5 Changes between 0.9.7f and 0.9.8 [xx XXX xxxx]
7 *) Add support for DER encoded private keys (SSL_FILETYPE_ASN1)
8 to SSL_CTX_use_PrivateKey_file() and SSL_use_PrivateKey_file()
11 *) Remove buggy and incompletet DH cert support from
12 ssl/ssl_rsa.c and ssl/s3_both.c
15 *) Use SHA-1 instead of MD5 as the default digest algorithm for
16 the apps/openssl applications.
19 *) Compile clean with "-Wall -Wmissing-prototypes
20 -Wstrict-prototypes -Wmissing-declarations -Werror". Currently
21 DEBUG_SAFESTACK must also be set.
24 *) Change ./Configure so that certain algorithms can be disabled by default.
25 The new counterpiece to "no-xxx" is "enable-xxx".
27 The patented RC5 and MDC2 algorithms will now be disabled unless
28 "enable-rc5" and "enable-mdc2", respectively, are specified.
30 (IDEA remains enabled despite being patented. This is because IDEA
31 is frequently required for interoperability, and there is no license
32 fee for non-commercial use. As before, "no-idea" can be used to
33 avoid this algorithm.)
37 *) Add processing of proxy certificates (see RFC 3820). This work was
38 sponsored by KTH (The Royal Institute of Technology in Stockholm) and
39 EGEE (Enabling Grids for E-science in Europe).
42 *) RC4 performance overhaul on modern architectures/implementations, such
43 as Intel P4, IA-64 and AMD64.
46 *) New utility extract-section.pl. This can be used specify an alternative
47 section number in a pod file instead of having to treat each file as
48 a separate case in Makefile. This can be done by adding two lines to the
51 =for comment openssl_section:XXX
53 The blank line is mandatory.
57 *) New arguments -certform, -keyform and -pass for s_client and s_server
58 to allow alternative format key and certificate files and passphrase
62 *) New structure X509_VERIFY_PARAM which combines current verify parameters,
63 update associated structures and add various utility functions.
65 Add new policy related verify parameters, include policy checking in
66 standard verify code. Enhance 'smime' application with extra parameters
67 to support policy checking and print out.
70 *) Add a new engine to support VIA PadLock ACE extensions in the VIA C3
71 Nehemiah processors. These extensions support AES encryption in hardware
72 as well as RNG (though RNG support is currently disabled).
73 [Michal Ludvig <michal@logix.cz>, with help from Andy Polyakov]
75 *) Deprecate BN_[get|set]_params() functions (they were ignored internally).
78 *) New FIPS 180-2 algorithms, SHA-224/-256/-384/-512 are implemented.
79 [Andy Polyakov and a number of other people]
81 *) Improved PowerPC platform support. Most notably BIGNUM assembler
82 implementation contributed by IBM.
83 [Suresh Chari, Peter Waltenberg, Andy Polyakov]
85 *) The new 'RSA_generate_key_ex' function now takes a BIGNUM for the public
86 exponent rather than 'unsigned long'. There is a corresponding change to
87 the new 'rsa_keygen' element of the RSA_METHOD structure.
88 [Jelte Jansen, Geoff Thorpe]
90 *) Functionality for creating the initial serial number file is now
91 moved from CA.pl to the 'ca' utility with a new option -create_serial.
93 (Before OpenSSL 0.9.7e, CA.pl used to initialize the serial
94 number file to 1, which is bound to cause problems. To avoid
95 the problems while respecting compatibility between different 0.9.7
96 patchlevels, 0.9.7e employed 'openssl x509 -next_serial' in
97 CA.pl for serial number initialization. With the new release 0.9.8,
98 we can fix the problem directly in the 'ca' utility.)
101 *) Reduced header interdepencies by declaring more opaque objects in
102 ossl_typ.h. As a consequence, including some headers (eg. engine.h) will
103 give fewer recursive includes, which could break lazy source code - so
104 this change is covered by the OPENSSL_NO_DEPRECATED symbol. As always,
105 developers should define this symbol when building and using openssl to
106 ensure they track the recommended behaviour, interfaces, [etc], but
107 backwards-compatible behaviour prevails when this isn't defined.
110 *) New function X509_POLICY_NODE_print() which prints out policy nodes.
113 *) Add new EVP function EVP_CIPHER_CTX_rand_key and associated functionality.
114 This will generate a random key of the appropriate length based on the
115 cipher context. The EVP_CIPHER can provide its own random key generation
116 routine to support keys of a specific form. This is used in the des and
117 3des routines to generate a key of the correct parity. Update S/MIME
118 code to use new functions and hence generate correct parity DES keys.
119 Add EVP_CHECK_DES_KEY #define to return an error if the key is not
120 valid (weak or incorrect parity).
123 *) Add a local set of CRLs that can be used by X509_verify_cert() as well
124 as looking them up. This is useful when the verified structure may contain
125 CRLs, for example PKCS#7 signedData. Modify PKCS7_verify() to use any CRLs
126 present unless the new PKCS7_NO_CRL flag is asserted.
129 *) Extend ASN1 oid configuration module. It now additionally accepts the
132 shortName = some long name, 1.2.3.4
135 *) Reimplemented the BN_CTX implementation. There is now no more static
136 limitation on the number of variables it can handle nor the depth of the
137 "stack" handling for BN_CTX_start()/BN_CTX_end() pairs. The stack
138 information can now expand as required, and rather than having a single
139 static array of bignums, BN_CTX now uses a linked-list of such arrays
140 allowing it to expand on demand whilst maintaining the usefulness of
144 *) Add a missing BN_CTX parameter to the 'rsa_mod_exp' callback in RSA_METHOD
145 to allow all RSA operations to function using a single BN_CTX.
148 *) Preliminary support for certificate policy evaluation and checking. This
149 is initially intended to pass the tests outlined in "Conformance Testing
150 of Relying Party Client Certificate Path Processing Logic" v1.07.
153 *) bn_dup_expand() has been deprecated, it was introduced in 0.9.7 and
154 remained unused and not that useful. A variety of other little bignum
155 tweaks and fixes have also been made continuing on from the audit (see
159 *) Constify all or almost all d2i, c2i, s2i and r2i functions, along with
160 associated ASN1, EVP and SSL functions and old ASN1 macros.
163 *) BN_zero() only needs to set 'top' and 'neg' to zero for correct results,
164 and this should never fail. So the return value from the use of
165 BN_set_word() (which can fail due to needless expansion) is now deprecated;
166 if OPENSSL_NO_DEPRECATED is defined, BN_zero() is a void macro.
169 *) BN_CTX_get() should return zero-valued bignums, providing the same
170 initialised value as BN_new().
171 [Geoff Thorpe, suggested by Ulf Möller]
173 *) Support for inhibitAnyPolicy certificate extension.
176 *) An audit of the BIGNUM code is underway, for which debugging code is
177 enabled when BN_DEBUG is defined. This makes stricter enforcements on what
178 is considered valid when processing BIGNUMs, and causes execution to
179 assert() when a problem is discovered. If BN_DEBUG_RAND is defined,
180 further steps are taken to deliberately pollute unused data in BIGNUM
181 structures to try and expose faulty code further on. For now, openssl will
182 (in its default mode of operation) continue to tolerate the inconsistent
183 forms that it has tolerated in the past, but authors and packagers should
184 consider trying openssl and their own applications when compiled with
185 these debugging symbols defined. It will help highlight potential bugs in
186 their own code, and will improve the test coverage for OpenSSL itself. At
187 some point, these tighter rules will become openssl's default to improve
188 maintainability, though the assert()s and other overheads will remain only
189 in debugging configurations. See bn.h for more details.
190 [Geoff Thorpe, Nils Larsch, Ulf Möller]
192 *) BN_CTX_init() has been deprecated, as BN_CTX is an opaque structure
193 that can only be obtained through BN_CTX_new() (which implicitly
194 initialises it). The presence of this function only made it possible
195 to overwrite an existing structure (and cause memory leaks).
198 *) Because of the callback-based approach for implementing LHASH as a
199 template type, lh_insert() adds opaque objects to hash-tables and
200 lh_doall() or lh_doall_arg() are typically used with a destructor callback
201 to clean up those corresponding objects before destroying the hash table
202 (and losing the object pointers). So some over-zealous constifications in
203 LHASH have been relaxed so that lh_insert() does not take (nor store) the
204 objects as "const" and the lh_doall[_arg] callback wrappers are not
205 prototyped to have "const" restrictions on the object pointers they are
206 given (and so aren't required to cast them away any more).
209 *) The tmdiff.h API was so ugly and minimal that our own timing utility
210 (speed) prefers to use its own implementation. The two implementations
211 haven't been consolidated as yet (volunteers?) but the tmdiff API has had
212 its object type properly exposed (MS_TM) instead of casting to/from "char
213 *". This may still change yet if someone realises MS_TM and "ms_time_***"
214 aren't necessarily the greatest nomenclatures - but this is what was used
215 internally to the implementation so I've used that for now.
218 *) Ensure that deprecated functions do not get compiled when
219 OPENSSL_NO_DEPRECATED is defined. Some "openssl" subcommands and a few of
220 the self-tests were still using deprecated key-generation functions so
221 these have been updated also.
224 *) Reorganise PKCS#7 code to separate the digest location functionality
225 into PKCS7_find_digest(), digest addtion into PKCS7_bio_add_digest().
226 New function PKCS7_set_digest() to set the digest type for PKCS#7
227 digestedData type. Add additional code to correctly generate the
228 digestedData type and add support for this type in PKCS7 initialization
232 *) New function PKCS7_set0_type_other() this initializes a PKCS7
233 structure of type "other".
236 *) Fix prime generation loop in crypto/bn/bn_prime.pl by making
237 sure the loop does correctly stop and breaking ("division by zero")
238 modulus operations are not performed. The (pre-generated) prime
239 table crypto/bn/bn_prime.h was already correct, but it could not be
240 re-generated on some platforms because of the "division by zero"
241 situation in the script.
242 [Ralf S. Engelschall]
244 *) Update support for ECC-based TLS ciphersuites according to
245 draft-ietf-tls-ecc-03.txt: the KDF1 key derivation function with
246 SHA-1 now is only used for "small" curves (where the
247 representation of a field element takes up to 24 bytes); for
248 larger curves, the field element resulting from ECDH is directly
249 used as premaster secret.
250 [Douglas Stebila (Sun Microsystems Laboratories)]
252 *) Add code for kP+lQ timings to crypto/ec/ectest.c, and add SEC2
253 curve secp160r1 to the tests.
254 [Douglas Stebila (Sun Microsystems Laboratories)]
256 *) Add the possibility to load symbols globally with DSO.
257 [Götz Babin-Ebell <babin-ebell@trustcenter.de> via Richard Levitte]
259 *) Add the functions ERR_set_mark() and ERR_pop_to_mark() for better
260 control of the error stack.
263 *) Add support for STORE in ENGINE.
266 *) Add the STORE type. The intention is to provide a common interface
267 to certificate and key stores, be they simple file-based stores, or
268 HSM-type store, or LDAP stores, or...
269 NOTE: The code is currently UNTESTED and isn't really used anywhere.
272 *) Add a generic structure called OPENSSL_ITEM. This can be used to
273 pass a list of arguments to any function as well as provide a way
274 for a function to pass data back to the caller.
277 *) Add the functions BUF_strndup() and BUF_memdup(). BUF_strndup()
278 works like BUF_strdup() but can be used to duplicate a portion of
279 a string. The copy gets NUL-terminated. BUF_memdup() duplicates
283 *) Add the function sk_find_ex() which works like sk_find(), but will
284 return an index to an element even if an exact match couldn't be
285 found. The index is guaranteed to point at the element where the
286 searched-for key would be inserted to preserve sorting order.
289 *) Add the function OBJ_bsearch_ex() which works like OBJ_bsearch() but
290 takes an extra flags argument for optional functionality. Currently,
291 the following flags are defined:
293 OBJ_BSEARCH_VALUE_ON_NOMATCH
294 This one gets OBJ_bsearch_ex() to return a pointer to the first
295 element where the comparing function returns a negative or zero
298 OBJ_BSEARCH_FIRST_VALUE_ON_MATCH
299 This one gets OBJ_bsearch_ex() to return a pointer to the first
300 element where the comparing function returns zero. This is useful
301 if there are more than one element where the comparing function
305 *) Make it possible to create self-signed certificates with 'openssl ca'
306 in such a way that the self-signed certificate becomes part of the
307 CA database and uses the same mechanisms for serial number generation
308 as all other certificate signing. The new flag '-selfsign' enables
309 this functionality. Adapt CA.sh and CA.pl.in.
312 *) Add functionality to check the public key of a certificate request
313 against a given private. This is useful to check that a certificate
314 request can be signed by that key (self-signing).
317 *) Make it possible to have multiple active certificates with the same
318 subject in the CA index file. This is done only if the keyword
319 'unique_subject' is set to 'no' in the main CA section (default
320 if 'CA_default') of the configuration file. The value is saved
321 with the database itself in a separate index attribute file,
322 named like the index file with '.attr' appended to the name.
325 *) Generate muti valued AVAs using '+' notation in config files for
329 *) Support for nameConstraints certificate extension.
332 *) Support for policyConstraints certificate extension.
335 *) Support for policyMappings certificate extension.
338 *) Make sure the default DSA_METHOD implementation only uses its
339 dsa_mod_exp() and/or bn_mod_exp() handlers if they are non-NULL,
340 and change its own handlers to be NULL so as to remove unnecessary
341 indirection. This lets alternative implementations fallback to the
342 default implementation more easily.
345 *) Support for directoryName in GeneralName related extensions
349 *) Make it possible to link applications using Makefile.shared.
350 Make that possible even when linking against static libraries!
353 *) Support for single pass processing for S/MIME signing. This now
354 means that S/MIME signing can be done from a pipe, in addition
355 cleartext signing (multipart/signed type) is effectively streaming
356 and the signed data does not need to be all held in memory.
358 This is done with a new flag PKCS7_STREAM. When this flag is set
359 PKCS7_sign() only initializes the PKCS7 structure and the actual signing
360 is done after the data is output (and digests calculated) in
364 *) Add full support for -rpath/-R, both in shared libraries and
365 applications, at least on the platforms where it's known how
369 *) In crypto/ec/ec_mult.c, implement fast point multiplication with
370 precomputation, based on wNAF splitting: EC_GROUP_precompute_mult()
371 will now compute a table of multiples of the generator that
372 makes subsequent invocations of EC_POINTs_mul() or EC_POINT_mul()
373 faster (notably in the case of a single point multiplication,
375 [Nils Larsch, Bodo Moeller]
377 *) IPv6 support for certificate extensions. The various extensions
378 which use the IP:a.b.c.d can now take IPv6 addresses using the
379 formats of RFC1884 2.2 . IPv6 addresses are now also displayed
383 *) Added an ENGINE that implements RSA by performing private key
384 exponentiations with the GMP library. The conversions to and from
385 GMP's mpz_t format aren't optimised nor are any montgomery forms
386 cached, and on x86 it appears OpenSSL's own performance has caught up.
387 However there are likely to be other architectures where GMP could
388 provide a boost. This ENGINE is not built in by default, but it can be
389 specified at Configure time and should be accompanied by the necessary
390 linker additions, eg;
391 ./config -DOPENSSL_USE_GMP -lgmp
394 *) "openssl engine" will not display ENGINE/DSO load failure errors when
395 testing availability of engines with "-t" - the old behaviour is
396 produced by increasing the feature's verbosity with "-tt".
399 *) ECDSA routines: under certain error conditions uninitialized BN objects
400 could be freed. Solution: make sure initialization is performed early
401 enough. (Reported and fix supplied by Nils Larsch <nla@trustcenter.de>
405 *) Key-generation can now be implemented in RSA_METHOD, DSA_METHOD
406 and DH_METHOD (eg. by ENGINE implementations) to override the normal
407 software implementations. For DSA and DH, parameter generation can
408 also be overriden by providing the appropriate method callbacks.
411 *) Change the "progress" mechanism used in key-generation and
412 primality testing to functions that take a new BN_GENCB pointer in
413 place of callback/argument pairs. The new API functions have "_ex"
414 postfixes and the older functions are reimplemented as wrappers for
415 the new ones. The OPENSSL_NO_DEPRECATED symbol can be used to hide
416 declarations of the old functions to help (graceful) attempts to
417 migrate to the new functions. Also, the new key-generation API
418 functions operate on a caller-supplied key-structure and return
419 success/failure rather than returning a key or NULL - this is to
420 help make "keygen" another member function of RSA_METHOD etc.
422 Example for using the new callback interface:
424 int (*my_callback)(int a, int b, BN_GENCB *cb) = ...;
428 BN_GENCB_set(&my_cb, my_callback, my_arg);
430 return BN_is_prime_ex(some_bignum, BN_prime_checks, NULL, &cb);
431 /* For the meaning of a, b in calls to my_callback(), see the
432 * documentation of the function that calls the callback.
433 * cb will point to my_cb; my_arg can be retrieved as cb->arg.
434 * my_callback should return 1 if it wants BN_is_prime_ex()
435 * to continue, or 0 to stop.
440 *) Change the ZLIB compression method to be stateful, and make it
441 available to TLS with the number defined in
442 draft-ietf-tls-compression-04.txt.
445 *) Add the ASN.1 structures and functions for CertificatePair, which
446 is defined as follows (according to X.509_4thEditionDraftV6.pdf):
448 CertificatePair ::= SEQUENCE {
449 forward [0] Certificate OPTIONAL,
450 reverse [1] Certificate OPTIONAL,
451 -- at least one of the pair shall be present -- }
453 Also implement the PEM functions to read and write certificate
454 pairs, and defined the PEM tag as "CERTIFICATE PAIR".
456 This needed to be defined, mostly for the sake of the LDAP
457 attribute crossCertificatePair, but may prove useful elsewhere as
461 *) Make it possible to inhibit symlinking of shared libraries in
462 Makefile.shared, for Cygwin's sake.
465 *) Extend the BIGNUM API by creating new macros that behave like
468 void BN_set_sign(BIGNUM *a, int neg);
469 int BN_get_sign(const BIGNUM *a);
471 and avoid the need to access 'a->neg' directly in applications.
472 [Nils Larsch <nla@trustcenter.de>]
474 *) Implement fast modular reduction for pseudo-Mersenne primes
475 used in NIST curves (crypto/bn/bn_nist.c, crypto/ec/ecp_nist.c).
476 EC_GROUP_new_curve_GFp() will now automatically use this
478 [Nils Larsch <nla@trustcenter.de>]
480 *) Add new lock type (CRYPTO_LOCK_BN).
483 *) Change the ENGINE framework to automatically load engines
484 dynamically from specific directories unless they could be
485 found to already be built in or loaded. Move all the
486 current engines except for the cryptodev one to a new
488 The engines in engines/ are built as shared libraries if
489 the "shared" options was given to ./Configure or ./config.
490 Otherwise, they are inserted in libcrypto.a.
491 /usr/local/ssl/engines is the default directory for dynamic
492 engines, but that can be overriden at configure time through
493 the usual use of --prefix and/or --openssldir, and at run
494 time with the environment variable OPENSSL_ENGINES.
495 [Geoff Thorpe and Richard Levitte]
497 *) Add Makefile.shared, a helper makefile to build shared
498 libraries. Addapt Makefile.org.
501 *) Add version info to Win32 DLLs.
502 [Peter 'Luna' Runestig" <peter@runestig.com>]
504 *) Add new 'medium level' PKCS#12 API. Certificates and keys
505 can be added using this API to created arbitrary PKCS#12
506 files while avoiding the low level API.
508 New options to PKCS12_create(), key or cert can be NULL and
509 will then be omitted from the output file. The encryption
510 algorithm NIDs can be set to -1 for no encryption, the mac
511 iteration count can be set to 0 to omit the mac.
513 Enhance pkcs12 utility by making the -nokeys and -nocerts
514 options work when creating a PKCS#12 file. New option -nomac
515 to omit the mac, NONE can be set for an encryption algorithm.
516 New code is modified to use the enhanced PKCS12_create()
517 instead of the low level API.
520 *) Extend ASN1 encoder to support indefinite length constructed
521 encoding. This can output sequences tags and octet strings in
522 this form. Modify pk7_asn1.c to support indefinite length
523 encoding. This is experimental and needs additional code to
524 be useful, such as an ASN1 bio and some enhanced streaming
527 Extend template encode functionality so that tagging is passed
528 down to the template encoder.
531 *) Let 'openssl req' fail if an argument to '-newkey' is not
532 recognized instead of using RSA as a default.
535 *) Add support for ECC-based ciphersuites from draft-ietf-tls-ecc-01.txt.
536 As these are not official, they are not included in "ALL";
537 the "ECCdraft" ciphersuite group alias can be used to select them.
538 [Vipul Gupta and Sumit Gupta (Sun Microsystems Laboratories)]
540 *) Add ECDH engine support.
541 [Nils Gura and Douglas Stebila (Sun Microsystems Laboratories)]
543 *) Add ECDH in new directory crypto/ecdh/.
544 [Douglas Stebila (Sun Microsystems Laboratories)]
546 *) Let BN_rand_range() abort with an error after 100 iterations
547 without success (which indicates a broken PRNG).
550 *) Change BN_mod_sqrt() so that it verifies that the input value
551 is really the square of the return value. (Previously,
552 BN_mod_sqrt would show GIGO behaviour.)
555 *) Add named elliptic curves over binary fields from X9.62, SECG,
556 and WAP/WTLS; add OIDs that were still missing.
558 [Sheueling Chang Shantz and Douglas Stebila
559 (Sun Microsystems Laboratories)]
561 *) Extend the EC library for elliptic curves over binary fields
562 (new files ec2_smpl.c, ec2_smpt.c, ec2_mult.c in crypto/ec/).
565 EC_GF2m_simple_method
569 EC_GROUP_new_curve_GF2m
570 EC_GROUP_set_curve_GF2m
571 EC_GROUP_get_curve_GF2m
572 EC_POINT_set_affine_coordinates_GF2m
573 EC_POINT_get_affine_coordinates_GF2m
574 EC_POINT_set_compressed_coordinates_GF2m
576 Point compression for binary fields is disabled by default for
577 patent reasons (compile with OPENSSL_EC_BIN_PT_COMP defined to
580 As binary polynomials are represented as BIGNUMs, various members
581 of the EC_GROUP and EC_POINT data structures can be shared
582 between the implementations for prime fields and binary fields;
583 the above ..._GF2m functions (except for EX_GROUP_new_curve_GF2m)
584 are essentially identical to their ..._GFp counterparts.
