5 Changes between 0.9.7 and 0.9.8 [xx XXX 2002]
7 *) Add the ASN.1 structures and functions for CertificatePair, which
8 is defined as follows (according to X.509_4thEditionDraftV6.pdf):
10 CertificatePair ::= SEQUENCE {
11 forward [0] Certificate OPTIONAL,
12 reverse [1] Certificate OPTIONAL,
13 -- at least one of the pair shall be present -- }
15 Also implement the PEM functions to read and write certificate
16 pairs, and defined the PEM tag as "CERTIFICATE PAIR".
18 This needed to be defined, mostly for the sake of the LDAP
19 attribute crossCertificatePair, but may prove useful elsewhere as
23 *) Make it possible to inhibit symlinking of shared libraries in
24 Makefile.shared, for Cygwin's sake.
27 *) Extend the BIGNUM API by creating new macros that behave like
30 void BN_set_sign(BIGNUM *a, int neg);
31 int BN_get_sign(const BIGNUM *a);
33 and avoid the need to access 'a->neg' directly in applications.
34 [Nils Larsch <nla@trustcenter.de>]
36 *) Implement fast modular reduction for pseudo-Mersenne primes
37 used in NIST curves (crypto/bn/bn_nist.c, crypto/ec/ecp_nist.c).
38 EC_GROUP_new_curve_GFp() will now automatically use this
40 [Nils Larsch <nla@trustcenter.de>]
42 *) Add new lock type (CRYPTO_LOCK_BN).
45 *) Change the ENGINE framework to automatically load engines
46 dynamically from specific directories unless they could be
47 found to already be built in or loaded. Move all the
48 current engines except for the cryptodev one to a new
50 The engines in engines/ are built as shared libraries if
51 the "shared" options was given to ./Configure or ./config.
52 Otherwise, they are inserted in libcrypto.a.
53 /usr/local/ssl/engines is the default directory for dynamic
54 engines, but that can be overriden at configure time through
55 the usual use of --prefix and/or --openssldir, and at run
56 time with the environment variable OPENSSL_ENGINES.
57 [Geoff Thorpe and Richard Levitte]
59 *) Add Makefile.shared, a helper makefile to build shared
60 libraries. Addapt Makefile.org.
63 *) Add version info to Win32 DLLs.
64 [Peter 'Luna' Runestig" <peter@runestig.com>]
66 *) Add new 'medium level' PKCS#12 API. Certificates and keys
67 can be added using this API to created arbitrary PKCS#12
68 files while avoiding the low level API.
70 New options to PKCS12_create(), key or cert can be NULL and
71 will then be omitted from the output file. The encryption
72 algorithm NIDs can be set to -1 for no encryption, the mac
73 iteration count can be set to 0 to omit the mac.
75 Enhance pkcs12 utility by making the -nokeys and -nocerts
76 options work when creating a PKCS#12 file. New option -nomac
77 to omit the mac, NONE can be set for an encryption algorithm.
78 New code is modified to use the enhanced PKCS12_create()
79 instead of the low level API.
82 *) Extend ASN1 encoder to support indefinite length constructed
83 encoding. This can output sequences tags and octet strings in
84 this form. Modify pk7_asn1.c to support indefinite length
85 encoding. This is experimental and needs additional code to
86 be useful, such as an ASN1 bio and some enhanced streaming
89 Extend template encode functionality so that tagging is passed
90 down to the template encoder.
93 *) Let 'openssl req' fail if an argument to '-newkey' is not
94 recognized instead of using RSA as a default.
97 *) Add support for ECC-based ciphersuites from draft-ietf-tls-ecc-01.txt.
98 As these are not official, they are not included in "ALL";
99 the "ECCdraft" ciphersuite group alias can be used to select them.
100 [Vipul Gupta and Sumit Gupta (Sun Microsystems Laboratories)]
102 *) Add ECDH engine support.
103 [Nils Gura and Douglas Stebila (Sun Microsystems Laboratories)]
105 *) Add ECDH in new directory crypto/ecdh/.
106 TODO: more general interface (return x coordinate, not its hash)
107 TODO: bug: pad x with leading zeros if necessary
108 [Douglas Stebila (Sun Microsystems Laboratories)]
110 *) Let BN_rand_range() abort with an error after 100 iterations
111 without success (which indicates a broken PRNG).
114 *) Change BN_mod_sqrt() so that it verifies that the input value
115 is really the square of the return value. (Previously,
116 BN_mod_sqrt would show GIGO behaviour.)
119 *) Add named elliptic curves over binary fields from X9.62, SECG,
120 and WAP/WTLS; add OIDs that were still missing.
122 [Sheueling Chang Shantz and Douglas Stebila
123 (Sun Microsystems Laboratories)]
125 *) Extend the EC library for elliptic curves over binary fields
126 (new files ec2_smpl.c, ec2_smpt.c, ec2_mult.c in crypto/ec/).
129 EC_GF2m_simple_method
133 EC_GROUP_new_curve_GF2m
134 EC_GROUP_set_curve_GF2m
135 EC_GROUP_get_curve_GF2m
136 EC_POINT_set_affine_coordinates_GF2m
137 EC_POINT_get_affine_coordinates_GF2m
138 EC_POINT_set_compressed_coordinates_GF2m
140 Point compression for binary fields is disabled by default for
141 patent reasons (compile with OPENSSL_EC_BIN_PT_COMP defined to
144 As binary polynomials are represented as BIGNUMs, various members
145 of the EC_GROUP and EC_POINT data structures can be shared
146 between the implementations for prime fields and binary fields;
147 the above ..._GF2m functions (except for EX_GROUP_new_curve_GF2m)
148 are essentially identical to their ..._GFp counterparts.
149 (For simplicity, the '..._GFp' prefix has been dropped from
150 various internal method names.)
152 An internal 'field_div' method (similar to 'field_mul' and
153 'field_sqr') has been added; this is used only for binary fields.
155 [Sheueling Chang Shantz and Douglas Stebila
156 (Sun Microsystems Laboratories)]
158 *) Optionally dispatch EC_POINT_mul(), EC_POINT_precompute_mult()
159 through methods ('mul', 'precompute_mult').
161 The generic implementations (now internally called 'ec_wNAF_mul'
162 and 'ec_wNAF_precomputed_mult') remain the default if these
163 methods are undefined.
165 [Sheueling Chang Shantz and Douglas Stebila
166 (Sun Microsystems Laboratories)]
168 *) New function EC_GROUP_get_degree, which is defined through
169 EC_METHOD. For curves over prime fields, this returns the bit
170 length of the modulus.
172 [Sheueling Chang Shantz and Douglas Stebila
173 (Sun Microsystems Laboratories)]
175 *) New functions EC_GROUP_dup, EC_POINT_dup.
176 (These simply call ..._new and ..._copy).
178 [Sheueling Chang Shantz and Douglas Stebila
179 (Sun Microsystems Laboratories)]
181 *) Add binary polynomial arithmetic software in crypto/bn/bn_gf2m.c.
182 Polynomials are represented as BIGNUMs (where the sign bit is not
183 used) in the following functions [macros]:
186 BN_GF2m_sub [= BN_GF2m_add]
187 BN_GF2m_mod [wrapper for BN_GF2m_mod_arr]
188 BN_GF2m_mod_mul [wrapper for BN_GF2m_mod_mul_arr]
189 BN_GF2m_mod_sqr [wrapper for BN_GF2m_mod_sqr_arr]
191 BN_GF2m_mod_exp [wrapper for BN_GF2m_mod_exp_arr]
192 BN_GF2m_mod_sqrt [wrapper for BN_GF2m_mod_sqrt_arr]
193 BN_GF2m_mod_solve_quad [wrapper for BN_GF2m_mod_solve_quad_arr]
194 BN_GF2m_cmp [= BN_ucmp]
196 (Note that only the 'mod' functions are actually for fields GF(2^m).
197 BN_GF2m_add() is misnomer, but this is for the sake of consistency.)
199 For some functions, an the irreducible polynomial defining a
200 field can be given as an 'unsigned int[]' with strictly
201 decreasing elements giving the indices of those bits that are set;
202 i.e., p[] represents the polynomial
203 f(t) = t^p[0] + t^p[1] + ... + t^p[k]
205 p[0] > p[1] > ... > p[k] = 0.
206 This applies to the following functions:
211 BN_GF2m_mod_inv_arr [wrapper for BN_GF2m_mod_inv]
212 BN_GF2m_mod_div_arr [wrapper for BN_GF2m_mod_div]
215 BN_GF2m_mod_solve_quad_arr
219 Conversion can be performed by the following functions:
224 bntest.c has additional tests for binary polynomial arithmetic.
226 Two implementations for BN_GF2m_mod_div() are available.
227 The default algorithm simply uses BN_GF2m_mod_inv() and
228 BN_GF2m_mod_mul(). The alternative algorithm is compiled in only
229 if OPENSSL_SUN_GF2M_DIV is defined (patent pending; read the
230 copyright notice in crypto/bn/bn_gf2m.c before enabling it).
232 [Sheueling Chang Shantz and Douglas Stebila
233 (Sun Microsystems Laboratories)]
235 *) Add new error code 'ERR_R_DISABLED' that can be used when some
236 functionality is disabled at compile-time.
237 [Douglas Stebila <douglas.stebila@sun.com>]
239 *) Change default behaviour of 'openssl asn1parse' so that more
240 information is visible when viewing, e.g., a certificate:
242 Modify asn1_parse2 (crypto/asn1/asn1_par.c) so that in non-'dump'
243 mode the content of non-printable OCTET STRINGs is output in a
244 style similar to INTEGERs, but with '[HEX DUMP]' prepended to
245 avoid the appearance of a printable string.
246 [Nils Larsch <nla@trustcenter.de>]
248 *) Add 'asn1_flag' and 'asn1_form' member to EC_GROUP with access
250 EC_GROUP_set_asn1_flag()
251 EC_GROUP_get_asn1_flag()
252 EC_GROUP_set_point_conversion_form()
253 EC_GROUP_get_point_conversion_form()
254 These control ASN1 encoding details:
255 - Curves (i.e., groups) are encoded explicitly unless asn1_flag
256 has been set to OPENSSL_EC_NAMED_CURVE.
257 - Points are encoded in uncompressed form by default; options for
258 asn1_for are as for point2oct, namely
259 POINT_CONVERSION_COMPRESSED
260 POINT_CONVERSION_UNCOMPRESSED
261 POINT_CONVERSION_HYBRID
263 Also add 'seed' and 'seed_len' members to EC_GROUP with access
267 EC_GROUP_get_seed_len()
268 This is used only for ASN1 purposes (so far).
269 [Nils Larsch <nla@trustcenter.de>]
271 *) Add 'field_type' member to EC_METHOD, which holds the NID
272 of the appropriate field type OID. The new function
273 EC_METHOD_get_field_type() returns this value.
274 [Nils Larsch <nla@trustcenter.de>]
281 providing useful interfaces to EC_POINT_point2oct() and
282 EC_POINT_oct2point().
283 [Nils Larsch <nla@trustcenter.de>]
285 *) Change internals of the EC library so that the functions
286 EC_GROUP_set_generator()
287 EC_GROUP_get_generator()
289 EC_GROUP_get_cofactor()
290 are implemented directly in crypto/ec/ec_lib.c and not dispatched
291 to methods, which would lead to unnecessary code duplication when
292 adding different types of curves.
293 [Nils Larsch <nla@trustcenter.de> with input by Bodo Moeller]
295 *) Implement compute_wNAF (crypto/ec/ec_mult.c) without BIGNUM
296 arithmetic, and such that modified wNAFs are generated
297 (which avoid length expansion in many cases).
300 *) Add a function EC_GROUP_check_discriminant() (defined via
301 EC_METHOD) that verifies that the curve discriminant is non-zero.
303 Add a function EC_GROUP_check() that makes some sanity tests
304 on a EC_GROUP, its generator and order. This includes
305 EC_GROUP_check_discriminant().
306 [Nils Larsch <nla@trustcenter.de>]
308 *) Add ECDSA in new directory crypto/ecdsa/.
310 Add applications 'openssl ecparam' and 'openssl ecdsa'
311 (these are based on 'openssl dsaparam' and 'openssl dsa').
313 ECDSA support is also included in various other files across the
314 library. Most notably,
315 - 'openssl req' now has a '-newkey ecdsa:file' option;
316 - EVP_PKCS82PKEY (crypto/evp/evp_pkey.c) now can handle ECDSA;
317 - X509_PUBKEY_get (crypto/asn1/x_pubkey.c) and
318 d2i_PublicKey (crypto/asn1/d2i_pu.c) have been modified to make
319 them suitable for ECDSA where domain parameters must be
320 extracted before the specific public key;
321 - ECDSA engine support has been added.
322 [Nils Larsch <nla@trustcenter.de>]
324 *) Include some named elliptic curves, and add OIDs from X9.62,
325 SECG, and WAP/WTLS. Each curve can be obtained from the new
327 EC_GROUP_new_by_nid(),
328 and the list of available named curves can be obtained with
329 EC_get_builtin_curves().
330 Also add a 'curve_name' member to EC_GROUP objects, which can be
334 [Nils Larsch <nla@trustcenter.de, Bodo Moeller]
336 Changes between 0.9.6h and 0.9.7 [XX xxx 2002]
338 *) Add the VC-CE target, introduce the WINCE sysname, and add
339 INSTALL.WCE and appropriate conditionals to make it build.
340 [Steven Reddie <smr@essemer.com.au> via Richard Levitte]
342 *) Change the DLL names for Cygwin to cygcrypto-x.y.z.dll and
343 cygssl-x.y.z.dll, where x, y and z are the major, minor and
344 edit numbers of the version.
345 [Corinna Vinschen <vinschen@redhat.com> and Richard Levitte]
347 *) Introduce safe string copy and catenation functions
348 (BUF_strlcpy() and BUF_strlcat()).
349 [Ben Laurie (CHATS) and Richard Levitte]
351 *) Avoid using fixed-size buffers for one-line DNs.
354 *) Add BUF_MEM_grow_clean() to avoid information leakage when
355 resizing buffers containing secrets, and use where appropriate.
358 *) Avoid using fixed size buffers for configuration file location.
361 *) Avoid filename truncation for various CA files.
364 *) Use sizeof in preference to magic numbers.
367 *) Avoid filename truncation in cert requests.
370 *) Add assertions to check for (supposedly impossible) buffer
374 *) Don't cache truncated DNS entries in the local cache (this could
375 potentially lead to a spoofing attack).
378 *) Fix various buffers to be large enough for hex/decimal
379 representations in a platform independent manner.
382 *) Add CRYPTO_realloc_clean() to avoid information leakage when
383 resizing buffers containing secrets, and use where appropriate.
386 *) Add BIO_indent() to avoid much slightly worrying code to do
390 *) Convert sprintf()/BIO_puts() to BIO_printf().
393 *) buffer_gets() could terminate with the buffer only half
397 *) Add assertions to prevent user-supplied crypto functions from
398 overflowing internal buffers by having large block sizes, etc.
401 *) New OPENSSL_assert() macro (similar to assert(), but enabled
405 *) Eliminate unused copy of key in RC4.
408 *) Eliminate unused and incorrectly sized buffers for IV in pem.h.
411 *) Fix off-by-one error in EGD path.
414 *) If RANDFILE path is too long, ignore instead of truncating.
417 *) Eliminate unused and incorrectly sized X.509 structure
421 *) Eliminate unused and dangerous function knumber().
424 *) Eliminate unused and dangerous structure, KSSL_ERR.
427 *) Protect against overlong session ID context length in an encoded
428 session object. Since these are local, this does not appear to be
432 *) Change from security patch (see 0.9.6e below) that did not affect
433 the 0.9.6 release series:
435 Remote buffer overflow in SSL3 protocol - an attacker could
436 supply an oversized master key in Kerberos-enabled versions.
440 *) Change the SSL kerb5 codes to match RFC 2712.
443 *) Make -nameopt work fully for req and add -reqopt switch.
444 [Michael Bell <michael.bell@rz.hu-berlin.de>, Steve Henson]
446 *) The "block size" for block ciphers in CFB and OFB mode should be 1.
447 [Steve Henson, reported by Yngve Nysaeter Pettersen <yngve@opera.com>]
449 *) Make sure tests can be performed even if the corresponding algorithms
450 have been removed entirely. This was also the last step to make
451 OpenSSL compilable with DJGPP under all reasonable conditions.
452 [Richard Levitte, Doug Kaufman <dkaufman@rahul.net>]
454 *) Add cipher selection rules COMPLEMENTOFALL and COMPLEMENTOFDEFAULT
455 to allow version independent disabling of normally unselected ciphers,
456 which may be activated as a side-effect of selecting a single cipher.
458 (E.g., cipher list string "RSA" enables ciphersuites that are left
459 out of "ALL" because they do not provide symmetric encryption.
460 "RSA:!COMPLEMEMENTOFALL" avoids these unsafe ciphersuites.)
461 [Lutz Jaenicke, Bodo Moeller]
463 *) Add appropriate support for separate platform-dependent build
464 directories. The recommended way to make a platform-dependent
465 build directory is the following (tested on Linux), maybe with
468 # Place yourself outside of the OpenSSL source tree. In
469 # this example, the environment variable OPENSSL_SOURCE
470 # is assumed to contain the absolute OpenSSL source directory.
471 mkdir -p objtree/"`uname -s`-`uname -r`-`uname -m`"
472 cd objtree/"`uname -s`-`uname -r`-`uname -m`"
473 (cd $OPENSSL_SOURCE; find . -type f -o -type l) | while read F; do
474 mkdir -p `dirname $F`
475 ln -s $OPENSSL_SOURCE/$F $F
478 To be absolutely sure not to disturb the source tree, a "make clean"
479 is a good thing. If it isn't successfull, don't worry about it,
480 it probably means the source directory is very clean.
483 *) Make sure any ENGINE control commands make local copies of string
484 pointers passed to them whenever necessary. Otherwise it is possible
485 the caller may have overwritten (or deallocated) the original string
486 data when a later ENGINE operation tries to use the stored values.
487 [Götz Babin-Ebell <babinebell@trustcenter.de>]
489 *) Improve diagnostics in file reading and command-line digests.
490 [Ben Laurie aided and abetted by Solar Designer <solar@openwall.com>]
492 *) Add AES modes CFB and OFB to the object database. Correct an
493 error in AES-CFB decryption.
