4 [09 Aug 1999]
8386
8387 *) Install libRSAglue.a when OpenSSL is built with RSAref.
8388 [Ralf S. Engelschall]
8389
8390 *) A few more ``#ifndef NO_FP_API / #endif'' pairs for consistency.
8391 [Andrija Antonijevic <TheAntony2@bigfoot.com>]
8392
8393 *) Fix -startdate and -enddate (which was missing) arguments to 'ca'
8394 program.
8395 [Steve Henson]
8396
8397 *) New function DSA_dup_DH, which duplicates DSA parameters/keys as
8398 DH parameters/keys (q is lost during that conversion, but the resulting
8399 DH parameters contain its length).
8400
8401 For 1024-bit p, DSA_generate_parameters followed by DSA_dup_DH is
8402 much faster than DH_generate_parameters (which creates parameters
8403 where p = 2*q + 1), and also the smaller q makes DH computations
8404 much more efficient (160-bit exponentiation instead of 1024-bit
8405 exponentiation); so this provides a convenient way to support DHE
8406 ciphersuites in SSL/TLS servers (see ssl/ssltest.c). It is of
8407 utter importance to use
8408 SSL_CTX_set_options(s_ctx, SSL_OP_SINGLE_DH_USE);
8409 or
8410 SSL_set_options(s_ctx, SSL_OP_SINGLE_DH_USE);
8411 when such DH parameters are used, because otherwise small subgroup
8412 attacks may become possible!
8413 [Bodo Moeller]
8414
8415 *) Avoid memory leak in i2d_DHparams.
8416 [Bodo Moeller]
8417
8418 *) Allow the -k option to be used more than once in the enc program:
8419 this allows the same encrypted message to be read by multiple recipients.
8420 [Steve Henson]
8421
8422 *) New function OBJ_obj2txt(buf, buf_len, a, no_name), this converts
8423 an ASN1_OBJECT to a text string. If the "no_name" parameter is set then
8424 it will always use the numerical form of the OID, even if it has a short
8425 or long name.
8426 [Steve Henson]
8427
8428 *) Added an extra RSA flag: RSA_FLAG_EXT_PKEY. Previously the rsa_mod_exp
8429 method only got called if p,q,dmp1,dmq1,iqmp components were present,
8430 otherwise bn_mod_exp was called. In the case of hardware keys for example
8431 no private key components need be present and it might store extra data
8432 in the RSA structure, which cannot be accessed from bn_mod_exp.
8433 By setting RSA_FLAG_EXT_PKEY rsa_mod_exp will always be called for
8434 private key operations.
8435 [Steve Henson]
8436
8437 *) Added support for SPARC Linux.
8438 [Andy Polyakov]
8439
8440 *) pem_password_cb function type incompatibly changed from
8441 typedef int pem_password_cb(char *buf, int size, int rwflag);
8442 to
8443 ....(char *buf, int size, int rwflag, void *userdata);
8444 so that applications can pass data to their callbacks:
8445 The PEM[_ASN1]_{read,write}... functions and macros now take an
8446 additional void * argument, which is just handed through whenever
8447 the password callback is called.
8448 [Damien Miller <dmiller@ilogic.com.au>; tiny changes by Bodo Moeller]
8449
8450 New function SSL_CTX_set_default_passwd_cb_userdata.
8451
8452 Compatibility note: As many C implementations push function arguments
8453 onto the stack in reverse order, the new library version is likely to
8454 interoperate with programs that have been compiled with the old
8455 pem_password_cb definition (PEM_whatever takes some data that
8456 happens to be on the stack as its last argument, and the callback
8457 just ignores this garbage); but there is no guarantee whatsoever that
8458 this will work.
8459
8460 *) The -DPLATFORM="\"$(PLATFORM)\"" definition and the similar -DCFLAGS=...
8461 (both in crypto/Makefile.ssl for use by crypto/cversion.c) caused
8462 problems not only on Windows, but also on some Unix platforms.
8463 To avoid problematic command lines, these definitions are now in an
8464 auto-generated file crypto/buildinf.h (created by crypto/Makefile.ssl
8465 for standard "make" builds, by util/mk1mf.pl for "mk1mf" builds).
8466 [Bodo Moeller]
8467
8468 *) MIPS III/IV assembler module is reimplemented.
8469 [Andy Polyakov]
8470
8471 *) More DES library cleanups: remove references to srand/rand and
8472 delete an unused file.
8473 [Ulf Möller]
8474
8475 *) Add support for the the free Netwide assembler (NASM) under Win32,
8476 since not many people have MASM (ml) and it can be hard to obtain.
8477 This is currently experimental but it seems to work OK and pass all
8478 the tests. Check out INSTALL.W32 for info.
8479 [Steve Henson]
8480
8481 *) Fix memory leaks in s3_clnt.c: All non-anonymous SSL3/TLS1 connections
8482 without temporary keys kept an extra copy of the server key,
8483 and connections with temporary keys did not free everything in case
8484 of an error.
8485 [Bodo Moeller]
8486
8487 *) New function RSA_check_key and new openssl rsa option -check
8488 for verifying the consistency of RSA keys.
8489 [Ulf Moeller, Bodo Moeller]
8490
8491 *) Various changes to make Win32 compile work:
8492 1. Casts to avoid "loss of data" warnings in p5_crpt2.c
8493 2. Change unsigned int to int in b_dump.c to avoid "signed/unsigned
8494 comparison" warnings.
8495 3. Add sk_<TYPE>_sort to DEF file generator and do make update.
8496 [Steve Henson]
8497
8498 *) Add a debugging option to PKCS#5 v2 key generation function: when
8499 you #define DEBUG_PKCS5V2 passwords, salts, iteration counts and
8500 derived keys are printed to stderr.
8501 [Steve Henson]
8502
8503 *) Copy the flags in ASN1_STRING_dup().
8504 [Roman E. Pavlov <pre@mo.msk.ru>]
8505
8506 *) The x509 application mishandled signing requests containing DSA
8507 keys when the signing key was also DSA and the parameters didn't match.
8508
8509 It was supposed to omit the parameters when they matched the signing key:
8510 the verifying software was then supposed to automatically use the CA's
8511 parameters if they were absent from the end user certificate.
8512
8513 Omitting parameters is no longer recommended. The test was also
8514 the wrong way round! This was probably due to unusual behaviour in
8515 EVP_cmp_parameters() which returns 1 if the parameters match.
8516 This meant that parameters were omitted when they *didn't* match and
8517 the certificate was useless. Certificates signed with 'ca' didn't have
8518 this bug.
8519 [Steve Henson, reported by Doug Erickson <Doug.Erickson@Part.NET>]
8520
8521 *) Memory leak checking (-DCRYPTO_MDEBUG) had some problems.
8522 The interface is as follows:
8523 Applications can use
8524 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_ON) aka MemCheck_start(),
8525 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_OFF) aka MemCheck_stop();
8526 "off" is now the default.
8527 The library internally uses
8528 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_DISABLE) aka MemCheck_off(),
8529 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_ENABLE) aka MemCheck_on()
8530 to disable memory-checking temporarily.
8531
8532 Some inconsistent states that previously were possible (and were
8533 even the default) are now avoided.
8534
8535 -DCRYPTO_MDEBUG_TIME is new and additionally stores the current time
8536 with each memory chunk allocated; this is occasionally more helpful
8537 than just having a counter.
8538
8539 -DCRYPTO_MDEBUG_THREAD is also new and adds the thread ID.
8540
8541 -DCRYPTO_MDEBUG_ALL enables all of the above, plus any future
8542 extensions.
8543 [Bodo Moeller]
8544
8545 *) Introduce "mode" for SSL structures (with defaults in SSL_CTX),
8546 which largely parallels "options", but is for changing API behaviour,
8547 whereas "options" are about protocol behaviour.
8548 Initial "mode" flags are:
8549
8550 SSL_MODE_ENABLE_PARTIAL_WRITE Allow SSL_write to report success when
8551 a single record has been written.
8552 SSL_MODE_ACCEPT_MOVING_WRITE_BUFFER Don't insist that SSL_write
8553 retries use the same buffer location.
8554 (But all of the contents must be
8555 copied!)
8556 [Bodo Moeller]
8557
8558 *) Bugfix: SSL_set_options ignored its parameter, only SSL_CTX_set_options
8559 worked.
8560
8561 *) Fix problems with no-hmac etc.
8562 [Ulf Möller, pointed out by Brian Wellington <bwelling@tislabs.com>]
8563
8564 *) New functions RSA_get_default_method(), RSA_set_method() and
8565 RSA_get_method(). These allows replacement of RSA_METHODs without having
8566 to mess around with the internals of an RSA structure.
8567 [Steve Henson]
8568
8569 *) Fix memory leaks in DSA_do_sign and DSA_is_prime.
8570 Also really enable memory leak checks in openssl.c and in some
8571 test programs.
8572 [Chad C. Mulligan, Bodo Moeller]
8573
8574 *) Fix a bug in d2i_ASN1_INTEGER() and i2d_ASN1_INTEGER() which can mess
8575 up the length of negative integers. This has now been simplified to just
8576 store the length when it is first determined and use it later, rather
8577 than trying to keep track of where data is copied and updating it to
8578 point to the end.