585 (For simplicity, the '..._GFp' prefix has been dropped from
586 various internal method names.)
588 An internal 'field_div' method (similar to 'field_mul' and
589 'field_sqr') has been added; this is used only for binary fields.
591 [Sheueling Chang Shantz and Douglas Stebila
592 (Sun Microsystems Laboratories)]
594 *) Optionally dispatch EC_POINT_mul(), EC_POINT_precompute_mult()
595 through methods ('mul', 'precompute_mult').
597 The generic implementations (now internally called 'ec_wNAF_mul'
598 and 'ec_wNAF_precomputed_mult') remain the default if these
599 methods are undefined.
601 [Sheueling Chang Shantz and Douglas Stebila
602 (Sun Microsystems Laboratories)]
604 *) New function EC_GROUP_get_degree, which is defined through
605 EC_METHOD. For curves over prime fields, this returns the bit
606 length of the modulus.
608 [Sheueling Chang Shantz and Douglas Stebila
609 (Sun Microsystems Laboratories)]
611 *) New functions EC_GROUP_dup, EC_POINT_dup.
612 (These simply call ..._new and ..._copy).
614 [Sheueling Chang Shantz and Douglas Stebila
615 (Sun Microsystems Laboratories)]
617 *) Add binary polynomial arithmetic software in crypto/bn/bn_gf2m.c.
618 Polynomials are represented as BIGNUMs (where the sign bit is not
619 used) in the following functions [macros]:
622 BN_GF2m_sub [= BN_GF2m_add]
623 BN_GF2m_mod [wrapper for BN_GF2m_mod_arr]
624 BN_GF2m_mod_mul [wrapper for BN_GF2m_mod_mul_arr]
625 BN_GF2m_mod_sqr [wrapper for BN_GF2m_mod_sqr_arr]
627 BN_GF2m_mod_exp [wrapper for BN_GF2m_mod_exp_arr]
628 BN_GF2m_mod_sqrt [wrapper for BN_GF2m_mod_sqrt_arr]
629 BN_GF2m_mod_solve_quad [wrapper for BN_GF2m_mod_solve_quad_arr]
630 BN_GF2m_cmp [= BN_ucmp]
632 (Note that only the 'mod' functions are actually for fields GF(2^m).
633 BN_GF2m_add() is misnomer, but this is for the sake of consistency.)
635 For some functions, an the irreducible polynomial defining a
636 field can be given as an 'unsigned int[]' with strictly
637 decreasing elements giving the indices of those bits that are set;
638 i.e., p[] represents the polynomial
639 f(t) = t^p[0] + t^p[1] + ... + t^p[k]
641 p[0] > p[1] > ... > p[k] = 0.
642 This applies to the following functions:
647 BN_GF2m_mod_inv_arr [wrapper for BN_GF2m_mod_inv]
648 BN_GF2m_mod_div_arr [wrapper for BN_GF2m_mod_div]
651 BN_GF2m_mod_solve_quad_arr
655 Conversion can be performed by the following functions:
660 bntest.c has additional tests for binary polynomial arithmetic.
662 Two implementations for BN_GF2m_mod_div() are available.
663 The default algorithm simply uses BN_GF2m_mod_inv() and
664 BN_GF2m_mod_mul(). The alternative algorithm is compiled in only
665 if OPENSSL_SUN_GF2M_DIV is defined (patent pending; read the
666 copyright notice in crypto/bn/bn_gf2m.c before enabling it).
668 [Sheueling Chang Shantz and Douglas Stebila
669 (Sun Microsystems Laboratories)]
671 *) Add new error code 'ERR_R_DISABLED' that can be used when some
672 functionality is disabled at compile-time.
673 [Douglas Stebila <douglas.stebila@sun.com>]
675 *) Change default behaviour of 'openssl asn1parse' so that more
676 information is visible when viewing, e.g., a certificate:
678 Modify asn1_parse2 (crypto/asn1/asn1_par.c) so that in non-'dump'
679 mode the content of non-printable OCTET STRINGs is output in a
680 style similar to INTEGERs, but with '[HEX DUMP]' prepended to
681 avoid the appearance of a printable string.
682 [Nils Larsch <nla@trustcenter.de>]
684 *) Add 'asn1_flag' and 'asn1_form' member to EC_GROUP with access
686 EC_GROUP_set_asn1_flag()
687 EC_GROUP_get_asn1_flag()
688 EC_GROUP_set_point_conversion_form()
689 EC_GROUP_get_point_conversion_form()
690 These control ASN1 encoding details:
691 - Curves (i.e., groups) are encoded explicitly unless asn1_flag
692 has been set to OPENSSL_EC_NAMED_CURVE.
693 - Points are encoded in uncompressed form by default; options for
694 asn1_for are as for point2oct, namely
695 POINT_CONVERSION_COMPRESSED
696 POINT_CONVERSION_UNCOMPRESSED
697 POINT_CONVERSION_HYBRID
699 Also add 'seed' and 'seed_len' members to EC_GROUP with access
703 EC_GROUP_get_seed_len()
704 This is used only for ASN1 purposes (so far).
705 [Nils Larsch <nla@trustcenter.de>]
707 *) Add 'field_type' member to EC_METHOD, which holds the NID
708 of the appropriate field type OID. The new function
709 EC_METHOD_get_field_type() returns this value.
710 [Nils Larsch <nla@trustcenter.de>]
717 providing useful interfaces to EC_POINT_point2oct() and
718 EC_POINT_oct2point().
719 [Nils Larsch <nla@trustcenter.de>]
721 *) Change internals of the EC library so that the functions
722 EC_GROUP_set_generator()
723 EC_GROUP_get_generator()
725 EC_GROUP_get_cofactor()
726 are implemented directly in crypto/ec/ec_lib.c and not dispatched
727 to methods, which would lead to unnecessary code duplication when
728 adding different types of curves.
729 [Nils Larsch <nla@trustcenter.de> with input by Bodo Moeller]
731 *) Implement compute_wNAF (crypto/ec/ec_mult.c) without BIGNUM
732 arithmetic, and such that modified wNAFs are generated
733 (which avoid length expansion in many cases).
736 *) Add a function EC_GROUP_check_discriminant() (defined via
737 EC_METHOD) that verifies that the curve discriminant is non-zero.
739 Add a function EC_GROUP_check() that makes some sanity tests
740 on a EC_GROUP, its generator and order. This includes
741 EC_GROUP_check_discriminant().
742 [Nils Larsch <nla@trustcenter.de>]
744 *) Add ECDSA in new directory crypto/ecdsa/.
746 Add applications 'openssl ecparam' and 'openssl ecdsa'
747 (these are based on 'openssl dsaparam' and 'openssl dsa').
749 ECDSA support is also included in various other files across the
750 library. Most notably,
751 - 'openssl req' now has a '-newkey ecdsa:file' option;
752 - EVP_PKCS82PKEY (crypto/evp/evp_pkey.c) now can handle ECDSA;
753 - X509_PUBKEY_get (crypto/asn1/x_pubkey.c) and
754 d2i_PublicKey (crypto/asn1/d2i_pu.c) have been modified to make
755 them suitable for ECDSA where domain parameters must be
756 extracted before the specific public key;
757 - ECDSA engine support has been added.
758 [Nils Larsch <nla@trustcenter.de>]
760 *) Include some named elliptic curves, and add OIDs from X9.62,
761 SECG, and WAP/WTLS. Each curve can be obtained from the new
763 EC_GROUP_new_by_nid(),
764 and the list of available named curves can be obtained with
765 EC_get_builtin_curves().
766 Also add a 'curve_name' member to EC_GROUP objects, which can be
770 [Nils Larsch <nla@trustcenter.de, Bodo Moeller]
772 *) Remove a few calls to bn_wexpand() in BN_sqr() (the one in there
773 was actually never needed) and in BN_mul(). The removal in BN_mul()
774 required a small change in bn_mul_part_recursive() and the addition
775 of the functions bn_cmp_part_words(), bn_sub_part_words() and
776 bn_add_part_words(), which do the same thing as bn_cmp_words(),
777 bn_sub_words() and bn_add_words() except they take arrays with
781 Changes between 0.9.7f and 0.9.7g [XX xxx xxxx]
783 *) Undo Cygwin change.
786 Changes between 0.9.7e and 0.9.7f [22 Mar 2005]
788 *) Use (SSL_RANDOM_VALUE - 4) bytes of pseudo random data when generating
789 server and client random values. Previously
790 (SSL_RANDOM_VALUE - sizeof(time_t)) would be used which would result in
791 less random data when sizeof(time_t) > 4 (some 64 bit platforms).
793 This change has negligible security impact because:
795 1. Server and client random values still have 24 bytes of pseudo random
798 2. Server and client random values are sent in the clear in the initial
801 3. The master secret is derived using the premaster secret (48 bytes in
802 size for static RSA ciphersuites) as well as client server and random
805 The OpenSSL team would like to thank the UK NISCC for bringing this issue
808 [Stephen Henson, reported by UK NISCC]
810 *) Use Windows randomness collection on Cygwin.
813 *) Fix hang in EGD/PRNGD query when communication socket is closed
814 prematurely by EGD/PRNGD.
815 [Darren Tucker <dtucker@zip.com.au> via Lutz Jänicke, resolves #1014]
817 *) Prompt for pass phrases when appropriate for PKCS12 input format.
820 *) Back-port of selected performance improvements from development
821 branch, as well as improved support for PowerPC platforms.
824 *) Add lots of checks for memory allocation failure, error codes to indicate
825 failure and freeing up memory if a failure occurs.
826 [Nauticus Networks SSL Team <openssl@nauticusnet.com>, Steve Henson]
828 *) Add new -passin argument to dgst.
831 *) Perform some character comparisons of different types in X509_NAME_cmp:
832 this is needed for some certificates that reencode DNs into UTF8Strings
833 (in violation of RFC3280) and can't or wont issue name rollover
837 *) Make an explicit check during certificate validation to see that
838 the CA setting in each certificate on the chain is correct. As a
839 side effect always do the following basic checks on extensions,
840 not just when there's an associated purpose to the check:
842 - if there is an unhandled critical extension (unless the user
843 has chosen to ignore this fault)
844 - if the path length has been exceeded (if one is set at all)
845 - that certain extensions fit the associated purpose (if one has
849 Changes between 0.9.7d and 0.9.7e [25 Oct 2004]
851 *) Avoid a race condition when CRLs are checked in a multi threaded
852 environment. This would happen due to the reordering of the revoked
853 entries during signature checking and serial number lookup. Now the
854 encoding is cached and the serial number sort performed under a lock.
855 Add new STACK function sk_is_sorted().
858 *) Add Delta CRL to the extension code.
861 *) Various fixes to s3_pkt.c so alerts are sent properly.
862 [David Holmes <d.holmes@f5.com>]
864 *) Reduce the chances of duplicate issuer name and serial numbers (in
865 violation of RFC3280) using the OpenSSL certificate creation utilities.
866 This is done by creating a random 64 bit value for the initial serial
867 number when a serial number file is created or when a self signed
868 certificate is created using 'openssl req -x509'. The initial serial
869 number file is created using 'openssl x509 -next_serial' in CA.pl
870 rather than being initialized to 1.
873 Changes between 0.9.7c and 0.9.7d [17 Mar 2004]
875 *) Fix null-pointer assignment in do_change_cipher_spec() revealed
876 by using the Codenomicon TLS Test Tool (CAN-2004-0079)
877 [Joe Orton, Steve Henson]
879 *) Fix flaw in SSL/TLS handshaking when using Kerberos ciphersuites
881 [Joe Orton, Steve Henson]
883 *) Make it possible to have multiple active certificates with the same
884 subject in the CA index file. This is done only if the keyword
885 'unique_subject' is set to 'no' in the main CA section (default
886 if 'CA_default') of the configuration file. The value is saved
887 with the database itself in a separate index attribute file,
888 named like the index file with '.attr' appended to the name.
891 *) X509 verify fixes. Disable broken certificate workarounds when
892 X509_V_FLAGS_X509_STRICT is set. Check CRL issuer has cRLSign set if
893 keyUsage extension present. Don't accept CRLs with unhandled critical
894 extensions: since verify currently doesn't process CRL extensions this
895 rejects a CRL with *any* critical extensions. Add new verify error codes
899 *) When creating an OCSP nonce use an OCTET STRING inside the extnValue.
900 A clarification of RFC2560 will require the use of OCTET STRINGs and
901 some implementations cannot handle the current raw format. Since OpenSSL
902 copies and compares OCSP nonces as opaque blobs without any attempt at
903 parsing them this should not create any compatibility issues.
906 *) New md flag EVP_MD_CTX_FLAG_REUSE this allows md_data to be reused when
907 calling EVP_MD_CTX_copy_ex() to avoid calling OPENSSL_malloc(). Without
908 this HMAC (and other) operations are several times slower than OpenSSL
912 *) Print out GeneralizedTime and UTCTime in ASN1_STRING_print_ex().
913 [Peter Sylvester <Peter.Sylvester@EdelWeb.fr>]
915 *) Use the correct content when signing type "other".
918 Changes between 0.9.7b and 0.9.7c [30 Sep 2003]
920 *) Fix various bugs revealed by running the NISCC test suite:
922 Stop out of bounds reads in the ASN1 code when presented with
923 invalid tags (CAN-2003-0543 and CAN-2003-0544).
925 Free up ASN1_TYPE correctly if ANY type is invalid (CAN-2003-0545).
927 If verify callback ignores invalid public key errors don't try to check
928 certificate signature with the NULL public key.
932 *) New -ignore_err option in ocsp application to stop the server
933 exiting on the first error in a request.
936 *) In ssl3_accept() (ssl/s3_srvr.c) only accept a client certificate
937 if the server requested one: as stated in TLS 1.0 and SSL 3.0
941 *) In ssl3_get_client_hello() (ssl/s3_srvr.c), tolerate additional
942 extra data after the compression methods not only for TLS 1.0
943 but also for SSL 3.0 (as required by the specification).
944 [Bodo Moeller; problem pointed out by Matthias Loepfe]
946 *) Change X509_certificate_type() to mark the key as exported/exportable
947 when it's 512 *bits* long, not 512 bytes.
950 *) Change AES_cbc_encrypt() so it outputs exact multiple of
951 blocks during encryption.
954 *) Various fixes to base64 BIO and non blocking I/O. On write
955 flushes were not handled properly if the BIO retried. On read
956 data was not being buffered properly and had various logic bugs.
957 This also affects blocking I/O when the data being decoded is a
961 *) Various S/MIME bugfixes and compatibility changes:
962 output correct application/pkcs7 MIME type if
963 PKCS7_NOOLDMIMETYPE is set. Tolerate some broken signatures.
964 Output CR+LF for EOL if PKCS7_CRLFEOL is set (this makes opening
965 of files as .eml work). Correctly handle very long lines in MIME
969 Changes between 0.9.7a and 0.9.7b [10 Apr 2003]
971 *) Countermeasure against the Klima-Pokorny-Rosa extension of
972 Bleichbacher's attack on PKCS #1 v1.5 padding: treat
973 a protocol version number mismatch like a decryption error
974 in ssl3_get_client_key_exchange (ssl/s3_srvr.c).
977 *) Turn on RSA blinding by default in the default implementation
978 to avoid a timing attack. Applications that don't want it can call
979 RSA_blinding_off() or use the new flag RSA_FLAG_NO_BLINDING.
980 They would be ill-advised to do so in most cases.
981 [Ben Laurie, Steve Henson, Geoff Thorpe, Bodo Moeller]
983 *) Change RSA blinding code so that it works when the PRNG is not
984 seeded (in this case, the secret RSA exponent is abused as
985 an unpredictable seed -- if it is not unpredictable, there
986 is no point in blinding anyway). Make RSA blinding thread-safe
987 by remembering the creator's thread ID in rsa->blinding and
988 having all other threads use local one-time blinding factors
989 (this requires more computation than sharing rsa->blinding, but
990 avoids excessive locking; and if an RSA object is not shared
991 between threads, blinding will still be very fast).
994 *) Fixed a typo bug that would cause ENGINE_set_default() to set an
995 ENGINE as defaults for all supported algorithms irrespective of
996 the 'flags' parameter. 'flags' is now honoured, so applications
997 should make sure they are passing it correctly.
1000 *) Target "mingw" now allows native Windows code to be generated in
1001 the Cygwin environment as well as with the MinGW compiler.
1004 Changes between 0.9.7 and 0.9.7a [19 Feb 2003]
1006 *) In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked
1007 via timing by performing a MAC computation even if incorrrect
1008 block cipher padding has been found. This is a countermeasure
1009 against active attacks where the attacker has to distinguish
1010 between bad padding and a MAC verification error. (CAN-2003-0078)
1012 [Bodo Moeller; problem pointed out by Brice Canvel (EPFL),
1013 Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and
1014 Martin Vuagnoux (EPFL, Ilion)]
1016 *) Make the no-err option work as intended. The intention with no-err
1017 is not to have the whole error stack handling routines removed from
1018 libcrypto, it's only intended to remove all the function name and
1019 reason texts, thereby removing some of the footprint that may not
1020 be interesting if those errors aren't displayed anyway.
1022 NOTE: it's still possible for any application or module to have it's
1023 own set of error texts inserted. The routines are there, just not
1024 used by default when no-err is given.
1027 *) Add support for FreeBSD on IA64.
1028 [dirk.meyer@dinoex.sub.org via Richard Levitte, resolves #454]
1030 *) Adjust DES_cbc_cksum() so it returns the same value as the MIT
1031 Kerberos function mit_des_cbc_cksum(). Before this change,
1032 the value returned by DES_cbc_cksum() was like the one from
1033 mit_des_cbc_cksum(), except the bytes were swapped.
1034 [Kevin Greaney <Kevin.Greaney@hp.com> and Richard Levitte]
1036 *) Allow an application to disable the automatic SSL chain building.
1037 Before this a rather primitive chain build was always performed in
1038 ssl3_output_cert_chain(): an application had no way to send the
1039 correct chain if the automatic operation produced an incorrect result.
1041 Now the chain builder is disabled if either:
1043 1. Extra certificates are added via SSL_CTX_add_extra_chain_cert().
1045 2. The mode flag SSL_MODE_NO_AUTO_CHAIN is set.
1047 The reasoning behind this is that an application would not want the
1048 auto chain building to take place if extra chain certificates are
1049 present and it might also want a means of sending no additional
1050 certificates (for example the chain has two certificates and the
1054 *) Add the possibility to build without the ENGINE framework.
1055 [Steven Reddie <smr@essemer.com.au> via Richard Levitte]
1057 *) Under Win32 gmtime() can return NULL: check return value in
1058 OPENSSL_gmtime(). Add error code for case where gmtime() fails.
1061 *) DSA routines: under certain error conditions uninitialized BN objects
1062 could be freed. Solution: make sure initialization is performed early
1063 enough. (Reported and fix supplied by Ivan D Nestlerode <nestler@MIT.EDU>,
1064 Nils Larsch <nla@trustcenter.de> via PR#459)
1067 *) Another fix for SSLv2 session ID handling: the session ID was incorrectly
1068 checked on reconnect on the client side, therefore session resumption
1069 could still fail with a "ssl session id is different" error. This
1070 behaviour is masked when SSL_OP_ALL is used due to
1071 SSL_OP_MICROSOFT_SESS_ID_BUG being set.
1072 Behaviour observed by Crispin Flowerday <crispin@flowerday.cx> as
1073 followup to PR #377.
1076 *) IA-32 assembler support enhancements: unified ELF targets, support
1077 for SCO/Caldera platforms, fix for Cygwin shared build.
1080 *) Add support for FreeBSD on sparc64. As a consequence, support for
1081 FreeBSD on non-x86 processors is separate from x86 processors on
1082 the config script, much like the NetBSD support.
1083 [Richard Levitte & Kris Kennaway <kris@obsecurity.org>]
1085 Changes between 0.9.6h and 0.9.7 [31 Dec 2002]
1087 [NB: OpenSSL 0.9.6i and later 0.9.6 patch levels were released after
1090 *) Fix session ID handling in SSLv2 client code: the SERVER FINISHED
1091 code (06) was taken as the first octet of the session ID and the last
1092 octet was ignored consequently. As a result SSLv2 client side session
1093 caching could not have worked due to the session ID mismatch between
1095 Behaviour observed by Crispin Flowerday <crispin@flowerday.cx> as
1099 *) Change the declaration of needed Kerberos libraries to use EX_LIBS
1100 instead of the special (and badly supported) LIBKRB5. LIBKRB5 is
1104 *) The hw_ncipher.c engine requires dynamic locks. Unfortunately, it
1105 seems that in spite of existing for more than a year, many application
1106 author have done nothing to provide the necessary callbacks, which
1107 means that this particular engine will not work properly anywhere.
1108 This is a very unfortunate situation which forces us, in the name
1109 of usability, to give the hw_ncipher.c a static lock, which is part
1111 NOTE: This is for the 0.9.7 series ONLY. This hack will never
1112 appear in 0.9.8 or later. We EXPECT application authors to have
1113 dealt properly with this when 0.9.8 is released (unless we actually
1114 make such changes in the libcrypto locking code that changes will
1115 have to be made anyway).
1118 *) In asn1_d2i_read_bio() repeatedly call BIO_read() until all content
1119 octets have been read, EOF or an error occurs. Without this change
1120 some truncated ASN1 structures will not produce an error.
1123 *) Disable Heimdal support, since it hasn't been fully implemented.
1124 Still give the possibility to force the use of Heimdal, but with
1125 warnings and a request that patches get sent to openssl-dev.
1128 *) Add the VC-CE target, introduce the WINCE sysname, and add
1129 INSTALL.WCE and appropriate conditionals to make it build.
1130 [Steven Reddie <smr@essemer.com.au> via Richard Levitte]
1132 *) Change the DLL names for Cygwin to cygcrypto-x.y.z.dll and
1133 cygssl-x.y.z.dll, where x, y and z are the major, minor and
1134 edit numbers of the version.
1135 [Corinna Vinschen <vinschen@redhat.com> and Richard Levitte]
1137 *) Introduce safe string copy and catenation functions
1138 (BUF_strlcpy() and BUF_strlcat()).
