5 Changes between 0.9.7c and 0.9.8 [xx XXX xxxx]
7 *) Constify all or almost all d2i, c2i, s2i and r2i functions, along with
8 associated ASN1, EVP and SSL functions and old ASN1 macros.
10 *) BN_zero() only needs to set 'top' and 'neg' to zero for correct results,
11 and this should never fail. So the return value from the use of
12 BN_set_word() (which can fail due to needless expansion) is now deprecated;
13 if OPENSSL_NO_DEPRECATED is defined, BN_zero() is a void macro.
16 *) BN_CTX_get() should return zero-valued bignums, providing the same
17 initialised value as BN_new().
18 [Geoff Thorpe, suggested by Ulf Möller]
20 *) Support for inhibitAnyPolicy certificate extension.
23 *) An audit of the BIGNUM code is underway, for which debugging code is
24 enabled when BN_DEBUG is defined. This makes stricter enforcements on what
25 is considered valid when processing BIGNUMs, and causes execution to
26 assert() when a problem is discovered. If BN_DEBUG_RAND is defined,
27 further steps are taken to deliberately pollute unused data in BIGNUM
28 structures to try and expose faulty code further on. For now, openssl will
29 (in its default mode of operation) continue to tolerate the inconsistent
30 forms that it has tolerated in the past, but authors and packagers should
31 consider trying openssl and their own applications when compiled with
32 these debugging symbols defined. It will help highlight potential bugs in
33 their own code, and will improve the test coverage for OpenSSL itself. At
34 some point, these tighter rules will become openssl's default to improve
35 maintainability, though the assert()s and other overheads will remain only
36 in debugging configurations. See bn.h for more details.
37 [Geoff Thorpe, Nils Larsch, Ulf Möller]
39 *) BN_CTX_init() has been deprecated, as BN_CTX is an opaque structure
40 that can only be obtained through BN_CTX_new() (which implicitly
41 initialises it). The presence of this function only made it possible
42 to overwrite an existing structure (and cause memory leaks).
45 *) Because of the callback-based approach for implementing LHASH as a
46 template type, lh_insert() adds opaque objects to hash-tables and
47 lh_doall() or lh_doall_arg() are typically used with a destructor callback
48 to clean up those corresponding objects before destroying the hash table
49 (and losing the object pointers). So some over-zealous constifications in
50 LHASH have been relaxed so that lh_insert() does not take (nor store) the
51 objects as "const" and the lh_doall[_arg] callback wrappers are not
52 prototyped to have "const" restrictions on the object pointers they are
53 given (and so aren't required to cast them away any more).
56 *) The tmdiff.h API was so ugly and minimal that our own timing utility
57 (speed) prefers to use its own implementation. The two implementations
58 haven't been consolidated as yet (volunteers?) but the tmdiff API has had
59 its object type properly exposed (MS_TM) instead of casting to/from "char
60 *". This may still change yet if someone realises MS_TM and "ms_time_***"
61 aren't necessarily the greatest nomenclatures - but this is what was used
62 internally to the implementation so I've used that for now.
65 *) Ensure that deprecated functions do not get compiled when
66 OPENSSL_NO_DEPRECATED is defined. Some "openssl" subcommands and a few of
67 the self-tests were still using deprecated key-generation functions so
68 these have been updated also.
71 *) Reorganise PKCS#7 code to separate the digest location functionality
72 into PKCS7_find_digest(), digest addtion into PKCS7_bio_add_digest().
73 New function PKCS7_set_digest() to set the digest type for PKCS#7
74 digestedData type. Add additional code to correctly generate the
75 digestedData type and add support for this type in PKCS7 initialization
79 *) New function PKCS7_set0_type_other() this initializes a PKCS7
80 structure of type "other".
83 *) Fix prime generation loop in crypto/bn/bn_prime.pl by making
84 sure the loop does correctly stop and breaking ("division by zero")
85 modulus operations are not performed. The (pre-generated) prime
86 table crypto/bn/bn_prime.h was already correct, but it could not be
87 re-generated on some platforms because of the "division by zero"
88 situation in the script.
91 *) Update support for ECC-based TLS ciphersuites according to
92 draft-ietf-tls-ecc-03.txt: the KDF1 key derivation function with
93 SHA-1 now is only used for "small" curves (where the
94 representation of a field element takes up to 24 bytes); for
95 larger curves, the field element resulting from ECDH is directly
96 used as premaster secret.
97 [Douglas Stebila (Sun Microsystems Laboratories)]
99 *) Add code for kP+lQ timings to crypto/ec/ectest.c, and add SEC2
100 curve secp160r1 to the tests.
101 [Douglas Stebila (Sun Microsystems Laboratories)]
103 *) Add the possibility to load symbols globally with DSO.
104 [Götz Babin-Ebell <babin-ebell@trustcenter.de> via Richard Levitte]
106 *) Add the functions ERR_set_mark() and ERR_pop_to_mark() for better
107 control of the error stack.
110 *) Add support for STORE in ENGINE.
113 *) Add the STORE type. The intention is to provide a common interface
114 to certificate and key stores, be they simple file-based stores, or
115 HSM-type store, or LDAP stores, or...
116 NOTE: The code is currently UNTESTED and isn't really used anywhere.
119 *) Add a generic structure called OPENSSL_ITEM. This can be used to
120 pass a list of arguments to any function as well as provide a way
121 for a function to pass data back to the caller.
124 *) Add the functions BUF_strndup() and BUF_memdup(). BUF_strndup()
125 works like BUF_strdup() but can be used to duplicate a portion of
126 a string. The copy gets NUL-terminated. BUF_memdup() duplicates
130 *) Add the function sk_find_ex() which works like sk_find(), but will
131 return an index to an element even if an exact match couldn't be
132 found. The index is guaranteed to point at the element where the
133 searched-for key would be inserted to preserve sorting order.
136 *) Add the function OBJ_bsearch_ex() which works like OBJ_bsearch() but
137 takes an extra flags argument for optional functionality. Currently,
138 the following flags are defined:
140 OBJ_BSEARCH_VALUE_ON_NOMATCH
141 This one gets OBJ_bsearch_ex() to return a pointer to the first
142 element where the comparing function returns a negative or zero
145 OBJ_BSEARCH_FIRST_VALUE_ON_MATCH
146 This one gets OBJ_bsearch_ex() to return a pointer to the first
147 element where the comparing function returns zero. This is useful
148 if there are more than one element where the comparing function
152 *) Make it possible to create self-signed certificates with 'openssl ca'
153 in such a way that the self-signed certificate becomes part of the
154 CA database and uses the same mechanisms for serial number generation
155 as all other certificate signing. The new flag '-selfsign' enables
156 this functionality. Adapt CA.sh and CA.pl.in.
159 *) Add functionality to check the public key of a certificate request
160 against a given private. This is useful to check that a certificate
161 request can be signed by that key (self-signing).
164 *) Make it possible to have multiple active certificates with the same
165 subject in the CA index file. This is done only if the keyword
166 'unique_subject' is set to 'no' in the main CA section (default
167 if 'CA_default') of the configuration file. The value is saved
168 with the database itself in a separate index attribute file,
169 named like the index file with '.attr' appended to the name.
172 *) Generate muti valued AVAs using '+' notation in config files for
176 *) Support for nameConstraints certificate extension.
179 *) Support for policyConstraints certificate extension.
182 *) Support for policyMappings certificate extension.
185 *) Fixed a typo bug that would cause ENGINE_set_default() to set an
186 ENGINE as defaults for all supported algorithms irrespective of
187 the 'flags' parameter. 'flags' is now honoured, so applications
188 should make sure they are passing it correctly.
191 *) Make sure the default DSA_METHOD implementation only uses its
192 dsa_mod_exp() and/or bn_mod_exp() handlers if they are non-NULL,
193 and change its own handlers to be NULL so as to remove unnecessary
194 indirection. This lets alternative implementations fallback to the
195 default implementation more easily.
198 *) Support for directoryName in GeneralName related extensions
202 *) Make it possible to link applications using Makefile.shared.
203 Make that possible even when linking against static libraries!
206 *) Support for single pass processing for S/MIME signing. This now
207 means that S/MIME signing can be done from a pipe, in addition
208 cleartext signing (multipart/signed type) is effectively streaming
209 and the signed data does not need to be all held in memory.
211 This is done with a new flag PKCS7_STREAM. When this flag is set
212 PKCS7_sign() only initializes the PKCS7 structure and the actual signing
213 is done after the data is output (and digests calculated) in
217 *) Add full support for -rpath/-R, both in shared libraries and
218 applications, at least on the platforms where it's known how
222 *) In crypto/ec/ec_mult.c, implement fast point multiplication with
223 precomputation, based on wNAF splitting: EC_GROUP_precompute_mult()
224 will now compute a table of multiples of the generator that
225 makes subsequent invocations of EC_POINTs_mul() or EC_POINT_mul()
226 faster (notably in the case of a single point multiplication,
228 [Nils Larsch, Bodo Moeller]
230 *) IPv6 support for certificate extensions. The various extensions
231 which use the IP:a.b.c.d can now take IPv6 addresses using the
232 formats of RFC1884 2.2 . IPv6 addresses are now also displayed
236 *) Added an ENGINE that implements RSA by performing private key
237 exponentiations with the GMP library. The conversions to and from
238 GMP's mpz_t format aren't optimised nor are any montgomery forms
239 cached, and on x86 it appears OpenSSL's own performance has caught up.
240 However there are likely to be other architectures where GMP could
241 provide a boost. This ENGINE is not built in by default, but it can be
242 specified at Configure time and should be accompanied by the necessary
243 linker additions, eg;
244 ./config -DOPENSSL_USE_GMP -lgmp
247 *) "openssl engine" will not display ENGINE/DSO load failure errors when
248 testing availability of engines with "-t" - the old behaviour is
249 produced by increasing the feature's verbosity with "-tt".
252 *) ECDSA routines: under certain error conditions uninitialized BN objects
253 could be freed. Solution: make sure initialization is performed early
254 enough. (Reported and fix supplied by Nils Larsch <nla@trustcenter.de>
258 *) Key-generation can now be implemented in RSA_METHOD, DSA_METHOD
259 and DH_METHOD (eg. by ENGINE implementations) to override the normal
260 software implementations. For DSA and DH, parameter generation can
261 also be overriden by providing the appropriate method callbacks.
264 *) Change the "progress" mechanism used in key-generation and
265 primality testing to functions that take a new BN_GENCB pointer in
266 place of callback/argument pairs. The new API functions have "_ex"
267 postfixes and the older functions are reimplemented as wrappers for
268 the new ones. The OPENSSL_NO_DEPRECATED symbol can be used to hide
269 declarations of the old functions to help (graceful) attempts to
270 migrate to the new functions. Also, the new key-generation API
271 functions operate on a caller-supplied key-structure and return
272 success/failure rather than returning a key or NULL - this is to
273 help make "keygen" another member function of RSA_METHOD etc.
275 Example for using the new callback interface:
277 int (*my_callback)(int a, int b, BN_GENCB *cb) = ...;
281 BN_GENCB_set(&my_cb, my_callback, my_arg);
283 return BN_is_prime_ex(some_bignum, BN_prime_checks, NULL, &cb);
284 /* For the meaning of a, b in calls to my_callback(), see the
285 * documentation of the function that calls the callback.
286 * cb will point to my_cb; my_arg can be retrieved as cb->arg.
287 * my_callback should return 1 if it wants BN_is_prime_ex()
288 * to continue, or 0 to stop.
293 *) Change the ZLIB compression method to be stateful, and make it
294 available to TLS with the number defined in
295 draft-ietf-tls-compression-04.txt.
298 *) Add the ASN.1 structures and functions for CertificatePair, which
299 is defined as follows (according to X.509_4thEditionDraftV6.pdf):
301 CertificatePair ::= SEQUENCE {
302 forward [0] Certificate OPTIONAL,
303 reverse [1] Certificate OPTIONAL,
304 -- at least one of the pair shall be present -- }
306 Also implement the PEM functions to read and write certificate
307 pairs, and defined the PEM tag as "CERTIFICATE PAIR".
309 This needed to be defined, mostly for the sake of the LDAP
310 attribute crossCertificatePair, but may prove useful elsewhere as
314 *) Make it possible to inhibit symlinking of shared libraries in
315 Makefile.shared, for Cygwin's sake.
318 *) Extend the BIGNUM API by creating new macros that behave like
321 void BN_set_sign(BIGNUM *a, int neg);
322 int BN_get_sign(const BIGNUM *a);
324 and avoid the need to access 'a->neg' directly in applications.
325 [Nils Larsch <nla@trustcenter.de>]
327 *) Implement fast modular reduction for pseudo-Mersenne primes
328 used in NIST curves (crypto/bn/bn_nist.c, crypto/ec/ecp_nist.c).
329 EC_GROUP_new_curve_GFp() will now automatically use this
331 [Nils Larsch <nla@trustcenter.de>]
333 *) Add new lock type (CRYPTO_LOCK_BN).
336 *) Change the ENGINE framework to automatically load engines
337 dynamically from specific directories unless they could be
338 found to already be built in or loaded. Move all the
339 current engines except for the cryptodev one to a new
341 The engines in engines/ are built as shared libraries if
342 the "shared" options was given to ./Configure or ./config.
343 Otherwise, they are inserted in libcrypto.a.
344 /usr/local/ssl/engines is the default directory for dynamic
345 engines, but that can be overriden at configure time through
346 the usual use of --prefix and/or --openssldir, and at run
347 time with the environment variable OPENSSL_ENGINES.
348 [Geoff Thorpe and Richard Levitte]
350 *) Add Makefile.shared, a helper makefile to build shared
351 libraries. Addapt Makefile.org.
354 *) Add version info to Win32 DLLs.
355 [Peter 'Luna' Runestig" <peter@runestig.com>]
357 *) Add new 'medium level' PKCS#12 API. Certificates and keys
358 can be added using this API to created arbitrary PKCS#12
359 files while avoiding the low level API.
361 New options to PKCS12_create(), key or cert can be NULL and
362 will then be omitted from the output file. The encryption
363 algorithm NIDs can be set to -1 for no encryption, the mac
364 iteration count can be set to 0 to omit the mac.
366 Enhance pkcs12 utility by making the -nokeys and -nocerts
367 options work when creating a PKCS#12 file. New option -nomac
368 to omit the mac, NONE can be set for an encryption algorithm.
369 New code is modified to use the enhanced PKCS12_create()
370 instead of the low level API.
373 *) Extend ASN1 encoder to support indefinite length constructed
374 encoding. This can output sequences tags and octet strings in
375 this form. Modify pk7_asn1.c to support indefinite length
376 encoding. This is experimental and needs additional code to
377 be useful, such as an ASN1 bio and some enhanced streaming
380 Extend template encode functionality so that tagging is passed
381 down to the template encoder.
384 *) Let 'openssl req' fail if an argument to '-newkey' is not
385 recognized instead of using RSA as a default.
388 *) Add support for ECC-based ciphersuites from draft-ietf-tls-ecc-01.txt.
389 As these are not official, they are not included in "ALL";
390 the "ECCdraft" ciphersuite group alias can be used to select them.
391 [Vipul Gupta and Sumit Gupta (Sun Microsystems Laboratories)]
393 *) Add ECDH engine support.
394 [Nils Gura and Douglas Stebila (Sun Microsystems Laboratories)]
396 *) Add ECDH in new directory crypto/ecdh/.
397 [Douglas Stebila (Sun Microsystems Laboratories)]
399 *) Let BN_rand_range() abort with an error after 100 iterations
400 without success (which indicates a broken PRNG).
403 *) Change BN_mod_sqrt() so that it verifies that the input value
404 is really the square of the return value. (Previously,
405 BN_mod_sqrt would show GIGO behaviour.)
408 *) Add named elliptic curves over binary fields from X9.62, SECG,
409 and WAP/WTLS; add OIDs that were still missing.
411 [Sheueling Chang Shantz and Douglas Stebila
412 (Sun Microsystems Laboratories)]
414 *) Extend the EC library for elliptic curves over binary fields
415 (new files ec2_smpl.c, ec2_smpt.c, ec2_mult.c in crypto/ec/).
418 EC_GF2m_simple_method
422 EC_GROUP_new_curve_GF2m
423 EC_GROUP_set_curve_GF2m
424 EC_GROUP_get_curve_GF2m
425 EC_POINT_set_affine_coordinates_GF2m
426 EC_POINT_get_affine_coordinates_GF2m
427 EC_POINT_set_compressed_coordinates_GF2m
429 Point compression for binary fields is disabled by default for
430 patent reasons (compile with OPENSSL_EC_BIN_PT_COMP defined to
433 As binary polynomials are represented as BIGNUMs, various members
434 of the EC_GROUP and EC_POINT data structures can be shared
435 between the implementations for prime fields and binary fields;
436 the above ..._GF2m functions (except for EX_GROUP_new_curve_GF2m)
437 are essentially identical to their ..._GFp counterparts.
438 (For simplicity, the '..._GFp' prefix has been dropped from
439 various internal method names.)
441 An internal 'field_div' method (similar to 'field_mul' and
442 'field_sqr') has been added; this is used only for binary fields.
444 [Sheueling Chang Shantz and Douglas Stebila
445 (Sun Microsystems Laboratories)]
447 *) Optionally dispatch EC_POINT_mul(), EC_POINT_precompute_mult()
448 through methods ('mul', 'precompute_mult').
450 The generic implementations (now internally called 'ec_wNAF_mul'
451 and 'ec_wNAF_precomputed_mult') remain the default if these
452 methods are undefined.
454 [Sheueling Chang Shantz and Douglas Stebila
455 (Sun Microsystems Laboratories)]
457 *) New function EC_GROUP_get_degree, which is defined through
458 EC_METHOD. For curves over prime fields, this returns the bit
459 length of the modulus.
461 [Sheueling Chang Shantz and Douglas Stebila
462 (Sun Microsystems Laboratories)]
464 *) New functions EC_GROUP_dup, EC_POINT_dup.
465 (These simply call ..._new and ..._copy).
467 [Sheueling Chang Shantz and Douglas Stebila
468 (Sun Microsystems Laboratories)]
470 *) Add binary polynomial arithmetic software in crypto/bn/bn_gf2m.c.
471 Polynomials are represented as BIGNUMs (where the sign bit is not
472 used) in the following functions [macros]:
475 BN_GF2m_sub [= BN_GF2m_add]
476 BN_GF2m_mod [wrapper for BN_GF2m_mod_arr]
477 BN_GF2m_mod_mul [wrapper for BN_GF2m_mod_mul_arr]
478 BN_GF2m_mod_sqr [wrapper for BN_GF2m_mod_sqr_arr]
480 BN_GF2m_mod_exp [wrapper for BN_GF2m_mod_exp_arr]
481 BN_GF2m_mod_sqrt [wrapper for BN_GF2m_mod_sqrt_arr]
482 BN_GF2m_mod_solve_quad [wrapper for BN_GF2m_mod_solve_quad_arr]
483 BN_GF2m_cmp [= BN_ucmp]
485 (Note that only the 'mod' functions are actually for fields GF(2^m).
486 BN_GF2m_add() is misnomer, but this is for the sake of consistency.)
488 For some functions, an the irreducible polynomial defining a
489 field can be given as an 'unsigned int[]' with strictly
490 decreasing elements giving the indices of those bits that are set;
491 i.e., p[] represents the polynomial
492 f(t) = t^p[0] + t^p[1] + ... + t^p[k]
494 p[0] > p[1] > ... > p[k] = 0.
495 This applies to the following functions:
500 BN_GF2m_mod_inv_arr [wrapper for BN_GF2m_mod_inv]
501 BN_GF2m_mod_div_arr [wrapper for BN_GF2m_mod_div]
504 BN_GF2m_mod_solve_quad_arr
508 Conversion can be performed by the following functions:
513 bntest.c has additional tests for binary polynomial arithmetic.
515 Two implementations for BN_GF2m_mod_div() are available.
516 The default algorithm simply uses BN_GF2m_mod_inv() and
517 BN_GF2m_mod_mul(). The alternative algorithm is compiled in only
518 if OPENSSL_SUN_GF2M_DIV is defined (patent pending; read the
519 copyright notice in crypto/bn/bn_gf2m.c before enabling it).
521 [Sheueling Chang Shantz and Douglas Stebila
522 (Sun Microsystems Laboratories)]
524 *) Add new error code 'ERR_R_DISABLED' that can be used when some
525 functionality is disabled at compile-time.
526 [Douglas Stebila <douglas.stebila@sun.com>]
528 *) Change default behaviour of 'openssl asn1parse' so that more
529 information is visible when viewing, e.g., a certificate:
531 Modify asn1_parse2 (crypto/asn1/asn1_par.c) so that in non-'dump'
532 mode the content of non-printable OCTET STRINGs is output in a
533 style similar to INTEGERs, but with '[HEX DUMP]' prepended to
534 avoid the appearance of a printable string.
535 [Nils Larsch <nla@trustcenter.de>]
537 *) Add 'asn1_flag' and 'asn1_form' member to EC_GROUP with access
539 EC_GROUP_set_asn1_flag()
540 EC_GROUP_get_asn1_flag()
541 EC_GROUP_set_point_conversion_form()
542 EC_GROUP_get_point_conversion_form()
543 These control ASN1 encoding details:
544 - Curves (i.e., groups) are encoded explicitly unless asn1_flag
545 has been set to OPENSSL_EC_NAMED_CURVE.
