5 Changes between 0.9.7 and 0.9.8 [xx XXX 2002]
7 *) Change the ENGINE framework to automatically load engines
8 dynamically from specific directories unless they could be
9 found to already be built in or loaded. Move all the
10 current engines except for the cryptodev one to a new
12 The engines in engines/ are built as shared libraries if
13 the "shared" options was given to ./Configure or ./config.
14 Otherwise, they are inserted in libcrypto.a.
15 /usr/local/ssl/engines is the default directory for dynamic
16 engines, but that can be overriden at configure time through
17 the usual use of --prefix and/or --openssldir, and at run
18 time with the environment variable OPENSSL_ENGINES.
19 [Geoff Thorpe and Richard Levitte]
21 *) Add Makefile.shared, a helper makefile to build shared
22 libraries. Addapt Makefile.org.
25 *) Add version info to Win32 DLLs.
26 [Peter 'Luna' Runestig" <peter@runestig.com>]
28 *) Add new 'medium level' PKCS#12 API. Certificates and keys
29 can be added using this API to created arbitrary PKCS#12
30 files while avoiding the low level API.
32 New options to PKCS12_create(), key or cert can be NULL and
33 will then be omitted from the output file. The encryption
34 algorithm NIDs can be set to -1 for no encryption, the mac
35 iteration count can be set to 0 to omit the mac.
37 Enhance pkcs12 utility by making the -nokeys and -nocerts
38 options work when creating a PKCS#12 file. New option -nomac
39 to omit the mac, NONE can be set for an encryption algorithm.
40 New code is modified to use the enhanced PKCS12_create()
41 instead of the low level API.
44 *) Extend ASN1 encoder to support indefinite length constructed
45 encoding. This can output sequences tags and octet strings in
46 this form. Modify pk7_asn1.c to support indefinite length
47 encoding. This is experimental and needs additional code to
48 be useful, such as an ASN1 bio and some enhanced streaming
51 Extend template encode functionality so that tagging is passed
52 down to the template encoder.
55 *) Let 'openssl req' fail if an argument to '-newkey' is not
56 recognized instead of using RSA as a default.
59 *) Add support for ECC-based ciphersuites from draft-ietf-tls-ecc-01.txt.
60 As these are not official, they are not included in "ALL";
61 the "ECCdraft" ciphersuite group alias can be used to select them.
62 [Vipul Gupta and Sumit Gupta (Sun Microsystems Laboratories)]
64 *) Add ECDH engine support.
65 [Nils Gura and Douglas Stebila (Sun Microsystems Laboratories)]
67 *) Add ECDH in new directory crypto/ecdh/.
68 TODO: more general interface (return x coordinate, not its hash)
69 TODO: bug: pad x with leading zeros if necessary
70 [Douglas Stebila (Sun Microsystems Laboratories)]
72 *) Let BN_rand_range() abort with an error after 100 iterations
73 without success (which indicates a broken PRNG).
76 *) Change BN_mod_sqrt() so that it verifies that the input value
77 is really the square of the return value. (Previously,
78 BN_mod_sqrt would show GIGO behaviour.)
81 *) Add named elliptic curves over binary fields from X9.62, SECG,
82 and WAP/WTLS; add OIDs that were still missing.
84 [Sheueling Chang Shantz and Douglas Stebila
85 (Sun Microsystems Laboratories)]
87 *) Extend the EC library for elliptic curves over binary fields
88 (new files ec2_smpl.c, ec2_smpt.c, ec2_mult.c in crypto/ec/).
95 EC_GROUP_new_curve_GF2m
96 EC_GROUP_set_curve_GF2m
97 EC_GROUP_get_curve_GF2m
98 EC_POINT_set_affine_coordinates_GF2m
99 EC_POINT_get_affine_coordinates_GF2m
100 EC_POINT_set_compressed_coordinates_GF2m
102 Point compression for binary fields is disabled by default for
103 patent reasons (compile with OPENSSL_EC_BIN_PT_COMP defined to
106 As binary polynomials are represented as BIGNUMs, various members
107 of the EC_GROUP and EC_POINT data structures can be shared
108 between the implementations for prime fields and binary fields;
109 the above ..._GF2m functions (except for EX_GROUP_new_curve_GF2m)
110 are essentially identical to their ..._GFp counterparts.
111 (For simplicity, the '..._GFp' prefix has been dropped from
112 various internal method names.)
114 An internal 'field_div' method (similar to 'field_mul' and
115 'field_sqr') has been added; this is used only for binary fields.
117 [Sheueling Chang Shantz and Douglas Stebila
118 (Sun Microsystems Laboratories)]
120 *) Optionally dispatch EC_POINT_mul(), EC_POINT_precompute_mult()
121 through methods ('mul', 'precompute_mult').
123 The generic implementations (now internally called 'ec_wNAF_mul'
124 and 'ec_wNAF_precomputed_mult') remain the default if these
125 methods are undefined.
127 [Sheueling Chang Shantz and Douglas Stebila
128 (Sun Microsystems Laboratories)]
130 *) New function EC_GROUP_get_degree, which is defined through
131 EC_METHOD. For curves over prime fields, this returns the bit
132 length of the modulus.
134 [Sheueling Chang Shantz and Douglas Stebila
135 (Sun Microsystems Laboratories)]
137 *) New functions EC_GROUP_dup, EC_POINT_dup.
138 (These simply call ..._new and ..._copy).
140 [Sheueling Chang Shantz and Douglas Stebila
141 (Sun Microsystems Laboratories)]
143 *) Add binary polynomial arithmetic software in crypto/bn/bn_gf2m.c.
144 Polynomials are represented as BIGNUMs (where the sign bit is not
145 used) in the following functions [macros]:
148 BN_GF2m_sub [= BN_GF2m_add]
149 BN_GF2m_mod [wrapper for BN_GF2m_mod_arr]
150 BN_GF2m_mod_mul [wrapper for BN_GF2m_mod_mul_arr]
151 BN_GF2m_mod_sqr [wrapper for BN_GF2m_mod_sqr_arr]
153 BN_GF2m_mod_exp [wrapper for BN_GF2m_mod_exp_arr]
154 BN_GF2m_mod_sqrt [wrapper for BN_GF2m_mod_sqrt_arr]
155 BN_GF2m_mod_solve_quad [wrapper for BN_GF2m_mod_solve_quad_arr]
156 BN_GF2m_cmp [= BN_ucmp]
158 (Note that only the 'mod' functions are actually for fields GF(2^m).
159 BN_GF2m_add() is misnomer, but this is for the sake of consistency.)
161 For some functions, an the irreducible polynomial defining a
162 field can be given as an 'unsigned int[]' with strictly
163 decreasing elements giving the indices of those bits that are set;
164 i.e., p[] represents the polynomial
165 f(t) = t^p[0] + t^p[1] + ... + t^p[k]
167 p[0] > p[1] > ... > p[k] = 0.
168 This applies to the following functions:
173 BN_GF2m_mod_inv_arr [wrapper for BN_GF2m_mod_inv]
174 BN_GF2m_mod_div_arr [wrapper for BN_GF2m_mod_div]
177 BN_GF2m_mod_solve_quad_arr
181 Conversion can be performed by the following functions:
186 bntest.c has additional tests for binary polynomial arithmetic.
188 Two implementations for BN_GF2m_mod_div() are available.
189 The default algorithm simply uses BN_GF2m_mod_inv() and
190 BN_GF2m_mod_mul(). The alternative algorithm is compiled in only
191 if OPENSSL_SUN_GF2M_DIV is defined (patent pending; read the
192 copyright notice in crypto/bn/bn_gf2m.c before enabling it).
194 [Sheueling Chang Shantz and Douglas Stebila
195 (Sun Microsystems Laboratories)]
197 *) Add new error code 'ERR_R_DISABLED' that can be used when some
198 functionality is disabled at compile-time.
199 [Douglas Stebila <douglas.stebila@sun.com>]
201 *) Change default behaviour of 'openssl asn1parse' so that more
202 information is visible when viewing, e.g., a certificate:
204 Modify asn1_parse2 (crypto/asn1/asn1_par.c) so that in non-'dump'
205 mode the content of non-printable OCTET STRINGs is output in a
206 style similar to INTEGERs, but with '[HEX DUMP]' prepended to
207 avoid the appearance of a printable string.
208 [Nils Larsch <nla@trustcenter.de>]
210 *) Add 'asn1_flag' and 'asn1_form' member to EC_GROUP with access
212 EC_GROUP_set_asn1_flag()
213 EC_GROUP_get_asn1_flag()
214 EC_GROUP_set_point_conversion_form()
215 EC_GROUP_get_point_conversion_form()
216 These control ASN1 encoding details:
217 - Curves (i.e., groups) are encoded explicitly unless asn1_flag
218 has been set to OPENSSL_EC_NAMED_CURVE.
219 - Points are encoded in uncompressed form by default; options for
220 asn1_for are as for point2oct, namely
221 POINT_CONVERSION_COMPRESSED
222 POINT_CONVERSION_UNCOMPRESSED
223 POINT_CONVERSION_HYBRID
225 Also add 'seed' and 'seed_len' members to EC_GROUP with access
229 EC_GROUP_get_seed_len()
230 This is used only for ASN1 purposes (so far).
231 [Nils Larsch <nla@trustcenter.de>]
233 *) Add 'field_type' member to EC_METHOD, which holds the NID
234 of the appropriate field type OID. The new function
235 EC_METHOD_get_field_type() returns this value.
236 [Nils Larsch <nla@trustcenter.de>]
243 providing useful interfaces to EC_POINT_point2oct() and
244 EC_POINT_oct2point().
245 [Nils Larsch <nla@trustcenter.de>]
247 *) Change internals of the EC library so that the functions
248 EC_GROUP_set_generator()
249 EC_GROUP_get_generator()
251 EC_GROUP_get_cofactor()
252 are implemented directly in crypto/ec/ec_lib.c and not dispatched
253 to methods, which would lead to unnecessary code duplication when
254 adding different types of curves.
255 [Nils Larsch <nla@trustcenter.de> with input by Bodo Moeller]
257 *) Implement compute_wNAF (crypto/ec/ec_mult.c) without BIGNUM
258 arithmetic, and such that modified wNAFs are generated
259 (which avoid length expansion in many cases).
262 *) Add a function EC_GROUP_check_discriminant() (defined via
263 EC_METHOD) that verifies that the curve discriminant is non-zero.
265 Add a function EC_GROUP_check() that makes some sanity tests
266 on a EC_GROUP, its generator and order. This includes
267 EC_GROUP_check_discriminant().
268 [Nils Larsch <nla@trustcenter.de>]
270 *) Add ECDSA in new directory crypto/ecdsa/.
272 Add applications 'openssl ecparam' and 'openssl ecdsa'
273 (these are based on 'openssl dsaparam' and 'openssl dsa').
275 ECDSA support is also included in various other files across the
276 library. Most notably,
277 - 'openssl req' now has a '-newkey ecdsa:file' option;
278 - EVP_PKCS82PKEY (crypto/evp/evp_pkey.c) now can handle ECDSA;
279 - X509_PUBKEY_get (crypto/asn1/x_pubkey.c) and
280 d2i_PublicKey (crypto/asn1/d2i_pu.c) have been modified to make
281 them suitable for ECDSA where domain parameters must be
282 extracted before the specific public key;
283 - ECDSA engine support has been added.
284 [Nils Larsch <nla@trustcenter.de>]
286 *) Include some named elliptic curves, and add OIDs from X9.62,
287 SECG, and WAP/WTLS. Each curve can be obtained from the new
289 EC_GROUP_new_by_nid(),
290 and the list of available named curves can be obtained with
291 EC_get_builtin_curves().
292 Also add a 'curve_name' member to EC_GROUP objects, which can be
296 [Nils Larsch <nla@trustcenter.de, Bodo Moeller]
298 Changes between 0.9.6h and 0.9.7 [XX xxx 2002]
300 *) Change from security patch (see 0.9.6e below) that did not affect
301 the 0.9.6 release series:
303 Remote buffer overflow in SSL3 protocol - an attacker could
304 supply an oversized master key in Kerberos-enabled versions.
308 *) Change the SSL kerb5 codes to match RFC 2712.
311 *) Make -nameopt work fully for req and add -reqopt switch.
312 [Michael Bell <michael.bell@rz.hu-berlin.de>, Steve Henson]
314 *) The "block size" for block ciphers in CFB and OFB mode should be 1.
315 [Steve Henson, reported by Yngve Nysaeter Pettersen <yngve@opera.com>]
317 *) Make sure tests can be performed even if the corresponding algorithms
318 have been removed entirely. This was also the last step to make
319 OpenSSL compilable with DJGPP under all reasonable conditions.
320 [Richard Levitte, Doug Kaufman <dkaufman@rahul.net>]
322 *) Add cipher selection rules COMPLEMENTOFALL and COMPLEMENTOFDEFAULT
323 to allow version independent disabling of normally unselected ciphers,
324 which may be activated as a side-effect of selecting a single cipher.
326 (E.g., cipher list string "RSA" enables ciphersuites that are left
327 out of "ALL" because they do not provide symmetric encryption.
328 "RSA:!COMPLEMEMENTOFALL" avoids these unsafe ciphersuites.)
329 [Lutz Jaenicke, Bodo Moeller]
331 *) Add appropriate support for separate platform-dependent build
332 directories. The recommended way to make a platform-dependent
333 build directory is the following (tested on Linux), maybe with
336 # Place yourself outside of the OpenSSL source tree. In
337 # this example, the environment variable OPENSSL_SOURCE
338 # is assumed to contain the absolute OpenSSL source directory.
339 mkdir -p objtree/"`uname -s`-`uname -r`-`uname -m`"
340 cd objtree/"`uname -s`-`uname -r`-`uname -m`"
341 (cd $OPENSSL_SOURCE; find . -type f -o -type l) | while read F; do
342 mkdir -p `dirname $F`
343 ln -s $OPENSSL_SOURCE/$F $F
346 To be absolutely sure not to disturb the source tree, a "make clean"
347 is a good thing. If it isn't successfull, don't worry about it,
348 it probably means the source directory is very clean.
351 *) Make sure any ENGINE control commands make local copies of string
352 pointers passed to them whenever necessary. Otherwise it is possible
353 the caller may have overwritten (or deallocated) the original string
354 data when a later ENGINE operation tries to use the stored values.
355 [Götz Babin-Ebell <babinebell@trustcenter.de>]
357 *) Improve diagnostics in file reading and command-line digests.
358 [Ben Laurie aided and abetted by Solar Designer <solar@openwall.com>]
360 *) Add AES modes CFB and OFB to the object database. Correct an
361 error in AES-CFB decryption.
364 *) Remove most calls to EVP_CIPHER_CTX_cleanup() in evp_enc.c, this
365 allows existing EVP_CIPHER_CTX structures to be reused after
366 calling EVP_*Final(). This behaviour is used by encryption
367 BIOs and some applications. This has the side effect that
368 applications must explicitly clean up cipher contexts with
369 EVP_CIPHER_CTX_cleanup() or they will leak memory.
372 *) Check the values of dna and dnb in bn_mul_recursive before calling
373 bn_mul_comba (a non zero value means the a or b arrays do not contain
374 n2 elements) and fallback to bn_mul_normal if either is not zero.
377 *) Fix escaping of non-ASCII characters when using the -subj option
378 of the "openssl req" command line tool. (Robert Joop <joop@fokus.gmd.de>)
381 *) Make object definitions compliant to LDAP (RFC2256): SN is the short
382 form for "surname", serialNumber has no short form.
383 Use "mail" as the short name for "rfc822Mailbox" according to RFC2798;
384 therefore remove "mail" short name for "internet 7".
385 The OID for unique identifiers in X509 certificates is
386 x500UniqueIdentifier, not uniqueIdentifier.
387 Some more OID additions. (Michael Bell <michael.bell@rz.hu-berlin.de>)
390 *) Add an "init" command to the ENGINE config module and auto initialize
391 ENGINEs. Without any "init" command the ENGINE will be initialized
392 after all ctrl commands have been executed on it. If init=1 the
393 ENGINE is initailized at that point (ctrls before that point are run
394 on the uninitialized ENGINE and after on the initialized one). If
395 init=0 then the ENGINE will not be iniatialized at all.
398 *) Fix the 'app_verify_callback' interface so that the user-defined
399 argument is actually passed to the callback: In the
400 SSL_CTX_set_cert_verify_callback() prototype, the callback
401 declaration has been changed from
404 int (*cb)(X509_STORE_CTX *,void *);
405 in ssl_verify_cert_chain (ssl/ssl_cert.c), the call
406 i=s->ctx->app_verify_callback(&ctx)
407 has been changed into
408 i=s->ctx->app_verify_callback(&ctx, s->ctx->app_verify_arg).
410 To update applications using SSL_CTX_set_cert_verify_callback(),
411 a dummy argument can be added to their callback functions.
412 [D. K. Smetters <smetters@parc.xerox.com>]
414 *) Added the '4758cca' ENGINE to support IBM 4758 cards.
415 [Maurice Gittens <maurice@gittens.nl>, touchups by Geoff Thorpe]
417 *) Add and OPENSSL_LOAD_CONF define which will cause
418 OpenSSL_add_all_algorithms() to load the openssl.cnf config file.
419 This allows older applications to transparently support certain
420 OpenSSL features: such as crypto acceleration and dynamic ENGINE loading.
421 Two new functions OPENSSL_add_all_algorithms_noconf() which will never
422 load the config file and OPENSSL_add_all_algorithms_conf() which will
423 always load it have also been added.
426 *) Add the OFB, CFB and CTR (all with 128 bit feedback) to AES.
427 Adjust NIDs and EVP layer.
428 [Stephen Sprunk <stephen@sprunk.org> and Richard Levitte]
430 *) Config modules support in openssl utility.
432 Most commands now load modules from the config file,
433 though in a few (such as version) this isn't done
434 because it couldn't be used for anything.
436 In the case of ca and req the config file used is
437 the same as the utility itself: that is the -config
438 command line option can be used to specify an
442 *) Move default behaviour from OPENSSL_config(). If appname is NULL
443 use "openssl_conf" if filename is NULL use default openssl config file.
446 *) Add an argument to OPENSSL_config() to allow the use of an alternative
447 config section name. Add a new flag to tolerate a missing config file
448 and move code to CONF_modules_load_file().
451 *) Support for crypto accelerator cards from Accelerated Encryption
452 Processing, www.aep.ie. (Use engine 'aep')
453 The support was copied from 0.9.6c [engine] and adapted/corrected
454 to work with the new engine framework.