1139 [Ben Laurie (CHATS) and Richard Levitte]
1141 *) Avoid using fixed-size buffers for one-line DNs.
1142 [Ben Laurie (CHATS)]
1144 *) Add BUF_MEM_grow_clean() to avoid information leakage when
1145 resizing buffers containing secrets, and use where appropriate.
1146 [Ben Laurie (CHATS)]
1148 *) Avoid using fixed size buffers for configuration file location.
1149 [Ben Laurie (CHATS)]
1151 *) Avoid filename truncation for various CA files.
1152 [Ben Laurie (CHATS)]
1154 *) Use sizeof in preference to magic numbers.
1155 [Ben Laurie (CHATS)]
1157 *) Avoid filename truncation in cert requests.
1158 [Ben Laurie (CHATS)]
1160 *) Add assertions to check for (supposedly impossible) buffer
1162 [Ben Laurie (CHATS)]
1164 *) Don't cache truncated DNS entries in the local cache (this could
1165 potentially lead to a spoofing attack).
1166 [Ben Laurie (CHATS)]
1168 *) Fix various buffers to be large enough for hex/decimal
1169 representations in a platform independent manner.
1170 [Ben Laurie (CHATS)]
1172 *) Add CRYPTO_realloc_clean() to avoid information leakage when
1173 resizing buffers containing secrets, and use where appropriate.
1174 [Ben Laurie (CHATS)]
1176 *) Add BIO_indent() to avoid much slightly worrying code to do
1178 [Ben Laurie (CHATS)]
1180 *) Convert sprintf()/BIO_puts() to BIO_printf().
1181 [Ben Laurie (CHATS)]
1183 *) buffer_gets() could terminate with the buffer only half
1185 [Ben Laurie (CHATS)]
1187 *) Add assertions to prevent user-supplied crypto functions from
1188 overflowing internal buffers by having large block sizes, etc.
1189 [Ben Laurie (CHATS)]
1191 *) New OPENSSL_assert() macro (similar to assert(), but enabled
1193 [Ben Laurie (CHATS)]
1195 *) Eliminate unused copy of key in RC4.
1196 [Ben Laurie (CHATS)]
1198 *) Eliminate unused and incorrectly sized buffers for IV in pem.h.
1199 [Ben Laurie (CHATS)]
1201 *) Fix off-by-one error in EGD path.
1202 [Ben Laurie (CHATS)]
1204 *) If RANDFILE path is too long, ignore instead of truncating.
1205 [Ben Laurie (CHATS)]
1207 *) Eliminate unused and incorrectly sized X.509 structure
1209 [Ben Laurie (CHATS)]
1211 *) Eliminate unused and dangerous function knumber().
1212 [Ben Laurie (CHATS)]
1214 *) Eliminate unused and dangerous structure, KSSL_ERR.
1215 [Ben Laurie (CHATS)]
1217 *) Protect against overlong session ID context length in an encoded
1218 session object. Since these are local, this does not appear to be
1220 [Ben Laurie (CHATS)]
1222 *) Change from security patch (see 0.9.6e below) that did not affect
1223 the 0.9.6 release series:
1225 Remote buffer overflow in SSL3 protocol - an attacker could
1226 supply an oversized master key in Kerberos-enabled versions.
1228 [Ben Laurie (CHATS)]
1230 *) Change the SSL kerb5 codes to match RFC 2712.
1233 *) Make -nameopt work fully for req and add -reqopt switch.
1234 [Michael Bell <michael.bell@rz.hu-berlin.de>, Steve Henson]
1236 *) The "block size" for block ciphers in CFB and OFB mode should be 1.
1237 [Steve Henson, reported by Yngve Nysaeter Pettersen <yngve@opera.com>]
1239 *) Make sure tests can be performed even if the corresponding algorithms
1240 have been removed entirely. This was also the last step to make
1241 OpenSSL compilable with DJGPP under all reasonable conditions.
1242 [Richard Levitte, Doug Kaufman <dkaufman@rahul.net>]
1244 *) Add cipher selection rules COMPLEMENTOFALL and COMPLEMENTOFDEFAULT
1245 to allow version independent disabling of normally unselected ciphers,
1246 which may be activated as a side-effect of selecting a single cipher.
1248 (E.g., cipher list string "RSA" enables ciphersuites that are left
1249 out of "ALL" because they do not provide symmetric encryption.
1250 "RSA:!COMPLEMEMENTOFALL" avoids these unsafe ciphersuites.)
1251 [Lutz Jaenicke, Bodo Moeller]
1253 *) Add appropriate support for separate platform-dependent build
1254 directories. The recommended way to make a platform-dependent
1255 build directory is the following (tested on Linux), maybe with
1258 # Place yourself outside of the OpenSSL source tree. In
1259 # this example, the environment variable OPENSSL_SOURCE
1260 # is assumed to contain the absolute OpenSSL source directory.
1261 mkdir -p objtree/"`uname -s`-`uname -r`-`uname -m`"
1262 cd objtree/"`uname -s`-`uname -r`-`uname -m`"
1263 (cd $OPENSSL_SOURCE; find . -type f) | while read F; do
1264 mkdir -p `dirname $F`
1265 ln -s $OPENSSL_SOURCE/$F $F
1268 To be absolutely sure not to disturb the source tree, a "make clean"
1269 is a good thing. If it isn't successfull, don't worry about it,
1270 it probably means the source directory is very clean.
1273 *) Make sure any ENGINE control commands make local copies of string
1274 pointers passed to them whenever necessary. Otherwise it is possible
1275 the caller may have overwritten (or deallocated) the original string
1276 data when a later ENGINE operation tries to use the stored values.
1277 [Götz Babin-Ebell <babinebell@trustcenter.de>]
1279 *) Improve diagnostics in file reading and command-line digests.
1280 [Ben Laurie aided and abetted by Solar Designer <solar@openwall.com>]
1282 *) Add AES modes CFB and OFB to the object database. Correct an
1283 error in AES-CFB decryption.
1286 *) Remove most calls to EVP_CIPHER_CTX_cleanup() in evp_enc.c, this
1287 allows existing EVP_CIPHER_CTX structures to be reused after
1288 calling EVP_*Final(). This behaviour is used by encryption
1289 BIOs and some applications. This has the side effect that
1290 applications must explicitly clean up cipher contexts with
1291 EVP_CIPHER_CTX_cleanup() or they will leak memory.
1294 *) Check the values of dna and dnb in bn_mul_recursive before calling
1295 bn_mul_comba (a non zero value means the a or b arrays do not contain
1296 n2 elements) and fallback to bn_mul_normal if either is not zero.
1299 *) Fix escaping of non-ASCII characters when using the -subj option
1300 of the "openssl req" command line tool. (Robert Joop <joop@fokus.gmd.de>)
1303 *) Make object definitions compliant to LDAP (RFC2256): SN is the short
1304 form for "surname", serialNumber has no short form.
1305 Use "mail" as the short name for "rfc822Mailbox" according to RFC2798;
1306 therefore remove "mail" short name for "internet 7".
1307 The OID for unique identifiers in X509 certificates is
1308 x500UniqueIdentifier, not uniqueIdentifier.
1309 Some more OID additions. (Michael Bell <michael.bell@rz.hu-berlin.de>)
1312 *) Add an "init" command to the ENGINE config module and auto initialize
1313 ENGINEs. Without any "init" command the ENGINE will be initialized
1314 after all ctrl commands have been executed on it. If init=1 the
1315 ENGINE is initailized at that point (ctrls before that point are run
1316 on the uninitialized ENGINE and after on the initialized one). If
1317 init=0 then the ENGINE will not be iniatialized at all.
1320 *) Fix the 'app_verify_callback' interface so that the user-defined
1321 argument is actually passed to the callback: In the
1322 SSL_CTX_set_cert_verify_callback() prototype, the callback
1323 declaration has been changed from
1326 int (*cb)(X509_STORE_CTX *,void *);
1327 in ssl_verify_cert_chain (ssl/ssl_cert.c), the call
1328 i=s->ctx->app_verify_callback(&ctx)
1329 has been changed into
1330 i=s->ctx->app_verify_callback(&ctx, s->ctx->app_verify_arg).
1332 To update applications using SSL_CTX_set_cert_verify_callback(),
1333 a dummy argument can be added to their callback functions.
1334 [D. K. Smetters <smetters@parc.xerox.com>]
1336 *) Added the '4758cca' ENGINE to support IBM 4758 cards.
1337 [Maurice Gittens <maurice@gittens.nl>, touchups by Geoff Thorpe]
1339 *) Add and OPENSSL_LOAD_CONF define which will cause
1340 OpenSSL_add_all_algorithms() to load the openssl.cnf config file.
1341 This allows older applications to transparently support certain
1342 OpenSSL features: such as crypto acceleration and dynamic ENGINE loading.
1343 Two new functions OPENSSL_add_all_algorithms_noconf() which will never
1344 load the config file and OPENSSL_add_all_algorithms_conf() which will
1345 always load it have also been added.
1348 *) Add the OFB, CFB and CTR (all with 128 bit feedback) to AES.
1349 Adjust NIDs and EVP layer.
1350 [Stephen Sprunk <stephen@sprunk.org> and Richard Levitte]
1352 *) Config modules support in openssl utility.
1354 Most commands now load modules from the config file,
1355 though in a few (such as version) this isn't done
1356 because it couldn't be used for anything.
1358 In the case of ca and req the config file used is
1359 the same as the utility itself: that is the -config
1360 command line option can be used to specify an
1364 *) Move default behaviour from OPENSSL_config(). If appname is NULL
1365 use "openssl_conf" if filename is NULL use default openssl config file.
1368 *) Add an argument to OPENSSL_config() to allow the use of an alternative
1369 config section name. Add a new flag to tolerate a missing config file
1370 and move code to CONF_modules_load_file().
1373 *) Support for crypto accelerator cards from Accelerated Encryption
1374 Processing, www.aep.ie. (Use engine 'aep')
1375 The support was copied from 0.9.6c [engine] and adapted/corrected
1376 to work with the new engine framework.
1377 [AEP Inc. and Richard Levitte]
1379 *) Support for SureWare crypto accelerator cards from Baltimore
1380 Technologies. (Use engine 'sureware')
1381 The support was copied from 0.9.6c [engine] and adapted
1382 to work with the new engine framework.
1385 *) Have the CHIL engine fork-safe (as defined by nCipher) and actually
1386 make the newer ENGINE framework commands for the CHIL engine work.
1387 [Toomas Kiisk <vix@cyber.ee> and Richard Levitte]
1389 *) Make it possible to produce shared libraries on ReliantUNIX.
1390 [Robert Dahlem <Robert.Dahlem@ffm2.siemens.de> via Richard Levitte]
1392 *) Add the configuration target debug-linux-ppro.
1393 Make 'openssl rsa' use the general key loading routines
1394 implemented in apps.c, and make those routines able to
1395 handle the key format FORMAT_NETSCAPE and the variant
1397 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte]
1399 *) Fix a crashbug and a logic bug in hwcrhk_load_pubkey().
1400 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte]
1402 *) Add -keyform to rsautl, and document -engine.
1403 [Richard Levitte, inspired by Toomas Kiisk <vix@cyber.ee>]
1405 *) Change BIO_new_file (crypto/bio/bss_file.c) to use new
1406 BIO_R_NO_SUCH_FILE error code rather than the generic
1407 ERR_R_SYS_LIB error code if fopen() fails with ENOENT.
1410 *) Add new functions
1412 ERR_peek_last_error_line
1413 ERR_peek_last_error_line_data.
1414 These are similar to
1417 ERR_peek_error_line_data,
1418 but report on the latest error recorded rather than the first one
1419 still in the error queue.
1420 [Ben Laurie, Bodo Moeller]
1422 *) default_algorithms option in ENGINE config module. This allows things
1424 default_algorithms = ALL
1425 default_algorithms = RSA, DSA, RAND, CIPHERS, DIGESTS
1428 *) Prelminary ENGINE config module.
1431 *) New experimental application configuration code.
1434 *) Change the AES code to follow the same name structure as all other
1435 symmetric ciphers, and behave the same way. Move everything to
1436 the directory crypto/aes, thereby obsoleting crypto/rijndael.
1437 [Stephen Sprunk <stephen@sprunk.org> and Richard Levitte]
1439 *) SECURITY: remove unsafe setjmp/signal interaction from ui_openssl.c.
1440 [Ben Laurie and Theo de Raadt]
1442 *) Add option to output public keys in req command.
1443 [Massimiliano Pala madwolf@openca.org]
1445 *) Use wNAFs in EC_POINTs_mul() for improved efficiency
1446 (up to about 10% better than before for P-192 and P-224).
1449 *) New functions/macros
1451 SSL_CTX_set_msg_callback(ctx, cb)
1452 SSL_CTX_set_msg_callback_arg(ctx, arg)
1453 SSL_set_msg_callback(ssl, cb)
1454 SSL_set_msg_callback_arg(ssl, arg)
1456 to request calling a callback function
1458 void cb(int write_p, int version, int content_type,
1459 const void *buf, size_t len, SSL *ssl, void *arg)
1461 whenever a protocol message has been completely received
1462 (write_p == 0) or sent (write_p == 1). Here 'version' is the
1463 protocol version according to which the SSL library interprets
1464 the current protocol message (SSL2_VERSION, SSL3_VERSION, or
1465 TLS1_VERSION). 'content_type' is 0 in the case of SSL 2.0, or
1466 the content type as defined in the SSL 3.0/TLS 1.0 protocol
1467 specification (change_cipher_spec(20), alert(21), handshake(22)).
1468 'buf' and 'len' point to the actual message, 'ssl' to the
1469 SSL object, and 'arg' is the application-defined value set by
1470 SSL[_CTX]_set_msg_callback_arg().
1472 'openssl s_client' and 'openssl s_server' have new '-msg' options
1473 to enable a callback that displays all protocol messages.
1476 *) Change the shared library support so shared libraries are built as
1477 soon as the corresponding static library is finished, and thereby get
1478 openssl and the test programs linked against the shared library.
1479 This still only happens when the keyword "shard" has been given to
1480 the configuration scripts.
1482 NOTE: shared library support is still an experimental thing, and
1483 backward binary compatibility is still not guaranteed.
1484 ["Maciej W. Rozycki" <macro@ds2.pg.gda.pl> and Richard Levitte]
1486 *) Add support for Subject Information Access extension.
1487 [Peter Sylvester <Peter.Sylvester@EdelWeb.fr>]
1489 *) Make BUF_MEM_grow() behaviour more consistent: Initialise to zero
1490 additional bytes when new memory had to be allocated, not just
1491 when reusing an existing buffer.
1494 *) New command line and configuration option 'utf8' for the req command.
1495 This allows field values to be specified as UTF8 strings.
1498 *) Add -multi and -mr options to "openssl speed" - giving multiple parallel
1499 runs for the former and machine-readable output for the latter.
1502 *) Add '-noemailDN' option to 'openssl ca'. This prevents inclusion
1503 of the e-mail address in the DN (i.e., it will go into a certificate
1504 extension only). The new configuration file option 'email_in_dn = no'
1505 has the same effect.
1506 [Massimiliano Pala madwolf@openca.org]
1508 *) Change all functions with names starting with des_ to be starting
1509 with DES_ instead. Add wrappers that are compatible with libdes,
1510 but are named _ossl_old_des_*. Finally, add macros that map the
1511 des_* symbols to the corresponding _ossl_old_des_* if libdes
1512 compatibility is desired. If OpenSSL 0.9.6c compatibility is
1513 desired, the des_* symbols will be mapped to DES_*, with one
1516 Since we provide two compatibility mappings, the user needs to
1517 define the macro OPENSSL_DES_LIBDES_COMPATIBILITY if libdes
1518 compatibility is desired. The default (i.e., when that macro
1519 isn't defined) is OpenSSL 0.9.6c compatibility.
1521 There are also macros that enable and disable the support of old
1522 des functions altogether. Those are OPENSSL_ENABLE_OLD_DES_SUPPORT
1523 and OPENSSL_DISABLE_OLD_DES_SUPPORT. If none or both of those
1524 are defined, the default will apply: to support the old des routines.
1526 In either case, one must include openssl/des.h to get the correct
1527 definitions. Do not try to just include openssl/des_old.h, that
1530 NOTE: This is a major break of an old API into a new one. Software
1531 authors are encouraged to switch to the DES_ style functions. Some
1532 time in the future, des_old.h and the libdes compatibility functions
1533 will be disable (i.e. OPENSSL_DISABLE_OLD_DES_SUPPORT will be the
1534 default), and then completely removed.
1537 *) Test for certificates which contain unsupported critical extensions.
1538 If such a certificate is found during a verify operation it is
1539 rejected by default: this behaviour can be overridden by either
1540 handling the new error X509_V_ERR_UNHANDLED_CRITICAL_EXTENSION or
1541 by setting the verify flag X509_V_FLAG_IGNORE_CRITICAL. A new function
1542 X509_supported_extension() has also been added which returns 1 if a
1543 particular extension is supported.
1546 *) Modify the behaviour of EVP cipher functions in similar way to digests
1547 to retain compatibility with existing code.
1550 *) Modify the behaviour of EVP_DigestInit() and EVP_DigestFinal() to retain
1551 compatibility with existing code. In particular the 'ctx' parameter does
1552 not have to be to be initialized before the call to EVP_DigestInit() and
1553 it is tidied up after a call to EVP_DigestFinal(). New function
1554 EVP_DigestFinal_ex() which does not tidy up the ctx. Similarly function
1555 EVP_MD_CTX_copy() changed to not require the destination to be
1556 initialized valid and new function EVP_MD_CTX_copy_ex() added which
1557 requires the destination to be valid.
1559 Modify all the OpenSSL digest calls to use EVP_DigestInit_ex(),
1560 EVP_DigestFinal_ex() and EVP_MD_CTX_copy_ex().
1563 *) Change ssl3_get_message (ssl/s3_both.c) and the functions using it
1564 so that complete 'Handshake' protocol structures are kept in memory
1565 instead of overwriting 'msg_type' and 'length' with 'body' data.
1568 *) Add an implementation of SSL_add_dir_cert_subjects_to_stack for Win32.
1569 [Massimo Santin via Richard Levitte]
1571 *) Major restructuring to the underlying ENGINE code. This includes
1572 reduction of linker bloat, separation of pure "ENGINE" manipulation
1573 (initialisation, etc) from functionality dealing with implementations
1574 of specific crypto iterfaces. This change also introduces integrated
1575 support for symmetric ciphers and digest implementations - so ENGINEs
1576 can now accelerate these by providing EVP_CIPHER and EVP_MD
1577 implementations of their own. This is detailed in crypto/engine/README
1578 as it couldn't be adequately described here. However, there are a few
1579 API changes worth noting - some RSA, DSA, DH, and RAND functions that
1580 were changed in the original introduction of ENGINE code have now
1581 reverted back - the hooking from this code to ENGINE is now a good
1582 deal more passive and at run-time, operations deal directly with
1583 RSA_METHODs, DSA_METHODs (etc) as they did before, rather than
1584 dereferencing through an ENGINE pointer any more. Also, the ENGINE
1585 functions dealing with BN_MOD_EXP[_CRT] handlers have been removed -
1586 they were not being used by the framework as there is no concept of a
1587 BIGNUM_METHOD and they could not be generalised to the new
1588 'ENGINE_TABLE' mechanism that underlies the new code. Similarly,
1589 ENGINE_cpy() has been removed as it cannot be consistently defined in
1593 *) Change ASN1_GENERALIZEDTIME_check() to allow fractional seconds.
1596 *) Change mkdef.pl to sort symbols that get the same entry number,
1597 and make sure the automatically generated functions ERR_load_*
1598 become part of libeay.num as well.
1601 *) New function SSL_renegotiate_pending(). This returns true once
1602 renegotiation has been requested (either SSL_renegotiate() call
1603 or HelloRequest/ClientHello receveived from the peer) and becomes
1604 false once a handshake has been completed.
1605 (For servers, SSL_renegotiate() followed by SSL_do_handshake()
1606 sends a HelloRequest, but does not ensure that a handshake takes
1607 place. SSL_renegotiate_pending() is useful for checking if the
1608 client has followed the request.)
1611 *) New SSL option SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION.
1612 By default, clients may request session resumption even during
1613 renegotiation (if session ID contexts permit); with this option,
1614 session resumption is possible only in the first handshake.
1616 SSL_OP_ALL is now 0x00000FFFL instead of 0x000FFFFFL. This makes
1617 more bits available for options that should not be part of
1618 SSL_OP_ALL (such as SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION).
1621 *) Add some demos for certificate and certificate request creation.
1624 *) Make maximum certificate chain size accepted from the peer application
1625 settable (SSL*_get/set_max_cert_list()), as proposed by
1626 "Douglas E. Engert" <deengert@anl.gov>.
1629 *) Add support for shared libraries for Unixware-7
1630 (Boyd Lynn Gerber <gerberb@zenez.com>).
1633 *) Add a "destroy" handler to ENGINEs that allows structural cleanup to
1634 be done prior to destruction. Use this to unload error strings from
1635 ENGINEs that load their own error strings. NB: This adds two new API
1636 functions to "get" and "set" this destroy handler in an ENGINE.
1639 *) Alter all existing ENGINE implementations (except "openssl" and
1640 "openbsd") to dynamically instantiate their own error strings. This
1641 makes them more flexible to be built both as statically-linked ENGINEs
1642 and self-contained shared-libraries loadable via the "dynamic" ENGINE.
1643 Also, add stub code to each that makes building them as self-contained
1644 shared-libraries easier (see README.ENGINE).
1647 *) Add a "dynamic" ENGINE that provides a mechanism for binding ENGINE
1648 implementations into applications that are completely implemented in
1649 self-contained shared-libraries. The "dynamic" ENGINE exposes control
1650 commands that can be used to configure what shared-library to load and
1651 to control aspects of the way it is handled. Also, made an update to
1652 the README.ENGINE file that brings its information up-to-date and
1653 provides some information and instructions on the "dynamic" ENGINE
1654 (ie. how to use it, how to build "dynamic"-loadable ENGINEs, etc).
1657 *) Make it possible to unload ranges of ERR strings with a new
1658 "ERR_unload_strings" function.
1661 *) Add a copy() function to EVP_MD.
1664 *) Make EVP_MD routines take a context pointer instead of just the
1665 md_data void pointer.
