5 Changes between 0.9.7a and 0.9.8 [xx XXX xxxx]
7 *) Generate muti valued AVAs using '+' notation in config files for
11 *) Support for nameConstraints certificate extension.
14 *) Support for policyConstraints certificate extension.
17 *) Support for policyMappings certificate extension.
20 *) Fixed a typo bug that would cause ENGINE_set_default() to set an
21 ENGINE as defaults for all supported algorithms irrespective of
22 the 'flags' parameter. 'flags' is now honoured, so applications
23 should make sure they are passing it correctly.
26 *) Make sure the default DSA_METHOD implementation only uses its
27 dsa_mod_exp() and/or bn_mod_exp() handlers if they are non-NULL,
28 and change its own handlers to be NULL so as to remove unnecessary
29 indirection. This lets alternative implementations fallback to the
30 default implementation more easily.
33 *) Support for directoryName in GeneralName related extensions
37 *) Make it possible to link applications using Makefile.shared.
38 Make that possible even when linking against static libraries!
41 *) Various fixes to base64 BIO and non blocking I/O. On write
42 flushes were not handled properly if the BIO retried. On read
43 data was not being buffered properly and had various logic bugs.
46 *) Support for single pass processing for S/MIME signing. This now
47 means that S/MIME signing can be done from a pipe, in addition
48 cleartext signing (multipart/signed type) is effectively streaming
49 and the signed data does not need to be all held in memory.
51 This is done with a new flag PKCS7_STREAM. When this flag is set
52 PKCS7_sign() only initializes the PKCS7 structure and the actual signing
53 is done after the data is output (and digests calculated) in
57 *) Add full support for -rpath/-R, both in shared libraries and
58 applications, at least on the platforms where it's known how
62 *) In crypto/ec/ec_mult.c, implement fast point multiplication with
63 precomputation, based on wNAF splitting: EC_GROUP_precompute_mult()
64 will now compute a table of multiples of the generator that
65 makes subsequent invocations of EC_POINTs_mul() or EC_POINT_mul()
66 faster (notably in the case of a single point multiplication,
68 [Nils Larsch, Bodo Moeller]
70 *) IPv6 support for certificate extensions. The various extensions
71 which use the IP:a.b.c.d can now take IPv6 addresses using the
72 formats of RFC1884 2.2 . IPv6 addresses are now also displayed
76 *) Added an ENGINE that implements RSA by performing private key
77 exponentiations with the GMP library. The conversions to and from
78 GMP's mpz_t format aren't optimised nor are any montgomery forms
79 cached, and on x86 it appears OpenSSL's own performance has caught up.
80 However there are likely to be other architectures where GMP could
81 provide a boost. This ENGINE is not built in by default, but it can be
82 specified at Configure time and should be accompanied by the necessary
84 ./config -DOPENSSL_USE_GMP -lgmp
87 *) "openssl engine" will not display ENGINE/DSO load failure errors when
88 testing availability of engines with "-t" - the old behaviour is
89 produced by increasing the feature's verbosity with "-tt".
92 *) ECDSA routines: under certain error conditions uninitialized BN objects
93 could be freed. Solution: make sure initialization is performed early
94 enough. (Reported and fix supplied by Nils Larsch <nla@trustcenter.de>
98 *) Key-generation can now be implemented in RSA_METHOD, DSA_METHOD
99 and DH_METHOD (eg. by ENGINE implementations) to override the normal
100 software implementations. For DSA and DH, parameter generation can
101 also be overriden by providing the appropriate method callbacks.
104 *) Change the "progress" mechanism used in key-generation and
105 primality testing to functions that take a new BN_GENCB pointer in
106 place of callback/argument pairs. The new API functions have "_ex"
107 postfixes and the older functions are reimplemented as wrappers for
108 the new ones. The OPENSSL_NO_DEPRECATED symbol can be used to hide
109 declarations of the old functions to help (graceful) attempts to
110 migrate to the new functions. Also, the new key-generation API
111 functions operate on a caller-supplied key-structure and return
112 success/failure rather than returning a key or NULL - this is to
113 help make "keygen" another member function of RSA_METHOD etc.
115 Example for using the new callback interface:
117 int (*my_callback)(int a, int b, BN_GENCB *cb) = ...;
121 BN_GENCB_set(&my_cb, my_callback, my_arg);
123 return BN_is_prime_ex(some_bignum, BN_prime_checks, NULL, &cb);
124 /* For the meaning of a, b in calls to my_callback(), see the
125 * documentation of the function that calls the callback.
126 * cb will point to my_cb; my_arg can be retrieved as cb->arg.
127 * my_callback should return 1 if it wants BN_is_prime_ex()
128 * to continue, or 0 to stop.
133 *) Change the ZLIB compression method to be stateful, and make it
134 available to TLS with the number defined in
135 draft-ietf-tls-compression-04.txt.
138 *) Add the ASN.1 structures and functions for CertificatePair, which
139 is defined as follows (according to X.509_4thEditionDraftV6.pdf):
141 CertificatePair ::= SEQUENCE {
142 forward [0] Certificate OPTIONAL,
143 reverse [1] Certificate OPTIONAL,
144 -- at least one of the pair shall be present -- }
146 Also implement the PEM functions to read and write certificate
147 pairs, and defined the PEM tag as "CERTIFICATE PAIR".
149 This needed to be defined, mostly for the sake of the LDAP
150 attribute crossCertificatePair, but may prove useful elsewhere as
154 *) Make it possible to inhibit symlinking of shared libraries in
155 Makefile.shared, for Cygwin's sake.
158 *) Extend the BIGNUM API by creating new macros that behave like
161 void BN_set_sign(BIGNUM *a, int neg);
162 int BN_get_sign(const BIGNUM *a);
164 and avoid the need to access 'a->neg' directly in applications.
165 [Nils Larsch <nla@trustcenter.de>]
167 *) Implement fast modular reduction for pseudo-Mersenne primes
168 used in NIST curves (crypto/bn/bn_nist.c, crypto/ec/ecp_nist.c).
169 EC_GROUP_new_curve_GFp() will now automatically use this
171 [Nils Larsch <nla@trustcenter.de>]
173 *) Add new lock type (CRYPTO_LOCK_BN).
176 *) Change the ENGINE framework to automatically load engines
177 dynamically from specific directories unless they could be
178 found to already be built in or loaded. Move all the
179 current engines except for the cryptodev one to a new
181 The engines in engines/ are built as shared libraries if
182 the "shared" options was given to ./Configure or ./config.
183 Otherwise, they are inserted in libcrypto.a.
184 /usr/local/ssl/engines is the default directory for dynamic
185 engines, but that can be overriden at configure time through
186 the usual use of --prefix and/or --openssldir, and at run
187 time with the environment variable OPENSSL_ENGINES.
188 [Geoff Thorpe and Richard Levitte]
190 *) Add Makefile.shared, a helper makefile to build shared
191 libraries. Addapt Makefile.org.
194 *) Add version info to Win32 DLLs.
195 [Peter 'Luna' Runestig" <peter@runestig.com>]
197 *) Add new 'medium level' PKCS#12 API. Certificates and keys
198 can be added using this API to created arbitrary PKCS#12
199 files while avoiding the low level API.
201 New options to PKCS12_create(), key or cert can be NULL and
202 will then be omitted from the output file. The encryption
203 algorithm NIDs can be set to -1 for no encryption, the mac
204 iteration count can be set to 0 to omit the mac.
206 Enhance pkcs12 utility by making the -nokeys and -nocerts
207 options work when creating a PKCS#12 file. New option -nomac
208 to omit the mac, NONE can be set for an encryption algorithm.
209 New code is modified to use the enhanced PKCS12_create()
210 instead of the low level API.
213 *) Extend ASN1 encoder to support indefinite length constructed
214 encoding. This can output sequences tags and octet strings in
215 this form. Modify pk7_asn1.c to support indefinite length
216 encoding. This is experimental and needs additional code to
217 be useful, such as an ASN1 bio and some enhanced streaming
220 Extend template encode functionality so that tagging is passed
221 down to the template encoder.
224 *) Let 'openssl req' fail if an argument to '-newkey' is not
225 recognized instead of using RSA as a default.
228 *) Add support for ECC-based ciphersuites from draft-ietf-tls-ecc-01.txt.
229 As these are not official, they are not included in "ALL";
230 the "ECCdraft" ciphersuite group alias can be used to select them.
231 [Vipul Gupta and Sumit Gupta (Sun Microsystems Laboratories)]
233 *) Add ECDH engine support.
234 [Nils Gura and Douglas Stebila (Sun Microsystems Laboratories)]
236 *) Add ECDH in new directory crypto/ecdh/.
237 [Douglas Stebila (Sun Microsystems Laboratories)]
239 *) Let BN_rand_range() abort with an error after 100 iterations
240 without success (which indicates a broken PRNG).
243 *) Change BN_mod_sqrt() so that it verifies that the input value
244 is really the square of the return value. (Previously,
245 BN_mod_sqrt would show GIGO behaviour.)
248 *) Add named elliptic curves over binary fields from X9.62, SECG,
249 and WAP/WTLS; add OIDs that were still missing.
251 [Sheueling Chang Shantz and Douglas Stebila
252 (Sun Microsystems Laboratories)]
254 *) Extend the EC library for elliptic curves over binary fields
255 (new files ec2_smpl.c, ec2_smpt.c, ec2_mult.c in crypto/ec/).
258 EC_GF2m_simple_method
262 EC_GROUP_new_curve_GF2m
263 EC_GROUP_set_curve_GF2m
264 EC_GROUP_get_curve_GF2m
265 EC_POINT_set_affine_coordinates_GF2m
266 EC_POINT_get_affine_coordinates_GF2m
267 EC_POINT_set_compressed_coordinates_GF2m
269 Point compression for binary fields is disabled by default for
270 patent reasons (compile with OPENSSL_EC_BIN_PT_COMP defined to
273 As binary polynomials are represented as BIGNUMs, various members
274 of the EC_GROUP and EC_POINT data structures can be shared
275 between the implementations for prime fields and binary fields;
276 the above ..._GF2m functions (except for EX_GROUP_new_curve_GF2m)
277 are essentially identical to their ..._GFp counterparts.
278 (For simplicity, the '..._GFp' prefix has been dropped from
279 various internal method names.)
281 An internal 'field_div' method (similar to 'field_mul' and
282 'field_sqr') has been added; this is used only for binary fields.
284 [Sheueling Chang Shantz and Douglas Stebila
285 (Sun Microsystems Laboratories)]
287 *) Optionally dispatch EC_POINT_mul(), EC_POINT_precompute_mult()
288 through methods ('mul', 'precompute_mult').
290 The generic implementations (now internally called 'ec_wNAF_mul'
291 and 'ec_wNAF_precomputed_mult') remain the default if these
292 methods are undefined.
294 [Sheueling Chang Shantz and Douglas Stebila
295 (Sun Microsystems Laboratories)]
297 *) New function EC_GROUP_get_degree, which is defined through
298 EC_METHOD. For curves over prime fields, this returns the bit
299 length of the modulus.
301 [Sheueling Chang Shantz and Douglas Stebila
302 (Sun Microsystems Laboratories)]
304 *) New functions EC_GROUP_dup, EC_POINT_dup.
305 (These simply call ..._new and ..._copy).
307 [Sheueling Chang Shantz and Douglas Stebila
308 (Sun Microsystems Laboratories)]
310 *) Add binary polynomial arithmetic software in crypto/bn/bn_gf2m.c.
311 Polynomials are represented as BIGNUMs (where the sign bit is not
312 used) in the following functions [macros]:
315 BN_GF2m_sub [= BN_GF2m_add]
316 BN_GF2m_mod [wrapper for BN_GF2m_mod_arr]
317 BN_GF2m_mod_mul [wrapper for BN_GF2m_mod_mul_arr]
318 BN_GF2m_mod_sqr [wrapper for BN_GF2m_mod_sqr_arr]
320 BN_GF2m_mod_exp [wrapper for BN_GF2m_mod_exp_arr]
321 BN_GF2m_mod_sqrt [wrapper for BN_GF2m_mod_sqrt_arr]
322 BN_GF2m_mod_solve_quad [wrapper for BN_GF2m_mod_solve_quad_arr]
323 BN_GF2m_cmp [= BN_ucmp]
325 (Note that only the 'mod' functions are actually for fields GF(2^m).
326 BN_GF2m_add() is misnomer, but this is for the sake of consistency.)
328 For some functions, an the irreducible polynomial defining a
329 field can be given as an 'unsigned int[]' with strictly
330 decreasing elements giving the indices of those bits that are set;
331 i.e., p[] represents the polynomial
332 f(t) = t^p[0] + t^p[1] + ... + t^p[k]
334 p[0] > p[1] > ... > p[k] = 0.
335 This applies to the following functions:
340 BN_GF2m_mod_inv_arr [wrapper for BN_GF2m_mod_inv]
341 BN_GF2m_mod_div_arr [wrapper for BN_GF2m_mod_div]
344 BN_GF2m_mod_solve_quad_arr
348 Conversion can be performed by the following functions:
353 bntest.c has additional tests for binary polynomial arithmetic.
355 Two implementations for BN_GF2m_mod_div() are available.
356 The default algorithm simply uses BN_GF2m_mod_inv() and
357 BN_GF2m_mod_mul(). The alternative algorithm is compiled in only
358 if OPENSSL_SUN_GF2M_DIV is defined (patent pending; read the
359 copyright notice in crypto/bn/bn_gf2m.c before enabling it).
361 [Sheueling Chang Shantz and Douglas Stebila
362 (Sun Microsystems Laboratories)]
364 *) Add new error code 'ERR_R_DISABLED' that can be used when some
365 functionality is disabled at compile-time.
366 [Douglas Stebila <douglas.stebila@sun.com>]
368 *) Change default behaviour of 'openssl asn1parse' so that more
369 information is visible when viewing, e.g., a certificate:
371 Modify asn1_parse2 (crypto/asn1/asn1_par.c) so that in non-'dump'
372 mode the content of non-printable OCTET STRINGs is output in a
373 style similar to INTEGERs, but with '[HEX DUMP]' prepended to
374 avoid the appearance of a printable string.
375 [Nils Larsch <nla@trustcenter.de>]
377 *) Add 'asn1_flag' and 'asn1_form' member to EC_GROUP with access
379 EC_GROUP_set_asn1_flag()
380 EC_GROUP_get_asn1_flag()
381 EC_GROUP_set_point_conversion_form()
382 EC_GROUP_get_point_conversion_form()
383 These control ASN1 encoding details:
384 - Curves (i.e., groups) are encoded explicitly unless asn1_flag
385 has been set to OPENSSL_EC_NAMED_CURVE.
386 - Points are encoded in uncompressed form by default; options for
387 asn1_for are as for point2oct, namely
388 POINT_CONVERSION_COMPRESSED
389 POINT_CONVERSION_UNCOMPRESSED
390 POINT_CONVERSION_HYBRID
392 Also add 'seed' and 'seed_len' members to EC_GROUP with access
396 EC_GROUP_get_seed_len()
397 This is used only for ASN1 purposes (so far).
398 [Nils Larsch <nla@trustcenter.de>]
400 *) Add 'field_type' member to EC_METHOD, which holds the NID
401 of the appropriate field type OID. The new function
402 EC_METHOD_get_field_type() returns this value.
403 [Nils Larsch <nla@trustcenter.de>]
410 providing useful interfaces to EC_POINT_point2oct() and
411 EC_POINT_oct2point().
412 [Nils Larsch <nla@trustcenter.de>]
414 *) Change internals of the EC library so that the functions
415 EC_GROUP_set_generator()
416 EC_GROUP_get_generator()
418 EC_GROUP_get_cofactor()
419 are implemented directly in crypto/ec/ec_lib.c and not dispatched
420 to methods, which would lead to unnecessary code duplication when
421 adding different types of curves.
422 [Nils Larsch <nla@trustcenter.de> with input by Bodo Moeller]
424 *) Implement compute_wNAF (crypto/ec/ec_mult.c) without BIGNUM
425 arithmetic, and such that modified wNAFs are generated
426 (which avoid length expansion in many cases).
429 *) Add a function EC_GROUP_check_discriminant() (defined via
430 EC_METHOD) that verifies that the curve discriminant is non-zero.
432 Add a function EC_GROUP_check() that makes some sanity tests
433 on a EC_GROUP, its generator and order. This includes
434 EC_GROUP_check_discriminant().
435 [Nils Larsch <nla@trustcenter.de>]
437 *) Add ECDSA in new directory crypto/ecdsa/.
439 Add applications 'openssl ecparam' and 'openssl ecdsa'
440 (these are based on 'openssl dsaparam' and 'openssl dsa').
442 ECDSA support is also included in various other files across the
443 library. Most notably,
444 - 'openssl req' now has a '-newkey ecdsa:file' option;
445 - EVP_PKCS82PKEY (crypto/evp/evp_pkey.c) now can handle ECDSA;
446 - X509_PUBKEY_get (crypto/asn1/x_pubkey.c) and
447 d2i_PublicKey (crypto/asn1/d2i_pu.c) have been modified to make
448 them suitable for ECDSA where domain parameters must be
449 extracted before the specific public key;
450 - ECDSA engine support has been added.
451 [Nils Larsch <nla@trustcenter.de>]
453 *) Include some named elliptic curves, and add OIDs from X9.62,
454 SECG, and WAP/WTLS. Each curve can be obtained from the new
456 EC_GROUP_new_by_nid(),
457 and the list of available named curves can be obtained with
458 EC_get_builtin_curves().
459 Also add a 'curve_name' member to EC_GROUP objects, which can be
463 [Nils Larsch <nla@trustcenter.de, Bodo Moeller]
465 *) Remove a few calls to bn_wexpand() in BN_sqr() (the one in there
466 was actually never needed) and in BN_mul(). The removal in BN_mul()
467 required a small change in bn_mul_part_recursive() and the addition
468 of the functions bn_cmp_part_words(), bn_sub_part_words() and
469 bn_add_part_words(), which do the same thing as bn_cmp_words(),
470 bn_sub_words() and bn_add_words() except they take arrays with
474 Changes between 0.9.7a and 0.9.7b [xx XXX 2003]
476 *) Countermeasure against the Klima-Pokorny-Rosa extension of
477 Bleichbacher's attack on PKCS #1 v1.5 padding: treat
478 a protocol version number mismatch like a decryption error
479 in ssl3_get_client_key_exchange (ssl/s3_srvr.c).
