5 Changes between 0.9.6 and 0.9.7 [xx XXX 2001]
7 OpenSSL 0.9.6a/0.9.6b (bugfix releases, 5 Apr 2001 and 9 July 2001)
8 and OpenSSL 0.9.7 were developped in parallel, based on OpenSSL 0.9.6.
10 Change log entries are tagged as follows:
11 -) applies to 0.9.6a/0.9.6b/0.9.6c only
12 *) applies to 0.9.6a/0.9.6b/0.9.6c and 0.9.7
13 +) applies to 0.9.7 only
15 +) Rationalise EVP so it can be extended: don't include a union of
16 cipher/digest structures, add init/cleanup functions. This also reduces
17 the number of header dependencies.
20 +) Make DES key schedule conform to the usual scheme, as well as
21 correcting its structure. This means that calls to DES functions
22 now have to pass a pointer to a des_key_schedule instead of a
23 plain des_key_schedule (which was actually always a pointer
27 +) Enhanced support for IA-64 Unix platforms (well, Linux and HP-UX).
30 *) Modified SSL library such that the verify_callback that has been set
31 specificly for an SSL object with SSL_set_verify() is actually being
32 used. Before the change, a verify_callback set with this function was
33 ignored and the verify_callback() set in the SSL_CTX at the time of
34 the call was used. New function X509_STORE_CTX_set_verify_cb() introduced
35 to allow the necessary settings.
38 +) Initial reduction of linker bloat: the use of some functions, such as
39 PEM causes large amounts of unused functions to be linked in due to
40 poor organisation. For example pem_all.c contains every PEM function
41 which has a knock on effect of linking in large amounts of (unused)
42 ASN1 code. Grouping together similar functions and splitting unrelated
43 functions prevents this.
46 *) Initialize static variable in crypto/dsa/dsa_lib.c and crypto/dh/dh_lib.c
47 explicitely to NULL, as at least on Solaris 8 this seems not always to be
48 done automatically (in contradiction to the requirements of the C
49 standard). This made problems when used from OpenSSH.
52 *) In OpenSSL 0.9.6a and 0.9.6b, crypto/dh/dh_key.c ignored
53 dh->length and always used
55 BN_rand_range(priv_key, dh->p).
57 BN_rand_range() is not necessary for Diffie-Hellman, and this
58 specific range makes Diffie-Hellman unnecessarily inefficient if
59 dh->length (recommended exponent length) is much smaller than the
60 length of dh->p. We could use BN_rand_range() if the order of
61 the subgroup was stored in the DH structure, but we only have
66 BN_rand(priv_key, l, ...)
68 where 'l' is dh->length if this is defined, or BN_num_bits(dh->p)-1
74 RSA_eay_public_encrypt
75 RSA_eay_private_decrypt
76 RSA_eay_private_encrypt (signing)
77 RSA_eay_public_decrypt (signature verification)
79 (default implementations for RSA_public_encrypt,
80 RSA_private_decrypt, RSA_private_encrypt, RSA_public_decrypt),
81 always reject numbers >= n.
84 *) In crypto/rand/md_rand.c, use a new short-time lock CRYPTO_LOCK_RAND2
85 to synchronize access to 'locking_thread'. This is necessary on
86 systems where access to 'locking_thread' (an 'unsigned long'
87 variable) is not atomic.
90 *) In crypto/rand/md_rand.c, set 'locking_thread' to current thread's ID
91 *before* setting the 'crypto_lock_rand' flag. The previous code had
92 a race condition if 0 is a valid thread ID.
93 [Travis Vitek <vitek@roguewave.com>]
95 +) Cleanup of EVP macros.
98 +) Change historical references to {NID,SN,LN}_des_ede and ede3 to add the
102 +) Add initial OCSP responder support to ocsp application. The
103 revocation information is handled using the text based index
104 use by the ca application. The responder can either handle
105 requests generated internally, supplied in files (for example
106 via a CGI script) or using an internal minimal server.
109 +) Add configuration choices to get zlib compression for TLS.
112 +) Changes to Kerberos SSL for RFC 2712 compliance:
113 1. Implemented real KerberosWrapper, instead of just using
114 KRB5 AP_REQ message. [Thanks to Simon Wilkinson <sxw@sxw.org.uk>]
115 2. Implemented optional authenticator field of KerberosWrapper.
117 Added openssl-style ASN.1 macros for Kerberos ticket, ap_req,
118 and authenticator structs; see crypto/krb5/.
120 Generalized Kerberos calls to support multiple Kerberos libraries.
121 [Vern Staats <staatsvr@asc.hpc.mil>,
122 Jeffrey Altman <jaltman@columbia.edu>
125 +) Cause 'openssl speed' to use fully hard-coded DSA keys as it
126 already does with RSA. testdsa.h now has 'priv_key/pub_key'
127 values for each of the key sizes rather than having just
128 parameters (and 'speed' generating keys each time).
131 -) OpenSSL 0.9.6b released [9 July 2001]
133 *) Change ssleay_rand_bytes (crypto/rand/md_rand.c)
134 to avoid a SSLeay/OpenSSL PRNG weakness pointed out by
135 Markku-Juhani O. Saarinen <markku-juhani.saarinen@nokia.com>:
136 PRNG state recovery was possible based on the output of
137 one PRNG request appropriately sized to gain knowledge on
138 'md' followed by enough consecutive 1-byte PRNG requests
139 to traverse all of 'state'.
141 1. When updating 'md_local' (the current thread's copy of 'md')
142 during PRNG output generation, hash all of the previous
143 'md_local' value, not just the half used for PRNG output.
145 2. Make the number of bytes from 'state' included into the hash
146 independent from the number of PRNG bytes requested.
148 The first measure alone would be sufficient to avoid
149 Markku-Juhani's attack. (Actually it had never occurred
150 to me that the half of 'md_local' used for chaining was the
151 half from which PRNG output bytes were taken -- I had always
152 assumed that the secret half would be used.) The second
153 measure makes sure that additional data from 'state' is never
154 mixed into 'md_local' in small portions; this heuristically
155 further strengthens the PRNG.
158 +) Speed up EVP routines.
161 type 8 bytes 64 bytes 256 bytes 1024 bytes 8192 bytes
162 des-cbc 4408.85k 5560.51k 5778.46k 5862.20k 5825.16k
163 des-cbc 4389.55k 5571.17k 5792.23k 5846.91k 5832.11k
164 des-cbc 4394.32k 5575.92k 5807.44k 5848.37k 5841.30k
166 des-cbc 3482.66k 5069.49k 5496.39k 5614.16k 5639.28k
167 des-cbc 3480.74k 5068.76k 5510.34k 5609.87k 5635.52k
168 des-cbc 3483.72k 5067.62k 5504.60k 5708.01k 5724.80k
171 des-cbc 4660.16k 5650.19k 5807.19k 5827.13k 5783.32k
173 des-cbc 3624.96k 5258.21k 5530.91k 5624.30k 5628.26k
176 *) Fix crypto/bn/asm/mips3.s.
179 *) When only the key is given to "enc", the IV is undefined. Print out
180 an error message in this case.
183 +) Added the OS2-EMX target.
184 ["Brian Havard" <brianh@kheldar.apana.org.au> and Richard Levitte]
186 +) Rewrite apps to use NCONF routines instead of the old CONF. New functions
187 to support NCONF routines in extension code. New function CONF_set_nconf()
188 to allow functions which take an NCONF to also handle the old LHASH
189 structure: this means that the old CONF compatible routines can be
190 retained (in particular wrt extensions) without having to duplicate the
191 code. New function X509V3_add_ext_nconf_sk to add extensions to a stack.
194 *) Handle special case when X509_NAME is empty in X509 printing routines.
197 *) In dsa_do_verify (crypto/dsa/dsa_ossl.c), verify that r and s are
198 positive and less than q.
201 +) Enhance the general user interface with mechanisms for inner control
202 and with pssibilities to have yes/no kind of prompts.
205 +) Change all calls to low level digest routines in the library and
206 applications to use EVP. Add missing calls to HMAC_cleanup() and
207 don't assume HMAC_CTX can be copied using memcpy().
208 [Verdon Walker <VWalker@novell.com>, Steve Henson]
210 +) Add the possibility to control engines through control names but with
211 arbitrary arguments instead of just a string.
212 Change the key loaders to take a UI_METHOD instead of a callback
213 function pointer. NOTE: this breaks binary compatibility with earlier
214 versions of OpenSSL [engine].
215 Addapt the nCipher code for these new conditions and add a card insertion
219 +) Enhance the general user interface with mechanisms to better support
220 dialog box interfaces, application-defined prompts, the possibility
221 to use defaults (for example default passwords from somewhere else)
222 and interrupts/cancelations.
225 *) Don't change *pointer in CRYPTO_add_lock() is add_lock_callback is
226 used: it isn't thread safe and the add_lock_callback should handle
228 [Paul Rose <Paul.Rose@bridge.com>]
230 *) Verify that incoming data obeys the block size in
231 ssl3_enc (ssl/s3_enc.c) and tls1_enc (ssl/t1_enc.c).
234 +) Tidy up PKCS#12 attribute handling. Add support for the CSP name
235 attribute in PKCS#12 files, add new -CSP option to pkcs12 utility.
239 [Ulf Möller, Bodo Möller]
241 *) The countermeasure against Bleichbacher's attack on PKCS #1 v1.5
242 RSA encryption was accidentily removed in s3_srvr.c in OpenSSL 0.9.5
243 when fixing the server behaviour for backwards-compatible 'client
244 hello' messages. (Note that the attack is impractical against
245 SSL 3.0 and TLS 1.0 anyway because length and version checking
246 means that the probability of guessing a valid ciphertext is
247 around 2^-40; see section 5 in Bleichenbacher's CRYPTO '98
250 Before 0.9.5, the countermeasure (hide the error by generating a
251 random 'decryption result') did not work properly because
252 ERR_clear_error() was missing, meaning that SSL_get_error() would
253 detect the supposedly ignored error.
255 Both problems are now fixed.
258 *) In crypto/bio/bf_buff.c, increase DEFAULT_BUFFER_SIZE to 4096
259 (previously it was 1024).
262 +) Fix a memory leak in 'sk_dup()' in the case reallocation fails. (Also
263 tidy up some unecessarily weird code in 'sk_new()').
264 [Geoff, reported by Diego Tartara <dtartara@novamens.com>]
266 +) Change the key loading routines for ENGINEs to use the same kind
267 callback (pem_password_cb) as all other routines that need this
271 *) Fix for compatibility mode trust settings: ignore trust settings
272 unless some valid trust or reject settings are present.
275 *) Fix for blowfish EVP: its a variable length cipher.
278 +) Increase ENTROPY_NEEDED to 32 bytes, as Rijndael can operate with
279 256 bit (=32 byte) keys. Of course seeding with more entropy bytes
280 than this minimum value is recommended.
283 +) New random seeder for OpenVMS, using the system process statistics
284 that are easily reachable.
287 +) Windows apparently can't transparently handle global
288 variables defined in DLLs. Initialisations such as:
290 const ASN1_ITEM *it = &ASN1_INTEGER_it;
292 wont compile. This is used by the any applications that need to
293 delcare their own ASN1 modules. This was fixed by adding the option
294 EXPORT_VAR_AS_FN to all Win32 platforms, although this isn't strictly
295 needed for static libraries under Win32.
298 +) New functions X509_PURPOSE_set() and X509_TRUST_set() to handle
299 setting of purpose and trust fields. New X509_STORE trust and
300 purpose functions and tidy up setting in other SSL functions.
303 +) Add copies of X509_STORE_CTX fields and callbacks to X509_STORE
304 structure. These are inherited by X509_STORE_CTX when it is
305 initialised. This allows various defaults to be set in the
306 X509_STORE structure (such as flags for CRL checking and custom
307 purpose or trust settings) for functions which only use X509_STORE_CTX
308 internally such as S/MIME.
310 Modify X509_STORE_CTX_purpose_inherit() so it only sets purposes and
311 trust settings if they are not set in X509_STORE. This allows X509_STORE
312 purposes and trust (in S/MIME for example) to override any set by default.
314 Add command line options for CRL checking to smime, s_client and s_server
318 +) Initial CRL based revocation checking. If the CRL checking flag(s)
319 are set then the CRL is looked up in the X509_STORE structure and
320 its validity and signature checked, then if the certificate is found
321 in the CRL the verify fails with a revoked error.
323 Various new CRL related callbacks added to X509_STORE_CTX structure.
325 Command line options added to 'verify' application to support this.
327 This needs some additional work, such as being able to handle multiple
328 CRLs with different times, extension based lookup (rather than just
329 by subject name) and ultimately more complete V2 CRL extension
333 +) Add a general user interface API (crypto/ui/). This is designed
334 to replace things like des_read_password and friends (backward
335 compatibility functions using this new API are provided).
336 The purpose is to remove prompting functions from the DES code
337 section as well as provide for prompting through dialog boxes in
338 a window system and the like.
341 *) Fix various bugs related to DSA S/MIME verification. Handle missing
342 parameters in DSA public key structures and return an error in the
343 DSA routines if parameters are absent.
346 *) In versions up to 0.9.6, RAND_file_name() resorted to file ".rnd"
347 in the current directory if neither $RANDFILE nor $HOME was set.
348 RAND_file_name() in 0.9.6a returned NULL in this case. This has
349 caused some confusion to Windows users who haven't defined $HOME.
350 Thus RAND_file_name() is changed again: e_os.h can define a
351 DEFAULT_HOME, which will be used if $HOME is not set.
352 For Windows, we use "C:"; on other platforms, we still require
353 environment variables.