1668 *) Add flags to EVP_MD and EVP_MD_CTX. EVP_MD_FLAG_ONESHOT indicates
1669 that the digest can only process a single chunk of data
1670 (typically because it is provided by a piece of
1671 hardware). EVP_MD_CTX_FLAG_ONESHOT indicates that the application
1672 is only going to provide a single chunk of data, and hence the
1673 framework needn't accumulate the data for oneshot drivers.
1676 *) As with "ERR", make it possible to replace the underlying "ex_data"
1677 functions. This change also alters the storage and management of global
1678 ex_data state - it's now all inside ex_data.c and all "class" code (eg.
1679 RSA, BIO, SSL_CTX, etc) no longer stores its own STACKS and per-class
1680 index counters. The API functions that use this state have been changed
1681 to take a "class_index" rather than pointers to the class's local STACK
1682 and counter, and there is now an API function to dynamically create new
1683 classes. This centralisation allows us to (a) plug a lot of the
1684 thread-safety problems that existed, and (b) makes it possible to clean
1685 up all allocated state using "CRYPTO_cleanup_all_ex_data()". W.r.t. (b)
1686 such data would previously have always leaked in application code and
1687 workarounds were in place to make the memory debugging turn a blind eye
1688 to it. Application code that doesn't use this new function will still
1689 leak as before, but their memory debugging output will announce it now
1690 rather than letting it slide.
1692 Besides the addition of CRYPTO_cleanup_all_ex_data(), another API change
1693 induced by the "ex_data" overhaul is that X509_STORE_CTX_init() now
1694 has a return value to indicate success or failure.
1697 *) Make it possible to replace the underlying "ERR" functions such that the
1698 global state (2 LHASH tables and 2 locks) is only used by the "default"
1699 implementation. This change also adds two functions to "get" and "set"
1700 the implementation prior to it being automatically set the first time
1701 any other ERR function takes place. Ie. an application can call "get",
1702 pass the return value to a module it has just loaded, and that module
1703 can call its own "set" function using that value. This means the
1704 module's "ERR" operations will use (and modify) the error state in the
1705 application and not in its own statically linked copy of OpenSSL code.
1708 *) Give DH, DSA, and RSA types their own "**_up_ref()" function to increment
1709 reference counts. This performs normal REF_PRINT/REF_CHECK macros on
1710 the operation, and provides a more encapsulated way for external code
1711 (crypto/evp/ and ssl/) to do this. Also changed the evp and ssl code
1712 to use these functions rather than manually incrementing the counts.
1714 Also rename "DSO_up()" function to more descriptive "DSO_up_ref()".
1717 *) Add EVP test program.
1720 *) Add symmetric cipher support to ENGINE. Expect the API to change!
1723 *) New CRL functions: X509_CRL_set_version(), X509_CRL_set_issuer_name()
1724 X509_CRL_set_lastUpdate(), X509_CRL_set_nextUpdate(), X509_CRL_sort(),
1725 X509_REVOKED_set_serialNumber(), and X509_REVOKED_set_revocationDate().
1726 These allow a CRL to be built without having to access X509_CRL fields
1727 directly. Modify 'ca' application to use new functions.
1730 *) Move SSL_OP_TLS_ROLLBACK_BUG out of the SSL_OP_ALL list of recommended
1731 bug workarounds. Rollback attack detection is a security feature.
1732 The problem will only arise on OpenSSL servers when TLSv1 is not
1733 available (sslv3_server_method() or SSL_OP_NO_TLSv1).
1734 Software authors not wanting to support TLSv1 will have special reasons
1735 for their choice and can explicitly enable this option.
1736 [Bodo Moeller, Lutz Jaenicke]
1738 *) Rationalise EVP so it can be extended: don't include a union of
1739 cipher/digest structures, add init/cleanup functions for EVP_MD_CTX
1740 (similar to those existing for EVP_CIPHER_CTX).
1745 EVP_MD_CTX_init(&md); /* new function call */
1746 EVP_DigestInit(&md, EVP_sha1());
1747 EVP_DigestUpdate(&md, in, len);
1748 EVP_DigestFinal(&md, out, NULL);
1749 EVP_MD_CTX_cleanup(&md); /* new function call */
1753 *) Make DES key schedule conform to the usual scheme, as well as
1754 correcting its structure. This means that calls to DES functions
1755 now have to pass a pointer to a des_key_schedule instead of a
1756 plain des_key_schedule (which was actually always a pointer
1759 des_key_schedule ks;
1761 des_set_key_checked(..., &ks);
1762 des_ncbc_encrypt(..., &ks, ...);
1764 (Note that a later change renames 'des_...' into 'DES_...'.)
1767 *) Initial reduction of linker bloat: the use of some functions, such as
1768 PEM causes large amounts of unused functions to be linked in due to
1769 poor organisation. For example pem_all.c contains every PEM function
1770 which has a knock on effect of linking in large amounts of (unused)
1771 ASN1 code. Grouping together similar functions and splitting unrelated
1772 functions prevents this.
1775 *) Cleanup of EVP macros.
1778 *) Change historical references to {NID,SN,LN}_des_ede and ede3 to add the
1779 correct _ecb suffix.
1782 *) Add initial OCSP responder support to ocsp application. The
1783 revocation information is handled using the text based index
1784 use by the ca application. The responder can either handle
1785 requests generated internally, supplied in files (for example
1786 via a CGI script) or using an internal minimal server.
1789 *) Add configuration choices to get zlib compression for TLS.
1792 *) Changes to Kerberos SSL for RFC 2712 compliance:
1793 1. Implemented real KerberosWrapper, instead of just using
1794 KRB5 AP_REQ message. [Thanks to Simon Wilkinson <sxw@sxw.org.uk>]
1795 2. Implemented optional authenticator field of KerberosWrapper.
1797 Added openssl-style ASN.1 macros for Kerberos ticket, ap_req,
1798 and authenticator structs; see crypto/krb5/.
1800 Generalized Kerberos calls to support multiple Kerberos libraries.
1801 [Vern Staats <staatsvr@asc.hpc.mil>,
1802 Jeffrey Altman <jaltman@columbia.edu>
1803 via Richard Levitte]
1805 *) Cause 'openssl speed' to use fully hard-coded DSA keys as it
1806 already does with RSA. testdsa.h now has 'priv_key/pub_key'
1807 values for each of the key sizes rather than having just
1808 parameters (and 'speed' generating keys each time).
1811 *) Speed up EVP routines.
1814 type 8 bytes 64 bytes 256 bytes 1024 bytes 8192 bytes
1815 des-cbc 4408.85k 5560.51k 5778.46k 5862.20k 5825.16k
1816 des-cbc 4389.55k 5571.17k 5792.23k 5846.91k 5832.11k
1817 des-cbc 4394.32k 5575.92k 5807.44k 5848.37k 5841.30k
1819 des-cbc 3482.66k 5069.49k 5496.39k 5614.16k 5639.28k
1820 des-cbc 3480.74k 5068.76k 5510.34k 5609.87k 5635.52k
1821 des-cbc 3483.72k 5067.62k 5504.60k 5708.01k 5724.80k
1824 des-cbc 4660.16k 5650.19k 5807.19k 5827.13k 5783.32k
1826 des-cbc 3624.96k 5258.21k 5530.91k 5624.30k 5628.26k
1829 *) Added the OS2-EMX target.
1830 ["Brian Havard" <brianh@kheldar.apana.org.au> and Richard Levitte]
1832 *) Rewrite apps to use NCONF routines instead of the old CONF. New functions
1833 to support NCONF routines in extension code. New function CONF_set_nconf()
1834 to allow functions which take an NCONF to also handle the old LHASH
1835 structure: this means that the old CONF compatible routines can be
1836 retained (in particular wrt extensions) without having to duplicate the
1837 code. New function X509V3_add_ext_nconf_sk to add extensions to a stack.
1840 *) Enhance the general user interface with mechanisms for inner control
1841 and with possibilities to have yes/no kind of prompts.
1844 *) Change all calls to low level digest routines in the library and
1845 applications to use EVP. Add missing calls to HMAC_cleanup() and
1846 don't assume HMAC_CTX can be copied using memcpy().
1847 [Verdon Walker <VWalker@novell.com>, Steve Henson]
1849 *) Add the possibility to control engines through control names but with
1850 arbitrary arguments instead of just a string.
1851 Change the key loaders to take a UI_METHOD instead of a callback
1852 function pointer. NOTE: this breaks binary compatibility with earlier
1853 versions of OpenSSL [engine].
1854 Adapt the nCipher code for these new conditions and add a card insertion
1858 *) Enhance the general user interface with mechanisms to better support
1859 dialog box interfaces, application-defined prompts, the possibility
1860 to use defaults (for example default passwords from somewhere else)
1861 and interrupts/cancellations.
1864 *) Tidy up PKCS#12 attribute handling. Add support for the CSP name
1865 attribute in PKCS#12 files, add new -CSP option to pkcs12 utility.
1868 *) Fix a memory leak in 'sk_dup()' in the case reallocation fails. (Also
1869 tidy up some unnecessarily weird code in 'sk_new()').
1870 [Geoff, reported by Diego Tartara <dtartara@novamens.com>]
1872 *) Change the key loading routines for ENGINEs to use the same kind
1873 callback (pem_password_cb) as all other routines that need this
1877 *) Increase ENTROPY_NEEDED to 32 bytes, as Rijndael can operate with
1878 256 bit (=32 byte) keys. Of course seeding with more entropy bytes
1879 than this minimum value is recommended.
1882 *) New random seeder for OpenVMS, using the system process statistics
1883 that are easily reachable.
1886 *) Windows apparently can't transparently handle global
1887 variables defined in DLLs. Initialisations such as:
1889 const ASN1_ITEM *it = &ASN1_INTEGER_it;
1891 wont compile. This is used by the any applications that need to
1892 declare their own ASN1 modules. This was fixed by adding the option
1893 EXPORT_VAR_AS_FN to all Win32 platforms, although this isn't strictly
1894 needed for static libraries under Win32.
1897 *) New functions X509_PURPOSE_set() and X509_TRUST_set() to handle
1898 setting of purpose and trust fields. New X509_STORE trust and
1899 purpose functions and tidy up setting in other SSL functions.
1902 *) Add copies of X509_STORE_CTX fields and callbacks to X509_STORE
1903 structure. These are inherited by X509_STORE_CTX when it is
1904 initialised. This allows various defaults to be set in the
1905 X509_STORE structure (such as flags for CRL checking and custom
1906 purpose or trust settings) for functions which only use X509_STORE_CTX
1907 internally such as S/MIME.
1909 Modify X509_STORE_CTX_purpose_inherit() so it only sets purposes and
1910 trust settings if they are not set in X509_STORE. This allows X509_STORE
1911 purposes and trust (in S/MIME for example) to override any set by default.
1913 Add command line options for CRL checking to smime, s_client and s_server
1917 *) Initial CRL based revocation checking. If the CRL checking flag(s)
1918 are set then the CRL is looked up in the X509_STORE structure and
1919 its validity and signature checked, then if the certificate is found
1920 in the CRL the verify fails with a revoked error.
1922 Various new CRL related callbacks added to X509_STORE_CTX structure.
1924 Command line options added to 'verify' application to support this.
1926 This needs some additional work, such as being able to handle multiple
1927 CRLs with different times, extension based lookup (rather than just
1928 by subject name) and ultimately more complete V2 CRL extension
1932 *) Add a general user interface API (crypto/ui/). This is designed
1933 to replace things like des_read_password and friends (backward
1934 compatibility functions using this new API are provided).
1935 The purpose is to remove prompting functions from the DES code
1936 section as well as provide for prompting through dialog boxes in
1937 a window system and the like.
1940 *) Add "ex_data" support to ENGINE so implementations can add state at a
1941 per-structure level rather than having to store it globally.
1944 *) Make it possible for ENGINE structures to be copied when retrieved by
1945 ENGINE_by_id() if the ENGINE specifies a new flag: ENGINE_FLAGS_BY_ID_COPY.
1946 This causes the "original" ENGINE structure to act like a template,
1947 analogous to the RSA vs. RSA_METHOD type of separation. Because of this
1948 operational state can be localised to each ENGINE structure, despite the
1949 fact they all share the same "methods". New ENGINE structures returned in
1950 this case have no functional references and the return value is the single
1951 structural reference. This matches the single structural reference returned
1952 by ENGINE_by_id() normally, when it is incremented on the pre-existing
1956 *) Fix ASN1 decoder when decoding type ANY and V_ASN1_OTHER: since this
1957 needs to match any other type at all we need to manually clear the
1961 *) Changes to the "openssl engine" utility to include;
1962 - verbosity levels ('-v', '-vv', and '-vvv') that provide information
1963 about an ENGINE's available control commands.
1964 - executing control commands from command line arguments using the
1965 '-pre' and '-post' switches. '-post' is only used if '-t' is
1966 specified and the ENGINE is successfully initialised. The syntax for
1967 the individual commands are colon-separated, for example;
1968 openssl engine chil -pre FORK_CHECK:0 -pre SO_PATH:/lib/test.so
1971 *) New dynamic control command support for ENGINEs. ENGINEs can now
1972 declare their own commands (numbers), names (strings), descriptions,
1973 and input types for run-time discovery by calling applications. A
1974 subset of these commands are implicitly classed as "executable"
1975 depending on their input type, and only these can be invoked through
1976 the new string-based API function ENGINE_ctrl_cmd_string(). (Eg. this
1977 can be based on user input, config files, etc). The distinction is
1978 that "executable" commands cannot return anything other than a boolean
1979 result and can only support numeric or string input, whereas some
1980 discoverable commands may only be for direct use through
1981 ENGINE_ctrl(), eg. supporting the exchange of binary data, function
1982 pointers, or other custom uses. The "executable" commands are to
1983 support parameterisations of ENGINE behaviour that can be
1984 unambiguously defined by ENGINEs and used consistently across any
1985 OpenSSL-based application. Commands have been added to all the
1986 existing hardware-supporting ENGINEs, noticeably "SO_PATH" to allow
1987 control over shared-library paths without source code alterations.
1990 *) Changed all ENGINE implementations to dynamically allocate their
1991 ENGINEs rather than declaring them statically. Apart from this being
1992 necessary with the removal of the ENGINE_FLAGS_MALLOCED distinction,
1993 this also allows the implementations to compile without using the
1994 internal engine_int.h header.
1997 *) Minor adjustment to "rand" code. RAND_get_rand_method() now returns a
1998 'const' value. Any code that should be able to modify a RAND_METHOD
1999 should already have non-const pointers to it (ie. they should only
2000 modify their own ones).
2003 *) Made a variety of little tweaks to the ENGINE code.
2004 - "atalla" and "ubsec" string definitions were moved from header files
2005 to C code. "nuron" string definitions were placed in variables
2006 rather than hard-coded - allowing parameterisation of these values
2007 later on via ctrl() commands.
2008 - Removed unused "#if 0"'d code.
2009 - Fixed engine list iteration code so it uses ENGINE_free() to release
2010 structural references.
2011 - Constified the RAND_METHOD element of ENGINE structures.
2012 - Constified various get/set functions as appropriate and added
2013 missing functions (including a catch-all ENGINE_cpy that duplicates
2014 all ENGINE values onto a new ENGINE except reference counts/state).
2015 - Removed NULL parameter checks in get/set functions. Setting a method
2016 or function to NULL is a way of cancelling out a previously set
2017 value. Passing a NULL ENGINE parameter is just plain stupid anyway
2018 and doesn't justify the extra error symbols and code.
2019 - Deprecate the ENGINE_FLAGS_MALLOCED define and move the area for
2020 flags from engine_int.h to engine.h.
2021 - Changed prototypes for ENGINE handler functions (init(), finish(),
2022 ctrl(), key-load functions, etc) to take an (ENGINE*) parameter.
2025 *) Implement binary inversion algorithm for BN_mod_inverse in addition
2026 to the algorithm using long division. The binary algorithm can be
2027 used only if the modulus is odd. On 32-bit systems, it is faster
2028 only for relatively small moduli (roughly 20-30% for 128-bit moduli,
2029 roughly 5-15% for 256-bit moduli), so we use it only for moduli
2030 up to 450 bits. In 64-bit environments, the binary algorithm
2031 appears to be advantageous for much longer moduli; here we use it
2032 for moduli up to 2048 bits.
2035 *) Rewrite CHOICE field setting in ASN1_item_ex_d2i(). The old code
2036 could not support the combine flag in choice fields.
2039 *) Add a 'copy_extensions' option to the 'ca' utility. This copies
2040 extensions from a certificate request to the certificate.
2043 *) Allow multiple 'certopt' and 'nameopt' options to be separated
2044 by commas. Add 'namopt' and 'certopt' options to the 'ca' config
2045 file: this allows the display of the certificate about to be
2046 signed to be customised, to allow certain fields to be included
2047 or excluded and extension details. The old system didn't display
2048 multicharacter strings properly, omitted fields not in the policy
2049 and couldn't display additional details such as extensions.
2052 *) Function EC_POINTs_mul for multiple scalar multiplication
2053 of an arbitrary number of elliptic curve points
2054 \sum scalars[i]*points[i],
2055 optionally including the generator defined for the EC_GROUP:
2056 scalar*generator + \sum scalars[i]*points[i].
2058 EC_POINT_mul is a simple wrapper function for the typical case
2059 that the point list has just one item (besides the optional
2063 *) First EC_METHODs for curves over GF(p):
2065 EC_GFp_simple_method() uses the basic BN_mod_mul and BN_mod_sqr
2066 operations and provides various method functions that can also
2067 operate with faster implementations of modular arithmetic.
2069 EC_GFp_mont_method() reuses most functions that are part of
2070 EC_GFp_simple_method, but uses Montgomery arithmetic.
2072 [Bodo Moeller; point addition and point doubling
2073 implementation directly derived from source code provided by
2074 Lenka Fibikova <fibikova@exp-math.uni-essen.de>]
2076 *) Framework for elliptic curves (crypto/ec/ec.h, crypto/ec/ec_lcl.h,
2077 crypto/ec/ec_lib.c):
2079 Curves are EC_GROUP objects (with an optional group generator)
2080 based on EC_METHODs that are built into the library.
2082 Points are EC_POINT objects based on EC_GROUP objects.
2084 Most of the framework would be able to handle curves over arbitrary
2085 finite fields, but as there are no obvious types for fields other
2086 than GF(p), some functions are limited to that for now.
2089 *) Add the -HTTP option to s_server. It is similar to -WWW, but requires
2090 that the file contains a complete HTTP response.
2093 *) Add the ec directory to mkdef.pl and mkfiles.pl. In mkdef.pl
2094 change the def and num file printf format specifier from "%-40sXXX"
2095 to "%-39s XXX". The latter will always guarantee a space after the
2096 field while the former will cause them to run together if the field
2097 is 40 of more characters long.
2100 *) Constify the cipher and digest 'method' functions and structures
2101 and modify related functions to take constant EVP_MD and EVP_CIPHER
2105 *) Hide BN_CTX structure details in bn_lcl.h instead of publishing them
2106 in <openssl/bn.h>. Also further increase BN_CTX_NUM to 32.
2109 *) Modify EVP_Digest*() routines so they now return values. Although the
2110 internal software routines can never fail additional hardware versions
2114 *) Clean up crypto/err/err.h and change some error codes to avoid conflicts:
2116 Previously ERR_R_FATAL was too small and coincided with ERR_LIB_PKCS7
2117 (= ERR_R_PKCS7_LIB); it is now 64 instead of 32.
2120 ERR_R_NESTED_ASN1_ERROR
2122 ERR_R_MISSING_ASN1_EOS
2123 were 4 .. 9, conflicting with
2124 ERR_LIB_RSA (= ERR_R_RSA_LIB)
2126 ERR_LIB_PEM (= ERR_R_PEM_LIB).
2127 They are now 58 .. 63 (i.e., just below ERR_R_FATAL).
2129 Add new error code 'ERR_R_INTERNAL_ERROR'.
2132 *) Don't overuse locks in crypto/err/err.c: For data retrieval, CRYPTO_r_lock
2136 *) New option '-subj arg' for 'openssl req' and 'openssl ca'. This
2137 sets the subject name for a new request or supersedes the
2138 subject name in a given request. Formats that can be parsed are
2139 'CN=Some Name, OU=myOU, C=IT'
2141 'CN=Some Name/OU=myOU/C=IT'.
2143 Add options '-batch' and '-verbose' to 'openssl req'.
2144 [Massimiliano Pala <madwolf@hackmasters.net>]
2146 *) Introduce the possibility to access global variables through
2147 functions on platform were that's the best way to handle exporting
2148 global variables in shared libraries. To enable this functionality,
2149 one must configure with "EXPORT_VAR_AS_FN" or defined the C macro
2150 "OPENSSL_EXPORT_VAR_AS_FUNCTION" in crypto/opensslconf.h (the latter
2151 is normally done by Configure or something similar).
2153 To implement a global variable, use the macro OPENSSL_IMPLEMENT_GLOBAL
2154 in the source file (foo.c) like this:
2156 OPENSSL_IMPLEMENT_GLOBAL(int,foo)=1;
2157 OPENSSL_IMPLEMENT_GLOBAL(double,bar);
2159 To declare a global variable, use the macros OPENSSL_DECLARE_GLOBAL
2160 and OPENSSL_GLOBAL_REF in the header file (foo.h) like this:
2162 OPENSSL_DECLARE_GLOBAL(int,foo);
2163 #define foo OPENSSL_GLOBAL_REF(foo)
2164 OPENSSL_DECLARE_GLOBAL(double,bar);
2165 #define bar OPENSSL_GLOBAL_REF(bar)
2167 The #defines are very important, and therefore so is including the
2168 header file everywhere where the defined globals are used.
2170 The macro OPENSSL_EXPORT_VAR_AS_FUNCTION also affects the definition
2171 of ASN.1 items, but that structure is a bit different.
2173 The largest change is in util/mkdef.pl which has been enhanced with
2174 better and easier to understand logic to choose which symbols should
2175 go into the Windows .def files as well as a number of fixes and code
2176 cleanup (among others, algorithm keywords are now sorted
2177 lexicographically to avoid constant rewrites).
2180 *) In BN_div() keep a copy of the sign of 'num' before writing the
2181 result to 'rm' because if rm==num the value will be overwritten
2182 and produce the wrong result if 'num' is negative: this caused
2183 problems with BN_mod() and BN_nnmod().