496 *) Remove most calls to EVP_CIPHER_CTX_cleanup() in evp_enc.c, this
497 allows existing EVP_CIPHER_CTX structures to be reused after
498 calling EVP_*Final(). This behaviour is used by encryption
499 BIOs and some applications. This has the side effect that
500 applications must explicitly clean up cipher contexts with
501 EVP_CIPHER_CTX_cleanup() or they will leak memory.
504 *) Check the values of dna and dnb in bn_mul_recursive before calling
505 bn_mul_comba (a non zero value means the a or b arrays do not contain
506 n2 elements) and fallback to bn_mul_normal if either is not zero.
509 *) Fix escaping of non-ASCII characters when using the -subj option
510 of the "openssl req" command line tool. (Robert Joop <joop@fokus.gmd.de>)
513 *) Make object definitions compliant to LDAP (RFC2256): SN is the short
514 form for "surname", serialNumber has no short form.
515 Use "mail" as the short name for "rfc822Mailbox" according to RFC2798;
516 therefore remove "mail" short name for "internet 7".
517 The OID for unique identifiers in X509 certificates is
518 x500UniqueIdentifier, not uniqueIdentifier.
519 Some more OID additions. (Michael Bell <michael.bell@rz.hu-berlin.de>)
522 *) Add an "init" command to the ENGINE config module and auto initialize
523 ENGINEs. Without any "init" command the ENGINE will be initialized
524 after all ctrl commands have been executed on it. If init=1 the
525 ENGINE is initailized at that point (ctrls before that point are run
526 on the uninitialized ENGINE and after on the initialized one). If
527 init=0 then the ENGINE will not be iniatialized at all.
530 *) Fix the 'app_verify_callback' interface so that the user-defined
531 argument is actually passed to the callback: In the
532 SSL_CTX_set_cert_verify_callback() prototype, the callback
533 declaration has been changed from
536 int (*cb)(X509_STORE_CTX *,void *);
537 in ssl_verify_cert_chain (ssl/ssl_cert.c), the call
538 i=s->ctx->app_verify_callback(&ctx)
539 has been changed into
540 i=s->ctx->app_verify_callback(&ctx, s->ctx->app_verify_arg).
542 To update applications using SSL_CTX_set_cert_verify_callback(),
543 a dummy argument can be added to their callback functions.
544 [D. K. Smetters <smetters@parc.xerox.com>]
546 *) Added the '4758cca' ENGINE to support IBM 4758 cards.
547 [Maurice Gittens <maurice@gittens.nl>, touchups by Geoff Thorpe]
549 *) Add and OPENSSL_LOAD_CONF define which will cause
550 OpenSSL_add_all_algorithms() to load the openssl.cnf config file.
551 This allows older applications to transparently support certain
552 OpenSSL features: such as crypto acceleration and dynamic ENGINE loading.
553 Two new functions OPENSSL_add_all_algorithms_noconf() which will never
554 load the config file and OPENSSL_add_all_algorithms_conf() which will
555 always load it have also been added.
558 *) Add the OFB, CFB and CTR (all with 128 bit feedback) to AES.
559 Adjust NIDs and EVP layer.
560 [Stephen Sprunk <stephen@sprunk.org> and Richard Levitte]
562 *) Config modules support in openssl utility.
564 Most commands now load modules from the config file,
565 though in a few (such as version) this isn't done
566 because it couldn't be used for anything.
568 In the case of ca and req the config file used is
569 the same as the utility itself: that is the -config
570 command line option can be used to specify an
574 *) Move default behaviour from OPENSSL_config(). If appname is NULL
575 use "openssl_conf" if filename is NULL use default openssl config file.
578 *) Add an argument to OPENSSL_config() to allow the use of an alternative
579 config section name. Add a new flag to tolerate a missing config file
580 and move code to CONF_modules_load_file().
583 *) Support for crypto accelerator cards from Accelerated Encryption
584 Processing, www.aep.ie. (Use engine 'aep')
585 The support was copied from 0.9.6c [engine] and adapted/corrected
586 to work with the new engine framework.
587 [AEP Inc. and Richard Levitte]
589 *) Support for SureWare crypto accelerator cards from Baltimore
590 Technologies. (Use engine 'sureware')
591 The support was copied from 0.9.6c [engine] and adapted
592 to work with the new engine framework.
595 *) Have the CHIL engine fork-safe (as defined by nCipher) and actually
596 make the newer ENGINE framework commands for the CHIL engine work.
597 [Toomas Kiisk <vix@cyber.ee> and Richard Levitte]
599 *) Make it possible to produce shared libraries on ReliantUNIX.
600 [Robert Dahlem <Robert.Dahlem@ffm2.siemens.de> via Richard Levitte]
602 *) Add the configuration target debug-linux-ppro.
603 Make 'openssl rsa' use the general key loading routines
604 implemented in apps.c, and make those routines able to
605 handle the key format FORMAT_NETSCAPE and the variant
607 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte]
609 *) Fix a crashbug and a logic bug in hwcrhk_load_pubkey().
610 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte]
612 *) Add -keyform to rsautl, and document -engine.
613 [Richard Levitte, inspired by Toomas Kiisk <vix@cyber.ee>]
615 *) Change BIO_new_file (crypto/bio/bss_file.c) to use new
616 BIO_R_NO_SUCH_FILE error code rather than the generic
617 ERR_R_SYS_LIB error code if fopen() fails with ENOENT.
622 ERR_peek_last_error_line
623 ERR_peek_last_error_line_data.
627 ERR_peek_error_line_data,
628 but report on the latest error recorded rather than the first one
629 still in the error queue.
630 [Ben Laurie, Bodo Moeller]
632 *) default_algorithms option in ENGINE config module. This allows things
634 default_algorithms = ALL
635 default_algorithms = RSA, DSA, RAND, CIPHERS, DIGESTS
638 *) Prelminary ENGINE config module.
641 *) New experimental application configuration code.
644 *) Change the AES code to follow the same name structure as all other
645 symmetric ciphers, and behave the same way. Move everything to
646 the directory crypto/aes, thereby obsoleting crypto/rijndael.
647 [Stephen Sprunk <stephen@sprunk.org> and Richard Levitte]
649 *) SECURITY: remove unsafe setjmp/signal interaction from ui_openssl.c.
650 [Ben Laurie and Theo de Raadt]
652 *) Add option to output public keys in req command.
653 [Massimiliano Pala madwolf@openca.org]
655 *) Use wNAFs in EC_POINTs_mul() for improved efficiency
656 (up to about 10% better than before for P-192 and P-224).
659 *) New functions/macros
661 SSL_CTX_set_msg_callback(ctx, cb)
662 SSL_CTX_set_msg_callback_arg(ctx, arg)
663 SSL_set_msg_callback(ssl, cb)
664 SSL_set_msg_callback_arg(ssl, arg)
666 to request calling a callback function
668 void cb(int write_p, int version, int content_type,
669 const void *buf, size_t len, SSL *ssl, void *arg)
671 whenever a protocol message has been completely received
672 (write_p == 0) or sent (write_p == 1). Here 'version' is the
673 protocol version according to which the SSL library interprets
674 the current protocol message (SSL2_VERSION, SSL3_VERSION, or
675 TLS1_VERSION). 'content_type' is 0 in the case of SSL 2.0, or
676 the content type as defined in the SSL 3.0/TLS 1.0 protocol
677 specification (change_cipher_spec(20), alert(21), handshake(22)).
678 'buf' and 'len' point to the actual message, 'ssl' to the
679 SSL object, and 'arg' is the application-defined value set by
680 SSL[_CTX]_set_msg_callback_arg().
682 'openssl s_client' and 'openssl s_server' have new '-msg' options
683 to enable a callback that displays all protocol messages.
686 *) Change the shared library support so shared libraries are built as
687 soon as the corresponding static library is finished, and thereby get
688 openssl and the test programs linked against the shared library.
689 This still only happens when the keyword "shard" has been given to
690 the configuration scripts.
692 NOTE: shared library support is still an experimental thing, and
693 backward binary compatibility is still not guaranteed.
694 ["Maciej W. Rozycki" <macro@ds2.pg.gda.pl> and Richard Levitte]
696 *) Add support for Subject Information Access extension.
697 [Peter Sylvester <Peter.Sylvester@EdelWeb.fr>]
699 *) Make BUF_MEM_grow() behaviour more consistent: Initialise to zero
700 additional bytes when new memory had to be allocated, not just
701 when reusing an existing buffer.
704 *) New command line and configuration option 'utf8' for the req command.
705 This allows field values to be specified as UTF8 strings.
708 *) Add -multi and -mr options to "openssl speed" - giving multiple parallel
709 runs for the former and machine-readable output for the latter.
712 *) Add '-noemailDN' option to 'openssl ca'. This prevents inclusion
713 of the e-mail address in the DN (i.e., it will go into a certificate
714 extension only). The new configuration file option 'email_in_dn = no'
716 [Massimiliano Pala madwolf@openca.org]
718 *) Change all functions with names starting with des_ to be starting
719 with DES_ instead. Add wrappers that are compatible with libdes,
720 but are named _ossl_old_des_*. Finally, add macros that map the
721 des_* symbols to the corresponding _ossl_old_des_* if libdes
722 compatibility is desired. If OpenSSL 0.9.6c compatibility is
723 desired, the des_* symbols will be mapped to DES_*, with one
726 Since we provide two compatibility mappings, the user needs to
727 define the macro OPENSSL_DES_LIBDES_COMPATIBILITY if libdes
728 compatibility is desired. The default (i.e., when that macro
729 isn't defined) is OpenSSL 0.9.6c compatibility.
731 There are also macros that enable and disable the support of old
732 des functions altogether. Those are OPENSSL_ENABLE_OLD_DES_SUPPORT
733 and OPENSSL_DISABLE_OLD_DES_SUPPORT. If none or both of those
734 are defined, the default will apply: to support the old des routines.
736 In either case, one must include openssl/des.h to get the correct
737 definitions. Do not try to just include openssl/des_old.h, that
740 NOTE: This is a major break of an old API into a new one. Software
741 authors are encouraged to switch to the DES_ style functions. Some
742 time in the future, des_old.h and the libdes compatibility functions
743 will be disable (i.e. OPENSSL_DISABLE_OLD_DES_SUPPORT will be the
744 default), and then completely removed.
747 *) Test for certificates which contain unsupported critical extensions.
748 If such a certificate is found during a verify operation it is
749 rejected by default: this behaviour can be overridden by either
750 handling the new error X509_V_ERR_UNHANDLED_CRITICAL_EXTENSION or
751 by setting the verify flag X509_V_FLAG_IGNORE_CRITICAL. A new function
752 X509_supported_extension() has also been added which returns 1 if a
753 particular extension is supported.
756 *) Modify the behaviour of EVP cipher functions in similar way to digests
757 to retain compatibility with existing code.
760 *) Modify the behaviour of EVP_DigestInit() and EVP_DigestFinal() to retain
761 compatibility with existing code. In particular the 'ctx' parameter does
762 not have to be to be initialized before the call to EVP_DigestInit() and
763 it is tidied up after a call to EVP_DigestFinal(). New function
764 EVP_DigestFinal_ex() which does not tidy up the ctx. Similarly function
765 EVP_MD_CTX_copy() changed to not require the destination to be
766 initialized valid and new function EVP_MD_CTX_copy_ex() added which
767 requires the destination to be valid.
769 Modify all the OpenSSL digest calls to use EVP_DigestInit_ex(),
770 EVP_DigestFinal_ex() and EVP_MD_CTX_copy_ex().
773 *) Change ssl3_get_message (ssl/s3_both.c) and the functions using it
774 so that complete 'Handshake' protocol structures are kept in memory
775 instead of overwriting 'msg_type' and 'length' with 'body' data.
778 *) Add an implementation of SSL_add_dir_cert_subjects_to_stack for Win32.
779 [Massimo Santin via Richard Levitte]
781 *) Major restructuring to the underlying ENGINE code. This includes
782 reduction of linker bloat, separation of pure "ENGINE" manipulation
783 (initialisation, etc) from functionality dealing with implementations
784 of specific crypto iterfaces. This change also introduces integrated
785 support for symmetric ciphers and digest implementations - so ENGINEs
786 can now accelerate these by providing EVP_CIPHER and EVP_MD
787 implementations of their own. This is detailed in crypto/engine/README
788 as it couldn't be adequately described here. However, there are a few
789 API changes worth noting - some RSA, DSA, DH, and RAND functions that
790 were changed in the original introduction of ENGINE code have now
791 reverted back - the hooking from this code to ENGINE is now a good
792 deal more passive and at run-time, operations deal directly with
793 RSA_METHODs, DSA_METHODs (etc) as they did before, rather than
794 dereferencing through an ENGINE pointer any more. Also, the ENGINE
795 functions dealing with BN_MOD_EXP[_CRT] handlers have been removed -
796 they were not being used by the framework as there is no concept of a
797 BIGNUM_METHOD and they could not be generalised to the new
798 'ENGINE_TABLE' mechanism that underlies the new code. Similarly,
799 ENGINE_cpy() has been removed as it cannot be consistently defined in
803 *) Change ASN1_GENERALIZEDTIME_check() to allow fractional seconds.
806 *) Change mkdef.pl to sort symbols that get the same entry number,
807 and make sure the automatically generated functions ERR_load_*
808 become part of libeay.num as well.
811 *) New function SSL_renegotiate_pending(). This returns true once
812 renegotiation has been requested (either SSL_renegotiate() call
813 or HelloRequest/ClientHello receveived from the peer) and becomes
814 false once a handshake has been completed.
815 (For servers, SSL_renegotiate() followed by SSL_do_handshake()
816 sends a HelloRequest, but does not ensure that a handshake takes
817 place. SSL_renegotiate_pending() is useful for checking if the
818 client has followed the request.)
821 *) New SSL option SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION.
822 By default, clients may request session resumption even during
823 renegotiation (if session ID contexts permit); with this option,
824 session resumption is possible only in the first handshake.
826 SSL_OP_ALL is now 0x00000FFFL instead of 0x000FFFFFL. This makes
827 more bits available for options that should not be part of
828 SSL_OP_ALL (such as SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION).
831 *) Add some demos for certificate and certificate request creation.
834 *) Make maximum certificate chain size accepted from the peer application
835 settable (SSL*_get/set_max_cert_list()), as proposed by
836 "Douglas E. Engert" <deengert@anl.gov>.
839 *) Add support for shared libraries for Unixware-7
840 (Boyd Lynn Gerber <gerberb@zenez.com>).
843 *) Add a "destroy" handler to ENGINEs that allows structural cleanup to
844 be done prior to destruction. Use this to unload error strings from
845 ENGINEs that load their own error strings. NB: This adds two new API
846 functions to "get" and "set" this destroy handler in an ENGINE.
849 *) Alter all existing ENGINE implementations (except "openssl" and
850 "openbsd") to dynamically instantiate their own error strings. This
851 makes them more flexible to be built both as statically-linked ENGINEs
852 and self-contained shared-libraries loadable via the "dynamic" ENGINE.
853 Also, add stub code to each that makes building them as self-contained
854 shared-libraries easier (see README.ENGINE).
857 *) Add a "dynamic" ENGINE that provides a mechanism for binding ENGINE
858 implementations into applications that are completely implemented in
859 self-contained shared-libraries. The "dynamic" ENGINE exposes control
860 commands that can be used to configure what shared-library to load and
861 to control aspects of the way it is handled. Also, made an update to
862 the README.ENGINE file that brings its information up-to-date and
863 provides some information and instructions on the "dynamic" ENGINE
864 (ie. how to use it, how to build "dynamic"-loadable ENGINEs, etc).
867 *) Make it possible to unload ranges of ERR strings with a new
868 "ERR_unload_strings" function.
871 *) Add a copy() function to EVP_MD.
874 *) Make EVP_MD routines take a context pointer instead of just the
875 md_data void pointer.
878 *) Add flags to EVP_MD and EVP_MD_CTX. EVP_MD_FLAG_ONESHOT indicates
879 that the digest can only process a single chunk of data
880 (typically because it is provided by a piece of
881 hardware). EVP_MD_CTX_FLAG_ONESHOT indicates that the application
882 is only going to provide a single chunk of data, and hence the
883 framework needn't accumulate the data for oneshot drivers.
886 *) As with "ERR", make it possible to replace the underlying "ex_data"
887 functions. This change also alters the storage and management of global
888 ex_data state - it's now all inside ex_data.c and all "class" code (eg.
889 RSA, BIO, SSL_CTX, etc) no longer stores its own STACKS and per-class
890 index counters. The API functions that use this state have been changed
891 to take a "class_index" rather than pointers to the class's local STACK
892 and counter, and there is now an API function to dynamically create new
893 classes. This centralisation allows us to (a) plug a lot of the
894 thread-safety problems that existed, and (b) makes it possible to clean
895 up all allocated state using "CRYPTO_cleanup_all_ex_data()". W.r.t. (b)
896 such data would previously have always leaked in application code and
897 workarounds were in place to make the memory debugging turn a blind eye
898 to it. Application code that doesn't use this new function will still
899 leak as before, but their memory debugging output will announce it now
900 rather than letting it slide.
902 Besides the addition of CRYPTO_cleanup_all_ex_data(), another API change
903 induced by the "ex_data" overhaul is that X509_STORE_CTX_init() now
904 has a return value to indicate success or failure.
907 *) Make it possible to replace the underlying "ERR" functions such that the
908 global state (2 LHASH tables and 2 locks) is only used by the "default"
909 implementation. This change also adds two functions to "get" and "set"
910 the implementation prior to it being automatically set the first time
911 any other ERR function takes place. Ie. an application can call "get",
912 pass the return value to a module it has just loaded, and that module
913 can call its own "set" function using that value. This means the
914 module's "ERR" operations will use (and modify) the error state in the
915 application and not in its own statically linked copy of OpenSSL code.
918 *) Give DH, DSA, and RSA types their own "**_up_ref()" function to increment
919 reference counts. This performs normal REF_PRINT/REF_CHECK macros on
920 the operation, and provides a more encapsulated way for external code
921 (crypto/evp/ and ssl/) to do this. Also changed the evp and ssl code
922 to use these functions rather than manually incrementing the counts.
924 Also rename "DSO_up()" function to more descriptive "DSO_up_ref()".
927 *) Add EVP test program.
930 *) Add symmetric cipher support to ENGINE. Expect the API to change!
