5 Changes between 0.9.7c and 0.9.8 [xx XXX xxxx]
7 *) Add the possibility to load symbols globally with DSO.
8 [Götz Babin-Ebell <babin-ebell@trustcenter.de> via Richard Levitte]
10 *) Add the functions ERR_set_mark() and ERR_pop_to_mark() for better
11 control of the error stack.
14 *) Add support for STORE in ENGINE.
17 *) Add the STORE type. The intention is to provide a common interface
18 to certificate and key stores, be they simple file-based stores, or
19 HSM-type store, or LDAP stores, or...
20 NOTE: The code is currently UNTESTED and isn't really used anywhere.
23 *) Add a generic structure called OPENSSL_ITEM. This can be used to
24 pass a list of arguments to any function as well as provide a way
25 for a function to pass data back to the caller.
28 *) Add the functions BUF_strndup() and BUF_memdup(). BUF_strndup()
29 works like BUF_strdup() but can be used to duplicate a portion of
30 a string. The copy gets NUL-terminated. BUF_memdup() duplicates
34 *) Add the function sk_find_ex() which works like sk_find(), but will
35 return an index to an element even if an exact match couldn't be
36 found. The index is guaranteed to point at the element where the
37 searched-for key would be inserted to preserve sorting order.
40 *) Add the function OBJ_bsearch_ex() which works like OBJ_bsearch() but
41 takes an extra flags argument for optional functionality. Currently,
42 the following flags are defined:
44 OBJ_BSEARCH_VALUE_ON_NOMATCH
45 This one gets OBJ_bsearch_ex() to return a pointer to the first
46 element where the comparing function returns a negative or zero
49 OBJ_BSEARCH_FIRST_VALUE_ON_MATCH
50 This one gets OBJ_bsearch_ex() to return a pointer to the first
51 element where the comparing function returns zero. This is useful
52 if there are more than one element where the comparing function
56 *) Make it possible to create self-signed certificates with 'openssl ca'
57 in such a way that the self-signed certificate becomes part of the
58 CA database and uses the same mechanisms for serial number generation
59 as all other certificate signing. The new flag '-selfsign' enables
60 this functionality. Adapt CA.sh and CA.pl.in.
63 *) Add functionality to check the public key of a certificate request
64 against a given private. This is useful to check that a certificate
65 request can be signed by that key (self-signing).
68 *) Make it possible to have multiple active certificates with the same
69 subject in the CA index file. This is done only if the keyword
70 'unique_subject' is set to 'no' in the main CA section (default
71 if 'CA_default') of the configuration file. The value is saved
72 with the database itself in a separate index attribute file,
73 named like the index file with '.attr' appended to the name.
76 *) Generate muti valued AVAs using '+' notation in config files for
80 *) Support for nameConstraints certificate extension.
83 *) Support for policyConstraints certificate extension.
86 *) Support for policyMappings certificate extension.
89 *) Fixed a typo bug that would cause ENGINE_set_default() to set an
90 ENGINE as defaults for all supported algorithms irrespective of
91 the 'flags' parameter. 'flags' is now honoured, so applications
92 should make sure they are passing it correctly.
95 *) Make sure the default DSA_METHOD implementation only uses its
96 dsa_mod_exp() and/or bn_mod_exp() handlers if they are non-NULL,
97 and change its own handlers to be NULL so as to remove unnecessary
98 indirection. This lets alternative implementations fallback to the
99 default implementation more easily.
102 *) Support for directoryName in GeneralName related extensions
106 *) Make it possible to link applications using Makefile.shared.
107 Make that possible even when linking against static libraries!
110 *) Support for single pass processing for S/MIME signing. This now
111 means that S/MIME signing can be done from a pipe, in addition
112 cleartext signing (multipart/signed type) is effectively streaming
113 and the signed data does not need to be all held in memory.
115 This is done with a new flag PKCS7_STREAM. When this flag is set
116 PKCS7_sign() only initializes the PKCS7 structure and the actual signing
117 is done after the data is output (and digests calculated) in
121 *) Add full support for -rpath/-R, both in shared libraries and
122 applications, at least on the platforms where it's known how
126 *) In crypto/ec/ec_mult.c, implement fast point multiplication with
127 precomputation, based on wNAF splitting: EC_GROUP_precompute_mult()
128 will now compute a table of multiples of the generator that
129 makes subsequent invocations of EC_POINTs_mul() or EC_POINT_mul()
130 faster (notably in the case of a single point multiplication,
132 [Nils Larsch, Bodo Moeller]
134 *) IPv6 support for certificate extensions. The various extensions
135 which use the IP:a.b.c.d can now take IPv6 addresses using the
136 formats of RFC1884 2.2 . IPv6 addresses are now also displayed
140 *) Added an ENGINE that implements RSA by performing private key
141 exponentiations with the GMP library. The conversions to and from
142 GMP's mpz_t format aren't optimised nor are any montgomery forms
143 cached, and on x86 it appears OpenSSL's own performance has caught up.
144 However there are likely to be other architectures where GMP could
145 provide a boost. This ENGINE is not built in by default, but it can be
146 specified at Configure time and should be accompanied by the necessary
147 linker additions, eg;
148 ./config -DOPENSSL_USE_GMP -lgmp
151 *) "openssl engine" will not display ENGINE/DSO load failure errors when
152 testing availability of engines with "-t" - the old behaviour is
153 produced by increasing the feature's verbosity with "-tt".
156 *) ECDSA routines: under certain error conditions uninitialized BN objects
157 could be freed. Solution: make sure initialization is performed early
158 enough. (Reported and fix supplied by Nils Larsch <nla@trustcenter.de>
162 *) Key-generation can now be implemented in RSA_METHOD, DSA_METHOD
163 and DH_METHOD (eg. by ENGINE implementations) to override the normal
164 software implementations. For DSA and DH, parameter generation can
165 also be overriden by providing the appropriate method callbacks.
168 *) Change the "progress" mechanism used in key-generation and
169 primality testing to functions that take a new BN_GENCB pointer in
170 place of callback/argument pairs. The new API functions have "_ex"
171 postfixes and the older functions are reimplemented as wrappers for
172 the new ones. The OPENSSL_NO_DEPRECATED symbol can be used to hide
173 declarations of the old functions to help (graceful) attempts to
174 migrate to the new functions. Also, the new key-generation API
175 functions operate on a caller-supplied key-structure and return
176 success/failure rather than returning a key or NULL - this is to
177 help make "keygen" another member function of RSA_METHOD etc.
179 Example for using the new callback interface:
181 int (*my_callback)(int a, int b, BN_GENCB *cb) = ...;
185 BN_GENCB_set(&my_cb, my_callback, my_arg);
187 return BN_is_prime_ex(some_bignum, BN_prime_checks, NULL, &cb);
188 /* For the meaning of a, b in calls to my_callback(), see the
189 * documentation of the function that calls the callback.
190 * cb will point to my_cb; my_arg can be retrieved as cb->arg.
191 * my_callback should return 1 if it wants BN_is_prime_ex()
192 * to continue, or 0 to stop.
197 *) Change the ZLIB compression method to be stateful, and make it
198 available to TLS with the number defined in
199 draft-ietf-tls-compression-04.txt.
202 *) Add the ASN.1 structures and functions for CertificatePair, which
203 is defined as follows (according to X.509_4thEditionDraftV6.pdf):
205 CertificatePair ::= SEQUENCE {
206 forward [0] Certificate OPTIONAL,
207 reverse [1] Certificate OPTIONAL,
208 -- at least one of the pair shall be present -- }
210 Also implement the PEM functions to read and write certificate
211 pairs, and defined the PEM tag as "CERTIFICATE PAIR".
213 This needed to be defined, mostly for the sake of the LDAP
214 attribute crossCertificatePair, but may prove useful elsewhere as
218 *) Make it possible to inhibit symlinking of shared libraries in
219 Makefile.shared, for Cygwin's sake.
222 *) Extend the BIGNUM API by creating new macros that behave like
225 void BN_set_sign(BIGNUM *a, int neg);
226 int BN_get_sign(const BIGNUM *a);
228 and avoid the need to access 'a->neg' directly in applications.
229 [Nils Larsch <nla@trustcenter.de>]
231 *) Implement fast modular reduction for pseudo-Mersenne primes
232 used in NIST curves (crypto/bn/bn_nist.c, crypto/ec/ecp_nist.c).
233 EC_GROUP_new_curve_GFp() will now automatically use this
235 [Nils Larsch <nla@trustcenter.de>]
237 *) Add new lock type (CRYPTO_LOCK_BN).
240 *) Change the ENGINE framework to automatically load engines
241 dynamically from specific directories unless they could be
242 found to already be built in or loaded. Move all the
243 current engines except for the cryptodev one to a new
245 The engines in engines/ are built as shared libraries if
246 the "shared" options was given to ./Configure or ./config.
247 Otherwise, they are inserted in libcrypto.a.
248 /usr/local/ssl/engines is the default directory for dynamic
249 engines, but that can be overriden at configure time through
250 the usual use of --prefix and/or --openssldir, and at run
251 time with the environment variable OPENSSL_ENGINES.
252 [Geoff Thorpe and Richard Levitte]
254 *) Add Makefile.shared, a helper makefile to build shared
255 libraries. Addapt Makefile.org.
258 *) Add version info to Win32 DLLs.
259 [Peter 'Luna' Runestig" <peter@runestig.com>]
261 *) Add new 'medium level' PKCS#12 API. Certificates and keys
262 can be added using this API to created arbitrary PKCS#12
263 files while avoiding the low level API.
265 New options to PKCS12_create(), key or cert can be NULL and
266 will then be omitted from the output file. The encryption
267 algorithm NIDs can be set to -1 for no encryption, the mac
268 iteration count can be set to 0 to omit the mac.
270 Enhance pkcs12 utility by making the -nokeys and -nocerts
271 options work when creating a PKCS#12 file. New option -nomac
272 to omit the mac, NONE can be set for an encryption algorithm.
273 New code is modified to use the enhanced PKCS12_create()
274 instead of the low level API.
277 *) Extend ASN1 encoder to support indefinite length constructed
278 encoding. This can output sequences tags and octet strings in
279 this form. Modify pk7_asn1.c to support indefinite length
280 encoding. This is experimental and needs additional code to
281 be useful, such as an ASN1 bio and some enhanced streaming
284 Extend template encode functionality so that tagging is passed
285 down to the template encoder.
288 *) Let 'openssl req' fail if an argument to '-newkey' is not
289 recognized instead of using RSA as a default.
292 *) Add support for ECC-based ciphersuites from draft-ietf-tls-ecc-01.txt.
293 As these are not official, they are not included in "ALL";
294 the "ECCdraft" ciphersuite group alias can be used to select them.
295 [Vipul Gupta and Sumit Gupta (Sun Microsystems Laboratories)]
297 *) Add ECDH engine support.
298 [Nils Gura and Douglas Stebila (Sun Microsystems Laboratories)]
300 *) Add ECDH in new directory crypto/ecdh/.
301 [Douglas Stebila (Sun Microsystems Laboratories)]
303 *) Let BN_rand_range() abort with an error after 100 iterations
304 without success (which indicates a broken PRNG).
307 *) Change BN_mod_sqrt() so that it verifies that the input value
308 is really the square of the return value. (Previously,
309 BN_mod_sqrt would show GIGO behaviour.)
312 *) Add named elliptic curves over binary fields from X9.62, SECG,
313 and WAP/WTLS; add OIDs that were still missing.
315 [Sheueling Chang Shantz and Douglas Stebila
316 (Sun Microsystems Laboratories)]
318 *) Extend the EC library for elliptic curves over binary fields
319 (new files ec2_smpl.c, ec2_smpt.c, ec2_mult.c in crypto/ec/).
322 EC_GF2m_simple_method
326 EC_GROUP_new_curve_GF2m
327 EC_GROUP_set_curve_GF2m
328 EC_GROUP_get_curve_GF2m
329 EC_POINT_set_affine_coordinates_GF2m
330 EC_POINT_get_affine_coordinates_GF2m
331 EC_POINT_set_compressed_coordinates_GF2m
333 Point compression for binary fields is disabled by default for
334 patent reasons (compile with OPENSSL_EC_BIN_PT_COMP defined to
337 As binary polynomials are represented as BIGNUMs, various members
338 of the EC_GROUP and EC_POINT data structures can be shared
339 between the implementations for prime fields and binary fields;
340 the above ..._GF2m functions (except for EX_GROUP_new_curve_GF2m)
341 are essentially identical to their ..._GFp counterparts.
342 (For simplicity, the '..._GFp' prefix has been dropped from
343 various internal method names.)
345 An internal 'field_div' method (similar to 'field_mul' and
346 'field_sqr') has been added; this is used only for binary fields.
348 [Sheueling Chang Shantz and Douglas Stebila
349 (Sun Microsystems Laboratories)]
351 *) Optionally dispatch EC_POINT_mul(), EC_POINT_precompute_mult()
352 through methods ('mul', 'precompute_mult').
354 The generic implementations (now internally called 'ec_wNAF_mul'
355 and 'ec_wNAF_precomputed_mult') remain the default if these
356 methods are undefined.
358 [Sheueling Chang Shantz and Douglas Stebila
359 (Sun Microsystems Laboratories)]
361 *) New function EC_GROUP_get_degree, which is defined through
362 EC_METHOD. For curves over prime fields, this returns the bit
363 length of the modulus.
365 [Sheueling Chang Shantz and Douglas Stebila
366 (Sun Microsystems Laboratories)]
368 *) New functions EC_GROUP_dup, EC_POINT_dup.
369 (These simply call ..._new and ..._copy).
371 [Sheueling Chang Shantz and Douglas Stebila
372 (Sun Microsystems Laboratories)]
374 *) Add binary polynomial arithmetic software in crypto/bn/bn_gf2m.c.
375 Polynomials are represented as BIGNUMs (where the sign bit is not
376 used) in the following functions [macros]:
379 BN_GF2m_sub [= BN_GF2m_add]
380 BN_GF2m_mod [wrapper for BN_GF2m_mod_arr]
381 BN_GF2m_mod_mul [wrapper for BN_GF2m_mod_mul_arr]
382 BN_GF2m_mod_sqr [wrapper for BN_GF2m_mod_sqr_arr]
384 BN_GF2m_mod_exp [wrapper for BN_GF2m_mod_exp_arr]
385 BN_GF2m_mod_sqrt [wrapper for BN_GF2m_mod_sqrt_arr]
386 BN_GF2m_mod_solve_quad [wrapper for BN_GF2m_mod_solve_quad_arr]
387 BN_GF2m_cmp [= BN_ucmp]
389 (Note that only the 'mod' functions are actually for fields GF(2^m).
390 BN_GF2m_add() is misnomer, but this is for the sake of consistency.)
392 For some functions, an the irreducible polynomial defining a
393 field can be given as an 'unsigned int[]' with strictly
394 decreasing elements giving the indices of those bits that are set;
395 i.e., p[] represents the polynomial
396 f(t) = t^p[0] + t^p[1] + ... + t^p[k]
398 p[0] > p[1] > ... > p[k] = 0.
399 This applies to the following functions:
404 BN_GF2m_mod_inv_arr [wrapper for BN_GF2m_mod_inv]
405 BN_GF2m_mod_div_arr [wrapper for BN_GF2m_mod_div]
408 BN_GF2m_mod_solve_quad_arr
412 Conversion can be performed by the following functions:
417 bntest.c has additional tests for binary polynomial arithmetic.
419 Two implementations for BN_GF2m_mod_div() are available.
420 The default algorithm simply uses BN_GF2m_mod_inv() and
421 BN_GF2m_mod_mul(). The alternative algorithm is compiled in only
422 if OPENSSL_SUN_GF2M_DIV is defined (patent pending; read the
423 copyright notice in crypto/bn/bn_gf2m.c before enabling it).
425 [Sheueling Chang Shantz and Douglas Stebila
426 (Sun Microsystems Laboratories)]
428 *) Add new error code 'ERR_R_DISABLED' that can be used when some
429 functionality is disabled at compile-time.
430 [Douglas Stebila <douglas.stebila@sun.com>]
432 *) Change default behaviour of 'openssl asn1parse' so that more
433 information is visible when viewing, e.g., a certificate:
435 Modify asn1_parse2 (crypto/asn1/asn1_par.c) so that in non-'dump'
436 mode the content of non-printable OCTET STRINGs is output in a
437 style similar to INTEGERs, but with '[HEX DUMP]' prepended to
438 avoid the appearance of a printable string.
439 [Nils Larsch <nla@trustcenter.de>]
441 *) Add 'asn1_flag' and 'asn1_form' member to EC_GROUP with access
443 EC_GROUP_set_asn1_flag()
444 EC_GROUP_get_asn1_flag()
445 EC_GROUP_set_point_conversion_form()
446 EC_GROUP_get_point_conversion_form()
447 These control ASN1 encoding details:
448 - Curves (i.e., groups) are encoded explicitly unless asn1_flag
449 has been set to OPENSSL_EC_NAMED_CURVE.
450 - Points are encoded in uncompressed form by default; options for
451 asn1_for are as for point2oct, namely
452 POINT_CONVERSION_COMPRESSED
453 POINT_CONVERSION_UNCOMPRESSED
454 POINT_CONVERSION_HYBRID
456 Also add 'seed' and 'seed_len' members to EC_GROUP with access
460 EC_GROUP_get_seed_len()
461 This is used only for ASN1 purposes (so far).
462 [Nils Larsch <nla@trustcenter.de>]
464 *) Add 'field_type' member to EC_METHOD, which holds the NID
465 of the appropriate field type OID. The new function
466 EC_METHOD_get_field_type() returns this value.
467 [Nils Larsch <nla@trustcenter.de>]
474 providing useful interfaces to EC_POINT_point2oct() and
475 EC_POINT_oct2point().
476 [Nils Larsch <nla@trustcenter.de>]
478 *) Change internals of the EC library so that the functions
479 EC_GROUP_set_generator()
480 EC_GROUP_get_generator()
482 EC_GROUP_get_cofactor()
483 are implemented directly in crypto/ec/ec_lib.c and not dispatched
484 to methods, which would lead to unnecessary code duplication when
485 adding different types of curves.
486 [Nils Larsch <nla@trustcenter.de> with input by Bodo Moeller]
488 *) Implement compute_wNAF (crypto/ec/ec_mult.c) without BIGNUM
489 arithmetic, and such that modified wNAFs are generated
490 (which avoid length expansion in many cases).