546 - Points are encoded in uncompressed form by default; options for
547 asn1_for are as for point2oct, namely
548 POINT_CONVERSION_COMPRESSED
549 POINT_CONVERSION_UNCOMPRESSED
550 POINT_CONVERSION_HYBRID
552 Also add 'seed' and 'seed_len' members to EC_GROUP with access
556 EC_GROUP_get_seed_len()
557 This is used only for ASN1 purposes (so far).
558 [Nils Larsch <nla@trustcenter.de>]
560 *) Add 'field_type' member to EC_METHOD, which holds the NID
561 of the appropriate field type OID. The new function
562 EC_METHOD_get_field_type() returns this value.
563 [Nils Larsch <nla@trustcenter.de>]
570 providing useful interfaces to EC_POINT_point2oct() and
571 EC_POINT_oct2point().
572 [Nils Larsch <nla@trustcenter.de>]
574 *) Change internals of the EC library so that the functions
575 EC_GROUP_set_generator()
576 EC_GROUP_get_generator()
578 EC_GROUP_get_cofactor()
579 are implemented directly in crypto/ec/ec_lib.c and not dispatched
580 to methods, which would lead to unnecessary code duplication when
581 adding different types of curves.
582 [Nils Larsch <nla@trustcenter.de> with input by Bodo Moeller]
584 *) Implement compute_wNAF (crypto/ec/ec_mult.c) without BIGNUM
585 arithmetic, and such that modified wNAFs are generated
586 (which avoid length expansion in many cases).
589 *) Add a function EC_GROUP_check_discriminant() (defined via
590 EC_METHOD) that verifies that the curve discriminant is non-zero.
592 Add a function EC_GROUP_check() that makes some sanity tests
593 on a EC_GROUP, its generator and order. This includes
594 EC_GROUP_check_discriminant().
595 [Nils Larsch <nla@trustcenter.de>]
597 *) Add ECDSA in new directory crypto/ecdsa/.
599 Add applications 'openssl ecparam' and 'openssl ecdsa'
600 (these are based on 'openssl dsaparam' and 'openssl dsa').
602 ECDSA support is also included in various other files across the
603 library. Most notably,
604 - 'openssl req' now has a '-newkey ecdsa:file' option;
605 - EVP_PKCS82PKEY (crypto/evp/evp_pkey.c) now can handle ECDSA;
606 - X509_PUBKEY_get (crypto/asn1/x_pubkey.c) and
607 d2i_PublicKey (crypto/asn1/d2i_pu.c) have been modified to make
608 them suitable for ECDSA where domain parameters must be
609 extracted before the specific public key;
610 - ECDSA engine support has been added.
611 [Nils Larsch <nla@trustcenter.de>]
613 *) Include some named elliptic curves, and add OIDs from X9.62,
614 SECG, and WAP/WTLS. Each curve can be obtained from the new
616 EC_GROUP_new_by_nid(),
617 and the list of available named curves can be obtained with
618 EC_get_builtin_curves().
619 Also add a 'curve_name' member to EC_GROUP objects, which can be
623 [Nils Larsch <nla@trustcenter.de, Bodo Moeller]
625 *) Remove a few calls to bn_wexpand() in BN_sqr() (the one in there
626 was actually never needed) and in BN_mul(). The removal in BN_mul()
627 required a small change in bn_mul_part_recursive() and the addition
628 of the functions bn_cmp_part_words(), bn_sub_part_words() and
629 bn_add_part_words(), which do the same thing as bn_cmp_words(),
630 bn_sub_words() and bn_add_words() except they take arrays with
634 Changes between 0.9.7c and 0.9.7d [xx XXX XXXX]
636 *) X509 verify fixes. Disable broken certificate workarounds when
637 X509_V_FLAGS_X509_STRICT is set. Check CRL issuer has cRLSign set if
638 keyUsage extension present. Don't accept CRLs with unhandled critical
639 extensions: since verify currently doesn't process CRL extensions this
640 rejects a CRL with *any* critical extensions. Add new verify error codes
644 *) When creating an OCSP nonce use an OCTET STRING inside the extnValue.
645 A clarification of RFC2560 will require the use of OCTET STRINGs and
646 some implementations cannot handle the current raw format. Since OpenSSL
647 copies and compares OCSP nonces as opaque blobs without any attempt at
648 parsing them this should not create any compatibility issues.
651 *) New md flag EVP_MD_CTX_FLAG_REUSE this allows md_data to be reused when
652 calling EVP_MD_CTX_copy_ex() to avoid calling OPENSSL_malloc(). Without
653 this HMAC (and other) operations are several times slower than OpenSSL
657 *) Print out GeneralizedTime and UTCTime in ASN1_STRING_print_ex().
658 [Peter Sylvester <Peter.Sylvester@EdelWeb.fr>]
660 *) Use the correct content when signing type "other".
663 Changes between 0.9.7b and 0.9.7c [30 Sep 2003]
665 *) Fix various bugs revealed by running the NISCC test suite:
667 Stop out of bounds reads in the ASN1 code when presented with
668 invalid tags (CAN-2003-0543 and CAN-2003-0544).
670 Free up ASN1_TYPE correctly if ANY type is invalid (CAN-2003-0545).
672 If verify callback ignores invalid public key errors don't try to check
673 certificate signature with the NULL public key.
677 *) New -ignore_err option in ocsp application to stop the server
678 exiting on the first error in a request.
681 *) In ssl3_accept() (ssl/s3_srvr.c) only accept a client certificate
682 if the server requested one: as stated in TLS 1.0 and SSL 3.0
686 *) In ssl3_get_client_hello() (ssl/s3_srvr.c), tolerate additional
687 extra data after the compression methods not only for TLS 1.0
688 but also for SSL 3.0 (as required by the specification).
689 [Bodo Moeller; problem pointed out by Matthias Loepfe]
691 *) Change X509_certificate_type() to mark the key as exported/exportable
692 when it's 512 *bits* long, not 512 bytes.
695 *) Change AES_cbc_encrypt() so it outputs exact multiple of
696 blocks during encryption.
699 *) Various fixes to base64 BIO and non blocking I/O. On write
700 flushes were not handled properly if the BIO retried. On read
701 data was not being buffered properly and had various logic bugs.
702 This also affects blocking I/O when the data being decoded is a
706 *) Various S/MIME bugfixes and compatibility changes:
707 output correct application/pkcs7 MIME type if
708 PKCS7_NOOLDMIMETYPE is set. Tolerate some broken signatures.
709 Output CR+LF for EOL if PKCS7_CRLFEOL is set (this makes opening
710 of files as .eml work). Correctly handle very long lines in MIME
714 Changes between 0.9.7a and 0.9.7b [10 Apr 2003]
716 *) Countermeasure against the Klima-Pokorny-Rosa extension of
717 Bleichbacher's attack on PKCS #1 v1.5 padding: treat
718 a protocol version number mismatch like a decryption error
719 in ssl3_get_client_key_exchange (ssl/s3_srvr.c).
722 *) Turn on RSA blinding by default in the default implementation
723 to avoid a timing attack. Applications that don't want it can call
724 RSA_blinding_off() or use the new flag RSA_FLAG_NO_BLINDING.
725 They would be ill-advised to do so in most cases.
726 [Ben Laurie, Steve Henson, Geoff Thorpe, Bodo Moeller]
728 *) Change RSA blinding code so that it works when the PRNG is not
729 seeded (in this case, the secret RSA exponent is abused as
730 an unpredictable seed -- if it is not unpredictable, there
731 is no point in blinding anyway). Make RSA blinding thread-safe
732 by remembering the creator's thread ID in rsa->blinding and
733 having all other threads use local one-time blinding factors
734 (this requires more computation than sharing rsa->blinding, but
735 avoids excessive locking; and if an RSA object is not shared
736 between threads, blinding will still be very fast).
739 yet to be integrated into this CVS branch:
740 - Geoff's ENGINE_set_default() fix
742 *) Target "mingw" now allows native Windows code to be generated in
743 the Cygwin environment as well as with the MinGW compiler.
746 Changes between 0.9.7 and 0.9.7a [19 Feb 2003]
748 *) In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked
749 via timing by performing a MAC computation even if incorrrect
750 block cipher padding has been found. This is a countermeasure
751 against active attacks where the attacker has to distinguish
752 between bad padding and a MAC verification error. (CAN-2003-0078)
754 [Bodo Moeller; problem pointed out by Brice Canvel (EPFL),
755 Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and
756 Martin Vuagnoux (EPFL, Ilion)]
758 *) Make the no-err option work as intended. The intention with no-err
759 is not to have the whole error stack handling routines removed from
760 libcrypto, it's only intended to remove all the function name and
761 reason texts, thereby removing some of the footprint that may not
762 be interesting if those errors aren't displayed anyway.
764 NOTE: it's still possible for any application or module to have it's
765 own set of error texts inserted. The routines are there, just not
766 used by default when no-err is given.
769 *) Add support for FreeBSD on IA64.
770 [dirk.meyer@dinoex.sub.org via Richard Levitte, resolves #454]
772 *) Adjust DES_cbc_cksum() so it returns the same value as the MIT
773 Kerberos function mit_des_cbc_cksum(). Before this change,
774 the value returned by DES_cbc_cksum() was like the one from
775 mit_des_cbc_cksum(), except the bytes were swapped.
776 [Kevin Greaney <Kevin.Greaney@hp.com> and Richard Levitte]
778 *) Allow an application to disable the automatic SSL chain building.
779 Before this a rather primitive chain build was always performed in
780 ssl3_output_cert_chain(): an application had no way to send the
781 correct chain if the automatic operation produced an incorrect result.
783 Now the chain builder is disabled if either:
785 1. Extra certificates are added via SSL_CTX_add_extra_chain_cert().
787 2. The mode flag SSL_MODE_NO_AUTO_CHAIN is set.
789 The reasoning behind this is that an application would not want the
790 auto chain building to take place if extra chain certificates are
791 present and it might also want a means of sending no additional
792 certificates (for example the chain has two certificates and the
796 *) Add the possibility to build without the ENGINE framework.
797 [Steven Reddie <smr@essemer.com.au> via Richard Levitte]
799 *) Under Win32 gmtime() can return NULL: check return value in
800 OPENSSL_gmtime(). Add error code for case where gmtime() fails.
803 *) DSA routines: under certain error conditions uninitialized BN objects
804 could be freed. Solution: make sure initialization is performed early
805 enough. (Reported and fix supplied by Ivan D Nestlerode <nestler@MIT.EDU>,
806 Nils Larsch <nla@trustcenter.de> via PR#459)
809 *) Another fix for SSLv2 session ID handling: the session ID was incorrectly
810 checked on reconnect on the client side, therefore session resumption
811 could still fail with a "ssl session id is different" error. This
812 behaviour is masked when SSL_OP_ALL is used due to
813 SSL_OP_MICROSOFT_SESS_ID_BUG being set.
814 Behaviour observed by Crispin Flowerday <crispin@flowerday.cx> as
818 *) IA-32 assembler support enhancements: unified ELF targets, support
819 for SCO/Caldera platforms, fix for Cygwin shared build.
822 *) Add support for FreeBSD on sparc64. As a consequence, support for
823 FreeBSD on non-x86 processors is separate from x86 processors on
824 the config script, much like the NetBSD support.
825 [Richard Levitte & Kris Kennaway <kris@obsecurity.org>]
827 Changes between 0.9.6h and 0.9.7 [31 Dec 2002]
829 [NB: OpenSSL 0.9.6i and later 0.9.6 patch levels were released after
832 *) Fix session ID handling in SSLv2 client code: the SERVER FINISHED
833 code (06) was taken as the first octet of the session ID and the last
834 octet was ignored consequently. As a result SSLv2 client side session
835 caching could not have worked due to the session ID mismatch between
837 Behaviour observed by Crispin Flowerday <crispin@flowerday.cx> as
841 *) Change the declaration of needed Kerberos libraries to use EX_LIBS
842 instead of the special (and badly supported) LIBKRB5. LIBKRB5 is
846 *) The hw_ncipher.c engine requires dynamic locks. Unfortunately, it
847 seems that in spite of existing for more than a year, many application
848 author have done nothing to provide the necessary callbacks, which
849 means that this particular engine will not work properly anywhere.
850 This is a very unfortunate situation which forces us, in the name
851 of usability, to give the hw_ncipher.c a static lock, which is part
853 NOTE: This is for the 0.9.7 series ONLY. This hack will never
854 appear in 0.9.8 or later. We EXPECT application authors to have
855 dealt properly with this when 0.9.8 is released (unless we actually
856 make such changes in the libcrypto locking code that changes will
857 have to be made anyway).
860 *) In asn1_d2i_read_bio() repeatedly call BIO_read() until all content
861 octets have been read, EOF or an error occurs. Without this change
862 some truncated ASN1 structures will not produce an error.
865 *) Disable Heimdal support, since it hasn't been fully implemented.
866 Still give the possibility to force the use of Heimdal, but with
867 warnings and a request that patches get sent to openssl-dev.
870 *) Add the VC-CE target, introduce the WINCE sysname, and add
871 INSTALL.WCE and appropriate conditionals to make it build.
872 [Steven Reddie <smr@essemer.com.au> via Richard Levitte]
874 *) Change the DLL names for Cygwin to cygcrypto-x.y.z.dll and
875 cygssl-x.y.z.dll, where x, y and z are the major, minor and
876 edit numbers of the version.
877 [Corinna Vinschen <vinschen@redhat.com> and Richard Levitte]
879 *) Introduce safe string copy and catenation functions
880 (BUF_strlcpy() and BUF_strlcat()).
881 [Ben Laurie (CHATS) and Richard Levitte]
883 *) Avoid using fixed-size buffers for one-line DNs.
886 *) Add BUF_MEM_grow_clean() to avoid information leakage when
887 resizing buffers containing secrets, and use where appropriate.
890 *) Avoid using fixed size buffers for configuration file location.
893 *) Avoid filename truncation for various CA files.
896 *) Use sizeof in preference to magic numbers.
899 *) Avoid filename truncation in cert requests.
902 *) Add assertions to check for (supposedly impossible) buffer
906 *) Don't cache truncated DNS entries in the local cache (this could
907 potentially lead to a spoofing attack).
910 *) Fix various buffers to be large enough for hex/decimal
911 representations in a platform independent manner.
914 *) Add CRYPTO_realloc_clean() to avoid information leakage when
915 resizing buffers containing secrets, and use where appropriate.
918 *) Add BIO_indent() to avoid much slightly worrying code to do
922 *) Convert sprintf()/BIO_puts() to BIO_printf().
925 *) buffer_gets() could terminate with the buffer only half
929 *) Add assertions to prevent user-supplied crypto functions from
930 overflowing internal buffers by having large block sizes, etc.
933 *) New OPENSSL_assert() macro (similar to assert(), but enabled
937 *) Eliminate unused copy of key in RC4.
940 *) Eliminate unused and incorrectly sized buffers for IV in pem.h.
943 *) Fix off-by-one error in EGD path.
946 *) If RANDFILE path is too long, ignore instead of truncating.
949 *) Eliminate unused and incorrectly sized X.509 structure
953 *) Eliminate unused and dangerous function knumber().
956 *) Eliminate unused and dangerous structure, KSSL_ERR.
959 *) Protect against overlong session ID context length in an encoded
960 session object. Since these are local, this does not appear to be
964 *) Change from security patch (see 0.9.6e below) that did not affect
965 the 0.9.6 release series:
967 Remote buffer overflow in SSL3 protocol - an attacker could
968 supply an oversized master key in Kerberos-enabled versions.
972 *) Change the SSL kerb5 codes to match RFC 2712.
975 *) Make -nameopt work fully for req and add -reqopt switch.
976 [Michael Bell <michael.bell@rz.hu-berlin.de>, Steve Henson]
978 *) The "block size" for block ciphers in CFB and OFB mode should be 1.
979 [Steve Henson, reported by Yngve Nysaeter Pettersen <yngve@opera.com>]
981 *) Make sure tests can be performed even if the corresponding algorithms
982 have been removed entirely. This was also the last step to make
983 OpenSSL compilable with DJGPP under all reasonable conditions.
984 [Richard Levitte, Doug Kaufman <dkaufman@rahul.net>]
986 *) Add cipher selection rules COMPLEMENTOFALL and COMPLEMENTOFDEFAULT
987 to allow version independent disabling of normally unselected ciphers,
988 which may be activated as a side-effect of selecting a single cipher.
990 (E.g., cipher list string "RSA" enables ciphersuites that are left
991 out of "ALL" because they do not provide symmetric encryption.
992 "RSA:!COMPLEMEMENTOFALL" avoids these unsafe ciphersuites.)
993 [Lutz Jaenicke, Bodo Moeller]
995 *) Add appropriate support for separate platform-dependent build
996 directories. The recommended way to make a platform-dependent
997 build directory is the following (tested on Linux), maybe with
1000 # Place yourself outside of the OpenSSL source tree. In
1001 # this example, the environment variable OPENSSL_SOURCE
1002 # is assumed to contain the absolute OpenSSL source directory.
1003 mkdir -p objtree/"`uname -s`-`uname -r`-`uname -m`"
1004 cd objtree/"`uname -s`-`uname -r`-`uname -m`"
1005 (cd $OPENSSL_SOURCE; find . -type f) | while read F; do
1006 mkdir -p `dirname $F`
1007 ln -s $OPENSSL_SOURCE/$F $F
1010 To be absolutely sure not to disturb the source tree, a "make clean"
1011 is a good thing. If it isn't successfull, don't worry about it,
1012 it probably means the source directory is very clean.
1015 *) Make sure any ENGINE control commands make local copies of string
1016 pointers passed to them whenever necessary. Otherwise it is possible
1017 the caller may have overwritten (or deallocated) the original string
1018 data when a later ENGINE operation tries to use the stored values.
1019 [Götz Babin-Ebell <babinebell@trustcenter.de>]
1021 *) Improve diagnostics in file reading and command-line digests.
1022 [Ben Laurie aided and abetted by Solar Designer <solar@openwall.com>]
1024 *) Add AES modes CFB and OFB to the object database. Correct an
1025 error in AES-CFB decryption.
1028 *) Remove most calls to EVP_CIPHER_CTX_cleanup() in evp_enc.c, this
1029 allows existing EVP_CIPHER_CTX structures to be reused after
1030 calling EVP_*Final(). This behaviour is used by encryption
1031 BIOs and some applications. This has the side effect that
1032 applications must explicitly clean up cipher contexts with
1033 EVP_CIPHER_CTX_cleanup() or they will leak memory.
1036 *) Check the values of dna and dnb in bn_mul_recursive before calling
1037 bn_mul_comba (a non zero value means the a or b arrays do not contain
1038 n2 elements) and fallback to bn_mul_normal if either is not zero.
1041 *) Fix escaping of non-ASCII characters when using the -subj option
1042 of the "openssl req" command line tool. (Robert Joop <joop@fokus.gmd.de>)
1045 *) Make object definitions compliant to LDAP (RFC2256): SN is the short
1046 form for "surname", serialNumber has no short form.
1047 Use "mail" as the short name for "rfc822Mailbox" according to RFC2798;
1048 therefore remove "mail" short name for "internet 7".
1049 The OID for unique identifiers in X509 certificates is
1050 x500UniqueIdentifier, not uniqueIdentifier.
1051 Some more OID additions. (Michael Bell <michael.bell@rz.hu-berlin.de>)
1054 *) Add an "init" command to the ENGINE config module and auto initialize
1055 ENGINEs. Without any "init" command the ENGINE will be initialized
1056 after all ctrl commands have been executed on it. If init=1 the
1057 ENGINE is initailized at that point (ctrls before that point are run
1058 on the uninitialized ENGINE and after on the initialized one). If
1059 init=0 then the ENGINE will not be iniatialized at all.
1062 *) Fix the 'app_verify_callback' interface so that the user-defined
1063 argument is actually passed to the callback: In the
1064 SSL_CTX_set_cert_verify_callback() prototype, the callback
1065 declaration has been changed from
1068 int (*cb)(X509_STORE_CTX *,void *);
1069 in ssl_verify_cert_chain (ssl/ssl_cert.c), the call
1070 i=s->ctx->app_verify_callback(&ctx)
1071 has been changed into
1072 i=s->ctx->app_verify_callback(&ctx, s->ctx->app_verify_arg).
1074 To update applications using SSL_CTX_set_cert_verify_callback(),
1075 a dummy argument can be added to their callback functions.
1076 [D. K. Smetters <smetters@parc.xerox.com>]
1078 *) Added the '4758cca' ENGINE to support IBM 4758 cards.
1079 [Maurice Gittens <maurice@gittens.nl>, touchups by Geoff Thorpe]
1081 *) Add and OPENSSL_LOAD_CONF define which will cause
1082 OpenSSL_add_all_algorithms() to load the openssl.cnf config file.
1083 This allows older applications to transparently support certain
1084 OpenSSL features: such as crypto acceleration and dynamic ENGINE loading.
1085 Two new functions OPENSSL_add_all_algorithms_noconf() which will never
1086 load the config file and OPENSSL_add_all_algorithms_conf() which will
1087 always load it have also been added.
1090 *) Add the OFB, CFB and CTR (all with 128 bit feedback) to AES.
1091 Adjust NIDs and EVP layer.
1092 [Stephen Sprunk <stephen@sprunk.org> and Richard Levitte]
1094 *) Config modules support in openssl utility.
1096 Most commands now load modules from the config file,
1097 though in a few (such as version) this isn't done
1098 because it couldn't be used for anything.
1100 In the case of ca and req the config file used is
1101 the same as the utility itself: that is the -config
1102 command line option can be used to specify an
1106 *) Move default behaviour from OPENSSL_config(). If appname is NULL
1107 use "openssl_conf" if filename is NULL use default openssl config file.
1110 *) Add an argument to OPENSSL_config() to allow the use of an alternative
1111 config section name. Add a new flag to tolerate a missing config file
1112 and move code to CONF_modules_load_file().
