5 Changes between 0.9.7f and 0.9.8 [xx XXX xxxx]
7 *) Use SHA-1 instead of MD5 as the default digest algorithm for
8 the apps/openssl applications.
11 *) Compile clean with "-Wall -Wmissing-prototypes
12 -Wstrict-prototypes -Wmissing-declarations -Werror". Currently
13 DEBUG_SAFESTACK must also be set.
16 *) Change ./Configure so that certain algorithms can be disabled by default.
17 The new counterpiece to "no-xxx" is "enable-xxx".
19 The patented RC5 and MDC2 algorithms will now be disabled unless
20 "enable-rc5" and "enable-mdc2", respectively, are specified.
22 (IDEA remains enabled despite being patented. This is because IDEA
23 is frequently required for interoperability, and there is no license
24 fee for non-commercial use. As before, "no-idea" can be used to
25 avoid this algorithm.)
29 *) Add processing of proxy certificates (see RFC 3820). This work was
30 sponsored by KTH (The Royal Institute of Technology in Stockholm) and
31 EGEE (Enabling Grids for E-science in Europe).
34 *) RC4 performance overhaul on modern architectures/implementations, such
35 as Intel P4, IA-64 and AMD64.
38 *) New utility extract-section.pl. This can be used specify an alternative
39 section number in a pod file instead of having to treat each file as
40 a separate case in Makefile. This can be done by adding two lines to the
43 =for comment openssl_section:XXX
45 The blank line is mandatory.
49 *) New arguments -certform, -keyform and -pass for s_client and s_server
50 to allow alternative format key and certificate files and passphrase
54 *) New structure X509_VERIFY_PARAM which combines current verify parameters,
55 update associated structures and add various utility functions.
57 Add new policy related verify parameters, include policy checking in
58 standard verify code. Enhance 'smime' application with extra parameters
59 to support policy checking and print out.
62 *) Add a new engine to support VIA PadLock ACE extensions in the VIA C3
63 Nehemiah processors. These extensions support AES encryption in hardware
64 as well as RNG (though RNG support is currently disabled).
65 [Michal Ludvig <michal@logix.cz>, with help from Andy Polyakov]
67 *) Deprecate BN_[get|set]_params() functions (they were ignored internally).
70 *) New FIPS 180-2 algorithms, SHA-224/-256/-384/-512 are implemented.
71 [Andy Polyakov and a number of other people]
73 *) Improved PowerPC platform support. Most notably BIGNUM assembler
74 implementation contributed by IBM.
75 [Suresh Chari, Peter Waltenberg, Andy Polyakov]
77 *) The new 'RSA_generate_key_ex' function now takes a BIGNUM for the public
78 exponent rather than 'unsigned long'. There is a corresponding change to
79 the new 'rsa_keygen' element of the RSA_METHOD structure.
80 [Jelte Jansen, Geoff Thorpe]
82 *) Functionality for creating the initial serial number file is now
83 moved from CA.pl to the 'ca' utility with a new option -create_serial.
85 (Before OpenSSL 0.9.7e, CA.pl used to initialize the serial
86 number file to 1, which is bound to cause problems. To avoid
87 the problems while respecting compatibility between different 0.9.7
88 patchlevels, 0.9.7e employed 'openssl x509 -next_serial' in
89 CA.pl for serial number initialization. With the new release 0.9.8,
90 we can fix the problem directly in the 'ca' utility.)
93 *) Reduced header interdepencies by declaring more opaque objects in
94 ossl_typ.h. As a consequence, including some headers (eg. engine.h) will
95 give fewer recursive includes, which could break lazy source code - so
96 this change is covered by the OPENSSL_NO_DEPRECATED symbol. As always,
97 developers should define this symbol when building and using openssl to
98 ensure they track the recommended behaviour, interfaces, [etc], but
99 backwards-compatible behaviour prevails when this isn't defined.
102 *) New function X509_POLICY_NODE_print() which prints out policy nodes.
105 *) Add new EVP function EVP_CIPHER_CTX_rand_key and associated functionality.
106 This will generate a random key of the appropriate length based on the
107 cipher context. The EVP_CIPHER can provide its own random key generation
108 routine to support keys of a specific form. This is used in the des and
109 3des routines to generate a key of the correct parity. Update S/MIME
110 code to use new functions and hence generate correct parity DES keys.
111 Add EVP_CHECK_DES_KEY #define to return an error if the key is not
112 valid (weak or incorrect parity).
115 *) Add a local set of CRLs that can be used by X509_verify_cert() as well
116 as looking them up. This is useful when the verified structure may contain
117 CRLs, for example PKCS#7 signedData. Modify PKCS7_verify() to use any CRLs
118 present unless the new PKCS7_NO_CRL flag is asserted.
121 *) Extend ASN1 oid configuration module. It now additionally accepts the
124 shortName = some long name, 1.2.3.4
127 *) Reimplemented the BN_CTX implementation. There is now no more static
128 limitation on the number of variables it can handle nor the depth of the
129 "stack" handling for BN_CTX_start()/BN_CTX_end() pairs. The stack
130 information can now expand as required, and rather than having a single
131 static array of bignums, BN_CTX now uses a linked-list of such arrays
132 allowing it to expand on demand whilst maintaining the usefulness of
136 *) Add a missing BN_CTX parameter to the 'rsa_mod_exp' callback in RSA_METHOD
137 to allow all RSA operations to function using a single BN_CTX.
140 *) Preliminary support for certificate policy evaluation and checking. This
141 is initially intended to pass the tests outlined in "Conformance Testing
142 of Relying Party Client Certificate Path Processing Logic" v1.07.
145 *) bn_dup_expand() has been deprecated, it was introduced in 0.9.7 and
146 remained unused and not that useful. A variety of other little bignum
147 tweaks and fixes have also been made continuing on from the audit (see
151 *) Constify all or almost all d2i, c2i, s2i and r2i functions, along with
152 associated ASN1, EVP and SSL functions and old ASN1 macros.
155 *) BN_zero() only needs to set 'top' and 'neg' to zero for correct results,
156 and this should never fail. So the return value from the use of
157 BN_set_word() (which can fail due to needless expansion) is now deprecated;
158 if OPENSSL_NO_DEPRECATED is defined, BN_zero() is a void macro.
161 *) BN_CTX_get() should return zero-valued bignums, providing the same
162 initialised value as BN_new().
163 [Geoff Thorpe, suggested by Ulf Möller]
165 *) Support for inhibitAnyPolicy certificate extension.
168 *) An audit of the BIGNUM code is underway, for which debugging code is
169 enabled when BN_DEBUG is defined. This makes stricter enforcements on what
170 is considered valid when processing BIGNUMs, and causes execution to
171 assert() when a problem is discovered. If BN_DEBUG_RAND is defined,
172 further steps are taken to deliberately pollute unused data in BIGNUM
173 structures to try and expose faulty code further on. For now, openssl will
174 (in its default mode of operation) continue to tolerate the inconsistent
175 forms that it has tolerated in the past, but authors and packagers should
176 consider trying openssl and their own applications when compiled with
177 these debugging symbols defined. It will help highlight potential bugs in
178 their own code, and will improve the test coverage for OpenSSL itself. At
179 some point, these tighter rules will become openssl's default to improve
180 maintainability, though the assert()s and other overheads will remain only
181 in debugging configurations. See bn.h for more details.
182 [Geoff Thorpe, Nils Larsch, Ulf Möller]
184 *) BN_CTX_init() has been deprecated, as BN_CTX is an opaque structure
185 that can only be obtained through BN_CTX_new() (which implicitly
186 initialises it). The presence of this function only made it possible
187 to overwrite an existing structure (and cause memory leaks).
190 *) Because of the callback-based approach for implementing LHASH as a
191 template type, lh_insert() adds opaque objects to hash-tables and
192 lh_doall() or lh_doall_arg() are typically used with a destructor callback
193 to clean up those corresponding objects before destroying the hash table
194 (and losing the object pointers). So some over-zealous constifications in
195 LHASH have been relaxed so that lh_insert() does not take (nor store) the
196 objects as "const" and the lh_doall[_arg] callback wrappers are not
197 prototyped to have "const" restrictions on the object pointers they are
198 given (and so aren't required to cast them away any more).
201 *) The tmdiff.h API was so ugly and minimal that our own timing utility
202 (speed) prefers to use its own implementation. The two implementations
203 haven't been consolidated as yet (volunteers?) but the tmdiff API has had
204 its object type properly exposed (MS_TM) instead of casting to/from "char
205 *". This may still change yet if someone realises MS_TM and "ms_time_***"
206 aren't necessarily the greatest nomenclatures - but this is what was used
207 internally to the implementation so I've used that for now.
210 *) Ensure that deprecated functions do not get compiled when
211 OPENSSL_NO_DEPRECATED is defined. Some "openssl" subcommands and a few of
212 the self-tests were still using deprecated key-generation functions so
213 these have been updated also.
216 *) Reorganise PKCS#7 code to separate the digest location functionality
217 into PKCS7_find_digest(), digest addtion into PKCS7_bio_add_digest().
218 New function PKCS7_set_digest() to set the digest type for PKCS#7
219 digestedData type. Add additional code to correctly generate the
220 digestedData type and add support for this type in PKCS7 initialization
224 *) New function PKCS7_set0_type_other() this initializes a PKCS7
225 structure of type "other".
228 *) Fix prime generation loop in crypto/bn/bn_prime.pl by making
229 sure the loop does correctly stop and breaking ("division by zero")
230 modulus operations are not performed. The (pre-generated) prime
231 table crypto/bn/bn_prime.h was already correct, but it could not be
232 re-generated on some platforms because of the "division by zero"
233 situation in the script.
234 [Ralf S. Engelschall]
236 *) Update support for ECC-based TLS ciphersuites according to
237 draft-ietf-tls-ecc-03.txt: the KDF1 key derivation function with
238 SHA-1 now is only used for "small" curves (where the
239 representation of a field element takes up to 24 bytes); for
240 larger curves, the field element resulting from ECDH is directly
241 used as premaster secret.
242 [Douglas Stebila (Sun Microsystems Laboratories)]
244 *) Add code for kP+lQ timings to crypto/ec/ectest.c, and add SEC2
245 curve secp160r1 to the tests.
246 [Douglas Stebila (Sun Microsystems Laboratories)]
248 *) Add the possibility to load symbols globally with DSO.
249 [Götz Babin-Ebell <babin-ebell@trustcenter.de> via Richard Levitte]
251 *) Add the functions ERR_set_mark() and ERR_pop_to_mark() for better
252 control of the error stack.
255 *) Add support for STORE in ENGINE.
258 *) Add the STORE type. The intention is to provide a common interface
259 to certificate and key stores, be they simple file-based stores, or
260 HSM-type store, or LDAP stores, or...
261 NOTE: The code is currently UNTESTED and isn't really used anywhere.
264 *) Add a generic structure called OPENSSL_ITEM. This can be used to
265 pass a list of arguments to any function as well as provide a way
266 for a function to pass data back to the caller.
269 *) Add the functions BUF_strndup() and BUF_memdup(). BUF_strndup()
270 works like BUF_strdup() but can be used to duplicate a portion of
271 a string. The copy gets NUL-terminated. BUF_memdup() duplicates
275 *) Add the function sk_find_ex() which works like sk_find(), but will
276 return an index to an element even if an exact match couldn't be
277 found. The index is guaranteed to point at the element where the
278 searched-for key would be inserted to preserve sorting order.
281 *) Add the function OBJ_bsearch_ex() which works like OBJ_bsearch() but
282 takes an extra flags argument for optional functionality. Currently,
283 the following flags are defined:
285 OBJ_BSEARCH_VALUE_ON_NOMATCH
286 This one gets OBJ_bsearch_ex() to return a pointer to the first
287 element where the comparing function returns a negative or zero
290 OBJ_BSEARCH_FIRST_VALUE_ON_MATCH
291 This one gets OBJ_bsearch_ex() to return a pointer to the first
292 element where the comparing function returns zero. This is useful
293 if there are more than one element where the comparing function
297 *) Make it possible to create self-signed certificates with 'openssl ca'
298 in such a way that the self-signed certificate becomes part of the
299 CA database and uses the same mechanisms for serial number generation
300 as all other certificate signing. The new flag '-selfsign' enables
301 this functionality. Adapt CA.sh and CA.pl.in.
304 *) Add functionality to check the public key of a certificate request
305 against a given private. This is useful to check that a certificate
306 request can be signed by that key (self-signing).
309 *) Make it possible to have multiple active certificates with the same
310 subject in the CA index file. This is done only if the keyword
311 'unique_subject' is set to 'no' in the main CA section (default
312 if 'CA_default') of the configuration file. The value is saved
313 with the database itself in a separate index attribute file,
314 named like the index file with '.attr' appended to the name.
317 *) Generate muti valued AVAs using '+' notation in config files for
321 *) Support for nameConstraints certificate extension.
324 *) Support for policyConstraints certificate extension.
327 *) Support for policyMappings certificate extension.
330 *) Make sure the default DSA_METHOD implementation only uses its
331 dsa_mod_exp() and/or bn_mod_exp() handlers if they are non-NULL,
332 and change its own handlers to be NULL so as to remove unnecessary
333 indirection. This lets alternative implementations fallback to the
334 default implementation more easily.
337 *) Support for directoryName in GeneralName related extensions
341 *) Make it possible to link applications using Makefile.shared.
342 Make that possible even when linking against static libraries!
345 *) Support for single pass processing for S/MIME signing. This now
346 means that S/MIME signing can be done from a pipe, in addition
347 cleartext signing (multipart/signed type) is effectively streaming
348 and the signed data does not need to be all held in memory.
350 This is done with a new flag PKCS7_STREAM. When this flag is set
351 PKCS7_sign() only initializes the PKCS7 structure and the actual signing
352 is done after the data is output (and digests calculated) in
356 *) Add full support for -rpath/-R, both in shared libraries and
357 applications, at least on the platforms where it's known how
361 *) In crypto/ec/ec_mult.c, implement fast point multiplication with
362 precomputation, based on wNAF splitting: EC_GROUP_precompute_mult()
363 will now compute a table of multiples of the generator that
364 makes subsequent invocations of EC_POINTs_mul() or EC_POINT_mul()
365 faster (notably in the case of a single point multiplication,
367 [Nils Larsch, Bodo Moeller]
369 *) IPv6 support for certificate extensions. The various extensions
370 which use the IP:a.b.c.d can now take IPv6 addresses using the
371 formats of RFC1884 2.2 . IPv6 addresses are now also displayed
375 *) Added an ENGINE that implements RSA by performing private key
376 exponentiations with the GMP library. The conversions to and from
377 GMP's mpz_t format aren't optimised nor are any montgomery forms
378 cached, and on x86 it appears OpenSSL's own performance has caught up.
379 However there are likely to be other architectures where GMP could
380 provide a boost. This ENGINE is not built in by default, but it can be
381 specified at Configure time and should be accompanied by the necessary
382 linker additions, eg;
383 ./config -DOPENSSL_USE_GMP -lgmp
386 *) "openssl engine" will not display ENGINE/DSO load failure errors when
387 testing availability of engines with "-t" - the old behaviour is
388 produced by increasing the feature's verbosity with "-tt".
391 *) ECDSA routines: under certain error conditions uninitialized BN objects
392 could be freed. Solution: make sure initialization is performed early
393 enough. (Reported and fix supplied by Nils Larsch <nla@trustcenter.de>
397 *) Key-generation can now be implemented in RSA_METHOD, DSA_METHOD
398 and DH_METHOD (eg. by ENGINE implementations) to override the normal
399 software implementations. For DSA and DH, parameter generation can
400 also be overriden by providing the appropriate method callbacks.
403 *) Change the "progress" mechanism used in key-generation and
404 primality testing to functions that take a new BN_GENCB pointer in
405 place of callback/argument pairs. The new API functions have "_ex"
406 postfixes and the older functions are reimplemented as wrappers for
407 the new ones. The OPENSSL_NO_DEPRECATED symbol can be used to hide
408 declarations of the old functions to help (graceful) attempts to
409 migrate to the new functions. Also, the new key-generation API
410 functions operate on a caller-supplied key-structure and return
411 success/failure rather than returning a key or NULL - this is to
412 help make "keygen" another member function of RSA_METHOD etc.
414 Example for using the new callback interface:
416 int (*my_callback)(int a, int b, BN_GENCB *cb) = ...;
420 BN_GENCB_set(&my_cb, my_callback, my_arg);
422 return BN_is_prime_ex(some_bignum, BN_prime_checks, NULL, &cb);
423 /* For the meaning of a, b in calls to my_callback(), see the
424 * documentation of the function that calls the callback.
425 * cb will point to my_cb; my_arg can be retrieved as cb->arg.
426 * my_callback should return 1 if it wants BN_is_prime_ex()
427 * to continue, or 0 to stop.
432 *) Change the ZLIB compression method to be stateful, and make it
433 available to TLS with the number defined in
434 draft-ietf-tls-compression-04.txt.
437 *) Add the ASN.1 structures and functions for CertificatePair, which
438 is defined as follows (according to X.509_4thEditionDraftV6.pdf):
440 CertificatePair ::= SEQUENCE {
441 forward [0] Certificate OPTIONAL,
442 reverse [1] Certificate OPTIONAL,
443 -- at least one of the pair shall be present -- }
445 Also implement the PEM functions to read and write certificate
446 pairs, and defined the PEM tag as "CERTIFICATE PAIR".
448 This needed to be defined, mostly for the sake of the LDAP
449 attribute crossCertificatePair, but may prove useful elsewhere as
453 *) Make it possible to inhibit symlinking of shared libraries in
454 Makefile.shared, for Cygwin's sake.
457 *) Extend the BIGNUM API by creating new macros that behave like
460 void BN_set_sign(BIGNUM *a, int neg);
461 int BN_get_sign(const BIGNUM *a);
463 and avoid the need to access 'a->neg' directly in applications.
464 [Nils Larsch <nla@trustcenter.de>]
466 *) Implement fast modular reduction for pseudo-Mersenne primes
467 used in NIST curves (crypto/bn/bn_nist.c, crypto/ec/ecp_nist.c).
468 EC_GROUP_new_curve_GFp() will now automatically use this
470 [Nils Larsch <nla@trustcenter.de>]
472 *) Add new lock type (CRYPTO_LOCK_BN).
475 *) Change the ENGINE framework to automatically load engines
476 dynamically from specific directories unless they could be
477 found to already be built in or loaded. Move all the
478 current engines except for the cryptodev one to a new
480 The engines in engines/ are built as shared libraries if
481 the "shared" options was given to ./Configure or ./config.
482 Otherwise, they are inserted in libcrypto.a.
483 /usr/local/ssl/engines is the default directory for dynamic
484 engines, but that can be overriden at configure time through
485 the usual use of --prefix and/or --openssldir, and at run
486 time with the environment variable OPENSSL_ENGINES.
487 [Geoff Thorpe and Richard Levitte]
489 *) Add Makefile.shared, a helper makefile to build shared
490 libraries. Addapt Makefile.org.
493 *) Add version info to Win32 DLLs.
494 [Peter 'Luna' Runestig" <peter@runestig.com>]
496 *) Add new 'medium level' PKCS#12 API. Certificates and keys
497 can be added using this API to created arbitrary PKCS#12
498 files while avoiding the low level API.
500 New options to PKCS12_create(), key or cert can be NULL and
501 will then be omitted from the output file. The encryption
502 algorithm NIDs can be set to -1 for no encryption, the mac
503 iteration count can be set to 0 to omit the mac.
505 Enhance pkcs12 utility by making the -nokeys and -nocerts
506 options work when creating a PKCS#12 file. New option -nomac
507 to omit the mac, NONE can be set for an encryption algorithm.
508 New code is modified to use the enhanced PKCS12_create()
509 instead of the low level API.
512 *) Extend ASN1 encoder to support indefinite length constructed
513 encoding. This can output sequences tags and octet strings in
514 this form. Modify pk7_asn1.c to support indefinite length
515 encoding. This is experimental and needs additional code to
516 be useful, such as an ASN1 bio and some enhanced streaming
519 Extend template encode functionality so that tagging is passed
520 down to the template encoder.
523 *) Let 'openssl req' fail if an argument to '-newkey' is not
524 recognized instead of using RSA as a default.
527 *) Add support for ECC-based ciphersuites from draft-ietf-tls-ecc-01.txt.
528 As these are not official, they are not included in "ALL";
529 the "ECCdraft" ciphersuite group alias can be used to select them.
530 [Vipul Gupta and Sumit Gupta (Sun Microsystems Laboratories)]
532 *) Add ECDH engine support.
533 [Nils Gura and Douglas Stebila (Sun Microsystems Laboratories)]
535 *) Add ECDH in new directory crypto/ecdh/.
536 [Douglas Stebila (Sun Microsystems Laboratories)]
538 *) Let BN_rand_range() abort with an error after 100 iterations
539 without success (which indicates a broken PRNG).
542 *) Change BN_mod_sqrt() so that it verifies that the input value
543 is really the square of the return value. (Previously,
544 BN_mod_sqrt would show GIGO behaviour.)
547 *) Add named elliptic curves over binary fields from X9.62, SECG,
548 and WAP/WTLS; add OIDs that were still missing.
550 [Sheueling Chang Shantz and Douglas Stebila
551 (Sun Microsystems Laboratories)]
553 *) Extend the EC library for elliptic curves over binary fields
554 (new files ec2_smpl.c, ec2_smpt.c, ec2_mult.c in crypto/ec/).