493 *) Add a function EC_GROUP_check_discriminant() (defined via
494 EC_METHOD) that verifies that the curve discriminant is non-zero.
496 Add a function EC_GROUP_check() that makes some sanity tests
497 on a EC_GROUP, its generator and order. This includes
498 EC_GROUP_check_discriminant().
499 [Nils Larsch <nla@trustcenter.de>]
501 *) Add ECDSA in new directory crypto/ecdsa/.
503 Add applications 'openssl ecparam' and 'openssl ecdsa'
504 (these are based on 'openssl dsaparam' and 'openssl dsa').
506 ECDSA support is also included in various other files across the
507 library. Most notably,
508 - 'openssl req' now has a '-newkey ecdsa:file' option;
509 - EVP_PKCS82PKEY (crypto/evp/evp_pkey.c) now can handle ECDSA;
510 - X509_PUBKEY_get (crypto/asn1/x_pubkey.c) and
511 d2i_PublicKey (crypto/asn1/d2i_pu.c) have been modified to make
512 them suitable for ECDSA where domain parameters must be
513 extracted before the specific public key;
514 - ECDSA engine support has been added.
515 [Nils Larsch <nla@trustcenter.de>]
517 *) Include some named elliptic curves, and add OIDs from X9.62,
518 SECG, and WAP/WTLS. Each curve can be obtained from the new
520 EC_GROUP_new_by_nid(),
521 and the list of available named curves can be obtained with
522 EC_get_builtin_curves().
523 Also add a 'curve_name' member to EC_GROUP objects, which can be
527 [Nils Larsch <nla@trustcenter.de, Bodo Moeller]
529 *) Remove a few calls to bn_wexpand() in BN_sqr() (the one in there
530 was actually never needed) and in BN_mul(). The removal in BN_mul()
531 required a small change in bn_mul_part_recursive() and the addition
532 of the functions bn_cmp_part_words(), bn_sub_part_words() and
533 bn_add_part_words(), which do the same thing as bn_cmp_words(),
534 bn_sub_words() and bn_add_words() except they take arrays with
538 Changes between 0.9.7b and 0.9.7c [xx XXX 2003]
540 *) Change AES_cbc_encrypt() so it outputs exact multiple of
541 blocks during encryption.
544 *) Various fixes to base64 BIO and non blocking I/O. On write
545 flushes were not handled properly if the BIO retried. On read
546 data was not being buffered properly and had various logic bugs.
547 This also affects blocking I/O when the data being decoded is a
551 *) Various S/MIME bugfixes and compatibility changes:
552 output correct application/pkcs7 MIME type if
553 PKCS7_NOOLDMIMETYPE is set. Tolerate some broken signatures.
554 Output CR+LF for EOL if PKCS7_CRLFEOL is set (this makes opening
555 of files as .eml work). Correctly handle very long lines in MIME
559 Changes between 0.9.7a and 0.9.7b [10 Apr 2003]
561 *) Countermeasure against the Klima-Pokorny-Rosa extension of
562 Bleichbacher's attack on PKCS #1 v1.5 padding: treat
563 a protocol version number mismatch like a decryption error
564 in ssl3_get_client_key_exchange (ssl/s3_srvr.c).
567 *) Turn on RSA blinding by default in the default implementation
568 to avoid a timing attack. Applications that don't want it can call
569 RSA_blinding_off() or use the new flag RSA_FLAG_NO_BLINDING.
570 They would be ill-advised to do so in most cases.
571 [Ben Laurie, Steve Henson, Geoff Thorpe, Bodo Moeller]
573 *) Change RSA blinding code so that it works when the PRNG is not
574 seeded (in this case, the secret RSA exponent is abused as
575 an unpredictable seed -- if it is not unpredictable, there
576 is no point in blinding anyway). Make RSA blinding thread-safe
577 by remembering the creator's thread ID in rsa->blinding and
578 having all other threads use local one-time blinding factors
579 (this requires more computation than sharing rsa->blinding, but
580 avoids excessive locking; and if an RSA object is not shared
581 between threads, blinding will still be very fast).
584 yet to be integrated into this CVS branch:
585 - Geoff's ENGINE_set_default() fix
587 *) Target "mingw" now allows native Windows code to be generated in
588 the Cygwin environment as well as with the MinGW compiler.
591 Changes between 0.9.7 and 0.9.7a [19 Feb 2003]
593 *) In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked
594 via timing by performing a MAC computation even if incorrrect
595 block cipher padding has been found. This is a countermeasure
596 against active attacks where the attacker has to distinguish
597 between bad padding and a MAC verification error. (CAN-2003-0078)
599 [Bodo Moeller; problem pointed out by Brice Canvel (EPFL),
600 Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and
601 Martin Vuagnoux (EPFL, Ilion)]
603 *) Make the no-err option work as intended. The intention with no-err
604 is not to have the whole error stack handling routines removed from
605 libcrypto, it's only intended to remove all the function name and
606 reason texts, thereby removing some of the footprint that may not
607 be interesting if those errors aren't displayed anyway.
609 NOTE: it's still possible for any application or module to have it's
610 own set of error texts inserted. The routines are there, just not
611 used by default when no-err is given.
614 *) Add support for FreeBSD on IA64.
615 [dirk.meyer@dinoex.sub.org via Richard Levitte, resolves #454]
617 *) Adjust DES_cbc_cksum() so it returns the same value as the MIT
618 Kerberos function mit_des_cbc_cksum(). Before this change,
619 the value returned by DES_cbc_cksum() was like the one from
620 mit_des_cbc_cksum(), except the bytes were swapped.
621 [Kevin Greaney <Kevin.Greaney@hp.com> and Richard Levitte]
623 *) Allow an application to disable the automatic SSL chain building.
624 Before this a rather primitive chain build was always performed in
625 ssl3_output_cert_chain(): an application had no way to send the
626 correct chain if the automatic operation produced an incorrect result.
628 Now the chain builder is disabled if either:
630 1. Extra certificates are added via SSL_CTX_add_extra_chain_cert().
632 2. The mode flag SSL_MODE_NO_AUTO_CHAIN is set.
634 The reasoning behind this is that an application would not want the
635 auto chain building to take place if extra chain certificates are
636 present and it might also want a means of sending no additional
637 certificates (for example the chain has two certificates and the
641 *) Add the possibility to build without the ENGINE framework.
642 [Steven Reddie <smr@essemer.com.au> via Richard Levitte]
644 *) Under Win32 gmtime() can return NULL: check return value in
645 OPENSSL_gmtime(). Add error code for case where gmtime() fails.
648 *) DSA routines: under certain error conditions uninitialized BN objects
649 could be freed. Solution: make sure initialization is performed early
650 enough. (Reported and fix supplied by Ivan D Nestlerode <nestler@MIT.EDU>,
651 Nils Larsch <nla@trustcenter.de> via PR#459)
654 *) Another fix for SSLv2 session ID handling: the session ID was incorrectly
655 checked on reconnect on the client side, therefore session resumption
656 could still fail with a "ssl session id is different" error. This
657 behaviour is masked when SSL_OP_ALL is used due to
658 SSL_OP_MICROSOFT_SESS_ID_BUG being set.
659 Behaviour observed by Crispin Flowerday <crispin@flowerday.cx> as
663 *) IA-32 assembler support enhancements: unified ELF targets, support
664 for SCO/Caldera platforms, fix for Cygwin shared build.
667 *) Add support for FreeBSD on sparc64. As a consequence, support for
668 FreeBSD on non-x86 processors is separate from x86 processors on
669 the config script, much like the NetBSD support.
670 [Richard Levitte & Kris Kennaway <kris@obsecurity.org>]
672 Changes between 0.9.6h and 0.9.7 [31 Dec 2002]
674 *) Fix session ID handling in SSLv2 client code: the SERVER FINISHED
675 code (06) was taken as the first octet of the session ID and the last
676 octet was ignored consequently. As a result SSLv2 client side session
677 caching could not have worked due to the session ID mismatch between
679 Behaviour observed by Crispin Flowerday <crispin@flowerday.cx> as
683 *) Change the declaration of needed Kerberos libraries to use EX_LIBS
684 instead of the special (and badly supported) LIBKRB5. LIBKRB5 is
688 *) The hw_ncipher.c engine requires dynamic locks. Unfortunately, it
689 seems that in spite of existing for more than a year, many application
690 author have done nothing to provide the necessary callbacks, which
691 means that this particular engine will not work properly anywhere.
692 This is a very unfortunate situation which forces us, in the name
693 of usability, to give the hw_ncipher.c a static lock, which is part
695 NOTE: This is for the 0.9.7 series ONLY. This hack will never
696 appear in 0.9.8 or later. We EXPECT application authors to have
697 dealt properly with this when 0.9.8 is released (unless we actually
698 make such changes in the libcrypto locking code that changes will
699 have to be made anyway).
702 *) In asn1_d2i_read_bio() repeatedly call BIO_read() until all content
703 octets have been read, EOF or an error occurs. Without this change
704 some truncated ASN1 structures will not produce an error.
707 *) Disable Heimdal support, since it hasn't been fully implemented.
708 Still give the possibility to force the use of Heimdal, but with
709 warnings and a request that patches get sent to openssl-dev.
712 *) Add the VC-CE target, introduce the WINCE sysname, and add
713 INSTALL.WCE and appropriate conditionals to make it build.
714 [Steven Reddie <smr@essemer.com.au> via Richard Levitte]
716 *) Change the DLL names for Cygwin to cygcrypto-x.y.z.dll and
717 cygssl-x.y.z.dll, where x, y and z are the major, minor and
718 edit numbers of the version.
719 [Corinna Vinschen <vinschen@redhat.com> and Richard Levitte]
721 *) Introduce safe string copy and catenation functions
722 (BUF_strlcpy() and BUF_strlcat()).
723 [Ben Laurie (CHATS) and Richard Levitte]
725 *) Avoid using fixed-size buffers for one-line DNs.
728 *) Add BUF_MEM_grow_clean() to avoid information leakage when
729 resizing buffers containing secrets, and use where appropriate.
732 *) Avoid using fixed size buffers for configuration file location.
735 *) Avoid filename truncation for various CA files.
738 *) Use sizeof in preference to magic numbers.
741 *) Avoid filename truncation in cert requests.
744 *) Add assertions to check for (supposedly impossible) buffer
748 *) Don't cache truncated DNS entries in the local cache (this could
749 potentially lead to a spoofing attack).
752 *) Fix various buffers to be large enough for hex/decimal
753 representations in a platform independent manner.
756 *) Add CRYPTO_realloc_clean() to avoid information leakage when
757 resizing buffers containing secrets, and use where appropriate.
760 *) Add BIO_indent() to avoid much slightly worrying code to do
764 *) Convert sprintf()/BIO_puts() to BIO_printf().
767 *) buffer_gets() could terminate with the buffer only half
771 *) Add assertions to prevent user-supplied crypto functions from
772 overflowing internal buffers by having large block sizes, etc.
775 *) New OPENSSL_assert() macro (similar to assert(), but enabled
779 *) Eliminate unused copy of key in RC4.
782 *) Eliminate unused and incorrectly sized buffers for IV in pem.h.
785 *) Fix off-by-one error in EGD path.
788 *) If RANDFILE path is too long, ignore instead of truncating.
791 *) Eliminate unused and incorrectly sized X.509 structure
795 *) Eliminate unused and dangerous function knumber().
798 *) Eliminate unused and dangerous structure, KSSL_ERR.
801 *) Protect against overlong session ID context length in an encoded
802 session object. Since these are local, this does not appear to be
806 *) Change from security patch (see 0.9.6e below) that did not affect
807 the 0.9.6 release series:
809 Remote buffer overflow in SSL3 protocol - an attacker could
810 supply an oversized master key in Kerberos-enabled versions.
814 *) Change the SSL kerb5 codes to match RFC 2712.
817 *) Make -nameopt work fully for req and add -reqopt switch.
818 [Michael Bell <michael.bell@rz.hu-berlin.de>, Steve Henson]
820 *) The "block size" for block ciphers in CFB and OFB mode should be 1.
821 [Steve Henson, reported by Yngve Nysaeter Pettersen <yngve@opera.com>]
823 *) Make sure tests can be performed even if the corresponding algorithms
824 have been removed entirely. This was also the last step to make
825 OpenSSL compilable with DJGPP under all reasonable conditions.
826 [Richard Levitte, Doug Kaufman <dkaufman@rahul.net>]
828 *) Add cipher selection rules COMPLEMENTOFALL and COMPLEMENTOFDEFAULT
829 to allow version independent disabling of normally unselected ciphers,
830 which may be activated as a side-effect of selecting a single cipher.
832 (E.g., cipher list string "RSA" enables ciphersuites that are left
833 out of "ALL" because they do not provide symmetric encryption.
834 "RSA:!COMPLEMEMENTOFALL" avoids these unsafe ciphersuites.)
835 [Lutz Jaenicke, Bodo Moeller]
837 *) Add appropriate support for separate platform-dependent build
838 directories. The recommended way to make a platform-dependent
839 build directory is the following (tested on Linux), maybe with
842 # Place yourself outside of the OpenSSL source tree. In
843 # this example, the environment variable OPENSSL_SOURCE
844 # is assumed to contain the absolute OpenSSL source directory.
845 mkdir -p objtree/"`uname -s`-`uname -r`-`uname -m`"
846 cd objtree/"`uname -s`-`uname -r`-`uname -m`"
847 (cd $OPENSSL_SOURCE; find . -type f) | while read F; do
848 mkdir -p `dirname $F`
849 ln -s $OPENSSL_SOURCE/$F $F
852 To be absolutely sure not to disturb the source tree, a "make clean"
853 is a good thing. If it isn't successfull, don't worry about it,
854 it probably means the source directory is very clean.
857 *) Make sure any ENGINE control commands make local copies of string
858 pointers passed to them whenever necessary. Otherwise it is possible
859 the caller may have overwritten (or deallocated) the original string
860 data when a later ENGINE operation tries to use the stored values.
861 [Götz Babin-Ebell <babinebell@trustcenter.de>]
863 *) Improve diagnostics in file reading and command-line digests.
864 [Ben Laurie aided and abetted by Solar Designer <solar@openwall.com>]
866 *) Add AES modes CFB and OFB to the object database. Correct an
867 error in AES-CFB decryption.
870 *) Remove most calls to EVP_CIPHER_CTX_cleanup() in evp_enc.c, this
871 allows existing EVP_CIPHER_CTX structures to be reused after
872 calling EVP_*Final(). This behaviour is used by encryption
873 BIOs and some applications. This has the side effect that
874 applications must explicitly clean up cipher contexts with
875 EVP_CIPHER_CTX_cleanup() or they will leak memory.
878 *) Check the values of dna and dnb in bn_mul_recursive before calling
879 bn_mul_comba (a non zero value means the a or b arrays do not contain
880 n2 elements) and fallback to bn_mul_normal if either is not zero.
883 *) Fix escaping of non-ASCII characters when using the -subj option
884 of the "openssl req" command line tool. (Robert Joop <joop@fokus.gmd.de>)
887 *) Make object definitions compliant to LDAP (RFC2256): SN is the short
888 form for "surname", serialNumber has no short form.
889 Use "mail" as the short name for "rfc822Mailbox" according to RFC2798;
890 therefore remove "mail" short name for "internet 7".
891 The OID for unique identifiers in X509 certificates is
892 x500UniqueIdentifier, not uniqueIdentifier.
893 Some more OID additions. (Michael Bell <michael.bell@rz.hu-berlin.de>)
896 *) Add an "init" command to the ENGINE config module and auto initialize
897 ENGINEs. Without any "init" command the ENGINE will be initialized
898 after all ctrl commands have been executed on it. If init=1 the
899 ENGINE is initailized at that point (ctrls before that point are run
900 on the uninitialized ENGINE and after on the initialized one). If
901 init=0 then the ENGINE will not be iniatialized at all.
904 *) Fix the 'app_verify_callback' interface so that the user-defined
905 argument is actually passed to the callback: In the
906 SSL_CTX_set_cert_verify_callback() prototype, the callback
907 declaration has been changed from
910 int (*cb)(X509_STORE_CTX *,void *);
911 in ssl_verify_cert_chain (ssl/ssl_cert.c), the call
912 i=s->ctx->app_verify_callback(&ctx)
913 has been changed into
914 i=s->ctx->app_verify_callback(&ctx, s->ctx->app_verify_arg).
916 To update applications using SSL_CTX_set_cert_verify_callback(),
917 a dummy argument can be added to their callback functions.
918 [D. K. Smetters <smetters@parc.xerox.com>]
920 *) Added the '4758cca' ENGINE to support IBM 4758 cards.
921 [Maurice Gittens <maurice@gittens.nl>, touchups by Geoff Thorpe]
923 *) Add and OPENSSL_LOAD_CONF define which will cause
924 OpenSSL_add_all_algorithms() to load the openssl.cnf config file.
925 This allows older applications to transparently support certain
926 OpenSSL features: such as crypto acceleration and dynamic ENGINE loading.
927 Two new functions OPENSSL_add_all_algorithms_noconf() which will never
928 load the config file and OPENSSL_add_all_algorithms_conf() which will
929 always load it have also been added.
932 *) Add the OFB, CFB and CTR (all with 128 bit feedback) to AES.
933 Adjust NIDs and EVP layer.
934 [Stephen Sprunk <stephen@sprunk.org> and Richard Levitte]
936 *) Config modules support in openssl utility.
938 Most commands now load modules from the config file,
939 though in a few (such as version) this isn't done
940 because it couldn't be used for anything.
942 In the case of ca and req the config file used is
943 the same as the utility itself: that is the -config
944 command line option can be used to specify an
948 *) Move default behaviour from OPENSSL_config(). If appname is NULL
949 use "openssl_conf" if filename is NULL use default openssl config file.
952 *) Add an argument to OPENSSL_config() to allow the use of an alternative
953 config section name. Add a new flag to tolerate a missing config file
954 and move code to CONF_modules_load_file().
957 *) Support for crypto accelerator cards from Accelerated Encryption
958 Processing, www.aep.ie. (Use engine 'aep')
959 The support was copied from 0.9.6c [engine] and adapted/corrected
960 to work with the new engine framework.
961 [AEP Inc. and Richard Levitte]
963 *) Support for SureWare crypto accelerator cards from Baltimore
964 Technologies. (Use engine 'sureware')
965 The support was copied from 0.9.6c [engine] and adapted
966 to work with the new engine framework.
