5 Changes between 0.9.6 and 0.9.7 [xx XXX 2001]
7 Both OpenSSL 0.9.6a (bugfix release, 5 Apr 2001) and OpenSSL 0.9.7
8 are based on OpenSSL 0.9.6.
9 Change log entries are tagged as follows:
10 -) applies to 0.9.6a (/0.9.6b) only
11 *) applies to 0.9.6a (/0.9.6b) and 0.9.7
12 +) applies to 0.9.7 only
14 +) Fix a memory leak in 'sk_dup()' in the case reallocation fails. (Also
15 tidy up some unecessarily weird code in 'sk_new()').
16 [Geoff, reported by Diego Tartara <dtartara@novamens.com>]
18 +) Change the key loading routines for ENGINEs to use the same kind
19 callback (pem_password_cb) as all other routines that need this
23 *) Fix for compatibility mode trust settings: ignore trust settings
24 unless some valid trust or reject settings are present.
27 *) Fix for blowfish EVP: its a variable length cipher.
30 +) Increase ENTROPY_NEEDED to 32 bytes, as Rijndael can operate with
31 256 bit (=32 byte) keys. Of course seeding with more entropy bytes
32 than this minimum value is recommended.
35 +) New random seeder for OpenVMS, using the system process statistics
36 that are easily reachable.
39 +) Windows apparently can't transparently handle global
40 variables defined in DLLs. Initialisations such as:
42 const ASN1_ITEM *it = &ASN1_INTEGER_it;
44 wont compile. This is used by the any applications that need to
45 delcare their own ASN1 modules. This was fixed by adding the option
46 EXPORT_VAR_AS_FN to all Win32 platforms, although this isn't strictly
47 needed for static libraries under Win32.
50 +) New functions X509_PURPOSE_set() and X509_TRUST_set() to handle
51 setting of purpose and trust fields. New X509_STORE trust and
52 purpose functions and tidy up setting in other SSL functions.
55 +) Add copies of X509_STORE_CTX fields and callbacks to X509_STORE
56 structure. These are inherited by X509_STORE_CTX when it is
57 initialised. This allows various defaults to be set in the
58 X509_STORE structure (such as flags for CRL checking and custom
59 purpose or trust settings) for functions which only use X509_STORE_CTX
60 internally such as S/MIME.
62 Modify X509_STORE_CTX_purpose_inherit() so it only sets purposes and
63 trust settings if they are not set in X509_STORE. This allows X509_STORE
64 purposes and trust (in S/MIME for example) to override any set by default.
66 Add command line options for CRL checking to smime, s_client and s_server
70 +) Initial CRL based revocation checking. If the CRL checking flag(s)
71 are set then the CRL is looked up in the X509_STORE structure and
72 its validity and signature checked, then if the certificate is found
73 in the CRL the verify fails with a revoked error.
75 Various new CRL related callbacks added to X509_STORE_CTX structure.
77 Command line options added to 'verify' application to support this.
79 This needs some additional work, such as being able to handle multiple
80 CRLs with different times, extension based lookup (rather than just
81 by subject name) and ultimately more complete V2 CRL extension
85 +) Add a general user interface API. This is designed to replace things
86 like des_read_password and friends (backward compatibility functions
87 using this new API are provided). The purpose is to remove prompting
88 functions from the DES code section as well as provide for prompting
89 through dialog boxes in a window system and the like.
92 *) In versions up to 0.9.6, RAND_file_name() resorted to file ".rnd"
93 in the current directory if neither $RANDFILE nor $HOME was set.
94 RAND_file_name() in 0.9.6a returned NULL in this case. This has
95 caused some confusion to Windows users who haven't defined $HOME.
96 Thus RAND_file_name() is changed again: e_os.h can define a
97 DEFAULT_HOME, which will be used if $HOME is not set.
98 For Windows, we use "C:"; on other platforms, we still require
99 environment variables.
101 +) Add "ex_data" support to ENGINE so implementations can add state at a
102 per-structure level rather than having to store it globally.
105 +) Make it possible for ENGINE structures to be copied when retrieved by
106 ENGINE_by_id() if the ENGINE specifies a new flag: ENGINE_FLAGS_BY_ID_COPY.
107 This causes the "original" ENGINE structure to act like a template,
108 analogous to the RSA vs. RSA_METHOD type of separation. Because of this
109 operational state can be localised to each ENGINE structure, despite the
110 fact they all share the same "methods". New ENGINE structures returned in
111 this case have no functional references and the return value is the single
112 structural reference. This matches the single structural reference returned
113 by ENGINE_by_id() normally, when it is incremented on the pre-existing
117 *) Fix various bugs related to DSA S/MIME verification. Handle missing
118 parameters in DSA public key structures and return an error in the
119 DSA routines if parameters are absent.
122 +) Fix ASN1 decoder when decoding type ANY and V_ASN1_OTHER: since this
123 needs to match any other type at all we need to manually clear the
127 +) Changes to the "openssl engine" utility to include;
128 - verbosity levels ('-v', '-vv', and '-vvv') that provide information
129 about an ENGINE's available control commands.
130 - executing control commands from command line arguments using the
131 '-pre' and '-post' switches. '-post' is only used if '-t' is
132 specified and the ENGINE is successfully initialised. The syntax for
133 the individual commands are colon-separated, for example;
134 openssl engine chil -pre FORK_CHECK:0 -pre SO_PATH:/lib/test.so
137 +) New dynamic control command support for ENGINEs. ENGINEs can now
138 declare their own commands (numbers), names (strings), descriptions,
139 and input types for run-time discovery by calling applications. A
140 subset of these commands are implicitly classed as "executable"
141 depending on their input type, and only these can be invoked through
142 the new string-based API function ENGINE_ctrl_cmd_string(). (Eg. this
143 can be based on user input, config files, etc). The distinction is
144 that "executable" commands cannot return anything other than a boolean
145 result and can only support numeric or string input, whereas some
146 discoverable commands may only be for direct use through
147 ENGINE_ctrl(), eg. supporting the exchange of binary data, function
148 pointers, or other custom uses. The "executable" commands are to
149 support parameterisations of ENGINE behaviour that can be
150 unambiguously defined by ENGINEs and used consistently across any
151 OpenSSL-based application. Commands have been added to all the
152 existing hardware-supporting ENGINEs, noticeably "SO_PATH" to allow
153 control over shared-library paths without source code alterations.
156 +) Changed all ENGINE implementations to dynamically allocate their
157 ENGINEs rather than declaring them statically. Apart from this being
158 necessary with the removal of the ENGINE_FLAGS_MALLOCED distinction,
159 this also allows the implementations to compile without using the
160 internal engine_int.h header.
163 +) Minor adjustment to "rand" code. RAND_get_rand_method() now returns a
164 'const' value. Any code that should be able to modify a RAND_METHOD
165 should already have non-const pointers to it (ie. they should only
166 modify their own ones).
169 +) Made a variety of little tweaks to the ENGINE code.
170 - "atalla" and "ubsec" string definitions were moved from header files
171 to C code. "nuron" string definitions were placed in variables
172 rather than hard-coded - allowing parameterisation of these values
173 later on via ctrl() commands.
174 - Removed unused "#if 0"'d code.
175 - Fixed engine list iteration code so it uses ENGINE_free() to release
176 structural references.
177 - Constified the RAND_METHOD element of ENGINE structures.
178 - Constified various get/set functions as appropriate and added
179 missing functions (including a catch-all ENGINE_cpy that duplicates
180 all ENGINE values onto a new ENGINE except reference counts/state).
181 - Removed NULL parameter checks in get/set functions. Setting a method
182 or function to NULL is a way of cancelling out a previously set
183 value. Passing a NULL ENGINE parameter is just plain stupid anyway
184 and doesn't justify the extra error symbols and code.
185 - Deprecate the ENGINE_FLAGS_MALLOCED define and move the area for
186 flags from engine_int.h to engine.h.
187 - Changed prototypes for ENGINE handler functions (init(), finish(),
188 ctrl(), key-load functions, etc) to take an (ENGINE*) parameter.
191 *) Move 'if (!initialized) RAND_poll()' into regions protected by
192 CRYPTO_LOCK_RAND. This is not strictly necessary, but avoids
193 having multiple threads call RAND_poll() concurrently.
196 *) In crypto/rand/md_rand.c, replace 'add_do_not_lock' flag by a
197 combination of a flag and a thread ID variable.
198 Otherwise while one thread is in ssleay_rand_bytes (which sets the
199 flag), *other* threads can enter ssleay_add_bytes without obeying
200 the CRYPTO_LOCK_RAND lock (and may even illegaly release the lock
201 that they do not hold after the first thread unsets add_do_not_lock).
204 +) Implement binary inversion algorithm for BN_mod_inverse in addition
205 to the algorithm using long divison. The binary algorithm can be
206 used only if the modulus is odd. On 32-bit systems, it is faster
207 only for relatively small moduli (roughly 20-30% for 128-bit moduli,
208 roughly 5-15% for 256-bit moduli), so we use it only for moduli
209 up to 450 bits. In 64-bit environments, the binary algorithm
210 appears to be advantageous for much longer moduli; here we use it
211 for moduli up to 2048 bits.
214 *) Change bctest again: '-x' expressions are not available in all
218 -) OpenSSL 0.9.6a released [5 Apr 2001]
220 *) Fix a couple of memory leaks in PKCS7_dataDecode()
221 [Steve Henson, reported by Heyun Zheng <hzheng@atdsprint.com>]
223 *) Change Configure and Makefiles to provide EXE_EXT, which will contain
224 the default extension for executables, if any. Also, make the perl
225 scripts that use symlink() to test if it really exists and use "cp"
226 if it doesn't. All this made OpenSSL compilable and installable in
230 +) Rewrite CHOICE field setting in ASN1_item_ex_d2i(). The old code
231 could not support the combine flag in choice fields.
234 -) Fix for asn1_GetSequence() for indefinite length constructed data.
235 If SEQUENCE is length is indefinite just set c->slen to the total
236 amount of data available.
237 [Steve Henson, reported by shige@FreeBSD.org]
238 [This change does not apply to 0.9.7.]
240 *) Change bctest to avoid here-documents inside command substitution
241 (workaround for FreeBSD /bin/sh bug).
242 For compatibility with Ultrix, avoid shell functions (introduced
243 in the bctest version that searches along $PATH).
246 *) Rename 'des_encrypt' to 'des_encrypt1'. This avoids the clashes
247 with des_encrypt() defined on some operating systems, like Solaris
251 *) Check the result of RSA-CRT (see D. Boneh, R. DeMillo, R. Lipton:
252 On the Importance of Eliminating Errors in Cryptographic
253 Computations, J. Cryptology 14 (2001) 2, 101-119,
254 http://theory.stanford.edu/~dabo/papers/faults.ps.gz).
257 *) MIPS assembler BIGNUM division bug fix.
260 *) Disabled incorrect Alpha assembler code.
263 -) Fix PKCS#7 decode routines so they correctly update the length
264 after reading an EOC for the EXPLICIT tag.
266 [This change does not apply to 0.9.7.]
268 *) Fix bug in PKCS#12 key generation routines. This was triggered
269 if a 3DES key was generated with a 0 initial byte. Include
270 PKCS12_BROKEN_KEYGEN compilation option to retain the old
271 (but broken) behaviour.
274 *) Enhance bctest to search for a working bc along $PATH and print
276 [Tim Rice <tim@multitalents.net> via Richard Levitte]
278 +) Add a 'copy_extensions' option to the 'ca' utility. This copies
279 extensions from a certificate request to the certificate.
282 +) Allow multiple 'certopt' and 'nameopt' options to be separated
283 by commas. Add 'namopt' and 'certopt' options to the 'ca' config
284 file: this allows the display of the certificate about to be
285 signed to be customised, to allow certain fields to be included
286 or excluded and extension details. The old system didn't display
287 multicharacter strings properly, omitted fields not in the policy
288 and couldn't display additional details such as extensions.
291 *) Fix memory leaks in err.c: free err_data string if necessary;
292 don't write to the wrong index in ERR_set_error_data.
295 +) Function EC_POINTs_mul for simultaneous scalar multiplication
296 of an arbitrary number of elliptic curve points, optionally
297 including the generator defined for the EC_GROUP.
298 EC_POINT_mul is a simple wrapper function for the typical case
299 that the point list has just one item (besides the optional
303 +) First EC_METHODs for curves over GF(p):
305 EC_GFp_simple_method() uses the basic BN_mod_mul and BN_mod_sqr
306 operations and provides various method functions that can also
307 operate with faster implementations of modular arithmetic.
309 EC_GFp_mont_method() reuses most functions that are part of
310 EC_GFp_simple_method, but uses Montgomery arithmetic.
312 [Bodo Moeller; point addition and point doubling
313 implementation directly derived from source code provided by
314 Lenka Fibikova <fibikova@exp-math.uni-essen.de>]
316 +) Framework for elliptic curves (crypto/ec/ec.h, crypto/ec/ec_lcl.h,
319 Curves are EC_GROUP objects (with an optional group generator)
320 based on EC_METHODs that are built into the library.
322 Points are EC_POINT objects based on EC_GROUP objects.
324 Most of the framework would be able to handle curves over arbitrary
325 finite fields, but as there are no obvious types for fields other
326 than GF(p), some functions are limited to that for now.
329 +) Add the -HTTP option to s_server. It is similar to -WWW, but requires
330 that the file contains a complete HTTP response.
333 +) Add the ec directory to mkdef.pl and mkfiles.pl. In mkdef.pl
334 change the def and num file printf format specifier from "%-40sXXX"
335 to "%-39s XXX". The latter will always guarantee a space after the
336 field while the former will cause them to run together if the field
337 is 40 of more characters long.
340 +) Constify the cipher and digest 'method' functions and structures
341 and modify related functions to take constant EVP_MD and EVP_CIPHER
345 *) Implement ssl23_peek (analogous to ssl23_read), which previously
349 *) Replace rdtsc with _emit statements for VC++ version 5.
350 [Jeremy Cooper <jeremy@baymoo.org>]
352 +) Hide BN_CTX structure details in bn_lcl.h instead of publishing them
353 in <openssl/bn.h>. Also further increase BN_CTX_NUM to 32.
356 +) Modify EVP_Digest*() routines so they now return values. Although the
357 internal software routines can never fail additional hardware versions
361 +) Clean up crypto/err/err.h and change some error codes to avoid conflicts:
363 Previously ERR_R_FATAL was too small and coincided with ERR_LIB_PKCS7
364 (= ERR_R_PKCS7_LIB); it is now 64 instead of 32.
367 ERR_R_NESTED_ASN1_ERROR
369 ERR_R_MISSING_ASN1_EOS
370 were 4 .. 9, conflicting with
371 ERR_LIB_RSA (= ERR_R_RSA_LIB)
373 ERR_LIB_PEM (= ERR_R_PEM_LIB).
374 They are now 58 .. 63 (i.e., just below ERR_R_FATAL).
376 Add new error code 'ERR_R_INTERNAL_ERROR'.
379 +) Don't overuse locks in crypto/err/err.c: For data retrieval, CRYPTO_r_lock
383 +) New option '-subj arg' for 'openssl req' and 'openssl ca'. This
384 sets the subject name for a new request or supersedes the
385 subject name in a given request. Formats that can be parsed are
386 'CN=Some Name, OU=myOU, C=IT'
388 'CN=Some Name/OU=myOU/C=IT'.
390 Add options '-batch' and '-verbose' to 'openssl req'.
391 [Massimiliano Pala <madwolf@hackmasters.net>]
393 *) Make it possible to reuse SSLv2 sessions.
396 +) Introduce the possibility to access global variables through
397 functions on platform were that's the best way to handle exporting
398 global variables in shared libraries. To enable this functionality,
399 one must configure with "EXPORT_VAR_AS_FN" or defined the C macro
400 "OPENSSL_EXPORT_VAR_AS_FUNCTION" in crypto/opensslconf.h (the latter
401 is normally done by Configure or something similar).
403 To implement a global variable, use the macro OPENSSL_IMPLEMENT_GLOBAL
404 in the source file (foo.c) like this:
406 OPENSSL_IMPLEMENT_GLOBAL(int,foo)=1;
407 OPENSSL_IMPLEMENT_GLOBAL(double,bar);
409 To declare a global variable, use the macros OPENSSL_DECLARE_GLOBAL
410 and OPENSSL_GLOBAL_REF in the header file (foo.h) like this:
412 OPENSSL_DECLARE_GLOBAL(int,foo);
413 #define foo OPENSSL_GLOBAL_REF(foo)
414 OPENSSL_DECLARE_GLOBAL(double,bar);
415 #define bar OPENSSL_GLOBAL_REF(bar)
417 The #defines are very important, and therefore so is including the
418 header file everywere where the defined globals are used.
420 The macro OPENSSL_EXPORT_VAR_AS_FUNCTION also affects the definition
421 of ASN.1 items, but that structure is a bt different.
423 The largest change is in util/mkdef.pl which has been enhanced with
424 better and easier to understand logic to choose which symbols should
425 go into the Windows .def files as well as a number of fixes and code
426 cleanup (among others, algorithm keywords are now sorted
427 lexicographically to avoid constant rewrites).
430 *) In copy_email() check for >= 0 as a return value for
431 X509_NAME_get_index_by_NID() since 0 is a valid index.
432 [Steve Henson reported by Massimiliano Pala <madwolf@opensca.org>]
434 +) In BN_div() keep a copy of the sign of 'num' before writing the
435 result to 'rm' because if rm==num the value will be overwritten
436 and produce the wrong result if 'num' is negative: this caused
437 problems with BN_mod() and BN_nnmod().
440 +) Function OCSP_request_verify(). This checks the signature on an
441 OCSP request and verifies the signer certificate. The signer
442 certificate is just checked for a generic purpose and OCSP request
446 +) Add OCSP_check_validity() function to check the validity of OCSP
447 responses. OCSP responses are prepared in real time and may only
448 be a few seconds old. Simply checking that the current time lies
449 between thisUpdate and nextUpdate max reject otherwise valid responses
450 caused by either OCSP responder or client clock innacuracy. Instead
451 we allow thisUpdate and nextUpdate to fall within a certain period of
452 the current time. The age of the response can also optionally be
453 checked. Two new options -validity_period and -status_age added to
457 +) If signature or public key algorithm is unrecognized print out its
458 OID rather that just UNKOWN.
461 *) Avoid coredump with unsupported or invalid public keys by checking if
462 X509_get_pubkey() fails in PKCS7_verify(). Fix memory leak when
463 PKCS7_verify() fails with non detached data.
466 +) Change OCSP_cert_to_id() to tolerate a NULL subject certificate and
467 OCSP_cert_id_new() a NULL serialNumber. This allows a partial certificate
468 ID to be generated from the issuer certificate alone which can then be
469 passed to OCSP_id_issuer_cmp().
