5 Changes between 0.9.7e and 0.9.8 [xx XXX xxxx]
7 *) The new 'RSA_generate_key_ex' function now takes a BIGNUM for the public
8 exponent rather than 'unsigned long'. There is a corresponding change to
9 the new 'rsa_keygen' element of the RSA_METHOD structure.
10 [Jelte Jansen, Geoff Thorpe]
12 *) Functionality for creating the initial serial number file is now
13 moved from CA.pl to the 'ca' utility with a new option -create_serial.
15 (Before OpenSSL 0.9.7e, CA.pl used to initialize the serial
16 number file to 1, which is bound to cause problems. To avoid
17 the problems while respecting compatibility between different 0.9.7
18 patchlevels, 0.9.7e employed 'openssl x509 -next_serial' in
19 CA.pl for serial number initialization. With the new release 0.9.8,
20 we can fix the problem directly in the 'ca' utility.)
23 *) Reduced header interdepencies by declaring more opaque objects in
24 ossl_typ.h. As a consequence, including some headers (eg. engine.h) will
25 give fewer recursive includes, which could break lazy source code - so
26 this change is covered by the OPENSSL_NO_DEPRECATED symbol. As always,
27 developers should define this symbol when building and using openssl to
28 ensure they track the recommended behaviour, interfaces, [etc], but
29 backwards-compatible behaviour prevails when this isn't defined.
32 *) New function X509_POLICY_NODE_print() which prints out policy nodes.
35 *) Add new EVP function EVP_CIPHER_CTX_rand_key and associated functionality.
36 This will generate a random key of the appropriate length based on the
37 cipher context. The EVP_CIPHER can provide its own random key generation
38 routine to support keys of a specific form. This is used in the des and
39 3des routines to generate a key of the correct parity. Update S/MIME
40 code to use new functions and hence generate correct parity DES keys.
41 Add EVP_CHECK_DES_KEY #define to return an error if the key is not
42 valid (weak or incorrect parity).
45 *) Add a local set of CRLs that can be used by X509_verify_cert() as well
46 as looking them up. This is useful when the verified structure may contain
47 CRLs, for example PKCS#7 signedData. Modify PKCS7_verify() to use any CRLs
48 present unless the new PKCS7_NO_CRL flag is asserted.
51 *) Extend ASN1 oid configuration module. It now additionally accepts the
54 shortName = some long name, 1.2.3.4
57 *) Reimplemented the BN_CTX implementation. There is now no more static
58 limitation on the number of variables it can handle nor the depth of the
59 "stack" handling for BN_CTX_start()/BN_CTX_end() pairs. The stack
60 information can now expand as required, and rather than having a single
61 static array of bignums, BN_CTX now uses a linked-list of such arrays
62 allowing it to expand on demand whilst maintaining the usefulness of
66 *) Add a missing BN_CTX parameter to the 'rsa_mod_exp' callback in RSA_METHOD
67 to allow all RSA operations to function using a single BN_CTX.
70 *) Preliminary support for certificate policy evaluation and checking. This
71 is initially intended to pass the tests outlined in "Conformance Testing
72 of Relying Party Client Certificate Path Processing Logic" v1.07.
75 *) bn_dup_expand() has been deprecated, it was introduced in 0.9.7 and
76 remained unused and not that useful. A variety of other little bignum
77 tweaks and fixes have also been made continuing on from the audit (see
81 *) Constify all or almost all d2i, c2i, s2i and r2i functions, along with
82 associated ASN1, EVP and SSL functions and old ASN1 macros.
85 *) BN_zero() only needs to set 'top' and 'neg' to zero for correct results,
86 and this should never fail. So the return value from the use of
87 BN_set_word() (which can fail due to needless expansion) is now deprecated;
88 if OPENSSL_NO_DEPRECATED is defined, BN_zero() is a void macro.
91 *) BN_CTX_get() should return zero-valued bignums, providing the same
92 initialised value as BN_new().
93 [Geoff Thorpe, suggested by Ulf Möller]
95 *) Support for inhibitAnyPolicy certificate extension.
98 *) An audit of the BIGNUM code is underway, for which debugging code is
99 enabled when BN_DEBUG is defined. This makes stricter enforcements on what
100 is considered valid when processing BIGNUMs, and causes execution to
101 assert() when a problem is discovered. If BN_DEBUG_RAND is defined,
102 further steps are taken to deliberately pollute unused data in BIGNUM
103 structures to try and expose faulty code further on. For now, openssl will
104 (in its default mode of operation) continue to tolerate the inconsistent
105 forms that it has tolerated in the past, but authors and packagers should
106 consider trying openssl and their own applications when compiled with
107 these debugging symbols defined. It will help highlight potential bugs in
108 their own code, and will improve the test coverage for OpenSSL itself. At
109 some point, these tighter rules will become openssl's default to improve
110 maintainability, though the assert()s and other overheads will remain only
111 in debugging configurations. See bn.h for more details.
112 [Geoff Thorpe, Nils Larsch, Ulf Möller]
114 *) BN_CTX_init() has been deprecated, as BN_CTX is an opaque structure
115 that can only be obtained through BN_CTX_new() (which implicitly
116 initialises it). The presence of this function only made it possible
117 to overwrite an existing structure (and cause memory leaks).
120 *) Because of the callback-based approach for implementing LHASH as a
121 template type, lh_insert() adds opaque objects to hash-tables and
122 lh_doall() or lh_doall_arg() are typically used with a destructor callback
123 to clean up those corresponding objects before destroying the hash table
124 (and losing the object pointers). So some over-zealous constifications in
125 LHASH have been relaxed so that lh_insert() does not take (nor store) the
126 objects as "const" and the lh_doall[_arg] callback wrappers are not
127 prototyped to have "const" restrictions on the object pointers they are
128 given (and so aren't required to cast them away any more).
131 *) The tmdiff.h API was so ugly and minimal that our own timing utility
132 (speed) prefers to use its own implementation. The two implementations
133 haven't been consolidated as yet (volunteers?) but the tmdiff API has had
134 its object type properly exposed (MS_TM) instead of casting to/from "char
135 *". This may still change yet if someone realises MS_TM and "ms_time_***"
136 aren't necessarily the greatest nomenclatures - but this is what was used
137 internally to the implementation so I've used that for now.
140 *) Ensure that deprecated functions do not get compiled when
141 OPENSSL_NO_DEPRECATED is defined. Some "openssl" subcommands and a few of
142 the self-tests were still using deprecated key-generation functions so
143 these have been updated also.
146 *) Reorganise PKCS#7 code to separate the digest location functionality
147 into PKCS7_find_digest(), digest addtion into PKCS7_bio_add_digest().
148 New function PKCS7_set_digest() to set the digest type for PKCS#7
149 digestedData type. Add additional code to correctly generate the
150 digestedData type and add support for this type in PKCS7 initialization
154 *) New function PKCS7_set0_type_other() this initializes a PKCS7
155 structure of type "other".
158 *) Fix prime generation loop in crypto/bn/bn_prime.pl by making
159 sure the loop does correctly stop and breaking ("division by zero")
160 modulus operations are not performed. The (pre-generated) prime
161 table crypto/bn/bn_prime.h was already correct, but it could not be
162 re-generated on some platforms because of the "division by zero"
163 situation in the script.
164 [Ralf S. Engelschall]
166 *) Update support for ECC-based TLS ciphersuites according to
167 draft-ietf-tls-ecc-03.txt: the KDF1 key derivation function with
168 SHA-1 now is only used for "small" curves (where the
169 representation of a field element takes up to 24 bytes); for
170 larger curves, the field element resulting from ECDH is directly
171 used as premaster secret.
172 [Douglas Stebila (Sun Microsystems Laboratories)]
174 *) Add code for kP+lQ timings to crypto/ec/ectest.c, and add SEC2
175 curve secp160r1 to the tests.
176 [Douglas Stebila (Sun Microsystems Laboratories)]
178 *) Add the possibility to load symbols globally with DSO.
179 [Götz Babin-Ebell <babin-ebell@trustcenter.de> via Richard Levitte]
181 *) Add the functions ERR_set_mark() and ERR_pop_to_mark() for better
182 control of the error stack.
185 *) Add support for STORE in ENGINE.
188 *) Add the STORE type. The intention is to provide a common interface
189 to certificate and key stores, be they simple file-based stores, or
190 HSM-type store, or LDAP stores, or...
191 NOTE: The code is currently UNTESTED and isn't really used anywhere.
194 *) Add a generic structure called OPENSSL_ITEM. This can be used to
195 pass a list of arguments to any function as well as provide a way
196 for a function to pass data back to the caller.
199 *) Add the functions BUF_strndup() and BUF_memdup(). BUF_strndup()
200 works like BUF_strdup() but can be used to duplicate a portion of
201 a string. The copy gets NUL-terminated. BUF_memdup() duplicates
205 *) Add the function sk_find_ex() which works like sk_find(), but will
206 return an index to an element even if an exact match couldn't be
207 found. The index is guaranteed to point at the element where the
208 searched-for key would be inserted to preserve sorting order.
211 *) Add the function OBJ_bsearch_ex() which works like OBJ_bsearch() but
212 takes an extra flags argument for optional functionality. Currently,
213 the following flags are defined:
215 OBJ_BSEARCH_VALUE_ON_NOMATCH
216 This one gets OBJ_bsearch_ex() to return a pointer to the first
217 element where the comparing function returns a negative or zero
220 OBJ_BSEARCH_FIRST_VALUE_ON_MATCH
221 This one gets OBJ_bsearch_ex() to return a pointer to the first
222 element where the comparing function returns zero. This is useful
223 if there are more than one element where the comparing function
227 *) Make it possible to create self-signed certificates with 'openssl ca'
228 in such a way that the self-signed certificate becomes part of the
229 CA database and uses the same mechanisms for serial number generation
230 as all other certificate signing. The new flag '-selfsign' enables
231 this functionality. Adapt CA.sh and CA.pl.in.
234 *) Add functionality to check the public key of a certificate request
235 against a given private. This is useful to check that a certificate
236 request can be signed by that key (self-signing).
239 *) Make it possible to have multiple active certificates with the same
240 subject in the CA index file. This is done only if the keyword
241 'unique_subject' is set to 'no' in the main CA section (default
242 if 'CA_default') of the configuration file. The value is saved
243 with the database itself in a separate index attribute file,
244 named like the index file with '.attr' appended to the name.
247 *) Generate muti valued AVAs using '+' notation in config files for
251 *) Support for nameConstraints certificate extension.
254 *) Support for policyConstraints certificate extension.
257 *) Support for policyMappings certificate extension.
260 *) Make sure the default DSA_METHOD implementation only uses its
261 dsa_mod_exp() and/or bn_mod_exp() handlers if they are non-NULL,
262 and change its own handlers to be NULL so as to remove unnecessary
263 indirection. This lets alternative implementations fallback to the
264 default implementation more easily.
267 *) Support for directoryName in GeneralName related extensions
271 *) Make it possible to link applications using Makefile.shared.
272 Make that possible even when linking against static libraries!
275 *) Support for single pass processing for S/MIME signing. This now
276 means that S/MIME signing can be done from a pipe, in addition
277 cleartext signing (multipart/signed type) is effectively streaming
278 and the signed data does not need to be all held in memory.
280 This is done with a new flag PKCS7_STREAM. When this flag is set
281 PKCS7_sign() only initializes the PKCS7 structure and the actual signing
282 is done after the data is output (and digests calculated) in
286 *) Add full support for -rpath/-R, both in shared libraries and
287 applications, at least on the platforms where it's known how
291 *) In crypto/ec/ec_mult.c, implement fast point multiplication with
292 precomputation, based on wNAF splitting: EC_GROUP_precompute_mult()
293 will now compute a table of multiples of the generator that
294 makes subsequent invocations of EC_POINTs_mul() or EC_POINT_mul()
295 faster (notably in the case of a single point multiplication,
297 [Nils Larsch, Bodo Moeller]
299 *) IPv6 support for certificate extensions. The various extensions
300 which use the IP:a.b.c.d can now take IPv6 addresses using the
301 formats of RFC1884 2.2 . IPv6 addresses are now also displayed
305 *) Added an ENGINE that implements RSA by performing private key
306 exponentiations with the GMP library. The conversions to and from
307 GMP's mpz_t format aren't optimised nor are any montgomery forms
308 cached, and on x86 it appears OpenSSL's own performance has caught up.
309 However there are likely to be other architectures where GMP could
310 provide a boost. This ENGINE is not built in by default, but it can be
311 specified at Configure time and should be accompanied by the necessary
312 linker additions, eg;
313 ./config -DOPENSSL_USE_GMP -lgmp
316 *) "openssl engine" will not display ENGINE/DSO load failure errors when
317 testing availability of engines with "-t" - the old behaviour is
318 produced by increasing the feature's verbosity with "-tt".
321 *) ECDSA routines: under certain error conditions uninitialized BN objects
322 could be freed. Solution: make sure initialization is performed early
323 enough. (Reported and fix supplied by Nils Larsch <nla@trustcenter.de>
327 *) Key-generation can now be implemented in RSA_METHOD, DSA_METHOD
328 and DH_METHOD (eg. by ENGINE implementations) to override the normal
329 software implementations. For DSA and DH, parameter generation can
330 also be overriden by providing the appropriate method callbacks.
333 *) Change the "progress" mechanism used in key-generation and
334 primality testing to functions that take a new BN_GENCB pointer in
335 place of callback/argument pairs. The new API functions have "_ex"
336 postfixes and the older functions are reimplemented as wrappers for
337 the new ones. The OPENSSL_NO_DEPRECATED symbol can be used to hide
338 declarations of the old functions to help (graceful) attempts to
339 migrate to the new functions. Also, the new key-generation API
340 functions operate on a caller-supplied key-structure and return
341 success/failure rather than returning a key or NULL - this is to
342 help make "keygen" another member function of RSA_METHOD etc.
344 Example for using the new callback interface:
346 int (*my_callback)(int a, int b, BN_GENCB *cb) = ...;
350 BN_GENCB_set(&my_cb, my_callback, my_arg);
352 return BN_is_prime_ex(some_bignum, BN_prime_checks, NULL, &cb);
353 /* For the meaning of a, b in calls to my_callback(), see the
354 * documentation of the function that calls the callback.
355 * cb will point to my_cb; my_arg can be retrieved as cb->arg.
356 * my_callback should return 1 if it wants BN_is_prime_ex()
357 * to continue, or 0 to stop.
362 *) Change the ZLIB compression method to be stateful, and make it
363 available to TLS with the number defined in
364 draft-ietf-tls-compression-04.txt.
367 *) Add the ASN.1 structures and functions for CertificatePair, which
368 is defined as follows (according to X.509_4thEditionDraftV6.pdf):
370 CertificatePair ::= SEQUENCE {
371 forward [0] Certificate OPTIONAL,
372 reverse [1] Certificate OPTIONAL,
373 -- at least one of the pair shall be present -- }
375 Also implement the PEM functions to read and write certificate
376 pairs, and defined the PEM tag as "CERTIFICATE PAIR".
378 This needed to be defined, mostly for the sake of the LDAP
379 attribute crossCertificatePair, but may prove useful elsewhere as
383 *) Make it possible to inhibit symlinking of shared libraries in
384 Makefile.shared, for Cygwin's sake.
387 *) Extend the BIGNUM API by creating new macros that behave like
390 void BN_set_sign(BIGNUM *a, int neg);
391 int BN_get_sign(const BIGNUM *a);
393 and avoid the need to access 'a->neg' directly in applications.
394 [Nils Larsch <nla@trustcenter.de>]
396 *) Implement fast modular reduction for pseudo-Mersenne primes
397 used in NIST curves (crypto/bn/bn_nist.c, crypto/ec/ecp_nist.c).
398 EC_GROUP_new_curve_GFp() will now automatically use this
400 [Nils Larsch <nla@trustcenter.de>]
402 *) Add new lock type (CRYPTO_LOCK_BN).
405 *) Change the ENGINE framework to automatically load engines
406 dynamically from specific directories unless they could be
407 found to already be built in or loaded. Move all the
408 current engines except for the cryptodev one to a new
410 The engines in engines/ are built as shared libraries if
411 the "shared" options was given to ./Configure or ./config.
412 Otherwise, they are inserted in libcrypto.a.
413 /usr/local/ssl/engines is the default directory for dynamic
414 engines, but that can be overriden at configure time through
415 the usual use of --prefix and/or --openssldir, and at run
416 time with the environment variable OPENSSL_ENGINES.
417 [Geoff Thorpe and Richard Levitte]
419 *) Add Makefile.shared, a helper makefile to build shared
420 libraries. Addapt Makefile.org.
423 *) Add version info to Win32 DLLs.
424 [Peter 'Luna' Runestig" <peter@runestig.com>]
426 *) Add new 'medium level' PKCS#12 API. Certificates and keys
427 can be added using this API to created arbitrary PKCS#12
428 files while avoiding the low level API.
430 New options to PKCS12_create(), key or cert can be NULL and
431 will then be omitted from the output file. The encryption
432 algorithm NIDs can be set to -1 for no encryption, the mac
433 iteration count can be set to 0 to omit the mac.
435 Enhance pkcs12 utility by making the -nokeys and -nocerts
436 options work when creating a PKCS#12 file. New option -nomac
437 to omit the mac, NONE can be set for an encryption algorithm.
438 New code is modified to use the enhanced PKCS12_create()
439 instead of the low level API.
442 *) Extend ASN1 encoder to support indefinite length constructed
443 encoding. This can output sequences tags and octet strings in
444 this form. Modify pk7_asn1.c to support indefinite length
445 encoding. This is experimental and needs additional code to
446 be useful, such as an ASN1 bio and some enhanced streaming
449 Extend template encode functionality so that tagging is passed
450 down to the template encoder.
453 *) Let 'openssl req' fail if an argument to '-newkey' is not
454 recognized instead of using RSA as a default.
457 *) Add support for ECC-based ciphersuites from draft-ietf-tls-ecc-01.txt.
458 As these are not official, they are not included in "ALL";
459 the "ECCdraft" ciphersuite group alias can be used to select them.
460 [Vipul Gupta and Sumit Gupta (Sun Microsystems Laboratories)]
462 *) Add ECDH engine support.
463 [Nils Gura and Douglas Stebila (Sun Microsystems Laboratories)]
465 *) Add ECDH in new directory crypto/ecdh/.
466 [Douglas Stebila (Sun Microsystems Laboratories)]
468 *) Let BN_rand_range() abort with an error after 100 iterations
469 without success (which indicates a broken PRNG).
472 *) Change BN_mod_sqrt() so that it verifies that the input value
473 is really the square of the return value. (Previously,
474 BN_mod_sqrt would show GIGO behaviour.)
477 *) Add named elliptic curves over binary fields from X9.62, SECG,
478 and WAP/WTLS; add OIDs that were still missing.
480 [Sheueling Chang Shantz and Douglas Stebila
481 (Sun Microsystems Laboratories)]
483 *) Extend the EC library for elliptic curves over binary fields
484 (new files ec2_smpl.c, ec2_smpt.c, ec2_mult.c in crypto/ec/).
487 EC_GF2m_simple_method
491 EC_GROUP_new_curve_GF2m
492 EC_GROUP_set_curve_GF2m
493 EC_GROUP_get_curve_GF2m
494 EC_POINT_set_affine_coordinates_GF2m
495 EC_POINT_get_affine_coordinates_GF2m
496 EC_POINT_set_compressed_coordinates_GF2m
498 Point compression for binary fields is disabled by default for
499 patent reasons (compile with OPENSSL_EC_BIN_PT_COMP defined to
502 As binary polynomials are represented as BIGNUMs, various members
503 of the EC_GROUP and EC_POINT data structures can be shared
504 between the implementations for prime fields and binary fields;
505 the above ..._GF2m functions (except for EX_GROUP_new_curve_GF2m)
506 are essentially identical to their ..._GFp counterparts.
507 (For simplicity, the '..._GFp' prefix has been dropped from
508 various internal method names.)
510 An internal 'field_div' method (similar to 'field_mul' and
511 'field_sqr') has been added; this is used only for binary fields.
513 [Sheueling Chang Shantz and Douglas Stebila
514 (Sun Microsystems Laboratories)]
516 *) Optionally dispatch EC_POINT_mul(), EC_POINT_precompute_mult()
517 through methods ('mul', 'precompute_mult').
519 The generic implementations (now internally called 'ec_wNAF_mul'
520 and 'ec_wNAF_precomputed_mult') remain the default if these
521 methods are undefined.
523 [Sheueling Chang Shantz and Douglas Stebila
524 (Sun Microsystems Laboratories)]
526 *) New function EC_GROUP_get_degree, which is defined through
527 EC_METHOD. For curves over prime fields, this returns the bit
528 length of the modulus.
530 [Sheueling Chang Shantz and Douglas Stebila
531 (Sun Microsystems Laboratories)]
533 *) New functions EC_GROUP_dup, EC_POINT_dup.
534 (These simply call ..._new and ..._copy).
536 [Sheueling Chang Shantz and Douglas Stebila
537 (Sun Microsystems Laboratories)]
539 *) Add binary polynomial arithmetic software in crypto/bn/bn_gf2m.c.
540 Polynomials are represented as BIGNUMs (where the sign bit is not
541 used) in the following functions [macros]:
544 BN_GF2m_sub [= BN_GF2m_add]
545 BN_GF2m_mod [wrapper for BN_GF2m_mod_arr]
546 BN_GF2m_mod_mul [wrapper for BN_GF2m_mod_mul_arr]
547 BN_GF2m_mod_sqr [wrapper for BN_GF2m_mod_sqr_arr]
549 BN_GF2m_mod_exp [wrapper for BN_GF2m_mod_exp_arr]
550 BN_GF2m_mod_sqrt [wrapper for BN_GF2m_mod_sqrt_arr]
551 BN_GF2m_mod_solve_quad [wrapper for BN_GF2m_mod_solve_quad_arr]
552 BN_GF2m_cmp [= BN_ucmp]
554 (Note that only the 'mod' functions are actually for fields GF(2^m).
555 BN_GF2m_add() is misnomer, but this is for the sake of consistency.)
557 For some functions, an the irreducible polynomial defining a
558 field can be given as an 'unsigned int[]' with strictly
559 decreasing elements giving the indices of those bits that are set;
560 i.e., p[] represents the polynomial
561 f(t) = t^p[0] + t^p[1] + ... + t^p[k]
563 p[0] > p[1] > ... > p[k] = 0.