557 EC_GF2m_simple_method
561 EC_GROUP_new_curve_GF2m
562 EC_GROUP_set_curve_GF2m
563 EC_GROUP_get_curve_GF2m
564 EC_POINT_set_affine_coordinates_GF2m
565 EC_POINT_get_affine_coordinates_GF2m
566 EC_POINT_set_compressed_coordinates_GF2m
568 Point compression for binary fields is disabled by default for
569 patent reasons (compile with OPENSSL_EC_BIN_PT_COMP defined to
572 As binary polynomials are represented as BIGNUMs, various members
573 of the EC_GROUP and EC_POINT data structures can be shared
574 between the implementations for prime fields and binary fields;
575 the above ..._GF2m functions (except for EX_GROUP_new_curve_GF2m)
576 are essentially identical to their ..._GFp counterparts.
577 (For simplicity, the '..._GFp' prefix has been dropped from
578 various internal method names.)
580 An internal 'field_div' method (similar to 'field_mul' and
581 'field_sqr') has been added; this is used only for binary fields.
583 [Sheueling Chang Shantz and Douglas Stebila
584 (Sun Microsystems Laboratories)]
586 *) Optionally dispatch EC_POINT_mul(), EC_POINT_precompute_mult()
587 through methods ('mul', 'precompute_mult').
589 The generic implementations (now internally called 'ec_wNAF_mul'
590 and 'ec_wNAF_precomputed_mult') remain the default if these
591 methods are undefined.
593 [Sheueling Chang Shantz and Douglas Stebila
594 (Sun Microsystems Laboratories)]
596 *) New function EC_GROUP_get_degree, which is defined through
597 EC_METHOD. For curves over prime fields, this returns the bit
598 length of the modulus.
600 [Sheueling Chang Shantz and Douglas Stebila
601 (Sun Microsystems Laboratories)]
603 *) New functions EC_GROUP_dup, EC_POINT_dup.
604 (These simply call ..._new and ..._copy).
606 [Sheueling Chang Shantz and Douglas Stebila
607 (Sun Microsystems Laboratories)]
609 *) Add binary polynomial arithmetic software in crypto/bn/bn_gf2m.c.
610 Polynomials are represented as BIGNUMs (where the sign bit is not
611 used) in the following functions [macros]:
614 BN_GF2m_sub [= BN_GF2m_add]
615 BN_GF2m_mod [wrapper for BN_GF2m_mod_arr]
616 BN_GF2m_mod_mul [wrapper for BN_GF2m_mod_mul_arr]
617 BN_GF2m_mod_sqr [wrapper for BN_GF2m_mod_sqr_arr]
619 BN_GF2m_mod_exp [wrapper for BN_GF2m_mod_exp_arr]
620 BN_GF2m_mod_sqrt [wrapper for BN_GF2m_mod_sqrt_arr]
621 BN_GF2m_mod_solve_quad [wrapper for BN_GF2m_mod_solve_quad_arr]
622 BN_GF2m_cmp [= BN_ucmp]
624 (Note that only the 'mod' functions are actually for fields GF(2^m).
625 BN_GF2m_add() is misnomer, but this is for the sake of consistency.)
627 For some functions, an the irreducible polynomial defining a
628 field can be given as an 'unsigned int[]' with strictly
629 decreasing elements giving the indices of those bits that are set;
630 i.e., p[] represents the polynomial
631 f(t) = t^p[0] + t^p[1] + ... + t^p[k]
633 p[0] > p[1] > ... > p[k] = 0.
634 This applies to the following functions:
639 BN_GF2m_mod_inv_arr [wrapper for BN_GF2m_mod_inv]
640 BN_GF2m_mod_div_arr [wrapper for BN_GF2m_mod_div]
643 BN_GF2m_mod_solve_quad_arr
647 Conversion can be performed by the following functions:
652 bntest.c has additional tests for binary polynomial arithmetic.
654 Two implementations for BN_GF2m_mod_div() are available.
655 The default algorithm simply uses BN_GF2m_mod_inv() and
656 BN_GF2m_mod_mul(). The alternative algorithm is compiled in only
657 if OPENSSL_SUN_GF2M_DIV is defined (patent pending; read the
658 copyright notice in crypto/bn/bn_gf2m.c before enabling it).
660 [Sheueling Chang Shantz and Douglas Stebila
661 (Sun Microsystems Laboratories)]
663 *) Add new error code 'ERR_R_DISABLED' that can be used when some
664 functionality is disabled at compile-time.
665 [Douglas Stebila <douglas.stebila@sun.com>]
667 *) Change default behaviour of 'openssl asn1parse' so that more
668 information is visible when viewing, e.g., a certificate:
670 Modify asn1_parse2 (crypto/asn1/asn1_par.c) so that in non-'dump'
671 mode the content of non-printable OCTET STRINGs is output in a
672 style similar to INTEGERs, but with '[HEX DUMP]' prepended to
673 avoid the appearance of a printable string.
674 [Nils Larsch <nla@trustcenter.de>]
676 *) Add 'asn1_flag' and 'asn1_form' member to EC_GROUP with access
678 EC_GROUP_set_asn1_flag()
679 EC_GROUP_get_asn1_flag()
680 EC_GROUP_set_point_conversion_form()
681 EC_GROUP_get_point_conversion_form()
682 These control ASN1 encoding details:
683 - Curves (i.e., groups) are encoded explicitly unless asn1_flag
684 has been set to OPENSSL_EC_NAMED_CURVE.
685 - Points are encoded in uncompressed form by default; options for
686 asn1_for are as for point2oct, namely
687 POINT_CONVERSION_COMPRESSED
688 POINT_CONVERSION_UNCOMPRESSED
689 POINT_CONVERSION_HYBRID
691 Also add 'seed' and 'seed_len' members to EC_GROUP with access
695 EC_GROUP_get_seed_len()
696 This is used only for ASN1 purposes (so far).
697 [Nils Larsch <nla@trustcenter.de>]
699 *) Add 'field_type' member to EC_METHOD, which holds the NID
700 of the appropriate field type OID. The new function
701 EC_METHOD_get_field_type() returns this value.
702 [Nils Larsch <nla@trustcenter.de>]
709 providing useful interfaces to EC_POINT_point2oct() and
710 EC_POINT_oct2point().
711 [Nils Larsch <nla@trustcenter.de>]
713 *) Change internals of the EC library so that the functions
714 EC_GROUP_set_generator()
715 EC_GROUP_get_generator()
717 EC_GROUP_get_cofactor()
718 are implemented directly in crypto/ec/ec_lib.c and not dispatched
719 to methods, which would lead to unnecessary code duplication when
720 adding different types of curves.
721 [Nils Larsch <nla@trustcenter.de> with input by Bodo Moeller]
723 *) Implement compute_wNAF (crypto/ec/ec_mult.c) without BIGNUM
724 arithmetic, and such that modified wNAFs are generated
725 (which avoid length expansion in many cases).
728 *) Add a function EC_GROUP_check_discriminant() (defined via
729 EC_METHOD) that verifies that the curve discriminant is non-zero.
731 Add a function EC_GROUP_check() that makes some sanity tests
732 on a EC_GROUP, its generator and order. This includes
733 EC_GROUP_check_discriminant().
734 [Nils Larsch <nla@trustcenter.de>]
736 *) Add ECDSA in new directory crypto/ecdsa/.
738 Add applications 'openssl ecparam' and 'openssl ecdsa'
739 (these are based on 'openssl dsaparam' and 'openssl dsa').
741 ECDSA support is also included in various other files across the
742 library. Most notably,
743 - 'openssl req' now has a '-newkey ecdsa:file' option;
744 - EVP_PKCS82PKEY (crypto/evp/evp_pkey.c) now can handle ECDSA;
745 - X509_PUBKEY_get (crypto/asn1/x_pubkey.c) and
746 d2i_PublicKey (crypto/asn1/d2i_pu.c) have been modified to make
747 them suitable for ECDSA where domain parameters must be
748 extracted before the specific public key;
749 - ECDSA engine support has been added.
750 [Nils Larsch <nla@trustcenter.de>]
752 *) Include some named elliptic curves, and add OIDs from X9.62,
753 SECG, and WAP/WTLS. Each curve can be obtained from the new
755 EC_GROUP_new_by_nid(),
756 and the list of available named curves can be obtained with
757 EC_get_builtin_curves().
758 Also add a 'curve_name' member to EC_GROUP objects, which can be
762 [Nils Larsch <nla@trustcenter.de, Bodo Moeller]
764 *) Remove a few calls to bn_wexpand() in BN_sqr() (the one in there
765 was actually never needed) and in BN_mul(). The removal in BN_mul()
766 required a small change in bn_mul_part_recursive() and the addition
767 of the functions bn_cmp_part_words(), bn_sub_part_words() and
768 bn_add_part_words(), which do the same thing as bn_cmp_words(),
769 bn_sub_words() and bn_add_words() except they take arrays with
773 Changes between 0.9.7f and 0.9.7g [XX xxx xxxx]
775 *) Undo Cygwin change.
778 Changes between 0.9.7e and 0.9.7f [22 Mar 2005]
780 *) Use (SSL_RANDOM_VALUE - 4) bytes of pseudo random data when generating
781 server and client random values. Previously
782 (SSL_RANDOM_VALUE - sizeof(time_t)) would be used which would result in
783 less random data when sizeof(time_t) > 4 (some 64 bit platforms).
785 This change has negligible security impact because:
787 1. Server and client random values still have 24 bytes of pseudo random
790 2. Server and client random values are sent in the clear in the initial
793 3. The master secret is derived using the premaster secret (48 bytes in
794 size for static RSA ciphersuites) as well as client server and random
797 The OpenSSL team would like to thank the UK NISCC for bringing this issue
800 [Stephen Henson, reported by UK NISCC]
802 *) Use Windows randomness collection on Cygwin.
805 *) Fix hang in EGD/PRNGD query when communication socket is closed
806 prematurely by EGD/PRNGD.
807 [Darren Tucker <dtucker@zip.com.au> via Lutz Jänicke, resolves #1014]
809 *) Prompt for pass phrases when appropriate for PKCS12 input format.
812 *) Back-port of selected performance improvements from development
813 branch, as well as improved support for PowerPC platforms.
816 *) Add lots of checks for memory allocation failure, error codes to indicate
817 failure and freeing up memory if a failure occurs.
818 [Nauticus Networks SSL Team <openssl@nauticusnet.com>, Steve Henson]
820 *) Add new -passin argument to dgst.
823 *) Perform some character comparisons of different types in X509_NAME_cmp:
824 this is needed for some certificates that reencode DNs into UTF8Strings
825 (in violation of RFC3280) and can't or wont issue name rollover
829 *) Make an explicit check during certificate validation to see that
830 the CA setting in each certificate on the chain is correct. As a
831 side effect always do the following basic checks on extensions,
832 not just when there's an associated purpose to the check:
834 - if there is an unhandled critical extension (unless the user
835 has chosen to ignore this fault)
836 - if the path length has been exceeded (if one is set at all)
837 - that certain extensions fit the associated purpose (if one has
841 Changes between 0.9.7d and 0.9.7e [25 Oct 2004]
843 *) Avoid a race condition when CRLs are checked in a multi threaded
844 environment. This would happen due to the reordering of the revoked
845 entries during signature checking and serial number lookup. Now the
846 encoding is cached and the serial number sort performed under a lock.
847 Add new STACK function sk_is_sorted().
850 *) Add Delta CRL to the extension code.
853 *) Various fixes to s3_pkt.c so alerts are sent properly.
854 [David Holmes <d.holmes@f5.com>]
856 *) Reduce the chances of duplicate issuer name and serial numbers (in
857 violation of RFC3280) using the OpenSSL certificate creation utilities.
858 This is done by creating a random 64 bit value for the initial serial
859 number when a serial number file is created or when a self signed
860 certificate is created using 'openssl req -x509'. The initial serial
861 number file is created using 'openssl x509 -next_serial' in CA.pl
862 rather than being initialized to 1.
865 Changes between 0.9.7c and 0.9.7d [17 Mar 2004]
867 *) Fix null-pointer assignment in do_change_cipher_spec() revealed
868 by using the Codenomicon TLS Test Tool (CAN-2004-0079)
869 [Joe Orton, Steve Henson]
871 *) Fix flaw in SSL/TLS handshaking when using Kerberos ciphersuites
873 [Joe Orton, Steve Henson]
875 *) Make it possible to have multiple active certificates with the same
876 subject in the CA index file. This is done only if the keyword
877 'unique_subject' is set to 'no' in the main CA section (default
878 if 'CA_default') of the configuration file. The value is saved
879 with the database itself in a separate index attribute file,
880 named like the index file with '.attr' appended to the name.
883 *) X509 verify fixes. Disable broken certificate workarounds when
884 X509_V_FLAGS_X509_STRICT is set. Check CRL issuer has cRLSign set if
885 keyUsage extension present. Don't accept CRLs with unhandled critical
886 extensions: since verify currently doesn't process CRL extensions this
887 rejects a CRL with *any* critical extensions. Add new verify error codes
891 *) When creating an OCSP nonce use an OCTET STRING inside the extnValue.
892 A clarification of RFC2560 will require the use of OCTET STRINGs and
893 some implementations cannot handle the current raw format. Since OpenSSL
894 copies and compares OCSP nonces as opaque blobs without any attempt at
895 parsing them this should not create any compatibility issues.
898 *) New md flag EVP_MD_CTX_FLAG_REUSE this allows md_data to be reused when
899 calling EVP_MD_CTX_copy_ex() to avoid calling OPENSSL_malloc(). Without
900 this HMAC (and other) operations are several times slower than OpenSSL
904 *) Print out GeneralizedTime and UTCTime in ASN1_STRING_print_ex().
905 [Peter Sylvester <Peter.Sylvester@EdelWeb.fr>]
907 *) Use the correct content when signing type "other".
910 Changes between 0.9.7b and 0.9.7c [30 Sep 2003]
912 *) Fix various bugs revealed by running the NISCC test suite:
914 Stop out of bounds reads in the ASN1 code when presented with
915 invalid tags (CAN-2003-0543 and CAN-2003-0544).
917 Free up ASN1_TYPE correctly if ANY type is invalid (CAN-2003-0545).
919 If verify callback ignores invalid public key errors don't try to check
920 certificate signature with the NULL public key.
924 *) New -ignore_err option in ocsp application to stop the server
925 exiting on the first error in a request.
928 *) In ssl3_accept() (ssl/s3_srvr.c) only accept a client certificate
929 if the server requested one: as stated in TLS 1.0 and SSL 3.0
933 *) In ssl3_get_client_hello() (ssl/s3_srvr.c), tolerate additional
934 extra data after the compression methods not only for TLS 1.0
935 but also for SSL 3.0 (as required by the specification).
936 [Bodo Moeller; problem pointed out by Matthias Loepfe]
938 *) Change X509_certificate_type() to mark the key as exported/exportable
939 when it's 512 *bits* long, not 512 bytes.
942 *) Change AES_cbc_encrypt() so it outputs exact multiple of
943 blocks during encryption.
946 *) Various fixes to base64 BIO and non blocking I/O. On write
947 flushes were not handled properly if the BIO retried. On read
948 data was not being buffered properly and had various logic bugs.
949 This also affects blocking I/O when the data being decoded is a
953 *) Various S/MIME bugfixes and compatibility changes:
954 output correct application/pkcs7 MIME type if
955 PKCS7_NOOLDMIMETYPE is set. Tolerate some broken signatures.
956 Output CR+LF for EOL if PKCS7_CRLFEOL is set (this makes opening
957 of files as .eml work). Correctly handle very long lines in MIME
961 Changes between 0.9.7a and 0.9.7b [10 Apr 2003]
963 *) Countermeasure against the Klima-Pokorny-Rosa extension of
964 Bleichbacher's attack on PKCS #1 v1.5 padding: treat
965 a protocol version number mismatch like a decryption error
966 in ssl3_get_client_key_exchange (ssl/s3_srvr.c).
969 *) Turn on RSA blinding by default in the default implementation
970 to avoid a timing attack. Applications that don't want it can call
971 RSA_blinding_off() or use the new flag RSA_FLAG_NO_BLINDING.
972 They would be ill-advised to do so in most cases.
973 [Ben Laurie, Steve Henson, Geoff Thorpe, Bodo Moeller]
975 *) Change RSA blinding code so that it works when the PRNG is not
976 seeded (in this case, the secret RSA exponent is abused as
977 an unpredictable seed -- if it is not unpredictable, there
978 is no point in blinding anyway). Make RSA blinding thread-safe
979 by remembering the creator's thread ID in rsa->blinding and
980 having all other threads use local one-time blinding factors
981 (this requires more computation than sharing rsa->blinding, but
982 avoids excessive locking; and if an RSA object is not shared
983 between threads, blinding will still be very fast).
986 *) Fixed a typo bug that would cause ENGINE_set_default() to set an
987 ENGINE as defaults for all supported algorithms irrespective of
988 the 'flags' parameter. 'flags' is now honoured, so applications
989 should make sure they are passing it correctly.
992 *) Target "mingw" now allows native Windows code to be generated in
993 the Cygwin environment as well as with the MinGW compiler.
996 Changes between 0.9.7 and 0.9.7a [19 Feb 2003]
998 *) In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked
999 via timing by performing a MAC computation even if incorrrect
1000 block cipher padding has been found. This is a countermeasure
1001 against active attacks where the attacker has to distinguish
1002 between bad padding and a MAC verification error. (CAN-2003-0078)
1004 [Bodo Moeller; problem pointed out by Brice Canvel (EPFL),
1005 Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and
1006 Martin Vuagnoux (EPFL, Ilion)]
1008 *) Make the no-err option work as intended. The intention with no-err
1009 is not to have the whole error stack handling routines removed from
1010 libcrypto, it's only intended to remove all the function name and
1011 reason texts, thereby removing some of the footprint that may not
1012 be interesting if those errors aren't displayed anyway.
1014 NOTE: it's still possible for any application or module to have it's
1015 own set of error texts inserted. The routines are there, just not
1016 used by default when no-err is given.
1019 *) Add support for FreeBSD on IA64.
1020 [dirk.meyer@dinoex.sub.org via Richard Levitte, resolves #454]
1022 *) Adjust DES_cbc_cksum() so it returns the same value as the MIT
1023 Kerberos function mit_des_cbc_cksum(). Before this change,
1024 the value returned by DES_cbc_cksum() was like the one from
1025 mit_des_cbc_cksum(), except the bytes were swapped.
1026 [Kevin Greaney <Kevin.Greaney@hp.com> and Richard Levitte]
1028 *) Allow an application to disable the automatic SSL chain building.
1029 Before this a rather primitive chain build was always performed in
1030 ssl3_output_cert_chain(): an application had no way to send the
1031 correct chain if the automatic operation produced an incorrect result.
1033 Now the chain builder is disabled if either:
1035 1. Extra certificates are added via SSL_CTX_add_extra_chain_cert().
1037 2. The mode flag SSL_MODE_NO_AUTO_CHAIN is set.
1039 The reasoning behind this is that an application would not want the
1040 auto chain building to take place if extra chain certificates are
1041 present and it might also want a means of sending no additional
1042 certificates (for example the chain has two certificates and the
1046 *) Add the possibility to build without the ENGINE framework.
1047 [Steven Reddie <smr@essemer.com.au> via Richard Levitte]
1049 *) Under Win32 gmtime() can return NULL: check return value in
1050 OPENSSL_gmtime(). Add error code for case where gmtime() fails.
1053 *) DSA routines: under certain error conditions uninitialized BN objects
1054 could be freed. Solution: make sure initialization is performed early
1055 enough. (Reported and fix supplied by Ivan D Nestlerode <nestler@MIT.EDU>,
1056 Nils Larsch <nla@trustcenter.de> via PR#459)
1059 *) Another fix for SSLv2 session ID handling: the session ID was incorrectly
1060 checked on reconnect on the client side, therefore session resumption
1061 could still fail with a "ssl session id is different" error. This
1062 behaviour is masked when SSL_OP_ALL is used due to
1063 SSL_OP_MICROSOFT_SESS_ID_BUG being set.
1064 Behaviour observed by Crispin Flowerday <crispin@flowerday.cx> as
1065 followup to PR #377.
1068 *) IA-32 assembler support enhancements: unified ELF targets, support
1069 for SCO/Caldera platforms, fix for Cygwin shared build.
1072 *) Add support for FreeBSD on sparc64. As a consequence, support for
1073 FreeBSD on non-x86 processors is separate from x86 processors on
1074 the config script, much like the NetBSD support.
1075 [Richard Levitte & Kris Kennaway <kris@obsecurity.org>]
1077 Changes between 0.9.6h and 0.9.7 [31 Dec 2002]
1079 [NB: OpenSSL 0.9.6i and later 0.9.6 patch levels were released after
1082 *) Fix session ID handling in SSLv2 client code: the SERVER FINISHED
1083 code (06) was taken as the first octet of the session ID and the last
1084 octet was ignored consequently. As a result SSLv2 client side session
1085 caching could not have worked due to the session ID mismatch between
1087 Behaviour observed by Crispin Flowerday <crispin@flowerday.cx> as
1091 *) Change the declaration of needed Kerberos libraries to use EX_LIBS
1092 instead of the special (and badly supported) LIBKRB5. LIBKRB5 is
1096 *) The hw_ncipher.c engine requires dynamic locks. Unfortunately, it
1097 seems that in spite of existing for more than a year, many application
1098 author have done nothing to provide the necessary callbacks, which
1099 means that this particular engine will not work properly anywhere.
1100 This is a very unfortunate situation which forces us, in the name
1101 of usability, to give the hw_ncipher.c a static lock, which is part
1103 NOTE: This is for the 0.9.7 series ONLY. This hack will never
1104 appear in 0.9.8 or later. We EXPECT application authors to have
1105 dealt properly with this when 0.9.8 is released (unless we actually
1106 make such changes in the libcrypto locking code that changes will
1107 have to be made anyway).