472 +) New compilation option ASN1_ITEM_FUNCTIONS. This causes the new
473 ASN1 modules to export functions returning ASN1_ITEM pointers
474 instead of the ASN1_ITEM structures themselves. This adds several
475 new macros which allow the underlying ASN1 function/structure to
476 be accessed transparently. As a result code should not use ASN1_ITEM
477 references directly (such as &X509_it) but instead use the relevant
478 macros (such as ASN1_ITEM_rptr(X509)). This option is to allow
479 use of the new ASN1 code on platforms where exporting structures
480 is problematical (for example in shared libraries) but exporting
481 functions returning pointers to structures is not.
484 +) Add support for overriding the generation of SSL/TLS session IDs.
485 These callbacks can be registered either in an SSL_CTX or per SSL.
486 The purpose of this is to allow applications to control, if they wish,
487 the arbitrary values chosen for use as session IDs, particularly as it
488 can be useful for session caching in multiple-server environments. A
489 command-line switch for testing this (and any client code that wishes
490 to use such a feature) has been added to "s_server".
491 [Geoff Thorpe, Lutz Jaenicke]
493 +) Modify mkdef.pl to recognise and parse prprocessor conditionals
494 of the form '#if defined(...) || defined(...) || ...' and
495 '#if !defined(...) && !defined(...) && ...'. This also avoids
496 the growing number of special cases it was previously handling.
499 *) Don't use getenv in library functions when run as setuid/setgid.
500 New function OPENSSL_issetugid().
503 +) Make all configuration macros available for application by making
504 sure they are available in opensslconf.h, by giving them names starting
505 with "OPENSSL_" to avoid conflicts with other packages and by making
506 sure e_os2.h will cover all platform-specific cases together with
508 Additionally, it is now possible to define configuration/platform-
509 specific names (called "system identities"). In the C code, these
510 are prefixed with "OPENSSL_SYSNAME_". e_os2.h will create another
511 macro with the name beginning with "OPENSSL_SYS_", which is determined
512 from "OPENSSL_SYSNAME_*" or compiler-specific macros depending on
516 +) New option -set_serial to 'req' and 'x509' this allows the serial
517 number to use to be specified on the command line. Previously self
518 signed certificates were hard coded with serial number 0 and the
519 CA options of 'x509' had to use a serial number in a file which was
523 *) Avoid false positives in memory leak detection code (crypto/mem_dbg.c)
524 due to incorrect handling of multi-threading:
526 1. Fix timing glitch in the MemCheck_off() portion of CRYPTO_mem_ctrl().
528 2. Fix logical glitch in is_MemCheck_on() aka CRYPTO_is_mem_check_on().
530 3. Count how many times MemCheck_off() has been called so that
531 nested use can be treated correctly. This also avoids
532 inband-signalling in the previous code (which relied on the
533 assumption that thread ID 0 is impossible).
536 +) New options to 'ca' utility to support V2 CRL entry extensions.
537 Currently CRL reason, invalidity date and hold instruction are
538 supported. Add new CRL extensions to V3 code and some new objects.
541 *) Add "-rand" option also to s_client and s_server.
544 +) New function EVP_CIPHER_CTX_set_padding() this is used to
545 disable standard block padding (aka PKCS#5 padding) in the EVP
546 API, which was previously mandatory. This means that the data is
547 not padded in any way and so the total length much be a multiple
548 of the block size, otherwise an error occurs.
551 +) Initial (incomplete) OCSP SSL support.
554 *) Fix CPU detection on Irix 6.x.
555 [Kurt Hockenbury <khockenb@stevens-tech.edu> and
556 "Bruce W. Forsberg" <bruce.forsberg@baesystems.com>]
558 +) New function OCSP_parse_url(). This splits up a URL into its host,
559 port and path components: primarily to parse OCSP URLs. New -url
560 option to ocsp utility.
563 +) New nonce behavior. The return value of OCSP_check_nonce() now
564 reflects the various checks performed. Applications can decide
565 whether to tolerate certain situations such as an absent nonce
566 in a response when one was present in a request: the ocsp application
567 just prints out a warning. New function OCSP_add1_basic_nonce()
568 this is to allow responders to include a nonce in a response even if
569 the request is nonce-less.
572 *) Fix X509_NAME bug which produced incorrect encoding if X509_NAME
574 [This change does not apply to 0.9.7.]
576 *) Use the cached encoding of an X509_NAME structure rather than
577 copying it. This is apparently the reason for the libsafe "errors"
578 but the code is actually correct.
581 +) Disable stdin buffering in load_cert (apps/apps.c) so that no certs are
582 skipped when using openssl x509 multiple times on a single input file,
583 e.g. "(openssl x509 -out cert1; openssl x509 -out cert2) <certs".
586 +) Make ASN1_UTCTIME_set_string() and ASN1_GENERALIZEDTIME_set_string()
587 set string type: to handle setting ASN1_TIME structures. Fix ca
588 utility to correctly initialize revocation date of CRLs.
591 +) New option SSL_OP_CIPHER_SERVER_PREFERENCE allows the server to override
592 the clients preferred ciphersuites and rather use its own preferences.
593 Should help to work around M$ SGC (Server Gated Cryptography) bug in
594 Internet Explorer by ensuring unchanged hash method during stepup.
597 +) Make mkdef.pl recognise all DECLARE_ASN1 macros, change rijndael
598 to aes and add a new 'exist' option to print out symbols that don't
602 +) Additional options to ocsp utility to allow flags to be set and
603 additional certificates supplied.
606 +) Add the option -VAfile to 'openssl ocsp', so the user can give the
607 OCSP client a number of certificate to only verify the response
611 *) Add new function BN_rand_range(), and fix DSA_sign_setup() to prevent
612 Bleichenbacher's DSA attack.
613 Extend BN_[pseudo_]rand: As before, top=1 forces the highest two bits
614 to be set and top=0 forces the highest bit to be set; top=-1 is new
615 and leaves the highest bit random.
616 [Ulf Moeller, Bodo Moeller]
618 +) Update Rijndael code to version 3.0 and change EVP AES ciphers to
619 handle the new API. Currently only ECB, CBC modes supported. Add new
620 AES OIDs. Add TLS AES ciphersuites as described in the "AES Ciphersuites
621 for TLS" draft-ietf-tls-ciphersuite-03.txt.
622 [Ben Laurie, Steve Henson]
624 *) In the NCONF_...-based implementations for CONF_... queries
625 (crypto/conf/conf_lib.c), if the input LHASH is NULL, avoid using
626 a temporary CONF structure with the data component set to NULL
627 (which gives segmentation faults in lh_retrieve).
628 Instead, use NULL for the CONF pointer in CONF_get_string and
629 CONF_get_number (which may use environment variables) and directly
630 return NULL from CONF_get_section.
633 *) Fix potential buffer overrun for EBCDIC.
636 +) New function OCSP_copy_nonce() to copy nonce value (if present) from
640 +) Functions for OCSP responders. OCSP_request_onereq_count(),
641 OCSP_request_onereq_get0(), OCSP_onereq_get0_id() and OCSP_id_get0_info()
642 extract information from a certificate request. OCSP_response_create()
643 creates a response and optionally adds a basic response structure.
644 OCSP_basic_add1_status() adds a complete single response to a basic
645 reponse and returns the OCSP_SINGLERESP structure just added (to allow
646 extensions to be included for example). OCSP_basic_add1_cert() adds a
647 certificate to a basic response and OCSP_basic_sign() signs a basic
648 response with various flags. New helper functions ASN1_TIME_check()
649 (checks validity of ASN1_TIME structure) and ASN1_TIME_to_generalizedtime()
650 (converts ASN1_TIME to GeneralizedTime).
653 +) Various new functions. EVP_Digest() combines EVP_Digest{Init,Update,Final}()
654 in a single operation. X509_get0_pubkey_bitstr() extracts the public_key
655 structure from a certificate. X509_pubkey_digest() digests tha public_key
656 contents: this is used in various key identifiers.
659 *) Tolerate nonRepudiation as being valid for S/MIME signing and certSign
660 keyUsage if basicConstraints absent for a CA.
663 *) Make SMIME_write_PKCS7() write mail header values with a format that
664 is more generally accepted (no spaces before the semicolon), since
665 some programs can't parse those values properly otherwise. Also make
666 sure BIO's that break lines after each write do not create invalid
670 +) Make sk_sort() tolerate a NULL argument.
671 [Steve Henson reported by Massimiliano Pala <madwolf@comune.modena.it>]
673 +) New OCSP verify flag OCSP_TRUSTOTHER. When set the "other" certificates
674 passed by the function are trusted implicitly. If any of them signed the
675 reponse then it is assumed to be valid and is not verified.
678 -) Make the CRL encoding routines work with empty SEQUENCE OF. The
679 macros previously used would not encode an empty SEQUENCE OF
680 and break the signature.
682 [This change does not apply to 0.9.7.]
684 *) Zero the premaster secret after deriving the master secret in
688 +) In PKCS7_set_type() initialise content_type in PKCS7_ENC_CONTENT
689 to data. This was previously part of the PKCS7 ASN1 code. This
690 was causing problems with OpenSSL created PKCS#12 and PKCS#7 structures.
691 [Steve Henson, reported by Kenneth R. Robinette
692 <support@securenetterm.com>]
694 +) Add CRYPTO_push_info() and CRYPTO_pop_info() calls to new ASN1
695 routines: without these tracing memory leaks is very painful.
696 Fix leaks in PKCS12 and PKCS7 routines.
699 *) Add some EVP_add_digest_alias registrations (as found in
700 OpenSSL_add_all_digests()) to SSL_library_init()
701 aka OpenSSL_add_ssl_algorithms(). This provides improved
702 compatibility with peers using X.509 certificates
703 with unconventional AlgorithmIdentifier OIDs.
706 *) Fix for Irix with NO_ASM.
707 ["Bruce W. Forsberg" <bruce.forsberg@baesystems.com>]
709 *) ./config script fixes.
710 [Ulf Moeller, Richard Levitte]
712 +) Make X509_time_adj() cope with the new behaviour of ASN1_TIME_new().
713 Previously it initialised the 'type' argument to V_ASN1_UTCTIME which
714 effectively meant GeneralizedTime would never be used. Now it
715 is initialised to -1 but X509_time_adj() now has to check the value
716 and use ASN1_TIME_set() if the value is not V_ASN1_UTCTIME or
717 V_ASN1_GENERALIZEDTIME, without this it always uses GeneralizedTime.
718 [Steve Henson, reported by Kenneth R. Robinette
719 <support@securenetterm.com>]
721 +) Fixes to BN_to_ASN1_INTEGER when bn is zero. This would previously
722 result in a zero length in the ASN1_INTEGER structure which was
723 not consistent with the structure when d2i_ASN1_INTEGER() was used
724 and would cause ASN1_INTEGER_cmp() to fail. Enhance s2i_ASN1_INTEGER()
725 to cope with hex and negative integers. Fix bug in i2a_ASN1_INTEGER()
726 where it did not print out a minus for negative ASN1_INTEGER.
729 *) Fix 'openssl passwd -1'.
732 +) Add summary printout to ocsp utility. The various functions which
733 convert status values to strings have been renamed to:
734 OCSP_response_status_str(), OCSP_cert_status_str() and
735 OCSP_crl_reason_str() and are no longer static. New options
736 to verify nonce values and to disable verification. OCSP response
737 printout format cleaned up.
740 +) Add additional OCSP certificate checks. These are those specified
741 in RFC2560. This consists of two separate checks: the CA of the
742 certificate being checked must either be the OCSP signer certificate
743 or the issuer of the OCSP signer certificate. In the latter case the
744 OCSP signer certificate must contain the OCSP signing extended key
745 usage. This check is performed by attempting to match the OCSP
746 signer or the OCSP signer CA to the issuerNameHash and issuerKeyHash
747 in the OCSP_CERTID structures of the response.
750 +) Initial OCSP certificate verification added to OCSP_basic_verify()
751 and related routines. This uses the standard OpenSSL certificate
752 verify routines to perform initial checks (just CA validity) and
753 to obtain the certificate chain. Then additional checks will be
754 performed on the chain. Currently the root CA is checked to see
755 if it is explicitly trusted for OCSP signing. This is used to set
756 a root CA as a global signing root: that is any certificate that
757 chains to that CA is an acceptable OCSP signing certificate.
760 +) New '-extfile ...' option to 'openssl ca' for reading X.509v3
761 extensions from a separate configuration file.
762 As when reading extensions from the main configuration file,
763 the '-extensions ...' option may be used for specifying the
765 [Massimiliano Pala <madwolf@comune.modena.it>]
767 *) Change PKCS12_key_gen_asc() so it can cope with non null
768 terminated strings whose length is passed in the passlen
769 parameter, for example from PEM callbacks. This was done
770 by adding an extra length parameter to asc2uni().
771 [Steve Henson, reported by <oddissey@samsung.co.kr>]
773 +) New OCSP utility. Allows OCSP requests to be generated or
774 read. The request can be sent to a responder and the output
775 parsed, outputed or printed in text form. Not complete yet:
776 still needs to check the OCSP response validity.
779 +) New subcommands for 'openssl ca':
780 'openssl ca -status <serial>' prints the status of the cert with
781 the given serial number (according to the index file).
782 'openssl ca -updatedb' updates the expiry status of certificates
784 [Massimiliano Pala <madwolf@comune.modena.it>]
786 +) New '-newreq-nodes' command option to CA.pl. This is like
787 '-newreq', but calls 'openssl req' with the '-nodes' option
788 so that the resulting key is not encrypted.
789 [Damien Miller <djm@mindrot.org>]
791 +) New configuration for the GNU Hurd.
792 [Jonathan Bartlett <johnnyb@wolfram.com> via Richard Levitte]
794 +) Initial code to implement OCSP basic response verify. This
795 is currently incomplete. Currently just finds the signer's
796 certificate and verifies the signature on the response.
799 +) New SSLeay_version code SSLEAY_DIR to determine the compiled-in
800 value of OPENSSLDIR. This is available via the new '-d' option
801 to 'openssl version', and is also included in 'openssl version -a'.
804 *) Fix C code generated by 'openssl dsaparam -C': If a BN_bin2bn
805 call failed, free the DSA structure.
808 +) Allowing defining memory allocation callbacks that will be given
809 file name and line number information in additional arguments
810 (a const char* and an int). The basic functionality remains, as
811 well as the original possibility to just replace malloc(),
812 realloc() and free() by functions that do not know about these
813 additional arguments. To register and find out the current
814 settings for extended allocation functions, the following
815 functions are provided:
817 CRYPTO_set_mem_ex_functions
818 CRYPTO_set_locked_mem_ex_functions
819 CRYPTO_get_mem_ex_functions
820 CRYPTO_get_locked_mem_ex_functions
822 These work the same way as CRYPTO_set_mem_functions and friends.
823 CRYPTO_get_[locked_]mem_functions now writes 0 where such an
824 extended allocation function is enabled.
825 Similarly, CRYPTO_get_[locked_]mem_ex_functions writes 0 where
826 a conventional allocation function is enabled.
827 [Richard Levitte, Bodo Moeller]
829 *) Fix to uni2asc() to cope with zero length Unicode strings.
830 These are present in some PKCS#12 files.
833 +) Finish off removing the remaining LHASH function pointer casts.
834 There should no longer be any prototype-casting required when using
835 the LHASH abstraction, and any casts that remain are "bugs". See
836 the callback types and macros at the head of lhash.h for details
837 (and "OBJ_cleanup" in crypto/objects/obj_dat.c as an example).
840 +) Add automatic query of EGD sockets in RAND_poll() for the unix variant.
841 If /dev/[u]random devices are not available or do not return enough
842 entropy, EGD style sockets (served by EGD or PRNGD) will automatically
844 The locations /var/run/egd-pool, /dev/egd-pool, /etc/egd-pool, and
845 /etc/entropy will be queried once each in this sequence, quering stops
846 when enough entropy was collected without querying more sockets.
849 +) Change the Unix RAND_poll() variant to be able to poll several
850 random devices, as specified by DEVRANDOM, until a sufficient amount
851 of data has been collected. We spend at most 10 ms on each file
852 (select timeout) and read in non-blocking mode. DEVRANDOM now
853 defaults to the list "/dev/urandom", "/dev/random", "/dev/srandom"
854 (previously it was just the string "/dev/urandom"), so on typical
855 platforms the 10 ms delay will never occur.
856 Also separate out the Unix variant to its own file, rand_unix.c.
857 For VMS, there's a currently-empty rand_vms.c.
860 +) Move OCSP client related routines to ocsp_cl.c. These
861 provide utility functions which an application needing
862 to issue a request to an OCSP responder and analyse the
863 response will typically need: as opposed to those which an
864 OCSP responder itself would need which will be added later.
866 OCSP_request_sign() signs an OCSP request with an API similar
867 to PKCS7_sign(). OCSP_response_status() returns status of OCSP
868 response. OCSP_response_get1_basic() extracts basic response
869 from response. OCSP_resp_find_status(): finds and extracts status
870 information from an OCSP_CERTID structure (which will be created
871 when the request structure is built). These are built from lower
872 level functions which work on OCSP_SINGLERESP structures but
873 wont normally be used unless the application wishes to examine
874 extensions in the OCSP response for example.
876 Replace nonce routines with a pair of functions.
877 OCSP_request_add1_nonce() adds a nonce value and optionally
878 generates a random value. OCSP_check_nonce() checks the
879 validity of the nonce in an OCSP response.
882 +) Change function OCSP_request_add() to OCSP_request_add0_id().
883 This doesn't copy the supplied OCSP_CERTID and avoids the
884 need to free up the newly created id. Change return type
885 to OCSP_ONEREQ to return the internal OCSP_ONEREQ structure.
886 This can then be used to add extensions to the request.
887 Deleted OCSP_request_new(), since most of its functionality
888 is now in OCSP_REQUEST_new() (and the case insensitive name
889 clash) apart from the ability to set the request name which
890 will be added elsewhere.