969 *) Have the CHIL engine fork-safe (as defined by nCipher) and actually
970 make the newer ENGINE framework commands for the CHIL engine work.
971 [Toomas Kiisk <vix@cyber.ee> and Richard Levitte]
973 *) Make it possible to produce shared libraries on ReliantUNIX.
974 [Robert Dahlem <Robert.Dahlem@ffm2.siemens.de> via Richard Levitte]
976 *) Add the configuration target debug-linux-ppro.
977 Make 'openssl rsa' use the general key loading routines
978 implemented in apps.c, and make those routines able to
979 handle the key format FORMAT_NETSCAPE and the variant
981 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte]
983 *) Fix a crashbug and a logic bug in hwcrhk_load_pubkey().
984 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte]
986 *) Add -keyform to rsautl, and document -engine.
987 [Richard Levitte, inspired by Toomas Kiisk <vix@cyber.ee>]
989 *) Change BIO_new_file (crypto/bio/bss_file.c) to use new
990 BIO_R_NO_SUCH_FILE error code rather than the generic
991 ERR_R_SYS_LIB error code if fopen() fails with ENOENT.
996 ERR_peek_last_error_line
997 ERR_peek_last_error_line_data.
1001 ERR_peek_error_line_data,
1002 but report on the latest error recorded rather than the first one
1003 still in the error queue.
1004 [Ben Laurie, Bodo Moeller]
1006 *) default_algorithms option in ENGINE config module. This allows things
1008 default_algorithms = ALL
1009 default_algorithms = RSA, DSA, RAND, CIPHERS, DIGESTS
1012 *) Prelminary ENGINE config module.
1015 *) New experimental application configuration code.
1018 *) Change the AES code to follow the same name structure as all other
1019 symmetric ciphers, and behave the same way. Move everything to
1020 the directory crypto/aes, thereby obsoleting crypto/rijndael.
1021 [Stephen Sprunk <stephen@sprunk.org> and Richard Levitte]
1023 *) SECURITY: remove unsafe setjmp/signal interaction from ui_openssl.c.
1024 [Ben Laurie and Theo de Raadt]
1026 *) Add option to output public keys in req command.
1027 [Massimiliano Pala madwolf@openca.org]
1029 *) Use wNAFs in EC_POINTs_mul() for improved efficiency
1030 (up to about 10% better than before for P-192 and P-224).
1033 *) New functions/macros
1035 SSL_CTX_set_msg_callback(ctx, cb)
1036 SSL_CTX_set_msg_callback_arg(ctx, arg)
1037 SSL_set_msg_callback(ssl, cb)
1038 SSL_set_msg_callback_arg(ssl, arg)
1040 to request calling a callback function
1042 void cb(int write_p, int version, int content_type,
1043 const void *buf, size_t len, SSL *ssl, void *arg)
1045 whenever a protocol message has been completely received
1046 (write_p == 0) or sent (write_p == 1). Here 'version' is the
1047 protocol version according to which the SSL library interprets
1048 the current protocol message (SSL2_VERSION, SSL3_VERSION, or
1049 TLS1_VERSION). 'content_type' is 0 in the case of SSL 2.0, or
1050 the content type as defined in the SSL 3.0/TLS 1.0 protocol
1051 specification (change_cipher_spec(20), alert(21), handshake(22)).
1052 'buf' and 'len' point to the actual message, 'ssl' to the
1053 SSL object, and 'arg' is the application-defined value set by
1054 SSL[_CTX]_set_msg_callback_arg().
1056 'openssl s_client' and 'openssl s_server' have new '-msg' options
1057 to enable a callback that displays all protocol messages.
1060 *) Change the shared library support so shared libraries are built as
1061 soon as the corresponding static library is finished, and thereby get
1062 openssl and the test programs linked against the shared library.
1063 This still only happens when the keyword "shard" has been given to
1064 the configuration scripts.
1066 NOTE: shared library support is still an experimental thing, and
1067 backward binary compatibility is still not guaranteed.
1068 ["Maciej W. Rozycki" <macro@ds2.pg.gda.pl> and Richard Levitte]
1070 *) Add support for Subject Information Access extension.
1071 [Peter Sylvester <Peter.Sylvester@EdelWeb.fr>]
1073 *) Make BUF_MEM_grow() behaviour more consistent: Initialise to zero
1074 additional bytes when new memory had to be allocated, not just
1075 when reusing an existing buffer.
1078 *) New command line and configuration option 'utf8' for the req command.
1079 This allows field values to be specified as UTF8 strings.
1082 *) Add -multi and -mr options to "openssl speed" - giving multiple parallel
1083 runs for the former and machine-readable output for the latter.
1086 *) Add '-noemailDN' option to 'openssl ca'. This prevents inclusion
1087 of the e-mail address in the DN (i.e., it will go into a certificate
1088 extension only). The new configuration file option 'email_in_dn = no'
1089 has the same effect.
1090 [Massimiliano Pala madwolf@openca.org]
1092 *) Change all functions with names starting with des_ to be starting
1093 with DES_ instead. Add wrappers that are compatible with libdes,
1094 but are named _ossl_old_des_*. Finally, add macros that map the
1095 des_* symbols to the corresponding _ossl_old_des_* if libdes
1096 compatibility is desired. If OpenSSL 0.9.6c compatibility is
1097 desired, the des_* symbols will be mapped to DES_*, with one
1100 Since we provide two compatibility mappings, the user needs to
1101 define the macro OPENSSL_DES_LIBDES_COMPATIBILITY if libdes
1102 compatibility is desired. The default (i.e., when that macro
1103 isn't defined) is OpenSSL 0.9.6c compatibility.
1105 There are also macros that enable and disable the support of old
1106 des functions altogether. Those are OPENSSL_ENABLE_OLD_DES_SUPPORT
1107 and OPENSSL_DISABLE_OLD_DES_SUPPORT. If none or both of those
1108 are defined, the default will apply: to support the old des routines.
1110 In either case, one must include openssl/des.h to get the correct
1111 definitions. Do not try to just include openssl/des_old.h, that
1114 NOTE: This is a major break of an old API into a new one. Software
1115 authors are encouraged to switch to the DES_ style functions. Some
1116 time in the future, des_old.h and the libdes compatibility functions
1117 will be disable (i.e. OPENSSL_DISABLE_OLD_DES_SUPPORT will be the
1118 default), and then completely removed.
1121 *) Test for certificates which contain unsupported critical extensions.
1122 If such a certificate is found during a verify operation it is
1123 rejected by default: this behaviour can be overridden by either
1124 handling the new error X509_V_ERR_UNHANDLED_CRITICAL_EXTENSION or
1125 by setting the verify flag X509_V_FLAG_IGNORE_CRITICAL. A new function
1126 X509_supported_extension() has also been added which returns 1 if a
1127 particular extension is supported.
1130 *) Modify the behaviour of EVP cipher functions in similar way to digests
1131 to retain compatibility with existing code.
1134 *) Modify the behaviour of EVP_DigestInit() and EVP_DigestFinal() to retain
1135 compatibility with existing code. In particular the 'ctx' parameter does
1136 not have to be to be initialized before the call to EVP_DigestInit() and
1137 it is tidied up after a call to EVP_DigestFinal(). New function
1138 EVP_DigestFinal_ex() which does not tidy up the ctx. Similarly function
1139 EVP_MD_CTX_copy() changed to not require the destination to be
1140 initialized valid and new function EVP_MD_CTX_copy_ex() added which
1141 requires the destination to be valid.
1143 Modify all the OpenSSL digest calls to use EVP_DigestInit_ex(),
1144 EVP_DigestFinal_ex() and EVP_MD_CTX_copy_ex().
1147 *) Change ssl3_get_message (ssl/s3_both.c) and the functions using it
1148 so that complete 'Handshake' protocol structures are kept in memory
1149 instead of overwriting 'msg_type' and 'length' with 'body' data.
1152 *) Add an implementation of SSL_add_dir_cert_subjects_to_stack for Win32.
1153 [Massimo Santin via Richard Levitte]
1155 *) Major restructuring to the underlying ENGINE code. This includes
1156 reduction of linker bloat, separation of pure "ENGINE" manipulation
1157 (initialisation, etc) from functionality dealing with implementations
1158 of specific crypto iterfaces. This change also introduces integrated
1159 support for symmetric ciphers and digest implementations - so ENGINEs
1160 can now accelerate these by providing EVP_CIPHER and EVP_MD
1161 implementations of their own. This is detailed in crypto/engine/README
1162 as it couldn't be adequately described here. However, there are a few
1163 API changes worth noting - some RSA, DSA, DH, and RAND functions that
1164 were changed in the original introduction of ENGINE code have now
1165 reverted back - the hooking from this code to ENGINE is now a good
1166 deal more passive and at run-time, operations deal directly with
1167 RSA_METHODs, DSA_METHODs (etc) as they did before, rather than
1168 dereferencing through an ENGINE pointer any more. Also, the ENGINE
1169 functions dealing with BN_MOD_EXP[_CRT] handlers have been removed -
1170 they were not being used by the framework as there is no concept of a
1171 BIGNUM_METHOD and they could not be generalised to the new
1172 'ENGINE_TABLE' mechanism that underlies the new code. Similarly,
1173 ENGINE_cpy() has been removed as it cannot be consistently defined in
1177 *) Change ASN1_GENERALIZEDTIME_check() to allow fractional seconds.
1180 *) Change mkdef.pl to sort symbols that get the same entry number,
1181 and make sure the automatically generated functions ERR_load_*
1182 become part of libeay.num as well.
1185 *) New function SSL_renegotiate_pending(). This returns true once
1186 renegotiation has been requested (either SSL_renegotiate() call
1187 or HelloRequest/ClientHello receveived from the peer) and becomes
1188 false once a handshake has been completed.
1189 (For servers, SSL_renegotiate() followed by SSL_do_handshake()
1190 sends a HelloRequest, but does not ensure that a handshake takes
1191 place. SSL_renegotiate_pending() is useful for checking if the
1192 client has followed the request.)
1195 *) New SSL option SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION.
1196 By default, clients may request session resumption even during
1197 renegotiation (if session ID contexts permit); with this option,
1198 session resumption is possible only in the first handshake.
1200 SSL_OP_ALL is now 0x00000FFFL instead of 0x000FFFFFL. This makes
1201 more bits available for options that should not be part of
1202 SSL_OP_ALL (such as SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION).
1205 *) Add some demos for certificate and certificate request creation.
1208 *) Make maximum certificate chain size accepted from the peer application
1209 settable (SSL*_get/set_max_cert_list()), as proposed by
1210 "Douglas E. Engert" <deengert@anl.gov>.
1213 *) Add support for shared libraries for Unixware-7
1214 (Boyd Lynn Gerber <gerberb@zenez.com>).
1217 *) Add a "destroy" handler to ENGINEs that allows structural cleanup to
1218 be done prior to destruction. Use this to unload error strings from
1219 ENGINEs that load their own error strings. NB: This adds two new API
1220 functions to "get" and "set" this destroy handler in an ENGINE.
1223 *) Alter all existing ENGINE implementations (except "openssl" and
1224 "openbsd") to dynamically instantiate their own error strings. This
1225 makes them more flexible to be built both as statically-linked ENGINEs
1226 and self-contained shared-libraries loadable via the "dynamic" ENGINE.
1227 Also, add stub code to each that makes building them as self-contained
1228 shared-libraries easier (see README.ENGINE).
1231 *) Add a "dynamic" ENGINE that provides a mechanism for binding ENGINE
1232 implementations into applications that are completely implemented in
1233 self-contained shared-libraries. The "dynamic" ENGINE exposes control
1234 commands that can be used to configure what shared-library to load and
1235 to control aspects of the way it is handled. Also, made an update to
1236 the README.ENGINE file that brings its information up-to-date and
1237 provides some information and instructions on the "dynamic" ENGINE
1238 (ie. how to use it, how to build "dynamic"-loadable ENGINEs, etc).
1241 *) Make it possible to unload ranges of ERR strings with a new
1242 "ERR_unload_strings" function.
1245 *) Add a copy() function to EVP_MD.
1248 *) Make EVP_MD routines take a context pointer instead of just the
1249 md_data void pointer.
1252 *) Add flags to EVP_MD and EVP_MD_CTX. EVP_MD_FLAG_ONESHOT indicates
1253 that the digest can only process a single chunk of data
1254 (typically because it is provided by a piece of
1255 hardware). EVP_MD_CTX_FLAG_ONESHOT indicates that the application
1256 is only going to provide a single chunk of data, and hence the
1257 framework needn't accumulate the data for oneshot drivers.
1260 *) As with "ERR", make it possible to replace the underlying "ex_data"
1261 functions. This change also alters the storage and management of global
1262 ex_data state - it's now all inside ex_data.c and all "class" code (eg.
1263 RSA, BIO, SSL_CTX, etc) no longer stores its own STACKS and per-class
1264 index counters. The API functions that use this state have been changed
1265 to take a "class_index" rather than pointers to the class's local STACK
1266 and counter, and there is now an API function to dynamically create new
1267 classes. This centralisation allows us to (a) plug a lot of the
1268 thread-safety problems that existed, and (b) makes it possible to clean
1269 up all allocated state using "CRYPTO_cleanup_all_ex_data()". W.r.t. (b)
1270 such data would previously have always leaked in application code and
1271 workarounds were in place to make the memory debugging turn a blind eye
1272 to it. Application code that doesn't use this new function will still
1273 leak as before, but their memory debugging output will announce it now
1274 rather than letting it slide.
1276 Besides the addition of CRYPTO_cleanup_all_ex_data(), another API change
1277 induced by the "ex_data" overhaul is that X509_STORE_CTX_init() now
1278 has a return value to indicate success or failure.
1281 *) Make it possible to replace the underlying "ERR" functions such that the
1282 global state (2 LHASH tables and 2 locks) is only used by the "default"
1283 implementation. This change also adds two functions to "get" and "set"
1284 the implementation prior to it being automatically set the first time
1285 any other ERR function takes place. Ie. an application can call "get",
1286 pass the return value to a module it has just loaded, and that module
1287 can call its own "set" function using that value. This means the
1288 module's "ERR" operations will use (and modify) the error state in the
1289 application and not in its own statically linked copy of OpenSSL code.
1292 *) Give DH, DSA, and RSA types their own "**_up_ref()" function to increment
1293 reference counts. This performs normal REF_PRINT/REF_CHECK macros on
1294 the operation, and provides a more encapsulated way for external code
1295 (crypto/evp/ and ssl/) to do this. Also changed the evp and ssl code
1296 to use these functions rather than manually incrementing the counts.
1298 Also rename "DSO_up()" function to more descriptive "DSO_up_ref()".
1301 *) Add EVP test program.
1304 *) Add symmetric cipher support to ENGINE. Expect the API to change!
1307 *) New CRL functions: X509_CRL_set_version(), X509_CRL_set_issuer_name()
1308 X509_CRL_set_lastUpdate(), X509_CRL_set_nextUpdate(), X509_CRL_sort(),
1309 X509_REVOKED_set_serialNumber(), and X509_REVOKED_set_revocationDate().
1310 These allow a CRL to be built without having to access X509_CRL fields
1311 directly. Modify 'ca' application to use new functions.
1314 *) Move SSL_OP_TLS_ROLLBACK_BUG out of the SSL_OP_ALL list of recommended
1315 bug workarounds. Rollback attack detection is a security feature.
1316 The problem will only arise on OpenSSL servers when TLSv1 is not
1317 available (sslv3_server_method() or SSL_OP_NO_TLSv1).
1318 Software authors not wanting to support TLSv1 will have special reasons
1319 for their choice and can explicitly enable this option.
1320 [Bodo Moeller, Lutz Jaenicke]
1322 *) Rationalise EVP so it can be extended: don't include a union of
1323 cipher/digest structures, add init/cleanup functions for EVP_MD_CTX
1324 (similar to those existing for EVP_CIPHER_CTX).
1329 EVP_MD_CTX_init(&md); /* new function call */
1330 EVP_DigestInit(&md, EVP_sha1());
1331 EVP_DigestUpdate(&md, in, len);
1332 EVP_DigestFinal(&md, out, NULL);
1333 EVP_MD_CTX_cleanup(&md); /* new function call */
1337 *) Make DES key schedule conform to the usual scheme, as well as
1338 correcting its structure. This means that calls to DES functions
1339 now have to pass a pointer to a des_key_schedule instead of a
1340 plain des_key_schedule (which was actually always a pointer
1343 des_key_schedule ks;
1345 des_set_key_checked(..., &ks);
1346 des_ncbc_encrypt(..., &ks, ...);
1348 (Note that a later change renames 'des_...' into 'DES_...'.)
1351 *) Initial reduction of linker bloat: the use of some functions, such as
1352 PEM causes large amounts of unused functions to be linked in due to
1353 poor organisation. For example pem_all.c contains every PEM function
1354 which has a knock on effect of linking in large amounts of (unused)
1355 ASN1 code. Grouping together similar functions and splitting unrelated
1356 functions prevents this.
1359 *) Cleanup of EVP macros.
1362 *) Change historical references to {NID,SN,LN}_des_ede and ede3 to add the
1363 correct _ecb suffix.
1366 *) Add initial OCSP responder support to ocsp application. The
1367 revocation information is handled using the text based index
1368 use by the ca application. The responder can either handle
1369 requests generated internally, supplied in files (for example
1370 via a CGI script) or using an internal minimal server.
1373 *) Add configuration choices to get zlib compression for TLS.
1376 *) Changes to Kerberos SSL for RFC 2712 compliance:
1377 1. Implemented real KerberosWrapper, instead of just using
1378 KRB5 AP_REQ message. [Thanks to Simon Wilkinson <sxw@sxw.org.uk>]
1379 2. Implemented optional authenticator field of KerberosWrapper.
1381 Added openssl-style ASN.1 macros for Kerberos ticket, ap_req,
1382 and authenticator structs; see crypto/krb5/.
1384 Generalized Kerberos calls to support multiple Kerberos libraries.
1385 [Vern Staats <staatsvr@asc.hpc.mil>,
1386 Jeffrey Altman <jaltman@columbia.edu>
1387 via Richard Levitte]
1389 *) Cause 'openssl speed' to use fully hard-coded DSA keys as it
1390 already does with RSA. testdsa.h now has 'priv_key/pub_key'
1391 values for each of the key sizes rather than having just
1392 parameters (and 'speed' generating keys each time).