1110 *) In asn1_d2i_read_bio() repeatedly call BIO_read() until all content
1111 octets have been read, EOF or an error occurs. Without this change
1112 some truncated ASN1 structures will not produce an error.
1115 *) Disable Heimdal support, since it hasn't been fully implemented.
1116 Still give the possibility to force the use of Heimdal, but with
1117 warnings and a request that patches get sent to openssl-dev.
1120 *) Add the VC-CE target, introduce the WINCE sysname, and add
1121 INSTALL.WCE and appropriate conditionals to make it build.
1122 [Steven Reddie <smr@essemer.com.au> via Richard Levitte]
1124 *) Change the DLL names for Cygwin to cygcrypto-x.y.z.dll and
1125 cygssl-x.y.z.dll, where x, y and z are the major, minor and
1126 edit numbers of the version.
1127 [Corinna Vinschen <vinschen@redhat.com> and Richard Levitte]
1129 *) Introduce safe string copy and catenation functions
1130 (BUF_strlcpy() and BUF_strlcat()).
1131 [Ben Laurie (CHATS) and Richard Levitte]
1133 *) Avoid using fixed-size buffers for one-line DNs.
1134 [Ben Laurie (CHATS)]
1136 *) Add BUF_MEM_grow_clean() to avoid information leakage when
1137 resizing buffers containing secrets, and use where appropriate.
1138 [Ben Laurie (CHATS)]
1140 *) Avoid using fixed size buffers for configuration file location.
1141 [Ben Laurie (CHATS)]
1143 *) Avoid filename truncation for various CA files.
1144 [Ben Laurie (CHATS)]
1146 *) Use sizeof in preference to magic numbers.
1147 [Ben Laurie (CHATS)]
1149 *) Avoid filename truncation in cert requests.
1150 [Ben Laurie (CHATS)]
1152 *) Add assertions to check for (supposedly impossible) buffer
1154 [Ben Laurie (CHATS)]
1156 *) Don't cache truncated DNS entries in the local cache (this could
1157 potentially lead to a spoofing attack).
1158 [Ben Laurie (CHATS)]
1160 *) Fix various buffers to be large enough for hex/decimal
1161 representations in a platform independent manner.
1162 [Ben Laurie (CHATS)]
1164 *) Add CRYPTO_realloc_clean() to avoid information leakage when
1165 resizing buffers containing secrets, and use where appropriate.
1166 [Ben Laurie (CHATS)]
1168 *) Add BIO_indent() to avoid much slightly worrying code to do
1170 [Ben Laurie (CHATS)]
1172 *) Convert sprintf()/BIO_puts() to BIO_printf().
1173 [Ben Laurie (CHATS)]
1175 *) buffer_gets() could terminate with the buffer only half
1177 [Ben Laurie (CHATS)]
1179 *) Add assertions to prevent user-supplied crypto functions from
1180 overflowing internal buffers by having large block sizes, etc.
1181 [Ben Laurie (CHATS)]
1183 *) New OPENSSL_assert() macro (similar to assert(), but enabled
1185 [Ben Laurie (CHATS)]
1187 *) Eliminate unused copy of key in RC4.
1188 [Ben Laurie (CHATS)]
1190 *) Eliminate unused and incorrectly sized buffers for IV in pem.h.
1191 [Ben Laurie (CHATS)]
1193 *) Fix off-by-one error in EGD path.
1194 [Ben Laurie (CHATS)]
1196 *) If RANDFILE path is too long, ignore instead of truncating.
1197 [Ben Laurie (CHATS)]
1199 *) Eliminate unused and incorrectly sized X.509 structure
1201 [Ben Laurie (CHATS)]
1203 *) Eliminate unused and dangerous function knumber().
1204 [Ben Laurie (CHATS)]
1206 *) Eliminate unused and dangerous structure, KSSL_ERR.
1207 [Ben Laurie (CHATS)]
1209 *) Protect against overlong session ID context length in an encoded
1210 session object. Since these are local, this does not appear to be
1212 [Ben Laurie (CHATS)]
1214 *) Change from security patch (see 0.9.6e below) that did not affect
1215 the 0.9.6 release series:
1217 Remote buffer overflow in SSL3 protocol - an attacker could
1218 supply an oversized master key in Kerberos-enabled versions.
1220 [Ben Laurie (CHATS)]
1222 *) Change the SSL kerb5 codes to match RFC 2712.
1225 *) Make -nameopt work fully for req and add -reqopt switch.
1226 [Michael Bell <michael.bell@rz.hu-berlin.de>, Steve Henson]
1228 *) The "block size" for block ciphers in CFB and OFB mode should be 1.
1229 [Steve Henson, reported by Yngve Nysaeter Pettersen <yngve@opera.com>]
1231 *) Make sure tests can be performed even if the corresponding algorithms
1232 have been removed entirely. This was also the last step to make
1233 OpenSSL compilable with DJGPP under all reasonable conditions.
1234 [Richard Levitte, Doug Kaufman <dkaufman@rahul.net>]
1236 *) Add cipher selection rules COMPLEMENTOFALL and COMPLEMENTOFDEFAULT
1237 to allow version independent disabling of normally unselected ciphers,
1238 which may be activated as a side-effect of selecting a single cipher.
1240 (E.g., cipher list string "RSA" enables ciphersuites that are left
1241 out of "ALL" because they do not provide symmetric encryption.
1242 "RSA:!COMPLEMEMENTOFALL" avoids these unsafe ciphersuites.)
1243 [Lutz Jaenicke, Bodo Moeller]
1245 *) Add appropriate support for separate platform-dependent build
1246 directories. The recommended way to make a platform-dependent
1247 build directory is the following (tested on Linux), maybe with
1250 # Place yourself outside of the OpenSSL source tree. In
1251 # this example, the environment variable OPENSSL_SOURCE
1252 # is assumed to contain the absolute OpenSSL source directory.
1253 mkdir -p objtree/"`uname -s`-`uname -r`-`uname -m`"
1254 cd objtree/"`uname -s`-`uname -r`-`uname -m`"
1255 (cd $OPENSSL_SOURCE; find . -type f) | while read F; do
1256 mkdir -p `dirname $F`
1257 ln -s $OPENSSL_SOURCE/$F $F
1260 To be absolutely sure not to disturb the source tree, a "make clean"
1261 is a good thing. If it isn't successfull, don't worry about it,
1262 it probably means the source directory is very clean.
1265 *) Make sure any ENGINE control commands make local copies of string
1266 pointers passed to them whenever necessary. Otherwise it is possible
1267 the caller may have overwritten (or deallocated) the original string
1268 data when a later ENGINE operation tries to use the stored values.
1269 [Götz Babin-Ebell <babinebell@trustcenter.de>]
1271 *) Improve diagnostics in file reading and command-line digests.
1272 [Ben Laurie aided and abetted by Solar Designer <solar@openwall.com>]
1274 *) Add AES modes CFB and OFB to the object database. Correct an
1275 error in AES-CFB decryption.
1278 *) Remove most calls to EVP_CIPHER_CTX_cleanup() in evp_enc.c, this
1279 allows existing EVP_CIPHER_CTX structures to be reused after
1280 calling EVP_*Final(). This behaviour is used by encryption
1281 BIOs and some applications. This has the side effect that
1282 applications must explicitly clean up cipher contexts with
1283 EVP_CIPHER_CTX_cleanup() or they will leak memory.
1286 *) Check the values of dna and dnb in bn_mul_recursive before calling
1287 bn_mul_comba (a non zero value means the a or b arrays do not contain
1288 n2 elements) and fallback to bn_mul_normal if either is not zero.
1291 *) Fix escaping of non-ASCII characters when using the -subj option
1292 of the "openssl req" command line tool. (Robert Joop <joop@fokus.gmd.de>)
1295 *) Make object definitions compliant to LDAP (RFC2256): SN is the short
1296 form for "surname", serialNumber has no short form.
1297 Use "mail" as the short name for "rfc822Mailbox" according to RFC2798;
1298 therefore remove "mail" short name for "internet 7".
1299 The OID for unique identifiers in X509 certificates is
1300 x500UniqueIdentifier, not uniqueIdentifier.
1301 Some more OID additions. (Michael Bell <michael.bell@rz.hu-berlin.de>)
1304 *) Add an "init" command to the ENGINE config module and auto initialize
1305 ENGINEs. Without any "init" command the ENGINE will be initialized
1306 after all ctrl commands have been executed on it. If init=1 the
1307 ENGINE is initailized at that point (ctrls before that point are run
1308 on the uninitialized ENGINE and after on the initialized one). If
1309 init=0 then the ENGINE will not be iniatialized at all.
1312 *) Fix the 'app_verify_callback' interface so that the user-defined
1313 argument is actually passed to the callback: In the
1314 SSL_CTX_set_cert_verify_callback() prototype, the callback
1315 declaration has been changed from
1318 int (*cb)(X509_STORE_CTX *,void *);
1319 in ssl_verify_cert_chain (ssl/ssl_cert.c), the call
1320 i=s->ctx->app_verify_callback(&ctx)
1321 has been changed into
1322 i=s->ctx->app_verify_callback(&ctx, s->ctx->app_verify_arg).
1324 To update applications using SSL_CTX_set_cert_verify_callback(),
1325 a dummy argument can be added to their callback functions.
1326 [D. K. Smetters <smetters@parc.xerox.com>]
1328 *) Added the '4758cca' ENGINE to support IBM 4758 cards.
1329 [Maurice Gittens <maurice@gittens.nl>, touchups by Geoff Thorpe]
1331 *) Add and OPENSSL_LOAD_CONF define which will cause
1332 OpenSSL_add_all_algorithms() to load the openssl.cnf config file.
1333 This allows older applications to transparently support certain
1334 OpenSSL features: such as crypto acceleration and dynamic ENGINE loading.
1335 Two new functions OPENSSL_add_all_algorithms_noconf() which will never
1336 load the config file and OPENSSL_add_all_algorithms_conf() which will
1337 always load it have also been added.
1340 *) Add the OFB, CFB and CTR (all with 128 bit feedback) to AES.
1341 Adjust NIDs and EVP layer.
1342 [Stephen Sprunk <stephen@sprunk.org> and Richard Levitte]
1344 *) Config modules support in openssl utility.
1346 Most commands now load modules from the config file,
1347 though in a few (such as version) this isn't done
1348 because it couldn't be used for anything.
1350 In the case of ca and req the config file used is
1351 the same as the utility itself: that is the -config
1352 command line option can be used to specify an
1356 *) Move default behaviour from OPENSSL_config(). If appname is NULL
1357 use "openssl_conf" if filename is NULL use default openssl config file.
1360 *) Add an argument to OPENSSL_config() to allow the use of an alternative
1361 config section name. Add a new flag to tolerate a missing config file
1362 and move code to CONF_modules_load_file().
1365 *) Support for crypto accelerator cards from Accelerated Encryption
1366 Processing, www.aep.ie. (Use engine 'aep')
1367 The support was copied from 0.9.6c [engine] and adapted/corrected
1368 to work with the new engine framework.
1369 [AEP Inc. and Richard Levitte]
1371 *) Support for SureWare crypto accelerator cards from Baltimore
1372 Technologies. (Use engine 'sureware')
1373 The support was copied from 0.9.6c [engine] and adapted
1374 to work with the new engine framework.
1377 *) Have the CHIL engine fork-safe (as defined by nCipher) and actually
1378 make the newer ENGINE framework commands for the CHIL engine work.
1379 [Toomas Kiisk <vix@cyber.ee> and Richard Levitte]
1381 *) Make it possible to produce shared libraries on ReliantUNIX.
1382 [Robert Dahlem <Robert.Dahlem@ffm2.siemens.de> via Richard Levitte]
1384 *) Add the configuration target debug-linux-ppro.
1385 Make 'openssl rsa' use the general key loading routines
1386 implemented in apps.c, and make those routines able to
1387 handle the key format FORMAT_NETSCAPE and the variant
1389 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte]
1391 *) Fix a crashbug and a logic bug in hwcrhk_load_pubkey().
1392 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte]
1394 *) Add -keyform to rsautl, and document -engine.
1395 [Richard Levitte, inspired by Toomas Kiisk <vix@cyber.ee>]
1397 *) Change BIO_new_file (crypto/bio/bss_file.c) to use new
1398 BIO_R_NO_SUCH_FILE error code rather than the generic
1399 ERR_R_SYS_LIB error code if fopen() fails with ENOENT.
1402 *) Add new functions
1404 ERR_peek_last_error_line
1405 ERR_peek_last_error_line_data.
1406 These are similar to
1409 ERR_peek_error_line_data,
1410 but report on the latest error recorded rather than the first one
1411 still in the error queue.
1412 [Ben Laurie, Bodo Moeller]
1414 *) default_algorithms option in ENGINE config module. This allows things
1416 default_algorithms = ALL
1417 default_algorithms = RSA, DSA, RAND, CIPHERS, DIGESTS
1420 *) Prelminary ENGINE config module.
1423 *) New experimental application configuration code.
1426 *) Change the AES code to follow the same name structure as all other
1427 symmetric ciphers, and behave the same way. Move everything to
1428 the directory crypto/aes, thereby obsoleting crypto/rijndael.
1429 [Stephen Sprunk <stephen@sprunk.org> and Richard Levitte]
1431 *) SECURITY: remove unsafe setjmp/signal interaction from ui_openssl.c.
1432 [Ben Laurie and Theo de Raadt]
1434 *) Add option to output public keys in req command.
1435 [Massimiliano Pala madwolf@openca.org]
1437 *) Use wNAFs in EC_POINTs_mul() for improved efficiency
1438 (up to about 10% better than before for P-192 and P-224).
1441 *) New functions/macros
1443 SSL_CTX_set_msg_callback(ctx, cb)
1444 SSL_CTX_set_msg_callback_arg(ctx, arg)
1445 SSL_set_msg_callback(ssl, cb)
1446 SSL_set_msg_callback_arg(ssl, arg)
1448 to request calling a callback function
1450 void cb(int write_p, int version, int content_type,
1451 const void *buf, size_t len, SSL *ssl, void *arg)
1453 whenever a protocol message has been completely received
1454 (write_p == 0) or sent (write_p == 1). Here 'version' is the
1455 protocol version according to which the SSL library interprets
1456 the current protocol message (SSL2_VERSION, SSL3_VERSION, or
1457 TLS1_VERSION). 'content_type' is 0 in the case of SSL 2.0, or
1458 the content type as defined in the SSL 3.0/TLS 1.0 protocol
1459 specification (change_cipher_spec(20), alert(21), handshake(22)).
1460 'buf' and 'len' point to the actual message, 'ssl' to the
1461 SSL object, and 'arg' is the application-defined value set by
1462 SSL[_CTX]_set_msg_callback_arg().
1464 'openssl s_client' and 'openssl s_server' have new '-msg' options
1465 to enable a callback that displays all protocol messages.
1468 *) Change the shared library support so shared libraries are built as
1469 soon as the corresponding static library is finished, and thereby get
1470 openssl and the test programs linked against the shared library.
1471 This still only happens when the keyword "shard" has been given to
1472 the configuration scripts.
1474 NOTE: shared library support is still an experimental thing, and
1475 backward binary compatibility is still not guaranteed.
1476 ["Maciej W. Rozycki" <macro@ds2.pg.gda.pl> and Richard Levitte]
1478 *) Add support for Subject Information Access extension.
1479 [Peter Sylvester <Peter.Sylvester@EdelWeb.fr>]
1481 *) Make BUF_MEM_grow() behaviour more consistent: Initialise to zero
1482 additional bytes when new memory had to be allocated, not just
1483 when reusing an existing buffer.
1486 *) New command line and configuration option 'utf8' for the req command.
1487 This allows field values to be specified as UTF8 strings.
1490 *) Add -multi and -mr options to "openssl speed" - giving multiple parallel
1491 runs for the former and machine-readable output for the latter.
1494 *) Add '-noemailDN' option to 'openssl ca'. This prevents inclusion
1495 of the e-mail address in the DN (i.e., it will go into a certificate
1496 extension only). The new configuration file option 'email_in_dn = no'
1497 has the same effect.
1498 [Massimiliano Pala madwolf@openca.org]
1500 *) Change all functions with names starting with des_ to be starting
1501 with DES_ instead. Add wrappers that are compatible with libdes,
1502 but are named _ossl_old_des_*. Finally, add macros that map the
1503 des_* symbols to the corresponding _ossl_old_des_* if libdes
1504 compatibility is desired. If OpenSSL 0.9.6c compatibility is
1505 desired, the des_* symbols will be mapped to DES_*, with one
1508 Since we provide two compatibility mappings, the user needs to
1509 define the macro OPENSSL_DES_LIBDES_COMPATIBILITY if libdes
1510 compatibility is desired. The default (i.e., when that macro
1511 isn't defined) is OpenSSL 0.9.6c compatibility.
1513 There are also macros that enable and disable the support of old
1514 des functions altogether. Those are OPENSSL_ENABLE_OLD_DES_SUPPORT
1515 and OPENSSL_DISABLE_OLD_DES_SUPPORT. If none or both of those
1516 are defined, the default will apply: to support the old des routines.
1518 In either case, one must include openssl/des.h to get the correct
1519 definitions. Do not try to just include openssl/des_old.h, that
1522 NOTE: This is a major break of an old API into a new one. Software
1523 authors are encouraged to switch to the DES_ style functions. Some
1524 time in the future, des_old.h and the libdes compatibility functions
1525 will be disable (i.e. OPENSSL_DISABLE_OLD_DES_SUPPORT will be the
1526 default), and then completely removed.
1529 *) Test for certificates which contain unsupported critical extensions.
1530 If such a certificate is found during a verify operation it is
1531 rejected by default: this behaviour can be overridden by either
1532 handling the new error X509_V_ERR_UNHANDLED_CRITICAL_EXTENSION or
1533 by setting the verify flag X509_V_FLAG_IGNORE_CRITICAL. A new function
1534 X509_supported_extension() has also been added which returns 1 if a
1535 particular extension is supported.
1538 *) Modify the behaviour of EVP cipher functions in similar way to digests
1539 to retain compatibility with existing code.
1542 *) Modify the behaviour of EVP_DigestInit() and EVP_DigestFinal() to retain
1543 compatibility with existing code. In particular the 'ctx' parameter does
1544 not have to be to be initialized before the call to EVP_DigestInit() and
1545 it is tidied up after a call to EVP_DigestFinal(). New function
1546 EVP_DigestFinal_ex() which does not tidy up the ctx. Similarly function
1547 EVP_MD_CTX_copy() changed to not require the destination to be
1548 initialized valid and new function EVP_MD_CTX_copy_ex() added which
1549 requires the destination to be valid.
1551 Modify all the OpenSSL digest calls to use EVP_DigestInit_ex(),
1552 EVP_DigestFinal_ex() and EVP_MD_CTX_copy_ex().
1555 *) Change ssl3_get_message (ssl/s3_both.c) and the functions using it
1556 so that complete 'Handshake' protocol structures are kept in memory
1557 instead of overwriting 'msg_type' and 'length' with 'body' data.
1560 *) Add an implementation of SSL_add_dir_cert_subjects_to_stack for Win32.
1561 [Massimo Santin via Richard Levitte]
1563 *) Major restructuring to the underlying ENGINE code. This includes
1564 reduction of linker bloat, separation of pure "ENGINE" manipulation
1565 (initialisation, etc) from functionality dealing with implementations
1566 of specific crypto iterfaces. This change also introduces integrated
1567 support for symmetric ciphers and digest implementations - so ENGINEs
1568 can now accelerate these by providing EVP_CIPHER and EVP_MD
1569 implementations of their own. This is detailed in crypto/engine/README
1570 as it couldn't be adequately described here. However, there are a few
1571 API changes worth noting - some RSA, DSA, DH, and RAND functions that
1572 were changed in the original introduction of ENGINE code have now
1573 reverted back - the hooking from this code to ENGINE is now a good
1574 deal more passive and at run-time, operations deal directly with
1575 RSA_METHODs, DSA_METHODs (etc) as they did before, rather than
1576 dereferencing through an ENGINE pointer any more. Also, the ENGINE
1577 functions dealing with BN_MOD_EXP[_CRT] handlers have been removed -
1578 they were not being used by the framework as there is no concept of a
1579 BIGNUM_METHOD and they could not be generalised to the new
1580 'ENGINE_TABLE' mechanism that underlies the new code. Similarly,
1581 ENGINE_cpy() has been removed as it cannot be consistently defined in
1585 *) Change ASN1_GENERALIZEDTIME_check() to allow fractional seconds.
1588 *) Change mkdef.pl to sort symbols that get the same entry number,
1589 and make sure the automatically generated functions ERR_load_*
1590 become part of libeay.num as well.
1593 *) New function SSL_renegotiate_pending(). This returns true once
1594 renegotiation has been requested (either SSL_renegotiate() call
1595 or HelloRequest/ClientHello receveived from the peer) and becomes
1596 false once a handshake has been completed.
1597 (For servers, SSL_renegotiate() followed by SSL_do_handshake()
1598 sends a HelloRequest, but does not ensure that a handshake takes
1599 place. SSL_renegotiate_pending() is useful for checking if the
1600 client has followed the request.)
1603 *) New SSL option SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION.
1604 By default, clients may request session resumption even during
1605 renegotiation (if session ID contexts permit); with this option,
1606 session resumption is possible only in the first handshake.
1608 SSL_OP_ALL is now 0x00000FFFL instead of 0x000FFFFFL. This makes
1609 more bits available for options that should not be part of
1610 SSL_OP_ALL (such as SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION).
1613 *) Add some demos for certificate and certificate request creation.