1115 *) Support for crypto accelerator cards from Accelerated Encryption
1116 Processing, www.aep.ie. (Use engine 'aep')
1117 The support was copied from 0.9.6c [engine] and adapted/corrected
1118 to work with the new engine framework.
1119 [AEP Inc. and Richard Levitte]
1121 *) Support for SureWare crypto accelerator cards from Baltimore
1122 Technologies. (Use engine 'sureware')
1123 The support was copied from 0.9.6c [engine] and adapted
1124 to work with the new engine framework.
1127 *) Have the CHIL engine fork-safe (as defined by nCipher) and actually
1128 make the newer ENGINE framework commands for the CHIL engine work.
1129 [Toomas Kiisk <vix@cyber.ee> and Richard Levitte]
1131 *) Make it possible to produce shared libraries on ReliantUNIX.
1132 [Robert Dahlem <Robert.Dahlem@ffm2.siemens.de> via Richard Levitte]
1134 *) Add the configuration target debug-linux-ppro.
1135 Make 'openssl rsa' use the general key loading routines
1136 implemented in apps.c, and make those routines able to
1137 handle the key format FORMAT_NETSCAPE and the variant
1139 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte]
1141 *) Fix a crashbug and a logic bug in hwcrhk_load_pubkey().
1142 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte]
1144 *) Add -keyform to rsautl, and document -engine.
1145 [Richard Levitte, inspired by Toomas Kiisk <vix@cyber.ee>]
1147 *) Change BIO_new_file (crypto/bio/bss_file.c) to use new
1148 BIO_R_NO_SUCH_FILE error code rather than the generic
1149 ERR_R_SYS_LIB error code if fopen() fails with ENOENT.
1152 *) Add new functions
1154 ERR_peek_last_error_line
1155 ERR_peek_last_error_line_data.
1156 These are similar to
1159 ERR_peek_error_line_data,
1160 but report on the latest error recorded rather than the first one
1161 still in the error queue.
1162 [Ben Laurie, Bodo Moeller]
1164 *) default_algorithms option in ENGINE config module. This allows things
1166 default_algorithms = ALL
1167 default_algorithms = RSA, DSA, RAND, CIPHERS, DIGESTS
1170 *) Prelminary ENGINE config module.
1173 *) New experimental application configuration code.
1176 *) Change the AES code to follow the same name structure as all other
1177 symmetric ciphers, and behave the same way. Move everything to
1178 the directory crypto/aes, thereby obsoleting crypto/rijndael.
1179 [Stephen Sprunk <stephen@sprunk.org> and Richard Levitte]
1181 *) SECURITY: remove unsafe setjmp/signal interaction from ui_openssl.c.
1182 [Ben Laurie and Theo de Raadt]
1184 *) Add option to output public keys in req command.
1185 [Massimiliano Pala madwolf@openca.org]
1187 *) Use wNAFs in EC_POINTs_mul() for improved efficiency
1188 (up to about 10% better than before for P-192 and P-224).
1191 *) New functions/macros
1193 SSL_CTX_set_msg_callback(ctx, cb)
1194 SSL_CTX_set_msg_callback_arg(ctx, arg)
1195 SSL_set_msg_callback(ssl, cb)
1196 SSL_set_msg_callback_arg(ssl, arg)
1198 to request calling a callback function
1200 void cb(int write_p, int version, int content_type,
1201 const void *buf, size_t len, SSL *ssl, void *arg)
1203 whenever a protocol message has been completely received
1204 (write_p == 0) or sent (write_p == 1). Here 'version' is the
1205 protocol version according to which the SSL library interprets
1206 the current protocol message (SSL2_VERSION, SSL3_VERSION, or
1207 TLS1_VERSION). 'content_type' is 0 in the case of SSL 2.0, or
1208 the content type as defined in the SSL 3.0/TLS 1.0 protocol
1209 specification (change_cipher_spec(20), alert(21), handshake(22)).
1210 'buf' and 'len' point to the actual message, 'ssl' to the
1211 SSL object, and 'arg' is the application-defined value set by
1212 SSL[_CTX]_set_msg_callback_arg().
1214 'openssl s_client' and 'openssl s_server' have new '-msg' options
1215 to enable a callback that displays all protocol messages.
1218 *) Change the shared library support so shared libraries are built as
1219 soon as the corresponding static library is finished, and thereby get
1220 openssl and the test programs linked against the shared library.
1221 This still only happens when the keyword "shard" has been given to
1222 the configuration scripts.
1224 NOTE: shared library support is still an experimental thing, and
1225 backward binary compatibility is still not guaranteed.
1226 ["Maciej W. Rozycki" <macro@ds2.pg.gda.pl> and Richard Levitte]
1228 *) Add support for Subject Information Access extension.
1229 [Peter Sylvester <Peter.Sylvester@EdelWeb.fr>]
1231 *) Make BUF_MEM_grow() behaviour more consistent: Initialise to zero
1232 additional bytes when new memory had to be allocated, not just
1233 when reusing an existing buffer.
1236 *) New command line and configuration option 'utf8' for the req command.
1237 This allows field values to be specified as UTF8 strings.
1240 *) Add -multi and -mr options to "openssl speed" - giving multiple parallel
1241 runs for the former and machine-readable output for the latter.
1244 *) Add '-noemailDN' option to 'openssl ca'. This prevents inclusion
1245 of the e-mail address in the DN (i.e., it will go into a certificate
1246 extension only). The new configuration file option 'email_in_dn = no'
1247 has the same effect.
1248 [Massimiliano Pala madwolf@openca.org]
1250 *) Change all functions with names starting with des_ to be starting
1251 with DES_ instead. Add wrappers that are compatible with libdes,
1252 but are named _ossl_old_des_*. Finally, add macros that map the
1253 des_* symbols to the corresponding _ossl_old_des_* if libdes
1254 compatibility is desired. If OpenSSL 0.9.6c compatibility is
1255 desired, the des_* symbols will be mapped to DES_*, with one
1258 Since we provide two compatibility mappings, the user needs to
1259 define the macro OPENSSL_DES_LIBDES_COMPATIBILITY if libdes
1260 compatibility is desired. The default (i.e., when that macro
1261 isn't defined) is OpenSSL 0.9.6c compatibility.
1263 There are also macros that enable and disable the support of old
1264 des functions altogether. Those are OPENSSL_ENABLE_OLD_DES_SUPPORT
1265 and OPENSSL_DISABLE_OLD_DES_SUPPORT. If none or both of those
1266 are defined, the default will apply: to support the old des routines.
1268 In either case, one must include openssl/des.h to get the correct
1269 definitions. Do not try to just include openssl/des_old.h, that
1272 NOTE: This is a major break of an old API into a new one. Software
1273 authors are encouraged to switch to the DES_ style functions. Some
1274 time in the future, des_old.h and the libdes compatibility functions
1275 will be disable (i.e. OPENSSL_DISABLE_OLD_DES_SUPPORT will be the
1276 default), and then completely removed.
1279 *) Test for certificates which contain unsupported critical extensions.
1280 If such a certificate is found during a verify operation it is
1281 rejected by default: this behaviour can be overridden by either
1282 handling the new error X509_V_ERR_UNHANDLED_CRITICAL_EXTENSION or
1283 by setting the verify flag X509_V_FLAG_IGNORE_CRITICAL. A new function
1284 X509_supported_extension() has also been added which returns 1 if a
1285 particular extension is supported.
1288 *) Modify the behaviour of EVP cipher functions in similar way to digests
1289 to retain compatibility with existing code.
1292 *) Modify the behaviour of EVP_DigestInit() and EVP_DigestFinal() to retain
1293 compatibility with existing code. In particular the 'ctx' parameter does
1294 not have to be to be initialized before the call to EVP_DigestInit() and
1295 it is tidied up after a call to EVP_DigestFinal(). New function
1296 EVP_DigestFinal_ex() which does not tidy up the ctx. Similarly function
1297 EVP_MD_CTX_copy() changed to not require the destination to be
1298 initialized valid and new function EVP_MD_CTX_copy_ex() added which
1299 requires the destination to be valid.
1301 Modify all the OpenSSL digest calls to use EVP_DigestInit_ex(),
1302 EVP_DigestFinal_ex() and EVP_MD_CTX_copy_ex().
1305 *) Change ssl3_get_message (ssl/s3_both.c) and the functions using it
1306 so that complete 'Handshake' protocol structures are kept in memory
1307 instead of overwriting 'msg_type' and 'length' with 'body' data.
1310 *) Add an implementation of SSL_add_dir_cert_subjects_to_stack for Win32.
1311 [Massimo Santin via Richard Levitte]
1313 *) Major restructuring to the underlying ENGINE code. This includes
1314 reduction of linker bloat, separation of pure "ENGINE" manipulation
1315 (initialisation, etc) from functionality dealing with implementations
1316 of specific crypto iterfaces. This change also introduces integrated
1317 support for symmetric ciphers and digest implementations - so ENGINEs
1318 can now accelerate these by providing EVP_CIPHER and EVP_MD
1319 implementations of their own. This is detailed in crypto/engine/README
1320 as it couldn't be adequately described here. However, there are a few
1321 API changes worth noting - some RSA, DSA, DH, and RAND functions that
1322 were changed in the original introduction of ENGINE code have now
1323 reverted back - the hooking from this code to ENGINE is now a good
1324 deal more passive and at run-time, operations deal directly with
1325 RSA_METHODs, DSA_METHODs (etc) as they did before, rather than
1326 dereferencing through an ENGINE pointer any more. Also, the ENGINE
1327 functions dealing with BN_MOD_EXP[_CRT] handlers have been removed -
1328 they were not being used by the framework as there is no concept of a
1329 BIGNUM_METHOD and they could not be generalised to the new
1330 'ENGINE_TABLE' mechanism that underlies the new code. Similarly,
1331 ENGINE_cpy() has been removed as it cannot be consistently defined in
1335 *) Change ASN1_GENERALIZEDTIME_check() to allow fractional seconds.
1338 *) Change mkdef.pl to sort symbols that get the same entry number,
1339 and make sure the automatically generated functions ERR_load_*
1340 become part of libeay.num as well.
1343 *) New function SSL_renegotiate_pending(). This returns true once
1344 renegotiation has been requested (either SSL_renegotiate() call
1345 or HelloRequest/ClientHello receveived from the peer) and becomes
1346 false once a handshake has been completed.
1347 (For servers, SSL_renegotiate() followed by SSL_do_handshake()
1348 sends a HelloRequest, but does not ensure that a handshake takes
1349 place. SSL_renegotiate_pending() is useful for checking if the
1350 client has followed the request.)
1353 *) New SSL option SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION.
1354 By default, clients may request session resumption even during
1355 renegotiation (if session ID contexts permit); with this option,
1356 session resumption is possible only in the first handshake.
1358 SSL_OP_ALL is now 0x00000FFFL instead of 0x000FFFFFL. This makes
1359 more bits available for options that should not be part of
1360 SSL_OP_ALL (such as SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION).
1363 *) Add some demos for certificate and certificate request creation.
1366 *) Make maximum certificate chain size accepted from the peer application
1367 settable (SSL*_get/set_max_cert_list()), as proposed by
1368 "Douglas E. Engert" <deengert@anl.gov>.
1371 *) Add support for shared libraries for Unixware-7
1372 (Boyd Lynn Gerber <gerberb@zenez.com>).
1375 *) Add a "destroy" handler to ENGINEs that allows structural cleanup to
1376 be done prior to destruction. Use this to unload error strings from
1377 ENGINEs that load their own error strings. NB: This adds two new API
1378 functions to "get" and "set" this destroy handler in an ENGINE.
1381 *) Alter all existing ENGINE implementations (except "openssl" and
1382 "openbsd") to dynamically instantiate their own error strings. This
1383 makes them more flexible to be built both as statically-linked ENGINEs
1384 and self-contained shared-libraries loadable via the "dynamic" ENGINE.
1385 Also, add stub code to each that makes building them as self-contained
1386 shared-libraries easier (see README.ENGINE).
1389 *) Add a "dynamic" ENGINE that provides a mechanism for binding ENGINE
1390 implementations into applications that are completely implemented in
1391 self-contained shared-libraries. The "dynamic" ENGINE exposes control
1392 commands that can be used to configure what shared-library to load and
1393 to control aspects of the way it is handled. Also, made an update to
1394 the README.ENGINE file that brings its information up-to-date and
1395 provides some information and instructions on the "dynamic" ENGINE
1396 (ie. how to use it, how to build "dynamic"-loadable ENGINEs, etc).
1399 *) Make it possible to unload ranges of ERR strings with a new
1400 "ERR_unload_strings" function.
1403 *) Add a copy() function to EVP_MD.
1406 *) Make EVP_MD routines take a context pointer instead of just the
1407 md_data void pointer.
1410 *) Add flags to EVP_MD and EVP_MD_CTX. EVP_MD_FLAG_ONESHOT indicates
1411 that the digest can only process a single chunk of data
1412 (typically because it is provided by a piece of
1413 hardware). EVP_MD_CTX_FLAG_ONESHOT indicates that the application
1414 is only going to provide a single chunk of data, and hence the
1415 framework needn't accumulate the data for oneshot drivers.
1418 *) As with "ERR", make it possible to replace the underlying "ex_data"
1419 functions. This change also alters the storage and management of global
1420 ex_data state - it's now all inside ex_data.c and all "class" code (eg.
1421 RSA, BIO, SSL_CTX, etc) no longer stores its own STACKS and per-class
1422 index counters. The API functions that use this state have been changed
1423 to take a "class_index" rather than pointers to the class's local STACK
1424 and counter, and there is now an API function to dynamically create new
1425 classes. This centralisation allows us to (a) plug a lot of the
1426 thread-safety problems that existed, and (b) makes it possible to clean
1427 up all allocated state using "CRYPTO_cleanup_all_ex_data()". W.r.t. (b)
1428 such data would previously have always leaked in application code and
1429 workarounds were in place to make the memory debugging turn a blind eye
1430 to it. Application code that doesn't use this new function will still
1431 leak as before, but their memory debugging output will announce it now
1432 rather than letting it slide.
1434 Besides the addition of CRYPTO_cleanup_all_ex_data(), another API change
1435 induced by the "ex_data" overhaul is that X509_STORE_CTX_init() now
1436 has a return value to indicate success or failure.
1439 *) Make it possible to replace the underlying "ERR" functions such that the
1440 global state (2 LHASH tables and 2 locks) is only used by the "default"
1441 implementation. This change also adds two functions to "get" and "set"
1442 the implementation prior to it being automatically set the first time
1443 any other ERR function takes place. Ie. an application can call "get",
1444 pass the return value to a module it has just loaded, and that module
1445 can call its own "set" function using that value. This means the
1446 module's "ERR" operations will use (and modify) the error state in the
1447 application and not in its own statically linked copy of OpenSSL code.
1450 *) Give DH, DSA, and RSA types their own "**_up_ref()" function to increment
1451 reference counts. This performs normal REF_PRINT/REF_CHECK macros on
1452 the operation, and provides a more encapsulated way for external code
1453 (crypto/evp/ and ssl/) to do this. Also changed the evp and ssl code
1454 to use these functions rather than manually incrementing the counts.
1456 Also rename "DSO_up()" function to more descriptive "DSO_up_ref()".
1459 *) Add EVP test program.
1462 *) Add symmetric cipher support to ENGINE. Expect the API to change!
1465 *) New CRL functions: X509_CRL_set_version(), X509_CRL_set_issuer_name()
1466 X509_CRL_set_lastUpdate(), X509_CRL_set_nextUpdate(), X509_CRL_sort(),
1467 X509_REVOKED_set_serialNumber(), and X509_REVOKED_set_revocationDate().
1468 These allow a CRL to be built without having to access X509_CRL fields
1469 directly. Modify 'ca' application to use new functions.
1472 *) Move SSL_OP_TLS_ROLLBACK_BUG out of the SSL_OP_ALL list of recommended
1473 bug workarounds. Rollback attack detection is a security feature.
1474 The problem will only arise on OpenSSL servers when TLSv1 is not
1475 available (sslv3_server_method() or SSL_OP_NO_TLSv1).
1476 Software authors not wanting to support TLSv1 will have special reasons
1477 for their choice and can explicitly enable this option.
1478 [Bodo Moeller, Lutz Jaenicke]
1480 *) Rationalise EVP so it can be extended: don't include a union of
1481 cipher/digest structures, add init/cleanup functions for EVP_MD_CTX
1482 (similar to those existing for EVP_CIPHER_CTX).
1487 EVP_MD_CTX_init(&md); /* new function call */
1488 EVP_DigestInit(&md, EVP_sha1());
1489 EVP_DigestUpdate(&md, in, len);
1490 EVP_DigestFinal(&md, out, NULL);
1491 EVP_MD_CTX_cleanup(&md); /* new function call */
1495 *) Make DES key schedule conform to the usual scheme, as well as
1496 correcting its structure. This means that calls to DES functions
1497 now have to pass a pointer to a des_key_schedule instead of a
1498 plain des_key_schedule (which was actually always a pointer
1501 des_key_schedule ks;
1503 des_set_key_checked(..., &ks);
1504 des_ncbc_encrypt(..., &ks, ...);
1506 (Note that a later change renames 'des_...' into 'DES_...'.)
1509 *) Initial reduction of linker bloat: the use of some functions, such as
1510 PEM causes large amounts of unused functions to be linked in due to
1511 poor organisation. For example pem_all.c contains every PEM function
1512 which has a knock on effect of linking in large amounts of (unused)
1513 ASN1 code. Grouping together similar functions and splitting unrelated
1514 functions prevents this.
1517 *) Cleanup of EVP macros.
1520 *) Change historical references to {NID,SN,LN}_des_ede and ede3 to add the
1521 correct _ecb suffix.
1524 *) Add initial OCSP responder support to ocsp application. The
1525 revocation information is handled using the text based index
1526 use by the ca application. The responder can either handle
1527 requests generated internally, supplied in files (for example
1528 via a CGI script) or using an internal minimal server.
1531 *) Add configuration choices to get zlib compression for TLS.
1534 *) Changes to Kerberos SSL for RFC 2712 compliance:
1535 1. Implemented real KerberosWrapper, instead of just using
1536 KRB5 AP_REQ message. [Thanks to Simon Wilkinson <sxw@sxw.org.uk>]
1537 2. Implemented optional authenticator field of KerberosWrapper.
1539 Added openssl-style ASN.1 macros for Kerberos ticket, ap_req,
1540 and authenticator structs; see crypto/krb5/.
1542 Generalized Kerberos calls to support multiple Kerberos libraries.
1543 [Vern Staats <staatsvr@asc.hpc.mil>,
1544 Jeffrey Altman <jaltman@columbia.edu>
1545 via Richard Levitte]
1547 *) Cause 'openssl speed' to use fully hard-coded DSA keys as it
1548 already does with RSA. testdsa.h now has 'priv_key/pub_key'
1549 values for each of the key sizes rather than having just
1550 parameters (and 'speed' generating keys each time).
1553 *) Speed up EVP routines.
1556 type 8 bytes 64 bytes 256 bytes 1024 bytes 8192 bytes
1557 des-cbc 4408.85k 5560.51k 5778.46k 5862.20k 5825.16k
1558 des-cbc 4389.55k 5571.17k 5792.23k 5846.91k 5832.11k
1559 des-cbc 4394.32k 5575.92k 5807.44k 5848.37k 5841.30k
1561 des-cbc 3482.66k 5069.49k 5496.39k 5614.16k 5639.28k
1562 des-cbc 3480.74k 5068.76k 5510.34k 5609.87k 5635.52k
1563 des-cbc 3483.72k 5067.62k 5504.60k 5708.01k 5724.80k
1566 des-cbc 4660.16k 5650.19k 5807.19k 5827.13k 5783.32k
1568 des-cbc 3624.96k 5258.21k 5530.91k 5624.30k 5628.26k
1571 *) Added the OS2-EMX target.
1572 ["Brian Havard" <brianh@kheldar.apana.org.au> and Richard Levitte]
1574 *) Rewrite apps to use NCONF routines instead of the old CONF. New functions
1575 to support NCONF routines in extension code. New function CONF_set_nconf()
1576 to allow functions which take an NCONF to also handle the old LHASH
1577 structure: this means that the old CONF compatible routines can be
1578 retained (in particular wrt extensions) without having to duplicate the
1579 code. New function X509V3_add_ext_nconf_sk to add extensions to a stack.
1582 *) Enhance the general user interface with mechanisms for inner control
1583 and with possibilities to have yes/no kind of prompts.
1586 *) Change all calls to low level digest routines in the library and
1587 applications to use EVP. Add missing calls to HMAC_cleanup() and
1588 don't assume HMAC_CTX can be copied using memcpy().
1589 [Verdon Walker <VWalker@novell.com>, Steve Henson]
1591 *) Add the possibility to control engines through control names but with
1592 arbitrary arguments instead of just a string.
1593 Change the key loaders to take a UI_METHOD instead of a callback
1594 function pointer. NOTE: this breaks binary compatibility with earlier
1595 versions of OpenSSL [engine].
1596 Adapt the nCipher code for these new conditions and add a card insertion
1600 *) Enhance the general user interface with mechanisms to better support
1601 dialog box interfaces, application-defined prompts, the possibility
1602 to use defaults (for example default passwords from somewhere else)
1603 and interrupts/cancellations.
1606 *) Tidy up PKCS#12 attribute handling. Add support for the CSP name
1607 attribute in PKCS#12 files, add new -CSP option to pkcs12 utility.
1610 *) Fix a memory leak in 'sk_dup()' in the case reallocation fails. (Also
1611 tidy up some unnecessarily weird code in 'sk_new()').
