5 Changes between 0.9.7c and 0.9.8 [xx XXX xxxx]
7 *) bn_dup_expand() has been deprecated, it was introduced in 0.9.7 and
8 remained unused and not that useful. A variety of other little bignum
9 tweaks and fixes have also been made continuing on from the audit (see
13 *) Constify all or almost all d2i, c2i, s2i and r2i functions, along with
14 associated ASN1, EVP and SSL functions and old ASN1 macros.
17 *) BN_zero() only needs to set 'top' and 'neg' to zero for correct results,
18 and this should never fail. So the return value from the use of
19 BN_set_word() (which can fail due to needless expansion) is now deprecated;
20 if OPENSSL_NO_DEPRECATED is defined, BN_zero() is a void macro.
23 *) BN_CTX_get() should return zero-valued bignums, providing the same
24 initialised value as BN_new().
25 [Geoff Thorpe, suggested by Ulf Möller]
27 *) Support for inhibitAnyPolicy certificate extension.
30 *) An audit of the BIGNUM code is underway, for which debugging code is
31 enabled when BN_DEBUG is defined. This makes stricter enforcements on what
32 is considered valid when processing BIGNUMs, and causes execution to
33 assert() when a problem is discovered. If BN_DEBUG_RAND is defined,
34 further steps are taken to deliberately pollute unused data in BIGNUM
35 structures to try and expose faulty code further on. For now, openssl will
36 (in its default mode of operation) continue to tolerate the inconsistent
37 forms that it has tolerated in the past, but authors and packagers should
38 consider trying openssl and their own applications when compiled with
39 these debugging symbols defined. It will help highlight potential bugs in
40 their own code, and will improve the test coverage for OpenSSL itself. At
41 some point, these tighter rules will become openssl's default to improve
42 maintainability, though the assert()s and other overheads will remain only
43 in debugging configurations. See bn.h for more details.
44 [Geoff Thorpe, Nils Larsch, Ulf Möller]
46 *) BN_CTX_init() has been deprecated, as BN_CTX is an opaque structure
47 that can only be obtained through BN_CTX_new() (which implicitly
48 initialises it). The presence of this function only made it possible
49 to overwrite an existing structure (and cause memory leaks).
52 *) Because of the callback-based approach for implementing LHASH as a
53 template type, lh_insert() adds opaque objects to hash-tables and
54 lh_doall() or lh_doall_arg() are typically used with a destructor callback
55 to clean up those corresponding objects before destroying the hash table
56 (and losing the object pointers). So some over-zealous constifications in
57 LHASH have been relaxed so that lh_insert() does not take (nor store) the
58 objects as "const" and the lh_doall[_arg] callback wrappers are not
59 prototyped to have "const" restrictions on the object pointers they are
60 given (and so aren't required to cast them away any more).
63 *) The tmdiff.h API was so ugly and minimal that our own timing utility
64 (speed) prefers to use its own implementation. The two implementations
65 haven't been consolidated as yet (volunteers?) but the tmdiff API has had
66 its object type properly exposed (MS_TM) instead of casting to/from "char
67 *". This may still change yet if someone realises MS_TM and "ms_time_***"
68 aren't necessarily the greatest nomenclatures - but this is what was used
69 internally to the implementation so I've used that for now.
72 *) Ensure that deprecated functions do not get compiled when
73 OPENSSL_NO_DEPRECATED is defined. Some "openssl" subcommands and a few of
74 the self-tests were still using deprecated key-generation functions so
75 these have been updated also.
78 *) Reorganise PKCS#7 code to separate the digest location functionality
79 into PKCS7_find_digest(), digest addtion into PKCS7_bio_add_digest().
80 New function PKCS7_set_digest() to set the digest type for PKCS#7
81 digestedData type. Add additional code to correctly generate the
82 digestedData type and add support for this type in PKCS7 initialization
86 *) New function PKCS7_set0_type_other() this initializes a PKCS7
87 structure of type "other".
90 *) Fix prime generation loop in crypto/bn/bn_prime.pl by making
91 sure the loop does correctly stop and breaking ("division by zero")
92 modulus operations are not performed. The (pre-generated) prime
93 table crypto/bn/bn_prime.h was already correct, but it could not be
94 re-generated on some platforms because of the "division by zero"
95 situation in the script.
98 *) Update support for ECC-based TLS ciphersuites according to
99 draft-ietf-tls-ecc-03.txt: the KDF1 key derivation function with
100 SHA-1 now is only used for "small" curves (where the
101 representation of a field element takes up to 24 bytes); for
102 larger curves, the field element resulting from ECDH is directly
103 used as premaster secret.
104 [Douglas Stebila (Sun Microsystems Laboratories)]
106 *) Add code for kP+lQ timings to crypto/ec/ectest.c, and add SEC2
107 curve secp160r1 to the tests.
108 [Douglas Stebila (Sun Microsystems Laboratories)]
110 *) Add the possibility to load symbols globally with DSO.
111 [Götz Babin-Ebell <babin-ebell@trustcenter.de> via Richard Levitte]
113 *) Add the functions ERR_set_mark() and ERR_pop_to_mark() for better
114 control of the error stack.
117 *) Add support for STORE in ENGINE.
120 *) Add the STORE type. The intention is to provide a common interface
121 to certificate and key stores, be they simple file-based stores, or
122 HSM-type store, or LDAP stores, or...
123 NOTE: The code is currently UNTESTED and isn't really used anywhere.
126 *) Add a generic structure called OPENSSL_ITEM. This can be used to
127 pass a list of arguments to any function as well as provide a way
128 for a function to pass data back to the caller.
131 *) Add the functions BUF_strndup() and BUF_memdup(). BUF_strndup()
132 works like BUF_strdup() but can be used to duplicate a portion of
133 a string. The copy gets NUL-terminated. BUF_memdup() duplicates
137 *) Add the function sk_find_ex() which works like sk_find(), but will
138 return an index to an element even if an exact match couldn't be
139 found. The index is guaranteed to point at the element where the
140 searched-for key would be inserted to preserve sorting order.
143 *) Add the function OBJ_bsearch_ex() which works like OBJ_bsearch() but
144 takes an extra flags argument for optional functionality. Currently,
145 the following flags are defined:
147 OBJ_BSEARCH_VALUE_ON_NOMATCH
148 This one gets OBJ_bsearch_ex() to return a pointer to the first
149 element where the comparing function returns a negative or zero
152 OBJ_BSEARCH_FIRST_VALUE_ON_MATCH
153 This one gets OBJ_bsearch_ex() to return a pointer to the first
154 element where the comparing function returns zero. This is useful
155 if there are more than one element where the comparing function
159 *) Make it possible to create self-signed certificates with 'openssl ca'
160 in such a way that the self-signed certificate becomes part of the
161 CA database and uses the same mechanisms for serial number generation
162 as all other certificate signing. The new flag '-selfsign' enables
163 this functionality. Adapt CA.sh and CA.pl.in.
166 *) Add functionality to check the public key of a certificate request
167 against a given private. This is useful to check that a certificate
168 request can be signed by that key (self-signing).
171 *) Make it possible to have multiple active certificates with the same
172 subject in the CA index file. This is done only if the keyword
173 'unique_subject' is set to 'no' in the main CA section (default
174 if 'CA_default') of the configuration file. The value is saved
175 with the database itself in a separate index attribute file,
176 named like the index file with '.attr' appended to the name.
179 *) Generate muti valued AVAs using '+' notation in config files for
183 *) Support for nameConstraints certificate extension.
186 *) Support for policyConstraints certificate extension.
189 *) Support for policyMappings certificate extension.
192 *) Fixed a typo bug that would cause ENGINE_set_default() to set an
193 ENGINE as defaults for all supported algorithms irrespective of
194 the 'flags' parameter. 'flags' is now honoured, so applications
195 should make sure they are passing it correctly.
198 *) Make sure the default DSA_METHOD implementation only uses its
199 dsa_mod_exp() and/or bn_mod_exp() handlers if they are non-NULL,
200 and change its own handlers to be NULL so as to remove unnecessary
201 indirection. This lets alternative implementations fallback to the
202 default implementation more easily.
205 *) Support for directoryName in GeneralName related extensions
209 *) Make it possible to link applications using Makefile.shared.
210 Make that possible even when linking against static libraries!
213 *) Support for single pass processing for S/MIME signing. This now
214 means that S/MIME signing can be done from a pipe, in addition
215 cleartext signing (multipart/signed type) is effectively streaming
216 and the signed data does not need to be all held in memory.
218 This is done with a new flag PKCS7_STREAM. When this flag is set
219 PKCS7_sign() only initializes the PKCS7 structure and the actual signing
220 is done after the data is output (and digests calculated) in
224 *) Add full support for -rpath/-R, both in shared libraries and
225 applications, at least on the platforms where it's known how
229 *) In crypto/ec/ec_mult.c, implement fast point multiplication with
230 precomputation, based on wNAF splitting: EC_GROUP_precompute_mult()
231 will now compute a table of multiples of the generator that
232 makes subsequent invocations of EC_POINTs_mul() or EC_POINT_mul()
233 faster (notably in the case of a single point multiplication,
235 [Nils Larsch, Bodo Moeller]
237 *) IPv6 support for certificate extensions. The various extensions
238 which use the IP:a.b.c.d can now take IPv6 addresses using the
239 formats of RFC1884 2.2 . IPv6 addresses are now also displayed
243 *) Added an ENGINE that implements RSA by performing private key
244 exponentiations with the GMP library. The conversions to and from
245 GMP's mpz_t format aren't optimised nor are any montgomery forms
246 cached, and on x86 it appears OpenSSL's own performance has caught up.
247 However there are likely to be other architectures where GMP could
248 provide a boost. This ENGINE is not built in by default, but it can be
249 specified at Configure time and should be accompanied by the necessary
250 linker additions, eg;
251 ./config -DOPENSSL_USE_GMP -lgmp
254 *) "openssl engine" will not display ENGINE/DSO load failure errors when
255 testing availability of engines with "-t" - the old behaviour is
256 produced by increasing the feature's verbosity with "-tt".
259 *) ECDSA routines: under certain error conditions uninitialized BN objects
260 could be freed. Solution: make sure initialization is performed early
261 enough. (Reported and fix supplied by Nils Larsch <nla@trustcenter.de>
265 *) Key-generation can now be implemented in RSA_METHOD, DSA_METHOD
266 and DH_METHOD (eg. by ENGINE implementations) to override the normal
267 software implementations. For DSA and DH, parameter generation can
268 also be overriden by providing the appropriate method callbacks.
271 *) Change the "progress" mechanism used in key-generation and
272 primality testing to functions that take a new BN_GENCB pointer in
273 place of callback/argument pairs. The new API functions have "_ex"
274 postfixes and the older functions are reimplemented as wrappers for
275 the new ones. The OPENSSL_NO_DEPRECATED symbol can be used to hide
276 declarations of the old functions to help (graceful) attempts to
277 migrate to the new functions. Also, the new key-generation API
278 functions operate on a caller-supplied key-structure and return
279 success/failure rather than returning a key or NULL - this is to
280 help make "keygen" another member function of RSA_METHOD etc.
282 Example for using the new callback interface:
284 int (*my_callback)(int a, int b, BN_GENCB *cb) = ...;
288 BN_GENCB_set(&my_cb, my_callback, my_arg);
290 return BN_is_prime_ex(some_bignum, BN_prime_checks, NULL, &cb);
291 /* For the meaning of a, b in calls to my_callback(), see the
292 * documentation of the function that calls the callback.
293 * cb will point to my_cb; my_arg can be retrieved as cb->arg.
294 * my_callback should return 1 if it wants BN_is_prime_ex()
295 * to continue, or 0 to stop.
300 *) Change the ZLIB compression method to be stateful, and make it
301 available to TLS with the number defined in
302 draft-ietf-tls-compression-04.txt.
305 *) Add the ASN.1 structures and functions for CertificatePair, which
306 is defined as follows (according to X.509_4thEditionDraftV6.pdf):
308 CertificatePair ::= SEQUENCE {
309 forward [0] Certificate OPTIONAL,
310 reverse [1] Certificate OPTIONAL,
311 -- at least one of the pair shall be present -- }
313 Also implement the PEM functions to read and write certificate
314 pairs, and defined the PEM tag as "CERTIFICATE PAIR".
316 This needed to be defined, mostly for the sake of the LDAP
317 attribute crossCertificatePair, but may prove useful elsewhere as
321 *) Make it possible to inhibit symlinking of shared libraries in
322 Makefile.shared, for Cygwin's sake.
325 *) Extend the BIGNUM API by creating new macros that behave like
328 void BN_set_sign(BIGNUM *a, int neg);
329 int BN_get_sign(const BIGNUM *a);
331 and avoid the need to access 'a->neg' directly in applications.
332 [Nils Larsch <nla@trustcenter.de>]
334 *) Implement fast modular reduction for pseudo-Mersenne primes
335 used in NIST curves (crypto/bn/bn_nist.c, crypto/ec/ecp_nist.c).
336 EC_GROUP_new_curve_GFp() will now automatically use this
338 [Nils Larsch <nla@trustcenter.de>]
340 *) Add new lock type (CRYPTO_LOCK_BN).
343 *) Change the ENGINE framework to automatically load engines
344 dynamically from specific directories unless they could be
345 found to already be built in or loaded. Move all the
346 current engines except for the cryptodev one to a new
348 The engines in engines/ are built as shared libraries if
349 the "shared" options was given to ./Configure or ./config.
350 Otherwise, they are inserted in libcrypto.a.
351 /usr/local/ssl/engines is the default directory for dynamic
352 engines, but that can be overriden at configure time through
353 the usual use of --prefix and/or --openssldir, and at run
354 time with the environment variable OPENSSL_ENGINES.
355 [Geoff Thorpe and Richard Levitte]
357 *) Add Makefile.shared, a helper makefile to build shared
358 libraries. Addapt Makefile.org.
361 *) Add version info to Win32 DLLs.
362 [Peter 'Luna' Runestig" <peter@runestig.com>]
364 *) Add new 'medium level' PKCS#12 API. Certificates and keys
365 can be added using this API to created arbitrary PKCS#12
366 files while avoiding the low level API.
368 New options to PKCS12_create(), key or cert can be NULL and
369 will then be omitted from the output file. The encryption
370 algorithm NIDs can be set to -1 for no encryption, the mac
371 iteration count can be set to 0 to omit the mac.
373 Enhance pkcs12 utility by making the -nokeys and -nocerts
374 options work when creating a PKCS#12 file. New option -nomac
375 to omit the mac, NONE can be set for an encryption algorithm.
376 New code is modified to use the enhanced PKCS12_create()
377 instead of the low level API.
380 *) Extend ASN1 encoder to support indefinite length constructed
381 encoding. This can output sequences tags and octet strings in
382 this form. Modify pk7_asn1.c to support indefinite length
383 encoding. This is experimental and needs additional code to
384 be useful, such as an ASN1 bio and some enhanced streaming
387 Extend template encode functionality so that tagging is passed
388 down to the template encoder.
391 *) Let 'openssl req' fail if an argument to '-newkey' is not
392 recognized instead of using RSA as a default.
395 *) Add support for ECC-based ciphersuites from draft-ietf-tls-ecc-01.txt.
396 As these are not official, they are not included in "ALL";
397 the "ECCdraft" ciphersuite group alias can be used to select them.
398 [Vipul Gupta and Sumit Gupta (Sun Microsystems Laboratories)]
400 *) Add ECDH engine support.
401 [Nils Gura and Douglas Stebila (Sun Microsystems Laboratories)]
403 *) Add ECDH in new directory crypto/ecdh/.
404 [Douglas Stebila (Sun Microsystems Laboratories)]
406 *) Let BN_rand_range() abort with an error after 100 iterations
407 without success (which indicates a broken PRNG).
410 *) Change BN_mod_sqrt() so that it verifies that the input value
411 is really the square of the return value. (Previously,
412 BN_mod_sqrt would show GIGO behaviour.)
415 *) Add named elliptic curves over binary fields from X9.62, SECG,
416 and WAP/WTLS; add OIDs that were still missing.
418 [Sheueling Chang Shantz and Douglas Stebila
419 (Sun Microsystems Laboratories)]
421 *) Extend the EC library for elliptic curves over binary fields
422 (new files ec2_smpl.c, ec2_smpt.c, ec2_mult.c in crypto/ec/).
425 EC_GF2m_simple_method
429 EC_GROUP_new_curve_GF2m
430 EC_GROUP_set_curve_GF2m
431 EC_GROUP_get_curve_GF2m
432 EC_POINT_set_affine_coordinates_GF2m
433 EC_POINT_get_affine_coordinates_GF2m
434 EC_POINT_set_compressed_coordinates_GF2m
436 Point compression for binary fields is disabled by default for
437 patent reasons (compile with OPENSSL_EC_BIN_PT_COMP defined to
440 As binary polynomials are represented as BIGNUMs, various members
441 of the EC_GROUP and EC_POINT data structures can be shared
442 between the implementations for prime fields and binary fields;
443 the above ..._GF2m functions (except for EX_GROUP_new_curve_GF2m)
444 are essentially identical to their ..._GFp counterparts.
445 (For simplicity, the '..._GFp' prefix has been dropped from
446 various internal method names.)
448 An internal 'field_div' method (similar to 'field_mul' and
449 'field_sqr') has been added; this is used only for binary fields.
451 [Sheueling Chang Shantz and Douglas Stebila
452 (Sun Microsystems Laboratories)]
454 *) Optionally dispatch EC_POINT_mul(), EC_POINT_precompute_mult()
455 through methods ('mul', 'precompute_mult').
457 The generic implementations (now internally called 'ec_wNAF_mul'
458 and 'ec_wNAF_precomputed_mult') remain the default if these
459 methods are undefined.
461 [Sheueling Chang Shantz and Douglas Stebila
462 (Sun Microsystems Laboratories)]
464 *) New function EC_GROUP_get_degree, which is defined through
465 EC_METHOD. For curves over prime fields, this returns the bit
466 length of the modulus.
468 [Sheueling Chang Shantz and Douglas Stebila
469 (Sun Microsystems Laboratories)]
471 *) New functions EC_GROUP_dup, EC_POINT_dup.
472 (These simply call ..._new and ..._copy).
474 [Sheueling Chang Shantz and Douglas Stebila
475 (Sun Microsystems Laboratories)]
477 *) Add binary polynomial arithmetic software in crypto/bn/bn_gf2m.c.