893 +) Update OCSP API. Remove obsolete extensions argument from
894 various functions. Extensions are now handled using the new
895 OCSP extension code. New simple OCSP HTTP function which
896 can be used to send requests and parse the response.
899 +) Fix the PKCS#7 (S/MIME) code to work with new ASN1. Two new
900 ASN1_ITEM structures help with sign and verify. PKCS7_ATTR_SIGN
901 uses the special reorder version of SET OF to sort the attributes
902 and reorder them to match the encoded order. This resolves a long
903 standing problem: a verify on a PKCS7 structure just after signing
904 it used to fail because the attribute order did not match the
905 encoded order. PKCS7_ATTR_VERIFY does not reorder the attributes:
906 it uses the received order. This is necessary to tolerate some broken
907 software that does not order SET OF. This is handled by encoding
908 as a SEQUENCE OF but using implicit tagging (with UNIVERSAL class)
909 to produce the required SET OF.
912 +) Have mk1mf.pl generate the macros OPENSSL_BUILD_SHLIBCRYPTO and
913 OPENSSL_BUILD_SHLIBSSL and use them appropriately in the header
914 files to get correct declarations of the ASN.1 item variables.
917 +) Rewrite of PKCS#12 code to use new ASN1 functionality. Replace many
918 PKCS#12 macros with real functions. Fix two unrelated ASN1 bugs:
919 asn1_check_tlen() would sometimes attempt to use 'ctx' when it was
920 NULL and ASN1_TYPE was not dereferenced properly in asn1_ex_c2i().
921 New ASN1 macro: DECLARE_ASN1_ITEM() which just declares the relevant
922 ASN1_ITEM and no wrapper functions.
925 +) New functions or ASN1_item_d2i_fp() and ASN1_item_d2i_bio(). These
926 replace the old function pointer based I/O routines. Change most of
927 the *_d2i_bio() and *_d2i_fp() functions to use these.
930 +) Enhance mkdef.pl to be more accepting about spacing in C preprocessor
931 lines, recognice more "algorithms" that can be deselected, and make
932 it complain about algorithm deselection that isn't recognised.
935 +) New ASN1 functions to handle dup, sign, verify, digest, pack and
936 unpack operations in terms of ASN1_ITEM. Modify existing wrappers
937 to use new functions. Add NO_ASN1_OLD which can be set to remove
938 some old style ASN1 functions: this can be used to determine if old
939 code will still work when these eventually go away.
942 +) New extension functions for OCSP structures, these follow the
943 same conventions as certificates and CRLs.
946 +) New function X509V3_add1_i2d(). This automatically encodes and
947 adds an extension. Its behaviour can be customised with various
948 flags to append, replace or delete. Various wrappers added for
949 certifcates and CRLs.
952 +) Fix to avoid calling the underlying ASN1 print routine when
953 an extension cannot be parsed. Correct a typo in the
954 OCSP_SERVICELOC extension. Tidy up print OCSP format.
957 *) Increase s2->wbuf allocation by one byte in ssl2_new (ssl/s2_lib.c).
958 Otherwise do_ssl_write (ssl/s2_pkt.c) will write beyond buffer limits
959 when writing a 32767 byte record.
960 [Bodo Moeller; problem reported by Eric Day <eday@concentric.net>]
962 *) In RSA_eay_public_{en,ed}crypt and RSA_eay_mod_exp (rsa_eay.c),
963 obtain lock CRYPTO_LOCK_RSA before setting rsa->_method_mod_{n,p,q}.
965 (RSA objects have a reference count access to which is protected
966 by CRYPTO_LOCK_RSA [see rsa_lib.c, s3_srvr.c, ssl_cert.c, ssl_rsa.c],
967 so they are meant to be shared between threads.)
968 [Bodo Moeller, Geoff Thorpe; original patch submitted by
969 "Reddie, Steven" <Steven.Reddie@ca.com>]
971 +) Make mkdef.pl parse some of the ASN1 macros and add apropriate
972 entries for variables.
975 *) Fix a deadlock in CRYPTO_mem_leaks().
978 +) Add functionality to apps/openssl.c for detecting locking
979 problems: As the program is single-threaded, all we have
980 to do is register a locking callback using an array for
981 storing which locks are currently held by the program.
984 +) Use a lock around the call to CRYPTO_get_ex_new_index() in
985 SSL_get_ex_data_X509_STORE_idx(), which is used in
986 ssl_verify_cert_chain() and thus can be called at any time
987 during TLS/SSL handshakes so that thread-safety is essential.
988 Unfortunately, the ex_data design is not at all suited
989 for multi-threaded use, so it probably should be abolished.
992 +) Added Broadcom "ubsec" ENGINE to OpenSSL.
993 [Broadcom, tweaked and integrated by Geoff Thorpe]
995 +) Move common extension printing code to new function
996 X509V3_print_extensions(). Reorganise OCSP print routines and
997 implement some needed OCSP ASN1 functions. Add OCSP extensions.
1000 +) New function X509_signature_print() to remove duplication in some
1004 +) Add a special meaning when SET OF and SEQUENCE OF flags are both
1005 set (this was treated exactly the same as SET OF previously). This
1006 is used to reorder the STACK representing the structure to match the
1007 encoding. This will be used to get round a problem where a PKCS7
1008 structure which was signed could not be verified because the STACK
1009 order did not reflect the encoded order.
1012 +) Reimplement the OCSP ASN1 module using the new code.
1015 +) Update the X509V3 code to permit the use of an ASN1_ITEM structure
1016 for its ASN1 operations. The old style function pointers still exist
1017 for now but they will eventually go away.
1020 +) Merge in replacement ASN1 code from the ASN1 branch. This almost
1021 completely replaces the old ASN1 functionality with a table driven
1022 encoder and decoder which interprets an ASN1_ITEM structure describing
1023 the ASN1 module. Compatibility with the existing ASN1 API (i2d,d2i) is
1024 largely maintained. Almost all of the old asn1_mac.h macro based ASN1
1025 has also been converted to the new form.
1028 +) Change BN_mod_exp_recp so that negative moduli are tolerated
1029 (the sign is ignored). Similarly, ignore the sign in BN_MONT_CTX_set
1030 so that BN_mod_exp_mont and BN_mod_exp_mont_word work
1031 for negative moduli.
1034 +) Fix BN_uadd and BN_usub: Always return non-negative results instead
1035 of not touching the result's sign bit.
1038 +) BN_div bugfix: If the result is 0, the sign (res->neg) must not be
1042 +) Changed the LHASH code to use prototypes for callbacks, and created
1043 macros to declare and implement thin (optionally static) functions
1044 that provide type-safety and avoid function pointer casting for the
1045 type-specific callbacks.
1048 *) Use better test patterns in bntest.
1051 +) Added Kerberos Cipher Suites to be used with TLS, as written in
1053 [Veers Staats <staatsvr@asc.hpc.mil>,
1054 Jeffrey Altman <jaltman@columbia.edu>, via Richard Levitte]
1056 *) rand_win.c fix for Borland C.
1059 *) BN_rshift bugfix for n == 0.
1062 +) Reformat the FAQ so the different questions and answers can be divided
1063 in sections depending on the subject.
1066 +) Have the zlib compression code load ZLIB.DLL dynamically under
1070 +) New function BN_mod_sqrt for computing square roots modulo a prime
1071 (using the probabilistic Tonelli-Shanks algorithm unless
1072 p == 3 (mod 4) or p == 5 (mod 8), which are cases that can
1073 be handled deterministically).
1074 [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller]
1076 *) Add a 'bctest' script that checks for some known 'bc' bugs
1077 so that 'make test' does not abort just because 'bc' is broken.
1080 *) Store verify_result within SSL_SESSION also for client side to
1081 avoid potential security hole. (Re-used sessions on the client side
1082 always resulted in verify_result==X509_V_OK, not using the original
1083 result of the server certificate verification.)
1086 +) Make BN_mod_inverse faster by explicitly handling small quotients
1087 in the Euclid loop. (Speed gain about 20% for small moduli [256 or
1088 512 bits], about 30% for larger ones [1024 or 2048 bits].)
1091 *) Fix ssl3_pending: If the record in s->s3->rrec is not of type
1092 SSL3_RT_APPLICATION_DATA, return 0.
1093 Similarly, change ssl2_pending to return 0 if SSL_in_init(s) is true.
1097 Both ssl2_peek and ssl3_peek, which were totally broken in earlier
1098 releases, have been re-implemented by renaming the previous
1099 implementations of ssl2_read and ssl3_read to ssl2_read_internal
1100 and ssl3_read_internal, respectively, and adding 'peek' parameters
1101 to them. The new ssl[23]_{read,peek} functions are calls to
1102 ssl[23]_read_internal with the 'peek' flag set appropriately.
1103 A 'peek' parameter has also been added to ssl3_read_bytes, which
1104 does the actual work for ssl3_read_internal.
1107 +) New function BN_kronecker.
1110 +) Fix BN_gcd so that it works on negative inputs; the result is
1111 positive unless both parameters are zero.
1112 Previously something reasonably close to an infinite loop was
1113 possible because numbers could be growing instead of shrinking
1114 in the implementation of Euclid's algorithm.
1117 +) Fix BN_is_word() and BN_is_one() macros to take into account the
1118 sign of the number in question.
1120 Fix BN_is_word(a,w) to work correctly for w == 0.
1122 The old BN_is_word(a,w) macro is now called BN_abs_is_word(a,w)
1123 because its test if the absolute value of 'a' equals 'w'.
1124 Note that BN_abs_is_word does *not* handle w == 0 reliably;
1125 it exists mostly for use in the implementations of BN_is_zero(),
1126 BN_is_one(), and BN_is_word().
1129 *) Initialise "ex_data" member of RSA/DSA/DH structures prior to calling
1130 the method-specific "init()" handler. Also clean up ex_data after
1131 calling the method-specific "finish()" handler. Previously, this was
1132 happening the other way round.
1135 +) New function BN_swap.
1138 +) Use BN_nnmod instead of BN_mod in crypto/bn/bn_exp.c so that
1139 the exponentiation functions are more likely to produce reasonable
1140 results on negative inputs.
1143 +) Change BN_mod_mul so that the result is always non-negative.
1144 Previously, it could be negative if one of the factors was negative;
1145 I don't think anyone really wanted that behaviour.
1148 +) Move BN_mod_... functions into new file crypto/bn/bn_mod.c
1149 (except for exponentiation, which stays in crypto/bn/bn_exp.c,
1150 and BN_mod_mul_reciprocal, which stays in crypto/bn/bn_recp.c)
1151 and add new functions:
1160 BN_mod_lshift1_quick
1164 These functions always generate non-negative results.
1166 BN_nnmod otherwise is like BN_mod (if BN_mod computes a remainder r
1167 such that |m| < r < 0, BN_nnmod will output rem + |m| instead).
1169 BN_mod_XXX_quick(r, a, [b,] m) generates the same result as
1170 BN_mod_XXX(r, a, [b,] m, ctx), but requires that a [and b]
1171 be reduced modulo m.
1172 [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller]
1174 +) Remove a few calls to bn_wexpand() in BN_sqr() (the one in there
1175 was actually never needed) and in BN_mul(). The removal in BN_mul()
1176 required a small change in bn_mul_part_recursive() and the addition
1177 of the functions bn_cmp_part_words(), bn_sub_part_words() and
1178 bn_add_part_words(), which do the same thing as bn_cmp_words(),
1179 bn_sub_words() and bn_add_words() except they take arrays with
1183 +) In 'openssl passwd', verify passwords read from the terminal
1184 unless the '-salt' option is used (which usually means that
1185 verification would just waste user's time since the resulting
1186 hash is going to be compared with some given password hash)
1187 or the new '-noverify' option is used.
1189 This is an incompatible change, but it does not affect
1190 non-interactive use of 'openssl passwd' (passwords on the command
1191 line, '-stdin' option, '-in ...' option) and thus should not
1195 +) Remove all references to RSAref, since there's no more need for it.
1198 *) Increase BN_CTX_NUM (the number of BIGNUMs in a BN_CTX) to 16.
1199 The previous value, 12, was not always sufficient for BN_mod_exp().
1202 +) Make DSO load along a path given through an environment variable
1203 (SHLIB_PATH) with shl_load().
1206 +) Constify the ENGINE code as a result of BIGNUM constification.
1207 Also constify the RSA code and most things related to it. In a
1208 few places, most notable in the depth of the ASN.1 code, ugly
1209 casts back to non-const were required (to be solved at a later
1213 +) Make it so the openssl application has all engines loaded by default.
1216 +) Constify the BIGNUM routines a little more.
1219 *) Make sure that shared libraries get the internal name engine with
1220 the full version number and not just 0. This should mark the
1221 shared libraries as not backward compatible. Of course, this should
1222 be changed again when we can guarantee backward binary compatibility.
1225 +) Add the following functions:
1227 ENGINE_load_cswift()
1229 ENGINE_load_atalla()
1231 ENGINE_load_builtin_engines()
1233 That way, an application can itself choose if external engines that
1234 are built-in in OpenSSL shall ever be used or not. The benefit is
1235 that applications won't have to be linked with libdl or other dso
1236 libraries unless it's really needed.
1238 Changed 'openssl engine' to load all engines on demand.
1239 Changed the engine header files to avoid the duplication of some
1240 declarations (they differed!).
1243 +) 'openssl engine' can now list capabilities.
1246 +) Better error reporting in 'openssl engine'.
1249 +) Never call load_dh_param(NULL) in s_server.
1252 +) Add engine application. It can currently list engines by name and
1253 identity, and test if they are actually available.
1256 +) Add support for shared libraries under Irix.
1257 [Albert Chin-A-Young <china@thewrittenword.com>]
1259 +) Improve RPM specification file by forcing symbolic linking and making
1260 sure the installed documentation is also owned by root.root.
1261 [Damien Miller <djm@mindrot.org>]
1263 +) Add configuration option to build on Linux on both big-endian and
1265 [Ralf Baechle <ralf@uni-koblenz.de>]
1267 +) Give the OpenSSL applications more possibilities to make use of
1268 keys (public as well as private) handled by engines.
1271 +) Add OCSP code that comes from CertCo.
1274 +) Add VMS support for the Rijndael code.
1277 +) Added untested support for Nuron crypto accelerator.
1280 +) Add support for external cryptographic devices. This code was
1281 previously distributed separately as the "engine" branch.
1282 [Geoff Thorpe, Richard Levitte]
1284 +) Rework the filename-translation in the DSO code. It is now possible to
1285 have far greater control over how a "name" is turned into a filename
1286 depending on the operating environment and any oddities about the
1287 different shared library filenames on each system.
1290 +) Support threads on FreeBSD-elf in Configure.
1293 +) Add the possibility to create shared libraries on HP-UX
1296 +) Fix for SHA1 assembly problem with MASM: it produces
1297 warnings about corrupt line number information when assembling
1298 with debugging information. This is caused by the overlapping
1300 [Bernd Matthes <mainbug@celocom.de>, Steve Henson]
1303 NCONF_get_number() has no error checking at all. As a replacement,
1304 NCONF_get_number_e() is defined (_e for "error checking") and is
1305 promoted strongly. The old NCONF_get_number is kept around for
1306 binary backward compatibility.
1307 Make it possible for methods to load from something other than a BIO,
1308 by providing a function pointer that is given a name instead of a BIO.
1309 For example, this could be used to load configuration data from an
1313 *) Fix typo in get_cert_by_subject() in by_dir.c
1314 [Jean-Marc Desperrier <jean-marc.desperrier@certplus.com>]
1316 *) Rework the system to generate shared libraries:
1318 - Make note of the expected extension for the shared libraries and
1319 if there is a need for symbolic links from for example libcrypto.so.0
1320 to libcrypto.so.0.9.7. There is extended info in Configure for
1323 - Make as few rebuilds of the shared libraries as possible.
1325 - Still avoid linking the OpenSSL programs with the shared libraries.
1327 - When installing, install the shared libraries separately from the
1331 +) Fix for non blocking accept BIOs. Added new I/O special reason
1332 BIO_RR_ACCEPT to cover this case. Previously use of accept BIOs
1333 with non blocking I/O was not possible because no retry code was
1334 implemented. Also added new SSL code SSL_WANT_ACCEPT to cover
1338 +) Added the beginnings of Rijndael support.
1341 +) Fix for bug in DirectoryString mask setting. Add support for
1342 X509_NAME_print_ex() in 'req' and X509_print_ex() function
1343 to allow certificate printing to more controllable, additional
1344 'certopt' option to 'x509' to allow new printing options to be
1348 +) Clean old EAY MD5 hack from e_os.h.
1351 *) Fix SSL_CTX_set_read_ahead macro to actually use its argument.
1353 Copy SSL_CTX's read_ahead flag to SSL object directly in SSL_new
1354 and not in SSL_clear because the latter is also used by the
1355 accept/connect functions; previously, the settings made by
1356 SSL_set_read_ahead would be lost during the handshake.
1357 [Bodo Moeller; problems reported by Anders Gertz <gertz@epact.se>]
1359 *) Correct util/mkdef.pl to be selective about disabled algorithms.
1360 Previously, it would create entries for disableed algorithms no
1364 *) Added several new manual pages for SSL_* function.
1367 Changes between 0.9.5a and 0.9.6 [24 Sep 2000]
1369 *) In ssl23_get_client_hello, generate an error message when faced
1370 with an initial SSL 3.0/TLS record that is too small to contain the
1371 first two bytes of the ClientHello message, i.e. client_version.
1372 (Note that this is a pathologic case that probably has never happened
1373 in real life.) The previous approach was to use the version number
1374 from the record header as a substitute; but our protocol choice
1375 should not depend on that one because it is not authenticated
1376 by the Finished messages.
1379 *) More robust randomness gathering functions for Windows.
1380 [Jeffrey Altman <jaltman@columbia.edu>]
1382 *) For compatibility reasons if the flag X509_V_FLAG_ISSUER_CHECK is
1383 not set then we don't setup the error code for issuer check errors
1384 to avoid possibly overwriting other errors which the callback does
1385 handle. If an application does set the flag then we assume it knows
1386 what it is doing and can handle the new informational codes
1390 *) Fix for a nasty bug in ASN1_TYPE handling. ASN1_TYPE is used for
1391 a general "ANY" type, as such it should be able to decode anything
1392 including tagged types. However it didn't check the class so it would
1393 wrongly interpret tagged types in the same way as their universal
1394 counterpart and unknown types were just rejected. Changed so that the
1395 tagged and unknown types are handled in the same way as a SEQUENCE:
1396 that is the encoding is stored intact. There is also a new type
1397 "V_ASN1_OTHER" which is used when the class is not universal, in this
1398 case we have no idea what the actual type is so we just lump them all
1402 *) On VMS, stdout may very well lead to a file that is written to
1403 in a record-oriented fashion. That means that every write() will
1404 write a separate record, which will be read separately by the
1405 programs trying to read from it. This can be very confusing.