1616 *) Make maximum certificate chain size accepted from the peer application
1617 settable (SSL*_get/set_max_cert_list()), as proposed by
1618 "Douglas E. Engert" <deengert@anl.gov>.
1621 *) Add support for shared libraries for Unixware-7
1622 (Boyd Lynn Gerber <gerberb@zenez.com>).
1625 *) Add a "destroy" handler to ENGINEs that allows structural cleanup to
1626 be done prior to destruction. Use this to unload error strings from
1627 ENGINEs that load their own error strings. NB: This adds two new API
1628 functions to "get" and "set" this destroy handler in an ENGINE.
1631 *) Alter all existing ENGINE implementations (except "openssl" and
1632 "openbsd") to dynamically instantiate their own error strings. This
1633 makes them more flexible to be built both as statically-linked ENGINEs
1634 and self-contained shared-libraries loadable via the "dynamic" ENGINE.
1635 Also, add stub code to each that makes building them as self-contained
1636 shared-libraries easier (see README.ENGINE).
1639 *) Add a "dynamic" ENGINE that provides a mechanism for binding ENGINE
1640 implementations into applications that are completely implemented in
1641 self-contained shared-libraries. The "dynamic" ENGINE exposes control
1642 commands that can be used to configure what shared-library to load and
1643 to control aspects of the way it is handled. Also, made an update to
1644 the README.ENGINE file that brings its information up-to-date and
1645 provides some information and instructions on the "dynamic" ENGINE
1646 (ie. how to use it, how to build "dynamic"-loadable ENGINEs, etc).
1649 *) Make it possible to unload ranges of ERR strings with a new
1650 "ERR_unload_strings" function.
1653 *) Add a copy() function to EVP_MD.
1656 *) Make EVP_MD routines take a context pointer instead of just the
1657 md_data void pointer.
1660 *) Add flags to EVP_MD and EVP_MD_CTX. EVP_MD_FLAG_ONESHOT indicates
1661 that the digest can only process a single chunk of data
1662 (typically because it is provided by a piece of
1663 hardware). EVP_MD_CTX_FLAG_ONESHOT indicates that the application
1664 is only going to provide a single chunk of data, and hence the
1665 framework needn't accumulate the data for oneshot drivers.
1668 *) As with "ERR", make it possible to replace the underlying "ex_data"
1669 functions. This change also alters the storage and management of global
1670 ex_data state - it's now all inside ex_data.c and all "class" code (eg.
1671 RSA, BIO, SSL_CTX, etc) no longer stores its own STACKS and per-class
1672 index counters. The API functions that use this state have been changed
1673 to take a "class_index" rather than pointers to the class's local STACK
1674 and counter, and there is now an API function to dynamically create new
1675 classes. This centralisation allows us to (a) plug a lot of the
1676 thread-safety problems that existed, and (b) makes it possible to clean
1677 up all allocated state using "CRYPTO_cleanup_all_ex_data()". W.r.t. (b)
1678 such data would previously have always leaked in application code and
1679 workarounds were in place to make the memory debugging turn a blind eye
1680 to it. Application code that doesn't use this new function will still
1681 leak as before, but their memory debugging output will announce it now
1682 rather than letting it slide.
1684 Besides the addition of CRYPTO_cleanup_all_ex_data(), another API change
1685 induced by the "ex_data" overhaul is that X509_STORE_CTX_init() now
1686 has a return value to indicate success or failure.
1689 *) Make it possible to replace the underlying "ERR" functions such that the
1690 global state (2 LHASH tables and 2 locks) is only used by the "default"
1691 implementation. This change also adds two functions to "get" and "set"
1692 the implementation prior to it being automatically set the first time
1693 any other ERR function takes place. Ie. an application can call "get",
1694 pass the return value to a module it has just loaded, and that module
1695 can call its own "set" function using that value. This means the
1696 module's "ERR" operations will use (and modify) the error state in the
1697 application and not in its own statically linked copy of OpenSSL code.
1700 *) Give DH, DSA, and RSA types their own "**_up_ref()" function to increment
1701 reference counts. This performs normal REF_PRINT/REF_CHECK macros on
1702 the operation, and provides a more encapsulated way for external code
1703 (crypto/evp/ and ssl/) to do this. Also changed the evp and ssl code
1704 to use these functions rather than manually incrementing the counts.
1706 Also rename "DSO_up()" function to more descriptive "DSO_up_ref()".
1709 *) Add EVP test program.
1712 *) Add symmetric cipher support to ENGINE. Expect the API to change!
1715 *) New CRL functions: X509_CRL_set_version(), X509_CRL_set_issuer_name()
1716 X509_CRL_set_lastUpdate(), X509_CRL_set_nextUpdate(), X509_CRL_sort(),
1717 X509_REVOKED_set_serialNumber(), and X509_REVOKED_set_revocationDate().
1718 These allow a CRL to be built without having to access X509_CRL fields
1719 directly. Modify 'ca' application to use new functions.
1722 *) Move SSL_OP_TLS_ROLLBACK_BUG out of the SSL_OP_ALL list of recommended
1723 bug workarounds. Rollback attack detection is a security feature.
1724 The problem will only arise on OpenSSL servers when TLSv1 is not
1725 available (sslv3_server_method() or SSL_OP_NO_TLSv1).
1726 Software authors not wanting to support TLSv1 will have special reasons
1727 for their choice and can explicitly enable this option.
1728 [Bodo Moeller, Lutz Jaenicke]
1730 *) Rationalise EVP so it can be extended: don't include a union of
1731 cipher/digest structures, add init/cleanup functions for EVP_MD_CTX
1732 (similar to those existing for EVP_CIPHER_CTX).
1737 EVP_MD_CTX_init(&md); /* new function call */
1738 EVP_DigestInit(&md, EVP_sha1());
1739 EVP_DigestUpdate(&md, in, len);
1740 EVP_DigestFinal(&md, out, NULL);
1741 EVP_MD_CTX_cleanup(&md); /* new function call */
1745 *) Make DES key schedule conform to the usual scheme, as well as
1746 correcting its structure. This means that calls to DES functions
1747 now have to pass a pointer to a des_key_schedule instead of a
1748 plain des_key_schedule (which was actually always a pointer
1751 des_key_schedule ks;
1753 des_set_key_checked(..., &ks);
1754 des_ncbc_encrypt(..., &ks, ...);
1756 (Note that a later change renames 'des_...' into 'DES_...'.)
1759 *) Initial reduction of linker bloat: the use of some functions, such as
1760 PEM causes large amounts of unused functions to be linked in due to
1761 poor organisation. For example pem_all.c contains every PEM function
1762 which has a knock on effect of linking in large amounts of (unused)
1763 ASN1 code. Grouping together similar functions and splitting unrelated
1764 functions prevents this.
1767 *) Cleanup of EVP macros.
1770 *) Change historical references to {NID,SN,LN}_des_ede and ede3 to add the
1771 correct _ecb suffix.
1774 *) Add initial OCSP responder support to ocsp application. The
1775 revocation information is handled using the text based index
1776 use by the ca application. The responder can either handle
1777 requests generated internally, supplied in files (for example
1778 via a CGI script) or using an internal minimal server.
1781 *) Add configuration choices to get zlib compression for TLS.
1784 *) Changes to Kerberos SSL for RFC 2712 compliance:
1785 1. Implemented real KerberosWrapper, instead of just using
1786 KRB5 AP_REQ message. [Thanks to Simon Wilkinson <sxw@sxw.org.uk>]
1787 2. Implemented optional authenticator field of KerberosWrapper.
1789 Added openssl-style ASN.1 macros for Kerberos ticket, ap_req,
1790 and authenticator structs; see crypto/krb5/.
1792 Generalized Kerberos calls to support multiple Kerberos libraries.
1793 [Vern Staats <staatsvr@asc.hpc.mil>,
1794 Jeffrey Altman <jaltman@columbia.edu>
1795 via Richard Levitte]
1797 *) Cause 'openssl speed' to use fully hard-coded DSA keys as it
1798 already does with RSA. testdsa.h now has 'priv_key/pub_key'
1799 values for each of the key sizes rather than having just
1800 parameters (and 'speed' generating keys each time).
1803 *) Speed up EVP routines.
1806 type 8 bytes 64 bytes 256 bytes 1024 bytes 8192 bytes
1807 des-cbc 4408.85k 5560.51k 5778.46k 5862.20k 5825.16k
1808 des-cbc 4389.55k 5571.17k 5792.23k 5846.91k 5832.11k
1809 des-cbc 4394.32k 5575.92k 5807.44k 5848.37k 5841.30k
1811 des-cbc 3482.66k 5069.49k 5496.39k 5614.16k 5639.28k
1812 des-cbc 3480.74k 5068.76k 5510.34k 5609.87k 5635.52k
1813 des-cbc 3483.72k 5067.62k 5504.60k 5708.01k 5724.80k
1816 des-cbc 4660.16k 5650.19k 5807.19k 5827.13k 5783.32k
1818 des-cbc 3624.96k 5258.21k 5530.91k 5624.30k 5628.26k
1821 *) Added the OS2-EMX target.
1822 ["Brian Havard" <brianh@kheldar.apana.org.au> and Richard Levitte]
1824 *) Rewrite apps to use NCONF routines instead of the old CONF. New functions
1825 to support NCONF routines in extension code. New function CONF_set_nconf()
1826 to allow functions which take an NCONF to also handle the old LHASH
1827 structure: this means that the old CONF compatible routines can be
1828 retained (in particular wrt extensions) without having to duplicate the
1829 code. New function X509V3_add_ext_nconf_sk to add extensions to a stack.
1832 *) Enhance the general user interface with mechanisms for inner control
1833 and with possibilities to have yes/no kind of prompts.
1836 *) Change all calls to low level digest routines in the library and
1837 applications to use EVP. Add missing calls to HMAC_cleanup() and
1838 don't assume HMAC_CTX can be copied using memcpy().
1839 [Verdon Walker <VWalker@novell.com>, Steve Henson]
1841 *) Add the possibility to control engines through control names but with
1842 arbitrary arguments instead of just a string.
1843 Change the key loaders to take a UI_METHOD instead of a callback
1844 function pointer. NOTE: this breaks binary compatibility with earlier
1845 versions of OpenSSL [engine].
1846 Adapt the nCipher code for these new conditions and add a card insertion
1850 *) Enhance the general user interface with mechanisms to better support
1851 dialog box interfaces, application-defined prompts, the possibility
1852 to use defaults (for example default passwords from somewhere else)
1853 and interrupts/cancellations.
1856 *) Tidy up PKCS#12 attribute handling. Add support for the CSP name
1857 attribute in PKCS#12 files, add new -CSP option to pkcs12 utility.
1860 *) Fix a memory leak in 'sk_dup()' in the case reallocation fails. (Also
1861 tidy up some unnecessarily weird code in 'sk_new()').
1862 [Geoff, reported by Diego Tartara <dtartara@novamens.com>]
1864 *) Change the key loading routines for ENGINEs to use the same kind
1865 callback (pem_password_cb) as all other routines that need this
1869 *) Increase ENTROPY_NEEDED to 32 bytes, as Rijndael can operate with
1870 256 bit (=32 byte) keys. Of course seeding with more entropy bytes
1871 than this minimum value is recommended.
1874 *) New random seeder for OpenVMS, using the system process statistics
1875 that are easily reachable.
1878 *) Windows apparently can't transparently handle global
1879 variables defined in DLLs. Initialisations such as:
1881 const ASN1_ITEM *it = &ASN1_INTEGER_it;
1883 wont compile. This is used by the any applications that need to
1884 declare their own ASN1 modules. This was fixed by adding the option
1885 EXPORT_VAR_AS_FN to all Win32 platforms, although this isn't strictly
1886 needed for static libraries under Win32.
1889 *) New functions X509_PURPOSE_set() and X509_TRUST_set() to handle
1890 setting of purpose and trust fields. New X509_STORE trust and
1891 purpose functions and tidy up setting in other SSL functions.
1894 *) Add copies of X509_STORE_CTX fields and callbacks to X509_STORE
1895 structure. These are inherited by X509_STORE_CTX when it is
1896 initialised. This allows various defaults to be set in the
1897 X509_STORE structure (such as flags for CRL checking and custom
1898 purpose or trust settings) for functions which only use X509_STORE_CTX
1899 internally such as S/MIME.
1901 Modify X509_STORE_CTX_purpose_inherit() so it only sets purposes and
1902 trust settings if they are not set in X509_STORE. This allows X509_STORE
1903 purposes and trust (in S/MIME for example) to override any set by default.
1905 Add command line options for CRL checking to smime, s_client and s_server
1909 *) Initial CRL based revocation checking. If the CRL checking flag(s)
1910 are set then the CRL is looked up in the X509_STORE structure and
1911 its validity and signature checked, then if the certificate is found
1912 in the CRL the verify fails with a revoked error.
1914 Various new CRL related callbacks added to X509_STORE_CTX structure.
1916 Command line options added to 'verify' application to support this.
1918 This needs some additional work, such as being able to handle multiple
1919 CRLs with different times, extension based lookup (rather than just
1920 by subject name) and ultimately more complete V2 CRL extension
1924 *) Add a general user interface API (crypto/ui/). This is designed
1925 to replace things like des_read_password and friends (backward
1926 compatibility functions using this new API are provided).
1927 The purpose is to remove prompting functions from the DES code
1928 section as well as provide for prompting through dialog boxes in
1929 a window system and the like.
1932 *) Add "ex_data" support to ENGINE so implementations can add state at a
1933 per-structure level rather than having to store it globally.
1936 *) Make it possible for ENGINE structures to be copied when retrieved by
1937 ENGINE_by_id() if the ENGINE specifies a new flag: ENGINE_FLAGS_BY_ID_COPY.
1938 This causes the "original" ENGINE structure to act like a template,
1939 analogous to the RSA vs. RSA_METHOD type of separation. Because of this
1940 operational state can be localised to each ENGINE structure, despite the
1941 fact they all share the same "methods". New ENGINE structures returned in
1942 this case have no functional references and the return value is the single
1943 structural reference. This matches the single structural reference returned
1944 by ENGINE_by_id() normally, when it is incremented on the pre-existing
1948 *) Fix ASN1 decoder when decoding type ANY and V_ASN1_OTHER: since this
1949 needs to match any other type at all we need to manually clear the
1953 *) Changes to the "openssl engine" utility to include;
1954 - verbosity levels ('-v', '-vv', and '-vvv') that provide information
1955 about an ENGINE's available control commands.
1956 - executing control commands from command line arguments using the
1957 '-pre' and '-post' switches. '-post' is only used if '-t' is
1958 specified and the ENGINE is successfully initialised. The syntax for
1959 the individual commands are colon-separated, for example;
1960 openssl engine chil -pre FORK_CHECK:0 -pre SO_PATH:/lib/test.so
1963 *) New dynamic control command support for ENGINEs. ENGINEs can now
1964 declare their own commands (numbers), names (strings), descriptions,
1965 and input types for run-time discovery by calling applications. A
1966 subset of these commands are implicitly classed as "executable"
1967 depending on their input type, and only these can be invoked through
1968 the new string-based API function ENGINE_ctrl_cmd_string(). (Eg. this
1969 can be based on user input, config files, etc). The distinction is
1970 that "executable" commands cannot return anything other than a boolean
1971 result and can only support numeric or string input, whereas some
1972 discoverable commands may only be for direct use through
1973 ENGINE_ctrl(), eg. supporting the exchange of binary data, function
1974 pointers, or other custom uses. The "executable" commands are to
1975 support parameterisations of ENGINE behaviour that can be
1976 unambiguously defined by ENGINEs and used consistently across any
1977 OpenSSL-based application. Commands have been added to all the
1978 existing hardware-supporting ENGINEs, noticeably "SO_PATH" to allow
1979 control over shared-library paths without source code alterations.
1982 *) Changed all ENGINE implementations to dynamically allocate their
1983 ENGINEs rather than declaring them statically. Apart from this being
1984 necessary with the removal of the ENGINE_FLAGS_MALLOCED distinction,
1985 this also allows the implementations to compile without using the
1986 internal engine_int.h header.
1989 *) Minor adjustment to "rand" code. RAND_get_rand_method() now returns a
1990 'const' value. Any code that should be able to modify a RAND_METHOD
1991 should already have non-const pointers to it (ie. they should only
1992 modify their own ones).
1995 *) Made a variety of little tweaks to the ENGINE code.
1996 - "atalla" and "ubsec" string definitions were moved from header files
1997 to C code. "nuron" string definitions were placed in variables
1998 rather than hard-coded - allowing parameterisation of these values
1999 later on via ctrl() commands.
2000 - Removed unused "#if 0"'d code.
2001 - Fixed engine list iteration code so it uses ENGINE_free() to release
2002 structural references.
2003 - Constified the RAND_METHOD element of ENGINE structures.
2004 - Constified various get/set functions as appropriate and added
2005 missing functions (including a catch-all ENGINE_cpy that duplicates
2006 all ENGINE values onto a new ENGINE except reference counts/state).
2007 - Removed NULL parameter checks in get/set functions. Setting a method
2008 or function to NULL is a way of cancelling out a previously set
2009 value. Passing a NULL ENGINE parameter is just plain stupid anyway
2010 and doesn't justify the extra error symbols and code.
2011 - Deprecate the ENGINE_FLAGS_MALLOCED define and move the area for
2012 flags from engine_int.h to engine.h.
2013 - Changed prototypes for ENGINE handler functions (init(), finish(),
2014 ctrl(), key-load functions, etc) to take an (ENGINE*) parameter.
2017 *) Implement binary inversion algorithm for BN_mod_inverse in addition
2018 to the algorithm using long division. The binary algorithm can be
2019 used only if the modulus is odd. On 32-bit systems, it is faster
2020 only for relatively small moduli (roughly 20-30% for 128-bit moduli,
2021 roughly 5-15% for 256-bit moduli), so we use it only for moduli
2022 up to 450 bits. In 64-bit environments, the binary algorithm
2023 appears to be advantageous for much longer moduli; here we use it
2024 for moduli up to 2048 bits.
2027 *) Rewrite CHOICE field setting in ASN1_item_ex_d2i(). The old code
2028 could not support the combine flag in choice fields.
2031 *) Add a 'copy_extensions' option to the 'ca' utility. This copies
2032 extensions from a certificate request to the certificate.
2035 *) Allow multiple 'certopt' and 'nameopt' options to be separated
2036 by commas. Add 'namopt' and 'certopt' options to the 'ca' config
2037 file: this allows the display of the certificate about to be
2038 signed to be customised, to allow certain fields to be included
2039 or excluded and extension details. The old system didn't display
2040 multicharacter strings properly, omitted fields not in the policy
2041 and couldn't display additional details such as extensions.
2044 *) Function EC_POINTs_mul for multiple scalar multiplication
2045 of an arbitrary number of elliptic curve points
2046 \sum scalars[i]*points[i],
2047 optionally including the generator defined for the EC_GROUP:
2048 scalar*generator + \sum scalars[i]*points[i].
2050 EC_POINT_mul is a simple wrapper function for the typical case
2051 that the point list has just one item (besides the optional
2055 *) First EC_METHODs for curves over GF(p):
2057 EC_GFp_simple_method() uses the basic BN_mod_mul and BN_mod_sqr
2058 operations and provides various method functions that can also
2059 operate with faster implementations of modular arithmetic.
2061 EC_GFp_mont_method() reuses most functions that are part of
2062 EC_GFp_simple_method, but uses Montgomery arithmetic.
2064 [Bodo Moeller; point addition and point doubling
2065 implementation directly derived from source code provided by
2066 Lenka Fibikova <fibikova@exp-math.uni-essen.de>]
2068 *) Framework for elliptic curves (crypto/ec/ec.h, crypto/ec/ec_lcl.h,
2069 crypto/ec/ec_lib.c):
2071 Curves are EC_GROUP objects (with an optional group generator)
2072 based on EC_METHODs that are built into the library.
2074 Points are EC_POINT objects based on EC_GROUP objects.
2076 Most of the framework would be able to handle curves over arbitrary
2077 finite fields, but as there are no obvious types for fields other
2078 than GF(p), some functions are limited to that for now.
2081 *) Add the -HTTP option to s_server. It is similar to -WWW, but requires
2082 that the file contains a complete HTTP response.
2085 *) Add the ec directory to mkdef.pl and mkfiles.pl. In mkdef.pl
2086 change the def and num file printf format specifier from "%-40sXXX"
2087 to "%-39s XXX". The latter will always guarantee a space after the
2088 field while the former will cause them to run together if the field
2089 is 40 of more characters long.
2092 *) Constify the cipher and digest 'method' functions and structures
2093 and modify related functions to take constant EVP_MD and EVP_CIPHER
2097 *) Hide BN_CTX structure details in bn_lcl.h instead of publishing them
2098 in <openssl/bn.h>. Also further increase BN_CTX_NUM to 32.
2101 *) Modify EVP_Digest*() routines so they now return values. Although the
2102 internal software routines can never fail additional hardware versions
2106 *) Clean up crypto/err/err.h and change some error codes to avoid conflicts:
2108 Previously ERR_R_FATAL was too small and coincided with ERR_LIB_PKCS7
2109 (= ERR_R_PKCS7_LIB); it is now 64 instead of 32.