478 Polynomials are represented as BIGNUMs (where the sign bit is not
479 used) in the following functions [macros]:
482 BN_GF2m_sub [= BN_GF2m_add]
483 BN_GF2m_mod [wrapper for BN_GF2m_mod_arr]
484 BN_GF2m_mod_mul [wrapper for BN_GF2m_mod_mul_arr]
485 BN_GF2m_mod_sqr [wrapper for BN_GF2m_mod_sqr_arr]
487 BN_GF2m_mod_exp [wrapper for BN_GF2m_mod_exp_arr]
488 BN_GF2m_mod_sqrt [wrapper for BN_GF2m_mod_sqrt_arr]
489 BN_GF2m_mod_solve_quad [wrapper for BN_GF2m_mod_solve_quad_arr]
490 BN_GF2m_cmp [= BN_ucmp]
492 (Note that only the 'mod' functions are actually for fields GF(2^m).
493 BN_GF2m_add() is misnomer, but this is for the sake of consistency.)
495 For some functions, an the irreducible polynomial defining a
496 field can be given as an 'unsigned int[]' with strictly
497 decreasing elements giving the indices of those bits that are set;
498 i.e., p[] represents the polynomial
499 f(t) = t^p[0] + t^p[1] + ... + t^p[k]
501 p[0] > p[1] > ... > p[k] = 0.
502 This applies to the following functions:
507 BN_GF2m_mod_inv_arr [wrapper for BN_GF2m_mod_inv]
508 BN_GF2m_mod_div_arr [wrapper for BN_GF2m_mod_div]
511 BN_GF2m_mod_solve_quad_arr
515 Conversion can be performed by the following functions:
520 bntest.c has additional tests for binary polynomial arithmetic.
522 Two implementations for BN_GF2m_mod_div() are available.
523 The default algorithm simply uses BN_GF2m_mod_inv() and
524 BN_GF2m_mod_mul(). The alternative algorithm is compiled in only
525 if OPENSSL_SUN_GF2M_DIV is defined (patent pending; read the
526 copyright notice in crypto/bn/bn_gf2m.c before enabling it).
528 [Sheueling Chang Shantz and Douglas Stebila
529 (Sun Microsystems Laboratories)]
531 *) Add new error code 'ERR_R_DISABLED' that can be used when some
532 functionality is disabled at compile-time.
533 [Douglas Stebila <douglas.stebila@sun.com>]
535 *) Change default behaviour of 'openssl asn1parse' so that more
536 information is visible when viewing, e.g., a certificate:
538 Modify asn1_parse2 (crypto/asn1/asn1_par.c) so that in non-'dump'
539 mode the content of non-printable OCTET STRINGs is output in a
540 style similar to INTEGERs, but with '[HEX DUMP]' prepended to
541 avoid the appearance of a printable string.
542 [Nils Larsch <nla@trustcenter.de>]
544 *) Add 'asn1_flag' and 'asn1_form' member to EC_GROUP with access
546 EC_GROUP_set_asn1_flag()
547 EC_GROUP_get_asn1_flag()
548 EC_GROUP_set_point_conversion_form()
549 EC_GROUP_get_point_conversion_form()
550 These control ASN1 encoding details:
551 - Curves (i.e., groups) are encoded explicitly unless asn1_flag
552 has been set to OPENSSL_EC_NAMED_CURVE.
553 - Points are encoded in uncompressed form by default; options for
554 asn1_for are as for point2oct, namely
555 POINT_CONVERSION_COMPRESSED
556 POINT_CONVERSION_UNCOMPRESSED
557 POINT_CONVERSION_HYBRID
559 Also add 'seed' and 'seed_len' members to EC_GROUP with access
563 EC_GROUP_get_seed_len()
564 This is used only for ASN1 purposes (so far).
565 [Nils Larsch <nla@trustcenter.de>]
567 *) Add 'field_type' member to EC_METHOD, which holds the NID
568 of the appropriate field type OID. The new function
569 EC_METHOD_get_field_type() returns this value.
570 [Nils Larsch <nla@trustcenter.de>]
577 providing useful interfaces to EC_POINT_point2oct() and
578 EC_POINT_oct2point().
579 [Nils Larsch <nla@trustcenter.de>]
581 *) Change internals of the EC library so that the functions
582 EC_GROUP_set_generator()
583 EC_GROUP_get_generator()
585 EC_GROUP_get_cofactor()
586 are implemented directly in crypto/ec/ec_lib.c and not dispatched
587 to methods, which would lead to unnecessary code duplication when
588 adding different types of curves.
589 [Nils Larsch <nla@trustcenter.de> with input by Bodo Moeller]
591 *) Implement compute_wNAF (crypto/ec/ec_mult.c) without BIGNUM
592 arithmetic, and such that modified wNAFs are generated
593 (which avoid length expansion in many cases).
596 *) Add a function EC_GROUP_check_discriminant() (defined via
597 EC_METHOD) that verifies that the curve discriminant is non-zero.
599 Add a function EC_GROUP_check() that makes some sanity tests
600 on a EC_GROUP, its generator and order. This includes
601 EC_GROUP_check_discriminant().
602 [Nils Larsch <nla@trustcenter.de>]
604 *) Add ECDSA in new directory crypto/ecdsa/.
606 Add applications 'openssl ecparam' and 'openssl ecdsa'
607 (these are based on 'openssl dsaparam' and 'openssl dsa').
609 ECDSA support is also included in various other files across the
610 library. Most notably,
611 - 'openssl req' now has a '-newkey ecdsa:file' option;
612 - EVP_PKCS82PKEY (crypto/evp/evp_pkey.c) now can handle ECDSA;
613 - X509_PUBKEY_get (crypto/asn1/x_pubkey.c) and
614 d2i_PublicKey (crypto/asn1/d2i_pu.c) have been modified to make
615 them suitable for ECDSA where domain parameters must be
616 extracted before the specific public key;
617 - ECDSA engine support has been added.
618 [Nils Larsch <nla@trustcenter.de>]
620 *) Include some named elliptic curves, and add OIDs from X9.62,
621 SECG, and WAP/WTLS. Each curve can be obtained from the new
623 EC_GROUP_new_by_nid(),
624 and the list of available named curves can be obtained with
625 EC_get_builtin_curves().
626 Also add a 'curve_name' member to EC_GROUP objects, which can be
630 [Nils Larsch <nla@trustcenter.de, Bodo Moeller]
632 *) Remove a few calls to bn_wexpand() in BN_sqr() (the one in there
633 was actually never needed) and in BN_mul(). The removal in BN_mul()
634 required a small change in bn_mul_part_recursive() and the addition
635 of the functions bn_cmp_part_words(), bn_sub_part_words() and
636 bn_add_part_words(), which do the same thing as bn_cmp_words(),
637 bn_sub_words() and bn_add_words() except they take arrays with
641 Changes between 0.9.7c and 0.9.7d [xx XXX XXXX]
643 *) X509 verify fixes. Disable broken certificate workarounds when
644 X509_V_FLAGS_X509_STRICT is set. Check CRL issuer has cRLSign set if
645 keyUsage extension present. Don't accept CRLs with unhandled critical
646 extensions: since verify currently doesn't process CRL extensions this
647 rejects a CRL with *any* critical extensions. Add new verify error codes
651 *) When creating an OCSP nonce use an OCTET STRING inside the extnValue.
652 A clarification of RFC2560 will require the use of OCTET STRINGs and
653 some implementations cannot handle the current raw format. Since OpenSSL
654 copies and compares OCSP nonces as opaque blobs without any attempt at
655 parsing them this should not create any compatibility issues.
658 *) New md flag EVP_MD_CTX_FLAG_REUSE this allows md_data to be reused when
659 calling EVP_MD_CTX_copy_ex() to avoid calling OPENSSL_malloc(). Without
660 this HMAC (and other) operations are several times slower than OpenSSL
664 *) Print out GeneralizedTime and UTCTime in ASN1_STRING_print_ex().
665 [Peter Sylvester <Peter.Sylvester@EdelWeb.fr>]
667 *) Use the correct content when signing type "other".
670 Changes between 0.9.7b and 0.9.7c [30 Sep 2003]
672 *) Fix various bugs revealed by running the NISCC test suite:
674 Stop out of bounds reads in the ASN1 code when presented with
675 invalid tags (CAN-2003-0543 and CAN-2003-0544).
677 Free up ASN1_TYPE correctly if ANY type is invalid (CAN-2003-0545).
679 If verify callback ignores invalid public key errors don't try to check
680 certificate signature with the NULL public key.
684 *) New -ignore_err option in ocsp application to stop the server
685 exiting on the first error in a request.
688 *) In ssl3_accept() (ssl/s3_srvr.c) only accept a client certificate
689 if the server requested one: as stated in TLS 1.0 and SSL 3.0
693 *) In ssl3_get_client_hello() (ssl/s3_srvr.c), tolerate additional
694 extra data after the compression methods not only for TLS 1.0
695 but also for SSL 3.0 (as required by the specification).
696 [Bodo Moeller; problem pointed out by Matthias Loepfe]
698 *) Change X509_certificate_type() to mark the key as exported/exportable
699 when it's 512 *bits* long, not 512 bytes.
702 *) Change AES_cbc_encrypt() so it outputs exact multiple of
703 blocks during encryption.
706 *) Various fixes to base64 BIO and non blocking I/O. On write
707 flushes were not handled properly if the BIO retried. On read
708 data was not being buffered properly and had various logic bugs.
709 This also affects blocking I/O when the data being decoded is a
713 *) Various S/MIME bugfixes and compatibility changes:
714 output correct application/pkcs7 MIME type if
715 PKCS7_NOOLDMIMETYPE is set. Tolerate some broken signatures.
716 Output CR+LF for EOL if PKCS7_CRLFEOL is set (this makes opening
717 of files as .eml work). Correctly handle very long lines in MIME
721 Changes between 0.9.7a and 0.9.7b [10 Apr 2003]
723 *) Countermeasure against the Klima-Pokorny-Rosa extension of
724 Bleichbacher's attack on PKCS #1 v1.5 padding: treat
725 a protocol version number mismatch like a decryption error
726 in ssl3_get_client_key_exchange (ssl/s3_srvr.c).
729 *) Turn on RSA blinding by default in the default implementation
730 to avoid a timing attack. Applications that don't want it can call
731 RSA_blinding_off() or use the new flag RSA_FLAG_NO_BLINDING.
732 They would be ill-advised to do so in most cases.
733 [Ben Laurie, Steve Henson, Geoff Thorpe, Bodo Moeller]
735 *) Change RSA blinding code so that it works when the PRNG is not
736 seeded (in this case, the secret RSA exponent is abused as
737 an unpredictable seed -- if it is not unpredictable, there
738 is no point in blinding anyway). Make RSA blinding thread-safe
739 by remembering the creator's thread ID in rsa->blinding and
740 having all other threads use local one-time blinding factors
741 (this requires more computation than sharing rsa->blinding, but
742 avoids excessive locking; and if an RSA object is not shared
743 between threads, blinding will still be very fast).
746 yet to be integrated into this CVS branch:
747 - Geoff's ENGINE_set_default() fix
749 *) Target "mingw" now allows native Windows code to be generated in
750 the Cygwin environment as well as with the MinGW compiler.
753 Changes between 0.9.7 and 0.9.7a [19 Feb 2003]
755 *) In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked
756 via timing by performing a MAC computation even if incorrrect
757 block cipher padding has been found. This is a countermeasure
758 against active attacks where the attacker has to distinguish
759 between bad padding and a MAC verification error. (CAN-2003-0078)
761 [Bodo Moeller; problem pointed out by Brice Canvel (EPFL),
762 Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and
763 Martin Vuagnoux (EPFL, Ilion)]
765 *) Make the no-err option work as intended. The intention with no-err
766 is not to have the whole error stack handling routines removed from
767 libcrypto, it's only intended to remove all the function name and
768 reason texts, thereby removing some of the footprint that may not
769 be interesting if those errors aren't displayed anyway.
771 NOTE: it's still possible for any application or module to have it's
772 own set of error texts inserted. The routines are there, just not
773 used by default when no-err is given.
776 *) Add support for FreeBSD on IA64.
777 [dirk.meyer@dinoex.sub.org via Richard Levitte, resolves #454]
779 *) Adjust DES_cbc_cksum() so it returns the same value as the MIT
780 Kerberos function mit_des_cbc_cksum(). Before this change,
781 the value returned by DES_cbc_cksum() was like the one from
782 mit_des_cbc_cksum(), except the bytes were swapped.
783 [Kevin Greaney <Kevin.Greaney@hp.com> and Richard Levitte]
785 *) Allow an application to disable the automatic SSL chain building.
786 Before this a rather primitive chain build was always performed in
787 ssl3_output_cert_chain(): an application had no way to send the
788 correct chain if the automatic operation produced an incorrect result.
790 Now the chain builder is disabled if either:
792 1. Extra certificates are added via SSL_CTX_add_extra_chain_cert().
794 2. The mode flag SSL_MODE_NO_AUTO_CHAIN is set.
796 The reasoning behind this is that an application would not want the
797 auto chain building to take place if extra chain certificates are
798 present and it might also want a means of sending no additional
799 certificates (for example the chain has two certificates and the
803 *) Add the possibility to build without the ENGINE framework.
804 [Steven Reddie <smr@essemer.com.au> via Richard Levitte]
806 *) Under Win32 gmtime() can return NULL: check return value in
807 OPENSSL_gmtime(). Add error code for case where gmtime() fails.
810 *) DSA routines: under certain error conditions uninitialized BN objects
811 could be freed. Solution: make sure initialization is performed early
812 enough. (Reported and fix supplied by Ivan D Nestlerode <nestler@MIT.EDU>,
813 Nils Larsch <nla@trustcenter.de> via PR#459)
816 *) Another fix for SSLv2 session ID handling: the session ID was incorrectly
817 checked on reconnect on the client side, therefore session resumption
818 could still fail with a "ssl session id is different" error. This
819 behaviour is masked when SSL_OP_ALL is used due to
820 SSL_OP_MICROSOFT_SESS_ID_BUG being set.
821 Behaviour observed by Crispin Flowerday <crispin@flowerday.cx> as
825 *) IA-32 assembler support enhancements: unified ELF targets, support
826 for SCO/Caldera platforms, fix for Cygwin shared build.
829 *) Add support for FreeBSD on sparc64. As a consequence, support for
830 FreeBSD on non-x86 processors is separate from x86 processors on
831 the config script, much like the NetBSD support.
832 [Richard Levitte & Kris Kennaway <kris@obsecurity.org>]
834 Changes between 0.9.6h and 0.9.7 [31 Dec 2002]
836 [NB: OpenSSL 0.9.6i and later 0.9.6 patch levels were released after
839 *) Fix session ID handling in SSLv2 client code: the SERVER FINISHED
840 code (06) was taken as the first octet of the session ID and the last
841 octet was ignored consequently. As a result SSLv2 client side session
842 caching could not have worked due to the session ID mismatch between
844 Behaviour observed by Crispin Flowerday <crispin@flowerday.cx> as
848 *) Change the declaration of needed Kerberos libraries to use EX_LIBS
849 instead of the special (and badly supported) LIBKRB5. LIBKRB5 is
853 *) The hw_ncipher.c engine requires dynamic locks. Unfortunately, it
854 seems that in spite of existing for more than a year, many application
855 author have done nothing to provide the necessary callbacks, which
856 means that this particular engine will not work properly anywhere.
857 This is a very unfortunate situation which forces us, in the name
858 of usability, to give the hw_ncipher.c a static lock, which is part
860 NOTE: This is for the 0.9.7 series ONLY. This hack will never
861 appear in 0.9.8 or later. We EXPECT application authors to have
862 dealt properly with this when 0.9.8 is released (unless we actually
863 make such changes in the libcrypto locking code that changes will
864 have to be made anyway).
867 *) In asn1_d2i_read_bio() repeatedly call BIO_read() until all content
868 octets have been read, EOF or an error occurs. Without this change
869 some truncated ASN1 structures will not produce an error.
872 *) Disable Heimdal support, since it hasn't been fully implemented.
873 Still give the possibility to force the use of Heimdal, but with
874 warnings and a request that patches get sent to openssl-dev.
877 *) Add the VC-CE target, introduce the WINCE sysname, and add
878 INSTALL.WCE and appropriate conditionals to make it build.
879 [Steven Reddie <smr@essemer.com.au> via Richard Levitte]
881 *) Change the DLL names for Cygwin to cygcrypto-x.y.z.dll and
882 cygssl-x.y.z.dll, where x, y and z are the major, minor and
883 edit numbers of the version.
884 [Corinna Vinschen <vinschen@redhat.com> and Richard Levitte]
886 *) Introduce safe string copy and catenation functions
887 (BUF_strlcpy() and BUF_strlcat()).
888 [Ben Laurie (CHATS) and Richard Levitte]
890 *) Avoid using fixed-size buffers for one-line DNs.
893 *) Add BUF_MEM_grow_clean() to avoid information leakage when
894 resizing buffers containing secrets, and use where appropriate.
897 *) Avoid using fixed size buffers for configuration file location.
900 *) Avoid filename truncation for various CA files.
903 *) Use sizeof in preference to magic numbers.
906 *) Avoid filename truncation in cert requests.
909 *) Add assertions to check for (supposedly impossible) buffer
913 *) Don't cache truncated DNS entries in the local cache (this could
914 potentially lead to a spoofing attack).
917 *) Fix various buffers to be large enough for hex/decimal
918 representations in a platform independent manner.
921 *) Add CRYPTO_realloc_clean() to avoid information leakage when
922 resizing buffers containing secrets, and use where appropriate.
925 *) Add BIO_indent() to avoid much slightly worrying code to do
929 *) Convert sprintf()/BIO_puts() to BIO_printf().
932 *) buffer_gets() could terminate with the buffer only half
936 *) Add assertions to prevent user-supplied crypto functions from
937 overflowing internal buffers by having large block sizes, etc.
940 *) New OPENSSL_assert() macro (similar to assert(), but enabled
944 *) Eliminate unused copy of key in RC4.
947 *) Eliminate unused and incorrectly sized buffers for IV in pem.h.
950 *) Fix off-by-one error in EGD path.
953 *) If RANDFILE path is too long, ignore instead of truncating.
956 *) Eliminate unused and incorrectly sized X.509 structure
960 *) Eliminate unused and dangerous function knumber().
963 *) Eliminate unused and dangerous structure, KSSL_ERR.
966 *) Protect against overlong session ID context length in an encoded
967 session object. Since these are local, this does not appear to be
971 *) Change from security patch (see 0.9.6e below) that did not affect
972 the 0.9.6 release series:
974 Remote buffer overflow in SSL3 protocol - an attacker could
975 supply an oversized master key in Kerberos-enabled versions.