933 *) New CRL functions: X509_CRL_set_version(), X509_CRL_set_issuer_name()
934 X509_CRL_set_lastUpdate(), X509_CRL_set_nextUpdate(), X509_CRL_sort(),
935 X509_REVOKED_set_serialNumber(), and X509_REVOKED_set_revocationDate().
936 These allow a CRL to be built without having to access X509_CRL fields
937 directly. Modify 'ca' application to use new functions.
940 *) Move SSL_OP_TLS_ROLLBACK_BUG out of the SSL_OP_ALL list of recommended
941 bug workarounds. Rollback attack detection is a security feature.
942 The problem will only arise on OpenSSL servers when TLSv1 is not
943 available (sslv3_server_method() or SSL_OP_NO_TLSv1).
944 Software authors not wanting to support TLSv1 will have special reasons
945 for their choice and can explicitly enable this option.
946 [Bodo Moeller, Lutz Jaenicke]
948 *) Rationalise EVP so it can be extended: don't include a union of
949 cipher/digest structures, add init/cleanup functions for EVP_MD_CTX
950 (similar to those existing for EVP_CIPHER_CTX).
955 EVP_MD_CTX_init(&md); /* new function call */
956 EVP_DigestInit(&md, EVP_sha1());
957 EVP_DigestUpdate(&md, in, len);
958 EVP_DigestFinal(&md, out, NULL);
959 EVP_MD_CTX_cleanup(&md); /* new function call */
963 *) Make DES key schedule conform to the usual scheme, as well as
964 correcting its structure. This means that calls to DES functions
965 now have to pass a pointer to a des_key_schedule instead of a
966 plain des_key_schedule (which was actually always a pointer
971 des_set_key_checked(..., &ks);
972 des_ncbc_encrypt(..., &ks, ...);
974 (Note that a later change renames 'des_...' into 'DES_...'.)
977 *) Initial reduction of linker bloat: the use of some functions, such as
978 PEM causes large amounts of unused functions to be linked in due to
979 poor organisation. For example pem_all.c contains every PEM function
980 which has a knock on effect of linking in large amounts of (unused)
981 ASN1 code. Grouping together similar functions and splitting unrelated
982 functions prevents this.
985 *) Cleanup of EVP macros.
988 *) Change historical references to {NID,SN,LN}_des_ede and ede3 to add the
992 *) Add initial OCSP responder support to ocsp application. The
993 revocation information is handled using the text based index
994 use by the ca application. The responder can either handle
995 requests generated internally, supplied in files (for example
996 via a CGI script) or using an internal minimal server.
999 *) Add configuration choices to get zlib compression for TLS.
1002 *) Changes to Kerberos SSL for RFC 2712 compliance:
1003 1. Implemented real KerberosWrapper, instead of just using
1004 KRB5 AP_REQ message. [Thanks to Simon Wilkinson <sxw@sxw.org.uk>]
1005 2. Implemented optional authenticator field of KerberosWrapper.
1007 Added openssl-style ASN.1 macros for Kerberos ticket, ap_req,
1008 and authenticator structs; see crypto/krb5/.
1010 Generalized Kerberos calls to support multiple Kerberos libraries.
1011 [Vern Staats <staatsvr@asc.hpc.mil>,
1012 Jeffrey Altman <jaltman@columbia.edu>
1013 via Richard Levitte]
1015 *) Cause 'openssl speed' to use fully hard-coded DSA keys as it
1016 already does with RSA. testdsa.h now has 'priv_key/pub_key'
1017 values for each of the key sizes rather than having just
1018 parameters (and 'speed' generating keys each time).
1021 *) Speed up EVP routines.
1024 type 8 bytes 64 bytes 256 bytes 1024 bytes 8192 bytes
1025 des-cbc 4408.85k 5560.51k 5778.46k 5862.20k 5825.16k
1026 des-cbc 4389.55k 5571.17k 5792.23k 5846.91k 5832.11k
1027 des-cbc 4394.32k 5575.92k 5807.44k 5848.37k 5841.30k
1029 des-cbc 3482.66k 5069.49k 5496.39k 5614.16k 5639.28k
1030 des-cbc 3480.74k 5068.76k 5510.34k 5609.87k 5635.52k
1031 des-cbc 3483.72k 5067.62k 5504.60k 5708.01k 5724.80k
1034 des-cbc 4660.16k 5650.19k 5807.19k 5827.13k 5783.32k
1036 des-cbc 3624.96k 5258.21k 5530.91k 5624.30k 5628.26k
1039 *) Added the OS2-EMX target.
1040 ["Brian Havard" <brianh@kheldar.apana.org.au> and Richard Levitte]
1042 *) Rewrite apps to use NCONF routines instead of the old CONF. New functions
1043 to support NCONF routines in extension code. New function CONF_set_nconf()
1044 to allow functions which take an NCONF to also handle the old LHASH
1045 structure: this means that the old CONF compatible routines can be
1046 retained (in particular wrt extensions) without having to duplicate the
1047 code. New function X509V3_add_ext_nconf_sk to add extensions to a stack.
1050 *) Enhance the general user interface with mechanisms for inner control
1051 and with possibilities to have yes/no kind of prompts.
1054 *) Change all calls to low level digest routines in the library and
1055 applications to use EVP. Add missing calls to HMAC_cleanup() and
1056 don't assume HMAC_CTX can be copied using memcpy().
1057 [Verdon Walker <VWalker@novell.com>, Steve Henson]
1059 *) Add the possibility to control engines through control names but with
1060 arbitrary arguments instead of just a string.
1061 Change the key loaders to take a UI_METHOD instead of a callback
1062 function pointer. NOTE: this breaks binary compatibility with earlier
1063 versions of OpenSSL [engine].
1064 Adapt the nCipher code for these new conditions and add a card insertion
1068 *) Enhance the general user interface with mechanisms to better support
1069 dialog box interfaces, application-defined prompts, the possibility
1070 to use defaults (for example default passwords from somewhere else)
1071 and interrupts/cancellations.
1074 *) Tidy up PKCS#12 attribute handling. Add support for the CSP name
1075 attribute in PKCS#12 files, add new -CSP option to pkcs12 utility.
1078 *) Fix a memory leak in 'sk_dup()' in the case reallocation fails. (Also
1079 tidy up some unnecessarily weird code in 'sk_new()').
1080 [Geoff, reported by Diego Tartara <dtartara@novamens.com>]
1082 *) Change the key loading routines for ENGINEs to use the same kind
1083 callback (pem_password_cb) as all other routines that need this
1087 *) Increase ENTROPY_NEEDED to 32 bytes, as Rijndael can operate with
1088 256 bit (=32 byte) keys. Of course seeding with more entropy bytes
1089 than this minimum value is recommended.
1092 *) New random seeder for OpenVMS, using the system process statistics
1093 that are easily reachable.
1096 *) Windows apparently can't transparently handle global
1097 variables defined in DLLs. Initialisations such as:
1099 const ASN1_ITEM *it = &ASN1_INTEGER_it;
1101 wont compile. This is used by the any applications that need to
1102 declare their own ASN1 modules. This was fixed by adding the option
1103 EXPORT_VAR_AS_FN to all Win32 platforms, although this isn't strictly
1104 needed for static libraries under Win32.
1107 *) New functions X509_PURPOSE_set() and X509_TRUST_set() to handle
1108 setting of purpose and trust fields. New X509_STORE trust and
1109 purpose functions and tidy up setting in other SSL functions.
1112 *) Add copies of X509_STORE_CTX fields and callbacks to X509_STORE
1113 structure. These are inherited by X509_STORE_CTX when it is
1114 initialised. This allows various defaults to be set in the
1115 X509_STORE structure (such as flags for CRL checking and custom
1116 purpose or trust settings) for functions which only use X509_STORE_CTX
1117 internally such as S/MIME.
1119 Modify X509_STORE_CTX_purpose_inherit() so it only sets purposes and
1120 trust settings if they are not set in X509_STORE. This allows X509_STORE
1121 purposes and trust (in S/MIME for example) to override any set by default.
1123 Add command line options for CRL checking to smime, s_client and s_server
1127 *) Initial CRL based revocation checking. If the CRL checking flag(s)
1128 are set then the CRL is looked up in the X509_STORE structure and
1129 its validity and signature checked, then if the certificate is found
1130 in the CRL the verify fails with a revoked error.
1132 Various new CRL related callbacks added to X509_STORE_CTX structure.
1134 Command line options added to 'verify' application to support this.
1136 This needs some additional work, such as being able to handle multiple
1137 CRLs with different times, extension based lookup (rather than just
1138 by subject name) and ultimately more complete V2 CRL extension
1142 *) Add a general user interface API (crypto/ui/). This is designed
1143 to replace things like des_read_password and friends (backward
1144 compatibility functions using this new API are provided).
1145 The purpose is to remove prompting functions from the DES code
1146 section as well as provide for prompting through dialog boxes in
1147 a window system and the like.
1150 *) Add "ex_data" support to ENGINE so implementations can add state at a
1151 per-structure level rather than having to store it globally.
1154 *) Make it possible for ENGINE structures to be copied when retrieved by
1155 ENGINE_by_id() if the ENGINE specifies a new flag: ENGINE_FLAGS_BY_ID_COPY.
1156 This causes the "original" ENGINE structure to act like a template,
1157 analogous to the RSA vs. RSA_METHOD type of separation. Because of this
1158 operational state can be localised to each ENGINE structure, despite the
1159 fact they all share the same "methods". New ENGINE structures returned in
1160 this case have no functional references and the return value is the single
1161 structural reference. This matches the single structural reference returned
1162 by ENGINE_by_id() normally, when it is incremented on the pre-existing
1166 *) Fix ASN1 decoder when decoding type ANY and V_ASN1_OTHER: since this
1167 needs to match any other type at all we need to manually clear the
1171 *) Changes to the "openssl engine" utility to include;
1172 - verbosity levels ('-v', '-vv', and '-vvv') that provide information
1173 about an ENGINE's available control commands.
1174 - executing control commands from command line arguments using the
1175 '-pre' and '-post' switches. '-post' is only used if '-t' is
1176 specified and the ENGINE is successfully initialised. The syntax for
1177 the individual commands are colon-separated, for example;
1178 openssl engine chil -pre FORK_CHECK:0 -pre SO_PATH:/lib/test.so
1181 *) New dynamic control command support for ENGINEs. ENGINEs can now
1182 declare their own commands (numbers), names (strings), descriptions,
1183 and input types for run-time discovery by calling applications. A
1184 subset of these commands are implicitly classed as "executable"
1185 depending on their input type, and only these can be invoked through
1186 the new string-based API function ENGINE_ctrl_cmd_string(). (Eg. this
1187 can be based on user input, config files, etc). The distinction is
1188 that "executable" commands cannot return anything other than a boolean
1189 result and can only support numeric or string input, whereas some
1190 discoverable commands may only be for direct use through
1191 ENGINE_ctrl(), eg. supporting the exchange of binary data, function
1192 pointers, or other custom uses. The "executable" commands are to
1193 support parameterisations of ENGINE behaviour that can be
1194 unambiguously defined by ENGINEs and used consistently across any
1195 OpenSSL-based application. Commands have been added to all the
1196 existing hardware-supporting ENGINEs, noticeably "SO_PATH" to allow
1197 control over shared-library paths without source code alterations.
1200 *) Changed all ENGINE implementations to dynamically allocate their
1201 ENGINEs rather than declaring them statically. Apart from this being
1202 necessary with the removal of the ENGINE_FLAGS_MALLOCED distinction,
1203 this also allows the implementations to compile without using the
1204 internal engine_int.h header.
1207 *) Minor adjustment to "rand" code. RAND_get_rand_method() now returns a
1208 'const' value. Any code that should be able to modify a RAND_METHOD
1209 should already have non-const pointers to it (ie. they should only
1210 modify their own ones).
1213 *) Made a variety of little tweaks to the ENGINE code.
1214 - "atalla" and "ubsec" string definitions were moved from header files
1215 to C code. "nuron" string definitions were placed in variables
1216 rather than hard-coded - allowing parameterisation of these values
1217 later on via ctrl() commands.
1218 - Removed unused "#if 0"'d code.
1219 - Fixed engine list iteration code so it uses ENGINE_free() to release
1220 structural references.
1221 - Constified the RAND_METHOD element of ENGINE structures.
1222 - Constified various get/set functions as appropriate and added
1223 missing functions (including a catch-all ENGINE_cpy that duplicates
1224 all ENGINE values onto a new ENGINE except reference counts/state).
1225 - Removed NULL parameter checks in get/set functions. Setting a method
1226 or function to NULL is a way of cancelling out a previously set
1227 value. Passing a NULL ENGINE parameter is just plain stupid anyway
1228 and doesn't justify the extra error symbols and code.
1229 - Deprecate the ENGINE_FLAGS_MALLOCED define and move the area for
1230 flags from engine_int.h to engine.h.
1231 - Changed prototypes for ENGINE handler functions (init(), finish(),
1232 ctrl(), key-load functions, etc) to take an (ENGINE*) parameter.
1235 *) Implement binary inversion algorithm for BN_mod_inverse in addition
1236 to the algorithm using long division. The binary algorithm can be
1237 used only if the modulus is odd. On 32-bit systems, it is faster
1238 only for relatively small moduli (roughly 20-30% for 128-bit moduli,
1239 roughly 5-15% for 256-bit moduli), so we use it only for moduli
1240 up to 450 bits. In 64-bit environments, the binary algorithm
1241 appears to be advantageous for much longer moduli; here we use it
1242 for moduli up to 2048 bits.
1245 *) Rewrite CHOICE field setting in ASN1_item_ex_d2i(). The old code
1246 could not support the combine flag in choice fields.
1249 *) Add a 'copy_extensions' option to the 'ca' utility. This copies
1250 extensions from a certificate request to the certificate.
1253 *) Allow multiple 'certopt' and 'nameopt' options to be separated
1254 by commas. Add 'namopt' and 'certopt' options to the 'ca' config
1255 file: this allows the display of the certificate about to be
1256 signed to be customised, to allow certain fields to be included
1257 or excluded and extension details. The old system didn't display
1258 multicharacter strings properly, omitted fields not in the policy
1259 and couldn't display additional details such as extensions.
1262 *) Function EC_POINTs_mul for multiple scalar multiplication
1263 of an arbitrary number of elliptic curve points
1264 \sum scalars[i]*points[i],
1265 optionally including the generator defined for the EC_GROUP:
1266 scalar*generator + \sum scalars[i]*points[i].
1268 EC_POINT_mul is a simple wrapper function for the typical case
1269 that the point list has just one item (besides the optional
1273 *) First EC_METHODs for curves over GF(p):
1275 EC_GFp_simple_method() uses the basic BN_mod_mul and BN_mod_sqr
1276 operations and provides various method functions that can also
1277 operate with faster implementations of modular arithmetic.
1279 EC_GFp_mont_method() reuses most functions that are part of
1280 EC_GFp_simple_method, but uses Montgomery arithmetic.
1282 [Bodo Moeller; point addition and point doubling
1283 implementation directly derived from source code provided by
1284 Lenka Fibikova <fibikova@exp-math.uni-essen.de>]
1286 *) Framework for elliptic curves (crypto/ec/ec.h, crypto/ec/ec_lcl.h,
1287 crypto/ec/ec_lib.c):
1289 Curves are EC_GROUP objects (with an optional group generator)
1290 based on EC_METHODs that are built into the library.
1292 Points are EC_POINT objects based on EC_GROUP objects.
1294 Most of the framework would be able to handle curves over arbitrary
1295 finite fields, but as there are no obvious types for fields other
1296 than GF(p), some functions are limited to that for now.
1299 *) Add the -HTTP option to s_server. It is similar to -WWW, but requires
1300 that the file contains a complete HTTP response.
1303 *) Add the ec directory to mkdef.pl and mkfiles.pl. In mkdef.pl
1304 change the def and num file printf format specifier from "%-40sXXX"
1305 to "%-39s XXX". The latter will always guarantee a space after the
1306 field while the former will cause them to run together if the field
1307 is 40 of more characters long.
1310 *) Constify the cipher and digest 'method' functions and structures
1311 and modify related functions to take constant EVP_MD and EVP_CIPHER
1315 *) Hide BN_CTX structure details in bn_lcl.h instead of publishing them
1316 in <openssl/bn.h>. Also further increase BN_CTX_NUM to 32.
1319 *) Modify EVP_Digest*() routines so they now return values. Although the
1320 internal software routines can never fail additional hardware versions
1324 *) Clean up crypto/err/err.h and change some error codes to avoid conflicts:
1326 Previously ERR_R_FATAL was too small and coincided with ERR_LIB_PKCS7
1327 (= ERR_R_PKCS7_LIB); it is now 64 instead of 32.
1330 ERR_R_NESTED_ASN1_ERROR
1332 ERR_R_MISSING_ASN1_EOS
1333 were 4 .. 9, conflicting with
1334 ERR_LIB_RSA (= ERR_R_RSA_LIB)
1336 ERR_LIB_PEM (= ERR_R_PEM_LIB).
1337 They are now 58 .. 63 (i.e., just below ERR_R_FATAL).
1339 Add new error code 'ERR_R_INTERNAL_ERROR'.
1342 *) Don't overuse locks in crypto/err/err.c: For data retrieval, CRYPTO_r_lock
1346 *) New option '-subj arg' for 'openssl req' and 'openssl ca'. This
1347 sets the subject name for a new request or supersedes the
1348 subject name in a given request. Formats that can be parsed are
1349 'CN=Some Name, OU=myOU, C=IT'
1351 'CN=Some Name/OU=myOU/C=IT'.
1353 Add options '-batch' and '-verbose' to 'openssl req'.
1354 [Massimiliano Pala <madwolf@hackmasters.net>]
1356 *) Introduce the possibility to access global variables through
1357 functions on platform were that's the best way to handle exporting
1358 global variables in shared libraries. To enable this functionality,
1359 one must configure with "EXPORT_VAR_AS_FN" or defined the C macro
1360 "OPENSSL_EXPORT_VAR_AS_FUNCTION" in crypto/opensslconf.h (the latter
1361 is normally done by Configure or something similar).