455 [AEP Inc. and Richard Levitte]
457 *) Support for SureWare crypto accelerator cards from Baltimore
458 Technologies. (Use engine 'sureware')
459 The support was copied from 0.9.6c [engine] and adapted
460 to work with the new engine framework.
463 *) Have the CHIL engine fork-safe (as defined by nCipher) and actually
464 make the newer ENGINE framework commands for the CHIL engine work.
465 [Toomas Kiisk <vix@cyber.ee> and Richard Levitte]
467 *) Make it possible to produce shared libraries on ReliantUNIX.
468 [Robert Dahlem <Robert.Dahlem@ffm2.siemens.de> via Richard Levitte]
470 *) Add the configuration target debug-linux-ppro.
471 Make 'openssl rsa' use the general key loading routines
472 implemented in apps.c, and make those routines able to
473 handle the key format FORMAT_NETSCAPE and the variant
475 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte]
477 *) Fix a crashbug and a logic bug in hwcrhk_load_pubkey().
478 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte]
480 *) Add -keyform to rsautl, and document -engine.
481 [Richard Levitte, inspired by Toomas Kiisk <vix@cyber.ee>]
483 *) Change BIO_new_file (crypto/bio/bss_file.c) to use new
484 BIO_R_NO_SUCH_FILE error code rather than the generic
485 ERR_R_SYS_LIB error code if fopen() fails with ENOENT.
490 ERR_peek_last_error_line
491 ERR_peek_last_error_line_data.
495 ERR_peek_error_line_data,
496 but report on the latest error recorded rather than the first one
497 still in the error queue.
498 [Ben Laurie, Bodo Moeller]
500 *) default_algorithms option in ENGINE config module. This allows things
502 default_algorithms = ALL
503 default_algorithms = RSA, DSA, RAND, CIPHERS, DIGESTS
506 *) Prelminary ENGINE config module.
509 *) New experimental application configuration code.
512 *) Change the AES code to follow the same name structure as all other
513 symmetric ciphers, and behave the same way. Move everything to
514 the directory crypto/aes, thereby obsoleting crypto/rijndael.
515 [Stephen Sprunk <stephen@sprunk.org> and Richard Levitte]
517 *) SECURITY: remove unsafe setjmp/signal interaction from ui_openssl.c.
518 [Ben Laurie and Theo de Raadt]
520 *) Add option to output public keys in req command.
521 [Massimiliano Pala madwolf@openca.org]
523 *) Use wNAFs in EC_POINTs_mul() for improved efficiency
524 (up to about 10% better than before for P-192 and P-224).
527 *) New functions/macros
529 SSL_CTX_set_msg_callback(ctx, cb)
530 SSL_CTX_set_msg_callback_arg(ctx, arg)
531 SSL_set_msg_callback(ssl, cb)
532 SSL_set_msg_callback_arg(ssl, arg)
534 to request calling a callback function
536 void cb(int write_p, int version, int content_type,
537 const void *buf, size_t len, SSL *ssl, void *arg)
539 whenever a protocol message has been completely received
540 (write_p == 0) or sent (write_p == 1). Here 'version' is the
541 protocol version according to which the SSL library interprets
542 the current protocol message (SSL2_VERSION, SSL3_VERSION, or
543 TLS1_VERSION). 'content_type' is 0 in the case of SSL 2.0, or
544 the content type as defined in the SSL 3.0/TLS 1.0 protocol
545 specification (change_cipher_spec(20), alert(21), handshake(22)).
546 'buf' and 'len' point to the actual message, 'ssl' to the
547 SSL object, and 'arg' is the application-defined value set by
548 SSL[_CTX]_set_msg_callback_arg().
550 'openssl s_client' and 'openssl s_server' have new '-msg' options
551 to enable a callback that displays all protocol messages.
554 *) Change the shared library support so shared libraries are built as
555 soon as the corresponding static library is finished, and thereby get
556 openssl and the test programs linked against the shared library.
557 This still only happens when the keyword "shard" has been given to
558 the configuration scripts.
560 NOTE: shared library support is still an experimental thing, and
561 backward binary compatibility is still not guaranteed.
562 ["Maciej W. Rozycki" <macro@ds2.pg.gda.pl> and Richard Levitte]
564 *) Add support for Subject Information Access extension.
565 [Peter Sylvester <Peter.Sylvester@EdelWeb.fr>]
567 *) Make BUF_MEM_grow() behaviour more consistent: Initialise to zero
568 additional bytes when new memory had to be allocated, not just
569 when reusing an existing buffer.
572 *) New command line and configuration option 'utf8' for the req command.
573 This allows field values to be specified as UTF8 strings.
576 *) Add -multi and -mr options to "openssl speed" - giving multiple parallel
577 runs for the former and machine-readable output for the latter.
580 *) Add '-noemailDN' option to 'openssl ca'. This prevents inclusion
581 of the e-mail address in the DN (i.e., it will go into a certificate
582 extension only). The new configuration file option 'email_in_dn = no'
584 [Massimiliano Pala madwolf@openca.org]
586 *) Change all functions with names starting with des_ to be starting
587 with DES_ instead. Add wrappers that are compatible with libdes,
588 but are named _ossl_old_des_*. Finally, add macros that map the
589 des_* symbols to the corresponding _ossl_old_des_* if libdes
590 compatibility is desired. If OpenSSL 0.9.6c compatibility is
591 desired, the des_* symbols will be mapped to DES_*, with one
594 Since we provide two compatibility mappings, the user needs to
595 define the macro OPENSSL_DES_LIBDES_COMPATIBILITY if libdes
596 compatibility is desired. The default (i.e., when that macro
597 isn't defined) is OpenSSL 0.9.6c compatibility.
599 There are also macros that enable and disable the support of old
600 des functions altogether. Those are OPENSSL_ENABLE_OLD_DES_SUPPORT
601 and OPENSSL_DISABLE_OLD_DES_SUPPORT. If none or both of those
602 are defined, the default will apply: to support the old des routines.
604 In either case, one must include openssl/des.h to get the correct
605 definitions. Do not try to just include openssl/des_old.h, that
608 NOTE: This is a major break of an old API into a new one. Software
609 authors are encouraged to switch to the DES_ style functions. Some
610 time in the future, des_old.h and the libdes compatibility functions
611 will be disable (i.e. OPENSSL_DISABLE_OLD_DES_SUPPORT will be the
612 default), and then completely removed.
615 *) Test for certificates which contain unsupported critical extensions.
616 If such a certificate is found during a verify operation it is
617 rejected by default: this behaviour can be overridden by either
618 handling the new error X509_V_ERR_UNHANDLED_CRITICAL_EXTENSION or
619 by setting the verify flag X509_V_FLAG_IGNORE_CRITICAL. A new function
620 X509_supported_extension() has also been added which returns 1 if a
621 particular extension is supported.
624 *) Modify the behaviour of EVP cipher functions in similar way to digests
625 to retain compatibility with existing code.
628 *) Modify the behaviour of EVP_DigestInit() and EVP_DigestFinal() to retain
629 compatibility with existing code. In particular the 'ctx' parameter does
630 not have to be to be initialized before the call to EVP_DigestInit() and
631 it is tidied up after a call to EVP_DigestFinal(). New function
632 EVP_DigestFinal_ex() which does not tidy up the ctx. Similarly function
633 EVP_MD_CTX_copy() changed to not require the destination to be
634 initialized valid and new function EVP_MD_CTX_copy_ex() added which
635 requires the destination to be valid.
637 Modify all the OpenSSL digest calls to use EVP_DigestInit_ex(),
638 EVP_DigestFinal_ex() and EVP_MD_CTX_copy_ex().
641 *) Change ssl3_get_message (ssl/s3_both.c) and the functions using it
642 so that complete 'Handshake' protocol structures are kept in memory
643 instead of overwriting 'msg_type' and 'length' with 'body' data.
646 *) Add an implementation of SSL_add_dir_cert_subjects_to_stack for Win32.
647 [Massimo Santin via Richard Levitte]
649 *) Major restructuring to the underlying ENGINE code. This includes
650 reduction of linker bloat, separation of pure "ENGINE" manipulation
651 (initialisation, etc) from functionality dealing with implementations
652 of specific crypto iterfaces. This change also introduces integrated
653 support for symmetric ciphers and digest implementations - so ENGINEs
654 can now accelerate these by providing EVP_CIPHER and EVP_MD
655 implementations of their own. This is detailed in crypto/engine/README
656 as it couldn't be adequately described here. However, there are a few
657 API changes worth noting - some RSA, DSA, DH, and RAND functions that
658 were changed in the original introduction of ENGINE code have now
659 reverted back - the hooking from this code to ENGINE is now a good
660 deal more passive and at run-time, operations deal directly with
661 RSA_METHODs, DSA_METHODs (etc) as they did before, rather than
662 dereferencing through an ENGINE pointer any more. Also, the ENGINE
663 functions dealing with BN_MOD_EXP[_CRT] handlers have been removed -
664 they were not being used by the framework as there is no concept of a
665 BIGNUM_METHOD and they could not be generalised to the new
666 'ENGINE_TABLE' mechanism that underlies the new code. Similarly,
667 ENGINE_cpy() has been removed as it cannot be consistently defined in
671 *) Change ASN1_GENERALIZEDTIME_check() to allow fractional seconds.
674 *) Change mkdef.pl to sort symbols that get the same entry number,
675 and make sure the automatically generated functions ERR_load_*
676 become part of libeay.num as well.
679 *) New function SSL_renegotiate_pending(). This returns true once
680 renegotiation has been requested (either SSL_renegotiate() call
681 or HelloRequest/ClientHello receveived from the peer) and becomes
682 false once a handshake has been completed.
683 (For servers, SSL_renegotiate() followed by SSL_do_handshake()
684 sends a HelloRequest, but does not ensure that a handshake takes
685 place. SSL_renegotiate_pending() is useful for checking if the
686 client has followed the request.)
689 *) New SSL option SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION.
690 By default, clients may request session resumption even during
691 renegotiation (if session ID contexts permit); with this option,
692 session resumption is possible only in the first handshake.
694 SSL_OP_ALL is now 0x00000FFFL instead of 0x000FFFFFL. This makes
695 more bits available for options that should not be part of
696 SSL_OP_ALL (such as SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION).
699 *) Add some demos for certificate and certificate request creation.
702 *) Make maximum certificate chain size accepted from the peer application
703 settable (SSL*_get/set_max_cert_list()), as proposed by
704 "Douglas E. Engert" <deengert@anl.gov>.
707 *) Add support for shared libraries for Unixware-7
708 (Boyd Lynn Gerber <gerberb@zenez.com>).
711 *) Add a "destroy" handler to ENGINEs that allows structural cleanup to
712 be done prior to destruction. Use this to unload error strings from
713 ENGINEs that load their own error strings. NB: This adds two new API
714 functions to "get" and "set" this destroy handler in an ENGINE.
717 *) Alter all existing ENGINE implementations (except "openssl" and
718 "openbsd") to dynamically instantiate their own error strings. This
719 makes them more flexible to be built both as statically-linked ENGINEs
720 and self-contained shared-libraries loadable via the "dynamic" ENGINE.
721 Also, add stub code to each that makes building them as self-contained
722 shared-libraries easier (see README.ENGINE).
725 *) Add a "dynamic" ENGINE that provides a mechanism for binding ENGINE
726 implementations into applications that are completely implemented in
727 self-contained shared-libraries. The "dynamic" ENGINE exposes control
728 commands that can be used to configure what shared-library to load and
729 to control aspects of the way it is handled. Also, made an update to
730 the README.ENGINE file that brings its information up-to-date and
731 provides some information and instructions on the "dynamic" ENGINE
732 (ie. how to use it, how to build "dynamic"-loadable ENGINEs, etc).
735 *) Make it possible to unload ranges of ERR strings with a new
736 "ERR_unload_strings" function.
739 *) Add a copy() function to EVP_MD.
742 *) Make EVP_MD routines take a context pointer instead of just the
743 md_data void pointer.
746 *) Add flags to EVP_MD and EVP_MD_CTX. EVP_MD_FLAG_ONESHOT indicates
747 that the digest can only process a single chunk of data
748 (typically because it is provided by a piece of
749 hardware). EVP_MD_CTX_FLAG_ONESHOT indicates that the application
750 is only going to provide a single chunk of data, and hence the
751 framework needn't accumulate the data for oneshot drivers.
754 *) As with "ERR", make it possible to replace the underlying "ex_data"
755 functions. This change also alters the storage and management of global
756 ex_data state - it's now all inside ex_data.c and all "class" code (eg.
757 RSA, BIO, SSL_CTX, etc) no longer stores its own STACKS and per-class
758 index counters. The API functions that use this state have been changed
759 to take a "class_index" rather than pointers to the class's local STACK
760 and counter, and there is now an API function to dynamically create new
761 classes. This centralisation allows us to (a) plug a lot of the
762 thread-safety problems that existed, and (b) makes it possible to clean
763 up all allocated state using "CRYPTO_cleanup_all_ex_data()". W.r.t. (b)
764 such data would previously have always leaked in application code and
765 workarounds were in place to make the memory debugging turn a blind eye
766 to it. Application code that doesn't use this new function will still
767 leak as before, but their memory debugging output will announce it now
768 rather than letting it slide.
770 Besides the addition of CRYPTO_cleanup_all_ex_data(), another API change
771 induced by the "ex_data" overhaul is that X509_STORE_CTX_init() now
772 has a return value to indicate success or failure.
775 *) Make it possible to replace the underlying "ERR" functions such that the
776 global state (2 LHASH tables and 2 locks) is only used by the "default"
777 implementation. This change also adds two functions to "get" and "set"
778 the implementation prior to it being automatically set the first time
779 any other ERR function takes place. Ie. an application can call "get",
780 pass the return value to a module it has just loaded, and that module
781 can call its own "set" function using that value. This means the
782 module's "ERR" operations will use (and modify) the error state in the
783 application and not in its own statically linked copy of OpenSSL code.
786 *) Give DH, DSA, and RSA types their own "**_up_ref()" function to increment
787 reference counts. This performs normal REF_PRINT/REF_CHECK macros on
788 the operation, and provides a more encapsulated way for external code
789 (crypto/evp/ and ssl/) to do this. Also changed the evp and ssl code
790 to use these functions rather than manually incrementing the counts.
792 Also rename "DSO_up()" function to more descriptive "DSO_up_ref()".
795 *) Add EVP test program.
798 *) Add symmetric cipher support to ENGINE. Expect the API to change!
801 *) New CRL functions: X509_CRL_set_version(), X509_CRL_set_issuer_name()
802 X509_CRL_set_lastUpdate(), X509_CRL_set_nextUpdate(), X509_CRL_sort(),
803 X509_REVOKED_set_serialNumber(), and X509_REVOKED_set_revocationDate().
804 These allow a CRL to be built without having to access X509_CRL fields
805 directly. Modify 'ca' application to use new functions.
808 *) Move SSL_OP_TLS_ROLLBACK_BUG out of the SSL_OP_ALL list of recommended
809 bug workarounds. Rollback attack detection is a security feature.
810 The problem will only arise on OpenSSL servers when TLSv1 is not
811 available (sslv3_server_method() or SSL_OP_NO_TLSv1).
812 Software authors not wanting to support TLSv1 will have special reasons
813 for their choice and can explicitly enable this option.
814 [Bodo Moeller, Lutz Jaenicke]
816 *) Rationalise EVP so it can be extended: don't include a union of
817 cipher/digest structures, add init/cleanup functions for EVP_MD_CTX
818 (similar to those existing for EVP_CIPHER_CTX).
823 EVP_MD_CTX_init(&md); /* new function call */
824 EVP_DigestInit(&md, EVP_sha1());
825 EVP_DigestUpdate(&md, in, len);
826 EVP_DigestFinal(&md, out, NULL);
827 EVP_MD_CTX_cleanup(&md); /* new function call */
831 *) Make DES key schedule conform to the usual scheme, as well as
832 correcting its structure. This means that calls to DES functions
833 now have to pass a pointer to a des_key_schedule instead of a
834 plain des_key_schedule (which was actually always a pointer
839 des_set_key_checked(..., &ks);
840 des_ncbc_encrypt(..., &ks, ...);
842 (Note that a later change renames 'des_...' into 'DES_...'.)
845 *) Initial reduction of linker bloat: the use of some functions, such as
846 PEM causes large amounts of unused functions to be linked in due to
847 poor organisation. For example pem_all.c contains every PEM function
848 which has a knock on effect of linking in large amounts of (unused)
849 ASN1 code. Grouping together similar functions and splitting unrelated
850 functions prevents this.
853 *) Cleanup of EVP macros.
856 *) Change historical references to {NID,SN,LN}_des_ede and ede3 to add the
860 *) Add initial OCSP responder support to ocsp application. The
861 revocation information is handled using the text based index
862 use by the ca application. The responder can either handle
863 requests generated internally, supplied in files (for example
864 via a CGI script) or using an internal minimal server.
867 *) Add configuration choices to get zlib compression for TLS.
870 *) Changes to Kerberos SSL for RFC 2712 compliance:
871 1. Implemented real KerberosWrapper, instead of just using
872 KRB5 AP_REQ message. [Thanks to Simon Wilkinson <sxw@sxw.org.uk>]
873 2. Implemented optional authenticator field of KerberosWrapper.
875 Added openssl-style ASN.1 macros for Kerberos ticket, ap_req,
876 and authenticator structs; see crypto/krb5/.
878 Generalized Kerberos calls to support multiple Kerberos libraries.
879 [Vern Staats <staatsvr@asc.hpc.mil>,
880 Jeffrey Altman <jaltman@columbia.edu>
883 *) Cause 'openssl speed' to use fully hard-coded DSA keys as it
884 already does with RSA. testdsa.h now has 'priv_key/pub_key'
885 values for each of the key sizes rather than having just
886 parameters (and 'speed' generating keys each time).
889 *) Speed up EVP routines.
892 type 8 bytes 64 bytes 256 bytes 1024 bytes 8192 bytes
893 des-cbc 4408.85k 5560.51k 5778.46k 5862.20k 5825.16k
894 des-cbc 4389.55k 5571.17k 5792.23k 5846.91k 5832.11k
895 des-cbc 4394.32k 5575.92k 5807.44k 5848.37k 5841.30k
897 des-cbc 3482.66k 5069.49k 5496.39k 5614.16k 5639.28k
898 des-cbc 3480.74k 5068.76k 5510.34k 5609.87k 5635.52k
899 des-cbc 3483.72k 5067.62k 5504.60k 5708.01k 5724.80k
902 des-cbc 4660.16k 5650.19k 5807.19k 5827.13k 5783.32k
904 des-cbc 3624.96k 5258.21k 5530.91k 5624.30k 5628.26k
907 *) Added the OS2-EMX target.