979 *) Change the SSL kerb5 codes to match RFC 2712.
982 *) Make -nameopt work fully for req and add -reqopt switch.
983 [Michael Bell <michael.bell@rz.hu-berlin.de>, Steve Henson]
985 *) The "block size" for block ciphers in CFB and OFB mode should be 1.
986 [Steve Henson, reported by Yngve Nysaeter Pettersen <yngve@opera.com>]
988 *) Make sure tests can be performed even if the corresponding algorithms
989 have been removed entirely. This was also the last step to make
990 OpenSSL compilable with DJGPP under all reasonable conditions.
991 [Richard Levitte, Doug Kaufman <dkaufman@rahul.net>]
993 *) Add cipher selection rules COMPLEMENTOFALL and COMPLEMENTOFDEFAULT
994 to allow version independent disabling of normally unselected ciphers,
995 which may be activated as a side-effect of selecting a single cipher.
997 (E.g., cipher list string "RSA" enables ciphersuites that are left
998 out of "ALL" because they do not provide symmetric encryption.
999 "RSA:!COMPLEMEMENTOFALL" avoids these unsafe ciphersuites.)
1000 [Lutz Jaenicke, Bodo Moeller]
1002 *) Add appropriate support for separate platform-dependent build
1003 directories. The recommended way to make a platform-dependent
1004 build directory is the following (tested on Linux), maybe with
1007 # Place yourself outside of the OpenSSL source tree. In
1008 # this example, the environment variable OPENSSL_SOURCE
1009 # is assumed to contain the absolute OpenSSL source directory.
1010 mkdir -p objtree/"`uname -s`-`uname -r`-`uname -m`"
1011 cd objtree/"`uname -s`-`uname -r`-`uname -m`"
1012 (cd $OPENSSL_SOURCE; find . -type f) | while read F; do
1013 mkdir -p `dirname $F`
1014 ln -s $OPENSSL_SOURCE/$F $F
1017 To be absolutely sure not to disturb the source tree, a "make clean"
1018 is a good thing. If it isn't successfull, don't worry about it,
1019 it probably means the source directory is very clean.
1022 *) Make sure any ENGINE control commands make local copies of string
1023 pointers passed to them whenever necessary. Otherwise it is possible
1024 the caller may have overwritten (or deallocated) the original string
1025 data when a later ENGINE operation tries to use the stored values.
1026 [Götz Babin-Ebell <babinebell@trustcenter.de>]
1028 *) Improve diagnostics in file reading and command-line digests.
1029 [Ben Laurie aided and abetted by Solar Designer <solar@openwall.com>]
1031 *) Add AES modes CFB and OFB to the object database. Correct an
1032 error in AES-CFB decryption.
1035 *) Remove most calls to EVP_CIPHER_CTX_cleanup() in evp_enc.c, this
1036 allows existing EVP_CIPHER_CTX structures to be reused after
1037 calling EVP_*Final(). This behaviour is used by encryption
1038 BIOs and some applications. This has the side effect that
1039 applications must explicitly clean up cipher contexts with
1040 EVP_CIPHER_CTX_cleanup() or they will leak memory.
1043 *) Check the values of dna and dnb in bn_mul_recursive before calling
1044 bn_mul_comba (a non zero value means the a or b arrays do not contain
1045 n2 elements) and fallback to bn_mul_normal if either is not zero.
1048 *) Fix escaping of non-ASCII characters when using the -subj option
1049 of the "openssl req" command line tool. (Robert Joop <joop@fokus.gmd.de>)
1052 *) Make object definitions compliant to LDAP (RFC2256): SN is the short
1053 form for "surname", serialNumber has no short form.
1054 Use "mail" as the short name for "rfc822Mailbox" according to RFC2798;
1055 therefore remove "mail" short name for "internet 7".
1056 The OID for unique identifiers in X509 certificates is
1057 x500UniqueIdentifier, not uniqueIdentifier.
1058 Some more OID additions. (Michael Bell <michael.bell@rz.hu-berlin.de>)
1061 *) Add an "init" command to the ENGINE config module and auto initialize
1062 ENGINEs. Without any "init" command the ENGINE will be initialized
1063 after all ctrl commands have been executed on it. If init=1 the
1064 ENGINE is initailized at that point (ctrls before that point are run
1065 on the uninitialized ENGINE and after on the initialized one). If
1066 init=0 then the ENGINE will not be iniatialized at all.
1069 *) Fix the 'app_verify_callback' interface so that the user-defined
1070 argument is actually passed to the callback: In the
1071 SSL_CTX_set_cert_verify_callback() prototype, the callback
1072 declaration has been changed from
1075 int (*cb)(X509_STORE_CTX *,void *);
1076 in ssl_verify_cert_chain (ssl/ssl_cert.c), the call
1077 i=s->ctx->app_verify_callback(&ctx)
1078 has been changed into
1079 i=s->ctx->app_verify_callback(&ctx, s->ctx->app_verify_arg).
1081 To update applications using SSL_CTX_set_cert_verify_callback(),
1082 a dummy argument can be added to their callback functions.
1083 [D. K. Smetters <smetters@parc.xerox.com>]
1085 *) Added the '4758cca' ENGINE to support IBM 4758 cards.
1086 [Maurice Gittens <maurice@gittens.nl>, touchups by Geoff Thorpe]
1088 *) Add and OPENSSL_LOAD_CONF define which will cause
1089 OpenSSL_add_all_algorithms() to load the openssl.cnf config file.
1090 This allows older applications to transparently support certain
1091 OpenSSL features: such as crypto acceleration and dynamic ENGINE loading.
1092 Two new functions OPENSSL_add_all_algorithms_noconf() which will never
1093 load the config file and OPENSSL_add_all_algorithms_conf() which will
1094 always load it have also been added.
1097 *) Add the OFB, CFB and CTR (all with 128 bit feedback) to AES.
1098 Adjust NIDs and EVP layer.
1099 [Stephen Sprunk <stephen@sprunk.org> and Richard Levitte]
1101 *) Config modules support in openssl utility.
1103 Most commands now load modules from the config file,
1104 though in a few (such as version) this isn't done
1105 because it couldn't be used for anything.
1107 In the case of ca and req the config file used is
1108 the same as the utility itself: that is the -config
1109 command line option can be used to specify an
1113 *) Move default behaviour from OPENSSL_config(). If appname is NULL
1114 use "openssl_conf" if filename is NULL use default openssl config file.
1117 *) Add an argument to OPENSSL_config() to allow the use of an alternative
1118 config section name. Add a new flag to tolerate a missing config file
1119 and move code to CONF_modules_load_file().
1122 *) Support for crypto accelerator cards from Accelerated Encryption
1123 Processing, www.aep.ie. (Use engine 'aep')
1124 The support was copied from 0.9.6c [engine] and adapted/corrected
1125 to work with the new engine framework.
1126 [AEP Inc. and Richard Levitte]
1128 *) Support for SureWare crypto accelerator cards from Baltimore
1129 Technologies. (Use engine 'sureware')
1130 The support was copied from 0.9.6c [engine] and adapted
1131 to work with the new engine framework.
1134 *) Have the CHIL engine fork-safe (as defined by nCipher) and actually
1135 make the newer ENGINE framework commands for the CHIL engine work.
1136 [Toomas Kiisk <vix@cyber.ee> and Richard Levitte]
1138 *) Make it possible to produce shared libraries on ReliantUNIX.
1139 [Robert Dahlem <Robert.Dahlem@ffm2.siemens.de> via Richard Levitte]
1141 *) Add the configuration target debug-linux-ppro.
1142 Make 'openssl rsa' use the general key loading routines
1143 implemented in apps.c, and make those routines able to
1144 handle the key format FORMAT_NETSCAPE and the variant
1146 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte]
1148 *) Fix a crashbug and a logic bug in hwcrhk_load_pubkey().
1149 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte]
1151 *) Add -keyform to rsautl, and document -engine.
1152 [Richard Levitte, inspired by Toomas Kiisk <vix@cyber.ee>]
1154 *) Change BIO_new_file (crypto/bio/bss_file.c) to use new
1155 BIO_R_NO_SUCH_FILE error code rather than the generic
1156 ERR_R_SYS_LIB error code if fopen() fails with ENOENT.
1159 *) Add new functions
1161 ERR_peek_last_error_line
1162 ERR_peek_last_error_line_data.
1163 These are similar to
1166 ERR_peek_error_line_data,
1167 but report on the latest error recorded rather than the first one
1168 still in the error queue.
1169 [Ben Laurie, Bodo Moeller]
1171 *) default_algorithms option in ENGINE config module. This allows things
1173 default_algorithms = ALL
1174 default_algorithms = RSA, DSA, RAND, CIPHERS, DIGESTS
1177 *) Prelminary ENGINE config module.
1180 *) New experimental application configuration code.
1183 *) Change the AES code to follow the same name structure as all other
1184 symmetric ciphers, and behave the same way. Move everything to
1185 the directory crypto/aes, thereby obsoleting crypto/rijndael.
1186 [Stephen Sprunk <stephen@sprunk.org> and Richard Levitte]
1188 *) SECURITY: remove unsafe setjmp/signal interaction from ui_openssl.c.
1189 [Ben Laurie and Theo de Raadt]
1191 *) Add option to output public keys in req command.
1192 [Massimiliano Pala madwolf@openca.org]
1194 *) Use wNAFs in EC_POINTs_mul() for improved efficiency
1195 (up to about 10% better than before for P-192 and P-224).
1198 *) New functions/macros
1200 SSL_CTX_set_msg_callback(ctx, cb)
1201 SSL_CTX_set_msg_callback_arg(ctx, arg)
1202 SSL_set_msg_callback(ssl, cb)
1203 SSL_set_msg_callback_arg(ssl, arg)
1205 to request calling a callback function
1207 void cb(int write_p, int version, int content_type,
1208 const void *buf, size_t len, SSL *ssl, void *arg)
1210 whenever a protocol message has been completely received
1211 (write_p == 0) or sent (write_p == 1). Here 'version' is the
1212 protocol version according to which the SSL library interprets
1213 the current protocol message (SSL2_VERSION, SSL3_VERSION, or
1214 TLS1_VERSION). 'content_type' is 0 in the case of SSL 2.0, or
1215 the content type as defined in the SSL 3.0/TLS 1.0 protocol
1216 specification (change_cipher_spec(20), alert(21), handshake(22)).
1217 'buf' and 'len' point to the actual message, 'ssl' to the
1218 SSL object, and 'arg' is the application-defined value set by
1219 SSL[_CTX]_set_msg_callback_arg().
1221 'openssl s_client' and 'openssl s_server' have new '-msg' options
1222 to enable a callback that displays all protocol messages.
1225 *) Change the shared library support so shared libraries are built as
1226 soon as the corresponding static library is finished, and thereby get
1227 openssl and the test programs linked against the shared library.
1228 This still only happens when the keyword "shard" has been given to
1229 the configuration scripts.
1231 NOTE: shared library support is still an experimental thing, and
1232 backward binary compatibility is still not guaranteed.
1233 ["Maciej W. Rozycki" <macro@ds2.pg.gda.pl> and Richard Levitte]
1235 *) Add support for Subject Information Access extension.
1236 [Peter Sylvester <Peter.Sylvester@EdelWeb.fr>]
1238 *) Make BUF_MEM_grow() behaviour more consistent: Initialise to zero
1239 additional bytes when new memory had to be allocated, not just
1240 when reusing an existing buffer.
1243 *) New command line and configuration option 'utf8' for the req command.
1244 This allows field values to be specified as UTF8 strings.
1247 *) Add -multi and -mr options to "openssl speed" - giving multiple parallel
1248 runs for the former and machine-readable output for the latter.
1251 *) Add '-noemailDN' option to 'openssl ca'. This prevents inclusion
1252 of the e-mail address in the DN (i.e., it will go into a certificate
1253 extension only). The new configuration file option 'email_in_dn = no'
1254 has the same effect.
1255 [Massimiliano Pala madwolf@openca.org]
1257 *) Change all functions with names starting with des_ to be starting
1258 with DES_ instead. Add wrappers that are compatible with libdes,
1259 but are named _ossl_old_des_*. Finally, add macros that map the
1260 des_* symbols to the corresponding _ossl_old_des_* if libdes
1261 compatibility is desired. If OpenSSL 0.9.6c compatibility is
1262 desired, the des_* symbols will be mapped to DES_*, with one
1265 Since we provide two compatibility mappings, the user needs to
1266 define the macro OPENSSL_DES_LIBDES_COMPATIBILITY if libdes
1267 compatibility is desired. The default (i.e., when that macro
1268 isn't defined) is OpenSSL 0.9.6c compatibility.
1270 There are also macros that enable and disable the support of old
1271 des functions altogether. Those are OPENSSL_ENABLE_OLD_DES_SUPPORT
1272 and OPENSSL_DISABLE_OLD_DES_SUPPORT. If none or both of those
1273 are defined, the default will apply: to support the old des routines.
1275 In either case, one must include openssl/des.h to get the correct
1276 definitions. Do not try to just include openssl/des_old.h, that
1279 NOTE: This is a major break of an old API into a new one. Software
1280 authors are encouraged to switch to the DES_ style functions. Some
1281 time in the future, des_old.h and the libdes compatibility functions
1282 will be disable (i.e. OPENSSL_DISABLE_OLD_DES_SUPPORT will be the
1283 default), and then completely removed.
1286 *) Test for certificates which contain unsupported critical extensions.
1287 If such a certificate is found during a verify operation it is
1288 rejected by default: this behaviour can be overridden by either
1289 handling the new error X509_V_ERR_UNHANDLED_CRITICAL_EXTENSION or
1290 by setting the verify flag X509_V_FLAG_IGNORE_CRITICAL. A new function
1291 X509_supported_extension() has also been added which returns 1 if a
1292 particular extension is supported.
1295 *) Modify the behaviour of EVP cipher functions in similar way to digests
1296 to retain compatibility with existing code.
1299 *) Modify the behaviour of EVP_DigestInit() and EVP_DigestFinal() to retain
1300 compatibility with existing code. In particular the 'ctx' parameter does
1301 not have to be to be initialized before the call to EVP_DigestInit() and
1302 it is tidied up after a call to EVP_DigestFinal(). New function
1303 EVP_DigestFinal_ex() which does not tidy up the ctx. Similarly function
1304 EVP_MD_CTX_copy() changed to not require the destination to be
1305 initialized valid and new function EVP_MD_CTX_copy_ex() added which
1306 requires the destination to be valid.
1308 Modify all the OpenSSL digest calls to use EVP_DigestInit_ex(),
1309 EVP_DigestFinal_ex() and EVP_MD_CTX_copy_ex().
1312 *) Change ssl3_get_message (ssl/s3_both.c) and the functions using it
1313 so that complete 'Handshake' protocol structures are kept in memory
1314 instead of overwriting 'msg_type' and 'length' with 'body' data.
1317 *) Add an implementation of SSL_add_dir_cert_subjects_to_stack for Win32.
1318 [Massimo Santin via Richard Levitte]
1320 *) Major restructuring to the underlying ENGINE code. This includes
1321 reduction of linker bloat, separation of pure "ENGINE" manipulation
1322 (initialisation, etc) from functionality dealing with implementations
1323 of specific crypto iterfaces. This change also introduces integrated
1324 support for symmetric ciphers and digest implementations - so ENGINEs
1325 can now accelerate these by providing EVP_CIPHER and EVP_MD
1326 implementations of their own. This is detailed in crypto/engine/README
1327 as it couldn't be adequately described here. However, there are a few
1328 API changes worth noting - some RSA, DSA, DH, and RAND functions that
1329 were changed in the original introduction of ENGINE code have now
1330 reverted back - the hooking from this code to ENGINE is now a good
1331 deal more passive and at run-time, operations deal directly with
1332 RSA_METHODs, DSA_METHODs (etc) as they did before, rather than
1333 dereferencing through an ENGINE pointer any more. Also, the ENGINE
1334 functions dealing with BN_MOD_EXP[_CRT] handlers have been removed -
1335 they were not being used by the framework as there is no concept of a
1336 BIGNUM_METHOD and they could not be generalised to the new
1337 'ENGINE_TABLE' mechanism that underlies the new code. Similarly,
1338 ENGINE_cpy() has been removed as it cannot be consistently defined in
1342 *) Change ASN1_GENERALIZEDTIME_check() to allow fractional seconds.
1345 *) Change mkdef.pl to sort symbols that get the same entry number,
1346 and make sure the automatically generated functions ERR_load_*
1347 become part of libeay.num as well.
1350 *) New function SSL_renegotiate_pending(). This returns true once
1351 renegotiation has been requested (either SSL_renegotiate() call
1352 or HelloRequest/ClientHello receveived from the peer) and becomes
1353 false once a handshake has been completed.
1354 (For servers, SSL_renegotiate() followed by SSL_do_handshake()
1355 sends a HelloRequest, but does not ensure that a handshake takes
1356 place. SSL_renegotiate_pending() is useful for checking if the
1357 client has followed the request.)
1360 *) New SSL option SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION.
1361 By default, clients may request session resumption even during
1362 renegotiation (if session ID contexts permit); with this option,
1363 session resumption is possible only in the first handshake.
1365 SSL_OP_ALL is now 0x00000FFFL instead of 0x000FFFFFL. This makes
1366 more bits available for options that should not be part of
1367 SSL_OP_ALL (such as SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION).
1370 *) Add some demos for certificate and certificate request creation.
1373 *) Make maximum certificate chain size accepted from the peer application
1374 settable (SSL*_get/set_max_cert_list()), as proposed by
1375 "Douglas E. Engert" <deengert@anl.gov>.
1378 *) Add support for shared libraries for Unixware-7
1379 (Boyd Lynn Gerber <gerberb@zenez.com>).
1382 *) Add a "destroy" handler to ENGINEs that allows structural cleanup to
1383 be done prior to destruction. Use this to unload error strings from
1384 ENGINEs that load their own error strings. NB: This adds two new API
1385 functions to "get" and "set" this destroy handler in an ENGINE.
1388 *) Alter all existing ENGINE implementations (except "openssl" and
1389 "openbsd") to dynamically instantiate their own error strings. This
1390 makes them more flexible to be built both as statically-linked ENGINEs
1391 and self-contained shared-libraries loadable via the "dynamic" ENGINE.
1392 Also, add stub code to each that makes building them as self-contained
1393 shared-libraries easier (see README.ENGINE).