355 +) Add "ex_data" support to ENGINE so implementations can add state at a
356 per-structure level rather than having to store it globally.
359 +) Make it possible for ENGINE structures to be copied when retrieved by
360 ENGINE_by_id() if the ENGINE specifies a new flag: ENGINE_FLAGS_BY_ID_COPY.
361 This causes the "original" ENGINE structure to act like a template,
362 analogous to the RSA vs. RSA_METHOD type of separation. Because of this
363 operational state can be localised to each ENGINE structure, despite the
364 fact they all share the same "methods". New ENGINE structures returned in
365 this case have no functional references and the return value is the single
366 structural reference. This matches the single structural reference returned
367 by ENGINE_by_id() normally, when it is incremented on the pre-existing
371 +) Fix ASN1 decoder when decoding type ANY and V_ASN1_OTHER: since this
372 needs to match any other type at all we need to manually clear the
376 +) Changes to the "openssl engine" utility to include;
377 - verbosity levels ('-v', '-vv', and '-vvv') that provide information
378 about an ENGINE's available control commands.
379 - executing control commands from command line arguments using the
380 '-pre' and '-post' switches. '-post' is only used if '-t' is
381 specified and the ENGINE is successfully initialised. The syntax for
382 the individual commands are colon-separated, for example;
383 openssl engine chil -pre FORK_CHECK:0 -pre SO_PATH:/lib/test.so
386 +) New dynamic control command support for ENGINEs. ENGINEs can now
387 declare their own commands (numbers), names (strings), descriptions,
388 and input types for run-time discovery by calling applications. A
389 subset of these commands are implicitly classed as "executable"
390 depending on their input type, and only these can be invoked through
391 the new string-based API function ENGINE_ctrl_cmd_string(). (Eg. this
392 can be based on user input, config files, etc). The distinction is
393 that "executable" commands cannot return anything other than a boolean
394 result and can only support numeric or string input, whereas some
395 discoverable commands may only be for direct use through
396 ENGINE_ctrl(), eg. supporting the exchange of binary data, function
397 pointers, or other custom uses. The "executable" commands are to
398 support parameterisations of ENGINE behaviour that can be
399 unambiguously defined by ENGINEs and used consistently across any
400 OpenSSL-based application. Commands have been added to all the
401 existing hardware-supporting ENGINEs, noticeably "SO_PATH" to allow
402 control over shared-library paths without source code alterations.
405 +) Changed all ENGINE implementations to dynamically allocate their
406 ENGINEs rather than declaring them statically. Apart from this being
407 necessary with the removal of the ENGINE_FLAGS_MALLOCED distinction,
408 this also allows the implementations to compile without using the
409 internal engine_int.h header.
412 +) Minor adjustment to "rand" code. RAND_get_rand_method() now returns a
413 'const' value. Any code that should be able to modify a RAND_METHOD
414 should already have non-const pointers to it (ie. they should only
415 modify their own ones).
418 +) Made a variety of little tweaks to the ENGINE code.
419 - "atalla" and "ubsec" string definitions were moved from header files
420 to C code. "nuron" string definitions were placed in variables
421 rather than hard-coded - allowing parameterisation of these values
422 later on via ctrl() commands.
423 - Removed unused "#if 0"'d code.
424 - Fixed engine list iteration code so it uses ENGINE_free() to release
425 structural references.
426 - Constified the RAND_METHOD element of ENGINE structures.
427 - Constified various get/set functions as appropriate and added
428 missing functions (including a catch-all ENGINE_cpy that duplicates
429 all ENGINE values onto a new ENGINE except reference counts/state).
430 - Removed NULL parameter checks in get/set functions. Setting a method
431 or function to NULL is a way of cancelling out a previously set
432 value. Passing a NULL ENGINE parameter is just plain stupid anyway
433 and doesn't justify the extra error symbols and code.
434 - Deprecate the ENGINE_FLAGS_MALLOCED define and move the area for
435 flags from engine_int.h to engine.h.
436 - Changed prototypes for ENGINE handler functions (init(), finish(),
437 ctrl(), key-load functions, etc) to take an (ENGINE*) parameter.
440 *) Move 'if (!initialized) RAND_poll()' into regions protected by
441 CRYPTO_LOCK_RAND. This is not strictly necessary, but avoids
442 having multiple threads call RAND_poll() concurrently.
445 *) In crypto/rand/md_rand.c, replace 'add_do_not_lock' flag by a
446 combination of a flag and a thread ID variable.
447 Otherwise while one thread is in ssleay_rand_bytes (which sets the
448 flag), *other* threads can enter ssleay_add_bytes without obeying
449 the CRYPTO_LOCK_RAND lock (and may even illegaly release the lock
450 that they do not hold after the first thread unsets add_do_not_lock).
453 +) Implement binary inversion algorithm for BN_mod_inverse in addition
454 to the algorithm using long divison. The binary algorithm can be
455 used only if the modulus is odd. On 32-bit systems, it is faster
456 only for relatively small moduli (roughly 20-30% for 128-bit moduli,
457 roughly 5-15% for 256-bit moduli), so we use it only for moduli
458 up to 450 bits. In 64-bit environments, the binary algorithm
459 appears to be advantageous for much longer moduli; here we use it
460 for moduli up to 2048 bits.
463 *) Change bctest again: '-x' expressions are not available in all
467 -) OpenSSL 0.9.6a released [5 Apr 2001]
469 *) Fix a couple of memory leaks in PKCS7_dataDecode()
470 [Steve Henson, reported by Heyun Zheng <hzheng@atdsprint.com>]
472 *) Change Configure and Makefiles to provide EXE_EXT, which will contain
473 the default extension for executables, if any. Also, make the perl
474 scripts that use symlink() to test if it really exists and use "cp"
475 if it doesn't. All this made OpenSSL compilable and installable in
479 +) Rewrite CHOICE field setting in ASN1_item_ex_d2i(). The old code
480 could not support the combine flag in choice fields.
483 -) Fix for asn1_GetSequence() for indefinite length constructed data.
484 If SEQUENCE is length is indefinite just set c->slen to the total
485 amount of data available.
486 [Steve Henson, reported by shige@FreeBSD.org]
487 [This change does not apply to 0.9.7.]
489 *) Change bctest to avoid here-documents inside command substitution
490 (workaround for FreeBSD /bin/sh bug).
491 For compatibility with Ultrix, avoid shell functions (introduced
492 in the bctest version that searches along $PATH).
495 *) Rename 'des_encrypt' to 'des_encrypt1'. This avoids the clashes
496 with des_encrypt() defined on some operating systems, like Solaris
500 *) Check the result of RSA-CRT (see D. Boneh, R. DeMillo, R. Lipton:
501 On the Importance of Eliminating Errors in Cryptographic
502 Computations, J. Cryptology 14 (2001) 2, 101-119,
503 http://theory.stanford.edu/~dabo/papers/faults.ps.gz).
506 *) MIPS assembler BIGNUM division bug fix.
509 *) Disabled incorrect Alpha assembler code.
512 -) Fix PKCS#7 decode routines so they correctly update the length
513 after reading an EOC for the EXPLICIT tag.
515 [This change does not apply to 0.9.7.]
517 *) Fix bug in PKCS#12 key generation routines. This was triggered
518 if a 3DES key was generated with a 0 initial byte. Include
519 PKCS12_BROKEN_KEYGEN compilation option to retain the old
520 (but broken) behaviour.
523 *) Enhance bctest to search for a working bc along $PATH and print
525 [Tim Rice <tim@multitalents.net> via Richard Levitte]
527 +) Add a 'copy_extensions' option to the 'ca' utility. This copies
528 extensions from a certificate request to the certificate.
531 +) Allow multiple 'certopt' and 'nameopt' options to be separated
532 by commas. Add 'namopt' and 'certopt' options to the 'ca' config
533 file: this allows the display of the certificate about to be
534 signed to be customised, to allow certain fields to be included
535 or excluded and extension details. The old system didn't display
536 multicharacter strings properly, omitted fields not in the policy
537 and couldn't display additional details such as extensions.
540 *) Fix memory leaks in err.c: free err_data string if necessary;
541 don't write to the wrong index in ERR_set_error_data.
544 +) Function EC_POINTs_mul for simultaneous scalar multiplication
545 of an arbitrary number of elliptic curve points, optionally
546 including the generator defined for the EC_GROUP.
547 EC_POINT_mul is a simple wrapper function for the typical case
548 that the point list has just one item (besides the optional
552 +) First EC_METHODs for curves over GF(p):
554 EC_GFp_simple_method() uses the basic BN_mod_mul and BN_mod_sqr
555 operations and provides various method functions that can also
556 operate with faster implementations of modular arithmetic.
558 EC_GFp_mont_method() reuses most functions that are part of
559 EC_GFp_simple_method, but uses Montgomery arithmetic.
561 [Bodo Moeller; point addition and point doubling
562 implementation directly derived from source code provided by
563 Lenka Fibikova <fibikova@exp-math.uni-essen.de>]
565 +) Framework for elliptic curves (crypto/ec/ec.h, crypto/ec/ec_lcl.h,
568 Curves are EC_GROUP objects (with an optional group generator)
569 based on EC_METHODs that are built into the library.
571 Points are EC_POINT objects based on EC_GROUP objects.
573 Most of the framework would be able to handle curves over arbitrary
574 finite fields, but as there are no obvious types for fields other
575 than GF(p), some functions are limited to that for now.
578 +) Add the -HTTP option to s_server. It is similar to -WWW, but requires
579 that the file contains a complete HTTP response.
582 +) Add the ec directory to mkdef.pl and mkfiles.pl. In mkdef.pl
583 change the def and num file printf format specifier from "%-40sXXX"
584 to "%-39s XXX". The latter will always guarantee a space after the
585 field while the former will cause them to run together if the field
586 is 40 of more characters long.
589 +) Constify the cipher and digest 'method' functions and structures
590 and modify related functions to take constant EVP_MD and EVP_CIPHER
594 *) Implement ssl23_peek (analogous to ssl23_read), which previously
598 *) Replace rdtsc with _emit statements for VC++ version 5.
599 [Jeremy Cooper <jeremy@baymoo.org>]
601 +) Hide BN_CTX structure details in bn_lcl.h instead of publishing them
602 in <openssl/bn.h>. Also further increase BN_CTX_NUM to 32.
605 +) Modify EVP_Digest*() routines so they now return values. Although the
606 internal software routines can never fail additional hardware versions
610 +) Clean up crypto/err/err.h and change some error codes to avoid conflicts:
612 Previously ERR_R_FATAL was too small and coincided with ERR_LIB_PKCS7
613 (= ERR_R_PKCS7_LIB); it is now 64 instead of 32.
616 ERR_R_NESTED_ASN1_ERROR
618 ERR_R_MISSING_ASN1_EOS
619 were 4 .. 9, conflicting with
620 ERR_LIB_RSA (= ERR_R_RSA_LIB)
622 ERR_LIB_PEM (= ERR_R_PEM_LIB).
623 They are now 58 .. 63 (i.e., just below ERR_R_FATAL).
625 Add new error code 'ERR_R_INTERNAL_ERROR'.
628 +) Don't overuse locks in crypto/err/err.c: For data retrieval, CRYPTO_r_lock
632 +) New option '-subj arg' for 'openssl req' and 'openssl ca'. This
633 sets the subject name for a new request or supersedes the
634 subject name in a given request. Formats that can be parsed are
635 'CN=Some Name, OU=myOU, C=IT'
637 'CN=Some Name/OU=myOU/C=IT'.
639 Add options '-batch' and '-verbose' to 'openssl req'.
640 [Massimiliano Pala <madwolf@hackmasters.net>]
642 *) Make it possible to reuse SSLv2 sessions.
645 +) Introduce the possibility to access global variables through
646 functions on platform were that's the best way to handle exporting
647 global variables in shared libraries. To enable this functionality,
648 one must configure with "EXPORT_VAR_AS_FN" or defined the C macro
649 "OPENSSL_EXPORT_VAR_AS_FUNCTION" in crypto/opensslconf.h (the latter
650 is normally done by Configure or something similar).
652 To implement a global variable, use the macro OPENSSL_IMPLEMENT_GLOBAL
653 in the source file (foo.c) like this:
655 OPENSSL_IMPLEMENT_GLOBAL(int,foo)=1;
656 OPENSSL_IMPLEMENT_GLOBAL(double,bar);
658 To declare a global variable, use the macros OPENSSL_DECLARE_GLOBAL
659 and OPENSSL_GLOBAL_REF in the header file (foo.h) like this:
661 OPENSSL_DECLARE_GLOBAL(int,foo);
662 #define foo OPENSSL_GLOBAL_REF(foo)
663 OPENSSL_DECLARE_GLOBAL(double,bar);
664 #define bar OPENSSL_GLOBAL_REF(bar)
666 The #defines are very important, and therefore so is including the
667 header file everywere where the defined globals are used.
669 The macro OPENSSL_EXPORT_VAR_AS_FUNCTION also affects the definition
670 of ASN.1 items, but that structure is a bt different.
672 The largest change is in util/mkdef.pl which has been enhanced with
673 better and easier to understand logic to choose which symbols should
674 go into the Windows .def files as well as a number of fixes and code
675 cleanup (among others, algorithm keywords are now sorted
676 lexicographically to avoid constant rewrites).