1407 The solution is to put a BIO filter in the way that will buffer
1408 text until a linefeed is reached, and then write everything a
1409 line at a time, so every record written will be an actual line,
1410 not chunks of lines and not (usually doesn't happen, but I've
1411 seen it once) several lines in one record. BIO_f_linebuffer() is
1414 Currently, it's a VMS-only method, because that's where it has
1415 been tested well enough.
1418 *) Remove 'optimized' squaring variant in BN_mod_mul_montgomery,
1419 it can return incorrect results.
1420 (Note: The buggy variant was not enabled in OpenSSL 0.9.5a,
1421 but it was in 0.9.6-beta[12].)
1424 *) Disable the check for content being present when verifying detached
1425 signatures in pk7_smime.c. Some versions of Netscape (wrongly)
1426 include zero length content when signing messages.
1429 *) New BIO_shutdown_wr macro, which invokes the BIO_C_SHUTDOWN_WR
1430 BIO_ctrl (for BIO pairs).
1433 *) Add DSO method for VMS.
1436 *) Bug fix: Montgomery multiplication could produce results with the
1440 *) Add RPM specification openssl.spec and modify it to build three
1441 packages. The default package contains applications, application
1442 documentation and run-time libraries. The devel package contains
1443 include files, static libraries and function documentation. The
1444 doc package contains the contents of the doc directory. The original
1445 openssl.spec was provided by Damien Miller <djm@mindrot.org>.
1448 *) Add a large number of documentation files for many SSL routines.
1449 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>]
1451 *) Add a configuration entry for Sony News 4.
1452 [NAKAJI Hiroyuki <nakaji@tutrp.tut.ac.jp>]
1454 *) Don't set the two most significant bits to one when generating a
1455 random number < q in the DSA library.
1458 *) New SSL API mode 'SSL_MODE_AUTO_RETRY'. This disables the default
1459 behaviour that SSL_read may result in SSL_ERROR_WANT_READ (even if
1460 the underlying transport is blocking) if a handshake took place.
1461 (The default behaviour is needed by applications such as s_client
1462 and s_server that use select() to determine when to use SSL_read;
1463 but for applications that know in advance when to expect data, it
1464 just makes things more complicated.)
1467 *) Add RAND_egd_bytes(), which gives control over the number of bytes read
1471 *) Add a few more EBCDIC conditionals that make `req' and `x509'
1472 work better on such systems.
1473 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>]
1475 *) Add two demo programs for PKCS12_parse() and PKCS12_create().
1476 Update PKCS12_parse() so it copies the friendlyName and the
1477 keyid to the certificates aux info.
1480 *) Fix bug in PKCS7_verify() which caused an infinite loop
1481 if there was more than one signature.
1482 [Sven Uszpelkat <su@celocom.de>]
1484 *) Major change in util/mkdef.pl to include extra information
1485 about each symbol, as well as presentig variables as well
1486 as functions. This change means that there's n more need
1487 to rebuild the .num files when some algorithms are excluded.
1490 *) Allow the verify time to be set by an application,
1491 rather than always using the current time.
1494 *) Phase 2 verify code reorganisation. The certificate
1495 verify code now looks up an issuer certificate by a
1496 number of criteria: subject name, authority key id
1497 and key usage. It also verifies self signed certificates
1498 by the same criteria. The main comparison function is
1499 X509_check_issued() which performs these checks.
1501 Lot of changes were necessary in order to support this
1502 without completely rewriting the lookup code.
1504 Authority and subject key identifier are now cached.
1506 The LHASH 'certs' is X509_STORE has now been replaced
1507 by a STACK_OF(X509_OBJECT). This is mainly because an
1508 LHASH can't store or retrieve multiple objects with
1509 the same hash value.
1511 As a result various functions (which were all internal
1512 use only) have changed to handle the new X509_STORE
1513 structure. This will break anything that messed round
1514 with X509_STORE internally.
1516 The functions X509_STORE_add_cert() now checks for an
1517 exact match, rather than just subject name.
1519 The X509_STORE API doesn't directly support the retrieval
1520 of multiple certificates matching a given criteria, however
1521 this can be worked round by performing a lookup first
1522 (which will fill the cache with candidate certificates)
1523 and then examining the cache for matches. This is probably
1524 the best we can do without throwing out X509_LOOKUP
1525 entirely (maybe later...).
1527 The X509_VERIFY_CTX structure has been enhanced considerably.
1529 All certificate lookup operations now go via a get_issuer()
1530 callback. Although this currently uses an X509_STORE it
1531 can be replaced by custom lookups. This is a simple way
1532 to bypass the X509_STORE hackery necessary to make this
1533 work and makes it possible to use more efficient techniques
1534 in future. A very simple version which uses a simple
1535 STACK for its trusted certificate store is also provided
1536 using X509_STORE_CTX_trusted_stack().
1538 The verify_cb() and verify() callbacks now have equivalents
1539 in the X509_STORE_CTX structure.
1541 X509_STORE_CTX also has a 'flags' field which can be used
1542 to customise the verify behaviour.
1545 *) Add new PKCS#7 signing option PKCS7_NOSMIMECAP which
1546 excludes S/MIME capabilities.
1549 *) When a certificate request is read in keep a copy of the
1550 original encoding of the signed data and use it when outputing
1551 again. Signatures then use the original encoding rather than
1552 a decoded, encoded version which may cause problems if the
1553 request is improperly encoded.
1556 *) For consistency with other BIO_puts implementations, call
1557 buffer_write(b, ...) directly in buffer_puts instead of calling
1560 In BIO_puts, increment b->num_write as in BIO_write.
1561 [Peter.Sylvester@EdelWeb.fr]
1563 *) Fix BN_mul_word for the case where the word is 0. (We have to use
1564 BN_zero, we may not return a BIGNUM with an array consisting of
1568 *) Avoid calling abort() from within the library when problems are
1569 detected, except if preprocessor symbols have been defined
1570 (such as REF_CHECK, BN_DEBUG etc.).
1573 *) New openssl application 'rsautl'. This utility can be
1574 used for low level RSA operations. DER public key
1575 BIO/fp routines also added.
1578 *) New Configure entry and patches for compiling on QNX 4.
1579 [Andreas Schneider <andreas@ds3.etech.fh-hamburg.de>]
1581 *) A demo state-machine implementation was sponsored by
1582 Nuron (http://www.nuron.com/) and is now available in
1583 demos/state_machine.
1586 *) New options added to the 'dgst' utility for signature
1587 generation and verification.
1590 *) Unrecognized PKCS#7 content types are now handled via a
1591 catch all ASN1_TYPE structure. This allows unsupported
1592 types to be stored as a "blob" and an application can
1593 encode and decode it manually.
1596 *) Fix various signed/unsigned issues to make a_strex.c
1598 [Oscar Jacobsson <oscar.jacobsson@celocom.com>]
1600 *) ASN1 fixes. i2d_ASN1_OBJECT was not returning the correct
1601 length if passed a buffer. ASN1_INTEGER_to_BN failed
1602 if passed a NULL BN and its argument was negative.
1603 [Steve Henson, pointed out by Sven Heiberg <sven@tartu.cyber.ee>]
1605 *) Modification to PKCS#7 encoding routines to output definite
1606 length encoding. Since currently the whole structures are in
1607 memory there's not real point in using indefinite length
1608 constructed encoding. However if OpenSSL is compiled with
1609 the flag PKCS7_INDEFINITE_ENCODING the old form is used.
1612 *) Added BIO_vprintf() and BIO_vsnprintf().
1615 *) Added more prefixes to parse for in the the strings written
1616 through a logging bio, to cover all the levels that are available
1617 through syslog. The prefixes are now:
1619 PANIC, EMERG, EMR => LOG_EMERG
1620 ALERT, ALR => LOG_ALERT
1621 CRIT, CRI => LOG_CRIT
1622 ERROR, ERR => LOG_ERR
1623 WARNING, WARN, WAR => LOG_WARNING
1624 NOTICE, NOTE, NOT => LOG_NOTICE
1625 INFO, INF => LOG_INFO
1626 DEBUG, DBG => LOG_DEBUG
1628 and as before, if none of those prefixes are present at the
1629 beginning of the string, LOG_ERR is chosen.
1631 On Win32, the LOG_* levels are mapped according to this:
1633 LOG_EMERG, LOG_ALERT, LOG_CRIT, LOG_ERR => EVENTLOG_ERROR_TYPE
1634 LOG_WARNING => EVENTLOG_WARNING_TYPE
1635 LOG_NOTICE, LOG_INFO, LOG_DEBUG => EVENTLOG_INFORMATION_TYPE
1639 *) Made it possible to reconfigure with just the configuration
1640 argument "reconf" or "reconfigure". The command line arguments
1641 are stored in Makefile.ssl in the variable CONFIGURE_ARGS,
1642 and are retrieved from there when reconfiguring.
1646 [Assar Westerlund <assar@sics.se>, Richard Levitte]
1648 *) Add the arguments -CAfile and -CApath to the pkcs12 utility.
1651 *) The obj_dat.pl script was messing up the sorting of object
1652 names. The reason was that it compared the quoted version
1653 of strings as a result "OCSP" > "OCSP Signing" because
1654 " > SPACE. Changed script to store unquoted versions of
1655 names and add quotes on output. It was also omitting some
1656 names from the lookup table if they were given a default
1657 value (that is if SN is missing it is given the same
1658 value as LN and vice versa), these are now added on the
1659 grounds that if an object has a name we should be able to
1660 look it up. Finally added warning output when duplicate
1661 short or long names are found.
1664 *) Changes needed for Tandem NSK.
1665 [Scott Uroff <scott@xypro.com>]
1667 *) Fix SSL 2.0 rollback checking: Due to an off-by-one error in
1668 RSA_padding_check_SSLv23(), special padding was never detected
1669 and thus the SSL 3.0/TLS 1.0 countermeasure against protocol
1670 version rollback attacks was not effective.
1672 In s23_clnt.c, don't use special rollback-attack detection padding
1673 (RSA_SSLV23_PADDING) if SSL 2.0 is the only protocol enabled in the
1674 client; similarly, in s23_srvr.c, don't do the rollback check if
1675 SSL 2.0 is the only protocol enabled in the server.
1678 *) Make it possible to get hexdumps of unprintable data with 'openssl
1679 asn1parse'. By implication, the functions ASN1_parse_dump() and
1680 BIO_dump_indent() are added.
1683 *) New functions ASN1_STRING_print_ex() and X509_NAME_print_ex()
1684 these print out strings and name structures based on various
1685 flags including RFC2253 support and proper handling of
1686 multibyte characters. Added options to the 'x509' utility
1687 to allow the various flags to be set.
1690 *) Various fixes to use ASN1_TIME instead of ASN1_UTCTIME.
1691 Also change the functions X509_cmp_current_time() and
1692 X509_gmtime_adj() work with an ASN1_TIME structure,
1693 this will enable certificates using GeneralizedTime in validity
1694 dates to be checked.
1697 *) Make the NEG_PUBKEY_BUG code (which tolerates invalid
1698 negative public key encodings) on by default,
1699 NO_NEG_PUBKEY_BUG can be set to disable it.
1702 *) New function c2i_ASN1_OBJECT() which acts on ASN1_OBJECT
1703 content octets. An i2c_ASN1_OBJECT is unnecessary because
1704 the encoding can be trivially obtained from the structure.
1707 *) crypto/err.c locking bugfix: Use write locks (CRYPTO_w_[un]lock),
1708 not read locks (CRYPTO_r_[un]lock).
1711 *) A first attempt at creating official support for shared
1712 libraries through configuration. I've kept it so the
1713 default is static libraries only, and the OpenSSL programs
1714 are always statically linked for now, but there are
1715 preparations for dynamic linking in place.
1716 This has been tested on Linux and True64.
1719 *) Randomness polling function for Win9x, as described in:
1720 Peter Gutmann, Software Generation of Practically Strong
1724 *) Fix so PRNG is seeded in req if using an already existing
1728 *) New options to smime application. -inform and -outform
1729 allow alternative formats for the S/MIME message including
1730 PEM and DER. The -content option allows the content to be
1731 specified separately. This should allow things like Netscape
1732 form signing output easier to verify.
1735 *) Fix the ASN1 encoding of tags using the 'long form'.
1738 *) New ASN1 functions, i2c_* and c2i_* for INTEGER and BIT
1739 STRING types. These convert content octets to and from the
1740 underlying type. The actual tag and length octets are
1741 already assumed to have been read in and checked. These
1742 are needed because all other string types have virtually
1743 identical handling apart from the tag. By having versions
1744 of the ASN1 functions that just operate on content octets
1745 IMPLICIT tagging can be handled properly. It also allows
1746 the ASN1_ENUMERATED code to be cut down because ASN1_ENUMERATED
1747 and ASN1_INTEGER are identical apart from the tag.
1750 *) Change the handling of OID objects as follows:
1752 - New object identifiers are inserted in objects.txt, following
1753 the syntax given in objects.README.
1754 - objects.pl is used to process obj_mac.num and create a new
1756 - obj_dat.pl is used to create a new obj_dat.h, using the data in
1759 This is currently kind of a hack, and the perl code in objects.pl
1760 isn't very elegant, but it works as I intended. The simplest way
1761 to check that it worked correctly is to look in obj_dat.h and
1762 check the array nid_objs and make sure the objects haven't moved
1763 around (this is important!). Additions are OK, as well as
1764 consistent name changes.
1767 *) Add BSD-style MD5-based passwords to 'openssl passwd' (option '-1').
1770 *) Addition of the command line parameter '-rand file' to 'openssl req'.
1771 The given file adds to whatever has already been seeded into the
1772 random pool through the RANDFILE configuration file option or
1773 environment variable, or the default random state file.
1776 *) mkstack.pl now sorts each macro group into lexical order.
1777 Previously the output order depended on the order the files
1778 appeared in the directory, resulting in needless rewriting
1782 *) Patches to make OpenSSL compile under Win32 again. Mostly
1783 work arounds for the VC++ problem that it treats func() as
1784 func(void). Also stripped out the parts of mkdef.pl that
1785 added extra typesafe functions: these no longer exist.
1788 *) Reorganisation of the stack code. The macros are now all
1789 collected in safestack.h . Each macro is defined in terms of
1790 a "stack macro" of the form SKM_<name>(type, a, b). The
1791 DEBUG_SAFESTACK is now handled in terms of function casts,
1792 this has the advantage of retaining type safety without the
1793 use of additional functions. If DEBUG_SAFESTACK is not defined
1794 then the non typesafe macros are used instead. Also modified the
1795 mkstack.pl script to handle the new form. Needs testing to see
1796 if which (if any) compilers it chokes and maybe make DEBUG_SAFESTACK
1797 the default if no major problems. Similar behaviour for ASN1_SET_OF
1798 and PKCS12_STACK_OF.
1801 *) When some versions of IIS use the 'NET' form of private key the
1802 key derivation algorithm is different. Normally MD5(password) is
1803 used as a 128 bit RC4 key. In the modified case
1804 MD5(MD5(password) + "SGCKEYSALT") is used insted. Added some
1805 new functions i2d_RSA_NET(), d2i_RSA_NET() etc which are the same
1806 as the old Netscape_RSA functions except they have an additional
1807 'sgckey' parameter which uses the modified algorithm. Also added
1808 an -sgckey command line option to the rsa utility. Thanks to
1809 Adrian Peck <bertie@ncipher.com> for posting details of the modified
1810 algorithm to openssl-dev.
1813 *) The evp_local.h macros were using 'c.##kname' which resulted in
1814 invalid expansion on some systems (SCO 5.0.5 for example).
1815 Corrected to 'c.kname'.
1816 [Phillip Porch <root@theporch.com>]
1818 *) New X509_get1_email() and X509_REQ_get1_email() functions that return
1819 a STACK of email addresses from a certificate or request, these look
1820 in the subject name and the subject alternative name extensions and
1821 omit any duplicate addresses.
1824 *) Re-implement BN_mod_exp2_mont using independent (and larger) windows.
1825 This makes DSA verification about 2 % faster.
1828 *) Increase maximum window size in BN_mod_exp_... to 6 bits instead of 5
1829 (meaning that now 2^5 values will be precomputed, which is only 4 KB
1830 plus overhead for 1024 bit moduli).
1831 This makes exponentiations about 0.5 % faster for 1024 bit
1832 exponents (as measured by "openssl speed rsa2048").
1835 *) Rename memory handling macros to avoid conflicts with other
1837 Malloc => OPENSSL_malloc
1838 Malloc_locked => OPENSSL_malloc_locked
1839 Realloc => OPENSSL_realloc
1840 Free => OPENSSL_free
1843 *) New function BN_mod_exp_mont_word for small bases (roughly 15%
1844 faster than BN_mod_exp_mont, i.e. 7% for a full DH exchange).
1847 *) CygWin32 support.
1848 [John Jarvie <jjarvie@newsguy.com>]
1850 *) The type-safe stack code has been rejigged. It is now only compiled
1851 in when OpenSSL is configured with the DEBUG_SAFESTACK option and
1852 by default all type-specific stack functions are "#define"d back to
1853 standard stack functions. This results in more streamlined output
1854 but retains the type-safety checking possibilities of the original
1858 *) The STACK code has been cleaned up, and certain type declarations
1859 that didn't make a lot of sense have been brought in line. This has
1860 also involved a cleanup of sorts in safestack.h to more correctly
1861 map type-safe stack functions onto their plain stack counterparts.
1862 This work has also resulted in a variety of "const"ifications of
1863 lots of the code, especially "_cmp" operations which should normally
1864 be prototyped with "const" parameters anyway.
1867 *) When generating bytes for the first time in md_rand.c, 'stir the pool'
1868 by seeding with STATE_SIZE dummy bytes (with zero entropy count).