908 ["Brian Havard" <brianh@kheldar.apana.org.au> and Richard Levitte]
910 *) Rewrite apps to use NCONF routines instead of the old CONF. New functions
911 to support NCONF routines in extension code. New function CONF_set_nconf()
912 to allow functions which take an NCONF to also handle the old LHASH
913 structure: this means that the old CONF compatible routines can be
914 retained (in particular wrt extensions) without having to duplicate the
915 code. New function X509V3_add_ext_nconf_sk to add extensions to a stack.
918 *) Enhance the general user interface with mechanisms for inner control
919 and with possibilities to have yes/no kind of prompts.
922 *) Change all calls to low level digest routines in the library and
923 applications to use EVP. Add missing calls to HMAC_cleanup() and
924 don't assume HMAC_CTX can be copied using memcpy().
925 [Verdon Walker <VWalker@novell.com>, Steve Henson]
927 *) Add the possibility to control engines through control names but with
928 arbitrary arguments instead of just a string.
929 Change the key loaders to take a UI_METHOD instead of a callback
930 function pointer. NOTE: this breaks binary compatibility with earlier
931 versions of OpenSSL [engine].
932 Adapt the nCipher code for these new conditions and add a card insertion
936 *) Enhance the general user interface with mechanisms to better support
937 dialog box interfaces, application-defined prompts, the possibility
938 to use defaults (for example default passwords from somewhere else)
939 and interrupts/cancellations.
942 *) Tidy up PKCS#12 attribute handling. Add support for the CSP name
943 attribute in PKCS#12 files, add new -CSP option to pkcs12 utility.
946 *) Fix a memory leak in 'sk_dup()' in the case reallocation fails. (Also
947 tidy up some unnecessarily weird code in 'sk_new()').
948 [Geoff, reported by Diego Tartara <dtartara@novamens.com>]
950 *) Change the key loading routines for ENGINEs to use the same kind
951 callback (pem_password_cb) as all other routines that need this
955 *) Increase ENTROPY_NEEDED to 32 bytes, as Rijndael can operate with
956 256 bit (=32 byte) keys. Of course seeding with more entropy bytes
957 than this minimum value is recommended.
960 *) New random seeder for OpenVMS, using the system process statistics
961 that are easily reachable.
964 *) Windows apparently can't transparently handle global
965 variables defined in DLLs. Initialisations such as:
967 const ASN1_ITEM *it = &ASN1_INTEGER_it;
969 wont compile. This is used by the any applications that need to
970 declare their own ASN1 modules. This was fixed by adding the option
971 EXPORT_VAR_AS_FN to all Win32 platforms, although this isn't strictly
972 needed for static libraries under Win32.
975 *) New functions X509_PURPOSE_set() and X509_TRUST_set() to handle
976 setting of purpose and trust fields. New X509_STORE trust and
977 purpose functions and tidy up setting in other SSL functions.
980 *) Add copies of X509_STORE_CTX fields and callbacks to X509_STORE
981 structure. These are inherited by X509_STORE_CTX when it is
982 initialised. This allows various defaults to be set in the
983 X509_STORE structure (such as flags for CRL checking and custom
984 purpose or trust settings) for functions which only use X509_STORE_CTX
985 internally such as S/MIME.
987 Modify X509_STORE_CTX_purpose_inherit() so it only sets purposes and
988 trust settings if they are not set in X509_STORE. This allows X509_STORE
989 purposes and trust (in S/MIME for example) to override any set by default.
991 Add command line options for CRL checking to smime, s_client and s_server
995 *) Initial CRL based revocation checking. If the CRL checking flag(s)
996 are set then the CRL is looked up in the X509_STORE structure and
997 its validity and signature checked, then if the certificate is found
998 in the CRL the verify fails with a revoked error.
1000 Various new CRL related callbacks added to X509_STORE_CTX structure.
1002 Command line options added to 'verify' application to support this.
1004 This needs some additional work, such as being able to handle multiple
1005 CRLs with different times, extension based lookup (rather than just
1006 by subject name) and ultimately more complete V2 CRL extension
1010 *) Add a general user interface API (crypto/ui/). This is designed
1011 to replace things like des_read_password and friends (backward
1012 compatibility functions using this new API are provided).
1013 The purpose is to remove prompting functions from the DES code
1014 section as well as provide for prompting through dialog boxes in
1015 a window system and the like.
1018 *) Add "ex_data" support to ENGINE so implementations can add state at a
1019 per-structure level rather than having to store it globally.
1022 *) Make it possible for ENGINE structures to be copied when retrieved by
1023 ENGINE_by_id() if the ENGINE specifies a new flag: ENGINE_FLAGS_BY_ID_COPY.
1024 This causes the "original" ENGINE structure to act like a template,
1025 analogous to the RSA vs. RSA_METHOD type of separation. Because of this
1026 operational state can be localised to each ENGINE structure, despite the
1027 fact they all share the same "methods". New ENGINE structures returned in
1028 this case have no functional references and the return value is the single
1029 structural reference. This matches the single structural reference returned
1030 by ENGINE_by_id() normally, when it is incremented on the pre-existing
1034 *) Fix ASN1 decoder when decoding type ANY and V_ASN1_OTHER: since this
1035 needs to match any other type at all we need to manually clear the
1039 *) Changes to the "openssl engine" utility to include;
1040 - verbosity levels ('-v', '-vv', and '-vvv') that provide information
1041 about an ENGINE's available control commands.
1042 - executing control commands from command line arguments using the
1043 '-pre' and '-post' switches. '-post' is only used if '-t' is
1044 specified and the ENGINE is successfully initialised. The syntax for
1045 the individual commands are colon-separated, for example;
1046 openssl engine chil -pre FORK_CHECK:0 -pre SO_PATH:/lib/test.so
1049 *) New dynamic control command support for ENGINEs. ENGINEs can now
1050 declare their own commands (numbers), names (strings), descriptions,
1051 and input types for run-time discovery by calling applications. A
1052 subset of these commands are implicitly classed as "executable"
1053 depending on their input type, and only these can be invoked through
1054 the new string-based API function ENGINE_ctrl_cmd_string(). (Eg. this
1055 can be based on user input, config files, etc). The distinction is
1056 that "executable" commands cannot return anything other than a boolean
1057 result and can only support numeric or string input, whereas some
1058 discoverable commands may only be for direct use through
1059 ENGINE_ctrl(), eg. supporting the exchange of binary data, function
1060 pointers, or other custom uses. The "executable" commands are to
1061 support parameterisations of ENGINE behaviour that can be
1062 unambiguously defined by ENGINEs and used consistently across any
1063 OpenSSL-based application. Commands have been added to all the
1064 existing hardware-supporting ENGINEs, noticeably "SO_PATH" to allow
1065 control over shared-library paths without source code alterations.
1068 *) Changed all ENGINE implementations to dynamically allocate their
1069 ENGINEs rather than declaring them statically. Apart from this being
1070 necessary with the removal of the ENGINE_FLAGS_MALLOCED distinction,
1071 this also allows the implementations to compile without using the
1072 internal engine_int.h header.
1075 *) Minor adjustment to "rand" code. RAND_get_rand_method() now returns a
1076 'const' value. Any code that should be able to modify a RAND_METHOD
1077 should already have non-const pointers to it (ie. they should only
1078 modify their own ones).
1081 *) Made a variety of little tweaks to the ENGINE code.
1082 - "atalla" and "ubsec" string definitions were moved from header files
1083 to C code. "nuron" string definitions were placed in variables
1084 rather than hard-coded - allowing parameterisation of these values
1085 later on via ctrl() commands.
1086 - Removed unused "#if 0"'d code.
1087 - Fixed engine list iteration code so it uses ENGINE_free() to release
1088 structural references.
1089 - Constified the RAND_METHOD element of ENGINE structures.
1090 - Constified various get/set functions as appropriate and added
1091 missing functions (including a catch-all ENGINE_cpy that duplicates
1092 all ENGINE values onto a new ENGINE except reference counts/state).
1093 - Removed NULL parameter checks in get/set functions. Setting a method
1094 or function to NULL is a way of cancelling out a previously set
1095 value. Passing a NULL ENGINE parameter is just plain stupid anyway
1096 and doesn't justify the extra error symbols and code.
1097 - Deprecate the ENGINE_FLAGS_MALLOCED define and move the area for
1098 flags from engine_int.h to engine.h.
1099 - Changed prototypes for ENGINE handler functions (init(), finish(),
1100 ctrl(), key-load functions, etc) to take an (ENGINE*) parameter.
1103 *) Implement binary inversion algorithm for BN_mod_inverse in addition
1104 to the algorithm using long division. The binary algorithm can be
1105 used only if the modulus is odd. On 32-bit systems, it is faster
1106 only for relatively small moduli (roughly 20-30% for 128-bit moduli,
1107 roughly 5-15% for 256-bit moduli), so we use it only for moduli
1108 up to 450 bits. In 64-bit environments, the binary algorithm
1109 appears to be advantageous for much longer moduli; here we use it
1110 for moduli up to 2048 bits.
1113 *) Rewrite CHOICE field setting in ASN1_item_ex_d2i(). The old code
1114 could not support the combine flag in choice fields.
1117 *) Add a 'copy_extensions' option to the 'ca' utility. This copies
1118 extensions from a certificate request to the certificate.
1121 *) Allow multiple 'certopt' and 'nameopt' options to be separated
1122 by commas. Add 'namopt' and 'certopt' options to the 'ca' config
1123 file: this allows the display of the certificate about to be
1124 signed to be customised, to allow certain fields to be included
1125 or excluded and extension details. The old system didn't display
1126 multicharacter strings properly, omitted fields not in the policy
1127 and couldn't display additional details such as extensions.
1130 *) Function EC_POINTs_mul for multiple scalar multiplication
1131 of an arbitrary number of elliptic curve points
1132 \sum scalars[i]*points[i],
1133 optionally including the generator defined for the EC_GROUP:
1134 scalar*generator + \sum scalars[i]*points[i].
1136 EC_POINT_mul is a simple wrapper function for the typical case
1137 that the point list has just one item (besides the optional
1141 *) First EC_METHODs for curves over GF(p):
1143 EC_GFp_simple_method() uses the basic BN_mod_mul and BN_mod_sqr
1144 operations and provides various method functions that can also
1145 operate with faster implementations of modular arithmetic.
1147 EC_GFp_mont_method() reuses most functions that are part of
1148 EC_GFp_simple_method, but uses Montgomery arithmetic.
1150 [Bodo Moeller; point addition and point doubling
1151 implementation directly derived from source code provided by
1152 Lenka Fibikova <fibikova@exp-math.uni-essen.de>]
1154 *) Framework for elliptic curves (crypto/ec/ec.h, crypto/ec/ec_lcl.h,
1155 crypto/ec/ec_lib.c):
1157 Curves are EC_GROUP objects (with an optional group generator)
1158 based on EC_METHODs that are built into the library.
1160 Points are EC_POINT objects based on EC_GROUP objects.
1162 Most of the framework would be able to handle curves over arbitrary
1163 finite fields, but as there are no obvious types for fields other
1164 than GF(p), some functions are limited to that for now.
1167 *) Add the -HTTP option to s_server. It is similar to -WWW, but requires
1168 that the file contains a complete HTTP response.
1171 *) Add the ec directory to mkdef.pl and mkfiles.pl. In mkdef.pl
1172 change the def and num file printf format specifier from "%-40sXXX"
1173 to "%-39s XXX". The latter will always guarantee a space after the
1174 field while the former will cause them to run together if the field
1175 is 40 of more characters long.
1178 *) Constify the cipher and digest 'method' functions and structures
1179 and modify related functions to take constant EVP_MD and EVP_CIPHER
1183 *) Hide BN_CTX structure details in bn_lcl.h instead of publishing them
1184 in <openssl/bn.h>. Also further increase BN_CTX_NUM to 32.
1187 *) Modify EVP_Digest*() routines so they now return values. Although the
1188 internal software routines can never fail additional hardware versions
1192 *) Clean up crypto/err/err.h and change some error codes to avoid conflicts:
1194 Previously ERR_R_FATAL was too small and coincided with ERR_LIB_PKCS7
1195 (= ERR_R_PKCS7_LIB); it is now 64 instead of 32.
1198 ERR_R_NESTED_ASN1_ERROR
1200 ERR_R_MISSING_ASN1_EOS
1201 were 4 .. 9, conflicting with
1202 ERR_LIB_RSA (= ERR_R_RSA_LIB)
1204 ERR_LIB_PEM (= ERR_R_PEM_LIB).
1205 They are now 58 .. 63 (i.e., just below ERR_R_FATAL).
1207 Add new error code 'ERR_R_INTERNAL_ERROR'.
1210 *) Don't overuse locks in crypto/err/err.c: For data retrieval, CRYPTO_r_lock
1214 *) New option '-subj arg' for 'openssl req' and 'openssl ca'. This
1215 sets the subject name for a new request or supersedes the
1216 subject name in a given request. Formats that can be parsed are
1217 'CN=Some Name, OU=myOU, C=IT'
1219 'CN=Some Name/OU=myOU/C=IT'.
1221 Add options '-batch' and '-verbose' to 'openssl req'.
1222 [Massimiliano Pala <madwolf@hackmasters.net>]
1224 *) Introduce the possibility to access global variables through
1225 functions on platform were that's the best way to handle exporting
1226 global variables in shared libraries. To enable this functionality,
1227 one must configure with "EXPORT_VAR_AS_FN" or defined the C macro
1228 "OPENSSL_EXPORT_VAR_AS_FUNCTION" in crypto/opensslconf.h (the latter
1229 is normally done by Configure or something similar).
1231 To implement a global variable, use the macro OPENSSL_IMPLEMENT_GLOBAL
1232 in the source file (foo.c) like this:
1234 OPENSSL_IMPLEMENT_GLOBAL(int,foo)=1;
1235 OPENSSL_IMPLEMENT_GLOBAL(double,bar);
1237 To declare a global variable, use the macros OPENSSL_DECLARE_GLOBAL
1238 and OPENSSL_GLOBAL_REF in the header file (foo.h) like this:
1240 OPENSSL_DECLARE_GLOBAL(int,foo);
1241 #define foo OPENSSL_GLOBAL_REF(foo)
1242 OPENSSL_DECLARE_GLOBAL(double,bar);
1243 #define bar OPENSSL_GLOBAL_REF(bar)
1245 The #defines are very important, and therefore so is including the
1246 header file everywhere where the defined globals are used.
1248 The macro OPENSSL_EXPORT_VAR_AS_FUNCTION also affects the definition
1249 of ASN.1 items, but that structure is a bit different.
1251 The largest change is in util/mkdef.pl which has been enhanced with
1252 better and easier to understand logic to choose which symbols should
1253 go into the Windows .def files as well as a number of fixes and code
1254 cleanup (among others, algorithm keywords are now sorted
1255 lexicographically to avoid constant rewrites).
1258 *) In BN_div() keep a copy of the sign of 'num' before writing the
1259 result to 'rm' because if rm==num the value will be overwritten
1260 and produce the wrong result if 'num' is negative: this caused
1261 problems with BN_mod() and BN_nnmod().
1264 *) Function OCSP_request_verify(). This checks the signature on an
1265 OCSP request and verifies the signer certificate. The signer
1266 certificate is just checked for a generic purpose and OCSP request
1270 *) Add OCSP_check_validity() function to check the validity of OCSP
1271 responses. OCSP responses are prepared in real time and may only
1272 be a few seconds old. Simply checking that the current time lies
1273 between thisUpdate and nextUpdate max reject otherwise valid responses
1274 caused by either OCSP responder or client clock inaccuracy. Instead
1275 we allow thisUpdate and nextUpdate to fall within a certain period of
1276 the current time. The age of the response can also optionally be
1277 checked. Two new options -validity_period and -status_age added to
1281 *) If signature or public key algorithm is unrecognized print out its
1282 OID rather that just UNKNOWN.
1285 *) Change OCSP_cert_to_id() to tolerate a NULL subject certificate and
1286 OCSP_cert_id_new() a NULL serialNumber. This allows a partial certificate
1287 ID to be generated from the issuer certificate alone which can then be
1288 passed to OCSP_id_issuer_cmp().
1291 *) New compilation option ASN1_ITEM_FUNCTIONS. This causes the new
1292 ASN1 modules to export functions returning ASN1_ITEM pointers
1293 instead of the ASN1_ITEM structures themselves. This adds several
1294 new macros which allow the underlying ASN1 function/structure to
1295 be accessed transparently. As a result code should not use ASN1_ITEM
1296 references directly (such as &X509_it) but instead use the relevant
1297 macros (such as ASN1_ITEM_rptr(X509)). This option is to allow
1298 use of the new ASN1 code on platforms where exporting structures
1299 is problematical (for example in shared libraries) but exporting
1300 functions returning pointers to structures is not.
1303 *) Add support for overriding the generation of SSL/TLS session IDs.
1304 These callbacks can be registered either in an SSL_CTX or per SSL.
1305 The purpose of this is to allow applications to control, if they wish,
1306 the arbitrary values chosen for use as session IDs, particularly as it
1307 can be useful for session caching in multiple-server environments. A
1308 command-line switch for testing this (and any client code that wishes
1309 to use such a feature) has been added to "s_server".
1310 [Geoff Thorpe, Lutz Jaenicke]
1312 *) Modify mkdef.pl to recognise and parse preprocessor conditionals
1313 of the form '#if defined(...) || defined(...) || ...' and
1314 '#if !defined(...) && !defined(...) && ...'. This also avoids
1315 the growing number of special cases it was previously handling.
1318 *) Make all configuration macros available for application by making
1319 sure they are available in opensslconf.h, by giving them names starting
1320 with "OPENSSL_" to avoid conflicts with other packages and by making
1321 sure e_os2.h will cover all platform-specific cases together with
1323 Additionally, it is now possible to define configuration/platform-
1324 specific names (called "system identities"). In the C code, these
1325 are prefixed with "OPENSSL_SYSNAME_". e_os2.h will create another
1326 macro with the name beginning with "OPENSSL_SYS_", which is determined
1327 from "OPENSSL_SYSNAME_*" or compiler-specific macros depending on
1331 *) New option -set_serial to 'req' and 'x509' this allows the serial
1332 number to use to be specified on the command line. Previously self
1333 signed certificates were hard coded with serial number 0 and the
1334 CA options of 'x509' had to use a serial number in a file which was
1338 *) New options to 'ca' utility to support V2 CRL entry extensions.