2112 ERR_R_NESTED_ASN1_ERROR
2114 ERR_R_MISSING_ASN1_EOS
2115 were 4 .. 9, conflicting with
2116 ERR_LIB_RSA (= ERR_R_RSA_LIB)
2118 ERR_LIB_PEM (= ERR_R_PEM_LIB).
2119 They are now 58 .. 63 (i.e., just below ERR_R_FATAL).
2121 Add new error code 'ERR_R_INTERNAL_ERROR'.
2124 *) Don't overuse locks in crypto/err/err.c: For data retrieval, CRYPTO_r_lock
2128 *) New option '-subj arg' for 'openssl req' and 'openssl ca'. This
2129 sets the subject name for a new request or supersedes the
2130 subject name in a given request. Formats that can be parsed are
2131 'CN=Some Name, OU=myOU, C=IT'
2133 'CN=Some Name/OU=myOU/C=IT'.
2135 Add options '-batch' and '-verbose' to 'openssl req'.
2136 [Massimiliano Pala <madwolf@hackmasters.net>]
2138 *) Introduce the possibility to access global variables through
2139 functions on platform were that's the best way to handle exporting
2140 global variables in shared libraries. To enable this functionality,
2141 one must configure with "EXPORT_VAR_AS_FN" or defined the C macro
2142 "OPENSSL_EXPORT_VAR_AS_FUNCTION" in crypto/opensslconf.h (the latter
2143 is normally done by Configure or something similar).
2145 To implement a global variable, use the macro OPENSSL_IMPLEMENT_GLOBAL
2146 in the source file (foo.c) like this:
2148 OPENSSL_IMPLEMENT_GLOBAL(int,foo)=1;
2149 OPENSSL_IMPLEMENT_GLOBAL(double,bar);
2151 To declare a global variable, use the macros OPENSSL_DECLARE_GLOBAL
2152 and OPENSSL_GLOBAL_REF in the header file (foo.h) like this:
2154 OPENSSL_DECLARE_GLOBAL(int,foo);
2155 #define foo OPENSSL_GLOBAL_REF(foo)
2156 OPENSSL_DECLARE_GLOBAL(double,bar);
2157 #define bar OPENSSL_GLOBAL_REF(bar)
2159 The #defines are very important, and therefore so is including the
2160 header file everywhere where the defined globals are used.
2162 The macro OPENSSL_EXPORT_VAR_AS_FUNCTION also affects the definition
2163 of ASN.1 items, but that structure is a bit different.
2165 The largest change is in util/mkdef.pl which has been enhanced with
2166 better and easier to understand logic to choose which symbols should
2167 go into the Windows .def files as well as a number of fixes and code
2168 cleanup (among others, algorithm keywords are now sorted
2169 lexicographically to avoid constant rewrites).
2172 *) In BN_div() keep a copy of the sign of 'num' before writing the
2173 result to 'rm' because if rm==num the value will be overwritten
2174 and produce the wrong result if 'num' is negative: this caused
2175 problems with BN_mod() and BN_nnmod().
2178 *) Function OCSP_request_verify(). This checks the signature on an
2179 OCSP request and verifies the signer certificate. The signer
2180 certificate is just checked for a generic purpose and OCSP request
2184 *) Add OCSP_check_validity() function to check the validity of OCSP
2185 responses. OCSP responses are prepared in real time and may only
2186 be a few seconds old. Simply checking that the current time lies
2187 between thisUpdate and nextUpdate max reject otherwise valid responses
2188 caused by either OCSP responder or client clock inaccuracy. Instead
2189 we allow thisUpdate and nextUpdate to fall within a certain period of
2190 the current time. The age of the response can also optionally be
2191 checked. Two new options -validity_period and -status_age added to
2195 *) If signature or public key algorithm is unrecognized print out its
2196 OID rather that just UNKNOWN.
2199 *) Change OCSP_cert_to_id() to tolerate a NULL subject certificate and
2200 OCSP_cert_id_new() a NULL serialNumber. This allows a partial certificate
2201 ID to be generated from the issuer certificate alone which can then be
2202 passed to OCSP_id_issuer_cmp().
2205 *) New compilation option ASN1_ITEM_FUNCTIONS. This causes the new
2206 ASN1 modules to export functions returning ASN1_ITEM pointers
2207 instead of the ASN1_ITEM structures themselves. This adds several
2208 new macros which allow the underlying ASN1 function/structure to
2209 be accessed transparently. As a result code should not use ASN1_ITEM
2210 references directly (such as &X509_it) but instead use the relevant
2211 macros (such as ASN1_ITEM_rptr(X509)). This option is to allow
2212 use of the new ASN1 code on platforms where exporting structures
2213 is problematical (for example in shared libraries) but exporting
2214 functions returning pointers to structures is not.
2217 *) Add support for overriding the generation of SSL/TLS session IDs.
2218 These callbacks can be registered either in an SSL_CTX or per SSL.
2219 The purpose of this is to allow applications to control, if they wish,
2220 the arbitrary values chosen for use as session IDs, particularly as it
2221 can be useful for session caching in multiple-server environments. A
2222 command-line switch for testing this (and any client code that wishes
2223 to use such a feature) has been added to "s_server".
2224 [Geoff Thorpe, Lutz Jaenicke]
2226 *) Modify mkdef.pl to recognise and parse preprocessor conditionals
2227 of the form '#if defined(...) || defined(...) || ...' and
2228 '#if !defined(...) && !defined(...) && ...'. This also avoids
2229 the growing number of special cases it was previously handling.
2232 *) Make all configuration macros available for application by making
2233 sure they are available in opensslconf.h, by giving them names starting
2234 with "OPENSSL_" to avoid conflicts with other packages and by making
2235 sure e_os2.h will cover all platform-specific cases together with
2237 Additionally, it is now possible to define configuration/platform-
2238 specific names (called "system identities"). In the C code, these
2239 are prefixed with "OPENSSL_SYSNAME_". e_os2.h will create another
2240 macro with the name beginning with "OPENSSL_SYS_", which is determined
2241 from "OPENSSL_SYSNAME_*" or compiler-specific macros depending on
2245 *) New option -set_serial to 'req' and 'x509' this allows the serial
2246 number to use to be specified on the command line. Previously self
2247 signed certificates were hard coded with serial number 0 and the
2248 CA options of 'x509' had to use a serial number in a file which was
2252 *) New options to 'ca' utility to support V2 CRL entry extensions.
2253 Currently CRL reason, invalidity date and hold instruction are
2254 supported. Add new CRL extensions to V3 code and some new objects.
2257 *) New function EVP_CIPHER_CTX_set_padding() this is used to
2258 disable standard block padding (aka PKCS#5 padding) in the EVP
2259 API, which was previously mandatory. This means that the data is
2260 not padded in any way and so the total length much be a multiple
2261 of the block size, otherwise an error occurs.
2264 *) Initial (incomplete) OCSP SSL support.
2267 *) New function OCSP_parse_url(). This splits up a URL into its host,
2268 port and path components: primarily to parse OCSP URLs. New -url
2269 option to ocsp utility.
2272 *) New nonce behavior. The return value of OCSP_check_nonce() now
2273 reflects the various checks performed. Applications can decide
2274 whether to tolerate certain situations such as an absent nonce
2275 in a response when one was present in a request: the ocsp application
2276 just prints out a warning. New function OCSP_add1_basic_nonce()
2277 this is to allow responders to include a nonce in a response even if
2278 the request is nonce-less.
2281 *) Disable stdin buffering in load_cert (apps/apps.c) so that no certs are
2282 skipped when using openssl x509 multiple times on a single input file,
2283 e.g. "(openssl x509 -out cert1; openssl x509 -out cert2) <certs".
2286 *) Make ASN1_UTCTIME_set_string() and ASN1_GENERALIZEDTIME_set_string()
2287 set string type: to handle setting ASN1_TIME structures. Fix ca
2288 utility to correctly initialize revocation date of CRLs.
2291 *) New option SSL_OP_CIPHER_SERVER_PREFERENCE allows the server to override
2292 the clients preferred ciphersuites and rather use its own preferences.
2293 Should help to work around M$ SGC (Server Gated Cryptography) bug in
2294 Internet Explorer by ensuring unchanged hash method during stepup.
2295 (Also replaces the broken/deactivated SSL_OP_NON_EXPORT_FIRST option.)
2298 *) Make mkdef.pl recognise all DECLARE_ASN1 macros, change rijndael
2299 to aes and add a new 'exist' option to print out symbols that don't
2303 *) Additional options to ocsp utility to allow flags to be set and
2304 additional certificates supplied.
2307 *) Add the option -VAfile to 'openssl ocsp', so the user can give the
2308 OCSP client a number of certificate to only verify the response
2312 *) Update Rijndael code to version 3.0 and change EVP AES ciphers to
2313 handle the new API. Currently only ECB, CBC modes supported. Add new
2316 Add TLS AES ciphersuites as described in RFC3268, "Advanced
2317 Encryption Standard (AES) Ciphersuites for Transport Layer
2318 Security (TLS)". (In beta versions of OpenSSL 0.9.7, these were
2319 not enabled by default and were not part of the "ALL" ciphersuite
2320 alias because they were not yet official; they could be
2321 explicitly requested by specifying the "AESdraft" ciphersuite
2322 group alias. In the final release of OpenSSL 0.9.7, the group
2323 alias is called "AES" and is part of "ALL".)
2324 [Ben Laurie, Steve Henson, Bodo Moeller]
2326 *) New function OCSP_copy_nonce() to copy nonce value (if present) from
2327 request to response.
2330 *) Functions for OCSP responders. OCSP_request_onereq_count(),
2331 OCSP_request_onereq_get0(), OCSP_onereq_get0_id() and OCSP_id_get0_info()
2332 extract information from a certificate request. OCSP_response_create()
2333 creates a response and optionally adds a basic response structure.
2334 OCSP_basic_add1_status() adds a complete single response to a basic
2335 response and returns the OCSP_SINGLERESP structure just added (to allow
2336 extensions to be included for example). OCSP_basic_add1_cert() adds a
2337 certificate to a basic response and OCSP_basic_sign() signs a basic
2338 response with various flags. New helper functions ASN1_TIME_check()
2339 (checks validity of ASN1_TIME structure) and ASN1_TIME_to_generalizedtime()
2340 (converts ASN1_TIME to GeneralizedTime).
2343 *) Various new functions. EVP_Digest() combines EVP_Digest{Init,Update,Final}()
2344 in a single operation. X509_get0_pubkey_bitstr() extracts the public_key
2345 structure from a certificate. X509_pubkey_digest() digests the public_key
2346 contents: this is used in various key identifiers.
2349 *) Make sk_sort() tolerate a NULL argument.
2350 [Steve Henson reported by Massimiliano Pala <madwolf@comune.modena.it>]
2352 *) New OCSP verify flag OCSP_TRUSTOTHER. When set the "other" certificates
2353 passed by the function are trusted implicitly. If any of them signed the
2354 response then it is assumed to be valid and is not verified.
2357 *) In PKCS7_set_type() initialise content_type in PKCS7_ENC_CONTENT
2358 to data. This was previously part of the PKCS7 ASN1 code. This
2359 was causing problems with OpenSSL created PKCS#12 and PKCS#7 structures.
2360 [Steve Henson, reported by Kenneth R. Robinette
2361 <support@securenetterm.com>]
2363 *) Add CRYPTO_push_info() and CRYPTO_pop_info() calls to new ASN1
2364 routines: without these tracing memory leaks is very painful.
2365 Fix leaks in PKCS12 and PKCS7 routines.
2368 *) Make X509_time_adj() cope with the new behaviour of ASN1_TIME_new().
2369 Previously it initialised the 'type' argument to V_ASN1_UTCTIME which
2370 effectively meant GeneralizedTime would never be used. Now it
2371 is initialised to -1 but X509_time_adj() now has to check the value
2372 and use ASN1_TIME_set() if the value is not V_ASN1_UTCTIME or
2373 V_ASN1_GENERALIZEDTIME, without this it always uses GeneralizedTime.
2374 [Steve Henson, reported by Kenneth R. Robinette
2375 <support@securenetterm.com>]
2377 *) Fixes to BN_to_ASN1_INTEGER when bn is zero. This would previously
2378 result in a zero length in the ASN1_INTEGER structure which was
2379 not consistent with the structure when d2i_ASN1_INTEGER() was used
2380 and would cause ASN1_INTEGER_cmp() to fail. Enhance s2i_ASN1_INTEGER()
2381 to cope with hex and negative integers. Fix bug in i2a_ASN1_INTEGER()
2382 where it did not print out a minus for negative ASN1_INTEGER.
2385 *) Add summary printout to ocsp utility. The various functions which
2386 convert status values to strings have been renamed to:
2387 OCSP_response_status_str(), OCSP_cert_status_str() and
2388 OCSP_crl_reason_str() and are no longer static. New options
2389 to verify nonce values and to disable verification. OCSP response
2390 printout format cleaned up.
2393 *) Add additional OCSP certificate checks. These are those specified
2394 in RFC2560. This consists of two separate checks: the CA of the
2395 certificate being checked must either be the OCSP signer certificate
2396 or the issuer of the OCSP signer certificate. In the latter case the
2397 OCSP signer certificate must contain the OCSP signing extended key
2398 usage. This check is performed by attempting to match the OCSP
2399 signer or the OCSP signer CA to the issuerNameHash and issuerKeyHash
2400 in the OCSP_CERTID structures of the response.
2403 *) Initial OCSP certificate verification added to OCSP_basic_verify()
2404 and related routines. This uses the standard OpenSSL certificate
2405 verify routines to perform initial checks (just CA validity) and
2406 to obtain the certificate chain. Then additional checks will be
2407 performed on the chain. Currently the root CA is checked to see
2408 if it is explicitly trusted for OCSP signing. This is used to set
2409 a root CA as a global signing root: that is any certificate that
2410 chains to that CA is an acceptable OCSP signing certificate.
2413 *) New '-extfile ...' option to 'openssl ca' for reading X.509v3
2414 extensions from a separate configuration file.
2415 As when reading extensions from the main configuration file,
2416 the '-extensions ...' option may be used for specifying the
2418 [Massimiliano Pala <madwolf@comune.modena.it>]
2420 *) New OCSP utility. Allows OCSP requests to be generated or
2421 read. The request can be sent to a responder and the output
2422 parsed, outputed or printed in text form. Not complete yet:
2423 still needs to check the OCSP response validity.
2426 *) New subcommands for 'openssl ca':
2427 'openssl ca -status <serial>' prints the status of the cert with
2428 the given serial number (according to the index file).
2429 'openssl ca -updatedb' updates the expiry status of certificates
2431 [Massimiliano Pala <madwolf@comune.modena.it>]
2433 *) New '-newreq-nodes' command option to CA.pl. This is like
2434 '-newreq', but calls 'openssl req' with the '-nodes' option
2435 so that the resulting key is not encrypted.
2436 [Damien Miller <djm@mindrot.org>]
2438 *) New configuration for the GNU Hurd.
2439 [Jonathan Bartlett <johnnyb@wolfram.com> via Richard Levitte]
2441 *) Initial code to implement OCSP basic response verify. This
2442 is currently incomplete. Currently just finds the signer's
2443 certificate and verifies the signature on the response.
2446 *) New SSLeay_version code SSLEAY_DIR to determine the compiled-in
2447 value of OPENSSLDIR. This is available via the new '-d' option
2448 to 'openssl version', and is also included in 'openssl version -a'.
2451 *) Allowing defining memory allocation callbacks that will be given
2452 file name and line number information in additional arguments
2453 (a const char* and an int). The basic functionality remains, as
2454 well as the original possibility to just replace malloc(),
2455 realloc() and free() by functions that do not know about these
2456 additional arguments. To register and find out the current
2457 settings for extended allocation functions, the following
2458 functions are provided:
2460 CRYPTO_set_mem_ex_functions
2461 CRYPTO_set_locked_mem_ex_functions
2462 CRYPTO_get_mem_ex_functions
2463 CRYPTO_get_locked_mem_ex_functions
2465 These work the same way as CRYPTO_set_mem_functions and friends.
2466 CRYPTO_get_[locked_]mem_functions now writes 0 where such an
2467 extended allocation function is enabled.
2468 Similarly, CRYPTO_get_[locked_]mem_ex_functions writes 0 where
2469 a conventional allocation function is enabled.
2470 [Richard Levitte, Bodo Moeller]
2472 *) Finish off removing the remaining LHASH function pointer casts.
2473 There should no longer be any prototype-casting required when using
2474 the LHASH abstraction, and any casts that remain are "bugs". See
2475 the callback types and macros at the head of lhash.h for details
2476 (and "OBJ_cleanup" in crypto/objects/obj_dat.c as an example).
2479 *) Add automatic query of EGD sockets in RAND_poll() for the unix variant.
2480 If /dev/[u]random devices are not available or do not return enough
2481 entropy, EGD style sockets (served by EGD or PRNGD) will automatically
2483 The locations /var/run/egd-pool, /dev/egd-pool, /etc/egd-pool, and
2484 /etc/entropy will be queried once each in this sequence, quering stops
2485 when enough entropy was collected without querying more sockets.
2488 *) Change the Unix RAND_poll() variant to be able to poll several
2489 random devices, as specified by DEVRANDOM, until a sufficient amount
2490 of data has been collected. We spend at most 10 ms on each file
2491 (select timeout) and read in non-blocking mode. DEVRANDOM now
2492 defaults to the list "/dev/urandom", "/dev/random", "/dev/srandom"
2493 (previously it was just the string "/dev/urandom"), so on typical
2494 platforms the 10 ms delay will never occur.
2495 Also separate out the Unix variant to its own file, rand_unix.c.
2496 For VMS, there's a currently-empty rand_vms.c.
2499 *) Move OCSP client related routines to ocsp_cl.c. These
2500 provide utility functions which an application needing
2501 to issue a request to an OCSP responder and analyse the
2502 response will typically need: as opposed to those which an
2503 OCSP responder itself would need which will be added later.
2505 OCSP_request_sign() signs an OCSP request with an API similar
2506 to PKCS7_sign(). OCSP_response_status() returns status of OCSP
2507 response. OCSP_response_get1_basic() extracts basic response
2508 from response. OCSP_resp_find_status(): finds and extracts status
2509 information from an OCSP_CERTID structure (which will be created
2510 when the request structure is built). These are built from lower
2511 level functions which work on OCSP_SINGLERESP structures but
2512 wont normally be used unless the application wishes to examine
2513 extensions in the OCSP response for example.
2515 Replace nonce routines with a pair of functions.
2516 OCSP_request_add1_nonce() adds a nonce value and optionally
2517 generates a random value. OCSP_check_nonce() checks the
2518 validity of the nonce in an OCSP response.
2521 *) Change function OCSP_request_add() to OCSP_request_add0_id().
2522 This doesn't copy the supplied OCSP_CERTID and avoids the
2523 need to free up the newly created id. Change return type
2524 to OCSP_ONEREQ to return the internal OCSP_ONEREQ structure.
2525 This can then be used to add extensions to the request.
2526 Deleted OCSP_request_new(), since most of its functionality
2527 is now in OCSP_REQUEST_new() (and the case insensitive name
2528 clash) apart from the ability to set the request name which
2529 will be added elsewhere.
2532 *) Update OCSP API. Remove obsolete extensions argument from
2533 various functions. Extensions are now handled using the new
2534 OCSP extension code. New simple OCSP HTTP function which
2535 can be used to send requests and parse the response.
2538 *) Fix the PKCS#7 (S/MIME) code to work with new ASN1. Two new
2539 ASN1_ITEM structures help with sign and verify. PKCS7_ATTR_SIGN
2540 uses the special reorder version of SET OF to sort the attributes
2541 and reorder them to match the encoded order. This resolves a long
2542 standing problem: a verify on a PKCS7 structure just after signing
2543 it used to fail because the attribute order did not match the
2544 encoded order. PKCS7_ATTR_VERIFY does not reorder the attributes:
2545 it uses the received order. This is necessary to tolerate some broken
2546 software that does not order SET OF. This is handled by encoding
2547 as a SEQUENCE OF but using implicit tagging (with UNIVERSAL class)
2548 to produce the required SET OF.
2551 *) Have mk1mf.pl generate the macros OPENSSL_BUILD_SHLIBCRYPTO and
2552 OPENSSL_BUILD_SHLIBSSL and use them appropriately in the header
2553 files to get correct declarations of the ASN.1 item variables.
2556 *) Rewrite of PKCS#12 code to use new ASN1 functionality. Replace many
2557 PKCS#12 macros with real functions. Fix two unrelated ASN1 bugs:
2558 asn1_check_tlen() would sometimes attempt to use 'ctx' when it was
2559 NULL and ASN1_TYPE was not dereferenced properly in asn1_ex_c2i().
2560 New ASN1 macro: DECLARE_ASN1_ITEM() which just declares the relevant
2561 ASN1_ITEM and no wrapper functions.
2564 *) New functions or ASN1_item_d2i_fp() and ASN1_item_d2i_bio(). These
2565 replace the old function pointer based I/O routines. Change most of
2566 the *_d2i_bio() and *_d2i_fp() functions to use these.
2569 *) Enhance mkdef.pl to be more accepting about spacing in C preprocessor
2570 lines, recognice more "algorithms" that can be deselected, and make
2571 it complain about algorithm deselection that isn't recognised.
2574 *) New ASN1 functions to handle dup, sign, verify, digest, pack and
2575 unpack operations in terms of ASN1_ITEM. Modify existing wrappers
2576 to use new functions. Add NO_ASN1_OLD which can be set to remove
2577 some old style ASN1 functions: this can be used to determine if old
2578 code will still work when these eventually go away.
2581 *) New extension functions for OCSP structures, these follow the
2582 same conventions as certificates and CRLs.
2585 *) New function X509V3_add1_i2d(). This automatically encodes and
2586 adds an extension. Its behaviour can be customised with various
2587 flags to append, replace or delete. Various wrappers added for
2588 certifcates and CRLs.