8579 [Steve Henson, reported by Brien Wheeler
8580 <bwheeler@authentica-security.com>]
8581
8582 *) Add a new function PKCS7_signatureVerify. This allows the verification
8583 of a PKCS#7 signature but with the signing certificate passed to the
8584 function itself. This contrasts with PKCS7_dataVerify which assumes the
8585 certificate is present in the PKCS#7 structure. This isn't always the
8586 case: certificates can be omitted from a PKCS#7 structure and be
8587 distributed by "out of band" means (such as a certificate database).
8588 [Steve Henson]
8589
8590 *) Complete the PEM_* macros with DECLARE_PEM versions to replace the
8591 function prototypes in pem.h, also change util/mkdef.pl to add the
8592 necessary function names.
8593 [Steve Henson]
8594
8595 *) mk1mf.pl (used by Windows builds) did not properly read the
8596 options set by Configure in the top level Makefile, and Configure
8597 was not even able to write more than one option correctly.
8598 Fixed, now "no-idea no-rc5 -DCRYPTO_MDEBUG" etc. works as intended.
8599 [Bodo Moeller]
8600
8601 *) New functions CONF_load_bio() and CONF_load_fp() to allow a config
8602 file to be loaded from a BIO or FILE pointer. The BIO version will
8603 for example allow memory BIOs to contain config info.
8604 [Steve Henson]
8605
8606 *) New function "CRYPTO_num_locks" that returns CRYPTO_NUM_LOCKS.
8607 Whoever hopes to achieve shared-library compatibility across versions
8608 must use this, not the compile-time macro.
8609 (Exercise 0.9.4: Which is the minimum library version required by
8610 such programs?)
8611 Note: All this applies only to multi-threaded programs, others don't
8612 need locks.
8613 [Bodo Moeller]
8614
8615 *) Add missing case to s3_clnt.c state machine -- one of the new SSL tests
8616 through a BIO pair triggered the default case, i.e.
8617 SSLerr(...,SSL_R_UNKNOWN_STATE).
8618 [Bodo Moeller]
8619
8620 *) New "BIO pair" concept (crypto/bio/bss_bio.c) so that applications
8621 can use the SSL library even if none of the specific BIOs is
8622 appropriate.
8623 [Bodo Moeller]
8624
8625 *) Fix a bug in i2d_DSAPublicKey() which meant it returned the wrong value
8626 for the encoded length.
8627 [Jeon KyoungHo <khjeon@sds.samsung.co.kr>]
8628
8629 *) Add initial documentation of the X509V3 functions.
8630 [Steve Henson]
8631
8632 *) Add a new pair of functions PEM_write_PKCS8PrivateKey() and
8633 PEM_write_bio_PKCS8PrivateKey() that are equivalent to
8634 PEM_write_PrivateKey() and PEM_write_bio_PrivateKey() but use the more
8635 secure PKCS#8 private key format with a high iteration count.
8636 [Steve Henson]
8637
8638 *) Fix determination of Perl interpreter: A perl or perl5
8639 _directory_ in $PATH was also accepted as the interpreter.
8640 [Ralf S. Engelschall]
8641
8642 *) Fix demos/sign/sign.c: well there wasn't anything strictly speaking
8643 wrong with it but it was very old and did things like calling
8644 PEM_ASN1_read() directly and used MD5 for the hash not to mention some
8645 unusual formatting.
8646 [Steve Henson]
8647
8648 *) Fix demos/selfsign.c: it used obsolete and deleted functions, changed
8649 to use the new extension code.
8650 [Steve Henson]
8651
8652 *) Implement the PEM_read/PEM_write functions in crypto/pem/pem_all.c
8653 with macros. This should make it easier to change their form, add extra
8654 arguments etc. Fix a few PEM prototypes which didn't have cipher as a
8655 constant.
8656 [Steve Henson]
8657
8658 *) Add to configuration table a new entry that can specify an alternative
8659 name for unistd.h (for pre-POSIX systems); we need this for NeXTstep,
8660 according to Mark Crispin <MRC@Panda.COM>.
8661 [Bodo Moeller]
8662
8663 #if 0
8664 *) DES CBC did not update the IV. Weird.
8665 [Ben Laurie]
8666 #else
8667 des_cbc_encrypt does not update the IV, but des_ncbc_encrypt does.
8668 Changing the behaviour of the former might break existing programs --
8669 where IV updating is needed, des_ncbc_encrypt can be used.
8670 #endif
8671
8672 *) When bntest is run from "make test" it drives bc to check its
8673 calculations, as well as internally checking them. If an internal check
8674 fails, it needs to cause bc to give a non-zero result or make test carries
8675 on without noticing the failure. Fixed.
8676 [Ben Laurie]
8677
8678 *) DES library cleanups.
8679 [Ulf Möller]
8680
8681 *) Add support for PKCS#5 v2.0 PBE algorithms. This will permit PKCS#8 to be
8682 used with any cipher unlike PKCS#5 v1.5 which can at most handle 64 bit
8683 ciphers. NOTE: although the key derivation function has been verified
8684 against some published test vectors it has not been extensively tested
8685 yet. Added a -v2 "cipher" option to pkcs8 application to allow the use
8686 of v2.0.
8687 [Steve Henson]
8688
8689 *) Instead of "mkdir -p", which is not fully portable, use new
8690 Perl script "util/mkdir-p.pl".
8691 [Bodo Moeller]
8692
8693 *) Rewrite the way password based encryption (PBE) is handled. It used to
8694 assume that the ASN1 AlgorithmIdentifier parameter was a PBEParameter
8695 structure. This was true for the PKCS#5 v1.5 and PKCS#12 PBE algorithms
8696 but doesn't apply to PKCS#5 v2.0 where it can be something else. Now
8697 the 'parameter' field of the AlgorithmIdentifier is passed to the
8698 underlying key generation function so it must do its own ASN1 parsing.
8699 This has also changed the EVP_PBE_CipherInit() function which now has a
8700 'parameter' argument instead of literal salt and iteration count values
8701 and the function EVP_PBE_ALGOR_CipherInit() has been deleted.
8702 [Steve Henson]
8703
8704 *) Support for PKCS#5 v1.5 compatible password based encryption algorithms
8705 and PKCS#8 functionality. New 'pkcs8' application linked to openssl.
8706 Needed to change the PEM_STRING_EVP_PKEY value which was just "PRIVATE
8707 KEY" because this clashed with PKCS#8 unencrypted string. Since this
8708 value was just used as a "magic string" and not used directly its
8709 value doesn't matter.
8710 [Steve Henson]
8711
8712 *) Introduce some semblance of const correctness to BN. Shame C doesn't
8713 support mutable.
8714 [Ben Laurie]
8715
8716 *) "linux-sparc64" configuration (ultrapenguin).
8717 [Ray Miller <ray.miller@oucs.ox.ac.uk>]
8718 "linux-sparc" configuration.
8719 [Christian Forster <fo@hawo.stw.uni-erlangen.de>]
8720
8721 *) config now generates no-xxx options for missing ciphers.
8722 [Ulf Möller]
8723
8724 *) Support the EBCDIC character set (work in progress).
8725 File ebcdic.c not yet included because it has a different license.
8726 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>]
8727
8728 *) Support BS2000/OSD-POSIX.
8729 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>]
8730
8731 *) Make callbacks for key generation use void * instead of char *.
8732 [Ben Laurie]
8733
8734 *) Make S/MIME samples compile (not yet tested).
8735 [Ben Laurie]
8736
8737 *) Additional typesafe stacks.
8738 [Ben Laurie]
8739
8740 *) New configuration variants "bsdi-elf-gcc" (BSD/OS 4.x).
8741 [Bodo Moeller]
8742
8743
8744 Changes between 0.9.3 and 0.9.3a [29 May 1999]
8745
8746 *) New configuration variant "sco5-gcc".
8747
8748 *) Updated some demos.
8749 [Sean O Riordain, Wade Scholine]
8750
8751 *) Add missing BIO_free at exit of pkcs12 application.
8752 [Wu Zhigang]
8753
8754 *) Fix memory leak in conf.c.
8755 [Steve Henson]
8756
8757 *) Updates for Win32 to assembler version of MD5.
8758 [Steve Henson]
8759
8760 *) Set #! path to perl in apps/der_chop to where we found it
8761 instead of using a fixed path.
8762 [Bodo Moeller]
8763
8764 *) SHA library changes for irix64-mips4-cc.
8765 [Andy Polyakov]
8766
8767 *) Improvements for VMS support.
8768 [Richard Levitte]
8769
8770
8771 Changes between 0.9.2b and 0.9.3 [24 May 1999]
8772
8773 *) Bignum library bug fix. IRIX 6 passes "make test" now!
8774 This also avoids the problems with SC4.2 and unpatched SC5.
8775 [Andy Polyakov <appro@fy.chalmers.se>]
8776
8777 *) New functions sk_num, sk_value and sk_set to replace the previous macros.
8778 These are required because of the typesafe stack would otherwise break
8779 existing code. If old code used a structure member which used to be STACK
8780 and is now STACK_OF (for example cert in a PKCS7_SIGNED structure) with
8781 sk_num or sk_value it would produce an error because the num, data members
8782 are not present in STACK_OF. Now it just produces a warning. sk_set
8783 replaces the old method of assigning a value to sk_value
8784 (e.g. sk_value(x, i) = y) which the library used in a few cases. Any code
8785 that does this will no longer work (and should use sk_set instead) but
8786 this could be regarded as a "questionable" behaviour anyway.
8787 [Steve Henson]
8788
8789 *) Fix most of the other PKCS#7 bugs. The "experimental" code can now
8790 correctly handle encrypted S/MIME data.