1612 [Geoff, reported by Diego Tartara <dtartara@novamens.com>]
1614 *) Change the key loading routines for ENGINEs to use the same kind
1615 callback (pem_password_cb) as all other routines that need this
1619 *) Increase ENTROPY_NEEDED to 32 bytes, as Rijndael can operate with
1620 256 bit (=32 byte) keys. Of course seeding with more entropy bytes
1621 than this minimum value is recommended.
1624 *) New random seeder for OpenVMS, using the system process statistics
1625 that are easily reachable.
1628 *) Windows apparently can't transparently handle global
1629 variables defined in DLLs. Initialisations such as:
1631 const ASN1_ITEM *it = &ASN1_INTEGER_it;
1633 wont compile. This is used by the any applications that need to
1634 declare their own ASN1 modules. This was fixed by adding the option
1635 EXPORT_VAR_AS_FN to all Win32 platforms, although this isn't strictly
1636 needed for static libraries under Win32.
1639 *) New functions X509_PURPOSE_set() and X509_TRUST_set() to handle
1640 setting of purpose and trust fields. New X509_STORE trust and
1641 purpose functions and tidy up setting in other SSL functions.
1644 *) Add copies of X509_STORE_CTX fields and callbacks to X509_STORE
1645 structure. These are inherited by X509_STORE_CTX when it is
1646 initialised. This allows various defaults to be set in the
1647 X509_STORE structure (such as flags for CRL checking and custom
1648 purpose or trust settings) for functions which only use X509_STORE_CTX
1649 internally such as S/MIME.
1651 Modify X509_STORE_CTX_purpose_inherit() so it only sets purposes and
1652 trust settings if they are not set in X509_STORE. This allows X509_STORE
1653 purposes and trust (in S/MIME for example) to override any set by default.
1655 Add command line options for CRL checking to smime, s_client and s_server
1659 *) Initial CRL based revocation checking. If the CRL checking flag(s)
1660 are set then the CRL is looked up in the X509_STORE structure and
1661 its validity and signature checked, then if the certificate is found
1662 in the CRL the verify fails with a revoked error.
1664 Various new CRL related callbacks added to X509_STORE_CTX structure.
1666 Command line options added to 'verify' application to support this.
1668 This needs some additional work, such as being able to handle multiple
1669 CRLs with different times, extension based lookup (rather than just
1670 by subject name) and ultimately more complete V2 CRL extension
1674 *) Add a general user interface API (crypto/ui/). This is designed
1675 to replace things like des_read_password and friends (backward
1676 compatibility functions using this new API are provided).
1677 The purpose is to remove prompting functions from the DES code
1678 section as well as provide for prompting through dialog boxes in
1679 a window system and the like.
1682 *) Add "ex_data" support to ENGINE so implementations can add state at a
1683 per-structure level rather than having to store it globally.
1686 *) Make it possible for ENGINE structures to be copied when retrieved by
1687 ENGINE_by_id() if the ENGINE specifies a new flag: ENGINE_FLAGS_BY_ID_COPY.
1688 This causes the "original" ENGINE structure to act like a template,
1689 analogous to the RSA vs. RSA_METHOD type of separation. Because of this
1690 operational state can be localised to each ENGINE structure, despite the
1691 fact they all share the same "methods". New ENGINE structures returned in
1692 this case have no functional references and the return value is the single
1693 structural reference. This matches the single structural reference returned
1694 by ENGINE_by_id() normally, when it is incremented on the pre-existing
1698 *) Fix ASN1 decoder when decoding type ANY and V_ASN1_OTHER: since this
1699 needs to match any other type at all we need to manually clear the
1703 *) Changes to the "openssl engine" utility to include;
1704 - verbosity levels ('-v', '-vv', and '-vvv') that provide information
1705 about an ENGINE's available control commands.
1706 - executing control commands from command line arguments using the
1707 '-pre' and '-post' switches. '-post' is only used if '-t' is
1708 specified and the ENGINE is successfully initialised. The syntax for
1709 the individual commands are colon-separated, for example;
1710 openssl engine chil -pre FORK_CHECK:0 -pre SO_PATH:/lib/test.so
1713 *) New dynamic control command support for ENGINEs. ENGINEs can now
1714 declare their own commands (numbers), names (strings), descriptions,
1715 and input types for run-time discovery by calling applications. A
1716 subset of these commands are implicitly classed as "executable"
1717 depending on their input type, and only these can be invoked through
1718 the new string-based API function ENGINE_ctrl_cmd_string(). (Eg. this
1719 can be based on user input, config files, etc). The distinction is
1720 that "executable" commands cannot return anything other than a boolean
1721 result and can only support numeric or string input, whereas some
1722 discoverable commands may only be for direct use through
1723 ENGINE_ctrl(), eg. supporting the exchange of binary data, function
1724 pointers, or other custom uses. The "executable" commands are to
1725 support parameterisations of ENGINE behaviour that can be
1726 unambiguously defined by ENGINEs and used consistently across any
1727 OpenSSL-based application. Commands have been added to all the
1728 existing hardware-supporting ENGINEs, noticeably "SO_PATH" to allow
1729 control over shared-library paths without source code alterations.
1732 *) Changed all ENGINE implementations to dynamically allocate their
1733 ENGINEs rather than declaring them statically. Apart from this being
1734 necessary with the removal of the ENGINE_FLAGS_MALLOCED distinction,
1735 this also allows the implementations to compile without using the
1736 internal engine_int.h header.
1739 *) Minor adjustment to "rand" code. RAND_get_rand_method() now returns a
1740 'const' value. Any code that should be able to modify a RAND_METHOD
1741 should already have non-const pointers to it (ie. they should only
1742 modify their own ones).
1745 *) Made a variety of little tweaks to the ENGINE code.
1746 - "atalla" and "ubsec" string definitions were moved from header files
1747 to C code. "nuron" string definitions were placed in variables
1748 rather than hard-coded - allowing parameterisation of these values
1749 later on via ctrl() commands.
1750 - Removed unused "#if 0"'d code.
1751 - Fixed engine list iteration code so it uses ENGINE_free() to release
1752 structural references.
1753 - Constified the RAND_METHOD element of ENGINE structures.
1754 - Constified various get/set functions as appropriate and added
1755 missing functions (including a catch-all ENGINE_cpy that duplicates
1756 all ENGINE values onto a new ENGINE except reference counts/state).
1757 - Removed NULL parameter checks in get/set functions. Setting a method
1758 or function to NULL is a way of cancelling out a previously set
1759 value. Passing a NULL ENGINE parameter is just plain stupid anyway
1760 and doesn't justify the extra error symbols and code.
1761 - Deprecate the ENGINE_FLAGS_MALLOCED define and move the area for
1762 flags from engine_int.h to engine.h.
1763 - Changed prototypes for ENGINE handler functions (init(), finish(),
1764 ctrl(), key-load functions, etc) to take an (ENGINE*) parameter.
1767 *) Implement binary inversion algorithm for BN_mod_inverse in addition
1768 to the algorithm using long division. The binary algorithm can be
1769 used only if the modulus is odd. On 32-bit systems, it is faster
1770 only for relatively small moduli (roughly 20-30% for 128-bit moduli,
1771 roughly 5-15% for 256-bit moduli), so we use it only for moduli
1772 up to 450 bits. In 64-bit environments, the binary algorithm
1773 appears to be advantageous for much longer moduli; here we use it
1774 for moduli up to 2048 bits.
1777 *) Rewrite CHOICE field setting in ASN1_item_ex_d2i(). The old code
1778 could not support the combine flag in choice fields.
1781 *) Add a 'copy_extensions' option to the 'ca' utility. This copies
1782 extensions from a certificate request to the certificate.
1785 *) Allow multiple 'certopt' and 'nameopt' options to be separated
1786 by commas. Add 'namopt' and 'certopt' options to the 'ca' config
1787 file: this allows the display of the certificate about to be
1788 signed to be customised, to allow certain fields to be included
1789 or excluded and extension details. The old system didn't display
1790 multicharacter strings properly, omitted fields not in the policy
1791 and couldn't display additional details such as extensions.
1794 *) Function EC_POINTs_mul for multiple scalar multiplication
1795 of an arbitrary number of elliptic curve points
1796 \sum scalars[i]*points[i],
1797 optionally including the generator defined for the EC_GROUP:
1798 scalar*generator + \sum scalars[i]*points[i].
1800 EC_POINT_mul is a simple wrapper function for the typical case
1801 that the point list has just one item (besides the optional
1805 *) First EC_METHODs for curves over GF(p):
1807 EC_GFp_simple_method() uses the basic BN_mod_mul and BN_mod_sqr
1808 operations and provides various method functions that can also
1809 operate with faster implementations of modular arithmetic.
1811 EC_GFp_mont_method() reuses most functions that are part of
1812 EC_GFp_simple_method, but uses Montgomery arithmetic.
1814 [Bodo Moeller; point addition and point doubling
1815 implementation directly derived from source code provided by
1816 Lenka Fibikova <fibikova@exp-math.uni-essen.de>]
1818 *) Framework for elliptic curves (crypto/ec/ec.h, crypto/ec/ec_lcl.h,
1819 crypto/ec/ec_lib.c):
1821 Curves are EC_GROUP objects (with an optional group generator)
1822 based on EC_METHODs that are built into the library.
1824 Points are EC_POINT objects based on EC_GROUP objects.
1826 Most of the framework would be able to handle curves over arbitrary
1827 finite fields, but as there are no obvious types for fields other
1828 than GF(p), some functions are limited to that for now.
1831 *) Add the -HTTP option to s_server. It is similar to -WWW, but requires
1832 that the file contains a complete HTTP response.
1835 *) Add the ec directory to mkdef.pl and mkfiles.pl. In mkdef.pl
1836 change the def and num file printf format specifier from "%-40sXXX"
1837 to "%-39s XXX". The latter will always guarantee a space after the
1838 field while the former will cause them to run together if the field
1839 is 40 of more characters long.
1842 *) Constify the cipher and digest 'method' functions and structures
1843 and modify related functions to take constant EVP_MD and EVP_CIPHER
1847 *) Hide BN_CTX structure details in bn_lcl.h instead of publishing them
1848 in <openssl/bn.h>. Also further increase BN_CTX_NUM to 32.
1851 *) Modify EVP_Digest*() routines so they now return values. Although the
1852 internal software routines can never fail additional hardware versions
1856 *) Clean up crypto/err/err.h and change some error codes to avoid conflicts:
1858 Previously ERR_R_FATAL was too small and coincided with ERR_LIB_PKCS7
1859 (= ERR_R_PKCS7_LIB); it is now 64 instead of 32.
1862 ERR_R_NESTED_ASN1_ERROR
1864 ERR_R_MISSING_ASN1_EOS
1865 were 4 .. 9, conflicting with
1866 ERR_LIB_RSA (= ERR_R_RSA_LIB)
1868 ERR_LIB_PEM (= ERR_R_PEM_LIB).
1869 They are now 58 .. 63 (i.e., just below ERR_R_FATAL).
1871 Add new error code 'ERR_R_INTERNAL_ERROR'.
1874 *) Don't overuse locks in crypto/err/err.c: For data retrieval, CRYPTO_r_lock
1878 *) New option '-subj arg' for 'openssl req' and 'openssl ca'. This
1879 sets the subject name for a new request or supersedes the
1880 subject name in a given request. Formats that can be parsed are
1881 'CN=Some Name, OU=myOU, C=IT'
1883 'CN=Some Name/OU=myOU/C=IT'.
1885 Add options '-batch' and '-verbose' to 'openssl req'.
1886 [Massimiliano Pala <madwolf@hackmasters.net>]
1888 *) Introduce the possibility to access global variables through
1889 functions on platform were that's the best way to handle exporting
1890 global variables in shared libraries. To enable this functionality,
1891 one must configure with "EXPORT_VAR_AS_FN" or defined the C macro
1892 "OPENSSL_EXPORT_VAR_AS_FUNCTION" in crypto/opensslconf.h (the latter
1893 is normally done by Configure or something similar).
1895 To implement a global variable, use the macro OPENSSL_IMPLEMENT_GLOBAL
1896 in the source file (foo.c) like this:
1898 OPENSSL_IMPLEMENT_GLOBAL(int,foo)=1;
1899 OPENSSL_IMPLEMENT_GLOBAL(double,bar);
1901 To declare a global variable, use the macros OPENSSL_DECLARE_GLOBAL
1902 and OPENSSL_GLOBAL_REF in the header file (foo.h) like this:
1904 OPENSSL_DECLARE_GLOBAL(int,foo);
1905 #define foo OPENSSL_GLOBAL_REF(foo)
1906 OPENSSL_DECLARE_GLOBAL(double,bar);
1907 #define bar OPENSSL_GLOBAL_REF(bar)
1909 The #defines are very important, and therefore so is including the
1910 header file everywhere where the defined globals are used.
1912 The macro OPENSSL_EXPORT_VAR_AS_FUNCTION also affects the definition
1913 of ASN.1 items, but that structure is a bit different.
1915 The largest change is in util/mkdef.pl which has been enhanced with
1916 better and easier to understand logic to choose which symbols should
1917 go into the Windows .def files as well as a number of fixes and code
1918 cleanup (among others, algorithm keywords are now sorted
1919 lexicographically to avoid constant rewrites).
1922 *) In BN_div() keep a copy of the sign of 'num' before writing the
1923 result to 'rm' because if rm==num the value will be overwritten
1924 and produce the wrong result if 'num' is negative: this caused
1925 problems with BN_mod() and BN_nnmod().
1928 *) Function OCSP_request_verify(). This checks the signature on an
1929 OCSP request and verifies the signer certificate. The signer
1930 certificate is just checked for a generic purpose and OCSP request
1934 *) Add OCSP_check_validity() function to check the validity of OCSP
1935 responses. OCSP responses are prepared in real time and may only
1936 be a few seconds old. Simply checking that the current time lies
1937 between thisUpdate and nextUpdate max reject otherwise valid responses
1938 caused by either OCSP responder or client clock inaccuracy. Instead
1939 we allow thisUpdate and nextUpdate to fall within a certain period of
1940 the current time. The age of the response can also optionally be
1941 checked. Two new options -validity_period and -status_age added to
1945 *) If signature or public key algorithm is unrecognized print out its
1946 OID rather that just UNKNOWN.
1949 *) Change OCSP_cert_to_id() to tolerate a NULL subject certificate and
1950 OCSP_cert_id_new() a NULL serialNumber. This allows a partial certificate
1951 ID to be generated from the issuer certificate alone which can then be
1952 passed to OCSP_id_issuer_cmp().
1955 *) New compilation option ASN1_ITEM_FUNCTIONS. This causes the new
1956 ASN1 modules to export functions returning ASN1_ITEM pointers
1957 instead of the ASN1_ITEM structures themselves. This adds several
1958 new macros which allow the underlying ASN1 function/structure to
1959 be accessed transparently. As a result code should not use ASN1_ITEM
1960 references directly (such as &X509_it) but instead use the relevant
1961 macros (such as ASN1_ITEM_rptr(X509)). This option is to allow
1962 use of the new ASN1 code on platforms where exporting structures
1963 is problematical (for example in shared libraries) but exporting
1964 functions returning pointers to structures is not.
1967 *) Add support for overriding the generation of SSL/TLS session IDs.
1968 These callbacks can be registered either in an SSL_CTX or per SSL.
1969 The purpose of this is to allow applications to control, if they wish,
1970 the arbitrary values chosen for use as session IDs, particularly as it
1971 can be useful for session caching in multiple-server environments. A
1972 command-line switch for testing this (and any client code that wishes
1973 to use such a feature) has been added to "s_server".
1974 [Geoff Thorpe, Lutz Jaenicke]
1976 *) Modify mkdef.pl to recognise and parse preprocessor conditionals
1977 of the form '#if defined(...) || defined(...) || ...' and
1978 '#if !defined(...) && !defined(...) && ...'. This also avoids
1979 the growing number of special cases it was previously handling.
1982 *) Make all configuration macros available for application by making
1983 sure they are available in opensslconf.h, by giving them names starting
1984 with "OPENSSL_" to avoid conflicts with other packages and by making
1985 sure e_os2.h will cover all platform-specific cases together with
1987 Additionally, it is now possible to define configuration/platform-
1988 specific names (called "system identities"). In the C code, these
1989 are prefixed with "OPENSSL_SYSNAME_". e_os2.h will create another
1990 macro with the name beginning with "OPENSSL_SYS_", which is determined
1991 from "OPENSSL_SYSNAME_*" or compiler-specific macros depending on
1995 *) New option -set_serial to 'req' and 'x509' this allows the serial
1996 number to use to be specified on the command line. Previously self
1997 signed certificates were hard coded with serial number 0 and the
1998 CA options of 'x509' had to use a serial number in a file which was
2002 *) New options to 'ca' utility to support V2 CRL entry extensions.
2003 Currently CRL reason, invalidity date and hold instruction are
2004 supported. Add new CRL extensions to V3 code and some new objects.
2007 *) New function EVP_CIPHER_CTX_set_padding() this is used to
2008 disable standard block padding (aka PKCS#5 padding) in the EVP
2009 API, which was previously mandatory. This means that the data is
2010 not padded in any way and so the total length much be a multiple
2011 of the block size, otherwise an error occurs.
2014 *) Initial (incomplete) OCSP SSL support.
2017 *) New function OCSP_parse_url(). This splits up a URL into its host,
2018 port and path components: primarily to parse OCSP URLs. New -url
2019 option to ocsp utility.
2022 *) New nonce behavior. The return value of OCSP_check_nonce() now
2023 reflects the various checks performed. Applications can decide
2024 whether to tolerate certain situations such as an absent nonce
2025 in a response when one was present in a request: the ocsp application
2026 just prints out a warning. New function OCSP_add1_basic_nonce()
2027 this is to allow responders to include a nonce in a response even if
2028 the request is nonce-less.
2031 *) Disable stdin buffering in load_cert (apps/apps.c) so that no certs are
2032 skipped when using openssl x509 multiple times on a single input file,
2033 e.g. "(openssl x509 -out cert1; openssl x509 -out cert2) <certs".
2036 *) Make ASN1_UTCTIME_set_string() and ASN1_GENERALIZEDTIME_set_string()
2037 set string type: to handle setting ASN1_TIME structures. Fix ca
2038 utility to correctly initialize revocation date of CRLs.
2041 *) New option SSL_OP_CIPHER_SERVER_PREFERENCE allows the server to override
2042 the clients preferred ciphersuites and rather use its own preferences.
2043 Should help to work around M$ SGC (Server Gated Cryptography) bug in
2044 Internet Explorer by ensuring unchanged hash method during stepup.
2045 (Also replaces the broken/deactivated SSL_OP_NON_EXPORT_FIRST option.)
2048 *) Make mkdef.pl recognise all DECLARE_ASN1 macros, change rijndael
2049 to aes and add a new 'exist' option to print out symbols that don't
2053 *) Additional options to ocsp utility to allow flags to be set and
2054 additional certificates supplied.
2057 *) Add the option -VAfile to 'openssl ocsp', so the user can give the
2058 OCSP client a number of certificate to only verify the response
2062 *) Update Rijndael code to version 3.0 and change EVP AES ciphers to
2063 handle the new API. Currently only ECB, CBC modes supported. Add new
2066 Add TLS AES ciphersuites as described in RFC3268, "Advanced
2067 Encryption Standard (AES) Ciphersuites for Transport Layer
2068 Security (TLS)". (In beta versions of OpenSSL 0.9.7, these were
2069 not enabled by default and were not part of the "ALL" ciphersuite
2070 alias because they were not yet official; they could be
2071 explicitly requested by specifying the "AESdraft" ciphersuite
2072 group alias. In the final release of OpenSSL 0.9.7, the group
2073 alias is called "AES" and is part of "ALL".)
2074 [Ben Laurie, Steve Henson, Bodo Moeller]
2076 *) New function OCSP_copy_nonce() to copy nonce value (if present) from
2077 request to response.
2080 *) Functions for OCSP responders. OCSP_request_onereq_count(),
2081 OCSP_request_onereq_get0(), OCSP_onereq_get0_id() and OCSP_id_get0_info()
2082 extract information from a certificate request. OCSP_response_create()
2083 creates a response and optionally adds a basic response structure.
2084 OCSP_basic_add1_status() adds a complete single response to a basic
2085 response and returns the OCSP_SINGLERESP structure just added (to allow
2086 extensions to be included for example). OCSP_basic_add1_cert() adds a
2087 certificate to a basic response and OCSP_basic_sign() signs a basic
2088 response with various flags. New helper functions ASN1_TIME_check()
2089 (checks validity of ASN1_TIME structure) and ASN1_TIME_to_generalizedtime()
2090 (converts ASN1_TIME to GeneralizedTime).
2093 *) Various new functions. EVP_Digest() combines EVP_Digest{Init,Update,Final}()
2094 in a single operation. X509_get0_pubkey_bitstr() extracts the public_key
2095 structure from a certificate. X509_pubkey_digest() digests the public_key
2096 contents: this is used in various key identifiers.
2099 *) Make sk_sort() tolerate a NULL argument.
2100 [Steve Henson reported by Massimiliano Pala <madwolf@comune.modena.it>]
2102 *) New OCSP verify flag OCSP_TRUSTOTHER. When set the "other" certificates
2103 passed by the function are trusted implicitly. If any of them signed the
2104 response then it is assumed to be valid and is not verified.
2107 *) In PKCS7_set_type() initialise content_type in PKCS7_ENC_CONTENT
2108 to data. This was previously part of the PKCS7 ASN1 code. This
2109 was causing problems with OpenSSL created PKCS#12 and PKCS#7 structures.