2591 *) Fix to avoid calling the underlying ASN1 print routine when
2592 an extension cannot be parsed. Correct a typo in the
2593 OCSP_SERVICELOC extension. Tidy up print OCSP format.
2596 *) Make mkdef.pl parse some of the ASN1 macros and add apropriate
2597 entries for variables.
2600 *) Add functionality to apps/openssl.c for detecting locking
2601 problems: As the program is single-threaded, all we have
2602 to do is register a locking callback using an array for
2603 storing which locks are currently held by the program.
2606 *) Use a lock around the call to CRYPTO_get_ex_new_index() in
2607 SSL_get_ex_data_X509_STORE_idx(), which is used in
2608 ssl_verify_cert_chain() and thus can be called at any time
2609 during TLS/SSL handshakes so that thread-safety is essential.
2610 Unfortunately, the ex_data design is not at all suited
2611 for multi-threaded use, so it probably should be abolished.
2614 *) Added Broadcom "ubsec" ENGINE to OpenSSL.
2615 [Broadcom, tweaked and integrated by Geoff Thorpe]
2617 *) Move common extension printing code to new function
2618 X509V3_print_extensions(). Reorganise OCSP print routines and
2619 implement some needed OCSP ASN1 functions. Add OCSP extensions.
2622 *) New function X509_signature_print() to remove duplication in some
2626 *) Add a special meaning when SET OF and SEQUENCE OF flags are both
2627 set (this was treated exactly the same as SET OF previously). This
2628 is used to reorder the STACK representing the structure to match the
2629 encoding. This will be used to get round a problem where a PKCS7
2630 structure which was signed could not be verified because the STACK
2631 order did not reflect the encoded order.
2634 *) Reimplement the OCSP ASN1 module using the new code.
2637 *) Update the X509V3 code to permit the use of an ASN1_ITEM structure
2638 for its ASN1 operations. The old style function pointers still exist
2639 for now but they will eventually go away.
2642 *) Merge in replacement ASN1 code from the ASN1 branch. This almost
2643 completely replaces the old ASN1 functionality with a table driven
2644 encoder and decoder which interprets an ASN1_ITEM structure describing
2645 the ASN1 module. Compatibility with the existing ASN1 API (i2d,d2i) is
2646 largely maintained. Almost all of the old asn1_mac.h macro based ASN1
2647 has also been converted to the new form.
2650 *) Change BN_mod_exp_recp so that negative moduli are tolerated
2651 (the sign is ignored). Similarly, ignore the sign in BN_MONT_CTX_set
2652 so that BN_mod_exp_mont and BN_mod_exp_mont_word work
2653 for negative moduli.
2656 *) Fix BN_uadd and BN_usub: Always return non-negative results instead
2657 of not touching the result's sign bit.
2660 *) BN_div bugfix: If the result is 0, the sign (res->neg) must not be
2664 *) Changed the LHASH code to use prototypes for callbacks, and created
2665 macros to declare and implement thin (optionally static) functions
2666 that provide type-safety and avoid function pointer casting for the
2667 type-specific callbacks.
2670 *) Added Kerberos Cipher Suites to be used with TLS, as written in
2672 [Veers Staats <staatsvr@asc.hpc.mil>,
2673 Jeffrey Altman <jaltman@columbia.edu>, via Richard Levitte]
2675 *) Reformat the FAQ so the different questions and answers can be divided
2676 in sections depending on the subject.
2679 *) Have the zlib compression code load ZLIB.DLL dynamically under
2683 *) New function BN_mod_sqrt for computing square roots modulo a prime
2684 (using the probabilistic Tonelli-Shanks algorithm unless
2685 p == 3 (mod 4) or p == 5 (mod 8), which are cases that can
2686 be handled deterministically).
2687 [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller]
2689 *) Make BN_mod_inverse faster by explicitly handling small quotients
2690 in the Euclid loop. (Speed gain about 20% for small moduli [256 or
2691 512 bits], about 30% for larger ones [1024 or 2048 bits].)
2694 *) New function BN_kronecker.
2697 *) Fix BN_gcd so that it works on negative inputs; the result is
2698 positive unless both parameters are zero.
2699 Previously something reasonably close to an infinite loop was
2700 possible because numbers could be growing instead of shrinking
2701 in the implementation of Euclid's algorithm.
2704 *) Fix BN_is_word() and BN_is_one() macros to take into account the
2705 sign of the number in question.
2707 Fix BN_is_word(a,w) to work correctly for w == 0.
2709 The old BN_is_word(a,w) macro is now called BN_abs_is_word(a,w)
2710 because its test if the absolute value of 'a' equals 'w'.
2711 Note that BN_abs_is_word does *not* handle w == 0 reliably;
2712 it exists mostly for use in the implementations of BN_is_zero(),
2713 BN_is_one(), and BN_is_word().
2716 *) New function BN_swap.
2719 *) Use BN_nnmod instead of BN_mod in crypto/bn/bn_exp.c so that
2720 the exponentiation functions are more likely to produce reasonable
2721 results on negative inputs.
2724 *) Change BN_mod_mul so that the result is always non-negative.
2725 Previously, it could be negative if one of the factors was negative;
2726 I don't think anyone really wanted that behaviour.
2729 *) Move BN_mod_... functions into new file crypto/bn/bn_mod.c
2730 (except for exponentiation, which stays in crypto/bn/bn_exp.c,
2731 and BN_mod_mul_reciprocal, which stays in crypto/bn/bn_recp.c)
2732 and add new functions:
2741 BN_mod_lshift1_quick
2745 These functions always generate non-negative results.
2747 BN_nnmod otherwise is like BN_mod (if BN_mod computes a remainder r
2748 such that |m| < r < 0, BN_nnmod will output rem + |m| instead).
2750 BN_mod_XXX_quick(r, a, [b,] m) generates the same result as
2751 BN_mod_XXX(r, a, [b,] m, ctx), but requires that a [and b]
2752 be reduced modulo m.
2753 [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller]
2756 The following entry accidentily appeared in the CHANGES file
2757 distributed with OpenSSL 0.9.7. The modifications described in
2758 it do *not* apply to OpenSSL 0.9.7.
2760 *) Remove a few calls to bn_wexpand() in BN_sqr() (the one in there
2761 was actually never needed) and in BN_mul(). The removal in BN_mul()
2762 required a small change in bn_mul_part_recursive() and the addition
2763 of the functions bn_cmp_part_words(), bn_sub_part_words() and
2764 bn_add_part_words(), which do the same thing as bn_cmp_words(),
2765 bn_sub_words() and bn_add_words() except they take arrays with
2770 *) In 'openssl passwd', verify passwords read from the terminal
2771 unless the '-salt' option is used (which usually means that
2772 verification would just waste user's time since the resulting
2773 hash is going to be compared with some given password hash)
2774 or the new '-noverify' option is used.
2776 This is an incompatible change, but it does not affect
2777 non-interactive use of 'openssl passwd' (passwords on the command
2778 line, '-stdin' option, '-in ...' option) and thus should not
2782 *) Remove all references to RSAref, since there's no more need for it.
2785 *) Make DSO load along a path given through an environment variable
2786 (SHLIB_PATH) with shl_load().
2789 *) Constify the ENGINE code as a result of BIGNUM constification.
2790 Also constify the RSA code and most things related to it. In a
2791 few places, most notable in the depth of the ASN.1 code, ugly
2792 casts back to non-const were required (to be solved at a later
2796 *) Make it so the openssl application has all engines loaded by default.
2799 *) Constify the BIGNUM routines a little more.
2802 *) Add the following functions:
2804 ENGINE_load_cswift()
2806 ENGINE_load_atalla()
2808 ENGINE_load_builtin_engines()
2810 That way, an application can itself choose if external engines that
2811 are built-in in OpenSSL shall ever be used or not. The benefit is
2812 that applications won't have to be linked with libdl or other dso
2813 libraries unless it's really needed.
2815 Changed 'openssl engine' to load all engines on demand.
2816 Changed the engine header files to avoid the duplication of some
2817 declarations (they differed!).
2820 *) 'openssl engine' can now list capabilities.
2823 *) Better error reporting in 'openssl engine'.
2826 *) Never call load_dh_param(NULL) in s_server.
2829 *) Add engine application. It can currently list engines by name and
2830 identity, and test if they are actually available.
2833 *) Improve RPM specification file by forcing symbolic linking and making
2834 sure the installed documentation is also owned by root.root.
2835 [Damien Miller <djm@mindrot.org>]
2837 *) Give the OpenSSL applications more possibilities to make use of
2838 keys (public as well as private) handled by engines.
2841 *) Add OCSP code that comes from CertCo.
2844 *) Add VMS support for the Rijndael code.
2847 *) Added untested support for Nuron crypto accelerator.
2850 *) Add support for external cryptographic devices. This code was
2851 previously distributed separately as the "engine" branch.
2852 [Geoff Thorpe, Richard Levitte]
2854 *) Rework the filename-translation in the DSO code. It is now possible to
2855 have far greater control over how a "name" is turned into a filename
2856 depending on the operating environment and any oddities about the
2857 different shared library filenames on each system.
2860 *) Support threads on FreeBSD-elf in Configure.
2863 *) Fix for SHA1 assembly problem with MASM: it produces
2864 warnings about corrupt line number information when assembling
2865 with debugging information. This is caused by the overlapping
2867 [Bernd Matthes <mainbug@celocom.de>, Steve Henson]
2870 NCONF_get_number() has no error checking at all. As a replacement,
2871 NCONF_get_number_e() is defined (_e for "error checking") and is
2872 promoted strongly. The old NCONF_get_number is kept around for
2873 binary backward compatibility.
2874 Make it possible for methods to load from something other than a BIO,
2875 by providing a function pointer that is given a name instead of a BIO.
2876 For example, this could be used to load configuration data from an
2880 *) Fix for non blocking accept BIOs. Added new I/O special reason
2881 BIO_RR_ACCEPT to cover this case. Previously use of accept BIOs
2882 with non blocking I/O was not possible because no retry code was
2883 implemented. Also added new SSL code SSL_WANT_ACCEPT to cover
2887 *) Added the beginnings of Rijndael support.
2890 *) Fix for bug in DirectoryString mask setting. Add support for
2891 X509_NAME_print_ex() in 'req' and X509_print_ex() function
2892 to allow certificate printing to more controllable, additional
2893 'certopt' option to 'x509' to allow new printing options to be
2897 *) Clean old EAY MD5 hack from e_os.h.
2900 Changes between 0.9.6l and 0.9.6m [17 Mar 2004]
2902 *) Fix null-pointer assignment in do_change_cipher_spec() revealed
2903 by using the Codenomicon TLS Test Tool (CAN-2004-0079)
2904 [Joe Orton, Steve Henson]
2906 Changes between 0.9.6k and 0.9.6l [04 Nov 2003]
2908 *) Fix additional bug revealed by the NISCC test suite:
2910 Stop bug triggering large recursion when presented with
2911 certain ASN.1 tags (CAN-2003-0851)
2914 Changes between 0.9.6j and 0.9.6k [30 Sep 2003]
2916 *) Fix various bugs revealed by running the NISCC test suite:
2918 Stop out of bounds reads in the ASN1 code when presented with
2919 invalid tags (CAN-2003-0543 and CAN-2003-0544).
2921 If verify callback ignores invalid public key errors don't try to check
2922 certificate signature with the NULL public key.
2926 *) In ssl3_accept() (ssl/s3_srvr.c) only accept a client certificate
2927 if the server requested one: as stated in TLS 1.0 and SSL 3.0
2931 *) In ssl3_get_client_hello() (ssl/s3_srvr.c), tolerate additional
2932 extra data after the compression methods not only for TLS 1.0
2933 but also for SSL 3.0 (as required by the specification).
2934 [Bodo Moeller; problem pointed out by Matthias Loepfe]
2936 *) Change X509_certificate_type() to mark the key as exported/exportable
2937 when it's 512 *bits* long, not 512 bytes.
2940 Changes between 0.9.6i and 0.9.6j [10 Apr 2003]
2942 *) Countermeasure against the Klima-Pokorny-Rosa extension of
2943 Bleichbacher's attack on PKCS #1 v1.5 padding: treat
2944 a protocol version number mismatch like a decryption error
2945 in ssl3_get_client_key_exchange (ssl/s3_srvr.c).
2948 *) Turn on RSA blinding by default in the default implementation
2949 to avoid a timing attack. Applications that don't want it can call
2950 RSA_blinding_off() or use the new flag RSA_FLAG_NO_BLINDING.
2951 They would be ill-advised to do so in most cases.
2952 [Ben Laurie, Steve Henson, Geoff Thorpe, Bodo Moeller]
2954 *) Change RSA blinding code so that it works when the PRNG is not
2955 seeded (in this case, the secret RSA exponent is abused as
2956 an unpredictable seed -- if it is not unpredictable, there
2957 is no point in blinding anyway). Make RSA blinding thread-safe
2958 by remembering the creator's thread ID in rsa->blinding and
2959 having all other threads use local one-time blinding factors
2960 (this requires more computation than sharing rsa->blinding, but
2961 avoids excessive locking; and if an RSA object is not shared
2962 between threads, blinding will still be very fast).
2965 Changes between 0.9.6h and 0.9.6i [19 Feb 2003]
2967 *) In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked
2968 via timing by performing a MAC computation even if incorrrect
2969 block cipher padding has been found. This is a countermeasure
2970 against active attacks where the attacker has to distinguish
2971 between bad padding and a MAC verification error. (CAN-2003-0078)
2973 [Bodo Moeller; problem pointed out by Brice Canvel (EPFL),
2974 Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and
2975 Martin Vuagnoux (EPFL, Ilion)]
2977 Changes between 0.9.6g and 0.9.6h [5 Dec 2002]
2979 *) New function OPENSSL_cleanse(), which is used to cleanse a section of
2980 memory from it's contents. This is done with a counter that will
2981 place alternating values in each byte. This can be used to solve
2982 two issues: 1) the removal of calls to memset() by highly optimizing
2983 compilers, and 2) cleansing with other values than 0, since those can
2984 be read through on certain media, for example a swap space on disk.
2987 *) Bugfix: client side session caching did not work with external caching,
2988 because the session->cipher setting was not restored when reloading
2989 from the external cache. This problem was masked, when
2990 SSL_OP_NETSCAPE_REUSE_CIPHER_CHANGE_BUG (part of SSL_OP_ALL) was set.
2991 (Found by Steve Haslam <steve@araqnid.ddts.net>.)
2994 *) Fix client_certificate (ssl/s2_clnt.c): The permissible total
2995 length of the REQUEST-CERTIFICATE message is 18 .. 34, not 17 .. 33.
2996 [Zeev Lieber <zeev-l@yahoo.com>]
2998 *) Undo an undocumented change introduced in 0.9.6e which caused
2999 repeated calls to OpenSSL_add_all_ciphers() and
3000 OpenSSL_add_all_digests() to be ignored, even after calling
3004 *) Change the default configuration reader to deal with last line not
3005 being properly terminated.
3008 *) Change X509_NAME_cmp() so it applies the special rules on handling
3009 DN values that are of type PrintableString, as well as RDNs of type
3010 emailAddress where the value has the type ia5String.
3011 [stefank@valicert.com via Richard Levitte]
3013 *) Add a SSL_SESS_CACHE_NO_INTERNAL_STORE flag to take over half
3014 the job SSL_SESS_CACHE_NO_INTERNAL_LOOKUP was inconsistently
3015 doing, define a new flag (SSL_SESS_CACHE_NO_INTERNAL) to be
3016 the bitwise-OR of the two for use by the majority of applications
3017 wanting this behaviour, and update the docs. The documented
3018 behaviour and actual behaviour were inconsistent and had been
3019 changing anyway, so this is more a bug-fix than a behavioural
3021 [Geoff Thorpe, diagnosed by Nadav Har'El]
3023 *) Don't impose a 16-byte length minimum on session IDs in ssl/s3_clnt.c
3024 (the SSL 3.0 and TLS 1.0 specifications allow any length up to 32 bytes).
3027 *) Fix initialization code race conditions in
3028 SSLv23_method(), SSLv23_client_method(), SSLv23_server_method(),
3029 SSLv2_method(), SSLv2_client_method(), SSLv2_server_method(),
3030 SSLv3_method(), SSLv3_client_method(), SSLv3_server_method(),
3031 TLSv1_method(), TLSv1_client_method(), TLSv1_server_method(),
3032 ssl2_get_cipher_by_char(),
3033 ssl3_get_cipher_by_char().
3034 [Patrick McCormick <patrick@tellme.com>, Bodo Moeller]
3036 *) Reorder cleanup sequence in SSL_CTX_free(): only remove the ex_data after
3037 the cached sessions are flushed, as the remove_cb() might use ex_data
3038 contents. Bug found by Sam Varshavchik <mrsam@courier-mta.com>
3039 (see [openssl.org #212]).
3040 [Geoff Thorpe, Lutz Jaenicke]
3042 *) Fix typo in OBJ_txt2obj which incorrectly passed the content
3043 length, instead of the encoding length to d2i_ASN1_OBJECT.
3046 Changes between 0.9.6f and 0.9.6g [9 Aug 2002]
3048 *) [In 0.9.6g-engine release:]
3049 Fix crypto/engine/vendor_defns/cswift.h for WIN32 (use '_stdcall').
3050 [Lynn Gazis <lgazis@rainbow.com>]
3052 Changes between 0.9.6e and 0.9.6f [8 Aug 2002]
3054 *) Fix ASN1 checks. Check for overflow by comparing with LONG_MAX
3055 and get fix the header length calculation.
3056 [Florian Weimer <Weimer@CERT.Uni-Stuttgart.DE>,
3057 Alon Kantor <alonk@checkpoint.com> (and others),
3060 *) Use proper error handling instead of 'assertions' in buffer
3061 overflow checks added in 0.9.6e. This prevents DoS (the
3062 assertions could call abort()).
3063 [Arne Ansper <arne@ats.cyber.ee>, Bodo Moeller]
3065 Changes between 0.9.6d and 0.9.6e [30 Jul 2002]
3067 *) Add various sanity checks to asn1_get_length() to reject
3068 the ASN1 length bytes if they exceed sizeof(long), will appear
3069 negative or the content length exceeds the length of the
3071 [Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>]
3073 *) Fix cipher selection routines: ciphers without encryption had no flags
3074 for the cipher strength set and where therefore not handled correctly
3075 by the selection routines (PR #130).
3078 *) Fix EVP_dsa_sha macro.
3082 SSL_OP_DONT_INSERT_EMPTY_FRAGMENTS
3083 for disabling the SSL 3.0/TLS 1.0 CBC vulnerability countermeasure
3084 that was added in OpenSSL 0.9.6d.
3086 As the countermeasure turned out to be incompatible with some
3087 broken SSL implementations, the new option is part of SSL_OP_ALL.
3088 SSL_OP_ALL is usually employed when compatibility with weird SSL
3089 implementations is desired (e.g. '-bugs' option to 's_client' and
3090 's_server'), so the new option is automatically set in many
3094 *) Changes in security patch:
3096 Changes marked "(CHATS)" were sponsored by the Defense Advanced
3097 Research Projects Agency (DARPA) and Air Force Research Laboratory,
3098 Air Force Materiel Command, USAF, under agreement number
3101 *) Add various sanity checks to asn1_get_length() to reject
3102 the ASN1 length bytes if they exceed sizeof(long), will appear
3103 negative or the content length exceeds the length of the
3104 supplied buffer. (CAN-2002-0659)
3105 [Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>]
3107 *) Assertions for various potential buffer overflows, not known to
3109 [Ben Laurie (CHATS)]
3111 *) Various temporary buffers to hold ASCII versions of integers were
3112 too small for 64 bit platforms. (CAN-2002-0655)
3113 [Matthew Byng-Maddick <mbm@aldigital.co.uk> and Ben Laurie (CHATS)>
3115 *) Remote buffer overflow in SSL3 protocol - an attacker could
3116 supply an oversized session ID to a client. (CAN-2002-0656)
3117 [Ben Laurie (CHATS)]
3119 *) Remote buffer overflow in SSL2 protocol - an attacker could
3120 supply an oversized client master key. (CAN-2002-0656)
3121 [Ben Laurie (CHATS)]
3123 Changes between 0.9.6c and 0.9.6d [9 May 2002]
3125 *) Fix crypto/asn1/a_sign.c so that 'parameters' is omitted (not
3126 encoded as NULL) with id-dsa-with-sha1.
3127 [Nils Larsch <nla@trustcenter.de>; problem pointed out by Bodo Moeller]
3129 *) Check various X509_...() return values in apps/req.c.
3130 [Nils Larsch <nla@trustcenter.de>]
3132 *) Fix BASE64 decode (EVP_DecodeUpdate) for data with CR/LF ended lines:
3133 an end-of-file condition would erronously be flagged, when the CRLF
3134 was just at the end of a processed block. The bug was discovered when
3135 processing data through a buffering memory BIO handing the data to a
3136 BASE64-decoding BIO. Bug fund and patch submitted by Pavel Tsekov
3137 <ptsekov@syntrex.com> and Nedelcho Stanev.
3140 *) Implement a countermeasure against a vulnerability recently found
3141 in CBC ciphersuites in SSL 3.0/TLS 1.0: Send an empty fragment
3142 before application data chunks to avoid the use of known IVs
3143 with data potentially chosen by the attacker.
3146 *) Fix length checks in ssl3_get_client_hello().
3149 *) TLS/SSL library bugfix: use s->s3->in_read_app_data differently
3150 to prevent ssl3_read_internal() from incorrectly assuming that
3151 ssl3_read_bytes() found application data while handshake
3152 processing was enabled when in fact s->s3->in_read_app_data was
3153 merely automatically cleared during the initial handshake.