1396 *) Add a "dynamic" ENGINE that provides a mechanism for binding ENGINE
1397 implementations into applications that are completely implemented in
1398 self-contained shared-libraries. The "dynamic" ENGINE exposes control
1399 commands that can be used to configure what shared-library to load and
1400 to control aspects of the way it is handled. Also, made an update to
1401 the README.ENGINE file that brings its information up-to-date and
1402 provides some information and instructions on the "dynamic" ENGINE
1403 (ie. how to use it, how to build "dynamic"-loadable ENGINEs, etc).
1406 *) Make it possible to unload ranges of ERR strings with a new
1407 "ERR_unload_strings" function.
1410 *) Add a copy() function to EVP_MD.
1413 *) Make EVP_MD routines take a context pointer instead of just the
1414 md_data void pointer.
1417 *) Add flags to EVP_MD and EVP_MD_CTX. EVP_MD_FLAG_ONESHOT indicates
1418 that the digest can only process a single chunk of data
1419 (typically because it is provided by a piece of
1420 hardware). EVP_MD_CTX_FLAG_ONESHOT indicates that the application
1421 is only going to provide a single chunk of data, and hence the
1422 framework needn't accumulate the data for oneshot drivers.
1425 *) As with "ERR", make it possible to replace the underlying "ex_data"
1426 functions. This change also alters the storage and management of global
1427 ex_data state - it's now all inside ex_data.c and all "class" code (eg.
1428 RSA, BIO, SSL_CTX, etc) no longer stores its own STACKS and per-class
1429 index counters. The API functions that use this state have been changed
1430 to take a "class_index" rather than pointers to the class's local STACK
1431 and counter, and there is now an API function to dynamically create new
1432 classes. This centralisation allows us to (a) plug a lot of the
1433 thread-safety problems that existed, and (b) makes it possible to clean
1434 up all allocated state using "CRYPTO_cleanup_all_ex_data()". W.r.t. (b)
1435 such data would previously have always leaked in application code and
1436 workarounds were in place to make the memory debugging turn a blind eye
1437 to it. Application code that doesn't use this new function will still
1438 leak as before, but their memory debugging output will announce it now
1439 rather than letting it slide.
1441 Besides the addition of CRYPTO_cleanup_all_ex_data(), another API change
1442 induced by the "ex_data" overhaul is that X509_STORE_CTX_init() now
1443 has a return value to indicate success or failure.
1446 *) Make it possible to replace the underlying "ERR" functions such that the
1447 global state (2 LHASH tables and 2 locks) is only used by the "default"
1448 implementation. This change also adds two functions to "get" and "set"
1449 the implementation prior to it being automatically set the first time
1450 any other ERR function takes place. Ie. an application can call "get",
1451 pass the return value to a module it has just loaded, and that module
1452 can call its own "set" function using that value. This means the
1453 module's "ERR" operations will use (and modify) the error state in the
1454 application and not in its own statically linked copy of OpenSSL code.
1457 *) Give DH, DSA, and RSA types their own "**_up_ref()" function to increment
1458 reference counts. This performs normal REF_PRINT/REF_CHECK macros on
1459 the operation, and provides a more encapsulated way for external code
1460 (crypto/evp/ and ssl/) to do this. Also changed the evp and ssl code
1461 to use these functions rather than manually incrementing the counts.
1463 Also rename "DSO_up()" function to more descriptive "DSO_up_ref()".
1466 *) Add EVP test program.
1469 *) Add symmetric cipher support to ENGINE. Expect the API to change!
1472 *) New CRL functions: X509_CRL_set_version(), X509_CRL_set_issuer_name()
1473 X509_CRL_set_lastUpdate(), X509_CRL_set_nextUpdate(), X509_CRL_sort(),
1474 X509_REVOKED_set_serialNumber(), and X509_REVOKED_set_revocationDate().
1475 These allow a CRL to be built without having to access X509_CRL fields
1476 directly. Modify 'ca' application to use new functions.
1479 *) Move SSL_OP_TLS_ROLLBACK_BUG out of the SSL_OP_ALL list of recommended
1480 bug workarounds. Rollback attack detection is a security feature.
1481 The problem will only arise on OpenSSL servers when TLSv1 is not
1482 available (sslv3_server_method() or SSL_OP_NO_TLSv1).
1483 Software authors not wanting to support TLSv1 will have special reasons
1484 for their choice and can explicitly enable this option.
1485 [Bodo Moeller, Lutz Jaenicke]
1487 *) Rationalise EVP so it can be extended: don't include a union of
1488 cipher/digest structures, add init/cleanup functions for EVP_MD_CTX
1489 (similar to those existing for EVP_CIPHER_CTX).
1494 EVP_MD_CTX_init(&md); /* new function call */
1495 EVP_DigestInit(&md, EVP_sha1());
1496 EVP_DigestUpdate(&md, in, len);
1497 EVP_DigestFinal(&md, out, NULL);
1498 EVP_MD_CTX_cleanup(&md); /* new function call */
1502 *) Make DES key schedule conform to the usual scheme, as well as
1503 correcting its structure. This means that calls to DES functions
1504 now have to pass a pointer to a des_key_schedule instead of a
1505 plain des_key_schedule (which was actually always a pointer
1508 des_key_schedule ks;
1510 des_set_key_checked(..., &ks);
1511 des_ncbc_encrypt(..., &ks, ...);
1513 (Note that a later change renames 'des_...' into 'DES_...'.)
1516 *) Initial reduction of linker bloat: the use of some functions, such as
1517 PEM causes large amounts of unused functions to be linked in due to
1518 poor organisation. For example pem_all.c contains every PEM function
1519 which has a knock on effect of linking in large amounts of (unused)
1520 ASN1 code. Grouping together similar functions and splitting unrelated
1521 functions prevents this.
1524 *) Cleanup of EVP macros.
1527 *) Change historical references to {NID,SN,LN}_des_ede and ede3 to add the
1528 correct _ecb suffix.
1531 *) Add initial OCSP responder support to ocsp application. The
1532 revocation information is handled using the text based index
1533 use by the ca application. The responder can either handle
1534 requests generated internally, supplied in files (for example
1535 via a CGI script) or using an internal minimal server.
1538 *) Add configuration choices to get zlib compression for TLS.
1541 *) Changes to Kerberos SSL for RFC 2712 compliance:
1542 1. Implemented real KerberosWrapper, instead of just using
1543 KRB5 AP_REQ message. [Thanks to Simon Wilkinson <sxw@sxw.org.uk>]
1544 2. Implemented optional authenticator field of KerberosWrapper.
1546 Added openssl-style ASN.1 macros for Kerberos ticket, ap_req,
1547 and authenticator structs; see crypto/krb5/.
1549 Generalized Kerberos calls to support multiple Kerberos libraries.
1550 [Vern Staats <staatsvr@asc.hpc.mil>,
1551 Jeffrey Altman <jaltman@columbia.edu>
1552 via Richard Levitte]
1554 *) Cause 'openssl speed' to use fully hard-coded DSA keys as it
1555 already does with RSA. testdsa.h now has 'priv_key/pub_key'
1556 values for each of the key sizes rather than having just
1557 parameters (and 'speed' generating keys each time).
1560 *) Speed up EVP routines.
1563 type 8 bytes 64 bytes 256 bytes 1024 bytes 8192 bytes
1564 des-cbc 4408.85k 5560.51k 5778.46k 5862.20k 5825.16k
1565 des-cbc 4389.55k 5571.17k 5792.23k 5846.91k 5832.11k
1566 des-cbc 4394.32k 5575.92k 5807.44k 5848.37k 5841.30k
1568 des-cbc 3482.66k 5069.49k 5496.39k 5614.16k 5639.28k
1569 des-cbc 3480.74k 5068.76k 5510.34k 5609.87k 5635.52k
1570 des-cbc 3483.72k 5067.62k 5504.60k 5708.01k 5724.80k
1573 des-cbc 4660.16k 5650.19k 5807.19k 5827.13k 5783.32k
1575 des-cbc 3624.96k 5258.21k 5530.91k 5624.30k 5628.26k
1578 *) Added the OS2-EMX target.
1579 ["Brian Havard" <brianh@kheldar.apana.org.au> and Richard Levitte]
1581 *) Rewrite apps to use NCONF routines instead of the old CONF. New functions
1582 to support NCONF routines in extension code. New function CONF_set_nconf()
1583 to allow functions which take an NCONF to also handle the old LHASH
1584 structure: this means that the old CONF compatible routines can be
1585 retained (in particular wrt extensions) without having to duplicate the
1586 code. New function X509V3_add_ext_nconf_sk to add extensions to a stack.
1589 *) Enhance the general user interface with mechanisms for inner control
1590 and with possibilities to have yes/no kind of prompts.
1593 *) Change all calls to low level digest routines in the library and
1594 applications to use EVP. Add missing calls to HMAC_cleanup() and
1595 don't assume HMAC_CTX can be copied using memcpy().
1596 [Verdon Walker <VWalker@novell.com>, Steve Henson]
1598 *) Add the possibility to control engines through control names but with
1599 arbitrary arguments instead of just a string.
1600 Change the key loaders to take a UI_METHOD instead of a callback
1601 function pointer. NOTE: this breaks binary compatibility with earlier
1602 versions of OpenSSL [engine].
1603 Adapt the nCipher code for these new conditions and add a card insertion
1607 *) Enhance the general user interface with mechanisms to better support
1608 dialog box interfaces, application-defined prompts, the possibility
1609 to use defaults (for example default passwords from somewhere else)
1610 and interrupts/cancellations.
1613 *) Tidy up PKCS#12 attribute handling. Add support for the CSP name
1614 attribute in PKCS#12 files, add new -CSP option to pkcs12 utility.
1617 *) Fix a memory leak in 'sk_dup()' in the case reallocation fails. (Also
1618 tidy up some unnecessarily weird code in 'sk_new()').
1619 [Geoff, reported by Diego Tartara <dtartara@novamens.com>]
1621 *) Change the key loading routines for ENGINEs to use the same kind
1622 callback (pem_password_cb) as all other routines that need this
1626 *) Increase ENTROPY_NEEDED to 32 bytes, as Rijndael can operate with
1627 256 bit (=32 byte) keys. Of course seeding with more entropy bytes
1628 than this minimum value is recommended.
1631 *) New random seeder for OpenVMS, using the system process statistics
1632 that are easily reachable.
1635 *) Windows apparently can't transparently handle global
1636 variables defined in DLLs. Initialisations such as:
1638 const ASN1_ITEM *it = &ASN1_INTEGER_it;
1640 wont compile. This is used by the any applications that need to
1641 declare their own ASN1 modules. This was fixed by adding the option
1642 EXPORT_VAR_AS_FN to all Win32 platforms, although this isn't strictly
1643 needed for static libraries under Win32.
1646 *) New functions X509_PURPOSE_set() and X509_TRUST_set() to handle
1647 setting of purpose and trust fields. New X509_STORE trust and
1648 purpose functions and tidy up setting in other SSL functions.
1651 *) Add copies of X509_STORE_CTX fields and callbacks to X509_STORE
1652 structure. These are inherited by X509_STORE_CTX when it is
1653 initialised. This allows various defaults to be set in the
1654 X509_STORE structure (such as flags for CRL checking and custom
1655 purpose or trust settings) for functions which only use X509_STORE_CTX
1656 internally such as S/MIME.
1658 Modify X509_STORE_CTX_purpose_inherit() so it only sets purposes and
1659 trust settings if they are not set in X509_STORE. This allows X509_STORE
1660 purposes and trust (in S/MIME for example) to override any set by default.
1662 Add command line options for CRL checking to smime, s_client and s_server
1666 *) Initial CRL based revocation checking. If the CRL checking flag(s)
1667 are set then the CRL is looked up in the X509_STORE structure and
1668 its validity and signature checked, then if the certificate is found
1669 in the CRL the verify fails with a revoked error.
1671 Various new CRL related callbacks added to X509_STORE_CTX structure.
1673 Command line options added to 'verify' application to support this.
1675 This needs some additional work, such as being able to handle multiple
1676 CRLs with different times, extension based lookup (rather than just
1677 by subject name) and ultimately more complete V2 CRL extension
1681 *) Add a general user interface API (crypto/ui/). This is designed
1682 to replace things like des_read_password and friends (backward
1683 compatibility functions using this new API are provided).
1684 The purpose is to remove prompting functions from the DES code
1685 section as well as provide for prompting through dialog boxes in
1686 a window system and the like.
1689 *) Add "ex_data" support to ENGINE so implementations can add state at a
1690 per-structure level rather than having to store it globally.
1693 *) Make it possible for ENGINE structures to be copied when retrieved by
1694 ENGINE_by_id() if the ENGINE specifies a new flag: ENGINE_FLAGS_BY_ID_COPY.
1695 This causes the "original" ENGINE structure to act like a template,
1696 analogous to the RSA vs. RSA_METHOD type of separation. Because of this
1697 operational state can be localised to each ENGINE structure, despite the
1698 fact they all share the same "methods". New ENGINE structures returned in
1699 this case have no functional references and the return value is the single
1700 structural reference. This matches the single structural reference returned
1701 by ENGINE_by_id() normally, when it is incremented on the pre-existing
1705 *) Fix ASN1 decoder when decoding type ANY and V_ASN1_OTHER: since this
1706 needs to match any other type at all we need to manually clear the
1710 *) Changes to the "openssl engine" utility to include;
1711 - verbosity levels ('-v', '-vv', and '-vvv') that provide information
1712 about an ENGINE's available control commands.
1713 - executing control commands from command line arguments using the
1714 '-pre' and '-post' switches. '-post' is only used if '-t' is
1715 specified and the ENGINE is successfully initialised. The syntax for
1716 the individual commands are colon-separated, for example;
1717 openssl engine chil -pre FORK_CHECK:0 -pre SO_PATH:/lib/test.so
1720 *) New dynamic control command support for ENGINEs. ENGINEs can now
1721 declare their own commands (numbers), names (strings), descriptions,
1722 and input types for run-time discovery by calling applications. A
1723 subset of these commands are implicitly classed as "executable"
1724 depending on their input type, and only these can be invoked through
1725 the new string-based API function ENGINE_ctrl_cmd_string(). (Eg. this
1726 can be based on user input, config files, etc). The distinction is
1727 that "executable" commands cannot return anything other than a boolean
1728 result and can only support numeric or string input, whereas some
1729 discoverable commands may only be for direct use through
1730 ENGINE_ctrl(), eg. supporting the exchange of binary data, function
1731 pointers, or other custom uses. The "executable" commands are to
1732 support parameterisations of ENGINE behaviour that can be
1733 unambiguously defined by ENGINEs and used consistently across any
1734 OpenSSL-based application. Commands have been added to all the
1735 existing hardware-supporting ENGINEs, noticeably "SO_PATH" to allow
1736 control over shared-library paths without source code alterations.
1739 *) Changed all ENGINE implementations to dynamically allocate their
1740 ENGINEs rather than declaring them statically. Apart from this being
1741 necessary with the removal of the ENGINE_FLAGS_MALLOCED distinction,
1742 this also allows the implementations to compile without using the
1743 internal engine_int.h header.
1746 *) Minor adjustment to "rand" code. RAND_get_rand_method() now returns a
1747 'const' value. Any code that should be able to modify a RAND_METHOD
1748 should already have non-const pointers to it (ie. they should only
1749 modify their own ones).
1752 *) Made a variety of little tweaks to the ENGINE code.
1753 - "atalla" and "ubsec" string definitions were moved from header files
1754 to C code. "nuron" string definitions were placed in variables
1755 rather than hard-coded - allowing parameterisation of these values
1756 later on via ctrl() commands.
1757 - Removed unused "#if 0"'d code.
1758 - Fixed engine list iteration code so it uses ENGINE_free() to release
1759 structural references.
1760 - Constified the RAND_METHOD element of ENGINE structures.
1761 - Constified various get/set functions as appropriate and added
1762 missing functions (including a catch-all ENGINE_cpy that duplicates
1763 all ENGINE values onto a new ENGINE except reference counts/state).
1764 - Removed NULL parameter checks in get/set functions. Setting a method
1765 or function to NULL is a way of cancelling out a previously set
1766 value. Passing a NULL ENGINE parameter is just plain stupid anyway
1767 and doesn't justify the extra error symbols and code.
1768 - Deprecate the ENGINE_FLAGS_MALLOCED define and move the area for
1769 flags from engine_int.h to engine.h.
1770 - Changed prototypes for ENGINE handler functions (init(), finish(),
1771 ctrl(), key-load functions, etc) to take an (ENGINE*) parameter.
1774 *) Implement binary inversion algorithm for BN_mod_inverse in addition
1775 to the algorithm using long division. The binary algorithm can be
1776 used only if the modulus is odd. On 32-bit systems, it is faster
1777 only for relatively small moduli (roughly 20-30% for 128-bit moduli,
1778 roughly 5-15% for 256-bit moduli), so we use it only for moduli
1779 up to 450 bits. In 64-bit environments, the binary algorithm
1780 appears to be advantageous for much longer moduli; here we use it
1781 for moduli up to 2048 bits.
1784 *) Rewrite CHOICE field setting in ASN1_item_ex_d2i(). The old code
1785 could not support the combine flag in choice fields.
1788 *) Add a 'copy_extensions' option to the 'ca' utility. This copies
1789 extensions from a certificate request to the certificate.
1792 *) Allow multiple 'certopt' and 'nameopt' options to be separated
1793 by commas. Add 'namopt' and 'certopt' options to the 'ca' config
1794 file: this allows the display of the certificate about to be
1795 signed to be customised, to allow certain fields to be included
1796 or excluded and extension details. The old system didn't display
1797 multicharacter strings properly, omitted fields not in the policy
1798 and couldn't display additional details such as extensions.
1801 *) Function EC_POINTs_mul for multiple scalar multiplication
1802 of an arbitrary number of elliptic curve points
1803 \sum scalars[i]*points[i],
1804 optionally including the generator defined for the EC_GROUP:
1805 scalar*generator + \sum scalars[i]*points[i].
1807 EC_POINT_mul is a simple wrapper function for the typical case
1808 that the point list has just one item (besides the optional
1812 *) First EC_METHODs for curves over GF(p):
1814 EC_GFp_simple_method() uses the basic BN_mod_mul and BN_mod_sqr
1815 operations and provides various method functions that can also
1816 operate with faster implementations of modular arithmetic.