679 *) In copy_email() check for >= 0 as a return value for
680 X509_NAME_get_index_by_NID() since 0 is a valid index.
681 [Steve Henson reported by Massimiliano Pala <madwolf@opensca.org>]
683 +) In BN_div() keep a copy of the sign of 'num' before writing the
684 result to 'rm' because if rm==num the value will be overwritten
685 and produce the wrong result if 'num' is negative: this caused
686 problems with BN_mod() and BN_nnmod().
689 +) Function OCSP_request_verify(). This checks the signature on an
690 OCSP request and verifies the signer certificate. The signer
691 certificate is just checked for a generic purpose and OCSP request
695 +) Add OCSP_check_validity() function to check the validity of OCSP
696 responses. OCSP responses are prepared in real time and may only
697 be a few seconds old. Simply checking that the current time lies
698 between thisUpdate and nextUpdate max reject otherwise valid responses
699 caused by either OCSP responder or client clock innacuracy. Instead
700 we allow thisUpdate and nextUpdate to fall within a certain period of
701 the current time. The age of the response can also optionally be
702 checked. Two new options -validity_period and -status_age added to
706 +) If signature or public key algorithm is unrecognized print out its
707 OID rather that just UNKOWN.
710 *) Avoid coredump with unsupported or invalid public keys by checking if
711 X509_get_pubkey() fails in PKCS7_verify(). Fix memory leak when
712 PKCS7_verify() fails with non detached data.
715 +) Change OCSP_cert_to_id() to tolerate a NULL subject certificate and
716 OCSP_cert_id_new() a NULL serialNumber. This allows a partial certificate
717 ID to be generated from the issuer certificate alone which can then be
718 passed to OCSP_id_issuer_cmp().
721 +) New compilation option ASN1_ITEM_FUNCTIONS. This causes the new
722 ASN1 modules to export functions returning ASN1_ITEM pointers
723 instead of the ASN1_ITEM structures themselves. This adds several
724 new macros which allow the underlying ASN1 function/structure to
725 be accessed transparently. As a result code should not use ASN1_ITEM
726 references directly (such as &X509_it) but instead use the relevant
727 macros (such as ASN1_ITEM_rptr(X509)). This option is to allow
728 use of the new ASN1 code on platforms where exporting structures
729 is problematical (for example in shared libraries) but exporting
730 functions returning pointers to structures is not.
733 +) Add support for overriding the generation of SSL/TLS session IDs.
734 These callbacks can be registered either in an SSL_CTX or per SSL.
735 The purpose of this is to allow applications to control, if they wish,
736 the arbitrary values chosen for use as session IDs, particularly as it
737 can be useful for session caching in multiple-server environments. A
738 command-line switch for testing this (and any client code that wishes
739 to use such a feature) has been added to "s_server".
740 [Geoff Thorpe, Lutz Jaenicke]
742 +) Modify mkdef.pl to recognise and parse prprocessor conditionals
743 of the form '#if defined(...) || defined(...) || ...' and
744 '#if !defined(...) && !defined(...) && ...'. This also avoids
745 the growing number of special cases it was previously handling.
748 *) Don't use getenv in library functions when run as setuid/setgid.
749 New function OPENSSL_issetugid().
752 +) Make all configuration macros available for application by making
753 sure they are available in opensslconf.h, by giving them names starting
754 with "OPENSSL_" to avoid conflicts with other packages and by making
755 sure e_os2.h will cover all platform-specific cases together with
757 Additionally, it is now possible to define configuration/platform-
758 specific names (called "system identities"). In the C code, these
759 are prefixed with "OPENSSL_SYSNAME_". e_os2.h will create another
760 macro with the name beginning with "OPENSSL_SYS_", which is determined
761 from "OPENSSL_SYSNAME_*" or compiler-specific macros depending on
765 +) New option -set_serial to 'req' and 'x509' this allows the serial
766 number to use to be specified on the command line. Previously self
767 signed certificates were hard coded with serial number 0 and the
768 CA options of 'x509' had to use a serial number in a file which was
772 *) Avoid false positives in memory leak detection code (crypto/mem_dbg.c)
773 due to incorrect handling of multi-threading:
775 1. Fix timing glitch in the MemCheck_off() portion of CRYPTO_mem_ctrl().
777 2. Fix logical glitch in is_MemCheck_on() aka CRYPTO_is_mem_check_on().
779 3. Count how many times MemCheck_off() has been called so that
780 nested use can be treated correctly. This also avoids
781 inband-signalling in the previous code (which relied on the
782 assumption that thread ID 0 is impossible).
785 +) New options to 'ca' utility to support V2 CRL entry extensions.
786 Currently CRL reason, invalidity date and hold instruction are
787 supported. Add new CRL extensions to V3 code and some new objects.
790 *) Add "-rand" option also to s_client and s_server.
793 +) New function EVP_CIPHER_CTX_set_padding() this is used to
794 disable standard block padding (aka PKCS#5 padding) in the EVP
795 API, which was previously mandatory. This means that the data is
796 not padded in any way and so the total length much be a multiple
797 of the block size, otherwise an error occurs.
800 +) Initial (incomplete) OCSP SSL support.
803 *) Fix CPU detection on Irix 6.x.
804 [Kurt Hockenbury <khockenb@stevens-tech.edu> and
805 "Bruce W. Forsberg" <bruce.forsberg@baesystems.com>]
807 +) New function OCSP_parse_url(). This splits up a URL into its host,
808 port and path components: primarily to parse OCSP URLs. New -url
809 option to ocsp utility.
812 +) New nonce behavior. The return value of OCSP_check_nonce() now
813 reflects the various checks performed. Applications can decide
814 whether to tolerate certain situations such as an absent nonce
815 in a response when one was present in a request: the ocsp application
816 just prints out a warning. New function OCSP_add1_basic_nonce()
817 this is to allow responders to include a nonce in a response even if
818 the request is nonce-less.
821 *) Fix X509_NAME bug which produced incorrect encoding if X509_NAME
824 [This change does not apply to 0.9.7.]
826 *) Use the cached encoding of an X509_NAME structure rather than
827 copying it. This is apparently the reason for the libsafe "errors"
828 but the code is actually correct.
831 +) Disable stdin buffering in load_cert (apps/apps.c) so that no certs are
832 skipped when using openssl x509 multiple times on a single input file,
833 e.g. "(openssl x509 -out cert1; openssl x509 -out cert2) <certs".
836 +) Make ASN1_UTCTIME_set_string() and ASN1_GENERALIZEDTIME_set_string()
837 set string type: to handle setting ASN1_TIME structures. Fix ca
838 utility to correctly initialize revocation date of CRLs.
841 +) New option SSL_OP_CIPHER_SERVER_PREFERENCE allows the server to override
842 the clients preferred ciphersuites and rather use its own preferences.
843 Should help to work around M$ SGC (Server Gated Cryptography) bug in
844 Internet Explorer by ensuring unchanged hash method during stepup.
845 (Also replaces the broken/deactivated SSL_OP_NON_EXPORT_FIRST option.)
848 +) Make mkdef.pl recognise all DECLARE_ASN1 macros, change rijndael
849 to aes and add a new 'exist' option to print out symbols that don't
853 +) Additional options to ocsp utility to allow flags to be set and
854 additional certificates supplied.
857 +) Add the option -VAfile to 'openssl ocsp', so the user can give the
858 OCSP client a number of certificate to only verify the response
862 *) Add new function BN_rand_range(), and fix DSA_sign_setup() to prevent
863 Bleichenbacher's DSA attack.
864 Extend BN_[pseudo_]rand: As before, top=1 forces the highest two bits
865 to be set and top=0 forces the highest bit to be set; top=-1 is new
866 and leaves the highest bit random.
867 [Ulf Moeller, Bodo Moeller]
869 +) Update Rijndael code to version 3.0 and change EVP AES ciphers to
870 handle the new API. Currently only ECB, CBC modes supported. Add new
871 AES OIDs. Add TLS AES ciphersuites as described in the "AES Ciphersuites
872 for TLS" draft-ietf-tls-ciphersuite-03.txt.
873 [Ben Laurie, Steve Henson]
875 *) In the NCONF_...-based implementations for CONF_... queries
876 (crypto/conf/conf_lib.c), if the input LHASH is NULL, avoid using
877 a temporary CONF structure with the data component set to NULL
878 (which gives segmentation faults in lh_retrieve).
879 Instead, use NULL for the CONF pointer in CONF_get_string and
880 CONF_get_number (which may use environment variables) and directly
881 return NULL from CONF_get_section.
884 *) Fix potential buffer overrun for EBCDIC.
887 +) New function OCSP_copy_nonce() to copy nonce value (if present) from
891 +) Functions for OCSP responders. OCSP_request_onereq_count(),
892 OCSP_request_onereq_get0(), OCSP_onereq_get0_id() and OCSP_id_get0_info()
893 extract information from a certificate request. OCSP_response_create()
894 creates a response and optionally adds a basic response structure.
895 OCSP_basic_add1_status() adds a complete single response to a basic
896 reponse and returns the OCSP_SINGLERESP structure just added (to allow
897 extensions to be included for example). OCSP_basic_add1_cert() adds a
898 certificate to a basic response and OCSP_basic_sign() signs a basic
899 response with various flags. New helper functions ASN1_TIME_check()
900 (checks validity of ASN1_TIME structure) and ASN1_TIME_to_generalizedtime()
901 (converts ASN1_TIME to GeneralizedTime).
904 +) Various new functions. EVP_Digest() combines EVP_Digest{Init,Update,Final}()
905 in a single operation. X509_get0_pubkey_bitstr() extracts the public_key
906 structure from a certificate. X509_pubkey_digest() digests tha public_key
907 contents: this is used in various key identifiers.
910 *) Tolerate nonRepudiation as being valid for S/MIME signing and certSign
911 keyUsage if basicConstraints absent for a CA.
914 *) Make SMIME_write_PKCS7() write mail header values with a format that
915 is more generally accepted (no spaces before the semicolon), since
916 some programs can't parse those values properly otherwise. Also make
917 sure BIO's that break lines after each write do not create invalid
921 +) Make sk_sort() tolerate a NULL argument.
922 [Steve Henson reported by Massimiliano Pala <madwolf@comune.modena.it>]
924 +) New OCSP verify flag OCSP_TRUSTOTHER. When set the "other" certificates
925 passed by the function are trusted implicitly. If any of them signed the
926 reponse then it is assumed to be valid and is not verified.
929 -) Make the CRL encoding routines work with empty SEQUENCE OF. The
930 macros previously used would not encode an empty SEQUENCE OF
931 and break the signature.
933 [This change does not apply to 0.9.7.]
935 *) Zero the premaster secret after deriving the master secret in
939 +) In PKCS7_set_type() initialise content_type in PKCS7_ENC_CONTENT
940 to data. This was previously part of the PKCS7 ASN1 code. This
941 was causing problems with OpenSSL created PKCS#12 and PKCS#7 structures.
942 [Steve Henson, reported by Kenneth R. Robinette
943 <support@securenetterm.com>]
945 +) Add CRYPTO_push_info() and CRYPTO_pop_info() calls to new ASN1
946 routines: without these tracing memory leaks is very painful.
947 Fix leaks in PKCS12 and PKCS7 routines.
950 *) Add some EVP_add_digest_alias registrations (as found in
951 OpenSSL_add_all_digests()) to SSL_library_init()
952 aka OpenSSL_add_ssl_algorithms(). This provides improved
953 compatibility with peers using X.509 certificates
954 with unconventional AlgorithmIdentifier OIDs.
957 *) Fix for Irix with NO_ASM.
958 ["Bruce W. Forsberg" <bruce.forsberg@baesystems.com>]
960 *) ./config script fixes.
961 [Ulf Moeller, Richard Levitte]
963 +) Make X509_time_adj() cope with the new behaviour of ASN1_TIME_new().
964 Previously it initialised the 'type' argument to V_ASN1_UTCTIME which
965 effectively meant GeneralizedTime would never be used. Now it
966 is initialised to -1 but X509_time_adj() now has to check the value
967 and use ASN1_TIME_set() if the value is not V_ASN1_UTCTIME or
968 V_ASN1_GENERALIZEDTIME, without this it always uses GeneralizedTime.
969 [Steve Henson, reported by Kenneth R. Robinette
970 <support@securenetterm.com>]
972 +) Fixes to BN_to_ASN1_INTEGER when bn is zero. This would previously
973 result in a zero length in the ASN1_INTEGER structure which was
974 not consistent with the structure when d2i_ASN1_INTEGER() was used
975 and would cause ASN1_INTEGER_cmp() to fail. Enhance s2i_ASN1_INTEGER()
976 to cope with hex and negative integers. Fix bug in i2a_ASN1_INTEGER()
977 where it did not print out a minus for negative ASN1_INTEGER.
980 *) Fix 'openssl passwd -1'.
983 +) Add summary printout to ocsp utility. The various functions which
984 convert status values to strings have been renamed to:
985 OCSP_response_status_str(), OCSP_cert_status_str() and
986 OCSP_crl_reason_str() and are no longer static. New options
987 to verify nonce values and to disable verification. OCSP response
988 printout format cleaned up.
991 +) Add additional OCSP certificate checks. These are those specified
992 in RFC2560. This consists of two separate checks: the CA of the
993 certificate being checked must either be the OCSP signer certificate
994 or the issuer of the OCSP signer certificate. In the latter case the
995 OCSP signer certificate must contain the OCSP signing extended key
996 usage. This check is performed by attempting to match the OCSP
997 signer or the OCSP signer CA to the issuerNameHash and issuerKeyHash
998 in the OCSP_CERTID structures of the response.