564 This applies to the following functions:
569 BN_GF2m_mod_inv_arr [wrapper for BN_GF2m_mod_inv]
570 BN_GF2m_mod_div_arr [wrapper for BN_GF2m_mod_div]
573 BN_GF2m_mod_solve_quad_arr
577 Conversion can be performed by the following functions:
582 bntest.c has additional tests for binary polynomial arithmetic.
584 Two implementations for BN_GF2m_mod_div() are available.
585 The default algorithm simply uses BN_GF2m_mod_inv() and
586 BN_GF2m_mod_mul(). The alternative algorithm is compiled in only
587 if OPENSSL_SUN_GF2M_DIV is defined (patent pending; read the
588 copyright notice in crypto/bn/bn_gf2m.c before enabling it).
590 [Sheueling Chang Shantz and Douglas Stebila
591 (Sun Microsystems Laboratories)]
593 *) Add new error code 'ERR_R_DISABLED' that can be used when some
594 functionality is disabled at compile-time.
595 [Douglas Stebila <douglas.stebila@sun.com>]
597 *) Change default behaviour of 'openssl asn1parse' so that more
598 information is visible when viewing, e.g., a certificate:
600 Modify asn1_parse2 (crypto/asn1/asn1_par.c) so that in non-'dump'
601 mode the content of non-printable OCTET STRINGs is output in a
602 style similar to INTEGERs, but with '[HEX DUMP]' prepended to
603 avoid the appearance of a printable string.
604 [Nils Larsch <nla@trustcenter.de>]
606 *) Add 'asn1_flag' and 'asn1_form' member to EC_GROUP with access
608 EC_GROUP_set_asn1_flag()
609 EC_GROUP_get_asn1_flag()
610 EC_GROUP_set_point_conversion_form()
611 EC_GROUP_get_point_conversion_form()
612 These control ASN1 encoding details:
613 - Curves (i.e., groups) are encoded explicitly unless asn1_flag
614 has been set to OPENSSL_EC_NAMED_CURVE.
615 - Points are encoded in uncompressed form by default; options for
616 asn1_for are as for point2oct, namely
617 POINT_CONVERSION_COMPRESSED
618 POINT_CONVERSION_UNCOMPRESSED
619 POINT_CONVERSION_HYBRID
621 Also add 'seed' and 'seed_len' members to EC_GROUP with access
625 EC_GROUP_get_seed_len()
626 This is used only for ASN1 purposes (so far).
627 [Nils Larsch <nla@trustcenter.de>]
629 *) Add 'field_type' member to EC_METHOD, which holds the NID
630 of the appropriate field type OID. The new function
631 EC_METHOD_get_field_type() returns this value.
632 [Nils Larsch <nla@trustcenter.de>]
639 providing useful interfaces to EC_POINT_point2oct() and
640 EC_POINT_oct2point().
641 [Nils Larsch <nla@trustcenter.de>]
643 *) Change internals of the EC library so that the functions
644 EC_GROUP_set_generator()
645 EC_GROUP_get_generator()
647 EC_GROUP_get_cofactor()
648 are implemented directly in crypto/ec/ec_lib.c and not dispatched
649 to methods, which would lead to unnecessary code duplication when
650 adding different types of curves.
651 [Nils Larsch <nla@trustcenter.de> with input by Bodo Moeller]
653 *) Implement compute_wNAF (crypto/ec/ec_mult.c) without BIGNUM
654 arithmetic, and such that modified wNAFs are generated
655 (which avoid length expansion in many cases).
658 *) Add a function EC_GROUP_check_discriminant() (defined via
659 EC_METHOD) that verifies that the curve discriminant is non-zero.
661 Add a function EC_GROUP_check() that makes some sanity tests
662 on a EC_GROUP, its generator and order. This includes
663 EC_GROUP_check_discriminant().
664 [Nils Larsch <nla@trustcenter.de>]
666 *) Add ECDSA in new directory crypto/ecdsa/.
668 Add applications 'openssl ecparam' and 'openssl ecdsa'
669 (these are based on 'openssl dsaparam' and 'openssl dsa').
671 ECDSA support is also included in various other files across the
672 library. Most notably,
673 - 'openssl req' now has a '-newkey ecdsa:file' option;
674 - EVP_PKCS82PKEY (crypto/evp/evp_pkey.c) now can handle ECDSA;
675 - X509_PUBKEY_get (crypto/asn1/x_pubkey.c) and
676 d2i_PublicKey (crypto/asn1/d2i_pu.c) have been modified to make
677 them suitable for ECDSA where domain parameters must be
678 extracted before the specific public key;
679 - ECDSA engine support has been added.
680 [Nils Larsch <nla@trustcenter.de>]
682 *) Include some named elliptic curves, and add OIDs from X9.62,
683 SECG, and WAP/WTLS. Each curve can be obtained from the new
685 EC_GROUP_new_by_nid(),
686 and the list of available named curves can be obtained with
687 EC_get_builtin_curves().
688 Also add a 'curve_name' member to EC_GROUP objects, which can be
692 [Nils Larsch <nla@trustcenter.de, Bodo Moeller]
694 *) Remove a few calls to bn_wexpand() in BN_sqr() (the one in there
695 was actually never needed) and in BN_mul(). The removal in BN_mul()
696 required a small change in bn_mul_part_recursive() and the addition
697 of the functions bn_cmp_part_words(), bn_sub_part_words() and
698 bn_add_part_words(), which do the same thing as bn_cmp_words(),
699 bn_sub_words() and bn_add_words() except they take arrays with
703 Changes between 0.9.7d and 0.9.7e [XX xxx XXXX]
705 *) Reduce the chances of duplicate issuer name and serial numbers (in
706 violation of RFC3280) using the OpenSSL certificate creation utilities.
707 This is done by creating a random 64 bit value for the initial serial
708 number when a serial number file is created or when a self signed
709 certificate is created using 'openssl req -x509'. The initial serial
710 number file is created using 'openssl x509 -next_serial' in CA.pl
711 rather than being initialized to 1.
714 Changes between 0.9.7c and 0.9.7d [17 Mar 2004]
716 *) Fix null-pointer assignment in do_change_cipher_spec() revealed
717 by using the Codenomicon TLS Test Tool (CAN-2004-0079)
718 [Joe Orton, Steve Henson]
720 *) Fix flaw in SSL/TLS handshaking when using Kerberos ciphersuites
722 [Joe Orton, Steve Henson]
724 *) Make it possible to have multiple active certificates with the same
725 subject in the CA index file. This is done only if the keyword
726 'unique_subject' is set to 'no' in the main CA section (default
727 if 'CA_default') of the configuration file. The value is saved
728 with the database itself in a separate index attribute file,
729 named like the index file with '.attr' appended to the name.
732 *) X509 verify fixes. Disable broken certificate workarounds when
733 X509_V_FLAGS_X509_STRICT is set. Check CRL issuer has cRLSign set if
734 keyUsage extension present. Don't accept CRLs with unhandled critical
735 extensions: since verify currently doesn't process CRL extensions this
736 rejects a CRL with *any* critical extensions. Add new verify error codes
740 *) When creating an OCSP nonce use an OCTET STRING inside the extnValue.
741 A clarification of RFC2560 will require the use of OCTET STRINGs and
742 some implementations cannot handle the current raw format. Since OpenSSL
743 copies and compares OCSP nonces as opaque blobs without any attempt at
744 parsing them this should not create any compatibility issues.
747 *) New md flag EVP_MD_CTX_FLAG_REUSE this allows md_data to be reused when
748 calling EVP_MD_CTX_copy_ex() to avoid calling OPENSSL_malloc(). Without
749 this HMAC (and other) operations are several times slower than OpenSSL
753 *) Print out GeneralizedTime and UTCTime in ASN1_STRING_print_ex().
754 [Peter Sylvester <Peter.Sylvester@EdelWeb.fr>]
756 *) Use the correct content when signing type "other".
759 Changes between 0.9.7b and 0.9.7c [30 Sep 2003]
761 *) Fix various bugs revealed by running the NISCC test suite:
763 Stop out of bounds reads in the ASN1 code when presented with
764 invalid tags (CAN-2003-0543 and CAN-2003-0544).
766 Free up ASN1_TYPE correctly if ANY type is invalid (CAN-2003-0545).
768 If verify callback ignores invalid public key errors don't try to check
769 certificate signature with the NULL public key.
773 *) New -ignore_err option in ocsp application to stop the server
774 exiting on the first error in a request.
777 *) In ssl3_accept() (ssl/s3_srvr.c) only accept a client certificate
778 if the server requested one: as stated in TLS 1.0 and SSL 3.0
782 *) In ssl3_get_client_hello() (ssl/s3_srvr.c), tolerate additional
783 extra data after the compression methods not only for TLS 1.0
784 but also for SSL 3.0 (as required by the specification).
785 [Bodo Moeller; problem pointed out by Matthias Loepfe]
787 *) Change X509_certificate_type() to mark the key as exported/exportable
788 when it's 512 *bits* long, not 512 bytes.
791 *) Change AES_cbc_encrypt() so it outputs exact multiple of
792 blocks during encryption.
795 *) Various fixes to base64 BIO and non blocking I/O. On write
796 flushes were not handled properly if the BIO retried. On read
797 data was not being buffered properly and had various logic bugs.
798 This also affects blocking I/O when the data being decoded is a
802 *) Various S/MIME bugfixes and compatibility changes:
803 output correct application/pkcs7 MIME type if
804 PKCS7_NOOLDMIMETYPE is set. Tolerate some broken signatures.
805 Output CR+LF for EOL if PKCS7_CRLFEOL is set (this makes opening
806 of files as .eml work). Correctly handle very long lines in MIME
810 Changes between 0.9.7a and 0.9.7b [10 Apr 2003]
812 *) Countermeasure against the Klima-Pokorny-Rosa extension of
813 Bleichbacher's attack on PKCS #1 v1.5 padding: treat
814 a protocol version number mismatch like a decryption error
815 in ssl3_get_client_key_exchange (ssl/s3_srvr.c).
818 *) Turn on RSA blinding by default in the default implementation
819 to avoid a timing attack. Applications that don't want it can call
820 RSA_blinding_off() or use the new flag RSA_FLAG_NO_BLINDING.
821 They would be ill-advised to do so in most cases.
822 [Ben Laurie, Steve Henson, Geoff Thorpe, Bodo Moeller]
824 *) Change RSA blinding code so that it works when the PRNG is not
825 seeded (in this case, the secret RSA exponent is abused as
826 an unpredictable seed -- if it is not unpredictable, there
827 is no point in blinding anyway). Make RSA blinding thread-safe
828 by remembering the creator's thread ID in rsa->blinding and
829 having all other threads use local one-time blinding factors
830 (this requires more computation than sharing rsa->blinding, but
831 avoids excessive locking; and if an RSA object is not shared
832 between threads, blinding will still be very fast).
835 *) Fixed a typo bug that would cause ENGINE_set_default() to set an
836 ENGINE as defaults for all supported algorithms irrespective of
837 the 'flags' parameter. 'flags' is now honoured, so applications
838 should make sure they are passing it correctly.
841 *) Target "mingw" now allows native Windows code to be generated in
842 the Cygwin environment as well as with the MinGW compiler.
845 Changes between 0.9.7 and 0.9.7a [19 Feb 2003]
847 *) In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked
848 via timing by performing a MAC computation even if incorrrect
849 block cipher padding has been found. This is a countermeasure
850 against active attacks where the attacker has to distinguish
851 between bad padding and a MAC verification error. (CAN-2003-0078)
853 [Bodo Moeller; problem pointed out by Brice Canvel (EPFL),
854 Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and
855 Martin Vuagnoux (EPFL, Ilion)]
857 *) Make the no-err option work as intended. The intention with no-err
858 is not to have the whole error stack handling routines removed from
859 libcrypto, it's only intended to remove all the function name and
860 reason texts, thereby removing some of the footprint that may not
861 be interesting if those errors aren't displayed anyway.
863 NOTE: it's still possible for any application or module to have it's
864 own set of error texts inserted. The routines are there, just not
865 used by default when no-err is given.
868 *) Add support for FreeBSD on IA64.
869 [dirk.meyer@dinoex.sub.org via Richard Levitte, resolves #454]
871 *) Adjust DES_cbc_cksum() so it returns the same value as the MIT
872 Kerberos function mit_des_cbc_cksum(). Before this change,
873 the value returned by DES_cbc_cksum() was like the one from
874 mit_des_cbc_cksum(), except the bytes were swapped.
875 [Kevin Greaney <Kevin.Greaney@hp.com> and Richard Levitte]
877 *) Allow an application to disable the automatic SSL chain building.
878 Before this a rather primitive chain build was always performed in
879 ssl3_output_cert_chain(): an application had no way to send the
880 correct chain if the automatic operation produced an incorrect result.
882 Now the chain builder is disabled if either:
884 1. Extra certificates are added via SSL_CTX_add_extra_chain_cert().
886 2. The mode flag SSL_MODE_NO_AUTO_CHAIN is set.
888 The reasoning behind this is that an application would not want the
889 auto chain building to take place if extra chain certificates are
890 present and it might also want a means of sending no additional
891 certificates (for example the chain has two certificates and the
895 *) Add the possibility to build without the ENGINE framework.
896 [Steven Reddie <smr@essemer.com.au> via Richard Levitte]
898 *) Under Win32 gmtime() can return NULL: check return value in
899 OPENSSL_gmtime(). Add error code for case where gmtime() fails.
902 *) DSA routines: under certain error conditions uninitialized BN objects
903 could be freed. Solution: make sure initialization is performed early
904 enough. (Reported and fix supplied by Ivan D Nestlerode <nestler@MIT.EDU>,
905 Nils Larsch <nla@trustcenter.de> via PR#459)
908 *) Another fix for SSLv2 session ID handling: the session ID was incorrectly
909 checked on reconnect on the client side, therefore session resumption
910 could still fail with a "ssl session id is different" error. This
911 behaviour is masked when SSL_OP_ALL is used due to
912 SSL_OP_MICROSOFT_SESS_ID_BUG being set.
913 Behaviour observed by Crispin Flowerday <crispin@flowerday.cx> as
917 *) IA-32 assembler support enhancements: unified ELF targets, support
918 for SCO/Caldera platforms, fix for Cygwin shared build.
921 *) Add support for FreeBSD on sparc64. As a consequence, support for
922 FreeBSD on non-x86 processors is separate from x86 processors on
923 the config script, much like the NetBSD support.
924 [Richard Levitte & Kris Kennaway <kris@obsecurity.org>]
926 Changes between 0.9.6h and 0.9.7 [31 Dec 2002]
928 [NB: OpenSSL 0.9.6i and later 0.9.6 patch levels were released after
931 *) Fix session ID handling in SSLv2 client code: the SERVER FINISHED
932 code (06) was taken as the first octet of the session ID and the last
933 octet was ignored consequently. As a result SSLv2 client side session
934 caching could not have worked due to the session ID mismatch between
936 Behaviour observed by Crispin Flowerday <crispin@flowerday.cx> as
940 *) Change the declaration of needed Kerberos libraries to use EX_LIBS
941 instead of the special (and badly supported) LIBKRB5. LIBKRB5 is
945 *) The hw_ncipher.c engine requires dynamic locks. Unfortunately, it
946 seems that in spite of existing for more than a year, many application
947 author have done nothing to provide the necessary callbacks, which
948 means that this particular engine will not work properly anywhere.
949 This is a very unfortunate situation which forces us, in the name
950 of usability, to give the hw_ncipher.c a static lock, which is part
952 NOTE: This is for the 0.9.7 series ONLY. This hack will never
953 appear in 0.9.8 or later. We EXPECT application authors to have
954 dealt properly with this when 0.9.8 is released (unless we actually
955 make such changes in the libcrypto locking code that changes will
956 have to be made anyway).
959 *) In asn1_d2i_read_bio() repeatedly call BIO_read() until all content
960 octets have been read, EOF or an error occurs. Without this change
961 some truncated ASN1 structures will not produce an error.
964 *) Disable Heimdal support, since it hasn't been fully implemented.
965 Still give the possibility to force the use of Heimdal, but with
966 warnings and a request that patches get sent to openssl-dev.
969 *) Add the VC-CE target, introduce the WINCE sysname, and add
970 INSTALL.WCE and appropriate conditionals to make it build.
971 [Steven Reddie <smr@essemer.com.au> via Richard Levitte]
973 *) Change the DLL names for Cygwin to cygcrypto-x.y.z.dll and
974 cygssl-x.y.z.dll, where x, y and z are the major, minor and
975 edit numbers of the version.
976 [Corinna Vinschen <vinschen@redhat.com> and Richard Levitte]
978 *) Introduce safe string copy and catenation functions
979 (BUF_strlcpy() and BUF_strlcat()).
980 [Ben Laurie (CHATS) and Richard Levitte]
982 *) Avoid using fixed-size buffers for one-line DNs.
985 *) Add BUF_MEM_grow_clean() to avoid information leakage when
986 resizing buffers containing secrets, and use where appropriate.
989 *) Avoid using fixed size buffers for configuration file location.
992 *) Avoid filename truncation for various CA files.
995 *) Use sizeof in preference to magic numbers.
998 *) Avoid filename truncation in cert requests.
1001 *) Add assertions to check for (supposedly impossible) buffer
1003 [Ben Laurie (CHATS)]
1005 *) Don't cache truncated DNS entries in the local cache (this could
1006 potentially lead to a spoofing attack).
1007 [Ben Laurie (CHATS)]
1009 *) Fix various buffers to be large enough for hex/decimal
1010 representations in a platform independent manner.
1011 [Ben Laurie (CHATS)]
1013 *) Add CRYPTO_realloc_clean() to avoid information leakage when
1014 resizing buffers containing secrets, and use where appropriate.
1015 [Ben Laurie (CHATS)]
1017 *) Add BIO_indent() to avoid much slightly worrying code to do
1019 [Ben Laurie (CHATS)]
1021 *) Convert sprintf()/BIO_puts() to BIO_printf().
1022 [Ben Laurie (CHATS)]
1024 *) buffer_gets() could terminate with the buffer only half
1026 [Ben Laurie (CHATS)]
1028 *) Add assertions to prevent user-supplied crypto functions from
1029 overflowing internal buffers by having large block sizes, etc.
1030 [Ben Laurie (CHATS)]
1032 *) New OPENSSL_assert() macro (similar to assert(), but enabled
1034 [Ben Laurie (CHATS)]
1036 *) Eliminate unused copy of key in RC4.
1037 [Ben Laurie (CHATS)]
1039 *) Eliminate unused and incorrectly sized buffers for IV in pem.h.
1040 [Ben Laurie (CHATS)]
1042 *) Fix off-by-one error in EGD path.
1043 [Ben Laurie (CHATS)]
1045 *) If RANDFILE path is too long, ignore instead of truncating.
1046 [Ben Laurie (CHATS)]
1048 *) Eliminate unused and incorrectly sized X.509 structure
1050 [Ben Laurie (CHATS)]
1052 *) Eliminate unused and dangerous function knumber().
1053 [Ben Laurie (CHATS)]
1055 *) Eliminate unused and dangerous structure, KSSL_ERR.
1056 [Ben Laurie (CHATS)]
1058 *) Protect against overlong session ID context length in an encoded
1059 session object. Since these are local, this does not appear to be
1061 [Ben Laurie (CHATS)]
1063 *) Change from security patch (see 0.9.6e below) that did not affect
1064 the 0.9.6 release series:
1066 Remote buffer overflow in SSL3 protocol - an attacker could
1067 supply an oversized master key in Kerberos-enabled versions.
1069 [Ben Laurie (CHATS)]
1071 *) Change the SSL kerb5 codes to match RFC 2712.
1074 *) Make -nameopt work fully for req and add -reqopt switch.
1075 [Michael Bell <michael.bell@rz.hu-berlin.de>, Steve Henson]
1077 *) The "block size" for block ciphers in CFB and OFB mode should be 1.
1078 [Steve Henson, reported by Yngve Nysaeter Pettersen <yngve@opera.com>]
1080 *) Make sure tests can be performed even if the corresponding algorithms
1081 have been removed entirely. This was also the last step to make
1082 OpenSSL compilable with DJGPP under all reasonable conditions.
1083 [Richard Levitte, Doug Kaufman <dkaufman@rahul.net>]
1085 *) Add cipher selection rules COMPLEMENTOFALL and COMPLEMENTOFDEFAULT
1086 to allow version independent disabling of normally unselected ciphers,
1087 which may be activated as a side-effect of selecting a single cipher.
1089 (E.g., cipher list string "RSA" enables ciphersuites that are left
1090 out of "ALL" because they do not provide symmetric encryption.
1091 "RSA:!COMPLEMEMENTOFALL" avoids these unsafe ciphersuites.)
1092 [Lutz Jaenicke, Bodo Moeller]
1094 *) Add appropriate support for separate platform-dependent build
1095 directories. The recommended way to make a platform-dependent
1096 build directory is the following (tested on Linux), maybe with
1099 # Place yourself outside of the OpenSSL source tree. In
1100 # this example, the environment variable OPENSSL_SOURCE
1101 # is assumed to contain the absolute OpenSSL source directory.
1102 mkdir -p objtree/"`uname -s`-`uname -r`-`uname -m`"
1103 cd objtree/"`uname -s`-`uname -r`-`uname -m`"
1104 (cd $OPENSSL_SOURCE; find . -type f) | while read F; do
1105 mkdir -p `dirname $F`
1106 ln -s $OPENSSL_SOURCE/$F $F
1109 To be absolutely sure not to disturb the source tree, a "make clean"
1110 is a good thing. If it isn't successfull, don't worry about it,
1111 it probably means the source directory is very clean.
1114 *) Make sure any ENGINE control commands make local copies of string
1115 pointers passed to them whenever necessary. Otherwise it is possible
1116 the caller may have overwritten (or deallocated) the original string
1117 data when a later ENGINE operation tries to use the stored values.
1118 [Götz Babin-Ebell <babinebell@trustcenter.de>]
1120 *) Improve diagnostics in file reading and command-line digests.
1121 [Ben Laurie aided and abetted by Solar Designer <solar@openwall.com>]
1123 *) Add AES modes CFB and OFB to the object database. Correct an
1124 error in AES-CFB decryption.