1395 *) Speed up EVP routines.
1398 type 8 bytes 64 bytes 256 bytes 1024 bytes 8192 bytes
1399 des-cbc 4408.85k 5560.51k 5778.46k 5862.20k 5825.16k
1400 des-cbc 4389.55k 5571.17k 5792.23k 5846.91k 5832.11k
1401 des-cbc 4394.32k 5575.92k 5807.44k 5848.37k 5841.30k
1403 des-cbc 3482.66k 5069.49k 5496.39k 5614.16k 5639.28k
1404 des-cbc 3480.74k 5068.76k 5510.34k 5609.87k 5635.52k
1405 des-cbc 3483.72k 5067.62k 5504.60k 5708.01k 5724.80k
1408 des-cbc 4660.16k 5650.19k 5807.19k 5827.13k 5783.32k
1410 des-cbc 3624.96k 5258.21k 5530.91k 5624.30k 5628.26k
1413 *) Added the OS2-EMX target.
1414 ["Brian Havard" <brianh@kheldar.apana.org.au> and Richard Levitte]
1416 *) Rewrite apps to use NCONF routines instead of the old CONF. New functions
1417 to support NCONF routines in extension code. New function CONF_set_nconf()
1418 to allow functions which take an NCONF to also handle the old LHASH
1419 structure: this means that the old CONF compatible routines can be
1420 retained (in particular wrt extensions) without having to duplicate the
1421 code. New function X509V3_add_ext_nconf_sk to add extensions to a stack.
1424 *) Enhance the general user interface with mechanisms for inner control
1425 and with possibilities to have yes/no kind of prompts.
1428 *) Change all calls to low level digest routines in the library and
1429 applications to use EVP. Add missing calls to HMAC_cleanup() and
1430 don't assume HMAC_CTX can be copied using memcpy().
1431 [Verdon Walker <VWalker@novell.com>, Steve Henson]
1433 *) Add the possibility to control engines through control names but with
1434 arbitrary arguments instead of just a string.
1435 Change the key loaders to take a UI_METHOD instead of a callback
1436 function pointer. NOTE: this breaks binary compatibility with earlier
1437 versions of OpenSSL [engine].
1438 Adapt the nCipher code for these new conditions and add a card insertion
1442 *) Enhance the general user interface with mechanisms to better support
1443 dialog box interfaces, application-defined prompts, the possibility
1444 to use defaults (for example default passwords from somewhere else)
1445 and interrupts/cancellations.
1448 *) Tidy up PKCS#12 attribute handling. Add support for the CSP name
1449 attribute in PKCS#12 files, add new -CSP option to pkcs12 utility.
1452 *) Fix a memory leak in 'sk_dup()' in the case reallocation fails. (Also
1453 tidy up some unnecessarily weird code in 'sk_new()').
1454 [Geoff, reported by Diego Tartara <dtartara@novamens.com>]
1456 *) Change the key loading routines for ENGINEs to use the same kind
1457 callback (pem_password_cb) as all other routines that need this
1461 *) Increase ENTROPY_NEEDED to 32 bytes, as Rijndael can operate with
1462 256 bit (=32 byte) keys. Of course seeding with more entropy bytes
1463 than this minimum value is recommended.
1466 *) New random seeder for OpenVMS, using the system process statistics
1467 that are easily reachable.
1470 *) Windows apparently can't transparently handle global
1471 variables defined in DLLs. Initialisations such as:
1473 const ASN1_ITEM *it = &ASN1_INTEGER_it;
1475 wont compile. This is used by the any applications that need to
1476 declare their own ASN1 modules. This was fixed by adding the option
1477 EXPORT_VAR_AS_FN to all Win32 platforms, although this isn't strictly
1478 needed for static libraries under Win32.
1481 *) New functions X509_PURPOSE_set() and X509_TRUST_set() to handle
1482 setting of purpose and trust fields. New X509_STORE trust and
1483 purpose functions and tidy up setting in other SSL functions.
1486 *) Add copies of X509_STORE_CTX fields and callbacks to X509_STORE
1487 structure. These are inherited by X509_STORE_CTX when it is
1488 initialised. This allows various defaults to be set in the
1489 X509_STORE structure (such as flags for CRL checking and custom
1490 purpose or trust settings) for functions which only use X509_STORE_CTX
1491 internally such as S/MIME.
1493 Modify X509_STORE_CTX_purpose_inherit() so it only sets purposes and
1494 trust settings if they are not set in X509_STORE. This allows X509_STORE
1495 purposes and trust (in S/MIME for example) to override any set by default.
1497 Add command line options for CRL checking to smime, s_client and s_server
1501 *) Initial CRL based revocation checking. If the CRL checking flag(s)
1502 are set then the CRL is looked up in the X509_STORE structure and
1503 its validity and signature checked, then if the certificate is found
1504 in the CRL the verify fails with a revoked error.
1506 Various new CRL related callbacks added to X509_STORE_CTX structure.
1508 Command line options added to 'verify' application to support this.
1510 This needs some additional work, such as being able to handle multiple
1511 CRLs with different times, extension based lookup (rather than just
1512 by subject name) and ultimately more complete V2 CRL extension
1516 *) Add a general user interface API (crypto/ui/). This is designed
1517 to replace things like des_read_password and friends (backward
1518 compatibility functions using this new API are provided).
1519 The purpose is to remove prompting functions from the DES code
1520 section as well as provide for prompting through dialog boxes in
1521 a window system and the like.
1524 *) Add "ex_data" support to ENGINE so implementations can add state at a
1525 per-structure level rather than having to store it globally.
1528 *) Make it possible for ENGINE structures to be copied when retrieved by
1529 ENGINE_by_id() if the ENGINE specifies a new flag: ENGINE_FLAGS_BY_ID_COPY.
1530 This causes the "original" ENGINE structure to act like a template,
1531 analogous to the RSA vs. RSA_METHOD type of separation. Because of this
1532 operational state can be localised to each ENGINE structure, despite the
1533 fact they all share the same "methods". New ENGINE structures returned in
1534 this case have no functional references and the return value is the single
1535 structural reference. This matches the single structural reference returned
1536 by ENGINE_by_id() normally, when it is incremented on the pre-existing
1540 *) Fix ASN1 decoder when decoding type ANY and V_ASN1_OTHER: since this
1541 needs to match any other type at all we need to manually clear the
1545 *) Changes to the "openssl engine" utility to include;
1546 - verbosity levels ('-v', '-vv', and '-vvv') that provide information
1547 about an ENGINE's available control commands.
1548 - executing control commands from command line arguments using the
1549 '-pre' and '-post' switches. '-post' is only used if '-t' is
1550 specified and the ENGINE is successfully initialised. The syntax for
1551 the individual commands are colon-separated, for example;
1552 openssl engine chil -pre FORK_CHECK:0 -pre SO_PATH:/lib/test.so
1555 *) New dynamic control command support for ENGINEs. ENGINEs can now
1556 declare their own commands (numbers), names (strings), descriptions,
1557 and input types for run-time discovery by calling applications. A
1558 subset of these commands are implicitly classed as "executable"
1559 depending on their input type, and only these can be invoked through
1560 the new string-based API function ENGINE_ctrl_cmd_string(). (Eg. this
1561 can be based on user input, config files, etc). The distinction is
1562 that "executable" commands cannot return anything other than a boolean
1563 result and can only support numeric or string input, whereas some
1564 discoverable commands may only be for direct use through
1565 ENGINE_ctrl(), eg. supporting the exchange of binary data, function
1566 pointers, or other custom uses. The "executable" commands are to
1567 support parameterisations of ENGINE behaviour that can be
1568 unambiguously defined by ENGINEs and used consistently across any
1569 OpenSSL-based application. Commands have been added to all the
1570 existing hardware-supporting ENGINEs, noticeably "SO_PATH" to allow
1571 control over shared-library paths without source code alterations.
1574 *) Changed all ENGINE implementations to dynamically allocate their
1575 ENGINEs rather than declaring them statically. Apart from this being
1576 necessary with the removal of the ENGINE_FLAGS_MALLOCED distinction,
1577 this also allows the implementations to compile without using the
1578 internal engine_int.h header.
1581 *) Minor adjustment to "rand" code. RAND_get_rand_method() now returns a
1582 'const' value. Any code that should be able to modify a RAND_METHOD
1583 should already have non-const pointers to it (ie. they should only
1584 modify their own ones).
1587 *) Made a variety of little tweaks to the ENGINE code.
1588 - "atalla" and "ubsec" string definitions were moved from header files
1589 to C code. "nuron" string definitions were placed in variables
1590 rather than hard-coded - allowing parameterisation of these values
1591 later on via ctrl() commands.
1592 - Removed unused "#if 0"'d code.
1593 - Fixed engine list iteration code so it uses ENGINE_free() to release
1594 structural references.
1595 - Constified the RAND_METHOD element of ENGINE structures.
1596 - Constified various get/set functions as appropriate and added
1597 missing functions (including a catch-all ENGINE_cpy that duplicates
1598 all ENGINE values onto a new ENGINE except reference counts/state).
1599 - Removed NULL parameter checks in get/set functions. Setting a method
1600 or function to NULL is a way of cancelling out a previously set
1601 value. Passing a NULL ENGINE parameter is just plain stupid anyway
1602 and doesn't justify the extra error symbols and code.
1603 - Deprecate the ENGINE_FLAGS_MALLOCED define and move the area for
1604 flags from engine_int.h to engine.h.
1605 - Changed prototypes for ENGINE handler functions (init(), finish(),
1606 ctrl(), key-load functions, etc) to take an (ENGINE*) parameter.
1609 *) Implement binary inversion algorithm for BN_mod_inverse in addition
1610 to the algorithm using long division. The binary algorithm can be
1611 used only if the modulus is odd. On 32-bit systems, it is faster
1612 only for relatively small moduli (roughly 20-30% for 128-bit moduli,
1613 roughly 5-15% for 256-bit moduli), so we use it only for moduli
1614 up to 450 bits. In 64-bit environments, the binary algorithm
1615 appears to be advantageous for much longer moduli; here we use it
1616 for moduli up to 2048 bits.
1619 *) Rewrite CHOICE field setting in ASN1_item_ex_d2i(). The old code
1620 could not support the combine flag in choice fields.
1623 *) Add a 'copy_extensions' option to the 'ca' utility. This copies
1624 extensions from a certificate request to the certificate.
1627 *) Allow multiple 'certopt' and 'nameopt' options to be separated
1628 by commas. Add 'namopt' and 'certopt' options to the 'ca' config
1629 file: this allows the display of the certificate about to be
1630 signed to be customised, to allow certain fields to be included
1631 or excluded and extension details. The old system didn't display
1632 multicharacter strings properly, omitted fields not in the policy
1633 and couldn't display additional details such as extensions.
1636 *) Function EC_POINTs_mul for multiple scalar multiplication
1637 of an arbitrary number of elliptic curve points
1638 \sum scalars[i]*points[i],
1639 optionally including the generator defined for the EC_GROUP:
1640 scalar*generator + \sum scalars[i]*points[i].
1642 EC_POINT_mul is a simple wrapper function for the typical case
1643 that the point list has just one item (besides the optional
1647 *) First EC_METHODs for curves over GF(p):
1649 EC_GFp_simple_method() uses the basic BN_mod_mul and BN_mod_sqr
1650 operations and provides various method functions that can also
1651 operate with faster implementations of modular arithmetic.
1653 EC_GFp_mont_method() reuses most functions that are part of
1654 EC_GFp_simple_method, but uses Montgomery arithmetic.
1656 [Bodo Moeller; point addition and point doubling
1657 implementation directly derived from source code provided by
1658 Lenka Fibikova <fibikova@exp-math.uni-essen.de>]
1660 *) Framework for elliptic curves (crypto/ec/ec.h, crypto/ec/ec_lcl.h,
1661 crypto/ec/ec_lib.c):
1663 Curves are EC_GROUP objects (with an optional group generator)
1664 based on EC_METHODs that are built into the library.
1666 Points are EC_POINT objects based on EC_GROUP objects.
1668 Most of the framework would be able to handle curves over arbitrary
1669 finite fields, but as there are no obvious types for fields other
1670 than GF(p), some functions are limited to that for now.
1673 *) Add the -HTTP option to s_server. It is similar to -WWW, but requires
1674 that the file contains a complete HTTP response.
1677 *) Add the ec directory to mkdef.pl and mkfiles.pl. In mkdef.pl
1678 change the def and num file printf format specifier from "%-40sXXX"
1679 to "%-39s XXX". The latter will always guarantee a space after the
1680 field while the former will cause them to run together if the field
1681 is 40 of more characters long.
1684 *) Constify the cipher and digest 'method' functions and structures
1685 and modify related functions to take constant EVP_MD and EVP_CIPHER
1689 *) Hide BN_CTX structure details in bn_lcl.h instead of publishing them
1690 in <openssl/bn.h>. Also further increase BN_CTX_NUM to 32.
1693 *) Modify EVP_Digest*() routines so they now return values. Although the
1694 internal software routines can never fail additional hardware versions
1698 *) Clean up crypto/err/err.h and change some error codes to avoid conflicts:
1700 Previously ERR_R_FATAL was too small and coincided with ERR_LIB_PKCS7
1701 (= ERR_R_PKCS7_LIB); it is now 64 instead of 32.
1704 ERR_R_NESTED_ASN1_ERROR
1706 ERR_R_MISSING_ASN1_EOS
1707 were 4 .. 9, conflicting with
1708 ERR_LIB_RSA (= ERR_R_RSA_LIB)
1710 ERR_LIB_PEM (= ERR_R_PEM_LIB).
1711 They are now 58 .. 63 (i.e., just below ERR_R_FATAL).
1713 Add new error code 'ERR_R_INTERNAL_ERROR'.
1716 *) Don't overuse locks in crypto/err/err.c: For data retrieval, CRYPTO_r_lock
1720 *) New option '-subj arg' for 'openssl req' and 'openssl ca'. This
1721 sets the subject name for a new request or supersedes the
1722 subject name in a given request. Formats that can be parsed are
1723 'CN=Some Name, OU=myOU, C=IT'
1725 'CN=Some Name/OU=myOU/C=IT'.
1727 Add options '-batch' and '-verbose' to 'openssl req'.
1728 [Massimiliano Pala <madwolf@hackmasters.net>]
1730 *) Introduce the possibility to access global variables through
1731 functions on platform were that's the best way to handle exporting
1732 global variables in shared libraries. To enable this functionality,
1733 one must configure with "EXPORT_VAR_AS_FN" or defined the C macro
1734 "OPENSSL_EXPORT_VAR_AS_FUNCTION" in crypto/opensslconf.h (the latter
1735 is normally done by Configure or something similar).
1737 To implement a global variable, use the macro OPENSSL_IMPLEMENT_GLOBAL
1738 in the source file (foo.c) like this:
1740 OPENSSL_IMPLEMENT_GLOBAL(int,foo)=1;
1741 OPENSSL_IMPLEMENT_GLOBAL(double,bar);
1743 To declare a global variable, use the macros OPENSSL_DECLARE_GLOBAL
1744 and OPENSSL_GLOBAL_REF in the header file (foo.h) like this:
1746 OPENSSL_DECLARE_GLOBAL(int,foo);
1747 #define foo OPENSSL_GLOBAL_REF(foo)
1748 OPENSSL_DECLARE_GLOBAL(double,bar);
1749 #define bar OPENSSL_GLOBAL_REF(bar)
1751 The #defines are very important, and therefore so is including the
1752 header file everywhere where the defined globals are used.
1754 The macro OPENSSL_EXPORT_VAR_AS_FUNCTION also affects the definition
1755 of ASN.1 items, but that structure is a bit different.
1757 The largest change is in util/mkdef.pl which has been enhanced with
1758 better and easier to understand logic to choose which symbols should
1759 go into the Windows .def files as well as a number of fixes and code
1760 cleanup (among others, algorithm keywords are now sorted
1761 lexicographically to avoid constant rewrites).
1764 *) In BN_div() keep a copy of the sign of 'num' before writing the
1765 result to 'rm' because if rm==num the value will be overwritten
1766 and produce the wrong result if 'num' is negative: this caused
1767 problems with BN_mod() and BN_nnmod().
1770 *) Function OCSP_request_verify(). This checks the signature on an
1771 OCSP request and verifies the signer certificate. The signer
1772 certificate is just checked for a generic purpose and OCSP request
1776 *) Add OCSP_check_validity() function to check the validity of OCSP
1777 responses. OCSP responses are prepared in real time and may only
1778 be a few seconds old. Simply checking that the current time lies
1779 between thisUpdate and nextUpdate max reject otherwise valid responses
1780 caused by either OCSP responder or client clock inaccuracy. Instead
1781 we allow thisUpdate and nextUpdate to fall within a certain period of
1782 the current time. The age of the response can also optionally be
1783 checked. Two new options -validity_period and -status_age added to
1787 *) If signature or public key algorithm is unrecognized print out its
1788 OID rather that just UNKNOWN.
1791 *) Change OCSP_cert_to_id() to tolerate a NULL subject certificate and
1792 OCSP_cert_id_new() a NULL serialNumber. This allows a partial certificate
1793 ID to be generated from the issuer certificate alone which can then be
1794 passed to OCSP_id_issuer_cmp().
1797 *) New compilation option ASN1_ITEM_FUNCTIONS. This causes the new
1798 ASN1 modules to export functions returning ASN1_ITEM pointers
1799 instead of the ASN1_ITEM structures themselves. This adds several
1800 new macros which allow the underlying ASN1 function/structure to
1801 be accessed transparently. As a result code should not use ASN1_ITEM
1802 references directly (such as &X509_it) but instead use the relevant
1803 macros (such as ASN1_ITEM_rptr(X509)). This option is to allow
1804 use of the new ASN1 code on platforms where exporting structures
1805 is problematical (for example in shared libraries) but exporting
1806 functions returning pointers to structures is not.
1809 *) Add support for overriding the generation of SSL/TLS session IDs.
1810 These callbacks can be registered either in an SSL_CTX or per SSL.
1811 The purpose of this is to allow applications to control, if they wish,
1812 the arbitrary values chosen for use as session IDs, particularly as it
1813 can be useful for session caching in multiple-server environments. A
1814 command-line switch for testing this (and any client code that wishes
1815 to use such a feature) has been added to "s_server".