1363 To implement a global variable, use the macro OPENSSL_IMPLEMENT_GLOBAL
1364 in the source file (foo.c) like this:
1366 OPENSSL_IMPLEMENT_GLOBAL(int,foo)=1;
1367 OPENSSL_IMPLEMENT_GLOBAL(double,bar);
1369 To declare a global variable, use the macros OPENSSL_DECLARE_GLOBAL
1370 and OPENSSL_GLOBAL_REF in the header file (foo.h) like this:
1372 OPENSSL_DECLARE_GLOBAL(int,foo);
1373 #define foo OPENSSL_GLOBAL_REF(foo)
1374 OPENSSL_DECLARE_GLOBAL(double,bar);
1375 #define bar OPENSSL_GLOBAL_REF(bar)
1377 The #defines are very important, and therefore so is including the
1378 header file everywhere where the defined globals are used.
1380 The macro OPENSSL_EXPORT_VAR_AS_FUNCTION also affects the definition
1381 of ASN.1 items, but that structure is a bit different.
1383 The largest change is in util/mkdef.pl which has been enhanced with
1384 better and easier to understand logic to choose which symbols should
1385 go into the Windows .def files as well as a number of fixes and code
1386 cleanup (among others, algorithm keywords are now sorted
1387 lexicographically to avoid constant rewrites).
1390 *) In BN_div() keep a copy of the sign of 'num' before writing the
1391 result to 'rm' because if rm==num the value will be overwritten
1392 and produce the wrong result if 'num' is negative: this caused
1393 problems with BN_mod() and BN_nnmod().
1396 *) Function OCSP_request_verify(). This checks the signature on an
1397 OCSP request and verifies the signer certificate. The signer
1398 certificate is just checked for a generic purpose and OCSP request
1402 *) Add OCSP_check_validity() function to check the validity of OCSP
1403 responses. OCSP responses are prepared in real time and may only
1404 be a few seconds old. Simply checking that the current time lies
1405 between thisUpdate and nextUpdate max reject otherwise valid responses
1406 caused by either OCSP responder or client clock inaccuracy. Instead
1407 we allow thisUpdate and nextUpdate to fall within a certain period of
1408 the current time. The age of the response can also optionally be
1409 checked. Two new options -validity_period and -status_age added to
1413 *) If signature or public key algorithm is unrecognized print out its
1414 OID rather that just UNKNOWN.
1417 *) Change OCSP_cert_to_id() to tolerate a NULL subject certificate and
1418 OCSP_cert_id_new() a NULL serialNumber. This allows a partial certificate
1419 ID to be generated from the issuer certificate alone which can then be
1420 passed to OCSP_id_issuer_cmp().
1423 *) New compilation option ASN1_ITEM_FUNCTIONS. This causes the new
1424 ASN1 modules to export functions returning ASN1_ITEM pointers
1425 instead of the ASN1_ITEM structures themselves. This adds several
1426 new macros which allow the underlying ASN1 function/structure to
1427 be accessed transparently. As a result code should not use ASN1_ITEM
1428 references directly (such as &X509_it) but instead use the relevant
1429 macros (such as ASN1_ITEM_rptr(X509)). This option is to allow
1430 use of the new ASN1 code on platforms where exporting structures
1431 is problematical (for example in shared libraries) but exporting
1432 functions returning pointers to structures is not.
1435 *) Add support for overriding the generation of SSL/TLS session IDs.
1436 These callbacks can be registered either in an SSL_CTX or per SSL.
1437 The purpose of this is to allow applications to control, if they wish,
1438 the arbitrary values chosen for use as session IDs, particularly as it
1439 can be useful for session caching in multiple-server environments. A
1440 command-line switch for testing this (and any client code that wishes
1441 to use such a feature) has been added to "s_server".
1442 [Geoff Thorpe, Lutz Jaenicke]
1444 *) Modify mkdef.pl to recognise and parse preprocessor conditionals
1445 of the form '#if defined(...) || defined(...) || ...' and
1446 '#if !defined(...) && !defined(...) && ...'. This also avoids
1447 the growing number of special cases it was previously handling.
1450 *) Make all configuration macros available for application by making
1451 sure they are available in opensslconf.h, by giving them names starting
1452 with "OPENSSL_" to avoid conflicts with other packages and by making
1453 sure e_os2.h will cover all platform-specific cases together with
1455 Additionally, it is now possible to define configuration/platform-
1456 specific names (called "system identities"). In the C code, these
1457 are prefixed with "OPENSSL_SYSNAME_". e_os2.h will create another
1458 macro with the name beginning with "OPENSSL_SYS_", which is determined
1459 from "OPENSSL_SYSNAME_*" or compiler-specific macros depending on
1463 *) New option -set_serial to 'req' and 'x509' this allows the serial
1464 number to use to be specified on the command line. Previously self
1465 signed certificates were hard coded with serial number 0 and the
1466 CA options of 'x509' had to use a serial number in a file which was
1470 *) New options to 'ca' utility to support V2 CRL entry extensions.
1471 Currently CRL reason, invalidity date and hold instruction are
1472 supported. Add new CRL extensions to V3 code and some new objects.
1475 *) New function EVP_CIPHER_CTX_set_padding() this is used to
1476 disable standard block padding (aka PKCS#5 padding) in the EVP
1477 API, which was previously mandatory. This means that the data is
1478 not padded in any way and so the total length much be a multiple
1479 of the block size, otherwise an error occurs.
1482 *) Initial (incomplete) OCSP SSL support.
1485 *) New function OCSP_parse_url(). This splits up a URL into its host,
1486 port and path components: primarily to parse OCSP URLs. New -url
1487 option to ocsp utility.
1490 *) New nonce behavior. The return value of OCSP_check_nonce() now
1491 reflects the various checks performed. Applications can decide
1492 whether to tolerate certain situations such as an absent nonce
1493 in a response when one was present in a request: the ocsp application
1494 just prints out a warning. New function OCSP_add1_basic_nonce()
1495 this is to allow responders to include a nonce in a response even if
1496 the request is nonce-less.
1499 *) Disable stdin buffering in load_cert (apps/apps.c) so that no certs are
1500 skipped when using openssl x509 multiple times on a single input file,
1501 e.g. "(openssl x509 -out cert1; openssl x509 -out cert2) <certs".
1504 *) Make ASN1_UTCTIME_set_string() and ASN1_GENERALIZEDTIME_set_string()
1505 set string type: to handle setting ASN1_TIME structures. Fix ca
1506 utility to correctly initialize revocation date of CRLs.
1509 *) New option SSL_OP_CIPHER_SERVER_PREFERENCE allows the server to override
1510 the clients preferred ciphersuites and rather use its own preferences.
1511 Should help to work around M$ SGC (Server Gated Cryptography) bug in
1512 Internet Explorer by ensuring unchanged hash method during stepup.
1513 (Also replaces the broken/deactivated SSL_OP_NON_EXPORT_FIRST option.)
1516 *) Make mkdef.pl recognise all DECLARE_ASN1 macros, change rijndael
1517 to aes and add a new 'exist' option to print out symbols that don't
1521 *) Additional options to ocsp utility to allow flags to be set and
1522 additional certificates supplied.
1525 *) Add the option -VAfile to 'openssl ocsp', so the user can give the
1526 OCSP client a number of certificate to only verify the response
1530 *) Update Rijndael code to version 3.0 and change EVP AES ciphers to
1531 handle the new API. Currently only ECB, CBC modes supported. Add new
1534 Add TLS AES ciphersuites as described in RFC3268, "Advanced
1535 Encryption Standard (AES) Ciphersuites for Transport Layer
1536 Security (TLS)". (In beta versions of OpenSSL 0.9.7, these were
1537 not enabled by default and were not part of the "ALL" ciphersuite
1538 alias because they were not yet official; they could be
1539 explicitly requested by specifying the "AESdraft" ciphersuite
1540 group alias. In the final release of OpenSSL 0.9.7, the group
1541 alias is called "AES" and is part of "ALL".)
1542 [Ben Laurie, Steve Henson, Bodo Moeller]
1544 *) New function OCSP_copy_nonce() to copy nonce value (if present) from
1545 request to response.
1548 *) Functions for OCSP responders. OCSP_request_onereq_count(),
1549 OCSP_request_onereq_get0(), OCSP_onereq_get0_id() and OCSP_id_get0_info()
1550 extract information from a certificate request. OCSP_response_create()
1551 creates a response and optionally adds a basic response structure.
1552 OCSP_basic_add1_status() adds a complete single response to a basic
1553 response and returns the OCSP_SINGLERESP structure just added (to allow
1554 extensions to be included for example). OCSP_basic_add1_cert() adds a
1555 certificate to a basic response and OCSP_basic_sign() signs a basic
1556 response with various flags. New helper functions ASN1_TIME_check()
1557 (checks validity of ASN1_TIME structure) and ASN1_TIME_to_generalizedtime()
1558 (converts ASN1_TIME to GeneralizedTime).
1561 *) Various new functions. EVP_Digest() combines EVP_Digest{Init,Update,Final}()
1562 in a single operation. X509_get0_pubkey_bitstr() extracts the public_key
1563 structure from a certificate. X509_pubkey_digest() digests the public_key
1564 contents: this is used in various key identifiers.
1567 *) Make sk_sort() tolerate a NULL argument.
1568 [Steve Henson reported by Massimiliano Pala <madwolf@comune.modena.it>]
1570 *) New OCSP verify flag OCSP_TRUSTOTHER. When set the "other" certificates
1571 passed by the function are trusted implicitly. If any of them signed the
1572 response then it is assumed to be valid and is not verified.
1575 *) In PKCS7_set_type() initialise content_type in PKCS7_ENC_CONTENT
1576 to data. This was previously part of the PKCS7 ASN1 code. This
1577 was causing problems with OpenSSL created PKCS#12 and PKCS#7 structures.
1578 [Steve Henson, reported by Kenneth R. Robinette
1579 <support@securenetterm.com>]
1581 *) Add CRYPTO_push_info() and CRYPTO_pop_info() calls to new ASN1
1582 routines: without these tracing memory leaks is very painful.
1583 Fix leaks in PKCS12 and PKCS7 routines.
1586 *) Make X509_time_adj() cope with the new behaviour of ASN1_TIME_new().
1587 Previously it initialised the 'type' argument to V_ASN1_UTCTIME which
1588 effectively meant GeneralizedTime would never be used. Now it
1589 is initialised to -1 but X509_time_adj() now has to check the value
1590 and use ASN1_TIME_set() if the value is not V_ASN1_UTCTIME or
1591 V_ASN1_GENERALIZEDTIME, without this it always uses GeneralizedTime.
1592 [Steve Henson, reported by Kenneth R. Robinette
1593 <support@securenetterm.com>]
1595 *) Fixes to BN_to_ASN1_INTEGER when bn is zero. This would previously
1596 result in a zero length in the ASN1_INTEGER structure which was
1597 not consistent with the structure when d2i_ASN1_INTEGER() was used
1598 and would cause ASN1_INTEGER_cmp() to fail. Enhance s2i_ASN1_INTEGER()
1599 to cope with hex and negative integers. Fix bug in i2a_ASN1_INTEGER()
1600 where it did not print out a minus for negative ASN1_INTEGER.
1603 *) Add summary printout to ocsp utility. The various functions which
1604 convert status values to strings have been renamed to:
1605 OCSP_response_status_str(), OCSP_cert_status_str() and
1606 OCSP_crl_reason_str() and are no longer static. New options
1607 to verify nonce values and to disable verification. OCSP response
1608 printout format cleaned up.
1611 *) Add additional OCSP certificate checks. These are those specified
1612 in RFC2560. This consists of two separate checks: the CA of the
1613 certificate being checked must either be the OCSP signer certificate
1614 or the issuer of the OCSP signer certificate. In the latter case the
1615 OCSP signer certificate must contain the OCSP signing extended key
1616 usage. This check is performed by attempting to match the OCSP
1617 signer or the OCSP signer CA to the issuerNameHash and issuerKeyHash
1618 in the OCSP_CERTID structures of the response.
1621 *) Initial OCSP certificate verification added to OCSP_basic_verify()
1622 and related routines. This uses the standard OpenSSL certificate
1623 verify routines to perform initial checks (just CA validity) and
1624 to obtain the certificate chain. Then additional checks will be
1625 performed on the chain. Currently the root CA is checked to see
1626 if it is explicitly trusted for OCSP signing. This is used to set
1627 a root CA as a global signing root: that is any certificate that
1628 chains to that CA is an acceptable OCSP signing certificate.
1631 *) New '-extfile ...' option to 'openssl ca' for reading X.509v3
1632 extensions from a separate configuration file.
1633 As when reading extensions from the main configuration file,
1634 the '-extensions ...' option may be used for specifying the
1636 [Massimiliano Pala <madwolf@comune.modena.it>]
1638 *) New OCSP utility. Allows OCSP requests to be generated or
1639 read. The request can be sent to a responder and the output
1640 parsed, outputed or printed in text form. Not complete yet:
1641 still needs to check the OCSP response validity.
1644 *) New subcommands for 'openssl ca':
1645 'openssl ca -status <serial>' prints the status of the cert with
1646 the given serial number (according to the index file).
1647 'openssl ca -updatedb' updates the expiry status of certificates
1649 [Massimiliano Pala <madwolf@comune.modena.it>]
1651 *) New '-newreq-nodes' command option to CA.pl. This is like
1652 '-newreq', but calls 'openssl req' with the '-nodes' option
1653 so that the resulting key is not encrypted.
1654 [Damien Miller <djm@mindrot.org>]
1656 *) New configuration for the GNU Hurd.
1657 [Jonathan Bartlett <johnnyb@wolfram.com> via Richard Levitte]
1659 *) Initial code to implement OCSP basic response verify. This
1660 is currently incomplete. Currently just finds the signer's
1661 certificate and verifies the signature on the response.
1664 *) New SSLeay_version code SSLEAY_DIR to determine the compiled-in
1665 value of OPENSSLDIR. This is available via the new '-d' option
1666 to 'openssl version', and is also included in 'openssl version -a'.
1669 *) Allowing defining memory allocation callbacks that will be given
1670 file name and line number information in additional arguments
1671 (a const char* and an int). The basic functionality remains, as
1672 well as the original possibility to just replace malloc(),
1673 realloc() and free() by functions that do not know about these
1674 additional arguments. To register and find out the current
1675 settings for extended allocation functions, the following
1676 functions are provided:
1678 CRYPTO_set_mem_ex_functions
1679 CRYPTO_set_locked_mem_ex_functions
1680 CRYPTO_get_mem_ex_functions
1681 CRYPTO_get_locked_mem_ex_functions
1683 These work the same way as CRYPTO_set_mem_functions and friends.
1684 CRYPTO_get_[locked_]mem_functions now writes 0 where such an
1685 extended allocation function is enabled.
1686 Similarly, CRYPTO_get_[locked_]mem_ex_functions writes 0 where
1687 a conventional allocation function is enabled.
1688 [Richard Levitte, Bodo Moeller]
1690 *) Finish off removing the remaining LHASH function pointer casts.
1691 There should no longer be any prototype-casting required when using
1692 the LHASH abstraction, and any casts that remain are "bugs". See
1693 the callback types and macros at the head of lhash.h for details
1694 (and "OBJ_cleanup" in crypto/objects/obj_dat.c as an example).
1697 *) Add automatic query of EGD sockets in RAND_poll() for the unix variant.
1698 If /dev/[u]random devices are not available or do not return enough
1699 entropy, EGD style sockets (served by EGD or PRNGD) will automatically
1701 The locations /var/run/egd-pool, /dev/egd-pool, /etc/egd-pool, and
1702 /etc/entropy will be queried once each in this sequence, quering stops
1703 when enough entropy was collected without querying more sockets.
1706 *) Change the Unix RAND_poll() variant to be able to poll several
1707 random devices, as specified by DEVRANDOM, until a sufficient amount
1708 of data has been collected. We spend at most 10 ms on each file
1709 (select timeout) and read in non-blocking mode. DEVRANDOM now
1710 defaults to the list "/dev/urandom", "/dev/random", "/dev/srandom"
1711 (previously it was just the string "/dev/urandom"), so on typical
1712 platforms the 10 ms delay will never occur.
1713 Also separate out the Unix variant to its own file, rand_unix.c.
1714 For VMS, there's a currently-empty rand_vms.c.
1717 *) Move OCSP client related routines to ocsp_cl.c. These
1718 provide utility functions which an application needing
1719 to issue a request to an OCSP responder and analyse the
1720 response will typically need: as opposed to those which an
1721 OCSP responder itself would need which will be added later.
1723 OCSP_request_sign() signs an OCSP request with an API similar
1724 to PKCS7_sign(). OCSP_response_status() returns status of OCSP
1725 response. OCSP_response_get1_basic() extracts basic response
1726 from response. OCSP_resp_find_status(): finds and extracts status
1727 information from an OCSP_CERTID structure (which will be created
1728 when the request structure is built). These are built from lower
1729 level functions which work on OCSP_SINGLERESP structures but
1730 wont normally be used unless the application wishes to examine
1731 extensions in the OCSP response for example.
1733 Replace nonce routines with a pair of functions.
1734 OCSP_request_add1_nonce() adds a nonce value and optionally
1735 generates a random value. OCSP_check_nonce() checks the
1736 validity of the nonce in an OCSP response.
1739 *) Change function OCSP_request_add() to OCSP_request_add0_id().
1740 This doesn't copy the supplied OCSP_CERTID and avoids the
1741 need to free up the newly created id. Change return type
1742 to OCSP_ONEREQ to return the internal OCSP_ONEREQ structure.
1743 This can then be used to add extensions to the request.
1744 Deleted OCSP_request_new(), since most of its functionality
1745 is now in OCSP_REQUEST_new() (and the case insensitive name
1746 clash) apart from the ability to set the request name which
1747 will be added elsewhere.
1750 *) Update OCSP API. Remove obsolete extensions argument from
1751 various functions. Extensions are now handled using the new
1752 OCSP extension code. New simple OCSP HTTP function which
1753 can be used to send requests and parse the response.
1756 *) Fix the PKCS#7 (S/MIME) code to work with new ASN1. Two new
1757 ASN1_ITEM structures help with sign and verify. PKCS7_ATTR_SIGN
1758 uses the special reorder version of SET OF to sort the attributes
1759 and reorder them to match the encoded order. This resolves a long
1760 standing problem: a verify on a PKCS7 structure just after signing
1761 it used to fail because the attribute order did not match the
1762 encoded order. PKCS7_ATTR_VERIFY does not reorder the attributes:
1763 it uses the received order. This is necessary to tolerate some broken
1764 software that does not order SET OF. This is handled by encoding
1765 as a SEQUENCE OF but using implicit tagging (with UNIVERSAL class)
1766 to produce the required SET OF.