2186 *) Function OCSP_request_verify(). This checks the signature on an
2187 OCSP request and verifies the signer certificate. The signer
2188 certificate is just checked for a generic purpose and OCSP request
2192 *) Add OCSP_check_validity() function to check the validity of OCSP
2193 responses. OCSP responses are prepared in real time and may only
2194 be a few seconds old. Simply checking that the current time lies
2195 between thisUpdate and nextUpdate max reject otherwise valid responses
2196 caused by either OCSP responder or client clock inaccuracy. Instead
2197 we allow thisUpdate and nextUpdate to fall within a certain period of
2198 the current time. The age of the response can also optionally be
2199 checked. Two new options -validity_period and -status_age added to
2203 *) If signature or public key algorithm is unrecognized print out its
2204 OID rather that just UNKNOWN.
2207 *) Change OCSP_cert_to_id() to tolerate a NULL subject certificate and
2208 OCSP_cert_id_new() a NULL serialNumber. This allows a partial certificate
2209 ID to be generated from the issuer certificate alone which can then be
2210 passed to OCSP_id_issuer_cmp().
2213 *) New compilation option ASN1_ITEM_FUNCTIONS. This causes the new
2214 ASN1 modules to export functions returning ASN1_ITEM pointers
2215 instead of the ASN1_ITEM structures themselves. This adds several
2216 new macros which allow the underlying ASN1 function/structure to
2217 be accessed transparently. As a result code should not use ASN1_ITEM
2218 references directly (such as &X509_it) but instead use the relevant
2219 macros (such as ASN1_ITEM_rptr(X509)). This option is to allow
2220 use of the new ASN1 code on platforms where exporting structures
2221 is problematical (for example in shared libraries) but exporting
2222 functions returning pointers to structures is not.
2225 *) Add support for overriding the generation of SSL/TLS session IDs.
2226 These callbacks can be registered either in an SSL_CTX or per SSL.
2227 The purpose of this is to allow applications to control, if they wish,
2228 the arbitrary values chosen for use as session IDs, particularly as it
2229 can be useful for session caching in multiple-server environments. A
2230 command-line switch for testing this (and any client code that wishes
2231 to use such a feature) has been added to "s_server".
2232 [Geoff Thorpe, Lutz Jaenicke]
2234 *) Modify mkdef.pl to recognise and parse preprocessor conditionals
2235 of the form '#if defined(...) || defined(...) || ...' and
2236 '#if !defined(...) && !defined(...) && ...'. This also avoids
2237 the growing number of special cases it was previously handling.
2240 *) Make all configuration macros available for application by making
2241 sure they are available in opensslconf.h, by giving them names starting
2242 with "OPENSSL_" to avoid conflicts with other packages and by making
2243 sure e_os2.h will cover all platform-specific cases together with
2245 Additionally, it is now possible to define configuration/platform-
2246 specific names (called "system identities"). In the C code, these
2247 are prefixed with "OPENSSL_SYSNAME_". e_os2.h will create another
2248 macro with the name beginning with "OPENSSL_SYS_", which is determined
2249 from "OPENSSL_SYSNAME_*" or compiler-specific macros depending on
2253 *) New option -set_serial to 'req' and 'x509' this allows the serial
2254 number to use to be specified on the command line. Previously self
2255 signed certificates were hard coded with serial number 0 and the
2256 CA options of 'x509' had to use a serial number in a file which was
2260 *) New options to 'ca' utility to support V2 CRL entry extensions.
2261 Currently CRL reason, invalidity date and hold instruction are
2262 supported. Add new CRL extensions to V3 code and some new objects.
2265 *) New function EVP_CIPHER_CTX_set_padding() this is used to
2266 disable standard block padding (aka PKCS#5 padding) in the EVP
2267 API, which was previously mandatory. This means that the data is
2268 not padded in any way and so the total length much be a multiple
2269 of the block size, otherwise an error occurs.
2272 *) Initial (incomplete) OCSP SSL support.
2275 *) New function OCSP_parse_url(). This splits up a URL into its host,
2276 port and path components: primarily to parse OCSP URLs. New -url
2277 option to ocsp utility.
2280 *) New nonce behavior. The return value of OCSP_check_nonce() now
2281 reflects the various checks performed. Applications can decide
2282 whether to tolerate certain situations such as an absent nonce
2283 in a response when one was present in a request: the ocsp application
2284 just prints out a warning. New function OCSP_add1_basic_nonce()
2285 this is to allow responders to include a nonce in a response even if
2286 the request is nonce-less.
2289 *) Disable stdin buffering in load_cert (apps/apps.c) so that no certs are
2290 skipped when using openssl x509 multiple times on a single input file,
2291 e.g. "(openssl x509 -out cert1; openssl x509 -out cert2) <certs".
2294 *) Make ASN1_UTCTIME_set_string() and ASN1_GENERALIZEDTIME_set_string()
2295 set string type: to handle setting ASN1_TIME structures. Fix ca
2296 utility to correctly initialize revocation date of CRLs.
2299 *) New option SSL_OP_CIPHER_SERVER_PREFERENCE allows the server to override
2300 the clients preferred ciphersuites and rather use its own preferences.
2301 Should help to work around M$ SGC (Server Gated Cryptography) bug in
2302 Internet Explorer by ensuring unchanged hash method during stepup.
2303 (Also replaces the broken/deactivated SSL_OP_NON_EXPORT_FIRST option.)
2306 *) Make mkdef.pl recognise all DECLARE_ASN1 macros, change rijndael
2307 to aes and add a new 'exist' option to print out symbols that don't
2311 *) Additional options to ocsp utility to allow flags to be set and
2312 additional certificates supplied.
2315 *) Add the option -VAfile to 'openssl ocsp', so the user can give the
2316 OCSP client a number of certificate to only verify the response
2320 *) Update Rijndael code to version 3.0 and change EVP AES ciphers to
2321 handle the new API. Currently only ECB, CBC modes supported. Add new
2324 Add TLS AES ciphersuites as described in RFC3268, "Advanced
2325 Encryption Standard (AES) Ciphersuites for Transport Layer
2326 Security (TLS)". (In beta versions of OpenSSL 0.9.7, these were
2327 not enabled by default and were not part of the "ALL" ciphersuite
2328 alias because they were not yet official; they could be
2329 explicitly requested by specifying the "AESdraft" ciphersuite
2330 group alias. In the final release of OpenSSL 0.9.7, the group
2331 alias is called "AES" and is part of "ALL".)
2332 [Ben Laurie, Steve Henson, Bodo Moeller]
2334 *) New function OCSP_copy_nonce() to copy nonce value (if present) from
2335 request to response.
2338 *) Functions for OCSP responders. OCSP_request_onereq_count(),
2339 OCSP_request_onereq_get0(), OCSP_onereq_get0_id() and OCSP_id_get0_info()
2340 extract information from a certificate request. OCSP_response_create()
2341 creates a response and optionally adds a basic response structure.
2342 OCSP_basic_add1_status() adds a complete single response to a basic
2343 response and returns the OCSP_SINGLERESP structure just added (to allow
2344 extensions to be included for example). OCSP_basic_add1_cert() adds a
2345 certificate to a basic response and OCSP_basic_sign() signs a basic
2346 response with various flags. New helper functions ASN1_TIME_check()
2347 (checks validity of ASN1_TIME structure) and ASN1_TIME_to_generalizedtime()
2348 (converts ASN1_TIME to GeneralizedTime).
2351 *) Various new functions. EVP_Digest() combines EVP_Digest{Init,Update,Final}()
2352 in a single operation. X509_get0_pubkey_bitstr() extracts the public_key
2353 structure from a certificate. X509_pubkey_digest() digests the public_key
2354 contents: this is used in various key identifiers.
2357 *) Make sk_sort() tolerate a NULL argument.
2358 [Steve Henson reported by Massimiliano Pala <madwolf@comune.modena.it>]
2360 *) New OCSP verify flag OCSP_TRUSTOTHER. When set the "other" certificates
2361 passed by the function are trusted implicitly. If any of them signed the
2362 response then it is assumed to be valid and is not verified.
2365 *) In PKCS7_set_type() initialise content_type in PKCS7_ENC_CONTENT
2366 to data. This was previously part of the PKCS7 ASN1 code. This
2367 was causing problems with OpenSSL created PKCS#12 and PKCS#7 structures.
2368 [Steve Henson, reported by Kenneth R. Robinette
2369 <support@securenetterm.com>]
2371 *) Add CRYPTO_push_info() and CRYPTO_pop_info() calls to new ASN1
2372 routines: without these tracing memory leaks is very painful.
2373 Fix leaks in PKCS12 and PKCS7 routines.
2376 *) Make X509_time_adj() cope with the new behaviour of ASN1_TIME_new().
2377 Previously it initialised the 'type' argument to V_ASN1_UTCTIME which
2378 effectively meant GeneralizedTime would never be used. Now it
2379 is initialised to -1 but X509_time_adj() now has to check the value
2380 and use ASN1_TIME_set() if the value is not V_ASN1_UTCTIME or
2381 V_ASN1_GENERALIZEDTIME, without this it always uses GeneralizedTime.
2382 [Steve Henson, reported by Kenneth R. Robinette
2383 <support@securenetterm.com>]
2385 *) Fixes to BN_to_ASN1_INTEGER when bn is zero. This would previously
2386 result in a zero length in the ASN1_INTEGER structure which was
2387 not consistent with the structure when d2i_ASN1_INTEGER() was used
2388 and would cause ASN1_INTEGER_cmp() to fail. Enhance s2i_ASN1_INTEGER()
2389 to cope with hex and negative integers. Fix bug in i2a_ASN1_INTEGER()
2390 where it did not print out a minus for negative ASN1_INTEGER.
2393 *) Add summary printout to ocsp utility. The various functions which
2394 convert status values to strings have been renamed to:
2395 OCSP_response_status_str(), OCSP_cert_status_str() and
2396 OCSP_crl_reason_str() and are no longer static. New options
2397 to verify nonce values and to disable verification. OCSP response
2398 printout format cleaned up.
2401 *) Add additional OCSP certificate checks. These are those specified
2402 in RFC2560. This consists of two separate checks: the CA of the
2403 certificate being checked must either be the OCSP signer certificate
2404 or the issuer of the OCSP signer certificate. In the latter case the
2405 OCSP signer certificate must contain the OCSP signing extended key
2406 usage. This check is performed by attempting to match the OCSP
2407 signer or the OCSP signer CA to the issuerNameHash and issuerKeyHash
2408 in the OCSP_CERTID structures of the response.
2411 *) Initial OCSP certificate verification added to OCSP_basic_verify()
2412 and related routines. This uses the standard OpenSSL certificate
2413 verify routines to perform initial checks (just CA validity) and
2414 to obtain the certificate chain. Then additional checks will be
2415 performed on the chain. Currently the root CA is checked to see
2416 if it is explicitly trusted for OCSP signing. This is used to set
2417 a root CA as a global signing root: that is any certificate that
2418 chains to that CA is an acceptable OCSP signing certificate.
2421 *) New '-extfile ...' option to 'openssl ca' for reading X.509v3
2422 extensions from a separate configuration file.
2423 As when reading extensions from the main configuration file,
2424 the '-extensions ...' option may be used for specifying the
2426 [Massimiliano Pala <madwolf@comune.modena.it>]
2428 *) New OCSP utility. Allows OCSP requests to be generated or
2429 read. The request can be sent to a responder and the output
2430 parsed, outputed or printed in text form. Not complete yet:
2431 still needs to check the OCSP response validity.
2434 *) New subcommands for 'openssl ca':
2435 'openssl ca -status <serial>' prints the status of the cert with
2436 the given serial number (according to the index file).
2437 'openssl ca -updatedb' updates the expiry status of certificates
2439 [Massimiliano Pala <madwolf@comune.modena.it>]
2441 *) New '-newreq-nodes' command option to CA.pl. This is like
2442 '-newreq', but calls 'openssl req' with the '-nodes' option
2443 so that the resulting key is not encrypted.
2444 [Damien Miller <djm@mindrot.org>]
2446 *) New configuration for the GNU Hurd.
2447 [Jonathan Bartlett <johnnyb@wolfram.com> via Richard Levitte]
2449 *) Initial code to implement OCSP basic response verify. This
2450 is currently incomplete. Currently just finds the signer's
2451 certificate and verifies the signature on the response.
2454 *) New SSLeay_version code SSLEAY_DIR to determine the compiled-in
2455 value of OPENSSLDIR. This is available via the new '-d' option
2456 to 'openssl version', and is also included in 'openssl version -a'.
2459 *) Allowing defining memory allocation callbacks that will be given
2460 file name and line number information in additional arguments
2461 (a const char* and an int). The basic functionality remains, as
2462 well as the original possibility to just replace malloc(),
2463 realloc() and free() by functions that do not know about these
2464 additional arguments. To register and find out the current
2465 settings for extended allocation functions, the following
2466 functions are provided:
2468 CRYPTO_set_mem_ex_functions
2469 CRYPTO_set_locked_mem_ex_functions
2470 CRYPTO_get_mem_ex_functions
2471 CRYPTO_get_locked_mem_ex_functions
2473 These work the same way as CRYPTO_set_mem_functions and friends.
2474 CRYPTO_get_[locked_]mem_functions now writes 0 where such an
2475 extended allocation function is enabled.
2476 Similarly, CRYPTO_get_[locked_]mem_ex_functions writes 0 where
2477 a conventional allocation function is enabled.
2478 [Richard Levitte, Bodo Moeller]
2480 *) Finish off removing the remaining LHASH function pointer casts.
2481 There should no longer be any prototype-casting required when using
2482 the LHASH abstraction, and any casts that remain are "bugs". See
2483 the callback types and macros at the head of lhash.h for details
2484 (and "OBJ_cleanup" in crypto/objects/obj_dat.c as an example).
2487 *) Add automatic query of EGD sockets in RAND_poll() for the unix variant.
2488 If /dev/[u]random devices are not available or do not return enough
2489 entropy, EGD style sockets (served by EGD or PRNGD) will automatically
2491 The locations /var/run/egd-pool, /dev/egd-pool, /etc/egd-pool, and
2492 /etc/entropy will be queried once each in this sequence, quering stops
2493 when enough entropy was collected without querying more sockets.
2496 *) Change the Unix RAND_poll() variant to be able to poll several
2497 random devices, as specified by DEVRANDOM, until a sufficient amount
2498 of data has been collected. We spend at most 10 ms on each file
2499 (select timeout) and read in non-blocking mode. DEVRANDOM now
2500 defaults to the list "/dev/urandom", "/dev/random", "/dev/srandom"
2501 (previously it was just the string "/dev/urandom"), so on typical
2502 platforms the 10 ms delay will never occur.
2503 Also separate out the Unix variant to its own file, rand_unix.c.
2504 For VMS, there's a currently-empty rand_vms.c.
2507 *) Move OCSP client related routines to ocsp_cl.c. These
2508 provide utility functions which an application needing
2509 to issue a request to an OCSP responder and analyse the
2510 response will typically need: as opposed to those which an
2511 OCSP responder itself would need which will be added later.
2513 OCSP_request_sign() signs an OCSP request with an API similar
2514 to PKCS7_sign(). OCSP_response_status() returns status of OCSP
2515 response. OCSP_response_get1_basic() extracts basic response
2516 from response. OCSP_resp_find_status(): finds and extracts status
2517 information from an OCSP_CERTID structure (which will be created
2518 when the request structure is built). These are built from lower
2519 level functions which work on OCSP_SINGLERESP structures but
2520 wont normally be used unless the application wishes to examine
2521 extensions in the OCSP response for example.
2523 Replace nonce routines with a pair of functions.
2524 OCSP_request_add1_nonce() adds a nonce value and optionally
2525 generates a random value. OCSP_check_nonce() checks the
2526 validity of the nonce in an OCSP response.
2529 *) Change function OCSP_request_add() to OCSP_request_add0_id().
2530 This doesn't copy the supplied OCSP_CERTID and avoids the
2531 need to free up the newly created id. Change return type
2532 to OCSP_ONEREQ to return the internal OCSP_ONEREQ structure.
2533 This can then be used to add extensions to the request.
2534 Deleted OCSP_request_new(), since most of its functionality
2535 is now in OCSP_REQUEST_new() (and the case insensitive name
2536 clash) apart from the ability to set the request name which
2537 will be added elsewhere.
2540 *) Update OCSP API. Remove obsolete extensions argument from
2541 various functions. Extensions are now handled using the new
2542 OCSP extension code. New simple OCSP HTTP function which
2543 can be used to send requests and parse the response.
2546 *) Fix the PKCS#7 (S/MIME) code to work with new ASN1. Two new
2547 ASN1_ITEM structures help with sign and verify. PKCS7_ATTR_SIGN
2548 uses the special reorder version of SET OF to sort the attributes
2549 and reorder them to match the encoded order. This resolves a long
2550 standing problem: a verify on a PKCS7 structure just after signing
2551 it used to fail because the attribute order did not match the
2552 encoded order. PKCS7_ATTR_VERIFY does not reorder the attributes:
2553 it uses the received order. This is necessary to tolerate some broken
2554 software that does not order SET OF. This is handled by encoding
2555 as a SEQUENCE OF but using implicit tagging (with UNIVERSAL class)
2556 to produce the required SET OF.
2559 *) Have mk1mf.pl generate the macros OPENSSL_BUILD_SHLIBCRYPTO and
2560 OPENSSL_BUILD_SHLIBSSL and use them appropriately in the header
2561 files to get correct declarations of the ASN.1 item variables.
2564 *) Rewrite of PKCS#12 code to use new ASN1 functionality. Replace many
2565 PKCS#12 macros with real functions. Fix two unrelated ASN1 bugs:
2566 asn1_check_tlen() would sometimes attempt to use 'ctx' when it was
2567 NULL and ASN1_TYPE was not dereferenced properly in asn1_ex_c2i().
2568 New ASN1 macro: DECLARE_ASN1_ITEM() which just declares the relevant
2569 ASN1_ITEM and no wrapper functions.
2572 *) New functions or ASN1_item_d2i_fp() and ASN1_item_d2i_bio(). These
2573 replace the old function pointer based I/O routines. Change most of
2574 the *_d2i_bio() and *_d2i_fp() functions to use these.
2577 *) Enhance mkdef.pl to be more accepting about spacing in C preprocessor
2578 lines, recognice more "algorithms" that can be deselected, and make
2579 it complain about algorithm deselection that isn't recognised.
2582 *) New ASN1 functions to handle dup, sign, verify, digest, pack and
2583 unpack operations in terms of ASN1_ITEM. Modify existing wrappers
2584 to use new functions. Add NO_ASN1_OLD which can be set to remove
2585 some old style ASN1 functions: this can be used to determine if old
2586 code will still work when these eventually go away.
2589 *) New extension functions for OCSP structures, these follow the
2590 same conventions as certificates and CRLs.
2593 *) New function X509V3_add1_i2d(). This automatically encodes and
2594 adds an extension. Its behaviour can be customised with various
2595 flags to append, replace or delete. Various wrappers added for
2596 certifcates and CRLs.
2599 *) Fix to avoid calling the underlying ASN1 print routine when
2600 an extension cannot be parsed. Correct a typo in the
2601 OCSP_SERVICELOC extension. Tidy up print OCSP format.
2604 *) Make mkdef.pl parse some of the ASN1 macros and add apropriate
2605 entries for variables.
2608 *) Add functionality to apps/openssl.c for detecting locking
2609 problems: As the program is single-threaded, all we have
2610 to do is register a locking callback using an array for
2611 storing which locks are currently held by the program.
2614 *) Use a lock around the call to CRYPTO_get_ex_new_index() in
2615 SSL_get_ex_data_X509_STORE_idx(), which is used in
2616 ssl_verify_cert_chain() and thus can be called at any time
2617 during TLS/SSL handshakes so that thread-safety is essential.
2618 Unfortunately, the ex_data design is not at all suited
2619 for multi-threaded use, so it probably should be abolished.
2622 *) Added Broadcom "ubsec" ENGINE to OpenSSL.
2623 [Broadcom, tweaked and integrated by Geoff Thorpe]
2625 *) Move common extension printing code to new function
2626 X509V3_print_extensions(). Reorganise OCSP print routines and
2627 implement some needed OCSP ASN1 functions. Add OCSP extensions.
2630 *) New function X509_signature_print() to remove duplication in some
2634 *) Add a special meaning when SET OF and SEQUENCE OF flags are both
2635 set (this was treated exactly the same as SET OF previously). This
2636 is used to reorder the STACK representing the structure to match the
2637 encoding. This will be used to get round a problem where a PKCS7
2638 structure which was signed could not be verified because the STACK
2639 order did not reflect the encoded order.
2642 *) Reimplement the OCSP ASN1 module using the new code.
2645 *) Update the X509V3 code to permit the use of an ASN1_ITEM structure
2646 for its ASN1 operations. The old style function pointers still exist
2647 for now but they will eventually go away.
2650 *) Merge in replacement ASN1 code from the ASN1 branch. This almost
2651 completely replaces the old ASN1 functionality with a table driven
2652 encoder and decoder which interprets an ASN1_ITEM structure describing
2653 the ASN1 module. Compatibility with the existing ASN1 API (i2d,d2i) is
2654 largely maintained. Almost all of the old asn1_mac.h macro based ASN1
2655 has also been converted to the new form.
2658 *) Change BN_mod_exp_recp so that negative moduli are tolerated
2659 (the sign is ignored). Similarly, ignore the sign in BN_MONT_CTX_set
2660 so that BN_mod_exp_mont and BN_mod_exp_mont_word work
2661 for negative moduli.
2664 *) Fix BN_uadd and BN_usub: Always return non-negative results instead
2665 of not touching the result's sign bit.
2668 *) BN_div bugfix: If the result is 0, the sign (res->neg) must not be
2672 *) Changed the LHASH code to use prototypes for callbacks, and created
2673 macros to declare and implement thin (optionally static) functions
2674 that provide type-safety and avoid function pointer casting for the
2675 type-specific callbacks.
2678 *) Added Kerberos Cipher Suites to be used with TLS, as written in
2680 [Veers Staats <staatsvr@asc.hpc.mil>,
2681 Jeffrey Altman <jaltman@columbia.edu>, via Richard Levitte]
2683 *) Reformat the FAQ so the different questions and answers can be divided
2684 in sections depending on the subject.