1816 [Geoff Thorpe, Lutz Jaenicke]
1818 *) Modify mkdef.pl to recognise and parse preprocessor conditionals
1819 of the form '#if defined(...) || defined(...) || ...' and
1820 '#if !defined(...) && !defined(...) && ...'. This also avoids
1821 the growing number of special cases it was previously handling.
1824 *) Make all configuration macros available for application by making
1825 sure they are available in opensslconf.h, by giving them names starting
1826 with "OPENSSL_" to avoid conflicts with other packages and by making
1827 sure e_os2.h will cover all platform-specific cases together with
1829 Additionally, it is now possible to define configuration/platform-
1830 specific names (called "system identities"). In the C code, these
1831 are prefixed with "OPENSSL_SYSNAME_". e_os2.h will create another
1832 macro with the name beginning with "OPENSSL_SYS_", which is determined
1833 from "OPENSSL_SYSNAME_*" or compiler-specific macros depending on
1837 *) New option -set_serial to 'req' and 'x509' this allows the serial
1838 number to use to be specified on the command line. Previously self
1839 signed certificates were hard coded with serial number 0 and the
1840 CA options of 'x509' had to use a serial number in a file which was
1844 *) New options to 'ca' utility to support V2 CRL entry extensions.
1845 Currently CRL reason, invalidity date and hold instruction are
1846 supported. Add new CRL extensions to V3 code and some new objects.
1849 *) New function EVP_CIPHER_CTX_set_padding() this is used to
1850 disable standard block padding (aka PKCS#5 padding) in the EVP
1851 API, which was previously mandatory. This means that the data is
1852 not padded in any way and so the total length much be a multiple
1853 of the block size, otherwise an error occurs.
1856 *) Initial (incomplete) OCSP SSL support.
1859 *) New function OCSP_parse_url(). This splits up a URL into its host,
1860 port and path components: primarily to parse OCSP URLs. New -url
1861 option to ocsp utility.
1864 *) New nonce behavior. The return value of OCSP_check_nonce() now
1865 reflects the various checks performed. Applications can decide
1866 whether to tolerate certain situations such as an absent nonce
1867 in a response when one was present in a request: the ocsp application
1868 just prints out a warning. New function OCSP_add1_basic_nonce()
1869 this is to allow responders to include a nonce in a response even if
1870 the request is nonce-less.
1873 *) Disable stdin buffering in load_cert (apps/apps.c) so that no certs are
1874 skipped when using openssl x509 multiple times on a single input file,
1875 e.g. "(openssl x509 -out cert1; openssl x509 -out cert2) <certs".
1878 *) Make ASN1_UTCTIME_set_string() and ASN1_GENERALIZEDTIME_set_string()
1879 set string type: to handle setting ASN1_TIME structures. Fix ca
1880 utility to correctly initialize revocation date of CRLs.
1883 *) New option SSL_OP_CIPHER_SERVER_PREFERENCE allows the server to override
1884 the clients preferred ciphersuites and rather use its own preferences.
1885 Should help to work around M$ SGC (Server Gated Cryptography) bug in
1886 Internet Explorer by ensuring unchanged hash method during stepup.
1887 (Also replaces the broken/deactivated SSL_OP_NON_EXPORT_FIRST option.)
1890 *) Make mkdef.pl recognise all DECLARE_ASN1 macros, change rijndael
1891 to aes and add a new 'exist' option to print out symbols that don't
1895 *) Additional options to ocsp utility to allow flags to be set and
1896 additional certificates supplied.
1899 *) Add the option -VAfile to 'openssl ocsp', so the user can give the
1900 OCSP client a number of certificate to only verify the response
1904 *) Update Rijndael code to version 3.0 and change EVP AES ciphers to
1905 handle the new API. Currently only ECB, CBC modes supported. Add new
1908 Add TLS AES ciphersuites as described in RFC3268, "Advanced
1909 Encryption Standard (AES) Ciphersuites for Transport Layer
1910 Security (TLS)". (In beta versions of OpenSSL 0.9.7, these were
1911 not enabled by default and were not part of the "ALL" ciphersuite
1912 alias because they were not yet official; they could be
1913 explicitly requested by specifying the "AESdraft" ciphersuite
1914 group alias. In the final release of OpenSSL 0.9.7, the group
1915 alias is called "AES" and is part of "ALL".)
1916 [Ben Laurie, Steve Henson, Bodo Moeller]
1918 *) New function OCSP_copy_nonce() to copy nonce value (if present) from
1919 request to response.
1922 *) Functions for OCSP responders. OCSP_request_onereq_count(),
1923 OCSP_request_onereq_get0(), OCSP_onereq_get0_id() and OCSP_id_get0_info()
1924 extract information from a certificate request. OCSP_response_create()
1925 creates a response and optionally adds a basic response structure.
1926 OCSP_basic_add1_status() adds a complete single response to a basic
1927 response and returns the OCSP_SINGLERESP structure just added (to allow
1928 extensions to be included for example). OCSP_basic_add1_cert() adds a
1929 certificate to a basic response and OCSP_basic_sign() signs a basic
1930 response with various flags. New helper functions ASN1_TIME_check()
1931 (checks validity of ASN1_TIME structure) and ASN1_TIME_to_generalizedtime()
1932 (converts ASN1_TIME to GeneralizedTime).
1935 *) Various new functions. EVP_Digest() combines EVP_Digest{Init,Update,Final}()
1936 in a single operation. X509_get0_pubkey_bitstr() extracts the public_key
1937 structure from a certificate. X509_pubkey_digest() digests the public_key
1938 contents: this is used in various key identifiers.
1941 *) Make sk_sort() tolerate a NULL argument.
1942 [Steve Henson reported by Massimiliano Pala <madwolf@comune.modena.it>]
1944 *) New OCSP verify flag OCSP_TRUSTOTHER. When set the "other" certificates
1945 passed by the function are trusted implicitly. If any of them signed the
1946 response then it is assumed to be valid and is not verified.
1949 *) In PKCS7_set_type() initialise content_type in PKCS7_ENC_CONTENT
1950 to data. This was previously part of the PKCS7 ASN1 code. This
1951 was causing problems with OpenSSL created PKCS#12 and PKCS#7 structures.
1952 [Steve Henson, reported by Kenneth R. Robinette
1953 <support@securenetterm.com>]
1955 *) Add CRYPTO_push_info() and CRYPTO_pop_info() calls to new ASN1
1956 routines: without these tracing memory leaks is very painful.
1957 Fix leaks in PKCS12 and PKCS7 routines.
1960 *) Make X509_time_adj() cope with the new behaviour of ASN1_TIME_new().
1961 Previously it initialised the 'type' argument to V_ASN1_UTCTIME which
1962 effectively meant GeneralizedTime would never be used. Now it
1963 is initialised to -1 but X509_time_adj() now has to check the value
1964 and use ASN1_TIME_set() if the value is not V_ASN1_UTCTIME or
1965 V_ASN1_GENERALIZEDTIME, without this it always uses GeneralizedTime.
1966 [Steve Henson, reported by Kenneth R. Robinette
1967 <support@securenetterm.com>]
1969 *) Fixes to BN_to_ASN1_INTEGER when bn is zero. This would previously
1970 result in a zero length in the ASN1_INTEGER structure which was
1971 not consistent with the structure when d2i_ASN1_INTEGER() was used
1972 and would cause ASN1_INTEGER_cmp() to fail. Enhance s2i_ASN1_INTEGER()
1973 to cope with hex and negative integers. Fix bug in i2a_ASN1_INTEGER()
1974 where it did not print out a minus for negative ASN1_INTEGER.
1977 *) Add summary printout to ocsp utility. The various functions which
1978 convert status values to strings have been renamed to:
1979 OCSP_response_status_str(), OCSP_cert_status_str() and
1980 OCSP_crl_reason_str() and are no longer static. New options
1981 to verify nonce values and to disable verification. OCSP response
1982 printout format cleaned up.
1985 *) Add additional OCSP certificate checks. These are those specified
1986 in RFC2560. This consists of two separate checks: the CA of the
1987 certificate being checked must either be the OCSP signer certificate
1988 or the issuer of the OCSP signer certificate. In the latter case the
1989 OCSP signer certificate must contain the OCSP signing extended key
1990 usage. This check is performed by attempting to match the OCSP
1991 signer or the OCSP signer CA to the issuerNameHash and issuerKeyHash
1992 in the OCSP_CERTID structures of the response.
1995 *) Initial OCSP certificate verification added to OCSP_basic_verify()
1996 and related routines. This uses the standard OpenSSL certificate
1997 verify routines to perform initial checks (just CA validity) and
1998 to obtain the certificate chain. Then additional checks will be
1999 performed on the chain. Currently the root CA is checked to see
2000 if it is explicitly trusted for OCSP signing. This is used to set
2001 a root CA as a global signing root: that is any certificate that
2002 chains to that CA is an acceptable OCSP signing certificate.
2005 *) New '-extfile ...' option to 'openssl ca' for reading X.509v3
2006 extensions from a separate configuration file.
2007 As when reading extensions from the main configuration file,
2008 the '-extensions ...' option may be used for specifying the
2010 [Massimiliano Pala <madwolf@comune.modena.it>]
2012 *) New OCSP utility. Allows OCSP requests to be generated or
2013 read. The request can be sent to a responder and the output
2014 parsed, outputed or printed in text form. Not complete yet:
2015 still needs to check the OCSP response validity.
2018 *) New subcommands for 'openssl ca':
2019 'openssl ca -status <serial>' prints the status of the cert with
2020 the given serial number (according to the index file).
2021 'openssl ca -updatedb' updates the expiry status of certificates
2023 [Massimiliano Pala <madwolf@comune.modena.it>]
2025 *) New '-newreq-nodes' command option to CA.pl. This is like
2026 '-newreq', but calls 'openssl req' with the '-nodes' option
2027 so that the resulting key is not encrypted.
2028 [Damien Miller <djm@mindrot.org>]
2030 *) New configuration for the GNU Hurd.
2031 [Jonathan Bartlett <johnnyb@wolfram.com> via Richard Levitte]
2033 *) Initial code to implement OCSP basic response verify. This
2034 is currently incomplete. Currently just finds the signer's
2035 certificate and verifies the signature on the response.
2038 *) New SSLeay_version code SSLEAY_DIR to determine the compiled-in
2039 value of OPENSSLDIR. This is available via the new '-d' option
2040 to 'openssl version', and is also included in 'openssl version -a'.
2043 *) Allowing defining memory allocation callbacks that will be given
2044 file name and line number information in additional arguments
2045 (a const char* and an int). The basic functionality remains, as
2046 well as the original possibility to just replace malloc(),
2047 realloc() and free() by functions that do not know about these
2048 additional arguments. To register and find out the current
2049 settings for extended allocation functions, the following
2050 functions are provided:
2052 CRYPTO_set_mem_ex_functions
2053 CRYPTO_set_locked_mem_ex_functions
2054 CRYPTO_get_mem_ex_functions
2055 CRYPTO_get_locked_mem_ex_functions
2057 These work the same way as CRYPTO_set_mem_functions and friends.
2058 CRYPTO_get_[locked_]mem_functions now writes 0 where such an
2059 extended allocation function is enabled.
2060 Similarly, CRYPTO_get_[locked_]mem_ex_functions writes 0 where
2061 a conventional allocation function is enabled.
2062 [Richard Levitte, Bodo Moeller]
2064 *) Finish off removing the remaining LHASH function pointer casts.
2065 There should no longer be any prototype-casting required when using
2066 the LHASH abstraction, and any casts that remain are "bugs". See
2067 the callback types and macros at the head of lhash.h for details
2068 (and "OBJ_cleanup" in crypto/objects/obj_dat.c as an example).
2071 *) Add automatic query of EGD sockets in RAND_poll() for the unix variant.
2072 If /dev/[u]random devices are not available or do not return enough
2073 entropy, EGD style sockets (served by EGD or PRNGD) will automatically
2075 The locations /var/run/egd-pool, /dev/egd-pool, /etc/egd-pool, and
2076 /etc/entropy will be queried once each in this sequence, quering stops
2077 when enough entropy was collected without querying more sockets.
2080 *) Change the Unix RAND_poll() variant to be able to poll several
2081 random devices, as specified by DEVRANDOM, until a sufficient amount
2082 of data has been collected. We spend at most 10 ms on each file
2083 (select timeout) and read in non-blocking mode. DEVRANDOM now
2084 defaults to the list "/dev/urandom", "/dev/random", "/dev/srandom"
2085 (previously it was just the string "/dev/urandom"), so on typical
2086 platforms the 10 ms delay will never occur.
2087 Also separate out the Unix variant to its own file, rand_unix.c.
2088 For VMS, there's a currently-empty rand_vms.c.
2091 *) Move OCSP client related routines to ocsp_cl.c. These
2092 provide utility functions which an application needing
2093 to issue a request to an OCSP responder and analyse the
2094 response will typically need: as opposed to those which an
2095 OCSP responder itself would need which will be added later.
2097 OCSP_request_sign() signs an OCSP request with an API similar
2098 to PKCS7_sign(). OCSP_response_status() returns status of OCSP
2099 response. OCSP_response_get1_basic() extracts basic response
2100 from response. OCSP_resp_find_status(): finds and extracts status
2101 information from an OCSP_CERTID structure (which will be created
2102 when the request structure is built). These are built from lower
2103 level functions which work on OCSP_SINGLERESP structures but
2104 wont normally be used unless the application wishes to examine
2105 extensions in the OCSP response for example.
2107 Replace nonce routines with a pair of functions.
2108 OCSP_request_add1_nonce() adds a nonce value and optionally
2109 generates a random value. OCSP_check_nonce() checks the
2110 validity of the nonce in an OCSP response.
2113 *) Change function OCSP_request_add() to OCSP_request_add0_id().
2114 This doesn't copy the supplied OCSP_CERTID and avoids the
2115 need to free up the newly created id. Change return type
2116 to OCSP_ONEREQ to return the internal OCSP_ONEREQ structure.
2117 This can then be used to add extensions to the request.
2118 Deleted OCSP_request_new(), since most of its functionality
2119 is now in OCSP_REQUEST_new() (and the case insensitive name
2120 clash) apart from the ability to set the request name which
2121 will be added elsewhere.
2124 *) Update OCSP API. Remove obsolete extensions argument from
2125 various functions. Extensions are now handled using the new
2126 OCSP extension code. New simple OCSP HTTP function which
2127 can be used to send requests and parse the response.
2130 *) Fix the PKCS#7 (S/MIME) code to work with new ASN1. Two new
2131 ASN1_ITEM structures help with sign and verify. PKCS7_ATTR_SIGN
2132 uses the special reorder version of SET OF to sort the attributes
2133 and reorder them to match the encoded order. This resolves a long
2134 standing problem: a verify on a PKCS7 structure just after signing
2135 it used to fail because the attribute order did not match the
2136 encoded order. PKCS7_ATTR_VERIFY does not reorder the attributes:
2137 it uses the received order. This is necessary to tolerate some broken
2138 software that does not order SET OF. This is handled by encoding
2139 as a SEQUENCE OF but using implicit tagging (with UNIVERSAL class)
2140 to produce the required SET OF.
2143 *) Have mk1mf.pl generate the macros OPENSSL_BUILD_SHLIBCRYPTO and
2144 OPENSSL_BUILD_SHLIBSSL and use them appropriately in the header
2145 files to get correct declarations of the ASN.1 item variables.
2148 *) Rewrite of PKCS#12 code to use new ASN1 functionality. Replace many
2149 PKCS#12 macros with real functions. Fix two unrelated ASN1 bugs:
2150 asn1_check_tlen() would sometimes attempt to use 'ctx' when it was
2151 NULL and ASN1_TYPE was not dereferenced properly in asn1_ex_c2i().
2152 New ASN1 macro: DECLARE_ASN1_ITEM() which just declares the relevant
2153 ASN1_ITEM and no wrapper functions.
2156 *) New functions or ASN1_item_d2i_fp() and ASN1_item_d2i_bio(). These
2157 replace the old function pointer based I/O routines. Change most of
2158 the *_d2i_bio() and *_d2i_fp() functions to use these.
2161 *) Enhance mkdef.pl to be more accepting about spacing in C preprocessor
2162 lines, recognice more "algorithms" that can be deselected, and make
2163 it complain about algorithm deselection that isn't recognised.
2166 *) New ASN1 functions to handle dup, sign, verify, digest, pack and
2167 unpack operations in terms of ASN1_ITEM. Modify existing wrappers
2168 to use new functions. Add NO_ASN1_OLD which can be set to remove
2169 some old style ASN1 functions: this can be used to determine if old
2170 code will still work when these eventually go away.
2173 *) New extension functions for OCSP structures, these follow the
2174 same conventions as certificates and CRLs.
2177 *) New function X509V3_add1_i2d(). This automatically encodes and
2178 adds an extension. Its behaviour can be customised with various
2179 flags to append, replace or delete. Various wrappers added for
2180 certifcates and CRLs.
2183 *) Fix to avoid calling the underlying ASN1 print routine when
2184 an extension cannot be parsed. Correct a typo in the
2185 OCSP_SERVICELOC extension. Tidy up print OCSP format.
2188 *) Make mkdef.pl parse some of the ASN1 macros and add apropriate
2189 entries for variables.
2192 *) Add functionality to apps/openssl.c for detecting locking
2193 problems: As the program is single-threaded, all we have
2194 to do is register a locking callback using an array for
2195 storing which locks are currently held by the program.
2198 *) Use a lock around the call to CRYPTO_get_ex_new_index() in
2199 SSL_get_ex_data_X509_STORE_idx(), which is used in
2200 ssl_verify_cert_chain() and thus can be called at any time
2201 during TLS/SSL handshakes so that thread-safety is essential.
2202 Unfortunately, the ex_data design is not at all suited
2203 for multi-threaded use, so it probably should be abolished.
2206 *) Added Broadcom "ubsec" ENGINE to OpenSSL.
2207 [Broadcom, tweaked and integrated by Geoff Thorpe]
2209 *) Move common extension printing code to new function
2210 X509V3_print_extensions(). Reorganise OCSP print routines and
2211 implement some needed OCSP ASN1 functions. Add OCSP extensions.