8791 [Steve Henson]
8792
8793 *) Change type of various DES function arguments from des_cblock
8794 (which means, in function argument declarations, pointer to char)
8795 to des_cblock * (meaning pointer to array with 8 char elements),
8796 which allows the compiler to do more typechecking; it was like
8797 that back in SSLeay, but with lots of ugly casts.
8798
8799 Introduce new type const_des_cblock.
8800 [Bodo Moeller]
8801
8802 *) Reorganise the PKCS#7 library and get rid of some of the more obvious
8803 problems: find RecipientInfo structure that matches recipient certificate
8804 and initialise the ASN1 structures properly based on passed cipher.
8805 [Steve Henson]
8806
8807 *) Belatedly make the BN tests actually check the results.
8808 [Ben Laurie]
8809
8810 *) Fix the encoding and decoding of negative ASN1 INTEGERS and conversion
8811 to and from BNs: it was completely broken. New compilation option
8812 NEG_PUBKEY_BUG to allow for some broken certificates that encode public
8813 key elements as negative integers.
8814 [Steve Henson]
8815
8816 *) Reorganize and speed up MD5.
8817 [Andy Polyakov <appro@fy.chalmers.se>]
8818
8819 *) VMS support.
8820 [Richard Levitte <richard@levitte.org>]
8821
8822 *) New option -out to asn1parse to allow the parsed structure to be
8823 output to a file. This is most useful when combined with the -strparse
8824 option to examine the output of things like OCTET STRINGS.
8825 [Steve Henson]
8826
8827 *) Make SSL library a little more fool-proof by not requiring any longer
8828 that SSL_set_{accept,connect}_state be called before
8829 SSL_{accept,connect} may be used (SSL_set_..._state is omitted
8830 in many applications because usually everything *appeared* to work as
8831 intended anyway -- now it really works as intended).
8832 [Bodo Moeller]
8833
8834 *) Move openssl.cnf out of lib/.
8835 [Ulf Möller]
8836
8837 *) Fix various things to let OpenSSL even pass ``egcc -pipe -O2 -Wall
8838 -Wshadow -Wpointer-arith -Wcast-align -Wmissing-prototypes
8839 -Wmissing-declarations -Wnested-externs -Winline'' with EGCS 1.1.2+
8840 [Ralf S. Engelschall]
8841
8842 *) Various fixes to the EVP and PKCS#7 code. It may now be able to
8843 handle PKCS#7 enveloped data properly.
8844 [Sebastian Akerman <sak@parallelconsulting.com>, modified by Steve]
8845
8846 *) Create a duplicate of the SSL_CTX's CERT in SSL_new instead of
8847 copying pointers. The cert_st handling is changed by this in
8848 various ways (and thus what used to be known as ctx->default_cert
8849 is now called ctx->cert, since we don't resort to s->ctx->[default_]cert
8850 any longer when s->cert does not give us what we need).
8851 ssl_cert_instantiate becomes obsolete by this change.
8852 As soon as we've got the new code right (possibly it already is?),
8853 we have solved a couple of bugs of the earlier code where s->cert
8854 was used as if it could not have been shared with other SSL structures.
8855
8856 Note that using the SSL API in certain dirty ways now will result
8857 in different behaviour than observed with earlier library versions:
8858 Changing settings for an SSL_CTX *ctx after having done s = SSL_new(ctx)
8859 does not influence s as it used to.
8860
8861 In order to clean up things more thoroughly, inside SSL_SESSION
8862 we don't use CERT any longer, but a new structure SESS_CERT
8863 that holds per-session data (if available); currently, this is
8864 the peer's certificate chain and, for clients, the server's certificate
8865 and temporary key. CERT holds only those values that can have
8866 meaningful defaults in an SSL_CTX.
8867 [Bodo Moeller]
8868
8869 *) New function X509V3_EXT_i2d() to create an X509_EXTENSION structure
8870 from the internal representation. Various PKCS#7 fixes: remove some
8871 evil casts and set the enc_dig_alg field properly based on the signing
8872 key type.
8873 [Steve Henson]
8874
8875 *) Allow PKCS#12 password to be set from the command line or the
8876 environment. Let 'ca' get its config file name from the environment
8877 variables "OPENSSL_CONF" or "SSLEAY_CONF" (for consistency with 'req'
8878 and 'x509').
8879 [Steve Henson]
8880
8881 *) Allow certificate policies extension to use an IA5STRING for the
8882 organization field. This is contrary to the PKIX definition but
8883 VeriSign uses it and IE5 only recognises this form. Document 'x509'
8884 extension option.
8885 [Steve Henson]
8886
8887 *) Add PEDANTIC compiler flag to allow compilation with gcc -pedantic,
8888 without disallowing inline assembler and the like for non-pedantic builds.
8889 [Ben Laurie]
8890
8891 *) Support Borland C++ builder.
8892 [Janez Jere <jj@void.si>, modified by Ulf Möller]
8893
8894 *) Support Mingw32.
8895 [Ulf Möller]
8896
8897 *) SHA-1 cleanups and performance enhancements.
8898 [Andy Polyakov <appro@fy.chalmers.se>]
8899
8900 *) Sparc v8plus assembler for the bignum library.
8901 [Andy Polyakov <appro@fy.chalmers.se>]
8902
8903 *) Accept any -xxx and +xxx compiler options in Configure.
8904 [Ulf Möller]
8905
8906 *) Update HPUX configuration.
8907 [Anonymous]
8908
8909 *) Add missing sk_<type>_unshift() function to safestack.h
8910 [Ralf S. Engelschall]
8911
8912 *) New function SSL_CTX_use_certificate_chain_file that sets the
8913 "extra_cert"s in addition to the certificate. (This makes sense
8914 only for "PEM" format files, as chains as a whole are not
8915 DER-encoded.)
8916 [Bodo Moeller]
8917
8918 *) Support verify_depth from the SSL API.
8919 x509_vfy.c had what can be considered an off-by-one-error:
8920 Its depth (which was not part of the external interface)
8921 was actually counting the number of certificates in a chain;
8922 now it really counts the depth.
8923 [Bodo Moeller]
8924
8925 *) Bugfix in crypto/x509/x509_cmp.c: The SSLerr macro was used
8926 instead of X509err, which often resulted in confusing error
8927 messages since the error codes are not globally unique
8928 (e.g. an alleged error in ssl3_accept when a certificate
8929 didn't match the private key).
8930
8931 *) New function SSL_CTX_set_session_id_context that allows to set a default
8932 value (so that you don't need SSL_set_session_id_context for each
8933 connection using the SSL_CTX).
8934 [Bodo Moeller]
8935
8936 *) OAEP decoding bug fix.
8937 [Ulf Möller]
8938
8939 *) Support INSTALL_PREFIX for package builders, as proposed by
8940 David Harris.
8941 [Bodo Moeller]
8942
8943 *) New Configure options "threads" and "no-threads". For systems
8944 where the proper compiler options are known (currently Solaris
8945 and Linux), "threads" is the default.
8946 [Bodo Moeller]
8947
8948 *) New script util/mklink.pl as a faster substitute for util/mklink.sh.
8949 [Bodo Moeller]
8950
8951 *) Install various scripts to $(OPENSSLDIR)/misc, not to
8952 $(INSTALLTOP)/bin -- they shouldn't clutter directories
8953 such as /usr/local/bin.
8954 [Bodo Moeller]
8955
8956 *) "make linux-shared" to build shared libraries.
8957 [Niels Poppe <niels@netbox.org>]
8958
8959 *) New Configure option no-<cipher> (rsa, idea, rc5, ...).
8960 [Ulf Möller]
8961
8962 *) Add the PKCS#12 API documentation to openssl.txt. Preliminary support for
8963 extension adding in x509 utility.
8964 [Steve Henson]
8965
8966 *) Remove NOPROTO sections and error code comments.
8967 [Ulf Möller]
8968
8969 *) Partial rewrite of the DEF file generator to now parse the ANSI
8970 prototypes.
8971 [Steve Henson]
8972
8973 *) New Configure options --prefix=DIR and --openssldir=DIR.
8974 [Ulf Möller]
8975
8976 *) Complete rewrite of the error code script(s). It is all now handled
8977 by one script at the top level which handles error code gathering,
8978 header rewriting and C source file generation. It should be much better
8979 than the old method: it now uses a modified version of Ulf's parser to
8980 read the ANSI prototypes in all header files (thus the old K&R definitions
8981 aren't needed for error creation any more) and do a better job of
8982 translating function codes into names. The old 'ASN1 error code imbedded
8983 in a comment' is no longer necessary and it doesn't use .err files which
8984 have now been deleted. Also the error code call doesn't have to appear all
8985 on one line (which resulted in some large lines...).
8986 [Steve Henson]
8987
8988 *) Change #include filenames from <foo.h> to <openssl/foo.h>.
8989 [Bodo Moeller]
8990
8991 *) Change behaviour of ssl2_read when facing length-0 packets: Don't return
8992 0 (which usually indicates a closed connection), but continue reading.
8993 [Bodo Moeller]
8994
8995 *) Fix some race conditions.
8996 [Bodo Moeller]
8997
8998 *) Add support for CRL distribution points extension. Add Certificate
8999 Policies and CRL distribution points documentation.
9000 [Steve Henson]
9001
9002 *) Move the autogenerated header file parts to crypto/opensslconf.h.