482 *) Turn on RSA blinding by default in the default implementation
483 to avoid a timing attack. Applications that don't want it can call
484 RSA_blinding_off() or use the new flag RSA_FLAG_NO_BLINDING.
485 They would be ill-advised to do so in most cases.
486 [Ben Laurie, Steve Henson, Geoff Thorpe]
488 *) Change RSA blinding code so that it works when the PRNG is not
489 seeded (in this case, the secret RSA exponent is abused as
490 an unpredictable seed -- if it is not unpredictable, there
491 is no point in blinding anyway).
494 yet to be integrated into this CVS branch:
495 - Geoff's ENGINE_set_default() fix
497 *) Target "mingw" now allows native Windows code to be generated in
498 the Cygwin environment as well as with the MinGW compiler.
501 Changes between 0.9.7 and 0.9.7a [19 Feb 2003]
503 *) In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked
504 via timing by performing a MAC computation even if incorrrect
505 block cipher padding has been found. This is a countermeasure
506 against active attacks where the attacker has to distinguish
507 between bad padding and a MAC verification error. (CAN-2003-0078)
509 [Bodo Moeller; problem pointed out by Brice Canvel (EPFL),
510 Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and
511 Martin Vuagnoux (EPFL, Ilion)]
513 *) Make the no-err option work as intended. The intention with no-err
514 is not to have the whole error stack handling routines removed from
515 libcrypto, it's only intended to remove all the function name and
516 reason texts, thereby removing some of the footprint that may not
517 be interesting if those errors aren't displayed anyway.
519 NOTE: it's still possible for any application or module to have it's
520 own set of error texts inserted. The routines are there, just not
521 used by default when no-err is given.
524 *) Add support for FreeBSD on IA64.
525 [dirk.meyer@dinoex.sub.org via Richard Levitte, resolves #454]
527 *) Adjust DES_cbc_cksum() so it returns the same value as the MIT
528 Kerberos function mit_des_cbc_cksum(). Before this change,
529 the value returned by DES_cbc_cksum() was like the one from
530 mit_des_cbc_cksum(), except the bytes were swapped.
531 [Kevin Greaney <Kevin.Greaney@hp.com> and Richard Levitte]
533 *) Allow an application to disable the automatic SSL chain building.
534 Before this a rather primitive chain build was always performed in
535 ssl3_output_cert_chain(): an application had no way to send the
536 correct chain if the automatic operation produced an incorrect result.
538 Now the chain builder is disabled if either:
540 1. Extra certificates are added via SSL_CTX_add_extra_chain_cert().
542 2. The mode flag SSL_MODE_NO_AUTO_CHAIN is set.
544 The reasoning behind this is that an application would not want the
545 auto chain building to take place if extra chain certificates are
546 present and it might also want a means of sending no additional
547 certificates (for example the chain has two certificates and the
551 *) Add the possibility to build without the ENGINE framework.
552 [Steven Reddie <smr@essemer.com.au> via Richard Levitte]
554 *) Under Win32 gmtime() can return NULL: check return value in
555 OPENSSL_gmtime(). Add error code for case where gmtime() fails.
558 *) DSA routines: under certain error conditions uninitialized BN objects
559 could be freed. Solution: make sure initialization is performed early
560 enough. (Reported and fix supplied by Ivan D Nestlerode <nestler@MIT.EDU>,
561 Nils Larsch <nla@trustcenter.de> via PR#459)
564 *) Another fix for SSLv2 session ID handling: the session ID was incorrectly
565 checked on reconnect on the client side, therefore session resumption
566 could still fail with a "ssl session id is different" error. This
567 behaviour is masked when SSL_OP_ALL is used due to
568 SSL_OP_MICROSOFT_SESS_ID_BUG being set.
569 Behaviour observed by Crispin Flowerday <crispin@flowerday.cx> as
573 *) IA-32 assembler support enhancements: unified ELF targets, support
574 for SCO/Caldera platforms, fix for Cygwin shared build.
577 *) Add support for FreeBSD on sparc64. As a consequence, support for
578 FreeBSD on non-x86 processors is separate from x86 processors on
579 the config script, much like the NetBSD support.
580 [Richard Levitte & Kris Kennaway <kris@obsecurity.org>]
582 Changes between 0.9.6h and 0.9.7 [31 Dec 2002]
584 *) Fix session ID handling in SSLv2 client code: the SERVER FINISHED
585 code (06) was taken as the first octet of the session ID and the last
586 octet was ignored consequently. As a result SSLv2 client side session
587 caching could not have worked due to the session ID mismatch between
589 Behaviour observed by Crispin Flowerday <crispin@flowerday.cx> as
593 *) Change the declaration of needed Kerberos libraries to use EX_LIBS
594 instead of the special (and badly supported) LIBKRB5. LIBKRB5 is
598 *) The hw_ncipher.c engine requires dynamic locks. Unfortunately, it
599 seems that in spite of existing for more than a year, many application
600 author have done nothing to provide the necessary callbacks, which
601 means that this particular engine will not work properly anywhere.
602 This is a very unfortunate situation which forces us, in the name
603 of usability, to give the hw_ncipher.c a static lock, which is part
605 NOTE: This is for the 0.9.7 series ONLY. This hack will never
606 appear in 0.9.8 or later. We EXPECT application authors to have
607 dealt properly with this when 0.9.8 is released (unless we actually
608 make such changes in the libcrypto locking code that changes will
609 have to be made anyway).
612 *) In asn1_d2i_read_bio() repeatedly call BIO_read() until all content
613 octets have been read, EOF or an error occurs. Without this change
614 some truncated ASN1 structures will not produce an error.
617 *) Disable Heimdal support, since it hasn't been fully implemented.
618 Still give the possibility to force the use of Heimdal, but with
619 warnings and a request that patches get sent to openssl-dev.
622 *) Add the VC-CE target, introduce the WINCE sysname, and add
623 INSTALL.WCE and appropriate conditionals to make it build.
624 [Steven Reddie <smr@essemer.com.au> via Richard Levitte]
626 *) Change the DLL names for Cygwin to cygcrypto-x.y.z.dll and
627 cygssl-x.y.z.dll, where x, y and z are the major, minor and
628 edit numbers of the version.
629 [Corinna Vinschen <vinschen@redhat.com> and Richard Levitte]
631 *) Introduce safe string copy and catenation functions
632 (BUF_strlcpy() and BUF_strlcat()).
633 [Ben Laurie (CHATS) and Richard Levitte]
635 *) Avoid using fixed-size buffers for one-line DNs.
638 *) Add BUF_MEM_grow_clean() to avoid information leakage when
639 resizing buffers containing secrets, and use where appropriate.
642 *) Avoid using fixed size buffers for configuration file location.
645 *) Avoid filename truncation for various CA files.
648 *) Use sizeof in preference to magic numbers.
651 *) Avoid filename truncation in cert requests.
654 *) Add assertions to check for (supposedly impossible) buffer
658 *) Don't cache truncated DNS entries in the local cache (this could
659 potentially lead to a spoofing attack).
662 *) Fix various buffers to be large enough for hex/decimal
663 representations in a platform independent manner.
666 *) Add CRYPTO_realloc_clean() to avoid information leakage when
667 resizing buffers containing secrets, and use where appropriate.
670 *) Add BIO_indent() to avoid much slightly worrying code to do
674 *) Convert sprintf()/BIO_puts() to BIO_printf().
677 *) buffer_gets() could terminate with the buffer only half
681 *) Add assertions to prevent user-supplied crypto functions from
682 overflowing internal buffers by having large block sizes, etc.
685 *) New OPENSSL_assert() macro (similar to assert(), but enabled
689 *) Eliminate unused copy of key in RC4.
692 *) Eliminate unused and incorrectly sized buffers for IV in pem.h.
695 *) Fix off-by-one error in EGD path.
698 *) If RANDFILE path is too long, ignore instead of truncating.
701 *) Eliminate unused and incorrectly sized X.509 structure
705 *) Eliminate unused and dangerous function knumber().
708 *) Eliminate unused and dangerous structure, KSSL_ERR.
711 *) Protect against overlong session ID context length in an encoded
712 session object. Since these are local, this does not appear to be
716 *) Change from security patch (see 0.9.6e below) that did not affect
717 the 0.9.6 release series:
719 Remote buffer overflow in SSL3 protocol - an attacker could
720 supply an oversized master key in Kerberos-enabled versions.
724 *) Change the SSL kerb5 codes to match RFC 2712.
727 *) Make -nameopt work fully for req and add -reqopt switch.
728 [Michael Bell <michael.bell@rz.hu-berlin.de>, Steve Henson]
730 *) The "block size" for block ciphers in CFB and OFB mode should be 1.
731 [Steve Henson, reported by Yngve Nysaeter Pettersen <yngve@opera.com>]
733 *) Make sure tests can be performed even if the corresponding algorithms
734 have been removed entirely. This was also the last step to make
735 OpenSSL compilable with DJGPP under all reasonable conditions.
736 [Richard Levitte, Doug Kaufman <dkaufman@rahul.net>]
738 *) Add cipher selection rules COMPLEMENTOFALL and COMPLEMENTOFDEFAULT
739 to allow version independent disabling of normally unselected ciphers,
740 which may be activated as a side-effect of selecting a single cipher.
742 (E.g., cipher list string "RSA" enables ciphersuites that are left
743 out of "ALL" because they do not provide symmetric encryption.
744 "RSA:!COMPLEMEMENTOFALL" avoids these unsafe ciphersuites.)
745 [Lutz Jaenicke, Bodo Moeller]
747 *) Add appropriate support for separate platform-dependent build
748 directories. The recommended way to make a platform-dependent
749 build directory is the following (tested on Linux), maybe with
752 # Place yourself outside of the OpenSSL source tree. In
753 # this example, the environment variable OPENSSL_SOURCE
754 # is assumed to contain the absolute OpenSSL source directory.
755 mkdir -p objtree/"`uname -s`-`uname -r`-`uname -m`"
756 cd objtree/"`uname -s`-`uname -r`-`uname -m`"
757 (cd $OPENSSL_SOURCE; find . -type f) | while read F; do
758 mkdir -p `dirname $F`
759 ln -s $OPENSSL_SOURCE/$F $F
762 To be absolutely sure not to disturb the source tree, a "make clean"
763 is a good thing. If it isn't successfull, don't worry about it,
764 it probably means the source directory is very clean.
767 *) Make sure any ENGINE control commands make local copies of string
768 pointers passed to them whenever necessary. Otherwise it is possible
769 the caller may have overwritten (or deallocated) the original string
770 data when a later ENGINE operation tries to use the stored values.
771 [Götz Babin-Ebell <babinebell@trustcenter.de>]
773 *) Improve diagnostics in file reading and command-line digests.
774 [Ben Laurie aided and abetted by Solar Designer <solar@openwall.com>]
776 *) Add AES modes CFB and OFB to the object database. Correct an
777 error in AES-CFB decryption.
780 *) Remove most calls to EVP_CIPHER_CTX_cleanup() in evp_enc.c, this
781 allows existing EVP_CIPHER_CTX structures to be reused after
782 calling EVP_*Final(). This behaviour is used by encryption
783 BIOs and some applications. This has the side effect that
784 applications must explicitly clean up cipher contexts with
785 EVP_CIPHER_CTX_cleanup() or they will leak memory.
788 *) Check the values of dna and dnb in bn_mul_recursive before calling
789 bn_mul_comba (a non zero value means the a or b arrays do not contain
790 n2 elements) and fallback to bn_mul_normal if either is not zero.
793 *) Fix escaping of non-ASCII characters when using the -subj option
794 of the "openssl req" command line tool. (Robert Joop <joop@fokus.gmd.de>)
797 *) Make object definitions compliant to LDAP (RFC2256): SN is the short
798 form for "surname", serialNumber has no short form.
799 Use "mail" as the short name for "rfc822Mailbox" according to RFC2798;
800 therefore remove "mail" short name for "internet 7".
801 The OID for unique identifiers in X509 certificates is
802 x500UniqueIdentifier, not uniqueIdentifier.
803 Some more OID additions. (Michael Bell <michael.bell@rz.hu-berlin.de>)
806 *) Add an "init" command to the ENGINE config module and auto initialize
807 ENGINEs. Without any "init" command the ENGINE will be initialized
808 after all ctrl commands have been executed on it. If init=1 the
809 ENGINE is initailized at that point (ctrls before that point are run
810 on the uninitialized ENGINE and after on the initialized one). If
811 init=0 then the ENGINE will not be iniatialized at all.
814 *) Fix the 'app_verify_callback' interface so that the user-defined
815 argument is actually passed to the callback: In the
816 SSL_CTX_set_cert_verify_callback() prototype, the callback
817 declaration has been changed from
820 int (*cb)(X509_STORE_CTX *,void *);
821 in ssl_verify_cert_chain (ssl/ssl_cert.c), the call
822 i=s->ctx->app_verify_callback(&ctx)
823 has been changed into
824 i=s->ctx->app_verify_callback(&ctx, s->ctx->app_verify_arg).
826 To update applications using SSL_CTX_set_cert_verify_callback(),
827 a dummy argument can be added to their callback functions.
828 [D. K. Smetters <smetters@parc.xerox.com>]
830 *) Added the '4758cca' ENGINE to support IBM 4758 cards.
831 [Maurice Gittens <maurice@gittens.nl>, touchups by Geoff Thorpe]
833 *) Add and OPENSSL_LOAD_CONF define which will cause
834 OpenSSL_add_all_algorithms() to load the openssl.cnf config file.
835 This allows older applications to transparently support certain
836 OpenSSL features: such as crypto acceleration and dynamic ENGINE loading.
837 Two new functions OPENSSL_add_all_algorithms_noconf() which will never
838 load the config file and OPENSSL_add_all_algorithms_conf() which will
839 always load it have also been added.
842 *) Add the OFB, CFB and CTR (all with 128 bit feedback) to AES.
843 Adjust NIDs and EVP layer.
844 [Stephen Sprunk <stephen@sprunk.org> and Richard Levitte]
846 *) Config modules support in openssl utility.
848 Most commands now load modules from the config file,
849 though in a few (such as version) this isn't done
850 because it couldn't be used for anything.
852 In the case of ca and req the config file used is
853 the same as the utility itself: that is the -config
854 command line option can be used to specify an
858 *) Move default behaviour from OPENSSL_config(). If appname is NULL
859 use "openssl_conf" if filename is NULL use default openssl config file.
862 *) Add an argument to OPENSSL_config() to allow the use of an alternative
863 config section name. Add a new flag to tolerate a missing config file
864 and move code to CONF_modules_load_file().
867 *) Support for crypto accelerator cards from Accelerated Encryption
868 Processing, www.aep.ie. (Use engine 'aep')
869 The support was copied from 0.9.6c [engine] and adapted/corrected
870 to work with the new engine framework.
871 [AEP Inc. and Richard Levitte]
873 *) Support for SureWare crypto accelerator cards from Baltimore
874 Technologies. (Use engine 'sureware')
875 The support was copied from 0.9.6c [engine] and adapted
876 to work with the new engine framework.
879 *) Have the CHIL engine fork-safe (as defined by nCipher) and actually
880 make the newer ENGINE framework commands for the CHIL engine work.
881 [Toomas Kiisk <vix@cyber.ee> and Richard Levitte]
883 *) Make it possible to produce shared libraries on ReliantUNIX.
884 [Robert Dahlem <Robert.Dahlem@ffm2.siemens.de> via Richard Levitte]
886 *) Add the configuration target debug-linux-ppro.
887 Make 'openssl rsa' use the general key loading routines
888 implemented in apps.c, and make those routines able to
889 handle the key format FORMAT_NETSCAPE and the variant
891 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte]
893 *) Fix a crashbug and a logic bug in hwcrhk_load_pubkey().
894 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte]
896 *) Add -keyform to rsautl, and document -engine.
897 [Richard Levitte, inspired by Toomas Kiisk <vix@cyber.ee>]
899 *) Change BIO_new_file (crypto/bio/bss_file.c) to use new
900 BIO_R_NO_SUCH_FILE error code rather than the generic
901 ERR_R_SYS_LIB error code if fopen() fails with ENOENT.
906 ERR_peek_last_error_line
907 ERR_peek_last_error_line_data.
911 ERR_peek_error_line_data,
912 but report on the latest error recorded rather than the first one
913 still in the error queue.
914 [Ben Laurie, Bodo Moeller]
916 *) default_algorithms option in ENGINE config module. This allows things
918 default_algorithms = ALL
919 default_algorithms = RSA, DSA, RAND, CIPHERS, DIGESTS
922 *) Prelminary ENGINE config module.
925 *) New experimental application configuration code.
928 *) Change the AES code to follow the same name structure as all other
929 symmetric ciphers, and behave the same way. Move everything to
930 the directory crypto/aes, thereby obsoleting crypto/rijndael.
931 [Stephen Sprunk <stephen@sprunk.org> and Richard Levitte]
933 *) SECURITY: remove unsafe setjmp/signal interaction from ui_openssl.c.
934 [Ben Laurie and Theo de Raadt]
936 *) Add option to output public keys in req command.
937 [Massimiliano Pala madwolf@openca.org]
939 *) Use wNAFs in EC_POINTs_mul() for improved efficiency
940 (up to about 10% better than before for P-192 and P-224).
943 *) New functions/macros
945 SSL_CTX_set_msg_callback(ctx, cb)
946 SSL_CTX_set_msg_callback_arg(ctx, arg)
947 SSL_set_msg_callback(ssl, cb)
948 SSL_set_msg_callback_arg(ssl, arg)
950 to request calling a callback function
952 void cb(int write_p, int version, int content_type,
953 const void *buf, size_t len, SSL *ssl, void *arg)
955 whenever a protocol message has been completely received
956 (write_p == 0) or sent (write_p == 1). Here 'version' is the
957 protocol version according to which the SSL library interprets
958 the current protocol message (SSL2_VERSION, SSL3_VERSION, or
959 TLS1_VERSION). 'content_type' is 0 in the case of SSL 2.0, or
960 the content type as defined in the SSL 3.0/TLS 1.0 protocol
961 specification (change_cipher_spec(20), alert(21), handshake(22)).