1127 *) Remove most calls to EVP_CIPHER_CTX_cleanup() in evp_enc.c, this
1128 allows existing EVP_CIPHER_CTX structures to be reused after
1129 calling EVP_*Final(). This behaviour is used by encryption
1130 BIOs and some applications. This has the side effect that
1131 applications must explicitly clean up cipher contexts with
1132 EVP_CIPHER_CTX_cleanup() or they will leak memory.
1135 *) Check the values of dna and dnb in bn_mul_recursive before calling
1136 bn_mul_comba (a non zero value means the a or b arrays do not contain
1137 n2 elements) and fallback to bn_mul_normal if either is not zero.
1140 *) Fix escaping of non-ASCII characters when using the -subj option
1141 of the "openssl req" command line tool. (Robert Joop <joop@fokus.gmd.de>)
1144 *) Make object definitions compliant to LDAP (RFC2256): SN is the short
1145 form for "surname", serialNumber has no short form.
1146 Use "mail" as the short name for "rfc822Mailbox" according to RFC2798;
1147 therefore remove "mail" short name for "internet 7".
1148 The OID for unique identifiers in X509 certificates is
1149 x500UniqueIdentifier, not uniqueIdentifier.
1150 Some more OID additions. (Michael Bell <michael.bell@rz.hu-berlin.de>)
1153 *) Add an "init" command to the ENGINE config module and auto initialize
1154 ENGINEs. Without any "init" command the ENGINE will be initialized
1155 after all ctrl commands have been executed on it. If init=1 the
1156 ENGINE is initailized at that point (ctrls before that point are run
1157 on the uninitialized ENGINE and after on the initialized one). If
1158 init=0 then the ENGINE will not be iniatialized at all.
1161 *) Fix the 'app_verify_callback' interface so that the user-defined
1162 argument is actually passed to the callback: In the
1163 SSL_CTX_set_cert_verify_callback() prototype, the callback
1164 declaration has been changed from
1167 int (*cb)(X509_STORE_CTX *,void *);
1168 in ssl_verify_cert_chain (ssl/ssl_cert.c), the call
1169 i=s->ctx->app_verify_callback(&ctx)
1170 has been changed into
1171 i=s->ctx->app_verify_callback(&ctx, s->ctx->app_verify_arg).
1173 To update applications using SSL_CTX_set_cert_verify_callback(),
1174 a dummy argument can be added to their callback functions.
1175 [D. K. Smetters <smetters@parc.xerox.com>]
1177 *) Added the '4758cca' ENGINE to support IBM 4758 cards.
1178 [Maurice Gittens <maurice@gittens.nl>, touchups by Geoff Thorpe]
1180 *) Add and OPENSSL_LOAD_CONF define which will cause
1181 OpenSSL_add_all_algorithms() to load the openssl.cnf config file.
1182 This allows older applications to transparently support certain
1183 OpenSSL features: such as crypto acceleration and dynamic ENGINE loading.
1184 Two new functions OPENSSL_add_all_algorithms_noconf() which will never
1185 load the config file and OPENSSL_add_all_algorithms_conf() which will
1186 always load it have also been added.
1189 *) Add the OFB, CFB and CTR (all with 128 bit feedback) to AES.
1190 Adjust NIDs and EVP layer.
1191 [Stephen Sprunk <stephen@sprunk.org> and Richard Levitte]
1193 *) Config modules support in openssl utility.
1195 Most commands now load modules from the config file,
1196 though in a few (such as version) this isn't done
1197 because it couldn't be used for anything.
1199 In the case of ca and req the config file used is
1200 the same as the utility itself: that is the -config
1201 command line option can be used to specify an
1205 *) Move default behaviour from OPENSSL_config(). If appname is NULL
1206 use "openssl_conf" if filename is NULL use default openssl config file.
1209 *) Add an argument to OPENSSL_config() to allow the use of an alternative
1210 config section name. Add a new flag to tolerate a missing config file
1211 and move code to CONF_modules_load_file().
1214 *) Support for crypto accelerator cards from Accelerated Encryption
1215 Processing, www.aep.ie. (Use engine 'aep')
1216 The support was copied from 0.9.6c [engine] and adapted/corrected
1217 to work with the new engine framework.
1218 [AEP Inc. and Richard Levitte]
1220 *) Support for SureWare crypto accelerator cards from Baltimore
1221 Technologies. (Use engine 'sureware')
1222 The support was copied from 0.9.6c [engine] and adapted
1223 to work with the new engine framework.
1226 *) Have the CHIL engine fork-safe (as defined by nCipher) and actually
1227 make the newer ENGINE framework commands for the CHIL engine work.
1228 [Toomas Kiisk <vix@cyber.ee> and Richard Levitte]
1230 *) Make it possible to produce shared libraries on ReliantUNIX.
1231 [Robert Dahlem <Robert.Dahlem@ffm2.siemens.de> via Richard Levitte]
1233 *) Add the configuration target debug-linux-ppro.
1234 Make 'openssl rsa' use the general key loading routines
1235 implemented in apps.c, and make those routines able to
1236 handle the key format FORMAT_NETSCAPE and the variant
1238 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte]
1240 *) Fix a crashbug and a logic bug in hwcrhk_load_pubkey().
1241 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte]
1243 *) Add -keyform to rsautl, and document -engine.
1244 [Richard Levitte, inspired by Toomas Kiisk <vix@cyber.ee>]
1246 *) Change BIO_new_file (crypto/bio/bss_file.c) to use new
1247 BIO_R_NO_SUCH_FILE error code rather than the generic
1248 ERR_R_SYS_LIB error code if fopen() fails with ENOENT.
1251 *) Add new functions
1253 ERR_peek_last_error_line
1254 ERR_peek_last_error_line_data.
1255 These are similar to
1258 ERR_peek_error_line_data,
1259 but report on the latest error recorded rather than the first one
1260 still in the error queue.
1261 [Ben Laurie, Bodo Moeller]
1263 *) default_algorithms option in ENGINE config module. This allows things
1265 default_algorithms = ALL
1266 default_algorithms = RSA, DSA, RAND, CIPHERS, DIGESTS
1269 *) Prelminary ENGINE config module.
1272 *) New experimental application configuration code.
1275 *) Change the AES code to follow the same name structure as all other
1276 symmetric ciphers, and behave the same way. Move everything to
1277 the directory crypto/aes, thereby obsoleting crypto/rijndael.
1278 [Stephen Sprunk <stephen@sprunk.org> and Richard Levitte]
1280 *) SECURITY: remove unsafe setjmp/signal interaction from ui_openssl.c.
1281 [Ben Laurie and Theo de Raadt]
1283 *) Add option to output public keys in req command.
1284 [Massimiliano Pala madwolf@openca.org]
1286 *) Use wNAFs in EC_POINTs_mul() for improved efficiency
1287 (up to about 10% better than before for P-192 and P-224).
1290 *) New functions/macros
1292 SSL_CTX_set_msg_callback(ctx, cb)
1293 SSL_CTX_set_msg_callback_arg(ctx, arg)
1294 SSL_set_msg_callback(ssl, cb)
1295 SSL_set_msg_callback_arg(ssl, arg)
1297 to request calling a callback function
1299 void cb(int write_p, int version, int content_type,
1300 const void *buf, size_t len, SSL *ssl, void *arg)
1302 whenever a protocol message has been completely received
1303 (write_p == 0) or sent (write_p == 1). Here 'version' is the
1304 protocol version according to which the SSL library interprets
1305 the current protocol message (SSL2_VERSION, SSL3_VERSION, or
1306 TLS1_VERSION). 'content_type' is 0 in the case of SSL 2.0, or
1307 the content type as defined in the SSL 3.0/TLS 1.0 protocol
1308 specification (change_cipher_spec(20), alert(21), handshake(22)).
1309 'buf' and 'len' point to the actual message, 'ssl' to the
1310 SSL object, and 'arg' is the application-defined value set by
1311 SSL[_CTX]_set_msg_callback_arg().
1313 'openssl s_client' and 'openssl s_server' have new '-msg' options
1314 to enable a callback that displays all protocol messages.
1317 *) Change the shared library support so shared libraries are built as
1318 soon as the corresponding static library is finished, and thereby get
1319 openssl and the test programs linked against the shared library.
1320 This still only happens when the keyword "shard" has been given to
1321 the configuration scripts.
1323 NOTE: shared library support is still an experimental thing, and
1324 backward binary compatibility is still not guaranteed.
1325 ["Maciej W. Rozycki" <macro@ds2.pg.gda.pl> and Richard Levitte]
1327 *) Add support for Subject Information Access extension.
1328 [Peter Sylvester <Peter.Sylvester@EdelWeb.fr>]
1330 *) Make BUF_MEM_grow() behaviour more consistent: Initialise to zero
1331 additional bytes when new memory had to be allocated, not just
1332 when reusing an existing buffer.
1335 *) New command line and configuration option 'utf8' for the req command.
1336 This allows field values to be specified as UTF8 strings.
1339 *) Add -multi and -mr options to "openssl speed" - giving multiple parallel
1340 runs for the former and machine-readable output for the latter.
1343 *) Add '-noemailDN' option to 'openssl ca'. This prevents inclusion
1344 of the e-mail address in the DN (i.e., it will go into a certificate
1345 extension only). The new configuration file option 'email_in_dn = no'
1346 has the same effect.
1347 [Massimiliano Pala madwolf@openca.org]
1349 *) Change all functions with names starting with des_ to be starting
1350 with DES_ instead. Add wrappers that are compatible with libdes,
1351 but are named _ossl_old_des_*. Finally, add macros that map the
1352 des_* symbols to the corresponding _ossl_old_des_* if libdes
1353 compatibility is desired. If OpenSSL 0.9.6c compatibility is
1354 desired, the des_* symbols will be mapped to DES_*, with one
1357 Since we provide two compatibility mappings, the user needs to
1358 define the macro OPENSSL_DES_LIBDES_COMPATIBILITY if libdes
1359 compatibility is desired. The default (i.e., when that macro
1360 isn't defined) is OpenSSL 0.9.6c compatibility.
1362 There are also macros that enable and disable the support of old
1363 des functions altogether. Those are OPENSSL_ENABLE_OLD_DES_SUPPORT
1364 and OPENSSL_DISABLE_OLD_DES_SUPPORT. If none or both of those
1365 are defined, the default will apply: to support the old des routines.
1367 In either case, one must include openssl/des.h to get the correct
1368 definitions. Do not try to just include openssl/des_old.h, that
1371 NOTE: This is a major break of an old API into a new one. Software
1372 authors are encouraged to switch to the DES_ style functions. Some
1373 time in the future, des_old.h and the libdes compatibility functions
1374 will be disable (i.e. OPENSSL_DISABLE_OLD_DES_SUPPORT will be the
1375 default), and then completely removed.
1378 *) Test for certificates which contain unsupported critical extensions.
1379 If such a certificate is found during a verify operation it is
1380 rejected by default: this behaviour can be overridden by either
1381 handling the new error X509_V_ERR_UNHANDLED_CRITICAL_EXTENSION or
1382 by setting the verify flag X509_V_FLAG_IGNORE_CRITICAL. A new function
1383 X509_supported_extension() has also been added which returns 1 if a
1384 particular extension is supported.
1387 *) Modify the behaviour of EVP cipher functions in similar way to digests
1388 to retain compatibility with existing code.
1391 *) Modify the behaviour of EVP_DigestInit() and EVP_DigestFinal() to retain
1392 compatibility with existing code. In particular the 'ctx' parameter does
1393 not have to be to be initialized before the call to EVP_DigestInit() and
1394 it is tidied up after a call to EVP_DigestFinal(). New function
1395 EVP_DigestFinal_ex() which does not tidy up the ctx. Similarly function
1396 EVP_MD_CTX_copy() changed to not require the destination to be
1397 initialized valid and new function EVP_MD_CTX_copy_ex() added which
1398 requires the destination to be valid.
1400 Modify all the OpenSSL digest calls to use EVP_DigestInit_ex(),
1401 EVP_DigestFinal_ex() and EVP_MD_CTX_copy_ex().
1404 *) Change ssl3_get_message (ssl/s3_both.c) and the functions using it
1405 so that complete 'Handshake' protocol structures are kept in memory
1406 instead of overwriting 'msg_type' and 'length' with 'body' data.
1409 *) Add an implementation of SSL_add_dir_cert_subjects_to_stack for Win32.
1410 [Massimo Santin via Richard Levitte]
1412 *) Major restructuring to the underlying ENGINE code. This includes
1413 reduction of linker bloat, separation of pure "ENGINE" manipulation
1414 (initialisation, etc) from functionality dealing with implementations
1415 of specific crypto iterfaces. This change also introduces integrated
1416 support for symmetric ciphers and digest implementations - so ENGINEs
1417 can now accelerate these by providing EVP_CIPHER and EVP_MD
1418 implementations of their own. This is detailed in crypto/engine/README
1419 as it couldn't be adequately described here. However, there are a few
1420 API changes worth noting - some RSA, DSA, DH, and RAND functions that
1421 were changed in the original introduction of ENGINE code have now
1422 reverted back - the hooking from this code to ENGINE is now a good
1423 deal more passive and at run-time, operations deal directly with
1424 RSA_METHODs, DSA_METHODs (etc) as they did before, rather than
1425 dereferencing through an ENGINE pointer any more. Also, the ENGINE
1426 functions dealing with BN_MOD_EXP[_CRT] handlers have been removed -
1427 they were not being used by the framework as there is no concept of a
1428 BIGNUM_METHOD and they could not be generalised to the new
1429 'ENGINE_TABLE' mechanism that underlies the new code. Similarly,
1430 ENGINE_cpy() has been removed as it cannot be consistently defined in
1434 *) Change ASN1_GENERALIZEDTIME_check() to allow fractional seconds.
1437 *) Change mkdef.pl to sort symbols that get the same entry number,
1438 and make sure the automatically generated functions ERR_load_*
1439 become part of libeay.num as well.
1442 *) New function SSL_renegotiate_pending(). This returns true once
1443 renegotiation has been requested (either SSL_renegotiate() call
1444 or HelloRequest/ClientHello receveived from the peer) and becomes
1445 false once a handshake has been completed.
1446 (For servers, SSL_renegotiate() followed by SSL_do_handshake()
1447 sends a HelloRequest, but does not ensure that a handshake takes
1448 place. SSL_renegotiate_pending() is useful for checking if the
1449 client has followed the request.)
1452 *) New SSL option SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION.
1453 By default, clients may request session resumption even during
1454 renegotiation (if session ID contexts permit); with this option,
1455 session resumption is possible only in the first handshake.
1457 SSL_OP_ALL is now 0x00000FFFL instead of 0x000FFFFFL. This makes
1458 more bits available for options that should not be part of
1459 SSL_OP_ALL (such as SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION).
1462 *) Add some demos for certificate and certificate request creation.
1465 *) Make maximum certificate chain size accepted from the peer application
1466 settable (SSL*_get/set_max_cert_list()), as proposed by
1467 "Douglas E. Engert" <deengert@anl.gov>.
1470 *) Add support for shared libraries for Unixware-7
1471 (Boyd Lynn Gerber <gerberb@zenez.com>).
1474 *) Add a "destroy" handler to ENGINEs that allows structural cleanup to
1475 be done prior to destruction. Use this to unload error strings from
1476 ENGINEs that load their own error strings. NB: This adds two new API
1477 functions to "get" and "set" this destroy handler in an ENGINE.
1480 *) Alter all existing ENGINE implementations (except "openssl" and
1481 "openbsd") to dynamically instantiate their own error strings. This
1482 makes them more flexible to be built both as statically-linked ENGINEs
1483 and self-contained shared-libraries loadable via the "dynamic" ENGINE.
1484 Also, add stub code to each that makes building them as self-contained
1485 shared-libraries easier (see README.ENGINE).
1488 *) Add a "dynamic" ENGINE that provides a mechanism for binding ENGINE
1489 implementations into applications that are completely implemented in
1490 self-contained shared-libraries. The "dynamic" ENGINE exposes control
1491 commands that can be used to configure what shared-library to load and
1492 to control aspects of the way it is handled. Also, made an update to
1493 the README.ENGINE file that brings its information up-to-date and
1494 provides some information and instructions on the "dynamic" ENGINE
1495 (ie. how to use it, how to build "dynamic"-loadable ENGINEs, etc).
1498 *) Make it possible to unload ranges of ERR strings with a new
1499 "ERR_unload_strings" function.
1502 *) Add a copy() function to EVP_MD.
1505 *) Make EVP_MD routines take a context pointer instead of just the
1506 md_data void pointer.
1509 *) Add flags to EVP_MD and EVP_MD_CTX. EVP_MD_FLAG_ONESHOT indicates
1510 that the digest can only process a single chunk of data
1511 (typically because it is provided by a piece of
1512 hardware). EVP_MD_CTX_FLAG_ONESHOT indicates that the application
1513 is only going to provide a single chunk of data, and hence the
1514 framework needn't accumulate the data for oneshot drivers.
1517 *) As with "ERR", make it possible to replace the underlying "ex_data"
1518 functions. This change also alters the storage and management of global
1519 ex_data state - it's now all inside ex_data.c and all "class" code (eg.
1520 RSA, BIO, SSL_CTX, etc) no longer stores its own STACKS and per-class
1521 index counters. The API functions that use this state have been changed
1522 to take a "class_index" rather than pointers to the class's local STACK
1523 and counter, and there is now an API function to dynamically create new
1524 classes. This centralisation allows us to (a) plug a lot of the
1525 thread-safety problems that existed, and (b) makes it possible to clean
1526 up all allocated state using "CRYPTO_cleanup_all_ex_data()". W.r.t. (b)
1527 such data would previously have always leaked in application code and
1528 workarounds were in place to make the memory debugging turn a blind eye
1529 to it. Application code that doesn't use this new function will still
1530 leak as before, but their memory debugging output will announce it now
1531 rather than letting it slide.
1533 Besides the addition of CRYPTO_cleanup_all_ex_data(), another API change
1534 induced by the "ex_data" overhaul is that X509_STORE_CTX_init() now
1535 has a return value to indicate success or failure.
1538 *) Make it possible to replace the underlying "ERR" functions such that the
1539 global state (2 LHASH tables and 2 locks) is only used by the "default"
1540 implementation. This change also adds two functions to "get" and "set"
1541 the implementation prior to it being automatically set the first time
1542 any other ERR function takes place. Ie. an application can call "get",
1543 pass the return value to a module it has just loaded, and that module
1544 can call its own "set" function using that value. This means the
1545 module's "ERR" operations will use (and modify) the error state in the
1546 application and not in its own statically linked copy of OpenSSL code.
1549 *) Give DH, DSA, and RSA types their own "**_up_ref()" function to increment
1550 reference counts. This performs normal REF_PRINT/REF_CHECK macros on
1551 the operation, and provides a more encapsulated way for external code
1552 (crypto/evp/ and ssl/) to do this. Also changed the evp and ssl code
1553 to use these functions rather than manually incrementing the counts.
1555 Also rename "DSO_up()" function to more descriptive "DSO_up_ref()".
1558 *) Add EVP test program.
1561 *) Add symmetric cipher support to ENGINE. Expect the API to change!
1564 *) New CRL functions: X509_CRL_set_version(), X509_CRL_set_issuer_name()
1565 X509_CRL_set_lastUpdate(), X509_CRL_set_nextUpdate(), X509_CRL_sort(),
1566 X509_REVOKED_set_serialNumber(), and X509_REVOKED_set_revocationDate().
1567 These allow a CRL to be built without having to access X509_CRL fields
1568 directly. Modify 'ca' application to use new functions.
1571 *) Move SSL_OP_TLS_ROLLBACK_BUG out of the SSL_OP_ALL list of recommended
1572 bug workarounds. Rollback attack detection is a security feature.
1573 The problem will only arise on OpenSSL servers when TLSv1 is not
1574 available (sslv3_server_method() or SSL_OP_NO_TLSv1).
1575 Software authors not wanting to support TLSv1 will have special reasons
1576 for their choice and can explicitly enable this option.
1577 [Bodo Moeller, Lutz Jaenicke]
1579 *) Rationalise EVP so it can be extended: don't include a union of
1580 cipher/digest structures, add init/cleanup functions for EVP_MD_CTX
1581 (similar to those existing for EVP_CIPHER_CTX).
1586 EVP_MD_CTX_init(&md); /* new function call */
1587 EVP_DigestInit(&md, EVP_sha1());
1588 EVP_DigestUpdate(&md, in, len);
1589 EVP_DigestFinal(&md, out, NULL);
1590 EVP_MD_CTX_cleanup(&md); /* new function call */
1594 *) Make DES key schedule conform to the usual scheme, as well as
1595 correcting its structure. This means that calls to DES functions
1596 now have to pass a pointer to a des_key_schedule instead of a
1597 plain des_key_schedule (which was actually always a pointer
1600 des_key_schedule ks;
1602 des_set_key_checked(..., &ks);
1603 des_ncbc_encrypt(..., &ks, ...);
1605 (Note that a later change renames 'des_...' into 'DES_...'.)
1608 *) Initial reduction of linker bloat: the use of some functions, such as
1609 PEM causes large amounts of unused functions to be linked in due to
1610 poor organisation. For example pem_all.c contains every PEM function
1611 which has a knock on effect of linking in large amounts of (unused)
1612 ASN1 code. Grouping together similar functions and splitting unrelated
1613 functions prevents this.
1616 *) Cleanup of EVP macros.
1619 *) Change historical references to {NID,SN,LN}_des_ede and ede3 to add the
1620 correct _ecb suffix.
1623 *) Add initial OCSP responder support to ocsp application. The
1624 revocation information is handled using the text based index
1625 use by the ca application. The responder can either handle
1626 requests generated internally, supplied in files (for example
1627 via a CGI script) or using an internal minimal server.
1630 *) Add configuration choices to get zlib compression for TLS.
1633 *) Changes to Kerberos SSL for RFC 2712 compliance:
1634 1. Implemented real KerberosWrapper, instead of just using
1635 KRB5 AP_REQ message. [Thanks to Simon Wilkinson <sxw@sxw.org.uk>]
1636 2. Implemented optional authenticator field of KerberosWrapper.
1638 Added openssl-style ASN.1 macros for Kerberos ticket, ap_req,
1639 and authenticator structs; see crypto/krb5/.