1818 EC_GFp_mont_method() reuses most functions that are part of
1819 EC_GFp_simple_method, but uses Montgomery arithmetic.
1821 [Bodo Moeller; point addition and point doubling
1822 implementation directly derived from source code provided by
1823 Lenka Fibikova <fibikova@exp-math.uni-essen.de>]
1825 *) Framework for elliptic curves (crypto/ec/ec.h, crypto/ec/ec_lcl.h,
1826 crypto/ec/ec_lib.c):
1828 Curves are EC_GROUP objects (with an optional group generator)
1829 based on EC_METHODs that are built into the library.
1831 Points are EC_POINT objects based on EC_GROUP objects.
1833 Most of the framework would be able to handle curves over arbitrary
1834 finite fields, but as there are no obvious types for fields other
1835 than GF(p), some functions are limited to that for now.
1838 *) Add the -HTTP option to s_server. It is similar to -WWW, but requires
1839 that the file contains a complete HTTP response.
1842 *) Add the ec directory to mkdef.pl and mkfiles.pl. In mkdef.pl
1843 change the def and num file printf format specifier from "%-40sXXX"
1844 to "%-39s XXX". The latter will always guarantee a space after the
1845 field while the former will cause them to run together if the field
1846 is 40 of more characters long.
1849 *) Constify the cipher and digest 'method' functions and structures
1850 and modify related functions to take constant EVP_MD and EVP_CIPHER
1854 *) Hide BN_CTX structure details in bn_lcl.h instead of publishing them
1855 in <openssl/bn.h>. Also further increase BN_CTX_NUM to 32.
1858 *) Modify EVP_Digest*() routines so they now return values. Although the
1859 internal software routines can never fail additional hardware versions
1863 *) Clean up crypto/err/err.h and change some error codes to avoid conflicts:
1865 Previously ERR_R_FATAL was too small and coincided with ERR_LIB_PKCS7
1866 (= ERR_R_PKCS7_LIB); it is now 64 instead of 32.
1869 ERR_R_NESTED_ASN1_ERROR
1871 ERR_R_MISSING_ASN1_EOS
1872 were 4 .. 9, conflicting with
1873 ERR_LIB_RSA (= ERR_R_RSA_LIB)
1875 ERR_LIB_PEM (= ERR_R_PEM_LIB).
1876 They are now 58 .. 63 (i.e., just below ERR_R_FATAL).
1878 Add new error code 'ERR_R_INTERNAL_ERROR'.
1881 *) Don't overuse locks in crypto/err/err.c: For data retrieval, CRYPTO_r_lock
1885 *) New option '-subj arg' for 'openssl req' and 'openssl ca'. This
1886 sets the subject name for a new request or supersedes the
1887 subject name in a given request. Formats that can be parsed are
1888 'CN=Some Name, OU=myOU, C=IT'
1890 'CN=Some Name/OU=myOU/C=IT'.
1892 Add options '-batch' and '-verbose' to 'openssl req'.
1893 [Massimiliano Pala <madwolf@hackmasters.net>]
1895 *) Introduce the possibility to access global variables through
1896 functions on platform were that's the best way to handle exporting
1897 global variables in shared libraries. To enable this functionality,
1898 one must configure with "EXPORT_VAR_AS_FN" or defined the C macro
1899 "OPENSSL_EXPORT_VAR_AS_FUNCTION" in crypto/opensslconf.h (the latter
1900 is normally done by Configure or something similar).
1902 To implement a global variable, use the macro OPENSSL_IMPLEMENT_GLOBAL
1903 in the source file (foo.c) like this:
1905 OPENSSL_IMPLEMENT_GLOBAL(int,foo)=1;
1906 OPENSSL_IMPLEMENT_GLOBAL(double,bar);
1908 To declare a global variable, use the macros OPENSSL_DECLARE_GLOBAL
1909 and OPENSSL_GLOBAL_REF in the header file (foo.h) like this:
1911 OPENSSL_DECLARE_GLOBAL(int,foo);
1912 #define foo OPENSSL_GLOBAL_REF(foo)
1913 OPENSSL_DECLARE_GLOBAL(double,bar);
1914 #define bar OPENSSL_GLOBAL_REF(bar)
1916 The #defines are very important, and therefore so is including the
1917 header file everywhere where the defined globals are used.
1919 The macro OPENSSL_EXPORT_VAR_AS_FUNCTION also affects the definition
1920 of ASN.1 items, but that structure is a bit different.
1922 The largest change is in util/mkdef.pl which has been enhanced with
1923 better and easier to understand logic to choose which symbols should
1924 go into the Windows .def files as well as a number of fixes and code
1925 cleanup (among others, algorithm keywords are now sorted
1926 lexicographically to avoid constant rewrites).
1929 *) In BN_div() keep a copy of the sign of 'num' before writing the
1930 result to 'rm' because if rm==num the value will be overwritten
1931 and produce the wrong result if 'num' is negative: this caused
1932 problems with BN_mod() and BN_nnmod().
1935 *) Function OCSP_request_verify(). This checks the signature on an
1936 OCSP request and verifies the signer certificate. The signer
1937 certificate is just checked for a generic purpose and OCSP request
1941 *) Add OCSP_check_validity() function to check the validity of OCSP
1942 responses. OCSP responses are prepared in real time and may only
1943 be a few seconds old. Simply checking that the current time lies
1944 between thisUpdate and nextUpdate max reject otherwise valid responses
1945 caused by either OCSP responder or client clock inaccuracy. Instead
1946 we allow thisUpdate and nextUpdate to fall within a certain period of
1947 the current time. The age of the response can also optionally be
1948 checked. Two new options -validity_period and -status_age added to
1952 *) If signature or public key algorithm is unrecognized print out its
1953 OID rather that just UNKNOWN.
1956 *) Change OCSP_cert_to_id() to tolerate a NULL subject certificate and
1957 OCSP_cert_id_new() a NULL serialNumber. This allows a partial certificate
1958 ID to be generated from the issuer certificate alone which can then be
1959 passed to OCSP_id_issuer_cmp().
1962 *) New compilation option ASN1_ITEM_FUNCTIONS. This causes the new
1963 ASN1 modules to export functions returning ASN1_ITEM pointers
1964 instead of the ASN1_ITEM structures themselves. This adds several
1965 new macros which allow the underlying ASN1 function/structure to
1966 be accessed transparently. As a result code should not use ASN1_ITEM
1967 references directly (such as &X509_it) but instead use the relevant
1968 macros (such as ASN1_ITEM_rptr(X509)). This option is to allow
1969 use of the new ASN1 code on platforms where exporting structures
1970 is problematical (for example in shared libraries) but exporting
1971 functions returning pointers to structures is not.
1974 *) Add support for overriding the generation of SSL/TLS session IDs.
1975 These callbacks can be registered either in an SSL_CTX or per SSL.
1976 The purpose of this is to allow applications to control, if they wish,
1977 the arbitrary values chosen for use as session IDs, particularly as it
1978 can be useful for session caching in multiple-server environments. A
1979 command-line switch for testing this (and any client code that wishes
1980 to use such a feature) has been added to "s_server".
1981 [Geoff Thorpe, Lutz Jaenicke]
1983 *) Modify mkdef.pl to recognise and parse preprocessor conditionals
1984 of the form '#if defined(...) || defined(...) || ...' and
1985 '#if !defined(...) && !defined(...) && ...'. This also avoids
1986 the growing number of special cases it was previously handling.
1989 *) Make all configuration macros available for application by making
1990 sure they are available in opensslconf.h, by giving them names starting
1991 with "OPENSSL_" to avoid conflicts with other packages and by making
1992 sure e_os2.h will cover all platform-specific cases together with
1994 Additionally, it is now possible to define configuration/platform-
1995 specific names (called "system identities"). In the C code, these
1996 are prefixed with "OPENSSL_SYSNAME_". e_os2.h will create another
1997 macro with the name beginning with "OPENSSL_SYS_", which is determined
1998 from "OPENSSL_SYSNAME_*" or compiler-specific macros depending on
2002 *) New option -set_serial to 'req' and 'x509' this allows the serial
2003 number to use to be specified on the command line. Previously self
2004 signed certificates were hard coded with serial number 0 and the
2005 CA options of 'x509' had to use a serial number in a file which was
2009 *) New options to 'ca' utility to support V2 CRL entry extensions.
2010 Currently CRL reason, invalidity date and hold instruction are
2011 supported. Add new CRL extensions to V3 code and some new objects.
2014 *) New function EVP_CIPHER_CTX_set_padding() this is used to
2015 disable standard block padding (aka PKCS#5 padding) in the EVP
2016 API, which was previously mandatory. This means that the data is
2017 not padded in any way and so the total length much be a multiple
2018 of the block size, otherwise an error occurs.
2021 *) Initial (incomplete) OCSP SSL support.
2024 *) New function OCSP_parse_url(). This splits up a URL into its host,
2025 port and path components: primarily to parse OCSP URLs. New -url
2026 option to ocsp utility.
2029 *) New nonce behavior. The return value of OCSP_check_nonce() now
2030 reflects the various checks performed. Applications can decide
2031 whether to tolerate certain situations such as an absent nonce
2032 in a response when one was present in a request: the ocsp application
2033 just prints out a warning. New function OCSP_add1_basic_nonce()
2034 this is to allow responders to include a nonce in a response even if
2035 the request is nonce-less.
2038 *) Disable stdin buffering in load_cert (apps/apps.c) so that no certs are
2039 skipped when using openssl x509 multiple times on a single input file,
2040 e.g. "(openssl x509 -out cert1; openssl x509 -out cert2) <certs".
2043 *) Make ASN1_UTCTIME_set_string() and ASN1_GENERALIZEDTIME_set_string()
2044 set string type: to handle setting ASN1_TIME structures. Fix ca
2045 utility to correctly initialize revocation date of CRLs.
2048 *) New option SSL_OP_CIPHER_SERVER_PREFERENCE allows the server to override
2049 the clients preferred ciphersuites and rather use its own preferences.
2050 Should help to work around M$ SGC (Server Gated Cryptography) bug in
2051 Internet Explorer by ensuring unchanged hash method during stepup.
2052 (Also replaces the broken/deactivated SSL_OP_NON_EXPORT_FIRST option.)
2055 *) Make mkdef.pl recognise all DECLARE_ASN1 macros, change rijndael
2056 to aes and add a new 'exist' option to print out symbols that don't
2060 *) Additional options to ocsp utility to allow flags to be set and
2061 additional certificates supplied.
2064 *) Add the option -VAfile to 'openssl ocsp', so the user can give the
2065 OCSP client a number of certificate to only verify the response
2069 *) Update Rijndael code to version 3.0 and change EVP AES ciphers to
2070 handle the new API. Currently only ECB, CBC modes supported. Add new
2073 Add TLS AES ciphersuites as described in RFC3268, "Advanced
2074 Encryption Standard (AES) Ciphersuites for Transport Layer
2075 Security (TLS)". (In beta versions of OpenSSL 0.9.7, these were
2076 not enabled by default and were not part of the "ALL" ciphersuite
2077 alias because they were not yet official; they could be
2078 explicitly requested by specifying the "AESdraft" ciphersuite
2079 group alias. In the final release of OpenSSL 0.9.7, the group
2080 alias is called "AES" and is part of "ALL".)
2081 [Ben Laurie, Steve Henson, Bodo Moeller]
2083 *) New function OCSP_copy_nonce() to copy nonce value (if present) from
2084 request to response.
2087 *) Functions for OCSP responders. OCSP_request_onereq_count(),
2088 OCSP_request_onereq_get0(), OCSP_onereq_get0_id() and OCSP_id_get0_info()
2089 extract information from a certificate request. OCSP_response_create()
2090 creates a response and optionally adds a basic response structure.
2091 OCSP_basic_add1_status() adds a complete single response to a basic
2092 response and returns the OCSP_SINGLERESP structure just added (to allow
2093 extensions to be included for example). OCSP_basic_add1_cert() adds a
2094 certificate to a basic response and OCSP_basic_sign() signs a basic
2095 response with various flags. New helper functions ASN1_TIME_check()
2096 (checks validity of ASN1_TIME structure) and ASN1_TIME_to_generalizedtime()
2097 (converts ASN1_TIME to GeneralizedTime).
2100 *) Various new functions. EVP_Digest() combines EVP_Digest{Init,Update,Final}()
2101 in a single operation. X509_get0_pubkey_bitstr() extracts the public_key
2102 structure from a certificate. X509_pubkey_digest() digests the public_key
2103 contents: this is used in various key identifiers.
2106 *) Make sk_sort() tolerate a NULL argument.
2107 [Steve Henson reported by Massimiliano Pala <madwolf@comune.modena.it>]
2109 *) New OCSP verify flag OCSP_TRUSTOTHER. When set the "other" certificates
2110 passed by the function are trusted implicitly. If any of them signed the
2111 response then it is assumed to be valid and is not verified.
2114 *) In PKCS7_set_type() initialise content_type in PKCS7_ENC_CONTENT
2115 to data. This was previously part of the PKCS7 ASN1 code. This
2116 was causing problems with OpenSSL created PKCS#12 and PKCS#7 structures.
2117 [Steve Henson, reported by Kenneth R. Robinette
2118 <support@securenetterm.com>]
2120 *) Add CRYPTO_push_info() and CRYPTO_pop_info() calls to new ASN1
2121 routines: without these tracing memory leaks is very painful.
2122 Fix leaks in PKCS12 and PKCS7 routines.
2125 *) Make X509_time_adj() cope with the new behaviour of ASN1_TIME_new().
2126 Previously it initialised the 'type' argument to V_ASN1_UTCTIME which
2127 effectively meant GeneralizedTime would never be used. Now it
2128 is initialised to -1 but X509_time_adj() now has to check the value
2129 and use ASN1_TIME_set() if the value is not V_ASN1_UTCTIME or
2130 V_ASN1_GENERALIZEDTIME, without this it always uses GeneralizedTime.
2131 [Steve Henson, reported by Kenneth R. Robinette
2132 <support@securenetterm.com>]
2134 *) Fixes to BN_to_ASN1_INTEGER when bn is zero. This would previously
2135 result in a zero length in the ASN1_INTEGER structure which was
2136 not consistent with the structure when d2i_ASN1_INTEGER() was used
2137 and would cause ASN1_INTEGER_cmp() to fail. Enhance s2i_ASN1_INTEGER()
2138 to cope with hex and negative integers. Fix bug in i2a_ASN1_INTEGER()
2139 where it did not print out a minus for negative ASN1_INTEGER.
2142 *) Add summary printout to ocsp utility. The various functions which
2143 convert status values to strings have been renamed to:
2144 OCSP_response_status_str(), OCSP_cert_status_str() and
2145 OCSP_crl_reason_str() and are no longer static. New options
2146 to verify nonce values and to disable verification. OCSP response
2147 printout format cleaned up.
2150 *) Add additional OCSP certificate checks. These are those specified
2151 in RFC2560. This consists of two separate checks: the CA of the
2152 certificate being checked must either be the OCSP signer certificate
2153 or the issuer of the OCSP signer certificate. In the latter case the
2154 OCSP signer certificate must contain the OCSP signing extended key
2155 usage. This check is performed by attempting to match the OCSP
2156 signer or the OCSP signer CA to the issuerNameHash and issuerKeyHash
2157 in the OCSP_CERTID structures of the response.
2160 *) Initial OCSP certificate verification added to OCSP_basic_verify()
2161 and related routines. This uses the standard OpenSSL certificate
2162 verify routines to perform initial checks (just CA validity) and
2163 to obtain the certificate chain. Then additional checks will be
2164 performed on the chain. Currently the root CA is checked to see
2165 if it is explicitly trusted for OCSP signing. This is used to set
2166 a root CA as a global signing root: that is any certificate that
2167 chains to that CA is an acceptable OCSP signing certificate.
2170 *) New '-extfile ...' option to 'openssl ca' for reading X.509v3
2171 extensions from a separate configuration file.
2172 As when reading extensions from the main configuration file,
2173 the '-extensions ...' option may be used for specifying the
2175 [Massimiliano Pala <madwolf@comune.modena.it>]
2177 *) New OCSP utility. Allows OCSP requests to be generated or
2178 read. The request can be sent to a responder and the output
2179 parsed, outputed or printed in text form. Not complete yet:
2180 still needs to check the OCSP response validity.
2183 *) New subcommands for 'openssl ca':
2184 'openssl ca -status <serial>' prints the status of the cert with
2185 the given serial number (according to the index file).
2186 'openssl ca -updatedb' updates the expiry status of certificates
2188 [Massimiliano Pala <madwolf@comune.modena.it>]
2190 *) New '-newreq-nodes' command option to CA.pl. This is like
2191 '-newreq', but calls 'openssl req' with the '-nodes' option
2192 so that the resulting key is not encrypted.
2193 [Damien Miller <djm@mindrot.org>]
2195 *) New configuration for the GNU Hurd.
2196 [Jonathan Bartlett <johnnyb@wolfram.com> via Richard Levitte]
2198 *) Initial code to implement OCSP basic response verify. This
2199 is currently incomplete. Currently just finds the signer's
2200 certificate and verifies the signature on the response.
2203 *) New SSLeay_version code SSLEAY_DIR to determine the compiled-in
2204 value of OPENSSLDIR. This is available via the new '-d' option
2205 to 'openssl version', and is also included in 'openssl version -a'.
2208 *) Allowing defining memory allocation callbacks that will be given
2209 file name and line number information in additional arguments
2210 (a const char* and an int). The basic functionality remains, as
2211 well as the original possibility to just replace malloc(),
2212 realloc() and free() by functions that do not know about these
2213 additional arguments. To register and find out the current
2214 settings for extended allocation functions, the following
2215 functions are provided:
2217 CRYPTO_set_mem_ex_functions
2218 CRYPTO_set_locked_mem_ex_functions
2219 CRYPTO_get_mem_ex_functions
2220 CRYPTO_get_locked_mem_ex_functions
2222 These work the same way as CRYPTO_set_mem_functions and friends.
2223 CRYPTO_get_[locked_]mem_functions now writes 0 where such an
2224 extended allocation function is enabled.
2225 Similarly, CRYPTO_get_[locked_]mem_ex_functions writes 0 where
2226 a conventional allocation function is enabled.
2227 [Richard Levitte, Bodo Moeller]
2229 *) Finish off removing the remaining LHASH function pointer casts.
2230 There should no longer be any prototype-casting required when using
2231 the LHASH abstraction, and any casts that remain are "bugs". See
2232 the callback types and macros at the head of lhash.h for details
2233 (and "OBJ_cleanup" in crypto/objects/obj_dat.c as an example).
2236 *) Add automatic query of EGD sockets in RAND_poll() for the unix variant.