3154 [Bodo Moeller; problem pointed out by Arne Ansper <arne@ats.cyber.ee>]
3156 *) Fix object definitions for Private and Enterprise: they were not
3157 recognized in their shortname (=lowercase) representation. Extend
3158 obj_dat.pl to issue an error when using undefined keywords instead
3159 of silently ignoring the problem (Svenning Sorensen
3160 <sss@sss.dnsalias.net>).
3163 *) Fix DH_generate_parameters() so that it works for 'non-standard'
3164 generators, i.e. generators other than 2 and 5. (Previously, the
3165 code did not properly initialise the 'add' and 'rem' values to
3166 BN_generate_prime().)
3168 In the new general case, we do not insist that 'generator' is
3169 actually a primitive root: This requirement is rather pointless;
3170 a generator of the order-q subgroup is just as good, if not
3174 *) Map new X509 verification errors to alerts. Discovered and submitted by
3175 Tom Wu <tom@arcot.com>.
3178 *) Fix ssl3_pending() (ssl/s3_lib.c) to prevent SSL_pending() from
3179 returning non-zero before the data has been completely received
3180 when using non-blocking I/O.
3181 [Bodo Moeller; problem pointed out by John Hughes]
3183 *) Some of the ciphers missed the strength entry (SSL_LOW etc).
3184 [Ben Laurie, Lutz Jaenicke]
3186 *) Fix bug in SSL_clear(): bad sessions were not removed (found by
3187 Yoram Zahavi <YoramZ@gilian.com>).
3190 *) Add information about CygWin 1.3 and on, and preserve proper
3191 configuration for the versions before that.
3192 [Corinna Vinschen <vinschen@redhat.com> and Richard Levitte]
3194 *) Make removal from session cache (SSL_CTX_remove_session()) more robust:
3195 check whether we deal with a copy of a session and do not delete from
3196 the cache in this case. Problem reported by "Izhar Shoshani Levi"
3197 <izhar@checkpoint.com>.
3200 *) Do not store session data into the internal session cache, if it
3201 is never intended to be looked up (SSL_SESS_CACHE_NO_INTERNAL_LOOKUP
3202 flag is set). Proposed by Aslam <aslam@funk.com>.
3205 *) Have ASN1_BIT_STRING_set_bit() really clear a bit when the requested
3209 *) [In 0.9.6d-engine release:]
3210 Fix a crashbug and a logic bug in hwcrhk_load_pubkey().
3211 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte]
3213 *) Add the configuration target linux-s390x.
3214 [Neale Ferguson <Neale.Ferguson@SoftwareAG-USA.com> via Richard Levitte]
3216 *) The earlier bugfix for the SSL3_ST_SW_HELLO_REQ_C case of
3217 ssl3_accept (ssl/s3_srvr.c) incorrectly used a local flag
3218 variable as an indication that a ClientHello message has been
3219 received. As the flag value will be lost between multiple
3220 invocations of ssl3_accept when using non-blocking I/O, the
3221 function may not be aware that a handshake has actually taken
3222 place, thus preventing a new session from being added to the
3225 To avoid this problem, we now set s->new_session to 2 instead of
3226 using a local variable.
3227 [Lutz Jaenicke, Bodo Moeller]
3229 *) Bugfix: Return -1 from ssl3_get_server_done (ssl3/s3_clnt.c)
3230 if the SSL_R_LENGTH_MISMATCH error is detected.
3231 [Geoff Thorpe, Bodo Moeller]
3233 *) New 'shared_ldflag' column in Configure platform table.
3236 *) Fix EVP_CIPHER_mode macro.
3237 ["Dan S. Camper" <dan@bti.net>]
3239 *) Fix ssl3_read_bytes (ssl/s3_pkt.c): To ignore messages of unknown
3240 type, we must throw them away by setting rr->length to 0.
3241 [D P Chang <dpc@qualys.com>]
3243 Changes between 0.9.6b and 0.9.6c [21 dec 2001]
3245 *) Fix BN_rand_range bug pointed out by Dominikus Scherkl
3246 <Dominikus.Scherkl@biodata.com>. (The previous implementation
3247 worked incorrectly for those cases where range = 10..._2 and
3248 3*range is two bits longer than range.)
3251 *) Only add signing time to PKCS7 structures if it is not already
3255 *) Fix crypto/objects/objects.h: "ld-ce" should be "id-ce",
3256 OBJ_ld_ce should be OBJ_id_ce.
3257 Also some ip-pda OIDs in crypto/objects/objects.txt were
3258 incorrect (cf. RFC 3039).
3259 [Matt Cooper, Frederic Giudicelli, Bodo Moeller]
3261 *) Release CRYPTO_LOCK_DYNLOCK when CRYPTO_destroy_dynlockid()
3262 returns early because it has nothing to do.
3263 [Andy Schneider <andy.schneider@bjss.co.uk>]
3265 *) [In 0.9.6c-engine release:]
3266 Fix mutex callback return values in crypto/engine/hw_ncipher.c.
3267 [Andy Schneider <andy.schneider@bjss.co.uk>]
3269 *) [In 0.9.6c-engine release:]
3270 Add support for Cryptographic Appliance's keyserver technology.
3271 (Use engine 'keyclient')
3272 [Cryptographic Appliances and Geoff Thorpe]
3274 *) Add a configuration entry for OS/390 Unix. The C compiler 'c89'
3275 is called via tools/c89.sh because arguments have to be
3276 rearranged (all '-L' options must appear before the first object
3278 [Richard Shapiro <rshapiro@abinitio.com>]
3280 *) [In 0.9.6c-engine release:]
3281 Add support for Broadcom crypto accelerator cards, backported
3283 [Broadcom, Nalin Dahyabhai <nalin@redhat.com>, Mark Cox]
3285 *) [In 0.9.6c-engine release:]
3286 Add support for SureWare crypto accelerator cards from
3287 Baltimore Technologies. (Use engine 'sureware')
3288 [Baltimore Technologies and Mark Cox]
3290 *) [In 0.9.6c-engine release:]
3291 Add support for crypto accelerator cards from Accelerated
3292 Encryption Processing, www.aep.ie. (Use engine 'aep')
3293 [AEP Inc. and Mark Cox]
3295 *) Add a configuration entry for gcc on UnixWare.
3296 [Gary Benson <gbenson@redhat.com>]
3298 *) Change ssl/s2_clnt.c and ssl/s2_srvr.c so that received handshake
3299 messages are stored in a single piece (fixed-length part and
3300 variable-length part combined) and fix various bugs found on the way.
3303 *) Disable caching in BIO_gethostbyname(), directly use gethostbyname()
3304 instead. BIO_gethostbyname() does not know what timeouts are
3305 appropriate, so entries would stay in cache even when they have
3307 [Bodo Moeller; problem pointed out by Rich Salz <rsalz@zolera.com>
3309 *) Change ssl23_get_client_hello (ssl/s23_srvr.c) behaviour when
3310 faced with a pathologically small ClientHello fragment that does
3311 not contain client_version: Instead of aborting with an error,
3312 simply choose the highest available protocol version (i.e.,
3313 TLS 1.0 unless it is disabled). In practice, ClientHello
3314 messages are never sent like this, but this change gives us
3315 strictly correct behaviour at least for TLS.
3318 *) Fix SSL handshake functions and SSL_clear() such that SSL_clear()
3319 never resets s->method to s->ctx->method when called from within
3320 one of the SSL handshake functions.
3321 [Bodo Moeller; problem pointed out by Niko Baric]
3323 *) In ssl3_get_client_hello (ssl/s3_srvr.c), generate a fatal alert
3324 (sent using the client's version number) if client_version is
3325 smaller than the protocol version in use. Also change
3326 ssl23_get_client_hello (ssl/s23_srvr.c) to select TLS 1.0 if
3327 the client demanded SSL 3.0 but only TLS 1.0 is enabled; then
3328 the client will at least see that alert.
3331 *) Fix ssl3_get_message (ssl/s3_both.c) to handle message fragmentation
3335 *) Avoid infinite loop in ssl3_get_message (ssl/s3_both.c) if a
3336 client receives HelloRequest while in a handshake.
3337 [Bodo Moeller; bug noticed by Andy Schneider <andy.schneider@bjss.co.uk>]
3339 *) Bugfix in ssl3_accept (ssl/s3_srvr.c): Case SSL3_ST_SW_HELLO_REQ_C
3340 should end in 'break', not 'goto end' which circuments various
3341 cleanups done in state SSL_ST_OK. But session related stuff
3342 must be disabled for SSL_ST_OK in the case that we just sent a
3345 Also avoid some overhead by not calling ssl_init_wbio_buffer()
3346 before just sending a HelloRequest.
3347 [Bodo Moeller, Eric Rescorla <ekr@rtfm.com>]
3349 *) Fix ssl/s3_enc.c, ssl/t1_enc.c and ssl/s3_pkt.c so that we don't
3350 reveal whether illegal block cipher padding was found or a MAC
3351 verification error occured. (Neither SSLerr() codes nor alerts
3352 are directly visible to potential attackers, but the information
3353 may leak via logfiles.)
3355 Similar changes are not required for the SSL 2.0 implementation
3356 because the number of padding bytes is sent in clear for SSL 2.0,
3357 and the extra bytes are just ignored. However ssl/s2_pkt.c
3358 failed to verify that the purported number of padding bytes is in
3362 *) Add OpenUNIX-8 support including shared libraries
3363 (Boyd Lynn Gerber <gerberb@zenez.com>).
3366 *) Improve RSA_padding_check_PKCS1_OAEP() check again to avoid
3367 'wristwatch attack' using huge encoding parameters (cf.
3368 James H. Manger's CRYPTO 2001 paper). Note that the
3369 RSA_PKCS1_OAEP_PADDING case of RSA_private_decrypt() does not use
3370 encoding parameters and hence was not vulnerable.
3373 *) BN_sqr() bug fix.
3374 [Ulf Möller, reported by Jim Ellis <jim.ellis@cavium.com>]
3376 *) Rabin-Miller test analyses assume uniformly distributed witnesses,
3377 so use BN_pseudo_rand_range() instead of using BN_pseudo_rand()
3378 followed by modular reduction.
3379 [Bodo Moeller; pointed out by Adam Young <AYoung1@NCSUS.JNJ.COM>]
3381 *) Add BN_pseudo_rand_range() with obvious functionality: BN_rand_range()
3382 equivalent based on BN_pseudo_rand() instead of BN_rand().
3385 *) s3_srvr.c: allow sending of large client certificate lists (> 16 kB).
3386 This function was broken, as the check for a new client hello message
3387 to handle SGC did not allow these large messages.
3388 (Tracked down by "Douglas E. Engert" <deengert@anl.gov>.)
3391 *) Add alert descriptions for TLSv1 to SSL_alert_desc_string[_long]().
3394 *) Fix buggy behaviour of BIO_get_num_renegotiates() and BIO_ctrl()
3395 for BIO_C_GET_WRITE_BUF_SIZE ("Stephen Hinton" <shinton@netopia.com>).
3398 *) Rework the configuration and shared library support for Tru64 Unix.
3399 The configuration part makes use of modern compiler features and
3400 still retains old compiler behavior for those that run older versions
3401 of the OS. The shared library support part includes a variant that
3402 uses the RPATH feature, and is available through the special
3403 configuration target "alpha-cc-rpath", which will never be selected
3405 [Tim Mooney <mooney@dogbert.cc.ndsu.NoDak.edu> via Richard Levitte]
3407 *) In ssl3_get_key_exchange (ssl/s3_clnt.c), call ssl3_get_message()
3408 with the same message size as in ssl3_get_certificate_request().
3409 Otherwise, if no ServerKeyExchange message occurs, CertificateRequest
3410 messages might inadvertently be reject as too long.
3411 [Petr Lampa <lampa@fee.vutbr.cz>]
3413 *) Enhanced support for IA-64 Unix platforms (well, Linux and HP-UX).
3416 *) Modified SSL library such that the verify_callback that has been set
3417 specificly for an SSL object with SSL_set_verify() is actually being
3418 used. Before the change, a verify_callback set with this function was
3419 ignored and the verify_callback() set in the SSL_CTX at the time of
3420 the call was used. New function X509_STORE_CTX_set_verify_cb() introduced
3421 to allow the necessary settings.
3424 *) Initialize static variable in crypto/dsa/dsa_lib.c and crypto/dh/dh_lib.c
3425 explicitly to NULL, as at least on Solaris 8 this seems not always to be
3426 done automatically (in contradiction to the requirements of the C
3427 standard). This made problems when used from OpenSSH.
3430 *) In OpenSSL 0.9.6a and 0.9.6b, crypto/dh/dh_key.c ignored
3431 dh->length and always used
3433 BN_rand_range(priv_key, dh->p).
3435 BN_rand_range() is not necessary for Diffie-Hellman, and this
3436 specific range makes Diffie-Hellman unnecessarily inefficient if
3437 dh->length (recommended exponent length) is much smaller than the
3438 length of dh->p. We could use BN_rand_range() if the order of
3439 the subgroup was stored in the DH structure, but we only have
3444 BN_rand(priv_key, l, ...)
3446 where 'l' is dh->length if this is defined, or BN_num_bits(dh->p)-1
3452 RSA_eay_public_encrypt
3453 RSA_eay_private_decrypt
3454 RSA_eay_private_encrypt (signing)
3455 RSA_eay_public_decrypt (signature verification)
3457 (default implementations for RSA_public_encrypt,
3458 RSA_private_decrypt, RSA_private_encrypt, RSA_public_decrypt),
3459 always reject numbers >= n.
3462 *) In crypto/rand/md_rand.c, use a new short-time lock CRYPTO_LOCK_RAND2
3463 to synchronize access to 'locking_thread'. This is necessary on
3464 systems where access to 'locking_thread' (an 'unsigned long'
3465 variable) is not atomic.
3468 *) In crypto/rand/md_rand.c, set 'locking_thread' to current thread's ID
3469 *before* setting the 'crypto_lock_rand' flag. The previous code had
3470 a race condition if 0 is a valid thread ID.
3471 [Travis Vitek <vitek@roguewave.com>]
3473 *) Add support for shared libraries under Irix.
3474 [Albert Chin-A-Young <china@thewrittenword.com>]
3476 *) Add configuration option to build on Linux on both big-endian and
3478 [Ralf Baechle <ralf@uni-koblenz.de>]
3480 *) Add the possibility to create shared libraries on HP-UX.
3483 Changes between 0.9.6a and 0.9.6b [9 Jul 2001]
3485 *) Change ssleay_rand_bytes (crypto/rand/md_rand.c)
3486 to avoid a SSLeay/OpenSSL PRNG weakness pointed out by
3487 Markku-Juhani O. Saarinen <markku-juhani.saarinen@nokia.com>:
3488 PRNG state recovery was possible based on the output of
3489 one PRNG request appropriately sized to gain knowledge on
3490 'md' followed by enough consecutive 1-byte PRNG requests
3491 to traverse all of 'state'.
3493 1. When updating 'md_local' (the current thread's copy of 'md')
3494 during PRNG output generation, hash all of the previous
3495 'md_local' value, not just the half used for PRNG output.
3497 2. Make the number of bytes from 'state' included into the hash
3498 independent from the number of PRNG bytes requested.
3500 The first measure alone would be sufficient to avoid
3501 Markku-Juhani's attack. (Actually it had never occurred
3502 to me that the half of 'md_local' used for chaining was the
3503 half from which PRNG output bytes were taken -- I had always
3504 assumed that the secret half would be used.) The second
3505 measure makes sure that additional data from 'state' is never
3506 mixed into 'md_local' in small portions; this heuristically
3507 further strengthens the PRNG.
3510 *) Fix crypto/bn/asm/mips3.s.
3513 *) When only the key is given to "enc", the IV is undefined. Print out
3514 an error message in this case.
3517 *) Handle special case when X509_NAME is empty in X509 printing routines.
3520 *) In dsa_do_verify (crypto/dsa/dsa_ossl.c), verify that r and s are
3521 positive and less than q.
3524 *) Don't change *pointer in CRYPTO_add_lock() is add_lock_callback is
3525 used: it isn't thread safe and the add_lock_callback should handle
3527 [Paul Rose <Paul.Rose@bridge.com>]
3529 *) Verify that incoming data obeys the block size in
3530 ssl3_enc (ssl/s3_enc.c) and tls1_enc (ssl/t1_enc.c).
3534 [Ulf Möller, Bodo Möller]
3536 *) The countermeasure against Bleichbacher's attack on PKCS #1 v1.5
3537 RSA encryption was accidentally removed in s3_srvr.c in OpenSSL 0.9.5
3538 when fixing the server behaviour for backwards-compatible 'client
3539 hello' messages. (Note that the attack is impractical against
3540 SSL 3.0 and TLS 1.0 anyway because length and version checking
3541 means that the probability of guessing a valid ciphertext is
3542 around 2^-40; see section 5 in Bleichenbacher's CRYPTO '98
3545 Before 0.9.5, the countermeasure (hide the error by generating a
3546 random 'decryption result') did not work properly because
3547 ERR_clear_error() was missing, meaning that SSL_get_error() would
3548 detect the supposedly ignored error.
3550 Both problems are now fixed.
3553 *) In crypto/bio/bf_buff.c, increase DEFAULT_BUFFER_SIZE to 4096
3554 (previously it was 1024).
3557 *) Fix for compatibility mode trust settings: ignore trust settings
3558 unless some valid trust or reject settings are present.
3561 *) Fix for blowfish EVP: its a variable length cipher.
3564 *) Fix various bugs related to DSA S/MIME verification. Handle missing
3565 parameters in DSA public key structures and return an error in the
3566 DSA routines if parameters are absent.
3569 *) In versions up to 0.9.6, RAND_file_name() resorted to file ".rnd"
3570 in the current directory if neither $RANDFILE nor $HOME was set.
3571 RAND_file_name() in 0.9.6a returned NULL in this case. This has
3572 caused some confusion to Windows users who haven't defined $HOME.
3573 Thus RAND_file_name() is changed again: e_os.h can define a
3574 DEFAULT_HOME, which will be used if $HOME is not set.
3575 For Windows, we use "C:"; on other platforms, we still require
3576 environment variables.
3578 *) Move 'if (!initialized) RAND_poll()' into regions protected by
3579 CRYPTO_LOCK_RAND. This is not strictly necessary, but avoids
3580 having multiple threads call RAND_poll() concurrently.
3583 *) In crypto/rand/md_rand.c, replace 'add_do_not_lock' flag by a
3584 combination of a flag and a thread ID variable.
3585 Otherwise while one thread is in ssleay_rand_bytes (which sets the
3586 flag), *other* threads can enter ssleay_add_bytes without obeying
3587 the CRYPTO_LOCK_RAND lock (and may even illegally release the lock
3588 that they do not hold after the first thread unsets add_do_not_lock).
3591 *) Change bctest again: '-x' expressions are not available in all
3595 Changes between 0.9.6 and 0.9.6a [5 Apr 2001]
3597 *) Fix a couple of memory leaks in PKCS7_dataDecode()
3598 [Steve Henson, reported by Heyun Zheng <hzheng@atdsprint.com>]
3600 *) Change Configure and Makefiles to provide EXE_EXT, which will contain
3601 the default extension for executables, if any. Also, make the perl
3602 scripts that use symlink() to test if it really exists and use "cp"
3603 if it doesn't. All this made OpenSSL compilable and installable in
3607 *) Fix for asn1_GetSequence() for indefinite length constructed data.
3608 If SEQUENCE is length is indefinite just set c->slen to the total
3609 amount of data available.
3610 [Steve Henson, reported by shige@FreeBSD.org]
3611 [This change does not apply to 0.9.7.]
3613 *) Change bctest to avoid here-documents inside command substitution
3614 (workaround for FreeBSD /bin/sh bug).
3615 For compatibility with Ultrix, avoid shell functions (introduced
3616 in the bctest version that searches along $PATH).
3619 *) Rename 'des_encrypt' to 'des_encrypt1'. This avoids the clashes
3620 with des_encrypt() defined on some operating systems, like Solaris
3624 *) Check the result of RSA-CRT (see D. Boneh, R. DeMillo, R. Lipton:
3625 On the Importance of Eliminating Errors in Cryptographic
3626 Computations, J. Cryptology 14 (2001) 2, 101-119,
3627 http://theory.stanford.edu/~dabo/papers/faults.ps.gz).
3630 *) MIPS assembler BIGNUM division bug fix.
3633 *) Disabled incorrect Alpha assembler code.
3636 *) Fix PKCS#7 decode routines so they correctly update the length
3637 after reading an EOC for the EXPLICIT tag.
3639 [This change does not apply to 0.9.7.]
3641 *) Fix bug in PKCS#12 key generation routines. This was triggered
3642 if a 3DES key was generated with a 0 initial byte. Include
3643 PKCS12_BROKEN_KEYGEN compilation option to retain the old
3644 (but broken) behaviour.
3647 *) Enhance bctest to search for a working bc along $PATH and print
3649 [Tim Rice <tim@multitalents.net> via Richard Levitte]
3651 *) Fix memory leaks in err.c: free err_data string if necessary;
3652 don't write to the wrong index in ERR_set_error_data.
3655 *) Implement ssl23_peek (analogous to ssl23_read), which previously
3659 *) Replace rdtsc with _emit statements for VC++ version 5.
3660 [Jeremy Cooper <jeremy@baymoo.org>]
3662 *) Make it possible to reuse SSLv2 sessions.