1641 Generalized Kerberos calls to support multiple Kerberos libraries.
1642 [Vern Staats <staatsvr@asc.hpc.mil>,
1643 Jeffrey Altman <jaltman@columbia.edu>
1644 via Richard Levitte]
1646 *) Cause 'openssl speed' to use fully hard-coded DSA keys as it
1647 already does with RSA. testdsa.h now has 'priv_key/pub_key'
1648 values for each of the key sizes rather than having just
1649 parameters (and 'speed' generating keys each time).
1652 *) Speed up EVP routines.
1655 type 8 bytes 64 bytes 256 bytes 1024 bytes 8192 bytes
1656 des-cbc 4408.85k 5560.51k 5778.46k 5862.20k 5825.16k
1657 des-cbc 4389.55k 5571.17k 5792.23k 5846.91k 5832.11k
1658 des-cbc 4394.32k 5575.92k 5807.44k 5848.37k 5841.30k
1660 des-cbc 3482.66k 5069.49k 5496.39k 5614.16k 5639.28k
1661 des-cbc 3480.74k 5068.76k 5510.34k 5609.87k 5635.52k
1662 des-cbc 3483.72k 5067.62k 5504.60k 5708.01k 5724.80k
1665 des-cbc 4660.16k 5650.19k 5807.19k 5827.13k 5783.32k
1667 des-cbc 3624.96k 5258.21k 5530.91k 5624.30k 5628.26k
1670 *) Added the OS2-EMX target.
1671 ["Brian Havard" <brianh@kheldar.apana.org.au> and Richard Levitte]
1673 *) Rewrite apps to use NCONF routines instead of the old CONF. New functions
1674 to support NCONF routines in extension code. New function CONF_set_nconf()
1675 to allow functions which take an NCONF to also handle the old LHASH
1676 structure: this means that the old CONF compatible routines can be
1677 retained (in particular wrt extensions) without having to duplicate the
1678 code. New function X509V3_add_ext_nconf_sk to add extensions to a stack.
1681 *) Enhance the general user interface with mechanisms for inner control
1682 and with possibilities to have yes/no kind of prompts.
1685 *) Change all calls to low level digest routines in the library and
1686 applications to use EVP. Add missing calls to HMAC_cleanup() and
1687 don't assume HMAC_CTX can be copied using memcpy().
1688 [Verdon Walker <VWalker@novell.com>, Steve Henson]
1690 *) Add the possibility to control engines through control names but with
1691 arbitrary arguments instead of just a string.
1692 Change the key loaders to take a UI_METHOD instead of a callback
1693 function pointer. NOTE: this breaks binary compatibility with earlier
1694 versions of OpenSSL [engine].
1695 Adapt the nCipher code for these new conditions and add a card insertion
1699 *) Enhance the general user interface with mechanisms to better support
1700 dialog box interfaces, application-defined prompts, the possibility
1701 to use defaults (for example default passwords from somewhere else)
1702 and interrupts/cancellations.
1705 *) Tidy up PKCS#12 attribute handling. Add support for the CSP name
1706 attribute in PKCS#12 files, add new -CSP option to pkcs12 utility.
1709 *) Fix a memory leak in 'sk_dup()' in the case reallocation fails. (Also
1710 tidy up some unnecessarily weird code in 'sk_new()').
1711 [Geoff, reported by Diego Tartara <dtartara@novamens.com>]
1713 *) Change the key loading routines for ENGINEs to use the same kind
1714 callback (pem_password_cb) as all other routines that need this
1718 *) Increase ENTROPY_NEEDED to 32 bytes, as Rijndael can operate with
1719 256 bit (=32 byte) keys. Of course seeding with more entropy bytes
1720 than this minimum value is recommended.
1723 *) New random seeder for OpenVMS, using the system process statistics
1724 that are easily reachable.
1727 *) Windows apparently can't transparently handle global
1728 variables defined in DLLs. Initialisations such as:
1730 const ASN1_ITEM *it = &ASN1_INTEGER_it;
1732 wont compile. This is used by the any applications that need to
1733 declare their own ASN1 modules. This was fixed by adding the option
1734 EXPORT_VAR_AS_FN to all Win32 platforms, although this isn't strictly
1735 needed for static libraries under Win32.
1738 *) New functions X509_PURPOSE_set() and X509_TRUST_set() to handle
1739 setting of purpose and trust fields. New X509_STORE trust and
1740 purpose functions and tidy up setting in other SSL functions.
1743 *) Add copies of X509_STORE_CTX fields and callbacks to X509_STORE
1744 structure. These are inherited by X509_STORE_CTX when it is
1745 initialised. This allows various defaults to be set in the
1746 X509_STORE structure (such as flags for CRL checking and custom
1747 purpose or trust settings) for functions which only use X509_STORE_CTX
1748 internally such as S/MIME.
1750 Modify X509_STORE_CTX_purpose_inherit() so it only sets purposes and
1751 trust settings if they are not set in X509_STORE. This allows X509_STORE
1752 purposes and trust (in S/MIME for example) to override any set by default.
1754 Add command line options for CRL checking to smime, s_client and s_server
1758 *) Initial CRL based revocation checking. If the CRL checking flag(s)
1759 are set then the CRL is looked up in the X509_STORE structure and
1760 its validity and signature checked, then if the certificate is found
1761 in the CRL the verify fails with a revoked error.
1763 Various new CRL related callbacks added to X509_STORE_CTX structure.
1765 Command line options added to 'verify' application to support this.
1767 This needs some additional work, such as being able to handle multiple
1768 CRLs with different times, extension based lookup (rather than just
1769 by subject name) and ultimately more complete V2 CRL extension
1773 *) Add a general user interface API (crypto/ui/). This is designed
1774 to replace things like des_read_password and friends (backward
1775 compatibility functions using this new API are provided).
1776 The purpose is to remove prompting functions from the DES code
1777 section as well as provide for prompting through dialog boxes in
1778 a window system and the like.
1781 *) Add "ex_data" support to ENGINE so implementations can add state at a
1782 per-structure level rather than having to store it globally.
1785 *) Make it possible for ENGINE structures to be copied when retrieved by
1786 ENGINE_by_id() if the ENGINE specifies a new flag: ENGINE_FLAGS_BY_ID_COPY.
1787 This causes the "original" ENGINE structure to act like a template,
1788 analogous to the RSA vs. RSA_METHOD type of separation. Because of this
1789 operational state can be localised to each ENGINE structure, despite the
1790 fact they all share the same "methods". New ENGINE structures returned in
1791 this case have no functional references and the return value is the single
1792 structural reference. This matches the single structural reference returned
1793 by ENGINE_by_id() normally, when it is incremented on the pre-existing
1797 *) Fix ASN1 decoder when decoding type ANY and V_ASN1_OTHER: since this
1798 needs to match any other type at all we need to manually clear the
1802 *) Changes to the "openssl engine" utility to include;
1803 - verbosity levels ('-v', '-vv', and '-vvv') that provide information
1804 about an ENGINE's available control commands.
1805 - executing control commands from command line arguments using the
1806 '-pre' and '-post' switches. '-post' is only used if '-t' is
1807 specified and the ENGINE is successfully initialised. The syntax for
1808 the individual commands are colon-separated, for example;
1809 openssl engine chil -pre FORK_CHECK:0 -pre SO_PATH:/lib/test.so
1812 *) New dynamic control command support for ENGINEs. ENGINEs can now
1813 declare their own commands (numbers), names (strings), descriptions,
1814 and input types for run-time discovery by calling applications. A
1815 subset of these commands are implicitly classed as "executable"
1816 depending on their input type, and only these can be invoked through
1817 the new string-based API function ENGINE_ctrl_cmd_string(). (Eg. this
1818 can be based on user input, config files, etc). The distinction is
1819 that "executable" commands cannot return anything other than a boolean
1820 result and can only support numeric or string input, whereas some
1821 discoverable commands may only be for direct use through
1822 ENGINE_ctrl(), eg. supporting the exchange of binary data, function
1823 pointers, or other custom uses. The "executable" commands are to
1824 support parameterisations of ENGINE behaviour that can be
1825 unambiguously defined by ENGINEs and used consistently across any
1826 OpenSSL-based application. Commands have been added to all the
1827 existing hardware-supporting ENGINEs, noticeably "SO_PATH" to allow
1828 control over shared-library paths without source code alterations.
1831 *) Changed all ENGINE implementations to dynamically allocate their
1832 ENGINEs rather than declaring them statically. Apart from this being
1833 necessary with the removal of the ENGINE_FLAGS_MALLOCED distinction,
1834 this also allows the implementations to compile without using the
1835 internal engine_int.h header.
1838 *) Minor adjustment to "rand" code. RAND_get_rand_method() now returns a
1839 'const' value. Any code that should be able to modify a RAND_METHOD
1840 should already have non-const pointers to it (ie. they should only
1841 modify their own ones).
1844 *) Made a variety of little tweaks to the ENGINE code.
1845 - "atalla" and "ubsec" string definitions were moved from header files
1846 to C code. "nuron" string definitions were placed in variables
1847 rather than hard-coded - allowing parameterisation of these values
1848 later on via ctrl() commands.
1849 - Removed unused "#if 0"'d code.
1850 - Fixed engine list iteration code so it uses ENGINE_free() to release
1851 structural references.
1852 - Constified the RAND_METHOD element of ENGINE structures.
1853 - Constified various get/set functions as appropriate and added
1854 missing functions (including a catch-all ENGINE_cpy that duplicates
1855 all ENGINE values onto a new ENGINE except reference counts/state).
1856 - Removed NULL parameter checks in get/set functions. Setting a method
1857 or function to NULL is a way of cancelling out a previously set
1858 value. Passing a NULL ENGINE parameter is just plain stupid anyway
1859 and doesn't justify the extra error symbols and code.
1860 - Deprecate the ENGINE_FLAGS_MALLOCED define and move the area for
1861 flags from engine_int.h to engine.h.
1862 - Changed prototypes for ENGINE handler functions (init(), finish(),
1863 ctrl(), key-load functions, etc) to take an (ENGINE*) parameter.
1866 *) Implement binary inversion algorithm for BN_mod_inverse in addition
1867 to the algorithm using long division. The binary algorithm can be
1868 used only if the modulus is odd. On 32-bit systems, it is faster
1869 only for relatively small moduli (roughly 20-30% for 128-bit moduli,
1870 roughly 5-15% for 256-bit moduli), so we use it only for moduli
1871 up to 450 bits. In 64-bit environments, the binary algorithm
1872 appears to be advantageous for much longer moduli; here we use it
1873 for moduli up to 2048 bits.
1876 *) Rewrite CHOICE field setting in ASN1_item_ex_d2i(). The old code
1877 could not support the combine flag in choice fields.
1880 *) Add a 'copy_extensions' option to the 'ca' utility. This copies
1881 extensions from a certificate request to the certificate.
1884 *) Allow multiple 'certopt' and 'nameopt' options to be separated
1885 by commas. Add 'namopt' and 'certopt' options to the 'ca' config
1886 file: this allows the display of the certificate about to be
1887 signed to be customised, to allow certain fields to be included
1888 or excluded and extension details. The old system didn't display
1889 multicharacter strings properly, omitted fields not in the policy
1890 and couldn't display additional details such as extensions.
1893 *) Function EC_POINTs_mul for multiple scalar multiplication
1894 of an arbitrary number of elliptic curve points
1895 \sum scalars[i]*points[i],
1896 optionally including the generator defined for the EC_GROUP:
1897 scalar*generator + \sum scalars[i]*points[i].
1899 EC_POINT_mul is a simple wrapper function for the typical case
1900 that the point list has just one item (besides the optional
1904 *) First EC_METHODs for curves over GF(p):
1906 EC_GFp_simple_method() uses the basic BN_mod_mul and BN_mod_sqr
1907 operations and provides various method functions that can also
1908 operate with faster implementations of modular arithmetic.
1910 EC_GFp_mont_method() reuses most functions that are part of
1911 EC_GFp_simple_method, but uses Montgomery arithmetic.
1913 [Bodo Moeller; point addition and point doubling
1914 implementation directly derived from source code provided by
1915 Lenka Fibikova <fibikova@exp-math.uni-essen.de>]
1917 *) Framework for elliptic curves (crypto/ec/ec.h, crypto/ec/ec_lcl.h,
1918 crypto/ec/ec_lib.c):
1920 Curves are EC_GROUP objects (with an optional group generator)
1921 based on EC_METHODs that are built into the library.
1923 Points are EC_POINT objects based on EC_GROUP objects.
1925 Most of the framework would be able to handle curves over arbitrary
1926 finite fields, but as there are no obvious types for fields other
1927 than GF(p), some functions are limited to that for now.
1930 *) Add the -HTTP option to s_server. It is similar to -WWW, but requires
1931 that the file contains a complete HTTP response.
1934 *) Add the ec directory to mkdef.pl and mkfiles.pl. In mkdef.pl
1935 change the def and num file printf format specifier from "%-40sXXX"
1936 to "%-39s XXX". The latter will always guarantee a space after the
1937 field while the former will cause them to run together if the field
1938 is 40 of more characters long.
1941 *) Constify the cipher and digest 'method' functions and structures
1942 and modify related functions to take constant EVP_MD and EVP_CIPHER
1946 *) Hide BN_CTX structure details in bn_lcl.h instead of publishing them
1947 in <openssl/bn.h>. Also further increase BN_CTX_NUM to 32.
1950 *) Modify EVP_Digest*() routines so they now return values. Although the
1951 internal software routines can never fail additional hardware versions
1955 *) Clean up crypto/err/err.h and change some error codes to avoid conflicts:
1957 Previously ERR_R_FATAL was too small and coincided with ERR_LIB_PKCS7
1958 (= ERR_R_PKCS7_LIB); it is now 64 instead of 32.
1961 ERR_R_NESTED_ASN1_ERROR
1963 ERR_R_MISSING_ASN1_EOS
1964 were 4 .. 9, conflicting with
1965 ERR_LIB_RSA (= ERR_R_RSA_LIB)
1967 ERR_LIB_PEM (= ERR_R_PEM_LIB).
1968 They are now 58 .. 63 (i.e., just below ERR_R_FATAL).
1970 Add new error code 'ERR_R_INTERNAL_ERROR'.
1973 *) Don't overuse locks in crypto/err/err.c: For data retrieval, CRYPTO_r_lock
1977 *) New option '-subj arg' for 'openssl req' and 'openssl ca'. This
1978 sets the subject name for a new request or supersedes the
1979 subject name in a given request. Formats that can be parsed are
1980 'CN=Some Name, OU=myOU, C=IT'
1982 'CN=Some Name/OU=myOU/C=IT'.
1984 Add options '-batch' and '-verbose' to 'openssl req'.
1985 [Massimiliano Pala <madwolf@hackmasters.net>]
1987 *) Introduce the possibility to access global variables through
1988 functions on platform were that's the best way to handle exporting
1989 global variables in shared libraries. To enable this functionality,
1990 one must configure with "EXPORT_VAR_AS_FN" or defined the C macro
1991 "OPENSSL_EXPORT_VAR_AS_FUNCTION" in crypto/opensslconf.h (the latter
1992 is normally done by Configure or something similar).
1994 To implement a global variable, use the macro OPENSSL_IMPLEMENT_GLOBAL
1995 in the source file (foo.c) like this:
1997 OPENSSL_IMPLEMENT_GLOBAL(int,foo)=1;
1998 OPENSSL_IMPLEMENT_GLOBAL(double,bar);
2000 To declare a global variable, use the macros OPENSSL_DECLARE_GLOBAL
2001 and OPENSSL_GLOBAL_REF in the header file (foo.h) like this:
2003 OPENSSL_DECLARE_GLOBAL(int,foo);
2004 #define foo OPENSSL_GLOBAL_REF(foo)
2005 OPENSSL_DECLARE_GLOBAL(double,bar);
2006 #define bar OPENSSL_GLOBAL_REF(bar)
2008 The #defines are very important, and therefore so is including the
2009 header file everywhere where the defined globals are used.
2011 The macro OPENSSL_EXPORT_VAR_AS_FUNCTION also affects the definition
2012 of ASN.1 items, but that structure is a bit different.
2014 The largest change is in util/mkdef.pl which has been enhanced with
2015 better and easier to understand logic to choose which symbols should
2016 go into the Windows .def files as well as a number of fixes and code
2017 cleanup (among others, algorithm keywords are now sorted
2018 lexicographically to avoid constant rewrites).
2021 *) In BN_div() keep a copy of the sign of 'num' before writing the
2022 result to 'rm' because if rm==num the value will be overwritten
2023 and produce the wrong result if 'num' is negative: this caused
2024 problems with BN_mod() and BN_nnmod().
2027 *) Function OCSP_request_verify(). This checks the signature on an
2028 OCSP request and verifies the signer certificate. The signer
2029 certificate is just checked for a generic purpose and OCSP request
2033 *) Add OCSP_check_validity() function to check the validity of OCSP
2034 responses. OCSP responses are prepared in real time and may only
2035 be a few seconds old. Simply checking that the current time lies
2036 between thisUpdate and nextUpdate max reject otherwise valid responses
2037 caused by either OCSP responder or client clock inaccuracy. Instead
2038 we allow thisUpdate and nextUpdate to fall within a certain period of
2039 the current time. The age of the response can also optionally be
2040 checked. Two new options -validity_period and -status_age added to
2044 *) If signature or public key algorithm is unrecognized print out its
2045 OID rather that just UNKNOWN.
2048 *) Change OCSP_cert_to_id() to tolerate a NULL subject certificate and
2049 OCSP_cert_id_new() a NULL serialNumber. This allows a partial certificate
2050 ID to be generated from the issuer certificate alone which can then be
2051 passed to OCSP_id_issuer_cmp().
2054 *) New compilation option ASN1_ITEM_FUNCTIONS. This causes the new
2055 ASN1 modules to export functions returning ASN1_ITEM pointers
2056 instead of the ASN1_ITEM structures themselves. This adds several
2057 new macros which allow the underlying ASN1 function/structure to
2058 be accessed transparently. As a result code should not use ASN1_ITEM
2059 references directly (such as &X509_it) but instead use the relevant
2060 macros (such as ASN1_ITEM_rptr(X509)). This option is to allow
2061 use of the new ASN1 code on platforms where exporting structures
2062 is problematical (for example in shared libraries) but exporting
2063 functions returning pointers to structures is not.
2066 *) Add support for overriding the generation of SSL/TLS session IDs.
2067 These callbacks can be registered either in an SSL_CTX or per SSL.
2068 The purpose of this is to allow applications to control, if they wish,
2069 the arbitrary values chosen for use as session IDs, particularly as it
2070 can be useful for session caching in multiple-server environments. A
2071 command-line switch for testing this (and any client code that wishes
2072 to use such a feature) has been added to "s_server".
2073 [Geoff Thorpe, Lutz Jaenicke]
2075 *) Modify mkdef.pl to recognise and parse preprocessor conditionals
2076 of the form '#if defined(...) || defined(...) || ...' and
2077 '#if !defined(...) && !defined(...) && ...'. This also avoids
2078 the growing number of special cases it was previously handling.
2081 *) Make all configuration macros available for application by making
2082 sure they are available in opensslconf.h, by giving them names starting
2083 with "OPENSSL_" to avoid conflicts with other packages and by making
2084 sure e_os2.h will cover all platform-specific cases together with
2086 Additionally, it is now possible to define configuration/platform-
2087 specific names (called "system identities"). In the C code, these
2088 are prefixed with "OPENSSL_SYSNAME_". e_os2.h will create another
2089 macro with the name beginning with "OPENSSL_SYS_", which is determined
2090 from "OPENSSL_SYSNAME_*" or compiler-specific macros depending on
2094 *) New option -set_serial to 'req' and 'x509' this allows the serial
2095 number to use to be specified on the command line. Previously self
2096 signed certificates were hard coded with serial number 0 and the
2097 CA options of 'x509' had to use a serial number in a file which was
2101 *) New options to 'ca' utility to support V2 CRL entry extensions.
2102 Currently CRL reason, invalidity date and hold instruction are
2103 supported. Add new CRL extensions to V3 code and some new objects.
2106 *) New function EVP_CIPHER_CTX_set_padding() this is used to
2107 disable standard block padding (aka PKCS#5 padding) in the EVP
2108 API, which was previously mandatory. This means that the data is
2109 not padded in any way and so the total length much be a multiple
2110 of the block size, otherwise an error occurs.
2113 *) Initial (incomplete) OCSP SSL support.
2116 *) New function OCSP_parse_url(). This splits up a URL into its host,
2117 port and path components: primarily to parse OCSP URLs. New -url
2118 option to ocsp utility.
2121 *) New nonce behavior. The return value of OCSP_check_nonce() now
2122 reflects the various checks performed. Applications can decide
2123 whether to tolerate certain situations such as an absent nonce
2124 in a response when one was present in a request: the ocsp application
2125 just prints out a warning. New function OCSP_add1_basic_nonce()
2126 this is to allow responders to include a nonce in a response even if
2127 the request is nonce-less.
2130 *) Disable stdin buffering in load_cert (apps/apps.c) so that no certs are
2131 skipped when using openssl x509 multiple times on a single input file,
2132 e.g. "(openssl x509 -out cert1; openssl x509 -out cert2) <certs".
2135 *) Make ASN1_UTCTIME_set_string() and ASN1_GENERALIZEDTIME_set_string()
2136 set string type: to handle setting ASN1_TIME structures. Fix ca
2137 utility to correctly initialize revocation date of CRLs.
2140 *) New option SSL_OP_CIPHER_SERVER_PREFERENCE allows the server to override
2141 the clients preferred ciphersuites and rather use its own preferences.
2142 Should help to work around M$ SGC (Server Gated Cryptography) bug in
2143 Internet Explorer by ensuring unchanged hash method during stepup.
2144 (Also replaces the broken/deactivated SSL_OP_NON_EXPORT_FIRST option.)
2147 *) Make mkdef.pl recognise all DECLARE_ASN1 macros, change rijndael
2148 to aes and add a new 'exist' option to print out symbols that don't
2152 *) Additional options to ocsp utility to allow flags to be set and
2153 additional certificates supplied.
2156 *) Add the option -VAfile to 'openssl ocsp', so the user can give the
2157 OCSP client a number of certificate to only verify the response
2161 *) Update Rijndael code to version 3.0 and change EVP AES ciphers to
2162 handle the new API. Currently only ECB, CBC modes supported. Add new
2165 Add TLS AES ciphersuites as described in RFC3268, "Advanced
2166 Encryption Standard (AES) Ciphersuites for Transport Layer
2167 Security (TLS)". (In beta versions of OpenSSL 0.9.7, these were
2168 not enabled by default and were not part of the "ALL" ciphersuite
2169 alias because they were not yet official; they could be
2170 explicitly requested by specifying the "AESdraft" ciphersuite
2171 group alias. In the final release of OpenSSL 0.9.7, the group
2172 alias is called "AES" and is part of "ALL".)