2214 *) New function X509_signature_print() to remove duplication in some
2218 *) Add a special meaning when SET OF and SEQUENCE OF flags are both
2219 set (this was treated exactly the same as SET OF previously). This
2220 is used to reorder the STACK representing the structure to match the
2221 encoding. This will be used to get round a problem where a PKCS7
2222 structure which was signed could not be verified because the STACK
2223 order did not reflect the encoded order.
2226 *) Reimplement the OCSP ASN1 module using the new code.
2229 *) Update the X509V3 code to permit the use of an ASN1_ITEM structure
2230 for its ASN1 operations. The old style function pointers still exist
2231 for now but they will eventually go away.
2234 *) Merge in replacement ASN1 code from the ASN1 branch. This almost
2235 completely replaces the old ASN1 functionality with a table driven
2236 encoder and decoder which interprets an ASN1_ITEM structure describing
2237 the ASN1 module. Compatibility with the existing ASN1 API (i2d,d2i) is
2238 largely maintained. Almost all of the old asn1_mac.h macro based ASN1
2239 has also been converted to the new form.
2242 *) Change BN_mod_exp_recp so that negative moduli are tolerated
2243 (the sign is ignored). Similarly, ignore the sign in BN_MONT_CTX_set
2244 so that BN_mod_exp_mont and BN_mod_exp_mont_word work
2245 for negative moduli.
2248 *) Fix BN_uadd and BN_usub: Always return non-negative results instead
2249 of not touching the result's sign bit.
2252 *) BN_div bugfix: If the result is 0, the sign (res->neg) must not be
2256 *) Changed the LHASH code to use prototypes for callbacks, and created
2257 macros to declare and implement thin (optionally static) functions
2258 that provide type-safety and avoid function pointer casting for the
2259 type-specific callbacks.
2262 *) Added Kerberos Cipher Suites to be used with TLS, as written in
2264 [Veers Staats <staatsvr@asc.hpc.mil>,
2265 Jeffrey Altman <jaltman@columbia.edu>, via Richard Levitte]
2267 *) Reformat the FAQ so the different questions and answers can be divided
2268 in sections depending on the subject.
2271 *) Have the zlib compression code load ZLIB.DLL dynamically under
2275 *) New function BN_mod_sqrt for computing square roots modulo a prime
2276 (using the probabilistic Tonelli-Shanks algorithm unless
2277 p == 3 (mod 4) or p == 5 (mod 8), which are cases that can
2278 be handled deterministically).
2279 [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller]
2281 *) Make BN_mod_inverse faster by explicitly handling small quotients
2282 in the Euclid loop. (Speed gain about 20% for small moduli [256 or
2283 512 bits], about 30% for larger ones [1024 or 2048 bits].)
2286 *) New function BN_kronecker.
2289 *) Fix BN_gcd so that it works on negative inputs; the result is
2290 positive unless both parameters are zero.
2291 Previously something reasonably close to an infinite loop was
2292 possible because numbers could be growing instead of shrinking
2293 in the implementation of Euclid's algorithm.
2296 *) Fix BN_is_word() and BN_is_one() macros to take into account the
2297 sign of the number in question.
2299 Fix BN_is_word(a,w) to work correctly for w == 0.
2301 The old BN_is_word(a,w) macro is now called BN_abs_is_word(a,w)
2302 because its test if the absolute value of 'a' equals 'w'.
2303 Note that BN_abs_is_word does *not* handle w == 0 reliably;
2304 it exists mostly for use in the implementations of BN_is_zero(),
2305 BN_is_one(), and BN_is_word().
2308 *) New function BN_swap.
2311 *) Use BN_nnmod instead of BN_mod in crypto/bn/bn_exp.c so that
2312 the exponentiation functions are more likely to produce reasonable
2313 results on negative inputs.
2316 *) Change BN_mod_mul so that the result is always non-negative.
2317 Previously, it could be negative if one of the factors was negative;
2318 I don't think anyone really wanted that behaviour.
2321 *) Move BN_mod_... functions into new file crypto/bn/bn_mod.c
2322 (except for exponentiation, which stays in crypto/bn/bn_exp.c,
2323 and BN_mod_mul_reciprocal, which stays in crypto/bn/bn_recp.c)
2324 and add new functions:
2333 BN_mod_lshift1_quick
2337 These functions always generate non-negative results.
2339 BN_nnmod otherwise is like BN_mod (if BN_mod computes a remainder r
2340 such that |m| < r < 0, BN_nnmod will output rem + |m| instead).
2342 BN_mod_XXX_quick(r, a, [b,] m) generates the same result as
2343 BN_mod_XXX(r, a, [b,] m, ctx), but requires that a [and b]
2344 be reduced modulo m.
2345 [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller]
2348 The following entry accidentily appeared in the CHANGES file
2349 distributed with OpenSSL 0.9.7. The modifications described in
2350 it do *not* apply to OpenSSL 0.9.7.
2352 *) Remove a few calls to bn_wexpand() in BN_sqr() (the one in there
2353 was actually never needed) and in BN_mul(). The removal in BN_mul()
2354 required a small change in bn_mul_part_recursive() and the addition
2355 of the functions bn_cmp_part_words(), bn_sub_part_words() and
2356 bn_add_part_words(), which do the same thing as bn_cmp_words(),
2357 bn_sub_words() and bn_add_words() except they take arrays with
2362 *) In 'openssl passwd', verify passwords read from the terminal
2363 unless the '-salt' option is used (which usually means that
2364 verification would just waste user's time since the resulting
2365 hash is going to be compared with some given password hash)
2366 or the new '-noverify' option is used.
2368 This is an incompatible change, but it does not affect
2369 non-interactive use of 'openssl passwd' (passwords on the command
2370 line, '-stdin' option, '-in ...' option) and thus should not
2374 *) Remove all references to RSAref, since there's no more need for it.
2377 *) Make DSO load along a path given through an environment variable
2378 (SHLIB_PATH) with shl_load().
2381 *) Constify the ENGINE code as a result of BIGNUM constification.
2382 Also constify the RSA code and most things related to it. In a
2383 few places, most notable in the depth of the ASN.1 code, ugly
2384 casts back to non-const were required (to be solved at a later
2388 *) Make it so the openssl application has all engines loaded by default.
2391 *) Constify the BIGNUM routines a little more.
2394 *) Add the following functions:
2396 ENGINE_load_cswift()
2398 ENGINE_load_atalla()
2400 ENGINE_load_builtin_engines()
2402 That way, an application can itself choose if external engines that
2403 are built-in in OpenSSL shall ever be used or not. The benefit is
2404 that applications won't have to be linked with libdl or other dso
2405 libraries unless it's really needed.
2407 Changed 'openssl engine' to load all engines on demand.
2408 Changed the engine header files to avoid the duplication of some
2409 declarations (they differed!).
2412 *) 'openssl engine' can now list capabilities.
2415 *) Better error reporting in 'openssl engine'.
2418 *) Never call load_dh_param(NULL) in s_server.
2421 *) Add engine application. It can currently list engines by name and
2422 identity, and test if they are actually available.
2425 *) Improve RPM specification file by forcing symbolic linking and making
2426 sure the installed documentation is also owned by root.root.
2427 [Damien Miller <djm@mindrot.org>]
2429 *) Give the OpenSSL applications more possibilities to make use of
2430 keys (public as well as private) handled by engines.
2433 *) Add OCSP code that comes from CertCo.
2436 *) Add VMS support for the Rijndael code.
2439 *) Added untested support for Nuron crypto accelerator.
2442 *) Add support for external cryptographic devices. This code was
2443 previously distributed separately as the "engine" branch.
2444 [Geoff Thorpe, Richard Levitte]
2446 *) Rework the filename-translation in the DSO code. It is now possible to
2447 have far greater control over how a "name" is turned into a filename
2448 depending on the operating environment and any oddities about the
2449 different shared library filenames on each system.
2452 *) Support threads on FreeBSD-elf in Configure.
2455 *) Fix for SHA1 assembly problem with MASM: it produces
2456 warnings about corrupt line number information when assembling
2457 with debugging information. This is caused by the overlapping
2459 [Bernd Matthes <mainbug@celocom.de>, Steve Henson]
2462 NCONF_get_number() has no error checking at all. As a replacement,
2463 NCONF_get_number_e() is defined (_e for "error checking") and is
2464 promoted strongly. The old NCONF_get_number is kept around for
2465 binary backward compatibility.
2466 Make it possible for methods to load from something other than a BIO,
2467 by providing a function pointer that is given a name instead of a BIO.
2468 For example, this could be used to load configuration data from an
2472 *) Fix for non blocking accept BIOs. Added new I/O special reason
2473 BIO_RR_ACCEPT to cover this case. Previously use of accept BIOs
2474 with non blocking I/O was not possible because no retry code was
2475 implemented. Also added new SSL code SSL_WANT_ACCEPT to cover
2479 *) Added the beginnings of Rijndael support.
2482 *) Fix for bug in DirectoryString mask setting. Add support for
2483 X509_NAME_print_ex() in 'req' and X509_print_ex() function
2484 to allow certificate printing to more controllable, additional
2485 'certopt' option to 'x509' to allow new printing options to be
2489 *) Clean old EAY MD5 hack from e_os.h.
2492 Changes between 0.9.6j and 0.9.6k [xx XXX 2003]
2494 *) Change X509_certificate_type() to mark the key as exported/exportable
2495 when it's 512 *bits* long, not 512 bytes.
2498 Changes between 0.9.6i and 0.9.6j [10 Apr 2003]
2500 *) Countermeasure against the Klima-Pokorny-Rosa extension of
2501 Bleichbacher's attack on PKCS #1 v1.5 padding: treat
2502 a protocol version number mismatch like a decryption error
2503 in ssl3_get_client_key_exchange (ssl/s3_srvr.c).
2506 *) Turn on RSA blinding by default in the default implementation
2507 to avoid a timing attack. Applications that don't want it can call
2508 RSA_blinding_off() or use the new flag RSA_FLAG_NO_BLINDING.
2509 They would be ill-advised to do so in most cases.
2510 [Ben Laurie, Steve Henson, Geoff Thorpe, Bodo Moeller]
2512 *) Change RSA blinding code so that it works when the PRNG is not
2513 seeded (in this case, the secret RSA exponent is abused as
2514 an unpredictable seed -- if it is not unpredictable, there
2515 is no point in blinding anyway). Make RSA blinding thread-safe
2516 by remembering the creator's thread ID in rsa->blinding and
2517 having all other threads use local one-time blinding factors
2518 (this requires more computation than sharing rsa->blinding, but
2519 avoids excessive locking; and if an RSA object is not shared
2520 between threads, blinding will still be very fast).
2523 Changes between 0.9.6h and 0.9.6i [19 Feb 2003]
2525 *) In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked
2526 via timing by performing a MAC computation even if incorrrect
2527 block cipher padding has been found. This is a countermeasure
2528 against active attacks where the attacker has to distinguish
2529 between bad padding and a MAC verification error. (CAN-2003-0078)
2531 [Bodo Moeller; problem pointed out by Brice Canvel (EPFL),
2532 Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and
2533 Martin Vuagnoux (EPFL, Ilion)]
2535 Changes between 0.9.6g and 0.9.6h [5 Dec 2002]
2537 *) New function OPENSSL_cleanse(), which is used to cleanse a section of
2538 memory from it's contents. This is done with a counter that will
2539 place alternating values in each byte. This can be used to solve
2540 two issues: 1) the removal of calls to memset() by highly optimizing
2541 compilers, and 2) cleansing with other values than 0, since those can
2542 be read through on certain media, for example a swap space on disk.
2545 *) Bugfix: client side session caching did not work with external caching,
2546 because the session->cipher setting was not restored when reloading
2547 from the external cache. This problem was masked, when
2548 SSL_OP_NETSCAPE_REUSE_CIPHER_CHANGE_BUG (part of SSL_OP_ALL) was set.
2549 (Found by Steve Haslam <steve@araqnid.ddts.net>.)
2552 *) Fix client_certificate (ssl/s2_clnt.c): The permissible total
2553 length of the REQUEST-CERTIFICATE message is 18 .. 34, not 17 .. 33.
2554 [Zeev Lieber <zeev-l@yahoo.com>]
2556 *) Undo an undocumented change introduced in 0.9.6e which caused
2557 repeated calls to OpenSSL_add_all_ciphers() and
2558 OpenSSL_add_all_digests() to be ignored, even after calling
2562 *) Change the default configuration reader to deal with last line not
2563 being properly terminated.
2566 *) Change X509_NAME_cmp() so it applies the special rules on handling
2567 DN values that are of type PrintableString, as well as RDNs of type
2568 emailAddress where the value has the type ia5String.
2569 [stefank@valicert.com via Richard Levitte]
2571 *) Add a SSL_SESS_CACHE_NO_INTERNAL_STORE flag to take over half
2572 the job SSL_SESS_CACHE_NO_INTERNAL_LOOKUP was inconsistently
2573 doing, define a new flag (SSL_SESS_CACHE_NO_INTERNAL) to be
2574 the bitwise-OR of the two for use by the majority of applications
2575 wanting this behaviour, and update the docs. The documented
2576 behaviour and actual behaviour were inconsistent and had been
2577 changing anyway, so this is more a bug-fix than a behavioural
2579 [Geoff Thorpe, diagnosed by Nadav Har'El]
2581 *) Don't impose a 16-byte length minimum on session IDs in ssl/s3_clnt.c
2582 (the SSL 3.0 and TLS 1.0 specifications allow any length up to 32 bytes).
2585 *) Fix initialization code race conditions in
2586 SSLv23_method(), SSLv23_client_method(), SSLv23_server_method(),
2587 SSLv2_method(), SSLv2_client_method(), SSLv2_server_method(),
2588 SSLv3_method(), SSLv3_client_method(), SSLv3_server_method(),
2589 TLSv1_method(), TLSv1_client_method(), TLSv1_server_method(),
2590 ssl2_get_cipher_by_char(),
2591 ssl3_get_cipher_by_char().
2592 [Patrick McCormick <patrick@tellme.com>, Bodo Moeller]
2594 *) Reorder cleanup sequence in SSL_CTX_free(): only remove the ex_data after
2595 the cached sessions are flushed, as the remove_cb() might use ex_data
2596 contents. Bug found by Sam Varshavchik <mrsam@courier-mta.com>
2597 (see [openssl.org #212]).
2598 [Geoff Thorpe, Lutz Jaenicke]
2600 *) Fix typo in OBJ_txt2obj which incorrectly passed the content
2601 length, instead of the encoding length to d2i_ASN1_OBJECT.
2604 Changes between 0.9.6f and 0.9.6g [9 Aug 2002]
2606 *) [In 0.9.6g-engine release:]
2607 Fix crypto/engine/vendor_defns/cswift.h for WIN32 (use '_stdcall').
2608 [Lynn Gazis <lgazis@rainbow.com>]
2610 Changes between 0.9.6e and 0.9.6f [8 Aug 2002]
2612 *) Fix ASN1 checks. Check for overflow by comparing with LONG_MAX
2613 and get fix the header length calculation.
2614 [Florian Weimer <Weimer@CERT.Uni-Stuttgart.DE>,
2615 Alon Kantor <alonk@checkpoint.com> (and others),
2618 *) Use proper error handling instead of 'assertions' in buffer
2619 overflow checks added in 0.9.6e. This prevents DoS (the
2620 assertions could call abort()).
2621 [Arne Ansper <arne@ats.cyber.ee>, Bodo Moeller]
2623 Changes between 0.9.6d and 0.9.6e [30 Jul 2002]
2625 *) Add various sanity checks to asn1_get_length() to reject
2626 the ASN1 length bytes if they exceed sizeof(long), will appear
2627 negative or the content length exceeds the length of the
2629 [Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>]
2631 *) Fix cipher selection routines: ciphers without encryption had no flags
2632 for the cipher strength set and where therefore not handled correctly
2633 by the selection routines (PR #130).
2636 *) Fix EVP_dsa_sha macro.
2640 SSL_OP_DONT_INSERT_EMPTY_FRAGMENTS
2641 for disabling the SSL 3.0/TLS 1.0 CBC vulnerability countermeasure
2642 that was added in OpenSSL 0.9.6d.
2644 As the countermeasure turned out to be incompatible with some
2645 broken SSL implementations, the new option is part of SSL_OP_ALL.
2646 SSL_OP_ALL is usually employed when compatibility with weird SSL
2647 implementations is desired (e.g. '-bugs' option to 's_client' and
2648 's_server'), so the new option is automatically set in many
2652 *) Changes in security patch:
2654 Changes marked "(CHATS)" were sponsored by the Defense Advanced
2655 Research Projects Agency (DARPA) and Air Force Research Laboratory,
2656 Air Force Materiel Command, USAF, under agreement number
2659 *) Add various sanity checks to asn1_get_length() to reject
2660 the ASN1 length bytes if they exceed sizeof(long), will appear
2661 negative or the content length exceeds the length of the
2662 supplied buffer. (CAN-2002-0659)
2663 [Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>]
2665 *) Assertions for various potential buffer overflows, not known to
2667 [Ben Laurie (CHATS)]
2669 *) Various temporary buffers to hold ASCII versions of integers were
2670 too small for 64 bit platforms. (CAN-2002-0655)
2671 [Matthew Byng-Maddick <mbm@aldigital.co.uk> and Ben Laurie (CHATS)>
2673 *) Remote buffer overflow in SSL3 protocol - an attacker could
2674 supply an oversized session ID to a client. (CAN-2002-0656)
2675 [Ben Laurie (CHATS)]
2677 *) Remote buffer overflow in SSL2 protocol - an attacker could
2678 supply an oversized client master key. (CAN-2002-0656)
2679 [Ben Laurie (CHATS)]
2681 Changes between 0.9.6c and 0.9.6d [9 May 2002]
2683 *) Fix crypto/asn1/a_sign.c so that 'parameters' is omitted (not
2684 encoded as NULL) with id-dsa-with-sha1.
2685 [Nils Larsch <nla@trustcenter.de>; problem pointed out by Bodo Moeller]
2687 *) Check various X509_...() return values in apps/req.c.
2688 [Nils Larsch <nla@trustcenter.de>]
2690 *) Fix BASE64 decode (EVP_DecodeUpdate) for data with CR/LF ended lines:
2691 an end-of-file condition would erronously be flagged, when the CRLF
2692 was just at the end of a processed block. The bug was discovered when
2693 processing data through a buffering memory BIO handing the data to a
2694 BASE64-decoding BIO. Bug fund and patch submitted by Pavel Tsekov
2695 <ptsekov@syntrex.com> and Nedelcho Stanev.