1869 (The PRNG state consists of two parts, the large pool 'state' and 'md',
1870 where all of 'md' is used each time the PRNG is used, but 'state'
1871 is used only indexed by a cyclic counter. As entropy may not be
1872 well distributed from the beginning, 'md' is important as a
1873 chaining variable. However, the output function chains only half
1874 of 'md', i.e. 80 bits. ssleay_rand_add, on the other hand, chains
1875 all of 'md', and seeding with STATE_SIZE dummy bytes will result
1876 in all of 'state' being rewritten, with the new values depending
1877 on virtually all of 'md'. This overcomes the 80 bit limitation.)
1880 *) In ssl/s2_clnt.c and ssl/s3_clnt.c, call ERR_clear_error() when
1881 the handshake is continued after ssl_verify_cert_chain();
1882 otherwise, if SSL_VERIFY_NONE is set, remaining error codes
1883 can lead to 'unexplainable' connection aborts later.
1884 [Bodo Moeller; problem tracked down by Lutz Jaenicke]
1886 *) Major EVP API cipher revision.
1887 Add hooks for extra EVP features. This allows various cipher
1888 parameters to be set in the EVP interface. Support added for variable
1889 key length ciphers via the EVP_CIPHER_CTX_set_key_length() function and
1890 setting of RC2 and RC5 parameters.
1892 Modify EVP_OpenInit() and EVP_SealInit() to cope with variable key length
1895 Remove lots of duplicated code from the EVP library. For example *every*
1896 cipher init() function handles the 'iv' in the same way according to the
1897 cipher mode. They also all do nothing if the 'key' parameter is NULL and
1898 for CFB and OFB modes they zero ctx->num.
1900 New functionality allows removal of S/MIME code RC2 hack.
1902 Most of the routines have the same form and so can be declared in terms
1905 By shifting this to the top level EVP_CipherInit() it can be removed from
1906 all individual ciphers. If the cipher wants to handle IVs or keys
1907 differently it can set the EVP_CIPH_CUSTOM_IV or EVP_CIPH_ALWAYS_CALL_INIT
1910 Change lots of functions like EVP_EncryptUpdate() to now return a
1911 value: although software versions of the algorithms cannot fail
1912 any installed hardware versions can.
1915 *) Implement SSL_OP_TLS_ROLLBACK_BUG: In ssl3_get_client_key_exchange, if
1916 this option is set, tolerate broken clients that send the negotiated
1917 protocol version number instead of the requested protocol version
1921 *) Call dh_tmp_cb (set by ..._TMP_DH_CB) with correct 'is_export' flag;
1922 i.e. non-zero for export ciphersuites, zero otherwise.
1923 Previous versions had this flag inverted, inconsistent with
1924 rsa_tmp_cb (..._TMP_RSA_CB).
1925 [Bodo Moeller; problem reported by Amit Chopra]
1927 *) Add missing DSA library text string. Work around for some IIS
1928 key files with invalid SEQUENCE encoding.
1931 *) Add a document (doc/standards.txt) that list all kinds of standards
1932 and so on that are implemented in OpenSSL.
1935 *) Enhance c_rehash script. Old version would mishandle certificates
1936 with the same subject name hash and wouldn't handle CRLs at all.
1937 Added -fingerprint option to crl utility, to support new c_rehash
1941 *) Eliminate non-ANSI declarations in crypto.h and stack.h.
1944 *) Fix for SSL server purpose checking. Server checking was
1945 rejecting certificates which had extended key usage present
1946 but no ssl client purpose.
1947 [Steve Henson, reported by Rene Grosser <grosser@hisolutions.com>]
1949 *) Make PKCS#12 code work with no password. The PKCS#12 spec
1950 is a little unclear about how a blank password is handled.
1951 Since the password in encoded as a BMPString with terminating
1952 double NULL a zero length password would end up as just the
1953 double NULL. However no password at all is different and is
1954 handled differently in the PKCS#12 key generation code. NS
1955 treats a blank password as zero length. MSIE treats it as no
1956 password on export: but it will try both on import. We now do
1957 the same: PKCS12_parse() tries zero length and no password if
1958 the password is set to "" or NULL (NULL is now a valid password:
1959 it wasn't before) as does the pkcs12 application.
1962 *) Bugfixes in apps/x509.c: Avoid a memory leak; and don't use
1963 perror when PEM_read_bio_X509_REQ fails, the error message must
1964 be obtained from the error queue.
1967 *) Avoid 'thread_hash' memory leak in crypto/err/err.c by freeing
1968 it in ERR_remove_state if appropriate, and change ERR_get_state
1969 accordingly to avoid race conditions (this is necessary because
1970 thread_hash is no longer constant once set).
1973 *) Bugfix for linux-elf makefile.one.
1976 *) RSA_get_default_method() will now cause a default
1977 RSA_METHOD to be chosen if one doesn't exist already.
1978 Previously this was only set during a call to RSA_new()
1979 or RSA_new_method(NULL) meaning it was possible for
1980 RSA_get_default_method() to return NULL.
1983 *) Added native name translation to the existing DSO code
1984 that will convert (if the flag to do so is set) filenames
1985 that are sufficiently small and have no path information
1986 into a canonical native form. Eg. "blah" converted to
1987 "libblah.so" or "blah.dll" etc.
1990 *) New function ERR_error_string_n(e, buf, len) which is like
1991 ERR_error_string(e, buf), but writes at most 'len' bytes
1992 including the 0 terminator. For ERR_error_string_n, 'buf'
1994 [Damien Miller <djm@mindrot.org>, Bodo Moeller]
1996 *) CONF library reworked to become more general. A new CONF
1997 configuration file reader "class" is implemented as well as a
1998 new functions (NCONF_*, for "New CONF") to handle it. The now
1999 old CONF_* functions are still there, but are reimplemented to
2000 work in terms of the new functions. Also, a set of functions
2001 to handle the internal storage of the configuration data is
2002 provided to make it easier to write new configuration file
2003 reader "classes" (I can definitely see something reading a
2004 configuration file in XML format, for example), called _CONF_*,
2005 or "the configuration storage API"...
2007 The new configuration file reading functions are:
2009 NCONF_new, NCONF_free, NCONF_load, NCONF_load_fp, NCONF_load_bio,
2010 NCONF_get_section, NCONF_get_string, NCONF_get_numbre
2012 NCONF_default, NCONF_WIN32
2014 NCONF_dump_fp, NCONF_dump_bio
2016 NCONF_default and NCONF_WIN32 are method (or "class") choosers,
2017 NCONF_new creates a new CONF object. This works in the same way
2018 as other interfaces in OpenSSL, like the BIO interface.
2019 NCONF_dump_* dump the internal storage of the configuration file,
2020 which is useful for debugging. All other functions take the same
2021 arguments as the old CONF_* functions wth the exception of the
2022 first that must be a `CONF *' instead of a `LHASH *'.
2024 To make it easer to use the new classes with the old CONF_* functions,
2025 the function CONF_set_default_method is provided.
2028 *) Add '-tls1' option to 'openssl ciphers', which was already
2029 mentioned in the documentation but had not been implemented.
2030 (This option is not yet really useful because even the additional
2031 experimental TLS 1.0 ciphers are currently treated as SSL 3.0 ciphers.)
2034 *) Initial DSO code added into libcrypto for letting OpenSSL (and
2035 OpenSSL-based applications) load shared libraries and bind to
2036 them in a portable way.
2037 [Geoff Thorpe, with contributions from Richard Levitte]
2039 Changes between 0.9.5 and 0.9.5a [1 Apr 2000]
2041 *) Make sure _lrotl and _lrotr are only used with MSVC.
2043 *) Use lock CRYPTO_LOCK_RAND correctly in ssleay_rand_status
2044 (the default implementation of RAND_status).
2046 *) Rename openssl x509 option '-crlext', which was added in 0.9.5,
2047 to '-clrext' (= clear extensions), as intended and documented.
2048 [Bodo Moeller; inconsistency pointed out by Michael Attili
2051 *) Fix for HMAC. It wasn't zeroing the rest of the block if the key length
2052 was larger than the MD block size.
2053 [Steve Henson, pointed out by Yost William <YostW@tce.com>]
2055 *) Modernise PKCS12_parse() so it uses STACK_OF(X509) for its ca argument
2056 fix a leak when the ca argument was passed as NULL. Stop X509_PUBKEY_set()
2057 using the passed key: if the passed key was a private key the result
2058 of X509_print(), for example, would be to print out all the private key
2062 *) des_quad_cksum() byte order bug fix.
2063 [Ulf Möller, using the problem description in krb4-0.9.7, where
2064 the solution is attributed to Derrick J Brashear <shadow@DEMENTIA.ORG>]
2066 *) Fix so V_ASN1_APP_CHOOSE works again: however its use is strongly
2068 [Steve Henson, pointed out by Brian Korver <briank@cs.stanford.edu>]
2070 *) For easily testing in shell scripts whether some command
2071 'openssl XXX' exists, the new pseudo-command 'openssl no-XXX'
2072 returns with exit code 0 iff no command of the given name is available.
2073 'no-XXX' is printed in this case, 'XXX' otherwise. In both cases,
2074 the output goes to stdout and nothing is printed to stderr.
2075 Additional arguments are always ignored.
2077 Since for each cipher there is a command of the same name,
2078 the 'no-cipher' compilation switches can be tested this way.
2080 ('openssl no-XXX' is not able to detect pseudo-commands such
2081 as 'quit', 'list-XXX-commands', or 'no-XXX' itself.)
2084 *) Update test suite so that 'make test' succeeds in 'no-rsa' configuration.
2087 *) For SSL_[CTX_]set_tmp_dh, don't create a DH key if SSL_OP_SINGLE_DH_USE
2088 is set; it will be thrown away anyway because each handshake creates
2090 ssl_cert_dup, which is used by SSL_new, now copies DH keys in addition
2091 to parameters -- in previous versions (since OpenSSL 0.9.3) the
2092 'default key' from SSL_CTX_set_tmp_dh would always be lost, meanining
2093 you effectivly got SSL_OP_SINGLE_DH_USE when using this macro.
2096 *) New s_client option -ign_eof: EOF at stdin is ignored, and
2097 'Q' and 'R' lose their special meanings (quit/renegotiate).
2098 This is part of what -quiet does; unlike -quiet, -ign_eof
2099 does not suppress any output.
2102 *) Add compatibility options to the purpose and trust code. The
2103 purpose X509_PURPOSE_ANY is "any purpose" which automatically
2104 accepts a certificate or CA, this was the previous behaviour,
2105 with all the associated security issues.
2107 X509_TRUST_COMPAT is the old trust behaviour: only and
2108 automatically trust self signed roots in certificate store. A
2109 new trust setting X509_TRUST_DEFAULT is used to specify that
2110 a purpose has no associated trust setting and it should instead
2111 use the value in the default purpose.
2114 *) Fix the PKCS#8 DSA private key code so it decodes keys again
2115 and fix a memory leak.
2118 *) In util/mkerr.pl (which implements 'make errors'), preserve
2119 reason strings from the previous version of the .c file, as
2120 the default to have only downcase letters (and digits) in
2121 automatically generated reasons codes is not always appropriate.
2124 *) In ERR_load_ERR_strings(), build an ERR_LIB_SYS error reason table
2125 using strerror. Previously, ERR_reason_error_string() returned
2126 library names as reason strings for SYSerr; but SYSerr is a special
2127 case where small numbers are errno values, not library numbers.
2130 *) Add '-dsaparam' option to 'openssl dhparam' application. This
2131 converts DSA parameters into DH parameters. (When creating parameters,
2132 DSA_generate_parameters is used.)
2135 *) Include 'length' (recommended exponent length) in C code generated
2136 by 'openssl dhparam -C'.
2139 *) The second argument to set_label in perlasm was already being used
2140 so couldn't be used as a "file scope" flag. Moved to third argument
2144 *) In PEM_ASN1_write_bio and some other functions, use RAND_pseudo_bytes
2145 instead of RAND_bytes for encryption IVs and salts.
2148 *) Include RAND_status() into RAND_METHOD instead of implementing
2149 it only for md_rand.c Otherwise replacing the PRNG by calling
2150 RAND_set_rand_method would be impossible.
2153 *) Don't let DSA_generate_key() enter an infinite loop if the random
2154 number generation fails.
2157 *) New 'rand' application for creating pseudo-random output.
2160 *) Added configuration support for Linux/IA64
2161 [Rolf Haberrecker <rolf@suse.de>]
2163 *) Assembler module support for Mingw32.
2166 *) Shared library support for HPUX (in shlib/).
2167 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE> and Anonymous]
2169 *) Shared library support for Solaris gcc.
2170 [Lutz Behnke <behnke@trustcenter.de>]
2172 Changes between 0.9.4 and 0.9.5 [28 Feb 2000]
2174 *) PKCS7_encrypt() was adding text MIME headers twice because they
2175 were added manually and by SMIME_crlf_copy().
2178 *) In bntest.c don't call BN_rand with zero bits argument.
2179 [Steve Henson, pointed out by Andrew W. Gray <agray@iconsinc.com>]
2181 *) BN_mul bugfix: In bn_mul_part_recursion() only the a>a[n] && b>b[n]
2182 case was implemented. This caused BN_div_recp() to fail occasionally.
2185 *) Add an optional second argument to the set_label() in the perl
2186 assembly language builder. If this argument exists and is set
2187 to 1 it signals that the assembler should use a symbol whose
2188 scope is the entire file, not just the current function. This
2189 is needed with MASM which uses the format label:: for this scope.
2190 [Steve Henson, pointed out by Peter Runestig <peter@runestig.com>]
2192 *) Change the ASN1 types so they are typedefs by default. Before
2193 almost all types were #define'd to ASN1_STRING which was causing
2194 STACK_OF() problems: you couldn't declare STACK_OF(ASN1_UTF8STRING)
2198 *) Change names of new functions to the new get1/get0 naming
2199 convention: After 'get1', the caller owns a reference count
2200 and has to call ..._free; 'get0' returns a pointer to some
2201 data structure without incrementing reference counters.
2202 (Some of the existing 'get' functions increment a reference
2203 counter, some don't.)
2204 Similarly, 'set1' and 'add1' functions increase reference
2205 counters or duplicate objects.
2208 *) Allow for the possibility of temp RSA key generation failure:
2209 the code used to assume it always worked and crashed on failure.
2212 *) Fix potential buffer overrun problem in BIO_printf().
2213 [Ulf Möller, using public domain code by Patrick Powell; problem
2214 pointed out by David Sacerdote <das33@cornell.edu>]
2216 *) Support EGD <http://www.lothar.com/tech/crypto/>. New functions
2217 RAND_egd() and RAND_status(). In the command line application,
2218 the EGD socket can be specified like a seed file using RANDFILE
2222 *) Allow the string CERTIFICATE to be tolerated in PKCS#7 structures.
2223 Some CAs (e.g. Verisign) distribute certificates in this form.
2226 *) Remove the SSL_ALLOW_ADH compile option and set the default cipher
2227 list to exclude them. This means that no special compilation option
2228 is needed to use anonymous DH: it just needs to be included in the
2232 *) Change the EVP_MD_CTX_type macro so its meaning consistent with
2233 EVP_MD_type. The old functionality is available in a new macro called
2234 EVP_MD_md(). Change code that uses it and update docs.
2237 *) ..._ctrl functions now have corresponding ..._callback_ctrl functions
2238 where the 'void *' argument is replaced by a function pointer argument.
2239 Previously 'void *' was abused to point to functions, which works on
2240 many platforms, but is not correct. As these functions are usually
2241 called by macros defined in OpenSSL header files, most source code
2242 should work without changes.
2245 *) <openssl/opensslconf.h> (which is created by Configure) now contains
2246 sections with information on -D... compiler switches used for
2247 compiling the library so that applications can see them. To enable
2248 one of these sections, a pre-processor symbol OPENSSL_..._DEFINES
2249 must be defined. E.g.,
2250 #define OPENSSL_ALGORITHM_DEFINES
2251 #include <openssl/opensslconf.h>
2252 defines all pertinent NO_<algo> symbols, such as NO_IDEA, NO_RSA, etc.
2253 [Richard Levitte, Ulf and Bodo Möller]
2255 *) Bugfix: Tolerate fragmentation and interleaving in the SSL 3/TLS
2259 *) Change the 'other' type in certificate aux info to a STACK_OF
2260 X509_ALGOR. Although not an AlgorithmIdentifier as such it has
2261 the required ASN1 format: arbitrary types determined by an OID.
2264 *) Add some PEM_write_X509_REQ_NEW() functions and a command line
2265 argument to 'req'. This is not because the function is newer or
2266 better than others it just uses the work 'NEW' in the certificate
2267 request header lines. Some software needs this.
2270 *) Reorganise password command line arguments: now passwords can be
2271 obtained from various sources. Delete the PEM_cb function and make
2272 it the default behaviour: i.e. if the callback is NULL and the
2273 usrdata argument is not NULL interpret it as a null terminated pass
2274 phrase. If usrdata and the callback are NULL then the pass phrase
2275 is prompted for as usual.
2278 *) Add support for the Compaq Atalla crypto accelerator. If it is installed,
2279 the support is automatically enabled. The resulting binaries will
2280 autodetect the card and use it if present.
2281 [Ben Laurie and Compaq Inc.]
2283 *) Work around for Netscape hang bug. This sends certificate request
2284 and server done in one record. Since this is perfectly legal in the
2285 SSL/TLS protocol it isn't a "bug" option and is on by default. See
2286 the bugs/SSLv3 entry for more info.
2289 *) HP-UX tune-up: new unified configs, HP C compiler bug workaround.
2292 *) Add -rand argument to smime and pkcs12 applications and read/write
2296 *) New 'passwd' tool for crypt(3) and apr1 password hashes.
2299 *) Add command line password options to the remaining applications.
2302 *) Bug fix for BN_div_recp() for numerators with an even number of
2306 *) More tests in bntest.c, and changed test_bn output.
2309 *) ./config recognizes MacOS X now.
2312 *) Bug fix for BN_div() when the first words of num and divsor are
2313 equal (it gave wrong results if (rem=(n1-q*d0)&BN_MASK2) < d0).
2316 *) Add support for various broken PKCS#8 formats, and command line
2317 options to produce them.
2320 *) New functions BN_CTX_start(), BN_CTX_get() and BT_CTX_end() to
2321 get temporary BIGNUMs from a BN_CTX.