2173 [Ben Laurie, Steve Henson, Bodo Moeller]
2175 *) New function OCSP_copy_nonce() to copy nonce value (if present) from
2176 request to response.
2179 *) Functions for OCSP responders. OCSP_request_onereq_count(),
2180 OCSP_request_onereq_get0(), OCSP_onereq_get0_id() and OCSP_id_get0_info()
2181 extract information from a certificate request. OCSP_response_create()
2182 creates a response and optionally adds a basic response structure.
2183 OCSP_basic_add1_status() adds a complete single response to a basic
2184 response and returns the OCSP_SINGLERESP structure just added (to allow
2185 extensions to be included for example). OCSP_basic_add1_cert() adds a
2186 certificate to a basic response and OCSP_basic_sign() signs a basic
2187 response with various flags. New helper functions ASN1_TIME_check()
2188 (checks validity of ASN1_TIME structure) and ASN1_TIME_to_generalizedtime()
2189 (converts ASN1_TIME to GeneralizedTime).
2192 *) Various new functions. EVP_Digest() combines EVP_Digest{Init,Update,Final}()
2193 in a single operation. X509_get0_pubkey_bitstr() extracts the public_key
2194 structure from a certificate. X509_pubkey_digest() digests the public_key
2195 contents: this is used in various key identifiers.
2198 *) Make sk_sort() tolerate a NULL argument.
2199 [Steve Henson reported by Massimiliano Pala <madwolf@comune.modena.it>]
2201 *) New OCSP verify flag OCSP_TRUSTOTHER. When set the "other" certificates
2202 passed by the function are trusted implicitly. If any of them signed the
2203 response then it is assumed to be valid and is not verified.
2206 *) In PKCS7_set_type() initialise content_type in PKCS7_ENC_CONTENT
2207 to data. This was previously part of the PKCS7 ASN1 code. This
2208 was causing problems with OpenSSL created PKCS#12 and PKCS#7 structures.
2209 [Steve Henson, reported by Kenneth R. Robinette
2210 <support@securenetterm.com>]
2212 *) Add CRYPTO_push_info() and CRYPTO_pop_info() calls to new ASN1
2213 routines: without these tracing memory leaks is very painful.
2214 Fix leaks in PKCS12 and PKCS7 routines.
2217 *) Make X509_time_adj() cope with the new behaviour of ASN1_TIME_new().
2218 Previously it initialised the 'type' argument to V_ASN1_UTCTIME which
2219 effectively meant GeneralizedTime would never be used. Now it
2220 is initialised to -1 but X509_time_adj() now has to check the value
2221 and use ASN1_TIME_set() if the value is not V_ASN1_UTCTIME or
2222 V_ASN1_GENERALIZEDTIME, without this it always uses GeneralizedTime.
2223 [Steve Henson, reported by Kenneth R. Robinette
2224 <support@securenetterm.com>]
2226 *) Fixes to BN_to_ASN1_INTEGER when bn is zero. This would previously
2227 result in a zero length in the ASN1_INTEGER structure which was
2228 not consistent with the structure when d2i_ASN1_INTEGER() was used
2229 and would cause ASN1_INTEGER_cmp() to fail. Enhance s2i_ASN1_INTEGER()
2230 to cope with hex and negative integers. Fix bug in i2a_ASN1_INTEGER()
2231 where it did not print out a minus for negative ASN1_INTEGER.
2234 *) Add summary printout to ocsp utility. The various functions which
2235 convert status values to strings have been renamed to:
2236 OCSP_response_status_str(), OCSP_cert_status_str() and
2237 OCSP_crl_reason_str() and are no longer static. New options
2238 to verify nonce values and to disable verification. OCSP response
2239 printout format cleaned up.
2242 *) Add additional OCSP certificate checks. These are those specified
2243 in RFC2560. This consists of two separate checks: the CA of the
2244 certificate being checked must either be the OCSP signer certificate
2245 or the issuer of the OCSP signer certificate. In the latter case the
2246 OCSP signer certificate must contain the OCSP signing extended key
2247 usage. This check is performed by attempting to match the OCSP
2248 signer or the OCSP signer CA to the issuerNameHash and issuerKeyHash
2249 in the OCSP_CERTID structures of the response.
2252 *) Initial OCSP certificate verification added to OCSP_basic_verify()
2253 and related routines. This uses the standard OpenSSL certificate
2254 verify routines to perform initial checks (just CA validity) and
2255 to obtain the certificate chain. Then additional checks will be
2256 performed on the chain. Currently the root CA is checked to see
2257 if it is explicitly trusted for OCSP signing. This is used to set
2258 a root CA as a global signing root: that is any certificate that
2259 chains to that CA is an acceptable OCSP signing certificate.
2262 *) New '-extfile ...' option to 'openssl ca' for reading X.509v3
2263 extensions from a separate configuration file.
2264 As when reading extensions from the main configuration file,
2265 the '-extensions ...' option may be used for specifying the
2267 [Massimiliano Pala <madwolf@comune.modena.it>]
2269 *) New OCSP utility. Allows OCSP requests to be generated or
2270 read. The request can be sent to a responder and the output
2271 parsed, outputed or printed in text form. Not complete yet:
2272 still needs to check the OCSP response validity.
2275 *) New subcommands for 'openssl ca':
2276 'openssl ca -status <serial>' prints the status of the cert with
2277 the given serial number (according to the index file).
2278 'openssl ca -updatedb' updates the expiry status of certificates
2280 [Massimiliano Pala <madwolf@comune.modena.it>]
2282 *) New '-newreq-nodes' command option to CA.pl. This is like
2283 '-newreq', but calls 'openssl req' with the '-nodes' option
2284 so that the resulting key is not encrypted.
2285 [Damien Miller <djm@mindrot.org>]
2287 *) New configuration for the GNU Hurd.
2288 [Jonathan Bartlett <johnnyb@wolfram.com> via Richard Levitte]
2290 *) Initial code to implement OCSP basic response verify. This
2291 is currently incomplete. Currently just finds the signer's
2292 certificate and verifies the signature on the response.
2295 *) New SSLeay_version code SSLEAY_DIR to determine the compiled-in
2296 value of OPENSSLDIR. This is available via the new '-d' option
2297 to 'openssl version', and is also included in 'openssl version -a'.
2300 *) Allowing defining memory allocation callbacks that will be given
2301 file name and line number information in additional arguments
2302 (a const char* and an int). The basic functionality remains, as
2303 well as the original possibility to just replace malloc(),
2304 realloc() and free() by functions that do not know about these
2305 additional arguments. To register and find out the current
2306 settings for extended allocation functions, the following
2307 functions are provided:
2309 CRYPTO_set_mem_ex_functions
2310 CRYPTO_set_locked_mem_ex_functions
2311 CRYPTO_get_mem_ex_functions
2312 CRYPTO_get_locked_mem_ex_functions
2314 These work the same way as CRYPTO_set_mem_functions and friends.
2315 CRYPTO_get_[locked_]mem_functions now writes 0 where such an
2316 extended allocation function is enabled.
2317 Similarly, CRYPTO_get_[locked_]mem_ex_functions writes 0 where
2318 a conventional allocation function is enabled.
2319 [Richard Levitte, Bodo Moeller]
2321 *) Finish off removing the remaining LHASH function pointer casts.
2322 There should no longer be any prototype-casting required when using
2323 the LHASH abstraction, and any casts that remain are "bugs". See
2324 the callback types and macros at the head of lhash.h for details
2325 (and "OBJ_cleanup" in crypto/objects/obj_dat.c as an example).
2328 *) Add automatic query of EGD sockets in RAND_poll() for the unix variant.
2329 If /dev/[u]random devices are not available or do not return enough
2330 entropy, EGD style sockets (served by EGD or PRNGD) will automatically
2332 The locations /var/run/egd-pool, /dev/egd-pool, /etc/egd-pool, and
2333 /etc/entropy will be queried once each in this sequence, quering stops
2334 when enough entropy was collected without querying more sockets.
2337 *) Change the Unix RAND_poll() variant to be able to poll several
2338 random devices, as specified by DEVRANDOM, until a sufficient amount
2339 of data has been collected. We spend at most 10 ms on each file
2340 (select timeout) and read in non-blocking mode. DEVRANDOM now
2341 defaults to the list "/dev/urandom", "/dev/random", "/dev/srandom"
2342 (previously it was just the string "/dev/urandom"), so on typical
2343 platforms the 10 ms delay will never occur.
2344 Also separate out the Unix variant to its own file, rand_unix.c.
2345 For VMS, there's a currently-empty rand_vms.c.
2348 *) Move OCSP client related routines to ocsp_cl.c. These
2349 provide utility functions which an application needing
2350 to issue a request to an OCSP responder and analyse the
2351 response will typically need: as opposed to those which an
2352 OCSP responder itself would need which will be added later.
2354 OCSP_request_sign() signs an OCSP request with an API similar
2355 to PKCS7_sign(). OCSP_response_status() returns status of OCSP
2356 response. OCSP_response_get1_basic() extracts basic response
2357 from response. OCSP_resp_find_status(): finds and extracts status
2358 information from an OCSP_CERTID structure (which will be created
2359 when the request structure is built). These are built from lower
2360 level functions which work on OCSP_SINGLERESP structures but
2361 wont normally be used unless the application wishes to examine
2362 extensions in the OCSP response for example.
2364 Replace nonce routines with a pair of functions.
2365 OCSP_request_add1_nonce() adds a nonce value and optionally
2366 generates a random value. OCSP_check_nonce() checks the
2367 validity of the nonce in an OCSP response.
2370 *) Change function OCSP_request_add() to OCSP_request_add0_id().
2371 This doesn't copy the supplied OCSP_CERTID and avoids the
2372 need to free up the newly created id. Change return type
2373 to OCSP_ONEREQ to return the internal OCSP_ONEREQ structure.
2374 This can then be used to add extensions to the request.
2375 Deleted OCSP_request_new(), since most of its functionality
2376 is now in OCSP_REQUEST_new() (and the case insensitive name
2377 clash) apart from the ability to set the request name which
2378 will be added elsewhere.
2381 *) Update OCSP API. Remove obsolete extensions argument from
2382 various functions. Extensions are now handled using the new
2383 OCSP extension code. New simple OCSP HTTP function which
2384 can be used to send requests and parse the response.
2387 *) Fix the PKCS#7 (S/MIME) code to work with new ASN1. Two new
2388 ASN1_ITEM structures help with sign and verify. PKCS7_ATTR_SIGN
2389 uses the special reorder version of SET OF to sort the attributes
2390 and reorder them to match the encoded order. This resolves a long
2391 standing problem: a verify on a PKCS7 structure just after signing
2392 it used to fail because the attribute order did not match the
2393 encoded order. PKCS7_ATTR_VERIFY does not reorder the attributes:
2394 it uses the received order. This is necessary to tolerate some broken
2395 software that does not order SET OF. This is handled by encoding
2396 as a SEQUENCE OF but using implicit tagging (with UNIVERSAL class)
2397 to produce the required SET OF.
2400 *) Have mk1mf.pl generate the macros OPENSSL_BUILD_SHLIBCRYPTO and
2401 OPENSSL_BUILD_SHLIBSSL and use them appropriately in the header
2402 files to get correct declarations of the ASN.1 item variables.
2405 *) Rewrite of PKCS#12 code to use new ASN1 functionality. Replace many
2406 PKCS#12 macros with real functions. Fix two unrelated ASN1 bugs:
2407 asn1_check_tlen() would sometimes attempt to use 'ctx' when it was
2408 NULL and ASN1_TYPE was not dereferenced properly in asn1_ex_c2i().
2409 New ASN1 macro: DECLARE_ASN1_ITEM() which just declares the relevant
2410 ASN1_ITEM and no wrapper functions.
2413 *) New functions or ASN1_item_d2i_fp() and ASN1_item_d2i_bio(). These
2414 replace the old function pointer based I/O routines. Change most of
2415 the *_d2i_bio() and *_d2i_fp() functions to use these.
2418 *) Enhance mkdef.pl to be more accepting about spacing in C preprocessor
2419 lines, recognice more "algorithms" that can be deselected, and make
2420 it complain about algorithm deselection that isn't recognised.
2423 *) New ASN1 functions to handle dup, sign, verify, digest, pack and
2424 unpack operations in terms of ASN1_ITEM. Modify existing wrappers
2425 to use new functions. Add NO_ASN1_OLD which can be set to remove
2426 some old style ASN1 functions: this can be used to determine if old
2427 code will still work when these eventually go away.
2430 *) New extension functions for OCSP structures, these follow the
2431 same conventions as certificates and CRLs.
2434 *) New function X509V3_add1_i2d(). This automatically encodes and
2435 adds an extension. Its behaviour can be customised with various
2436 flags to append, replace or delete. Various wrappers added for
2437 certifcates and CRLs.
2440 *) Fix to avoid calling the underlying ASN1 print routine when
2441 an extension cannot be parsed. Correct a typo in the
2442 OCSP_SERVICELOC extension. Tidy up print OCSP format.
2445 *) Make mkdef.pl parse some of the ASN1 macros and add apropriate
2446 entries for variables.
2449 *) Add functionality to apps/openssl.c for detecting locking
2450 problems: As the program is single-threaded, all we have
2451 to do is register a locking callback using an array for
2452 storing which locks are currently held by the program.
2455 *) Use a lock around the call to CRYPTO_get_ex_new_index() in
2456 SSL_get_ex_data_X509_STORE_idx(), which is used in
2457 ssl_verify_cert_chain() and thus can be called at any time
2458 during TLS/SSL handshakes so that thread-safety is essential.
2459 Unfortunately, the ex_data design is not at all suited
2460 for multi-threaded use, so it probably should be abolished.
2463 *) Added Broadcom "ubsec" ENGINE to OpenSSL.
2464 [Broadcom, tweaked and integrated by Geoff Thorpe]
2466 *) Move common extension printing code to new function
2467 X509V3_print_extensions(). Reorganise OCSP print routines and
2468 implement some needed OCSP ASN1 functions. Add OCSP extensions.
2471 *) New function X509_signature_print() to remove duplication in some
2475 *) Add a special meaning when SET OF and SEQUENCE OF flags are both
2476 set (this was treated exactly the same as SET OF previously). This
2477 is used to reorder the STACK representing the structure to match the
2478 encoding. This will be used to get round a problem where a PKCS7
2479 structure which was signed could not be verified because the STACK
2480 order did not reflect the encoded order.
2483 *) Reimplement the OCSP ASN1 module using the new code.
2486 *) Update the X509V3 code to permit the use of an ASN1_ITEM structure
2487 for its ASN1 operations. The old style function pointers still exist
2488 for now but they will eventually go away.
2491 *) Merge in replacement ASN1 code from the ASN1 branch. This almost
2492 completely replaces the old ASN1 functionality with a table driven
2493 encoder and decoder which interprets an ASN1_ITEM structure describing
2494 the ASN1 module. Compatibility with the existing ASN1 API (i2d,d2i) is
2495 largely maintained. Almost all of the old asn1_mac.h macro based ASN1
2496 has also been converted to the new form.
2499 *) Change BN_mod_exp_recp so that negative moduli are tolerated
2500 (the sign is ignored). Similarly, ignore the sign in BN_MONT_CTX_set
2501 so that BN_mod_exp_mont and BN_mod_exp_mont_word work
2502 for negative moduli.
2505 *) Fix BN_uadd and BN_usub: Always return non-negative results instead
2506 of not touching the result's sign bit.
2509 *) BN_div bugfix: If the result is 0, the sign (res->neg) must not be
2513 *) Changed the LHASH code to use prototypes for callbacks, and created
2514 macros to declare and implement thin (optionally static) functions
2515 that provide type-safety and avoid function pointer casting for the
2516 type-specific callbacks.
2519 *) Added Kerberos Cipher Suites to be used with TLS, as written in
2521 [Veers Staats <staatsvr@asc.hpc.mil>,
2522 Jeffrey Altman <jaltman@columbia.edu>, via Richard Levitte]
2524 *) Reformat the FAQ so the different questions and answers can be divided
2525 in sections depending on the subject.
2528 *) Have the zlib compression code load ZLIB.DLL dynamically under
2532 *) New function BN_mod_sqrt for computing square roots modulo a prime
2533 (using the probabilistic Tonelli-Shanks algorithm unless
2534 p == 3 (mod 4) or p == 5 (mod 8), which are cases that can
2535 be handled deterministically).
2536 [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller]
2538 *) Make BN_mod_inverse faster by explicitly handling small quotients
2539 in the Euclid loop. (Speed gain about 20% for small moduli [256 or
2540 512 bits], about 30% for larger ones [1024 or 2048 bits].)
2543 *) New function BN_kronecker.
2546 *) Fix BN_gcd so that it works on negative inputs; the result is
2547 positive unless both parameters are zero.
2548 Previously something reasonably close to an infinite loop was
2549 possible because numbers could be growing instead of shrinking
2550 in the implementation of Euclid's algorithm.
2553 *) Fix BN_is_word() and BN_is_one() macros to take into account the
2554 sign of the number in question.
2556 Fix BN_is_word(a,w) to work correctly for w == 0.
2558 The old BN_is_word(a,w) macro is now called BN_abs_is_word(a,w)
2559 because its test if the absolute value of 'a' equals 'w'.
2560 Note that BN_abs_is_word does *not* handle w == 0 reliably;
2561 it exists mostly for use in the implementations of BN_is_zero(),
2562 BN_is_one(), and BN_is_word().
2565 *) New function BN_swap.
2568 *) Use BN_nnmod instead of BN_mod in crypto/bn/bn_exp.c so that
2569 the exponentiation functions are more likely to produce reasonable
2570 results on negative inputs.
2573 *) Change BN_mod_mul so that the result is always non-negative.
2574 Previously, it could be negative if one of the factors was negative;
2575 I don't think anyone really wanted that behaviour.
2578 *) Move BN_mod_... functions into new file crypto/bn/bn_mod.c
2579 (except for exponentiation, which stays in crypto/bn/bn_exp.c,
2580 and BN_mod_mul_reciprocal, which stays in crypto/bn/bn_recp.c)
2581 and add new functions:
2590 BN_mod_lshift1_quick
2594 These functions always generate non-negative results.
2596 BN_nnmod otherwise is like BN_mod (if BN_mod computes a remainder r
2597 such that |m| < r < 0, BN_nnmod will output rem + |m| instead).
2599 BN_mod_XXX_quick(r, a, [b,] m) generates the same result as
2600 BN_mod_XXX(r, a, [b,] m, ctx), but requires that a [and b]
2601 be reduced modulo m.
2602 [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller]
2605 The following entry accidentily appeared in the CHANGES file
2606 distributed with OpenSSL 0.9.7. The modifications described in
2607 it do *not* apply to OpenSSL 0.9.7.
2609 *) Remove a few calls to bn_wexpand() in BN_sqr() (the one in there
2610 was actually never needed) and in BN_mul(). The removal in BN_mul()
2611 required a small change in bn_mul_part_recursive() and the addition
2612 of the functions bn_cmp_part_words(), bn_sub_part_words() and
2613 bn_add_part_words(), which do the same thing as bn_cmp_words(),
2614 bn_sub_words() and bn_add_words() except they take arrays with
2619 *) In 'openssl passwd', verify passwords read from the terminal
2620 unless the '-salt' option is used (which usually means that
2621 verification would just waste user's time since the resulting
2622 hash is going to be compared with some given password hash)
2623 or the new '-noverify' option is used.
2625 This is an incompatible change, but it does not affect
2626 non-interactive use of 'openssl passwd' (passwords on the command
2627 line, '-stdin' option, '-in ...' option) and thus should not
2631 *) Remove all references to RSAref, since there's no more need for it.
2634 *) Make DSO load along a path given through an environment variable
2635 (SHLIB_PATH) with shl_load().
2638 *) Constify the ENGINE code as a result of BIGNUM constification.
2639 Also constify the RSA code and most things related to it. In a
2640 few places, most notable in the depth of the ASN.1 code, ugly
2641 casts back to non-const were required (to be solved at a later
2645 *) Make it so the openssl application has all engines loaded by default.
2648 *) Constify the BIGNUM routines a little more.
2651 *) Add the following functions:
2653 ENGINE_load_cswift()
2655 ENGINE_load_atalla()
2657 ENGINE_load_builtin_engines()
2659 That way, an application can itself choose if external engines that
2660 are built-in in OpenSSL shall ever be used or not. The benefit is
2661 that applications won't have to be linked with libdl or other dso
2662 libraries unless it's really needed.
2664 Changed 'openssl engine' to load all engines on demand.
2665 Changed the engine header files to avoid the duplication of some
2666 declarations (they differed!).
2669 *) 'openssl engine' can now list capabilities.
2672 *) Better error reporting in 'openssl engine'.
2675 *) Never call load_dh_param(NULL) in s_server.
2678 *) Add engine application. It can currently list engines by name and
2679 identity, and test if they are actually available.
2682 *) Improve RPM specification file by forcing symbolic linking and making
2683 sure the installed documentation is also owned by root.root.
2684 [Damien Miller <djm@mindrot.org>]
2686 *) Give the OpenSSL applications more possibilities to make use of
2687 keys (public as well as private) handled by engines.
2690 *) Add OCSP code that comes from CertCo.
2693 *) Add VMS support for the Rijndael code.
2696 *) Added untested support for Nuron crypto accelerator.
2699 *) Add support for external cryptographic devices. This code was
2700 previously distributed separately as the "engine" branch.
2701 [Geoff Thorpe, Richard Levitte]
2703 *) Rework the filename-translation in the DSO code. It is now possible to
2704 have far greater control over how a "name" is turned into a filename
2705 depending on the operating environment and any oddities about the
2706 different shared library filenames on each system.
2709 *) Support threads on FreeBSD-elf in Configure.
2712 *) Fix for SHA1 assembly problem with MASM: it produces
2713 warnings about corrupt line number information when assembling
2714 with debugging information. This is caused by the overlapping
2716 [Bernd Matthes <mainbug@celocom.de>, Steve Henson]
2719 NCONF_get_number() has no error checking at all. As a replacement,
2720 NCONF_get_number_e() is defined (_e for "error checking") and is
2721 promoted strongly. The old NCONF_get_number is kept around for
2722 binary backward compatibility.