2237 If /dev/[u]random devices are not available or do not return enough
2238 entropy, EGD style sockets (served by EGD or PRNGD) will automatically
2240 The locations /var/run/egd-pool, /dev/egd-pool, /etc/egd-pool, and
2241 /etc/entropy will be queried once each in this sequence, quering stops
2242 when enough entropy was collected without querying more sockets.
2245 *) Change the Unix RAND_poll() variant to be able to poll several
2246 random devices, as specified by DEVRANDOM, until a sufficient amount
2247 of data has been collected. We spend at most 10 ms on each file
2248 (select timeout) and read in non-blocking mode. DEVRANDOM now
2249 defaults to the list "/dev/urandom", "/dev/random", "/dev/srandom"
2250 (previously it was just the string "/dev/urandom"), so on typical
2251 platforms the 10 ms delay will never occur.
2252 Also separate out the Unix variant to its own file, rand_unix.c.
2253 For VMS, there's a currently-empty rand_vms.c.
2256 *) Move OCSP client related routines to ocsp_cl.c. These
2257 provide utility functions which an application needing
2258 to issue a request to an OCSP responder and analyse the
2259 response will typically need: as opposed to those which an
2260 OCSP responder itself would need which will be added later.
2262 OCSP_request_sign() signs an OCSP request with an API similar
2263 to PKCS7_sign(). OCSP_response_status() returns status of OCSP
2264 response. OCSP_response_get1_basic() extracts basic response
2265 from response. OCSP_resp_find_status(): finds and extracts status
2266 information from an OCSP_CERTID structure (which will be created
2267 when the request structure is built). These are built from lower
2268 level functions which work on OCSP_SINGLERESP structures but
2269 wont normally be used unless the application wishes to examine
2270 extensions in the OCSP response for example.
2272 Replace nonce routines with a pair of functions.
2273 OCSP_request_add1_nonce() adds a nonce value and optionally
2274 generates a random value. OCSP_check_nonce() checks the
2275 validity of the nonce in an OCSP response.
2278 *) Change function OCSP_request_add() to OCSP_request_add0_id().
2279 This doesn't copy the supplied OCSP_CERTID and avoids the
2280 need to free up the newly created id. Change return type
2281 to OCSP_ONEREQ to return the internal OCSP_ONEREQ structure.
2282 This can then be used to add extensions to the request.
2283 Deleted OCSP_request_new(), since most of its functionality
2284 is now in OCSP_REQUEST_new() (and the case insensitive name
2285 clash) apart from the ability to set the request name which
2286 will be added elsewhere.
2289 *) Update OCSP API. Remove obsolete extensions argument from
2290 various functions. Extensions are now handled using the new
2291 OCSP extension code. New simple OCSP HTTP function which
2292 can be used to send requests and parse the response.
2295 *) Fix the PKCS#7 (S/MIME) code to work with new ASN1. Two new
2296 ASN1_ITEM structures help with sign and verify. PKCS7_ATTR_SIGN
2297 uses the special reorder version of SET OF to sort the attributes
2298 and reorder them to match the encoded order. This resolves a long
2299 standing problem: a verify on a PKCS7 structure just after signing
2300 it used to fail because the attribute order did not match the
2301 encoded order. PKCS7_ATTR_VERIFY does not reorder the attributes:
2302 it uses the received order. This is necessary to tolerate some broken
2303 software that does not order SET OF. This is handled by encoding
2304 as a SEQUENCE OF but using implicit tagging (with UNIVERSAL class)
2305 to produce the required SET OF.
2308 *) Have mk1mf.pl generate the macros OPENSSL_BUILD_SHLIBCRYPTO and
2309 OPENSSL_BUILD_SHLIBSSL and use them appropriately in the header
2310 files to get correct declarations of the ASN.1 item variables.
2313 *) Rewrite of PKCS#12 code to use new ASN1 functionality. Replace many
2314 PKCS#12 macros with real functions. Fix two unrelated ASN1 bugs:
2315 asn1_check_tlen() would sometimes attempt to use 'ctx' when it was
2316 NULL and ASN1_TYPE was not dereferenced properly in asn1_ex_c2i().
2317 New ASN1 macro: DECLARE_ASN1_ITEM() which just declares the relevant
2318 ASN1_ITEM and no wrapper functions.
2321 *) New functions or ASN1_item_d2i_fp() and ASN1_item_d2i_bio(). These
2322 replace the old function pointer based I/O routines. Change most of
2323 the *_d2i_bio() and *_d2i_fp() functions to use these.
2326 *) Enhance mkdef.pl to be more accepting about spacing in C preprocessor
2327 lines, recognice more "algorithms" that can be deselected, and make
2328 it complain about algorithm deselection that isn't recognised.
2331 *) New ASN1 functions to handle dup, sign, verify, digest, pack and
2332 unpack operations in terms of ASN1_ITEM. Modify existing wrappers
2333 to use new functions. Add NO_ASN1_OLD which can be set to remove
2334 some old style ASN1 functions: this can be used to determine if old
2335 code will still work when these eventually go away.
2338 *) New extension functions for OCSP structures, these follow the
2339 same conventions as certificates and CRLs.
2342 *) New function X509V3_add1_i2d(). This automatically encodes and
2343 adds an extension. Its behaviour can be customised with various
2344 flags to append, replace or delete. Various wrappers added for
2345 certifcates and CRLs.
2348 *) Fix to avoid calling the underlying ASN1 print routine when
2349 an extension cannot be parsed. Correct a typo in the
2350 OCSP_SERVICELOC extension. Tidy up print OCSP format.
2353 *) Make mkdef.pl parse some of the ASN1 macros and add apropriate
2354 entries for variables.
2357 *) Add functionality to apps/openssl.c for detecting locking
2358 problems: As the program is single-threaded, all we have
2359 to do is register a locking callback using an array for
2360 storing which locks are currently held by the program.
2363 *) Use a lock around the call to CRYPTO_get_ex_new_index() in
2364 SSL_get_ex_data_X509_STORE_idx(), which is used in
2365 ssl_verify_cert_chain() and thus can be called at any time
2366 during TLS/SSL handshakes so that thread-safety is essential.
2367 Unfortunately, the ex_data design is not at all suited
2368 for multi-threaded use, so it probably should be abolished.
2371 *) Added Broadcom "ubsec" ENGINE to OpenSSL.
2372 [Broadcom, tweaked and integrated by Geoff Thorpe]
2374 *) Move common extension printing code to new function
2375 X509V3_print_extensions(). Reorganise OCSP print routines and
2376 implement some needed OCSP ASN1 functions. Add OCSP extensions.
2379 *) New function X509_signature_print() to remove duplication in some
2383 *) Add a special meaning when SET OF and SEQUENCE OF flags are both
2384 set (this was treated exactly the same as SET OF previously). This
2385 is used to reorder the STACK representing the structure to match the
2386 encoding. This will be used to get round a problem where a PKCS7
2387 structure which was signed could not be verified because the STACK
2388 order did not reflect the encoded order.
2391 *) Reimplement the OCSP ASN1 module using the new code.
2394 *) Update the X509V3 code to permit the use of an ASN1_ITEM structure
2395 for its ASN1 operations. The old style function pointers still exist
2396 for now but they will eventually go away.
2399 *) Merge in replacement ASN1 code from the ASN1 branch. This almost
2400 completely replaces the old ASN1 functionality with a table driven
2401 encoder and decoder which interprets an ASN1_ITEM structure describing
2402 the ASN1 module. Compatibility with the existing ASN1 API (i2d,d2i) is
2403 largely maintained. Almost all of the old asn1_mac.h macro based ASN1
2404 has also been converted to the new form.
2407 *) Change BN_mod_exp_recp so that negative moduli are tolerated
2408 (the sign is ignored). Similarly, ignore the sign in BN_MONT_CTX_set
2409 so that BN_mod_exp_mont and BN_mod_exp_mont_word work
2410 for negative moduli.
2413 *) Fix BN_uadd and BN_usub: Always return non-negative results instead
2414 of not touching the result's sign bit.
2417 *) BN_div bugfix: If the result is 0, the sign (res->neg) must not be
2421 *) Changed the LHASH code to use prototypes for callbacks, and created
2422 macros to declare and implement thin (optionally static) functions
2423 that provide type-safety and avoid function pointer casting for the
2424 type-specific callbacks.
2427 *) Added Kerberos Cipher Suites to be used with TLS, as written in
2429 [Veers Staats <staatsvr@asc.hpc.mil>,
2430 Jeffrey Altman <jaltman@columbia.edu>, via Richard Levitte]
2432 *) Reformat the FAQ so the different questions and answers can be divided
2433 in sections depending on the subject.
2436 *) Have the zlib compression code load ZLIB.DLL dynamically under
2440 *) New function BN_mod_sqrt for computing square roots modulo a prime
2441 (using the probabilistic Tonelli-Shanks algorithm unless
2442 p == 3 (mod 4) or p == 5 (mod 8), which are cases that can
2443 be handled deterministically).
2444 [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller]
2446 *) Make BN_mod_inverse faster by explicitly handling small quotients
2447 in the Euclid loop. (Speed gain about 20% for small moduli [256 or
2448 512 bits], about 30% for larger ones [1024 or 2048 bits].)
2451 *) New function BN_kronecker.
2454 *) Fix BN_gcd so that it works on negative inputs; the result is
2455 positive unless both parameters are zero.
2456 Previously something reasonably close to an infinite loop was
2457 possible because numbers could be growing instead of shrinking
2458 in the implementation of Euclid's algorithm.
2461 *) Fix BN_is_word() and BN_is_one() macros to take into account the
2462 sign of the number in question.
2464 Fix BN_is_word(a,w) to work correctly for w == 0.
2466 The old BN_is_word(a,w) macro is now called BN_abs_is_word(a,w)
2467 because its test if the absolute value of 'a' equals 'w'.
2468 Note that BN_abs_is_word does *not* handle w == 0 reliably;
2469 it exists mostly for use in the implementations of BN_is_zero(),
2470 BN_is_one(), and BN_is_word().
2473 *) New function BN_swap.
2476 *) Use BN_nnmod instead of BN_mod in crypto/bn/bn_exp.c so that
2477 the exponentiation functions are more likely to produce reasonable
2478 results on negative inputs.
2481 *) Change BN_mod_mul so that the result is always non-negative.
2482 Previously, it could be negative if one of the factors was negative;
2483 I don't think anyone really wanted that behaviour.
2486 *) Move BN_mod_... functions into new file crypto/bn/bn_mod.c
2487 (except for exponentiation, which stays in crypto/bn/bn_exp.c,
2488 and BN_mod_mul_reciprocal, which stays in crypto/bn/bn_recp.c)
2489 and add new functions:
2498 BN_mod_lshift1_quick
2502 These functions always generate non-negative results.
2504 BN_nnmod otherwise is like BN_mod (if BN_mod computes a remainder r
2505 such that |m| < r < 0, BN_nnmod will output rem + |m| instead).
2507 BN_mod_XXX_quick(r, a, [b,] m) generates the same result as
2508 BN_mod_XXX(r, a, [b,] m, ctx), but requires that a [and b]
2509 be reduced modulo m.
2510 [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller]
2513 The following entry accidentily appeared in the CHANGES file
2514 distributed with OpenSSL 0.9.7. The modifications described in
2515 it do *not* apply to OpenSSL 0.9.7.
2517 *) Remove a few calls to bn_wexpand() in BN_sqr() (the one in there
2518 was actually never needed) and in BN_mul(). The removal in BN_mul()
2519 required a small change in bn_mul_part_recursive() and the addition
2520 of the functions bn_cmp_part_words(), bn_sub_part_words() and
2521 bn_add_part_words(), which do the same thing as bn_cmp_words(),
2522 bn_sub_words() and bn_add_words() except they take arrays with
2527 *) In 'openssl passwd', verify passwords read from the terminal
2528 unless the '-salt' option is used (which usually means that
2529 verification would just waste user's time since the resulting
2530 hash is going to be compared with some given password hash)
2531 or the new '-noverify' option is used.
2533 This is an incompatible change, but it does not affect
2534 non-interactive use of 'openssl passwd' (passwords on the command
2535 line, '-stdin' option, '-in ...' option) and thus should not
2539 *) Remove all references to RSAref, since there's no more need for it.
2542 *) Make DSO load along a path given through an environment variable
2543 (SHLIB_PATH) with shl_load().
2546 *) Constify the ENGINE code as a result of BIGNUM constification.
2547 Also constify the RSA code and most things related to it. In a
2548 few places, most notable in the depth of the ASN.1 code, ugly
2549 casts back to non-const were required (to be solved at a later
2553 *) Make it so the openssl application has all engines loaded by default.
2556 *) Constify the BIGNUM routines a little more.
2559 *) Add the following functions:
2561 ENGINE_load_cswift()
2563 ENGINE_load_atalla()
2565 ENGINE_load_builtin_engines()
2567 That way, an application can itself choose if external engines that
2568 are built-in in OpenSSL shall ever be used or not. The benefit is
2569 that applications won't have to be linked with libdl or other dso
2570 libraries unless it's really needed.
2572 Changed 'openssl engine' to load all engines on demand.
2573 Changed the engine header files to avoid the duplication of some
2574 declarations (they differed!).
2577 *) 'openssl engine' can now list capabilities.
2580 *) Better error reporting in 'openssl engine'.
2583 *) Never call load_dh_param(NULL) in s_server.
2586 *) Add engine application. It can currently list engines by name and
2587 identity, and test if they are actually available.
2590 *) Improve RPM specification file by forcing symbolic linking and making
2591 sure the installed documentation is also owned by root.root.
2592 [Damien Miller <djm@mindrot.org>]
2594 *) Give the OpenSSL applications more possibilities to make use of
2595 keys (public as well as private) handled by engines.
2598 *) Add OCSP code that comes from CertCo.
2601 *) Add VMS support for the Rijndael code.
2604 *) Added untested support for Nuron crypto accelerator.
2607 *) Add support for external cryptographic devices. This code was
2608 previously distributed separately as the "engine" branch.
2609 [Geoff Thorpe, Richard Levitte]
2611 *) Rework the filename-translation in the DSO code. It is now possible to
2612 have far greater control over how a "name" is turned into a filename
2613 depending on the operating environment and any oddities about the
2614 different shared library filenames on each system.
2617 *) Support threads on FreeBSD-elf in Configure.
2620 *) Fix for SHA1 assembly problem with MASM: it produces
2621 warnings about corrupt line number information when assembling
2622 with debugging information. This is caused by the overlapping
2624 [Bernd Matthes <mainbug@celocom.de>, Steve Henson]
2627 NCONF_get_number() has no error checking at all. As a replacement,
2628 NCONF_get_number_e() is defined (_e for "error checking") and is
2629 promoted strongly. The old NCONF_get_number is kept around for
2630 binary backward compatibility.
2631 Make it possible for methods to load from something other than a BIO,
2632 by providing a function pointer that is given a name instead of a BIO.
2633 For example, this could be used to load configuration data from an
2637 *) Fix for non blocking accept BIOs. Added new I/O special reason
2638 BIO_RR_ACCEPT to cover this case. Previously use of accept BIOs
2639 with non blocking I/O was not possible because no retry code was
2640 implemented. Also added new SSL code SSL_WANT_ACCEPT to cover
2644 *) Added the beginnings of Rijndael support.
2647 *) Fix for bug in DirectoryString mask setting. Add support for
2648 X509_NAME_print_ex() in 'req' and X509_print_ex() function
2649 to allow certificate printing to more controllable, additional
2650 'certopt' option to 'x509' to allow new printing options to be
2654 *) Clean old EAY MD5 hack from e_os.h.
2657 Changes between 0.9.6j and 0.9.6k [30 Sep 2003]
2659 *) Fix various bugs revealed by running the NISCC test suite:
2661 Stop out of bounds reads in the ASN1 code when presented with
2662 invalid tags (CAN-2003-0543 and CAN-2003-0544).
2664 If verify callback ignores invalid public key errors don't try to check
2665 certificate signature with the NULL public key.
2669 *) Fix various bugs revealed by running the NISCC test suite:
2671 Stop out of bounds reads in the ASN1 code when presented with
2672 invalid tags (CAN-2003-0543 and CAN-2003-0544).
2674 If verify callback ignores invalid public key errors don't try to check
2675 certificate signature with the NULL public key.
2679 *) In ssl3_accept() (ssl/s3_srvr.c) only accept a client certificate
2680 if the server requested one: as stated in TLS 1.0 and SSL 3.0
2684 *) In ssl3_get_client_hello() (ssl/s3_srvr.c), tolerate additional
2685 extra data after the compression methods not only for TLS 1.0
2686 but also for SSL 3.0 (as required by the specification).
2687 [Bodo Moeller; problem pointed out by Matthias Loepfe]
2689 *) Change X509_certificate_type() to mark the key as exported/exportable
2690 when it's 512 *bits* long, not 512 bytes.
2693 Changes between 0.9.6i and 0.9.6j [10 Apr 2003]
2695 *) Countermeasure against the Klima-Pokorny-Rosa extension of
2696 Bleichbacher's attack on PKCS #1 v1.5 padding: treat
2697 a protocol version number mismatch like a decryption error
2698 in ssl3_get_client_key_exchange (ssl/s3_srvr.c).
2701 *) Turn on RSA blinding by default in the default implementation
2702 to avoid a timing attack. Applications that don't want it can call
2703 RSA_blinding_off() or use the new flag RSA_FLAG_NO_BLINDING.
2704 They would be ill-advised to do so in most cases.
2705 [Ben Laurie, Steve Henson, Geoff Thorpe, Bodo Moeller]
2707 *) Change RSA blinding code so that it works when the PRNG is not
2708 seeded (in this case, the secret RSA exponent is abused as
2709 an unpredictable seed -- if it is not unpredictable, there
2710 is no point in blinding anyway). Make RSA blinding thread-safe
2711 by remembering the creator's thread ID in rsa->blinding and
2712 having all other threads use local one-time blinding factors
2713 (this requires more computation than sharing rsa->blinding, but
2714 avoids excessive locking; and if an RSA object is not shared
2715 between threads, blinding will still be very fast).