1769 *) Have mk1mf.pl generate the macros OPENSSL_BUILD_SHLIBCRYPTO and
1770 OPENSSL_BUILD_SHLIBSSL and use them appropriately in the header
1771 files to get correct declarations of the ASN.1 item variables.
1774 *) Rewrite of PKCS#12 code to use new ASN1 functionality. Replace many
1775 PKCS#12 macros with real functions. Fix two unrelated ASN1 bugs:
1776 asn1_check_tlen() would sometimes attempt to use 'ctx' when it was
1777 NULL and ASN1_TYPE was not dereferenced properly in asn1_ex_c2i().
1778 New ASN1 macro: DECLARE_ASN1_ITEM() which just declares the relevant
1779 ASN1_ITEM and no wrapper functions.
1782 *) New functions or ASN1_item_d2i_fp() and ASN1_item_d2i_bio(). These
1783 replace the old function pointer based I/O routines. Change most of
1784 the *_d2i_bio() and *_d2i_fp() functions to use these.
1787 *) Enhance mkdef.pl to be more accepting about spacing in C preprocessor
1788 lines, recognice more "algorithms" that can be deselected, and make
1789 it complain about algorithm deselection that isn't recognised.
1792 *) New ASN1 functions to handle dup, sign, verify, digest, pack and
1793 unpack operations in terms of ASN1_ITEM. Modify existing wrappers
1794 to use new functions. Add NO_ASN1_OLD which can be set to remove
1795 some old style ASN1 functions: this can be used to determine if old
1796 code will still work when these eventually go away.
1799 *) New extension functions for OCSP structures, these follow the
1800 same conventions as certificates and CRLs.
1803 *) New function X509V3_add1_i2d(). This automatically encodes and
1804 adds an extension. Its behaviour can be customised with various
1805 flags to append, replace or delete. Various wrappers added for
1806 certifcates and CRLs.
1809 *) Fix to avoid calling the underlying ASN1 print routine when
1810 an extension cannot be parsed. Correct a typo in the
1811 OCSP_SERVICELOC extension. Tidy up print OCSP format.
1814 *) Make mkdef.pl parse some of the ASN1 macros and add apropriate
1815 entries for variables.
1818 *) Add functionality to apps/openssl.c for detecting locking
1819 problems: As the program is single-threaded, all we have
1820 to do is register a locking callback using an array for
1821 storing which locks are currently held by the program.
1824 *) Use a lock around the call to CRYPTO_get_ex_new_index() in
1825 SSL_get_ex_data_X509_STORE_idx(), which is used in
1826 ssl_verify_cert_chain() and thus can be called at any time
1827 during TLS/SSL handshakes so that thread-safety is essential.
1828 Unfortunately, the ex_data design is not at all suited
1829 for multi-threaded use, so it probably should be abolished.
1832 *) Added Broadcom "ubsec" ENGINE to OpenSSL.
1833 [Broadcom, tweaked and integrated by Geoff Thorpe]
1835 *) Move common extension printing code to new function
1836 X509V3_print_extensions(). Reorganise OCSP print routines and
1837 implement some needed OCSP ASN1 functions. Add OCSP extensions.
1840 *) New function X509_signature_print() to remove duplication in some
1844 *) Add a special meaning when SET OF and SEQUENCE OF flags are both
1845 set (this was treated exactly the same as SET OF previously). This
1846 is used to reorder the STACK representing the structure to match the
1847 encoding. This will be used to get round a problem where a PKCS7
1848 structure which was signed could not be verified because the STACK
1849 order did not reflect the encoded order.
1852 *) Reimplement the OCSP ASN1 module using the new code.
1855 *) Update the X509V3 code to permit the use of an ASN1_ITEM structure
1856 for its ASN1 operations. The old style function pointers still exist
1857 for now but they will eventually go away.
1860 *) Merge in replacement ASN1 code from the ASN1 branch. This almost
1861 completely replaces the old ASN1 functionality with a table driven
1862 encoder and decoder which interprets an ASN1_ITEM structure describing
1863 the ASN1 module. Compatibility with the existing ASN1 API (i2d,d2i) is
1864 largely maintained. Almost all of the old asn1_mac.h macro based ASN1
1865 has also been converted to the new form.
1868 *) Change BN_mod_exp_recp so that negative moduli are tolerated
1869 (the sign is ignored). Similarly, ignore the sign in BN_MONT_CTX_set
1870 so that BN_mod_exp_mont and BN_mod_exp_mont_word work
1871 for negative moduli.
1874 *) Fix BN_uadd and BN_usub: Always return non-negative results instead
1875 of not touching the result's sign bit.
1878 *) BN_div bugfix: If the result is 0, the sign (res->neg) must not be
1882 *) Changed the LHASH code to use prototypes for callbacks, and created
1883 macros to declare and implement thin (optionally static) functions
1884 that provide type-safety and avoid function pointer casting for the
1885 type-specific callbacks.
1888 *) Added Kerberos Cipher Suites to be used with TLS, as written in
1890 [Veers Staats <staatsvr@asc.hpc.mil>,
1891 Jeffrey Altman <jaltman@columbia.edu>, via Richard Levitte]
1893 *) Reformat the FAQ so the different questions and answers can be divided
1894 in sections depending on the subject.
1897 *) Have the zlib compression code load ZLIB.DLL dynamically under
1901 *) New function BN_mod_sqrt for computing square roots modulo a prime
1902 (using the probabilistic Tonelli-Shanks algorithm unless
1903 p == 3 (mod 4) or p == 5 (mod 8), which are cases that can
1904 be handled deterministically).
1905 [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller]
1907 *) Make BN_mod_inverse faster by explicitly handling small quotients
1908 in the Euclid loop. (Speed gain about 20% for small moduli [256 or
1909 512 bits], about 30% for larger ones [1024 or 2048 bits].)
1912 *) New function BN_kronecker.
1915 *) Fix BN_gcd so that it works on negative inputs; the result is
1916 positive unless both parameters are zero.
1917 Previously something reasonably close to an infinite loop was
1918 possible because numbers could be growing instead of shrinking
1919 in the implementation of Euclid's algorithm.
1922 *) Fix BN_is_word() and BN_is_one() macros to take into account the
1923 sign of the number in question.
1925 Fix BN_is_word(a,w) to work correctly for w == 0.
1927 The old BN_is_word(a,w) macro is now called BN_abs_is_word(a,w)
1928 because its test if the absolute value of 'a' equals 'w'.
1929 Note that BN_abs_is_word does *not* handle w == 0 reliably;
1930 it exists mostly for use in the implementations of BN_is_zero(),
1931 BN_is_one(), and BN_is_word().
1934 *) New function BN_swap.
1937 *) Use BN_nnmod instead of BN_mod in crypto/bn/bn_exp.c so that
1938 the exponentiation functions are more likely to produce reasonable
1939 results on negative inputs.
1942 *) Change BN_mod_mul so that the result is always non-negative.
1943 Previously, it could be negative if one of the factors was negative;
1944 I don't think anyone really wanted that behaviour.
1947 *) Move BN_mod_... functions into new file crypto/bn/bn_mod.c
1948 (except for exponentiation, which stays in crypto/bn/bn_exp.c,
1949 and BN_mod_mul_reciprocal, which stays in crypto/bn/bn_recp.c)
1950 and add new functions:
1959 BN_mod_lshift1_quick
1963 These functions always generate non-negative results.
1965 BN_nnmod otherwise is like BN_mod (if BN_mod computes a remainder r
1966 such that |m| < r < 0, BN_nnmod will output rem + |m| instead).
1968 BN_mod_XXX_quick(r, a, [b,] m) generates the same result as
1969 BN_mod_XXX(r, a, [b,] m, ctx), but requires that a [and b]
1970 be reduced modulo m.
1971 [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller]
1973 *) Remove a few calls to bn_wexpand() in BN_sqr() (the one in there
1974 was actually never needed) and in BN_mul(). The removal in BN_mul()
1975 required a small change in bn_mul_part_recursive() and the addition
1976 of the functions bn_cmp_part_words(), bn_sub_part_words() and
1977 bn_add_part_words(), which do the same thing as bn_cmp_words(),
1978 bn_sub_words() and bn_add_words() except they take arrays with
1982 *) In 'openssl passwd', verify passwords read from the terminal
1983 unless the '-salt' option is used (which usually means that
1984 verification would just waste user's time since the resulting
1985 hash is going to be compared with some given password hash)
1986 or the new '-noverify' option is used.
1988 This is an incompatible change, but it does not affect
1989 non-interactive use of 'openssl passwd' (passwords on the command
1990 line, '-stdin' option, '-in ...' option) and thus should not
1994 *) Remove all references to RSAref, since there's no more need for it.
1997 *) Make DSO load along a path given through an environment variable
1998 (SHLIB_PATH) with shl_load().
2001 *) Constify the ENGINE code as a result of BIGNUM constification.
2002 Also constify the RSA code and most things related to it. In a
2003 few places, most notable in the depth of the ASN.1 code, ugly
2004 casts back to non-const were required (to be solved at a later
2008 *) Make it so the openssl application has all engines loaded by default.
2011 *) Constify the BIGNUM routines a little more.
2014 *) Add the following functions:
2016 ENGINE_load_cswift()
2018 ENGINE_load_atalla()
2020 ENGINE_load_builtin_engines()
2022 That way, an application can itself choose if external engines that
2023 are built-in in OpenSSL shall ever be used or not. The benefit is
2024 that applications won't have to be linked with libdl or other dso
2025 libraries unless it's really needed.
2027 Changed 'openssl engine' to load all engines on demand.
2028 Changed the engine header files to avoid the duplication of some
2029 declarations (they differed!).
2032 *) 'openssl engine' can now list capabilities.
2035 *) Better error reporting in 'openssl engine'.
2038 *) Never call load_dh_param(NULL) in s_server.
2041 *) Add engine application. It can currently list engines by name and
2042 identity, and test if they are actually available.
2045 *) Improve RPM specification file by forcing symbolic linking and making
2046 sure the installed documentation is also owned by root.root.
2047 [Damien Miller <djm@mindrot.org>]
2049 *) Give the OpenSSL applications more possibilities to make use of
2050 keys (public as well as private) handled by engines.
2053 *) Add OCSP code that comes from CertCo.
2056 *) Add VMS support for the Rijndael code.
2059 *) Added untested support for Nuron crypto accelerator.
2062 *) Add support for external cryptographic devices. This code was
2063 previously distributed separately as the "engine" branch.
2064 [Geoff Thorpe, Richard Levitte]
2066 *) Rework the filename-translation in the DSO code. It is now possible to
2067 have far greater control over how a "name" is turned into a filename
2068 depending on the operating environment and any oddities about the
2069 different shared library filenames on each system.
2072 *) Support threads on FreeBSD-elf in Configure.
2075 *) Fix for SHA1 assembly problem with MASM: it produces
2076 warnings about corrupt line number information when assembling
2077 with debugging information. This is caused by the overlapping
2079 [Bernd Matthes <mainbug@celocom.de>, Steve Henson]
2082 NCONF_get_number() has no error checking at all. As a replacement,
2083 NCONF_get_number_e() is defined (_e for "error checking") and is
2084 promoted strongly. The old NCONF_get_number is kept around for
2085 binary backward compatibility.
2086 Make it possible for methods to load from something other than a BIO,
2087 by providing a function pointer that is given a name instead of a BIO.
2088 For example, this could be used to load configuration data from an
2092 *) Fix for non blocking accept BIOs. Added new I/O special reason
2093 BIO_RR_ACCEPT to cover this case. Previously use of accept BIOs
2094 with non blocking I/O was not possible because no retry code was
2095 implemented. Also added new SSL code SSL_WANT_ACCEPT to cover
2099 *) Added the beginnings of Rijndael support.
2102 *) Fix for bug in DirectoryString mask setting. Add support for
2103 X509_NAME_print_ex() in 'req' and X509_print_ex() function
2104 to allow certificate printing to more controllable, additional
2105 'certopt' option to 'x509' to allow new printing options to be
2109 *) Clean old EAY MD5 hack from e_os.h.
2112 Changes between 0.9.6g and 0.9.6h [xx XXX xxxx]
2114 *) Bugfix: client side session caching did not work with external caching,
2115 because the session->cipher setting was not restored when reloading
2116 from the external cache. This problem was masked, when
2117 SSL_OP_NETSCAPE_REUSE_CIPHER_CHANGE_BUG (part of SSL_OP_ALL) was set.
2118 (Found by Steve Haslam <steve@araqnid.ddts.net>.)
2121 *) Fix client_certificate (ssl/s2_clnt.c): The permissible total
2122 length of the REQUEST-CERTIFICATE message is 18 .. 34, not 17 .. 33.
2123 [Zeev Lieber <zeev-l@yahoo.com>]
2125 *) Undo an undocumented change introduced in 0.9.6e which caused
2126 repeated calls to OpenSSL_add_all_ciphers() and
2127 OpenSSL_add_all_digests() to be ignored, even after calling
2131 *) Change the default configuration reader to deal with last line not
2132 being properly terminated.
2135 *) Change X509_NAME_cmp() so it applies the special rules on handling
2136 DN values that are of type PrintableString, as well as RDNs of type
2137 emailAddress where the value has the type ia5String.
2138 [stefank@valicert.com via Richard Levitte]
2140 *) Add a SSL_SESS_CACHE_NO_INTERNAL_STORE flag to take over half
2141 the job SSL_SESS_CACHE_NO_INTERNAL_LOOKUP was inconsistently
2142 doing, define a new flag (SSL_SESS_CACHE_NO_INTERNAL) to be
2143 the bitwise-OR of the two for use by the majority of applications
2144 wanting this behaviour, and update the docs. The documented
2145 behaviour and actual behaviour were inconsistent and had been
2146 changing anyway, so this is more a bug-fix than a behavioural
2148 [Geoff Thorpe, diagnosed by Nadav Har'El]
2150 *) Don't impose a 16-byte length minimum on session IDs in ssl/s3_clnt.c
2151 (the SSL 3.0 and TLS 1.0 specifications allow any length up to 32 bytes).
2154 *) Fix initialization code race conditions in
2155 SSLv23_method(), SSLv23_client_method(), SSLv23_server_method(),
2156 SSLv2_method(), SSLv2_client_method(), SSLv2_server_method(),
2157 SSLv3_method(), SSLv3_client_method(), SSLv3_server_method(),
2158 TLSv1_method(), TLSv1_client_method(), TLSv1_server_method(),
2159 ssl2_get_cipher_by_char(),
2160 ssl3_get_cipher_by_char().
2161 [Patrick McCormick <patrick@tellme.com>, Bodo Moeller]
2163 *) Reorder cleanup sequence in SSL_CTX_free(): only remove the ex_data after
2164 the cached sessions are flushed, as the remove_cb() might use ex_data
2165 contents. Bug found by Sam Varshavchik <mrsam@courier-mta.com>
2166 (see [openssl.org #212]).
2167 [Geoff Thorpe, Lutz Jaenicke]
2169 *) Fix typo in OBJ_txt2obj which incorrectly passed the content
2170 length, instead of the encoding length to d2i_ASN1_OBJECT.
2173 Changes between 0.9.6f and 0.9.6g [9 Aug 2002]
2175 *) [In 0.9.6g-engine release:]
2176 Fix crypto/engine/vendor_defns/cswift.h for WIN32 (use '_stdcall').
2177 [Lynn Gazis <lgazis@rainbow.com>]
2179 Changes between 0.9.6e and 0.9.6f [8 Aug 2002]
2181 *) Fix ASN1 checks. Check for overflow by comparing with LONG_MAX
2182 and get fix the header length calculation.
2183 [Florian Weimer <Weimer@CERT.Uni-Stuttgart.DE>,
2184 Alon Kantor <alonk@checkpoint.com> (and others),
2187 *) Use proper error handling instead of 'assertions' in buffer
2188 overflow checks added in 0.9.6e. This prevents DoS (the
2189 assertions could call abort()).
2190 [Arne Ansper <arne@ats.cyber.ee>, Bodo Moeller]
2192 Changes between 0.9.6d and 0.9.6e [30 Jul 2002]
2194 *) Add various sanity checks to asn1_get_length() to reject
2195 the ASN1 length bytes if they exceed sizeof(long), will appear
2196 negative or the content length exceeds the length of the
2198 [Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>]
2200 *) Fix cipher selection routines: ciphers without encryption had no flags
2201 for the cipher strength set and where therefore not handled correctly
2202 by the selection routines (PR #130).
2205 *) Fix EVP_dsa_sha macro.
2209 SSL_OP_DONT_INSERT_EMPTY_FRAGMENTS
2210 for disabling the SSL 3.0/TLS 1.0 CBC vulnerability countermeasure
2211 that was added in OpenSSL 0.9.6d.
2213 As the countermeasure turned out to be incompatible with some
2214 broken SSL implementations, the new option is part of SSL_OP_ALL.
2215 SSL_OP_ALL is usually employed when compatibility with weird SSL
2216 implementations is desired (e.g. '-bugs' option to 's_client' and
2217 's_server'), so the new option is automatically set in many
2221 *) Changes in security patch:
2223 Changes marked "(CHATS)" were sponsored by the Defense Advanced
2224 Research Projects Agency (DARPA) and Air Force Research Laboratory,
2225 Air Force Materiel Command, USAF, under agreement number
2228 *) Add various sanity checks to asn1_get_length() to reject
2229 the ASN1 length bytes if they exceed sizeof(long), will appear
2230 negative or the content length exceeds the length of the
2231 supplied buffer. (CAN-2002-0659)
2232 [Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>]
2234 *) Assertions for various potential buffer overflows, not known to
2236 [Ben Laurie (CHATS)]
2238 *) Various temporary buffers to hold ASCII versions of integers were
2239 too small for 64 bit platforms. (CAN-2002-0655)
2240 [Matthew Byng-Maddick <mbm@aldigital.co.uk> and Ben Laurie (CHATS)>
2242 *) Remote buffer overflow in SSL3 protocol - an attacker could
2243 supply an oversized session ID to a client. (CAN-2002-0656)
2244 [Ben Laurie (CHATS)]
2246 *) Remote buffer overflow in SSL2 protocol - an attacker could
2247 supply an oversized client master key. (CAN-2002-0656)
2248 [Ben Laurie (CHATS)]
2250 Changes between 0.9.6c and 0.9.6d [9 May 2002]
2252 *) Fix crypto/asn1/a_sign.c so that 'parameters' is omitted (not
2253 encoded as NULL) with id-dsa-with-sha1.