2723 Make it possible for methods to load from something other than a BIO,
2724 by providing a function pointer that is given a name instead of a BIO.
2725 For example, this could be used to load configuration data from an
2729 *) Fix for non blocking accept BIOs. Added new I/O special reason
2730 BIO_RR_ACCEPT to cover this case. Previously use of accept BIOs
2731 with non blocking I/O was not possible because no retry code was
2732 implemented. Also added new SSL code SSL_WANT_ACCEPT to cover
2736 *) Added the beginnings of Rijndael support.
2739 *) Fix for bug in DirectoryString mask setting. Add support for
2740 X509_NAME_print_ex() in 'req' and X509_print_ex() function
2741 to allow certificate printing to more controllable, additional
2742 'certopt' option to 'x509' to allow new printing options to be
2746 *) Clean old EAY MD5 hack from e_os.h.
2749 Changes between 0.9.6l and 0.9.6m [17 Mar 2004]
2751 *) Fix null-pointer assignment in do_change_cipher_spec() revealed
2752 by using the Codenomicon TLS Test Tool (CAN-2004-0079)
2753 [Joe Orton, Steve Henson]
2755 Changes between 0.9.6k and 0.9.6l [04 Nov 2003]
2757 *) Fix additional bug revealed by the NISCC test suite:
2759 Stop bug triggering large recursion when presented with
2760 certain ASN.1 tags (CAN-2003-0851)
2763 Changes between 0.9.6j and 0.9.6k [30 Sep 2003]
2765 *) Fix various bugs revealed by running the NISCC test suite:
2767 Stop out of bounds reads in the ASN1 code when presented with
2768 invalid tags (CAN-2003-0543 and CAN-2003-0544).
2770 If verify callback ignores invalid public key errors don't try to check
2771 certificate signature with the NULL public key.
2775 *) In ssl3_accept() (ssl/s3_srvr.c) only accept a client certificate
2776 if the server requested one: as stated in TLS 1.0 and SSL 3.0
2780 *) In ssl3_get_client_hello() (ssl/s3_srvr.c), tolerate additional
2781 extra data after the compression methods not only for TLS 1.0
2782 but also for SSL 3.0 (as required by the specification).
2783 [Bodo Moeller; problem pointed out by Matthias Loepfe]
2785 *) Change X509_certificate_type() to mark the key as exported/exportable
2786 when it's 512 *bits* long, not 512 bytes.
2789 Changes between 0.9.6i and 0.9.6j [10 Apr 2003]
2791 *) Countermeasure against the Klima-Pokorny-Rosa extension of
2792 Bleichbacher's attack on PKCS #1 v1.5 padding: treat
2793 a protocol version number mismatch like a decryption error
2794 in ssl3_get_client_key_exchange (ssl/s3_srvr.c).
2797 *) Turn on RSA blinding by default in the default implementation
2798 to avoid a timing attack. Applications that don't want it can call
2799 RSA_blinding_off() or use the new flag RSA_FLAG_NO_BLINDING.
2800 They would be ill-advised to do so in most cases.
2801 [Ben Laurie, Steve Henson, Geoff Thorpe, Bodo Moeller]
2803 *) Change RSA blinding code so that it works when the PRNG is not
2804 seeded (in this case, the secret RSA exponent is abused as
2805 an unpredictable seed -- if it is not unpredictable, there
2806 is no point in blinding anyway). Make RSA blinding thread-safe
2807 by remembering the creator's thread ID in rsa->blinding and
2808 having all other threads use local one-time blinding factors
2809 (this requires more computation than sharing rsa->blinding, but
2810 avoids excessive locking; and if an RSA object is not shared
2811 between threads, blinding will still be very fast).
2814 Changes between 0.9.6h and 0.9.6i [19 Feb 2003]
2816 *) In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked
2817 via timing by performing a MAC computation even if incorrrect
2818 block cipher padding has been found. This is a countermeasure
2819 against active attacks where the attacker has to distinguish
2820 between bad padding and a MAC verification error. (CAN-2003-0078)
2822 [Bodo Moeller; problem pointed out by Brice Canvel (EPFL),
2823 Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and
2824 Martin Vuagnoux (EPFL, Ilion)]
2826 Changes between 0.9.6g and 0.9.6h [5 Dec 2002]
2828 *) New function OPENSSL_cleanse(), which is used to cleanse a section of
2829 memory from it's contents. This is done with a counter that will
2830 place alternating values in each byte. This can be used to solve
2831 two issues: 1) the removal of calls to memset() by highly optimizing
2832 compilers, and 2) cleansing with other values than 0, since those can
2833 be read through on certain media, for example a swap space on disk.
2836 *) Bugfix: client side session caching did not work with external caching,
2837 because the session->cipher setting was not restored when reloading
2838 from the external cache. This problem was masked, when
2839 SSL_OP_NETSCAPE_REUSE_CIPHER_CHANGE_BUG (part of SSL_OP_ALL) was set.
2840 (Found by Steve Haslam <steve@araqnid.ddts.net>.)
2843 *) Fix client_certificate (ssl/s2_clnt.c): The permissible total
2844 length of the REQUEST-CERTIFICATE message is 18 .. 34, not 17 .. 33.
2845 [Zeev Lieber <zeev-l@yahoo.com>]
2847 *) Undo an undocumented change introduced in 0.9.6e which caused
2848 repeated calls to OpenSSL_add_all_ciphers() and
2849 OpenSSL_add_all_digests() to be ignored, even after calling
2853 *) Change the default configuration reader to deal with last line not
2854 being properly terminated.
2857 *) Change X509_NAME_cmp() so it applies the special rules on handling
2858 DN values that are of type PrintableString, as well as RDNs of type
2859 emailAddress where the value has the type ia5String.
2860 [stefank@valicert.com via Richard Levitte]
2862 *) Add a SSL_SESS_CACHE_NO_INTERNAL_STORE flag to take over half
2863 the job SSL_SESS_CACHE_NO_INTERNAL_LOOKUP was inconsistently
2864 doing, define a new flag (SSL_SESS_CACHE_NO_INTERNAL) to be
2865 the bitwise-OR of the two for use by the majority of applications
2866 wanting this behaviour, and update the docs. The documented
2867 behaviour and actual behaviour were inconsistent and had been
2868 changing anyway, so this is more a bug-fix than a behavioural
2870 [Geoff Thorpe, diagnosed by Nadav Har'El]
2872 *) Don't impose a 16-byte length minimum on session IDs in ssl/s3_clnt.c
2873 (the SSL 3.0 and TLS 1.0 specifications allow any length up to 32 bytes).
2876 *) Fix initialization code race conditions in
2877 SSLv23_method(), SSLv23_client_method(), SSLv23_server_method(),
2878 SSLv2_method(), SSLv2_client_method(), SSLv2_server_method(),
2879 SSLv3_method(), SSLv3_client_method(), SSLv3_server_method(),
2880 TLSv1_method(), TLSv1_client_method(), TLSv1_server_method(),
2881 ssl2_get_cipher_by_char(),
2882 ssl3_get_cipher_by_char().
2883 [Patrick McCormick <patrick@tellme.com>, Bodo Moeller]
2885 *) Reorder cleanup sequence in SSL_CTX_free(): only remove the ex_data after
2886 the cached sessions are flushed, as the remove_cb() might use ex_data
2887 contents. Bug found by Sam Varshavchik <mrsam@courier-mta.com>
2888 (see [openssl.org #212]).
2889 [Geoff Thorpe, Lutz Jaenicke]
2891 *) Fix typo in OBJ_txt2obj which incorrectly passed the content
2892 length, instead of the encoding length to d2i_ASN1_OBJECT.
2895 Changes between 0.9.6f and 0.9.6g [9 Aug 2002]
2897 *) [In 0.9.6g-engine release:]
2898 Fix crypto/engine/vendor_defns/cswift.h for WIN32 (use '_stdcall').
2899 [Lynn Gazis <lgazis@rainbow.com>]
2901 Changes between 0.9.6e and 0.9.6f [8 Aug 2002]
2903 *) Fix ASN1 checks. Check for overflow by comparing with LONG_MAX
2904 and get fix the header length calculation.
2905 [Florian Weimer <Weimer@CERT.Uni-Stuttgart.DE>,
2906 Alon Kantor <alonk@checkpoint.com> (and others),
2909 *) Use proper error handling instead of 'assertions' in buffer
2910 overflow checks added in 0.9.6e. This prevents DoS (the
2911 assertions could call abort()).
2912 [Arne Ansper <arne@ats.cyber.ee>, Bodo Moeller]
2914 Changes between 0.9.6d and 0.9.6e [30 Jul 2002]
2916 *) Add various sanity checks to asn1_get_length() to reject
2917 the ASN1 length bytes if they exceed sizeof(long), will appear
2918 negative or the content length exceeds the length of the
2920 [Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>]
2922 *) Fix cipher selection routines: ciphers without encryption had no flags
2923 for the cipher strength set and where therefore not handled correctly
2924 by the selection routines (PR #130).
2927 *) Fix EVP_dsa_sha macro.
2931 SSL_OP_DONT_INSERT_EMPTY_FRAGMENTS
2932 for disabling the SSL 3.0/TLS 1.0 CBC vulnerability countermeasure
2933 that was added in OpenSSL 0.9.6d.
2935 As the countermeasure turned out to be incompatible with some
2936 broken SSL implementations, the new option is part of SSL_OP_ALL.
2937 SSL_OP_ALL is usually employed when compatibility with weird SSL
2938 implementations is desired (e.g. '-bugs' option to 's_client' and
2939 's_server'), so the new option is automatically set in many
2943 *) Changes in security patch:
2945 Changes marked "(CHATS)" were sponsored by the Defense Advanced
2946 Research Projects Agency (DARPA) and Air Force Research Laboratory,
2947 Air Force Materiel Command, USAF, under agreement number
2950 *) Add various sanity checks to asn1_get_length() to reject
2951 the ASN1 length bytes if they exceed sizeof(long), will appear
2952 negative or the content length exceeds the length of the
2953 supplied buffer. (CAN-2002-0659)
2954 [Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>]
2956 *) Assertions for various potential buffer overflows, not known to
2958 [Ben Laurie (CHATS)]
2960 *) Various temporary buffers to hold ASCII versions of integers were
2961 too small for 64 bit platforms. (CAN-2002-0655)
2962 [Matthew Byng-Maddick <mbm@aldigital.co.uk> and Ben Laurie (CHATS)>
2964 *) Remote buffer overflow in SSL3 protocol - an attacker could
2965 supply an oversized session ID to a client. (CAN-2002-0656)
2966 [Ben Laurie (CHATS)]
2968 *) Remote buffer overflow in SSL2 protocol - an attacker could
2969 supply an oversized client master key. (CAN-2002-0656)
2970 [Ben Laurie (CHATS)]
2972 Changes between 0.9.6c and 0.9.6d [9 May 2002]
2974 *) Fix crypto/asn1/a_sign.c so that 'parameters' is omitted (not
2975 encoded as NULL) with id-dsa-with-sha1.
2976 [Nils Larsch <nla@trustcenter.de>; problem pointed out by Bodo Moeller]
2978 *) Check various X509_...() return values in apps/req.c.
2979 [Nils Larsch <nla@trustcenter.de>]
2981 *) Fix BASE64 decode (EVP_DecodeUpdate) for data with CR/LF ended lines:
2982 an end-of-file condition would erronously be flagged, when the CRLF
2983 was just at the end of a processed block. The bug was discovered when
2984 processing data through a buffering memory BIO handing the data to a
2985 BASE64-decoding BIO. Bug fund and patch submitted by Pavel Tsekov
2986 <ptsekov@syntrex.com> and Nedelcho Stanev.
2989 *) Implement a countermeasure against a vulnerability recently found
2990 in CBC ciphersuites in SSL 3.0/TLS 1.0: Send an empty fragment
2991 before application data chunks to avoid the use of known IVs
2992 with data potentially chosen by the attacker.
2995 *) Fix length checks in ssl3_get_client_hello().
2998 *) TLS/SSL library bugfix: use s->s3->in_read_app_data differently
2999 to prevent ssl3_read_internal() from incorrectly assuming that
3000 ssl3_read_bytes() found application data while handshake
3001 processing was enabled when in fact s->s3->in_read_app_data was
3002 merely automatically cleared during the initial handshake.
3003 [Bodo Moeller; problem pointed out by Arne Ansper <arne@ats.cyber.ee>]
3005 *) Fix object definitions for Private and Enterprise: they were not
3006 recognized in their shortname (=lowercase) representation. Extend
3007 obj_dat.pl to issue an error when using undefined keywords instead
3008 of silently ignoring the problem (Svenning Sorensen
3009 <sss@sss.dnsalias.net>).
3012 *) Fix DH_generate_parameters() so that it works for 'non-standard'
3013 generators, i.e. generators other than 2 and 5. (Previously, the
3014 code did not properly initialise the 'add' and 'rem' values to
3015 BN_generate_prime().)
3017 In the new general case, we do not insist that 'generator' is
3018 actually a primitive root: This requirement is rather pointless;
3019 a generator of the order-q subgroup is just as good, if not
3023 *) Map new X509 verification errors to alerts. Discovered and submitted by
3024 Tom Wu <tom@arcot.com>.
3027 *) Fix ssl3_pending() (ssl/s3_lib.c) to prevent SSL_pending() from
3028 returning non-zero before the data has been completely received
3029 when using non-blocking I/O.
3030 [Bodo Moeller; problem pointed out by John Hughes]
3032 *) Some of the ciphers missed the strength entry (SSL_LOW etc).
3033 [Ben Laurie, Lutz Jaenicke]
3035 *) Fix bug in SSL_clear(): bad sessions were not removed (found by
3036 Yoram Zahavi <YoramZ@gilian.com>).
3039 *) Add information about CygWin 1.3 and on, and preserve proper
3040 configuration for the versions before that.
3041 [Corinna Vinschen <vinschen@redhat.com> and Richard Levitte]
3043 *) Make removal from session cache (SSL_CTX_remove_session()) more robust:
3044 check whether we deal with a copy of a session and do not delete from
3045 the cache in this case. Problem reported by "Izhar Shoshani Levi"
3046 <izhar@checkpoint.com>.
3049 *) Do not store session data into the internal session cache, if it
3050 is never intended to be looked up (SSL_SESS_CACHE_NO_INTERNAL_LOOKUP
3051 flag is set). Proposed by Aslam <aslam@funk.com>.
3054 *) Have ASN1_BIT_STRING_set_bit() really clear a bit when the requested
3058 *) [In 0.9.6d-engine release:]
3059 Fix a crashbug and a logic bug in hwcrhk_load_pubkey().
3060 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte]
3062 *) Add the configuration target linux-s390x.
3063 [Neale Ferguson <Neale.Ferguson@SoftwareAG-USA.com> via Richard Levitte]
3065 *) The earlier bugfix for the SSL3_ST_SW_HELLO_REQ_C case of
3066 ssl3_accept (ssl/s3_srvr.c) incorrectly used a local flag
3067 variable as an indication that a ClientHello message has been
3068 received. As the flag value will be lost between multiple
3069 invocations of ssl3_accept when using non-blocking I/O, the
3070 function may not be aware that a handshake has actually taken
3071 place, thus preventing a new session from being added to the
3074 To avoid this problem, we now set s->new_session to 2 instead of
3075 using a local variable.
3076 [Lutz Jaenicke, Bodo Moeller]
3078 *) Bugfix: Return -1 from ssl3_get_server_done (ssl3/s3_clnt.c)
3079 if the SSL_R_LENGTH_MISMATCH error is detected.
3080 [Geoff Thorpe, Bodo Moeller]
3082 *) New 'shared_ldflag' column in Configure platform table.
3085 *) Fix EVP_CIPHER_mode macro.
3086 ["Dan S. Camper" <dan@bti.net>]
3088 *) Fix ssl3_read_bytes (ssl/s3_pkt.c): To ignore messages of unknown
3089 type, we must throw them away by setting rr->length to 0.
3090 [D P Chang <dpc@qualys.com>]
3092 Changes between 0.9.6b and 0.9.6c [21 dec 2001]
3094 *) Fix BN_rand_range bug pointed out by Dominikus Scherkl
3095 <Dominikus.Scherkl@biodata.com>. (The previous implementation
3096 worked incorrectly for those cases where range = 10..._2 and
3097 3*range is two bits longer than range.)
3100 *) Only add signing time to PKCS7 structures if it is not already
3104 *) Fix crypto/objects/objects.h: "ld-ce" should be "id-ce",
3105 OBJ_ld_ce should be OBJ_id_ce.
3106 Also some ip-pda OIDs in crypto/objects/objects.txt were
3107 incorrect (cf. RFC 3039).
3108 [Matt Cooper, Frederic Giudicelli, Bodo Moeller]
3110 *) Release CRYPTO_LOCK_DYNLOCK when CRYPTO_destroy_dynlockid()
3111 returns early because it has nothing to do.
3112 [Andy Schneider <andy.schneider@bjss.co.uk>]
3114 *) [In 0.9.6c-engine release:]
3115 Fix mutex callback return values in crypto/engine/hw_ncipher.c.
3116 [Andy Schneider <andy.schneider@bjss.co.uk>]
3118 *) [In 0.9.6c-engine release:]
3119 Add support for Cryptographic Appliance's keyserver technology.
3120 (Use engine 'keyclient')
3121 [Cryptographic Appliances and Geoff Thorpe]
3123 *) Add a configuration entry for OS/390 Unix. The C compiler 'c89'
3124 is called via tools/c89.sh because arguments have to be
3125 rearranged (all '-L' options must appear before the first object
3127 [Richard Shapiro <rshapiro@abinitio.com>]
3129 *) [In 0.9.6c-engine release:]
3130 Add support for Broadcom crypto accelerator cards, backported
3132 [Broadcom, Nalin Dahyabhai <nalin@redhat.com>, Mark Cox]
3134 *) [In 0.9.6c-engine release:]
3135 Add support for SureWare crypto accelerator cards from
3136 Baltimore Technologies. (Use engine 'sureware')
3137 [Baltimore Technologies and Mark Cox]
3139 *) [In 0.9.6c-engine release:]
3140 Add support for crypto accelerator cards from Accelerated
3141 Encryption Processing, www.aep.ie. (Use engine 'aep')
3142 [AEP Inc. and Mark Cox]
3144 *) Add a configuration entry for gcc on UnixWare.
3145 [Gary Benson <gbenson@redhat.com>]
3147 *) Change ssl/s2_clnt.c and ssl/s2_srvr.c so that received handshake
3148 messages are stored in a single piece (fixed-length part and
3149 variable-length part combined) and fix various bugs found on the way.
3152 *) Disable caching in BIO_gethostbyname(), directly use gethostbyname()
3153 instead. BIO_gethostbyname() does not know what timeouts are
3154 appropriate, so entries would stay in cache even when they have
3156 [Bodo Moeller; problem pointed out by Rich Salz <rsalz@zolera.com>
3158 *) Change ssl23_get_client_hello (ssl/s23_srvr.c) behaviour when
3159 faced with a pathologically small ClientHello fragment that does
3160 not contain client_version: Instead of aborting with an error,
3161 simply choose the highest available protocol version (i.e.,
3162 TLS 1.0 unless it is disabled). In practice, ClientHello
3163 messages are never sent like this, but this change gives us
3164 strictly correct behaviour at least for TLS.
3167 *) Fix SSL handshake functions and SSL_clear() such that SSL_clear()
3168 never resets s->method to s->ctx->method when called from within
3169 one of the SSL handshake functions.
3170 [Bodo Moeller; problem pointed out by Niko Baric]
3172 *) In ssl3_get_client_hello (ssl/s3_srvr.c), generate a fatal alert
3173 (sent using the client's version number) if client_version is
3174 smaller than the protocol version in use. Also change
3175 ssl23_get_client_hello (ssl/s23_srvr.c) to select TLS 1.0 if
3176 the client demanded SSL 3.0 but only TLS 1.0 is enabled; then
3177 the client will at least see that alert.
3180 *) Fix ssl3_get_message (ssl/s3_both.c) to handle message fragmentation
3184 *) Avoid infinite loop in ssl3_get_message (ssl/s3_both.c) if a
3185 client receives HelloRequest while in a handshake.
3186 [Bodo Moeller; bug noticed by Andy Schneider <andy.schneider@bjss.co.uk>]
3188 *) Bugfix in ssl3_accept (ssl/s3_srvr.c): Case SSL3_ST_SW_HELLO_REQ_C
3189 should end in 'break', not 'goto end' which circuments various
3190 cleanups done in state SSL_ST_OK. But session related stuff
3191 must be disabled for SSL_ST_OK in the case that we just sent a
3194 Also avoid some overhead by not calling ssl_init_wbio_buffer()
3195 before just sending a HelloRequest.
3196 [Bodo Moeller, Eric Rescorla <ekr@rtfm.com>]
3198 *) Fix ssl/s3_enc.c, ssl/t1_enc.c and ssl/s3_pkt.c so that we don't
3199 reveal whether illegal block cipher padding was found or a MAC
3200 verification error occured. (Neither SSLerr() codes nor alerts
3201 are directly visible to potential attackers, but the information
3202 may leak via logfiles.)
3204 Similar changes are not required for the SSL 2.0 implementation
3205 because the number of padding bytes is sent in clear for SSL 2.0,
3206 and the extra bytes are just ignored. However ssl/s2_pkt.c
3207 failed to verify that the purported number of padding bytes is in
3211 *) Add OpenUNIX-8 support including shared libraries
3212 (Boyd Lynn Gerber <gerberb@zenez.com>).
3215 *) Improve RSA_padding_check_PKCS1_OAEP() check again to avoid
3216 'wristwatch attack' using huge encoding parameters (cf.
3217 James H. Manger's CRYPTO 2001 paper). Note that the
3218 RSA_PKCS1_OAEP_PADDING case of RSA_private_decrypt() does not use
3219 encoding parameters and hence was not vulnerable.
3222 *) BN_sqr() bug fix.
3223 [Ulf Möller, reported by Jim Ellis <jim.ellis@cavium.com>]
3225 *) Rabin-Miller test analyses assume uniformly distributed witnesses,
3226 so use BN_pseudo_rand_range() instead of using BN_pseudo_rand()
3227 followed by modular reduction.