2324 *) Correct return values in BN_mod_exp_mont() and BN_mod_exp2_mont()
2328 *) Change the SSLeay_add_all_*() functions to OpenSSL_add_all_*() and
2329 include a #define from the old name to the new. The original intent
2330 was that statically linked binaries could for example just call
2331 SSLeay_add_all_ciphers() to just add ciphers to the table and not
2332 link with digests. This never worked becayse SSLeay_add_all_digests()
2333 and SSLeay_add_all_ciphers() were in the same source file so calling
2334 one would link with the other. They are now in separate source files.
2337 *) Add a new -notext option to 'ca' and a -pubkey option to 'spkac'.
2340 *) Use a less unusual form of the Miller-Rabin primality test (it used
2341 a binary algorithm for exponentiation integrated into the Miller-Rabin
2342 loop, our standard modexp algorithms are faster).
2345 *) Support for the EBCDIC character set completed.
2346 [Martin Kraemer <Martin.Kraemer@Mch.SNI.De>]
2348 *) Source code cleanups: use const where appropriate, eliminate casts,
2349 use void * instead of char * in lhash.
2352 *) Bugfix: ssl3_send_server_key_exchange was not restartable
2353 (the state was not changed to SSL3_ST_SW_KEY_EXCH_B, and because of
2354 this the server could overwrite ephemeral keys that the client
2358 *) Turn DSA_is_prime into a macro that calls BN_is_prime,
2359 using 50 iterations of the Rabin-Miller test.
2361 DSA_generate_parameters now uses BN_is_prime_fasttest (with 50
2362 iterations of the Rabin-Miller test as required by the appendix
2363 to FIPS PUB 186[-1]) instead of DSA_is_prime.
2364 As BN_is_prime_fasttest includes trial division, DSA parameter
2365 generation becomes much faster.
2367 This implies a change for the callback functions in DSA_is_prime
2368 and DSA_generate_parameters: The callback function is called once
2369 for each positive witness in the Rabin-Miller test, not just
2370 occasionally in the inner loop; and the parameters to the
2371 callback function now provide an iteration count for the outer
2372 loop rather than for the current invocation of the inner loop.
2373 DSA_generate_parameters additionally can call the callback
2374 function with an 'iteration count' of -1, meaning that a
2375 candidate has passed the trial division test (when q is generated
2376 from an application-provided seed, trial division is skipped).
2379 *) New function BN_is_prime_fasttest that optionally does trial
2380 division before starting the Rabin-Miller test and has
2381 an additional BN_CTX * argument (whereas BN_is_prime always
2382 has to allocate at least one BN_CTX).
2383 'callback(1, -1, cb_arg)' is called when a number has passed the
2384 trial division stage.
2387 *) Fix for bug in CRL encoding. The validity dates weren't being handled
2391 *) New -pkcs12 option to CA.pl script to write out a PKCS#12 file.
2394 *) New function BN_pseudo_rand().
2397 *) Clean up BN_mod_mul_montgomery(): replace the broken (and unreadable)
2398 bignum version of BN_from_montgomery() with the working code from
2399 SSLeay 0.9.0 (the word based version is faster anyway), and clean up
2403 *) Avoid a race condition in s2_clnt.c (function get_server_hello) that
2404 made it impossible to use the same SSL_SESSION data structure in
2405 SSL2 clients in multiple threads.
2408 *) The return value of RAND_load_file() no longer counts bytes obtained
2409 by stat(). RAND_load_file(..., -1) is new and uses the complete file
2410 to seed the PRNG (previously an explicit byte count was required).
2411 [Ulf Möller, Bodo Möller]
2413 *) Clean up CRYPTO_EX_DATA functions, some of these didn't have prototypes
2414 used (char *) instead of (void *) and had casts all over the place.
2417 *) Make BN_generate_prime() return NULL on error if ret!=NULL.
2420 *) Retain source code compatibility for BN_prime_checks macro:
2421 BN_is_prime(..., BN_prime_checks, ...) now uses
2422 BN_prime_checks_for_size to determine the appropriate number of
2423 Rabin-Miller iterations.
2426 *) Diffie-Hellman uses "safe" primes: DH_check() return code renamed to
2427 DH_CHECK_P_NOT_SAFE_PRIME.
2428 (Check if this is true? OpenPGP calls them "strong".)
2431 *) Merge the functionality of "dh" and "gendh" programs into a new program
2432 "dhparam". The old programs are retained for now but will handle DH keys
2433 (instead of parameters) in future.
2436 *) Make the ciphers, s_server and s_client programs check the return values
2437 when a new cipher list is set.
2440 *) Enhance the SSL/TLS cipher mechanism to correctly handle the TLS 56bit
2441 ciphers. Before when the 56bit ciphers were enabled the sorting was
2444 The syntax for the cipher sorting has been extended to support sorting by
2445 cipher-strength (using the strength_bits hard coded in the tables).
2446 The new command is "@STRENGTH" (see also doc/apps/ciphers.pod).
2448 Fix a bug in the cipher-command parser: when supplying a cipher command
2449 string with an "undefined" symbol (neither command nor alphanumeric
2450 [A-Za-z0-9], ssl_set_cipher_list used to hang in an endless loop. Now
2451 an error is flagged.
2453 Due to the strength-sorting extension, the code of the
2454 ssl_create_cipher_list() function was completely rearranged. I hope that
2455 the readability was also increased :-)
2456 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>]
2458 *) Minor change to 'x509' utility. The -CAcreateserial option now uses 1
2459 for the first serial number and places 2 in the serial number file. This
2460 avoids problems when the root CA is created with serial number zero and
2461 the first user certificate has the same issuer name and serial number
2465 *) Fixes to X509_ATTRIBUTE utilities, change the 'req' program so it uses
2466 the new code. Add documentation for this stuff.
2469 *) Changes to X509_ATTRIBUTE utilities. These have been renamed from
2470 X509_*() to X509at_*() on the grounds that they don't handle X509
2471 structures and behave in an analagous way to the X509v3 functions:
2472 they shouldn't be called directly but wrapper functions should be used
2475 So we also now have some wrapper functions that call the X509at functions
2476 when passed certificate requests. (TO DO: similar things can be done with
2477 PKCS#7 signed and unsigned attributes, PKCS#12 attributes and a few other
2478 things. Some of these need some d2i or i2d and print functionality
2479 because they handle more complex structures.)
2482 *) Add missing #ifndefs that caused missing symbols when building libssl
2483 as a shared library without RSA. Use #ifndef NO_SSL2 instead of
2484 NO_RSA in ssl/s2*.c.
2485 [Kris Kennaway <kris@hub.freebsd.org>, modified by Ulf Möller]
2487 *) Precautions against using the PRNG uninitialized: RAND_bytes() now
2488 has a return value which indicates the quality of the random data
2489 (1 = ok, 0 = not seeded). Also an error is recorded on the thread's
2490 error queue. New function RAND_pseudo_bytes() generates output that is
2491 guaranteed to be unique but not unpredictable. RAND_add is like
2492 RAND_seed, but takes an extra argument for an entropy estimate
2493 (RAND_seed always assumes full entropy).
2496 *) Do more iterations of Rabin-Miller probable prime test (specifically,
2497 3 for 1024-bit primes, 6 for 512-bit primes, 12 for 256-bit primes
2498 instead of only 2 for all lengths; see BN_prime_checks_for_size definition
2499 in crypto/bn/bn_prime.c for the complete table). This guarantees a
2500 false-positive rate of at most 2^-80 for random input.
2503 *) Rewrite ssl3_read_n (ssl/s3_pkt.c) avoiding a couple of bugs.
2506 *) New function X509_CTX_rget_chain() (renamed to X509_CTX_get1_chain
2507 in the 0.9.5 release), this returns the chain
2508 from an X509_CTX structure with a dup of the stack and all
2509 the X509 reference counts upped: so the stack will exist
2510 after X509_CTX_cleanup() has been called. Modify pkcs12.c
2513 Also make SSL_SESSION_print() print out the verify return
2517 *) Add manpage for the pkcs12 command. Also change the default
2518 behaviour so MAC iteration counts are used unless the new
2519 -nomaciter option is used. This improves file security and
2520 only older versions of MSIE (4.0 for example) need it.
2523 *) Honor the no-xxx Configure options when creating .DEF files.
2526 *) Add PKCS#10 attributes to field table: challengePassword,
2527 unstructuredName and unstructuredAddress. These are taken from
2528 draft PKCS#9 v2.0 but are compatible with v1.2 provided no
2529 international characters are used.
2531 More changes to X509_ATTRIBUTE code: allow the setting of types
2532 based on strings. Remove the 'loc' parameter when adding
2533 attributes because these will be a SET OF encoding which is sorted
2537 *) Initial changes to the 'req' utility to allow request generation
2538 automation. This will allow an application to just generate a template
2539 file containing all the field values and have req construct the
2542 Initial support for X509_ATTRIBUTE handling. Stacks of these are
2543 used all over the place including certificate requests and PKCS#7
2544 structures. They are currently handled manually where necessary with
2545 some primitive wrappers for PKCS#7. The new functions behave in a
2546 manner analogous to the X509 extension functions: they allow
2547 attributes to be looked up by NID and added.
2549 Later something similar to the X509V3 code would be desirable to
2550 automatically handle the encoding, decoding and printing of the
2551 more complex types. The string types like challengePassword can
2552 be handled by the string table functions.
2554 Also modified the multi byte string table handling. Now there is
2555 a 'global mask' which masks out certain types. The table itself
2556 can use the flag STABLE_NO_MASK to ignore the mask setting: this
2557 is useful when for example there is only one permissible type
2558 (as in countryName) and using the mask might result in no valid
2562 *) Clean up 'Finished' handling, and add functions SSL_get_finished and
2563 SSL_get_peer_finished to allow applications to obtain the latest
2564 Finished messages sent to the peer or expected from the peer,
2565 respectively. (SSL_get_peer_finished is usually the Finished message
2566 actually received from the peer, otherwise the protocol will be aborted.)
2568 As the Finished message are message digests of the complete handshake
2569 (with a total of 192 bits for TLS 1.0 and more for SSL 3.0), they can
2570 be used for external authentication procedures when the authentication
2571 provided by SSL/TLS is not desired or is not enough.
2574 *) Enhanced support for Alpha Linux is added. Now ./config checks if
2575 the host supports BWX extension and if Compaq C is present on the
2576 $PATH. Just exploiting of the BWX extension results in 20-30%
2577 performance kick for some algorithms, e.g. DES and RC4 to mention
2578 a couple. Compaq C in turn generates ~20% faster code for MD5 and
2582 *) Add support for MS "fast SGC". This is arguably a violation of the
2583 SSL3/TLS protocol. Netscape SGC does two handshakes: the first with
2584 weak crypto and after checking the certificate is SGC a second one
2585 with strong crypto. MS SGC stops the first handshake after receiving
2586 the server certificate message and sends a second client hello. Since
2587 a server will typically do all the time consuming operations before
2588 expecting any further messages from the client (server key exchange
2589 is the most expensive) there is little difference between the two.
2591 To get OpenSSL to support MS SGC we have to permit a second client
2592 hello message after we have sent server done. In addition we have to
2593 reset the MAC if we do get this second client hello.
2596 *) Add a function 'd2i_AutoPrivateKey()' this will automatically decide
2597 if a DER encoded private key is RSA or DSA traditional format. Changed
2598 d2i_PrivateKey_bio() to use it. This is only needed for the "traditional"
2599 format DER encoded private key. Newer code should use PKCS#8 format which
2600 has the key type encoded in the ASN1 structure. Added DER private key
2601 support to pkcs8 application.
2604 *) SSL 3/TLS 1 servers now don't request certificates when an anonymous
2605 ciphersuites has been selected (as required by the SSL 3/TLS 1
2606 specifications). Exception: When SSL_VERIFY_FAIL_IF_NO_PEER_CERT
2607 is set, we interpret this as a request to violate the specification
2608 (the worst that can happen is a handshake failure, and 'correct'
2609 behaviour would result in a handshake failure anyway).
2612 *) In SSL_CTX_add_session, take into account that there might be multiple
2613 SSL_SESSION structures with the same session ID (e.g. when two threads
2614 concurrently obtain them from an external cache).
2615 The internal cache can handle only one SSL_SESSION with a given ID,
2616 so if there's a conflict, we now throw out the old one to achieve
2620 *) Add OIDs for idea and blowfish in CBC mode. This will allow both
2621 to be used in PKCS#5 v2.0 and S/MIME. Also add checking to
2622 some routines that use cipher OIDs: some ciphers do not have OIDs
2623 defined and so they cannot be used for S/MIME and PKCS#5 v2.0 for
2627 *) Simplify the trust setting structure and code. Now we just have
2628 two sequences of OIDs for trusted and rejected settings. These will
2629 typically have values the same as the extended key usage extension
2630 and any application specific purposes.
2632 The trust checking code now has a default behaviour: it will just
2633 check for an object with the same NID as the passed id. Functions can
2634 be provided to override either the default behaviour or the behaviour
2635 for a given id. SSL client, server and email already have functions
2636 in place for compatibility: they check the NID and also return "trusted"
2637 if the certificate is self signed.
2640 *) Add d2i,i2d bio/fp functions for PrivateKey: these convert the
2641 traditional format into an EVP_PKEY structure.
2644 *) Add a password callback function PEM_cb() which either prompts for
2645 a password if usr_data is NULL or otherwise assumes it is a null
2646 terminated password. Allow passwords to be passed on command line
2647 environment or config files in a few more utilities.
2650 *) Add a bunch of DER and PEM functions to handle PKCS#8 format private
2651 keys. Add some short names for PKCS#8 PBE algorithms and allow them
2652 to be specified on the command line for the pkcs8 and pkcs12 utilities.
2653 Update documentation.
2656 *) Support for ASN1 "NULL" type. This could be handled before by using
2657 ASN1_TYPE but there wasn't any function that would try to read a NULL
2658 and produce an error if it couldn't. For compatibility we also have
2659 ASN1_NULL_new() and ASN1_NULL_free() functions but these are faked and
2660 don't allocate anything because they don't need to.
2663 *) Initial support for MacOS is now provided. Examine INSTALL.MacOS
2665 [Andy Polyakov, Roy Woods <roy@centicsystems.ca>]
2667 *) Rebuild of the memory allocation routines used by OpenSSL code and
2668 possibly others as well. The purpose is to make an interface that
2669 provide hooks so anyone can build a separate set of allocation and
2670 deallocation routines to be used by OpenSSL, for example memory
2671 pool implementations, or something else, which was previously hard
2672 since Malloc(), Realloc() and Free() were defined as macros having
2673 the values malloc, realloc and free, respectively (except for Win32
2674 compilations). The same is provided for memory debugging code.
2675 OpenSSL already comes with functionality to find memory leaks, but
2676 this gives people a chance to debug other memory problems.
2678 With these changes, a new set of functions and macros have appeared:
2680 CRYPTO_set_mem_debug_functions() [F]
2681 CRYPTO_get_mem_debug_functions() [F]
2682 CRYPTO_dbg_set_options() [F]
2683 CRYPTO_dbg_get_options() [F]
2684 CRYPTO_malloc_debug_init() [M]
2686 The memory debug functions are NULL by default, unless the library
2687 is compiled with CRYPTO_MDEBUG or friends is defined. If someone
2688 wants to debug memory anyway, CRYPTO_malloc_debug_init() (which
2689 gives the standard debugging functions that come with OpenSSL) or
2690 CRYPTO_set_mem_debug_functions() (tells OpenSSL to use functions
2691 provided by the library user) must be used. When the standard
2692 debugging functions are used, CRYPTO_dbg_set_options can be used to
2693 request additional information:
2694 CRYPTO_dbg_set_options(V_CYRPTO_MDEBUG_xxx) corresponds to setting
2695 the CRYPTO_MDEBUG_xxx macro when compiling the library.
2697 Also, things like CRYPTO_set_mem_functions will always give the
2698 expected result (the new set of functions is used for allocation
2699 and deallocation) at all times, regardless of platform and compiler
2702 To finish it up, some functions that were never use in any other
2703 way than through macros have a new API and new semantic:
2706 CRYPTO_dbg_realloc()
2709 All macros of value have retained their old syntax.
2710 [Richard Levitte and Bodo Moeller]
2712 *) Some S/MIME fixes. The OID for SMIMECapabilities was wrong, the
2713 ordering of SMIMECapabilities wasn't in "strength order" and there
2714 was a missing NULL in the AlgorithmIdentifier for the SHA1 signature
2718 *) Some ASN1 types with illegal zero length encoding (INTEGER,
2719 ENUMERATED and OBJECT IDENTIFIER) choked the ASN1 routines.
2720 [Frans Heymans <fheymans@isaserver.be>, modified by Steve Henson]
2722 *) Merge in my S/MIME library for OpenSSL. This provides a simple
2723 S/MIME API on top of the PKCS#7 code, a MIME parser (with enough
2724 functionality to handle multipart/signed properly) and a utility
2725 called 'smime' to call all this stuff. This is based on code I
2726 originally wrote for Celo who have kindly allowed it to be
2727 included in OpenSSL.
2730 *) Add variants des_set_key_checked and des_set_key_unchecked of
2731 des_set_key (aka des_key_sched). Global variable des_check_key
2732 decides which of these is called by des_set_key; this way
2733 des_check_key behaves as it always did, but applications and
2734 the library itself, which was buggy for des_check_key == 1,
2735 have a cleaner way to pick the version they need.
2738 *) New function PKCS12_newpass() which changes the password of a
2742 *) Modify X509_TRUST and X509_PURPOSE so it also uses a static and
2743 dynamic mix. In both cases the ids can be used as an index into the
2744 table. Also modified the X509_TRUST_add() and X509_PURPOSE_add()
2745 functions so they accept a list of the field values and the
2746 application doesn't need to directly manipulate the X509_TRUST
2750 *) Modify the ASN1_STRING_TABLE stuff so it also uses bsearch and doesn't
2754 *) Modify the way the V3 extension code looks up extensions. This now
2755 works in a similar way to the object code: we have some "standard"
2756 extensions in a static table which is searched with OBJ_bsearch()
2757 and the application can add dynamic ones if needed. The file
2758 crypto/x509v3/ext_dat.h now has the info: this file needs to be
2759 updated whenever a new extension is added to the core code and kept
2760 in ext_nid order. There is a simple program 'tabtest.c' which checks
2761 this. New extensions are not added too often so this file can readily
2762 be maintained manually.