2687 *) Have the zlib compression code load ZLIB.DLL dynamically under
2691 *) New function BN_mod_sqrt for computing square roots modulo a prime
2692 (using the probabilistic Tonelli-Shanks algorithm unless
2693 p == 3 (mod 4) or p == 5 (mod 8), which are cases that can
2694 be handled deterministically).
2695 [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller]
2697 *) Make BN_mod_inverse faster by explicitly handling small quotients
2698 in the Euclid loop. (Speed gain about 20% for small moduli [256 or
2699 512 bits], about 30% for larger ones [1024 or 2048 bits].)
2702 *) New function BN_kronecker.
2705 *) Fix BN_gcd so that it works on negative inputs; the result is
2706 positive unless both parameters are zero.
2707 Previously something reasonably close to an infinite loop was
2708 possible because numbers could be growing instead of shrinking
2709 in the implementation of Euclid's algorithm.
2712 *) Fix BN_is_word() and BN_is_one() macros to take into account the
2713 sign of the number in question.
2715 Fix BN_is_word(a,w) to work correctly for w == 0.
2717 The old BN_is_word(a,w) macro is now called BN_abs_is_word(a,w)
2718 because its test if the absolute value of 'a' equals 'w'.
2719 Note that BN_abs_is_word does *not* handle w == 0 reliably;
2720 it exists mostly for use in the implementations of BN_is_zero(),
2721 BN_is_one(), and BN_is_word().
2724 *) New function BN_swap.
2727 *) Use BN_nnmod instead of BN_mod in crypto/bn/bn_exp.c so that
2728 the exponentiation functions are more likely to produce reasonable
2729 results on negative inputs.
2732 *) Change BN_mod_mul so that the result is always non-negative.
2733 Previously, it could be negative if one of the factors was negative;
2734 I don't think anyone really wanted that behaviour.
2737 *) Move BN_mod_... functions into new file crypto/bn/bn_mod.c
2738 (except for exponentiation, which stays in crypto/bn/bn_exp.c,
2739 and BN_mod_mul_reciprocal, which stays in crypto/bn/bn_recp.c)
2740 and add new functions:
2749 BN_mod_lshift1_quick
2753 These functions always generate non-negative results.
2755 BN_nnmod otherwise is like BN_mod (if BN_mod computes a remainder r
2756 such that |m| < r < 0, BN_nnmod will output rem + |m| instead).
2758 BN_mod_XXX_quick(r, a, [b,] m) generates the same result as
2759 BN_mod_XXX(r, a, [b,] m, ctx), but requires that a [and b]
2760 be reduced modulo m.
2761 [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller]
2764 The following entry accidentily appeared in the CHANGES file
2765 distributed with OpenSSL 0.9.7. The modifications described in
2766 it do *not* apply to OpenSSL 0.9.7.
2768 *) Remove a few calls to bn_wexpand() in BN_sqr() (the one in there
2769 was actually never needed) and in BN_mul(). The removal in BN_mul()
2770 required a small change in bn_mul_part_recursive() and the addition
2771 of the functions bn_cmp_part_words(), bn_sub_part_words() and
2772 bn_add_part_words(), which do the same thing as bn_cmp_words(),
2773 bn_sub_words() and bn_add_words() except they take arrays with
2778 *) In 'openssl passwd', verify passwords read from the terminal
2779 unless the '-salt' option is used (which usually means that
2780 verification would just waste user's time since the resulting
2781 hash is going to be compared with some given password hash)
2782 or the new '-noverify' option is used.
2784 This is an incompatible change, but it does not affect
2785 non-interactive use of 'openssl passwd' (passwords on the command
2786 line, '-stdin' option, '-in ...' option) and thus should not
2790 *) Remove all references to RSAref, since there's no more need for it.
2793 *) Make DSO load along a path given through an environment variable
2794 (SHLIB_PATH) with shl_load().
2797 *) Constify the ENGINE code as a result of BIGNUM constification.
2798 Also constify the RSA code and most things related to it. In a
2799 few places, most notable in the depth of the ASN.1 code, ugly
2800 casts back to non-const were required (to be solved at a later
2804 *) Make it so the openssl application has all engines loaded by default.
2807 *) Constify the BIGNUM routines a little more.
2810 *) Add the following functions:
2812 ENGINE_load_cswift()
2814 ENGINE_load_atalla()
2816 ENGINE_load_builtin_engines()
2818 That way, an application can itself choose if external engines that
2819 are built-in in OpenSSL shall ever be used or not. The benefit is
2820 that applications won't have to be linked with libdl or other dso
2821 libraries unless it's really needed.
2823 Changed 'openssl engine' to load all engines on demand.
2824 Changed the engine header files to avoid the duplication of some
2825 declarations (they differed!).
2828 *) 'openssl engine' can now list capabilities.
2831 *) Better error reporting in 'openssl engine'.
2834 *) Never call load_dh_param(NULL) in s_server.
2837 *) Add engine application. It can currently list engines by name and
2838 identity, and test if they are actually available.
2841 *) Improve RPM specification file by forcing symbolic linking and making
2842 sure the installed documentation is also owned by root.root.
2843 [Damien Miller <djm@mindrot.org>]
2845 *) Give the OpenSSL applications more possibilities to make use of
2846 keys (public as well as private) handled by engines.
2849 *) Add OCSP code that comes from CertCo.
2852 *) Add VMS support for the Rijndael code.
2855 *) Added untested support for Nuron crypto accelerator.
2858 *) Add support for external cryptographic devices. This code was
2859 previously distributed separately as the "engine" branch.
2860 [Geoff Thorpe, Richard Levitte]
2862 *) Rework the filename-translation in the DSO code. It is now possible to
2863 have far greater control over how a "name" is turned into a filename
2864 depending on the operating environment and any oddities about the
2865 different shared library filenames on each system.
2868 *) Support threads on FreeBSD-elf in Configure.
2871 *) Fix for SHA1 assembly problem with MASM: it produces
2872 warnings about corrupt line number information when assembling
2873 with debugging information. This is caused by the overlapping
2875 [Bernd Matthes <mainbug@celocom.de>, Steve Henson]
2878 NCONF_get_number() has no error checking at all. As a replacement,
2879 NCONF_get_number_e() is defined (_e for "error checking") and is
2880 promoted strongly. The old NCONF_get_number is kept around for
2881 binary backward compatibility.
2882 Make it possible for methods to load from something other than a BIO,
2883 by providing a function pointer that is given a name instead of a BIO.
2884 For example, this could be used to load configuration data from an
2888 *) Fix for non blocking accept BIOs. Added new I/O special reason
2889 BIO_RR_ACCEPT to cover this case. Previously use of accept BIOs
2890 with non blocking I/O was not possible because no retry code was
2891 implemented. Also added new SSL code SSL_WANT_ACCEPT to cover
2895 *) Added the beginnings of Rijndael support.
2898 *) Fix for bug in DirectoryString mask setting. Add support for
2899 X509_NAME_print_ex() in 'req' and X509_print_ex() function
2900 to allow certificate printing to more controllable, additional
2901 'certopt' option to 'x509' to allow new printing options to be
2905 *) Clean old EAY MD5 hack from e_os.h.
2908 Changes between 0.9.6l and 0.9.6m [17 Mar 2004]
2910 *) Fix null-pointer assignment in do_change_cipher_spec() revealed
2911 by using the Codenomicon TLS Test Tool (CAN-2004-0079)
2912 [Joe Orton, Steve Henson]
2914 Changes between 0.9.6k and 0.9.6l [04 Nov 2003]
2916 *) Fix additional bug revealed by the NISCC test suite:
2918 Stop bug triggering large recursion when presented with
2919 certain ASN.1 tags (CAN-2003-0851)
2922 Changes between 0.9.6j and 0.9.6k [30 Sep 2003]
2924 *) Fix various bugs revealed by running the NISCC test suite:
2926 Stop out of bounds reads in the ASN1 code when presented with
2927 invalid tags (CAN-2003-0543 and CAN-2003-0544).
2929 If verify callback ignores invalid public key errors don't try to check
2930 certificate signature with the NULL public key.
2934 *) In ssl3_accept() (ssl/s3_srvr.c) only accept a client certificate
2935 if the server requested one: as stated in TLS 1.0 and SSL 3.0
2939 *) In ssl3_get_client_hello() (ssl/s3_srvr.c), tolerate additional
2940 extra data after the compression methods not only for TLS 1.0
2941 but also for SSL 3.0 (as required by the specification).
2942 [Bodo Moeller; problem pointed out by Matthias Loepfe]
2944 *) Change X509_certificate_type() to mark the key as exported/exportable
2945 when it's 512 *bits* long, not 512 bytes.
2948 Changes between 0.9.6i and 0.9.6j [10 Apr 2003]
2950 *) Countermeasure against the Klima-Pokorny-Rosa extension of
2951 Bleichbacher's attack on PKCS #1 v1.5 padding: treat
2952 a protocol version number mismatch like a decryption error
2953 in ssl3_get_client_key_exchange (ssl/s3_srvr.c).
2956 *) Turn on RSA blinding by default in the default implementation
2957 to avoid a timing attack. Applications that don't want it can call
2958 RSA_blinding_off() or use the new flag RSA_FLAG_NO_BLINDING.
2959 They would be ill-advised to do so in most cases.
2960 [Ben Laurie, Steve Henson, Geoff Thorpe, Bodo Moeller]
2962 *) Change RSA blinding code so that it works when the PRNG is not
2963 seeded (in this case, the secret RSA exponent is abused as
2964 an unpredictable seed -- if it is not unpredictable, there
2965 is no point in blinding anyway). Make RSA blinding thread-safe
2966 by remembering the creator's thread ID in rsa->blinding and
2967 having all other threads use local one-time blinding factors
2968 (this requires more computation than sharing rsa->blinding, but
2969 avoids excessive locking; and if an RSA object is not shared
2970 between threads, blinding will still be very fast).
2973 Changes between 0.9.6h and 0.9.6i [19 Feb 2003]
2975 *) In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked
2976 via timing by performing a MAC computation even if incorrrect
2977 block cipher padding has been found. This is a countermeasure
2978 against active attacks where the attacker has to distinguish
2979 between bad padding and a MAC verification error. (CAN-2003-0078)
2981 [Bodo Moeller; problem pointed out by Brice Canvel (EPFL),
2982 Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and
2983 Martin Vuagnoux (EPFL, Ilion)]
2985 Changes between 0.9.6g and 0.9.6h [5 Dec 2002]
2987 *) New function OPENSSL_cleanse(), which is used to cleanse a section of
2988 memory from it's contents. This is done with a counter that will
2989 place alternating values in each byte. This can be used to solve
2990 two issues: 1) the removal of calls to memset() by highly optimizing
2991 compilers, and 2) cleansing with other values than 0, since those can
2992 be read through on certain media, for example a swap space on disk.
2995 *) Bugfix: client side session caching did not work with external caching,
2996 because the session->cipher setting was not restored when reloading
2997 from the external cache. This problem was masked, when
2998 SSL_OP_NETSCAPE_REUSE_CIPHER_CHANGE_BUG (part of SSL_OP_ALL) was set.
2999 (Found by Steve Haslam <steve@araqnid.ddts.net>.)
3002 *) Fix client_certificate (ssl/s2_clnt.c): The permissible total
3003 length of the REQUEST-CERTIFICATE message is 18 .. 34, not 17 .. 33.
3004 [Zeev Lieber <zeev-l@yahoo.com>]
3006 *) Undo an undocumented change introduced in 0.9.6e which caused
3007 repeated calls to OpenSSL_add_all_ciphers() and
3008 OpenSSL_add_all_digests() to be ignored, even after calling
3012 *) Change the default configuration reader to deal with last line not
3013 being properly terminated.
3016 *) Change X509_NAME_cmp() so it applies the special rules on handling
3017 DN values that are of type PrintableString, as well as RDNs of type
3018 emailAddress where the value has the type ia5String.
3019 [stefank@valicert.com via Richard Levitte]
3021 *) Add a SSL_SESS_CACHE_NO_INTERNAL_STORE flag to take over half
3022 the job SSL_SESS_CACHE_NO_INTERNAL_LOOKUP was inconsistently
3023 doing, define a new flag (SSL_SESS_CACHE_NO_INTERNAL) to be
3024 the bitwise-OR of the two for use by the majority of applications
3025 wanting this behaviour, and update the docs. The documented
3026 behaviour and actual behaviour were inconsistent and had been
3027 changing anyway, so this is more a bug-fix than a behavioural
3029 [Geoff Thorpe, diagnosed by Nadav Har'El]
3031 *) Don't impose a 16-byte length minimum on session IDs in ssl/s3_clnt.c
3032 (the SSL 3.0 and TLS 1.0 specifications allow any length up to 32 bytes).
3035 *) Fix initialization code race conditions in
3036 SSLv23_method(), SSLv23_client_method(), SSLv23_server_method(),
3037 SSLv2_method(), SSLv2_client_method(), SSLv2_server_method(),
3038 SSLv3_method(), SSLv3_client_method(), SSLv3_server_method(),
3039 TLSv1_method(), TLSv1_client_method(), TLSv1_server_method(),
3040 ssl2_get_cipher_by_char(),
3041 ssl3_get_cipher_by_char().
3042 [Patrick McCormick <patrick@tellme.com>, Bodo Moeller]
3044 *) Reorder cleanup sequence in SSL_CTX_free(): only remove the ex_data after
3045 the cached sessions are flushed, as the remove_cb() might use ex_data
3046 contents. Bug found by Sam Varshavchik <mrsam@courier-mta.com>
3047 (see [openssl.org #212]).
3048 [Geoff Thorpe, Lutz Jaenicke]
3050 *) Fix typo in OBJ_txt2obj which incorrectly passed the content
3051 length, instead of the encoding length to d2i_ASN1_OBJECT.
3054 Changes between 0.9.6f and 0.9.6g [9 Aug 2002]
3056 *) [In 0.9.6g-engine release:]
3057 Fix crypto/engine/vendor_defns/cswift.h for WIN32 (use '_stdcall').
3058 [Lynn Gazis <lgazis@rainbow.com>]
3060 Changes between 0.9.6e and 0.9.6f [8 Aug 2002]
3062 *) Fix ASN1 checks. Check for overflow by comparing with LONG_MAX
3063 and get fix the header length calculation.
3064 [Florian Weimer <Weimer@CERT.Uni-Stuttgart.DE>,
3065 Alon Kantor <alonk@checkpoint.com> (and others),
3068 *) Use proper error handling instead of 'assertions' in buffer
3069 overflow checks added in 0.9.6e. This prevents DoS (the
3070 assertions could call abort()).
3071 [Arne Ansper <arne@ats.cyber.ee>, Bodo Moeller]
3073 Changes between 0.9.6d and 0.9.6e [30 Jul 2002]
3075 *) Add various sanity checks to asn1_get_length() to reject
3076 the ASN1 length bytes if they exceed sizeof(long), will appear
3077 negative or the content length exceeds the length of the
3079 [Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>]
3081 *) Fix cipher selection routines: ciphers without encryption had no flags
3082 for the cipher strength set and where therefore not handled correctly
3083 by the selection routines (PR #130).
3086 *) Fix EVP_dsa_sha macro.
3090 SSL_OP_DONT_INSERT_EMPTY_FRAGMENTS
3091 for disabling the SSL 3.0/TLS 1.0 CBC vulnerability countermeasure
3092 that was added in OpenSSL 0.9.6d.
3094 As the countermeasure turned out to be incompatible with some
3095 broken SSL implementations, the new option is part of SSL_OP_ALL.
3096 SSL_OP_ALL is usually employed when compatibility with weird SSL
3097 implementations is desired (e.g. '-bugs' option to 's_client' and
3098 's_server'), so the new option is automatically set in many
3102 *) Changes in security patch:
3104 Changes marked "(CHATS)" were sponsored by the Defense Advanced
3105 Research Projects Agency (DARPA) and Air Force Research Laboratory,
3106 Air Force Materiel Command, USAF, under agreement number
3109 *) Add various sanity checks to asn1_get_length() to reject
3110 the ASN1 length bytes if they exceed sizeof(long), will appear
3111 negative or the content length exceeds the length of the
3112 supplied buffer. (CAN-2002-0659)
3113 [Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>]
3115 *) Assertions for various potential buffer overflows, not known to
3117 [Ben Laurie (CHATS)]
3119 *) Various temporary buffers to hold ASCII versions of integers were
3120 too small for 64 bit platforms. (CAN-2002-0655)
3121 [Matthew Byng-Maddick <mbm@aldigital.co.uk> and Ben Laurie (CHATS)>
3123 *) Remote buffer overflow in SSL3 protocol - an attacker could
3124 supply an oversized session ID to a client. (CAN-2002-0656)
3125 [Ben Laurie (CHATS)]
3127 *) Remote buffer overflow in SSL2 protocol - an attacker could
3128 supply an oversized client master key. (CAN-2002-0656)
3129 [Ben Laurie (CHATS)]
3131 Changes between 0.9.6c and 0.9.6d [9 May 2002]
3133 *) Fix crypto/asn1/a_sign.c so that 'parameters' is omitted (not
3134 encoded as NULL) with id-dsa-with-sha1.
3135 [Nils Larsch <nla@trustcenter.de>; problem pointed out by Bodo Moeller]
3137 *) Check various X509_...() return values in apps/req.c.
3138 [Nils Larsch <nla@trustcenter.de>]
3140 *) Fix BASE64 decode (EVP_DecodeUpdate) for data with CR/LF ended lines:
3141 an end-of-file condition would erronously be flagged, when the CRLF
3142 was just at the end of a processed block. The bug was discovered when
3143 processing data through a buffering memory BIO handing the data to a
3144 BASE64-decoding BIO. Bug fund and patch submitted by Pavel Tsekov
3145 <ptsekov@syntrex.com> and Nedelcho Stanev.
3148 *) Implement a countermeasure against a vulnerability recently found
3149 in CBC ciphersuites in SSL 3.0/TLS 1.0: Send an empty fragment
3150 before application data chunks to avoid the use of known IVs
3151 with data potentially chosen by the attacker.
3154 *) Fix length checks in ssl3_get_client_hello().
3157 *) TLS/SSL library bugfix: use s->s3->in_read_app_data differently
3158 to prevent ssl3_read_internal() from incorrectly assuming that
3159 ssl3_read_bytes() found application data while handshake
3160 processing was enabled when in fact s->s3->in_read_app_data was
3161 merely automatically cleared during the initial handshake.
3162 [Bodo Moeller; problem pointed out by Arne Ansper <arne@ats.cyber.ee>]
3164 *) Fix object definitions for Private and Enterprise: they were not
3165 recognized in their shortname (=lowercase) representation. Extend
3166 obj_dat.pl to issue an error when using undefined keywords instead
3167 of silently ignoring the problem (Svenning Sorensen
3168 <sss@sss.dnsalias.net>).
3171 *) Fix DH_generate_parameters() so that it works for 'non-standard'
3172 generators, i.e. generators other than 2 and 5. (Previously, the
3173 code did not properly initialise the 'add' and 'rem' values to
3174 BN_generate_prime().)
3176 In the new general case, we do not insist that 'generator' is
3177 actually a primitive root: This requirement is rather pointless;
3178 a generator of the order-q subgroup is just as good, if not
3182 *) Map new X509 verification errors to alerts. Discovered and submitted by
3183 Tom Wu <tom@arcot.com>.
3186 *) Fix ssl3_pending() (ssl/s3_lib.c) to prevent SSL_pending() from
3187 returning non-zero before the data has been completely received
3188 when using non-blocking I/O.
3189 [Bodo Moeller; problem pointed out by John Hughes]
3191 *) Some of the ciphers missed the strength entry (SSL_LOW etc).
3192 [Ben Laurie, Lutz Jaenicke]
3194 *) Fix bug in SSL_clear(): bad sessions were not removed (found by
3195 Yoram Zahavi <YoramZ@gilian.com>).
3198 *) Add information about CygWin 1.3 and on, and preserve proper
3199 configuration for the versions before that.
3200 [Corinna Vinschen <vinschen@redhat.com> and Richard Levitte]
3202 *) Make removal from session cache (SSL_CTX_remove_session()) more robust:
3203 check whether we deal with a copy of a session and do not delete from
3204 the cache in this case. Problem reported by "Izhar Shoshani Levi"
3205 <izhar@checkpoint.com>.
3208 *) Do not store session data into the internal session cache, if it
3209 is never intended to be looked up (SSL_SESS_CACHE_NO_INTERNAL_LOOKUP
3210 flag is set). Proposed by Aslam <aslam@funk.com>.
3213 *) Have ASN1_BIT_STRING_set_bit() really clear a bit when the requested
3217 *) [In 0.9.6d-engine release:]
3218 Fix a crashbug and a logic bug in hwcrhk_load_pubkey().
3219 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte]
3221 *) Add the configuration target linux-s390x.
3222 [Neale Ferguson <Neale.Ferguson@SoftwareAG-USA.com> via Richard Levitte]
3224 *) The earlier bugfix for the SSL3_ST_SW_HELLO_REQ_C case of
3225 ssl3_accept (ssl/s3_srvr.c) incorrectly used a local flag
3226 variable as an indication that a ClientHello message has been
3227 received. As the flag value will be lost between multiple
3228 invocations of ssl3_accept when using non-blocking I/O, the
3229 function may not be aware that a handshake has actually taken
3230 place, thus preventing a new session from being added to the
3233 To avoid this problem, we now set s->new_session to 2 instead of
3234 using a local variable.
3235 [Lutz Jaenicke, Bodo Moeller]
3237 *) Bugfix: Return -1 from ssl3_get_server_done (ssl3/s3_clnt.c)
3238 if the SSL_R_LENGTH_MISMATCH error is detected.
3239 [Geoff Thorpe, Bodo Moeller]
3241 *) New 'shared_ldflag' column in Configure platform table.
3244 *) Fix EVP_CIPHER_mode macro.
3245 ["Dan S. Camper" <dan@bti.net>]
3247 *) Fix ssl3_read_bytes (ssl/s3_pkt.c): To ignore messages of unknown
3248 type, we must throw them away by setting rr->length to 0.