2254 [Nils Larsch <nla@trustcenter.de>; problem pointed out by Bodo Moeller]
2256 *) Check various X509_...() return values in apps/req.c.
2257 [Nils Larsch <nla@trustcenter.de>]
2259 *) Fix BASE64 decode (EVP_DecodeUpdate) for data with CR/LF ended lines:
2260 an end-of-file condition would erronously be flagged, when the CRLF
2261 was just at the end of a processed block. The bug was discovered when
2262 processing data through a buffering memory BIO handing the data to a
2263 BASE64-decoding BIO. Bug fund and patch submitted by Pavel Tsekov
2264 <ptsekov@syntrex.com> and Nedelcho Stanev.
2267 *) Implement a countermeasure against a vulnerability recently found
2268 in CBC ciphersuites in SSL 3.0/TLS 1.0: Send an empty fragment
2269 before application data chunks to avoid the use of known IVs
2270 with data potentially chosen by the attacker.
2273 *) Fix length checks in ssl3_get_client_hello().
2276 *) TLS/SSL library bugfix: use s->s3->in_read_app_data differently
2277 to prevent ssl3_read_internal() from incorrectly assuming that
2278 ssl3_read_bytes() found application data while handshake
2279 processing was enabled when in fact s->s3->in_read_app_data was
2280 merely automatically cleared during the initial handshake.
2281 [Bodo Moeller; problem pointed out by Arne Ansper <arne@ats.cyber.ee>]
2283 *) Fix object definitions for Private and Enterprise: they were not
2284 recognized in their shortname (=lowercase) representation. Extend
2285 obj_dat.pl to issue an error when using undefined keywords instead
2286 of silently ignoring the problem (Svenning Sorensen
2287 <sss@sss.dnsalias.net>).
2290 *) Fix DH_generate_parameters() so that it works for 'non-standard'
2291 generators, i.e. generators other than 2 and 5. (Previously, the
2292 code did not properly initialise the 'add' and 'rem' values to
2293 BN_generate_prime().)
2295 In the new general case, we do not insist that 'generator' is
2296 actually a primitive root: This requirement is rather pointless;
2297 a generator of the order-q subgroup is just as good, if not
2301 *) Map new X509 verification errors to alerts. Discovered and submitted by
2302 Tom Wu <tom@arcot.com>.
2305 *) Fix ssl3_pending() (ssl/s3_lib.c) to prevent SSL_pending() from
2306 returning non-zero before the data has been completely received
2307 when using non-blocking I/O.
2308 [Bodo Moeller; problem pointed out by John Hughes]
2310 *) Some of the ciphers missed the strength entry (SSL_LOW etc).
2311 [Ben Laurie, Lutz Jaenicke]
2313 *) Fix bug in SSL_clear(): bad sessions were not removed (found by
2314 Yoram Zahavi <YoramZ@gilian.com>).
2317 *) Add information about CygWin 1.3 and on, and preserve proper
2318 configuration for the versions before that.
2319 [Corinna Vinschen <vinschen@redhat.com> and Richard Levitte]
2321 *) Make removal from session cache (SSL_CTX_remove_session()) more robust:
2322 check whether we deal with a copy of a session and do not delete from
2323 the cache in this case. Problem reported by "Izhar Shoshani Levi"
2324 <izhar@checkpoint.com>.
2327 *) Do not store session data into the internal session cache, if it
2328 is never intended to be looked up (SSL_SESS_CACHE_NO_INTERNAL_LOOKUP
2329 flag is set). Proposed by Aslam <aslam@funk.com>.
2332 *) Have ASN1_BIT_STRING_set_bit() really clear a bit when the requested
2336 *) [In 0.9.6d-engine release:]
2337 Fix a crashbug and a logic bug in hwcrhk_load_pubkey().
2338 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte]
2340 *) Add the configuration target linux-s390x.
2341 [Neale Ferguson <Neale.Ferguson@SoftwareAG-USA.com> via Richard Levitte]
2343 *) The earlier bugfix for the SSL3_ST_SW_HELLO_REQ_C case of
2344 ssl3_accept (ssl/s3_srvr.c) incorrectly used a local flag
2345 variable as an indication that a ClientHello message has been
2346 received. As the flag value will be lost between multiple
2347 invocations of ssl3_accept when using non-blocking I/O, the
2348 function may not be aware that a handshake has actually taken
2349 place, thus preventing a new session from being added to the
2352 To avoid this problem, we now set s->new_session to 2 instead of
2353 using a local variable.
2354 [Lutz Jaenicke, Bodo Moeller]
2356 *) Bugfix: Return -1 from ssl3_get_server_done (ssl3/s3_clnt.c)
2357 if the SSL_R_LENGTH_MISMATCH error is detected.
2358 [Geoff Thorpe, Bodo Moeller]
2360 *) New 'shared_ldflag' column in Configure platform table.
2363 *) Fix EVP_CIPHER_mode macro.
2364 ["Dan S. Camper" <dan@bti.net>]
2366 *) Fix ssl3_read_bytes (ssl/s3_pkt.c): To ignore messages of unknown
2367 type, we must throw them away by setting rr->length to 0.
2368 [D P Chang <dpc@qualys.com>]
2370 Changes between 0.9.6b and 0.9.6c [21 dec 2001]
2372 *) Fix BN_rand_range bug pointed out by Dominikus Scherkl
2373 <Dominikus.Scherkl@biodata.com>. (The previous implementation
2374 worked incorrectly for those cases where range = 10..._2 and
2375 3*range is two bits longer than range.)
2378 *) Only add signing time to PKCS7 structures if it is not already
2382 *) Fix crypto/objects/objects.h: "ld-ce" should be "id-ce",
2383 OBJ_ld_ce should be OBJ_id_ce.
2384 Also some ip-pda OIDs in crypto/objects/objects.txt were
2385 incorrect (cf. RFC 3039).
2386 [Matt Cooper, Frederic Giudicelli, Bodo Moeller]
2388 *) Release CRYPTO_LOCK_DYNLOCK when CRYPTO_destroy_dynlockid()
2389 returns early because it has nothing to do.
2390 [Andy Schneider <andy.schneider@bjss.co.uk>]
2392 *) [In 0.9.6c-engine release:]
2393 Fix mutex callback return values in crypto/engine/hw_ncipher.c.
2394 [Andy Schneider <andy.schneider@bjss.co.uk>]
2396 *) [In 0.9.6c-engine release:]
2397 Add support for Cryptographic Appliance's keyserver technology.
2398 (Use engine 'keyclient')
2399 [Cryptographic Appliances and Geoff Thorpe]
2401 *) Add a configuration entry for OS/390 Unix. The C compiler 'c89'
2402 is called via tools/c89.sh because arguments have to be
2403 rearranged (all '-L' options must appear before the first object
2405 [Richard Shapiro <rshapiro@abinitio.com>]
2407 *) [In 0.9.6c-engine release:]
2408 Add support for Broadcom crypto accelerator cards, backported
2410 [Broadcom, Nalin Dahyabhai <nalin@redhat.com>, Mark Cox]
2412 *) [In 0.9.6c-engine release:]
2413 Add support for SureWare crypto accelerator cards from
2414 Baltimore Technologies. (Use engine 'sureware')
2415 [Baltimore Technologies and Mark Cox]
2417 *) [In 0.9.6c-engine release:]
2418 Add support for crypto accelerator cards from Accelerated
2419 Encryption Processing, www.aep.ie. (Use engine 'aep')
2420 [AEP Inc. and Mark Cox]
2422 *) Add a configuration entry for gcc on UnixWare.
2423 [Gary Benson <gbenson@redhat.com>]
2425 *) Change ssl/s2_clnt.c and ssl/s2_srvr.c so that received handshake
2426 messages are stored in a single piece (fixed-length part and
2427 variable-length part combined) and fix various bugs found on the way.
2430 *) Disable caching in BIO_gethostbyname(), directly use gethostbyname()
2431 instead. BIO_gethostbyname() does not know what timeouts are
2432 appropriate, so entries would stay in cache even when they have
2434 [Bodo Moeller; problem pointed out by Rich Salz <rsalz@zolera.com>
2436 *) Change ssl23_get_client_hello (ssl/s23_srvr.c) behaviour when
2437 faced with a pathologically small ClientHello fragment that does
2438 not contain client_version: Instead of aborting with an error,
2439 simply choose the highest available protocol version (i.e.,
2440 TLS 1.0 unless it is disabled). In practice, ClientHello
2441 messages are never sent like this, but this change gives us
2442 strictly correct behaviour at least for TLS.
2445 *) Fix SSL handshake functions and SSL_clear() such that SSL_clear()
2446 never resets s->method to s->ctx->method when called from within
2447 one of the SSL handshake functions.
2448 [Bodo Moeller; problem pointed out by Niko Baric]
2450 *) In ssl3_get_client_hello (ssl/s3_srvr.c), generate a fatal alert
2451 (sent using the client's version number) if client_version is
2452 smaller than the protocol version in use. Also change
2453 ssl23_get_client_hello (ssl/s23_srvr.c) to select TLS 1.0 if
2454 the client demanded SSL 3.0 but only TLS 1.0 is enabled; then
2455 the client will at least see that alert.
2458 *) Fix ssl3_get_message (ssl/s3_both.c) to handle message fragmentation
2462 *) Avoid infinite loop in ssl3_get_message (ssl/s3_both.c) if a
2463 client receives HelloRequest while in a handshake.
2464 [Bodo Moeller; bug noticed by Andy Schneider <andy.schneider@bjss.co.uk>]
2466 *) Bugfix in ssl3_accept (ssl/s3_srvr.c): Case SSL3_ST_SW_HELLO_REQ_C
2467 should end in 'break', not 'goto end' which circuments various
2468 cleanups done in state SSL_ST_OK. But session related stuff
2469 must be disabled for SSL_ST_OK in the case that we just sent a
2472 Also avoid some overhead by not calling ssl_init_wbio_buffer()
2473 before just sending a HelloRequest.
2474 [Bodo Moeller, Eric Rescorla <ekr@rtfm.com>]
2476 *) Fix ssl/s3_enc.c, ssl/t1_enc.c and ssl/s3_pkt.c so that we don't
2477 reveal whether illegal block cipher padding was found or a MAC
2478 verification error occured. (Neither SSLerr() codes nor alerts
2479 are directly visible to potential attackers, but the information
2480 may leak via logfiles.)
2482 Similar changes are not required for the SSL 2.0 implementation
2483 because the number of padding bytes is sent in clear for SSL 2.0,
2484 and the extra bytes are just ignored. However ssl/s2_pkt.c
2485 failed to verify that the purported number of padding bytes is in
2489 *) Add OpenUNIX-8 support including shared libraries
2490 (Boyd Lynn Gerber <gerberb@zenez.com>).
2493 *) Improve RSA_padding_check_PKCS1_OAEP() check again to avoid
2494 'wristwatch attack' using huge encoding parameters (cf.
2495 James H. Manger's CRYPTO 2001 paper). Note that the
2496 RSA_PKCS1_OAEP_PADDING case of RSA_private_decrypt() does not use
2497 encoding parameters and hence was not vulnerable.
2500 *) BN_sqr() bug fix.
2501 [Ulf Möller, reported by Jim Ellis <jim.ellis@cavium.com>]
2503 *) Rabin-Miller test analyses assume uniformly distributed witnesses,
2504 so use BN_pseudo_rand_range() instead of using BN_pseudo_rand()
2505 followed by modular reduction.
2506 [Bodo Moeller; pointed out by Adam Young <AYoung1@NCSUS.JNJ.COM>]
2508 *) Add BN_pseudo_rand_range() with obvious functionality: BN_rand_range()
2509 equivalent based on BN_pseudo_rand() instead of BN_rand().
2512 *) s3_srvr.c: allow sending of large client certificate lists (> 16 kB).
2513 This function was broken, as the check for a new client hello message
2514 to handle SGC did not allow these large messages.
2515 (Tracked down by "Douglas E. Engert" <deengert@anl.gov>.)
2518 *) Add alert descriptions for TLSv1 to SSL_alert_desc_string[_long]().
2521 *) Fix buggy behaviour of BIO_get_num_renegotiates() and BIO_ctrl()
2522 for BIO_C_GET_WRITE_BUF_SIZE ("Stephen Hinton" <shinton@netopia.com>).
2525 *) Rework the configuration and shared library support for Tru64 Unix.
2526 The configuration part makes use of modern compiler features and
2527 still retains old compiler behavior for those that run older versions
2528 of the OS. The shared library support part includes a variant that
2529 uses the RPATH feature, and is available through the special
2530 configuration target "alpha-cc-rpath", which will never be selected
2532 [Tim Mooney <mooney@dogbert.cc.ndsu.NoDak.edu> via Richard Levitte]
2534 *) In ssl3_get_key_exchange (ssl/s3_clnt.c), call ssl3_get_message()
2535 with the same message size as in ssl3_get_certificate_request().
2536 Otherwise, if no ServerKeyExchange message occurs, CertificateRequest
2537 messages might inadvertently be reject as too long.
2538 [Petr Lampa <lampa@fee.vutbr.cz>]
2540 *) Enhanced support for IA-64 Unix platforms (well, Linux and HP-UX).
2543 *) Modified SSL library such that the verify_callback that has been set
2544 specificly for an SSL object with SSL_set_verify() is actually being
2545 used. Before the change, a verify_callback set with this function was
2546 ignored and the verify_callback() set in the SSL_CTX at the time of
2547 the call was used. New function X509_STORE_CTX_set_verify_cb() introduced
2548 to allow the necessary settings.
2551 *) Initialize static variable in crypto/dsa/dsa_lib.c and crypto/dh/dh_lib.c
2552 explicitly to NULL, as at least on Solaris 8 this seems not always to be
2553 done automatically (in contradiction to the requirements of the C
2554 standard). This made problems when used from OpenSSH.
2557 *) In OpenSSL 0.9.6a and 0.9.6b, crypto/dh/dh_key.c ignored
2558 dh->length and always used
2560 BN_rand_range(priv_key, dh->p).
2562 BN_rand_range() is not necessary for Diffie-Hellman, and this
2563 specific range makes Diffie-Hellman unnecessarily inefficient if
2564 dh->length (recommended exponent length) is much smaller than the
2565 length of dh->p. We could use BN_rand_range() if the order of
2566 the subgroup was stored in the DH structure, but we only have
2571 BN_rand(priv_key, l, ...)
2573 where 'l' is dh->length if this is defined, or BN_num_bits(dh->p)-1
2579 RSA_eay_public_encrypt
2580 RSA_eay_private_decrypt
2581 RSA_eay_private_encrypt (signing)
2582 RSA_eay_public_decrypt (signature verification)
2584 (default implementations for RSA_public_encrypt,
2585 RSA_private_decrypt, RSA_private_encrypt, RSA_public_decrypt),
2586 always reject numbers >= n.
2589 *) In crypto/rand/md_rand.c, use a new short-time lock CRYPTO_LOCK_RAND2
2590 to synchronize access to 'locking_thread'. This is necessary on
2591 systems where access to 'locking_thread' (an 'unsigned long'
2592 variable) is not atomic.
2595 *) In crypto/rand/md_rand.c, set 'locking_thread' to current thread's ID
2596 *before* setting the 'crypto_lock_rand' flag. The previous code had
2597 a race condition if 0 is a valid thread ID.
2598 [Travis Vitek <vitek@roguewave.com>]
2600 *) Add support for shared libraries under Irix.
2601 [Albert Chin-A-Young <china@thewrittenword.com>]
2603 *) Add configuration option to build on Linux on both big-endian and
2605 [Ralf Baechle <ralf@uni-koblenz.de>]
2607 *) Add the possibility to create shared libraries on HP-UX.
2610 Changes between 0.9.6a and 0.9.6b [9 Jul 2001]
2612 *) Change ssleay_rand_bytes (crypto/rand/md_rand.c)
2613 to avoid a SSLeay/OpenSSL PRNG weakness pointed out by
2614 Markku-Juhani O. Saarinen <markku-juhani.saarinen@nokia.com>:
2615 PRNG state recovery was possible based on the output of
2616 one PRNG request appropriately sized to gain knowledge on
2617 'md' followed by enough consecutive 1-byte PRNG requests
2618 to traverse all of 'state'.
2620 1. When updating 'md_local' (the current thread's copy of 'md')
2621 during PRNG output generation, hash all of the previous
2622 'md_local' value, not just the half used for PRNG output.
2624 2. Make the number of bytes from 'state' included into the hash
2625 independent from the number of PRNG bytes requested.
2627 The first measure alone would be sufficient to avoid
2628 Markku-Juhani's attack. (Actually it had never occurred
2629 to me that the half of 'md_local' used for chaining was the
2630 half from which PRNG output bytes were taken -- I had always
2631 assumed that the secret half would be used.) The second
2632 measure makes sure that additional data from 'state' is never
2633 mixed into 'md_local' in small portions; this heuristically
2634 further strengthens the PRNG.
2637 *) Fix crypto/bn/asm/mips3.s.
2640 *) When only the key is given to "enc", the IV is undefined. Print out
2641 an error message in this case.
2644 *) Handle special case when X509_NAME is empty in X509 printing routines.
2647 *) In dsa_do_verify (crypto/dsa/dsa_ossl.c), verify that r and s are
2648 positive and less than q.
2651 *) Don't change *pointer in CRYPTO_add_lock() is add_lock_callback is
2652 used: it isn't thread safe and the add_lock_callback should handle
2654 [Paul Rose <Paul.Rose@bridge.com>]
2656 *) Verify that incoming data obeys the block size in
2657 ssl3_enc (ssl/s3_enc.c) and tls1_enc (ssl/t1_enc.c).
2661 [Ulf Möller, Bodo Möller]
2663 *) The countermeasure against Bleichbacher's attack on PKCS #1 v1.5
2664 RSA encryption was accidentally removed in s3_srvr.c in OpenSSL 0.9.5
2665 when fixing the server behaviour for backwards-compatible 'client
2666 hello' messages. (Note that the attack is impractical against
2667 SSL 3.0 and TLS 1.0 anyway because length and version checking
2668 means that the probability of guessing a valid ciphertext is
2669 around 2^-40; see section 5 in Bleichenbacher's CRYPTO '98
2672 Before 0.9.5, the countermeasure (hide the error by generating a
2673 random 'decryption result') did not work properly because
2674 ERR_clear_error() was missing, meaning that SSL_get_error() would
2675 detect the supposedly ignored error.