962 'buf' and 'len' point to the actual message, 'ssl' to the
963 SSL object, and 'arg' is the application-defined value set by
964 SSL[_CTX]_set_msg_callback_arg().
966 'openssl s_client' and 'openssl s_server' have new '-msg' options
967 to enable a callback that displays all protocol messages.
970 *) Change the shared library support so shared libraries are built as
971 soon as the corresponding static library is finished, and thereby get
972 openssl and the test programs linked against the shared library.
973 This still only happens when the keyword "shard" has been given to
974 the configuration scripts.
976 NOTE: shared library support is still an experimental thing, and
977 backward binary compatibility is still not guaranteed.
978 ["Maciej W. Rozycki" <macro@ds2.pg.gda.pl> and Richard Levitte]
980 *) Add support for Subject Information Access extension.
981 [Peter Sylvester <Peter.Sylvester@EdelWeb.fr>]
983 *) Make BUF_MEM_grow() behaviour more consistent: Initialise to zero
984 additional bytes when new memory had to be allocated, not just
985 when reusing an existing buffer.
988 *) New command line and configuration option 'utf8' for the req command.
989 This allows field values to be specified as UTF8 strings.
992 *) Add -multi and -mr options to "openssl speed" - giving multiple parallel
993 runs for the former and machine-readable output for the latter.
996 *) Add '-noemailDN' option to 'openssl ca'. This prevents inclusion
997 of the e-mail address in the DN (i.e., it will go into a certificate
998 extension only). The new configuration file option 'email_in_dn = no'
1000 [Massimiliano Pala madwolf@openca.org]
1002 *) Change all functions with names starting with des_ to be starting
1003 with DES_ instead. Add wrappers that are compatible with libdes,
1004 but are named _ossl_old_des_*. Finally, add macros that map the
1005 des_* symbols to the corresponding _ossl_old_des_* if libdes
1006 compatibility is desired. If OpenSSL 0.9.6c compatibility is
1007 desired, the des_* symbols will be mapped to DES_*, with one
1010 Since we provide two compatibility mappings, the user needs to
1011 define the macro OPENSSL_DES_LIBDES_COMPATIBILITY if libdes
1012 compatibility is desired. The default (i.e., when that macro
1013 isn't defined) is OpenSSL 0.9.6c compatibility.
1015 There are also macros that enable and disable the support of old
1016 des functions altogether. Those are OPENSSL_ENABLE_OLD_DES_SUPPORT
1017 and OPENSSL_DISABLE_OLD_DES_SUPPORT. If none or both of those
1018 are defined, the default will apply: to support the old des routines.
1020 In either case, one must include openssl/des.h to get the correct
1021 definitions. Do not try to just include openssl/des_old.h, that
1024 NOTE: This is a major break of an old API into a new one. Software
1025 authors are encouraged to switch to the DES_ style functions. Some
1026 time in the future, des_old.h and the libdes compatibility functions
1027 will be disable (i.e. OPENSSL_DISABLE_OLD_DES_SUPPORT will be the
1028 default), and then completely removed.
1031 *) Test for certificates which contain unsupported critical extensions.
1032 If such a certificate is found during a verify operation it is
1033 rejected by default: this behaviour can be overridden by either
1034 handling the new error X509_V_ERR_UNHANDLED_CRITICAL_EXTENSION or
1035 by setting the verify flag X509_V_FLAG_IGNORE_CRITICAL. A new function
1036 X509_supported_extension() has also been added which returns 1 if a
1037 particular extension is supported.
1040 *) Modify the behaviour of EVP cipher functions in similar way to digests
1041 to retain compatibility with existing code.
1044 *) Modify the behaviour of EVP_DigestInit() and EVP_DigestFinal() to retain
1045 compatibility with existing code. In particular the 'ctx' parameter does
1046 not have to be to be initialized before the call to EVP_DigestInit() and
1047 it is tidied up after a call to EVP_DigestFinal(). New function
1048 EVP_DigestFinal_ex() which does not tidy up the ctx. Similarly function
1049 EVP_MD_CTX_copy() changed to not require the destination to be
1050 initialized valid and new function EVP_MD_CTX_copy_ex() added which
1051 requires the destination to be valid.
1053 Modify all the OpenSSL digest calls to use EVP_DigestInit_ex(),
1054 EVP_DigestFinal_ex() and EVP_MD_CTX_copy_ex().
1057 *) Change ssl3_get_message (ssl/s3_both.c) and the functions using it
1058 so that complete 'Handshake' protocol structures are kept in memory
1059 instead of overwriting 'msg_type' and 'length' with 'body' data.
1062 *) Add an implementation of SSL_add_dir_cert_subjects_to_stack for Win32.
1063 [Massimo Santin via Richard Levitte]
1065 *) Major restructuring to the underlying ENGINE code. This includes
1066 reduction of linker bloat, separation of pure "ENGINE" manipulation
1067 (initialisation, etc) from functionality dealing with implementations
1068 of specific crypto iterfaces. This change also introduces integrated
1069 support for symmetric ciphers and digest implementations - so ENGINEs
1070 can now accelerate these by providing EVP_CIPHER and EVP_MD
1071 implementations of their own. This is detailed in crypto/engine/README
1072 as it couldn't be adequately described here. However, there are a few
1073 API changes worth noting - some RSA, DSA, DH, and RAND functions that
1074 were changed in the original introduction of ENGINE code have now
1075 reverted back - the hooking from this code to ENGINE is now a good
1076 deal more passive and at run-time, operations deal directly with
1077 RSA_METHODs, DSA_METHODs (etc) as they did before, rather than
1078 dereferencing through an ENGINE pointer any more. Also, the ENGINE
1079 functions dealing with BN_MOD_EXP[_CRT] handlers have been removed -
1080 they were not being used by the framework as there is no concept of a
1081 BIGNUM_METHOD and they could not be generalised to the new
1082 'ENGINE_TABLE' mechanism that underlies the new code. Similarly,
1083 ENGINE_cpy() has been removed as it cannot be consistently defined in
1087 *) Change ASN1_GENERALIZEDTIME_check() to allow fractional seconds.
1090 *) Change mkdef.pl to sort symbols that get the same entry number,
1091 and make sure the automatically generated functions ERR_load_*
1092 become part of libeay.num as well.
1095 *) New function SSL_renegotiate_pending(). This returns true once
1096 renegotiation has been requested (either SSL_renegotiate() call
1097 or HelloRequest/ClientHello receveived from the peer) and becomes
1098 false once a handshake has been completed.
1099 (For servers, SSL_renegotiate() followed by SSL_do_handshake()
1100 sends a HelloRequest, but does not ensure that a handshake takes
1101 place. SSL_renegotiate_pending() is useful for checking if the
1102 client has followed the request.)
1105 *) New SSL option SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION.
1106 By default, clients may request session resumption even during
1107 renegotiation (if session ID contexts permit); with this option,
1108 session resumption is possible only in the first handshake.
1110 SSL_OP_ALL is now 0x00000FFFL instead of 0x000FFFFFL. This makes
1111 more bits available for options that should not be part of
1112 SSL_OP_ALL (such as SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION).
1115 *) Add some demos for certificate and certificate request creation.
1118 *) Make maximum certificate chain size accepted from the peer application
1119 settable (SSL*_get/set_max_cert_list()), as proposed by
1120 "Douglas E. Engert" <deengert@anl.gov>.
1123 *) Add support for shared libraries for Unixware-7
1124 (Boyd Lynn Gerber <gerberb@zenez.com>).
1127 *) Add a "destroy" handler to ENGINEs that allows structural cleanup to
1128 be done prior to destruction. Use this to unload error strings from
1129 ENGINEs that load their own error strings. NB: This adds two new API
1130 functions to "get" and "set" this destroy handler in an ENGINE.
1133 *) Alter all existing ENGINE implementations (except "openssl" and
1134 "openbsd") to dynamically instantiate their own error strings. This
1135 makes them more flexible to be built both as statically-linked ENGINEs
1136 and self-contained shared-libraries loadable via the "dynamic" ENGINE.
1137 Also, add stub code to each that makes building them as self-contained
1138 shared-libraries easier (see README.ENGINE).
1141 *) Add a "dynamic" ENGINE that provides a mechanism for binding ENGINE
1142 implementations into applications that are completely implemented in
1143 self-contained shared-libraries. The "dynamic" ENGINE exposes control
1144 commands that can be used to configure what shared-library to load and
1145 to control aspects of the way it is handled. Also, made an update to
1146 the README.ENGINE file that brings its information up-to-date and
1147 provides some information and instructions on the "dynamic" ENGINE
1148 (ie. how to use it, how to build "dynamic"-loadable ENGINEs, etc).
1151 *) Make it possible to unload ranges of ERR strings with a new
1152 "ERR_unload_strings" function.
1155 *) Add a copy() function to EVP_MD.
1158 *) Make EVP_MD routines take a context pointer instead of just the
1159 md_data void pointer.
1162 *) Add flags to EVP_MD and EVP_MD_CTX. EVP_MD_FLAG_ONESHOT indicates
1163 that the digest can only process a single chunk of data
1164 (typically because it is provided by a piece of
1165 hardware). EVP_MD_CTX_FLAG_ONESHOT indicates that the application
1166 is only going to provide a single chunk of data, and hence the
1167 framework needn't accumulate the data for oneshot drivers.
1170 *) As with "ERR", make it possible to replace the underlying "ex_data"
1171 functions. This change also alters the storage and management of global
1172 ex_data state - it's now all inside ex_data.c and all "class" code (eg.
1173 RSA, BIO, SSL_CTX, etc) no longer stores its own STACKS and per-class
1174 index counters. The API functions that use this state have been changed
1175 to take a "class_index" rather than pointers to the class's local STACK
1176 and counter, and there is now an API function to dynamically create new
1177 classes. This centralisation allows us to (a) plug a lot of the
1178 thread-safety problems that existed, and (b) makes it possible to clean
1179 up all allocated state using "CRYPTO_cleanup_all_ex_data()". W.r.t. (b)
1180 such data would previously have always leaked in application code and
1181 workarounds were in place to make the memory debugging turn a blind eye
1182 to it. Application code that doesn't use this new function will still
1183 leak as before, but their memory debugging output will announce it now
1184 rather than letting it slide.
1186 Besides the addition of CRYPTO_cleanup_all_ex_data(), another API change
1187 induced by the "ex_data" overhaul is that X509_STORE_CTX_init() now
1188 has a return value to indicate success or failure.
1191 *) Make it possible to replace the underlying "ERR" functions such that the
1192 global state (2 LHASH tables and 2 locks) is only used by the "default"
1193 implementation. This change also adds two functions to "get" and "set"
1194 the implementation prior to it being automatically set the first time
1195 any other ERR function takes place. Ie. an application can call "get",
1196 pass the return value to a module it has just loaded, and that module
1197 can call its own "set" function using that value. This means the
1198 module's "ERR" operations will use (and modify) the error state in the
1199 application and not in its own statically linked copy of OpenSSL code.
1202 *) Give DH, DSA, and RSA types their own "**_up_ref()" function to increment
1203 reference counts. This performs normal REF_PRINT/REF_CHECK macros on
1204 the operation, and provides a more encapsulated way for external code
1205 (crypto/evp/ and ssl/) to do this. Also changed the evp and ssl code
1206 to use these functions rather than manually incrementing the counts.
1208 Also rename "DSO_up()" function to more descriptive "DSO_up_ref()".
1211 *) Add EVP test program.
1214 *) Add symmetric cipher support to ENGINE. Expect the API to change!
1217 *) New CRL functions: X509_CRL_set_version(), X509_CRL_set_issuer_name()
1218 X509_CRL_set_lastUpdate(), X509_CRL_set_nextUpdate(), X509_CRL_sort(),
1219 X509_REVOKED_set_serialNumber(), and X509_REVOKED_set_revocationDate().
1220 These allow a CRL to be built without having to access X509_CRL fields
1221 directly. Modify 'ca' application to use new functions.
1224 *) Move SSL_OP_TLS_ROLLBACK_BUG out of the SSL_OP_ALL list of recommended
1225 bug workarounds. Rollback attack detection is a security feature.
1226 The problem will only arise on OpenSSL servers when TLSv1 is not
1227 available (sslv3_server_method() or SSL_OP_NO_TLSv1).
1228 Software authors not wanting to support TLSv1 will have special reasons
1229 for their choice and can explicitly enable this option.
1230 [Bodo Moeller, Lutz Jaenicke]
1232 *) Rationalise EVP so it can be extended: don't include a union of
1233 cipher/digest structures, add init/cleanup functions for EVP_MD_CTX
1234 (similar to those existing for EVP_CIPHER_CTX).
1239 EVP_MD_CTX_init(&md); /* new function call */
1240 EVP_DigestInit(&md, EVP_sha1());
1241 EVP_DigestUpdate(&md, in, len);
1242 EVP_DigestFinal(&md, out, NULL);
1243 EVP_MD_CTX_cleanup(&md); /* new function call */
1247 *) Make DES key schedule conform to the usual scheme, as well as
1248 correcting its structure. This means that calls to DES functions
1249 now have to pass a pointer to a des_key_schedule instead of a
1250 plain des_key_schedule (which was actually always a pointer
1253 des_key_schedule ks;
1255 des_set_key_checked(..., &ks);
1256 des_ncbc_encrypt(..., &ks, ...);
1258 (Note that a later change renames 'des_...' into 'DES_...'.)
1261 *) Initial reduction of linker bloat: the use of some functions, such as
1262 PEM causes large amounts of unused functions to be linked in due to
1263 poor organisation. For example pem_all.c contains every PEM function
1264 which has a knock on effect of linking in large amounts of (unused)
1265 ASN1 code. Grouping together similar functions and splitting unrelated
1266 functions prevents this.
1269 *) Cleanup of EVP macros.
1272 *) Change historical references to {NID,SN,LN}_des_ede and ede3 to add the
1273 correct _ecb suffix.
1276 *) Add initial OCSP responder support to ocsp application. The
1277 revocation information is handled using the text based index
1278 use by the ca application. The responder can either handle
1279 requests generated internally, supplied in files (for example
1280 via a CGI script) or using an internal minimal server.
1283 *) Add configuration choices to get zlib compression for TLS.
1286 *) Changes to Kerberos SSL for RFC 2712 compliance:
1287 1. Implemented real KerberosWrapper, instead of just using
1288 KRB5 AP_REQ message. [Thanks to Simon Wilkinson <sxw@sxw.org.uk>]
1289 2. Implemented optional authenticator field of KerberosWrapper.
1291 Added openssl-style ASN.1 macros for Kerberos ticket, ap_req,
1292 and authenticator structs; see crypto/krb5/.
1294 Generalized Kerberos calls to support multiple Kerberos libraries.
1295 [Vern Staats <staatsvr@asc.hpc.mil>,
1296 Jeffrey Altman <jaltman@columbia.edu>
1297 via Richard Levitte]
1299 *) Cause 'openssl speed' to use fully hard-coded DSA keys as it
1300 already does with RSA. testdsa.h now has 'priv_key/pub_key'
1301 values for each of the key sizes rather than having just
1302 parameters (and 'speed' generating keys each time).
1305 *) Speed up EVP routines.
1308 type 8 bytes 64 bytes 256 bytes 1024 bytes 8192 bytes
1309 des-cbc 4408.85k 5560.51k 5778.46k 5862.20k 5825.16k
1310 des-cbc 4389.55k 5571.17k 5792.23k 5846.91k 5832.11k
1311 des-cbc 4394.32k 5575.92k 5807.44k 5848.37k 5841.30k
1313 des-cbc 3482.66k 5069.49k 5496.39k 5614.16k 5639.28k
1314 des-cbc 3480.74k 5068.76k 5510.34k 5609.87k 5635.52k
1315 des-cbc 3483.72k 5067.62k 5504.60k 5708.01k 5724.80k
1318 des-cbc 4660.16k 5650.19k 5807.19k 5827.13k 5783.32k
1320 des-cbc 3624.96k 5258.21k 5530.91k 5624.30k 5628.26k
1323 *) Added the OS2-EMX target.
1324 ["Brian Havard" <brianh@kheldar.apana.org.au> and Richard Levitte]
1326 *) Rewrite apps to use NCONF routines instead of the old CONF. New functions
1327 to support NCONF routines in extension code. New function CONF_set_nconf()
1328 to allow functions which take an NCONF to also handle the old LHASH
1329 structure: this means that the old CONF compatible routines can be
1330 retained (in particular wrt extensions) without having to duplicate the
1331 code. New function X509V3_add_ext_nconf_sk to add extensions to a stack.
1334 *) Enhance the general user interface with mechanisms for inner control
1335 and with possibilities to have yes/no kind of prompts.
1338 *) Change all calls to low level digest routines in the library and
1339 applications to use EVP. Add missing calls to HMAC_cleanup() and
1340 don't assume HMAC_CTX can be copied using memcpy().
1341 [Verdon Walker <VWalker@novell.com>, Steve Henson]
1343 *) Add the possibility to control engines through control names but with
1344 arbitrary arguments instead of just a string.
1345 Change the key loaders to take a UI_METHOD instead of a callback
1346 function pointer. NOTE: this breaks binary compatibility with earlier
1347 versions of OpenSSL [engine].
1348 Adapt the nCipher code for these new conditions and add a card insertion
1352 *) Enhance the general user interface with mechanisms to better support
1353 dialog box interfaces, application-defined prompts, the possibility
1354 to use defaults (for example default passwords from somewhere else)
1355 and interrupts/cancellations.
1358 *) Tidy up PKCS#12 attribute handling. Add support for the CSP name
1359 attribute in PKCS#12 files, add new -CSP option to pkcs12 utility.
1362 *) Fix a memory leak in 'sk_dup()' in the case reallocation fails. (Also
1363 tidy up some unnecessarily weird code in 'sk_new()').
1364 [Geoff, reported by Diego Tartara <dtartara@novamens.com>]
1366 *) Change the key loading routines for ENGINEs to use the same kind
1367 callback (pem_password_cb) as all other routines that need this
1371 *) Increase ENTROPY_NEEDED to 32 bytes, as Rijndael can operate with
1372 256 bit (=32 byte) keys. Of course seeding with more entropy bytes
1373 than this minimum value is recommended.
1376 *) New random seeder for OpenVMS, using the system process statistics
1377 that are easily reachable.