1001 +) Initial OCSP certificate verification added to OCSP_basic_verify()
1002 and related routines. This uses the standard OpenSSL certificate
1003 verify routines to perform initial checks (just CA validity) and
1004 to obtain the certificate chain. Then additional checks will be
1005 performed on the chain. Currently the root CA is checked to see
1006 if it is explicitly trusted for OCSP signing. This is used to set
1007 a root CA as a global signing root: that is any certificate that
1008 chains to that CA is an acceptable OCSP signing certificate.
1011 +) New '-extfile ...' option to 'openssl ca' for reading X.509v3
1012 extensions from a separate configuration file.
1013 As when reading extensions from the main configuration file,
1014 the '-extensions ...' option may be used for specifying the
1016 [Massimiliano Pala <madwolf@comune.modena.it>]
1018 *) Change PKCS12_key_gen_asc() so it can cope with non null
1019 terminated strings whose length is passed in the passlen
1020 parameter, for example from PEM callbacks. This was done
1021 by adding an extra length parameter to asc2uni().
1022 [Steve Henson, reported by <oddissey@samsung.co.kr>]
1024 +) New OCSP utility. Allows OCSP requests to be generated or
1025 read. The request can be sent to a responder and the output
1026 parsed, outputed or printed in text form. Not complete yet:
1027 still needs to check the OCSP response validity.
1030 +) New subcommands for 'openssl ca':
1031 'openssl ca -status <serial>' prints the status of the cert with
1032 the given serial number (according to the index file).
1033 'openssl ca -updatedb' updates the expiry status of certificates
1035 [Massimiliano Pala <madwolf@comune.modena.it>]
1037 +) New '-newreq-nodes' command option to CA.pl. This is like
1038 '-newreq', but calls 'openssl req' with the '-nodes' option
1039 so that the resulting key is not encrypted.
1040 [Damien Miller <djm@mindrot.org>]
1042 +) New configuration for the GNU Hurd.
1043 [Jonathan Bartlett <johnnyb@wolfram.com> via Richard Levitte]
1045 +) Initial code to implement OCSP basic response verify. This
1046 is currently incomplete. Currently just finds the signer's
1047 certificate and verifies the signature on the response.
1050 +) New SSLeay_version code SSLEAY_DIR to determine the compiled-in
1051 value of OPENSSLDIR. This is available via the new '-d' option
1052 to 'openssl version', and is also included in 'openssl version -a'.
1055 *) Fix C code generated by 'openssl dsaparam -C': If a BN_bin2bn
1056 call failed, free the DSA structure.
1059 +) Allowing defining memory allocation callbacks that will be given
1060 file name and line number information in additional arguments
1061 (a const char* and an int). The basic functionality remains, as
1062 well as the original possibility to just replace malloc(),
1063 realloc() and free() by functions that do not know about these
1064 additional arguments. To register and find out the current
1065 settings for extended allocation functions, the following
1066 functions are provided:
1068 CRYPTO_set_mem_ex_functions
1069 CRYPTO_set_locked_mem_ex_functions
1070 CRYPTO_get_mem_ex_functions
1071 CRYPTO_get_locked_mem_ex_functions
1073 These work the same way as CRYPTO_set_mem_functions and friends.
1074 CRYPTO_get_[locked_]mem_functions now writes 0 where such an
1075 extended allocation function is enabled.
1076 Similarly, CRYPTO_get_[locked_]mem_ex_functions writes 0 where
1077 a conventional allocation function is enabled.
1078 [Richard Levitte, Bodo Moeller]
1080 *) Fix to uni2asc() to cope with zero length Unicode strings.
1081 These are present in some PKCS#12 files.
1084 +) Finish off removing the remaining LHASH function pointer casts.
1085 There should no longer be any prototype-casting required when using
1086 the LHASH abstraction, and any casts that remain are "bugs". See
1087 the callback types and macros at the head of lhash.h for details
1088 (and "OBJ_cleanup" in crypto/objects/obj_dat.c as an example).
1091 +) Add automatic query of EGD sockets in RAND_poll() for the unix variant.
1092 If /dev/[u]random devices are not available or do not return enough
1093 entropy, EGD style sockets (served by EGD or PRNGD) will automatically
1095 The locations /var/run/egd-pool, /dev/egd-pool, /etc/egd-pool, and
1096 /etc/entropy will be queried once each in this sequence, quering stops
1097 when enough entropy was collected without querying more sockets.
1100 +) Change the Unix RAND_poll() variant to be able to poll several
1101 random devices, as specified by DEVRANDOM, until a sufficient amount
1102 of data has been collected. We spend at most 10 ms on each file
1103 (select timeout) and read in non-blocking mode. DEVRANDOM now
1104 defaults to the list "/dev/urandom", "/dev/random", "/dev/srandom"
1105 (previously it was just the string "/dev/urandom"), so on typical
1106 platforms the 10 ms delay will never occur.
1107 Also separate out the Unix variant to its own file, rand_unix.c.
1108 For VMS, there's a currently-empty rand_vms.c.
1111 +) Move OCSP client related routines to ocsp_cl.c. These
1112 provide utility functions which an application needing
1113 to issue a request to an OCSP responder and analyse the
1114 response will typically need: as opposed to those which an
1115 OCSP responder itself would need which will be added later.
1117 OCSP_request_sign() signs an OCSP request with an API similar
1118 to PKCS7_sign(). OCSP_response_status() returns status of OCSP
1119 response. OCSP_response_get1_basic() extracts basic response
1120 from response. OCSP_resp_find_status(): finds and extracts status
1121 information from an OCSP_CERTID structure (which will be created
1122 when the request structure is built). These are built from lower
1123 level functions which work on OCSP_SINGLERESP structures but
1124 wont normally be used unless the application wishes to examine
1125 extensions in the OCSP response for example.
1127 Replace nonce routines with a pair of functions.
1128 OCSP_request_add1_nonce() adds a nonce value and optionally
1129 generates a random value. OCSP_check_nonce() checks the
1130 validity of the nonce in an OCSP response.
1133 +) Change function OCSP_request_add() to OCSP_request_add0_id().
1134 This doesn't copy the supplied OCSP_CERTID and avoids the
1135 need to free up the newly created id. Change return type
1136 to OCSP_ONEREQ to return the internal OCSP_ONEREQ structure.
1137 This can then be used to add extensions to the request.
1138 Deleted OCSP_request_new(), since most of its functionality
1139 is now in OCSP_REQUEST_new() (and the case insensitive name
1140 clash) apart from the ability to set the request name which
1141 will be added elsewhere.
1144 +) Update OCSP API. Remove obsolete extensions argument from
1145 various functions. Extensions are now handled using the new
1146 OCSP extension code. New simple OCSP HTTP function which
1147 can be used to send requests and parse the response.
1150 +) Fix the PKCS#7 (S/MIME) code to work with new ASN1. Two new
1151 ASN1_ITEM structures help with sign and verify. PKCS7_ATTR_SIGN
1152 uses the special reorder version of SET OF to sort the attributes
1153 and reorder them to match the encoded order. This resolves a long
1154 standing problem: a verify on a PKCS7 structure just after signing
1155 it used to fail because the attribute order did not match the
1156 encoded order. PKCS7_ATTR_VERIFY does not reorder the attributes:
1157 it uses the received order. This is necessary to tolerate some broken
1158 software that does not order SET OF. This is handled by encoding
1159 as a SEQUENCE OF but using implicit tagging (with UNIVERSAL class)
1160 to produce the required SET OF.
1163 +) Have mk1mf.pl generate the macros OPENSSL_BUILD_SHLIBCRYPTO and
1164 OPENSSL_BUILD_SHLIBSSL and use them appropriately in the header
1165 files to get correct declarations of the ASN.1 item variables.
1168 +) Rewrite of PKCS#12 code to use new ASN1 functionality. Replace many
1169 PKCS#12 macros with real functions. Fix two unrelated ASN1 bugs:
1170 asn1_check_tlen() would sometimes attempt to use 'ctx' when it was
1171 NULL and ASN1_TYPE was not dereferenced properly in asn1_ex_c2i().
1172 New ASN1 macro: DECLARE_ASN1_ITEM() which just declares the relevant
1173 ASN1_ITEM and no wrapper functions.
1176 +) New functions or ASN1_item_d2i_fp() and ASN1_item_d2i_bio(). These
1177 replace the old function pointer based I/O routines. Change most of
1178 the *_d2i_bio() and *_d2i_fp() functions to use these.
1181 +) Enhance mkdef.pl to be more accepting about spacing in C preprocessor
1182 lines, recognice more "algorithms" that can be deselected, and make
1183 it complain about algorithm deselection that isn't recognised.
1186 +) New ASN1 functions to handle dup, sign, verify, digest, pack and
1187 unpack operations in terms of ASN1_ITEM. Modify existing wrappers
1188 to use new functions. Add NO_ASN1_OLD which can be set to remove
1189 some old style ASN1 functions: this can be used to determine if old
1190 code will still work when these eventually go away.
1193 +) New extension functions for OCSP structures, these follow the
1194 same conventions as certificates and CRLs.
1197 +) New function X509V3_add1_i2d(). This automatically encodes and
1198 adds an extension. Its behaviour can be customised with various
1199 flags to append, replace or delete. Various wrappers added for
1200 certifcates and CRLs.
1203 +) Fix to avoid calling the underlying ASN1 print routine when
1204 an extension cannot be parsed. Correct a typo in the
1205 OCSP_SERVICELOC extension. Tidy up print OCSP format.
1208 *) Increase s2->wbuf allocation by one byte in ssl2_new (ssl/s2_lib.c).
1209 Otherwise do_ssl_write (ssl/s2_pkt.c) will write beyond buffer limits
1210 when writing a 32767 byte record.
1211 [Bodo Moeller; problem reported by Eric Day <eday@concentric.net>]
1213 *) In RSA_eay_public_{en,ed}crypt and RSA_eay_mod_exp (rsa_eay.c),
1214 obtain lock CRYPTO_LOCK_RSA before setting rsa->_method_mod_{n,p,q}.
1216 (RSA objects have a reference count access to which is protected
1217 by CRYPTO_LOCK_RSA [see rsa_lib.c, s3_srvr.c, ssl_cert.c, ssl_rsa.c],
1218 so they are meant to be shared between threads.)
1219 [Bodo Moeller, Geoff Thorpe; original patch submitted by
1220 "Reddie, Steven" <Steven.Reddie@ca.com>]
1222 +) Make mkdef.pl parse some of the ASN1 macros and add apropriate
1223 entries for variables.
1226 *) Fix a deadlock in CRYPTO_mem_leaks().
1229 +) Add functionality to apps/openssl.c for detecting locking
1230 problems: As the program is single-threaded, all we have
1231 to do is register a locking callback using an array for
1232 storing which locks are currently held by the program.
1235 +) Use a lock around the call to CRYPTO_get_ex_new_index() in
1236 SSL_get_ex_data_X509_STORE_idx(), which is used in
1237 ssl_verify_cert_chain() and thus can be called at any time
1238 during TLS/SSL handshakes so that thread-safety is essential.
1239 Unfortunately, the ex_data design is not at all suited
1240 for multi-threaded use, so it probably should be abolished.
1243 +) Added Broadcom "ubsec" ENGINE to OpenSSL.
1244 [Broadcom, tweaked and integrated by Geoff Thorpe]
1246 +) Move common extension printing code to new function
1247 X509V3_print_extensions(). Reorganise OCSP print routines and
1248 implement some needed OCSP ASN1 functions. Add OCSP extensions.
1251 +) New function X509_signature_print() to remove duplication in some
1255 +) Add a special meaning when SET OF and SEQUENCE OF flags are both
1256 set (this was treated exactly the same as SET OF previously). This
1257 is used to reorder the STACK representing the structure to match the
1258 encoding. This will be used to get round a problem where a PKCS7
1259 structure which was signed could not be verified because the STACK
1260 order did not reflect the encoded order.
1263 +) Reimplement the OCSP ASN1 module using the new code.
1266 +) Update the X509V3 code to permit the use of an ASN1_ITEM structure
1267 for its ASN1 operations. The old style function pointers still exist
1268 for now but they will eventually go away.
1271 +) Merge in replacement ASN1 code from the ASN1 branch. This almost
1272 completely replaces the old ASN1 functionality with a table driven
1273 encoder and decoder which interprets an ASN1_ITEM structure describing
1274 the ASN1 module. Compatibility with the existing ASN1 API (i2d,d2i) is
1275 largely maintained. Almost all of the old asn1_mac.h macro based ASN1
1276 has also been converted to the new form.
1279 +) Change BN_mod_exp_recp so that negative moduli are tolerated
1280 (the sign is ignored). Similarly, ignore the sign in BN_MONT_CTX_set
1281 so that BN_mod_exp_mont and BN_mod_exp_mont_word work
1282 for negative moduli.
1285 +) Fix BN_uadd and BN_usub: Always return non-negative results instead
1286 of not touching the result's sign bit.
1289 +) BN_div bugfix: If the result is 0, the sign (res->neg) must not be
1293 +) Changed the LHASH code to use prototypes for callbacks, and created
1294 macros to declare and implement thin (optionally static) functions
1295 that provide type-safety and avoid function pointer casting for the
1296 type-specific callbacks.
1299 *) Use better test patterns in bntest.
1302 +) Added Kerberos Cipher Suites to be used with TLS, as written in
1304 [Veers Staats <staatsvr@asc.hpc.mil>,
1305 Jeffrey Altman <jaltman@columbia.edu>, via Richard Levitte]
1307 *) rand_win.c fix for Borland C.
1310 *) BN_rshift bugfix for n == 0.
1313 +) Reformat the FAQ so the different questions and answers can be divided
1314 in sections depending on the subject.