1339 Currently CRL reason, invalidity date and hold instruction are
1340 supported. Add new CRL extensions to V3 code and some new objects.
1343 *) New function EVP_CIPHER_CTX_set_padding() this is used to
1344 disable standard block padding (aka PKCS#5 padding) in the EVP
1345 API, which was previously mandatory. This means that the data is
1346 not padded in any way and so the total length much be a multiple
1347 of the block size, otherwise an error occurs.
1350 *) Initial (incomplete) OCSP SSL support.
1353 *) New function OCSP_parse_url(). This splits up a URL into its host,
1354 port and path components: primarily to parse OCSP URLs. New -url
1355 option to ocsp utility.
1358 *) New nonce behavior. The return value of OCSP_check_nonce() now
1359 reflects the various checks performed. Applications can decide
1360 whether to tolerate certain situations such as an absent nonce
1361 in a response when one was present in a request: the ocsp application
1362 just prints out a warning. New function OCSP_add1_basic_nonce()
1363 this is to allow responders to include a nonce in a response even if
1364 the request is nonce-less.
1367 *) Disable stdin buffering in load_cert (apps/apps.c) so that no certs are
1368 skipped when using openssl x509 multiple times on a single input file,
1369 e.g. "(openssl x509 -out cert1; openssl x509 -out cert2) <certs".
1372 *) Make ASN1_UTCTIME_set_string() and ASN1_GENERALIZEDTIME_set_string()
1373 set string type: to handle setting ASN1_TIME structures. Fix ca
1374 utility to correctly initialize revocation date of CRLs.
1377 *) New option SSL_OP_CIPHER_SERVER_PREFERENCE allows the server to override
1378 the clients preferred ciphersuites and rather use its own preferences.
1379 Should help to work around M$ SGC (Server Gated Cryptography) bug in
1380 Internet Explorer by ensuring unchanged hash method during stepup.
1381 (Also replaces the broken/deactivated SSL_OP_NON_EXPORT_FIRST option.)
1384 *) Make mkdef.pl recognise all DECLARE_ASN1 macros, change rijndael
1385 to aes and add a new 'exist' option to print out symbols that don't
1389 *) Additional options to ocsp utility to allow flags to be set and
1390 additional certificates supplied.
1393 *) Add the option -VAfile to 'openssl ocsp', so the user can give the
1394 OCSP client a number of certificate to only verify the response
1398 *) Update Rijndael code to version 3.0 and change EVP AES ciphers to
1399 handle the new API. Currently only ECB, CBC modes supported. Add new
1402 Add TLS AES ciphersuites as described in RFC3268, "Advanced
1403 Encryption Standard (AES) Ciphersuites for Transport Layer
1404 Security (TLS)". (In beta versions of OpenSSL 0.9.7, these were
1405 not enabled by default and were not part of the "ALL" ciphersuite
1406 alias because they were not yet official; they could be
1407 explicitly requested by specifying the "AESdraft" ciphersuite
1408 group alias. In the final release of OpenSSL 0.9.7, the group
1409 alias is called "AES" and is part of "ALL".)
1410 [Ben Laurie, Steve Henson, Bodo Moeller]
1412 *) New function OCSP_copy_nonce() to copy nonce value (if present) from
1413 request to response.
1416 *) Functions for OCSP responders. OCSP_request_onereq_count(),
1417 OCSP_request_onereq_get0(), OCSP_onereq_get0_id() and OCSP_id_get0_info()
1418 extract information from a certificate request. OCSP_response_create()
1419 creates a response and optionally adds a basic response structure.
1420 OCSP_basic_add1_status() adds a complete single response to a basic
1421 response and returns the OCSP_SINGLERESP structure just added (to allow
1422 extensions to be included for example). OCSP_basic_add1_cert() adds a
1423 certificate to a basic response and OCSP_basic_sign() signs a basic
1424 response with various flags. New helper functions ASN1_TIME_check()
1425 (checks validity of ASN1_TIME structure) and ASN1_TIME_to_generalizedtime()
1426 (converts ASN1_TIME to GeneralizedTime).
1429 *) Various new functions. EVP_Digest() combines EVP_Digest{Init,Update,Final}()
1430 in a single operation. X509_get0_pubkey_bitstr() extracts the public_key
1431 structure from a certificate. X509_pubkey_digest() digests the public_key
1432 contents: this is used in various key identifiers.
1435 *) Make sk_sort() tolerate a NULL argument.
1436 [Steve Henson reported by Massimiliano Pala <madwolf@comune.modena.it>]
1438 *) New OCSP verify flag OCSP_TRUSTOTHER. When set the "other" certificates
1439 passed by the function are trusted implicitly. If any of them signed the
1440 response then it is assumed to be valid and is not verified.
1443 *) In PKCS7_set_type() initialise content_type in PKCS7_ENC_CONTENT
1444 to data. This was previously part of the PKCS7 ASN1 code. This
1445 was causing problems with OpenSSL created PKCS#12 and PKCS#7 structures.
1446 [Steve Henson, reported by Kenneth R. Robinette
1447 <support@securenetterm.com>]
1449 *) Add CRYPTO_push_info() and CRYPTO_pop_info() calls to new ASN1
1450 routines: without these tracing memory leaks is very painful.
1451 Fix leaks in PKCS12 and PKCS7 routines.
1454 *) Make X509_time_adj() cope with the new behaviour of ASN1_TIME_new().
1455 Previously it initialised the 'type' argument to V_ASN1_UTCTIME which
1456 effectively meant GeneralizedTime would never be used. Now it
1457 is initialised to -1 but X509_time_adj() now has to check the value
1458 and use ASN1_TIME_set() if the value is not V_ASN1_UTCTIME or
1459 V_ASN1_GENERALIZEDTIME, without this it always uses GeneralizedTime.
1460 [Steve Henson, reported by Kenneth R. Robinette
1461 <support@securenetterm.com>]
1463 *) Fixes to BN_to_ASN1_INTEGER when bn is zero. This would previously
1464 result in a zero length in the ASN1_INTEGER structure which was
1465 not consistent with the structure when d2i_ASN1_INTEGER() was used
1466 and would cause ASN1_INTEGER_cmp() to fail. Enhance s2i_ASN1_INTEGER()
1467 to cope with hex and negative integers. Fix bug in i2a_ASN1_INTEGER()
1468 where it did not print out a minus for negative ASN1_INTEGER.
1471 *) Add summary printout to ocsp utility. The various functions which
1472 convert status values to strings have been renamed to:
1473 OCSP_response_status_str(), OCSP_cert_status_str() and
1474 OCSP_crl_reason_str() and are no longer static. New options
1475 to verify nonce values and to disable verification. OCSP response
1476 printout format cleaned up.
1479 *) Add additional OCSP certificate checks. These are those specified
1480 in RFC2560. This consists of two separate checks: the CA of the
1481 certificate being checked must either be the OCSP signer certificate
1482 or the issuer of the OCSP signer certificate. In the latter case the
1483 OCSP signer certificate must contain the OCSP signing extended key
1484 usage. This check is performed by attempting to match the OCSP
1485 signer or the OCSP signer CA to the issuerNameHash and issuerKeyHash
1486 in the OCSP_CERTID structures of the response.
1489 *) Initial OCSP certificate verification added to OCSP_basic_verify()
1490 and related routines. This uses the standard OpenSSL certificate
1491 verify routines to perform initial checks (just CA validity) and
1492 to obtain the certificate chain. Then additional checks will be
1493 performed on the chain. Currently the root CA is checked to see
1494 if it is explicitly trusted for OCSP signing. This is used to set
1495 a root CA as a global signing root: that is any certificate that
1496 chains to that CA is an acceptable OCSP signing certificate.
1499 *) New '-extfile ...' option to 'openssl ca' for reading X.509v3
1500 extensions from a separate configuration file.
1501 As when reading extensions from the main configuration file,
1502 the '-extensions ...' option may be used for specifying the
1504 [Massimiliano Pala <madwolf@comune.modena.it>]
1506 *) New OCSP utility. Allows OCSP requests to be generated or
1507 read. The request can be sent to a responder and the output
1508 parsed, outputed or printed in text form. Not complete yet:
1509 still needs to check the OCSP response validity.
1512 *) New subcommands for 'openssl ca':
1513 'openssl ca -status <serial>' prints the status of the cert with
1514 the given serial number (according to the index file).
1515 'openssl ca -updatedb' updates the expiry status of certificates
1517 [Massimiliano Pala <madwolf@comune.modena.it>]
1519 *) New '-newreq-nodes' command option to CA.pl. This is like
1520 '-newreq', but calls 'openssl req' with the '-nodes' option
1521 so that the resulting key is not encrypted.
1522 [Damien Miller <djm@mindrot.org>]
1524 *) New configuration for the GNU Hurd.
1525 [Jonathan Bartlett <johnnyb@wolfram.com> via Richard Levitte]
1527 *) Initial code to implement OCSP basic response verify. This
1528 is currently incomplete. Currently just finds the signer's
1529 certificate and verifies the signature on the response.
1532 *) New SSLeay_version code SSLEAY_DIR to determine the compiled-in
1533 value of OPENSSLDIR. This is available via the new '-d' option
1534 to 'openssl version', and is also included in 'openssl version -a'.
1537 *) Allowing defining memory allocation callbacks that will be given
1538 file name and line number information in additional arguments
1539 (a const char* and an int). The basic functionality remains, as
1540 well as the original possibility to just replace malloc(),
1541 realloc() and free() by functions that do not know about these
1542 additional arguments. To register and find out the current
1543 settings for extended allocation functions, the following
1544 functions are provided:
1546 CRYPTO_set_mem_ex_functions
1547 CRYPTO_set_locked_mem_ex_functions
1548 CRYPTO_get_mem_ex_functions
1549 CRYPTO_get_locked_mem_ex_functions
1551 These work the same way as CRYPTO_set_mem_functions and friends.
1552 CRYPTO_get_[locked_]mem_functions now writes 0 where such an
1553 extended allocation function is enabled.
1554 Similarly, CRYPTO_get_[locked_]mem_ex_functions writes 0 where
1555 a conventional allocation function is enabled.
1556 [Richard Levitte, Bodo Moeller]
1558 *) Finish off removing the remaining LHASH function pointer casts.
1559 There should no longer be any prototype-casting required when using
1560 the LHASH abstraction, and any casts that remain are "bugs". See
1561 the callback types and macros at the head of lhash.h for details
1562 (and "OBJ_cleanup" in crypto/objects/obj_dat.c as an example).
1565 *) Add automatic query of EGD sockets in RAND_poll() for the unix variant.
1566 If /dev/[u]random devices are not available or do not return enough
1567 entropy, EGD style sockets (served by EGD or PRNGD) will automatically
1569 The locations /var/run/egd-pool, /dev/egd-pool, /etc/egd-pool, and
1570 /etc/entropy will be queried once each in this sequence, quering stops
1571 when enough entropy was collected without querying more sockets.
1574 *) Change the Unix RAND_poll() variant to be able to poll several
1575 random devices, as specified by DEVRANDOM, until a sufficient amount
1576 of data has been collected. We spend at most 10 ms on each file
1577 (select timeout) and read in non-blocking mode. DEVRANDOM now
1578 defaults to the list "/dev/urandom", "/dev/random", "/dev/srandom"
1579 (previously it was just the string "/dev/urandom"), so on typical
1580 platforms the 10 ms delay will never occur.
1581 Also separate out the Unix variant to its own file, rand_unix.c.
1582 For VMS, there's a currently-empty rand_vms.c.
1585 *) Move OCSP client related routines to ocsp_cl.c. These
1586 provide utility functions which an application needing
1587 to issue a request to an OCSP responder and analyse the
1588 response will typically need: as opposed to those which an
1589 OCSP responder itself would need which will be added later.
1591 OCSP_request_sign() signs an OCSP request with an API similar
1592 to PKCS7_sign(). OCSP_response_status() returns status of OCSP
1593 response. OCSP_response_get1_basic() extracts basic response
1594 from response. OCSP_resp_find_status(): finds and extracts status
1595 information from an OCSP_CERTID structure (which will be created
1596 when the request structure is built). These are built from lower
1597 level functions which work on OCSP_SINGLERESP structures but
1598 wont normally be used unless the application wishes to examine
1599 extensions in the OCSP response for example.
1601 Replace nonce routines with a pair of functions.
1602 OCSP_request_add1_nonce() adds a nonce value and optionally
1603 generates a random value. OCSP_check_nonce() checks the
1604 validity of the nonce in an OCSP response.
1607 *) Change function OCSP_request_add() to OCSP_request_add0_id().
1608 This doesn't copy the supplied OCSP_CERTID and avoids the
1609 need to free up the newly created id. Change return type
1610 to OCSP_ONEREQ to return the internal OCSP_ONEREQ structure.
1611 This can then be used to add extensions to the request.
1612 Deleted OCSP_request_new(), since most of its functionality
1613 is now in OCSP_REQUEST_new() (and the case insensitive name
1614 clash) apart from the ability to set the request name which
1615 will be added elsewhere.
1618 *) Update OCSP API. Remove obsolete extensions argument from
1619 various functions. Extensions are now handled using the new
1620 OCSP extension code. New simple OCSP HTTP function which
1621 can be used to send requests and parse the response.
1624 *) Fix the PKCS#7 (S/MIME) code to work with new ASN1. Two new
1625 ASN1_ITEM structures help with sign and verify. PKCS7_ATTR_SIGN
1626 uses the special reorder version of SET OF to sort the attributes
1627 and reorder them to match the encoded order. This resolves a long
1628 standing problem: a verify on a PKCS7 structure just after signing
1629 it used to fail because the attribute order did not match the
1630 encoded order. PKCS7_ATTR_VERIFY does not reorder the attributes:
1631 it uses the received order. This is necessary to tolerate some broken
1632 software that does not order SET OF. This is handled by encoding
1633 as a SEQUENCE OF but using implicit tagging (with UNIVERSAL class)
1634 to produce the required SET OF.
1637 *) Have mk1mf.pl generate the macros OPENSSL_BUILD_SHLIBCRYPTO and
1638 OPENSSL_BUILD_SHLIBSSL and use them appropriately in the header
1639 files to get correct declarations of the ASN.1 item variables.
1642 *) Rewrite of PKCS#12 code to use new ASN1 functionality. Replace many
1643 PKCS#12 macros with real functions. Fix two unrelated ASN1 bugs:
1644 asn1_check_tlen() would sometimes attempt to use 'ctx' when it was
1645 NULL and ASN1_TYPE was not dereferenced properly in asn1_ex_c2i().
1646 New ASN1 macro: DECLARE_ASN1_ITEM() which just declares the relevant
1647 ASN1_ITEM and no wrapper functions.
1650 *) New functions or ASN1_item_d2i_fp() and ASN1_item_d2i_bio(). These
1651 replace the old function pointer based I/O routines. Change most of
1652 the *_d2i_bio() and *_d2i_fp() functions to use these.
1655 *) Enhance mkdef.pl to be more accepting about spacing in C preprocessor
1656 lines, recognice more "algorithms" that can be deselected, and make
1657 it complain about algorithm deselection that isn't recognised.
1660 *) New ASN1 functions to handle dup, sign, verify, digest, pack and
1661 unpack operations in terms of ASN1_ITEM. Modify existing wrappers
1662 to use new functions. Add NO_ASN1_OLD which can be set to remove
1663 some old style ASN1 functions: this can be used to determine if old
1664 code will still work when these eventually go away.
1667 *) New extension functions for OCSP structures, these follow the
1668 same conventions as certificates and CRLs.
1671 *) New function X509V3_add1_i2d(). This automatically encodes and
1672 adds an extension. Its behaviour can be customised with various
1673 flags to append, replace or delete. Various wrappers added for
1674 certifcates and CRLs.
1677 *) Fix to avoid calling the underlying ASN1 print routine when
1678 an extension cannot be parsed. Correct a typo in the
1679 OCSP_SERVICELOC extension. Tidy up print OCSP format.
1682 *) Make mkdef.pl parse some of the ASN1 macros and add apropriate
1683 entries for variables.
1686 *) Add functionality to apps/openssl.c for detecting locking
1687 problems: As the program is single-threaded, all we have
1688 to do is register a locking callback using an array for
1689 storing which locks are currently held by the program.
1692 *) Use a lock around the call to CRYPTO_get_ex_new_index() in
1693 SSL_get_ex_data_X509_STORE_idx(), which is used in
1694 ssl_verify_cert_chain() and thus can be called at any time
1695 during TLS/SSL handshakes so that thread-safety is essential.
1696 Unfortunately, the ex_data design is not at all suited
1697 for multi-threaded use, so it probably should be abolished.
1700 *) Added Broadcom "ubsec" ENGINE to OpenSSL.
1701 [Broadcom, tweaked and integrated by Geoff Thorpe]
1703 *) Move common extension printing code to new function
1704 X509V3_print_extensions(). Reorganise OCSP print routines and
1705 implement some needed OCSP ASN1 functions. Add OCSP extensions.
1708 *) New function X509_signature_print() to remove duplication in some
1712 *) Add a special meaning when SET OF and SEQUENCE OF flags are both
1713 set (this was treated exactly the same as SET OF previously). This
1714 is used to reorder the STACK representing the structure to match the
1715 encoding. This will be used to get round a problem where a PKCS7
1716 structure which was signed could not be verified because the STACK
1717 order did not reflect the encoded order.
1720 *) Reimplement the OCSP ASN1 module using the new code.
1723 *) Update the X509V3 code to permit the use of an ASN1_ITEM structure
1724 for its ASN1 operations. The old style function pointers still exist
1725 for now but they will eventually go away.
1728 *) Merge in replacement ASN1 code from the ASN1 branch. This almost
1729 completely replaces the old ASN1 functionality with a table driven
1730 encoder and decoder which interprets an ASN1_ITEM structure describing
1731 the ASN1 module. Compatibility with the existing ASN1 API (i2d,d2i) is
1732 largely maintained. Almost all of the old asn1_mac.h macro based ASN1
1733 has also been converted to the new form.
1736 *) Change BN_mod_exp_recp so that negative moduli are tolerated
1737 (the sign is ignored). Similarly, ignore the sign in BN_MONT_CTX_set
1738 so that BN_mod_exp_mont and BN_mod_exp_mont_word work
1739 for negative moduli.