1380 *) Windows apparently can't transparently handle global
1381 variables defined in DLLs. Initialisations such as:
1383 const ASN1_ITEM *it = &ASN1_INTEGER_it;
1385 wont compile. This is used by the any applications that need to
1386 declare their own ASN1 modules. This was fixed by adding the option
1387 EXPORT_VAR_AS_FN to all Win32 platforms, although this isn't strictly
1388 needed for static libraries under Win32.
1391 *) New functions X509_PURPOSE_set() and X509_TRUST_set() to handle
1392 setting of purpose and trust fields. New X509_STORE trust and
1393 purpose functions and tidy up setting in other SSL functions.
1396 *) Add copies of X509_STORE_CTX fields and callbacks to X509_STORE
1397 structure. These are inherited by X509_STORE_CTX when it is
1398 initialised. This allows various defaults to be set in the
1399 X509_STORE structure (such as flags for CRL checking and custom
1400 purpose or trust settings) for functions which only use X509_STORE_CTX
1401 internally such as S/MIME.
1403 Modify X509_STORE_CTX_purpose_inherit() so it only sets purposes and
1404 trust settings if they are not set in X509_STORE. This allows X509_STORE
1405 purposes and trust (in S/MIME for example) to override any set by default.
1407 Add command line options for CRL checking to smime, s_client and s_server
1411 *) Initial CRL based revocation checking. If the CRL checking flag(s)
1412 are set then the CRL is looked up in the X509_STORE structure and
1413 its validity and signature checked, then if the certificate is found
1414 in the CRL the verify fails with a revoked error.
1416 Various new CRL related callbacks added to X509_STORE_CTX structure.
1418 Command line options added to 'verify' application to support this.
1420 This needs some additional work, such as being able to handle multiple
1421 CRLs with different times, extension based lookup (rather than just
1422 by subject name) and ultimately more complete V2 CRL extension
1426 *) Add a general user interface API (crypto/ui/). This is designed
1427 to replace things like des_read_password and friends (backward
1428 compatibility functions using this new API are provided).
1429 The purpose is to remove prompting functions from the DES code
1430 section as well as provide for prompting through dialog boxes in
1431 a window system and the like.
1434 *) Add "ex_data" support to ENGINE so implementations can add state at a
1435 per-structure level rather than having to store it globally.
1438 *) Make it possible for ENGINE structures to be copied when retrieved by
1439 ENGINE_by_id() if the ENGINE specifies a new flag: ENGINE_FLAGS_BY_ID_COPY.
1440 This causes the "original" ENGINE structure to act like a template,
1441 analogous to the RSA vs. RSA_METHOD type of separation. Because of this
1442 operational state can be localised to each ENGINE structure, despite the
1443 fact they all share the same "methods". New ENGINE structures returned in
1444 this case have no functional references and the return value is the single
1445 structural reference. This matches the single structural reference returned
1446 by ENGINE_by_id() normally, when it is incremented on the pre-existing
1450 *) Fix ASN1 decoder when decoding type ANY and V_ASN1_OTHER: since this
1451 needs to match any other type at all we need to manually clear the
1455 *) Changes to the "openssl engine" utility to include;
1456 - verbosity levels ('-v', '-vv', and '-vvv') that provide information
1457 about an ENGINE's available control commands.
1458 - executing control commands from command line arguments using the
1459 '-pre' and '-post' switches. '-post' is only used if '-t' is
1460 specified and the ENGINE is successfully initialised. The syntax for
1461 the individual commands are colon-separated, for example;
1462 openssl engine chil -pre FORK_CHECK:0 -pre SO_PATH:/lib/test.so
1465 *) New dynamic control command support for ENGINEs. ENGINEs can now
1466 declare their own commands (numbers), names (strings), descriptions,
1467 and input types for run-time discovery by calling applications. A
1468 subset of these commands are implicitly classed as "executable"
1469 depending on their input type, and only these can be invoked through
1470 the new string-based API function ENGINE_ctrl_cmd_string(). (Eg. this
1471 can be based on user input, config files, etc). The distinction is
1472 that "executable" commands cannot return anything other than a boolean
1473 result and can only support numeric or string input, whereas some
1474 discoverable commands may only be for direct use through
1475 ENGINE_ctrl(), eg. supporting the exchange of binary data, function
1476 pointers, or other custom uses. The "executable" commands are to
1477 support parameterisations of ENGINE behaviour that can be
1478 unambiguously defined by ENGINEs and used consistently across any
1479 OpenSSL-based application. Commands have been added to all the
1480 existing hardware-supporting ENGINEs, noticeably "SO_PATH" to allow
1481 control over shared-library paths without source code alterations.
1484 *) Changed all ENGINE implementations to dynamically allocate their
1485 ENGINEs rather than declaring them statically. Apart from this being
1486 necessary with the removal of the ENGINE_FLAGS_MALLOCED distinction,
1487 this also allows the implementations to compile without using the
1488 internal engine_int.h header.
1491 *) Minor adjustment to "rand" code. RAND_get_rand_method() now returns a
1492 'const' value. Any code that should be able to modify a RAND_METHOD
1493 should already have non-const pointers to it (ie. they should only
1494 modify their own ones).
1497 *) Made a variety of little tweaks to the ENGINE code.
1498 - "atalla" and "ubsec" string definitions were moved from header files
1499 to C code. "nuron" string definitions were placed in variables
1500 rather than hard-coded - allowing parameterisation of these values
1501 later on via ctrl() commands.
1502 - Removed unused "#if 0"'d code.
1503 - Fixed engine list iteration code so it uses ENGINE_free() to release
1504 structural references.
1505 - Constified the RAND_METHOD element of ENGINE structures.
1506 - Constified various get/set functions as appropriate and added
1507 missing functions (including a catch-all ENGINE_cpy that duplicates
1508 all ENGINE values onto a new ENGINE except reference counts/state).
1509 - Removed NULL parameter checks in get/set functions. Setting a method
1510 or function to NULL is a way of cancelling out a previously set
1511 value. Passing a NULL ENGINE parameter is just plain stupid anyway
1512 and doesn't justify the extra error symbols and code.
1513 - Deprecate the ENGINE_FLAGS_MALLOCED define and move the area for
1514 flags from engine_int.h to engine.h.
1515 - Changed prototypes for ENGINE handler functions (init(), finish(),
1516 ctrl(), key-load functions, etc) to take an (ENGINE*) parameter.
1519 *) Implement binary inversion algorithm for BN_mod_inverse in addition
1520 to the algorithm using long division. The binary algorithm can be
1521 used only if the modulus is odd. On 32-bit systems, it is faster
1522 only for relatively small moduli (roughly 20-30% for 128-bit moduli,
1523 roughly 5-15% for 256-bit moduli), so we use it only for moduli
1524 up to 450 bits. In 64-bit environments, the binary algorithm
1525 appears to be advantageous for much longer moduli; here we use it
1526 for moduli up to 2048 bits.
1529 *) Rewrite CHOICE field setting in ASN1_item_ex_d2i(). The old code
1530 could not support the combine flag in choice fields.
1533 *) Add a 'copy_extensions' option to the 'ca' utility. This copies
1534 extensions from a certificate request to the certificate.
1537 *) Allow multiple 'certopt' and 'nameopt' options to be separated
1538 by commas. Add 'namopt' and 'certopt' options to the 'ca' config
1539 file: this allows the display of the certificate about to be
1540 signed to be customised, to allow certain fields to be included
1541 or excluded and extension details. The old system didn't display
1542 multicharacter strings properly, omitted fields not in the policy
1543 and couldn't display additional details such as extensions.
1546 *) Function EC_POINTs_mul for multiple scalar multiplication
1547 of an arbitrary number of elliptic curve points
1548 \sum scalars[i]*points[i],
1549 optionally including the generator defined for the EC_GROUP:
1550 scalar*generator + \sum scalars[i]*points[i].
1552 EC_POINT_mul is a simple wrapper function for the typical case
1553 that the point list has just one item (besides the optional
1557 *) First EC_METHODs for curves over GF(p):
1559 EC_GFp_simple_method() uses the basic BN_mod_mul and BN_mod_sqr
1560 operations and provides various method functions that can also
1561 operate with faster implementations of modular arithmetic.
1563 EC_GFp_mont_method() reuses most functions that are part of
1564 EC_GFp_simple_method, but uses Montgomery arithmetic.
1566 [Bodo Moeller; point addition and point doubling
1567 implementation directly derived from source code provided by
1568 Lenka Fibikova <fibikova@exp-math.uni-essen.de>]
1570 *) Framework for elliptic curves (crypto/ec/ec.h, crypto/ec/ec_lcl.h,
1571 crypto/ec/ec_lib.c):
1573 Curves are EC_GROUP objects (with an optional group generator)
1574 based on EC_METHODs that are built into the library.
1576 Points are EC_POINT objects based on EC_GROUP objects.
1578 Most of the framework would be able to handle curves over arbitrary
1579 finite fields, but as there are no obvious types for fields other
1580 than GF(p), some functions are limited to that for now.
1583 *) Add the -HTTP option to s_server. It is similar to -WWW, but requires
1584 that the file contains a complete HTTP response.
1587 *) Add the ec directory to mkdef.pl and mkfiles.pl. In mkdef.pl
1588 change the def and num file printf format specifier from "%-40sXXX"
1589 to "%-39s XXX". The latter will always guarantee a space after the
1590 field while the former will cause them to run together if the field
1591 is 40 of more characters long.
1594 *) Constify the cipher and digest 'method' functions and structures
1595 and modify related functions to take constant EVP_MD and EVP_CIPHER
1599 *) Hide BN_CTX structure details in bn_lcl.h instead of publishing them
1600 in <openssl/bn.h>. Also further increase BN_CTX_NUM to 32.
1603 *) Modify EVP_Digest*() routines so they now return values. Although the
1604 internal software routines can never fail additional hardware versions
1608 *) Clean up crypto/err/err.h and change some error codes to avoid conflicts:
1610 Previously ERR_R_FATAL was too small and coincided with ERR_LIB_PKCS7
1611 (= ERR_R_PKCS7_LIB); it is now 64 instead of 32.
1614 ERR_R_NESTED_ASN1_ERROR
1616 ERR_R_MISSING_ASN1_EOS
1617 were 4 .. 9, conflicting with
1618 ERR_LIB_RSA (= ERR_R_RSA_LIB)
1620 ERR_LIB_PEM (= ERR_R_PEM_LIB).
1621 They are now 58 .. 63 (i.e., just below ERR_R_FATAL).
1623 Add new error code 'ERR_R_INTERNAL_ERROR'.
1626 *) Don't overuse locks in crypto/err/err.c: For data retrieval, CRYPTO_r_lock
1630 *) New option '-subj arg' for 'openssl req' and 'openssl ca'. This
1631 sets the subject name for a new request or supersedes the
1632 subject name in a given request. Formats that can be parsed are
1633 'CN=Some Name, OU=myOU, C=IT'
1635 'CN=Some Name/OU=myOU/C=IT'.
1637 Add options '-batch' and '-verbose' to 'openssl req'.
1638 [Massimiliano Pala <madwolf@hackmasters.net>]
1640 *) Introduce the possibility to access global variables through
1641 functions on platform were that's the best way to handle exporting
1642 global variables in shared libraries. To enable this functionality,
1643 one must configure with "EXPORT_VAR_AS_FN" or defined the C macro
1644 "OPENSSL_EXPORT_VAR_AS_FUNCTION" in crypto/opensslconf.h (the latter
1645 is normally done by Configure or something similar).
1647 To implement a global variable, use the macro OPENSSL_IMPLEMENT_GLOBAL
1648 in the source file (foo.c) like this:
1650 OPENSSL_IMPLEMENT_GLOBAL(int,foo)=1;
1651 OPENSSL_IMPLEMENT_GLOBAL(double,bar);
1653 To declare a global variable, use the macros OPENSSL_DECLARE_GLOBAL
1654 and OPENSSL_GLOBAL_REF in the header file (foo.h) like this:
1656 OPENSSL_DECLARE_GLOBAL(int,foo);
1657 #define foo OPENSSL_GLOBAL_REF(foo)
1658 OPENSSL_DECLARE_GLOBAL(double,bar);
1659 #define bar OPENSSL_GLOBAL_REF(bar)
1661 The #defines are very important, and therefore so is including the
1662 header file everywhere where the defined globals are used.
1664 The macro OPENSSL_EXPORT_VAR_AS_FUNCTION also affects the definition
1665 of ASN.1 items, but that structure is a bit different.
1667 The largest change is in util/mkdef.pl which has been enhanced with
1668 better and easier to understand logic to choose which symbols should
1669 go into the Windows .def files as well as a number of fixes and code
1670 cleanup (among others, algorithm keywords are now sorted
1671 lexicographically to avoid constant rewrites).
1674 *) In BN_div() keep a copy of the sign of 'num' before writing the
1675 result to 'rm' because if rm==num the value will be overwritten
1676 and produce the wrong result if 'num' is negative: this caused
1677 problems with BN_mod() and BN_nnmod().
1680 *) Function OCSP_request_verify(). This checks the signature on an
1681 OCSP request and verifies the signer certificate. The signer
1682 certificate is just checked for a generic purpose and OCSP request
1686 *) Add OCSP_check_validity() function to check the validity of OCSP
1687 responses. OCSP responses are prepared in real time and may only
1688 be a few seconds old. Simply checking that the current time lies
1689 between thisUpdate and nextUpdate max reject otherwise valid responses
1690 caused by either OCSP responder or client clock inaccuracy. Instead
1691 we allow thisUpdate and nextUpdate to fall within a certain period of
1692 the current time. The age of the response can also optionally be
1693 checked. Two new options -validity_period and -status_age added to
1697 *) If signature or public key algorithm is unrecognized print out its
1698 OID rather that just UNKNOWN.
1701 *) Change OCSP_cert_to_id() to tolerate a NULL subject certificate and
1702 OCSP_cert_id_new() a NULL serialNumber. This allows a partial certificate
1703 ID to be generated from the issuer certificate alone which can then be
1704 passed to OCSP_id_issuer_cmp().
1707 *) New compilation option ASN1_ITEM_FUNCTIONS. This causes the new
1708 ASN1 modules to export functions returning ASN1_ITEM pointers
1709 instead of the ASN1_ITEM structures themselves. This adds several
1710 new macros which allow the underlying ASN1 function/structure to
1711 be accessed transparently. As a result code should not use ASN1_ITEM
1712 references directly (such as &X509_it) but instead use the relevant
1713 macros (such as ASN1_ITEM_rptr(X509)). This option is to allow
1714 use of the new ASN1 code on platforms where exporting structures
1715 is problematical (for example in shared libraries) but exporting
1716 functions returning pointers to structures is not.
1719 *) Add support for overriding the generation of SSL/TLS session IDs.
1720 These callbacks can be registered either in an SSL_CTX or per SSL.
1721 The purpose of this is to allow applications to control, if they wish,
1722 the arbitrary values chosen for use as session IDs, particularly as it
1723 can be useful for session caching in multiple-server environments. A
1724 command-line switch for testing this (and any client code that wishes
1725 to use such a feature) has been added to "s_server".
1726 [Geoff Thorpe, Lutz Jaenicke]
1728 *) Modify mkdef.pl to recognise and parse preprocessor conditionals
1729 of the form '#if defined(...) || defined(...) || ...' and
1730 '#if !defined(...) && !defined(...) && ...'. This also avoids
1731 the growing number of special cases it was previously handling.
1734 *) Make all configuration macros available for application by making
1735 sure they are available in opensslconf.h, by giving them names starting
1736 with "OPENSSL_" to avoid conflicts with other packages and by making
1737 sure e_os2.h will cover all platform-specific cases together with
1739 Additionally, it is now possible to define configuration/platform-
1740 specific names (called "system identities"). In the C code, these
1741 are prefixed with "OPENSSL_SYSNAME_". e_os2.h will create another
1742 macro with the name beginning with "OPENSSL_SYS_", which is determined
1743 from "OPENSSL_SYSNAME_*" or compiler-specific macros depending on
1747 *) New option -set_serial to 'req' and 'x509' this allows the serial
1748 number to use to be specified on the command line. Previously self
1749 signed certificates were hard coded with serial number 0 and the
1750 CA options of 'x509' had to use a serial number in a file which was
1754 *) New options to 'ca' utility to support V2 CRL entry extensions.
1755 Currently CRL reason, invalidity date and hold instruction are
1756 supported. Add new CRL extensions to V3 code and some new objects.
1759 *) New function EVP_CIPHER_CTX_set_padding() this is used to
1760 disable standard block padding (aka PKCS#5 padding) in the EVP
1761 API, which was previously mandatory. This means that the data is
1762 not padded in any way and so the total length much be a multiple
1763 of the block size, otherwise an error occurs.
1766 *) Initial (incomplete) OCSP SSL support.
1769 *) New function OCSP_parse_url(). This splits up a URL into its host,
1770 port and path components: primarily to parse OCSP URLs. New -url
1771 option to ocsp utility.
1774 *) New nonce behavior. The return value of OCSP_check_nonce() now
1775 reflects the various checks performed. Applications can decide
1776 whether to tolerate certain situations such as an absent nonce
1777 in a response when one was present in a request: the ocsp application
1778 just prints out a warning. New function OCSP_add1_basic_nonce()
1779 this is to allow responders to include a nonce in a response even if
1780 the request is nonce-less.
1783 *) Disable stdin buffering in load_cert (apps/apps.c) so that no certs are
1784 skipped when using openssl x509 multiple times on a single input file,
1785 e.g. "(openssl x509 -out cert1; openssl x509 -out cert2) <certs".
1788 *) Make ASN1_UTCTIME_set_string() and ASN1_GENERALIZEDTIME_set_string()
1789 set string type: to handle setting ASN1_TIME structures. Fix ca
1790 utility to correctly initialize revocation date of CRLs.
1793 *) New option SSL_OP_CIPHER_SERVER_PREFERENCE allows the server to override
1794 the clients preferred ciphersuites and rather use its own preferences.
1795 Should help to work around M$ SGC (Server Gated Cryptography) bug in
1796 Internet Explorer by ensuring unchanged hash method during stepup.
1797 (Also replaces the broken/deactivated SSL_OP_NON_EXPORT_FIRST option.)
1800 *) Make mkdef.pl recognise all DECLARE_ASN1 macros, change rijndael
1801 to aes and add a new 'exist' option to print out symbols that don't
1805 *) Additional options to ocsp utility to allow flags to be set and
1806 additional certificates supplied.