1317 +) Have the zlib compression code load ZLIB.DLL dynamically under
1321 +) New function BN_mod_sqrt for computing square roots modulo a prime
1322 (using the probabilistic Tonelli-Shanks algorithm unless
1323 p == 3 (mod 4) or p == 5 (mod 8), which are cases that can
1324 be handled deterministically).
1325 [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller]
1327 *) Add a 'bctest' script that checks for some known 'bc' bugs
1328 so that 'make test' does not abort just because 'bc' is broken.
1331 *) Store verify_result within SSL_SESSION also for client side to
1332 avoid potential security hole. (Re-used sessions on the client side
1333 always resulted in verify_result==X509_V_OK, not using the original
1334 result of the server certificate verification.)
1337 +) Make BN_mod_inverse faster by explicitly handling small quotients
1338 in the Euclid loop. (Speed gain about 20% for small moduli [256 or
1339 512 bits], about 30% for larger ones [1024 or 2048 bits].)
1342 *) Fix ssl3_pending: If the record in s->s3->rrec is not of type
1343 SSL3_RT_APPLICATION_DATA, return 0.
1344 Similarly, change ssl2_pending to return 0 if SSL_in_init(s) is true.
1348 Both ssl2_peek and ssl3_peek, which were totally broken in earlier
1349 releases, have been re-implemented by renaming the previous
1350 implementations of ssl2_read and ssl3_read to ssl2_read_internal
1351 and ssl3_read_internal, respectively, and adding 'peek' parameters
1352 to them. The new ssl[23]_{read,peek} functions are calls to
1353 ssl[23]_read_internal with the 'peek' flag set appropriately.
1354 A 'peek' parameter has also been added to ssl3_read_bytes, which
1355 does the actual work for ssl3_read_internal.
1358 +) New function BN_kronecker.
1361 +) Fix BN_gcd so that it works on negative inputs; the result is
1362 positive unless both parameters are zero.
1363 Previously something reasonably close to an infinite loop was
1364 possible because numbers could be growing instead of shrinking
1365 in the implementation of Euclid's algorithm.
1368 +) Fix BN_is_word() and BN_is_one() macros to take into account the
1369 sign of the number in question.
1371 Fix BN_is_word(a,w) to work correctly for w == 0.
1373 The old BN_is_word(a,w) macro is now called BN_abs_is_word(a,w)
1374 because its test if the absolute value of 'a' equals 'w'.
1375 Note that BN_abs_is_word does *not* handle w == 0 reliably;
1376 it exists mostly for use in the implementations of BN_is_zero(),
1377 BN_is_one(), and BN_is_word().
1380 *) Initialise "ex_data" member of RSA/DSA/DH structures prior to calling
1381 the method-specific "init()" handler. Also clean up ex_data after
1382 calling the method-specific "finish()" handler. Previously, this was
1383 happening the other way round.
1386 +) New function BN_swap.
1389 +) Use BN_nnmod instead of BN_mod in crypto/bn/bn_exp.c so that
1390 the exponentiation functions are more likely to produce reasonable
1391 results on negative inputs.
1394 +) Change BN_mod_mul so that the result is always non-negative.
1395 Previously, it could be negative if one of the factors was negative;
1396 I don't think anyone really wanted that behaviour.
1399 +) Move BN_mod_... functions into new file crypto/bn/bn_mod.c
1400 (except for exponentiation, which stays in crypto/bn/bn_exp.c,
1401 and BN_mod_mul_reciprocal, which stays in crypto/bn/bn_recp.c)
1402 and add new functions:
1411 BN_mod_lshift1_quick
1415 These functions always generate non-negative results.
1417 BN_nnmod otherwise is like BN_mod (if BN_mod computes a remainder r
1418 such that |m| < r < 0, BN_nnmod will output rem + |m| instead).
1420 BN_mod_XXX_quick(r, a, [b,] m) generates the same result as
1421 BN_mod_XXX(r, a, [b,] m, ctx), but requires that a [and b]
1422 be reduced modulo m.
1423 [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller]
1425 +) Remove a few calls to bn_wexpand() in BN_sqr() (the one in there
1426 was actually never needed) and in BN_mul(). The removal in BN_mul()
1427 required a small change in bn_mul_part_recursive() and the addition
1428 of the functions bn_cmp_part_words(), bn_sub_part_words() and
1429 bn_add_part_words(), which do the same thing as bn_cmp_words(),
1430 bn_sub_words() and bn_add_words() except they take arrays with
1434 +) In 'openssl passwd', verify passwords read from the terminal
1435 unless the '-salt' option is used (which usually means that
1436 verification would just waste user's time since the resulting
1437 hash is going to be compared with some given password hash)
1438 or the new '-noverify' option is used.
1440 This is an incompatible change, but it does not affect
1441 non-interactive use of 'openssl passwd' (passwords on the command
1442 line, '-stdin' option, '-in ...' option) and thus should not
1446 +) Remove all references to RSAref, since there's no more need for it.
1449 *) Increase BN_CTX_NUM (the number of BIGNUMs in a BN_CTX) to 16.
1450 The previous value, 12, was not always sufficient for BN_mod_exp().
1453 +) Make DSO load along a path given through an environment variable
1454 (SHLIB_PATH) with shl_load().
1457 +) Constify the ENGINE code as a result of BIGNUM constification.
1458 Also constify the RSA code and most things related to it. In a
1459 few places, most notable in the depth of the ASN.1 code, ugly
1460 casts back to non-const were required (to be solved at a later
1464 +) Make it so the openssl application has all engines loaded by default.
1467 +) Constify the BIGNUM routines a little more.
1470 *) Make sure that shared libraries get the internal name engine with
1471 the full version number and not just 0. This should mark the
1472 shared libraries as not backward compatible. Of course, this should
1473 be changed again when we can guarantee backward binary compatibility.
1476 +) Add the following functions:
1478 ENGINE_load_cswift()
1480 ENGINE_load_atalla()
1482 ENGINE_load_builtin_engines()
1484 That way, an application can itself choose if external engines that
1485 are built-in in OpenSSL shall ever be used or not. The benefit is
1486 that applications won't have to be linked with libdl or other dso
1487 libraries unless it's really needed.
1489 Changed 'openssl engine' to load all engines on demand.
1490 Changed the engine header files to avoid the duplication of some
1491 declarations (they differed!).
1494 +) 'openssl engine' can now list capabilities.
1497 +) Better error reporting in 'openssl engine'.
1500 +) Never call load_dh_param(NULL) in s_server.
1503 +) Add engine application. It can currently list engines by name and
1504 identity, and test if they are actually available.
1507 +) Add support for shared libraries under Irix.
1508 [Albert Chin-A-Young <china@thewrittenword.com>]
1510 +) Improve RPM specification file by forcing symbolic linking and making
1511 sure the installed documentation is also owned by root.root.
1512 [Damien Miller <djm@mindrot.org>]
1514 +) Add configuration option to build on Linux on both big-endian and
1516 [Ralf Baechle <ralf@uni-koblenz.de>]
1518 +) Give the OpenSSL applications more possibilities to make use of
1519 keys (public as well as private) handled by engines.
1522 +) Add OCSP code that comes from CertCo.
1525 +) Add VMS support for the Rijndael code.
1528 +) Added untested support for Nuron crypto accelerator.
1531 +) Add support for external cryptographic devices. This code was
1532 previously distributed separately as the "engine" branch.
1533 [Geoff Thorpe, Richard Levitte]
1535 +) Rework the filename-translation in the DSO code. It is now possible to
1536 have far greater control over how a "name" is turned into a filename
1537 depending on the operating environment and any oddities about the
1538 different shared library filenames on each system.
1541 +) Support threads on FreeBSD-elf in Configure.
1544 +) Add the possibility to create shared libraries on HP-UX
1547 +) Fix for SHA1 assembly problem with MASM: it produces
1548 warnings about corrupt line number information when assembling
1549 with debugging information. This is caused by the overlapping
1551 [Bernd Matthes <mainbug@celocom.de>, Steve Henson]
1554 NCONF_get_number() has no error checking at all. As a replacement,
1555 NCONF_get_number_e() is defined (_e for "error checking") and is
1556 promoted strongly. The old NCONF_get_number is kept around for
1557 binary backward compatibility.
1558 Make it possible for methods to load from something other than a BIO,
1559 by providing a function pointer that is given a name instead of a BIO.
1560 For example, this could be used to load configuration data from an
1564 *) Fix typo in get_cert_by_subject() in by_dir.c
1565 [Jean-Marc Desperrier <jean-marc.desperrier@certplus.com>]
1567 *) Rework the system to generate shared libraries:
1569 - Make note of the expected extension for the shared libraries and
1570 if there is a need for symbolic links from for example libcrypto.so.0
1571 to libcrypto.so.0.9.7. There is extended info in Configure for
1574 - Make as few rebuilds of the shared libraries as possible.
1576 - Still avoid linking the OpenSSL programs with the shared libraries.
1578 - When installing, install the shared libraries separately from the
1582 +) Fix for non blocking accept BIOs. Added new I/O special reason
1583 BIO_RR_ACCEPT to cover this case. Previously use of accept BIOs
1584 with non blocking I/O was not possible because no retry code was
1585 implemented. Also added new SSL code SSL_WANT_ACCEPT to cover
1589 +) Added the beginnings of Rijndael support.
1592 +) Fix for bug in DirectoryString mask setting. Add support for
1593 X509_NAME_print_ex() in 'req' and X509_print_ex() function
1594 to allow certificate printing to more controllable, additional
1595 'certopt' option to 'x509' to allow new printing options to be
1599 +) Clean old EAY MD5 hack from e_os.h.
1602 *) Fix SSL_CTX_set_read_ahead macro to actually use its argument.
1604 Copy SSL_CTX's read_ahead flag to SSL object directly in SSL_new
1605 and not in SSL_clear because the latter is also used by the
1606 accept/connect functions; previously, the settings made by
1607 SSL_set_read_ahead would be lost during the handshake.
1608 [Bodo Moeller; problems reported by Anders Gertz <gertz@epact.se>]
1610 *) Correct util/mkdef.pl to be selective about disabled algorithms.
1611 Previously, it would create entries for disableed algorithms no
1615 *) Added several new manual pages for SSL_* function.
1618 Changes between 0.9.5a and 0.9.6 [24 Sep 2000]
1620 *) In ssl23_get_client_hello, generate an error message when faced
1621 with an initial SSL 3.0/TLS record that is too small to contain the
1622 first two bytes of the ClientHello message, i.e. client_version.
1623 (Note that this is a pathologic case that probably has never happened
1624 in real life.) The previous approach was to use the version number
1625 from the record header as a substitute; but our protocol choice
1626 should not depend on that one because it is not authenticated
1627 by the Finished messages.
1630 *) More robust randomness gathering functions for Windows.
1631 [Jeffrey Altman <jaltman@columbia.edu>]
1633 *) For compatibility reasons if the flag X509_V_FLAG_ISSUER_CHECK is
1634 not set then we don't setup the error code for issuer check errors
1635 to avoid possibly overwriting other errors which the callback does
1636 handle. If an application does set the flag then we assume it knows
1637 what it is doing and can handle the new informational codes
1641 *) Fix for a nasty bug in ASN1_TYPE handling. ASN1_TYPE is used for
1642 a general "ANY" type, as such it should be able to decode anything
1643 including tagged types. However it didn't check the class so it would
1644 wrongly interpret tagged types in the same way as their universal
1645 counterpart and unknown types were just rejected. Changed so that the
1646 tagged and unknown types are handled in the same way as a SEQUENCE:
1647 that is the encoding is stored intact. There is also a new type
1648 "V_ASN1_OTHER" which is used when the class is not universal, in this
1649 case we have no idea what the actual type is so we just lump them all
1653 *) On VMS, stdout may very well lead to a file that is written to
1654 in a record-oriented fashion. That means that every write() will
1655 write a separate record, which will be read separately by the
1656 programs trying to read from it. This can be very confusing.
1658 The solution is to put a BIO filter in the way that will buffer
1659 text until a linefeed is reached, and then write everything a
1660 line at a time, so every record written will be an actual line,
1661 not chunks of lines and not (usually doesn't happen, but I've
1662 seen it once) several lines in one record. BIO_f_linebuffer() is
1665 Currently, it's a VMS-only method, because that's where it has
1666 been tested well enough.
1669 *) Remove 'optimized' squaring variant in BN_mod_mul_montgomery,
1670 it can return incorrect results.
1671 (Note: The buggy variant was not enabled in OpenSSL 0.9.5a,
1672 but it was in 0.9.6-beta[12].)
1675 *) Disable the check for content being present when verifying detached
1676 signatures in pk7_smime.c. Some versions of Netscape (wrongly)
1677 include zero length content when signing messages.
1680 *) New BIO_shutdown_wr macro, which invokes the BIO_C_SHUTDOWN_WR
1681 BIO_ctrl (for BIO pairs).
1684 *) Add DSO method for VMS.
1687 *) Bug fix: Montgomery multiplication could produce results with the
1691 *) Add RPM specification openssl.spec and modify it to build three
1692 packages. The default package contains applications, application
1693 documentation and run-time libraries. The devel package contains
1694 include files, static libraries and function documentation. The
1695 doc package contains the contents of the doc directory. The original
1696 openssl.spec was provided by Damien Miller <djm@mindrot.org>.
1699 *) Add a large number of documentation files for many SSL routines.
1700 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>]
1702 *) Add a configuration entry for Sony News 4.