1742 *) Fix BN_uadd and BN_usub: Always return non-negative results instead
1743 of not touching the result's sign bit.
1746 *) BN_div bugfix: If the result is 0, the sign (res->neg) must not be
1750 *) Changed the LHASH code to use prototypes for callbacks, and created
1751 macros to declare and implement thin (optionally static) functions
1752 that provide type-safety and avoid function pointer casting for the
1753 type-specific callbacks.
1756 *) Added Kerberos Cipher Suites to be used with TLS, as written in
1758 [Veers Staats <staatsvr@asc.hpc.mil>,
1759 Jeffrey Altman <jaltman@columbia.edu>, via Richard Levitte]
1761 *) Reformat the FAQ so the different questions and answers can be divided
1762 in sections depending on the subject.
1765 *) Have the zlib compression code load ZLIB.DLL dynamically under
1769 *) New function BN_mod_sqrt for computing square roots modulo a prime
1770 (using the probabilistic Tonelli-Shanks algorithm unless
1771 p == 3 (mod 4) or p == 5 (mod 8), which are cases that can
1772 be handled deterministically).
1773 [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller]
1775 *) Make BN_mod_inverse faster by explicitly handling small quotients
1776 in the Euclid loop. (Speed gain about 20% for small moduli [256 or
1777 512 bits], about 30% for larger ones [1024 or 2048 bits].)
1780 *) New function BN_kronecker.
1783 *) Fix BN_gcd so that it works on negative inputs; the result is
1784 positive unless both parameters are zero.
1785 Previously something reasonably close to an infinite loop was
1786 possible because numbers could be growing instead of shrinking
1787 in the implementation of Euclid's algorithm.
1790 *) Fix BN_is_word() and BN_is_one() macros to take into account the
1791 sign of the number in question.
1793 Fix BN_is_word(a,w) to work correctly for w == 0.
1795 The old BN_is_word(a,w) macro is now called BN_abs_is_word(a,w)
1796 because its test if the absolute value of 'a' equals 'w'.
1797 Note that BN_abs_is_word does *not* handle w == 0 reliably;
1798 it exists mostly for use in the implementations of BN_is_zero(),
1799 BN_is_one(), and BN_is_word().
1802 *) New function BN_swap.
1805 *) Use BN_nnmod instead of BN_mod in crypto/bn/bn_exp.c so that
1806 the exponentiation functions are more likely to produce reasonable
1807 results on negative inputs.
1810 *) Change BN_mod_mul so that the result is always non-negative.
1811 Previously, it could be negative if one of the factors was negative;
1812 I don't think anyone really wanted that behaviour.
1815 *) Move BN_mod_... functions into new file crypto/bn/bn_mod.c
1816 (except for exponentiation, which stays in crypto/bn/bn_exp.c,
1817 and BN_mod_mul_reciprocal, which stays in crypto/bn/bn_recp.c)
1818 and add new functions:
1827 BN_mod_lshift1_quick
1831 These functions always generate non-negative results.
1833 BN_nnmod otherwise is like BN_mod (if BN_mod computes a remainder r
1834 such that |m| < r < 0, BN_nnmod will output rem + |m| instead).
1836 BN_mod_XXX_quick(r, a, [b,] m) generates the same result as
1837 BN_mod_XXX(r, a, [b,] m, ctx), but requires that a [and b]
1838 be reduced modulo m.
1839 [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller]
1841 *) Remove a few calls to bn_wexpand() in BN_sqr() (the one in there
1842 was actually never needed) and in BN_mul(). The removal in BN_mul()
1843 required a small change in bn_mul_part_recursive() and the addition
1844 of the functions bn_cmp_part_words(), bn_sub_part_words() and
1845 bn_add_part_words(), which do the same thing as bn_cmp_words(),
1846 bn_sub_words() and bn_add_words() except they take arrays with
1850 *) In 'openssl passwd', verify passwords read from the terminal
1851 unless the '-salt' option is used (which usually means that
1852 verification would just waste user's time since the resulting
1853 hash is going to be compared with some given password hash)
1854 or the new '-noverify' option is used.
1856 This is an incompatible change, but it does not affect
1857 non-interactive use of 'openssl passwd' (passwords on the command
1858 line, '-stdin' option, '-in ...' option) and thus should not
1862 *) Remove all references to RSAref, since there's no more need for it.
1865 *) Make DSO load along a path given through an environment variable
1866 (SHLIB_PATH) with shl_load().
1869 *) Constify the ENGINE code as a result of BIGNUM constification.
1870 Also constify the RSA code and most things related to it. In a
1871 few places, most notable in the depth of the ASN.1 code, ugly
1872 casts back to non-const were required (to be solved at a later
1876 *) Make it so the openssl application has all engines loaded by default.
1879 *) Constify the BIGNUM routines a little more.
1882 *) Add the following functions:
1884 ENGINE_load_cswift()
1886 ENGINE_load_atalla()
1888 ENGINE_load_builtin_engines()
1890 That way, an application can itself choose if external engines that
1891 are built-in in OpenSSL shall ever be used or not. The benefit is
1892 that applications won't have to be linked with libdl or other dso
1893 libraries unless it's really needed.
1895 Changed 'openssl engine' to load all engines on demand.
1896 Changed the engine header files to avoid the duplication of some
1897 declarations (they differed!).
1900 *) 'openssl engine' can now list capabilities.
1903 *) Better error reporting in 'openssl engine'.
1906 *) Never call load_dh_param(NULL) in s_server.
1909 *) Add engine application. It can currently list engines by name and
1910 identity, and test if they are actually available.
1913 *) Improve RPM specification file by forcing symbolic linking and making
1914 sure the installed documentation is also owned by root.root.
1915 [Damien Miller <djm@mindrot.org>]
1917 *) Give the OpenSSL applications more possibilities to make use of
1918 keys (public as well as private) handled by engines.
1921 *) Add OCSP code that comes from CertCo.
1924 *) Add VMS support for the Rijndael code.
1927 *) Added untested support for Nuron crypto accelerator.
1930 *) Add support for external cryptographic devices. This code was
1931 previously distributed separately as the "engine" branch.
1932 [Geoff Thorpe, Richard Levitte]
1934 *) Rework the filename-translation in the DSO code. It is now possible to
1935 have far greater control over how a "name" is turned into a filename
1936 depending on the operating environment and any oddities about the
1937 different shared library filenames on each system.
1940 *) Support threads on FreeBSD-elf in Configure.
1943 *) Fix for SHA1 assembly problem with MASM: it produces
1944 warnings about corrupt line number information when assembling
1945 with debugging information. This is caused by the overlapping
1947 [Bernd Matthes <mainbug@celocom.de>, Steve Henson]
1950 NCONF_get_number() has no error checking at all. As a replacement,
1951 NCONF_get_number_e() is defined (_e for "error checking") and is
1952 promoted strongly. The old NCONF_get_number is kept around for
1953 binary backward compatibility.
1954 Make it possible for methods to load from something other than a BIO,
1955 by providing a function pointer that is given a name instead of a BIO.
1956 For example, this could be used to load configuration data from an
1960 *) Fix for non blocking accept BIOs. Added new I/O special reason
1961 BIO_RR_ACCEPT to cover this case. Previously use of accept BIOs
1962 with non blocking I/O was not possible because no retry code was
1963 implemented. Also added new SSL code SSL_WANT_ACCEPT to cover
1967 *) Added the beginnings of Rijndael support.
1970 *) Fix for bug in DirectoryString mask setting. Add support for
1971 X509_NAME_print_ex() in 'req' and X509_print_ex() function
1972 to allow certificate printing to more controllable, additional
1973 'certopt' option to 'x509' to allow new printing options to be
1977 *) Clean old EAY MD5 hack from e_os.h.
1980 Changes between 0.9.6g and 0.9.6h [xx XXX xxxx]
1982 *) Don't impose a 16-byte length minimum on session IDs in ssl/s3_clnt.c
1983 (the SSL 3.0 and TLS 1.0 specifications allow any length up to 32 bytes).
1986 *) Fix initialization code race conditions in
1987 SSLv23_method(), SSLv23_client_method(), SSLv23_server_method(),
1988 SSLv2_method(), SSLv2_client_method(), SSLv2_server_method(),
1989 SSLv3_method(), SSLv3_client_method(), SSLv3_server_method(),
1990 TLSv1_method(), TLSv1_client_method(), TLSv1_server_method(),
1991 ssl2_get_cipher_by_char(),
1992 ssl3_get_cipher_by_char().
1993 [Patrick McCormick <patrick@tellme.com>, Bodo Moeller]
1995 *) Reorder cleanup sequence in SSL_CTX_free(): only remove the ex_data after
1996 the cached sessions are flushed, as the remove_cb() might use ex_data
1997 contents. Bug found by Sam Varshavchik <mrsam@courier-mta.com>
1998 (see [openssl.org #212]).
1999 [Geoff Thorpe, Lutz Jaenicke]
2001 *) Fix typo in OBJ_txt2obj which incorrectly passed the content
2002 length, instead of the encoding length to d2i_ASN1_OBJECT.
2005 Changes between 0.9.6f and 0.9.6g [9 Aug 2002]
2007 *) [In 0.9.6g-engine release:]
2008 Fix crypto/engine/vendor_defns/cswift.h for WIN32 (use '_stdcall').
2009 [Lynn Gazis <lgazis@rainbow.com>]
2011 Changes between 0.9.6e and 0.9.6f [8 Aug 2002]
2013 *) Fix ASN1 checks. Check for overflow by comparing with LONG_MAX
2014 and get fix the header length calculation.
2015 [Florian Weimer <Weimer@CERT.Uni-Stuttgart.DE>,
2016 Alon Kantor <alonk@checkpoint.com> (and others),
2019 *) Use proper error handling instead of 'assertions' in buffer
2020 overflow checks added in 0.9.6e. This prevents DoS (the
2021 assertions could call abort()).
2022 [Arne Ansper <arne@ats.cyber.ee>, Bodo Moeller]
2024 Changes between 0.9.6d and 0.9.6e [30 Jul 2002]
2026 *) Add various sanity checks to asn1_get_length() to reject
2027 the ASN1 length bytes if they exceed sizeof(long), will appear
2028 negative or the content length exceeds the length of the
2030 [Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>]
2032 *) Fix cipher selection routines: ciphers without encryption had no flags
2033 for the cipher strength set and where therefore not handled correctly
2034 by the selection routines (PR #130).
2037 *) Fix EVP_dsa_sha macro.
2041 SSL_OP_DONT_INSERT_EMPTY_FRAGMENTS
2042 for disabling the SSL 3.0/TLS 1.0 CBC vulnerability countermeasure
2043 that was added in OpenSSL 0.9.6d.
2045 As the countermeasure turned out to be incompatible with some
2046 broken SSL implementations, the new option is part of SSL_OP_ALL.
2047 SSL_OP_ALL is usually employed when compatibility with weird SSL
2048 implementations is desired (e.g. '-bugs' option to 's_client' and
2049 's_server'), so the new option is automatically set in many
2053 *) Changes in security patch:
2055 Changes marked "(CHATS)" were sponsored by the Defense Advanced
2056 Research Projects Agency (DARPA) and Air Force Research Laboratory,
2057 Air Force Materiel Command, USAF, under agreement number
2060 *) Add various sanity checks to asn1_get_length() to reject
2061 the ASN1 length bytes if they exceed sizeof(long), will appear
2062 negative or the content length exceeds the length of the
2063 supplied buffer. (CAN-2002-0659)
2064 [Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>]
2066 *) Assertions for various potential buffer overflows, not known to
2068 [Ben Laurie (CHATS)]
2070 *) Various temporary buffers to hold ASCII versions of integers were
2071 too small for 64 bit platforms. (CAN-2002-0655)
2072 [Matthew Byng-Maddick <mbm@aldigital.co.uk> and Ben Laurie (CHATS)>
2074 *) Remote buffer overflow in SSL3 protocol - an attacker could
2075 supply an oversized session ID to a client. (CAN-2002-0656)
2076 [Ben Laurie (CHATS)]
2078 *) Remote buffer overflow in SSL2 protocol - an attacker could
2079 supply an oversized client master key. (CAN-2002-0656)
2080 [Ben Laurie (CHATS)]
2082 Changes between 0.9.6c and 0.9.6d [9 May 2002]
2084 *) Fix crypto/asn1/a_sign.c so that 'parameters' is omitted (not
2085 encoded as NULL) with id-dsa-with-sha1.
2086 [Nils Larsch <nla@trustcenter.de>; problem pointed out by Bodo Moeller]
2088 *) Check various X509_...() return values in apps/req.c.
2089 [Nils Larsch <nla@trustcenter.de>]
2091 *) Fix BASE64 decode (EVP_DecodeUpdate) for data with CR/LF ended lines:
2092 an end-of-file condition would erronously be flagged, when the CRLF
2093 was just at the end of a processed block. The bug was discovered when
2094 processing data through a buffering memory BIO handing the data to a
2095 BASE64-decoding BIO. Bug fund and patch submitted by Pavel Tsekov
2096 <ptsekov@syntrex.com> and Nedelcho Stanev.
2099 *) Implement a countermeasure against a vulnerability recently found
2100 in CBC ciphersuites in SSL 3.0/TLS 1.0: Send an empty fragment
2101 before application data chunks to avoid the use of known IVs
2102 with data potentially chosen by the attacker.
2105 *) Fix length checks in ssl3_get_client_hello().
2108 *) TLS/SSL library bugfix: use s->s3->in_read_app_data differently
2109 to prevent ssl3_read_internal() from incorrectly assuming that
2110 ssl3_read_bytes() found application data while handshake
2111 processing was enabled when in fact s->s3->in_read_app_data was
2112 merely automatically cleared during the initial handshake.
2113 [Bodo Moeller; problem pointed out by Arne Ansper <arne@ats.cyber.ee>]
2115 *) Fix object definitions for Private and Enterprise: they were not
2116 recognized in their shortname (=lowercase) representation. Extend
2117 obj_dat.pl to issue an error when using undefined keywords instead
2118 of silently ignoring the problem (Svenning Sorensen
2119 <sss@sss.dnsalias.net>).
2122 *) Fix DH_generate_parameters() so that it works for 'non-standard'
2123 generators, i.e. generators other than 2 and 5. (Previously, the
2124 code did not properly initialise the 'add' and 'rem' values to
2125 BN_generate_prime().)
2127 In the new general case, we do not insist that 'generator' is
2128 actually a primitive root: This requirement is rather pointless;
2129 a generator of the order-q subgroup is just as good, if not
2133 *) Map new X509 verification errors to alerts. Discovered and submitted by
2134 Tom Wu <tom@arcot.com>.
2137 *) Fix ssl3_pending() (ssl/s3_lib.c) to prevent SSL_pending() from
2138 returning non-zero before the data has been completely received
2139 when using non-blocking I/O.
2140 [Bodo Moeller; problem pointed out by John Hughes]
2142 *) Some of the ciphers missed the strength entry (SSL_LOW etc).
2143 [Ben Laurie, Lutz Jaenicke]
2145 *) Fix bug in SSL_clear(): bad sessions were not removed (found by
2146 Yoram Zahavi <YoramZ@gilian.com>).
2149 *) Add information about CygWin 1.3 and on, and preserve proper
2150 configuration for the versions before that.
2151 [Corinna Vinschen <vinschen@redhat.com> and Richard Levitte]
2153 *) Make removal from session cache (SSL_CTX_remove_session()) more robust:
2154 check whether we deal with a copy of a session and do not delete from
2155 the cache in this case. Problem reported by "Izhar Shoshani Levi"
2156 <izhar@checkpoint.com>.
2159 *) Do not store session data into the internal session cache, if it
2160 is never intended to be looked up (SSL_SESS_CACHE_NO_INTERNAL_LOOKUP
2161 flag is set). Proposed by Aslam <aslam@funk.com>.
2164 *) Have ASN1_BIT_STRING_set_bit() really clear a bit when the requested
2168 *) [In 0.9.6d-engine release:]
2169 Fix a crashbug and a logic bug in hwcrhk_load_pubkey().
2170 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte]
2172 *) Add the configuration target linux-s390x.
2173 [Neale Ferguson <Neale.Ferguson@SoftwareAG-USA.com> via Richard Levitte]
2175 *) The earlier bugfix for the SSL3_ST_SW_HELLO_REQ_C case of
2176 ssl3_accept (ssl/s3_srvr.c) incorrectly used a local flag
2177 variable as an indication that a ClientHello message has been
2178 received. As the flag value will be lost between multiple
2179 invocations of ssl3_accept when using non-blocking I/O, the
2180 function may not be aware that a handshake has actually taken
2181 place, thus preventing a new session from being added to the
2184 To avoid this problem, we now set s->new_session to 2 instead of
2185 using a local variable.
2186 [Lutz Jaenicke, Bodo Moeller]
2188 *) Bugfix: Return -1 from ssl3_get_server_done (ssl3/s3_clnt.c)
2189 if the SSL_R_LENGTH_MISMATCH error is detected.
2190 [Geoff Thorpe, Bodo Moeller]
2192 *) New 'shared_ldflag' column in Configure platform table.
2195 *) Fix EVP_CIPHER_mode macro.
2196 ["Dan S. Camper" <dan@bti.net>]
2198 *) Fix ssl3_read_bytes (ssl/s3_pkt.c): To ignore messages of unknown
2199 type, we must throw them away by setting rr->length to 0.
2200 [D P Chang <dpc@qualys.com>]
2202 Changes between 0.9.6b and 0.9.6c [21 dec 2001]
2204 *) Fix BN_rand_range bug pointed out by Dominikus Scherkl
2205 <Dominikus.Scherkl@biodata.com>. (The previous implementation
2206 worked incorrectly for those cases where range = 10..._2 and
2207 3*range is two bits longer than range.)
2210 *) Only add signing time to PKCS7 structures if it is not already
2214 *) Fix crypto/objects/objects.h: "ld-ce" should be "id-ce",
2215 OBJ_ld_ce should be OBJ_id_ce.
2216 Also some ip-pda OIDs in crypto/objects/objects.txt were
2217 incorrect (cf. RFC 3039).
2218 [Matt Cooper, Frederic Giudicelli, Bodo Moeller]
2220 *) Release CRYPTO_LOCK_DYNLOCK when CRYPTO_destroy_dynlockid()
2221 returns early because it has nothing to do.