2110 [Steve Henson, reported by Kenneth R. Robinette
2111 <support@securenetterm.com>]
2113 *) Add CRYPTO_push_info() and CRYPTO_pop_info() calls to new ASN1
2114 routines: without these tracing memory leaks is very painful.
2115 Fix leaks in PKCS12 and PKCS7 routines.
2118 *) Make X509_time_adj() cope with the new behaviour of ASN1_TIME_new().
2119 Previously it initialised the 'type' argument to V_ASN1_UTCTIME which
2120 effectively meant GeneralizedTime would never be used. Now it
2121 is initialised to -1 but X509_time_adj() now has to check the value
2122 and use ASN1_TIME_set() if the value is not V_ASN1_UTCTIME or
2123 V_ASN1_GENERALIZEDTIME, without this it always uses GeneralizedTime.
2124 [Steve Henson, reported by Kenneth R. Robinette
2125 <support@securenetterm.com>]
2127 *) Fixes to BN_to_ASN1_INTEGER when bn is zero. This would previously
2128 result in a zero length in the ASN1_INTEGER structure which was
2129 not consistent with the structure when d2i_ASN1_INTEGER() was used
2130 and would cause ASN1_INTEGER_cmp() to fail. Enhance s2i_ASN1_INTEGER()
2131 to cope with hex and negative integers. Fix bug in i2a_ASN1_INTEGER()
2132 where it did not print out a minus for negative ASN1_INTEGER.
2135 *) Add summary printout to ocsp utility. The various functions which
2136 convert status values to strings have been renamed to:
2137 OCSP_response_status_str(), OCSP_cert_status_str() and
2138 OCSP_crl_reason_str() and are no longer static. New options
2139 to verify nonce values and to disable verification. OCSP response
2140 printout format cleaned up.
2143 *) Add additional OCSP certificate checks. These are those specified
2144 in RFC2560. This consists of two separate checks: the CA of the
2145 certificate being checked must either be the OCSP signer certificate
2146 or the issuer of the OCSP signer certificate. In the latter case the
2147 OCSP signer certificate must contain the OCSP signing extended key
2148 usage. This check is performed by attempting to match the OCSP
2149 signer or the OCSP signer CA to the issuerNameHash and issuerKeyHash
2150 in the OCSP_CERTID structures of the response.
2153 *) Initial OCSP certificate verification added to OCSP_basic_verify()
2154 and related routines. This uses the standard OpenSSL certificate
2155 verify routines to perform initial checks (just CA validity) and
2156 to obtain the certificate chain. Then additional checks will be
2157 performed on the chain. Currently the root CA is checked to see
2158 if it is explicitly trusted for OCSP signing. This is used to set
2159 a root CA as a global signing root: that is any certificate that
2160 chains to that CA is an acceptable OCSP signing certificate.
2163 *) New '-extfile ...' option to 'openssl ca' for reading X.509v3
2164 extensions from a separate configuration file.
2165 As when reading extensions from the main configuration file,
2166 the '-extensions ...' option may be used for specifying the
2168 [Massimiliano Pala <madwolf@comune.modena.it>]
2170 *) New OCSP utility. Allows OCSP requests to be generated or
2171 read. The request can be sent to a responder and the output
2172 parsed, outputed or printed in text form. Not complete yet:
2173 still needs to check the OCSP response validity.
2176 *) New subcommands for 'openssl ca':
2177 'openssl ca -status <serial>' prints the status of the cert with
2178 the given serial number (according to the index file).
2179 'openssl ca -updatedb' updates the expiry status of certificates
2181 [Massimiliano Pala <madwolf@comune.modena.it>]
2183 *) New '-newreq-nodes' command option to CA.pl. This is like
2184 '-newreq', but calls 'openssl req' with the '-nodes' option
2185 so that the resulting key is not encrypted.
2186 [Damien Miller <djm@mindrot.org>]
2188 *) New configuration for the GNU Hurd.
2189 [Jonathan Bartlett <johnnyb@wolfram.com> via Richard Levitte]
2191 *) Initial code to implement OCSP basic response verify. This
2192 is currently incomplete. Currently just finds the signer's
2193 certificate and verifies the signature on the response.
2196 *) New SSLeay_version code SSLEAY_DIR to determine the compiled-in
2197 value of OPENSSLDIR. This is available via the new '-d' option
2198 to 'openssl version', and is also included in 'openssl version -a'.
2201 *) Allowing defining memory allocation callbacks that will be given
2202 file name and line number information in additional arguments
2203 (a const char* and an int). The basic functionality remains, as
2204 well as the original possibility to just replace malloc(),
2205 realloc() and free() by functions that do not know about these
2206 additional arguments. To register and find out the current
2207 settings for extended allocation functions, the following
2208 functions are provided:
2210 CRYPTO_set_mem_ex_functions
2211 CRYPTO_set_locked_mem_ex_functions
2212 CRYPTO_get_mem_ex_functions
2213 CRYPTO_get_locked_mem_ex_functions
2215 These work the same way as CRYPTO_set_mem_functions and friends.
2216 CRYPTO_get_[locked_]mem_functions now writes 0 where such an
2217 extended allocation function is enabled.
2218 Similarly, CRYPTO_get_[locked_]mem_ex_functions writes 0 where
2219 a conventional allocation function is enabled.
2220 [Richard Levitte, Bodo Moeller]
2222 *) Finish off removing the remaining LHASH function pointer casts.
2223 There should no longer be any prototype-casting required when using
2224 the LHASH abstraction, and any casts that remain are "bugs". See
2225 the callback types and macros at the head of lhash.h for details
2226 (and "OBJ_cleanup" in crypto/objects/obj_dat.c as an example).
2229 *) Add automatic query of EGD sockets in RAND_poll() for the unix variant.
2230 If /dev/[u]random devices are not available or do not return enough
2231 entropy, EGD style sockets (served by EGD or PRNGD) will automatically
2233 The locations /var/run/egd-pool, /dev/egd-pool, /etc/egd-pool, and
2234 /etc/entropy will be queried once each in this sequence, quering stops
2235 when enough entropy was collected without querying more sockets.
2238 *) Change the Unix RAND_poll() variant to be able to poll several
2239 random devices, as specified by DEVRANDOM, until a sufficient amount
2240 of data has been collected. We spend at most 10 ms on each file
2241 (select timeout) and read in non-blocking mode. DEVRANDOM now
2242 defaults to the list "/dev/urandom", "/dev/random", "/dev/srandom"
2243 (previously it was just the string "/dev/urandom"), so on typical
2244 platforms the 10 ms delay will never occur.
2245 Also separate out the Unix variant to its own file, rand_unix.c.
2246 For VMS, there's a currently-empty rand_vms.c.
2249 *) Move OCSP client related routines to ocsp_cl.c. These
2250 provide utility functions which an application needing
2251 to issue a request to an OCSP responder and analyse the
2252 response will typically need: as opposed to those which an
2253 OCSP responder itself would need which will be added later.
2255 OCSP_request_sign() signs an OCSP request with an API similar
2256 to PKCS7_sign(). OCSP_response_status() returns status of OCSP
2257 response. OCSP_response_get1_basic() extracts basic response
2258 from response. OCSP_resp_find_status(): finds and extracts status
2259 information from an OCSP_CERTID structure (which will be created
2260 when the request structure is built). These are built from lower
2261 level functions which work on OCSP_SINGLERESP structures but
2262 wont normally be used unless the application wishes to examine
2263 extensions in the OCSP response for example.
2265 Replace nonce routines with a pair of functions.
2266 OCSP_request_add1_nonce() adds a nonce value and optionally
2267 generates a random value. OCSP_check_nonce() checks the
2268 validity of the nonce in an OCSP response.
2271 *) Change function OCSP_request_add() to OCSP_request_add0_id().
2272 This doesn't copy the supplied OCSP_CERTID and avoids the
2273 need to free up the newly created id. Change return type
2274 to OCSP_ONEREQ to return the internal OCSP_ONEREQ structure.
2275 This can then be used to add extensions to the request.
2276 Deleted OCSP_request_new(), since most of its functionality
2277 is now in OCSP_REQUEST_new() (and the case insensitive name
2278 clash) apart from the ability to set the request name which
2279 will be added elsewhere.
2282 *) Update OCSP API. Remove obsolete extensions argument from
2283 various functions. Extensions are now handled using the new
2284 OCSP extension code. New simple OCSP HTTP function which
2285 can be used to send requests and parse the response.
2288 *) Fix the PKCS#7 (S/MIME) code to work with new ASN1. Two new
2289 ASN1_ITEM structures help with sign and verify. PKCS7_ATTR_SIGN
2290 uses the special reorder version of SET OF to sort the attributes
2291 and reorder them to match the encoded order. This resolves a long
2292 standing problem: a verify on a PKCS7 structure just after signing
2293 it used to fail because the attribute order did not match the
2294 encoded order. PKCS7_ATTR_VERIFY does not reorder the attributes:
2295 it uses the received order. This is necessary to tolerate some broken
2296 software that does not order SET OF. This is handled by encoding
2297 as a SEQUENCE OF but using implicit tagging (with UNIVERSAL class)
2298 to produce the required SET OF.
2301 *) Have mk1mf.pl generate the macros OPENSSL_BUILD_SHLIBCRYPTO and
2302 OPENSSL_BUILD_SHLIBSSL and use them appropriately in the header
2303 files to get correct declarations of the ASN.1 item variables.
2306 *) Rewrite of PKCS#12 code to use new ASN1 functionality. Replace many
2307 PKCS#12 macros with real functions. Fix two unrelated ASN1 bugs:
2308 asn1_check_tlen() would sometimes attempt to use 'ctx' when it was
2309 NULL and ASN1_TYPE was not dereferenced properly in asn1_ex_c2i().
2310 New ASN1 macro: DECLARE_ASN1_ITEM() which just declares the relevant
2311 ASN1_ITEM and no wrapper functions.
2314 *) New functions or ASN1_item_d2i_fp() and ASN1_item_d2i_bio(). These
2315 replace the old function pointer based I/O routines. Change most of
2316 the *_d2i_bio() and *_d2i_fp() functions to use these.
2319 *) Enhance mkdef.pl to be more accepting about spacing in C preprocessor
2320 lines, recognice more "algorithms" that can be deselected, and make
2321 it complain about algorithm deselection that isn't recognised.
2324 *) New ASN1 functions to handle dup, sign, verify, digest, pack and
2325 unpack operations in terms of ASN1_ITEM. Modify existing wrappers
2326 to use new functions. Add NO_ASN1_OLD which can be set to remove
2327 some old style ASN1 functions: this can be used to determine if old
2328 code will still work when these eventually go away.
2331 *) New extension functions for OCSP structures, these follow the
2332 same conventions as certificates and CRLs.
2335 *) New function X509V3_add1_i2d(). This automatically encodes and
2336 adds an extension. Its behaviour can be customised with various
2337 flags to append, replace or delete. Various wrappers added for
2338 certifcates and CRLs.
2341 *) Fix to avoid calling the underlying ASN1 print routine when
2342 an extension cannot be parsed. Correct a typo in the
2343 OCSP_SERVICELOC extension. Tidy up print OCSP format.
2346 *) Make mkdef.pl parse some of the ASN1 macros and add apropriate
2347 entries for variables.
2350 *) Add functionality to apps/openssl.c for detecting locking
2351 problems: As the program is single-threaded, all we have
2352 to do is register a locking callback using an array for
2353 storing which locks are currently held by the program.
2356 *) Use a lock around the call to CRYPTO_get_ex_new_index() in
2357 SSL_get_ex_data_X509_STORE_idx(), which is used in
2358 ssl_verify_cert_chain() and thus can be called at any time
2359 during TLS/SSL handshakes so that thread-safety is essential.
2360 Unfortunately, the ex_data design is not at all suited
2361 for multi-threaded use, so it probably should be abolished.
2364 *) Added Broadcom "ubsec" ENGINE to OpenSSL.
2365 [Broadcom, tweaked and integrated by Geoff Thorpe]
2367 *) Move common extension printing code to new function
2368 X509V3_print_extensions(). Reorganise OCSP print routines and
2369 implement some needed OCSP ASN1 functions. Add OCSP extensions.
2372 *) New function X509_signature_print() to remove duplication in some
2376 *) Add a special meaning when SET OF and SEQUENCE OF flags are both
2377 set (this was treated exactly the same as SET OF previously). This
2378 is used to reorder the STACK representing the structure to match the
2379 encoding. This will be used to get round a problem where a PKCS7
2380 structure which was signed could not be verified because the STACK
2381 order did not reflect the encoded order.
2384 *) Reimplement the OCSP ASN1 module using the new code.
2387 *) Update the X509V3 code to permit the use of an ASN1_ITEM structure
2388 for its ASN1 operations. The old style function pointers still exist
2389 for now but they will eventually go away.
2392 *) Merge in replacement ASN1 code from the ASN1 branch. This almost
2393 completely replaces the old ASN1 functionality with a table driven
2394 encoder and decoder which interprets an ASN1_ITEM structure describing
2395 the ASN1 module. Compatibility with the existing ASN1 API (i2d,d2i) is
2396 largely maintained. Almost all of the old asn1_mac.h macro based ASN1
2397 has also been converted to the new form.
2400 *) Change BN_mod_exp_recp so that negative moduli are tolerated
2401 (the sign is ignored). Similarly, ignore the sign in BN_MONT_CTX_set
2402 so that BN_mod_exp_mont and BN_mod_exp_mont_word work
2403 for negative moduli.
2406 *) Fix BN_uadd and BN_usub: Always return non-negative results instead
2407 of not touching the result's sign bit.
2410 *) BN_div bugfix: If the result is 0, the sign (res->neg) must not be
2414 *) Changed the LHASH code to use prototypes for callbacks, and created
2415 macros to declare and implement thin (optionally static) functions
2416 that provide type-safety and avoid function pointer casting for the
2417 type-specific callbacks.
2420 *) Added Kerberos Cipher Suites to be used with TLS, as written in
2422 [Veers Staats <staatsvr@asc.hpc.mil>,
2423 Jeffrey Altman <jaltman@columbia.edu>, via Richard Levitte]
2425 *) Reformat the FAQ so the different questions and answers can be divided
2426 in sections depending on the subject.
2429 *) Have the zlib compression code load ZLIB.DLL dynamically under
2433 *) New function BN_mod_sqrt for computing square roots modulo a prime
2434 (using the probabilistic Tonelli-Shanks algorithm unless
2435 p == 3 (mod 4) or p == 5 (mod 8), which are cases that can
2436 be handled deterministically).
2437 [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller]
2439 *) Make BN_mod_inverse faster by explicitly handling small quotients
2440 in the Euclid loop. (Speed gain about 20% for small moduli [256 or
2441 512 bits], about 30% for larger ones [1024 or 2048 bits].)
2444 *) New function BN_kronecker.
2447 *) Fix BN_gcd so that it works on negative inputs; the result is
2448 positive unless both parameters are zero.
2449 Previously something reasonably close to an infinite loop was
2450 possible because numbers could be growing instead of shrinking
2451 in the implementation of Euclid's algorithm.
2454 *) Fix BN_is_word() and BN_is_one() macros to take into account the
2455 sign of the number in question.
2457 Fix BN_is_word(a,w) to work correctly for w == 0.
2459 The old BN_is_word(a,w) macro is now called BN_abs_is_word(a,w)
2460 because its test if the absolute value of 'a' equals 'w'.
2461 Note that BN_abs_is_word does *not* handle w == 0 reliably;
2462 it exists mostly for use in the implementations of BN_is_zero(),
2463 BN_is_one(), and BN_is_word().
2466 *) New function BN_swap.
2469 *) Use BN_nnmod instead of BN_mod in crypto/bn/bn_exp.c so that
2470 the exponentiation functions are more likely to produce reasonable
2471 results on negative inputs.
2474 *) Change BN_mod_mul so that the result is always non-negative.
2475 Previously, it could be negative if one of the factors was negative;
2476 I don't think anyone really wanted that behaviour.
2479 *) Move BN_mod_... functions into new file crypto/bn/bn_mod.c
2480 (except for exponentiation, which stays in crypto/bn/bn_exp.c,
2481 and BN_mod_mul_reciprocal, which stays in crypto/bn/bn_recp.c)
2482 and add new functions:
2491 BN_mod_lshift1_quick
2495 These functions always generate non-negative results.
2497 BN_nnmod otherwise is like BN_mod (if BN_mod computes a remainder r
2498 such that |m| < r < 0, BN_nnmod will output rem + |m| instead).
2500 BN_mod_XXX_quick(r, a, [b,] m) generates the same result as
2501 BN_mod_XXX(r, a, [b,] m, ctx), but requires that a [and b]
2502 be reduced modulo m.
2503 [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller]
2506 The following entry accidentily appeared in the CHANGES file
2507 distributed with OpenSSL 0.9.7. The modifications described in
2508 it do *not* apply to OpenSSL 0.9.7.
2510 *) Remove a few calls to bn_wexpand() in BN_sqr() (the one in there
2511 was actually never needed) and in BN_mul(). The removal in BN_mul()
2512 required a small change in bn_mul_part_recursive() and the addition
2513 of the functions bn_cmp_part_words(), bn_sub_part_words() and
2514 bn_add_part_words(), which do the same thing as bn_cmp_words(),
2515 bn_sub_words() and bn_add_words() except they take arrays with
2520 *) In 'openssl passwd', verify passwords read from the terminal
2521 unless the '-salt' option is used (which usually means that
2522 verification would just waste user's time since the resulting
2523 hash is going to be compared with some given password hash)
2524 or the new '-noverify' option is used.
2526 This is an incompatible change, but it does not affect
2527 non-interactive use of 'openssl passwd' (passwords on the command
2528 line, '-stdin' option, '-in ...' option) and thus should not
2532 *) Remove all references to RSAref, since there's no more need for it.
2535 *) Make DSO load along a path given through an environment variable
2536 (SHLIB_PATH) with shl_load().
2539 *) Constify the ENGINE code as a result of BIGNUM constification.
2540 Also constify the RSA code and most things related to it. In a
2541 few places, most notable in the depth of the ASN.1 code, ugly
2542 casts back to non-const were required (to be solved at a later
2546 *) Make it so the openssl application has all engines loaded by default.
2549 *) Constify the BIGNUM routines a little more.
2552 *) Add the following functions:
2554 ENGINE_load_cswift()
2556 ENGINE_load_atalla()
2558 ENGINE_load_builtin_engines()
2560 That way, an application can itself choose if external engines that
2561 are built-in in OpenSSL shall ever be used or not. The benefit is
2562 that applications won't have to be linked with libdl or other dso
2563 libraries unless it's really needed.
2565 Changed 'openssl engine' to load all engines on demand.
2566 Changed the engine header files to avoid the duplication of some
2567 declarations (they differed!).
2570 *) 'openssl engine' can now list capabilities.
2573 *) Better error reporting in 'openssl engine'.
2576 *) Never call load_dh_param(NULL) in s_server.
2579 *) Add engine application. It can currently list engines by name and
2580 identity, and test if they are actually available.
2583 *) Improve RPM specification file by forcing symbolic linking and making
2584 sure the installed documentation is also owned by root.root.
2585 [Damien Miller <djm@mindrot.org>]
2587 *) Give the OpenSSL applications more possibilities to make use of
2588 keys (public as well as private) handled by engines.
2591 *) Add OCSP code that comes from CertCo.
2594 *) Add VMS support for the Rijndael code.
2597 *) Added untested support for Nuron crypto accelerator.
2600 *) Add support for external cryptographic devices. This code was
2601 previously distributed separately as the "engine" branch.
2602 [Geoff Thorpe, Richard Levitte]
2604 *) Rework the filename-translation in the DSO code. It is now possible to
2605 have far greater control over how a "name" is turned into a filename
2606 depending on the operating environment and any oddities about the
2607 different shared library filenames on each system.
2610 *) Support threads on FreeBSD-elf in Configure.
2613 *) Fix for SHA1 assembly problem with MASM: it produces
2614 warnings about corrupt line number information when assembling
2615 with debugging information. This is caused by the overlapping
2617 [Bernd Matthes <mainbug@celocom.de>, Steve Henson]
2620 NCONF_get_number() has no error checking at all. As a replacement,
2621 NCONF_get_number_e() is defined (_e for "error checking") and is
2622 promoted strongly. The old NCONF_get_number is kept around for
2623 binary backward compatibility.
2624 Make it possible for methods to load from something other than a BIO,
2625 by providing a function pointer that is given a name instead of a BIO.
2626 For example, this could be used to load configuration data from an
2630 *) Fix for non blocking accept BIOs. Added new I/O special reason
2631 BIO_RR_ACCEPT to cover this case. Previously use of accept BIOs
2632 with non blocking I/O was not possible because no retry code was
2633 implemented. Also added new SSL code SSL_WANT_ACCEPT to cover
2637 *) Added the beginnings of Rijndael support.
2640 *) Fix for bug in DirectoryString mask setting. Add support for
2641 X509_NAME_print_ex() in 'req' and X509_print_ex() function
2642 to allow certificate printing to more controllable, additional
2643 'certopt' option to 'x509' to allow new printing options to be
2647 *) Clean old EAY MD5 hack from e_os.h.
2650 Changes between 0.9.6j and 0.9.6k [30 Sep 2003]
2652 *) Fix various bugs revealed by running the NISCC test suite:
2654 Stop out of bounds reads in the ASN1 code when presented with
2655 invalid tags (CAN-2003-0543 and CAN-2003-0544).
2657 If verify callback ignores invalid public key errors don't try to check
2658 certificate signature with the NULL public key.