2718 Changes between 0.9.6h and 0.9.6i [19 Feb 2003]
2720 *) In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked
2721 via timing by performing a MAC computation even if incorrrect
2722 block cipher padding has been found. This is a countermeasure
2723 against active attacks where the attacker has to distinguish
2724 between bad padding and a MAC verification error. (CAN-2003-0078)
2726 [Bodo Moeller; problem pointed out by Brice Canvel (EPFL),
2727 Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and
2728 Martin Vuagnoux (EPFL, Ilion)]
2730 Changes between 0.9.6g and 0.9.6h [5 Dec 2002]
2732 *) New function OPENSSL_cleanse(), which is used to cleanse a section of
2733 memory from it's contents. This is done with a counter that will
2734 place alternating values in each byte. This can be used to solve
2735 two issues: 1) the removal of calls to memset() by highly optimizing
2736 compilers, and 2) cleansing with other values than 0, since those can
2737 be read through on certain media, for example a swap space on disk.
2740 *) Bugfix: client side session caching did not work with external caching,
2741 because the session->cipher setting was not restored when reloading
2742 from the external cache. This problem was masked, when
2743 SSL_OP_NETSCAPE_REUSE_CIPHER_CHANGE_BUG (part of SSL_OP_ALL) was set.
2744 (Found by Steve Haslam <steve@araqnid.ddts.net>.)
2747 *) Fix client_certificate (ssl/s2_clnt.c): The permissible total
2748 length of the REQUEST-CERTIFICATE message is 18 .. 34, not 17 .. 33.
2749 [Zeev Lieber <zeev-l@yahoo.com>]
2751 *) Undo an undocumented change introduced in 0.9.6e which caused
2752 repeated calls to OpenSSL_add_all_ciphers() and
2753 OpenSSL_add_all_digests() to be ignored, even after calling
2757 *) Change the default configuration reader to deal with last line not
2758 being properly terminated.
2761 *) Change X509_NAME_cmp() so it applies the special rules on handling
2762 DN values that are of type PrintableString, as well as RDNs of type
2763 emailAddress where the value has the type ia5String.
2764 [stefank@valicert.com via Richard Levitte]
2766 *) Add a SSL_SESS_CACHE_NO_INTERNAL_STORE flag to take over half
2767 the job SSL_SESS_CACHE_NO_INTERNAL_LOOKUP was inconsistently
2768 doing, define a new flag (SSL_SESS_CACHE_NO_INTERNAL) to be
2769 the bitwise-OR of the two for use by the majority of applications
2770 wanting this behaviour, and update the docs. The documented
2771 behaviour and actual behaviour were inconsistent and had been
2772 changing anyway, so this is more a bug-fix than a behavioural
2774 [Geoff Thorpe, diagnosed by Nadav Har'El]
2776 *) Don't impose a 16-byte length minimum on session IDs in ssl/s3_clnt.c
2777 (the SSL 3.0 and TLS 1.0 specifications allow any length up to 32 bytes).
2780 *) Fix initialization code race conditions in
2781 SSLv23_method(), SSLv23_client_method(), SSLv23_server_method(),
2782 SSLv2_method(), SSLv2_client_method(), SSLv2_server_method(),
2783 SSLv3_method(), SSLv3_client_method(), SSLv3_server_method(),
2784 TLSv1_method(), TLSv1_client_method(), TLSv1_server_method(),
2785 ssl2_get_cipher_by_char(),
2786 ssl3_get_cipher_by_char().
2787 [Patrick McCormick <patrick@tellme.com>, Bodo Moeller]
2789 *) Reorder cleanup sequence in SSL_CTX_free(): only remove the ex_data after
2790 the cached sessions are flushed, as the remove_cb() might use ex_data
2791 contents. Bug found by Sam Varshavchik <mrsam@courier-mta.com>
2792 (see [openssl.org #212]).
2793 [Geoff Thorpe, Lutz Jaenicke]
2795 *) Fix typo in OBJ_txt2obj which incorrectly passed the content
2796 length, instead of the encoding length to d2i_ASN1_OBJECT.
2799 Changes between 0.9.6f and 0.9.6g [9 Aug 2002]
2801 *) [In 0.9.6g-engine release:]
2802 Fix crypto/engine/vendor_defns/cswift.h for WIN32 (use '_stdcall').
2803 [Lynn Gazis <lgazis@rainbow.com>]
2805 Changes between 0.9.6e and 0.9.6f [8 Aug 2002]
2807 *) Fix ASN1 checks. Check for overflow by comparing with LONG_MAX
2808 and get fix the header length calculation.
2809 [Florian Weimer <Weimer@CERT.Uni-Stuttgart.DE>,
2810 Alon Kantor <alonk@checkpoint.com> (and others),
2813 *) Use proper error handling instead of 'assertions' in buffer
2814 overflow checks added in 0.9.6e. This prevents DoS (the
2815 assertions could call abort()).
2816 [Arne Ansper <arne@ats.cyber.ee>, Bodo Moeller]
2818 Changes between 0.9.6d and 0.9.6e [30 Jul 2002]
2820 *) Add various sanity checks to asn1_get_length() to reject
2821 the ASN1 length bytes if they exceed sizeof(long), will appear
2822 negative or the content length exceeds the length of the
2824 [Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>]
2826 *) Fix cipher selection routines: ciphers without encryption had no flags
2827 for the cipher strength set and where therefore not handled correctly
2828 by the selection routines (PR #130).
2831 *) Fix EVP_dsa_sha macro.
2835 SSL_OP_DONT_INSERT_EMPTY_FRAGMENTS
2836 for disabling the SSL 3.0/TLS 1.0 CBC vulnerability countermeasure
2837 that was added in OpenSSL 0.9.6d.
2839 As the countermeasure turned out to be incompatible with some
2840 broken SSL implementations, the new option is part of SSL_OP_ALL.
2841 SSL_OP_ALL is usually employed when compatibility with weird SSL
2842 implementations is desired (e.g. '-bugs' option to 's_client' and
2843 's_server'), so the new option is automatically set in many
2847 *) Changes in security patch:
2849 Changes marked "(CHATS)" were sponsored by the Defense Advanced
2850 Research Projects Agency (DARPA) and Air Force Research Laboratory,
2851 Air Force Materiel Command, USAF, under agreement number
2854 *) Add various sanity checks to asn1_get_length() to reject
2855 the ASN1 length bytes if they exceed sizeof(long), will appear
2856 negative or the content length exceeds the length of the
2857 supplied buffer. (CAN-2002-0659)
2858 [Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>]
2860 *) Assertions for various potential buffer overflows, not known to
2862 [Ben Laurie (CHATS)]
2864 *) Various temporary buffers to hold ASCII versions of integers were
2865 too small for 64 bit platforms. (CAN-2002-0655)
2866 [Matthew Byng-Maddick <mbm@aldigital.co.uk> and Ben Laurie (CHATS)>
2868 *) Remote buffer overflow in SSL3 protocol - an attacker could
2869 supply an oversized session ID to a client. (CAN-2002-0656)
2870 [Ben Laurie (CHATS)]
2872 *) Remote buffer overflow in SSL2 protocol - an attacker could
2873 supply an oversized client master key. (CAN-2002-0656)
2874 [Ben Laurie (CHATS)]
2876 Changes between 0.9.6c and 0.9.6d [9 May 2002]
2878 *) Fix crypto/asn1/a_sign.c so that 'parameters' is omitted (not
2879 encoded as NULL) with id-dsa-with-sha1.
2880 [Nils Larsch <nla@trustcenter.de>; problem pointed out by Bodo Moeller]
2882 *) Check various X509_...() return values in apps/req.c.
2883 [Nils Larsch <nla@trustcenter.de>]
2885 *) Fix BASE64 decode (EVP_DecodeUpdate) for data with CR/LF ended lines:
2886 an end-of-file condition would erronously be flagged, when the CRLF
2887 was just at the end of a processed block. The bug was discovered when
2888 processing data through a buffering memory BIO handing the data to a
2889 BASE64-decoding BIO. Bug fund and patch submitted by Pavel Tsekov
2890 <ptsekov@syntrex.com> and Nedelcho Stanev.
2893 *) Implement a countermeasure against a vulnerability recently found
2894 in CBC ciphersuites in SSL 3.0/TLS 1.0: Send an empty fragment
2895 before application data chunks to avoid the use of known IVs
2896 with data potentially chosen by the attacker.
2899 *) Fix length checks in ssl3_get_client_hello().
2902 *) TLS/SSL library bugfix: use s->s3->in_read_app_data differently
2903 to prevent ssl3_read_internal() from incorrectly assuming that
2904 ssl3_read_bytes() found application data while handshake
2905 processing was enabled when in fact s->s3->in_read_app_data was
2906 merely automatically cleared during the initial handshake.
2907 [Bodo Moeller; problem pointed out by Arne Ansper <arne@ats.cyber.ee>]
2909 *) Fix object definitions for Private and Enterprise: they were not
2910 recognized in their shortname (=lowercase) representation. Extend
2911 obj_dat.pl to issue an error when using undefined keywords instead
2912 of silently ignoring the problem (Svenning Sorensen
2913 <sss@sss.dnsalias.net>).
2916 *) Fix DH_generate_parameters() so that it works for 'non-standard'
2917 generators, i.e. generators other than 2 and 5. (Previously, the
2918 code did not properly initialise the 'add' and 'rem' values to
2919 BN_generate_prime().)
2921 In the new general case, we do not insist that 'generator' is
2922 actually a primitive root: This requirement is rather pointless;
2923 a generator of the order-q subgroup is just as good, if not
2927 *) Map new X509 verification errors to alerts. Discovered and submitted by
2928 Tom Wu <tom@arcot.com>.
2931 *) Fix ssl3_pending() (ssl/s3_lib.c) to prevent SSL_pending() from
2932 returning non-zero before the data has been completely received
2933 when using non-blocking I/O.
2934 [Bodo Moeller; problem pointed out by John Hughes]
2936 *) Some of the ciphers missed the strength entry (SSL_LOW etc).
2937 [Ben Laurie, Lutz Jaenicke]
2939 *) Fix bug in SSL_clear(): bad sessions were not removed (found by
2940 Yoram Zahavi <YoramZ@gilian.com>).
2943 *) Add information about CygWin 1.3 and on, and preserve proper
2944 configuration for the versions before that.
2945 [Corinna Vinschen <vinschen@redhat.com> and Richard Levitte]
2947 *) Make removal from session cache (SSL_CTX_remove_session()) more robust:
2948 check whether we deal with a copy of a session and do not delete from
2949 the cache in this case. Problem reported by "Izhar Shoshani Levi"
2950 <izhar@checkpoint.com>.
2953 *) Do not store session data into the internal session cache, if it
2954 is never intended to be looked up (SSL_SESS_CACHE_NO_INTERNAL_LOOKUP
2955 flag is set). Proposed by Aslam <aslam@funk.com>.
2958 *) Have ASN1_BIT_STRING_set_bit() really clear a bit when the requested
2962 *) [In 0.9.6d-engine release:]
2963 Fix a crashbug and a logic bug in hwcrhk_load_pubkey().
2964 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte]
2966 *) Add the configuration target linux-s390x.
2967 [Neale Ferguson <Neale.Ferguson@SoftwareAG-USA.com> via Richard Levitte]
2969 *) The earlier bugfix for the SSL3_ST_SW_HELLO_REQ_C case of
2970 ssl3_accept (ssl/s3_srvr.c) incorrectly used a local flag
2971 variable as an indication that a ClientHello message has been
2972 received. As the flag value will be lost between multiple
2973 invocations of ssl3_accept when using non-blocking I/O, the
2974 function may not be aware that a handshake has actually taken
2975 place, thus preventing a new session from being added to the
2978 To avoid this problem, we now set s->new_session to 2 instead of
2979 using a local variable.
2980 [Lutz Jaenicke, Bodo Moeller]
2982 *) Bugfix: Return -1 from ssl3_get_server_done (ssl3/s3_clnt.c)
2983 if the SSL_R_LENGTH_MISMATCH error is detected.
2984 [Geoff Thorpe, Bodo Moeller]
2986 *) New 'shared_ldflag' column in Configure platform table.
2989 *) Fix EVP_CIPHER_mode macro.
2990 ["Dan S. Camper" <dan@bti.net>]
2992 *) Fix ssl3_read_bytes (ssl/s3_pkt.c): To ignore messages of unknown
2993 type, we must throw them away by setting rr->length to 0.
2994 [D P Chang <dpc@qualys.com>]
2996 Changes between 0.9.6b and 0.9.6c [21 dec 2001]
2998 *) Fix BN_rand_range bug pointed out by Dominikus Scherkl
2999 <Dominikus.Scherkl@biodata.com>. (The previous implementation
3000 worked incorrectly for those cases where range = 10..._2 and
3001 3*range is two bits longer than range.)
3004 *) Only add signing time to PKCS7 structures if it is not already
3008 *) Fix crypto/objects/objects.h: "ld-ce" should be "id-ce",
3009 OBJ_ld_ce should be OBJ_id_ce.
3010 Also some ip-pda OIDs in crypto/objects/objects.txt were
3011 incorrect (cf. RFC 3039).
3012 [Matt Cooper, Frederic Giudicelli, Bodo Moeller]
3014 *) Release CRYPTO_LOCK_DYNLOCK when CRYPTO_destroy_dynlockid()
3015 returns early because it has nothing to do.
3016 [Andy Schneider <andy.schneider@bjss.co.uk>]
3018 *) [In 0.9.6c-engine release:]
3019 Fix mutex callback return values in crypto/engine/hw_ncipher.c.
3020 [Andy Schneider <andy.schneider@bjss.co.uk>]
3022 *) [In 0.9.6c-engine release:]
3023 Add support for Cryptographic Appliance's keyserver technology.
3024 (Use engine 'keyclient')
3025 [Cryptographic Appliances and Geoff Thorpe]
3027 *) Add a configuration entry for OS/390 Unix. The C compiler 'c89'
3028 is called via tools/c89.sh because arguments have to be
3029 rearranged (all '-L' options must appear before the first object
3031 [Richard Shapiro <rshapiro@abinitio.com>]
3033 *) [In 0.9.6c-engine release:]
3034 Add support for Broadcom crypto accelerator cards, backported
3036 [Broadcom, Nalin Dahyabhai <nalin@redhat.com>, Mark Cox]
3038 *) [In 0.9.6c-engine release:]
3039 Add support for SureWare crypto accelerator cards from
3040 Baltimore Technologies. (Use engine 'sureware')
3041 [Baltimore Technologies and Mark Cox]
3043 *) [In 0.9.6c-engine release:]
3044 Add support for crypto accelerator cards from Accelerated
3045 Encryption Processing, www.aep.ie. (Use engine 'aep')
3046 [AEP Inc. and Mark Cox]
3048 *) Add a configuration entry for gcc on UnixWare.
3049 [Gary Benson <gbenson@redhat.com>]
3051 *) Change ssl/s2_clnt.c and ssl/s2_srvr.c so that received handshake
3052 messages are stored in a single piece (fixed-length part and
3053 variable-length part combined) and fix various bugs found on the way.
3056 *) Disable caching in BIO_gethostbyname(), directly use gethostbyname()
3057 instead. BIO_gethostbyname() does not know what timeouts are
3058 appropriate, so entries would stay in cache even when they have
3060 [Bodo Moeller; problem pointed out by Rich Salz <rsalz@zolera.com>
3062 *) Change ssl23_get_client_hello (ssl/s23_srvr.c) behaviour when
3063 faced with a pathologically small ClientHello fragment that does
3064 not contain client_version: Instead of aborting with an error,
3065 simply choose the highest available protocol version (i.e.,
3066 TLS 1.0 unless it is disabled). In practice, ClientHello
3067 messages are never sent like this, but this change gives us
3068 strictly correct behaviour at least for TLS.
3071 *) Fix SSL handshake functions and SSL_clear() such that SSL_clear()
3072 never resets s->method to s->ctx->method when called from within
3073 one of the SSL handshake functions.
3074 [Bodo Moeller; problem pointed out by Niko Baric]
3076 *) In ssl3_get_client_hello (ssl/s3_srvr.c), generate a fatal alert
3077 (sent using the client's version number) if client_version is
3078 smaller than the protocol version in use. Also change
3079 ssl23_get_client_hello (ssl/s23_srvr.c) to select TLS 1.0 if
3080 the client demanded SSL 3.0 but only TLS 1.0 is enabled; then
3081 the client will at least see that alert.
3084 *) Fix ssl3_get_message (ssl/s3_both.c) to handle message fragmentation
3088 *) Avoid infinite loop in ssl3_get_message (ssl/s3_both.c) if a
3089 client receives HelloRequest while in a handshake.
3090 [Bodo Moeller; bug noticed by Andy Schneider <andy.schneider@bjss.co.uk>]
3092 *) Bugfix in ssl3_accept (ssl/s3_srvr.c): Case SSL3_ST_SW_HELLO_REQ_C
3093 should end in 'break', not 'goto end' which circuments various
3094 cleanups done in state SSL_ST_OK. But session related stuff
3095 must be disabled for SSL_ST_OK in the case that we just sent a
3098 Also avoid some overhead by not calling ssl_init_wbio_buffer()
3099 before just sending a HelloRequest.
3100 [Bodo Moeller, Eric Rescorla <ekr@rtfm.com>]
3102 *) Fix ssl/s3_enc.c, ssl/t1_enc.c and ssl/s3_pkt.c so that we don't
3103 reveal whether illegal block cipher padding was found or a MAC
3104 verification error occured. (Neither SSLerr() codes nor alerts
3105 are directly visible to potential attackers, but the information
3106 may leak via logfiles.)
3108 Similar changes are not required for the SSL 2.0 implementation
3109 because the number of padding bytes is sent in clear for SSL 2.0,
3110 and the extra bytes are just ignored. However ssl/s2_pkt.c
3111 failed to verify that the purported number of padding bytes is in
3115 *) Add OpenUNIX-8 support including shared libraries
3116 (Boyd Lynn Gerber <gerberb@zenez.com>).
3119 *) Improve RSA_padding_check_PKCS1_OAEP() check again to avoid
3120 'wristwatch attack' using huge encoding parameters (cf.
3121 James H. Manger's CRYPTO 2001 paper). Note that the
3122 RSA_PKCS1_OAEP_PADDING case of RSA_private_decrypt() does not use
3123 encoding parameters and hence was not vulnerable.
3126 *) BN_sqr() bug fix.
3127 [Ulf Möller, reported by Jim Ellis <jim.ellis@cavium.com>]
3129 *) Rabin-Miller test analyses assume uniformly distributed witnesses,
3130 so use BN_pseudo_rand_range() instead of using BN_pseudo_rand()
3131 followed by modular reduction.
3132 [Bodo Moeller; pointed out by Adam Young <AYoung1@NCSUS.JNJ.COM>]
3134 *) Add BN_pseudo_rand_range() with obvious functionality: BN_rand_range()
3135 equivalent based on BN_pseudo_rand() instead of BN_rand().