3228 [Bodo Moeller; pointed out by Adam Young <AYoung1@NCSUS.JNJ.COM>]
3230 *) Add BN_pseudo_rand_range() with obvious functionality: BN_rand_range()
3231 equivalent based on BN_pseudo_rand() instead of BN_rand().
3234 *) s3_srvr.c: allow sending of large client certificate lists (> 16 kB).
3235 This function was broken, as the check for a new client hello message
3236 to handle SGC did not allow these large messages.
3237 (Tracked down by "Douglas E. Engert" <deengert@anl.gov>.)
3240 *) Add alert descriptions for TLSv1 to SSL_alert_desc_string[_long]().
3243 *) Fix buggy behaviour of BIO_get_num_renegotiates() and BIO_ctrl()
3244 for BIO_C_GET_WRITE_BUF_SIZE ("Stephen Hinton" <shinton@netopia.com>).
3247 *) Rework the configuration and shared library support for Tru64 Unix.
3248 The configuration part makes use of modern compiler features and
3249 still retains old compiler behavior for those that run older versions
3250 of the OS. The shared library support part includes a variant that
3251 uses the RPATH feature, and is available through the special
3252 configuration target "alpha-cc-rpath", which will never be selected
3254 [Tim Mooney <mooney@dogbert.cc.ndsu.NoDak.edu> via Richard Levitte]
3256 *) In ssl3_get_key_exchange (ssl/s3_clnt.c), call ssl3_get_message()
3257 with the same message size as in ssl3_get_certificate_request().
3258 Otherwise, if no ServerKeyExchange message occurs, CertificateRequest
3259 messages might inadvertently be reject as too long.
3260 [Petr Lampa <lampa@fee.vutbr.cz>]
3262 *) Enhanced support for IA-64 Unix platforms (well, Linux and HP-UX).
3265 *) Modified SSL library such that the verify_callback that has been set
3266 specificly for an SSL object with SSL_set_verify() is actually being
3267 used. Before the change, a verify_callback set with this function was
3268 ignored and the verify_callback() set in the SSL_CTX at the time of
3269 the call was used. New function X509_STORE_CTX_set_verify_cb() introduced
3270 to allow the necessary settings.
3273 *) Initialize static variable in crypto/dsa/dsa_lib.c and crypto/dh/dh_lib.c
3274 explicitly to NULL, as at least on Solaris 8 this seems not always to be
3275 done automatically (in contradiction to the requirements of the C
3276 standard). This made problems when used from OpenSSH.
3279 *) In OpenSSL 0.9.6a and 0.9.6b, crypto/dh/dh_key.c ignored
3280 dh->length and always used
3282 BN_rand_range(priv_key, dh->p).
3284 BN_rand_range() is not necessary for Diffie-Hellman, and this
3285 specific range makes Diffie-Hellman unnecessarily inefficient if
3286 dh->length (recommended exponent length) is much smaller than the
3287 length of dh->p. We could use BN_rand_range() if the order of
3288 the subgroup was stored in the DH structure, but we only have
3293 BN_rand(priv_key, l, ...)
3295 where 'l' is dh->length if this is defined, or BN_num_bits(dh->p)-1
3301 RSA_eay_public_encrypt
3302 RSA_eay_private_decrypt
3303 RSA_eay_private_encrypt (signing)
3304 RSA_eay_public_decrypt (signature verification)
3306 (default implementations for RSA_public_encrypt,
3307 RSA_private_decrypt, RSA_private_encrypt, RSA_public_decrypt),
3308 always reject numbers >= n.
3311 *) In crypto/rand/md_rand.c, use a new short-time lock CRYPTO_LOCK_RAND2
3312 to synchronize access to 'locking_thread'. This is necessary on
3313 systems where access to 'locking_thread' (an 'unsigned long'
3314 variable) is not atomic.
3317 *) In crypto/rand/md_rand.c, set 'locking_thread' to current thread's ID
3318 *before* setting the 'crypto_lock_rand' flag. The previous code had
3319 a race condition if 0 is a valid thread ID.
3320 [Travis Vitek <vitek@roguewave.com>]
3322 *) Add support for shared libraries under Irix.
3323 [Albert Chin-A-Young <china@thewrittenword.com>]
3325 *) Add configuration option to build on Linux on both big-endian and
3327 [Ralf Baechle <ralf@uni-koblenz.de>]
3329 *) Add the possibility to create shared libraries on HP-UX.
3332 Changes between 0.9.6a and 0.9.6b [9 Jul 2001]
3334 *) Change ssleay_rand_bytes (crypto/rand/md_rand.c)
3335 to avoid a SSLeay/OpenSSL PRNG weakness pointed out by
3336 Markku-Juhani O. Saarinen <markku-juhani.saarinen@nokia.com>:
3337 PRNG state recovery was possible based on the output of
3338 one PRNG request appropriately sized to gain knowledge on
3339 'md' followed by enough consecutive 1-byte PRNG requests
3340 to traverse all of 'state'.
3342 1. When updating 'md_local' (the current thread's copy of 'md')
3343 during PRNG output generation, hash all of the previous
3344 'md_local' value, not just the half used for PRNG output.
3346 2. Make the number of bytes from 'state' included into the hash
3347 independent from the number of PRNG bytes requested.
3349 The first measure alone would be sufficient to avoid
3350 Markku-Juhani's attack. (Actually it had never occurred
3351 to me that the half of 'md_local' used for chaining was the
3352 half from which PRNG output bytes were taken -- I had always
3353 assumed that the secret half would be used.) The second
3354 measure makes sure that additional data from 'state' is never
3355 mixed into 'md_local' in small portions; this heuristically
3356 further strengthens the PRNG.
3359 *) Fix crypto/bn/asm/mips3.s.
3362 *) When only the key is given to "enc", the IV is undefined. Print out
3363 an error message in this case.
3366 *) Handle special case when X509_NAME is empty in X509 printing routines.
3369 *) In dsa_do_verify (crypto/dsa/dsa_ossl.c), verify that r and s are
3370 positive and less than q.
3373 *) Don't change *pointer in CRYPTO_add_lock() is add_lock_callback is
3374 used: it isn't thread safe and the add_lock_callback should handle
3376 [Paul Rose <Paul.Rose@bridge.com>]
3378 *) Verify that incoming data obeys the block size in
3379 ssl3_enc (ssl/s3_enc.c) and tls1_enc (ssl/t1_enc.c).
3383 [Ulf Möller, Bodo Möller]
3385 *) The countermeasure against Bleichbacher's attack on PKCS #1 v1.5
3386 RSA encryption was accidentally removed in s3_srvr.c in OpenSSL 0.9.5
3387 when fixing the server behaviour for backwards-compatible 'client
3388 hello' messages. (Note that the attack is impractical against
3389 SSL 3.0 and TLS 1.0 anyway because length and version checking
3390 means that the probability of guessing a valid ciphertext is
3391 around 2^-40; see section 5 in Bleichenbacher's CRYPTO '98
3394 Before 0.9.5, the countermeasure (hide the error by generating a
3395 random 'decryption result') did not work properly because
3396 ERR_clear_error() was missing, meaning that SSL_get_error() would
3397 detect the supposedly ignored error.
3399 Both problems are now fixed.
3402 *) In crypto/bio/bf_buff.c, increase DEFAULT_BUFFER_SIZE to 4096
3403 (previously it was 1024).
3406 *) Fix for compatibility mode trust settings: ignore trust settings
3407 unless some valid trust or reject settings are present.
3410 *) Fix for blowfish EVP: its a variable length cipher.
3413 *) Fix various bugs related to DSA S/MIME verification. Handle missing
3414 parameters in DSA public key structures and return an error in the
3415 DSA routines if parameters are absent.
3418 *) In versions up to 0.9.6, RAND_file_name() resorted to file ".rnd"
3419 in the current directory if neither $RANDFILE nor $HOME was set.
3420 RAND_file_name() in 0.9.6a returned NULL in this case. This has
3421 caused some confusion to Windows users who haven't defined $HOME.
3422 Thus RAND_file_name() is changed again: e_os.h can define a
3423 DEFAULT_HOME, which will be used if $HOME is not set.
3424 For Windows, we use "C:"; on other platforms, we still require
3425 environment variables.
3427 *) Move 'if (!initialized) RAND_poll()' into regions protected by
3428 CRYPTO_LOCK_RAND. This is not strictly necessary, but avoids
3429 having multiple threads call RAND_poll() concurrently.
3432 *) In crypto/rand/md_rand.c, replace 'add_do_not_lock' flag by a
3433 combination of a flag and a thread ID variable.
3434 Otherwise while one thread is in ssleay_rand_bytes (which sets the
3435 flag), *other* threads can enter ssleay_add_bytes without obeying
3436 the CRYPTO_LOCK_RAND lock (and may even illegally release the lock
3437 that they do not hold after the first thread unsets add_do_not_lock).
3440 *) Change bctest again: '-x' expressions are not available in all
3444 Changes between 0.9.6 and 0.9.6a [5 Apr 2001]
3446 *) Fix a couple of memory leaks in PKCS7_dataDecode()
3447 [Steve Henson, reported by Heyun Zheng <hzheng@atdsprint.com>]
3449 *) Change Configure and Makefiles to provide EXE_EXT, which will contain
3450 the default extension for executables, if any. Also, make the perl
3451 scripts that use symlink() to test if it really exists and use "cp"
3452 if it doesn't. All this made OpenSSL compilable and installable in
3456 *) Fix for asn1_GetSequence() for indefinite length constructed data.
3457 If SEQUENCE is length is indefinite just set c->slen to the total
3458 amount of data available.
3459 [Steve Henson, reported by shige@FreeBSD.org]
3460 [This change does not apply to 0.9.7.]
3462 *) Change bctest to avoid here-documents inside command substitution
3463 (workaround for FreeBSD /bin/sh bug).
3464 For compatibility with Ultrix, avoid shell functions (introduced
3465 in the bctest version that searches along $PATH).
3468 *) Rename 'des_encrypt' to 'des_encrypt1'. This avoids the clashes
3469 with des_encrypt() defined on some operating systems, like Solaris
3473 *) Check the result of RSA-CRT (see D. Boneh, R. DeMillo, R. Lipton:
3474 On the Importance of Eliminating Errors in Cryptographic
3475 Computations, J. Cryptology 14 (2001) 2, 101-119,
3476 http://theory.stanford.edu/~dabo/papers/faults.ps.gz).
3479 *) MIPS assembler BIGNUM division bug fix.
3482 *) Disabled incorrect Alpha assembler code.
3485 *) Fix PKCS#7 decode routines so they correctly update the length
3486 after reading an EOC for the EXPLICIT tag.
3488 [This change does not apply to 0.9.7.]
3490 *) Fix bug in PKCS#12 key generation routines. This was triggered
3491 if a 3DES key was generated with a 0 initial byte. Include
3492 PKCS12_BROKEN_KEYGEN compilation option to retain the old
3493 (but broken) behaviour.
3496 *) Enhance bctest to search for a working bc along $PATH and print
3498 [Tim Rice <tim@multitalents.net> via Richard Levitte]
3500 *) Fix memory leaks in err.c: free err_data string if necessary;
3501 don't write to the wrong index in ERR_set_error_data.
3504 *) Implement ssl23_peek (analogous to ssl23_read), which previously
3508 *) Replace rdtsc with _emit statements for VC++ version 5.
3509 [Jeremy Cooper <jeremy@baymoo.org>]
3511 *) Make it possible to reuse SSLv2 sessions.
3514 *) In copy_email() check for >= 0 as a return value for
3515 X509_NAME_get_index_by_NID() since 0 is a valid index.
3516 [Steve Henson reported by Massimiliano Pala <madwolf@opensca.org>]
3518 *) Avoid coredump with unsupported or invalid public keys by checking if
3519 X509_get_pubkey() fails in PKCS7_verify(). Fix memory leak when
3520 PKCS7_verify() fails with non detached data.
3523 *) Don't use getenv in library functions when run as setuid/setgid.
3524 New function OPENSSL_issetugid().
3527 *) Avoid false positives in memory leak detection code (crypto/mem_dbg.c)
3528 due to incorrect handling of multi-threading:
3530 1. Fix timing glitch in the MemCheck_off() portion of CRYPTO_mem_ctrl().
3532 2. Fix logical glitch in is_MemCheck_on() aka CRYPTO_is_mem_check_on().
3534 3. Count how many times MemCheck_off() has been called so that
3535 nested use can be treated correctly. This also avoids
3536 inband-signalling in the previous code (which relied on the
3537 assumption that thread ID 0 is impossible).
3540 *) Add "-rand" option also to s_client and s_server.
3543 *) Fix CPU detection on Irix 6.x.
3544 [Kurt Hockenbury <khockenb@stevens-tech.edu> and
3545 "Bruce W. Forsberg" <bruce.forsberg@baesystems.com>]
3547 *) Fix X509_NAME bug which produced incorrect encoding if X509_NAME
3550 [This change does not apply to 0.9.7.]
3552 *) Use the cached encoding of an X509_NAME structure rather than
3553 copying it. This is apparently the reason for the libsafe "errors"
3554 but the code is actually correct.
3557 *) Add new function BN_rand_range(), and fix DSA_sign_setup() to prevent
3558 Bleichenbacher's DSA attack.
3559 Extend BN_[pseudo_]rand: As before, top=1 forces the highest two bits
3560 to be set and top=0 forces the highest bit to be set; top=-1 is new
3561 and leaves the highest bit random.
3562 [Ulf Moeller, Bodo Moeller]
3564 *) In the NCONF_...-based implementations for CONF_... queries
3565 (crypto/conf/conf_lib.c), if the input LHASH is NULL, avoid using
3566 a temporary CONF structure with the data component set to NULL
3567 (which gives segmentation faults in lh_retrieve).
3568 Instead, use NULL for the CONF pointer in CONF_get_string and
3569 CONF_get_number (which may use environment variables) and directly
3570 return NULL from CONF_get_section.
3573 *) Fix potential buffer overrun for EBCDIC.
3576 *) Tolerate nonRepudiation as being valid for S/MIME signing and certSign
3577 keyUsage if basicConstraints absent for a CA.
3580 *) Make SMIME_write_PKCS7() write mail header values with a format that
3581 is more generally accepted (no spaces before the semicolon), since
3582 some programs can't parse those values properly otherwise. Also make
3583 sure BIO's that break lines after each write do not create invalid
3587 *) Make the CRL encoding routines work with empty SEQUENCE OF. The
3588 macros previously used would not encode an empty SEQUENCE OF
3589 and break the signature.
3591 [This change does not apply to 0.9.7.]
3593 *) Zero the premaster secret after deriving the master secret in
3597 *) Add some EVP_add_digest_alias registrations (as found in
3598 OpenSSL_add_all_digests()) to SSL_library_init()
3599 aka OpenSSL_add_ssl_algorithms(). This provides improved
3600 compatibility with peers using X.509 certificates
3601 with unconventional AlgorithmIdentifier OIDs.
3604 *) Fix for Irix with NO_ASM.
3605 ["Bruce W. Forsberg" <bruce.forsberg@baesystems.com>]
3607 *) ./config script fixes.
3608 [Ulf Moeller, Richard Levitte]
3610 *) Fix 'openssl passwd -1'.
3613 *) Change PKCS12_key_gen_asc() so it can cope with non null
3614 terminated strings whose length is passed in the passlen
3615 parameter, for example from PEM callbacks. This was done
3616 by adding an extra length parameter to asc2uni().
3617 [Steve Henson, reported by <oddissey@samsung.co.kr>]
3619 *) Fix C code generated by 'openssl dsaparam -C': If a BN_bin2bn
3620 call failed, free the DSA structure.
3623 *) Fix to uni2asc() to cope with zero length Unicode strings.
3624 These are present in some PKCS#12 files.
3627 *) Increase s2->wbuf allocation by one byte in ssl2_new (ssl/s2_lib.c).
3628 Otherwise do_ssl_write (ssl/s2_pkt.c) will write beyond buffer limits
3629 when writing a 32767 byte record.
3630 [Bodo Moeller; problem reported by Eric Day <eday@concentric.net>]
3632 *) In RSA_eay_public_{en,ed}crypt and RSA_eay_mod_exp (rsa_eay.c),
3633 obtain lock CRYPTO_LOCK_RSA before setting rsa->_method_mod_{n,p,q}.
3635 (RSA objects have a reference count access to which is protected
3636 by CRYPTO_LOCK_RSA [see rsa_lib.c, s3_srvr.c, ssl_cert.c, ssl_rsa.c],
3637 so they are meant to be shared between threads.)
3638 [Bodo Moeller, Geoff Thorpe; original patch submitted by
3639 "Reddie, Steven" <Steven.Reddie@ca.com>]
3641 *) Fix a deadlock in CRYPTO_mem_leaks().
3644 *) Use better test patterns in bntest.
3647 *) rand_win.c fix for Borland C.
3650 *) BN_rshift bugfix for n == 0.
3653 *) Add a 'bctest' script that checks for some known 'bc' bugs
3654 so that 'make test' does not abort just because 'bc' is broken.
3657 *) Store verify_result within SSL_SESSION also for client side to
3658 avoid potential security hole. (Re-used sessions on the client side
3659 always resulted in verify_result==X509_V_OK, not using the original
3660 result of the server certificate verification.)
3663 *) Fix ssl3_pending: If the record in s->s3->rrec is not of type
3664 SSL3_RT_APPLICATION_DATA, return 0.
3665 Similarly, change ssl2_pending to return 0 if SSL_in_init(s) is true.
3669 Both ssl2_peek and ssl3_peek, which were totally broken in earlier
3670 releases, have been re-implemented by renaming the previous
3671 implementations of ssl2_read and ssl3_read to ssl2_read_internal
3672 and ssl3_read_internal, respectively, and adding 'peek' parameters
3673 to them. The new ssl[23]_{read,peek} functions are calls to
3674 ssl[23]_read_internal with the 'peek' flag set appropriately.
3675 A 'peek' parameter has also been added to ssl3_read_bytes, which
3676 does the actual work for ssl3_read_internal.
3679 *) Initialise "ex_data" member of RSA/DSA/DH structures prior to calling
3680 the method-specific "init()" handler. Also clean up ex_data after
3681 calling the method-specific "finish()" handler. Previously, this was
3682 happening the other way round.
3685 *) Increase BN_CTX_NUM (the number of BIGNUMs in a BN_CTX) to 16.
3686 The previous value, 12, was not always sufficient for BN_mod_exp().
3689 *) Make sure that shared libraries get the internal name engine with
3690 the full version number and not just 0. This should mark the
3691 shared libraries as not backward compatible. Of course, this should
3692 be changed again when we can guarantee backward binary compatibility.
3695 *) Fix typo in get_cert_by_subject() in by_dir.c
3696 [Jean-Marc Desperrier <jean-marc.desperrier@certplus.com>]
3698 *) Rework the system to generate shared libraries:
3700 - Make note of the expected extension for the shared libraries and
3701 if there is a need for symbolic links from for example libcrypto.so.0
3702 to libcrypto.so.0.9.7. There is extended info in Configure for
3705 - Make as few rebuilds of the shared libraries as possible.
3707 - Still avoid linking the OpenSSL programs with the shared libraries.
3709 - When installing, install the shared libraries separately from the
3713 *) Fix SSL_CTX_set_read_ahead macro to actually use its argument.
3715 Copy SSL_CTX's read_ahead flag to SSL object directly in SSL_new
3716 and not in SSL_clear because the latter is also used by the
3717 accept/connect functions; previously, the settings made by
3718 SSL_set_read_ahead would be lost during the handshake.
3719 [Bodo Moeller; problems reported by Anders Gertz <gertz@epact.se>]
3721 *) Correct util/mkdef.pl to be selective about disabled algorithms.
3722 Previously, it would create entries for disableed algorithms no
3726 *) Added several new manual pages for SSL_* function.
3729 Changes between 0.9.5a and 0.9.6 [24 Sep 2000]
3731 *) In ssl23_get_client_hello, generate an error message when faced
3732 with an initial SSL 3.0/TLS record that is too small to contain the
3733 first two bytes of the ClientHello message, i.e. client_version.
3734 (Note that this is a pathologic case that probably has never happened
3735 in real life.) The previous approach was to use the version number
3736 from the record header as a substitute; but our protocol choice
3737 should not depend on that one because it is not authenticated
3738 by the Finished messages.
3741 *) More robust randomness gathering functions for Windows.
3742 [Jeffrey Altman <jaltman@columbia.edu>]
3744 *) For compatibility reasons if the flag X509_V_FLAG_ISSUER_CHECK is
3745 not set then we don't setup the error code for issuer check errors
3746 to avoid possibly overwriting other errors which the callback does
3747 handle. If an application does set the flag then we assume it knows
3748 what it is doing and can handle the new informational codes
3752 *) Fix for a nasty bug in ASN1_TYPE handling. ASN1_TYPE is used for
3753 a general "ANY" type, as such it should be able to decode anything
3754 including tagged types. However it didn't check the class so it would
3755 wrongly interpret tagged types in the same way as their universal
3756 counterpart and unknown types were just rejected. Changed so that the
3757 tagged and unknown types are handled in the same way as a SEQUENCE:
3758 that is the encoding is stored intact. There is also a new type
3759 "V_ASN1_OTHER" which is used when the class is not universal, in this
3760 case we have no idea what the actual type is so we just lump them all
3764 *) On VMS, stdout may very well lead to a file that is written to
3765 in a record-oriented fashion. That means that every write() will
3766 write a separate record, which will be read separately by the
3767 programs trying to read from it. This can be very confusing.
3769 The solution is to put a BIO filter in the way that will buffer
3770 text until a linefeed is reached, and then write everything a
3771 line at a time, so every record written will be an actual line,
3772 not chunks of lines and not (usually doesn't happen, but I've
3773 seen it once) several lines in one record. BIO_f_linebuffer() is
3776 Currently, it's a VMS-only method, because that's where it has
3777 been tested well enough.
3780 *) Remove 'optimized' squaring variant in BN_mod_mul_montgomery,
3781 it can return incorrect results.
3782 (Note: The buggy variant was not enabled in OpenSSL 0.9.5a,
3783 but it was in 0.9.6-beta[12].)
3786 *) Disable the check for content being present when verifying detached
3787 signatures in pk7_smime.c. Some versions of Netscape (wrongly)
3788 include zero length content when signing messages.
3791 *) New BIO_shutdown_wr macro, which invokes the BIO_C_SHUTDOWN_WR
3792 BIO_ctrl (for BIO pairs).
3795 *) Add DSO method for VMS.