2662 *) Fix various bugs revealed by running the NISCC test suite:
2664 Stop out of bounds reads in the ASN1 code when presented with
2665 invalid tags (CAN-2003-0543 and CAN-2003-0544).
2667 If verify callback ignores invalid public key errors don't try to check
2668 certificate signature with the NULL public key.
2672 *) In ssl3_accept() (ssl/s3_srvr.c) only accept a client certificate
2673 if the server requested one: as stated in TLS 1.0 and SSL 3.0
2677 *) In ssl3_get_client_hello() (ssl/s3_srvr.c), tolerate additional
2678 extra data after the compression methods not only for TLS 1.0
2679 but also for SSL 3.0 (as required by the specification).
2680 [Bodo Moeller; problem pointed out by Matthias Loepfe]
2682 *) Change X509_certificate_type() to mark the key as exported/exportable
2683 when it's 512 *bits* long, not 512 bytes.
2686 Changes between 0.9.6i and 0.9.6j [10 Apr 2003]
2688 *) Countermeasure against the Klima-Pokorny-Rosa extension of
2689 Bleichbacher's attack on PKCS #1 v1.5 padding: treat
2690 a protocol version number mismatch like a decryption error
2691 in ssl3_get_client_key_exchange (ssl/s3_srvr.c).
2694 *) Turn on RSA blinding by default in the default implementation
2695 to avoid a timing attack. Applications that don't want it can call
2696 RSA_blinding_off() or use the new flag RSA_FLAG_NO_BLINDING.
2697 They would be ill-advised to do so in most cases.
2698 [Ben Laurie, Steve Henson, Geoff Thorpe, Bodo Moeller]
2700 *) Change RSA blinding code so that it works when the PRNG is not
2701 seeded (in this case, the secret RSA exponent is abused as
2702 an unpredictable seed -- if it is not unpredictable, there
2703 is no point in blinding anyway). Make RSA blinding thread-safe
2704 by remembering the creator's thread ID in rsa->blinding and
2705 having all other threads use local one-time blinding factors
2706 (this requires more computation than sharing rsa->blinding, but
2707 avoids excessive locking; and if an RSA object is not shared
2708 between threads, blinding will still be very fast).
2711 Changes between 0.9.6h and 0.9.6i [19 Feb 2003]
2713 *) In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked
2714 via timing by performing a MAC computation even if incorrrect
2715 block cipher padding has been found. This is a countermeasure
2716 against active attacks where the attacker has to distinguish
2717 between bad padding and a MAC verification error. (CAN-2003-0078)
2719 [Bodo Moeller; problem pointed out by Brice Canvel (EPFL),
2720 Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and
2721 Martin Vuagnoux (EPFL, Ilion)]
2723 Changes between 0.9.6g and 0.9.6h [5 Dec 2002]
2725 *) New function OPENSSL_cleanse(), which is used to cleanse a section of
2726 memory from it's contents. This is done with a counter that will
2727 place alternating values in each byte. This can be used to solve
2728 two issues: 1) the removal of calls to memset() by highly optimizing
2729 compilers, and 2) cleansing with other values than 0, since those can
2730 be read through on certain media, for example a swap space on disk.
2733 *) Bugfix: client side session caching did not work with external caching,
2734 because the session->cipher setting was not restored when reloading
2735 from the external cache. This problem was masked, when
2736 SSL_OP_NETSCAPE_REUSE_CIPHER_CHANGE_BUG (part of SSL_OP_ALL) was set.
2737 (Found by Steve Haslam <steve@araqnid.ddts.net>.)
2740 *) Fix client_certificate (ssl/s2_clnt.c): The permissible total
2741 length of the REQUEST-CERTIFICATE message is 18 .. 34, not 17 .. 33.
2742 [Zeev Lieber <zeev-l@yahoo.com>]
2744 *) Undo an undocumented change introduced in 0.9.6e which caused
2745 repeated calls to OpenSSL_add_all_ciphers() and
2746 OpenSSL_add_all_digests() to be ignored, even after calling
2750 *) Change the default configuration reader to deal with last line not
2751 being properly terminated.
2754 *) Change X509_NAME_cmp() so it applies the special rules on handling
2755 DN values that are of type PrintableString, as well as RDNs of type
2756 emailAddress where the value has the type ia5String.
2757 [stefank@valicert.com via Richard Levitte]
2759 *) Add a SSL_SESS_CACHE_NO_INTERNAL_STORE flag to take over half
2760 the job SSL_SESS_CACHE_NO_INTERNAL_LOOKUP was inconsistently
2761 doing, define a new flag (SSL_SESS_CACHE_NO_INTERNAL) to be
2762 the bitwise-OR of the two for use by the majority of applications
2763 wanting this behaviour, and update the docs. The documented
2764 behaviour and actual behaviour were inconsistent and had been
2765 changing anyway, so this is more a bug-fix than a behavioural
2767 [Geoff Thorpe, diagnosed by Nadav Har'El]
2769 *) Don't impose a 16-byte length minimum on session IDs in ssl/s3_clnt.c
2770 (the SSL 3.0 and TLS 1.0 specifications allow any length up to 32 bytes).
2773 *) Fix initialization code race conditions in
2774 SSLv23_method(), SSLv23_client_method(), SSLv23_server_method(),
2775 SSLv2_method(), SSLv2_client_method(), SSLv2_server_method(),
2776 SSLv3_method(), SSLv3_client_method(), SSLv3_server_method(),
2777 TLSv1_method(), TLSv1_client_method(), TLSv1_server_method(),
2778 ssl2_get_cipher_by_char(),
2779 ssl3_get_cipher_by_char().
2780 [Patrick McCormick <patrick@tellme.com>, Bodo Moeller]
2782 *) Reorder cleanup sequence in SSL_CTX_free(): only remove the ex_data after
2783 the cached sessions are flushed, as the remove_cb() might use ex_data
2784 contents. Bug found by Sam Varshavchik <mrsam@courier-mta.com>
2785 (see [openssl.org #212]).
2786 [Geoff Thorpe, Lutz Jaenicke]
2788 *) Fix typo in OBJ_txt2obj which incorrectly passed the content
2789 length, instead of the encoding length to d2i_ASN1_OBJECT.
2792 Changes between 0.9.6f and 0.9.6g [9 Aug 2002]
2794 *) [In 0.9.6g-engine release:]
2795 Fix crypto/engine/vendor_defns/cswift.h for WIN32 (use '_stdcall').
2796 [Lynn Gazis <lgazis@rainbow.com>]
2798 Changes between 0.9.6e and 0.9.6f [8 Aug 2002]
2800 *) Fix ASN1 checks. Check for overflow by comparing with LONG_MAX
2801 and get fix the header length calculation.
2802 [Florian Weimer <Weimer@CERT.Uni-Stuttgart.DE>,
2803 Alon Kantor <alonk@checkpoint.com> (and others),
2806 *) Use proper error handling instead of 'assertions' in buffer
2807 overflow checks added in 0.9.6e. This prevents DoS (the
2808 assertions could call abort()).
2809 [Arne Ansper <arne@ats.cyber.ee>, Bodo Moeller]
2811 Changes between 0.9.6d and 0.9.6e [30 Jul 2002]
2813 *) Add various sanity checks to asn1_get_length() to reject
2814 the ASN1 length bytes if they exceed sizeof(long), will appear
2815 negative or the content length exceeds the length of the
2817 [Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>]
2819 *) Fix cipher selection routines: ciphers without encryption had no flags
2820 for the cipher strength set and where therefore not handled correctly
2821 by the selection routines (PR #130).
2824 *) Fix EVP_dsa_sha macro.
2828 SSL_OP_DONT_INSERT_EMPTY_FRAGMENTS
2829 for disabling the SSL 3.0/TLS 1.0 CBC vulnerability countermeasure
2830 that was added in OpenSSL 0.9.6d.
2832 As the countermeasure turned out to be incompatible with some
2833 broken SSL implementations, the new option is part of SSL_OP_ALL.
2834 SSL_OP_ALL is usually employed when compatibility with weird SSL
2835 implementations is desired (e.g. '-bugs' option to 's_client' and
2836 's_server'), so the new option is automatically set in many
2840 *) Changes in security patch:
2842 Changes marked "(CHATS)" were sponsored by the Defense Advanced
2843 Research Projects Agency (DARPA) and Air Force Research Laboratory,
2844 Air Force Materiel Command, USAF, under agreement number
2847 *) Add various sanity checks to asn1_get_length() to reject
2848 the ASN1 length bytes if they exceed sizeof(long), will appear
2849 negative or the content length exceeds the length of the
2850 supplied buffer. (CAN-2002-0659)
2851 [Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>]
2853 *) Assertions for various potential buffer overflows, not known to
2855 [Ben Laurie (CHATS)]
2857 *) Various temporary buffers to hold ASCII versions of integers were
2858 too small for 64 bit platforms. (CAN-2002-0655)
2859 [Matthew Byng-Maddick <mbm@aldigital.co.uk> and Ben Laurie (CHATS)>
2861 *) Remote buffer overflow in SSL3 protocol - an attacker could
2862 supply an oversized session ID to a client. (CAN-2002-0656)
2863 [Ben Laurie (CHATS)]
2865 *) Remote buffer overflow in SSL2 protocol - an attacker could
2866 supply an oversized client master key. (CAN-2002-0656)
2867 [Ben Laurie (CHATS)]
2869 Changes between 0.9.6c and 0.9.6d [9 May 2002]
2871 *) Fix crypto/asn1/a_sign.c so that 'parameters' is omitted (not
2872 encoded as NULL) with id-dsa-with-sha1.
2873 [Nils Larsch <nla@trustcenter.de>; problem pointed out by Bodo Moeller]
2875 *) Check various X509_...() return values in apps/req.c.
2876 [Nils Larsch <nla@trustcenter.de>]
2878 *) Fix BASE64 decode (EVP_DecodeUpdate) for data with CR/LF ended lines:
2879 an end-of-file condition would erronously be flagged, when the CRLF
2880 was just at the end of a processed block. The bug was discovered when
2881 processing data through a buffering memory BIO handing the data to a
2882 BASE64-decoding BIO. Bug fund and patch submitted by Pavel Tsekov
2883 <ptsekov@syntrex.com> and Nedelcho Stanev.
2886 *) Implement a countermeasure against a vulnerability recently found
2887 in CBC ciphersuites in SSL 3.0/TLS 1.0: Send an empty fragment
2888 before application data chunks to avoid the use of known IVs
2889 with data potentially chosen by the attacker.
2892 *) Fix length checks in ssl3_get_client_hello().
2895 *) TLS/SSL library bugfix: use s->s3->in_read_app_data differently
2896 to prevent ssl3_read_internal() from incorrectly assuming that
2897 ssl3_read_bytes() found application data while handshake
2898 processing was enabled when in fact s->s3->in_read_app_data was
2899 merely automatically cleared during the initial handshake.
2900 [Bodo Moeller; problem pointed out by Arne Ansper <arne@ats.cyber.ee>]
2902 *) Fix object definitions for Private and Enterprise: they were not
2903 recognized in their shortname (=lowercase) representation. Extend
2904 obj_dat.pl to issue an error when using undefined keywords instead
2905 of silently ignoring the problem (Svenning Sorensen
2906 <sss@sss.dnsalias.net>).
2909 *) Fix DH_generate_parameters() so that it works for 'non-standard'
2910 generators, i.e. generators other than 2 and 5. (Previously, the
2911 code did not properly initialise the 'add' and 'rem' values to
2912 BN_generate_prime().)
2914 In the new general case, we do not insist that 'generator' is
2915 actually a primitive root: This requirement is rather pointless;
2916 a generator of the order-q subgroup is just as good, if not
2920 *) Map new X509 verification errors to alerts. Discovered and submitted by
2921 Tom Wu <tom@arcot.com>.
2924 *) Fix ssl3_pending() (ssl/s3_lib.c) to prevent SSL_pending() from
2925 returning non-zero before the data has been completely received
2926 when using non-blocking I/O.
2927 [Bodo Moeller; problem pointed out by John Hughes]
2929 *) Some of the ciphers missed the strength entry (SSL_LOW etc).
2930 [Ben Laurie, Lutz Jaenicke]
2932 *) Fix bug in SSL_clear(): bad sessions were not removed (found by
2933 Yoram Zahavi <YoramZ@gilian.com>).
2936 *) Add information about CygWin 1.3 and on, and preserve proper
2937 configuration for the versions before that.
2938 [Corinna Vinschen <vinschen@redhat.com> and Richard Levitte]
2940 *) Make removal from session cache (SSL_CTX_remove_session()) more robust:
2941 check whether we deal with a copy of a session and do not delete from
2942 the cache in this case. Problem reported by "Izhar Shoshani Levi"
2943 <izhar@checkpoint.com>.
2946 *) Do not store session data into the internal session cache, if it
2947 is never intended to be looked up (SSL_SESS_CACHE_NO_INTERNAL_LOOKUP
2948 flag is set). Proposed by Aslam <aslam@funk.com>.
2951 *) Have ASN1_BIT_STRING_set_bit() really clear a bit when the requested
2955 *) [In 0.9.6d-engine release:]
2956 Fix a crashbug and a logic bug in hwcrhk_load_pubkey().
2957 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte]
2959 *) Add the configuration target linux-s390x.
2960 [Neale Ferguson <Neale.Ferguson@SoftwareAG-USA.com> via Richard Levitte]
2962 *) The earlier bugfix for the SSL3_ST_SW_HELLO_REQ_C case of
2963 ssl3_accept (ssl/s3_srvr.c) incorrectly used a local flag
2964 variable as an indication that a ClientHello message has been
2965 received. As the flag value will be lost between multiple
2966 invocations of ssl3_accept when using non-blocking I/O, the
2967 function may not be aware that a handshake has actually taken
2968 place, thus preventing a new session from being added to the
2971 To avoid this problem, we now set s->new_session to 2 instead of
2972 using a local variable.
2973 [Lutz Jaenicke, Bodo Moeller]
2975 *) Bugfix: Return -1 from ssl3_get_server_done (ssl3/s3_clnt.c)
2976 if the SSL_R_LENGTH_MISMATCH error is detected.
2977 [Geoff Thorpe, Bodo Moeller]
2979 *) New 'shared_ldflag' column in Configure platform table.
2982 *) Fix EVP_CIPHER_mode macro.
2983 ["Dan S. Camper" <dan@bti.net>]
2985 *) Fix ssl3_read_bytes (ssl/s3_pkt.c): To ignore messages of unknown
2986 type, we must throw them away by setting rr->length to 0.
2987 [D P Chang <dpc@qualys.com>]
2989 Changes between 0.9.6b and 0.9.6c [21 dec 2001]
2991 *) Fix BN_rand_range bug pointed out by Dominikus Scherkl
2992 <Dominikus.Scherkl@biodata.com>. (The previous implementation
2993 worked incorrectly for those cases where range = 10..._2 and
2994 3*range is two bits longer than range.)
2997 *) Only add signing time to PKCS7 structures if it is not already
3001 *) Fix crypto/objects/objects.h: "ld-ce" should be "id-ce",
3002 OBJ_ld_ce should be OBJ_id_ce.
3003 Also some ip-pda OIDs in crypto/objects/objects.txt were
3004 incorrect (cf. RFC 3039).
3005 [Matt Cooper, Frederic Giudicelli, Bodo Moeller]
3007 *) Release CRYPTO_LOCK_DYNLOCK when CRYPTO_destroy_dynlockid()
3008 returns early because it has nothing to do.
3009 [Andy Schneider <andy.schneider@bjss.co.uk>]
3011 *) [In 0.9.6c-engine release:]
3012 Fix mutex callback return values in crypto/engine/hw_ncipher.c.
3013 [Andy Schneider <andy.schneider@bjss.co.uk>]
3015 *) [In 0.9.6c-engine release:]
3016 Add support for Cryptographic Appliance's keyserver technology.
3017 (Use engine 'keyclient')
3018 [Cryptographic Appliances and Geoff Thorpe]
3020 *) Add a configuration entry for OS/390 Unix. The C compiler 'c89'
3021 is called via tools/c89.sh because arguments have to be
3022 rearranged (all '-L' options must appear before the first object
3024 [Richard Shapiro <rshapiro@abinitio.com>]
3026 *) [In 0.9.6c-engine release:]
3027 Add support for Broadcom crypto accelerator cards, backported
3029 [Broadcom, Nalin Dahyabhai <nalin@redhat.com>, Mark Cox]
3031 *) [In 0.9.6c-engine release:]
3032 Add support for SureWare crypto accelerator cards from
3033 Baltimore Technologies. (Use engine 'sureware')
3034 [Baltimore Technologies and Mark Cox]
3036 *) [In 0.9.6c-engine release:]
3037 Add support for crypto accelerator cards from Accelerated
3038 Encryption Processing, www.aep.ie. (Use engine 'aep')
3039 [AEP Inc. and Mark Cox]
3041 *) Add a configuration entry for gcc on UnixWare.
3042 [Gary Benson <gbenson@redhat.com>]
3044 *) Change ssl/s2_clnt.c and ssl/s2_srvr.c so that received handshake
3045 messages are stored in a single piece (fixed-length part and
3046 variable-length part combined) and fix various bugs found on the way.
3049 *) Disable caching in BIO_gethostbyname(), directly use gethostbyname()
3050 instead. BIO_gethostbyname() does not know what timeouts are
3051 appropriate, so entries would stay in cache even when they have
3053 [Bodo Moeller; problem pointed out by Rich Salz <rsalz@zolera.com>
3055 *) Change ssl23_get_client_hello (ssl/s23_srvr.c) behaviour when
3056 faced with a pathologically small ClientHello fragment that does
3057 not contain client_version: Instead of aborting with an error,
3058 simply choose the highest available protocol version (i.e.,
3059 TLS 1.0 unless it is disabled). In practice, ClientHello
3060 messages are never sent like this, but this change gives us
3061 strictly correct behaviour at least for TLS.
3064 *) Fix SSL handshake functions and SSL_clear() such that SSL_clear()
3065 never resets s->method to s->ctx->method when called from within
3066 one of the SSL handshake functions.
3067 [Bodo Moeller; problem pointed out by Niko Baric]
3069 *) In ssl3_get_client_hello (ssl/s3_srvr.c), generate a fatal alert
3070 (sent using the client's version number) if client_version is
3071 smaller than the protocol version in use. Also change
3072 ssl23_get_client_hello (ssl/s23_srvr.c) to select TLS 1.0 if
3073 the client demanded SSL 3.0 but only TLS 1.0 is enabled; then
3074 the client will at least see that alert.
3077 *) Fix ssl3_get_message (ssl/s3_both.c) to handle message fragmentation
3081 *) Avoid infinite loop in ssl3_get_message (ssl/s3_both.c) if a
3082 client receives HelloRequest while in a handshake.
3083 [Bodo Moeller; bug noticed by Andy Schneider <andy.schneider@bjss.co.uk>]
3085 *) Bugfix in ssl3_accept (ssl/s3_srvr.c): Case SSL3_ST_SW_HELLO_REQ_C
3086 should end in 'break', not 'goto end' which circuments various
3087 cleanups done in state SSL_ST_OK. But session related stuff
3088 must be disabled for SSL_ST_OK in the case that we just sent a
3091 Also avoid some overhead by not calling ssl_init_wbio_buffer()
3092 before just sending a HelloRequest.
3093 [Bodo Moeller, Eric Rescorla <ekr@rtfm.com>]
3095 *) Fix ssl/s3_enc.c, ssl/t1_enc.c and ssl/s3_pkt.c so that we don't
3096 reveal whether illegal block cipher padding was found or a MAC
3097 verification error occured. (Neither SSLerr() codes nor alerts
3098 are directly visible to potential attackers, but the information
3099 may leak via logfiles.)
3101 Similar changes are not required for the SSL 2.0 implementation
3102 because the number of padding bytes is sent in clear for SSL 2.0,
3103 and the extra bytes are just ignored. However ssl/s2_pkt.c
3104 failed to verify that the purported number of padding bytes is in
3108 *) Add OpenUNIX-8 support including shared libraries
3109 (Boyd Lynn Gerber <gerberb@zenez.com>).
3112 *) Improve RSA_padding_check_PKCS1_OAEP() check again to avoid
3113 'wristwatch attack' using huge encoding parameters (cf.
3114 James H. Manger's CRYPTO 2001 paper). Note that the
3115 RSA_PKCS1_OAEP_PADDING case of RSA_private_decrypt() does not use
3116 encoding parameters and hence was not vulnerable.
3119 *) BN_sqr() bug fix.
3120 [Ulf Möller, reported by Jim Ellis <jim.ellis@cavium.com>]
3122 *) Rabin-Miller test analyses assume uniformly distributed witnesses,
3123 so use BN_pseudo_rand_range() instead of using BN_pseudo_rand()
3124 followed by modular reduction.
3125 [Bodo Moeller; pointed out by Adam Young <AYoung1@NCSUS.JNJ.COM>]
3127 *) Add BN_pseudo_rand_range() with obvious functionality: BN_rand_range()
3128 equivalent based on BN_pseudo_rand() instead of BN_rand().
3131 *) s3_srvr.c: allow sending of large client certificate lists (> 16 kB).
3132 This function was broken, as the check for a new client hello message
3133 to handle SGC did not allow these large messages.
3134 (Tracked down by "Douglas E. Engert" <deengert@anl.gov>.)
3137 *) Add alert descriptions for TLSv1 to SSL_alert_desc_string[_long]().
3140 *) Fix buggy behaviour of BIO_get_num_renegotiates() and BIO_ctrl()
3141 for BIO_C_GET_WRITE_BUF_SIZE ("Stephen Hinton" <shinton@netopia.com>).
3144 *) Rework the configuration and shared library support for Tru64 Unix.