3138 *) s3_srvr.c: allow sending of large client certificate lists (> 16 kB).
3139 This function was broken, as the check for a new client hello message
3140 to handle SGC did not allow these large messages.
3141 (Tracked down by "Douglas E. Engert" <deengert@anl.gov>.)
3144 *) Add alert descriptions for TLSv1 to SSL_alert_desc_string[_long]().
3147 *) Fix buggy behaviour of BIO_get_num_renegotiates() and BIO_ctrl()
3148 for BIO_C_GET_WRITE_BUF_SIZE ("Stephen Hinton" <shinton@netopia.com>).
3151 *) Rework the configuration and shared library support for Tru64 Unix.
3152 The configuration part makes use of modern compiler features and
3153 still retains old compiler behavior for those that run older versions
3154 of the OS. The shared library support part includes a variant that
3155 uses the RPATH feature, and is available through the special
3156 configuration target "alpha-cc-rpath", which will never be selected
3158 [Tim Mooney <mooney@dogbert.cc.ndsu.NoDak.edu> via Richard Levitte]
3160 *) In ssl3_get_key_exchange (ssl/s3_clnt.c), call ssl3_get_message()
3161 with the same message size as in ssl3_get_certificate_request().
3162 Otherwise, if no ServerKeyExchange message occurs, CertificateRequest
3163 messages might inadvertently be reject as too long.
3164 [Petr Lampa <lampa@fee.vutbr.cz>]
3166 *) Enhanced support for IA-64 Unix platforms (well, Linux and HP-UX).
3169 *) Modified SSL library such that the verify_callback that has been set
3170 specificly for an SSL object with SSL_set_verify() is actually being
3171 used. Before the change, a verify_callback set with this function was
3172 ignored and the verify_callback() set in the SSL_CTX at the time of
3173 the call was used. New function X509_STORE_CTX_set_verify_cb() introduced
3174 to allow the necessary settings.
3177 *) Initialize static variable in crypto/dsa/dsa_lib.c and crypto/dh/dh_lib.c
3178 explicitly to NULL, as at least on Solaris 8 this seems not always to be
3179 done automatically (in contradiction to the requirements of the C
3180 standard). This made problems when used from OpenSSH.
3183 *) In OpenSSL 0.9.6a and 0.9.6b, crypto/dh/dh_key.c ignored
3184 dh->length and always used
3186 BN_rand_range(priv_key, dh->p).
3188 BN_rand_range() is not necessary for Diffie-Hellman, and this
3189 specific range makes Diffie-Hellman unnecessarily inefficient if
3190 dh->length (recommended exponent length) is much smaller than the
3191 length of dh->p. We could use BN_rand_range() if the order of
3192 the subgroup was stored in the DH structure, but we only have
3197 BN_rand(priv_key, l, ...)
3199 where 'l' is dh->length if this is defined, or BN_num_bits(dh->p)-1
3205 RSA_eay_public_encrypt
3206 RSA_eay_private_decrypt
3207 RSA_eay_private_encrypt (signing)
3208 RSA_eay_public_decrypt (signature verification)
3210 (default implementations for RSA_public_encrypt,
3211 RSA_private_decrypt, RSA_private_encrypt, RSA_public_decrypt),
3212 always reject numbers >= n.
3215 *) In crypto/rand/md_rand.c, use a new short-time lock CRYPTO_LOCK_RAND2
3216 to synchronize access to 'locking_thread'. This is necessary on
3217 systems where access to 'locking_thread' (an 'unsigned long'
3218 variable) is not atomic.
3221 *) In crypto/rand/md_rand.c, set 'locking_thread' to current thread's ID
3222 *before* setting the 'crypto_lock_rand' flag. The previous code had
3223 a race condition if 0 is a valid thread ID.
3224 [Travis Vitek <vitek@roguewave.com>]
3226 *) Add support for shared libraries under Irix.
3227 [Albert Chin-A-Young <china@thewrittenword.com>]
3229 *) Add configuration option to build on Linux on both big-endian and
3231 [Ralf Baechle <ralf@uni-koblenz.de>]
3233 *) Add the possibility to create shared libraries on HP-UX.
3236 Changes between 0.9.6a and 0.9.6b [9 Jul 2001]
3238 *) Change ssleay_rand_bytes (crypto/rand/md_rand.c)
3239 to avoid a SSLeay/OpenSSL PRNG weakness pointed out by
3240 Markku-Juhani O. Saarinen <markku-juhani.saarinen@nokia.com>:
3241 PRNG state recovery was possible based on the output of
3242 one PRNG request appropriately sized to gain knowledge on
3243 'md' followed by enough consecutive 1-byte PRNG requests
3244 to traverse all of 'state'.
3246 1. When updating 'md_local' (the current thread's copy of 'md')
3247 during PRNG output generation, hash all of the previous
3248 'md_local' value, not just the half used for PRNG output.
3250 2. Make the number of bytes from 'state' included into the hash
3251 independent from the number of PRNG bytes requested.
3253 The first measure alone would be sufficient to avoid
3254 Markku-Juhani's attack. (Actually it had never occurred
3255 to me that the half of 'md_local' used for chaining was the
3256 half from which PRNG output bytes were taken -- I had always
3257 assumed that the secret half would be used.) The second
3258 measure makes sure that additional data from 'state' is never
3259 mixed into 'md_local' in small portions; this heuristically
3260 further strengthens the PRNG.
3263 *) Fix crypto/bn/asm/mips3.s.
3266 *) When only the key is given to "enc", the IV is undefined. Print out
3267 an error message in this case.
3270 *) Handle special case when X509_NAME is empty in X509 printing routines.
3273 *) In dsa_do_verify (crypto/dsa/dsa_ossl.c), verify that r and s are
3274 positive and less than q.
3277 *) Don't change *pointer in CRYPTO_add_lock() is add_lock_callback is
3278 used: it isn't thread safe and the add_lock_callback should handle
3280 [Paul Rose <Paul.Rose@bridge.com>]
3282 *) Verify that incoming data obeys the block size in
3283 ssl3_enc (ssl/s3_enc.c) and tls1_enc (ssl/t1_enc.c).
3287 [Ulf Möller, Bodo Möller]
3289 *) The countermeasure against Bleichbacher's attack on PKCS #1 v1.5
3290 RSA encryption was accidentally removed in s3_srvr.c in OpenSSL 0.9.5
3291 when fixing the server behaviour for backwards-compatible 'client
3292 hello' messages. (Note that the attack is impractical against
3293 SSL 3.0 and TLS 1.0 anyway because length and version checking
3294 means that the probability of guessing a valid ciphertext is
3295 around 2^-40; see section 5 in Bleichenbacher's CRYPTO '98
3298 Before 0.9.5, the countermeasure (hide the error by generating a
3299 random 'decryption result') did not work properly because
3300 ERR_clear_error() was missing, meaning that SSL_get_error() would
3301 detect the supposedly ignored error.
3303 Both problems are now fixed.
3306 *) In crypto/bio/bf_buff.c, increase DEFAULT_BUFFER_SIZE to 4096
3307 (previously it was 1024).
3310 *) Fix for compatibility mode trust settings: ignore trust settings
3311 unless some valid trust or reject settings are present.
3314 *) Fix for blowfish EVP: its a variable length cipher.
3317 *) Fix various bugs related to DSA S/MIME verification. Handle missing
3318 parameters in DSA public key structures and return an error in the
3319 DSA routines if parameters are absent.
3322 *) In versions up to 0.9.6, RAND_file_name() resorted to file ".rnd"
3323 in the current directory if neither $RANDFILE nor $HOME was set.
3324 RAND_file_name() in 0.9.6a returned NULL in this case. This has
3325 caused some confusion to Windows users who haven't defined $HOME.
3326 Thus RAND_file_name() is changed again: e_os.h can define a
3327 DEFAULT_HOME, which will be used if $HOME is not set.
3328 For Windows, we use "C:"; on other platforms, we still require
3329 environment variables.
3331 *) Move 'if (!initialized) RAND_poll()' into regions protected by
3332 CRYPTO_LOCK_RAND. This is not strictly necessary, but avoids
3333 having multiple threads call RAND_poll() concurrently.
3336 *) In crypto/rand/md_rand.c, replace 'add_do_not_lock' flag by a
3337 combination of a flag and a thread ID variable.
3338 Otherwise while one thread is in ssleay_rand_bytes (which sets the
3339 flag), *other* threads can enter ssleay_add_bytes without obeying
3340 the CRYPTO_LOCK_RAND lock (and may even illegally release the lock
3341 that they do not hold after the first thread unsets add_do_not_lock).
3344 *) Change bctest again: '-x' expressions are not available in all
3348 Changes between 0.9.6 and 0.9.6a [5 Apr 2001]
3350 *) Fix a couple of memory leaks in PKCS7_dataDecode()
3351 [Steve Henson, reported by Heyun Zheng <hzheng@atdsprint.com>]
3353 *) Change Configure and Makefiles to provide EXE_EXT, which will contain
3354 the default extension for executables, if any. Also, make the perl
3355 scripts that use symlink() to test if it really exists and use "cp"
3356 if it doesn't. All this made OpenSSL compilable and installable in
3360 *) Fix for asn1_GetSequence() for indefinite length constructed data.
3361 If SEQUENCE is length is indefinite just set c->slen to the total
3362 amount of data available.
3363 [Steve Henson, reported by shige@FreeBSD.org]
3364 [This change does not apply to 0.9.7.]
3366 *) Change bctest to avoid here-documents inside command substitution
3367 (workaround for FreeBSD /bin/sh bug).
3368 For compatibility with Ultrix, avoid shell functions (introduced
3369 in the bctest version that searches along $PATH).
3372 *) Rename 'des_encrypt' to 'des_encrypt1'. This avoids the clashes
3373 with des_encrypt() defined on some operating systems, like Solaris
3377 *) Check the result of RSA-CRT (see D. Boneh, R. DeMillo, R. Lipton:
3378 On the Importance of Eliminating Errors in Cryptographic
3379 Computations, J. Cryptology 14 (2001) 2, 101-119,
3380 http://theory.stanford.edu/~dabo/papers/faults.ps.gz).
3383 *) MIPS assembler BIGNUM division bug fix.
3386 *) Disabled incorrect Alpha assembler code.
3389 *) Fix PKCS#7 decode routines so they correctly update the length
3390 after reading an EOC for the EXPLICIT tag.
3392 [This change does not apply to 0.9.7.]
3394 *) Fix bug in PKCS#12 key generation routines. This was triggered
3395 if a 3DES key was generated with a 0 initial byte. Include
3396 PKCS12_BROKEN_KEYGEN compilation option to retain the old
3397 (but broken) behaviour.
3400 *) Enhance bctest to search for a working bc along $PATH and print
3402 [Tim Rice <tim@multitalents.net> via Richard Levitte]
3404 *) Fix memory leaks in err.c: free err_data string if necessary;
3405 don't write to the wrong index in ERR_set_error_data.
3408 *) Implement ssl23_peek (analogous to ssl23_read), which previously
3412 *) Replace rdtsc with _emit statements for VC++ version 5.
3413 [Jeremy Cooper <jeremy@baymoo.org>]
3415 *) Make it possible to reuse SSLv2 sessions.
3418 *) In copy_email() check for >= 0 as a return value for
3419 X509_NAME_get_index_by_NID() since 0 is a valid index.
3420 [Steve Henson reported by Massimiliano Pala <madwolf@opensca.org>]
3422 *) Avoid coredump with unsupported or invalid public keys by checking if
3423 X509_get_pubkey() fails in PKCS7_verify(). Fix memory leak when
3424 PKCS7_verify() fails with non detached data.
3427 *) Don't use getenv in library functions when run as setuid/setgid.
3428 New function OPENSSL_issetugid().
3431 *) Avoid false positives in memory leak detection code (crypto/mem_dbg.c)
3432 due to incorrect handling of multi-threading:
3434 1. Fix timing glitch in the MemCheck_off() portion of CRYPTO_mem_ctrl().
3436 2. Fix logical glitch in is_MemCheck_on() aka CRYPTO_is_mem_check_on().
3438 3. Count how many times MemCheck_off() has been called so that
3439 nested use can be treated correctly. This also avoids
3440 inband-signalling in the previous code (which relied on the
3441 assumption that thread ID 0 is impossible).
3444 *) Add "-rand" option also to s_client and s_server.
3447 *) Fix CPU detection on Irix 6.x.
3448 [Kurt Hockenbury <khockenb@stevens-tech.edu> and
3449 "Bruce W. Forsberg" <bruce.forsberg@baesystems.com>]
3451 *) Fix X509_NAME bug which produced incorrect encoding if X509_NAME
3454 [This change does not apply to 0.9.7.]
3456 *) Use the cached encoding of an X509_NAME structure rather than
3457 copying it. This is apparently the reason for the libsafe "errors"
3458 but the code is actually correct.
3461 *) Add new function BN_rand_range(), and fix DSA_sign_setup() to prevent
3462 Bleichenbacher's DSA attack.
3463 Extend BN_[pseudo_]rand: As before, top=1 forces the highest two bits
3464 to be set and top=0 forces the highest bit to be set; top=-1 is new
3465 and leaves the highest bit random.
3466 [Ulf Moeller, Bodo Moeller]
3468 *) In the NCONF_...-based implementations for CONF_... queries
3469 (crypto/conf/conf_lib.c), if the input LHASH is NULL, avoid using
3470 a temporary CONF structure with the data component set to NULL
3471 (which gives segmentation faults in lh_retrieve).
3472 Instead, use NULL for the CONF pointer in CONF_get_string and
3473 CONF_get_number (which may use environment variables) and directly
3474 return NULL from CONF_get_section.
3477 *) Fix potential buffer overrun for EBCDIC.
3480 *) Tolerate nonRepudiation as being valid for S/MIME signing and certSign
3481 keyUsage if basicConstraints absent for a CA.
3484 *) Make SMIME_write_PKCS7() write mail header values with a format that
3485 is more generally accepted (no spaces before the semicolon), since
3486 some programs can't parse those values properly otherwise. Also make
3487 sure BIO's that break lines after each write do not create invalid
3491 *) Make the CRL encoding routines work with empty SEQUENCE OF. The
3492 macros previously used would not encode an empty SEQUENCE OF
3493 and break the signature.
3495 [This change does not apply to 0.9.7.]
3497 *) Zero the premaster secret after deriving the master secret in
3501 *) Add some EVP_add_digest_alias registrations (as found in
3502 OpenSSL_add_all_digests()) to SSL_library_init()
3503 aka OpenSSL_add_ssl_algorithms(). This provides improved
3504 compatibility with peers using X.509 certificates
3505 with unconventional AlgorithmIdentifier OIDs.
3508 *) Fix for Irix with NO_ASM.
3509 ["Bruce W. Forsberg" <bruce.forsberg@baesystems.com>]
3511 *) ./config script fixes.
3512 [Ulf Moeller, Richard Levitte]
3514 *) Fix 'openssl passwd -1'.
3517 *) Change PKCS12_key_gen_asc() so it can cope with non null
3518 terminated strings whose length is passed in the passlen
3519 parameter, for example from PEM callbacks. This was done
3520 by adding an extra length parameter to asc2uni().
3521 [Steve Henson, reported by <oddissey@samsung.co.kr>]
3523 *) Fix C code generated by 'openssl dsaparam -C': If a BN_bin2bn
3524 call failed, free the DSA structure.
3527 *) Fix to uni2asc() to cope with zero length Unicode strings.
3528 These are present in some PKCS#12 files.
3531 *) Increase s2->wbuf allocation by one byte in ssl2_new (ssl/s2_lib.c).
3532 Otherwise do_ssl_write (ssl/s2_pkt.c) will write beyond buffer limits
3533 when writing a 32767 byte record.
3534 [Bodo Moeller; problem reported by Eric Day <eday@concentric.net>]
3536 *) In RSA_eay_public_{en,ed}crypt and RSA_eay_mod_exp (rsa_eay.c),
3537 obtain lock CRYPTO_LOCK_RSA before setting rsa->_method_mod_{n,p,q}.
3539 (RSA objects have a reference count access to which is protected
3540 by CRYPTO_LOCK_RSA [see rsa_lib.c, s3_srvr.c, ssl_cert.c, ssl_rsa.c],
3541 so they are meant to be shared between threads.)
3542 [Bodo Moeller, Geoff Thorpe; original patch submitted by
3543 "Reddie, Steven" <Steven.Reddie@ca.com>]
3545 *) Fix a deadlock in CRYPTO_mem_leaks().
3548 *) Use better test patterns in bntest.
3551 *) rand_win.c fix for Borland C.
3554 *) BN_rshift bugfix for n == 0.
3557 *) Add a 'bctest' script that checks for some known 'bc' bugs
3558 so that 'make test' does not abort just because 'bc' is broken.
3561 *) Store verify_result within SSL_SESSION also for client side to
3562 avoid potential security hole. (Re-used sessions on the client side
3563 always resulted in verify_result==X509_V_OK, not using the original
3564 result of the server certificate verification.)
3567 *) Fix ssl3_pending: If the record in s->s3->rrec is not of type
3568 SSL3_RT_APPLICATION_DATA, return 0.
3569 Similarly, change ssl2_pending to return 0 if SSL_in_init(s) is true.
3573 Both ssl2_peek and ssl3_peek, which were totally broken in earlier
3574 releases, have been re-implemented by renaming the previous
3575 implementations of ssl2_read and ssl3_read to ssl2_read_internal
3576 and ssl3_read_internal, respectively, and adding 'peek' parameters
3577 to them. The new ssl[23]_{read,peek} functions are calls to
3578 ssl[23]_read_internal with the 'peek' flag set appropriately.
3579 A 'peek' parameter has also been added to ssl3_read_bytes, which
3580 does the actual work for ssl3_read_internal.
3583 *) Initialise "ex_data" member of RSA/DSA/DH structures prior to calling
3584 the method-specific "init()" handler. Also clean up ex_data after
3585 calling the method-specific "finish()" handler. Previously, this was
3586 happening the other way round.
3589 *) Increase BN_CTX_NUM (the number of BIGNUMs in a BN_CTX) to 16.
3590 The previous value, 12, was not always sufficient for BN_mod_exp().
3593 *) Make sure that shared libraries get the internal name engine with
3594 the full version number and not just 0. This should mark the
3595 shared libraries as not backward compatible. Of course, this should
3596 be changed again when we can guarantee backward binary compatibility.
3599 *) Fix typo in get_cert_by_subject() in by_dir.c
projects / openssl.git / blob