2698 *) Implement a countermeasure against a vulnerability recently found
2699 in CBC ciphersuites in SSL 3.0/TLS 1.0: Send an empty fragment
2700 before application data chunks to avoid the use of known IVs
2701 with data potentially chosen by the attacker.
2704 *) Fix length checks in ssl3_get_client_hello().
2707 *) TLS/SSL library bugfix: use s->s3->in_read_app_data differently
2708 to prevent ssl3_read_internal() from incorrectly assuming that
2709 ssl3_read_bytes() found application data while handshake
2710 processing was enabled when in fact s->s3->in_read_app_data was
2711 merely automatically cleared during the initial handshake.
2712 [Bodo Moeller; problem pointed out by Arne Ansper <arne@ats.cyber.ee>]
2714 *) Fix object definitions for Private and Enterprise: they were not
2715 recognized in their shortname (=lowercase) representation. Extend
2716 obj_dat.pl to issue an error when using undefined keywords instead
2717 of silently ignoring the problem (Svenning Sorensen
2718 <sss@sss.dnsalias.net>).
2721 *) Fix DH_generate_parameters() so that it works for 'non-standard'
2722 generators, i.e. generators other than 2 and 5. (Previously, the
2723 code did not properly initialise the 'add' and 'rem' values to
2724 BN_generate_prime().)
2726 In the new general case, we do not insist that 'generator' is
2727 actually a primitive root: This requirement is rather pointless;
2728 a generator of the order-q subgroup is just as good, if not
2732 *) Map new X509 verification errors to alerts. Discovered and submitted by
2733 Tom Wu <tom@arcot.com>.
2736 *) Fix ssl3_pending() (ssl/s3_lib.c) to prevent SSL_pending() from
2737 returning non-zero before the data has been completely received
2738 when using non-blocking I/O.
2739 [Bodo Moeller; problem pointed out by John Hughes]
2741 *) Some of the ciphers missed the strength entry (SSL_LOW etc).
2742 [Ben Laurie, Lutz Jaenicke]
2744 *) Fix bug in SSL_clear(): bad sessions were not removed (found by
2745 Yoram Zahavi <YoramZ@gilian.com>).
2748 *) Add information about CygWin 1.3 and on, and preserve proper
2749 configuration for the versions before that.
2750 [Corinna Vinschen <vinschen@redhat.com> and Richard Levitte]
2752 *) Make removal from session cache (SSL_CTX_remove_session()) more robust:
2753 check whether we deal with a copy of a session and do not delete from
2754 the cache in this case. Problem reported by "Izhar Shoshani Levi"
2755 <izhar@checkpoint.com>.
2758 *) Do not store session data into the internal session cache, if it
2759 is never intended to be looked up (SSL_SESS_CACHE_NO_INTERNAL_LOOKUP
2760 flag is set). Proposed by Aslam <aslam@funk.com>.
2763 *) Have ASN1_BIT_STRING_set_bit() really clear a bit when the requested
2767 *) [In 0.9.6d-engine release:]
2768 Fix a crashbug and a logic bug in hwcrhk_load_pubkey().
2769 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte]
2771 *) Add the configuration target linux-s390x.
2772 [Neale Ferguson <Neale.Ferguson@SoftwareAG-USA.com> via Richard Levitte]
2774 *) The earlier bugfix for the SSL3_ST_SW_HELLO_REQ_C case of
2775 ssl3_accept (ssl/s3_srvr.c) incorrectly used a local flag
2776 variable as an indication that a ClientHello message has been
2777 received. As the flag value will be lost between multiple
2778 invocations of ssl3_accept when using non-blocking I/O, the
2779 function may not be aware that a handshake has actually taken
2780 place, thus preventing a new session from being added to the
2783 To avoid this problem, we now set s->new_session to 2 instead of
2784 using a local variable.
2785 [Lutz Jaenicke, Bodo Moeller]
2787 *) Bugfix: Return -1 from ssl3_get_server_done (ssl3/s3_clnt.c)
2788 if the SSL_R_LENGTH_MISMATCH error is detected.
2789 [Geoff Thorpe, Bodo Moeller]
2791 *) New 'shared_ldflag' column in Configure platform table.
2794 *) Fix EVP_CIPHER_mode macro.
2795 ["Dan S. Camper" <dan@bti.net>]
2797 *) Fix ssl3_read_bytes (ssl/s3_pkt.c): To ignore messages of unknown
2798 type, we must throw them away by setting rr->length to 0.
2799 [D P Chang <dpc@qualys.com>]
2801 Changes between 0.9.6b and 0.9.6c [21 dec 2001]
2803 *) Fix BN_rand_range bug pointed out by Dominikus Scherkl
2804 <Dominikus.Scherkl@biodata.com>. (The previous implementation
2805 worked incorrectly for those cases where range = 10..._2 and
2806 3*range is two bits longer than range.)
2809 *) Only add signing time to PKCS7 structures if it is not already
2813 *) Fix crypto/objects/objects.h: "ld-ce" should be "id-ce",
2814 OBJ_ld_ce should be OBJ_id_ce.
2815 Also some ip-pda OIDs in crypto/objects/objects.txt were
2816 incorrect (cf. RFC 3039).
2817 [Matt Cooper, Frederic Giudicelli, Bodo Moeller]
2819 *) Release CRYPTO_LOCK_DYNLOCK when CRYPTO_destroy_dynlockid()
2820 returns early because it has nothing to do.
2821 [Andy Schneider <andy.schneider@bjss.co.uk>]
2823 *) [In 0.9.6c-engine release:]
2824 Fix mutex callback return values in crypto/engine/hw_ncipher.c.
2825 [Andy Schneider <andy.schneider@bjss.co.uk>]
2827 *) [In 0.9.6c-engine release:]
2828 Add support for Cryptographic Appliance's keyserver technology.
2829 (Use engine 'keyclient')
2830 [Cryptographic Appliances and Geoff Thorpe]
2832 *) Add a configuration entry for OS/390 Unix. The C compiler 'c89'
2833 is called via tools/c89.sh because arguments have to be
2834 rearranged (all '-L' options must appear before the first object
2836 [Richard Shapiro <rshapiro@abinitio.com>]
2838 *) [In 0.9.6c-engine release:]
2839 Add support for Broadcom crypto accelerator cards, backported
2841 [Broadcom, Nalin Dahyabhai <nalin@redhat.com>, Mark Cox]
2843 *) [In 0.9.6c-engine release:]
2844 Add support for SureWare crypto accelerator cards from
2845 Baltimore Technologies. (Use engine 'sureware')
2846 [Baltimore Technologies and Mark Cox]
2848 *) [In 0.9.6c-engine release:]
2849 Add support for crypto accelerator cards from Accelerated
2850 Encryption Processing, www.aep.ie. (Use engine 'aep')
2851 [AEP Inc. and Mark Cox]
2853 *) Add a configuration entry for gcc on UnixWare.
2854 [Gary Benson <gbenson@redhat.com>]
2856 *) Change ssl/s2_clnt.c and ssl/s2_srvr.c so that received handshake
2857 messages are stored in a single piece (fixed-length part and
2858 variable-length part combined) and fix various bugs found on the way.
2861 *) Disable caching in BIO_gethostbyname(), directly use gethostbyname()
2862 instead. BIO_gethostbyname() does not know what timeouts are
2863 appropriate, so entries would stay in cache even when they have
2865 [Bodo Moeller; problem pointed out by Rich Salz <rsalz@zolera.com>
2867 *) Change ssl23_get_client_hello (ssl/s23_srvr.c) behaviour when
2868 faced with a pathologically small ClientHello fragment that does
2869 not contain client_version: Instead of aborting with an error,
2870 simply choose the highest available protocol version (i.e.,
2871 TLS 1.0 unless it is disabled). In practice, ClientHello
2872 messages are never sent like this, but this change gives us
2873 strictly correct behaviour at least for TLS.
2876 *) Fix SSL handshake functions and SSL_clear() such that SSL_clear()
2877 never resets s->method to s->ctx->method when called from within
2878 one of the SSL handshake functions.
2879 [Bodo Moeller; problem pointed out by Niko Baric]
2881 *) In ssl3_get_client_hello (ssl/s3_srvr.c), generate a fatal alert
2882 (sent using the client's version number) if client_version is
2883 smaller than the protocol version in use. Also change
2884 ssl23_get_client_hello (ssl/s23_srvr.c) to select TLS 1.0 if
2885 the client demanded SSL 3.0 but only TLS 1.0 is enabled; then
2886 the client will at least see that alert.
2889 *) Fix ssl3_get_message (ssl/s3_both.c) to handle message fragmentation
2893 *) Avoid infinite loop in ssl3_get_message (ssl/s3_both.c) if a
2894 client receives HelloRequest while in a handshake.
2895 [Bodo Moeller; bug noticed by Andy Schneider <andy.schneider@bjss.co.uk>]
2897 *) Bugfix in ssl3_accept (ssl/s3_srvr.c): Case SSL3_ST_SW_HELLO_REQ_C
2898 should end in 'break', not 'goto end' which circuments various
2899 cleanups done in state SSL_ST_OK. But session related stuff
2900 must be disabled for SSL_ST_OK in the case that we just sent a
2903 Also avoid some overhead by not calling ssl_init_wbio_buffer()
2904 before just sending a HelloRequest.
2905 [Bodo Moeller, Eric Rescorla <ekr@rtfm.com>]
2907 *) Fix ssl/s3_enc.c, ssl/t1_enc.c and ssl/s3_pkt.c so that we don't
2908 reveal whether illegal block cipher padding was found or a MAC
2909 verification error occured. (Neither SSLerr() codes nor alerts
2910 are directly visible to potential attackers, but the information
2911 may leak via logfiles.)
2913 Similar changes are not required for the SSL 2.0 implementation
2914 because the number of padding bytes is sent in clear for SSL 2.0,
2915 and the extra bytes are just ignored. However ssl/s2_pkt.c
2916 failed to verify that the purported number of padding bytes is in
2920 *) Add OpenUNIX-8 support including shared libraries
2921 (Boyd Lynn Gerber <gerberb@zenez.com>).
2924 *) Improve RSA_padding_check_PKCS1_OAEP() check again to avoid
2925 'wristwatch attack' using huge encoding parameters (cf.
2926 James H. Manger's CRYPTO 2001 paper). Note that the
2927 RSA_PKCS1_OAEP_PADDING case of RSA_private_decrypt() does not use
2928 encoding parameters and hence was not vulnerable.
2931 *) BN_sqr() bug fix.
2932 [Ulf Möller, reported by Jim Ellis <jim.ellis@cavium.com>]
2934 *) Rabin-Miller test analyses assume uniformly distributed witnesses,
2935 so use BN_pseudo_rand_range() instead of using BN_pseudo_rand()
2936 followed by modular reduction.
2937 [Bodo Moeller; pointed out by Adam Young <AYoung1@NCSUS.JNJ.COM>]
2939 *) Add BN_pseudo_rand_range() with obvious functionality: BN_rand_range()
2940 equivalent based on BN_pseudo_rand() instead of BN_rand().
2943 *) s3_srvr.c: allow sending of large client certificate lists (> 16 kB).
2944 This function was broken, as the check for a new client hello message
2945 to handle SGC did not allow these large messages.
2946 (Tracked down by "Douglas E. Engert" <deengert@anl.gov>.)
2949 *) Add alert descriptions for TLSv1 to SSL_alert_desc_string[_long]().
2952 *) Fix buggy behaviour of BIO_get_num_renegotiates() and BIO_ctrl()
2953 for BIO_C_GET_WRITE_BUF_SIZE ("Stephen Hinton" <shinton@netopia.com>).
2956 *) Rework the configuration and shared library support for Tru64 Unix.
2957 The configuration part makes use of modern compiler features and
2958 still retains old compiler behavior for those that run older versions
2959 of the OS. The shared library support part includes a variant that
2960 uses the RPATH feature, and is available through the special
2961 configuration target "alpha-cc-rpath", which will never be selected
2963 [Tim Mooney <mooney@dogbert.cc.ndsu.NoDak.edu> via Richard Levitte]
2965 *) In ssl3_get_key_exchange (ssl/s3_clnt.c), call ssl3_get_message()
2966 with the same message size as in ssl3_get_certificate_request().
2967 Otherwise, if no ServerKeyExchange message occurs, CertificateRequest
2968 messages might inadvertently be reject as too long.
2969 [Petr Lampa <lampa@fee.vutbr.cz>]
2971 *) Enhanced support for IA-64 Unix platforms (well, Linux and HP-UX).
2974 *) Modified SSL library such that the verify_callback that has been set
2975 specificly for an SSL object with SSL_set_verify() is actually being
2976 used. Before the change, a verify_callback set with this function was
2977 ignored and the verify_callback() set in the SSL_CTX at the time of
2978 the call was used. New function X509_STORE_CTX_set_verify_cb() introduced
2979 to allow the necessary settings.
2982 *) Initialize static variable in crypto/dsa/dsa_lib.c and crypto/dh/dh_lib.c
2983 explicitly to NULL, as at least on Solaris 8 this seems not always to be
2984 done automatically (in contradiction to the requirements of the C
2985 standard). This made problems when used from OpenSSH.
2988 *) In OpenSSL 0.9.6a and 0.9.6b, crypto/dh/dh_key.c ignored
2989 dh->length and always used
2991 BN_rand_range(priv_key, dh->p).
2993 BN_rand_range() is not necessary for Diffie-Hellman, and this
2994 specific range makes Diffie-Hellman unnecessarily inefficient if
2995 dh->length (recommended exponent length) is much smaller than the
2996 length of dh->p. We could use BN_rand_range() if the order of
2997 the subgroup was stored in the DH structure, but we only have
3002 BN_rand(priv_key, l, ...)
3004 where 'l' is dh->length if this is defined, or BN_num_bits(dh->p)-1
3010 RSA_eay_public_encrypt
3011 RSA_eay_private_decrypt
3012 RSA_eay_private_encrypt (signing)
3013 RSA_eay_public_decrypt (signature verification)
3015 (default implementations for RSA_public_encrypt,
3016 RSA_private_decrypt, RSA_private_encrypt, RSA_public_decrypt),
3017 always reject numbers >= n.
3020 *) In crypto/rand/md_rand.c, use a new short-time lock CRYPTO_LOCK_RAND2
3021 to synchronize access to 'locking_thread'. This is necessary on
3022 systems where access to 'locking_thread' (an 'unsigned long'
3023 variable) is not atomic.
3026 *) In crypto/rand/md_rand.c, set 'locking_thread' to current thread's ID
3027 *before* setting the 'crypto_lock_rand' flag. The previous code had
3028 a race condition if 0 is a valid thread ID.
3029 [Travis Vitek <vitek@roguewave.com>]
3031 *) Add support for shared libraries under Irix.
3032 [Albert Chin-A-Young <china@thewrittenword.com>]
3034 *) Add configuration option to build on Linux on both big-endian and
3036 [Ralf Baechle <ralf@uni-koblenz.de>]
3038 *) Add the possibility to create shared libraries on HP-UX.
3041 Changes between 0.9.6a and 0.9.6b [9 Jul 2001]
3043 *) Change ssleay_rand_bytes (crypto/rand/md_rand.c)
3044 to avoid a SSLeay/OpenSSL PRNG weakness pointed out by
3045 Markku-Juhani O. Saarinen <markku-juhani.saarinen@nokia.com>:
3046 PRNG state recovery was possible based on the output of
3047 one PRNG request appropriately sized to gain knowledge on
3048 'md' followed by enough consecutive 1-byte PRNG requests
3049 to traverse all of 'state'.
3051 1. When updating 'md_local' (the current thread's copy of 'md')
3052 during PRNG output generation, hash all of the previous
3053 'md_local' value, not just the half used for PRNG output.
3055 2. Make the number of bytes from 'state' included into the hash
3056 independent from the number of PRNG bytes requested.
3058 The first measure alone would be sufficient to avoid
3059 Markku-Juhani's attack. (Actually it had never occurred
3060 to me that the half of 'md_local' used for chaining was the
3061 half from which PRNG output bytes were taken -- I had always
3062 assumed that the secret half would be used.) The second
3063 measure makes sure that additional data from 'state' is never
3064 mixed into 'md_local' in small portions; this heuristically
3065 further strengthens the PRNG.
3068 *) Fix crypto/bn/asm/mips3.s.
3071 *) When only the key is given to "enc", the IV is undefined. Print out
3072 an error message in this case.
3075 *) Handle special case when X509_NAME is empty in X509 printing routines.
3078 *) In dsa_do_verify (crypto/dsa/dsa_ossl.c), verify that r and s are
3079 positive and less than q.
3082 *) Don't change *pointer in CRYPTO_add_lock() is add_lock_callback is
3083 used: it isn't thread safe and the add_lock_callback should handle
3085 [Paul Rose <Paul.Rose@bridge.com>]
3087 *) Verify that incoming data obeys the block size in
3088 ssl3_enc (ssl/s3_enc.c) and tls1_enc (ssl/t1_enc.c).
3092 [Ulf Möller, Bodo Möller]
3094 *) The countermeasure against Bleichbacher's attack on PKCS #1 v1.5
3095 RSA encryption was accidentally removed in s3_srvr.c in OpenSSL 0.9.5
3096 when fixing the server behaviour for backwards-compatible 'client
3097 hello' messages. (Note that the attack is impractical against
3098 SSL 3.0 and TLS 1.0 anyway because length and version checking
3099 means that the probability of guessing a valid ciphertext is
3100 around 2^-40; see section 5 in Bleichenbacher's CRYPTO '98
3103 Before 0.9.5, the countermeasure (hide the error by generating a
3104 random 'decryption result') did not work properly because
3105 ERR_clear_error() was missing, meaning that SSL_get_error() would
3106 detect the supposedly ignored error.
3108 Both problems are now fixed.
3111 *) In crypto/bio/bf_buff.c, increase DEFAULT_BUFFER_SIZE to 4096
3112 (previously it was 1024).
3115 *) Fix for compatibility mode trust settings: ignore trust settings
3116 unless some valid trust or reject settings are present.
3119 *) Fix for blowfish EVP: its a variable length cipher.
3122 *) Fix various bugs related to DSA S/MIME verification. Handle missing
3123 parameters in DSA public key structures and return an error in the
3124 DSA routines if parameters are absent.
3127 *) In versions up to 0.9.6, RAND_file_name() resorted to file ".rnd"
3128 in the current directory if neither $RANDFILE nor $HOME was set.