1703 [NAKAJI Hiroyuki <nakaji@tutrp.tut.ac.jp>]
1705 *) Don't set the two most significant bits to one when generating a
1706 random number < q in the DSA library.
1709 *) New SSL API mode 'SSL_MODE_AUTO_RETRY'. This disables the default
1710 behaviour that SSL_read may result in SSL_ERROR_WANT_READ (even if
1711 the underlying transport is blocking) if a handshake took place.
1712 (The default behaviour is needed by applications such as s_client
1713 and s_server that use select() to determine when to use SSL_read;
1714 but for applications that know in advance when to expect data, it
1715 just makes things more complicated.)
1718 *) Add RAND_egd_bytes(), which gives control over the number of bytes read
1722 *) Add a few more EBCDIC conditionals that make `req' and `x509'
1723 work better on such systems.
1724 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>]
1726 *) Add two demo programs for PKCS12_parse() and PKCS12_create().
1727 Update PKCS12_parse() so it copies the friendlyName and the
1728 keyid to the certificates aux info.
1731 *) Fix bug in PKCS7_verify() which caused an infinite loop
1732 if there was more than one signature.
1733 [Sven Uszpelkat <su@celocom.de>]
1735 *) Major change in util/mkdef.pl to include extra information
1736 about each symbol, as well as presentig variables as well
1737 as functions. This change means that there's n more need
1738 to rebuild the .num files when some algorithms are excluded.
1741 *) Allow the verify time to be set by an application,
1742 rather than always using the current time.
1745 *) Phase 2 verify code reorganisation. The certificate
1746 verify code now looks up an issuer certificate by a
1747 number of criteria: subject name, authority key id
1748 and key usage. It also verifies self signed certificates
1749 by the same criteria. The main comparison function is
1750 X509_check_issued() which performs these checks.
1752 Lot of changes were necessary in order to support this
1753 without completely rewriting the lookup code.
1755 Authority and subject key identifier are now cached.
1757 The LHASH 'certs' is X509_STORE has now been replaced
1758 by a STACK_OF(X509_OBJECT). This is mainly because an
1759 LHASH can't store or retrieve multiple objects with
1760 the same hash value.
1762 As a result various functions (which were all internal
1763 use only) have changed to handle the new X509_STORE
1764 structure. This will break anything that messed round
1765 with X509_STORE internally.
1767 The functions X509_STORE_add_cert() now checks for an
1768 exact match, rather than just subject name.
1770 The X509_STORE API doesn't directly support the retrieval
1771 of multiple certificates matching a given criteria, however
1772 this can be worked round by performing a lookup first
1773 (which will fill the cache with candidate certificates)
1774 and then examining the cache for matches. This is probably
1775 the best we can do without throwing out X509_LOOKUP
1776 entirely (maybe later...).
1778 The X509_VERIFY_CTX structure has been enhanced considerably.
1780 All certificate lookup operations now go via a get_issuer()
1781 callback. Although this currently uses an X509_STORE it
1782 can be replaced by custom lookups. This is a simple way
1783 to bypass the X509_STORE hackery necessary to make this
1784 work and makes it possible to use more efficient techniques
1785 in future. A very simple version which uses a simple
1786 STACK for its trusted certificate store is also provided
1787 using X509_STORE_CTX_trusted_stack().
1789 The verify_cb() and verify() callbacks now have equivalents
1790 in the X509_STORE_CTX structure.
1792 X509_STORE_CTX also has a 'flags' field which can be used
1793 to customise the verify behaviour.
1796 *) Add new PKCS#7 signing option PKCS7_NOSMIMECAP which
1797 excludes S/MIME capabilities.
1800 *) When a certificate request is read in keep a copy of the
1801 original encoding of the signed data and use it when outputing
1802 again. Signatures then use the original encoding rather than
1803 a decoded, encoded version which may cause problems if the
1804 request is improperly encoded.
1807 *) For consistency with other BIO_puts implementations, call
1808 buffer_write(b, ...) directly in buffer_puts instead of calling
1811 In BIO_puts, increment b->num_write as in BIO_write.
1812 [Peter.Sylvester@EdelWeb.fr]
1814 *) Fix BN_mul_word for the case where the word is 0. (We have to use
1815 BN_zero, we may not return a BIGNUM with an array consisting of
1819 *) Avoid calling abort() from within the library when problems are
1820 detected, except if preprocessor symbols have been defined
1821 (such as REF_CHECK, BN_DEBUG etc.).
1824 *) New openssl application 'rsautl'. This utility can be
1825 used for low level RSA operations. DER public key
1826 BIO/fp routines also added.
1829 *) New Configure entry and patches for compiling on QNX 4.
1830 [Andreas Schneider <andreas@ds3.etech.fh-hamburg.de>]
1832 *) A demo state-machine implementation was sponsored by
1833 Nuron (http://www.nuron.com/) and is now available in
1834 demos/state_machine.
1837 *) New options added to the 'dgst' utility for signature
1838 generation and verification.
1841 *) Unrecognized PKCS#7 content types are now handled via a
1842 catch all ASN1_TYPE structure. This allows unsupported
1843 types to be stored as a "blob" and an application can
1844 encode and decode it manually.
1847 *) Fix various signed/unsigned issues to make a_strex.c
1849 [Oscar Jacobsson <oscar.jacobsson@celocom.com>]
1851 *) ASN1 fixes. i2d_ASN1_OBJECT was not returning the correct
1852 length if passed a buffer. ASN1_INTEGER_to_BN failed
1853 if passed a NULL BN and its argument was negative.
1854 [Steve Henson, pointed out by Sven Heiberg <sven@tartu.cyber.ee>]
1856 *) Modification to PKCS#7 encoding routines to output definite
1857 length encoding. Since currently the whole structures are in
1858 memory there's not real point in using indefinite length
1859 constructed encoding. However if OpenSSL is compiled with
1860 the flag PKCS7_INDEFINITE_ENCODING the old form is used.
1863 *) Added BIO_vprintf() and BIO_vsnprintf().
1866 *) Added more prefixes to parse for in the the strings written
1867 through a logging bio, to cover all the levels that are available
1868 through syslog. The prefixes are now:
1870 PANIC, EMERG, EMR => LOG_EMERG
1871 ALERT, ALR => LOG_ALERT
1872 CRIT, CRI => LOG_CRIT
1873 ERROR, ERR => LOG_ERR
1874 WARNING, WARN, WAR => LOG_WARNING
1875 NOTICE, NOTE, NOT => LOG_NOTICE
1876 INFO, INF => LOG_INFO
1877 DEBUG, DBG => LOG_DEBUG
1879 and as before, if none of those prefixes are present at the
1880 beginning of the string, LOG_ERR is chosen.
1882 On Win32, the LOG_* levels are mapped according to this:
1884 LOG_EMERG, LOG_ALERT, LOG_CRIT, LOG_ERR => EVENTLOG_ERROR_TYPE
1885 LOG_WARNING => EVENTLOG_WARNING_TYPE
1886 LOG_NOTICE, LOG_INFO, LOG_DEBUG => EVENTLOG_INFORMATION_TYPE
1890 *) Made it possible to reconfigure with just the configuration
1891 argument "reconf" or "reconfigure". The command line arguments
1892 are stored in Makefile.ssl in the variable CONFIGURE_ARGS,
1893 and are retrieved from there when reconfiguring.
1897 [Assar Westerlund <assar@sics.se>, Richard Levitte]
1899 *) Add the arguments -CAfile and -CApath to the pkcs12 utility.
1902 *) The obj_dat.pl script was messing up the sorting of object
1903 names. The reason was that it compared the quoted version
1904 of strings as a result "OCSP" > "OCSP Signing" because
1905 " > SPACE. Changed script to store unquoted versions of
1906 names and add quotes on output. It was also omitting some
1907 names from the lookup table if they were given a default
1908 value (that is if SN is missing it is given the same
1909 value as LN and vice versa), these are now added on the
1910 grounds that if an object has a name we should be able to
1911 look it up. Finally added warning output when duplicate
1912 short or long names are found.
1915 *) Changes needed for Tandem NSK.
1916 [Scott Uroff <scott@xypro.com>]
1918 *) Fix SSL 2.0 rollback checking: Due to an off-by-one error in
1919 RSA_padding_check_SSLv23(), special padding was never detected
1920 and thus the SSL 3.0/TLS 1.0 countermeasure against protocol
1921 version rollback attacks was not effective.
1923 In s23_clnt.c, don't use special rollback-attack detection padding
1924 (RSA_SSLV23_PADDING) if SSL 2.0 is the only protocol enabled in the
1925 client; similarly, in s23_srvr.c, don't do the rollback check if
1926 SSL 2.0 is the only protocol enabled in the server.
1929 *) Make it possible to get hexdumps of unprintable data with 'openssl
1930 asn1parse'. By implication, the functions ASN1_parse_dump() and
1931 BIO_dump_indent() are added.
1934 *) New functions ASN1_STRING_print_ex() and X509_NAME_print_ex()
1935 these print out strings and name structures based on various
1936 flags including RFC2253 support and proper handling of
1937 multibyte characters. Added options to the 'x509' utility
1938 to allow the various flags to be set.
1941 *) Various fixes to use ASN1_TIME instead of ASN1_UTCTIME.
1942 Also change the functions X509_cmp_current_time() and
1943 X509_gmtime_adj() work with an ASN1_TIME structure,
1944 this will enable certificates using GeneralizedTime in validity
1945 dates to be checked.
1948 *) Make the NEG_PUBKEY_BUG code (which tolerates invalid
1949 negative public key encodings) on by default,
1950 NO_NEG_PUBKEY_BUG can be set to disable it.
1953 *) New function c2i_ASN1_OBJECT() which acts on ASN1_OBJECT
1954 content octets. An i2c_ASN1_OBJECT is unnecessary because
1955 the encoding can be trivially obtained from the structure.
1958 *) crypto/err.c locking bugfix: Use write locks (CRYPTO_w_[un]lock),
1959 not read locks (CRYPTO_r_[un]lock).
1962 *) A first attempt at creating official support for shared
1963 libraries through configuration. I've kept it so the
1964 default is static libraries only, and the OpenSSL programs
1965 are always statically linked for now, but there are
1966 preparations for dynamic linking in place.
1967 This has been tested on Linux and True64.
1970 *) Randomness polling function for Win9x, as described in:
1971 Peter Gutmann, Software Generation of Practically Strong
1975 *) Fix so PRNG is seeded in req if using an already existing
1979 *) New options to smime application. -inform and -outform
1980 allow alternative formats for the S/MIME message including
1981 PEM and DER. The -content option allows the content to be
1982 specified separately. This should allow things like Netscape
1983 form signing output easier to verify.
1986 *) Fix the ASN1 encoding of tags using the 'long form'.
1989 *) New ASN1 functions, i2c_* and c2i_* for INTEGER and BIT
1990 STRING types. These convert content octets to and from the
1991 underlying type. The actual tag and length octets are
1992 already assumed to have been read in and checked. These
1993 are needed because all other string types have virtually
1994 identical handling apart from the tag. By having versions
1995 of the ASN1 functions that just operate on content octets
1996 IMPLICIT tagging can be handled properly. It also allows
1997 the ASN1_ENUMERATED code to be cut down because ASN1_ENUMERATED
1998 and ASN1_INTEGER are identical apart from the tag.
2001 *) Change the handling of OID objects as follows:
2003 - New object identifiers are inserted in objects.txt, following
2004 the syntax given in objects.README.
2005 - objects.pl is used to process obj_mac.num and create a new
2007 - obj_dat.pl is used to create a new obj_dat.h, using the data in
2010 This is currently kind of a hack, and the perl code in objects.pl
2011 isn't very elegant, but it works as I intended. The simplest way
2012 to check that it worked correctly is to look in obj_dat.h and
2013 check the array nid_objs and make sure the objects haven't moved
2014 around (this is important!). Additions are OK, as well as
2015 consistent name changes.
2018 *) Add BSD-style MD5-based passwords to 'openssl passwd' (option '-1').
2021 *) Addition of the command line parameter '-rand file' to 'openssl req'.
2022 The given file adds to whatever has already been seeded into the
2023 random pool through the RANDFILE configuration file option or
2024 environment variable, or the default random state file.
2027 *) mkstack.pl now sorts each macro group into lexical order.
2028 Previously the output order depended on the order the files
2029 appeared in the directory, resulting in needless rewriting
2033 *) Patches to make OpenSSL compile under Win32 again. Mostly
2034 work arounds for the VC++ problem that it treats func() as
2035 func(void). Also stripped out the parts of mkdef.pl that
2036 added extra typesafe functions: these no longer exist.
2039 *) Reorganisation of the stack code. The macros are now all
2040 collected in safestack.h . Each macro is defined in terms of
2041 a "stack macro" of the form SKM_<name>(type, a, b). The
2042 DEBUG_SAFESTACK is now handled in terms of function casts,
2043 this has the advantage of retaining type safety without the
2044 use of additional functions. If DEBUG_SAFESTACK is not defined
2045 then the non typesafe macros are used instead. Also modified the
2046 mkstack.pl script to handle the new form. Needs testing to see
2047 if which (if any) compilers it chokes and maybe make DEBUG_SAFESTACK
2048 the default if no major problems. Similar behaviour for ASN1_SET_OF
2049 and PKCS12_STACK_OF.