9003 [Ulf Möller]
9004
9005 *) Fix new 56-bit DES export ciphersuites: they were using 7 bytes instead of
9006 8 of keying material. Merlin has also confirmed interop with this fix
9007 between OpenSSL and Baltimore C/SSL 2.0 and J/SSL 2.0.
9008 [Merlin Hughes <merlin@baltimore.ie>]
9009
9010 *) Fix lots of warnings.
9011 [Richard Levitte <levitte@stacken.kth.se>]
9012
9013 *) In add_cert_dir() in crypto/x509/by_dir.c, break out of the loop if
9014 the directory spec didn't end with a LIST_SEPARATOR_CHAR.
9015 [Richard Levitte <levitte@stacken.kth.se>]
9016
9017 *) Fix problems with sizeof(long) == 8.
9018 [Andy Polyakov <appro@fy.chalmers.se>]
9019
9020 *) Change functions to ANSI C.
9021 [Ulf Möller]
9022
9023 *) Fix typos in error codes.
9024 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>, Ulf Möller]
9025
9026 *) Remove defunct assembler files from Configure.
9027 [Ulf Möller]
9028
9029 *) SPARC v8 assembler BIGNUM implementation.
9030 [Andy Polyakov <appro@fy.chalmers.se>]
9031
9032 *) Support for Certificate Policies extension: both print and set.
9033 Various additions to support the r2i method this uses.
9034 [Steve Henson]
9035
9036 *) A lot of constification, and fix a bug in X509_NAME_oneline() that could
9037 return a const string when you are expecting an allocated buffer.
9038 [Ben Laurie]
9039
9040 *) Add support for ASN1 types UTF8String and VISIBLESTRING, also the CHOICE
9041 types DirectoryString and DisplayText.
9042 [Steve Henson]
9043
9044 *) Add code to allow r2i extensions to access the configuration database,
9045 add an LHASH database driver and add several ctx helper functions.
9046 [Steve Henson]
9047
9048 *) Fix an evil bug in bn_expand2() which caused various BN functions to
9049 fail when they extended the size of a BIGNUM.
9050 [Steve Henson]
9051
9052 *) Various utility functions to handle SXNet extension. Modify mkdef.pl to
9053 support typesafe stack.
9054 [Steve Henson]
9055
9056 *) Fix typo in SSL_[gs]et_options().
9057 [Nils Frostberg <nils@medcom.se>]
9058
9059 *) Delete various functions and files that belonged to the (now obsolete)
9060 old X509V3 handling code.
9061 [Steve Henson]
9062
9063 *) New Configure option "rsaref".
9064 [Ulf Möller]
9065
9066 *) Don't auto-generate pem.h.
9067 [Bodo Moeller]
9068
9069 *) Introduce type-safe ASN.1 SETs.
9070 [Ben Laurie]
9071
9072 *) Convert various additional casted stacks to type-safe STACK_OF() variants.
9073 [Ben Laurie, Ralf S. Engelschall, Steve Henson]
9074
9075 *) Introduce type-safe STACKs. This will almost certainly break lots of code
9076 that links with OpenSSL (well at least cause lots of warnings), but fear
9077 not: the conversion is trivial, and it eliminates loads of evil casts. A
9078 few STACKed things have been converted already. Feel free to convert more.
9079 In the fullness of time, I'll do away with the STACK type altogether.
9080 [Ben Laurie]
9081
9082 *) Add `openssl ca -revoke <certfile>' facility which revokes a certificate
9083 specified in <certfile> by updating the entry in the index.txt file.
9084 This way one no longer has to edit the index.txt file manually for
9085 revoking a certificate. The -revoke option does the gory details now.
9086 [Massimiliano Pala <madwolf@openca.org>, Ralf S. Engelschall]
9087
9088 *) Fix `openssl crl -noout -text' combination where `-noout' killed the
9089 `-text' option at all and this way the `-noout -text' combination was
9090 inconsistent in `openssl crl' with the friends in `openssl x509|rsa|dsa'.
9091 [Ralf S. Engelschall]
9092
9093 *) Make sure a corresponding plain text error message exists for the
9094 X509_V_ERR_CERT_REVOKED/23 error number which can occur when a
9095 verify callback function determined that a certificate was revoked.
9096 [Ralf S. Engelschall]
9097
9098 *) Bugfix: In test/testenc, don't test "openssl <cipher>" for
9099 ciphers that were excluded, e.g. by -DNO_IDEA. Also, test
9100 all available cipers including rc5, which was forgotten until now.
9101 In order to let the testing shell script know which algorithms
9102 are available, a new (up to now undocumented) command
9103 "openssl list-cipher-commands" is used.
9104 [Bodo Moeller]
9105
9106 *) Bugfix: s_client occasionally would sleep in select() when
9107 it should have checked SSL_pending() first.
9108 [Bodo Moeller]
9109
9110 *) New functions DSA_do_sign and DSA_do_verify to provide access to
9111 the raw DSA values prior to ASN.1 encoding.
9112 [Ulf Möller]
9113
9114 *) Tweaks to Configure
9115 [Niels Poppe <niels@netbox.org>]
9116
9117 *) Add support for PKCS#5 v2.0 ASN1 PBES2 structures. No other support,
9118 yet...
9119 [Steve Henson]
9120
9121 *) New variables $(RANLIB) and $(PERL) in the Makefiles.
9122 [Ulf Möller]
9123
9124 *) New config option to avoid instructions that are illegal on the 80386.
9125 The default code is faster, but requires at least a 486.
9126 [Ulf Möller]
9127
9128 *) Got rid of old SSL2_CLIENT_VERSION (inconsistently used) and
9129 SSL2_SERVER_VERSION (not used at all) macros, which are now the
9130 same as SSL2_VERSION anyway.
9131 [Bodo Moeller]
9132
9133 *) New "-showcerts" option for s_client.
9134 [Bodo Moeller]
9135
9136 *) Still more PKCS#12 integration. Add pkcs12 application to openssl
9137 application. Various cleanups and fixes.
9138 [Steve Henson]
9139
9140 *) More PKCS#12 integration. Add new pkcs12 directory with Makefile.ssl and
9141 modify error routines to work internally. Add error codes and PBE init
9142 to library startup routines.
9143 [Steve Henson]
9144
9145 *) Further PKCS#12 integration. Added password based encryption, PKCS#8 and
9146 packing functions to asn1 and evp. Changed function names and error
9147 codes along the way.
9148 [Steve Henson]
9149
9150 *) PKCS12 integration: and so it begins... First of several patches to
9151 slowly integrate PKCS#12 functionality into OpenSSL. Add PKCS#12
9152 objects to objects.h
9153 [Steve Henson]
9154
9155 *) Add a new 'indent' option to some X509V3 extension code. Initial ASN1
9156 and display support for Thawte strong extranet extension.
9157 [Steve Henson]
9158
9159 *) Add LinuxPPC support.
9160 [Jeff Dubrule <igor@pobox.org>]
9161
9162 *) Get rid of redundant BN file bn_mulw.c, and rename bn_div64 to
9163 bn_div_words in alpha.s.
9164 [Hannes Reinecke <H.Reinecke@hw.ac.uk> and Ben Laurie]
9165
9166 *) Make sure the RSA OAEP test is skipped under -DRSAref because
9167 OAEP isn't supported when OpenSSL is built with RSAref.
9168 [Ulf Moeller <ulf@fitug.de>]
9169
9170 *) Move definitions of IS_SET/IS_SEQUENCE inside crypto/asn1/asn1.h
9171 so they no longer are missing under -DNOPROTO.
9172 [Soren S. Jorvang <soren@t.dk>]
9173
9174
9175 Changes between 0.9.1c and 0.9.2b [22 Mar 1999]
9176
9177 *) Make SSL_get_peer_cert_chain() work in servers. Unfortunately, it still
9178 doesn't work when the session is reused. Coming soon!
9179 [Ben Laurie]
9180
9181 *) Fix a security hole, that allows sessions to be reused in the wrong
9182 context thus bypassing client cert protection! All software that uses
9183 client certs and session caches in multiple contexts NEEDS PATCHING to
9184 allow session reuse! A fuller solution is in the works.
9185 [Ben Laurie, problem pointed out by Holger Reif, Bodo Moeller (and ???)]
9186
9187 *) Some more source tree cleanups (removed obsolete files
9188 crypto/bf/asm/bf586.pl, test/test.txt and crypto/sha/asm/f.s; changed
9189 permission on "config" script to be executable) and a fix for the INSTALL
9190 document.
9191 [Ulf Moeller <ulf@fitug.de>]
9192
9193 *) Remove some legacy and erroneous uses of malloc, free instead of
9194 Malloc, Free.
9195 [Lennart Bang <lob@netstream.se>, with minor changes by Steve]
9196
9197 *) Make rsa_oaep_test return non-zero on error.
9198 [Ulf Moeller <ulf@fitug.de>]
9199
9200 *) Add support for native Solaris shared libraries. Configure
9201 solaris-sparc-sc4-pic, make, then run shlib/solaris-sc4.sh. It'd be nice
9202 if someone would make that last step automatic.
9203 [Matthias Loepfe <Matthias.Loepfe@AdNovum.CH>]
9204
9205 *) ctx_size was not built with the right compiler during "make links". Fixed.
9206 [Ben Laurie]
9207
9208 *) Change the meaning of 'ALL' in the cipher list. It now means "everything
9209 except NULL ciphers". This means the default cipher list will no longer
9210 enable NULL ciphers. They need to be specifically enabled e.g. with
9211 the string "DEFAULT:eNULL".