3145 The configuration part makes use of modern compiler features and
3146 still retains old compiler behavior for those that run older versions
3147 of the OS. The shared library support part includes a variant that
3148 uses the RPATH feature, and is available through the special
3149 configuration target "alpha-cc-rpath", which will never be selected
3151 [Tim Mooney <mooney@dogbert.cc.ndsu.NoDak.edu> via Richard Levitte]
3153 *) In ssl3_get_key_exchange (ssl/s3_clnt.c), call ssl3_get_message()
3154 with the same message size as in ssl3_get_certificate_request().
3155 Otherwise, if no ServerKeyExchange message occurs, CertificateRequest
3156 messages might inadvertently be reject as too long.
3157 [Petr Lampa <lampa@fee.vutbr.cz>]
3159 *) Enhanced support for IA-64 Unix platforms (well, Linux and HP-UX).
3162 *) Modified SSL library such that the verify_callback that has been set
3163 specificly for an SSL object with SSL_set_verify() is actually being
3164 used. Before the change, a verify_callback set with this function was
3165 ignored and the verify_callback() set in the SSL_CTX at the time of
3166 the call was used. New function X509_STORE_CTX_set_verify_cb() introduced
3167 to allow the necessary settings.
3170 *) Initialize static variable in crypto/dsa/dsa_lib.c and crypto/dh/dh_lib.c
3171 explicitly to NULL, as at least on Solaris 8 this seems not always to be
3172 done automatically (in contradiction to the requirements of the C
3173 standard). This made problems when used from OpenSSH.
3176 *) In OpenSSL 0.9.6a and 0.9.6b, crypto/dh/dh_key.c ignored
3177 dh->length and always used
3179 BN_rand_range(priv_key, dh->p).
3181 BN_rand_range() is not necessary for Diffie-Hellman, and this
3182 specific range makes Diffie-Hellman unnecessarily inefficient if
3183 dh->length (recommended exponent length) is much smaller than the
3184 length of dh->p. We could use BN_rand_range() if the order of
3185 the subgroup was stored in the DH structure, but we only have
3190 BN_rand(priv_key, l, ...)
3192 where 'l' is dh->length if this is defined, or BN_num_bits(dh->p)-1
3198 RSA_eay_public_encrypt
3199 RSA_eay_private_decrypt
3200 RSA_eay_private_encrypt (signing)
3201 RSA_eay_public_decrypt (signature verification)
3203 (default implementations for RSA_public_encrypt,
3204 RSA_private_decrypt, RSA_private_encrypt, RSA_public_decrypt),
3205 always reject numbers >= n.
3208 *) In crypto/rand/md_rand.c, use a new short-time lock CRYPTO_LOCK_RAND2
3209 to synchronize access to 'locking_thread'. This is necessary on
3210 systems where access to 'locking_thread' (an 'unsigned long'
3211 variable) is not atomic.
3214 *) In crypto/rand/md_rand.c, set 'locking_thread' to current thread's ID
3215 *before* setting the 'crypto_lock_rand' flag. The previous code had
3216 a race condition if 0 is a valid thread ID.
3217 [Travis Vitek <vitek@roguewave.com>]
3219 *) Add support for shared libraries under Irix.
3220 [Albert Chin-A-Young <china@thewrittenword.com>]
3222 *) Add configuration option to build on Linux on both big-endian and
3224 [Ralf Baechle <ralf@uni-koblenz.de>]
3226 *) Add the possibility to create shared libraries on HP-UX.
3229 Changes between 0.9.6a and 0.9.6b [9 Jul 2001]
3231 *) Change ssleay_rand_bytes (crypto/rand/md_rand.c)
3232 to avoid a SSLeay/OpenSSL PRNG weakness pointed out by
3233 Markku-Juhani O. Saarinen <markku-juhani.saarinen@nokia.com>:
3234 PRNG state recovery was possible based on the output of
3235 one PRNG request appropriately sized to gain knowledge on
3236 'md' followed by enough consecutive 1-byte PRNG requests
3237 to traverse all of 'state'.
3239 1. When updating 'md_local' (the current thread's copy of 'md')
3240 during PRNG output generation, hash all of the previous
3241 'md_local' value, not just the half used for PRNG output.
3243 2. Make the number of bytes from 'state' included into the hash
3244 independent from the number of PRNG bytes requested.
3246 The first measure alone would be sufficient to avoid
3247 Markku-Juhani's attack. (Actually it had never occurred
3248 to me that the half of 'md_local' used for chaining was the
3249 half from which PRNG output bytes were taken -- I had always
3250 assumed that the secret half would be used.) The second
3251 measure makes sure that additional data from 'state' is never
3252 mixed into 'md_local' in small portions; this heuristically
3253 further strengthens the PRNG.
3256 *) Fix crypto/bn/asm/mips3.s.
3259 *) When only the key is given to "enc", the IV is undefined. Print out
3260 an error message in this case.
3263 *) Handle special case when X509_NAME is empty in X509 printing routines.
3266 *) In dsa_do_verify (crypto/dsa/dsa_ossl.c), verify that r and s are
3267 positive and less than q.
3270 *) Don't change *pointer in CRYPTO_add_lock() is add_lock_callback is
3271 used: it isn't thread safe and the add_lock_callback should handle
3273 [Paul Rose <Paul.Rose@bridge.com>]
3275 *) Verify that incoming data obeys the block size in
3276 ssl3_enc (ssl/s3_enc.c) and tls1_enc (ssl/t1_enc.c).
3280 [Ulf Möller, Bodo Möller]
3282 *) The countermeasure against Bleichbacher's attack on PKCS #1 v1.5
3283 RSA encryption was accidentally removed in s3_srvr.c in OpenSSL 0.9.5
3284 when fixing the server behaviour for backwards-compatible 'client
3285 hello' messages. (Note that the attack is impractical against
3286 SSL 3.0 and TLS 1.0 anyway because length and version checking
3287 means that the probability of guessing a valid ciphertext is
3288 around 2^-40; see section 5 in Bleichenbacher's CRYPTO '98
3291 Before 0.9.5, the countermeasure (hide the error by generating a
3292 random 'decryption result') did not work properly because
3293 ERR_clear_error() was missing, meaning that SSL_get_error() would
3294 detect the supposedly ignored error.
3296 Both problems are now fixed.
3299 *) In crypto/bio/bf_buff.c, increase DEFAULT_BUFFER_SIZE to 4096
3300 (previously it was 1024).
3303 *) Fix for compatibility mode trust settings: ignore trust settings
3304 unless some valid trust or reject settings are present.
3307 *) Fix for blowfish EVP: its a variable length cipher.
3310 *) Fix various bugs related to DSA S/MIME verification. Handle missing
3311 parameters in DSA public key structures and return an error in the
3312 DSA routines if parameters are absent.
3315 *) In versions up to 0.9.6, RAND_file_name() resorted to file ".rnd"
3316 in the current directory if neither $RANDFILE nor $HOME was set.
3317 RAND_file_name() in 0.9.6a returned NULL in this case. This has
3318 caused some confusion to Windows users who haven't defined $HOME.
3319 Thus RAND_file_name() is changed again: e_os.h can define a
3320 DEFAULT_HOME, which will be used if $HOME is not set.
3321 For Windows, we use "C:"; on other platforms, we still require
3322 environment variables.
3324 *) Move 'if (!initialized) RAND_poll()' into regions protected by
3325 CRYPTO_LOCK_RAND. This is not strictly necessary, but avoids
3326 having multiple threads call RAND_poll() concurrently.
3329 *) In crypto/rand/md_rand.c, replace 'add_do_not_lock' flag by a
3330 combination of a flag and a thread ID variable.
3331 Otherwise while one thread is in ssleay_rand_bytes (which sets the
3332 flag), *other* threads can enter ssleay_add_bytes without obeying
3333 the CRYPTO_LOCK_RAND lock (and may even illegally release the lock
3334 that they do not hold after the first thread unsets add_do_not_lock).
3337 *) Change bctest again: '-x' expressions are not available in all
3341 Changes between 0.9.6 and 0.9.6a [5 Apr 2001]
3343 *) Fix a couple of memory leaks in PKCS7_dataDecode()
3344 [Steve Henson, reported by Heyun Zheng <hzheng@atdsprint.com>]
3346 *) Change Configure and Makefiles to provide EXE_EXT, which will contain
3347 the default extension for executables, if any. Also, make the perl
3348 scripts that use symlink() to test if it really exists and use "cp"
3349 if it doesn't. All this made OpenSSL compilable and installable in
3353 *) Fix for asn1_GetSequence() for indefinite length constructed data.
3354 If SEQUENCE is length is indefinite just set c->slen to the total
3355 amount of data available.
3356 [Steve Henson, reported by shige@FreeBSD.org]
3357 [This change does not apply to 0.9.7.]
3359 *) Change bctest to avoid here-documents inside command substitution
3360 (workaround for FreeBSD /bin/sh bug).
3361 For compatibility with Ultrix, avoid shell functions (introduced
3362 in the bctest version that searches along $PATH).
3365 *) Rename 'des_encrypt' to 'des_encrypt1'. This avoids the clashes
3366 with des_encrypt() defined on some operating systems, like Solaris
3370 *) Check the result of RSA-CRT (see D. Boneh, R. DeMillo, R. Lipton:
3371 On the Importance of Eliminating Errors in Cryptographic
3372 Computations, J. Cryptology 14 (2001) 2, 101-119,
3373 http://theory.stanford.edu/~dabo/papers/faults.ps.gz).
3376 *) MIPS assembler BIGNUM division bug fix.
3379 *) Disabled incorrect Alpha assembler code.
3382 *) Fix PKCS#7 decode routines so they correctly update the length
3383 after reading an EOC for the EXPLICIT tag.
3385 [This change does not apply to 0.9.7.]
3387 *) Fix bug in PKCS#12 key generation routines. This was triggered
3388 if a 3DES key was generated with a 0 initial byte. Include
3389 PKCS12_BROKEN_KEYGEN compilation option to retain the old
3390 (but broken) behaviour.
3393 *) Enhance bctest to search for a working bc along $PATH and print
3395 [Tim Rice <tim@multitalents.net> via Richard Levitte]
3397 *) Fix memory leaks in err.c: free err_data string if necessary;
3398 don't write to the wrong index in ERR_set_error_data.
3401 *) Implement ssl23_peek (analogous to ssl23_read), which previously
3405 *) Replace rdtsc with _emit statements for VC++ version 5.
3406 [Jeremy Cooper <jeremy@baymoo.org>]
3408 *) Make it possible to reuse SSLv2 sessions.
3411 *) In copy_email() check for >= 0 as a return value for
3412 X509_NAME_get_index_by_NID() since 0 is a valid index.
3413 [Steve Henson reported by Massimiliano Pala <madwolf@opensca.org>]
3415 *) Avoid coredump with unsupported or invalid public keys by checking if
3416 X509_get_pubkey() fails in PKCS7_verify(). Fix memory leak when
3417 PKCS7_verify() fails with non detached data.
3420 *) Don't use getenv in library functions when run as setuid/setgid.
3421 New function OPENSSL_issetugid().
3424 *) Avoid false positives in memory leak detection code (crypto/mem_dbg.c)
3425 due to incorrect handling of multi-threading:
3427 1. Fix timing glitch in the MemCheck_off() portion of CRYPTO_mem_ctrl().
3429 2. Fix logical glitch in is_MemCheck_on() aka CRYPTO_is_mem_check_on().
3431 3. Count how many times MemCheck_off() has been called so that
3432 nested use can be treated correctly. This also avoids
3433 inband-signalling in the previous code (which relied on the
3434 assumption that thread ID 0 is impossible).
3437 *) Add "-rand" option also to s_client and s_server.
3440 *) Fix CPU detection on Irix 6.x.
3441 [Kurt Hockenbury <khockenb@stevens-tech.edu> and
3442 "Bruce W. Forsberg" <bruce.forsberg@baesystems.com>]
3444 *) Fix X509_NAME bug which produced incorrect encoding if X509_NAME
3447 [This change does not apply to 0.9.7.]
3449 *) Use the cached encoding of an X509_NAME structure rather than
3450 copying it. This is apparently the reason for the libsafe "errors"
3451 but the code is actually correct.
3454 *) Add new function BN_rand_range(), and fix DSA_sign_setup() to prevent
3455 Bleichenbacher's DSA attack.
3456 Extend BN_[pseudo_]rand: As before, top=1 forces the highest two bits
3457 to be set and top=0 forces the highest bit to be set; top=-1 is new
3458 and leaves the highest bit random.
3459 [Ulf Moeller, Bodo Moeller]
3461 *) In the NCONF_...-based implementations for CONF_... queries
3462 (crypto/conf/conf_lib.c), if the input LHASH is NULL, avoid using
3463 a temporary CONF structure with the data component set to NULL
3464 (which gives segmentation faults in lh_retrieve).
3465 Instead, use NULL for the CONF pointer in CONF_get_string and
3466 CONF_get_number (which may use environment variables) and directly
3467 return NULL from CONF_get_section.
3470 *) Fix potential buffer overrun for EBCDIC.
3473 *) Tolerate nonRepudiation as being valid for S/MIME signing and certSign
3474 keyUsage if basicConstraints absent for a CA.
3477 *) Make SMIME_write_PKCS7() write mail header values with a format that
3478 is more generally accepted (no spaces before the semicolon), since
3479 some programs can't parse those values properly otherwise. Also make
3480 sure BIO's that break lines after each write do not create invalid
3484 *) Make the CRL encoding routines work with empty SEQUENCE OF. The
3485 macros previously used would not encode an empty SEQUENCE OF
3486 and break the signature.
3488 [This change does not apply to 0.9.7.]
3490 *) Zero the premaster secret after deriving the master secret in
3494 *) Add some EVP_add_digest_alias registrations (as found in
3495 OpenSSL_add_all_digests()) to SSL_library_init()
3496 aka OpenSSL_add_ssl_algorithms(). This provides improved
3497 compatibility with peers using X.509 certificates
3498 with unconventional AlgorithmIdentifier OIDs.
3501 *) Fix for Irix with NO_ASM.
3502 ["Bruce W. Forsberg" <bruce.forsberg@baesystems.com>]
3504 *) ./config script fixes.
3505 [Ulf Moeller, Richard Levitte]
3507 *) Fix 'openssl passwd -1'.
3510 *) Change PKCS12_key_gen_asc() so it can cope with non null
3511 terminated strings whose length is passed in the passlen
3512 parameter, for example from PEM callbacks. This was done
3513 by adding an extra length parameter to asc2uni().
3514 [Steve Henson, reported by <oddissey@samsung.co.kr>]
3516 *) Fix C code generated by 'openssl dsaparam -C': If a BN_bin2bn
3517 call failed, free the DSA structure.
3520 *) Fix to uni2asc() to cope with zero length Unicode strings.
3521 These are present in some PKCS#12 files.
3524 *) Increase s2->wbuf allocation by one byte in ssl2_new (ssl/s2_lib.c).
3525 Otherwise do_ssl_write (ssl/s2_pkt.c) will write beyond buffer limits
3526 when writing a 32767 byte record.
3527 [Bodo Moeller; problem reported by Eric Day <eday@concentric.net>]
3529 *) In RSA_eay_public_{en,ed}crypt and RSA_eay_mod_exp (rsa_eay.c),
3530 obtain lock CRYPTO_LOCK_RSA before setting rsa->_method_mod_{n,p,q}.
3532 (RSA objects have a reference count access to which is protected
3533 by CRYPTO_LOCK_RSA [see rsa_lib.c, s3_srvr.c, ssl_cert.c, ssl_rsa.c],
3534 so they are meant to be shared between threads.)
3535 [Bodo Moeller, Geoff Thorpe; original patch submitted by
3536 "Reddie, Steven" <Steven.Reddie@ca.com>]
3538 *) Fix a deadlock in CRYPTO_mem_leaks().
3541 *) Use better test patterns in bntest.
3544 *) rand_win.c fix for Borland C.
3547 *) BN_rshift bugfix for n == 0.
3550 *) Add a 'bctest' script that checks for some known 'bc' bugs
3551 so that 'make test' does not abort just because 'bc' is broken.
3554 *) Store verify_result within SSL_SESSION also for client side to
3555 avoid potential security hole. (Re-used sessions on the client side
3556 always resulted in verify_result==X509_V_OK, not using the original
3557 result of the server certificate verification.)
3560 *) Fix ssl3_pending: If the record in s->s3->rrec is not of type
3561 SSL3_RT_APPLICATION_DATA, return 0.
3562 Similarly, change ssl2_pending to return 0 if SSL_in_init(s) is true.
3566 Both ssl2_peek and ssl3_peek, which were totally broken in earlier
3567 releases, have been re-implemented by renaming the previous
3568 implementations of ssl2_read and ssl3_read to ssl2_read_internal
3569 and ssl3_read_internal, respectively, and adding 'peek' parameters
3570 to them. The new ssl[23]_{read,peek} functions are calls to
3571 ssl[23]_read_internal with the 'peek' flag set appropriately.
3572 A 'peek' parameter has also been added to ssl3_read_bytes, which
3573 does the actual work for ssl3_read_internal.
3576 *) Initialise "ex_data" member of RSA/DSA/DH structures prior to calling
3577 the method-specific "init()" handler. Also clean up ex_data after
3578 calling the method-specific "finish()" handler. Previously, this was
3579 happening the other way round.
3582 *) Increase BN_CTX_NUM (the number of BIGNUMs in a BN_CTX) to 16.
3583 The previous value, 12, was not always sufficient for BN_mod_exp().
3586 *) Make sure that shared libraries get the internal name engine with
3587 the full version number and not just 0. This should mark the
3588 shared libraries as not backward compatible. Of course, this should
3589 be changed again when we can guarantee backward binary compatibility.
3592 *) Fix typo in get_cert_by_subject() in by_dir.c
3593 [Jean-Marc Desperrier <jean-marc.desperrier@certplus.com>]
3595 *) Rework the system to generate shared libraries:
3597 - Make note of the expected extension for the shared libraries and
3598 if there is a need for symbolic links from for example libcrypto.so.0
3599 to libcrypto.so.0.9.7. There is extended info in Configure for
3602 - Make as few rebuilds of the shared libraries as possible.
3604 - Still avoid linking the OpenSSL programs with the shared libraries.
3606 - When installing, install the shared libraries separately from the
3610 *) Fix SSL_CTX_set_read_ahead macro to actually use its argument.
3612 Copy SSL_CTX's read_ahead flag to SSL object directly in SSL_new
3613 and not in SSL_clear because the latter is also used by the
3614 accept/connect functions; previously, the settings made by
3615 SSL_set_read_ahead would be lost during the handshake.
3616 [Bodo Moeller; problems reported by Anders Gertz <gertz@epact.se>]
3618 *) Correct util/mkdef.pl to be selective about disabled algorithms.
3619 Previously, it would create entries for disableed algorithms no
3623 *) Added several new manual pages for SSL_* function.
3626 Changes between 0.9.5a and 0.9.6 [24 Sep 2000]
3628 *) In ssl23_get_client_hello, generate an error message when faced
3629 with an initial SSL 3.0/TLS record that is too small to contain the
3630 first two bytes of the ClientHello message, i.e. client_version.
3631 (Note that this is a pathologic case that probably has never happened
3632 in real life.) The previous approach was to use the version number
3633 from the record header as a substitute; but our protocol choice
3634 should not depend on that one because it is not authenticated
3635 by the Finished messages.
3638 *) More robust randomness gathering functions for Windows.
3639 [Jeffrey Altman <jaltman@columbia.edu>]
3641 *) For compatibility reasons if the flag X509_V_FLAG_ISSUER_CHECK is
3642 not set then we don't setup the error code for issuer check errors
3643 to avoid possibly overwriting other errors which the callback does
3644 handle. If an application does set the flag then we assume it knows
3645 what it is doing and can handle the new informational codes
3649 *) Fix for a nasty bug in ASN1_TYPE handling. ASN1_TYPE is used for
3650 a general "ANY" type, as such it should be able to decode anything
3651 including tagged types. However it didn't check the class so it would
3652 wrongly interpret tagged types in the same way as their universal
3653 counterpart and unknown types were just rejected. Changed so that the
3654 tagged and unknown types are handled in the same way as a SEQUENCE:
3655 that is the encoding is stored intact. There is also a new type
3656 "V_ASN1_OTHER" which is used when the class is not universal, in this
3657 case we have no idea what the actual type is so we just lump them all
3661 *) On VMS, stdout may very well lead to a file that is written to
3662 in a record-oriented fashion. That means that every write() will
3663 write a separate record, which will be read separately by the
3664 programs trying to read from it. This can be very confusing.
3666 The solution is to put a BIO filter in the way that will buffer
3667 text until a linefeed is reached, and then write everything a
3668 line at a time, so every record written will be an actual line,
3669 not chunks of lines and not (usually doesn't happen, but I've
3670 seen it once) several lines in one record. BIO_f_linebuffer() is
3673 Currently, it's a VMS-only method, because that's where it has
3674 been tested well enough.
3677 *) Remove 'optimized' squaring variant in BN_mod_mul_montgomery,
3678 it can return incorrect results.
3679 (Note: The buggy variant was not enabled in OpenSSL 0.9.5a,
3680 but it was in 0.9.6-beta[12].)
3683 *) Disable the check for content being present when verifying detached
3684 signatures in pk7_smime.c. Some versions of Netscape (wrongly)
3685 include zero length content when signing messages.
3688 *) New BIO_shutdown_wr macro, which invokes the BIO_C_SHUTDOWN_WR
3689 BIO_ctrl (for BIO pairs).
3692 *) Add DSO method for VMS.