2677 Both problems are now fixed.
2680 *) In crypto/bio/bf_buff.c, increase DEFAULT_BUFFER_SIZE to 4096
2681 (previously it was 1024).
2684 *) Fix for compatibility mode trust settings: ignore trust settings
2685 unless some valid trust or reject settings are present.
2688 *) Fix for blowfish EVP: its a variable length cipher.
2691 *) Fix various bugs related to DSA S/MIME verification. Handle missing
2692 parameters in DSA public key structures and return an error in the
2693 DSA routines if parameters are absent.
2696 *) In versions up to 0.9.6, RAND_file_name() resorted to file ".rnd"
2697 in the current directory if neither $RANDFILE nor $HOME was set.
2698 RAND_file_name() in 0.9.6a returned NULL in this case. This has
2699 caused some confusion to Windows users who haven't defined $HOME.
2700 Thus RAND_file_name() is changed again: e_os.h can define a
2701 DEFAULT_HOME, which will be used if $HOME is not set.
2702 For Windows, we use "C:"; on other platforms, we still require
2703 environment variables.
2705 *) Move 'if (!initialized) RAND_poll()' into regions protected by
2706 CRYPTO_LOCK_RAND. This is not strictly necessary, but avoids
2707 having multiple threads call RAND_poll() concurrently.
2710 *) In crypto/rand/md_rand.c, replace 'add_do_not_lock' flag by a
2711 combination of a flag and a thread ID variable.
2712 Otherwise while one thread is in ssleay_rand_bytes (which sets the
2713 flag), *other* threads can enter ssleay_add_bytes without obeying
2714 the CRYPTO_LOCK_RAND lock (and may even illegally release the lock
2715 that they do not hold after the first thread unsets add_do_not_lock).
2718 *) Change bctest again: '-x' expressions are not available in all
2722 Changes between 0.9.6 and 0.9.6a [5 Apr 2001]
2724 *) Fix a couple of memory leaks in PKCS7_dataDecode()
2725 [Steve Henson, reported by Heyun Zheng <hzheng@atdsprint.com>]
2727 *) Change Configure and Makefiles to provide EXE_EXT, which will contain
2728 the default extension for executables, if any. Also, make the perl
2729 scripts that use symlink() to test if it really exists and use "cp"
2730 if it doesn't. All this made OpenSSL compilable and installable in
2734 *) Fix for asn1_GetSequence() for indefinite length constructed data.
2735 If SEQUENCE is length is indefinite just set c->slen to the total
2736 amount of data available.
2737 [Steve Henson, reported by shige@FreeBSD.org]
2738 [This change does not apply to 0.9.7.]
2740 *) Change bctest to avoid here-documents inside command substitution
2741 (workaround for FreeBSD /bin/sh bug).
2742 For compatibility with Ultrix, avoid shell functions (introduced
2743 in the bctest version that searches along $PATH).
2746 *) Rename 'des_encrypt' to 'des_encrypt1'. This avoids the clashes
2747 with des_encrypt() defined on some operating systems, like Solaris
2751 *) Check the result of RSA-CRT (see D. Boneh, R. DeMillo, R. Lipton:
2752 On the Importance of Eliminating Errors in Cryptographic
2753 Computations, J. Cryptology 14 (2001) 2, 101-119,
2754 http://theory.stanford.edu/~dabo/papers/faults.ps.gz).
2757 *) MIPS assembler BIGNUM division bug fix.
2760 *) Disabled incorrect Alpha assembler code.
2763 *) Fix PKCS#7 decode routines so they correctly update the length
2764 after reading an EOC for the EXPLICIT tag.
2766 [This change does not apply to 0.9.7.]
2768 *) Fix bug in PKCS#12 key generation routines. This was triggered
2769 if a 3DES key was generated with a 0 initial byte. Include
2770 PKCS12_BROKEN_KEYGEN compilation option to retain the old
2771 (but broken) behaviour.
2774 *) Enhance bctest to search for a working bc along $PATH and print
2776 [Tim Rice <tim@multitalents.net> via Richard Levitte]
2778 *) Fix memory leaks in err.c: free err_data string if necessary;
2779 don't write to the wrong index in ERR_set_error_data.
2782 *) Implement ssl23_peek (analogous to ssl23_read), which previously
2786 *) Replace rdtsc with _emit statements for VC++ version 5.
2787 [Jeremy Cooper <jeremy@baymoo.org>]
2789 *) Make it possible to reuse SSLv2 sessions.
2792 *) In copy_email() check for >= 0 as a return value for
2793 X509_NAME_get_index_by_NID() since 0 is a valid index.
2794 [Steve Henson reported by Massimiliano Pala <madwolf@opensca.org>]
2796 *) Avoid coredump with unsupported or invalid public keys by checking if
2797 X509_get_pubkey() fails in PKCS7_verify(). Fix memory leak when
2798 PKCS7_verify() fails with non detached data.
2801 *) Don't use getenv in library functions when run as setuid/setgid.
2802 New function OPENSSL_issetugid().
2805 *) Avoid false positives in memory leak detection code (crypto/mem_dbg.c)
2806 due to incorrect handling of multi-threading:
2808 1. Fix timing glitch in the MemCheck_off() portion of CRYPTO_mem_ctrl().
2810 2. Fix logical glitch in is_MemCheck_on() aka CRYPTO_is_mem_check_on().
2812 3. Count how many times MemCheck_off() has been called so that
2813 nested use can be treated correctly. This also avoids
2814 inband-signalling in the previous code (which relied on the
2815 assumption that thread ID 0 is impossible).
2818 *) Add "-rand" option also to s_client and s_server.
2821 *) Fix CPU detection on Irix 6.x.
2822 [Kurt Hockenbury <khockenb@stevens-tech.edu> and
2823 "Bruce W. Forsberg" <bruce.forsberg@baesystems.com>]
2825 *) Fix X509_NAME bug which produced incorrect encoding if X509_NAME
2828 [This change does not apply to 0.9.7.]
2830 *) Use the cached encoding of an X509_NAME structure rather than
2831 copying it. This is apparently the reason for the libsafe "errors"
2832 but the code is actually correct.
2835 *) Add new function BN_rand_range(), and fix DSA_sign_setup() to prevent
2836 Bleichenbacher's DSA attack.
2837 Extend BN_[pseudo_]rand: As before, top=1 forces the highest two bits
2838 to be set and top=0 forces the highest bit to be set; top=-1 is new
2839 and leaves the highest bit random.
2840 [Ulf Moeller, Bodo Moeller]
2842 *) In the NCONF_...-based implementations for CONF_... queries
2843 (crypto/conf/conf_lib.c), if the input LHASH is NULL, avoid using
2844 a temporary CONF structure with the data component set to NULL
2845 (which gives segmentation faults in lh_retrieve).
2846 Instead, use NULL for the CONF pointer in CONF_get_string and
2847 CONF_get_number (which may use environment variables) and directly
2848 return NULL from CONF_get_section.
2851 *) Fix potential buffer overrun for EBCDIC.
2854 *) Tolerate nonRepudiation as being valid for S/MIME signing and certSign
2855 keyUsage if basicConstraints absent for a CA.
2858 *) Make SMIME_write_PKCS7() write mail header values with a format that
2859 is more generally accepted (no spaces before the semicolon), since
2860 some programs can't parse those values properly otherwise. Also make
2861 sure BIO's that break lines after each write do not create invalid
2865 *) Make the CRL encoding routines work with empty SEQUENCE OF. The
2866 macros previously used would not encode an empty SEQUENCE OF
2867 and break the signature.
2869 [This change does not apply to 0.9.7.]
2871 *) Zero the premaster secret after deriving the master secret in
2875 *) Add some EVP_add_digest_alias registrations (as found in
2876 OpenSSL_add_all_digests()) to SSL_library_init()
2877 aka OpenSSL_add_ssl_algorithms(). This provides improved
2878 compatibility with peers using X.509 certificates
2879 with unconventional AlgorithmIdentifier OIDs.
2882 *) Fix for Irix with NO_ASM.
2883 ["Bruce W. Forsberg" <bruce.forsberg@baesystems.com>]
2885 *) ./config script fixes.
2886 [Ulf Moeller, Richard Levitte]
2888 *) Fix 'openssl passwd -1'.
2891 *) Change PKCS12_key_gen_asc() so it can cope with non null
2892 terminated strings whose length is passed in the passlen
2893 parameter, for example from PEM callbacks. This was done
2894 by adding an extra length parameter to asc2uni().
2895 [Steve Henson, reported by <oddissey@samsung.co.kr>]
2897 *) Fix C code generated by 'openssl dsaparam -C': If a BN_bin2bn
2898 call failed, free the DSA structure.
2901 *) Fix to uni2asc() to cope with zero length Unicode strings.
2902 These are present in some PKCS#12 files.
2905 *) Increase s2->wbuf allocation by one byte in ssl2_new (ssl/s2_lib.c).
2906 Otherwise do_ssl_write (ssl/s2_pkt.c) will write beyond buffer limits
2907 when writing a 32767 byte record.
2908 [Bodo Moeller; problem reported by Eric Day <eday@concentric.net>]
2910 *) In RSA_eay_public_{en,ed}crypt and RSA_eay_mod_exp (rsa_eay.c),
2911 obtain lock CRYPTO_LOCK_RSA before setting rsa->_method_mod_{n,p,q}.
2913 (RSA objects have a reference count access to which is protected
2914 by CRYPTO_LOCK_RSA [see rsa_lib.c, s3_srvr.c, ssl_cert.c, ssl_rsa.c],
2915 so they are meant to be shared between threads.)
2916 [Bodo Moeller, Geoff Thorpe; original patch submitted by
2917 "Reddie, Steven" <Steven.Reddie@ca.com>]
2919 *) Fix a deadlock in CRYPTO_mem_leaks().
2922 *) Use better test patterns in bntest.
2925 *) rand_win.c fix for Borland C.
2928 *) BN_rshift bugfix for n == 0.
2931 *) Add a 'bctest' script that checks for some known 'bc' bugs
2932 so that 'make test' does not abort just because 'bc' is broken.
2935 *) Store verify_result within SSL_SESSION also for client side to
2936 avoid potential security hole. (Re-used sessions on the client side
2937 always resulted in verify_result==X509_V_OK, not using the original
2938 result of the server certificate verification.)
2941 *) Fix ssl3_pending: If the record in s->s3->rrec is not of type
2942 SSL3_RT_APPLICATION_DATA, return 0.
2943 Similarly, change ssl2_pending to return 0 if SSL_in_init(s) is true.
2947 Both ssl2_peek and ssl3_peek, which were totally broken in earlier
2948 releases, have been re-implemented by renaming the previous
2949 implementations of ssl2_read and ssl3_read to ssl2_read_internal
2950 and ssl3_read_internal, respectively, and adding 'peek' parameters
2951 to them. The new ssl[23]_{read,peek} functions are calls to
2952 ssl[23]_read_internal with the 'peek' flag set appropriately.
2953 A 'peek' parameter has also been added to ssl3_read_bytes, which
2954 does the actual work for ssl3_read_internal.
2957 *) Initialise "ex_data" member of RSA/DSA/DH structures prior to calling
2958 the method-specific "init()" handler. Also clean up ex_data after
2959 calling the method-specific "finish()" handler. Previously, this was
2960 happening the other way round.
2963 *) Increase BN_CTX_NUM (the number of BIGNUMs in a BN_CTX) to 16.
2964 The previous value, 12, was not always sufficient for BN_mod_exp().
2967 *) Make sure that shared libraries get the internal name engine with
2968 the full version number and not just 0. This should mark the
2969 shared libraries as not backward compatible. Of course, this should
2970 be changed again when we can guarantee backward binary compatibility.
2973 *) Fix typo in get_cert_by_subject() in by_dir.c
2974 [Jean-Marc Desperrier <jean-marc.desperrier@certplus.com>]
2976 *) Rework the system to generate shared libraries:
2978 - Make note of the expected extension for the shared libraries and
2979 if there is a need for symbolic links from for example libcrypto.so.0
2980 to libcrypto.so.0.9.7. There is extended info in Configure for
2983 - Make as few rebuilds of the shared libraries as possible.
2985 - Still avoid linking the OpenSSL programs with the shared libraries.
2987 - When installing, install the shared libraries separately from the
2991 *) Fix SSL_CTX_set_read_ahead macro to actually use its argument.
2993 Copy SSL_CTX's read_ahead flag to SSL object directly in SSL_new
2994 and not in SSL_clear because the latter is also used by the
2995 accept/connect functions; previously, the settings made by
2996 SSL_set_read_ahead would be lost during the handshake.
2997 [Bodo Moeller; problems reported by Anders Gertz <gertz@epact.se>]
2999 *) Correct util/mkdef.pl to be selective about disabled algorithms.
3000 Previously, it would create entries for disableed algorithms no
3004 *) Added several new manual pages for SSL_* function.
3007 Changes between 0.9.5a and 0.9.6 [24 Sep 2000]
3009 *) In ssl23_get_client_hello, generate an error message when faced
3010 with an initial SSL 3.0/TLS record that is too small to contain the
3011 first two bytes of the ClientHello message, i.e. client_version.
3012 (Note that this is a pathologic case that probably has never happened
3013 in real life.) The previous approach was to use the version number
3014 from the record header as a substitute; but our protocol choice
3015 should not depend on that one because it is not authenticated
3016 by the Finished messages.
3019 *) More robust randomness gathering functions for Windows.
3020 [Jeffrey Altman <jaltman@columbia.edu>]
3022 *) For compatibility reasons if the flag X509_V_FLAG_ISSUER_CHECK is
3023 not set then we don't setup the error code for issuer check errors
3024 to avoid possibly overwriting other errors which the callback does
3025 handle. If an application does set the flag then we assume it knows
3026 what it is doing and can handle the new informational codes
3030 *) Fix for a nasty bug in ASN1_TYPE handling. ASN1_TYPE is used for
3031 a general "ANY" type, as such it should be able to decode anything
3032 including tagged types. However it didn't check the class so it would
3033 wrongly interpret tagged types in the same way as their universal
3034 counterpart and unknown types were just rejected. Changed so that the
3035 tagged and unknown types are handled in the same way as a SEQUENCE:
3036 that is the encoding is stored intact. There is also a new type
3037 "V_ASN1_OTHER" which is used when the class is not universal, in this
3038 case we have no idea what the actual type is so we just lump them all
3042 *) On VMS, stdout may very well lead to a file that is written to
3043 in a record-oriented fashion. That means that every write() will
3044 write a separate record, which will be read separately by the
3045 programs trying to read from it. This can be very confusing.
3047 The solution is to put a BIO filter in the way that will buffer
3048 text until a linefeed is reached, and then write everything a
3049 line at a time, so every record written will be an actual line,
3050 not chunks of lines and not (usually doesn't happen, but I've
3051 seen it once) several lines in one record. BIO_f_linebuffer() is
3054 Currently, it's a VMS-only method, because that's where it has
3055 been tested well enough.
3058 *) Remove 'optimized' squaring variant in BN_mod_mul_montgomery,
3059 it can return incorrect results.
3060 (Note: The buggy variant was not enabled in OpenSSL 0.9.5a,
3061 but it was in 0.9.6-beta[12].)
3064 *) Disable the check for content being present when verifying detached
3065 signatures in pk7_smime.c. Some versions of Netscape (wrongly)
3066 include zero length content when signing messages.
3069 *) New BIO_shutdown_wr macro, which invokes the BIO_C_SHUTDOWN_WR
3070 BIO_ctrl (for BIO pairs).
3073 *) Add DSO method for VMS.
3076 *) Bug fix: Montgomery multiplication could produce results with the
3080 *) Add RPM specification openssl.spec and modify it to build three
3081 packages. The default package contains applications, application
3082 documentation and run-time libraries. The devel package contains
3083 include files, static libraries and function documentation. The
3084 doc package contains the contents of the doc directory. The original
3085 openssl.spec was provided by Damien Miller <djm@mindrot.org>.
3088 *) Add a large number of documentation files for many SSL routines.
3089 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>]
3091 *) Add a configuration entry for Sony News 4.
3092 [NAKAJI Hiroyuki <nakaji@tutrp.tut.ac.jp>]
3094 *) Don't set the two most significant bits to one when generating a
3095 random number < q in the DSA library.
3098 *) New SSL API mode 'SSL_MODE_AUTO_RETRY'. This disables the default
3099 behaviour that SSL_read may result in SSL_ERROR_WANT_READ (even if
3100 the underlying transport is blocking) if a handshake took place.
3101 (The default behaviour is needed by applications such as s_client
3102 and s_server that use select() to determine when to use SSL_read;
3103 but for applications that know in advance when to expect data, it
3104 just makes things more complicated.)
3107 *) Add RAND_egd_bytes(), which gives control over the number of bytes read
3111 *) Add a few more EBCDIC conditionals that make `req' and `x509'
3112 work better on such systems.
3113 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>]
3115 *) Add two demo programs for PKCS12_parse() and PKCS12_create().
3116 Update PKCS12_parse() so it copies the friendlyName and the
3117 keyid to the certificates aux info.
3120 *) Fix bug in PKCS7_verify() which caused an infinite loop
3121 if there was more than one signature.
3122 [Sven Uszpelkat <su@celocom.de>]
3124 *) Major change in util/mkdef.pl to include extra information
3125 about each symbol, as well as presentig variables as well
3126 as functions. This change means that there's n more need
3127 to rebuild the .num files when some algorithms are excluded.
3130 *) Allow the verify time to be set by an application,
3131 rather than always using the current time.
3134 *) Phase 2 verify code reorganisation. The certificate
3135 verify code now looks up an issuer certificate by a
3136 number of criteria: subject name, authority key id
3137 and key usage. It also verifies self signed certificates
3138 by the same criteria. The main comparison function is
3139 X509_check_issued() which performs these checks.
3141 Lot of changes were necessary in order to support this
3142 without completely rewriting the lookup code.
3144 Authority and subject key identifier are now cached.
3146 The LHASH 'certs' is X509_STORE has now been replaced
3147 by a STACK_OF(X509_OBJECT). This is mainly because an
3148 LHASH can't store or retrieve multiple objects with
3149 the same hash value.