2052 *) When some versions of IIS use the 'NET' form of private key the
2053 key derivation algorithm is different. Normally MD5(password) is
2054 used as a 128 bit RC4 key. In the modified case
2055 MD5(MD5(password) + "SGCKEYSALT") is used insted. Added some
2056 new functions i2d_RSA_NET(), d2i_RSA_NET() etc which are the same
2057 as the old Netscape_RSA functions except they have an additional
2058 'sgckey' parameter which uses the modified algorithm. Also added
2059 an -sgckey command line option to the rsa utility. Thanks to
2060 Adrian Peck <bertie@ncipher.com> for posting details of the modified
2061 algorithm to openssl-dev.
2064 *) The evp_local.h macros were using 'c.##kname' which resulted in
2065 invalid expansion on some systems (SCO 5.0.5 for example).
2066 Corrected to 'c.kname'.
2067 [Phillip Porch <root@theporch.com>]
2069 *) New X509_get1_email() and X509_REQ_get1_email() functions that return
2070 a STACK of email addresses from a certificate or request, these look
2071 in the subject name and the subject alternative name extensions and
2072 omit any duplicate addresses.
2075 *) Re-implement BN_mod_exp2_mont using independent (and larger) windows.
2076 This makes DSA verification about 2 % faster.
2079 *) Increase maximum window size in BN_mod_exp_... to 6 bits instead of 5
2080 (meaning that now 2^5 values will be precomputed, which is only 4 KB
2081 plus overhead for 1024 bit moduli).
2082 This makes exponentiations about 0.5 % faster for 1024 bit
2083 exponents (as measured by "openssl speed rsa2048").
2086 *) Rename memory handling macros to avoid conflicts with other
2088 Malloc => OPENSSL_malloc
2089 Malloc_locked => OPENSSL_malloc_locked
2090 Realloc => OPENSSL_realloc
2091 Free => OPENSSL_free
2094 *) New function BN_mod_exp_mont_word for small bases (roughly 15%
2095 faster than BN_mod_exp_mont, i.e. 7% for a full DH exchange).
2098 *) CygWin32 support.
2099 [John Jarvie <jjarvie@newsguy.com>]
2101 *) The type-safe stack code has been rejigged. It is now only compiled
2102 in when OpenSSL is configured with the DEBUG_SAFESTACK option and
2103 by default all type-specific stack functions are "#define"d back to
2104 standard stack functions. This results in more streamlined output
2105 but retains the type-safety checking possibilities of the original
2109 *) The STACK code has been cleaned up, and certain type declarations
2110 that didn't make a lot of sense have been brought in line. This has
2111 also involved a cleanup of sorts in safestack.h to more correctly
2112 map type-safe stack functions onto their plain stack counterparts.
2113 This work has also resulted in a variety of "const"ifications of
2114 lots of the code, especially "_cmp" operations which should normally
2115 be prototyped with "const" parameters anyway.
2118 *) When generating bytes for the first time in md_rand.c, 'stir the pool'
2119 by seeding with STATE_SIZE dummy bytes (with zero entropy count).
2120 (The PRNG state consists of two parts, the large pool 'state' and 'md',
2121 where all of 'md' is used each time the PRNG is used, but 'state'
2122 is used only indexed by a cyclic counter. As entropy may not be
2123 well distributed from the beginning, 'md' is important as a
2124 chaining variable. However, the output function chains only half
2125 of 'md', i.e. 80 bits. ssleay_rand_add, on the other hand, chains
2126 all of 'md', and seeding with STATE_SIZE dummy bytes will result
2127 in all of 'state' being rewritten, with the new values depending
2128 on virtually all of 'md'. This overcomes the 80 bit limitation.)
2131 *) In ssl/s2_clnt.c and ssl/s3_clnt.c, call ERR_clear_error() when
2132 the handshake is continued after ssl_verify_cert_chain();
2133 otherwise, if SSL_VERIFY_NONE is set, remaining error codes
2134 can lead to 'unexplainable' connection aborts later.
2135 [Bodo Moeller; problem tracked down by Lutz Jaenicke]
2137 *) Major EVP API cipher revision.
2138 Add hooks for extra EVP features. This allows various cipher
2139 parameters to be set in the EVP interface. Support added for variable
2140 key length ciphers via the EVP_CIPHER_CTX_set_key_length() function and
2141 setting of RC2 and RC5 parameters.
2143 Modify EVP_OpenInit() and EVP_SealInit() to cope with variable key length
2146 Remove lots of duplicated code from the EVP library. For example *every*
2147 cipher init() function handles the 'iv' in the same way according to the
2148 cipher mode. They also all do nothing if the 'key' parameter is NULL and
2149 for CFB and OFB modes they zero ctx->num.
2151 New functionality allows removal of S/MIME code RC2 hack.
2153 Most of the routines have the same form and so can be declared in terms
2156 By shifting this to the top level EVP_CipherInit() it can be removed from
2157 all individual ciphers. If the cipher wants to handle IVs or keys
2158 differently it can set the EVP_CIPH_CUSTOM_IV or EVP_CIPH_ALWAYS_CALL_INIT
2161 Change lots of functions like EVP_EncryptUpdate() to now return a
2162 value: although software versions of the algorithms cannot fail
2163 any installed hardware versions can.
2166 *) Implement SSL_OP_TLS_ROLLBACK_BUG: In ssl3_get_client_key_exchange, if
2167 this option is set, tolerate broken clients that send the negotiated
2168 protocol version number instead of the requested protocol version
2172 *) Call dh_tmp_cb (set by ..._TMP_DH_CB) with correct 'is_export' flag;
2173 i.e. non-zero for export ciphersuites, zero otherwise.
2174 Previous versions had this flag inverted, inconsistent with
2175 rsa_tmp_cb (..._TMP_RSA_CB).
2176 [Bodo Moeller; problem reported by Amit Chopra]
2178 *) Add missing DSA library text string. Work around for some IIS
2179 key files with invalid SEQUENCE encoding.
2182 *) Add a document (doc/standards.txt) that list all kinds of standards
2183 and so on that are implemented in OpenSSL.
2186 *) Enhance c_rehash script. Old version would mishandle certificates
2187 with the same subject name hash and wouldn't handle CRLs at all.
2188 Added -fingerprint option to crl utility, to support new c_rehash
2192 *) Eliminate non-ANSI declarations in crypto.h and stack.h.
2195 *) Fix for SSL server purpose checking. Server checking was
2196 rejecting certificates which had extended key usage present
2197 but no ssl client purpose.
2198 [Steve Henson, reported by Rene Grosser <grosser@hisolutions.com>]
2200 *) Make PKCS#12 code work with no password. The PKCS#12 spec
2201 is a little unclear about how a blank password is handled.
2202 Since the password in encoded as a BMPString with terminating
2203 double NULL a zero length password would end up as just the
2204 double NULL. However no password at all is different and is
2205 handled differently in the PKCS#12 key generation code. NS
2206 treats a blank password as zero length. MSIE treats it as no
2207 password on export: but it will try both on import. We now do
2208 the same: PKCS12_parse() tries zero length and no password if
2209 the password is set to "" or NULL (NULL is now a valid password:
2210 it wasn't before) as does the pkcs12 application.
2213 *) Bugfixes in apps/x509.c: Avoid a memory leak; and don't use
2214 perror when PEM_read_bio_X509_REQ fails, the error message must
2215 be obtained from the error queue.
2218 *) Avoid 'thread_hash' memory leak in crypto/err/err.c by freeing
2219 it in ERR_remove_state if appropriate, and change ERR_get_state
2220 accordingly to avoid race conditions (this is necessary because
2221 thread_hash is no longer constant once set).
2224 *) Bugfix for linux-elf makefile.one.
2227 *) RSA_get_default_method() will now cause a default
2228 RSA_METHOD to be chosen if one doesn't exist already.
2229 Previously this was only set during a call to RSA_new()
2230 or RSA_new_method(NULL) meaning it was possible for
2231 RSA_get_default_method() to return NULL.
2234 *) Added native name translation to the existing DSO code
2235 that will convert (if the flag to do so is set) filenames
2236 that are sufficiently small and have no path information
2237 into a canonical native form. Eg. "blah" converted to
2238 "libblah.so" or "blah.dll" etc.
2241 *) New function ERR_error_string_n(e, buf, len) which is like
2242 ERR_error_string(e, buf), but writes at most 'len' bytes
2243 including the 0 terminator. For ERR_error_string_n, 'buf'
2245 [Damien Miller <djm@mindrot.org>, Bodo Moeller]
2247 *) CONF library reworked to become more general. A new CONF
2248 configuration file reader "class" is implemented as well as a
2249 new functions (NCONF_*, for "New CONF") to handle it. The now
2250 old CONF_* functions are still there, but are reimplemented to
2251 work in terms of the new functions. Also, a set of functions
2252 to handle the internal storage of the configuration data is
2253 provided to make it easier to write new configuration file
2254 reader "classes" (I can definitely see something reading a
2255 configuration file in XML format, for example), called _CONF_*,
2256 or "the configuration storage API"...
2258 The new configuration file reading functions are:
2260 NCONF_new, NCONF_free, NCONF_load, NCONF_load_fp, NCONF_load_bio,
2261 NCONF_get_section, NCONF_get_string, NCONF_get_numbre
2263 NCONF_default, NCONF_WIN32
2265 NCONF_dump_fp, NCONF_dump_bio
2267 NCONF_default and NCONF_WIN32 are method (or "class") choosers,
2268 NCONF_new creates a new CONF object. This works in the same way
2269 as other interfaces in OpenSSL, like the BIO interface.
2270 NCONF_dump_* dump the internal storage of the configuration file,
2271 which is useful for debugging. All other functions take the same
2272 arguments as the old CONF_* functions wth the exception of the
2273 first that must be a `CONF *' instead of a `LHASH *'.
2275 To make it easer to use the new classes with the old CONF_* functions,
2276 the function CONF_set_default_method is provided.
2279 *) Add '-tls1' option to 'openssl ciphers', which was already
2280 mentioned in the documentation but had not been implemented.
2281 (This option is not yet really useful because even the additional
2282 experimental TLS 1.0 ciphers are currently treated as SSL 3.0 ciphers.)
2285 *) Initial DSO code added into libcrypto for letting OpenSSL (and
2286 OpenSSL-based applications) load shared libraries and bind to
2287 them in a portable way.
2288 [Geoff Thorpe, with contributions from Richard Levitte]
2290 Changes between 0.9.5 and 0.9.5a [1 Apr 2000]
2292 *) Make sure _lrotl and _lrotr are only used with MSVC.
2294 *) Use lock CRYPTO_LOCK_RAND correctly in ssleay_rand_status
2295 (the default implementation of RAND_status).
2297 *) Rename openssl x509 option '-crlext', which was added in 0.9.5,
2298 to '-clrext' (= clear extensions), as intended and documented.
2299 [Bodo Moeller; inconsistency pointed out by Michael Attili
2302 *) Fix for HMAC. It wasn't zeroing the rest of the block if the key length
2303 was larger than the MD block size.
2304 [Steve Henson, pointed out by Yost William <YostW@tce.com>]
2306 *) Modernise PKCS12_parse() so it uses STACK_OF(X509) for its ca argument
2307 fix a leak when the ca argument was passed as NULL. Stop X509_PUBKEY_set()
2308 using the passed key: if the passed key was a private key the result
2309 of X509_print(), for example, would be to print out all the private key
2313 *) des_quad_cksum() byte order bug fix.
2314 [Ulf Möller, using the problem description in krb4-0.9.7, where
2315 the solution is attributed to Derrick J Brashear <shadow@DEMENTIA.ORG>]
2317 *) Fix so V_ASN1_APP_CHOOSE works again: however its use is strongly
2319 [Steve Henson, pointed out by Brian Korver <briank@cs.stanford.edu>]
2321 *) For easily testing in shell scripts whether some command
2322 'openssl XXX' exists, the new pseudo-command 'openssl no-XXX'
2323 returns with exit code 0 iff no command of the given name is available.
2324 'no-XXX' is printed in this case, 'XXX' otherwise. In both cases,
2325 the output goes to stdout and nothing is printed to stderr.
2326 Additional arguments are always ignored.
2328 Since for each cipher there is a command of the same name,
2329 the 'no-cipher' compilation switches can be tested this way.
2331 ('openssl no-XXX' is not able to detect pseudo-commands such
2332 as 'quit', 'list-XXX-commands', or 'no-XXX' itself.)
2335 *) Update test suite so that 'make test' succeeds in 'no-rsa' configuration.
2338 *) For SSL_[CTX_]set_tmp_dh, don't create a DH key if SSL_OP_SINGLE_DH_USE
2339 is set; it will be thrown away anyway because each handshake creates
2341 ssl_cert_dup, which is used by SSL_new, now copies DH keys in addition
2342 to parameters -- in previous versions (since OpenSSL 0.9.3) the
2343 'default key' from SSL_CTX_set_tmp_dh would always be lost, meanining
2344 you effectivly got SSL_OP_SINGLE_DH_USE when using this macro.
2347 *) New s_client option -ign_eof: EOF at stdin is ignored, and
2348 'Q' and 'R' lose their special meanings (quit/renegotiate).
2349 This is part of what -quiet does; unlike -quiet, -ign_eof
2350 does not suppress any output.
2353 *) Add compatibility options to the purpose and trust code. The
2354 purpose X509_PURPOSE_ANY is "any purpose" which automatically
2355 accepts a certificate or CA, this was the previous behaviour,
2356 with all the associated security issues.
2358 X509_TRUST_COMPAT is the old trust behaviour: only and
2359 automatically trust self signed roots in certificate store. A
2360 new trust setting X509_TRUST_DEFAULT is used to specify that
2361 a purpose has no associated trust setting and it should instead
2362 use the value in the default purpose.
2365 *) Fix the PKCS#8 DSA private key code so it decodes keys again
2366 and fix a memory leak.