2764 There are two big advantages in doing things this way. The extensions
2765 can be looked up immediately and no longer need to be "added" using
2766 X509V3_add_standard_extensions(): this function now does nothing.
2767 [Side note: I get *lots* of email saying the extension code doesn't
2768 work because people forget to call this function]
2769 Also no dynamic allocation is done unless new extensions are added:
2770 so if we don't add custom extensions there is no need to call
2771 X509V3_EXT_cleanup().
2774 *) Modify enc utility's salting as follows: make salting the default. Add a
2775 magic header, so unsalted files fail gracefully instead of just decrypting
2776 to garbage. This is because not salting is a big security hole, so people
2777 should be discouraged from doing it.
2780 *) Fixes and enhancements to the 'x509' utility. It allowed a message
2781 digest to be passed on the command line but it only used this
2782 parameter when signing a certificate. Modified so all relevant
2783 operations are affected by the digest parameter including the
2784 -fingerprint and -x509toreq options. Also -x509toreq choked if a
2785 DSA key was used because it didn't fix the digest.
2788 *) Initial certificate chain verify code. Currently tests the untrusted
2789 certificates for consistency with the verify purpose (which is set
2790 when the X509_STORE_CTX structure is set up) and checks the pathlength.
2792 There is a NO_CHAIN_VERIFY compilation option to keep the old behaviour:
2793 this is because it will reject chains with invalid extensions whereas
2794 every previous version of OpenSSL and SSLeay made no checks at all.
2796 Trust code: checks the root CA for the relevant trust settings. Trust
2797 settings have an initial value consistent with the verify purpose: e.g.
2798 if the verify purpose is for SSL client use it expects the CA to be
2799 trusted for SSL client use. However the default value can be changed to
2800 permit custom trust settings: one example of this would be to only trust
2801 certificates from a specific "secure" set of CAs.
2803 Also added X509_STORE_CTX_new() and X509_STORE_CTX_free() functions
2804 which should be used for version portability: especially since the
2805 verify structure is likely to change more often now.
2807 SSL integration. Add purpose and trust to SSL_CTX and SSL and functions
2808 to set them. If not set then assume SSL clients will verify SSL servers
2811 Two new options to the verify program: -untrusted allows a set of
2812 untrusted certificates to be passed in and -purpose which sets the
2813 intended purpose of the certificate. If a purpose is set then the
2814 new chain verify code is used to check extension consistency.
2817 *) Support for the authority information access extension.
2820 *) Modify RSA and DSA PEM read routines to transparently handle
2821 PKCS#8 format private keys. New *_PUBKEY_* functions that handle
2822 public keys in a format compatible with certificate
2823 SubjectPublicKeyInfo structures. Unfortunately there were already
2824 functions called *_PublicKey_* which used various odd formats so
2825 these are retained for compatibility: however the DSA variants were
2826 never in a public release so they have been deleted. Changed dsa/rsa
2827 utilities to handle the new format: note no releases ever handled public
2828 keys so we should be OK.
2830 The primary motivation for this change is to avoid the same fiasco
2831 that dogs private keys: there are several incompatible private key
2832 formats some of which are standard and some OpenSSL specific and
2833 require various evil hacks to allow partial transparent handling and
2834 even then it doesn't work with DER formats. Given the option anything
2835 other than PKCS#8 should be dumped: but the other formats have to
2836 stay in the name of compatibility.
2838 With public keys and the benefit of hindsight one standard format
2839 is used which works with EVP_PKEY, RSA or DSA structures: though
2840 it clearly returns an error if you try to read the wrong kind of key.
2842 Added a -pubkey option to the 'x509' utility to output the public key.
2843 Also rename the EVP_PKEY_get_*() to EVP_PKEY_rget_*()
2844 (renamed to EVP_PKEY_get1_*() in the OpenSSL 0.9.5 release) and add
2845 EVP_PKEY_rset_*() functions (renamed to EVP_PKEY_set1_*())
2846 that do the same as the EVP_PKEY_assign_*() except they up the
2847 reference count of the added key (they don't "swallow" the
2851 *) Fixes to crypto/x509/by_file.c the code to read in certificates and
2852 CRLs would fail if the file contained no certificates or no CRLs:
2853 added a new function to read in both types and return the number
2854 read: this means that if none are read it will be an error. The
2855 DER versions of the certificate and CRL reader would always fail
2856 because it isn't possible to mix certificates and CRLs in DER format
2857 without choking one or the other routine. Changed this to just read
2858 a certificate: this is the best we can do. Also modified the code
2859 in apps/verify.c to take notice of return codes: it was previously
2860 attempting to read in certificates from NULL pointers and ignoring
2861 any errors: this is one reason why the cert and CRL reader seemed
2862 to work. It doesn't check return codes from the default certificate
2863 routines: these may well fail if the certificates aren't installed.
2866 *) Code to support otherName option in GeneralName.
2869 *) First update to verify code. Change the verify utility
2870 so it warns if it is passed a self signed certificate:
2871 for consistency with the normal behaviour. X509_verify
2872 has been modified to it will now verify a self signed
2873 certificate if *exactly* the same certificate appears
2874 in the store: it was previously impossible to trust a
2875 single self signed certificate. This means that:
2876 openssl verify ss.pem
2877 now gives a warning about a self signed certificate but
2878 openssl verify -CAfile ss.pem ss.pem
2882 *) For servers, store verify_result in SSL_SESSION data structure
2883 (and add it to external session representation).
2884 This is needed when client certificate verifications fails,
2885 but an application-provided verification callback (set by
2886 SSL_CTX_set_cert_verify_callback) allows accepting the session
2887 anyway (i.e. leaves x509_store_ctx->error != X509_V_OK
2888 but returns 1): When the session is reused, we have to set
2889 ssl->verify_result to the appropriate error code to avoid
2891 [Bodo Moeller, problem pointed out by Lutz Jaenicke]
2893 *) Fix a bug in the new PKCS#7 code: it didn't consider the
2894 case in PKCS7_dataInit() where the signed PKCS7 structure
2895 didn't contain any existing data because it was being created.
2896 [Po-Cheng Chen <pocheng@nst.com.tw>, slightly modified by Steve Henson]
2898 *) Add a salt to the key derivation routines in enc.c. This
2899 forms the first 8 bytes of the encrypted file. Also add a
2900 -S option to allow a salt to be input on the command line.
2903 *) New function X509_cmp(). Oddly enough there wasn't a function
2904 to compare two certificates. We do this by working out the SHA1
2905 hash and comparing that. X509_cmp() will be needed by the trust
2909 *) SSL_get1_session() is like SSL_get_session(), but increments
2910 the reference count in the SSL_SESSION returned.
2911 [Geoff Thorpe <geoff@eu.c2.net>]
2913 *) Fix for 'req': it was adding a null to request attributes.
2914 Also change the X509_LOOKUP and X509_INFO code to handle
2915 certificate auxiliary information.
2918 *) Add support for 40 and 64 bit RC2 and RC4 algorithms: document
2922 *) Add the possibility to add extra information to the memory leak
2923 detecting output, to form tracebacks, showing from where each
2924 allocation was originated: CRYPTO_push_info("constant string") adds
2925 the string plus current file name and line number to a per-thread
2926 stack, CRYPTO_pop_info() does the obvious, CRYPTO_remove_all_info()
2927 is like calling CYRPTO_pop_info() until the stack is empty.
2928 Also updated memory leak detection code to be multi-thread-safe.
2931 *) Add options -text and -noout to pkcs7 utility and delete the
2932 encryption options which never did anything. Update docs.
2935 *) Add options to some of the utilities to allow the pass phrase
2936 to be included on either the command line (not recommended on
2937 OSes like Unix) or read from the environment. Update the
2938 manpages and fix a few bugs.
2941 *) Add a few manpages for some of the openssl commands.
2944 *) Fix the -revoke option in ca. It was freeing up memory twice,
2945 leaking and not finding already revoked certificates.
2948 *) Extensive changes to support certificate auxiliary information.
2949 This involves the use of X509_CERT_AUX structure and X509_AUX
2950 functions. An X509_AUX function such as PEM_read_X509_AUX()
2951 can still read in a certificate file in the usual way but it
2952 will also read in any additional "auxiliary information". By
2953 doing things this way a fair degree of compatibility can be
2954 retained: existing certificates can have this information added
2955 using the new 'x509' options.
2957 Current auxiliary information includes an "alias" and some trust
2958 settings. The trust settings will ultimately be used in enhanced
2959 certificate chain verification routines: currently a certificate
2960 can only be trusted if it is self signed and then it is trusted
2964 *) Fix assembler for Alpha (tested only on DEC OSF not Linux or *BSD).
2965 The problem was that one of the replacement routines had not been working
2966 since SSLeay releases. For now the offending routine has been replaced
2967 with non-optimised assembler. Even so, this now gives around 95%
2968 performance improvement for 1024 bit RSA signs.
2971 *) Hack to fix PKCS#7 decryption when used with some unorthodox RC2
2972 handling. Most clients have the effective key size in bits equal to
2973 the key length in bits: so a 40 bit RC2 key uses a 40 bit (5 byte) key.
2974 A few however don't do this and instead use the size of the decrypted key
2975 to determine the RC2 key length and the AlgorithmIdentifier to determine
2976 the effective key length. In this case the effective key length can still
2977 be 40 bits but the key length can be 168 bits for example. This is fixed
2978 by manually forcing an RC2 key into the EVP_PKEY structure because the
2979 EVP code can't currently handle unusual RC2 key sizes: it always assumes
2980 the key length and effective key length are equal.
2983 *) Add a bunch of functions that should simplify the creation of
2984 X509_NAME structures. Now you should be able to do:
2985 X509_NAME_add_entry_by_txt(nm, "CN", MBSTRING_ASC, "Steve", -1, -1, 0);
2986 and have it automatically work out the correct field type and fill in
2987 the structures. The more adventurous can try:
2988 X509_NAME_add_entry_by_txt(nm, field, MBSTRING_UTF8, str, -1, -1, 0);
2989 and it will (hopefully) work out the correct multibyte encoding.
2992 *) Change the 'req' utility to use the new field handling and multibyte
2993 copy routines. Before the DN field creation was handled in an ad hoc
2994 way in req, ca, and x509 which was rather broken and didn't support
2995 BMPStrings or UTF8Strings. Since some software doesn't implement
2996 BMPStrings or UTF8Strings yet, they can be enabled using the config file
2997 using the dirstring_type option. See the new comment in the default
2998 openssl.cnf for more info.
3001 *) Make crypto/rand/md_rand.c more robust:
3002 - Assure unique random numbers after fork().
3003 - Make sure that concurrent threads access the global counter and
3004 md serializably so that we never lose entropy in them
3005 or use exactly the same state in multiple threads.
3006 Access to the large state is not always serializable because
3007 the additional locking could be a performance killer, and
3008 md should be large enough anyway.
3011 *) New file apps/app_rand.c with commonly needed functionality
3012 for handling the random seed file.
3014 Use the random seed file in some applications that previously did not:
3016 dsaparam -genkey (which also ignored its '-rand' option),
3019 x509 (when signing).
3020 Except on systems with /dev/urandom, it is crucial to have a random
3021 seed file at least for key creation, DSA signing, and for DH exchanges;
3022 for RSA signatures we could do without one.
3024 gendh and gendsa (unlike genrsa) used to read only the first byte
3025 of each file listed in the '-rand' option. The function as previously
3026 found in genrsa is now in app_rand.c and is used by all programs
3027 that support '-rand'.
3030 *) In RAND_write_file, use mode 0600 for creating files;
3031 don't just chmod when it may be too late.
3034 *) Report an error from X509_STORE_load_locations
3035 when X509_LOOKUP_load_file or X509_LOOKUP_add_dir failed.
3038 *) New function ASN1_mbstring_copy() this copies a string in either
3039 ASCII, Unicode, Universal (4 bytes per character) or UTF8 format
3040 into an ASN1_STRING type. A mask of permissible types is passed
3041 and it chooses the "minimal" type to use or an error if not type
3045 *) Add function equivalents to the various macros in asn1.h. The old
3046 macros are retained with an M_ prefix. Code inside the library can
3047 use the M_ macros. External code (including the openssl utility)
3048 should *NOT* in order to be "shared library friendly".
3051 *) Add various functions that can check a certificate's extensions
3052 to see if it usable for various purposes such as SSL client,
3053 server or S/MIME and CAs of these types. This is currently
3054 VERY EXPERIMENTAL but will ultimately be used for certificate chain
3055 verification. Also added a -purpose flag to x509 utility to
3056 print out all the purposes.
3059 *) Add a CRYPTO_EX_DATA to X509 certificate structure and associated
3063 *) New X509V3_{X509,CRL,REVOKED}_get_d2i() functions. These will search
3064 for, obtain and decode and extension and obtain its critical flag.
3065 This allows all the necessary extension code to be handled in a
3066 single function call.
3069 *) RC4 tune-up featuring 30-40% performance improvement on most RISC
3070 platforms. See crypto/rc4/rc4_enc.c for further details.
3073 *) New -noout option to asn1parse. This causes no output to be produced
3074 its main use is when combined with -strparse and -out to extract data
3075 from a file (which may not be in ASN.1 format).
3078 *) Fix for pkcs12 program. It was hashing an invalid certificate pointer
3079 when producing the local key id.
3080 [Richard Levitte <levitte@stacken.kth.se>]
3082 *) New option -dhparam in s_server. This allows a DH parameter file to be
3083 stated explicitly. If it is not stated then it tries the first server
3084 certificate file. The previous behaviour hard coded the filename
3088 *) Add -pubin and -pubout options to the rsa and dsa commands. These allow
3089 a public key to be input or output. For example:
3090 openssl rsa -in key.pem -pubout -out pubkey.pem
3091 Also added necessary DSA public key functions to handle this.
3094 *) Fix so PKCS7_dataVerify() doesn't crash if no certificates are contained
3095 in the message. This was handled by allowing
3096 X509_find_by_issuer_and_serial() to tolerate a NULL passed to it.
3097 [Steve Henson, reported by Sampo Kellomaki <sampo@mail.neuronio.pt>]
3099 *) Fix for bug in d2i_ASN1_bytes(): other ASN1 functions add an extra null
3100 to the end of the strings whereas this didn't. This would cause problems
3101 if strings read with d2i_ASN1_bytes() were later modified.
3102 [Steve Henson, reported by Arne Ansper <arne@ats.cyber.ee>]
3104 *) Fix for base64 decode bug. When a base64 bio reads only one line of
3105 data and it contains EOF it will end up returning an error. This is
3106 caused by input 46 bytes long. The cause is due to the way base64
3107 BIOs find the start of base64 encoded data. They do this by trying a
3108 trial decode on each line until they find one that works. When they
3109 do a flag is set and it starts again knowing it can pass all the
3110 data directly through the decoder. Unfortunately it doesn't reset
3111 the context it uses. This means that if EOF is reached an attempt
3112 is made to pass two EOFs through the context and this causes the
3113 resulting error. This can also cause other problems as well. As is
3114 usual with these problems it takes *ages* to find and the fix is
3115 trivial: move one line.
3116 [Steve Henson, reported by ian@uns.ns.ac.yu (Ivan Nejgebauer) ]
3118 *) Ugly workaround to get s_client and s_server working under Windows. The
3119 old code wouldn't work because it needed to select() on sockets and the
3120 tty (for keypresses and to see if data could be written). Win32 only
3121 supports select() on sockets so we select() with a 1s timeout on the
3122 sockets and then see if any characters are waiting to be read, if none
3123 are present then we retry, we also assume we can always write data to
3124 the tty. This isn't nice because the code then blocks until we've
3125 received a complete line of data and it is effectively polling the
3126 keyboard at 1s intervals: however it's quite a bit better than not
3127 working at all :-) A dedicated Windows application might handle this
3128 with an event loop for example.
3131 *) Enhance RSA_METHOD structure. Now there are two extra methods, rsa_sign
3132 and rsa_verify. When the RSA_FLAGS_SIGN_VER option is set these functions
3133 will be called when RSA_sign() and RSA_verify() are used. This is useful
3134 if rsa_pub_dec() and rsa_priv_enc() equivalents are not available.
3135 For this to work properly RSA_public_decrypt() and RSA_private_encrypt()
3136 should *not* be used: RSA_sign() and RSA_verify() must be used instead.
3137 This necessitated the support of an extra signature type NID_md5_sha1
3138 for SSL signatures and modifications to the SSL library to use it instead
3139 of calling RSA_public_decrypt() and RSA_private_encrypt().
3142 *) Add new -verify -CAfile and -CApath options to the crl program, these
3143 will lookup a CRL issuers certificate and verify the signature in a
3144 similar way to the verify program. Tidy up the crl program so it
3145 no longer accesses structures directly. Make the ASN1 CRL parsing a bit
3146 less strict. It will now permit CRL extensions even if it is not
3147 a V2 CRL: this will allow it to tolerate some broken CRLs.
3150 *) Initialize all non-automatic variables each time one of the openssl
3151 sub-programs is started (this is necessary as they may be started
3152 multiple times from the "OpenSSL>" prompt).
3153 [Lennart Bang, Bodo Moeller]
3155 *) Preliminary compilation option RSA_NULL which disables RSA crypto without
3156 removing all other RSA functionality (this is what NO_RSA does). This
3157 is so (for example) those in the US can disable those operations covered
3158 by the RSA patent while allowing storage and parsing of RSA keys and RSA
3162 *) Non-copying interface to BIO pairs.
3163 (still largely untested)
3166 *) New function ANS1_tag2str() to convert an ASN1 tag to a descriptive
3167 ASCII string. This was handled independently in various places before.
3170 *) New functions UTF8_getc() and UTF8_putc() that parse and generate
3171 UTF8 strings a character at a time.
3174 *) Use client_version from client hello to select the protocol
3175 (s23_srvr.c) and for RSA client key exchange verification
3176 (s3_srvr.c), as required by the SSL 3.0/TLS 1.0 specifications.
3179 *) Add various utility functions to handle SPKACs, these were previously
3180 handled by poking round in the structure internals. Added new function
3181 NETSCAPE_SPKI_print() to print out SPKAC and a new utility 'spkac' to
3182 print, verify and generate SPKACs. Based on an original idea from
3183 Massimiliano Pala <madwolf@comune.modena.it> but extensively modified.
3186 *) RIPEMD160 is operational on all platforms and is back in 'make test'.