3798 *) Bug fix: Montgomery multiplication could produce results with the
3802 *) Add RPM specification openssl.spec and modify it to build three
3803 packages. The default package contains applications, application
3804 documentation and run-time libraries. The devel package contains
3805 include files, static libraries and function documentation. The
3806 doc package contains the contents of the doc directory. The original
3807 openssl.spec was provided by Damien Miller <djm@mindrot.org>.
3810 *) Add a large number of documentation files for many SSL routines.
3811 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>]
3813 *) Add a configuration entry for Sony News 4.
3814 [NAKAJI Hiroyuki <nakaji@tutrp.tut.ac.jp>]
3816 *) Don't set the two most significant bits to one when generating a
3817 random number < q in the DSA library.
3820 *) New SSL API mode 'SSL_MODE_AUTO_RETRY'. This disables the default
3821 behaviour that SSL_read may result in SSL_ERROR_WANT_READ (even if
3822 the underlying transport is blocking) if a handshake took place.
3823 (The default behaviour is needed by applications such as s_client
3824 and s_server that use select() to determine when to use SSL_read;
3825 but for applications that know in advance when to expect data, it
3826 just makes things more complicated.)
3829 *) Add RAND_egd_bytes(), which gives control over the number of bytes read
3833 *) Add a few more EBCDIC conditionals that make `req' and `x509'
3834 work better on such systems.
3835 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>]
3837 *) Add two demo programs for PKCS12_parse() and PKCS12_create().
3838 Update PKCS12_parse() so it copies the friendlyName and the
3839 keyid to the certificates aux info.
3842 *) Fix bug in PKCS7_verify() which caused an infinite loop
3843 if there was more than one signature.
3844 [Sven Uszpelkat <su@celocom.de>]
3846 *) Major change in util/mkdef.pl to include extra information
3847 about each symbol, as well as presentig variables as well
3848 as functions. This change means that there's n more need
3849 to rebuild the .num files when some algorithms are excluded.
3852 *) Allow the verify time to be set by an application,
3853 rather than always using the current time.
3856 *) Phase 2 verify code reorganisation. The certificate
3857 verify code now looks up an issuer certificate by a
3858 number of criteria: subject name, authority key id
3859 and key usage. It also verifies self signed certificates
3860 by the same criteria. The main comparison function is
3861 X509_check_issued() which performs these checks.
3863 Lot of changes were necessary in order to support this
3864 without completely rewriting the lookup code.
3866 Authority and subject key identifier are now cached.
3868 The LHASH 'certs' is X509_STORE has now been replaced
3869 by a STACK_OF(X509_OBJECT). This is mainly because an
3870 LHASH can't store or retrieve multiple objects with
3871 the same hash value.
3873 As a result various functions (which were all internal
3874 use only) have changed to handle the new X509_STORE
3875 structure. This will break anything that messed round
3876 with X509_STORE internally.
3878 The functions X509_STORE_add_cert() now checks for an
3879 exact match, rather than just subject name.
3881 The X509_STORE API doesn't directly support the retrieval
3882 of multiple certificates matching a given criteria, however
3883 this can be worked round by performing a lookup first
3884 (which will fill the cache with candidate certificates)
3885 and then examining the cache for matches. This is probably
3886 the best we can do without throwing out X509_LOOKUP
3887 entirely (maybe later...).
3889 The X509_VERIFY_CTX structure has been enhanced considerably.
3891 All certificate lookup operations now go via a get_issuer()
3892 callback. Although this currently uses an X509_STORE it
3893 can be replaced by custom lookups. This is a simple way
3894 to bypass the X509_STORE hackery necessary to make this
3895 work and makes it possible to use more efficient techniques
3896 in future. A very simple version which uses a simple
3897 STACK for its trusted certificate store is also provided
3898 using X509_STORE_CTX_trusted_stack().
3900 The verify_cb() and verify() callbacks now have equivalents
3901 in the X509_STORE_CTX structure.
3903 X509_STORE_CTX also has a 'flags' field which can be used
3904 to customise the verify behaviour.
3907 *) Add new PKCS#7 signing option PKCS7_NOSMIMECAP which
3908 excludes S/MIME capabilities.
3911 *) When a certificate request is read in keep a copy of the
3912 original encoding of the signed data and use it when outputing
3913 again. Signatures then use the original encoding rather than
3914 a decoded, encoded version which may cause problems if the
3915 request is improperly encoded.
3918 *) For consistency with other BIO_puts implementations, call
3919 buffer_write(b, ...) directly in buffer_puts instead of calling
3922 In BIO_puts, increment b->num_write as in BIO_write.
3923 [Peter.Sylvester@EdelWeb.fr]
3925 *) Fix BN_mul_word for the case where the word is 0. (We have to use
3926 BN_zero, we may not return a BIGNUM with an array consisting of
3930 *) Avoid calling abort() from within the library when problems are
3931 detected, except if preprocessor symbols have been defined
3932 (such as REF_CHECK, BN_DEBUG etc.).
3935 *) New openssl application 'rsautl'. This utility can be
3936 used for low level RSA operations. DER public key
3937 BIO/fp routines also added.
3940 *) New Configure entry and patches for compiling on QNX 4.
3941 [Andreas Schneider <andreas@ds3.etech.fh-hamburg.de>]
3943 *) A demo state-machine implementation was sponsored by
3944 Nuron (http://www.nuron.com/) and is now available in
3945 demos/state_machine.
3948 *) New options added to the 'dgst' utility for signature
3949 generation and verification.
3952 *) Unrecognized PKCS#7 content types are now handled via a
3953 catch all ASN1_TYPE structure. This allows unsupported
3954 types to be stored as a "blob" and an application can
3955 encode and decode it manually.
3958 *) Fix various signed/unsigned issues to make a_strex.c
3960 [Oscar Jacobsson <oscar.jacobsson@celocom.com>]
3962 *) ASN1 fixes. i2d_ASN1_OBJECT was not returning the correct
3963 length if passed a buffer. ASN1_INTEGER_to_BN failed
3964 if passed a NULL BN and its argument was negative.
3965 [Steve Henson, pointed out by Sven Heiberg <sven@tartu.cyber.ee>]
3967 *) Modification to PKCS#7 encoding routines to output definite
3968 length encoding. Since currently the whole structures are in
3969 memory there's not real point in using indefinite length
3970 constructed encoding. However if OpenSSL is compiled with
3971 the flag PKCS7_INDEFINITE_ENCODING the old form is used.
3974 *) Added BIO_vprintf() and BIO_vsnprintf().
3977 *) Added more prefixes to parse for in the the strings written
3978 through a logging bio, to cover all the levels that are available
3979 through syslog. The prefixes are now:
3981 PANIC, EMERG, EMR => LOG_EMERG
3982 ALERT, ALR => LOG_ALERT
3983 CRIT, CRI => LOG_CRIT
3984 ERROR, ERR => LOG_ERR
3985 WARNING, WARN, WAR => LOG_WARNING
3986 NOTICE, NOTE, NOT => LOG_NOTICE
3987 INFO, INF => LOG_INFO
3988 DEBUG, DBG => LOG_DEBUG
3990 and as before, if none of those prefixes are present at the
3991 beginning of the string, LOG_ERR is chosen.
3993 On Win32, the LOG_* levels are mapped according to this:
3995 LOG_EMERG, LOG_ALERT, LOG_CRIT, LOG_ERR => EVENTLOG_ERROR_TYPE
3996 LOG_WARNING => EVENTLOG_WARNING_TYPE
3997 LOG_NOTICE, LOG_INFO, LOG_DEBUG => EVENTLOG_INFORMATION_TYPE
4001 *) Made it possible to reconfigure with just the configuration
4002 argument "reconf" or "reconfigure". The command line arguments
4003 are stored in Makefile.ssl in the variable CONFIGURE_ARGS,
4004 and are retrieved from there when reconfiguring.
4008 [Assar Westerlund <assar@sics.se>, Richard Levitte]
4010 *) Add the arguments -CAfile and -CApath to the pkcs12 utility.
4013 *) The obj_dat.pl script was messing up the sorting of object
4014 names. The reason was that it compared the quoted version
4015 of strings as a result "OCSP" > "OCSP Signing" because
4016 " > SPACE. Changed script to store unquoted versions of
4017 names and add quotes on output. It was also omitting some
4018 names from the lookup table if they were given a default
4019 value (that is if SN is missing it is given the same
4020 value as LN and vice versa), these are now added on the
4021 grounds that if an object has a name we should be able to
4022 look it up. Finally added warning output when duplicate
4023 short or long names are found.
4026 *) Changes needed for Tandem NSK.
4027 [Scott Uroff <scott@xypro.com>]
4029 *) Fix SSL 2.0 rollback checking: Due to an off-by-one error in
4030 RSA_padding_check_SSLv23(), special padding was never detected
4031 and thus the SSL 3.0/TLS 1.0 countermeasure against protocol
4032 version rollback attacks was not effective.
4034 In s23_clnt.c, don't use special rollback-attack detection padding
4035 (RSA_SSLV23_PADDING) if SSL 2.0 is the only protocol enabled in the
4036 client; similarly, in s23_srvr.c, don't do the rollback check if
4037 SSL 2.0 is the only protocol enabled in the server.
4040 *) Make it possible to get hexdumps of unprintable data with 'openssl
4041 asn1parse'. By implication, the functions ASN1_parse_dump() and
4042 BIO_dump_indent() are added.
4045 *) New functions ASN1_STRING_print_ex() and X509_NAME_print_ex()
4046 these print out strings and name structures based on various
4047 flags including RFC2253 support and proper handling of
4048 multibyte characters. Added options to the 'x509' utility
4049 to allow the various flags to be set.
4052 *) Various fixes to use ASN1_TIME instead of ASN1_UTCTIME.
4053 Also change the functions X509_cmp_current_time() and
4054 X509_gmtime_adj() work with an ASN1_TIME structure,
4055 this will enable certificates using GeneralizedTime in validity
4056 dates to be checked.
4059 *) Make the NEG_PUBKEY_BUG code (which tolerates invalid
4060 negative public key encodings) on by default,
4061 NO_NEG_PUBKEY_BUG can be set to disable it.
4064 *) New function c2i_ASN1_OBJECT() which acts on ASN1_OBJECT
4065 content octets. An i2c_ASN1_OBJECT is unnecessary because
4066 the encoding can be trivially obtained from the structure.
4069 *) crypto/err.c locking bugfix: Use write locks (CRYPTO_w_[un]lock),
4070 not read locks (CRYPTO_r_[un]lock).
4073 *) A first attempt at creating official support for shared
4074 libraries through configuration. I've kept it so the
4075 default is static libraries only, and the OpenSSL programs
4076 are always statically linked for now, but there are
4077 preparations for dynamic linking in place.
4078 This has been tested on Linux and Tru64.
4081 *) Randomness polling function for Win9x, as described in:
4082 Peter Gutmann, Software Generation of Practically Strong
4086 *) Fix so PRNG is seeded in req if using an already existing
4090 *) New options to smime application. -inform and -outform
4091 allow alternative formats for the S/MIME message including
4092 PEM and DER. The -content option allows the content to be
4093 specified separately. This should allow things like Netscape
4094 form signing output easier to verify.
4097 *) Fix the ASN1 encoding of tags using the 'long form'.
4100 *) New ASN1 functions, i2c_* and c2i_* for INTEGER and BIT
4101 STRING types. These convert content octets to and from the
4102 underlying type. The actual tag and length octets are
4103 already assumed to have been read in and checked. These
4104 are needed because all other string types have virtually
4105 identical handling apart from the tag. By having versions
4106 of the ASN1 functions that just operate on content octets
4107 IMPLICIT tagging can be handled properly. It also allows
4108 the ASN1_ENUMERATED code to be cut down because ASN1_ENUMERATED
4109 and ASN1_INTEGER are identical apart from the tag.
4112 *) Change the handling of OID objects as follows:
4114 - New object identifiers are inserted in objects.txt, following
4115 the syntax given in objects.README.
4116 - objects.pl is used to process obj_mac.num and create a new
4118 - obj_dat.pl is used to create a new obj_dat.h, using the data in
4121 This is currently kind of a hack, and the perl code in objects.pl
4122 isn't very elegant, but it works as I intended. The simplest way
4123 to check that it worked correctly is to look in obj_dat.h and
4124 check the array nid_objs and make sure the objects haven't moved
4125 around (this is important!). Additions are OK, as well as
4126 consistent name changes.
4129 *) Add BSD-style MD5-based passwords to 'openssl passwd' (option '-1').
4132 *) Addition of the command line parameter '-rand file' to 'openssl req'.
4133 The given file adds to whatever has already been seeded into the
4134 random pool through the RANDFILE configuration file option or
4135 environment variable, or the default random state file.
4138 *) mkstack.pl now sorts each macro group into lexical order.
4139 Previously the output order depended on the order the files
4140 appeared in the directory, resulting in needless rewriting
4144 *) Patches to make OpenSSL compile under Win32 again. Mostly
4145 work arounds for the VC++ problem that it treats func() as
4146 func(void). Also stripped out the parts of mkdef.pl that
4147 added extra typesafe functions: these no longer exist.
4150 *) Reorganisation of the stack code. The macros are now all
4151 collected in safestack.h . Each macro is defined in terms of
4152 a "stack macro" of the form SKM_<name>(type, a, b). The
4153 DEBUG_SAFESTACK is now handled in terms of function casts,
4154 this has the advantage of retaining type safety without the
4155 use of additional functions. If DEBUG_SAFESTACK is not defined
4156 then the non typesafe macros are used instead. Also modified the
4157 mkstack.pl script to handle the new form. Needs testing to see
4158 if which (if any) compilers it chokes and maybe make DEBUG_SAFESTACK
4159 the default if no major problems. Similar behaviour for ASN1_SET_OF
4160 and PKCS12_STACK_OF.
4163 *) When some versions of IIS use the 'NET' form of private key the
4164 key derivation algorithm is different. Normally MD5(password) is
4165 used as a 128 bit RC4 key. In the modified case
4166 MD5(MD5(password) + "SGCKEYSALT") is used insted. Added some
4167 new functions i2d_RSA_NET(), d2i_RSA_NET() etc which are the same
4168 as the old Netscape_RSA functions except they have an additional
4169 'sgckey' parameter which uses the modified algorithm. Also added
4170 an -sgckey command line option to the rsa utility. Thanks to
4171 Adrian Peck <bertie@ncipher.com> for posting details of the modified
4172 algorithm to openssl-dev.
4175 *) The evp_local.h macros were using 'c.##kname' which resulted in
4176 invalid expansion on some systems (SCO 5.0.5 for example).
4177 Corrected to 'c.kname'.
4178 [Phillip Porch <root@theporch.com>]
4180 *) New X509_get1_email() and X509_REQ_get1_email() functions that return
4181 a STACK of email addresses from a certificate or request, these look
4182 in the subject name and the subject alternative name extensions and
4183 omit any duplicate addresses.
4186 *) Re-implement BN_mod_exp2_mont using independent (and larger) windows.
4187 This makes DSA verification about 2 % faster.
4190 *) Increase maximum window size in BN_mod_exp_... to 6 bits instead of 5
4191 (meaning that now 2^5 values will be precomputed, which is only 4 KB
4192 plus overhead for 1024 bit moduli).
4193 This makes exponentiations about 0.5 % faster for 1024 bit
4194 exponents (as measured by "openssl speed rsa2048").
4197 *) Rename memory handling macros to avoid conflicts with other
4199 Malloc => OPENSSL_malloc
4200 Malloc_locked => OPENSSL_malloc_locked
4201 Realloc => OPENSSL_realloc
4202 Free => OPENSSL_free
4205 *) New function BN_mod_exp_mont_word for small bases (roughly 15%
4206 faster than BN_mod_exp_mont, i.e. 7% for a full DH exchange).
4209 *) CygWin32 support.
4210 [John Jarvie <jjarvie@newsguy.com>]
4212 *) The type-safe stack code has been rejigged. It is now only compiled
4213 in when OpenSSL is configured with the DEBUG_SAFESTACK option and
4214 by default all type-specific stack functions are "#define"d back to
4215 standard stack functions. This results in more streamlined output
4216 but retains the type-safety checking possibilities of the original
4220 *) The STACK code has been cleaned up, and certain type declarations
4221 that didn't make a lot of sense have been brought in line. This has
4222 also involved a cleanup of sorts in safestack.h to more correctly
4223 map type-safe stack functions onto their plain stack counterparts.
4224 This work has also resulted in a variety of "const"ifications of
4225 lots of the code, especially "_cmp" operations which should normally
4226 be prototyped with "const" parameters anyway.
4229 *) When generating bytes for the first time in md_rand.c, 'stir the pool'
4230 by seeding with STATE_SIZE dummy bytes (with zero entropy count).
4231 (The PRNG state consists of two parts, the large pool 'state' and 'md',
4232 where all of 'md' is used each time the PRNG is used, but 'state'
4233 is used only indexed by a cyclic counter. As entropy may not be
4234 well distributed from the beginning, 'md' is important as a
4235 chaining variable. However, the output function chains only half
4236 of 'md', i.e. 80 bits. ssleay_rand_add, on the other hand, chains
4237 all of 'md', and seeding with STATE_SIZE dummy bytes will result
4238 in all of 'state' being rewritten, with the new values depending
4239 on virtually all of 'md'. This overcomes the 80 bit limitation.)
4242 *) In ssl/s2_clnt.c and ssl/s3_clnt.c, call ERR_clear_error() when
4243 the handshake is continued after ssl_verify_cert_chain();
4244 otherwise, if SSL_VERIFY_NONE is set, remaining error codes
4245 can lead to 'unexplainable' connection aborts later.
4246 [Bodo Moeller; problem tracked down by Lutz Jaenicke]
4248 *) Major EVP API cipher revision.
4249 Add hooks for extra EVP features. This allows various cipher
4250 parameters to be set in the EVP interface. Support added for variable
4251 key length ciphers via the EVP_CIPHER_CTX_set_key_length() function and
4252 setting of RC2 and RC5 parameters.
4254 Modify EVP_OpenInit() and EVP_SealInit() to cope with variable key length
4257 Remove lots of duplicated code from the EVP library. For example *every*
4258 cipher init() function handles the 'iv' in the same way according to the
4259 cipher mode. They also all do nothing if the 'key' parameter is NULL and
4260 for CFB and OFB modes they zero ctx->num.
4262 New functionality allows removal of S/MIME code RC2 hack.
4264 Most of the routines have the same form and so can be declared in terms
4267 By shifting this to the top level EVP_CipherInit() it can be removed from
4268 all individual ciphers. If the cipher wants to handle IVs or keys
4269 differently it can set the EVP_CIPH_CUSTOM_IV or EVP_CIPH_ALWAYS_CALL_INIT
4272 Change lots of functions like EVP_EncryptUpdate() to now return a
4273 value: although software versions of the algorithms cannot fail
4274 any installed hardware versions can.
4277 *) Implement SSL_OP_TLS_ROLLBACK_BUG: In ssl3_get_client_key_exchange, if
4278 this option is set, tolerate broken clients that send the negotiated
4279 protocol version number instead of the requested protocol version
4283 *) Call dh_tmp_cb (set by ..._TMP_DH_CB) with correct 'is_export' flag;
4284 i.e. non-zero for export ciphersuites, zero otherwise.
4285 Previous versions had this flag inverted, inconsistent with
4286 rsa_tmp_cb (..._TMP_RSA_CB).
4287 [Bodo Moeller; problem reported by Amit Chopra]
4289 *) Add missing DSA library text string. Work around for some IIS
4290 key files with invalid SEQUENCE encoding.
4293 *) Add a document (doc/standards.txt) that list all kinds of standards
4294 and so on that are implemented in OpenSSL.
4297 *) Enhance c_rehash script. Old version would mishandle certificates
4298 with the same subject name hash and wouldn't handle CRLs at all.
4299 Added -fingerprint option to crl utility, to support new c_rehash
4303 *) Eliminate non-ANSI declarations in crypto.h and stack.h.
4306 *) Fix for SSL server purpose checking. Server checking was
4307 rejecting certificates which had extended key usage present
4308 but no ssl client purpose.
4309 [Steve Henson, reported by Rene Grosser <grosser@hisolutions.com>]
4311 *) Make PKCS#12 code work with no password. The PKCS#12 spec
4312 is a little unclear about how a blank password is handled.
4313 Since the password in encoded as a BMPString with terminating
4314 double NULL a zero length password would end up as just the
4315 double NULL. However no password at all is different and is
4316 handled differently in the PKCS#12 key generation code. NS
4317 treats a blank password as zero length. MSIE treats it as no
4318 password on export: but it will try both on import. We now do
4319 the same: PKCS12_parse() tries zero length and no password if
4320 the password is set to "" or NULL (NULL is now a valid password:
4321 it wasn't before) as does the pkcs12 application.
4324 *) Bugfixes in apps/x509.c: Avoid a memory leak; and don't use
4325 perror when PEM_read_bio_X509_REQ fails, the error message must
4326 be obtained from the error queue.
4329 *) Avoid 'thread_hash' memory leak in crypto/err/err.c by freeing
4330 it in ERR_remove_state if appropriate, and change ERR_get_state
4331 accordingly to avoid race conditions (this is necessary because
4332 thread_hash is no longer constant once set).
4335 *) Bugfix for linux-elf makefile.one.
4338 *) RSA_get_default_method() will now cause a default
4339 RSA_METHOD to be chosen if one doesn't exist already.
4340 Previously this was only set during a call to RSA_new()
4341 or RSA_new_method(NULL) meaning it was possible for
4342 RSA_get_default_method() to return NULL.
4345 *) Added native name translation to the existing DSO code
4346 that will convert (if the flag to do so is set) filenames
4347 that are sufficiently small and have no path information
4348 into a canonical native form. Eg. "blah" converted to
4349 "libblah.so" or "blah.dll" etc.
4352 *) New function ERR_error_string_n(e, buf, len) which is like
4353 ERR_error_string(e, buf), but writes at most 'len' bytes
4354 including the 0 terminator. For ERR_error_string_n, 'buf'
4356 [Damien Miller <djm@mindrot.org>, Bodo Moeller]
4358 *) CONF library reworked to become more general. A new CONF
4359 configuration file reader "class" is implemented as well as a
4360 new functions (NCONF_*, for "New CONF") to handle it. The now
projects / openssl.git / blob