3129 RAND_file_name() in 0.9.6a returned NULL in this case. This has
3130 caused some confusion to Windows users who haven't defined $HOME.
3131 Thus RAND_file_name() is changed again: e_os.h can define a
3132 DEFAULT_HOME, which will be used if $HOME is not set.
3133 For Windows, we use "C:"; on other platforms, we still require
3134 environment variables.
3136 *) Move 'if (!initialized) RAND_poll()' into regions protected by
3137 CRYPTO_LOCK_RAND. This is not strictly necessary, but avoids
3138 having multiple threads call RAND_poll() concurrently.
3141 *) In crypto/rand/md_rand.c, replace 'add_do_not_lock' flag by a
3142 combination of a flag and a thread ID variable.
3143 Otherwise while one thread is in ssleay_rand_bytes (which sets the
3144 flag), *other* threads can enter ssleay_add_bytes without obeying
3145 the CRYPTO_LOCK_RAND lock (and may even illegally release the lock
3146 that they do not hold after the first thread unsets add_do_not_lock).
3149 *) Change bctest again: '-x' expressions are not available in all
3153 Changes between 0.9.6 and 0.9.6a [5 Apr 2001]
3155 *) Fix a couple of memory leaks in PKCS7_dataDecode()
3156 [Steve Henson, reported by Heyun Zheng <hzheng@atdsprint.com>]
3158 *) Change Configure and Makefiles to provide EXE_EXT, which will contain
3159 the default extension for executables, if any. Also, make the perl
3160 scripts that use symlink() to test if it really exists and use "cp"
3161 if it doesn't. All this made OpenSSL compilable and installable in
3165 *) Fix for asn1_GetSequence() for indefinite length constructed data.
3166 If SEQUENCE is length is indefinite just set c->slen to the total
3167 amount of data available.
3168 [Steve Henson, reported by shige@FreeBSD.org]
3169 [This change does not apply to 0.9.7.]
3171 *) Change bctest to avoid here-documents inside command substitution
3172 (workaround for FreeBSD /bin/sh bug).
3173 For compatibility with Ultrix, avoid shell functions (introduced
3174 in the bctest version that searches along $PATH).
3177 *) Rename 'des_encrypt' to 'des_encrypt1'. This avoids the clashes
3178 with des_encrypt() defined on some operating systems, like Solaris
3182 *) Check the result of RSA-CRT (see D. Boneh, R. DeMillo, R. Lipton:
3183 On the Importance of Eliminating Errors in Cryptographic
3184 Computations, J. Cryptology 14 (2001) 2, 101-119,
3185 http://theory.stanford.edu/~dabo/papers/faults.ps.gz).
3188 *) MIPS assembler BIGNUM division bug fix.
3191 *) Disabled incorrect Alpha assembler code.
3194 *) Fix PKCS#7 decode routines so they correctly update the length
3195 after reading an EOC for the EXPLICIT tag.
3197 [This change does not apply to 0.9.7.]
3199 *) Fix bug in PKCS#12 key generation routines. This was triggered
3200 if a 3DES key was generated with a 0 initial byte. Include
3201 PKCS12_BROKEN_KEYGEN compilation option to retain the old
3202 (but broken) behaviour.
3205 *) Enhance bctest to search for a working bc along $PATH and print
3207 [Tim Rice <tim@multitalents.net> via Richard Levitte]
3209 *) Fix memory leaks in err.c: free err_data string if necessary;
3210 don't write to the wrong index in ERR_set_error_data.
3213 *) Implement ssl23_peek (analogous to ssl23_read), which previously
3217 *) Replace rdtsc with _emit statements for VC++ version 5.
3218 [Jeremy Cooper <jeremy@baymoo.org>]
3220 *) Make it possible to reuse SSLv2 sessions.
3223 *) In copy_email() check for >= 0 as a return value for
3224 X509_NAME_get_index_by_NID() since 0 is a valid index.
3225 [Steve Henson reported by Massimiliano Pala <madwolf@opensca.org>]
3227 *) Avoid coredump with unsupported or invalid public keys by checking if
3228 X509_get_pubkey() fails in PKCS7_verify(). Fix memory leak when
3229 PKCS7_verify() fails with non detached data.
3232 *) Don't use getenv in library functions when run as setuid/setgid.
3233 New function OPENSSL_issetugid().
3236 *) Avoid false positives in memory leak detection code (crypto/mem_dbg.c)
3237 due to incorrect handling of multi-threading:
3239 1. Fix timing glitch in the MemCheck_off() portion of CRYPTO_mem_ctrl().
3241 2. Fix logical glitch in is_MemCheck_on() aka CRYPTO_is_mem_check_on().
3243 3. Count how many times MemCheck_off() has been called so that
3244 nested use can be treated correctly. This also avoids
3245 inband-signalling in the previous code (which relied on the
3246 assumption that thread ID 0 is impossible).
3249 *) Add "-rand" option also to s_client and s_server.
3252 *) Fix CPU detection on Irix 6.x.
3253 [Kurt Hockenbury <khockenb@stevens-tech.edu> and
3254 "Bruce W. Forsberg" <bruce.forsberg@baesystems.com>]
3256 *) Fix X509_NAME bug which produced incorrect encoding if X509_NAME
3259 [This change does not apply to 0.9.7.]
3261 *) Use the cached encoding of an X509_NAME structure rather than
3262 copying it. This is apparently the reason for the libsafe "errors"
3263 but the code is actually correct.
3266 *) Add new function BN_rand_range(), and fix DSA_sign_setup() to prevent
3267 Bleichenbacher's DSA attack.
3268 Extend BN_[pseudo_]rand: As before, top=1 forces the highest two bits
3269 to be set and top=0 forces the highest bit to be set; top=-1 is new
3270 and leaves the highest bit random.
3271 [Ulf Moeller, Bodo Moeller]
3273 *) In the NCONF_...-based implementations for CONF_... queries
3274 (crypto/conf/conf_lib.c), if the input LHASH is NULL, avoid using
3275 a temporary CONF structure with the data component set to NULL
3276 (which gives segmentation faults in lh_retrieve).
3277 Instead, use NULL for the CONF pointer in CONF_get_string and
3278 CONF_get_number (which may use environment variables) and directly
3279 return NULL from CONF_get_section.
3282 *) Fix potential buffer overrun for EBCDIC.
3285 *) Tolerate nonRepudiation as being valid for S/MIME signing and certSign
3286 keyUsage if basicConstraints absent for a CA.
3289 *) Make SMIME_write_PKCS7() write mail header values with a format that
3290 is more generally accepted (no spaces before the semicolon), since
3291 some programs can't parse those values properly otherwise. Also make
3292 sure BIO's that break lines after each write do not create invalid
3296 *) Make the CRL encoding routines work with empty SEQUENCE OF. The
3297 macros previously used would not encode an empty SEQUENCE OF
3298 and break the signature.
3300 [This change does not apply to 0.9.7.]
3302 *) Zero the premaster secret after deriving the master secret in
3306 *) Add some EVP_add_digest_alias registrations (as found in
3307 OpenSSL_add_all_digests()) to SSL_library_init()
3308 aka OpenSSL_add_ssl_algorithms(). This provides improved
3309 compatibility with peers using X.509 certificates
3310 with unconventional AlgorithmIdentifier OIDs.
3313 *) Fix for Irix with NO_ASM.
3314 ["Bruce W. Forsberg" <bruce.forsberg@baesystems.com>]
3316 *) ./config script fixes.
3317 [Ulf Moeller, Richard Levitte]
3319 *) Fix 'openssl passwd -1'.
3322 *) Change PKCS12_key_gen_asc() so it can cope with non null
3323 terminated strings whose length is passed in the passlen
3324 parameter, for example from PEM callbacks. This was done
3325 by adding an extra length parameter to asc2uni().
3326 [Steve Henson, reported by <oddissey@samsung.co.kr>]
3328 *) Fix C code generated by 'openssl dsaparam -C': If a BN_bin2bn
3329 call failed, free the DSA structure.
3332 *) Fix to uni2asc() to cope with zero length Unicode strings.
3333 These are present in some PKCS#12 files.
3336 *) Increase s2->wbuf allocation by one byte in ssl2_new (ssl/s2_lib.c).
3337 Otherwise do_ssl_write (ssl/s2_pkt.c) will write beyond buffer limits
3338 when writing a 32767 byte record.
3339 [Bodo Moeller; problem reported by Eric Day <eday@concentric.net>]
3341 *) In RSA_eay_public_{en,ed}crypt and RSA_eay_mod_exp (rsa_eay.c),
3342 obtain lock CRYPTO_LOCK_RSA before setting rsa->_method_mod_{n,p,q}.
3344 (RSA objects have a reference count access to which is protected
3345 by CRYPTO_LOCK_RSA [see rsa_lib.c, s3_srvr.c, ssl_cert.c, ssl_rsa.c],
3346 so they are meant to be shared between threads.)
3347 [Bodo Moeller, Geoff Thorpe; original patch submitted by
3348 "Reddie, Steven" <Steven.Reddie@ca.com>]
3350 *) Fix a deadlock in CRYPTO_mem_leaks().
3353 *) Use better test patterns in bntest.
3356 *) rand_win.c fix for Borland C.
3359 *) BN_rshift bugfix for n == 0.
3362 *) Add a 'bctest' script that checks for some known 'bc' bugs
3363 so that 'make test' does not abort just because 'bc' is broken.
3366 *) Store verify_result within SSL_SESSION also for client side to
3367 avoid potential security hole. (Re-used sessions on the client side
3368 always resulted in verify_result==X509_V_OK, not using the original
3369 result of the server certificate verification.)
3372 *) Fix ssl3_pending: If the record in s->s3->rrec is not of type
3373 SSL3_RT_APPLICATION_DATA, return 0.
3374 Similarly, change ssl2_pending to return 0 if SSL_in_init(s) is true.
3378 Both ssl2_peek and ssl3_peek, which were totally broken in earlier
3379 releases, have been re-implemented by renaming the previous
3380 implementations of ssl2_read and ssl3_read to ssl2_read_internal
3381 and ssl3_read_internal, respectively, and adding 'peek' parameters
3382 to them. The new ssl[23]_{read,peek} functions are calls to
3383 ssl[23]_read_internal with the 'peek' flag set appropriately.
3384 A 'peek' parameter has also been added to ssl3_read_bytes, which
3385 does the actual work for ssl3_read_internal.
3388 *) Initialise "ex_data" member of RSA/DSA/DH structures prior to calling
3389 the method-specific "init()" handler. Also clean up ex_data after
3390 calling the method-specific "finish()" handler. Previously, this was
3391 happening the other way round.
3394 *) Increase BN_CTX_NUM (the number of BIGNUMs in a BN_CTX) to 16.
3395 The previous value, 12, was not always sufficient for BN_mod_exp().
3398 *) Make sure that shared libraries get the internal name engine with
3399 the full version number and not just 0. This should mark the
3400 shared libraries as not backward compatible. Of course, this should
3401 be changed again when we can guarantee backward binary compatibility.
3404 *) Fix typo in get_cert_by_subject() in by_dir.c
3405 [Jean-Marc Desperrier <jean-marc.desperrier@certplus.com>]
3407 *) Rework the system to generate shared libraries:
3409 - Make note of the expected extension for the shared libraries and
3410 if there is a need for symbolic links from for example libcrypto.so.0
3411 to libcrypto.so.0.9.7. There is extended info in Configure for
3414 - Make as few rebuilds of the shared libraries as possible.
3416 - Still avoid linking the OpenSSL programs with the shared libraries.
3418 - When installing, install the shared libraries separately from the
3422 *) Fix SSL_CTX_set_read_ahead macro to actually use its argument.
3424 Copy SSL_CTX's read_ahead flag to SSL object directly in SSL_new
3425 and not in SSL_clear because the latter is also used by the
3426 accept/connect functions; previously, the settings made by
3427 SSL_set_read_ahead would be lost during the handshake.
3428 [Bodo Moeller; problems reported by Anders Gertz <gertz@epact.se>]
3430 *) Correct util/mkdef.pl to be selective about disabled algorithms.
3431 Previously, it would create entries for disableed algorithms no
3435 *) Added several new manual pages for SSL_* function.
3438 Changes between 0.9.5a and 0.9.6 [24 Sep 2000]
3440 *) In ssl23_get_client_hello, generate an error message when faced
3441 with an initial SSL 3.0/TLS record that is too small to contain the
3442 first two bytes of the ClientHello message, i.e. client_version.
3443 (Note that this is a pathologic case that probably has never happened
3444 in real life.) The previous approach was to use the version number
3445 from the record header as a substitute; but our protocol choice
3446 should not depend on that one because it is not authenticated
3447 by the Finished messages.
3450 *) More robust randomness gathering functions for Windows.
3451 [Jeffrey Altman <jaltman@columbia.edu>]
3453 *) For compatibility reasons if the flag X509_V_FLAG_ISSUER_CHECK is
3454 not set then we don't setup the error code for issuer check errors
3455 to avoid possibly overwriting other errors which the callback does
3456 handle. If an application does set the flag then we assume it knows
3457 what it is doing and can handle the new informational codes
3461 *) Fix for a nasty bug in ASN1_TYPE handling. ASN1_TYPE is used for
3462 a general "ANY" type, as such it should be able to decode anything
3463 including tagged types. However it didn't check the class so it would
3464 wrongly interpret tagged types in the same way as their universal
3465 counterpart and unknown types were just rejected. Changed so that the
3466 tagged and unknown types are handled in the same way as a SEQUENCE:
3467 that is the encoding is stored intact. There is also a new type
3468 "V_ASN1_OTHER" which is used when the class is not universal, in this
3469 case we have no idea what the actual type is so we just lump them all
3473 *) On VMS, stdout may very well lead to a file that is written to
3474 in a record-oriented fashion. That means that every write() will
3475 write a separate record, which will be read separately by the
3476 programs trying to read from it. This can be very confusing.
3478 The solution is to put a BIO filter in the way that will buffer
3479 text until a linefeed is reached, and then write everything a
3480 line at a time, so every record written will be an actual line,
3481 not chunks of lines and not (usually doesn't happen, but I've
3482 seen it once) several lines in one record. BIO_f_linebuffer() is
3485 Currently, it's a VMS-only method, because that's where it has
3486 been tested well enough.
3489 *) Remove 'optimized' squaring variant in BN_mod_mul_montgomery,
3490 it can return incorrect results.
3491 (Note: The buggy variant was not enabled in OpenSSL 0.9.5a,
3492 but it was in 0.9.6-beta[12].)
3495 *) Disable the check for content being present when verifying detached
3496 signatures in pk7_smime.c. Some versions of Netscape (wrongly)
3497 include zero length content when signing messages.
3500 *) New BIO_shutdown_wr macro, which invokes the BIO_C_SHUTDOWN_WR
3501 BIO_ctrl (for BIO pairs).
3504 *) Add DSO method for VMS.
3507 *) Bug fix: Montgomery multiplication could produce results with the
3511 *) Add RPM specification openssl.spec and modify it to build three
3512 packages. The default package contains applications, application
3513 documentation and run-time libraries. The devel package contains
3514 include files, static libraries and function documentation. The
3515 doc package contains the contents of the doc directory. The original
3516 openssl.spec was provided by Damien Miller <djm@mindrot.org>.
3519 *) Add a large number of documentation files for many SSL routines.
3520 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>]
3522 *) Add a configuration entry for Sony News 4.
3523 [NAKAJI Hiroyuki <nakaji@tutrp.tut.ac.jp>]
3525 *) Don't set the two most significant bits to one when generating a
3526 random number < q in the DSA library.
3529 *) New SSL API mode 'SSL_MODE_AUTO_RETRY'. This disables the default
3530 behaviour that SSL_read may result in SSL_ERROR_WANT_READ (even if
3531 the underlying transport is blocking) if a handshake took place.
3532 (The default behaviour is needed by applications such as s_client
3533 and s_server that use select() to determine when to use SSL_read;
3534 but for applications that know in advance when to expect data, it
3535 just makes things more complicated.)
3538 *) Add RAND_egd_bytes(), which gives control over the number of bytes read
3542 *) Add a few more EBCDIC conditionals that make `req' and `x509'
3543 work better on such systems.
3544 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>]
3546 *) Add two demo programs for PKCS12_parse() and PKCS12_create().
3547 Update PKCS12_parse() so it copies the friendlyName and the
3548 keyid to the certificates aux info.
3551 *) Fix bug in PKCS7_verify() which caused an infinite loop
3552 if there was more than one signature.
3553 [Sven Uszpelkat <su@celocom.de>]
3555 *) Major change in util/mkdef.pl to include extra information
3556 about each symbol, as well as presentig variables as well
3557 as functions. This change means that there's n more need
3558 to rebuild the .num files when some algorithms are excluded.
3561 *) Allow the verify time to be set by an application,
3562 rather than always using the current time.
3565 *) Phase 2 verify code reorganisation. The certificate
3566 verify code now looks up an issuer certificate by a
3567 number of criteria: subject name, authority key id
3568 and key usage. It also verifies self signed certificates
3569 by the same criteria. The main comparison function is
3570 X509_check_issued() which performs these checks.
3572 Lot of changes were necessary in order to support this
3573 without completely rewriting the lookup code.
3575 Authority and subject key identifier are now cached.
3577 The LHASH 'certs' is X509_STORE has now been replaced
3578 by a STACK_OF(X509_OBJECT). This is mainly because an
3579 LHASH can't store or retrieve multiple objects with
3580 the same hash value.
3582 As a result various functions (which were all internal
3583 use only) have changed to handle the new X509_STORE
3584 structure. This will break anything that messed round
3585 with X509_STORE internally.
3587 The functions X509_STORE_add_cert() now checks for an
3588 exact match, rather than just subject name.
3590 The X509_STORE API doesn't directly support the retrieval
3591 of multiple certificates matching a given criteria, however
3592 this can be worked round by performing a lookup first
3593 (which will fill the cache with candidate certificates)
3594 and then examining the cache for matches. This is probably
3595 the best we can do without throwing out X509_LOOKUP
3596 entirely (maybe later...).
3598 The X509_VERIFY_CTX structure has been enhanced considerably.
3600 All certificate lookup operations now go via a get_issuer()
3601 callback. Although this currently uses an X509_STORE it
3602 can be replaced by custom lookups. This is a simple way
projects / openssl.git / blob