3665 *) In copy_email() check for >= 0 as a return value for
3666 X509_NAME_get_index_by_NID() since 0 is a valid index.
3667 [Steve Henson reported by Massimiliano Pala <madwolf@opensca.org>]
3669 *) Avoid coredump with unsupported or invalid public keys by checking if
3670 X509_get_pubkey() fails in PKCS7_verify(). Fix memory leak when
3671 PKCS7_verify() fails with non detached data.
3674 *) Don't use getenv in library functions when run as setuid/setgid.
3675 New function OPENSSL_issetugid().
3678 *) Avoid false positives in memory leak detection code (crypto/mem_dbg.c)
3679 due to incorrect handling of multi-threading:
3681 1. Fix timing glitch in the MemCheck_off() portion of CRYPTO_mem_ctrl().
3683 2. Fix logical glitch in is_MemCheck_on() aka CRYPTO_is_mem_check_on().
3685 3. Count how many times MemCheck_off() has been called so that
3686 nested use can be treated correctly. This also avoids
3687 inband-signalling in the previous code (which relied on the
3688 assumption that thread ID 0 is impossible).
3691 *) Add "-rand" option also to s_client and s_server.
3694 *) Fix CPU detection on Irix 6.x.
3695 [Kurt Hockenbury <khockenb@stevens-tech.edu> and
3696 "Bruce W. Forsberg" <bruce.forsberg@baesystems.com>]
3698 *) Fix X509_NAME bug which produced incorrect encoding if X509_NAME
3701 [This change does not apply to 0.9.7.]
3703 *) Use the cached encoding of an X509_NAME structure rather than
3704 copying it. This is apparently the reason for the libsafe "errors"
3705 but the code is actually correct.
3708 *) Add new function BN_rand_range(), and fix DSA_sign_setup() to prevent
3709 Bleichenbacher's DSA attack.
3710 Extend BN_[pseudo_]rand: As before, top=1 forces the highest two bits
3711 to be set and top=0 forces the highest bit to be set; top=-1 is new
3712 and leaves the highest bit random.
3713 [Ulf Moeller, Bodo Moeller]
3715 *) In the NCONF_...-based implementations for CONF_... queries
3716 (crypto/conf/conf_lib.c), if the input LHASH is NULL, avoid using
3717 a temporary CONF structure with the data component set to NULL
3718 (which gives segmentation faults in lh_retrieve).
3719 Instead, use NULL for the CONF pointer in CONF_get_string and
3720 CONF_get_number (which may use environment variables) and directly
3721 return NULL from CONF_get_section.
3724 *) Fix potential buffer overrun for EBCDIC.
3727 *) Tolerate nonRepudiation as being valid for S/MIME signing and certSign
3728 keyUsage if basicConstraints absent for a CA.
3731 *) Make SMIME_write_PKCS7() write mail header values with a format that
3732 is more generally accepted (no spaces before the semicolon), since
3733 some programs can't parse those values properly otherwise. Also make
3734 sure BIO's that break lines after each write do not create invalid
3738 *) Make the CRL encoding routines work with empty SEQUENCE OF. The
3739 macros previously used would not encode an empty SEQUENCE OF
3740 and break the signature.
3742 [This change does not apply to 0.9.7.]
3744 *) Zero the premaster secret after deriving the master secret in
3748 *) Add some EVP_add_digest_alias registrations (as found in
3749 OpenSSL_add_all_digests()) to SSL_library_init()
3750 aka OpenSSL_add_ssl_algorithms(). This provides improved
3751 compatibility with peers using X.509 certificates
3752 with unconventional AlgorithmIdentifier OIDs.
3755 *) Fix for Irix with NO_ASM.
3756 ["Bruce W. Forsberg" <bruce.forsberg@baesystems.com>]
3758 *) ./config script fixes.
3759 [Ulf Moeller, Richard Levitte]
3761 *) Fix 'openssl passwd -1'.
3764 *) Change PKCS12_key_gen_asc() so it can cope with non null
3765 terminated strings whose length is passed in the passlen
3766 parameter, for example from PEM callbacks. This was done
3767 by adding an extra length parameter to asc2uni().
3768 [Steve Henson, reported by <oddissey@samsung.co.kr>]
3770 *) Fix C code generated by 'openssl dsaparam -C': If a BN_bin2bn
3771 call failed, free the DSA structure.
3774 *) Fix to uni2asc() to cope with zero length Unicode strings.
3775 These are present in some PKCS#12 files.
3778 *) Increase s2->wbuf allocation by one byte in ssl2_new (ssl/s2_lib.c).
3779 Otherwise do_ssl_write (ssl/s2_pkt.c) will write beyond buffer limits
3780 when writing a 32767 byte record.
3781 [Bodo Moeller; problem reported by Eric Day <eday@concentric.net>]
3783 *) In RSA_eay_public_{en,ed}crypt and RSA_eay_mod_exp (rsa_eay.c),
3784 obtain lock CRYPTO_LOCK_RSA before setting rsa->_method_mod_{n,p,q}.
3786 (RSA objects have a reference count access to which is protected
3787 by CRYPTO_LOCK_RSA [see rsa_lib.c, s3_srvr.c, ssl_cert.c, ssl_rsa.c],
3788 so they are meant to be shared between threads.)
3789 [Bodo Moeller, Geoff Thorpe; original patch submitted by
3790 "Reddie, Steven" <Steven.Reddie@ca.com>]
3792 *) Fix a deadlock in CRYPTO_mem_leaks().
3795 *) Use better test patterns in bntest.
3798 *) rand_win.c fix for Borland C.
3801 *) BN_rshift bugfix for n == 0.
3804 *) Add a 'bctest' script that checks for some known 'bc' bugs
3805 so that 'make test' does not abort just because 'bc' is broken.
3808 *) Store verify_result within SSL_SESSION also for client side to
3809 avoid potential security hole. (Re-used sessions on the client side
3810 always resulted in verify_result==X509_V_OK, not using the original
3811 result of the server certificate verification.)
3814 *) Fix ssl3_pending: If the record in s->s3->rrec is not of type
3815 SSL3_RT_APPLICATION_DATA, return 0.
3816 Similarly, change ssl2_pending to return 0 if SSL_in_init(s) is true.
3820 Both ssl2_peek and ssl3_peek, which were totally broken in earlier
3821 releases, have been re-implemented by renaming the previous
3822 implementations of ssl2_read and ssl3_read to ssl2_read_internal
3823 and ssl3_read_internal, respectively, and adding 'peek' parameters
3824 to them. The new ssl[23]_{read,peek} functions are calls to
3825 ssl[23]_read_internal with the 'peek' flag set appropriately.
3826 A 'peek' parameter has also been added to ssl3_read_bytes, which
3827 does the actual work for ssl3_read_internal.
3830 *) Initialise "ex_data" member of RSA/DSA/DH structures prior to calling
3831 the method-specific "init()" handler. Also clean up ex_data after
3832 calling the method-specific "finish()" handler. Previously, this was
3833 happening the other way round.
3836 *) Increase BN_CTX_NUM (the number of BIGNUMs in a BN_CTX) to 16.
3837 The previous value, 12, was not always sufficient for BN_mod_exp().
3840 *) Make sure that shared libraries get the internal name engine with
3841 the full version number and not just 0. This should mark the
3842 shared libraries as not backward compatible. Of course, this should
3843 be changed again when we can guarantee backward binary compatibility.
3846 *) Fix typo in get_cert_by_subject() in by_dir.c
3847 [Jean-Marc Desperrier <jean-marc.desperrier@certplus.com>]
3849 *) Rework the system to generate shared libraries:
3851 - Make note of the expected extension for the shared libraries and
3852 if there is a need for symbolic links from for example libcrypto.so.0
3853 to libcrypto.so.0.9.7. There is extended info in Configure for
3856 - Make as few rebuilds of the shared libraries as possible.
3858 - Still avoid linking the OpenSSL programs with the shared libraries.
3860 - When installing, install the shared libraries separately from the
3864 *) Fix SSL_CTX_set_read_ahead macro to actually use its argument.
3866 Copy SSL_CTX's read_ahead flag to SSL object directly in SSL_new
3867 and not in SSL_clear because the latter is also used by the
3868 accept/connect functions; previously, the settings made by
3869 SSL_set_read_ahead would be lost during the handshake.
3870 [Bodo Moeller; problems reported by Anders Gertz <gertz@epact.se>]
3872 *) Correct util/mkdef.pl to be selective about disabled algorithms.
3873 Previously, it would create entries for disableed algorithms no
3877 *) Added several new manual pages for SSL_* function.
3880 Changes between 0.9.5a and 0.9.6 [24 Sep 2000]
3882 *) In ssl23_get_client_hello, generate an error message when faced
3883 with an initial SSL 3.0/TLS record that is too small to contain the
3884 first two bytes of the ClientHello message, i.e. client_version.
3885 (Note that this is a pathologic case that probably has never happened
3886 in real life.) The previous approach was to use the version number
3887 from the record header as a substitute; but our protocol choice
3888 should not depend on that one because it is not authenticated
3889 by the Finished messages.
3892 *) More robust randomness gathering functions for Windows.
3893 [Jeffrey Altman <jaltman@columbia.edu>]
3895 *) For compatibility reasons if the flag X509_V_FLAG_ISSUER_CHECK is
3896 not set then we don't setup the error code for issuer check errors
3897 to avoid possibly overwriting other errors which the callback does
3898 handle. If an application does set the flag then we assume it knows
3899 what it is doing and can handle the new informational codes
3903 *) Fix for a nasty bug in ASN1_TYPE handling. ASN1_TYPE is used for
3904 a general "ANY" type, as such it should be able to decode anything
3905 including tagged types. However it didn't check the class so it would
3906 wrongly interpret tagged types in the same way as their universal
3907 counterpart and unknown types were just rejected. Changed so that the
3908 tagged and unknown types are handled in the same way as a SEQUENCE:
3909 that is the encoding is stored intact. There is also a new type
3910 "V_ASN1_OTHER" which is used when the class is not universal, in this
3911 case we have no idea what the actual type is so we just lump them all
3915 *) On VMS, stdout may very well lead to a file that is written to
3916 in a record-oriented fashion. That means that every write() will
3917 write a separate record, which will be read separately by the
3918 programs trying to read from it. This can be very confusing.
3920 The solution is to put a BIO filter in the way that will buffer
3921 text until a linefeed is reached, and then write everything a
3922 line at a time, so every record written will be an actual line,
3923 not chunks of lines and not (usually doesn't happen, but I've
3924 seen it once) several lines in one record. BIO_f_linebuffer() is
3927 Currently, it's a VMS-only method, because that's where it has
3928 been tested well enough.
3931 *) Remove 'optimized' squaring variant in BN_mod_mul_montgomery,
3932 it can return incorrect results.
3933 (Note: The buggy variant was not enabled in OpenSSL 0.9.5a,
3934 but it was in 0.9.6-beta[12].)
3937 *) Disable the check for content being present when verifying detached
3938 signatures in pk7_smime.c. Some versions of Netscape (wrongly)
3939 include zero length content when signing messages.
3942 *) New BIO_shutdown_wr macro, which invokes the BIO_C_SHUTDOWN_WR
3943 BIO_ctrl (for BIO pairs).
3946 *) Add DSO method for VMS.
3949 *) Bug fix: Montgomery multiplication could produce results with the
3953 *) Add RPM specification openssl.spec and modify it to build three
3954 packages. The default package contains applications, application
3955 documentation and run-time libraries. The devel package contains
3956 include files, static libraries and function documentation. The
3957 doc package contains the contents of the doc directory. The original
3958 openssl.spec was provided by Damien Miller <djm@mindrot.org>.
3961 *) Add a large number of documentation files for many SSL routines.
3962 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>]
3964 *) Add a configuration entry for Sony News 4.
3965 [NAKAJI Hiroyuki <nakaji@tutrp.tut.ac.jp>]
3967 *) Don't set the two most significant bits to one when generating a
3968 random number < q in the DSA library.
3971 *) New SSL API mode 'SSL_MODE_AUTO_RETRY'. This disables the default
3972 behaviour that SSL_read may result in SSL_ERROR_WANT_READ (even if
3973 the underlying transport is blocking) if a handshake took place.
3974 (The default behaviour is needed by applications such as s_client
3975 and s_server that use select() to determine when to use SSL_read;
3976 but for applications that know in advance when to expect data, it
3977 just makes things more complicated.)
3980 *) Add RAND_egd_bytes(), which gives control over the number of bytes read
3984 *) Add a few more EBCDIC conditionals that make `req' and `x509'
3985 work better on such systems.
3986 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>]
3988 *) Add two demo programs for PKCS12_parse() and PKCS12_create().
3989 Update PKCS12_parse() so it copies the friendlyName and the
3990 keyid to the certificates aux info.
3993 *) Fix bug in PKCS7_verify() which caused an infinite loop
3994 if there was more than one signature.
3995 [Sven Uszpelkat <su@celocom.de>]
3997 *) Major change in util/mkdef.pl to include extra information
3998 about each symbol, as well as presentig variables as well
3999 as functions. This change means that there's n more need
4000 to rebuild the .num files when some algorithms are excluded.
4003 *) Allow the verify time to be set by an application,
4004 rather than always using the current time.
4007 *) Phase 2 verify code reorganisation. The certificate
4008 verify code now looks up an issuer certificate by a
4009 number of criteria: subject name, authority key id
4010 and key usage. It also verifies self signed certificates
4011 by the same criteria. The main comparison function is
4012 X509_check_issued() which performs these checks.
4014 Lot of changes were necessary in order to support this
4015 without completely rewriting the lookup code.
4017 Authority and subject key identifier are now cached.
4019 The LHASH 'certs' is X509_STORE has now been replaced
4020 by a STACK_OF(X509_OBJECT). This is mainly because an
4021 LHASH can't store or retrieve multiple objects with
4022 the same hash value.
4024 As a result various functions (which were all internal
4025 use only) have changed to handle the new X509_STORE
4026 structure. This will break anything that messed round
4027 with X509_STORE internally.
4029 The functions X509_STORE_add_cert() now checks for an
4030 exact match, rather than just subject name.
4032 The X509_STORE API doesn't directly support the retrieval
4033 of multiple certificates matching a given criteria, however
4034 this can be worked round by performing a lookup first
4035 (which will fill the cache with candidate certificates)
4036 and then examining the cache for matches. This is probably
4037 the best we can do without throwing out X509_LOOKUP
4038 entirely (maybe later...).
4040 The X509_VERIFY_CTX structure has been enhanced considerably.
4042 All certificate lookup operations now go via a get_issuer()
4043 callback. Although this currently uses an X509_STORE it
4044 can be replaced by custom lookups. This is a simple way
4045 to bypass the X509_STORE hackery necessary to make this
4046 work and makes it possible to use more efficient techniques
4047 in future. A very simple version which uses a simple
4048 STACK for its trusted certificate store is also provided
4049 using X509_STORE_CTX_trusted_stack().
4051 The verify_cb() and verify() callbacks now have equivalents
4052 in the X509_STORE_CTX structure.
4054 X509_STORE_CTX also has a 'flags' field which can be used
4055 to customise the verify behaviour.
4058 *) Add new PKCS#7 signing option PKCS7_NOSMIMECAP which
4059 excludes S/MIME capabilities.
4062 *) When a certificate request is read in keep a copy of the
4063 original encoding of the signed data and use it when outputing
4064 again. Signatures then use the original encoding rather than
4065 a decoded, encoded version which may cause problems if the
4066 request is improperly encoded.
4069 *) For consistency with other BIO_puts implementations, call
4070 buffer_write(b, ...) directly in buffer_puts instead of calling
4073 In BIO_puts, increment b->num_write as in BIO_write.
4074 [Peter.Sylvester@EdelWeb.fr]
4076 *) Fix BN_mul_word for the case where the word is 0. (We have to use
4077 BN_zero, we may not return a BIGNUM with an array consisting of
4081 *) Avoid calling abort() from within the library when problems are
4082 detected, except if preprocessor symbols have been defined
4083 (such as REF_CHECK, BN_DEBUG etc.).
4086 *) New openssl application 'rsautl'. This utility can be
4087 used for low level RSA operations. DER public key
4088 BIO/fp routines also added.
4091 *) New Configure entry and patches for compiling on QNX 4.
4092 [Andreas Schneider <andreas@ds3.etech.fh-hamburg.de>]
4094 *) A demo state-machine implementation was sponsored by
4095 Nuron (http://www.nuron.com/) and is now available in
4096 demos/state_machine.
4099 *) New options added to the 'dgst' utility for signature
4100 generation and verification.
4103 *) Unrecognized PKCS#7 content types are now handled via a
4104 catch all ASN1_TYPE structure. This allows unsupported
4105 types to be stored as a "blob" and an application can
4106 encode and decode it manually.
4109 *) Fix various signed/unsigned issues to make a_strex.c
4111 [Oscar Jacobsson <oscar.jacobsson@celocom.com>]
4113 *) ASN1 fixes. i2d_ASN1_OBJECT was not returning the correct
4114 length if passed a buffer. ASN1_INTEGER_to_BN failed
4115 if passed a NULL BN and its argument was negative.
4116 [Steve Henson, pointed out by Sven Heiberg <sven@tartu.cyber.ee>]
4118 *) Modification to PKCS#7 encoding routines to output definite
4119 length encoding. Since currently the whole structures are in
4120 memory there's not real point in using indefinite length
4121 constructed encoding. However if OpenSSL is compiled with
4122 the flag PKCS7_INDEFINITE_ENCODING the old form is used.
4125 *) Added BIO_vprintf() and BIO_vsnprintf().
4128 *) Added more prefixes to parse for in the the strings written
4129 through a logging bio, to cover all the levels that are available
4130 through syslog. The prefixes are now:
4132 PANIC, EMERG, EMR => LOG_EMERG
4133 ALERT, ALR => LOG_ALERT
4134 CRIT, CRI => LOG_CRIT
4135 ERROR, ERR => LOG_ERR
4136 WARNING, WARN, WAR => LOG_WARNING
4137 NOTICE, NOTE, NOT => LOG_NOTICE
4138 INFO, INF => LOG_INFO
4139 DEBUG, DBG => LOG_DEBUG
4141 and as before, if none of those prefixes are present at the
4142 beginning of the string, LOG_ERR is chosen.
4144 On Win32, the LOG_* levels are mapped according to this:
4146 LOG_EMERG, LOG_ALERT, LOG_CRIT, LOG_ERR => EVENTLOG_ERROR_TYPE
4147 LOG_WARNING => EVENTLOG_WARNING_TYPE
4148 LOG_NOTICE, LOG_INFO, LOG_DEBUG => EVENTLOG_INFORMATION_TYPE
4152 *) Made it possible to reconfigure with just the configuration
4153 argument "reconf" or "reconfigure". The command line arguments
4154 are stored in Makefile.ssl in the variable CONFIGURE_ARGS,
4155 and are retrieved from there when reconfiguring.
4159 [Assar Westerlund <assar@sics.se>, Richard Levitte]
4161 *) Add the arguments -CAfile and -CApath to the pkcs12 utility.
4164 *) The obj_dat.pl script was messing up the sorting of object
4165 names. The reason was that it compared the quoted version
4166 of strings as a result "OCSP" > "OCSP Signing" because
4167 " > SPACE. Changed script to store unquoted versions of
4168 names and add quotes on output. It was also omitting some
4169 names from the lookup table if they were given a default
4170 value (that is if SN is missing it is given the same
4171 value as LN and vice versa), these are now added on the
4172 grounds that if an object has a name we should be able to
4173 look it up. Finally added warning output when duplicate
4174 short or long names are found.
4177 *) Changes needed for Tandem NSK.
4178 [Scott Uroff <scott@xypro.com>]
4180 *) Fix SSL 2.0 rollback checking: Due to an off-by-one error in
4181 RSA_padding_check_SSLv23(), special padding was never detected
4182 and thus the SSL 3.0/TLS 1.0 countermeasure against protocol
4183 version rollback attacks was not effective.
4185 In s23_clnt.c, don't use special rollback-attack detection padding
4186 (RSA_SSLV23_PADDING) if SSL 2.0 is the only protocol enabled in the
4187 client; similarly, in s23_srvr.c, don't do the rollback check if
4188 SSL 2.0 is the only protocol enabled in the server.
4191 *) Make it possible to get hexdumps of unprintable data with 'openssl
4192 asn1parse'. By implication, the functions ASN1_parse_dump() and
4193 BIO_dump_indent() are added.
4196 *) New functions ASN1_STRING_print_ex() and X509_NAME_print_ex()
4197 these print out strings and name structures based on various
4198 flags including RFC2253 support and proper handling of
4199 multibyte characters. Added options to the 'x509' utility
4200 to allow the various flags to be set.
4203 *) Various fixes to use ASN1_TIME instead of ASN1_UTCTIME.
4204 Also change the functions X509_cmp_current_time() and
4205 X509_gmtime_adj() work with an ASN1_TIME structure,
4206 this will enable certificates using GeneralizedTime in validity
4207 dates to be checked.
4210 *) Make the NEG_PUBKEY_BUG code (which tolerates invalid
4211 negative public key encodings) on by default,
4212 NO_NEG_PUBKEY_BUG can be set to disable it.
4215 *) New function c2i_ASN1_OBJECT() which acts on ASN1_OBJECT
4216 content octets. An i2c_ASN1_OBJECT is unnecessary because
4217 the encoding can be trivially obtained from the structure.
4220 *) crypto/err.c locking bugfix: Use write locks (CRYPTO_w_[un]lock),
4221 not read locks (CRYPTO_r_[un]lock).
4224 *) A first attempt at creating official support for shared
4225 libraries through configuration. I've kept it so the
4226 default is static libraries only, and the OpenSSL programs
4227 are always statically linked for now, but there are