3189 *) Allow the config file extension section to be overwritten on the
3190 command line. Based on an original idea from Massimiliano Pala
3191 <madwolf@comune.modena.it>. The new option is called -extensions
3192 and can be applied to ca, req and x509. Also -reqexts to override
3193 the request extensions in req and -crlexts to override the crl extensions
3197 *) Add new feature to the SPKAC handling in ca. Now you can include
3198 the same field multiple times by preceding it by "XXXX." for example:
3201 this is the same syntax as used in the req config file.
3204 *) Allow certificate extensions to be added to certificate requests. These
3205 are specified in a 'req_extensions' option of the req section of the
3206 config file. They can be printed out with the -text option to req but
3207 are otherwise ignored at present.
3210 *) Fix a horrible bug in enc_read() in crypto/evp/bio_enc.c: if the first
3211 data read consists of only the final block it would not decrypted because
3212 EVP_CipherUpdate() would correctly report zero bytes had been decrypted.
3213 A misplaced 'break' also meant the decrypted final block might not be
3214 copied until the next read.
3217 *) Initial support for DH_METHOD. Again based on RSA_METHOD. Also added
3218 a few extra parameters to the DH structure: these will be useful if
3219 for example we want the value of 'q' or implement X9.42 DH.
3222 *) Initial support for DSA_METHOD. This is based on the RSA_METHOD and
3223 provides hooks that allow the default DSA functions or functions on a
3224 "per key" basis to be replaced. This allows hardware acceleration and
3225 hardware key storage to be handled without major modification to the
3226 library. Also added low level modexp hooks and CRYPTO_EX structure and
3227 associated functions.
3230 *) Add a new flag to memory BIOs, BIO_FLAG_MEM_RDONLY. This marks the BIO
3231 as "read only": it can't be written to and the buffer it points to will
3232 not be freed. Reading from a read only BIO is much more efficient than
3233 a normal memory BIO. This was added because there are several times when
3234 an area of memory needs to be read from a BIO. The previous method was
3235 to create a memory BIO and write the data to it, this results in two
3236 copies of the data and an O(n^2) reading algorithm. There is a new
3237 function BIO_new_mem_buf() which creates a read only memory BIO from
3238 an area of memory. Also modified the PKCS#7 routines to use read only
3242 *) Bugfix: ssl23_get_client_hello did not work properly when called in
3243 state SSL23_ST_SR_CLNT_HELLO_B, i.e. when the first 7 bytes of
3244 a SSLv2-compatible client hello for SSLv3 or TLSv1 could be read,
3245 but a retry condition occured while trying to read the rest.
3248 *) The PKCS7_ENC_CONTENT_new() function was setting the content type as
3249 NID_pkcs7_encrypted by default: this was wrong since this should almost
3250 always be NID_pkcs7_data. Also modified the PKCS7_set_type() to handle
3251 the encrypted data type: this is a more sensible place to put it and it
3252 allows the PKCS#12 code to be tidied up that duplicated this
3256 *) Changed obj_dat.pl script so it takes its input and output files on
3257 the command line. This should avoid shell escape redirection problems
3261 *) Initial support for certificate extension requests, these are included
3262 in things like Xenroll certificate requests. Included functions to allow
3263 extensions to be obtained and added.
3266 *) -crlf option to s_client and s_server for sending newlines as
3267 CRLF (as required by many protocols).
3270 Changes between 0.9.3a and 0.9.4 [09 Aug 1999]
3272 *) Install libRSAglue.a when OpenSSL is built with RSAref.
3273 [Ralf S. Engelschall]
3275 *) A few more ``#ifndef NO_FP_API / #endif'' pairs for consistency.
3276 [Andrija Antonijevic <TheAntony2@bigfoot.com>]
3278 *) Fix -startdate and -enddate (which was missing) arguments to 'ca'
3282 *) New function DSA_dup_DH, which duplicates DSA parameters/keys as
3283 DH parameters/keys (q is lost during that conversion, but the resulting
3284 DH parameters contain its length).
3286 For 1024-bit p, DSA_generate_parameters followed by DSA_dup_DH is
3287 much faster than DH_generate_parameters (which creates parameters
3288 where p = 2*q + 1), and also the smaller q makes DH computations
3289 much more efficient (160-bit exponentiation instead of 1024-bit
3290 exponentiation); so this provides a convenient way to support DHE
3291 ciphersuites in SSL/TLS servers (see ssl/ssltest.c). It is of
3292 utter importance to use
3293 SSL_CTX_set_options(s_ctx, SSL_OP_SINGLE_DH_USE);
3295 SSL_set_options(s_ctx, SSL_OP_SINGLE_DH_USE);
3296 when such DH parameters are used, because otherwise small subgroup
3297 attacks may become possible!
3300 *) Avoid memory leak in i2d_DHparams.
3303 *) Allow the -k option to be used more than once in the enc program:
3304 this allows the same encrypted message to be read by multiple recipients.
3307 *) New function OBJ_obj2txt(buf, buf_len, a, no_name), this converts
3308 an ASN1_OBJECT to a text string. If the "no_name" parameter is set then
3309 it will always use the numerical form of the OID, even if it has a short
3313 *) Added an extra RSA flag: RSA_FLAG_EXT_PKEY. Previously the rsa_mod_exp
3314 method only got called if p,q,dmp1,dmq1,iqmp components were present,
3315 otherwise bn_mod_exp was called. In the case of hardware keys for example
3316 no private key components need be present and it might store extra data
3317 in the RSA structure, which cannot be accessed from bn_mod_exp.
3318 By setting RSA_FLAG_EXT_PKEY rsa_mod_exp will always be called for
3319 private key operations.
3322 *) Added support for SPARC Linux.
3325 *) pem_password_cb function type incompatibly changed from
3326 typedef int pem_password_cb(char *buf, int size, int rwflag);
3328 ....(char *buf, int size, int rwflag, void *userdata);
3329 so that applications can pass data to their callbacks:
3330 The PEM[_ASN1]_{read,write}... functions and macros now take an
3331 additional void * argument, which is just handed through whenever
3332 the password callback is called.
3333 [Damien Miller <dmiller@ilogic.com.au>; tiny changes by Bodo Moeller]
3335 New function SSL_CTX_set_default_passwd_cb_userdata.
3337 Compatibility note: As many C implementations push function arguments
3338 onto the stack in reverse order, the new library version is likely to
3339 interoperate with programs that have been compiled with the old
3340 pem_password_cb definition (PEM_whatever takes some data that
3341 happens to be on the stack as its last argument, and the callback
3342 just ignores this garbage); but there is no guarantee whatsoever that
3345 *) The -DPLATFORM="\"$(PLATFORM)\"" definition and the similar -DCFLAGS=...
3346 (both in crypto/Makefile.ssl for use by crypto/cversion.c) caused
3347 problems not only on Windows, but also on some Unix platforms.
3348 To avoid problematic command lines, these definitions are now in an
3349 auto-generated file crypto/buildinf.h (created by crypto/Makefile.ssl
3350 for standard "make" builds, by util/mk1mf.pl for "mk1mf" builds).
3353 *) MIPS III/IV assembler module is reimplemented.
3356 *) More DES library cleanups: remove references to srand/rand and
3357 delete an unused file.
3360 *) Add support for the the free Netwide assembler (NASM) under Win32,
3361 since not many people have MASM (ml) and it can be hard to obtain.
3362 This is currently experimental but it seems to work OK and pass all
3363 the tests. Check out INSTALL.W32 for info.
3366 *) Fix memory leaks in s3_clnt.c: All non-anonymous SSL3/TLS1 connections
3367 without temporary keys kept an extra copy of the server key,
3368 and connections with temporary keys did not free everything in case
3372 *) New function RSA_check_key and new openssl rsa option -check
3373 for verifying the consistency of RSA keys.
3374 [Ulf Moeller, Bodo Moeller]
3376 *) Various changes to make Win32 compile work:
3377 1. Casts to avoid "loss of data" warnings in p5_crpt2.c
3378 2. Change unsigned int to int in b_dump.c to avoid "signed/unsigned
3379 comparison" warnings.
3380 3. Add sk_<TYPE>_sort to DEF file generator and do make update.
3383 *) Add a debugging option to PKCS#5 v2 key generation function: when
3384 you #define DEBUG_PKCS5V2 passwords, salts, iteration counts and
3385 derived keys are printed to stderr.
3388 *) Copy the flags in ASN1_STRING_dup().
3389 [Roman E. Pavlov <pre@mo.msk.ru>]
3391 *) The x509 application mishandled signing requests containing DSA
3392 keys when the signing key was also DSA and the parameters didn't match.
3394 It was supposed to omit the parameters when they matched the signing key:
3395 the verifying software was then supposed to automatically use the CA's
3396 parameters if they were absent from the end user certificate.
3398 Omitting parameters is no longer recommended. The test was also
3399 the wrong way round! This was probably due to unusual behaviour in
3400 EVP_cmp_parameters() which returns 1 if the parameters match.
3401 This meant that parameters were omitted when they *didn't* match and
3402 the certificate was useless. Certificates signed with 'ca' didn't have
3404 [Steve Henson, reported by Doug Erickson <Doug.Erickson@Part.NET>]
3406 *) Memory leak checking (-DCRYPTO_MDEBUG) had some problems.
3407 The interface is as follows:
3408 Applications can use
3409 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_ON) aka MemCheck_start(),
3410 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_OFF) aka MemCheck_stop();
3411 "off" is now the default.
3412 The library internally uses
3413 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_DISABLE) aka MemCheck_off(),
3414 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_ENABLE) aka MemCheck_on()
3415 to disable memory-checking temporarily.
3417 Some inconsistent states that previously were possible (and were
3418 even the default) are now avoided.
3420 -DCRYPTO_MDEBUG_TIME is new and additionally stores the current time
3421 with each memory chunk allocated; this is occasionally more helpful
3422 than just having a counter.
3424 -DCRYPTO_MDEBUG_THREAD is also new and adds the thread ID.
3426 -DCRYPTO_MDEBUG_ALL enables all of the above, plus any future
3430 *) Introduce "mode" for SSL structures (with defaults in SSL_CTX),
3431 which largely parallels "options", but is for changing API behaviour,
3432 whereas "options" are about protocol behaviour.
3433 Initial "mode" flags are:
3435 SSL_MODE_ENABLE_PARTIAL_WRITE Allow SSL_write to report success when
3436 a single record has been written.
3437 SSL_MODE_ACCEPT_MOVING_WRITE_BUFFER Don't insist that SSL_write
3438 retries use the same buffer location.
3439 (But all of the contents must be
3443 *) Bugfix: SSL_set_options ignored its parameter, only SSL_CTX_set_options
3446 *) Fix problems with no-hmac etc.
3447 [Ulf Möller, pointed out by Brian Wellington <bwelling@tislabs.com>]
3449 *) New functions RSA_get_default_method(), RSA_set_method() and
3450 RSA_get_method(). These allows replacement of RSA_METHODs without having
3451 to mess around with the internals of an RSA structure.
3454 *) Fix memory leaks in DSA_do_sign and DSA_is_prime.
3455 Also really enable memory leak checks in openssl.c and in some
3457 [Chad C. Mulligan, Bodo Moeller]
3459 *) Fix a bug in d2i_ASN1_INTEGER() and i2d_ASN1_INTEGER() which can mess
3460 up the length of negative integers. This has now been simplified to just
3461 store the length when it is first determined and use it later, rather
3462 than trying to keep track of where data is copied and updating it to
3464 [Steve Henson, reported by Brien Wheeler
3465 <bwheeler@authentica-security.com>]
3467 *) Add a new function PKCS7_signatureVerify. This allows the verification
3468 of a PKCS#7 signature but with the signing certificate passed to the
3469 function itself. This contrasts with PKCS7_dataVerify which assumes the
3470 certificate is present in the PKCS#7 structure. This isn't always the
3471 case: certificates can be omitted from a PKCS#7 structure and be
3472 distributed by "out of band" means (such as a certificate database).
3475 *) Complete the PEM_* macros with DECLARE_PEM versions to replace the
3476 function prototypes in pem.h, also change util/mkdef.pl to add the
3477 necessary function names.
3480 *) mk1mf.pl (used by Windows builds) did not properly read the
3481 options set by Configure in the top level Makefile, and Configure
3482 was not even able to write more than one option correctly.
3483 Fixed, now "no-idea no-rc5 -DCRYPTO_MDEBUG" etc. works as intended.
3486 *) New functions CONF_load_bio() and CONF_load_fp() to allow a config
3487 file to be loaded from a BIO or FILE pointer. The BIO version will
3488 for example allow memory BIOs to contain config info.
3491 *) New function "CRYPTO_num_locks" that returns CRYPTO_NUM_LOCKS.
3492 Whoever hopes to achieve shared-library compatibility across versions
3493 must use this, not the compile-time macro.
3494 (Exercise 0.9.4: Which is the minimum library version required by
3496 Note: All this applies only to multi-threaded programs, others don't
3500 *) Add missing case to s3_clnt.c state machine -- one of the new SSL tests
3501 through a BIO pair triggered the default case, i.e.
3502 SSLerr(...,SSL_R_UNKNOWN_STATE).
3505 *) New "BIO pair" concept (crypto/bio/bss_bio.c) so that applications
3506 can use the SSL library even if none of the specific BIOs is
3510 *) Fix a bug in i2d_DSAPublicKey() which meant it returned the wrong value
3511 for the encoded length.
3512 [Jeon KyoungHo <khjeon@sds.samsung.co.kr>]
3514 *) Add initial documentation of the X509V3 functions.
3517 *) Add a new pair of functions PEM_write_PKCS8PrivateKey() and
3518 PEM_write_bio_PKCS8PrivateKey() that are equivalent to
3519 PEM_write_PrivateKey() and PEM_write_bio_PrivateKey() but use the more
3520 secure PKCS#8 private key format with a high iteration count.
3523 *) Fix determination of Perl interpreter: A perl or perl5
3524 _directory_ in $PATH was also accepted as the interpreter.
3525 [Ralf S. Engelschall]
3527 *) Fix demos/sign/sign.c: well there wasn't anything strictly speaking
3528 wrong with it but it was very old and did things like calling
3529 PEM_ASN1_read() directly and used MD5 for the hash not to mention some
3533 *) Fix demos/selfsign.c: it used obsolete and deleted functions, changed
3534 to use the new extension code.
3537 *) Implement the PEM_read/PEM_write functions in crypto/pem/pem_all.c
3538 with macros. This should make it easier to change their form, add extra
3539 arguments etc. Fix a few PEM prototypes which didn't have cipher as a
3543 *) Add to configuration table a new entry that can specify an alternative
3544 name for unistd.h (for pre-POSIX systems); we need this for NeXTstep,
3545 according to Mark Crispin <MRC@Panda.COM>.
3549 *) DES CBC did not update the IV. Weird.
3552 des_cbc_encrypt does not update the IV, but des_ncbc_encrypt does.
3553 Changing the behaviour of the former might break existing programs --
3554 where IV updating is needed, des_ncbc_encrypt can be used.
3557 *) When bntest is run from "make test" it drives bc to check its
3558 calculations, as well as internally checking them. If an internal check
3559 fails, it needs to cause bc to give a non-zero result or make test carries
3560 on without noticing the failure. Fixed.
3563 *) DES library cleanups.
3566 *) Add support for PKCS#5 v2.0 PBE algorithms. This will permit PKCS#8 to be
3567 used with any cipher unlike PKCS#5 v1.5 which can at most handle 64 bit
3568 ciphers. NOTE: although the key derivation function has been verified
3569 against some published test vectors it has not been extensively tested
3570 yet. Added a -v2 "cipher" option to pkcs8 application to allow the use
3574 *) Instead of "mkdir -p", which is not fully portable, use new
3575 Perl script "util/mkdir-p.pl".
3578 *) Rewrite the way password based encryption (PBE) is handled. It used to
3579 assume that the ASN1 AlgorithmIdentifier parameter was a PBEParameter
3580 structure. This was true for the PKCS#5 v1.5 and PKCS#12 PBE algorithms
3581 but doesn't apply to PKCS#5 v2.0 where it can be something else. Now
3582 the 'parameter' field of the AlgorithmIdentifier is passed to the
3583 underlying key generation function so it must do its own ASN1 parsing.
3584 This has also changed the EVP_PBE_CipherInit() function which now has a
3585 'parameter' argument instead of literal salt and iteration count values
3586 and the function EVP_PBE_ALGOR_CipherInit() has been deleted.
3589 *) Support for PKCS#5 v1.5 compatible password based encryption algorithms
3590 and PKCS#8 functionality. New 'pkcs8' application linked to openssl.
3591 Needed to change the PEM_STRING_EVP_PKEY value which was just "PRIVATE
3592 KEY" because this clashed with PKCS#8 unencrypted string. Since this
3593 value was just used as a "magic string" and not used directly its
3594 value doesn't matter.
3597 *) Introduce some semblance of const correctness to BN. Shame C doesn't
3601 *) "linux-sparc64" configuration (ultrapenguin).
3602 [Ray Miller <ray.miller@oucs.ox.ac.uk>]
3603 "linux-sparc" configuration.
3604 [Christian Forster <fo@hawo.stw.uni-erlangen.de>]
3606 *) config now generates no-xxx options for missing ciphers.
3609 *) Support the EBCDIC character set (work in progress).
3610 File ebcdic.c not yet included because it has a different license.
3611 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>]
3613 *) Support BS2000/OSD-POSIX.
3614 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>]
3616 *) Make callbacks for key generation use void * instead of char *.
3619 *) Make S/MIME samples compile (not yet tested).
3622 *) Additional typesafe stacks.
3625 *) New configuration variants "bsdi-elf-gcc" (BSD/OS 4.x).
3629 Changes between 0.9.3 and 0.9.3a [29 May 1999]
3631 *) New configuration variant "sco5-gcc".
3633 *) Updated some demos.
3634 [Sean O Riordain, Wade Scholine]
3636 *) Add missing BIO_free at exit of pkcs12 application.
3639 *) Fix memory leak in conf.c.
3642 *) Updates for Win32 to assembler version of MD5.
3645 *) Set #! path to perl in apps/der_chop to where we found it
3646 instead of using a fixed path.
3649 *) SHA library changes for irix64-mips4-cc.
3652 *) Improvements for VMS support.
3656 Changes between 0.9.2b and 0.9.3 [24 May 1999]
projects / openssl.git / blob