1809 *) Add the option -VAfile to 'openssl ocsp', so the user can give the
1810 OCSP client a number of certificate to only verify the response
1814 *) Update Rijndael code to version 3.0 and change EVP AES ciphers to
1815 handle the new API. Currently only ECB, CBC modes supported. Add new
1818 Add TLS AES ciphersuites as described in RFC3268, "Advanced
1819 Encryption Standard (AES) Ciphersuites for Transport Layer
1820 Security (TLS)". (In beta versions of OpenSSL 0.9.7, these were
1821 not enabled by default and were not part of the "ALL" ciphersuite
1822 alias because they were not yet official; they could be
1823 explicitly requested by specifying the "AESdraft" ciphersuite
1824 group alias. In the final release of OpenSSL 0.9.7, the group
1825 alias is called "AES" and is part of "ALL".)
1826 [Ben Laurie, Steve Henson, Bodo Moeller]
1828 *) New function OCSP_copy_nonce() to copy nonce value (if present) from
1829 request to response.
1832 *) Functions for OCSP responders. OCSP_request_onereq_count(),
1833 OCSP_request_onereq_get0(), OCSP_onereq_get0_id() and OCSP_id_get0_info()
1834 extract information from a certificate request. OCSP_response_create()
1835 creates a response and optionally adds a basic response structure.
1836 OCSP_basic_add1_status() adds a complete single response to a basic
1837 response and returns the OCSP_SINGLERESP structure just added (to allow
1838 extensions to be included for example). OCSP_basic_add1_cert() adds a
1839 certificate to a basic response and OCSP_basic_sign() signs a basic
1840 response with various flags. New helper functions ASN1_TIME_check()
1841 (checks validity of ASN1_TIME structure) and ASN1_TIME_to_generalizedtime()
1842 (converts ASN1_TIME to GeneralizedTime).
1845 *) Various new functions. EVP_Digest() combines EVP_Digest{Init,Update,Final}()
1846 in a single operation. X509_get0_pubkey_bitstr() extracts the public_key
1847 structure from a certificate. X509_pubkey_digest() digests the public_key
1848 contents: this is used in various key identifiers.
1851 *) Make sk_sort() tolerate a NULL argument.
1852 [Steve Henson reported by Massimiliano Pala <madwolf@comune.modena.it>]
1854 *) New OCSP verify flag OCSP_TRUSTOTHER. When set the "other" certificates
1855 passed by the function are trusted implicitly. If any of them signed the
1856 response then it is assumed to be valid and is not verified.
1859 *) In PKCS7_set_type() initialise content_type in PKCS7_ENC_CONTENT
1860 to data. This was previously part of the PKCS7 ASN1 code. This
1861 was causing problems with OpenSSL created PKCS#12 and PKCS#7 structures.
1862 [Steve Henson, reported by Kenneth R. Robinette
1863 <support@securenetterm.com>]
1865 *) Add CRYPTO_push_info() and CRYPTO_pop_info() calls to new ASN1
1866 routines: without these tracing memory leaks is very painful.
1867 Fix leaks in PKCS12 and PKCS7 routines.
1870 *) Make X509_time_adj() cope with the new behaviour of ASN1_TIME_new().
1871 Previously it initialised the 'type' argument to V_ASN1_UTCTIME which
1872 effectively meant GeneralizedTime would never be used. Now it
1873 is initialised to -1 but X509_time_adj() now has to check the value
1874 and use ASN1_TIME_set() if the value is not V_ASN1_UTCTIME or
1875 V_ASN1_GENERALIZEDTIME, without this it always uses GeneralizedTime.
1876 [Steve Henson, reported by Kenneth R. Robinette
1877 <support@securenetterm.com>]
1879 *) Fixes to BN_to_ASN1_INTEGER when bn is zero. This would previously
1880 result in a zero length in the ASN1_INTEGER structure which was
1881 not consistent with the structure when d2i_ASN1_INTEGER() was used
1882 and would cause ASN1_INTEGER_cmp() to fail. Enhance s2i_ASN1_INTEGER()
1883 to cope with hex and negative integers. Fix bug in i2a_ASN1_INTEGER()
1884 where it did not print out a minus for negative ASN1_INTEGER.
1887 *) Add summary printout to ocsp utility. The various functions which
1888 convert status values to strings have been renamed to:
1889 OCSP_response_status_str(), OCSP_cert_status_str() and
1890 OCSP_crl_reason_str() and are no longer static. New options
1891 to verify nonce values and to disable verification. OCSP response
1892 printout format cleaned up.
1895 *) Add additional OCSP certificate checks. These are those specified
1896 in RFC2560. This consists of two separate checks: the CA of the
1897 certificate being checked must either be the OCSP signer certificate
1898 or the issuer of the OCSP signer certificate. In the latter case the
1899 OCSP signer certificate must contain the OCSP signing extended key
1900 usage. This check is performed by attempting to match the OCSP
1901 signer or the OCSP signer CA to the issuerNameHash and issuerKeyHash
1902 in the OCSP_CERTID structures of the response.
1905 *) Initial OCSP certificate verification added to OCSP_basic_verify()
1906 and related routines. This uses the standard OpenSSL certificate
1907 verify routines to perform initial checks (just CA validity) and
1908 to obtain the certificate chain. Then additional checks will be
1909 performed on the chain. Currently the root CA is checked to see
1910 if it is explicitly trusted for OCSP signing. This is used to set
1911 a root CA as a global signing root: that is any certificate that
1912 chains to that CA is an acceptable OCSP signing certificate.
1915 *) New '-extfile ...' option to 'openssl ca' for reading X.509v3
1916 extensions from a separate configuration file.
1917 As when reading extensions from the main configuration file,
1918 the '-extensions ...' option may be used for specifying the
1920 [Massimiliano Pala <madwolf@comune.modena.it>]
1922 *) New OCSP utility. Allows OCSP requests to be generated or
1923 read. The request can be sent to a responder and the output
1924 parsed, outputed or printed in text form. Not complete yet:
1925 still needs to check the OCSP response validity.
1928 *) New subcommands for 'openssl ca':
1929 'openssl ca -status <serial>' prints the status of the cert with
1930 the given serial number (according to the index file).
1931 'openssl ca -updatedb' updates the expiry status of certificates
1933 [Massimiliano Pala <madwolf@comune.modena.it>]
1935 *) New '-newreq-nodes' command option to CA.pl. This is like
1936 '-newreq', but calls 'openssl req' with the '-nodes' option
1937 so that the resulting key is not encrypted.
1938 [Damien Miller <djm@mindrot.org>]
1940 *) New configuration for the GNU Hurd.
1941 [Jonathan Bartlett <johnnyb@wolfram.com> via Richard Levitte]
1943 *) Initial code to implement OCSP basic response verify. This
1944 is currently incomplete. Currently just finds the signer's
1945 certificate and verifies the signature on the response.
1948 *) New SSLeay_version code SSLEAY_DIR to determine the compiled-in
1949 value of OPENSSLDIR. This is available via the new '-d' option
1950 to 'openssl version', and is also included in 'openssl version -a'.
1953 *) Allowing defining memory allocation callbacks that will be given
1954 file name and line number information in additional arguments
1955 (a const char* and an int). The basic functionality remains, as
1956 well as the original possibility to just replace malloc(),
1957 realloc() and free() by functions that do not know about these
1958 additional arguments. To register and find out the current
1959 settings for extended allocation functions, the following
1960 functions are provided:
1962 CRYPTO_set_mem_ex_functions
1963 CRYPTO_set_locked_mem_ex_functions
1964 CRYPTO_get_mem_ex_functions
1965 CRYPTO_get_locked_mem_ex_functions
1967 These work the same way as CRYPTO_set_mem_functions and friends.
1968 CRYPTO_get_[locked_]mem_functions now writes 0 where such an
1969 extended allocation function is enabled.
1970 Similarly, CRYPTO_get_[locked_]mem_ex_functions writes 0 where
1971 a conventional allocation function is enabled.
1972 [Richard Levitte, Bodo Moeller]
1974 *) Finish off removing the remaining LHASH function pointer casts.
1975 There should no longer be any prototype-casting required when using
1976 the LHASH abstraction, and any casts that remain are "bugs". See
1977 the callback types and macros at the head of lhash.h for details
1978 (and "OBJ_cleanup" in crypto/objects/obj_dat.c as an example).
1981 *) Add automatic query of EGD sockets in RAND_poll() for the unix variant.
1982 If /dev/[u]random devices are not available or do not return enough
1983 entropy, EGD style sockets (served by EGD or PRNGD) will automatically
1985 The locations /var/run/egd-pool, /dev/egd-pool, /etc/egd-pool, and
1986 /etc/entropy will be queried once each in this sequence, quering stops
1987 when enough entropy was collected without querying more sockets.
1990 *) Change the Unix RAND_poll() variant to be able to poll several
1991 random devices, as specified by DEVRANDOM, until a sufficient amount
1992 of data has been collected. We spend at most 10 ms on each file
1993 (select timeout) and read in non-blocking mode. DEVRANDOM now
1994 defaults to the list "/dev/urandom", "/dev/random", "/dev/srandom"
1995 (previously it was just the string "/dev/urandom"), so on typical
1996 platforms the 10 ms delay will never occur.
1997 Also separate out the Unix variant to its own file, rand_unix.c.
1998 For VMS, there's a currently-empty rand_vms.c.
2001 *) Move OCSP client related routines to ocsp_cl.c. These
2002 provide utility functions which an application needing
2003 to issue a request to an OCSP responder and analyse the
2004 response will typically need: as opposed to those which an
2005 OCSP responder itself would need which will be added later.
2007 OCSP_request_sign() signs an OCSP request with an API similar
2008 to PKCS7_sign(). OCSP_response_status() returns status of OCSP
2009 response. OCSP_response_get1_basic() extracts basic response
2010 from response. OCSP_resp_find_status(): finds and extracts status
2011 information from an OCSP_CERTID structure (which will be created
2012 when the request structure is built). These are built from lower
2013 level functions which work on OCSP_SINGLERESP structures but
2014 wont normally be used unless the application wishes to examine
2015 extensions in the OCSP response for example.
2017 Replace nonce routines with a pair of functions.
2018 OCSP_request_add1_nonce() adds a nonce value and optionally
2019 generates a random value. OCSP_check_nonce() checks the
2020 validity of the nonce in an OCSP response.
2023 *) Change function OCSP_request_add() to OCSP_request_add0_id().
2024 This doesn't copy the supplied OCSP_CERTID and avoids the
2025 need to free up the newly created id. Change return type
2026 to OCSP_ONEREQ to return the internal OCSP_ONEREQ structure.
2027 This can then be used to add extensions to the request.
2028 Deleted OCSP_request_new(), since most of its functionality
2029 is now in OCSP_REQUEST_new() (and the case insensitive name
2030 clash) apart from the ability to set the request name which
2031 will be added elsewhere.
2034 *) Update OCSP API. Remove obsolete extensions argument from
2035 various functions. Extensions are now handled using the new
2036 OCSP extension code. New simple OCSP HTTP function which
2037 can be used to send requests and parse the response.
2040 *) Fix the PKCS#7 (S/MIME) code to work with new ASN1. Two new
2041 ASN1_ITEM structures help with sign and verify. PKCS7_ATTR_SIGN
2042 uses the special reorder version of SET OF to sort the attributes
2043 and reorder them to match the encoded order. This resolves a long
2044 standing problem: a verify on a PKCS7 structure just after signing
2045 it used to fail because the attribute order did not match the
2046 encoded order. PKCS7_ATTR_VERIFY does not reorder the attributes:
2047 it uses the received order. This is necessary to tolerate some broken
2048 software that does not order SET OF. This is handled by encoding
2049 as a SEQUENCE OF but using implicit tagging (with UNIVERSAL class)
2050 to produce the required SET OF.
2053 *) Have mk1mf.pl generate the macros OPENSSL_BUILD_SHLIBCRYPTO and
2054 OPENSSL_BUILD_SHLIBSSL and use them appropriately in the header
2055 files to get correct declarations of the ASN.1 item variables.
2058 *) Rewrite of PKCS#12 code to use new ASN1 functionality. Replace many
2059 PKCS#12 macros with real functions. Fix two unrelated ASN1 bugs:
2060 asn1_check_tlen() would sometimes attempt to use 'ctx' when it was
2061 NULL and ASN1_TYPE was not dereferenced properly in asn1_ex_c2i().
2062 New ASN1 macro: DECLARE_ASN1_ITEM() which just declares the relevant
2063 ASN1_ITEM and no wrapper functions.
2066 *) New functions or ASN1_item_d2i_fp() and ASN1_item_d2i_bio(). These
2067 replace the old function pointer based I/O routines. Change most of
2068 the *_d2i_bio() and *_d2i_fp() functions to use these.
2071 *) Enhance mkdef.pl to be more accepting about spacing in C preprocessor
2072 lines, recognice more "algorithms" that can be deselected, and make
2073 it complain about algorithm deselection that isn't recognised.
2076 *) New ASN1 functions to handle dup, sign, verify, digest, pack and
2077 unpack operations in terms of ASN1_ITEM. Modify existing wrappers
2078 to use new functions. Add NO_ASN1_OLD which can be set to remove
2079 some old style ASN1 functions: this can be used to determine if old
2080 code will still work when these eventually go away.
2083 *) New extension functions for OCSP structures, these follow the
2084 same conventions as certificates and CRLs.
2087 *) New function X509V3_add1_i2d(). This automatically encodes and
2088 adds an extension. Its behaviour can be customised with various
2089 flags to append, replace or delete. Various wrappers added for
2090 certifcates and CRLs.
2093 *) Fix to avoid calling the underlying ASN1 print routine when
2094 an extension cannot be parsed. Correct a typo in the
2095 OCSP_SERVICELOC extension. Tidy up print OCSP format.
2098 *) Make mkdef.pl parse some of the ASN1 macros and add apropriate
2099 entries for variables.
2102 *) Add functionality to apps/openssl.c for detecting locking
2103 problems: As the program is single-threaded, all we have
2104 to do is register a locking callback using an array for
2105 storing which locks are currently held by the program.
2108 *) Use a lock around the call to CRYPTO_get_ex_new_index() in
2109 SSL_get_ex_data_X509_STORE_idx(), which is used in
2110 ssl_verify_cert_chain() and thus can be called at any time
2111 during TLS/SSL handshakes so that thread-safety is essential.
2112 Unfortunately, the ex_data design is not at all suited
2113 for multi-threaded use, so it probably should be abolished.
2116 *) Added Broadcom "ubsec" ENGINE to OpenSSL.
2117 [Broadcom, tweaked and integrated by Geoff Thorpe]
2119 *) Move common extension printing code to new function
2120 X509V3_print_extensions(). Reorganise OCSP print routines and
2121 implement some needed OCSP ASN1 functions. Add OCSP extensions.
2124 *) New function X509_signature_print() to remove duplication in some
2128 *) Add a special meaning when SET OF and SEQUENCE OF flags are both
2129 set (this was treated exactly the same as SET OF previously). This
2130 is used to reorder the STACK representing the structure to match the
2131 encoding. This will be used to get round a problem where a PKCS7
2132 structure which was signed could not be verified because the STACK
2133 order did not reflect the encoded order.
2136 *) Reimplement the OCSP ASN1 module using the new code.
2139 *) Update the X509V3 code to permit the use of an ASN1_ITEM structure
2140 for its ASN1 operations. The old style function pointers still exist
2141 for now but they will eventually go away.
2144 *) Merge in replacement ASN1 code from the ASN1 branch. This almost
2145 completely replaces the old ASN1 functionality with a table driven
2146 encoder and decoder which interprets an ASN1_ITEM structure describing
2147 the ASN1 module. Compatibility with the existing ASN1 API (i2d,d2i) is
2148 largely maintained. Almost all of the old asn1_mac.h macro based ASN1
2149 has also been converted to the new form.
2152 *) Change BN_mod_exp_recp so that negative moduli are tolerated
2153 (the sign is ignored). Similarly, ignore the sign in BN_MONT_CTX_set
2154 so that BN_mod_exp_mont and BN_mod_exp_mont_word work
2155 for negative moduli.
2158 *) Fix BN_uadd and BN_usub: Always return non-negative results instead
2159 of not touching the result's sign bit.
2162 *) BN_div bugfix: If the result is 0, the sign (res->neg) must not be
2166 *) Changed the LHASH code to use prototypes for callbacks, and created
2167 macros to declare and implement thin (optionally static) functions
2168 that provide type-safety and avoid function pointer casting for the
2169 type-specific callbacks.
2172 *) Added Kerberos Cipher Suites to be used with TLS, as written in
2174 [Veers Staats <staatsvr@asc.hpc.mil>,
2175 Jeffrey Altman <jaltman@columbia.edu>, via Richard Levitte]
2177 *) Reformat the FAQ so the different questions and answers can be divided
2178 in sections depending on the subject.
2181 *) Have the zlib compression code load ZLIB.DLL dynamically under
2185 *) New function BN_mod_sqrt for computing square roots modulo a prime
2186 (using the probabilistic Tonelli-Shanks algorithm unless
2187 p == 3 (mod 4) or p == 5 (mod 8), which are cases that can
2188 be handled deterministically).
2189 [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller]
2191 *) Make BN_mod_inverse faster by explicitly handling small quotients
2192 in the Euclid loop. (Speed gain about 20% for small moduli [256 or
2193 512 bits], about 30% for larger ones [1024 or 2048 bits].)
2196 *) New function BN_kronecker.
2199 *) Fix BN_gcd so that it works on negative inputs; the result is
2200 positive unless both parameters are zero.
2201 Previously something reasonably close to an infinite loop was
2202 possible because numbers could be growing instead of shrinking
2203 in the implementation of Euclid's algorithm.
2206 *) Fix BN_is_word() and BN_is_one() macros to take into account the
2207 sign of the number in question.
2209 Fix BN_is_word(a,w) to work correctly for w == 0.
2211 The old BN_is_word(a,w) macro is now called BN_abs_is_word(a,w)
2212 because its test if the absolute value of 'a' equals 'w'.
2213 Note that BN_abs_is_word does *not* handle w == 0 reliably;
2214 it exists mostly for use in the implementations of BN_is_zero(),
2215 BN_is_one(), and BN_is_word().
2218 *) New function BN_swap.
2221 *) Use BN_nnmod instead of BN_mod in crypto/bn/bn_exp.c so that
2222 the exponentiation functions are more likely to produce reasonable
2223 results on negative inputs.