2222 [Andy Schneider <andy.schneider@bjss.co.uk>]
2224 *) [In 0.9.6c-engine release:]
2225 Fix mutex callback return values in crypto/engine/hw_ncipher.c.
2226 [Andy Schneider <andy.schneider@bjss.co.uk>]
2228 *) [In 0.9.6c-engine release:]
2229 Add support for Cryptographic Appliance's keyserver technology.
2230 (Use engine 'keyclient')
2231 [Cryptographic Appliances and Geoff Thorpe]
2233 *) Add a configuration entry for OS/390 Unix. The C compiler 'c89'
2234 is called via tools/c89.sh because arguments have to be
2235 rearranged (all '-L' options must appear before the first object
2237 [Richard Shapiro <rshapiro@abinitio.com>]
2239 *) [In 0.9.6c-engine release:]
2240 Add support for Broadcom crypto accelerator cards, backported
2242 [Broadcom, Nalin Dahyabhai <nalin@redhat.com>, Mark Cox]
2244 *) [In 0.9.6c-engine release:]
2245 Add support for SureWare crypto accelerator cards from
2246 Baltimore Technologies. (Use engine 'sureware')
2247 [Baltimore Technologies and Mark Cox]
2249 *) [In 0.9.6c-engine release:]
2250 Add support for crypto accelerator cards from Accelerated
2251 Encryption Processing, www.aep.ie. (Use engine 'aep')
2252 [AEP Inc. and Mark Cox]
2254 *) Add a configuration entry for gcc on UnixWare.
2255 [Gary Benson <gbenson@redhat.com>]
2257 *) Change ssl/s2_clnt.c and ssl/s2_srvr.c so that received handshake
2258 messages are stored in a single piece (fixed-length part and
2259 variable-length part combined) and fix various bugs found on the way.
2262 *) Disable caching in BIO_gethostbyname(), directly use gethostbyname()
2263 instead. BIO_gethostbyname() does not know what timeouts are
2264 appropriate, so entries would stay in cache even when they have
2266 [Bodo Moeller; problem pointed out by Rich Salz <rsalz@zolera.com>
2268 *) Change ssl23_get_client_hello (ssl/s23_srvr.c) behaviour when
2269 faced with a pathologically small ClientHello fragment that does
2270 not contain client_version: Instead of aborting with an error,
2271 simply choose the highest available protocol version (i.e.,
2272 TLS 1.0 unless it is disabled). In practice, ClientHello
2273 messages are never sent like this, but this change gives us
2274 strictly correct behaviour at least for TLS.
2277 *) Fix SSL handshake functions and SSL_clear() such that SSL_clear()
2278 never resets s->method to s->ctx->method when called from within
2279 one of the SSL handshake functions.
2280 [Bodo Moeller; problem pointed out by Niko Baric]
2282 *) In ssl3_get_client_hello (ssl/s3_srvr.c), generate a fatal alert
2283 (sent using the client's version number) if client_version is
2284 smaller than the protocol version in use. Also change
2285 ssl23_get_client_hello (ssl/s23_srvr.c) to select TLS 1.0 if
2286 the client demanded SSL 3.0 but only TLS 1.0 is enabled; then
2287 the client will at least see that alert.
2290 *) Fix ssl3_get_message (ssl/s3_both.c) to handle message fragmentation
2294 *) Avoid infinite loop in ssl3_get_message (ssl/s3_both.c) if a
2295 client receives HelloRequest while in a handshake.
2296 [Bodo Moeller; bug noticed by Andy Schneider <andy.schneider@bjss.co.uk>]
2298 *) Bugfix in ssl3_accept (ssl/s3_srvr.c): Case SSL3_ST_SW_HELLO_REQ_C
2299 should end in 'break', not 'goto end' which circuments various
2300 cleanups done in state SSL_ST_OK. But session related stuff
2301 must be disabled for SSL_ST_OK in the case that we just sent a
2304 Also avoid some overhead by not calling ssl_init_wbio_buffer()
2305 before just sending a HelloRequest.
2306 [Bodo Moeller, Eric Rescorla <ekr@rtfm.com>]
2308 *) Fix ssl/s3_enc.c, ssl/t1_enc.c and ssl/s3_pkt.c so that we don't
2309 reveal whether illegal block cipher padding was found or a MAC
2310 verification error occured. (Neither SSLerr() codes nor alerts
2311 are directly visible to potential attackers, but the information
2312 may leak via logfiles.)
2314 Similar changes are not required for the SSL 2.0 implementation
2315 because the number of padding bytes is sent in clear for SSL 2.0,
2316 and the extra bytes are just ignored. However ssl/s2_pkt.c
2317 failed to verify that the purported number of padding bytes is in
2321 *) Add OpenUNIX-8 support including shared libraries
2322 (Boyd Lynn Gerber <gerberb@zenez.com>).
2325 *) Improve RSA_padding_check_PKCS1_OAEP() check again to avoid
2326 'wristwatch attack' using huge encoding parameters (cf.
2327 James H. Manger's CRYPTO 2001 paper). Note that the
2328 RSA_PKCS1_OAEP_PADDING case of RSA_private_decrypt() does not use
2329 encoding parameters and hence was not vulnerable.
2332 *) BN_sqr() bug fix.
2333 [Ulf Möller, reported by Jim Ellis <jim.ellis@cavium.com>]
2335 *) Rabin-Miller test analyses assume uniformly distributed witnesses,
2336 so use BN_pseudo_rand_range() instead of using BN_pseudo_rand()
2337 followed by modular reduction.
2338 [Bodo Moeller; pointed out by Adam Young <AYoung1@NCSUS.JNJ.COM>]
2340 *) Add BN_pseudo_rand_range() with obvious functionality: BN_rand_range()
2341 equivalent based on BN_pseudo_rand() instead of BN_rand().
2344 *) s3_srvr.c: allow sending of large client certificate lists (> 16 kB).
2345 This function was broken, as the check for a new client hello message
2346 to handle SGC did not allow these large messages.
2347 (Tracked down by "Douglas E. Engert" <deengert@anl.gov>.)
2350 *) Add alert descriptions for TLSv1 to SSL_alert_desc_string[_long]().
2353 *) Fix buggy behaviour of BIO_get_num_renegotiates() and BIO_ctrl()
2354 for BIO_C_GET_WRITE_BUF_SIZE ("Stephen Hinton" <shinton@netopia.com>).
2357 *) Rework the configuration and shared library support for Tru64 Unix.
2358 The configuration part makes use of modern compiler features and
2359 still retains old compiler behavior for those that run older versions
2360 of the OS. The shared library support part includes a variant that
2361 uses the RPATH feature, and is available through the special
2362 configuration target "alpha-cc-rpath", which will never be selected
2364 [Tim Mooney <mooney@dogbert.cc.ndsu.NoDak.edu> via Richard Levitte]
2366 *) In ssl3_get_key_exchange (ssl/s3_clnt.c), call ssl3_get_message()
2367 with the same message size as in ssl3_get_certificate_request().
2368 Otherwise, if no ServerKeyExchange message occurs, CertificateRequest
2369 messages might inadvertently be reject as too long.
2370 [Petr Lampa <lampa@fee.vutbr.cz>]
2372 *) Enhanced support for IA-64 Unix platforms (well, Linux and HP-UX).
2375 *) Modified SSL library such that the verify_callback that has been set
2376 specificly for an SSL object with SSL_set_verify() is actually being
2377 used. Before the change, a verify_callback set with this function was
2378 ignored and the verify_callback() set in the SSL_CTX at the time of
2379 the call was used. New function X509_STORE_CTX_set_verify_cb() introduced
2380 to allow the necessary settings.
2383 *) Initialize static variable in crypto/dsa/dsa_lib.c and crypto/dh/dh_lib.c
2384 explicitly to NULL, as at least on Solaris 8 this seems not always to be
2385 done automatically (in contradiction to the requirements of the C
2386 standard). This made problems when used from OpenSSH.
2389 *) In OpenSSL 0.9.6a and 0.9.6b, crypto/dh/dh_key.c ignored
2390 dh->length and always used
2392 BN_rand_range(priv_key, dh->p).
2394 BN_rand_range() is not necessary for Diffie-Hellman, and this
2395 specific range makes Diffie-Hellman unnecessarily inefficient if
2396 dh->length (recommended exponent length) is much smaller than the
2397 length of dh->p. We could use BN_rand_range() if the order of
2398 the subgroup was stored in the DH structure, but we only have
2403 BN_rand(priv_key, l, ...)
2405 where 'l' is dh->length if this is defined, or BN_num_bits(dh->p)-1
2411 RSA_eay_public_encrypt
2412 RSA_eay_private_decrypt
2413 RSA_eay_private_encrypt (signing)
2414 RSA_eay_public_decrypt (signature verification)
2416 (default implementations for RSA_public_encrypt,
2417 RSA_private_decrypt, RSA_private_encrypt, RSA_public_decrypt),
2418 always reject numbers >= n.
2421 *) In crypto/rand/md_rand.c, use a new short-time lock CRYPTO_LOCK_RAND2
2422 to synchronize access to 'locking_thread'. This is necessary on
2423 systems where access to 'locking_thread' (an 'unsigned long'
2424 variable) is not atomic.
2427 *) In crypto/rand/md_rand.c, set 'locking_thread' to current thread's ID
2428 *before* setting the 'crypto_lock_rand' flag. The previous code had
2429 a race condition if 0 is a valid thread ID.
2430 [Travis Vitek <vitek@roguewave.com>]
2432 *) Add support for shared libraries under Irix.
2433 [Albert Chin-A-Young <china@thewrittenword.com>]
2435 *) Add configuration option to build on Linux on both big-endian and
2437 [Ralf Baechle <ralf@uni-koblenz.de>]
2439 *) Add the possibility to create shared libraries on HP-UX.
2442 Changes between 0.9.6a and 0.9.6b [9 Jul 2001]
2444 *) Change ssleay_rand_bytes (crypto/rand/md_rand.c)
2445 to avoid a SSLeay/OpenSSL PRNG weakness pointed out by
2446 Markku-Juhani O. Saarinen <markku-juhani.saarinen@nokia.com>:
2447 PRNG state recovery was possible based on the output of
2448 one PRNG request appropriately sized to gain knowledge on
2449 'md' followed by enough consecutive 1-byte PRNG requests
2450 to traverse all of 'state'.
2452 1. When updating 'md_local' (the current thread's copy of 'md')
2453 during PRNG output generation, hash all of the previous
2454 'md_local' value, not just the half used for PRNG output.
2456 2. Make the number of bytes from 'state' included into the hash
2457 independent from the number of PRNG bytes requested.
2459 The first measure alone would be sufficient to avoid
2460 Markku-Juhani's attack. (Actually it had never occurred
2461 to me that the half of 'md_local' used for chaining was the
2462 half from which PRNG output bytes were taken -- I had always
2463 assumed that the secret half would be used.) The second
2464 measure makes sure that additional data from 'state' is never
2465 mixed into 'md_local' in small portions; this heuristically
2466 further strengthens the PRNG.
2469 *) Fix crypto/bn/asm/mips3.s.
2472 *) When only the key is given to "enc", the IV is undefined. Print out
2473 an error message in this case.
2476 *) Handle special case when X509_NAME is empty in X509 printing routines.
2479 *) In dsa_do_verify (crypto/dsa/dsa_ossl.c), verify that r and s are
2480 positive and less than q.
2483 *) Don't change *pointer in CRYPTO_add_lock() is add_lock_callback is
2484 used: it isn't thread safe and the add_lock_callback should handle
2486 [Paul Rose <Paul.Rose@bridge.com>]
2488 *) Verify that incoming data obeys the block size in
2489 ssl3_enc (ssl/s3_enc.c) and tls1_enc (ssl/t1_enc.c).
2493 [Ulf Möller, Bodo Möller]
2495 *) The countermeasure against Bleichbacher's attack on PKCS #1 v1.5
2496 RSA encryption was accidentally removed in s3_srvr.c in OpenSSL 0.9.5
2497 when fixing the server behaviour for backwards-compatible 'client
2498 hello' messages. (Note that the attack is impractical against
2499 SSL 3.0 and TLS 1.0 anyway because length and version checking
2500 means that the probability of guessing a valid ciphertext is
2501 around 2^-40; see section 5 in Bleichenbacher's CRYPTO '98
2504 Before 0.9.5, the countermeasure (hide the error by generating a
2505 random 'decryption result') did not work properly because
2506 ERR_clear_error() was missing, meaning that SSL_get_error() would
2507 detect the supposedly ignored error.
2509 Both problems are now fixed.
2512 *) In crypto/bio/bf_buff.c, increase DEFAULT_BUFFER_SIZE to 4096
2513 (previously it was 1024).
2516 *) Fix for compatibility mode trust settings: ignore trust settings
2517 unless some valid trust or reject settings are present.
2520 *) Fix for blowfish EVP: its a variable length cipher.
2523 *) Fix various bugs related to DSA S/MIME verification. Handle missing
2524 parameters in DSA public key structures and return an error in the
2525 DSA routines if parameters are absent.
2528 *) In versions up to 0.9.6, RAND_file_name() resorted to file ".rnd"
2529 in the current directory if neither $RANDFILE nor $HOME was set.
2530 RAND_file_name() in 0.9.6a returned NULL in this case. This has
2531 caused some confusion to Windows users who haven't defined $HOME.
2532 Thus RAND_file_name() is changed again: e_os.h can define a
2533 DEFAULT_HOME, which will be used if $HOME is not set.
2534 For Windows, we use "C:"; on other platforms, we still require
2535 environment variables.
2537 *) Move 'if (!initialized) RAND_poll()' into regions protected by
2538 CRYPTO_LOCK_RAND. This is not strictly necessary, but avoids
2539 having multiple threads call RAND_poll() concurrently.
2542 *) In crypto/rand/md_rand.c, replace 'add_do_not_lock' flag by a
2543 combination of a flag and a thread ID variable.
2544 Otherwise while one thread is in ssleay_rand_bytes (which sets the
2545 flag), *other* threads can enter ssleay_add_bytes without obeying
2546 the CRYPTO_LOCK_RAND lock (and may even illegally release the lock
2547 that they do not hold after the first thread unsets add_do_not_lock).
2550 *) Change bctest again: '-x' expressions are not available in all
2554 Changes between 0.9.6 and 0.9.6a [5 Apr 2001]
2556 *) Fix a couple of memory leaks in PKCS7_dataDecode()
2557 [Steve Henson, reported by Heyun Zheng <hzheng@atdsprint.com>]
2559 *) Change Configure and Makefiles to provide EXE_EXT, which will contain
2560 the default extension for executables, if any. Also, make the perl
2561 scripts that use symlink() to test if it really exists and use "cp"
2562 if it doesn't. All this made OpenSSL compilable and installable in
2566 *) Fix for asn1_GetSequence() for indefinite length constructed data.
2567 If SEQUENCE is length is indefinite just set c->slen to the total
2568 amount of data available.
2569 [Steve Henson, reported by shige@FreeBSD.org]
2570 [This change does not apply to 0.9.7.]
2572 *) Change bctest to avoid here-documents inside command substitution
2573 (workaround for FreeBSD /bin/sh bug).
2574 For compatibility with Ultrix, avoid shell functions (introduced
2575 in the bctest version that searches along $PATH).
2578 *) Rename 'des_encrypt' to 'des_encrypt1'. This avoids the clashes
2579 with des_encrypt() defined on some operating systems, like Solaris
2583 *) Check the result of RSA-CRT (see D. Boneh, R. DeMillo, R. Lipton:
2584 On the Importance of Eliminating Errors in Cryptographic
2585 Computations, J. Cryptology 14 (2001) 2, 101-119,
2586 http://theory.stanford.edu/~dabo/papers/faults.ps.gz).
2589 *) MIPS assembler BIGNUM division bug fix.
2592 *) Disabled incorrect Alpha assembler code.
2595 *) Fix PKCS#7 decode routines so they correctly update the length
2596 after reading an EOC for the EXPLICIT tag.
2598 [This change does not apply to 0.9.7.]
2600 *) Fix bug in PKCS#12 key generation routines. This was triggered
2601 if a 3DES key was generated with a 0 initial byte. Include
2602 PKCS12_BROKEN_KEYGEN compilation option to retain the old
2603 (but broken) behaviour.
2606 *) Enhance bctest to search for a working bc along $PATH and print
2608 [Tim Rice <tim@multitalents.net> via Richard Levitte]
2610 *) Fix memory leaks in err.c: free err_data string if necessary;
2611 don't write to the wrong index in ERR_set_error_data.
2614 *) Implement ssl23_peek (analogous to ssl23_read), which previously
2618 *) Replace rdtsc with _emit statements for VC++ version 5.
2619 [Jeremy Cooper <jeremy@baymoo.org>]
2621 *) Make it possible to reuse SSLv2 sessions.
2624 *) In copy_email() check for >= 0 as a return value for
2625 X509_NAME_get_index_by_NID() since 0 is a valid index.
2626 [Steve Henson reported by Massimiliano Pala <madwolf@opensca.org>]
2628 *) Avoid coredump with unsupported or invalid public keys by checking if
2629 X509_get_pubkey() fails in PKCS7_verify(). Fix memory leak when
2630 PKCS7_verify() fails with non detached data.
2633 *) Don't use getenv in library functions when run as setuid/setgid.
2634 New function OPENSSL_issetugid().
2637 *) Avoid false positives in memory leak detection code (crypto/mem_dbg.c)
2638 due to incorrect handling of multi-threading:
2640 1. Fix timing glitch in the MemCheck_off() portion of CRYPTO_mem_ctrl().
2642 2. Fix logical glitch in is_MemCheck_on() aka CRYPTO_is_mem_check_on().
2644 3. Count how many times MemCheck_off() has been called so that
2645 nested use can be treated correctly. This also avoids
2646 inband-signalling in the previous code (which relied on the
2647 assumption that thread ID 0 is impossible).
2650 *) Add "-rand" option also to s_client and s_server.
2653 *) Fix CPU detection on Irix 6.x.
2654 [Kurt Hockenbury <khockenb@stevens-tech.edu> and
2655 "Bruce W. Forsberg" <bruce.forsberg@baesystems.com>]
2657 *) Fix X509_NAME bug which produced incorrect encoding if X509_NAME
2660 [This change does not apply to 0.9.7.]
2662 *) Use the cached encoding of an X509_NAME structure rather than
2663 copying it. This is apparently the reason for the libsafe "errors"
2664 but the code is actually correct.
2667 *) Add new function BN_rand_range(), and fix DSA_sign_setup() to prevent
2668 Bleichenbacher's DSA attack.
2669 Extend BN_[pseudo_]rand: As before, top=1 forces the highest two bits
2670 to be set and top=0 forces the highest bit to be set; top=-1 is new
2671 and leaves the highest bit random.