3249 [D P Chang <dpc@qualys.com>]
3251 Changes between 0.9.6b and 0.9.6c [21 dec 2001]
3253 *) Fix BN_rand_range bug pointed out by Dominikus Scherkl
3254 <Dominikus.Scherkl@biodata.com>. (The previous implementation
3255 worked incorrectly for those cases where range = 10..._2 and
3256 3*range is two bits longer than range.)
3259 *) Only add signing time to PKCS7 structures if it is not already
3263 *) Fix crypto/objects/objects.h: "ld-ce" should be "id-ce",
3264 OBJ_ld_ce should be OBJ_id_ce.
3265 Also some ip-pda OIDs in crypto/objects/objects.txt were
3266 incorrect (cf. RFC 3039).
3267 [Matt Cooper, Frederic Giudicelli, Bodo Moeller]
3269 *) Release CRYPTO_LOCK_DYNLOCK when CRYPTO_destroy_dynlockid()
3270 returns early because it has nothing to do.
3271 [Andy Schneider <andy.schneider@bjss.co.uk>]
3273 *) [In 0.9.6c-engine release:]
3274 Fix mutex callback return values in crypto/engine/hw_ncipher.c.
3275 [Andy Schneider <andy.schneider@bjss.co.uk>]
3277 *) [In 0.9.6c-engine release:]
3278 Add support for Cryptographic Appliance's keyserver technology.
3279 (Use engine 'keyclient')
3280 [Cryptographic Appliances and Geoff Thorpe]
3282 *) Add a configuration entry for OS/390 Unix. The C compiler 'c89'
3283 is called via tools/c89.sh because arguments have to be
3284 rearranged (all '-L' options must appear before the first object
3286 [Richard Shapiro <rshapiro@abinitio.com>]
3288 *) [In 0.9.6c-engine release:]
3289 Add support for Broadcom crypto accelerator cards, backported
3291 [Broadcom, Nalin Dahyabhai <nalin@redhat.com>, Mark Cox]
3293 *) [In 0.9.6c-engine release:]
3294 Add support for SureWare crypto accelerator cards from
3295 Baltimore Technologies. (Use engine 'sureware')
3296 [Baltimore Technologies and Mark Cox]
3298 *) [In 0.9.6c-engine release:]
3299 Add support for crypto accelerator cards from Accelerated
3300 Encryption Processing, www.aep.ie. (Use engine 'aep')
3301 [AEP Inc. and Mark Cox]
3303 *) Add a configuration entry for gcc on UnixWare.
3304 [Gary Benson <gbenson@redhat.com>]
3306 *) Change ssl/s2_clnt.c and ssl/s2_srvr.c so that received handshake
3307 messages are stored in a single piece (fixed-length part and
3308 variable-length part combined) and fix various bugs found on the way.
3311 *) Disable caching in BIO_gethostbyname(), directly use gethostbyname()
3312 instead. BIO_gethostbyname() does not know what timeouts are
3313 appropriate, so entries would stay in cache even when they have
3315 [Bodo Moeller; problem pointed out by Rich Salz <rsalz@zolera.com>
3317 *) Change ssl23_get_client_hello (ssl/s23_srvr.c) behaviour when
3318 faced with a pathologically small ClientHello fragment that does
3319 not contain client_version: Instead of aborting with an error,
3320 simply choose the highest available protocol version (i.e.,
3321 TLS 1.0 unless it is disabled). In practice, ClientHello
3322 messages are never sent like this, but this change gives us
3323 strictly correct behaviour at least for TLS.
3326 *) Fix SSL handshake functions and SSL_clear() such that SSL_clear()
3327 never resets s->method to s->ctx->method when called from within
3328 one of the SSL handshake functions.
3329 [Bodo Moeller; problem pointed out by Niko Baric]
3331 *) In ssl3_get_client_hello (ssl/s3_srvr.c), generate a fatal alert
3332 (sent using the client's version number) if client_version is
3333 smaller than the protocol version in use. Also change
3334 ssl23_get_client_hello (ssl/s23_srvr.c) to select TLS 1.0 if
3335 the client demanded SSL 3.0 but only TLS 1.0 is enabled; then
3336 the client will at least see that alert.
3339 *) Fix ssl3_get_message (ssl/s3_both.c) to handle message fragmentation
3343 *) Avoid infinite loop in ssl3_get_message (ssl/s3_both.c) if a
3344 client receives HelloRequest while in a handshake.
3345 [Bodo Moeller; bug noticed by Andy Schneider <andy.schneider@bjss.co.uk>]
3347 *) Bugfix in ssl3_accept (ssl/s3_srvr.c): Case SSL3_ST_SW_HELLO_REQ_C
3348 should end in 'break', not 'goto end' which circuments various
3349 cleanups done in state SSL_ST_OK. But session related stuff
3350 must be disabled for SSL_ST_OK in the case that we just sent a
3353 Also avoid some overhead by not calling ssl_init_wbio_buffer()
3354 before just sending a HelloRequest.
3355 [Bodo Moeller, Eric Rescorla <ekr@rtfm.com>]
3357 *) Fix ssl/s3_enc.c, ssl/t1_enc.c and ssl/s3_pkt.c so that we don't
3358 reveal whether illegal block cipher padding was found or a MAC
3359 verification error occured. (Neither SSLerr() codes nor alerts
3360 are directly visible to potential attackers, but the information
3361 may leak via logfiles.)
3363 Similar changes are not required for the SSL 2.0 implementation
3364 because the number of padding bytes is sent in clear for SSL 2.0,
3365 and the extra bytes are just ignored. However ssl/s2_pkt.c
3366 failed to verify that the purported number of padding bytes is in
3370 *) Add OpenUNIX-8 support including shared libraries
3371 (Boyd Lynn Gerber <gerberb@zenez.com>).
3374 *) Improve RSA_padding_check_PKCS1_OAEP() check again to avoid
3375 'wristwatch attack' using huge encoding parameters (cf.
3376 James H. Manger's CRYPTO 2001 paper). Note that the
3377 RSA_PKCS1_OAEP_PADDING case of RSA_private_decrypt() does not use
3378 encoding parameters and hence was not vulnerable.
3381 *) BN_sqr() bug fix.
3382 [Ulf Möller, reported by Jim Ellis <jim.ellis@cavium.com>]
3384 *) Rabin-Miller test analyses assume uniformly distributed witnesses,
3385 so use BN_pseudo_rand_range() instead of using BN_pseudo_rand()
3386 followed by modular reduction.
3387 [Bodo Moeller; pointed out by Adam Young <AYoung1@NCSUS.JNJ.COM>]
3389 *) Add BN_pseudo_rand_range() with obvious functionality: BN_rand_range()
3390 equivalent based on BN_pseudo_rand() instead of BN_rand().
3393 *) s3_srvr.c: allow sending of large client certificate lists (> 16 kB).
3394 This function was broken, as the check for a new client hello message
3395 to handle SGC did not allow these large messages.
3396 (Tracked down by "Douglas E. Engert" <deengert@anl.gov>.)
3399 *) Add alert descriptions for TLSv1 to SSL_alert_desc_string[_long]().
3402 *) Fix buggy behaviour of BIO_get_num_renegotiates() and BIO_ctrl()
3403 for BIO_C_GET_WRITE_BUF_SIZE ("Stephen Hinton" <shinton@netopia.com>).
3406 *) Rework the configuration and shared library support for Tru64 Unix.
3407 The configuration part makes use of modern compiler features and
3408 still retains old compiler behavior for those that run older versions
3409 of the OS. The shared library support part includes a variant that
3410 uses the RPATH feature, and is available through the special
3411 configuration target "alpha-cc-rpath", which will never be selected
3413 [Tim Mooney <mooney@dogbert.cc.ndsu.NoDak.edu> via Richard Levitte]
3415 *) In ssl3_get_key_exchange (ssl/s3_clnt.c), call ssl3_get_message()
3416 with the same message size as in ssl3_get_certificate_request().
3417 Otherwise, if no ServerKeyExchange message occurs, CertificateRequest
3418 messages might inadvertently be reject as too long.
3419 [Petr Lampa <lampa@fee.vutbr.cz>]
3421 *) Enhanced support for IA-64 Unix platforms (well, Linux and HP-UX).
3424 *) Modified SSL library such that the verify_callback that has been set
3425 specificly for an SSL object with SSL_set_verify() is actually being
3426 used. Before the change, a verify_callback set with this function was
3427 ignored and the verify_callback() set in the SSL_CTX at the time of
3428 the call was used. New function X509_STORE_CTX_set_verify_cb() introduced
3429 to allow the necessary settings.
3432 *) Initialize static variable in crypto/dsa/dsa_lib.c and crypto/dh/dh_lib.c
3433 explicitly to NULL, as at least on Solaris 8 this seems not always to be
3434 done automatically (in contradiction to the requirements of the C
3435 standard). This made problems when used from OpenSSH.
3438 *) In OpenSSL 0.9.6a and 0.9.6b, crypto/dh/dh_key.c ignored
3439 dh->length and always used
3441 BN_rand_range(priv_key, dh->p).
3443 BN_rand_range() is not necessary for Diffie-Hellman, and this
3444 specific range makes Diffie-Hellman unnecessarily inefficient if
3445 dh->length (recommended exponent length) is much smaller than the
3446 length of dh->p. We could use BN_rand_range() if the order of
3447 the subgroup was stored in the DH structure, but we only have
3452 BN_rand(priv_key, l, ...)
3454 where 'l' is dh->length if this is defined, or BN_num_bits(dh->p)-1
3460 RSA_eay_public_encrypt
3461 RSA_eay_private_decrypt
3462 RSA_eay_private_encrypt (signing)
3463 RSA_eay_public_decrypt (signature verification)
3465 (default implementations for RSA_public_encrypt,
3466 RSA_private_decrypt, RSA_private_encrypt, RSA_public_decrypt),
3467 always reject numbers >= n.
3470 *) In crypto/rand/md_rand.c, use a new short-time lock CRYPTO_LOCK_RAND2
3471 to synchronize access to 'locking_thread'. This is necessary on
3472 systems where access to 'locking_thread' (an 'unsigned long'
3473 variable) is not atomic.
3476 *) In crypto/rand/md_rand.c, set 'locking_thread' to current thread's ID
3477 *before* setting the 'crypto_lock_rand' flag. The previous code had
3478 a race condition if 0 is a valid thread ID.
3479 [Travis Vitek <vitek@roguewave.com>]
3481 *) Add support for shared libraries under Irix.
3482 [Albert Chin-A-Young <china@thewrittenword.com>]
3484 *) Add configuration option to build on Linux on both big-endian and
3486 [Ralf Baechle <ralf@uni-koblenz.de>]
3488 *) Add the possibility to create shared libraries on HP-UX.
3491 Changes between 0.9.6a and 0.9.6b [9 Jul 2001]
3493 *) Change ssleay_rand_bytes (crypto/rand/md_rand.c)
3494 to avoid a SSLeay/OpenSSL PRNG weakness pointed out by
3495 Markku-Juhani O. Saarinen <markku-juhani.saarinen@nokia.com>:
3496 PRNG state recovery was possible based on the output of
3497 one PRNG request appropriately sized to gain knowledge on
3498 'md' followed by enough consecutive 1-byte PRNG requests
3499 to traverse all of 'state'.
3501 1. When updating 'md_local' (the current thread's copy of 'md')
3502 during PRNG output generation, hash all of the previous
3503 'md_local' value, not just the half used for PRNG output.
3505 2. Make the number of bytes from 'state' included into the hash
3506 independent from the number of PRNG bytes requested.
3508 The first measure alone would be sufficient to avoid
3509 Markku-Juhani's attack. (Actually it had never occurred
3510 to me that the half of 'md_local' used for chaining was the
3511 half from which PRNG output bytes were taken -- I had always
3512 assumed that the secret half would be used.) The second
3513 measure makes sure that additional data from 'state' is never
3514 mixed into 'md_local' in small portions; this heuristically
3515 further strengthens the PRNG.
3518 *) Fix crypto/bn/asm/mips3.s.
3521 *) When only the key is given to "enc", the IV is undefined. Print out
3522 an error message in this case.
3525 *) Handle special case when X509_NAME is empty in X509 printing routines.
3528 *) In dsa_do_verify (crypto/dsa/dsa_ossl.c), verify that r and s are
3529 positive and less than q.
3532 *) Don't change *pointer in CRYPTO_add_lock() is add_lock_callback is
3533 used: it isn't thread safe and the add_lock_callback should handle
3535 [Paul Rose <Paul.Rose@bridge.com>]
3537 *) Verify that incoming data obeys the block size in
3538 ssl3_enc (ssl/s3_enc.c) and tls1_enc (ssl/t1_enc.c).
3542 [Ulf Möller, Bodo Möller]
3544 *) The countermeasure against Bleichbacher's attack on PKCS #1 v1.5
3545 RSA encryption was accidentally removed in s3_srvr.c in OpenSSL 0.9.5
3546 when fixing the server behaviour for backwards-compatible 'client
3547 hello' messages. (Note that the attack is impractical against
3548 SSL 3.0 and TLS 1.0 anyway because length and version checking
3549 means that the probability of guessing a valid ciphertext is
3550 around 2^-40; see section 5 in Bleichenbacher's CRYPTO '98
3553 Before 0.9.5, the countermeasure (hide the error by generating a
3554 random 'decryption result') did not work properly because
3555 ERR_clear_error() was missing, meaning that SSL_get_error() would
3556 detect the supposedly ignored error.
3558 Both problems are now fixed.
3561 *) In crypto/bio/bf_buff.c, increase DEFAULT_BUFFER_SIZE to 4096
3562 (previously it was 1024).
3565 *) Fix for compatibility mode trust settings: ignore trust settings
3566 unless some valid trust or reject settings are present.
3569 *) Fix for blowfish EVP: its a variable length cipher.
3572 *) Fix various bugs related to DSA S/MIME verification. Handle missing
3573 parameters in DSA public key structures and return an error in the
3574 DSA routines if parameters are absent.
3577 *) In versions up to 0.9.6, RAND_file_name() resorted to file ".rnd"
3578 in the current directory if neither $RANDFILE nor $HOME was set.
3579 RAND_file_name() in 0.9.6a returned NULL in this case. This has
3580 caused some confusion to Windows users who haven't defined $HOME.
3581 Thus RAND_file_name() is changed again: e_os.h can define a
3582 DEFAULT_HOME, which will be used if $HOME is not set.
3583 For Windows, we use "C:"; on other platforms, we still require
3584 environment variables.
3586 *) Move 'if (!initialized) RAND_poll()' into regions protected by
3587 CRYPTO_LOCK_RAND. This is not strictly necessary, but avoids
3588 having multiple threads call RAND_poll() concurrently.
3591 *) In crypto/rand/md_rand.c, replace 'add_do_not_lock' flag by a
3592 combination of a flag and a thread ID variable.
3593 Otherwise while one thread is in ssleay_rand_bytes (which sets the
3594 flag), *other* threads can enter ssleay_add_bytes without obeying
3595 the CRYPTO_LOCK_RAND lock (and may even illegally release the lock
3596 that they do not hold after the first thread unsets add_do_not_lock).
3599 *) Change bctest again: '-x' expressions are not available in all
3603 Changes between 0.9.6 and 0.9.6a [5 Apr 2001]
3605 *) Fix a couple of memory leaks in PKCS7_dataDecode()
3606 [Steve Henson, reported by Heyun Zheng <hzheng@atdsprint.com>]
3608 *) Change Configure and Makefiles to provide EXE_EXT, which will contain
3609 the default extension for executables, if any. Also, make the perl
3610 scripts that use symlink() to test if it really exists and use "cp"
3611 if it doesn't. All this made OpenSSL compilable and installable in
3615 *) Fix for asn1_GetSequence() for indefinite length constructed data.
3616 If SEQUENCE is length is indefinite just set c->slen to the total
3617 amount of data available.
3618 [Steve Henson, reported by shige@FreeBSD.org]
3619 [This change does not apply to 0.9.7.]
3621 *) Change bctest to avoid here-documents inside command substitution
3622 (workaround for FreeBSD /bin/sh bug).
3623 For compatibility with Ultrix, avoid shell functions (introduced
3624 in the bctest version that searches along $PATH).
3627 *) Rename 'des_encrypt' to 'des_encrypt1'. This avoids the clashes
3628 with des_encrypt() defined on some operating systems, like Solaris
3632 *) Check the result of RSA-CRT (see D. Boneh, R. DeMillo, R. Lipton:
3633 On the Importance of Eliminating Errors in Cryptographic
3634 Computations, J. Cryptology 14 (2001) 2, 101-119,
3635 http://theory.stanford.edu/~dabo/papers/faults.ps.gz).
3638 *) MIPS assembler BIGNUM division bug fix.
3641 *) Disabled incorrect Alpha assembler code.
3644 *) Fix PKCS#7 decode routines so they correctly update the length
3645 after reading an EOC for the EXPLICIT tag.
3647 [This change does not apply to 0.9.7.]
3649 *) Fix bug in PKCS#12 key generation routines. This was triggered
3650 if a 3DES key was generated with a 0 initial byte. Include
3651 PKCS12_BROKEN_KEYGEN compilation option to retain the old
3652 (but broken) behaviour.
3655 *) Enhance bctest to search for a working bc along $PATH and print
3657 [Tim Rice <tim@multitalents.net> via Richard Levitte]
3659 *) Fix memory leaks in err.c: free err_data string if necessary;
3660 don't write to the wrong index in ERR_set_error_data.
3663 *) Implement ssl23_peek (analogous to ssl23_read), which previously
3667 *) Replace rdtsc with _emit statements for VC++ version 5.
3668 [Jeremy Cooper <jeremy@baymoo.org>]
3670 *) Make it possible to reuse SSLv2 sessions.
3673 *) In copy_email() check for >= 0 as a return value for
3674 X509_NAME_get_index_by_NID() since 0 is a valid index.
3675 [Steve Henson reported by Massimiliano Pala <madwolf@opensca.org>]
3677 *) Avoid coredump with unsupported or invalid public keys by checking if
3678 X509_get_pubkey() fails in PKCS7_verify(). Fix memory leak when
3679 PKCS7_verify() fails with non detached data.
3682 *) Don't use getenv in library functions when run as setuid/setgid.
3683 New function OPENSSL_issetugid().
3686 *) Avoid false positives in memory leak detection code (crypto/mem_dbg.c)
3687 due to incorrect handling of multi-threading:
3689 1. Fix timing glitch in the MemCheck_off() portion of CRYPTO_mem_ctrl().
3691 2. Fix logical glitch in is_MemCheck_on() aka CRYPTO_is_mem_check_on().
3693 3. Count how many times MemCheck_off() has been called so that
3694 nested use can be treated correctly. This also avoids
3695 inband-signalling in the previous code (which relied on the
3696 assumption that thread ID 0 is impossible).
3699 *) Add "-rand" option also to s_client and s_server.
3702 *) Fix CPU detection on Irix 6.x.
3703 [Kurt Hockenbury <khockenb@stevens-tech.edu> and
3704 "Bruce W. Forsberg" <bruce.forsberg@baesystems.com>]
3706 *) Fix X509_NAME bug which produced incorrect encoding if X509_NAME
3709 [This change does not apply to 0.9.7.]
3711 *) Use the cached encoding of an X509_NAME structure rather than
3712 copying it. This is apparently the reason for the libsafe "errors"
3713 but the code is actually correct.
3716 *) Add new function BN_rand_range(), and fix DSA_sign_setup() to prevent
3717 Bleichenbacher's DSA attack.
3718 Extend BN_[pseudo_]rand: As before, top=1 forces the highest two bits
3719 to be set and top=0 forces the highest bit to be set; top=-1 is new
3720 and leaves the highest bit random.
3721 [Ulf Moeller, Bodo Moeller]
3723 *) In the NCONF_...-based implementations for CONF_... queries
3724 (crypto/conf/conf_lib.c), if the input LHASH is NULL, avoid using
3725 a temporary CONF structure with the data component set to NULL
3726 (which gives segmentation faults in lh_retrieve).
3727 Instead, use NULL for the CONF pointer in CONF_get_string and
3728 CONF_get_number (which may use environment variables) and directly
3729 return NULL from CONF_get_section.
3732 *) Fix potential buffer overrun for EBCDIC.
3735 *) Tolerate nonRepudiation as being valid for S/MIME signing and certSign
3736 keyUsage if basicConstraints absent for a CA.
3739 *) Make SMIME_write_PKCS7() write mail header values with a format that
3740 is more generally accepted (no spaces before the semicolon), since
3741 some programs can't parse those values properly otherwise. Also make
3742 sure BIO's that break lines after each write do not create invalid
3746 *) Make the CRL encoding routines work with empty SEQUENCE OF. The
3747 macros previously used would not encode an empty SEQUENCE OF
3748 and break the signature.
3750 [This change does not apply to 0.9.7.]
3752 *) Zero the premaster secret after deriving the master secret in
3756 *) Add some EVP_add_digest_alias registrations (as found in
3757 OpenSSL_add_all_digests()) to SSL_library_init()
3758 aka OpenSSL_add_ssl_algorithms(). This provides improved
3759 compatibility with peers using X.509 certificates
3760 with unconventional AlgorithmIdentifier OIDs.
3763 *) Fix for Irix with NO_ASM.
3764 ["Bruce W. Forsberg" <bruce.forsberg@baesystems.com>]
3766 *) ./config script fixes.
3767 [Ulf Moeller, Richard Levitte]
3769 *) Fix 'openssl passwd -1'.
3772 *) Change PKCS12_key_gen_asc() so it can cope with non null
3773 terminated strings whose length is passed in the passlen
3774 parameter, for example from PEM callbacks. This was done
3775 by adding an extra length parameter to asc2uni().
3776 [Steve Henson, reported by <oddissey@samsung.co.kr>]
3778 *) Fix C code generated by 'openssl dsaparam -C': If a BN_bin2bn
3779 call failed, free the DSA structure.
3782 *) Fix to uni2asc() to cope with zero length Unicode strings.
3783 These are present in some PKCS#12 files.
3786 *) Increase s2->wbuf allocation by one byte in ssl2_new (ssl/s2_lib.c).
3787 Otherwise do_ssl_write (ssl/s2_pkt.c) will write beyond buffer limits
3788 when writing a 32767 byte record.
3789 [Bodo Moeller; problem reported by Eric Day <eday@concentric.net>]
3791 *) In RSA_eay_public_{en,ed}crypt and RSA_eay_mod_exp (rsa_eay.c),
3792 obtain lock CRYPTO_LOCK_RSA before setting rsa->_method_mod_{n,p,q}.