9212 [Steve Henson]
9213
9214 *) Fix to RSA private encryption routines: if p < q then it would
9215 occasionally produce an invalid result. This will only happen with
9216 externally generated keys because OpenSSL (and SSLeay) ensure p > q.
9217 [Steve Henson]
9218
9219 *) Be less restrictive and allow also `perl util/perlpath.pl
9220 /path/to/bin/perl' in addition to `perl util/perlpath.pl /path/to/bin',
9221 because this way one can also use an interpreter named `perl5' (which is
9222 usually the name of Perl 5.xxx on platforms where an Perl 4.x is still
9223 installed as `perl').
9224 [Matthias Loepfe <Matthias.Loepfe@adnovum.ch>]
9225
9226 *) Let util/clean-depend.pl work also with older Perl 5.00x versions.
9227 [Matthias Loepfe <Matthias.Loepfe@adnovum.ch>]
9228
9229 *) Fix Makefile.org so CC,CFLAG etc are passed to 'make links' add
9230 advapi32.lib to Win32 build and change the pem test comparision
9231 to fc.exe (thanks to Ulrich Kroener <kroneru@yahoo.com> for the
9232 suggestion). Fix misplaced ASNI prototypes and declarations in evp.h
9233 and crypto/des/ede_cbcm_enc.c.
9234 [Steve Henson]
9235
9236 *) DES quad checksum was broken on big-endian architectures. Fixed.
9237 [Ben Laurie]
9238
9239 *) Comment out two functions in bio.h that aren't implemented. Fix up the
9240 Win32 test batch file so it (might) work again. The Win32 test batch file
9241 is horrible: I feel ill....
9242 [Steve Henson]
9243
9244 *) Move various #ifdefs around so NO_SYSLOG, NO_DIRENT etc are now selected
9245 in e_os.h. Audit of header files to check ANSI and non ANSI
9246 sections: 10 functions were absent from non ANSI section and not exported
9247 from Windows DLLs. Fixed up libeay.num for new functions.
9248 [Steve Henson]
9249
9250 *) Make `openssl version' output lines consistent.
9251 [Ralf S. Engelschall]
9252
9253 *) Fix Win32 symbol export lists for BIO functions: Added
9254 BIO_get_ex_new_index, BIO_get_ex_num, BIO_get_ex_data and BIO_set_ex_data
9255 to ms/libeay{16,32}.def.
9256 [Ralf S. Engelschall]
9257
9258 *) Second round of fixing the OpenSSL perl/ stuff. It now at least compiled
9259 fine under Unix and passes some trivial tests I've now added. But the
9260 whole stuff is horribly incomplete, so a README.1ST with a disclaimer was
9261 added to make sure no one expects that this stuff really works in the
9262 OpenSSL 0.9.2 release. Additionally I've started to clean the XS sources
9263 up and fixed a few little bugs and inconsistencies in OpenSSL.{pm,xs} and
9264 openssl_bio.xs.
9265 [Ralf S. Engelschall]
9266
9267 *) Fix the generation of two part addresses in perl.
9268 [Kenji Miyake <kenji@miyake.org>, integrated by Ben Laurie]
9269
9270 *) Add config entry for Linux on MIPS.
9271 [John Tobey <jtobey@channel1.com>]
9272
9273 *) Make links whenever Configure is run, unless we are on Windoze.
9274 [Ben Laurie]
9275
9276 *) Permit extensions to be added to CRLs using crl_section in openssl.cnf.
9277 Currently only issuerAltName and AuthorityKeyIdentifier make any sense
9278 in CRLs.
9279 [Steve Henson]
9280
9281 *) Add a useful kludge to allow package maintainers to specify compiler and
9282 other platforms details on the command line without having to patch the
9283 Configure script everytime: One now can use ``perl Configure
9284 <id>:<details>'', i.e. platform ids are allowed to have details appended
9285 to them (seperated by colons). This is treated as there would be a static
9286 pre-configured entry in Configure's %table under key <id> with value
9287 <details> and ``perl Configure <id>'' is called. So, when you want to
9288 perform a quick test-compile under FreeBSD 3.1 with pgcc and without
9289 assembler stuff you can use ``perl Configure "FreeBSD-elf:pgcc:-O6:::"''
9290 now, which overrides the FreeBSD-elf entry on-the-fly.
9291 [Ralf S. Engelschall]
9292
9293 *) Disable new TLS1 ciphersuites by default: they aren't official yet.
9294 [Ben Laurie]
9295
9296 *) Allow DSO flags like -fpic, -fPIC, -KPIC etc. to be specified
9297 on the `perl Configure ...' command line. This way one can compile
9298 OpenSSL libraries with Position Independent Code (PIC) which is needed
9299 for linking it into DSOs.
9300 [Ralf S. Engelschall]
9301
9302 *) Remarkably, export ciphers were totally broken and no-one had noticed!
9303 Fixed.
9304 [Ben Laurie]
9305
9306 *) Cleaned up the LICENSE document: The official contact for any license
9307 questions now is the OpenSSL core team under openssl-core@openssl.org.
9308 And add a paragraph about the dual-license situation to make sure people
9309 recognize that _BOTH_ the OpenSSL license _AND_ the SSLeay license apply
9310 to the OpenSSL toolkit.
9311 [Ralf S. Engelschall]
9312
9313 *) General source tree makefile cleanups: Made `making xxx in yyy...'
9314 display consistent in the source tree and replaced `/bin/rm' by `rm'.
9315 Additonally cleaned up the `make links' target: Remove unnecessary
9316 semicolons, subsequent redundant removes, inline point.sh into mklink.sh
9317 to speed processing and no longer clutter the display with confusing
9318 stuff. Instead only the actually done links are displayed.
9319 [Ralf S. Engelschall]
9320
9321 *) Permit null encryption ciphersuites, used for authentication only. It used
9322 to be necessary to set the preprocessor define SSL_ALLOW_ENULL to do this.
9323 It is now necessary to set SSL_FORBID_ENULL to prevent the use of null
9324 encryption.
9325 [Ben Laurie]
9326
9327 *) Add a bunch of fixes to the PKCS#7 stuff. It used to sometimes reorder
9328 signed attributes when verifying signatures (this would break them),
9329 the detached data encoding was wrong and public keys obtained using
9330 X509_get_pubkey() weren't freed.
9331 [Steve Henson]
9332
9333 *) Add text documentation for the BUFFER functions. Also added a work around
9334 to a Win95 console bug. This was triggered by the password read stuff: the
9335 last character typed gets carried over to the next fread(). If you were
9336 generating a new cert request using 'req' for example then the last
9337 character of the passphrase would be CR which would then enter the first
9338 field as blank.
9339 [Steve Henson]
9340
9341 *) Added the new `Includes OpenSSL Cryptography Software' button as
9342 doc/openssl_button.{gif,html} which is similar in style to the old SSLeay
9343 button and can be used by applications based on OpenSSL to show the
9344 relationship to the OpenSSL project.
9345 [Ralf S. Engelschall]
9346
9347 *) Remove confusing variables in function signatures in files
9348 ssl/ssl_lib.c and ssl/ssl.h.
9349 [Lennart Bong <lob@kulthea.stacken.kth.se>]
9350
9351 *) Don't install bss_file.c under PREFIX/include/
9352 [Lennart Bong <lob@kulthea.stacken.kth.se>]
9353
9354 *) Get the Win32 compile working again. Modify mkdef.pl so it can handle
9355 functions that return function pointers and has support for NT specific
9356 stuff. Fix mk1mf.pl and VC-32.pl to support NT differences also. Various
9357 #ifdef WIN32 and WINNTs sprinkled about the place and some changes from
9358 unsigned to signed types: this was killing the Win32 compile.
9359 [Steve Henson]
9360
9361 *) Add new certificate file to stack functions,
9362 SSL_add_dir_cert_subjects_to_stack() and
9363 SSL_add_file_cert_subjects_to_stack(). These largely supplant
9364 SSL_load_client_CA_file(), and can be used to add multiple certs easily
9365 to a stack (usually this is then handed to SSL_CTX_set_client_CA_list()).
9366 This means that Apache-SSL and similar packages don't have to mess around
9367 to add as many CAs as they want to the preferred list.
9368 [Ben Laurie]
9369
9370 *) Experiment with doxygen documentation. Currently only partially applied to
9371 ssl/ssl_lib.c.
9372 See http://www.stack.nl/~dimitri/doxygen/index.html, and run doxygen with
9373 openssl.doxy as the configuration file.
9374 [Ben Laurie]
9375
9376 *) Get rid of remaining C++-style comments which strict C compilers hate.
9377 [Ralf S. Engelschall, pointed out by Carlos Amengual]
9378
9379 *) Changed BN_RECURSION in bn_mont.c to BN_RECURSION_MONT so it is not
9380 compiled in by default: it has problems with large keys.
9381 [Steve Henson]
9382
9383 *) Add a bunch of SSL_xxx() functions for configuring the temporary RSA and
9384 DH private keys and/or callback functions which directly correspond to
9385 their SSL_CTX_xxx() counterparts but work on a per-connection basis. This
9386 is needed for applications which have to configure certificates on a
9387 per-connection basis (e.g. Apache+mod_ssl) instead of a per-context basis
9388 (e.g. s_server).
9389 For the RSA certificate situation is makes no difference, but
9390 for the DSA certificate situation this fixes the "no shared cipher"
9391 problem where the OpenSSL cipher selection procedure failed because the
9392 temporary keys were not overtaken from the context and the API provided
9393 no way to reconfigure them.