2672 [Ulf Moeller, Bodo Moeller]
2674 *) In the NCONF_...-based implementations for CONF_... queries
2675 (crypto/conf/conf_lib.c), if the input LHASH is NULL, avoid using
2676 a temporary CONF structure with the data component set to NULL
2677 (which gives segmentation faults in lh_retrieve).
2678 Instead, use NULL for the CONF pointer in CONF_get_string and
2679 CONF_get_number (which may use environment variables) and directly
2680 return NULL from CONF_get_section.
2683 *) Fix potential buffer overrun for EBCDIC.
2686 *) Tolerate nonRepudiation as being valid for S/MIME signing and certSign
2687 keyUsage if basicConstraints absent for a CA.
2690 *) Make SMIME_write_PKCS7() write mail header values with a format that
2691 is more generally accepted (no spaces before the semicolon), since
2692 some programs can't parse those values properly otherwise. Also make
2693 sure BIO's that break lines after each write do not create invalid
2697 *) Make the CRL encoding routines work with empty SEQUENCE OF. The
2698 macros previously used would not encode an empty SEQUENCE OF
2699 and break the signature.
2701 [This change does not apply to 0.9.7.]
2703 *) Zero the premaster secret after deriving the master secret in
2707 *) Add some EVP_add_digest_alias registrations (as found in
2708 OpenSSL_add_all_digests()) to SSL_library_init()
2709 aka OpenSSL_add_ssl_algorithms(). This provides improved
2710 compatibility with peers using X.509 certificates
2711 with unconventional AlgorithmIdentifier OIDs.
2714 *) Fix for Irix with NO_ASM.
2715 ["Bruce W. Forsberg" <bruce.forsberg@baesystems.com>]
2717 *) ./config script fixes.
2718 [Ulf Moeller, Richard Levitte]
2720 *) Fix 'openssl passwd -1'.
2723 *) Change PKCS12_key_gen_asc() so it can cope with non null
2724 terminated strings whose length is passed in the passlen
2725 parameter, for example from PEM callbacks. This was done
2726 by adding an extra length parameter to asc2uni().
2727 [Steve Henson, reported by <oddissey@samsung.co.kr>]
2729 *) Fix C code generated by 'openssl dsaparam -C': If a BN_bin2bn
2730 call failed, free the DSA structure.
2733 *) Fix to uni2asc() to cope with zero length Unicode strings.
2734 These are present in some PKCS#12 files.
2737 *) Increase s2->wbuf allocation by one byte in ssl2_new (ssl/s2_lib.c).
2738 Otherwise do_ssl_write (ssl/s2_pkt.c) will write beyond buffer limits
2739 when writing a 32767 byte record.
2740 [Bodo Moeller; problem reported by Eric Day <eday@concentric.net>]
2742 *) In RSA_eay_public_{en,ed}crypt and RSA_eay_mod_exp (rsa_eay.c),
2743 obtain lock CRYPTO_LOCK_RSA before setting rsa->_method_mod_{n,p,q}.
2745 (RSA objects have a reference count access to which is protected
2746 by CRYPTO_LOCK_RSA [see rsa_lib.c, s3_srvr.c, ssl_cert.c, ssl_rsa.c],
2747 so they are meant to be shared between threads.)
2748 [Bodo Moeller, Geoff Thorpe; original patch submitted by
2749 "Reddie, Steven" <Steven.Reddie@ca.com>]
2751 *) Fix a deadlock in CRYPTO_mem_leaks().
2754 *) Use better test patterns in bntest.
2757 *) rand_win.c fix for Borland C.
2760 *) BN_rshift bugfix for n == 0.
2763 *) Add a 'bctest' script that checks for some known 'bc' bugs
2764 so that 'make test' does not abort just because 'bc' is broken.
2767 *) Store verify_result within SSL_SESSION also for client side to
2768 avoid potential security hole. (Re-used sessions on the client side
2769 always resulted in verify_result==X509_V_OK, not using the original
2770 result of the server certificate verification.)
2773 *) Fix ssl3_pending: If the record in s->s3->rrec is not of type
2774 SSL3_RT_APPLICATION_DATA, return 0.
2775 Similarly, change ssl2_pending to return 0 if SSL_in_init(s) is true.
2779 Both ssl2_peek and ssl3_peek, which were totally broken in earlier
2780 releases, have been re-implemented by renaming the previous
2781 implementations of ssl2_read and ssl3_read to ssl2_read_internal
2782 and ssl3_read_internal, respectively, and adding 'peek' parameters
2783 to them. The new ssl[23]_{read,peek} functions are calls to
2784 ssl[23]_read_internal with the 'peek' flag set appropriately.
2785 A 'peek' parameter has also been added to ssl3_read_bytes, which
2786 does the actual work for ssl3_read_internal.
2789 *) Initialise "ex_data" member of RSA/DSA/DH structures prior to calling
2790 the method-specific "init()" handler. Also clean up ex_data after
2791 calling the method-specific "finish()" handler. Previously, this was
2792 happening the other way round.
2795 *) Increase BN_CTX_NUM (the number of BIGNUMs in a BN_CTX) to 16.
2796 The previous value, 12, was not always sufficient for BN_mod_exp().
2799 *) Make sure that shared libraries get the internal name engine with
2800 the full version number and not just 0. This should mark the
2801 shared libraries as not backward compatible. Of course, this should
2802 be changed again when we can guarantee backward binary compatibility.
2805 *) Fix typo in get_cert_by_subject() in by_dir.c
2806 [Jean-Marc Desperrier <jean-marc.desperrier@certplus.com>]
2808 *) Rework the system to generate shared libraries:
2810 - Make note of the expected extension for the shared libraries and
2811 if there is a need for symbolic links from for example libcrypto.so.0
2812 to libcrypto.so.0.9.7. There is extended info in Configure for
2815 - Make as few rebuilds of the shared libraries as possible.
2817 - Still avoid linking the OpenSSL programs with the shared libraries.
2819 - When installing, install the shared libraries separately from the
2823 *) Fix SSL_CTX_set_read_ahead macro to actually use its argument.
2825 Copy SSL_CTX's read_ahead flag to SSL object directly in SSL_new
2826 and not in SSL_clear because the latter is also used by the
2827 accept/connect functions; previously, the settings made by
2828 SSL_set_read_ahead would be lost during the handshake.
2829 [Bodo Moeller; problems reported by Anders Gertz <gertz@epact.se>]
2831 *) Correct util/mkdef.pl to be selective about disabled algorithms.
2832 Previously, it would create entries for disableed algorithms no
2836 *) Added several new manual pages for SSL_* function.
2839 Changes between 0.9.5a and 0.9.6 [24 Sep 2000]
2841 *) In ssl23_get_client_hello, generate an error message when faced
2842 with an initial SSL 3.0/TLS record that is too small to contain the
2843 first two bytes of the ClientHello message, i.e. client_version.
2844 (Note that this is a pathologic case that probably has never happened
2845 in real life.) The previous approach was to use the version number
2846 from the record header as a substitute; but our protocol choice
2847 should not depend on that one because it is not authenticated
2848 by the Finished messages.
2851 *) More robust randomness gathering functions for Windows.
2852 [Jeffrey Altman <jaltman@columbia.edu>]
2854 *) For compatibility reasons if the flag X509_V_FLAG_ISSUER_CHECK is
2855 not set then we don't setup the error code for issuer check errors
2856 to avoid possibly overwriting other errors which the callback does
2857 handle. If an application does set the flag then we assume it knows
2858 what it is doing and can handle the new informational codes
2862 *) Fix for a nasty bug in ASN1_TYPE handling. ASN1_TYPE is used for
2863 a general "ANY" type, as such it should be able to decode anything
2864 including tagged types. However it didn't check the class so it would
2865 wrongly interpret tagged types in the same way as their universal
2866 counterpart and unknown types were just rejected. Changed so that the
2867 tagged and unknown types are handled in the same way as a SEQUENCE:
2868 that is the encoding is stored intact. There is also a new type
2869 "V_ASN1_OTHER" which is used when the class is not universal, in this
2870 case we have no idea what the actual type is so we just lump them all
2874 *) On VMS, stdout may very well lead to a file that is written to
2875 in a record-oriented fashion. That means that every write() will
2876 write a separate record, which will be read separately by the
2877 programs trying to read from it. This can be very confusing.
2879 The solution is to put a BIO filter in the way that will buffer
2880 text until a linefeed is reached, and then write everything a
2881 line at a time, so every record written will be an actual line,
2882 not chunks of lines and not (usually doesn't happen, but I've
2883 seen it once) several lines in one record. BIO_f_linebuffer() is
2886 Currently, it's a VMS-only method, because that's where it has
2887 been tested well enough.
2890 *) Remove 'optimized' squaring variant in BN_mod_mul_montgomery,
2891 it can return incorrect results.
2892 (Note: The buggy variant was not enabled in OpenSSL 0.9.5a,
2893 but it was in 0.9.6-beta[12].)
2896 *) Disable the check for content being present when verifying detached
2897 signatures in pk7_smime.c. Some versions of Netscape (wrongly)
2898 include zero length content when signing messages.
2901 *) New BIO_shutdown_wr macro, which invokes the BIO_C_SHUTDOWN_WR
2902 BIO_ctrl (for BIO pairs).
2905 *) Add DSO method for VMS.
2908 *) Bug fix: Montgomery multiplication could produce results with the
2912 *) Add RPM specification openssl.spec and modify it to build three
2913 packages. The default package contains applications, application
2914 documentation and run-time libraries. The devel package contains
2915 include files, static libraries and function documentation. The
2916 doc package contains the contents of the doc directory. The original
2917 openssl.spec was provided by Damien Miller <djm@mindrot.org>.
2920 *) Add a large number of documentation files for many SSL routines.
2921 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>]
2923 *) Add a configuration entry for Sony News 4.
2924 [NAKAJI Hiroyuki <nakaji@tutrp.tut.ac.jp>]
2926 *) Don't set the two most significant bits to one when generating a
2927 random number < q in the DSA library.
2930 *) New SSL API mode 'SSL_MODE_AUTO_RETRY'. This disables the default
2931 behaviour that SSL_read may result in SSL_ERROR_WANT_READ (even if
2932 the underlying transport is blocking) if a handshake took place.
2933 (The default behaviour is needed by applications such as s_client
2934 and s_server that use select() to determine when to use SSL_read;
2935 but for applications that know in advance when to expect data, it
2936 just makes things more complicated.)
2939 *) Add RAND_egd_bytes(), which gives control over the number of bytes read
2943 *) Add a few more EBCDIC conditionals that make `req' and `x509'
2944 work better on such systems.
2945 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>]
2947 *) Add two demo programs for PKCS12_parse() and PKCS12_create().
2948 Update PKCS12_parse() so it copies the friendlyName and the
2949 keyid to the certificates aux info.
2952 *) Fix bug in PKCS7_verify() which caused an infinite loop
2953 if there was more than one signature.
2954 [Sven Uszpelkat <su@celocom.de>]
2956 *) Major change in util/mkdef.pl to include extra information
2957 about each symbol, as well as presentig variables as well
2958 as functions. This change means that there's n more need
2959 to rebuild the .num files when some algorithms are excluded.
2962 *) Allow the verify time to be set by an application,
2963 rather than always using the current time.
2966 *) Phase 2 verify code reorganisation. The certificate
2967 verify code now looks up an issuer certificate by a
2968 number of criteria: subject name, authority key id
2969 and key usage. It also verifies self signed certificates
2970 by the same criteria. The main comparison function is
2971 X509_check_issued() which performs these checks.
2973 Lot of changes were necessary in order to support this
2974 without completely rewriting the lookup code.
2976 Authority and subject key identifier are now cached.
2978 The LHASH 'certs' is X509_STORE has now been replaced
2979 by a STACK_OF(X509_OBJECT). This is mainly because an
2980 LHASH can't store or retrieve multiple objects with
2981 the same hash value.
2983 As a result various functions (which were all internal
2984 use only) have changed to handle the new X509_STORE
2985 structure. This will break anything that messed round
2986 with X509_STORE internally.
2988 The functions X509_STORE_add_cert() now checks for an
2989 exact match, rather than just subject name.
2991 The X509_STORE API doesn't directly support the retrieval
2992 of multiple certificates matching a given criteria, however
2993 this can be worked round by performing a lookup first
2994 (which will fill the cache with candidate certificates)
2995 and then examining the cache for matches. This is probably
2996 the best we can do without throwing out X509_LOOKUP
2997 entirely (maybe later...).
2999 The X509_VERIFY_CTX structure has been enhanced considerably.
3001 All certificate lookup operations now go via a get_issuer()
3002 callback. Although this currently uses an X509_STORE it
3003 can be replaced by custom lookups. This is a simple way
3004 to bypass the X509_STORE hackery necessary to make this
3005 work and makes it possible to use more efficient techniques
3006 in future. A very simple version which uses a simple
3007 STACK for its trusted certificate store is also provided
3008 using X509_STORE_CTX_trusted_stack().
3010 The verify_cb() and verify() callbacks now have equivalents
3011 in the X509_STORE_CTX structure.
3013 X509_STORE_CTX also has a 'flags' field which can be used
3014 to customise the verify behaviour.
3017 *) Add new PKCS#7 signing option PKCS7_NOSMIMECAP which
3018 excludes S/MIME capabilities.
3021 *) When a certificate request is read in keep a copy of the
3022 original encoding of the signed data and use it when outputing
3023 again. Signatures then use the original encoding rather than
3024 a decoded, encoded version which may cause problems if the
3025 request is improperly encoded.
3028 *) For consistency with other BIO_puts implementations, call
3029 buffer_write(b, ...) directly in buffer_puts instead of calling
3032 In BIO_puts, increment b->num_write as in BIO_write.
3033 [Peter.Sylvester@EdelWeb.fr]
3035 *) Fix BN_mul_word for the case where the word is 0. (We have to use
3036 BN_zero, we may not return a BIGNUM with an array consisting of
3040 *) Avoid calling abort() from within the library when problems are
3041 detected, except if preprocessor symbols have been defined
3042 (such as REF_CHECK, BN_DEBUG etc.).
3045 *) New openssl application 'rsautl'. This utility can be
3046 used for low level RSA operations. DER public key
3047 BIO/fp routines also added.
3050 *) New Configure entry and patches for compiling on QNX 4.
3051 [Andreas Schneider <andreas@ds3.etech.fh-hamburg.de>]
3053 *) A demo state-machine implementation was sponsored by
3054 Nuron (http://www.nuron.com/) and is now available in
3055 demos/state_machine.
3058 *) New options added to the 'dgst' utility for signature
3059 generation and verification.
3062 *) Unrecognized PKCS#7 content types are now handled via a
3063 catch all ASN1_TYPE structure. This allows unsupported
3064 types to be stored as a "blob" and an application can
3065 encode and decode it manually.
3068 *) Fix various signed/unsigned issues to make a_strex.c
3070 [Oscar Jacobsson <oscar.jacobsson@celocom.com>]
3072 *) ASN1 fixes. i2d_ASN1_OBJECT was not returning the correct
3073 length if passed a buffer. ASN1_INTEGER_to_BN failed
3074 if passed a NULL BN and its argument was negative.
3075 [Steve Henson, pointed out by Sven Heiberg <sven@tartu.cyber.ee>]
3077 *) Modification to PKCS#7 encoding routines to output definite
3078 length encoding. Since currently the whole structures are in
3079 memory there's not real point in using indefinite length
3080 constructed encoding. However if OpenSSL is compiled with
3081 the flag PKCS7_INDEFINITE_ENCODING the old form is used.
3084 *) Added BIO_vprintf() and BIO_vsnprintf().
3087 *) Added more prefixes to parse for in the the strings written
3088 through a logging bio, to cover all the levels that are available
3089 through syslog. The prefixes are now:
3091 PANIC, EMERG, EMR => LOG_EMERG
3092 ALERT, ALR => LOG_ALERT
3093 CRIT, CRI => LOG_CRIT
3094 ERROR, ERR => LOG_ERR
3095 WARNING, WARN, WAR => LOG_WARNING
3096 NOTICE, NOTE, NOT => LOG_NOTICE
3097 INFO, INF => LOG_INFO
3098 DEBUG, DBG => LOG_DEBUG
3100 and as before, if none of those prefixes are present at the
3101 beginning of the string, LOG_ERR is chosen.
3103 On Win32, the LOG_* levels are mapped according to this:
3105 LOG_EMERG, LOG_ALERT, LOG_CRIT, LOG_ERR => EVENTLOG_ERROR_TYPE
3106 LOG_WARNING => EVENTLOG_WARNING_TYPE
3107 LOG_NOTICE, LOG_INFO, LOG_DEBUG => EVENTLOG_INFORMATION_TYPE
3111 *) Made it possible to reconfigure with just the configuration
3112 argument "reconf" or "reconfigure". The command line arguments
3113 are stored in Makefile.ssl in the variable CONFIGURE_ARGS,
3114 and are retrieved from there when reconfiguring.
3118 [Assar Westerlund <assar@sics.se>, Richard Levitte]
3120 *) Add the arguments -CAfile and -CApath to the pkcs12 utility.
3123 *) The obj_dat.pl script was messing up the sorting of object
3124 names. The reason was that it compared the quoted version
3125 of strings as a result "OCSP" > "OCSP Signing" because
3126 " > SPACE. Changed script to store unquoted versions of
3127 names and add quotes on output. It was also omitting some
3128 names from the lookup table if they were given a default
3129 value (that is if SN is missing it is given the same
3130 value as LN and vice versa), these are now added on the
3131 grounds that if an object has a name we should be able to
3132 look it up. Finally added warning output when duplicate
3133 short or long names are found.
3136 *) Changes needed for Tandem NSK.
3137 [Scott Uroff <scott@xypro.com>]
3139 *) Fix SSL 2.0 rollback checking: Due to an off-by-one error in
3140 RSA_padding_check_SSLv23(), special padding was never detected
3141 and thus the SSL 3.0/TLS 1.0 countermeasure against protocol
3142 version rollback attacks was not effective.
3144 In s23_clnt.c, don't use special rollback-attack detection padding
3145 (RSA_SSLV23_PADDING) if SSL 2.0 is the only protocol enabled in the
3146 client; similarly, in s23_srvr.c, don't do the rollback check if
3147 SSL 2.0 is the only protocol enabled in the server.
3150 *) Make it possible to get hexdumps of unprintable data with 'openssl
3151 asn1parse'. By implication, the functions ASN1_parse_dump() and
3152 BIO_dump_indent() are added.
3155 *) New functions ASN1_STRING_print_ex() and X509_NAME_print_ex()
3156 these print out strings and name structures based on various
3157 flags including RFC2253 support and proper handling of
3158 multibyte characters. Added options to the 'x509' utility
3159 to allow the various flags to be set.