3794 (RSA objects have a reference count access to which is protected
3795 by CRYPTO_LOCK_RSA [see rsa_lib.c, s3_srvr.c, ssl_cert.c, ssl_rsa.c],
3796 so they are meant to be shared between threads.)
3797 [Bodo Moeller, Geoff Thorpe; original patch submitted by
3798 "Reddie, Steven" <Steven.Reddie@ca.com>]
3800 *) Fix a deadlock in CRYPTO_mem_leaks().
3803 *) Use better test patterns in bntest.
3806 *) rand_win.c fix for Borland C.
3809 *) BN_rshift bugfix for n == 0.
3812 *) Add a 'bctest' script that checks for some known 'bc' bugs
3813 so that 'make test' does not abort just because 'bc' is broken.
3816 *) Store verify_result within SSL_SESSION also for client side to
3817 avoid potential security hole. (Re-used sessions on the client side
3818 always resulted in verify_result==X509_V_OK, not using the original
3819 result of the server certificate verification.)
3822 *) Fix ssl3_pending: If the record in s->s3->rrec is not of type
3823 SSL3_RT_APPLICATION_DATA, return 0.
3824 Similarly, change ssl2_pending to return 0 if SSL_in_init(s) is true.
3828 Both ssl2_peek and ssl3_peek, which were totally broken in earlier
3829 releases, have been re-implemented by renaming the previous
3830 implementations of ssl2_read and ssl3_read to ssl2_read_internal
3831 and ssl3_read_internal, respectively, and adding 'peek' parameters
3832 to them. The new ssl[23]_{read,peek} functions are calls to
3833 ssl[23]_read_internal with the 'peek' flag set appropriately.
3834 A 'peek' parameter has also been added to ssl3_read_bytes, which
3835 does the actual work for ssl3_read_internal.
3838 *) Initialise "ex_data" member of RSA/DSA/DH structures prior to calling
3839 the method-specific "init()" handler. Also clean up ex_data after
3840 calling the method-specific "finish()" handler. Previously, this was
3841 happening the other way round.
3844 *) Increase BN_CTX_NUM (the number of BIGNUMs in a BN_CTX) to 16.
3845 The previous value, 12, was not always sufficient for BN_mod_exp().
3848 *) Make sure that shared libraries get the internal name engine with
3849 the full version number and not just 0. This should mark the
3850 shared libraries as not backward compatible. Of course, this should
3851 be changed again when we can guarantee backward binary compatibility.
3854 *) Fix typo in get_cert_by_subject() in by_dir.c
3855 [Jean-Marc Desperrier <jean-marc.desperrier@certplus.com>]
3857 *) Rework the system to generate shared libraries:
3859 - Make note of the expected extension for the shared libraries and
3860 if there is a need for symbolic links from for example libcrypto.so.0
3861 to libcrypto.so.0.9.7. There is extended info in Configure for
3864 - Make as few rebuilds of the shared libraries as possible.
3866 - Still avoid linking the OpenSSL programs with the shared libraries.
3868 - When installing, install the shared libraries separately from the
3872 *) Fix SSL_CTX_set_read_ahead macro to actually use its argument.
3874 Copy SSL_CTX's read_ahead flag to SSL object directly in SSL_new
3875 and not in SSL_clear because the latter is also used by the
3876 accept/connect functions; previously, the settings made by
3877 SSL_set_read_ahead would be lost during the handshake.
3878 [Bodo Moeller; problems reported by Anders Gertz <gertz@epact.se>]
3880 *) Correct util/mkdef.pl to be selective about disabled algorithms.
3881 Previously, it would create entries for disableed algorithms no
3885 *) Added several new manual pages for SSL_* function.
3888 Changes between 0.9.5a and 0.9.6 [24 Sep 2000]
3890 *) In ssl23_get_client_hello, generate an error message when faced
3891 with an initial SSL 3.0/TLS record that is too small to contain the
3892 first two bytes of the ClientHello message, i.e. client_version.
3893 (Note that this is a pathologic case that probably has never happened
3894 in real life.) The previous approach was to use the version number
3895 from the record header as a substitute; but our protocol choice
3896 should not depend on that one because it is not authenticated
3897 by the Finished messages.
3900 *) More robust randomness gathering functions for Windows.
3901 [Jeffrey Altman <jaltman@columbia.edu>]
3903 *) For compatibility reasons if the flag X509_V_FLAG_ISSUER_CHECK is
3904 not set then we don't setup the error code for issuer check errors
3905 to avoid possibly overwriting other errors which the callback does
3906 handle. If an application does set the flag then we assume it knows
3907 what it is doing and can handle the new informational codes
3911 *) Fix for a nasty bug in ASN1_TYPE handling. ASN1_TYPE is used for
3912 a general "ANY" type, as such it should be able to decode anything
3913 including tagged types. However it didn't check the class so it would
3914 wrongly interpret tagged types in the same way as their universal
3915 counterpart and unknown types were just rejected. Changed so that the
3916 tagged and unknown types are handled in the same way as a SEQUENCE:
3917 that is the encoding is stored intact. There is also a new type
3918 "V_ASN1_OTHER" which is used when the class is not universal, in this
3919 case we have no idea what the actual type is so we just lump them all
3923 *) On VMS, stdout may very well lead to a file that is written to
3924 in a record-oriented fashion. That means that every write() will
3925 write a separate record, which will be read separately by the
3926 programs trying to read from it. This can be very confusing.
3928 The solution is to put a BIO filter in the way that will buffer
3929 text until a linefeed is reached, and then write everything a
3930 line at a time, so every record written will be an actual line,
3931 not chunks of lines and not (usually doesn't happen, but I've
3932 seen it once) several lines in one record. BIO_f_linebuffer() is
3935 Currently, it's a VMS-only method, because that's where it has
3936 been tested well enough.
3939 *) Remove 'optimized' squaring variant in BN_mod_mul_montgomery,
3940 it can return incorrect results.
3941 (Note: The buggy variant was not enabled in OpenSSL 0.9.5a,
3942 but it was in 0.9.6-beta[12].)
3945 *) Disable the check for content being present when verifying detached
3946 signatures in pk7_smime.c. Some versions of Netscape (wrongly)
3947 include zero length content when signing messages.
3950 *) New BIO_shutdown_wr macro, which invokes the BIO_C_SHUTDOWN_WR
3951 BIO_ctrl (for BIO pairs).
3954 *) Add DSO method for VMS.
3957 *) Bug fix: Montgomery multiplication could produce results with the
3961 *) Add RPM specification openssl.spec and modify it to build three
3962 packages. The default package contains applications, application
3963 documentation and run-time libraries. The devel package contains
3964 include files, static libraries and function documentation. The
3965 doc package contains the contents of the doc directory. The original
3966 openssl.spec was provided by Damien Miller <djm@mindrot.org>.
3969 *) Add a large number of documentation files for many SSL routines.
3970 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>]
3972 *) Add a configuration entry for Sony News 4.
3973 [NAKAJI Hiroyuki <nakaji@tutrp.tut.ac.jp>]
3975 *) Don't set the two most significant bits to one when generating a
3976 random number < q in the DSA library.
3979 *) New SSL API mode 'SSL_MODE_AUTO_RETRY'. This disables the default
3980 behaviour that SSL_read may result in SSL_ERROR_WANT_READ (even if
3981 the underlying transport is blocking) if a handshake took place.
3982 (The default behaviour is needed by applications such as s_client
3983 and s_server that use select() to determine when to use SSL_read;
3984 but for applications that know in advance when to expect data, it
3985 just makes things more complicated.)
3988 *) Add RAND_egd_bytes(), which gives control over the number of bytes read
3992 *) Add a few more EBCDIC conditionals that make `req' and `x509'
3993 work better on such systems.
3994 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>]
3996 *) Add two demo programs for PKCS12_parse() and PKCS12_create().
3997 Update PKCS12_parse() so it copies the friendlyName and the
3998 keyid to the certificates aux info.
4001 *) Fix bug in PKCS7_verify() which caused an infinite loop
4002 if there was more than one signature.
4003 [Sven Uszpelkat <su@celocom.de>]
4005 *) Major change in util/mkdef.pl to include extra information
4006 about each symbol, as well as presentig variables as well
4007 as functions. This change means that there's n more need
4008 to rebuild the .num files when some algorithms are excluded.
4011 *) Allow the verify time to be set by an application,
4012 rather than always using the current time.
4015 *) Phase 2 verify code reorganisation. The certificate
4016 verify code now looks up an issuer certificate by a
4017 number of criteria: subject name, authority key id
4018 and key usage. It also verifies self signed certificates
4019 by the same criteria. The main comparison function is
4020 X509_check_issued() which performs these checks.
4022 Lot of changes were necessary in order to support this
4023 without completely rewriting the lookup code.
4025 Authority and subject key identifier are now cached.
4027 The LHASH 'certs' is X509_STORE has now been replaced
4028 by a STACK_OF(X509_OBJECT). This is mainly because an
4029 LHASH can't store or retrieve multiple objects with
4030 the same hash value.
4032 As a result various functions (which were all internal
4033 use only) have changed to handle the new X509_STORE
4034 structure. This will break anything that messed round
4035 with X509_STORE internally.
4037 The functions X509_STORE_add_cert() now checks for an
4038 exact match, rather than just subject name.
4040 The X509_STORE API doesn't directly support the retrieval
4041 of multiple certificates matching a given criteria, however
4042 this can be worked round by performing a lookup first
4043 (which will fill the cache with candidate certificates)
4044 and then examining the cache for matches. This is probably
4045 the best we can do without throwing out X509_LOOKUP
4046 entirely (maybe later...).
4048 The X509_VERIFY_CTX structure has been enhanced considerably.
4050 All certificate lookup operations now go via a get_issuer()
4051 callback. Although this currently uses an X509_STORE it
4052 can be replaced by custom lookups. This is a simple way
4053 to bypass the X509_STORE hackery necessary to make this
4054 work and makes it possible to use more efficient techniques
4055 in future. A very simple version which uses a simple
4056 STACK for its trusted certificate store is also provided
4057 using X509_STORE_CTX_trusted_stack().
4059 The verify_cb() and verify() callbacks now have equivalents
4060 in the X509_STORE_CTX structure.
4062 X509_STORE_CTX also has a 'flags' field which can be used
4063 to customise the verify behaviour.
4066 *) Add new PKCS#7 signing option PKCS7_NOSMIMECAP which
4067 excludes S/MIME capabilities.
4070 *) When a certificate request is read in keep a copy of the
4071 original encoding of the signed data and use it when outputing
4072 again. Signatures then use the original encoding rather than
4073 a decoded, encoded version which may cause problems if the
4074 request is improperly encoded.
4077 *) For consistency with other BIO_puts implementations, call
4078 buffer_write(b, ...) directly in buffer_puts instead of calling
4081 In BIO_puts, increment b->num_write as in BIO_write.
4082 [Peter.Sylvester@EdelWeb.fr]
4084 *) Fix BN_mul_word for the case where the word is 0. (We have to use
4085 BN_zero, we may not return a BIGNUM with an array consisting of
4089 *) Avoid calling abort() from within the library when problems are
4090 detected, except if preprocessor symbols have been defined
4091 (such as REF_CHECK, BN_DEBUG etc.).
4094 *) New openssl application 'rsautl'. This utility can be
4095 used for low level RSA operations. DER public key
4096 BIO/fp routines also added.
4099 *) New Configure entry and patches for compiling on QNX 4.
4100 [Andreas Schneider <andreas@ds3.etech.fh-hamburg.de>]
4102 *) A demo state-machine implementation was sponsored by
4103 Nuron (http://www.nuron.com/) and is now available in
4104 demos/state_machine.
4107 *) New options added to the 'dgst' utility for signature
4108 generation and verification.
4111 *) Unrecognized PKCS#7 content types are now handled via a
4112 catch all ASN1_TYPE structure. This allows unsupported
4113 types to be stored as a "blob" and an application can
4114 encode and decode it manually.
4117 *) Fix various signed/unsigned issues to make a_strex.c
4119 [Oscar Jacobsson <oscar.jacobsson@celocom.com>]
4121 *) ASN1 fixes. i2d_ASN1_OBJECT was not returning the correct
4122 length if passed a buffer. ASN1_INTEGER_to_BN failed
4123 if passed a NULL BN and its argument was negative.
4124 [Steve Henson, pointed out by Sven Heiberg <sven@tartu.cyber.ee>]
4126 *) Modification to PKCS#7 encoding routines to output definite
4127 length encoding. Since currently the whole structures are in
4128 memory there's not real point in using indefinite length
4129 constructed encoding. However if OpenSSL is compiled with
4130 the flag PKCS7_INDEFINITE_ENCODING the old form is used.
4133 *) Added BIO_vprintf() and BIO_vsnprintf().
4136 *) Added more prefixes to parse for in the the strings written
4137 through a logging bio, to cover all the levels that are available
4138 through syslog. The prefixes are now:
4140 PANIC, EMERG, EMR => LOG_EMERG
4141 ALERT, ALR => LOG_ALERT
4142 CRIT, CRI => LOG_CRIT
4143 ERROR, ERR => LOG_ERR
4144 WARNING, WARN, WAR => LOG_WARNING
4145 NOTICE, NOTE, NOT => LOG_NOTICE
4146 INFO, INF => LOG_INFO
4147 DEBUG, DBG => LOG_DEBUG
4149 and as before, if none of those prefixes are present at the
4150 beginning of the string, LOG_ERR is chosen.
4152 On Win32, the LOG_* levels are mapped according to this:
4154 LOG_EMERG, LOG_ALERT, LOG_CRIT, LOG_ERR => EVENTLOG_ERROR_TYPE
4155 LOG_WARNING => EVENTLOG_WARNING_TYPE
4156 LOG_NOTICE, LOG_INFO, LOG_DEBUG => EVENTLOG_INFORMATION_TYPE
4160 *) Made it possible to reconfigure with just the configuration
4161 argument "reconf" or "reconfigure". The command line arguments
4162 are stored in Makefile.ssl in the variable CONFIGURE_ARGS,
4163 and are retrieved from there when reconfiguring.
4167 [Assar Westerlund <assar@sics.se>, Richard Levitte]
4169 *) Add the arguments -CAfile and -CApath to the pkcs12 utility.
4172 *) The obj_dat.pl script was messing up the sorting of object
4173 names. The reason was that it compared the quoted version
4174 of strings as a result "OCSP" > "OCSP Signing" because
4175 " > SPACE. Changed script to store unquoted versions of
4176 names and add quotes on output. It was also omitting some
4177 names from the lookup table if they were given a default
4178 value (that is if SN is missing it is given the same
4179 value as LN and vice versa), these are now added on the
4180 grounds that if an object has a name we should be able to
4181 look it up. Finally added warning output when duplicate
4182 short or long names are found.
4185 *) Changes needed for Tandem NSK.
4186 [Scott Uroff <scott@xypro.com>]
4188 *) Fix SSL 2.0 rollback checking: Due to an off-by-one error in
4189 RSA_padding_check_SSLv23(), special padding was never detected
4190 and thus the SSL 3.0/TLS 1.0 countermeasure against protocol
4191 version rollback attacks was not effective.
4193 In s23_clnt.c, don't use special rollback-attack detection padding
4194 (RSA_SSLV23_PADDING) if SSL 2.0 is the only protocol enabled in the
4195 client; similarly, in s23_srvr.c, don't do the rollback check if
4196 SSL 2.0 is the only protocol enabled in the server.
4199 *) Make it possible to get hexdumps of unprintable data with 'openssl
4200 asn1parse'. By implication, the functions ASN1_parse_dump() and
4201 BIO_dump_indent() are added.
4204 *) New functions ASN1_STRING_print_ex() and X509_NAME_print_ex()
4205 these print out strings and name structures based on various
4206 flags including RFC2253 support and proper handling of
4207 multibyte characters. Added options to the 'x509' utility
4208 to allow the various flags to be set.
4211 *) Various fixes to use ASN1_TIME instead of ASN1_UTCTIME.
4212 Also change the functions X509_cmp_current_time() and
4213 X509_gmtime_adj() work with an ASN1_TIME structure,
4214 this will enable certificates using GeneralizedTime in validity
4215 dates to be checked.
4218 *) Make the NEG_PUBKEY_BUG code (which tolerates invalid
4219 negative public key encodings) on by default,
4220 NO_NEG_PUBKEY_BUG can be set to disable it.
4223 *) New function c2i_ASN1_OBJECT() which acts on ASN1_OBJECT
4224 content octets. An i2c_ASN1_OBJECT is unnecessary because
4225 the encoding can be trivially obtained from the structure.
4228 *) crypto/err.c locking bugfix: Use write locks (CRYPTO_w_[un]lock),
4229 not read locks (CRYPTO_r_[un]lock).
4232 *) A first attempt at creating official support for shared
4233 libraries through configuration. I've kept it so the
4234 default is static libraries only, and the OpenSSL programs
4235 are always statically linked for now, but there are
4236 preparations for dynamic linking in place.
4237 This has been tested on Linux and Tru64.
4240 *) Randomness polling function for Win9x, as described in:
4241 Peter Gutmann, Software Generation of Practically Strong
4245 *) Fix so PRNG is seeded in req if using an already existing
4249 *) New options to smime application. -inform and -outform
4250 allow alternative formats for the S/MIME message including
4251 PEM and DER. The -content option allows the content to be
4252 specified separately. This should allow things like Netscape
4253 form signing output easier to verify.
4256 *) Fix the ASN1 encoding of tags using the 'long form'.
4259 *) New ASN1 functions, i2c_* and c2i_* for INTEGER and BIT
4260 STRING types. These convert content octets to and from the
4261 underlying type. The actual tag and length octets are
4262 already assumed to have been read in and checked. These
4263 are needed because all other string types have virtually
4264 identical handling apart from the tag. By having versions
4265 of the ASN1 functions that just operate on content octets
4266 IMPLICIT tagging can be handled properly. It also allows
4267 the ASN1_ENUMERATED code to be cut down because ASN1_ENUMERATED
4268 and ASN1_INTEGER are identical apart from the tag.
4271 *) Change the handling of OID objects as follows:
4273 - New object identifiers are inserted in objects.txt, following
4274 the syntax given in objects.README.
4275 - objects.pl is used to process obj_mac.num and create a new
4277 - obj_dat.pl is used to create a new obj_dat.h, using the data in
4280 This is currently kind of a hack, and the perl code in objects.pl
4281 isn't very elegant, but it works as I intended. The simplest way
4282 to check that it worked correctly is to look in obj_dat.h and
4283 check the array nid_objs and make sure the objects haven't moved
4284 around (this is important!). Additions are OK, as well as
4285 consistent name changes.
4288 *) Add BSD-style MD5-based passwords to 'openssl passwd' (option '-1').
4291 *) Addition of the command line parameter '-rand file' to 'openssl req'.
4292 The given file adds to whatever has already been seeded into the
4293 random pool through the RANDFILE configuration file option or
4294 environment variable, or the default random state file.
4297 *) mkstack.pl now sorts each macro group into lexical order.
4298 Previously the output order depended on the order the files
4299 appeared in the directory, resulting in needless rewriting
4303 *) Patches to make OpenSSL compile under Win32 again. Mostly
4304 work arounds for the VC++ problem that it treats func() as
4305 func(void). Also stripped out the parts of mkdef.pl that
4306 added extra typesafe functions: these no longer exist.
4309 *) Reorganisation of the stack code. The macros are now all
4310 collected in safestack.h . Each macro is defined in terms of
4311 a "stack macro" of the form SKM_<name>(type, a, b). The
4312 DEBUG_SAFESTACK is now handled in terms of function casts,
4313 this has the advantage of retaining type safety without the
4314 use of additional functions. If DEBUG_SAFESTACK is not defined
4315 then the non typesafe macros are used instead. Also modified the
4316 mkstack.pl script to handle the new form. Needs testing to see
4317 if which (if any) compilers it chokes and maybe make DEBUG_SAFESTACK
4318 the default if no major problems. Similar behaviour for ASN1_SET_OF
4319 and PKCS12_STACK_OF.
4322 *) When some versions of IIS use the 'NET' form of private key the
4323 key derivation algorithm is different. Normally MD5(password) is
4324 used as a 128 bit RC4 key. In the modified case
4325 MD5(MD5(password) + "SGCKEYSALT") is used insted. Added some
4326 new functions i2d_RSA_NET(), d2i_RSA_NET() etc which are the same
4327 as the old Netscape_RSA functions except they have an additional
4328 'sgckey' parameter which uses the modified algorithm. Also added
4329 an -sgckey command line option to the rsa utility. Thanks to
4330 Adrian Peck <bertie@ncipher.com> for posting details of the modified
4331 algorithm to openssl-dev.
4334 *) The evp_local.h macros were using 'c.##kname' which resulted in
4335 invalid expansion on some systems (SCO 5.0.5 for example).
4336 Corrected to 'c.kname'.
4337 [Phillip Porch <root@theporch.com>]
4339 *) New X509_get1_email() and X509_REQ_get1_email() functions that return
4340 a STACK of email addresses from a certificate or request, these look
4341 in the subject name and the subject alternative name extensions and
4342 omit any duplicate addresses.
4345 *) Re-implement BN_mod_exp2_mont using independent (and larger) windows.
4346 This makes DSA verification about 2 % faster.
4349 *) Increase maximum window size in BN_mod_exp_... to 6 bits instead of 5
4350 (meaning that now 2^5 values will be precomputed, which is only 4 KB
4351 plus overhead for 1024 bit moduli).
4352 This makes exponentiations about 0.5 % faster for 1024 bit
4353 exponents (as measured by "openssl speed rsa2048").
4356 *) Rename memory handling macros to avoid conflicts with other
4358 Malloc => OPENSSL_malloc
4359 Malloc_locked => OPENSSL_malloc_locked
4360 Realloc => OPENSSL_realloc
4361 Free => OPENSSL_free
4364 *) New function BN_mod_exp_mont_word for small bases (roughly 15%
4365 faster than BN_mod_exp_mont, i.e. 7% for a full DH exchange).
4368 *) CygWin32 support.
4369 [John Jarvie <jjarvie@newsguy.com>]
4371 *) The type-safe stack code has been rejigged. It is now only compiled
4372 in when OpenSSL is configured with the DEBUG_SAFESTACK option and
projects / openssl.git / blob