2226 *) Change BN_mod_mul so that the result is always non-negative.
2227 Previously, it could be negative if one of the factors was negative;
2228 I don't think anyone really wanted that behaviour.
2231 *) Move BN_mod_... functions into new file crypto/bn/bn_mod.c
2232 (except for exponentiation, which stays in crypto/bn/bn_exp.c,
2233 and BN_mod_mul_reciprocal, which stays in crypto/bn/bn_recp.c)
2234 and add new functions:
2243 BN_mod_lshift1_quick
2247 These functions always generate non-negative results.
2249 BN_nnmod otherwise is like BN_mod (if BN_mod computes a remainder r
2250 such that |m| < r < 0, BN_nnmod will output rem + |m| instead).
2252 BN_mod_XXX_quick(r, a, [b,] m) generates the same result as
2253 BN_mod_XXX(r, a, [b,] m, ctx), but requires that a [and b]
2254 be reduced modulo m.
2255 [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller]
2258 The following entry accidentily appeared in the CHANGES file
2259 distributed with OpenSSL 0.9.7. The modifications described in
2260 it do *not* apply to OpenSSL 0.9.7.
2262 *) Remove a few calls to bn_wexpand() in BN_sqr() (the one in there
2263 was actually never needed) and in BN_mul(). The removal in BN_mul()
2264 required a small change in bn_mul_part_recursive() and the addition
2265 of the functions bn_cmp_part_words(), bn_sub_part_words() and
2266 bn_add_part_words(), which do the same thing as bn_cmp_words(),
2267 bn_sub_words() and bn_add_words() except they take arrays with
2272 *) In 'openssl passwd', verify passwords read from the terminal
2273 unless the '-salt' option is used (which usually means that
2274 verification would just waste user's time since the resulting
2275 hash is going to be compared with some given password hash)
2276 or the new '-noverify' option is used.
2278 This is an incompatible change, but it does not affect
2279 non-interactive use of 'openssl passwd' (passwords on the command
2280 line, '-stdin' option, '-in ...' option) and thus should not
2284 *) Remove all references to RSAref, since there's no more need for it.
2287 *) Make DSO load along a path given through an environment variable
2288 (SHLIB_PATH) with shl_load().
2291 *) Constify the ENGINE code as a result of BIGNUM constification.
2292 Also constify the RSA code and most things related to it. In a
2293 few places, most notable in the depth of the ASN.1 code, ugly
2294 casts back to non-const were required (to be solved at a later
2298 *) Make it so the openssl application has all engines loaded by default.
2301 *) Constify the BIGNUM routines a little more.
2304 *) Add the following functions:
2306 ENGINE_load_cswift()
2308 ENGINE_load_atalla()
2310 ENGINE_load_builtin_engines()
2312 That way, an application can itself choose if external engines that
2313 are built-in in OpenSSL shall ever be used or not. The benefit is
2314 that applications won't have to be linked with libdl or other dso
2315 libraries unless it's really needed.
2317 Changed 'openssl engine' to load all engines on demand.
2318 Changed the engine header files to avoid the duplication of some
2319 declarations (they differed!).
2322 *) 'openssl engine' can now list capabilities.
2325 *) Better error reporting in 'openssl engine'.
2328 *) Never call load_dh_param(NULL) in s_server.
2331 *) Add engine application. It can currently list engines by name and
2332 identity, and test if they are actually available.
2335 *) Improve RPM specification file by forcing symbolic linking and making
2336 sure the installed documentation is also owned by root.root.
2337 [Damien Miller <djm@mindrot.org>]
2339 *) Give the OpenSSL applications more possibilities to make use of
2340 keys (public as well as private) handled by engines.
2343 *) Add OCSP code that comes from CertCo.
2346 *) Add VMS support for the Rijndael code.
2349 *) Added untested support for Nuron crypto accelerator.
2352 *) Add support for external cryptographic devices. This code was
2353 previously distributed separately as the "engine" branch.
2354 [Geoff Thorpe, Richard Levitte]
2356 *) Rework the filename-translation in the DSO code. It is now possible to
2357 have far greater control over how a "name" is turned into a filename
2358 depending on the operating environment and any oddities about the
2359 different shared library filenames on each system.
2362 *) Support threads on FreeBSD-elf in Configure.
2365 *) Fix for SHA1 assembly problem with MASM: it produces
2366 warnings about corrupt line number information when assembling
2367 with debugging information. This is caused by the overlapping
2369 [Bernd Matthes <mainbug@celocom.de>, Steve Henson]
2372 NCONF_get_number() has no error checking at all. As a replacement,
2373 NCONF_get_number_e() is defined (_e for "error checking") and is
2374 promoted strongly. The old NCONF_get_number is kept around for
2375 binary backward compatibility.
2376 Make it possible for methods to load from something other than a BIO,
2377 by providing a function pointer that is given a name instead of a BIO.
2378 For example, this could be used to load configuration data from an
2382 *) Fix for non blocking accept BIOs. Added new I/O special reason
2383 BIO_RR_ACCEPT to cover this case. Previously use of accept BIOs
2384 with non blocking I/O was not possible because no retry code was
2385 implemented. Also added new SSL code SSL_WANT_ACCEPT to cover
2389 *) Added the beginnings of Rijndael support.
2392 *) Fix for bug in DirectoryString mask setting. Add support for
2393 X509_NAME_print_ex() in 'req' and X509_print_ex() function
2394 to allow certificate printing to more controllable, additional
2395 'certopt' option to 'x509' to allow new printing options to be
2399 *) Clean old EAY MD5 hack from e_os.h.
2402 Changes between 0.9.6h and 0.9.6i [19 Feb 2003]
2404 *) In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked
2405 via timing by performing a MAC computation even if incorrrect
2406 block cipher padding has been found. This is a countermeasure
2407 against active attacks where the attacker has to distinguish
2408 between bad padding and a MAC verification error. (CAN-2003-0078)
2410 [Bodo Moeller; problem pointed out by Brice Canvel (EPFL),
2411 Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and
2412 Martin Vuagnoux (EPFL, Ilion)]
2414 Changes between 0.9.6g and 0.9.6h [5 Dec 2002]
2416 *) New function OPENSSL_cleanse(), which is used to cleanse a section of
2417 memory from it's contents. This is done with a counter that will
2418 place alternating values in each byte. This can be used to solve
2419 two issues: 1) the removal of calls to memset() by highly optimizing
2420 compilers, and 2) cleansing with other values than 0, since those can
2421 be read through on certain media, for example a swap space on disk.
2424 *) Bugfix: client side session caching did not work with external caching,
2425 because the session->cipher setting was not restored when reloading
2426 from the external cache. This problem was masked, when
2427 SSL_OP_NETSCAPE_REUSE_CIPHER_CHANGE_BUG (part of SSL_OP_ALL) was set.
2428 (Found by Steve Haslam <steve@araqnid.ddts.net>.)
2431 *) Fix client_certificate (ssl/s2_clnt.c): The permissible total
2432 length of the REQUEST-CERTIFICATE message is 18 .. 34, not 17 .. 33.
2433 [Zeev Lieber <zeev-l@yahoo.com>]
2435 *) Undo an undocumented change introduced in 0.9.6e which caused
2436 repeated calls to OpenSSL_add_all_ciphers() and
2437 OpenSSL_add_all_digests() to be ignored, even after calling
2441 *) Change the default configuration reader to deal with last line not
2442 being properly terminated.
2445 *) Change X509_NAME_cmp() so it applies the special rules on handling
2446 DN values that are of type PrintableString, as well as RDNs of type
2447 emailAddress where the value has the type ia5String.
2448 [stefank@valicert.com via Richard Levitte]
2450 *) Add a SSL_SESS_CACHE_NO_INTERNAL_STORE flag to take over half
2451 the job SSL_SESS_CACHE_NO_INTERNAL_LOOKUP was inconsistently
2452 doing, define a new flag (SSL_SESS_CACHE_NO_INTERNAL) to be
2453 the bitwise-OR of the two for use by the majority of applications
2454 wanting this behaviour, and update the docs. The documented
2455 behaviour and actual behaviour were inconsistent and had been
2456 changing anyway, so this is more a bug-fix than a behavioural
2458 [Geoff Thorpe, diagnosed by Nadav Har'El]
2460 *) Don't impose a 16-byte length minimum on session IDs in ssl/s3_clnt.c
2461 (the SSL 3.0 and TLS 1.0 specifications allow any length up to 32 bytes).
2464 *) Fix initialization code race conditions in
2465 SSLv23_method(), SSLv23_client_method(), SSLv23_server_method(),
2466 SSLv2_method(), SSLv2_client_method(), SSLv2_server_method(),
2467 SSLv3_method(), SSLv3_client_method(), SSLv3_server_method(),
2468 TLSv1_method(), TLSv1_client_method(), TLSv1_server_method(),
2469 ssl2_get_cipher_by_char(),
2470 ssl3_get_cipher_by_char().
2471 [Patrick McCormick <patrick@tellme.com>, Bodo Moeller]
2473 *) Reorder cleanup sequence in SSL_CTX_free(): only remove the ex_data after
2474 the cached sessions are flushed, as the remove_cb() might use ex_data
2475 contents. Bug found by Sam Varshavchik <mrsam@courier-mta.com>
2476 (see [openssl.org #212]).
2477 [Geoff Thorpe, Lutz Jaenicke]
2479 *) Fix typo in OBJ_txt2obj which incorrectly passed the content
2480 length, instead of the encoding length to d2i_ASN1_OBJECT.
2483 Changes between 0.9.6f and 0.9.6g [9 Aug 2002]
2485 *) [In 0.9.6g-engine release:]
2486 Fix crypto/engine/vendor_defns/cswift.h for WIN32 (use '_stdcall').
2487 [Lynn Gazis <lgazis@rainbow.com>]
2489 Changes between 0.9.6e and 0.9.6f [8 Aug 2002]
2491 *) Fix ASN1 checks. Check for overflow by comparing with LONG_MAX
2492 and get fix the header length calculation.
2493 [Florian Weimer <Weimer@CERT.Uni-Stuttgart.DE>,
2494 Alon Kantor <alonk@checkpoint.com> (and others),
2497 *) Use proper error handling instead of 'assertions' in buffer
2498 overflow checks added in 0.9.6e. This prevents DoS (the
2499 assertions could call abort()).
2500 [Arne Ansper <arne@ats.cyber.ee>, Bodo Moeller]
2502 Changes between 0.9.6d and 0.9.6e [30 Jul 2002]
2504 *) Add various sanity checks to asn1_get_length() to reject
2505 the ASN1 length bytes if they exceed sizeof(long), will appear
2506 negative or the content length exceeds the length of the
2508 [Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>]
2510 *) Fix cipher selection routines: ciphers without encryption had no flags
2511 for the cipher strength set and where therefore not handled correctly
2512 by the selection routines (PR #130).
2515 *) Fix EVP_dsa_sha macro.
2519 SSL_OP_DONT_INSERT_EMPTY_FRAGMENTS
2520 for disabling the SSL 3.0/TLS 1.0 CBC vulnerability countermeasure
2521 that was added in OpenSSL 0.9.6d.
2523 As the countermeasure turned out to be incompatible with some
2524 broken SSL implementations, the new option is part of SSL_OP_ALL.
2525 SSL_OP_ALL is usually employed when compatibility with weird SSL
2526 implementations is desired (e.g. '-bugs' option to 's_client' and
2527 's_server'), so the new option is automatically set in many
2531 *) Changes in security patch:
2533 Changes marked "(CHATS)" were sponsored by the Defense Advanced
2534 Research Projects Agency (DARPA) and Air Force Research Laboratory,
2535 Air Force Materiel Command, USAF, under agreement number
2538 *) Add various sanity checks to asn1_get_length() to reject
2539 the ASN1 length bytes if they exceed sizeof(long), will appear
2540 negative or the content length exceeds the length of the
2541 supplied buffer. (CAN-2002-0659)
2542 [Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>]
2544 *) Assertions for various potential buffer overflows, not known to
2546 [Ben Laurie (CHATS)]
2548 *) Various temporary buffers to hold ASCII versions of integers were
2549 too small for 64 bit platforms. (CAN-2002-0655)
2550 [Matthew Byng-Maddick <mbm@aldigital.co.uk> and Ben Laurie (CHATS)>
2552 *) Remote buffer overflow in SSL3 protocol - an attacker could
2553 supply an oversized session ID to a client. (CAN-2002-0656)
2554 [Ben Laurie (CHATS)]
2556 *) Remote buffer overflow in SSL2 protocol - an attacker could
2557 supply an oversized client master key. (CAN-2002-0656)
2558 [Ben Laurie (CHATS)]
2560 Changes between 0.9.6c and 0.9.6d [9 May 2002]
2562 *) Fix crypto/asn1/a_sign.c so that 'parameters' is omitted (not
2563 encoded as NULL) with id-dsa-with-sha1.
2564 [Nils Larsch <nla@trustcenter.de>; problem pointed out by Bodo Moeller]
2566 *) Check various X509_...() return values in apps/req.c.
2567 [Nils Larsch <nla@trustcenter.de>]
2569 *) Fix BASE64 decode (EVP_DecodeUpdate) for data with CR/LF ended lines:
2570 an end-of-file condition would erronously be flagged, when the CRLF
2571 was just at the end of a processed block. The bug was discovered when
2572 processing data through a buffering memory BIO handing the data to a
2573 BASE64-decoding BIO. Bug fund and patch submitted by Pavel Tsekov
2574 <ptsekov@syntrex.com> and Nedelcho Stanev.
2577 *) Implement a countermeasure against a vulnerability recently found
2578 in CBC ciphersuites in SSL 3.0/TLS 1.0: Send an empty fragment
2579 before application data chunks to avoid the use of known IVs
2580 with data potentially chosen by the attacker.
2583 *) Fix length checks in ssl3_get_client_hello().
2586 *) TLS/SSL library bugfix: use s->s3->in_read_app_data differently
2587 to prevent ssl3_read_internal() from incorrectly assuming that
2588 ssl3_read_bytes() found application data while handshake
2589 processing was enabled when in fact s->s3->in_read_app_data was
2590 merely automatically cleared during the initial handshake.
2591 [Bodo Moeller; problem pointed out by Arne Ansper <arne@ats.cyber.ee>]
2593 *) Fix object definitions for Private and Enterprise: they were not
2594 recognized in their shortname (=lowercase) representation. Extend
2595 obj_dat.pl to issue an error when using undefined keywords instead
2596 of silently ignoring the problem (Svenning Sorensen
2597 <sss@sss.dnsalias.net>).
2600 *) Fix DH_generate_parameters() so that it works for 'non-standard'
2601 generators, i.e. generators other than 2 and 5. (Previously, the
2602 code did not properly initialise the 'add' and 'rem' values to
2603 BN_generate_prime().)
2605 In the new general case, we do not insist that 'generator' is
2606 actually a primitive root: This requirement is rather pointless;
2607 a generator of the order-q subgroup is just as good, if not
2611 *) Map new X509 verification errors to alerts. Discovered and submitted by
2612 Tom Wu <tom@arcot.com>.
2615 *) Fix ssl3_pending() (ssl/s3_lib.c) to prevent SSL_pending() from
2616 returning non-zero before the data has been completely received
2617 when using non-blocking I/O.
2618 [Bodo Moeller; problem pointed out by John Hughes]
2620 *) Some of the ciphers missed the strength entry (SSL_LOW etc).
2621 [Ben Laurie, Lutz Jaenicke]
2623 *) Fix bug in SSL_clear(): bad sessions were not removed (found by
2624 Yoram Zahavi <YoramZ@gilian.com>).
2627 *) Add information about CygWin 1.3 and on, and preserve proper
2628 configuration for the versions before that.
2629 [Corinna Vinschen <vinschen@redhat.com> and Richard Levitte]
2631 *) Make removal from session cache (SSL_CTX_remove_session()) more robust:
2632 check whether we deal with a copy of a session and do not delete from
2633 the cache in this case. Problem reported by "Izhar Shoshani Levi"
2634 <izhar@checkpoint.com>.
2637 *) Do not store session data into the internal session cache, if it
2638 is never intended to be looked up (SSL_SESS_CACHE_NO_INTERNAL_LOOKUP
2639 flag is set). Proposed by Aslam <aslam@funk.com>.
2642 *) Have ASN1_BIT_STRING_set_bit() really clear a bit when the requested
2646 *) [In 0.9.6d-engine release:]
2647 Fix a crashbug and a logic bug in hwcrhk_load_pubkey().
2648 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte]
2650 *) Add the configuration target linux-s390x.
2651 [Neale Ferguson <Neale.Ferguson@SoftwareAG-USA.com> via Richard Levitte]
2653 *) The earlier bugfix for the SSL3_ST_SW_HELLO_REQ_C case of
2654 ssl3_accept (ssl/s3_srvr.c) incorrectly used a local flag
2655 variable as an indication that a ClientHello message has been
2656 received. As the flag value will be lost between multiple
2657 invocations of ssl3_accept when using non-blocking I/O, the
2658 function may not be aware that a handshake has actually taken
2659 place, thus preventing a new session from being added to the
2662 To avoid this problem, we now set s->new_session to 2 instead of
2663 using a local variable.
2664 [Lutz Jaenicke, Bodo Moeller]
2666 *) Bugfix: Return -1 from ssl3_get_server_done (ssl3/s3_clnt.c)
2667 if the SSL_R_LENGTH_MISMATCH error is detected.
2668 [Geoff Thorpe, Bodo Moeller]
2670 *) New 'shared_ldflag' column in Configure platform table.
2673 *) Fix EVP_CIPHER_mode macro.
2674 ["Dan S. Camper" <dan@bti.net>]
2676 *) Fix ssl3_read_bytes (ssl/s3_pkt.c): To ignore messages of unknown
2677 type, we must throw them away by setting rr->length to 0.
2678 [D P Chang <dpc@qualys.com>]
2680 Changes between 0.9.6b and 0.9.6c [21 dec 2001]
2682 *) Fix BN_rand_range bug pointed out by Dominikus Scherkl
2683 <Dominikus.Scherkl@biodata.com>. (The previous implementation
2684 worked incorrectly for those cases where range = 10..._2 and
2685 3*range is two bits longer than range.)