3695 *) Bug fix: Montgomery multiplication could produce results with the
3699 *) Add RPM specification openssl.spec and modify it to build three
3700 packages. The default package contains applications, application
3701 documentation and run-time libraries. The devel package contains
3702 include files, static libraries and function documentation. The
3703 doc package contains the contents of the doc directory. The original
3704 openssl.spec was provided by Damien Miller <djm@mindrot.org>.
3707 *) Add a large number of documentation files for many SSL routines.
3708 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>]
3710 *) Add a configuration entry for Sony News 4.
3711 [NAKAJI Hiroyuki <nakaji@tutrp.tut.ac.jp>]
3713 *) Don't set the two most significant bits to one when generating a
3714 random number < q in the DSA library.
3717 *) New SSL API mode 'SSL_MODE_AUTO_RETRY'. This disables the default
3718 behaviour that SSL_read may result in SSL_ERROR_WANT_READ (even if
3719 the underlying transport is blocking) if a handshake took place.
3720 (The default behaviour is needed by applications such as s_client
3721 and s_server that use select() to determine when to use SSL_read;
3722 but for applications that know in advance when to expect data, it
3723 just makes things more complicated.)
3726 *) Add RAND_egd_bytes(), which gives control over the number of bytes read
3730 *) Add a few more EBCDIC conditionals that make `req' and `x509'
3731 work better on such systems.
3732 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>]
3734 *) Add two demo programs for PKCS12_parse() and PKCS12_create().
3735 Update PKCS12_parse() so it copies the friendlyName and the
3736 keyid to the certificates aux info.
3739 *) Fix bug in PKCS7_verify() which caused an infinite loop
3740 if there was more than one signature.
3741 [Sven Uszpelkat <su@celocom.de>]
3743 *) Major change in util/mkdef.pl to include extra information
3744 about each symbol, as well as presentig variables as well
3745 as functions. This change means that there's n more need
3746 to rebuild the .num files when some algorithms are excluded.
3749 *) Allow the verify time to be set by an application,
3750 rather than always using the current time.
3753 *) Phase 2 verify code reorganisation. The certificate
3754 verify code now looks up an issuer certificate by a
3755 number of criteria: subject name, authority key id
3756 and key usage. It also verifies self signed certificates
3757 by the same criteria. The main comparison function is
3758 X509_check_issued() which performs these checks.
3760 Lot of changes were necessary in order to support this
3761 without completely rewriting the lookup code.
3763 Authority and subject key identifier are now cached.
3765 The LHASH 'certs' is X509_STORE has now been replaced
3766 by a STACK_OF(X509_OBJECT). This is mainly because an
3767 LHASH can't store or retrieve multiple objects with
3768 the same hash value.
3770 As a result various functions (which were all internal
3771 use only) have changed to handle the new X509_STORE
3772 structure. This will break anything that messed round
3773 with X509_STORE internally.
3775 The functions X509_STORE_add_cert() now checks for an
3776 exact match, rather than just subject name.
3778 The X509_STORE API doesn't directly support the retrieval
3779 of multiple certificates matching a given criteria, however
3780 this can be worked round by performing a lookup first
3781 (which will fill the cache with candidate certificates)
3782 and then examining the cache for matches. This is probably
3783 the best we can do without throwing out X509_LOOKUP
3784 entirely (maybe later...).
3786 The X509_VERIFY_CTX structure has been enhanced considerably.
3788 All certificate lookup operations now go via a get_issuer()
3789 callback. Although this currently uses an X509_STORE it
3790 can be replaced by custom lookups. This is a simple way
3791 to bypass the X509_STORE hackery necessary to make this
3792 work and makes it possible to use more efficient techniques
3793 in future. A very simple version which uses a simple
3794 STACK for its trusted certificate store is also provided
3795 using X509_STORE_CTX_trusted_stack().
3797 The verify_cb() and verify() callbacks now have equivalents
3798 in the X509_STORE_CTX structure.
3800 X509_STORE_CTX also has a 'flags' field which can be used
3801 to customise the verify behaviour.
3804 *) Add new PKCS#7 signing option PKCS7_NOSMIMECAP which
3805 excludes S/MIME capabilities.
3808 *) When a certificate request is read in keep a copy of the
3809 original encoding of the signed data and use it when outputing
3810 again. Signatures then use the original encoding rather than
3811 a decoded, encoded version which may cause problems if the
3812 request is improperly encoded.
3815 *) For consistency with other BIO_puts implementations, call
3816 buffer_write(b, ...) directly in buffer_puts instead of calling
3819 In BIO_puts, increment b->num_write as in BIO_write.
3820 [Peter.Sylvester@EdelWeb.fr]
3822 *) Fix BN_mul_word for the case where the word is 0. (We have to use
3823 BN_zero, we may not return a BIGNUM with an array consisting of
3827 *) Avoid calling abort() from within the library when problems are
3828 detected, except if preprocessor symbols have been defined
3829 (such as REF_CHECK, BN_DEBUG etc.).
3832 *) New openssl application 'rsautl'. This utility can be
3833 used for low level RSA operations. DER public key
3834 BIO/fp routines also added.
3837 *) New Configure entry and patches for compiling on QNX 4.
3838 [Andreas Schneider <andreas@ds3.etech.fh-hamburg.de>]
3840 *) A demo state-machine implementation was sponsored by
3841 Nuron (http://www.nuron.com/) and is now available in
3842 demos/state_machine.
3845 *) New options added to the 'dgst' utility for signature
3846 generation and verification.
3849 *) Unrecognized PKCS#7 content types are now handled via a
3850 catch all ASN1_TYPE structure. This allows unsupported
3851 types to be stored as a "blob" and an application can
3852 encode and decode it manually.
3855 *) Fix various signed/unsigned issues to make a_strex.c
3857 [Oscar Jacobsson <oscar.jacobsson@celocom.com>]
3859 *) ASN1 fixes. i2d_ASN1_OBJECT was not returning the correct
3860 length if passed a buffer. ASN1_INTEGER_to_BN failed
3861 if passed a NULL BN and its argument was negative.
3862 [Steve Henson, pointed out by Sven Heiberg <sven@tartu.cyber.ee>]
3864 *) Modification to PKCS#7 encoding routines to output definite
3865 length encoding. Since currently the whole structures are in
3866 memory there's not real point in using indefinite length
3867 constructed encoding. However if OpenSSL is compiled with
3868 the flag PKCS7_INDEFINITE_ENCODING the old form is used.
3871 *) Added BIO_vprintf() and BIO_vsnprintf().
3874 *) Added more prefixes to parse for in the the strings written
3875 through a logging bio, to cover all the levels that are available
3876 through syslog. The prefixes are now:
3878 PANIC, EMERG, EMR => LOG_EMERG
3879 ALERT, ALR => LOG_ALERT
3880 CRIT, CRI => LOG_CRIT
3881 ERROR, ERR => LOG_ERR
3882 WARNING, WARN, WAR => LOG_WARNING
3883 NOTICE, NOTE, NOT => LOG_NOTICE
3884 INFO, INF => LOG_INFO
3885 DEBUG, DBG => LOG_DEBUG
3887 and as before, if none of those prefixes are present at the
3888 beginning of the string, LOG_ERR is chosen.
3890 On Win32, the LOG_* levels are mapped according to this:
3892 LOG_EMERG, LOG_ALERT, LOG_CRIT, LOG_ERR => EVENTLOG_ERROR_TYPE
3893 LOG_WARNING => EVENTLOG_WARNING_TYPE
3894 LOG_NOTICE, LOG_INFO, LOG_DEBUG => EVENTLOG_INFORMATION_TYPE
3898 *) Made it possible to reconfigure with just the configuration
3899 argument "reconf" or "reconfigure". The command line arguments
3900 are stored in Makefile.ssl in the variable CONFIGURE_ARGS,
3901 and are retrieved from there when reconfiguring.
3905 [Assar Westerlund <assar@sics.se>, Richard Levitte]
3907 *) Add the arguments -CAfile and -CApath to the pkcs12 utility.
3910 *) The obj_dat.pl script was messing up the sorting of object
3911 names. The reason was that it compared the quoted version
3912 of strings as a result "OCSP" > "OCSP Signing" because
3913 " > SPACE. Changed script to store unquoted versions of
3914 names and add quotes on output. It was also omitting some
3915 names from the lookup table if they were given a default
3916 value (that is if SN is missing it is given the same
3917 value as LN and vice versa), these are now added on the
3918 grounds that if an object has a name we should be able to
3919 look it up. Finally added warning output when duplicate
3920 short or long names are found.
3923 *) Changes needed for Tandem NSK.
3924 [Scott Uroff <scott@xypro.com>]
3926 *) Fix SSL 2.0 rollback checking: Due to an off-by-one error in
3927 RSA_padding_check_SSLv23(), special padding was never detected
3928 and thus the SSL 3.0/TLS 1.0 countermeasure against protocol
3929 version rollback attacks was not effective.
3931 In s23_clnt.c, don't use special rollback-attack detection padding
3932 (RSA_SSLV23_PADDING) if SSL 2.0 is the only protocol enabled in the
3933 client; similarly, in s23_srvr.c, don't do the rollback check if
3934 SSL 2.0 is the only protocol enabled in the server.
3937 *) Make it possible to get hexdumps of unprintable data with 'openssl
3938 asn1parse'. By implication, the functions ASN1_parse_dump() and
3939 BIO_dump_indent() are added.
3942 *) New functions ASN1_STRING_print_ex() and X509_NAME_print_ex()
3943 these print out strings and name structures based on various
3944 flags including RFC2253 support and proper handling of
3945 multibyte characters. Added options to the 'x509' utility
3946 to allow the various flags to be set.
3949 *) Various fixes to use ASN1_TIME instead of ASN1_UTCTIME.
3950 Also change the functions X509_cmp_current_time() and
3951 X509_gmtime_adj() work with an ASN1_TIME structure,
3952 this will enable certificates using GeneralizedTime in validity
3953 dates to be checked.
3956 *) Make the NEG_PUBKEY_BUG code (which tolerates invalid
3957 negative public key encodings) on by default,
3958 NO_NEG_PUBKEY_BUG can be set to disable it.
3961 *) New function c2i_ASN1_OBJECT() which acts on ASN1_OBJECT
3962 content octets. An i2c_ASN1_OBJECT is unnecessary because
3963 the encoding can be trivially obtained from the structure.
3966 *) crypto/err.c locking bugfix: Use write locks (CRYPTO_w_[un]lock),
3967 not read locks (CRYPTO_r_[un]lock).
3970 *) A first attempt at creating official support for shared
3971 libraries through configuration. I've kept it so the
3972 default is static libraries only, and the OpenSSL programs
3973 are always statically linked for now, but there are
3974 preparations for dynamic linking in place.
3975 This has been tested on Linux and Tru64.
3978 *) Randomness polling function for Win9x, as described in:
3979 Peter Gutmann, Software Generation of Practically Strong
3983 *) Fix so PRNG is seeded in req if using an already existing
3987 *) New options to smime application. -inform and -outform
3988 allow alternative formats for the S/MIME message including
3989 PEM and DER. The -content option allows the content to be
3990 specified separately. This should allow things like Netscape
3991 form signing output easier to verify.
3994 *) Fix the ASN1 encoding of tags using the 'long form'.
3997 *) New ASN1 functions, i2c_* and c2i_* for INTEGER and BIT
3998 STRING types. These convert content octets to and from the
3999 underlying type. The actual tag and length octets are
4000 already assumed to have been read in and checked. These
4001 are needed because all other string types have virtually
4002 identical handling apart from the tag. By having versions
4003 of the ASN1 functions that just operate on content octets
4004 IMPLICIT tagging can be handled properly. It also allows
4005 the ASN1_ENUMERATED code to be cut down because ASN1_ENUMERATED
4006 and ASN1_INTEGER are identical apart from the tag.
4009 *) Change the handling of OID objects as follows:
4011 - New object identifiers are inserted in objects.txt, following
4012 the syntax given in objects.README.
4013 - objects.pl is used to process obj_mac.num and create a new
4015 - obj_dat.pl is used to create a new obj_dat.h, using the data in
4018 This is currently kind of a hack, and the perl code in objects.pl
4019 isn't very elegant, but it works as I intended. The simplest way
4020 to check that it worked correctly is to look in obj_dat.h and
4021 check the array nid_objs and make sure the objects haven't moved
4022 around (this is important!). Additions are OK, as well as
4023 consistent name changes.
4026 *) Add BSD-style MD5-based passwords to 'openssl passwd' (option '-1').
4029 *) Addition of the command line parameter '-rand file' to 'openssl req'.
4030 The given file adds to whatever has already been seeded into the
4031 random pool through the RANDFILE configuration file option or
4032 environment variable, or the default random state file.
4035 *) mkstack.pl now sorts each macro group into lexical order.
4036 Previously the output order depended on the order the files
4037 appeared in the directory, resulting in needless rewriting
4041 *) Patches to make OpenSSL compile under Win32 again. Mostly
4042 work arounds for the VC++ problem that it treats func() as
4043 func(void). Also stripped out the parts of mkdef.pl that
4044 added extra typesafe functions: these no longer exist.
4047 *) Reorganisation of the stack code. The macros are now all
4048 collected in safestack.h . Each macro is defined in terms of
4049 a "stack macro" of the form SKM_<name>(type, a, b). The
4050 DEBUG_SAFESTACK is now handled in terms of function casts,
4051 this has the advantage of retaining type safety without the
4052 use of additional functions. If DEBUG_SAFESTACK is not defined
4053 then the non typesafe macros are used instead. Also modified the
4054 mkstack.pl script to handle the new form. Needs testing to see
4055 if which (if any) compilers it chokes and maybe make DEBUG_SAFESTACK
4056 the default if no major problems. Similar behaviour for ASN1_SET_OF
4057 and PKCS12_STACK_OF.
4060 *) When some versions of IIS use the 'NET' form of private key the
4061 key derivation algorithm is different. Normally MD5(password) is
4062 used as a 128 bit RC4 key. In the modified case
4063 MD5(MD5(password) + "SGCKEYSALT") is used insted. Added some
4064 new functions i2d_RSA_NET(), d2i_RSA_NET() etc which are the same
4065 as the old Netscape_RSA functions except they have an additional
4066 'sgckey' parameter which uses the modified algorithm. Also added
4067 an -sgckey command line option to the rsa utility. Thanks to
4068 Adrian Peck <bertie@ncipher.com> for posting details of the modified
4069 algorithm to openssl-dev.
4072 *) The evp_local.h macros were using 'c.##kname' which resulted in
4073 invalid expansion on some systems (SCO 5.0.5 for example).
4074 Corrected to 'c.kname'.
4075 [Phillip Porch <root@theporch.com>]
4077 *) New X509_get1_email() and X509_REQ_get1_email() functions that return
4078 a STACK of email addresses from a certificate or request, these look
4079 in the subject name and the subject alternative name extensions and
4080 omit any duplicate addresses.
4083 *) Re-implement BN_mod_exp2_mont using independent (and larger) windows.
4084 This makes DSA verification about 2 % faster.
4087 *) Increase maximum window size in BN_mod_exp_... to 6 bits instead of 5
4088 (meaning that now 2^5 values will be precomputed, which is only 4 KB
4089 plus overhead for 1024 bit moduli).
4090 This makes exponentiations about 0.5 % faster for 1024 bit
4091 exponents (as measured by "openssl speed rsa2048").
4094 *) Rename memory handling macros to avoid conflicts with other
4096 Malloc => OPENSSL_malloc
4097 Malloc_locked => OPENSSL_malloc_locked
4098 Realloc => OPENSSL_realloc
4099 Free => OPENSSL_free
4102 *) New function BN_mod_exp_mont_word for small bases (roughly 15%
4103 faster than BN_mod_exp_mont, i.e. 7% for a full DH exchange).
4106 *) CygWin32 support.
4107 [John Jarvie <jjarvie@newsguy.com>]
4109 *) The type-safe stack code has been rejigged. It is now only compiled
4110 in when OpenSSL is configured with the DEBUG_SAFESTACK option and
4111 by default all type-specific stack functions are "#define"d back to
4112 standard stack functions. This results in more streamlined output
4113 but retains the type-safety checking possibilities of the original
4117 *) The STACK code has been cleaned up, and certain type declarations
4118 that didn't make a lot of sense have been brought in line. This has
4119 also involved a cleanup of sorts in safestack.h to more correctly
4120 map type-safe stack functions onto their plain stack counterparts.
4121 This work has also resulted in a variety of "const"ifications of
4122 lots of the code, especially "_cmp" operations which should normally
4123 be prototyped with "const" parameters anyway.
4126 *) When generating bytes for the first time in md_rand.c, 'stir the pool'
4127 by seeding with STATE_SIZE dummy bytes (with zero entropy count).
4128 (The PRNG state consists of two parts, the large pool 'state' and 'md',
4129 where all of 'md' is used each time the PRNG is used, but 'state'
4130 is used only indexed by a cyclic counter. As entropy may not be
4131 well distributed from the beginning, 'md' is important as a
4132 chaining variable. However, the output function chains only half
4133 of 'md', i.e. 80 bits. ssleay_rand_add, on the other hand, chains
4134 all of 'md', and seeding with STATE_SIZE dummy bytes will result
4135 in all of 'state' being rewritten, with the new values depending
4136 on virtually all of 'md'. This overcomes the 80 bit limitation.)
4139 *) In ssl/s2_clnt.c and ssl/s3_clnt.c, call ERR_clear_error() when
4140 the handshake is continued after ssl_verify_cert_chain();
4141 otherwise, if SSL_VERIFY_NONE is set, remaining error codes
4142 can lead to 'unexplainable' connection aborts later.
4143 [Bodo Moeller; problem tracked down by Lutz Jaenicke]
4145 *) Major EVP API cipher revision.
4146 Add hooks for extra EVP features. This allows various cipher
4147 parameters to be set in the EVP interface. Support added for variable
4148 key length ciphers via the EVP_CIPHER_CTX_set_key_length() function and
4149 setting of RC2 and RC5 parameters.
4151 Modify EVP_OpenInit() and EVP_SealInit() to cope with variable key length
4154 Remove lots of duplicated code from the EVP library. For example *every*
4155 cipher init() function handles the 'iv' in the same way according to the
4156 cipher mode. They also all do nothing if the 'key' parameter is NULL and
4157 for CFB and OFB modes they zero ctx->num.
4159 New functionality allows removal of S/MIME code RC2 hack.
4161 Most of the routines have the same form and so can be declared in terms
4164 By shifting this to the top level EVP_CipherInit() it can be removed from
4165 all individual ciphers. If the cipher wants to handle IVs or keys
4166 differently it can set the EVP_CIPH_CUSTOM_IV or EVP_CIPH_ALWAYS_CALL_INIT
4169 Change lots of functions like EVP_EncryptUpdate() to now return a
4170 value: although software versions of the algorithms cannot fail
4171 any installed hardware versions can.
4174 *) Implement SSL_OP_TLS_ROLLBACK_BUG: In ssl3_get_client_key_exchange, if
4175 this option is set, tolerate broken clients that send the negotiated
4176 protocol version number instead of the requested protocol version
4180 *) Call dh_tmp_cb (set by ..._TMP_DH_CB) with correct 'is_export' flag;
4181 i.e. non-zero for export ciphersuites, zero otherwise.
4182 Previous versions had this flag inverted, inconsistent with
4183 rsa_tmp_cb (..._TMP_RSA_CB).
4184 [Bodo Moeller; problem reported by Amit Chopra]
4186 *) Add missing DSA library text string. Work around for some IIS
4187 key files with invalid SEQUENCE encoding.
4190 *) Add a document (doc/standards.txt) that list all kinds of standards
4191 and so on that are implemented in OpenSSL.
4194 *) Enhance c_rehash script. Old version would mishandle certificates
4195 with the same subject name hash and wouldn't handle CRLs at all.
4196 Added -fingerprint option to crl utility, to support new c_rehash
4200 *) Eliminate non-ANSI declarations in crypto.h and stack.h.
4203 *) Fix for SSL server purpose checking. Server checking was
4204 rejecting certificates which had extended key usage present
4205 but no ssl client purpose.
4206 [Steve Henson, reported by Rene Grosser <grosser@hisolutions.com>]
4208 *) Make PKCS#12 code work with no password. The PKCS#12 spec
4209 is a little unclear about how a blank password is handled.
4210 Since the password in encoded as a BMPString with terminating
4211 double NULL a zero length password would end up as just the
4212 double NULL. However no password at all is different and is
4213 handled differently in the PKCS#12 key generation code. NS
4214 treats a blank password as zero length. MSIE treats it as no
4215 password on export: but it will try both on import. We now do
4216 the same: PKCS12_parse() tries zero length and no password if
4217 the password is set to "" or NULL (NULL is now a valid password:
4218 it wasn't before) as does the pkcs12 application.
4221 *) Bugfixes in apps/x509.c: Avoid a memory leak; and don't use
4222 perror when PEM_read_bio_X509_REQ fails, the error message must
4223 be obtained from the error queue.
4226 *) Avoid 'thread_hash' memory leak in crypto/err/err.c by freeing
4227 it in ERR_remove_state if appropriate, and change ERR_get_state
4228 accordingly to avoid race conditions (this is necessary because
4229 thread_hash is no longer constant once set).
4232 *) Bugfix for linux-elf makefile.one.
4235 *) RSA_get_default_method() will now cause a default
4236 RSA_METHOD to be chosen if one doesn't exist already.
4237 Previously this was only set during a call to RSA_new()
4238 or RSA_new_method(NULL) meaning it was possible for
4239 RSA_get_default_method() to return NULL.
4242 *) Added native name tra
projects / openssl.git / blob