3162 *) Various fixes to use ASN1_TIME instead of ASN1_UTCTIME.
3163 Also change the functions X509_cmp_current_time() and
3164 X509_gmtime_adj() work with an ASN1_TIME structure,
3165 this will enable certificates using GeneralizedTime in validity
3166 dates to be checked.
3169 *) Make the NEG_PUBKEY_BUG code (which tolerates invalid
3170 negative public key encodings) on by default,
3171 NO_NEG_PUBKEY_BUG can be set to disable it.
3174 *) New function c2i_ASN1_OBJECT() which acts on ASN1_OBJECT
3175 content octets. An i2c_ASN1_OBJECT is unnecessary because
3176 the encoding can be trivially obtained from the structure.
3179 *) crypto/err.c locking bugfix: Use write locks (CRYPTO_w_[un]lock),
3180 not read locks (CRYPTO_r_[un]lock).
3183 *) A first attempt at creating official support for shared
3184 libraries through configuration. I've kept it so the
3185 default is static libraries only, and the OpenSSL programs
3186 are always statically linked for now, but there are
3187 preparations for dynamic linking in place.
3188 This has been tested on Linux and Tru64.
3191 *) Randomness polling function for Win9x, as described in:
3192 Peter Gutmann, Software Generation of Practically Strong
3196 *) Fix so PRNG is seeded in req if using an already existing
3200 *) New options to smime application. -inform and -outform
3201 allow alternative formats for the S/MIME message including
3202 PEM and DER. The -content option allows the content to be
3203 specified separately. This should allow things like Netscape
3204 form signing output easier to verify.
3207 *) Fix the ASN1 encoding of tags using the 'long form'.
3210 *) New ASN1 functions, i2c_* and c2i_* for INTEGER and BIT
3211 STRING types. These convert content octets to and from the
3212 underlying type. The actual tag and length octets are
3213 already assumed to have been read in and checked. These
3214 are needed because all other string types have virtually
3215 identical handling apart from the tag. By having versions
3216 of the ASN1 functions that just operate on content octets
3217 IMPLICIT tagging can be handled properly. It also allows
3218 the ASN1_ENUMERATED code to be cut down because ASN1_ENUMERATED
3219 and ASN1_INTEGER are identical apart from the tag.
3222 *) Change the handling of OID objects as follows:
3224 - New object identifiers are inserted in objects.txt, following
3225 the syntax given in objects.README.
3226 - objects.pl is used to process obj_mac.num and create a new
3228 - obj_dat.pl is used to create a new obj_dat.h, using the data in
3231 This is currently kind of a hack, and the perl code in objects.pl
3232 isn't very elegant, but it works as I intended. The simplest way
3233 to check that it worked correctly is to look in obj_dat.h and
3234 check the array nid_objs and make sure the objects haven't moved
3235 around (this is important!). Additions are OK, as well as
3236 consistent name changes.
3239 *) Add BSD-style MD5-based passwords to 'openssl passwd' (option '-1').
3242 *) Addition of the command line parameter '-rand file' to 'openssl req'.
3243 The given file adds to whatever has already been seeded into the
3244 random pool through the RANDFILE configuration file option or
3245 environment variable, or the default random state file.
3248 *) mkstack.pl now sorts each macro group into lexical order.
3249 Previously the output order depended on the order the files
3250 appeared in the directory, resulting in needless rewriting
3254 *) Patches to make OpenSSL compile under Win32 again. Mostly
3255 work arounds for the VC++ problem that it treats func() as
3256 func(void). Also stripped out the parts of mkdef.pl that
3257 added extra typesafe functions: these no longer exist.
3260 *) Reorganisation of the stack code. The macros are now all
3261 collected in safestack.h . Each macro is defined in terms of
3262 a "stack macro" of the form SKM_<name>(type, a, b). The
3263 DEBUG_SAFESTACK is now handled in terms of function casts,
3264 this has the advantage of retaining type safety without the
3265 use of additional functions. If DEBUG_SAFESTACK is not defined
3266 then the non typesafe macros are used instead. Also modified the
3267 mkstack.pl script to handle the new form. Needs testing to see
3268 if which (if any) compilers it chokes and maybe make DEBUG_SAFESTACK
3269 the default if no major problems. Similar behaviour for ASN1_SET_OF
3270 and PKCS12_STACK_OF.
3273 *) When some versions of IIS use the 'NET' form of private key the
3274 key derivation algorithm is different. Normally MD5(password) is
3275 used as a 128 bit RC4 key. In the modified case
3276 MD5(MD5(password) + "SGCKEYSALT") is used insted. Added some
3277 new functions i2d_RSA_NET(), d2i_RSA_NET() etc which are the same
3278 as the old Netscape_RSA functions except they have an additional
3279 'sgckey' parameter which uses the modified algorithm. Also added
3280 an -sgckey command line option to the rsa utility. Thanks to
3281 Adrian Peck <bertie@ncipher.com> for posting details of the modified
3282 algorithm to openssl-dev.
3285 *) The evp_local.h macros were using 'c.##kname' which resulted in
3286 invalid expansion on some systems (SCO 5.0.5 for example).
3287 Corrected to 'c.kname'.
3288 [Phillip Porch <root@theporch.com>]
3290 *) New X509_get1_email() and X509_REQ_get1_email() functions that return
3291 a STACK of email addresses from a certificate or request, these look
3292 in the subject name and the subject alternative name extensions and
3293 omit any duplicate addresses.
3296 *) Re-implement BN_mod_exp2_mont using independent (and larger) windows.
3297 This makes DSA verification about 2 % faster.
3300 *) Increase maximum window size in BN_mod_exp_... to 6 bits instead of 5
3301 (meaning that now 2^5 values will be precomputed, which is only 4 KB
3302 plus overhead for 1024 bit moduli).
3303 This makes exponentiations about 0.5 % faster for 1024 bit
3304 exponents (as measured by "openssl speed rsa2048").
3307 *) Rename memory handling macros to avoid conflicts with other
3309 Malloc => OPENSSL_malloc
3310 Malloc_locked => OPENSSL_malloc_locked
3311 Realloc => OPENSSL_realloc
3312 Free => OPENSSL_free
3315 *) New function BN_mod_exp_mont_word for small bases (roughly 15%
3316 faster than BN_mod_exp_mont, i.e. 7% for a full DH exchange).
3319 *) CygWin32 support.
3320 [John Jarvie <jjarvie@newsguy.com>]
3322 *) The type-safe stack code has been rejigged. It is now only compiled
3323 in when OpenSSL is configured with the DEBUG_SAFESTACK option and
3324 by default all type-specific stack functions are "#define"d back to
3325 standard stack functions. This results in more streamlined output
3326 but retains the type-safety checking possibilities of the original
3330 *) The STACK code has been cleaned up, and certain type declarations
3331 that didn't make a lot of sense have been brought in line. This has
3332 also involved a cleanup of sorts in safestack.h to more correctly
3333 map type-safe stack functions onto their plain stack counterparts.
3334 This work has also resulted in a variety of "const"ifications of
3335 lots of the code, especially "_cmp" operations which should normally
3336 be prototyped with "const" parameters anyway.
3339 *) When generating bytes for the first time in md_rand.c, 'stir the pool'
3340 by seeding with STATE_SIZE dummy bytes (with zero entropy count).
3341 (The PRNG state consists of two parts, the large pool 'state' and 'md',
3342 where all of 'md' is used each time the PRNG is used, but 'state'
3343 is used only indexed by a cyclic counter. As entropy may not be
3344 well distributed from the beginning, 'md' is important as a
3345 chaining variable. However, the output function chains only half
3346 of 'md', i.e. 80 bits. ssleay_rand_add, on the other hand, chains
3347 all of 'md', and seeding with STATE_SIZE dummy bytes will result
3348 in all of 'state' being rewritten, with the new values depending
3349 on virtually all of 'md'. This overcomes the 80 bit limitation.)
3352 *) In ssl/s2_clnt.c and ssl/s3_clnt.c, call ERR_clear_error() when
3353 the handshake is continued after ssl_verify_cert_chain();
3354 otherwise, if SSL_VERIFY_NONE is set, remaining error codes
3355 can lead to 'unexplainable' connection aborts later.
3356 [Bodo Moeller; problem tracked down by Lutz Jaenicke]
3358 *) Major EVP API cipher revision.
3359 Add hooks for extra EVP features. This allows various cipher
3360 parameters to be set in the EVP interface. Support added for variable
3361 key length ciphers via the EVP_CIPHER_CTX_set_key_length() function and
3362 setting of RC2 and RC5 parameters.
3364 Modify EVP_OpenInit() and EVP_SealInit() to cope with variable key length
3367 Remove lots of duplicated code from the EVP library. For example *every*
3368 cipher init() function handles the 'iv' in the same way according to the
3369 cipher mode. They also all do nothing if the 'key' parameter is NULL and
3370 for CFB and OFB modes they zero ctx->num.
3372 New functionality allows removal of S/MIME code RC2 hack.
3374 Most of the routines have the same form and so can be declared in terms
3377 By shifting this to the top level EVP_CipherInit() it can be removed from
3378 all individual ciphers. If the cipher wants to handle IVs or keys
3379 differently it can set the EVP_CIPH_CUSTOM_IV or EVP_CIPH_ALWAYS_CALL_INIT
3382 Change lots of functions like EVP_EncryptUpdate() to now return a
3383 value: although software versions of the algorithms cannot fail
3384 any installed hardware versions can.
3387 *) Implement SSL_OP_TLS_ROLLBACK_BUG: In ssl3_get_client_key_exchange, if
3388 this option is set, tolerate broken clients that send the negotiated
3389 protocol version number instead of the requested protocol version
3393 *) Call dh_tmp_cb (set by ..._TMP_DH_CB) with correct 'is_export' flag;
3394 i.e. non-zero for export ciphersuites, zero otherwise.
3395 Previous versions had this flag inverted, inconsistent with
3396 rsa_tmp_cb (..._TMP_RSA_CB).
3397 [Bodo Moeller; problem reported by Amit Chopra]
3399 *) Add missing DSA library text string. Work around for some IIS
3400 key files with invalid SEQUENCE encoding.
3403 *) Add a document (doc/standards.txt) that list all kinds of standards
3404 and so on that are implemented in OpenSSL.
3407 *) Enhance c_rehash script. Old version would mishandle certificates
3408 with the same subject name hash and wouldn't handle CRLs at all.
3409 Added -fingerprint option to crl utility, to support new c_rehash
3413 *) Eliminate non-ANSI declarations in crypto.h and stack.h.
3416 *) Fix for SSL server purpose checking. Server checking was
3417 rejecting certificates which had extended key usage present
3418 but no ssl client purpose.
3419 [Steve Henson, reported by Rene Grosser <grosser@hisolutions.com>]
3421 *) Make PKCS#12 code work with no password. The PKCS#12 spec
3422 is a little unclear about how a blank password is handled.
3423 Since the password in encoded as a BMPString with terminating
3424 double NULL a zero length password would end up as just the
3425 double NULL. However no password at all is different and is
3426 handled differently in the PKCS#12 key generation code. NS
3427 treats a blank password as zero length. MSIE treats it as no
3428 password on export: but it will try both on import. We now do
3429 the same: PKCS12_parse() tries zero length and no password if
3430 the password is set to "" or NULL (NULL is now a valid password:
3431 it wasn't before) as does the pkcs12 application.
3434 *) Bugfixes in apps/x509.c: Avoid a memory leak; and don't use
3435 perror when PEM_read_bio_X509_REQ fails, the error message must
3436 be obtained from the error queue.
3439 *) Avoid 'thread_hash' memory leak in crypto/err/err.c by freeing
3440 it in ERR_remove_state if appropriate, and change ERR_get_state
3441 accordingly to avoid race conditions (this is necessary because
3442 thread_hash is no longer constant once set).
3445 *) Bugfix for linux-elf makefile.one.
3448 *) RSA_get_default_method() will now cause a default
3449 RSA_METHOD to be chosen if one doesn't exist already.
3450 Previously this was only set during a call to RSA_new()
3451 or RSA_new_method(NULL) meaning it was possible for
3452 RSA_get_default_method() to return NULL.
3455 *) Added native name translation to the existing DSO code
3456 that will convert (if the flag to do so is set) filenames
3457 that are sufficiently small and have no path information
3458 into a canonical native form. Eg. "blah" converted to
3459 "libblah.so" or "blah.dll" etc.
3462 *) New function ERR_error_string_n(e, buf, len) which is like
3463 ERR_error_string(e, buf), but writes at most 'len' bytes
3464 including the 0 terminator. For ERR_error_string_n, 'buf'
3466 [Damien Miller <djm@mindrot.org>, Bodo Moeller]
3468 *) CONF library reworked to become more general. A new CONF
3469 configuration file reader "class" is implemented as well as a
3470 new functions (NCONF_*, for "New CONF") to handle it. The now
3471 old CONF_* functions are still there, but are reimplemented to
3472 work in terms of the new functions. Also, a set of functions
3473 to handle the internal storage of the configuration data is
3474 provided to make it easier to write new configuration file
3475 reader "classes" (I can definitely see something reading a
3476 configuration file in XML format, for example), called _CONF_*,
3477 or "the configuration storage API"...
3479 The new configuration file reading functions are:
3481 NCONF_new, NCONF_free, NCONF_load, NCONF_load_fp, NCONF_load_bio,
3482 NCONF_get_section, NCONF_get_string, NCONF_get_numbre
3484 NCONF_default, NCONF_WIN32
3486 NCONF_dump_fp, NCONF_dump_bio
3488 NCONF_default and NCONF_WIN32 are method (or "class") choosers,
3489 NCONF_new creates a new CONF object. This works in the same way
3490 as other interfaces in OpenSSL, like the BIO interface.
3491 NCONF_dump_* dump the internal storage of the configuration file,
3492 which is useful for debugging. All other functions take the same
3493 arguments as the old CONF_* functions wth the exception of the
3494 first that must be a `CONF *' instead of a `LHASH *'.
3496 To make it easer to use the new classes with the old CONF_* functions,
3497 the function CONF_set_default_method is provided.
3500 *) Add '-tls1' option to 'openssl ciphers', which was already
3501 mentioned in the documentation but had not been implemented.
3502 (This option is not yet really useful because even the additional
3503 experimental TLS 1.0 ciphers are currently treated as SSL 3.0 ciphers.)
3506 *) Initial DSO code added into libcrypto for letting OpenSSL (and
3507 OpenSSL-based applications) load shared libraries and bind to
3508 them in a portable way.
3509 [Geoff Thorpe, with contributions from Richard Levitte]
3511 Changes between 0.9.5 and 0.9.5a [1 Apr 2000]
3513 *) Make sure _lrotl and _lrotr are only used with MSVC.
3515 *) Use lock CRYPTO_LOCK_RAND correctly in ssleay_rand_status
3516 (the default implementation of RAND_status).
3518 *) Rename openssl x509 option '-crlext', which was added in 0.9.5,
3519 to '-clrext' (= clear extensions), as intended and documented.
3520 [Bodo Moeller; inconsistency pointed out by Michael Attili
3523 *) Fix for HMAC. It wasn't zeroing the rest of the block if the key length
3524 was larger than the MD block size.
3525 [Steve Henson, pointed out by Yost William <YostW@tce.com>]
3527 *) Modernise PKCS12_parse() so it uses STACK_OF(X509) for its ca argument
3528 fix a leak when the ca argument was passed as NULL. Stop X509_PUBKEY_set()
3529 using the passed key: if the passed key was a private key the result
3530 of X509_print(), for example, would be to print out all the private key
3534 *) des_quad_cksum() byte order bug fix.
3535 [Ulf Möller, using the problem description in krb4-0.9.7, where
3536 the solution is attributed to Derrick J Brashear <shadow@DEMENTIA.ORG>]
3538 *) Fix so V_ASN1_APP_CHOOSE works again: however its use is strongly
3540 [Steve Henson, pointed out by Brian Korver <briank@cs.stanford.edu>]
3542 *) For easily testing in shell scripts whether some command
3543 'openssl XXX' exists, the new pseudo-command 'openssl no-XXX'
3544 returns with exit code 0 iff no command of the given name is available.
3545 'no-XXX' is printed in this case, 'XXX' otherwise. In both cases,
3546 the output goes to stdout and nothing is printed to stderr.
3547 Additional arguments are always ignored.
3549 Since for each cipher there is a command of the same name,
3550 the 'no-cipher' compilation switches can be tested this way.
3552 ('openssl no-XXX' is not able to detect pseudo-commands such
3553 as 'quit', 'list-XXX-commands', or 'no-XXX' itself.)
3556 *) Update test suite so that 'make test' succeeds in 'no-rsa' configuration.
3559 *) For SSL_[CTX_]set_tmp_dh, don't create a DH key if SSL_OP_SINGLE_DH_USE
3560 is set; it will be thrown away anyway because each handshake creates
3562 ssl_cert_dup, which is used by SSL_new, now copies DH keys in addition
3563 to parameters -- in previous versions (since OpenSSL 0.9.3) the
3564 'default key' from SSL_CTX_set_tmp_dh would always be lost, meanining
3565 you effectivly got SSL_OP_SINGLE_DH_USE when using this macro.
3568 *) New s_client option -ign_eof: EOF at stdin is ignored, and
3569 'Q' and 'R' lose their special meanings (quit/renegotiate).
3570 This is part of what -quiet does; unlike -quiet, -ign_eof
3571 does not suppress any output.
3574 *) Add compatibility options to the purpose and trust code. The
3575 purpose X509_PURPOSE_ANY is "any purpose" which automatically
3576 accepts a certificate or CA, this was the previous behaviour,
3577 with all the associated security issues.
3579 X509_TRUST_COMPAT is the old trust behaviour: only and
3580 automatically trust self signed roots in certificate store. A
3581 new trust setting X509_TRUST_DEFAULT is used to specify that
3582 a purpose has no associated trust setting and it should instead
3583 use the value in the default purpose.
3586 *) Fix the PKCS#8 DSA private key code so it decodes keys again
3587 and fix a memory leak.
3590 *) In util/mkerr.pl (which implements 'make errors'), preserve
3591 reason strings from the previous version of the .c file, as
projects / openssl.git / blob