9394 The new functions now let applications reconfigure the stuff and they
9395 are in detail: SSL_need_tmp_RSA, SSL_set_tmp_rsa, SSL_set_tmp_dh,
9396 SSL_set_tmp_rsa_callback and SSL_set_tmp_dh_callback. Additionally a new
9397 non-public-API function ssl_cert_instantiate() is used as a helper
9398 function and also to reduce code redundancy inside ssl_rsa.c.
9399 [Ralf S. Engelschall]
9400
9401 *) Move s_server -dcert and -dkey options out of the undocumented feature
9402 area because they are useful for the DSA situation and should be
9403 recognized by the users.
9404 [Ralf S. Engelschall]
9405
9406 *) Fix the cipher decision scheme for export ciphers: the export bits are
9407 *not* within SSL_MKEY_MASK or SSL_AUTH_MASK, they are within
9408 SSL_EXP_MASK. So, the original variable has to be used instead of the
9409 already masked variable.
9410 [Richard Levitte <levitte@stacken.kth.se>]
9411
9412 *) Fix 'port' variable from `int' to `unsigned int' in crypto/bio/b_sock.c
9413 [Richard Levitte <levitte@stacken.kth.se>]
9414
9415 *) Change type of another md_len variable in pk7_doit.c:PKCS7_dataFinal()
9416 from `int' to `unsigned int' because it's a length and initialized by
9417 EVP_DigestFinal() which expects an `unsigned int *'.
9418 [Richard Levitte <levitte@stacken.kth.se>]
9419
9420 *) Don't hard-code path to Perl interpreter on shebang line of Configure
9421 script. Instead use the usual Shell->Perl transition trick.
9422 [Ralf S. Engelschall]
9423
9424 *) Make `openssl x509 -noout -modulus' functional also for DSA certificates
9425 (in addition to RSA certificates) to match the behaviour of `openssl dsa
9426 -noout -modulus' as it's already the case for `openssl rsa -noout
9427 -modulus'. For RSA the -modulus is the real "modulus" while for DSA
9428 currently the public key is printed (a decision which was already done by
9429 `openssl dsa -modulus' in the past) which serves a similar purpose.
9430 Additionally the NO_RSA no longer completely removes the whole -modulus
9431 option; it now only avoids using the RSA stuff. Same applies to NO_DSA
9432 now, too.
9433 [Ralf S. Engelschall]
9434
9435 *) Add Arne Ansper's reliable BIO - this is an encrypted, block-digested
9436 BIO. See the source (crypto/evp/bio_ok.c) for more info.
9437 [Arne Ansper <arne@ats.cyber.ee>]
9438
9439 *) Dump the old yucky req code that tried (and failed) to allow raw OIDs
9440 to be added. Now both 'req' and 'ca' can use new objects defined in the
9441 config file.
9442 [Steve Henson]
9443
9444 *) Add cool BIO that does syslog (or event log on NT).
9445 [Arne Ansper <arne@ats.cyber.ee>, integrated by Ben Laurie]
9446
9447 *) Add support for new TLS ciphersuites, TLS_RSA_EXPORT56_WITH_RC4_56_MD5,
9448 TLS_RSA_EXPORT56_WITH_RC2_CBC_56_MD5 and
9449 TLS_RSA_EXPORT56_WITH_DES_CBC_SHA, as specified in "56-bit Export Cipher
9450 Suites For TLS", draft-ietf-tls-56-bit-ciphersuites-00.txt.
9451 [Ben Laurie]
9452
9453 *) Add preliminary config info for new extension code.
9454 [Steve Henson]
9455
9456 *) Make RSA_NO_PADDING really use no padding.
9457 [Ulf Moeller <ulf@fitug.de>]
9458
9459 *) Generate errors when private/public key check is done.
9460 [Ben Laurie]
9461
9462 *) Overhaul for 'crl' utility. New function X509_CRL_print. Partial support
9463 for some CRL extensions and new objects added.
9464 [Steve Henson]
9465
9466 *) Really fix the ASN1 IMPLICIT bug this time... Partial support for private
9467 key usage extension and fuller support for authority key id.
9468 [Steve Henson]
9469
9470 *) Add OAEP encryption for the OpenSSL crypto library. OAEP is the improved
9471 padding method for RSA, which is recommended for new applications in PKCS
9472 #1 v2.0 (RFC 2437, October 1998).
9473 OAEP (Optimal Asymmetric Encryption Padding) has better theoretical
9474 foundations than the ad-hoc padding used in PKCS #1 v1.5. It is secure
9475 against Bleichbacher's attack on RSA.
9476 [Ulf Moeller <ulf@fitug.de>, reformatted, corrected and integrated by
9477 Ben Laurie]
9478
9479 *) Updates to the new SSL compression code
9480 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)]
9481
9482 *) Fix so that the version number in the master secret, when passed
9483 via RSA, checks that if TLS was proposed, but we roll back to SSLv3
9484 (because the server will not accept higher), that the version number
9485 is 0x03,0x01, not 0x03,0x00
9486 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)]
9487
9488 *) Run extensive memory leak checks on SSL apps. Fixed *lots* of memory
9489 leaks in ssl/ relating to new X509_get_pubkey() behaviour. Also fixes
9490 in apps/ and an unrelated leak in crypto/dsa/dsa_vrf.c
9491 [Steve Henson]
9492
9493 *) Support for RAW extensions where an arbitrary extension can be
9494 created by including its DER encoding. See apps/openssl.cnf for
9495 an example.
9496 [Steve Henson]
9497
9498 *) Make sure latest Perl versions don't interpret some generated C array
9499 code as Perl array code in the crypto/err/err_genc.pl script.
9500 [Lars Weber <3weber@informatik.uni-hamburg.de>]
9501
9502 *) Modify ms/do_ms.bat to not generate assembly language makefiles since
9503 not many people have the assembler. Various Win32 compilation fixes and
9504 update to the INSTALL.W32 file with (hopefully) more accurate Win32
9505 build instructions.
9506 [Steve Henson]
9507
9508 *) Modify configure script 'Configure' to automatically create crypto/date.h
9509 file under Win32 and also build pem.h from pem.org. New script
9510 util/mkfiles.pl to create the MINFO file on environments that can't do a
9511 'make files': perl util/mkfiles.pl >MINFO should work.
9512 [Steve Henson]
9513
9514 *) Major rework of DES function declarations, in the pursuit of correctness
9515 and purity. As a result, many evil casts evaporated, and some weirdness,
9516 too. You may find this causes warnings in your code. Zapping your evil
9517 casts will probably fix them. Mostly.
9518 [Ben Laurie]
9519
9520 *) Fix for a typo in asn1.h. Bug fix to object creation script
9521 obj_dat.pl. It considered a zero in an object definition to mean
9522 "end of object": none of the objects in objects.h have any zeros
9523 so it wasn't spotted.
9524 [Steve Henson, reported by Erwann ABALEA <eabalea@certplus.com>]
9525
9526 *) Add support for Triple DES Cipher Block Chaining with Output Feedback
9527 Masking (CBCM). In the absence of test vectors, the best I have been able
9528 to do is check that the decrypt undoes the encrypt, so far. Send me test
9529 vectors if you have them.
9530 [Ben Laurie]
9531
9532 *) Correct calculation of key length for export ciphers (too much space was
9533 allocated for null ciphers). This has not been tested!
9534 [Ben Laurie]
9535
9536 *) Modifications to the mkdef.pl for Win32 DEF file creation. The usage
9537 message is now correct (it understands "crypto" and "ssl" on its
9538 command line). There is also now an "update" option. This will update
9539 the util/ssleay.num and util/libeay.num files with any new functions.
9540 If you do a:
9541 perl util/mkdef.pl crypto ssl update
9542 it will update them.
9543 [Steve Henson]
9544
9545 *) Overhauled the Perl interface (perl/*):
9546 - ported BN stuff to OpenSSL's different BN library
9547 - made the perl/ source tree CVS-aware
9548 - renamed the package from SSLeay to OpenSSL (the files still contain
9549 their history because I've copied them in the repository)
9550 - removed obsolete files (the test scripts will be replaced
9551 by better Test::Harness variants in the future)
9552 [Ralf S. Engelschall]
9553
9554 *) First cut for a very conservative source tree cleanup:
9555 1. merge various obsolete readme texts into doc/ssleay.txt
9556 where we collect the old documents and readme texts.
9557 2. remove the first part of files where I'm already sure that we no
9558 longer need them because of three reasons: either they are just temporary
9559 files which were left by Eric or they are preserved original files where
9560 I've verified that the diff is also available in the CVS via "cvs diff
9561 -rSSLeay_0_8_1b" or they were renamed (as it was definitely the case for
9562 the crypto/md/ stuff).
9563 [Ralf S. Engelschall]
9564
9565 *) More extension code. Incomplete support for subject and issuer alt
9566 name, issuer and authority key id. Change the i2v function parameters
9567 and add an extra 'crl' parameter in the X509V3_CTX structure: guess
9568 what that's for :-) Fix to ASN1 macro which messed up
9569 IMPLICIT tag and add f_enum.c which adds a2i, i2a for ENUMERATED.
9570 [Steve Henson]
9571
9572 *) Preliminary support for ENUMERATED type. This is largely copied from the
9573 INTEGER code.
9574 [Steve Henson]
9575
9576 *) Add new function, EVP_MD_CTX_copy() to replace frequent use of memcpy.