2688 *) Only add signing time to PKCS7 structures if it is not already
2692 *) Fix crypto/objects/objects.h: "ld-ce" should be "id-ce",
2693 OBJ_ld_ce should be OBJ_id_ce.
2694 Also some ip-pda OIDs in crypto/objects/objects.txt were
2695 incorrect (cf. RFC 3039).
2696 [Matt Cooper, Frederic Giudicelli, Bodo Moeller]
2698 *) Release CRYPTO_LOCK_DYNLOCK when CRYPTO_destroy_dynlockid()
2699 returns early because it has nothing to do.
2700 [Andy Schneider <andy.schneider@bjss.co.uk>]
2702 *) [In 0.9.6c-engine release:]
2703 Fix mutex callback return values in crypto/engine/hw_ncipher.c.
2704 [Andy Schneider <andy.schneider@bjss.co.uk>]
2706 *) [In 0.9.6c-engine release:]
2707 Add support for Cryptographic Appliance's keyserver technology.
2708 (Use engine 'keyclient')
2709 [Cryptographic Appliances and Geoff Thorpe]
2711 *) Add a configuration entry for OS/390 Unix. The C compiler 'c89'
2712 is called via tools/c89.sh because arguments have to be
2713 rearranged (all '-L' options must appear before the first object
2715 [Richard Shapiro <rshapiro@abinitio.com>]
2717 *) [In 0.9.6c-engine release:]
2718 Add support for Broadcom crypto accelerator cards, backported
2720 [Broadcom, Nalin Dahyabhai <nalin@redhat.com>, Mark Cox]
2722 *) [In 0.9.6c-engine release:]
2723 Add support for SureWare crypto accelerator cards from
2724 Baltimore Technologies. (Use engine 'sureware')
2725 [Baltimore Technologies and Mark Cox]
2727 *) [In 0.9.6c-engine release:]
2728 Add support for crypto accelerator cards from Accelerated
2729 Encryption Processing, www.aep.ie. (Use engine 'aep')
2730 [AEP Inc. and Mark Cox]
2732 *) Add a configuration entry for gcc on UnixWare.
2733 [Gary Benson <gbenson@redhat.com>]
2735 *) Change ssl/s2_clnt.c and ssl/s2_srvr.c so that received handshake
2736 messages are stored in a single piece (fixed-length part and
2737 variable-length part combined) and fix various bugs found on the way.
2740 *) Disable caching in BIO_gethostbyname(), directly use gethostbyname()
2741 instead. BIO_gethostbyname() does not know what timeouts are
2742 appropriate, so entries would stay in cache even when they have
2744 [Bodo Moeller; problem pointed out by Rich Salz <rsalz@zolera.com>
2746 *) Change ssl23_get_client_hello (ssl/s23_srvr.c) behaviour when
2747 faced with a pathologically small ClientHello fragment that does
2748 not contain client_version: Instead of aborting with an error,
2749 simply choose the highest available protocol version (i.e.,
2750 TLS 1.0 unless it is disabled). In practice, ClientHello
2751 messages are never sent like this, but this change gives us
2752 strictly correct behaviour at least for TLS.
2755 *) Fix SSL handshake functions and SSL_clear() such that SSL_clear()
2756 never resets s->method to s->ctx->method when called from within
2757 one of the SSL handshake functions.
2758 [Bodo Moeller; problem pointed out by Niko Baric]
2760 *) In ssl3_get_client_hello (ssl/s3_srvr.c), generate a fatal alert
2761 (sent using the client's version number) if client_version is
2762 smaller than the protocol version in use. Also change
2763 ssl23_get_client_hello (ssl/s23_srvr.c) to select TLS 1.0 if
2764 the client demanded SSL 3.0 but only TLS 1.0 is enabled; then
2765 the client will at least see that alert.
2768 *) Fix ssl3_get_message (ssl/s3_both.c) to handle message fragmentation
2772 *) Avoid infinite loop in ssl3_get_message (ssl/s3_both.c) if a
2773 client receives HelloRequest while in a handshake.
2774 [Bodo Moeller; bug noticed by Andy Schneider <andy.schneider@bjss.co.uk>]
2776 *) Bugfix in ssl3_accept (ssl/s3_srvr.c): Case SSL3_ST_SW_HELLO_REQ_C
2777 should end in 'break', not 'goto end' which circuments various
2778 cleanups done in state SSL_ST_OK. But session related stuff
2779 must be disabled for SSL_ST_OK in the case that we just sent a
2782 Also avoid some overhead by not calling ssl_init_wbio_buffer()
2783 before just sending a HelloRequest.
2784 [Bodo Moeller, Eric Rescorla <ekr@rtfm.com>]
2786 *) Fix ssl/s3_enc.c, ssl/t1_enc.c and ssl/s3_pkt.c so that we don't
2787 reveal whether illegal block cipher padding was found or a MAC
2788 verification error occured. (Neither SSLerr() codes nor alerts
2789 are directly visible to potential attackers, but the information
2790 may leak via logfiles.)
2792 Similar changes are not required for the SSL 2.0 implementation
2793 because the number of padding bytes is sent in clear for SSL 2.0,
2794 and the extra bytes are just ignored. However ssl/s2_pkt.c
2795 failed to verify that the purported number of padding bytes is in
2799 *) Add OpenUNIX-8 support including shared libraries
2800 (Boyd Lynn Gerber <gerberb@zenez.com>).
2803 *) Improve RSA_padding_check_PKCS1_OAEP() check again to avoid
2804 'wristwatch attack' using huge encoding parameters (cf.
2805 James H. Manger's CRYPTO 2001 paper). Note that the
2806 RSA_PKCS1_OAEP_PADDING case of RSA_private_decrypt() does not use
2807 encoding parameters and hence was not vulnerable.
2810 *) BN_sqr() bug fix.
2811 [Ulf Möller, reported by Jim Ellis <jim.ellis@cavium.com>]
2813 *) Rabin-Miller test analyses assume uniformly distributed witnesses,
2814 so use BN_pseudo_rand_range() instead of using BN_pseudo_rand()
2815 followed by modular reduction.
2816 [Bodo Moeller; pointed out by Adam Young <AYoung1@NCSUS.JNJ.COM>]
2818 *) Add BN_pseudo_rand_range() with obvious functionality: BN_rand_range()
2819 equivalent based on BN_pseudo_rand() instead of BN_rand().
2822 *) s3_srvr.c: allow sending of large client certificate lists (> 16 kB).
2823 This function was broken, as the check for a new client hello message
2824 to handle SGC did not allow these large messages.
2825 (Tracked down by "Douglas E. Engert" <deengert@anl.gov>.)
2828 *) Add alert descriptions for TLSv1 to SSL_alert_desc_string[_long]().
2831 *) Fix buggy behaviour of BIO_get_num_renegotiates() and BIO_ctrl()
2832 for BIO_C_GET_WRITE_BUF_SIZE ("Stephen Hinton" <shinton@netopia.com>).
2835 *) Rework the configuration and shared library support for Tru64 Unix.
2836 The configuration part makes use of modern compiler features and
2837 still retains old compiler behavior for those that run older versions
2838 of the OS. The shared library support part includes a variant that
2839 uses the RPATH feature, and is available through the special
2840 configuration target "alpha-cc-rpath", which will never be selected
2842 [Tim Mooney <mooney@dogbert.cc.ndsu.NoDak.edu> via Richard Levitte]
2844 *) In ssl3_get_key_exchange (ssl/s3_clnt.c), call ssl3_get_message()
2845 with the same message size as in ssl3_get_certificate_request().
2846 Otherwise, if no ServerKeyExchange message occurs, CertificateRequest
2847 messages might inadvertently be reject as too long.
2848 [Petr Lampa <lampa@fee.vutbr.cz>]
2850 *) Enhanced support for IA-64 Unix platforms (well, Linux and HP-UX).
2853 *) Modified SSL library such that the verify_callback that has been set
2854 specificly for an SSL object with SSL_set_verify() is actually being
2855 used. Before the change, a verify_callback set with this function was
2856 ignored and the verify_callback() set in the SSL_CTX at the time of
2857 the call was used. New function X509_STORE_CTX_set_verify_cb() introduced
2858 to allow the necessary settings.
2861 *) Initialize static variable in crypto/dsa/dsa_lib.c and crypto/dh/dh_lib.c
2862 explicitly to NULL, as at least on Solaris 8 this seems not always to be
2863 done automatically (in contradiction to the requirements of the C
2864 standard). This made problems when used from OpenSSH.
2867 *) In OpenSSL 0.9.6a and 0.9.6b, crypto/dh/dh_key.c ignored
2868 dh->length and always used
2870 BN_rand_range(priv_key, dh->p).
2872 BN_rand_range() is not necessary for Diffie-Hellman, and this
2873 specific range makes Diffie-Hellman unnecessarily inefficient if
2874 dh->length (recommended exponent length) is much smaller than the
2875 length of dh->p. We could use BN_rand_range() if the order of
2876 the subgroup was stored in the DH structure, but we only have
2881 BN_rand(priv_key, l, ...)
2883 where 'l' is dh->length if this is defined, or BN_num_bits(dh->p)-1
2889 RSA_eay_public_encrypt
2890 RSA_eay_private_decrypt
2891 RSA_eay_private_encrypt (signing)
2892 RSA_eay_public_decrypt (signature verification)
2894 (default implementations for RSA_public_encrypt,
2895 RSA_private_decrypt, RSA_private_encrypt, RSA_public_decrypt),
2896 always reject numbers >= n.
2899 *) In crypto/rand/md_rand.c, use a new short-time lock CRYPTO_LOCK_RAND2
2900 to synchronize access to 'locking_thread'. This is necessary on
2901 systems where access to 'locking_thread' (an 'unsigned long'
2902 variable) is not atomic.
2905 *) In crypto/rand/md_rand.c, set 'locking_thread' to current thread's ID
2906 *before* setting the 'crypto_lock_rand' flag. The previous code had
2907 a race condition if 0 is a valid thread ID.
2908 [Travis Vitek <vitek@roguewave.com>]
2910 *) Add support for shared libraries under Irix.
2911 [Albert Chin-A-Young <china@thewrittenword.com>]
2913 *) Add configuration option to build on Linux on both big-endian and
2915 [Ralf Baechle <ralf@uni-koblenz.de>]
2917 *) Add the possibility to create shared libraries on HP-UX.
2920 Changes between 0.9.6a and 0.9.6b [9 Jul 2001]
2922 *) Change ssleay_rand_bytes (crypto/rand/md_rand.c)
2923 to avoid a SSLeay/OpenSSL PRNG weakness pointed out by
2924 Markku-Juhani O. Saarinen <markku-juhani.saarinen@nokia.com>:
2925 PRNG state recovery was possible based on the output of
2926 one PRNG request appropriately sized to gain knowledge on
2927 'md' followed by enough consecutive 1-byte PRNG requests
2928 to traverse all of 'state'.
2930 1. When updating 'md_local' (the current thread's copy of 'md')
2931 during PRNG output generation, hash all of the previous
2932 'md_local' value, not just the half used for PRNG output.
2934 2. Make the number of bytes from 'state' included into the hash
2935 independent from the number of PRNG bytes requested.
2937 The first measure alone would be sufficient to avoid
2938 Markku-Juhani's attack. (Actually it had never occurred
2939 to me that the half of 'md_local' used for chaining was the
2940 half from which PRNG output bytes were taken -- I had always
2941 assumed that the secret half would be used.) The second
2942 measure makes sure that additional data from 'state' is never
2943 mixed into 'md_local' in small portions; this heuristically
2944 further strengthens the PRNG.
2947 *) Fix crypto/bn/asm/mips3.s.
2950 *) When only the key is given to "enc", the IV is undefined. Print out
2951 an error message in this case.
2954 *) Handle special case when X509_NAME is empty in X509 printing routines.
2957 *) In dsa_do_verify (crypto/dsa/dsa_ossl.c), verify that r and s are
2958 positive and less than q.
2961 *) Don't change *pointer in CRYPTO_add_lock() is add_lock_callback is
2962 used: it isn't thread safe and the add_lock_callback should handle
2964 [Paul Rose <Paul.Rose@bridge.com>]
2966 *) Verify that incoming data obeys the block size in
2967 ssl3_enc (ssl/s3_enc.c) and tls1_enc (ssl/t1_enc.c).
2971 [Ulf Möller, Bodo Möller]
2973 *) The countermeasure against Bleichbacher's attack on PKCS #1 v1.5
2974 RSA encryption was accidentally removed in s3_srvr.c in OpenSSL 0.9.5
2975 when fixing the server behaviour for backwards-compatible 'client
2976 hello' messages. (Note that the attack is impractical against
2977 SSL 3.0 and TLS 1.0 anyway because length and version checking
2978 means that the probability of guessing a valid ciphertext is
2979 around 2^-40; see section 5 in Bleichenbacher's CRYPTO '98
2982 Before 0.9.5, the countermeasure (hide the error by generating a
2983 random 'decryption result') did not work properly because
2984 ERR_clear_error() was missing, meaning that SSL_get_error() would
2985 detect the supposedly ignored error.
2987 Both problems are now fixed.
2990 *) In crypto/bio/bf_buff.c, increase DEFAULT_BUFFER_SIZE to 4096
2991 (previously it was 1024).
2994 *) Fix for compatibility mode trust settings: ignore trust settings
2995 unless some valid trust or reject settings are present.
2998 *) Fix for blowfish EVP: its a variable length cipher.
3001 *) Fix various bugs related to DSA S/MIME verification. Handle missing
3002 parameters in DSA public key structures and return an error in the
3003 DSA routines if parameters are absent.
3006 *) In versions up to 0.9.6, RAND_file_name() resorted to file ".rnd"
3007 in the current directory if neither $RANDFILE nor $HOME was set.
3008 RAND_file_name() in 0.9.6a returned NULL in this case. This has
3009 caused some confusion to Windows users who haven't defined $HOME.
3010 Thus RAND_file_name() is changed again: e_os.h can define a
3011 DEFAULT_HOME, which will be used if $HOME is not set.
3012 For Windows, we use "C:"; on other platforms, we still require
3013 environment variables.
3015 *) Move 'if (!initialized) RAND_poll()' into regions protected by
3016 CRYPTO_LOCK_RAND. This is not strictly necessary, but avoids
3017 having multiple threads call RAND_poll() concurrently.
3020 *) In crypto/rand/md_rand.c, replace 'add_do_not_lock' flag by a
3021 combination of a flag and a thread ID variable.
3022 Otherwise while one thread is in ssleay_rand_bytes (which sets the
3023 flag), *other* threads can enter ssleay_add_bytes without obeying
3024 the CRYPTO_LOCK_RAND lock (and may even illegally release the lock
3025 that they do not hold after the first thread unsets add_do_not_lock).
3028 *) Change bctest again: '-x' expressions are not available in all
3032 Changes between 0.9.6 and 0.9.6a [5 Apr 2001]
3034 *) Fix a couple of memory leaks in PKCS7_dataDecode()
3035 [Steve Henson, reported by Heyun Zheng <hzheng@atdsprint.com>]
3037 *) Change Configure and Makefiles to provide EXE_EXT, which will contain
3038 the default extension for executables, if any. Also, make the perl
3039 scripts that use symlink() to test if it really exists and use "cp"
3040 if it doesn't. All this made OpenSSL compilable and installable in
3044 *) Fix for asn1_GetSequence() for indefinite length constructed data.
3045 If SEQUENCE is length is indefinite just set c->slen to the total
3046 amount of data available.
3047 [Steve Henson, reported by shige@FreeBSD.org]
3048 [This change does not apply to 0.9.7.]
3050 *) Change bctest to avoid here-documents inside command substitution
3051 (workaround for FreeBSD /bin/sh bug).
3052 For compatibility with Ultrix, avoid shell functions (introduced
3053 in the bctest version that searches along $PATH).
3056 *) Rename 'des_encrypt' to 'des_encrypt1'. This avoids the clashes
3057 with des_encrypt() defined on some operating systems, like Solaris
3061 *) Check the result of RSA-CRT (see D. Boneh, R. DeMillo, R. Lipton:
3062 On the Importance of Eliminating Errors in Cryptographic
3063 Computations, J. Cryptology 14 (2001) 2, 101-119,
3064 http://theory.stanford.edu/~dabo/papers/faults.ps.gz).
3067 *) MIPS assembler BIGNUM division bug fix.
3070 *) Disabled incorrect Alpha assembler code.
3073 *) Fix PKCS#7 decode routines so they correctly update the length
3074 after reading an EOC for the EXPLICIT tag.
3076 [This change does not apply to 0.9.7.]
3078 *) Fix bug in PKCS#12 key generation routines. This was triggered
3079 if a 3DES key was generated with a 0 initial byte. Include
3080 PKCS12_BROKEN_KEYGEN compilation option to retain the old
3081 (but broken) behaviour.
3084 *) Enhance bctest to search for a working bc along $PATH and print
3086 [Tim Rice <tim@multitalents.net> via Richard Levitte]
3088 *) Fix memory leaks in err.c: free err_data string if necessary;
3089 don't write to the wrong index in ERR_set_error_data.
3092 *) Implement ssl23_peek (analogous to ssl23_read), which previously
3096 *) Replace rdtsc with _emit statements for VC++ version 5.
3097 [Jeremy Cooper <jeremy@baymoo.org>]
3099 *) Make it possible to reuse SSLv2 sessions.
3102 *) In copy_email() check for >= 0 as a return value for
3103 X509_NAME_get_index_by_NID() since 0 is a valid index.
3104 [Steve Henson reported by Massimiliano Pala <madwolf@opensca.org>]
3106 *) Avoid coredump with unsupported or invalid public keys by checking if
3107 X509_get_pubkey() fails in PKCS7_verify(). Fix memory leak when
3108 PKCS7_verify() fails with non detached data.
3111 *) Don't use getenv in library functions when run as setuid/setgid.
3112 New function OPENSSL_issetugid().
3115 *) Avoid false positives in memory leak detection code (crypto/mem_dbg.c)
3116 due to incorrect handling of multi-threading:
3118 1. Fix timing glitch in the MemCheck_off() portion of CRYPTO_mem_ctrl().
3120 2. Fix logical glitch in is_MemCheck_on() aka CRYPTO_is_mem_check_on().
3122 3. Count how many times MemCheck_off() has been called so that
3123 nested use can be treated correctly. This also avoids
3124 inband-signalling in the previous code (which relied on the
3125 assumption that thread ID 0 is impossible).