3151 As a result various functions (which were all internal
3152 use only) have changed to handle the new X509_STORE
3153 structure. This will break anything that messed round
3154 with X509_STORE internally.
3156 The functions X509_STORE_add_cert() now checks for an
3157 exact match, rather than just subject name.
3159 The X509_STORE API doesn't directly support the retrieval
3160 of multiple certificates matching a given criteria, however
3161 this can be worked round by performing a lookup first
3162 (which will fill the cache with candidate certificates)
3163 and then examining the cache for matches. This is probably
3164 the best we can do without throwing out X509_LOOKUP
3165 entirely (maybe later...).
3167 The X509_VERIFY_CTX structure has been enhanced considerably.
3169 All certificate lookup operations now go via a get_issuer()
3170 callback. Although this currently uses an X509_STORE it
3171 can be replaced by custom lookups. This is a simple way
3172 to bypass the X509_STORE hackery necessary to make this
3173 work and makes it possible to use more efficient techniques
3174 in future. A very simple version which uses a simple
3175 STACK for its trusted certificate store is also provided
3176 using X509_STORE_CTX_trusted_stack().
3178 The verify_cb() and verify() callbacks now have equivalents
3179 in the X509_STORE_CTX structure.
3181 X509_STORE_CTX also has a 'flags' field which can be used
3182 to customise the verify behaviour.
3185 *) Add new PKCS#7 signing option PKCS7_NOSMIMECAP which
3186 excludes S/MIME capabilities.
3189 *) When a certificate request is read in keep a copy of the
3190 original encoding of the signed data and use it when outputing
3191 again. Signatures then use the original encoding rather than
3192 a decoded, encoded version which may cause problems if the
3193 request is improperly encoded.
3196 *) For consistency with other BIO_puts implementations, call
3197 buffer_write(b, ...) directly in buffer_puts instead of calling
3200 In BIO_puts, increment b->num_write as in BIO_write.
3201 [Peter.Sylvester@EdelWeb.fr]
3203 *) Fix BN_mul_word for the case where the word is 0. (We have to use
3204 BN_zero, we may not return a BIGNUM with an array consisting of
3208 *) Avoid calling abort() from within the library when problems are
3209 detected, except if preprocessor symbols have been defined
3210 (such as REF_CHECK, BN_DEBUG etc.).
3213 *) New openssl application 'rsautl'. This utility can be
3214 used for low level RSA operations. DER public key
3215 BIO/fp routines also added.
3218 *) New Configure entry and patches for compiling on QNX 4.
3219 [Andreas Schneider <andreas@ds3.etech.fh-hamburg.de>]
3221 *) A demo state-machine implementation was sponsored by
3222 Nuron (http://www.nuron.com/) and is now available in
3223 demos/state_machine.
3226 *) New options added to the 'dgst' utility for signature
3227 generation and verification.
3230 *) Unrecognized PKCS#7 content types are now handled via a
3231 catch all ASN1_TYPE structure. This allows unsupported
3232 types to be stored as a "blob" and an application can
3233 encode and decode it manually.
3236 *) Fix various signed/unsigned issues to make a_strex.c
3238 [Oscar Jacobsson <oscar.jacobsson@celocom.com>]
3240 *) ASN1 fixes. i2d_ASN1_OBJECT was not returning the correct
3241 length if passed a buffer. ASN1_INTEGER_to_BN failed
3242 if passed a NULL BN and its argument was negative.
3243 [Steve Henson, pointed out by Sven Heiberg <sven@tartu.cyber.ee>]
3245 *) Modification to PKCS#7 encoding routines to output definite
3246 length encoding. Since currently the whole structures are in
3247 memory there's not real point in using indefinite length
3248 constructed encoding. However if OpenSSL is compiled with
3249 the flag PKCS7_INDEFINITE_ENCODING the old form is used.
3252 *) Added BIO_vprintf() and BIO_vsnprintf().
3255 *) Added more prefixes to parse for in the the strings written
3256 through a logging bio, to cover all the levels that are available
3257 through syslog. The prefixes are now:
3259 PANIC, EMERG, EMR => LOG_EMERG
3260 ALERT, ALR => LOG_ALERT
3261 CRIT, CRI => LOG_CRIT
3262 ERROR, ERR => LOG_ERR
3263 WARNING, WARN, WAR => LOG_WARNING
3264 NOTICE, NOTE, NOT => LOG_NOTICE
3265 INFO, INF => LOG_INFO
3266 DEBUG, DBG => LOG_DEBUG
3268 and as before, if none of those prefixes are present at the
3269 beginning of the string, LOG_ERR is chosen.
3271 On Win32, the LOG_* levels are mapped according to this:
3273 LOG_EMERG, LOG_ALERT, LOG_CRIT, LOG_ERR => EVENTLOG_ERROR_TYPE
3274 LOG_WARNING => EVENTLOG_WARNING_TYPE
3275 LOG_NOTICE, LOG_INFO, LOG_DEBUG => EVENTLOG_INFORMATION_TYPE
3279 *) Made it possible to reconfigure with just the configuration
3280 argument "reconf" or "reconfigure". The command line arguments
3281 are stored in Makefile.ssl in the variable CONFIGURE_ARGS,
3282 and are retrieved from there when reconfiguring.
3286 [Assar Westerlund <assar@sics.se>, Richard Levitte]
3288 *) Add the arguments -CAfile and -CApath to the pkcs12 utility.
3291 *) The obj_dat.pl script was messing up the sorting of object
3292 names. The reason was that it compared the quoted version
3293 of strings as a result "OCSP" > "OCSP Signing" because
3294 " > SPACE. Changed script to store unquoted versions of
3295 names and add quotes on output. It was also omitting some
3296 names from the lookup table if they were given a default
3297 value (that is if SN is missing it is given the same
3298 value as LN and vice versa), these are now added on the
3299 grounds that if an object has a name we should be able to
3300 look it up. Finally added warning output when duplicate
3301 short or long names are found.
3304 *) Changes needed for Tandem NSK.
3305 [Scott Uroff <scott@xypro.com>]
3307 *) Fix SSL 2.0 rollback checking: Due to an off-by-one error in
3308 RSA_padding_check_SSLv23(), special padding was never detected
3309 and thus the SSL 3.0/TLS 1.0 countermeasure against protocol
3310 version rollback attacks was not effective.
3312 In s23_clnt.c, don't use special rollback-attack detection padding
3313 (RSA_SSLV23_PADDING) if SSL 2.0 is the only protocol enabled in the
3314 client; similarly, in s23_srvr.c, don't do the rollback check if
3315 SSL 2.0 is the only protocol enabled in the server.
3318 *) Make it possible to get hexdumps of unprintable data with 'openssl
3319 asn1parse'. By implication, the functions ASN1_parse_dump() and
3320 BIO_dump_indent() are added.
3323 *) New functions ASN1_STRING_print_ex() and X509_NAME_print_ex()
3324 these print out strings and name structures based on various
3325 flags including RFC2253 support and proper handling of
3326 multibyte characters. Added options to the 'x509' utility
3327 to allow the various flags to be set.
3330 *) Various fixes to use ASN1_TIME instead of ASN1_UTCTIME.
3331 Also change the functions X509_cmp_current_time() and
3332 X509_gmtime_adj() work with an ASN1_TIME structure,
3333 this will enable certificates using GeneralizedTime in validity
3334 dates to be checked.
3337 *) Make the NEG_PUBKEY_BUG code (which tolerates invalid
3338 negative public key encodings) on by default,
3339 NO_NEG_PUBKEY_BUG can be set to disable it.
3342 *) New function c2i_ASN1_OBJECT() which acts on ASN1_OBJECT
3343 content octets. An i2c_ASN1_OBJECT is unnecessary because
3344 the encoding can be trivially obtained from the structure.
3347 *) crypto/err.c locking bugfix: Use write locks (CRYPTO_w_[un]lock),
3348 not read locks (CRYPTO_r_[un]lock).
3351 *) A first attempt at creating official support for shared
3352 libraries through configuration. I've kept it so the
3353 default is static libraries only, and the OpenSSL programs
3354 are always statically linked for now, but there are
3355 preparations for dynamic linking in place.
3356 This has been tested on Linux and Tru64.
3359 *) Randomness polling function for Win9x, as described in:
3360 Peter Gutmann, Software Generation of Practically Strong
3364 *) Fix so PRNG is seeded in req if using an already existing
3368 *) New options to smime application. -inform and -outform
3369 allow alternative formats for the S/MIME message including
3370 PEM and DER. The -content option allows the content to be
3371 specified separately. This should allow things like Netscape
3372 form signing output easier to verify.
3375 *) Fix the ASN1 encoding of tags using the 'long form'.
3378 *) New ASN1 functions, i2c_* and c2i_* for INTEGER and BIT
3379 STRING types. These convert content octets to and from the
3380 underlying type. The actual tag and length octets are
3381 already assumed to have been read in and checked. These
3382 are needed because all other string types have virtually
3383 identical handling apart from the tag. By having versions
3384 of the ASN1 functions that just operate on content octets
3385 IMPLICIT tagging can be handled properly. It also allows
3386 the ASN1_ENUMERATED code to be cut down because ASN1_ENUMERATED
3387 and ASN1_INTEGER are identical apart from the tag.
3390 *) Change the handling of OID objects as follows:
3392 - New object identifiers are inserted in objects.txt, following
3393 the syntax given in objects.README.
3394 - objects.pl is used to process obj_mac.num and create a new
3396 - obj_dat.pl is used to create a new obj_dat.h, using the data in
3399 This is currently kind of a hack, and the perl code in objects.pl
3400 isn't very elegant, but it works as I intended. The simplest way
3401 to check that it worked correctly is to look in obj_dat.h and
3402 check the array nid_objs and make sure the objects haven't moved
3403 around (this is important!). Additions are OK, as well as
3404 consistent name changes.
3407 *) Add BSD-style MD5-based passwords to 'openssl passwd' (option '-1').
3410 *) Addition of the command line parameter '-rand file' to 'openssl req'.
3411 The given file adds to whatever has already been seeded into the
3412 random pool through the RANDFILE configuration file option or
3413 environment variable, or the default random state file.
3416 *) mkstack.pl now sorts each macro group into lexical order.
3417 Previously the output order depended on the order the files
3418 appeared in the directory, resulting in needless rewriting
3422 *) Patches to make OpenSSL compile under Win32 again. Mostly
3423 work arounds for the VC++ problem that it treats func() as
3424 func(void). Also stripped out the parts of mkdef.pl that
3425 added extra typesafe functions: these no longer exist.
3428 *) Reorganisation of the stack code. The macros are now all
3429 collected in safestack.h . Each macro is defined in terms of
3430 a "stack macro" of the form SKM_<name>(type, a, b). The
3431 DEBUG_SAFESTACK is now handled in terms of function casts,
3432 this has the advantage of retaining type safety without the
3433 use of additional functions. If DEBUG_SAFESTACK is not defined
3434 then the non typesafe macros are used instead. Also modified the
3435 mkstack.pl script to handle the new form. Needs testing to see
3436 if which (if any) compilers it chokes and maybe make DEBUG_SAFESTACK
3437 the default if no major problems. Similar behaviour for ASN1_SET_OF
3438 and PKCS12_STACK_OF.
3441 *) When some versions of IIS use the 'NET' form of private key the
3442 key derivation algorithm is different. Normally MD5(password) is
3443 used as a 128 bit RC4 key. In the modified case
3444 MD5(MD5(password) + "SGCKEYSALT") is used insted. Added some
3445 new functions i2d_RSA_NET(), d2i_RSA_NET() etc which are the same
3446 as the old Netscape_RSA functions except they have an additional
3447 'sgckey' parameter which uses the modified algorithm. Also added
3448 an -sgckey command line option to the rsa utility. Thanks to
3449 Adrian Peck <bertie@ncipher.com> for posting details of the modified
3450 algorithm to openssl-dev.
3453 *) The evp_local.h macros were using 'c.##kname' which resulted in
3454 invalid expansion on some systems (SCO 5.0.5 for example).
3455 Corrected to 'c.kname'.
3456 [Phillip Porch <root@theporch.com>]
3458 *) New X509_get1_email() and X509_REQ_get1_email() functions that return
3459 a STACK of email addresses from a certificate or request, these look
3460 in the subject name and the subject alternative name extensions and
3461 omit any duplicate addresses.
3464 *) Re-implement BN_mod_exp2_mont using independent (and larger) windows.
3465 This makes DSA verification about 2 % faster.
3468 *) Increase maximum window size in BN_mod_exp_... to 6 bits instead of 5
3469 (meaning that now 2^5 values will be precomputed, which is only 4 KB
3470 plus overhead for 1024 bit moduli).
3471 This makes exponentiations about 0.5 % faster for 1024 bit
3472 exponents (as measured by "openssl speed rsa2048").
3475 *) Rename memory handling macros to avoid conflicts with other
3477 Malloc => OPENSSL_malloc
3478 Malloc_locked => OPENSSL_malloc_locked
3479 Realloc => OPENSSL_realloc
3480 Free => OPENSSL_free
3483 *) New function BN_mod_exp_mont_word for small bases (roughly 15%
3484 faster than BN_mod_exp_mont, i.e. 7% for a full DH exchange).
3487 *) CygWin32 support.
3488 [John Jarvie <jjarvie@newsguy.com>]
3490 *) The type-safe stack code has been rejigged. It is now only compiled
3491 in when OpenSSL is configured with the DEBUG_SAFESTACK option and
3492 by default all type-specific stack functions are "#define"d back to
3493 standard stack functions. This results in more streamlined output
3494 but retains the type-safety checking possibilities of the original
3498 *) The STACK code has been cleaned up, and certain type declarations
3499 that didn't make a lot of sense have been brought in line. This has
3500 also involved a cleanup of sorts in safestack.h to more correctly
3501 map type-safe stack functions onto their plain stack counterparts.
3502 This work has also resulted in a variety of "const"ifications of
3503 lots of the code, especially "_cmp" operations which should normally
3504 be prototyped with "const" parameters anyway.
3507 *) When generating bytes for the first time in md_rand.c, 'stir the pool'
3508 by seeding with STATE_SIZE dummy bytes (with zero entropy count).
3509 (The PRNG state consists of two parts, the large pool 'state' and 'md',
3510 where all of 'md' is used each time the PRNG is used, but 'state'
3511 is used only indexed by a cyclic counter. As entropy may not be
3512 well distributed from the beginning, 'md' is important as a
3513 chaining variable. However, the output function chains only half
3514 of 'md', i.e. 80 bits. ssleay_rand_add, on the other hand, chains
3515 all of 'md', and seeding with STATE_SIZE dummy bytes will result
3516 in all of 'state' being rewritten, with the new values depending
3517 on virtually all of 'md'. This overcomes the 80 bit limitation.)
3520 *) In ssl/s2_clnt.c and ssl/s3_clnt.c, call ERR_clear_error() when
3521 the handshake is continued after ssl_verify_cert_chain();
3522 otherwise, if SSL_VERIFY_NONE is set, remaining error codes
3523 can lead to 'unexplainable' connection aborts later.
3524 [Bodo Moeller; problem tracked down by Lutz Jaenicke]
3526 *) Major EVP API cipher revision.
3527 Add hooks for extra EVP features. This allows various cipher
3528 parameters to be set in the EVP interface. Support added for variable
3529 key length ciphers via the EVP_CIPHER_CTX_set_key_length() function and
3530 setting of RC2 and RC5 parameters.
3532 Modify EVP_OpenInit() and EVP_SealInit() to cope with variable key length
3535 Remove lots of duplicated code from the EVP library. For example *every*
3536 cipher init() function handles the 'iv' in the same way according to the
3537 cipher mode. They also all do nothing if the 'key' parameter is NULL and
3538 for CFB and OFB modes they zero ctx->num.
3540 New functionality allows removal of S/MIME code RC2 hack.
3542 Most of the routines have the same form and so can be declared in terms
3545 By shifting this to the top level EVP_CipherInit() it can be removed from
3546 all individual ciphers. If the cipher wants to handle IVs or keys
3547 differently it can set the EVP_CIPH_CUSTOM_IV or EVP_CIPH_ALWAYS_CALL_INIT
3550 Change lots of functions like EVP_EncryptUpdate() to now return a
3551 value: although software versions of the algorithms cannot fail
3552 any installed hardware versions can.
3555 *) Implement SSL_OP_TLS_ROLLBACK_BUG: In ssl3_get_client_key_exchange, if
3556 this option is set, tolerate broken clients that send the negotiated
3557 protocol version number instead of the requested protocol version
3561 *) Call dh_tmp_cb (set by ..._TMP_DH_CB) with correct 'is_export' flag;
3562 i.e. non-zero for export ciphersuites, zero otherwise.
3563 Previous versions had this flag inverted, inconsistent with
3564 rsa_tmp_cb (..._TMP_RSA_CB).
3565 [Bodo Moeller; problem reported by Amit Chopra]
3567 *) Add missing DSA library text string. Work around for some IIS
3568 key files with invalid SEQUENCE encoding.
3571 *) Add a document (doc/standards.txt) that list all kinds of standards
3572 and so on that are implemented in OpenSSL.
3575 *) Enhance c_rehash script. Old version would mishandle certificates
3576 with the same subject name hash and wouldn't handle CRLs at all.
3577 Added -fingerprint option to crl utility, to support new c_rehash
3581 *) Eliminate non-ANSI declarations in crypto.h and stack.h.
3584 *) Fix for SSL server purpose checking. Server checking was
3585 rejecting certificates which had extended key usage present
3586 but no ssl client purpose.
3587 [Steve Henson, reported by Rene Grosser <grosser@hisolutions.com>]
3589 *) Make PKCS#12 code work with no password. The PKCS#12 spec
3590 is a little unclear about how a blank password is handled.
3591 Since the password in encoded as a BMPString with terminating
3592 double NULL a zero length password would end up as just the
3593 double NULL. However no password at all is different and is
3594 handled differently in the PKCS#12 key generation code. NS
3595 treats a blank password as zero length. MSIE treats it as no
3596 password on export: but it will try both on import. We now do
3597 the same: PKCS12_parse() tries zero length and no password if
3598 the password is set to "" or NULL (NULL is now a valid password:
3599 it wasn't before) as does the pkcs12 application.
3602 *) Bugfixes in apps/x509.c: Avoid a memory leak; and don't use
projects / openssl.git / blob