9577 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)]
9578
9579 *) Make sure `make rehash' target really finds the `openssl' program.
9580 [Ralf S. Engelschall, Matthias Loepfe <Matthias.Loepfe@adnovum.ch>]
9581
9582 *) Squeeze another 7% of speed out of MD5 assembler, at least on a P2. I'd
9583 like to hear about it if this slows down other processors.
9584 [Ben Laurie]
9585
9586 *) Add CygWin32 platform information to Configure script.
9587 [Alan Batie <batie@aahz.jf.intel.com>]
9588
9589 *) Fixed ms/32all.bat script: `no_asm' -> `no-asm'
9590 [Rainer W. Gerling <gerling@mpg-gv.mpg.de>]
9591
9592 *) New program nseq to manipulate netscape certificate sequences
9593 [Steve Henson]
9594
9595 *) Modify crl2pkcs7 so it supports multiple -certfile arguments. Fix a
9596 few typos.
9597 [Steve Henson]
9598
9599 *) Fixes to BN code. Previously the default was to define BN_RECURSION
9600 but the BN code had some problems that would cause failures when
9601 doing certificate verification and some other functions.
9602 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)]
9603
9604 *) Add ASN1 and PEM code to support netscape certificate sequences.
9605 [Steve Henson]
9606
9607 *) Add ASN1 and PEM code to support netscape certificate sequences.
9608 [Steve Henson]
9609
9610 *) Add several PKIX and private extended key usage OIDs.
9611 [Steve Henson]
9612
9613 *) Modify the 'ca' program to handle the new extension code. Modify
9614 openssl.cnf for new extension format, add comments.
9615 [Steve Henson]
9616
9617 *) More X509 V3 changes. Fix typo in v3_bitstr.c. Add support to 'req'
9618 and add a sample to openssl.cnf so req -x509 now adds appropriate
9619 CA extensions.
9620 [Steve Henson]
9621
9622 *) Continued X509 V3 changes. Add to other makefiles, integrate with the
9623 error code, add initial support to X509_print() and x509 application.
9624 [Steve Henson]
9625
9626 *) Takes a deep breath and start addding X509 V3 extension support code. Add
9627 files in crypto/x509v3. Move original stuff to crypto/x509v3/old. All this
9628 stuff is currently isolated and isn't even compiled yet.
9629 [Steve Henson]
9630
9631 *) Continuing patches for GeneralizedTime. Fix up certificate and CRL
9632 ASN1 to use ASN1_TIME and modify print routines to use ASN1_TIME_print.
9633 Removed the versions check from X509 routines when loading extensions:
9634 this allows certain broken certificates that don't set the version
9635 properly to be processed.
9636 [Steve Henson]
9637
9638 *) Deal with irritating shit to do with dependencies, in YAAHW (Yet Another
9639 Ad Hoc Way) - Makefile.ssls now all contain local dependencies, which
9640 can still be regenerated with "make depend".
9641 [Ben Laurie]
9642
9643 *) Spelling mistake in C version of CAST-128.
9644 [Ben Laurie, reported by Jeremy Hylton <jeremy@cnri.reston.va.us>]
9645
9646 *) Changes to the error generation code. The perl script err-code.pl
9647 now reads in the old error codes and retains the old numbers, only
9648 adding new ones if necessary. It also only changes the .err files if new
9649 codes are added. The makefiles have been modified to only insert errors
9650 when needed (to avoid needlessly modifying header files). This is done
9651 by only inserting errors if the .err file is newer than the auto generated
9652 C file. To rebuild all the error codes from scratch (the old behaviour)
9653 either modify crypto/Makefile.ssl to pass the -regen flag to err_code.pl
9654 or delete all the .err files.
9655 [Steve Henson]
9656
9657 *) CAST-128 was incorrectly implemented for short keys. The C version has
9658 been fixed, but is untested. The assembler versions are also fixed, but
9659 new assembler HAS NOT BEEN GENERATED FOR WIN32 - the Makefile needs fixing
9660 to regenerate it if needed.
9661 [Ben Laurie, reported (with fix for C version) by Jun-ichiro itojun
9662 Hagino <itojun@kame.net>]
9663
9664 *) File was opened incorrectly in randfile.c.
9665 [Ulf Möller <ulf@fitug.de>]
9666
9667 *) Beginning of support for GeneralizedTime. d2i, i2d, check and print
9668 functions. Also ASN1_TIME suite which is a CHOICE of UTCTime or
9669 GeneralizedTime. ASN1_TIME is the proper type used in certificates et
9670 al: it's just almost always a UTCTime. Note this patch adds new error
9671 codes so do a "make errors" if there are problems.
9672 [Steve Henson]
9673
9674 *) Correct Linux 1 recognition in config.
9675 [Ulf Möller <ulf@fitug.de>]
9676
9677 *) Remove pointless MD5 hash when using DSA keys in ca.
9678 [Anonymous <nobody@replay.com>]
9679
9680 *) Generate an error if given an empty string as a cert directory. Also
9681 generate an error if handed NULL (previously returned 0 to indicate an
9682 error, but didn't set one).
9683 [Ben Laurie, reported by Anonymous <nobody@replay.com>]
9684
9685 *) Add prototypes to SSL methods. Make SSL_write's buffer const, at last.
9686 [Ben Laurie]
9687
9688 *) Fix the dummy function BN_ref_mod_exp() in rsaref.c to have the correct
9689 parameters. This was causing a warning which killed off the Win32 compile.
9690 [Steve Henson]
9691
9692 *) Remove C++ style comments from crypto/bn/bn_local.h.
9693 [Neil Costigan <neil.costigan@celocom.com>]
9694
9695 *) The function OBJ_txt2nid was broken. It was supposed to return a nid
9696 based on a text string, looking up short and long names and finally
9697 "dot" format. The "dot" format stuff didn't work. Added new function
9698 OBJ_txt2obj to do the same but return an ASN1_OBJECT and rewrote
9699 OBJ_txt2nid to use it. OBJ_txt2obj can also return objects even if the
9700 OID is not part of the table.
9701 [Steve Henson]
9702
9703 *) Add prototypes to X509 lookup/verify methods, fixing a bug in
9704 X509_LOOKUP_by_alias().
9705 [Ben Laurie]
9706
9707 *) Sort openssl functions by name.
9708 [Ben Laurie]
9709
9710 *) Get the gendsa program working (hopefully) and add it to app list. Remove
9711 encryption from sample DSA keys (in case anyone is interested the password
9712 was "1234").
9713 [Steve Henson]
9714
9715 *) Make _all_ *_free functions accept a NULL pointer.
9716 [Frans Heymans <fheymans@isaserver.be>]
9717
9718 *) If a DH key is generated in s3_srvr.c, don't blow it by trying to use
9719 NULL pointers.
9720 [Anonymous <nobody@replay.com>]
9721
9722 *) s_server should send the CAfile as acceptable CAs, not its own cert.
9723 [Bodo Moeller <3moeller@informatik.uni-hamburg.de>]
9724
9725 *) Don't blow it for numeric -newkey arguments to apps/req.
9726 [Bodo Moeller <3moeller@informatik.uni-hamburg.de>]
9727
9728 *) Temp key "for export" tests were wrong in s3_srvr.c.
9729 [Anonymous <nobody@replay.com>]
9730
9731 *) Add prototype for temp key callback functions
9732 SSL_CTX_set_tmp_{rsa,dh}_callback().
9733 [Ben Laurie]
9734
9735 *) Make DH_free() tolerate being passed a NULL pointer (like RSA_free() and
9736 DSA_free()). Make X509_PUBKEY_set() check for errors in d2i_PublicKey().
9737 [Steve Henson]
9738
9739 *) X509_name_add_entry() freed the wrong thing after an error.
9740 [Arne Ansper <arne@ats.cyber.ee>]
9741
9742 *) rsa_eay.c would attempt to free a NULL context.
9743 [Arne Ansper <arne@ats.cyber.ee>]
9744
9745 *) BIO_s_socket() had a broken should_retry() on Windoze.
9746 [Arne Ansper <arne@ats.cyber.ee>]
9747
9748 *) BIO_f_buffer() didn't pass on BIO_CTRL_FLUSH.
9749 [Arne Ansper <arne@ats.cyber.ee>]
9750
9751 *) Make sure the already existing X509_STORE->depth variable is initialized
9752 in X509_STORE_new(), but document the fact that this variable is still
9753 unused in the certificate verification process.
9754 [Ralf S. Engelschall]
9755
9756 *) Fix the various library and apps files to free up pkeys obtained from
9757 X509_PUBKEY_get() et al. Also allow x509.c to handle netscape extensions.
9758 [Steve Henson]
9759
9760 *) Fix reference counting in X509_PUBKEY_get(). This makes
9761 demos/maurice/example2.c work, amongst others, probably.
9762 [Steve Henson and Ben Laurie]
9763
9764 *) First cut of a cleanup for apps/. First the `ssleay' program is now named
9765 `openssl' and second, the shortcut symlinks for the `openssl <command>'
9766 are no longer created. This way we have a single and consistent command
9767 line interface `openssl <command>', similar to `cvs <command>'.
9768 [Ralf S. Engelschall, Paul Sutton and Ben Laurie]
9769
9770 *) ca.c: move test for DSA keys inside #ifndef NO_DSA. Make pubkey
9771 BIT STRING wrapper always have zero unused bits.
9772 [Steve Henson]
9773
9774 *) Add CA.pl, perl version of CA.sh, add extended key usage OID.