3128 *) Add "-rand" option also to s_client and s_server.
3131 *) Fix CPU detection on Irix 6.x.
3132 [Kurt Hockenbury <khockenb@stevens-tech.edu> and
3133 "Bruce W. Forsberg" <bruce.forsberg@baesystems.com>]
3135 *) Fix X509_NAME bug which produced incorrect encoding if X509_NAME
3138 [This change does not apply to 0.9.7.]
3140 *) Use the cached encoding of an X509_NAME structure rather than
3141 copying it. This is apparently the reason for the libsafe "errors"
3142 but the code is actually correct.
3145 *) Add new function BN_rand_range(), and fix DSA_sign_setup() to prevent
3146 Bleichenbacher's DSA attack.
3147 Extend BN_[pseudo_]rand: As before, top=1 forces the highest two bits
3148 to be set and top=0 forces the highest bit to be set; top=-1 is new
3149 and leaves the highest bit random.
3150 [Ulf Moeller, Bodo Moeller]
3152 *) In the NCONF_...-based implementations for CONF_... queries
3153 (crypto/conf/conf_lib.c), if the input LHASH is NULL, avoid using
3154 a temporary CONF structure with the data component set to NULL
3155 (which gives segmentation faults in lh_retrieve).
3156 Instead, use NULL for the CONF pointer in CONF_get_string and
3157 CONF_get_number (which may use environment variables) and directly
3158 return NULL from CONF_get_section.
3161 *) Fix potential buffer overrun for EBCDIC.
3164 *) Tolerate nonRepudiation as being valid for S/MIME signing and certSign
3165 keyUsage if basicConstraints absent for a CA.
3168 *) Make SMIME_write_PKCS7() write mail header values with a format that
3169 is more generally accepted (no spaces before the semicolon), since
3170 some programs can't parse those values properly otherwise. Also make
3171 sure BIO's that break lines after each write do not create invalid
3175 *) Make the CRL encoding routines work with empty SEQUENCE OF. The
3176 macros previously used would not encode an empty SEQUENCE OF
3177 and break the signature.
3179 [This change does not apply to 0.9.7.]
3181 *) Zero the premaster secret after deriving the master secret in
3185 *) Add some EVP_add_digest_alias registrations (as found in
3186 OpenSSL_add_all_digests()) to SSL_library_init()
3187 aka OpenSSL_add_ssl_algorithms(). This provides improved
3188 compatibility with peers using X.509 certificates
3189 with unconventional AlgorithmIdentifier OIDs.
3192 *) Fix for Irix with NO_ASM.
3193 ["Bruce W. Forsberg" <bruce.forsberg@baesystems.com>]
3195 *) ./config script fixes.
3196 [Ulf Moeller, Richard Levitte]
3198 *) Fix 'openssl passwd -1'.
3201 *) Change PKCS12_key_gen_asc() so it can cope with non null
3202 terminated strings whose length is passed in the passlen
3203 parameter, for example from PEM callbacks. This was done
3204 by adding an extra length parameter to asc2uni().
3205 [Steve Henson, reported by <oddissey@samsung.co.kr>]
3207 *) Fix C code generated by 'openssl dsaparam -C': If a BN_bin2bn
3208 call failed, free the DSA structure.
3211 *) Fix to uni2asc() to cope with zero length Unicode strings.
3212 These are present in some PKCS#12 files.
3215 *) Increase s2->wbuf allocation by one byte in ssl2_new (ssl/s2_lib.c).
3216 Otherwise do_ssl_write (ssl/s2_pkt.c) will write beyond buffer limits
3217 when writing a 32767 byte record.
3218 [Bodo Moeller; problem reported by Eric Day <eday@concentric.net>]
3220 *) In RSA_eay_public_{en,ed}crypt and RSA_eay_mod_exp (rsa_eay.c),
3221 obtain lock CRYPTO_LOCK_RSA before setting rsa->_method_mod_{n,p,q}.
3223 (RSA objects have a reference count access to which is protected
3224 by CRYPTO_LOCK_RSA [see rsa_lib.c, s3_srvr.c, ssl_cert.c, ssl_rsa.c],
3225 so they are meant to be shared between threads.)
3226 [Bodo Moeller, Geoff Thorpe; original patch submitted by
3227 "Reddie, Steven" <Steven.Reddie@ca.com>]
3229 *) Fix a deadlock in CRYPTO_mem_leaks().
3232 *) Use better test patterns in bntest.
3235 *) rand_win.c fix for Borland C.
3238 *) BN_rshift bugfix for n == 0.
3241 *) Add a 'bctest' script that checks for some known 'bc' bugs
3242 so that 'make test' does not abort just because 'bc' is broken.
3245 *) Store verify_result within SSL_SESSION also for client side to
3246 avoid potential security hole. (Re-used sessions on the client side
3247 always resulted in verify_result==X509_V_OK, not using the original
3248 result of the server certificate verification.)
3251 *) Fix ssl3_pending: If the record in s->s3->rrec is not of type
3252 SSL3_RT_APPLICATION_DATA, return 0.
3253 Similarly, change ssl2_pending to return 0 if SSL_in_init(s) is true.
3257 Both ssl2_peek and ssl3_peek, which were totally broken in earlier
3258 releases, have been re-implemented by renaming the previous
3259 implementations of ssl2_read and ssl3_read to ssl2_read_internal
3260 and ssl3_read_internal, respectively, and adding 'peek' parameters
3261 to them. The new ssl[23]_{read,peek} functions are calls to
3262 ssl[23]_read_internal with the 'peek' flag set appropriately.
3263 A 'peek' parameter has also been added to ssl3_read_bytes, which
3264 does the actual work for ssl3_read_internal.
3267 *) Initialise "ex_data" member of RSA/DSA/DH structures prior to calling
3268 the method-specific "init()" handler. Also clean up ex_data after
3269 calling the method-specific "finish()" handler. Previously, this was
3270 happening the other way round.
3273 *) Increase BN_CTX_NUM (the number of BIGNUMs in a BN_CTX) to 16.
3274 The previous value, 12, was not always sufficient for BN_mod_exp().
3277 *) Make sure that shared libraries get the internal name engine with
3278 the full version number and not just 0. This should mark the
3279 shared libraries as not backward compatible. Of course, this should
3280 be changed again when we can guarantee backward binary compatibility.
3283 *) Fix typo in get_cert_by_subject() in by_dir.c
3284 [Jean-Marc Desperrier <jean-marc.desperrier@certplus.com>]
3286 *) Rework the system to generate shared libraries:
3288 - Make note of the expected extension for the shared libraries and
3289 if there is a need for symbolic links from for example libcrypto.so.0
3290 to libcrypto.so.0.9.7. There is extended info in Configure for
3293 - Make as few rebuilds of the shared libraries as possible.
3295 - Still avoid linking the OpenSSL programs with the shared libraries.
3297 - When installing, install the shared libraries separately from the
3301 *) Fix SSL_CTX_set_read_ahead macro to actually use its argument.
3303 Copy SSL_CTX's read_ahead flag to SSL object directly in SSL_new
3304 and not in SSL_clear because the latter is also used by the
3305 accept/connect functions; previously, the settings made by
3306 SSL_set_read_ahead would be lost during the handshake.
3307 [Bodo Moeller; problems reported by Anders Gertz <gertz@epact.se>]
3309 *) Correct util/mkdef.pl to be selective about disabled algorithms.
3310 Previously, it would create entries for disableed algorithms no
3314 *) Added several new manual pages for SSL_* function.
3317 Changes between 0.9.5a and 0.9.6 [24 Sep 2000]
3319 *) In ssl23_get_client_hello, generate an error message when faced
3320 with an initial SSL 3.0/TLS record that is too small to contain the
3321 first two bytes of the ClientHello message, i.e. client_version.
3322 (Note that this is a pathologic case that probably has never happened
3323 in real life.) The previous approach was to use the version number
3324 from the record header as a substitute; but our protocol choice
3325 should not depend on that one because it is not authenticated
3326 by the Finished messages.
3329 *) More robust randomness gathering functions for Windows.
3330 [Jeffrey Altman <jaltman@columbia.edu>]
3332 *) For compatibility reasons if the flag X509_V_FLAG_ISSUER_CHECK is
3333 not set then we don't setup the error code for issuer check errors
3334 to avoid possibly overwriting other errors which the callback does
3335 handle. If an application does set the flag then we assume it knows
3336 what it is doing and can handle the new informational codes
3340 *) Fix for a nasty bug in ASN1_TYPE handling. ASN1_TYPE is used for
3341 a general "ANY" type, as such it should be able to decode anything
3342 including tagged types. However it didn't check the class so it would
3343 wrongly interpret tagged types in the same way as their universal
3344 counterpart and unknown types were just rejected. Changed so that the
3345 tagged and unknown types are handled in the same way as a SEQUENCE:
3346 that is the encoding is stored intact. There is also a new type
3347 "V_ASN1_OTHER" which is used when the class is not universal, in this
3348 case we have no idea what the actual type is so we just lump them all
3352 *) On VMS, stdout may very well lead to a file that is written to
3353 in a record-oriented fashion. That means that every write() will
3354 write a separate record, which will be read separately by the
3355 programs trying to read from it. This can be very confusing.
3357 The solution is to put a BIO filter in the way that will buffer
3358 text until a linefeed is reached, and then write everything a
3359 line at a time, so every record written will be an actual line,
3360 not chunks of lines and not (usually doesn't happen, but I've
3361 seen it once) several lines in one record. BIO_f_linebuffer() is
3364 Currently, it's a VMS-only method, because that's where it has
3365 been tested well enough.
3368 *) Remove 'optimized' squaring variant in BN_mod_mul_montgomery,
3369 it can return incorrect results.
3370 (Note: The buggy variant was not enabled in OpenSSL 0.9.5a,
3371 but it was in 0.9.6-beta[12].)
3374 *) Disable the check for content being present when verifying detached
3375 signatures in pk7_smime.c. Some versions of Netscape (wrongly)
3376 include zero length content when signing messages.
3379 *) New BIO_shutdown_wr macro, which invokes the BIO_C_SHUTDOWN_WR
3380 BIO_ctrl (for BIO pairs).
3383 *) Add DSO method for VMS.
3386 *) Bug fix: Montgomery multiplication could produce results with the
3390 *) Add RPM specification openssl.spec and modify it to build three
3391 packages. The default package contains applications, application
3392 documentation and run-time libraries. The devel package contains
3393 include files, static libraries and function documentation. The
3394 doc package contains the contents of the doc directory. The original
3395 openssl.spec was provided by Damien Miller <djm@mindrot.org>.
3398 *) Add a large number of documentation files for many SSL routines.
3399 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>]
3401 *) Add a configuration entry for Sony News 4.
3402 [NAKAJI Hiroyuki <nakaji@tutrp.tut.ac.jp>]
3404 *) Don't set the two most significant bits to one when generating a
3405 random number < q in the DSA library.
3408 *) New SSL API mode 'SSL_MODE_AUTO_RETRY'. This disables the default
3409 behaviour that SSL_read may result in SSL_ERROR_WANT_READ (even if
3410 the underlying transport is blocking) if a handshake took place.
3411 (The default behaviour is needed by applications such as s_client
3412 and s_server that use select() to determine when to use SSL_read;
3413 but for applications that know in advance when to expect data, it
3414 just makes things more complicated.)
3417 *) Add RAND_egd_bytes(), which gives control over the number of bytes read
3421 *) Add a few more EBCDIC conditionals that make `req' and `x509'
3422 work better on such systems.
3423 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>]
3425 *) Add two demo programs for PKCS12_parse() and PKCS12_create().
3426 Update PKCS12_parse() so it copies the friendlyName and the
3427 keyid to the certificates aux info.
3430 *) Fix bug in PKCS7_verify() which caused an infinite loop
3431 if there was more than one signature.
3432 [Sven Uszpelkat <su@celocom.de>]
3434 *) Major change in util/mkdef.pl to include extra information
3435 about each symbol, as well as presentig variables as well
3436 as functions. This change means that there's n more need
3437 to rebuild the .num files when some algorithms are excluded.
3440 *) Allow the verify time to be set by an application,
3441 rather than always using the current time.
3444 *) Phase 2 verify code reorganisation. The certificate
3445 verify code now looks up an issuer certificate by a
3446 number of criteria: subject name, authority key id
3447 and key usage. It also verifies self signed certificates
3448 by the same criteria. The main comparison function is
3449 X509_check_issued() which performs these checks.
3451 Lot of changes were necessary in order to support this
3452 without completely rewriting the lookup code.
3454 Authority and subject key identifier are now cached.
3456 The LHASH 'certs' is X509_STORE has now been replaced
3457 by a STACK_OF(X509_OBJECT). This is mainly because an
3458 LHASH can't store or retrieve multiple objects with
3459 the same hash value.
3461 As a result various functions (which were all internal
3462 use only) have changed to handle the new X509_STORE
3463 structure. This will break anything that messed round
3464 with X509_STORE internally.
3466 The functions X509_STORE_add_cert() now checks for an
3467 exact match, rather than just subject name.
3469 The X509_STORE API doesn't directly support the retrieval
3470 of multiple certificates matching a given criteria, however
3471 this can be worked round by performing a lookup first
3472 (which will fill the cache with candidate certificates)
3473 and then examining the cache for matches. This is probably
3474 the best we can do without throwing out X509_LOOKUP
3475 entirely (maybe later...).
3477 The X509_VERIFY_CTX structure has been enhanced considerably.
3479 All certificate lookup operations now go via a get_issuer()
3480 callback. Although this currently uses an X509_STORE it
3481 can be replaced by custom lookups. This is a simple way
3482 to bypass the X509_STORE hackery necessary to make this
3483 work and makes it possible to use more efficient techniques
3484 in future. A very simple version which uses a simple
3485 STACK for its trusted certificate store is also provided
3486 using X509_STORE_CTX_trusted_stack().
3488 The verify_cb() and verify() callbacks now have equivalents
3489 in the X509_STORE_CTX structure.
3491 X509_STORE_CTX also has a 'flags' field which can be used
3492 to customise the verify behaviour.
3495 *) Add new PKCS#7 signing option PKCS7_NOSMIMECAP which
3496 excludes S/MIME capabilities.
3499 *) When a certificate request is read in keep a copy of the
3500 original encoding of the signed data and use it when outputing
3501 again. Signatures then use the original encoding rather than
3502 a decoded, encoded version which may cause problems if the
3503 request is improperly encoded.
3506 *) For consistency with other BIO_puts implementations, call
3507 buffer_write(b, ...) directly in buffer_puts instead of calling
3510 In BIO_puts, increment b->num_write as in BIO_write.
3511 [Peter.Sylvester@EdelWeb.fr]
3513 *) Fix BN_mul_word for the case where the word is 0. (We have to use
3514 BN_zero, we may not return a BIGNUM with an array consisting of
3518 *) Avoid calling abort() from within the library when problems are
3519 detected, except if preprocessor symbols have been defined
3520 (such as REF_CHECK, BN_DEBUG etc.).
3523 *) New openssl application 'rsautl'. This utility can be
3524 used for low level RSA operations. DER public key
3525 BIO/fp routines also added.
3528 *) New Configure entry and patches for compiling on QNX 4.
3529 [Andreas Schneider <andreas@ds3.etech.fh-hamburg.de>]
3531 *) A demo state-machine implementation was sponsored by
3532 Nuron (http://www.nuron.com/) and is now available in
3533 demos/state_machine.
3536 *) New options added to the 'dgst' utility for signature
3537 generation and verification.
3540 *) Unrecognized PKCS#7 content types are now handled via a
3541 catch all ASN1_TYPE structure. This allows unsupported
3542 types to be stored as a "blob" and an application can
3543 encode and decode it manually.
3546 *) Fix various signed/unsigned issues to make a_strex.c
3548 [Oscar Jacobsson <oscar.jacobsson@celocom.com>]
3550 *) ASN1 fixes. i2d_ASN1_OBJECT was not returning the correct
3551 length if passed a buffer. ASN1_INTEGER_to_BN failed
3552 if passed a NULL BN and its argument was negative.
3553 [Steve Henson, pointed out by Sven Heiberg <sven@tartu.cyber.ee>]
3555 *) Modification to PKCS#7 encoding routines to output definite
3556 length encoding. Since currently the whole structures are in
3557 memory there's not real point in using indefinite length
3558 constructed encoding. However if OpenSSL is compiled with
3559 the flag PKCS7_INDEFINITE_ENCODING the old form is used.
3562 *) Added BIO_vprintf() and BIO_vsnprintf().
3565 *) Added more prefixes to parse for in the the strings written
3566 through a logging bio, to cover all the levels that are available
3567 through syslog. The prefixes are now:
3569 PANIC, EMERG, EMR => LOG_EMERG
3570 ALERT, ALR => LOG_ALERT
3571 CRIT, CRI => LOG_CRIT
3572 ERROR, ERR => LOG_ERR
3573 WARNING, WARN, WAR => LOG_WARNING
3574 NOTICE, NOTE, NOT => LOG_NOTICE
3575 INFO, INF => LOG_INFO
3576 DEBUG, DBG => LOG_DEBUG
3578 and as before, if none of those prefixes are present at the
3579 beginning of the string, LOG_ERR is chosen.
3581 On Win32, the LOG_* levels are mapped according to this:
3583 LOG_EMERG, LOG_ALERT, LOG_CRIT, LOG_ERR => EVENTLOG_ERROR_TYPE
3584 LOG_WARNING => EVENTLOG_WARNING_TYPE
3585 LOG_NOTICE, LOG_INFO, LOG_DEBUG => EVENTLOG_INFORMATION_TYPE
3589 *) Made it possible to reconfigure with just the configuration
3590 argument "reconf" or "reconfigure". The command line arguments
3591 are stored in Makefile.ssl in the variable CONFIGURE_ARGS,
3592 and are retrieved from there when reconfiguring.
3596 [Assar Westerlund <assar@sics.se>, Richard Levitte]
3598 *) Add the arguments -CAfile and -CApath to the pkcs12 utility.
3601 *) The obj_dat.pl script was messing up the sorting of object
3602 names. The reason was that it compared the quoted version
3603 of strings as a result "OCSP" > "OCSP Signing" because
3604 " > SPACE. Changed script to store unquoted versions of
3605 names and add quotes on output. It was also omitting&
projects / openssl.git / blob