9775 [Steve Henson]
9776
9777 *) Make the top-level INSTALL documentation easier to understand.
9778 [Paul Sutton]
9779
9780 *) Makefiles updated to exit if an error occurs in a sub-directory
9781 make (including if user presses ^C) [Paul Sutton]
9782
9783 *) Make Montgomery context stuff explicit in RSA data structure.
9784 [Ben Laurie]
9785
9786 *) Fix build order of pem and err to allow for generated pem.h.
9787 [Ben Laurie]
9788
9789 *) Fix renumbering bug in X509_NAME_delete_entry().
9790 [Ben Laurie]
9791
9792 *) Enhanced the err-ins.pl script so it makes the error library number
9793 global and can add a library name. This is needed for external ASN1 and
9794 other error libraries.
9795 [Steve Henson]
9796
9797 *) Fixed sk_insert which never worked properly.
9798 [Steve Henson]
9799
9800 *) Fix ASN1 macros so they can handle indefinite length construted
9801 EXPLICIT tags. Some non standard certificates use these: they can now
9802 be read in.
9803 [Steve Henson]
9804
9805 *) Merged the various old/obsolete SSLeay documentation files (doc/xxx.doc)
9806 into a single doc/ssleay.txt bundle. This way the information is still
9807 preserved but no longer messes up this directory. Now it's new room for
9808 the new set of documenation files.
9809 [Ralf S. Engelschall]
9810
9811 *) SETs were incorrectly DER encoded. This was a major pain, because they
9812 shared code with SEQUENCEs, which aren't coded the same. This means that
9813 almost everything to do with SETs or SEQUENCEs has either changed name or
9814 number of arguments.
9815 [Ben Laurie, based on a partial fix by GP Jayan <gp@nsj.co.jp>]
9816
9817 *) Fix test data to work with the above.
9818 [Ben Laurie]
9819
9820 *) Fix the RSA header declarations that hid a bug I fixed in 0.9.0b but
9821 was already fixed by Eric for 0.9.1 it seems.
9822 [Ben Laurie - pointed out by Ulf Möller <ulf@fitug.de>]
9823
9824 *) Autodetect FreeBSD3.
9825 [Ben Laurie]
9826
9827 *) Fix various bugs in Configure. This affects the following platforms:
9828 nextstep
9829 ncr-scde
9830 unixware-2.0
9831 unixware-2.0-pentium
9832 sco5-cc.
9833 [Ben Laurie]
9834
9835 *) Eliminate generated files from CVS. Reorder tests to regenerate files
9836 before they are needed.
9837 [Ben Laurie]
9838
9839 *) Generate Makefile.ssl from Makefile.org (to keep CVS happy).
9840 [Ben Laurie]
9841
9842
9843 Changes between 0.9.1b and 0.9.1c [23-Dec-1998]
9844
9845 *) Added OPENSSL_VERSION_NUMBER to crypto/crypto.h and
9846 changed SSLeay to OpenSSL in version strings.
9847 [Ralf S. Engelschall]
9848
9849 *) Some fixups to the top-level documents.
9850 [Paul Sutton]
9851
9852 *) Fixed the nasty bug where rsaref.h was not found under compile-time
9853 because the symlink to include/ was missing.
9854 [Ralf S. Engelschall]
9855
9856 *) Incorporated the popular no-RSA/DSA-only patches
9857 which allow to compile a RSA-free SSLeay.
9858 [Andrew Cooke / Interrader Ldt., Ralf S. Engelschall]
9859
9860 *) Fixed nasty rehash problem under `make -f Makefile.ssl links'
9861 when "ssleay" is still not found.
9862 [Ralf S. Engelschall]
9863
9864 *) Added more platforms to Configure: Cray T3E, HPUX 11,
9865 [Ralf S. Engelschall, Beckmann <beckman@acl.lanl.gov>]
9866
9867 *) Updated the README file.
9868 [Ralf S. Engelschall]
9869
9870 *) Added various .cvsignore files in the CVS repository subdirs
9871 to make a "cvs update" really silent.
9872 [Ralf S. Engelschall]
9873
9874 *) Recompiled the error-definition header files and added
9875 missing symbols to the Win32 linker tables.
9876 [Ralf S. Engelschall]
9877
9878 *) Cleaned up the top-level documents;
9879 o new files: CHANGES and LICENSE
9880 o merged VERSION, HISTORY* and README* files a CHANGES.SSLeay
9881 o merged COPYRIGHT into LICENSE
9882 o removed obsolete TODO file
9883 o renamed MICROSOFT to INSTALL.W32
9884 [Ralf S. Engelschall]
9885
9886 *) Removed dummy files from the 0.9.1b source tree:
9887 crypto/asn1/x crypto/bio/cd crypto/bio/fg crypto/bio/grep crypto/bio/vi
9888 crypto/bn/asm/......add.c crypto/bn/asm/a.out crypto/dsa/f crypto/md5/f
9889 crypto/pem/gmon.out crypto/perlasm/f crypto/pkcs7/build crypto/rsa/f
9890 crypto/sha/asm/f crypto/threads/f ms/zzz ssl/f ssl/f.mak test/f
9891 util/f.mak util/pl/f util/pl/f.mak crypto/bf/bf_locl.old apps/f
9892 [Ralf S. Engelschall]
9893
9894 *) Added various platform portability fixes.
9895 [Mark J. Cox]
9896
9897 *) The Genesis of the OpenSSL rpject:
9898 We start with the latest (unreleased) SSLeay version 0.9.1b which Eric A.
9899 Young and Tim J. Hudson created while they were working for C2Net until
9900 summer 1998.
9901 [The OpenSSL Project]
9902
9903
9904 Changes between 0.9.0b and 0.9.1b [not released]
9905
9906 *) Updated a few CA certificates under certs/
9907 [Eric A. Young]
9908
9909 *) Changed some BIGNUM api stuff.
9910 [Eric A. Young]
9911
9912 *) Various platform ports: OpenBSD, Ultrix, IRIX 64bit, NetBSD,
9913 DGUX x86, Linux Alpha, etc.
9914 [Eric A. Young]
9915
9916 *) New COMP library [crypto/comp/] for SSL Record Layer Compression:
9917 RLE (dummy implemented) and ZLIB (really implemented when ZLIB is
9918 available).
9919 [Eric A. Young]
9920
9921 *) Add -strparse option to asn1pars program which parses nested
9922 binary structures
9923 [Dr Stephen Henson <shenson@bigfoot.com>]
9924
9925 *) Added "oid_file" to ssleay.cnf for "ca" and "req" programs.
9926 [Eric A. Young]
9927
9928 *) DSA fix for "ca" program.
9929 [Eric A. Young]
9930
9931 *) Added "-genkey" option to "dsaparam" program.
9932 [Eric A. Young]
9933
9934 *) Added RIPE MD160 (rmd160) message digest.
9935 [Eric A. Young]
9936
9937 *) Added -a (all) option to "ssleay version" command.
9938 [Eric A. Young]
9939
9940 *) Added PLATFORM define which is the id given to Configure.
9941 [Eric A. Young]
9942
9943 *) Added MemCheck_XXXX functions to crypto/mem.c for memory checking.
9944 [Eric A. Young]
9945
9946 *) Extended the ASN.1 parser routines.
9947 [Eric A. Young]
9948
9949 *) Extended BIO routines to support REUSEADDR, seek, tell, etc.
9950 [Eric A. Young]
9951
9952 *) Added a BN_CTX to the BN library.
9953 [Eric A. Young]
9954
9955 *) Fixed the weak key values in DES library
9956 [Eric A. Young]
9957
9958 *) Changed API in EVP library for cipher aliases.
9959 [Eric A. Young]
9960
9961 *) Added support for RC2/64bit cipher.
9962 [Eric A. Young]
9963
9964 *) Converted the lhash library to the crypto/mem.c functions.
9965 [Eric A. Young]
9966
9967 *) Added more recognized ASN.1 object ids.
9968 [Eric A. Young]
9969
9970 *) Added more RSA padding checks for SSL/TLS.
9971 [Eric A. Young]
9972
9973 *) Added BIO proxy/filter functionality.
9974 [Eric A. Young]
9975
9976 *) Added extra_certs to SSL_CTX which can be used
9977 send extra CA certificates to the client in the CA cert chain sending
9978 process. It can be configured with SSL_CTX_add_extra_chain_cert().
9979 [Eric A. Young]
9980
9981 *) Now Fortezza is denied in the authentication phase because
9982 this is key exchange mechanism is not supported by SSLeay at all.
9983 [Eric A. Young]
9984
9985 *) Additional PKCS1 checks.
9986 [Eric A. Young]
9987
9988 *) Support the string "TLSv1" for all TLS v1 ciphers.
9989 [Eric A. Young]
9990
9991 *) Added function SSL_get_ex_data_X509_STORE_CTX_idx() which gives the
9992 ex_data index of the SSL context in the X509_STORE_CTX ex_data.
9993 [Eric A. Young]
9994
9995 *) Fixed a few memory leaks.
9996 [Eric A. Young]
9997
9998 *) Fixed various code and comment typos.
9999 [Eric A. Young]
10000
10001 *) A minor bug in ssl/s3_clnt.c where there would always be 4 0
10002 bytes sent in the client random.
10003 [Edward Bishop <ebishop@spyglass.com>]
10004