2369 *) In util/mkerr.pl (which implements 'make errors'), preserve
2370 reason strings from the previous version of the .c file, as
2371 the default to have only downcase letters (and digits) in
2372 automatically generated reasons codes is not always appropriate.
2375 *) In ERR_load_ERR_strings(), build an ERR_LIB_SYS error reason table
2376 using strerror. Previously, ERR_reason_error_string() returned
2377 library names as reason strings for SYSerr; but SYSerr is a special
2378 case where small numbers are errno values, not library numbers.
2381 *) Add '-dsaparam' option to 'openssl dhparam' application. This
2382 converts DSA parameters into DH parameters. (When creating parameters,
2383 DSA_generate_parameters is used.)
2386 *) Include 'length' (recommended exponent length) in C code generated
2387 by 'openssl dhparam -C'.
2390 *) The second argument to set_label in perlasm was already being used
2391 so couldn't be used as a "file scope" flag. Moved to third argument
2395 *) In PEM_ASN1_write_bio and some other functions, use RAND_pseudo_bytes
2396 instead of RAND_bytes for encryption IVs and salts.
2399 *) Include RAND_status() into RAND_METHOD instead of implementing
2400 it only for md_rand.c Otherwise replacing the PRNG by calling
2401 RAND_set_rand_method would be impossible.
2404 *) Don't let DSA_generate_key() enter an infinite loop if the random
2405 number generation fails.
2408 *) New 'rand' application for creating pseudo-random output.
2411 *) Added configuration support for Linux/IA64
2412 [Rolf Haberrecker <rolf@suse.de>]
2414 *) Assembler module support for Mingw32.
2417 *) Shared library support for HPUX (in shlib/).
2418 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE> and Anonymous]
2420 *) Shared library support for Solaris gcc.
2421 [Lutz Behnke <behnke@trustcenter.de>]
2423 Changes between 0.9.4 and 0.9.5 [28 Feb 2000]
2425 *) PKCS7_encrypt() was adding text MIME headers twice because they
2426 were added manually and by SMIME_crlf_copy().
2429 *) In bntest.c don't call BN_rand with zero bits argument.
2430 [Steve Henson, pointed out by Andrew W. Gray <agray@iconsinc.com>]
2432 *) BN_mul bugfix: In bn_mul_part_recursion() only the a>a[n] && b>b[n]
2433 case was implemented. This caused BN_div_recp() to fail occasionally.
2436 *) Add an optional second argument to the set_label() in the perl
2437 assembly language builder. If this argument exists and is set
2438 to 1 it signals that the assembler should use a symbol whose
2439 scope is the entire file, not just the current function. This
2440 is needed with MASM which uses the format label:: for this scope.
2441 [Steve Henson, pointed out by Peter Runestig <peter@runestig.com>]
2443 *) Change the ASN1 types so they are typedefs by default. Before
2444 almost all types were #define'd to ASN1_STRING which was causing
2445 STACK_OF() problems: you couldn't declare STACK_OF(ASN1_UTF8STRING)
2449 *) Change names of new functions to the new get1/get0 naming
2450 convention: After 'get1', the caller owns a reference count
2451 and has to call ..._free; 'get0' returns a pointer to some
2452 data structure without incrementing reference counters.
2453 (Some of the existing 'get' functions increment a reference
2454 counter, some don't.)
2455 Similarly, 'set1' and 'add1' functions increase reference
2456 counters or duplicate objects.
2459 *) Allow for the possibility of temp RSA key generation failure:
2460 the code used to assume it always worked and crashed on failure.
2463 *) Fix potential buffer overrun problem in BIO_printf().
2464 [Ulf Möller, using public domain code by Patrick Powell; problem
2465 pointed out by David Sacerdote <das33@cornell.edu>]
2467 *) Support EGD <http://www.lothar.com/tech/crypto/>. New functions
2468 RAND_egd() and RAND_status(). In the command line application,
2469 the EGD socket can be specified like a seed file using RANDFILE
2473 *) Allow the string CERTIFICATE to be tolerated in PKCS#7 structures.
2474 Some CAs (e.g. Verisign) distribute certificates in this form.
2477 *) Remove the SSL_ALLOW_ADH compile option and set the default cipher
2478 list to exclude them. This means that no special compilation option
2479 is needed to use anonymous DH: it just needs to be included in the
2483 *) Change the EVP_MD_CTX_type macro so its meaning consistent with
2484 EVP_MD_type. The old functionality is available in a new macro called
2485 EVP_MD_md(). Change code that uses it and update docs.
2488 *) ..._ctrl functions now have corresponding ..._callback_ctrl functions
2489 where the 'void *' argument is replaced by a function pointer argument.
2490 Previously 'void *' was abused to point to functions, which works on
2491 many platforms, but is not correct. As these functions are usually
2492 called by macros defined in OpenSSL header files, most source code
2493 should work without changes.
2496 *) <openssl/opensslconf.h> (which is created by Configure) now contains
2497 sections with information on -D... compiler switches used for
2498 compiling the library so that applications can see them. To enable
2499 one of these sections, a pre-processor symbol OPENSSL_..._DEFINES
2500 must be defined. E.g.,
2501 #define OPENSSL_ALGORITHM_DEFINES
2502 #include <openssl/opensslconf.h>
2503 defines all pertinent NO_<algo> symbols, such as NO_IDEA, NO_RSA, etc.
2504 [Richard Levitte, Ulf and Bodo Möller]
2506 *) Bugfix: Tolerate fragmentation and interleaving in the SSL 3/TLS
2510 *) Change the 'other' type in certificate aux info to a STACK_OF
2511 X509_ALGOR. Although not an AlgorithmIdentifier as such it has
2512 the required ASN1 format: arbitrary types determined by an OID.
2515 *) Add some PEM_write_X509_REQ_NEW() functions and a command line
2516 argument to 'req'. This is not because the function is newer or
2517 better than others it just uses the work 'NEW' in the certificate
2518 request header lines. Some software needs this.
2521 *) Reorganise password command line arguments: now passwords can be
2522 obtained from various sources. Delete the PEM_cb function and make
2523 it the default behaviour: i.e. if the callback is NULL and the
2524 usrdata argument is not NULL interpret it as a null terminated pass
2525 phrase. If usrdata and the callback are NULL then the pass phrase
2526 is prompted for as usual.
2529 *) Add support for the Compaq Atalla crypto accelerator. If it is installed,
2530 the support is automatically enabled. The resulting binaries will
2531 autodetect the card and use it if present.
2532 [Ben Laurie and Compaq Inc.]
2534 *) Work around for Netscape hang bug. This sends certificate request
2535 and server done in one record. Since this is perfectly legal in the
2536 SSL/TLS protocol it isn't a "bug" option and is on by default. See
2537 the bugs/SSLv3 entry for more info.
2540 *) HP-UX tune-up: new unified configs, HP C compiler bug workaround.
2543 *) Add -rand argument to smime and pkcs12 applications and read/write
2547 *) New 'passwd' tool for crypt(3) and apr1 password hashes.
2550 *) Add command line password options to the remaining applications.
2553 *) Bug fix for BN_div_recp() for numerators with an even number of
2557 *) More tests in bntest.c, and changed test_bn output.
2560 *) ./config recognizes MacOS X now.
2563 *) Bug fix for BN_div() when the first words of num and divsor are
2564 equal (it gave wrong results if (rem=(n1-q*d0)&BN_MASK2) < d0).
2567 *) Add support for various broken PKCS#8 formats, and command line
2568 options to produce them.
2571 *) New functions BN_CTX_start(), BN_CTX_get() and BT_CTX_end() to
2572 get temporary BIGNUMs from a BN_CTX.
2575 *) Correct return values in BN_mod_exp_mont() and BN_mod_exp2_mont()
2579 *) Change the SSLeay_add_all_*() functions to OpenSSL_add_all_*() and
2580 include a #define from the old name to the new. The original intent
2581 was that statically linked binaries could for example just call
2582 SSLeay_add_all_ciphers() to just add ciphers to the table and not
2583 link with digests. This never worked becayse SSLeay_add_all_digests()
2584 and SSLeay_add_all_ciphers() were in the same source file so calling
2585 one would link with the other. They are now in separate source files.
2588 *) Add a new -notext option to 'ca' and a -pubkey option to 'spkac'.
2591 *) Use a less unusual form of the Miller-Rabin primality test (it used
2592 a binary algorithm for exponentiation integrated into the Miller-Rabin
2593 loop, our standard modexp algorithms are faster).
2596 *) Support for the EBCDIC character set completed.
2597 [Martin Kraemer <Martin.Kraemer@Mch.SNI.De>]
2599 *) Source code cleanups: use const where appropriate, eliminate casts,
2600 use void * instead of char * in lhash.
2603 *) Bugfix: ssl3_send_server_key_exchange was not restartable
2604 (the state was not changed to SSL3_ST_SW_KEY_EXCH_B, and because of
2605 this the server could overwrite ephemeral keys that the client
2609 *) Turn DSA_is_prime into a macro that calls BN_is_prime,
2610 using 50 iterations of the Rabin-Miller test.
2612 DSA_generate_parameters now uses BN_is_prime_fasttest (with 50
2613 iterations of the Rabin-Miller test as required by the appendix
2614 to FIPS PUB 186[-1]) instead of DSA_is_prime.
2615 As BN_is_prime_fasttest includes trial division, DSA parameter
2616 generation becomes much faster.
2618 This implies a change for the callback functions in DSA_is_prime
2619 and DSA_generate_parameters: The callback function is called once
2620 for each positive witness in the Rabin-Miller test, not just
2621 occasionally in the inner loop; and the parameters to the
2622 callback function now provide an iteration count for the outer
2623 loop rather than for the current invocation of the inner loop.
2624 DSA_generate_parameters additionally can call the callback
2625 function with an 'iteration count' of -1, meaning that a
2626 candidate has passed the trial division test (when q is generated
2627 from an application-provided seed, trial division is skipped).
2630 *) New function BN_is_prime_fasttest that optionally does trial
2631 division before starting the Rabin-Miller test and has
2632 an additional BN_CTX * argument (whereas BN_is_prime always
2633 has to allocate at least one BN_CTX).
2634 'callback(1, -1, cb_arg)' is called when a number has passed the
2635 trial division stage.
2638 *) Fix for bug in CRL encoding. The validity dates weren't being handled
2642 *) New -pkcs12 option to CA.pl script to write out a PKCS#12 file.
2645 *) New function BN_pseudo_rand().
2648 *) Clean up BN_mod_mul_montgomery(): replace the broken (and unreadable)
2649 bignum version of BN_from_montgomery() with the working code from
2650 SSLeay 0.9.0 (the word based version is faster anyway), and clean up
2654 *) Avoid a race condition in s2_clnt.c (function get_server_hello) that
2655 made it impossible to use the same SSL_SESSION data structure in
2656 SSL2 clients in multiple threads.
2659 *) The return value of RAND_load_file() no longer counts bytes obtained
2660 by stat(). RAND_load_file(..., -1) is new and uses the complete file
2661 to seed the PRNG (previously an explicit byte count was required).
2662 [Ulf Möller, Bodo Möller]
2664 *) Clean up CRYPTO_EX_DATA functions, some of these didn't have prototypes
2665 used (char *) instead of (void *) and had casts all over the place.
2668 *) Make BN_generate_prime() return NULL on error if ret!=NULL.
2671 *) Retain source code compatibility for BN_prime_checks macro:
2672 BN_is_prime(..., BN_prime_checks, ...) now uses
2673 BN_prime_checks_for_size to determine the appropriate number of
2674 Rabin-Miller iterations.
2677 *) Diffie-Hellman uses "safe" primes: DH_check() return code renamed to
2678 DH_CHECK_P_NOT_SAFE_PRIME.
2679 (Check if this is true? OpenPGP calls them "strong".)
2682 *) Merge the functionality of "dh" and "gendh" programs into a new program
2683 "dhparam". The old programs are retained for now but will handle DH keys
2684 (instead of parameters) in future.
2687 *) Make the ciphers, s_server and s_client programs check the return values
2688 when a new cipher list is set.
2691 *) Enhance the SSL/TLS cipher mechanism to correctly handle the TLS 56bit
2692 ciphers. Before when the 56bit ciphers were enabled the sorting was
2695 The syntax for the cipher sorting has been extended to support sorting by
2696 cipher-strength (using the strength_bits hard coded in the tables).
2697 The new command is "@STRENGTH" (see also doc/apps/ciphers.pod).
2699 Fix a bug in the cipher-command parser: when supplying a cipher command
2700 string with an "undefined" symbol (neither command nor alphanumeric
2701 [A-Za-z0-9], ssl_set_cipher_list used to hang in an endless loop. Now
2702 an error is flagged.
2704 Due to the strength-sorting extension, the code of the
2705 ssl_create_cipher_list() function was completely rearranged. I hope that
2706 the readability was also increased :-)
2707 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>]
2709 *) Minor change to 'x509' utility. The -CAcreateserial option now uses 1
2710 for the first serial number and places 2 in the serial number file. This
2711 avoids problems when the root CA is created with serial number zero and
2712 the first user certificate has the same issuer name and serial number
2716 *) Fixes to X509_ATTRIBUTE utilities, change the 'req' program so it uses
2717 the new code. Add documentation for this stuff.
2720 *) Changes to X509_ATTRIBUTE utilities. These have been renamed from
2721 X509_*() to X509at_*() on the grounds that they don't handle X509
2722 structures and behave in an analagous way to the X509v3 functions:
2723 they shouldn't be called directly but wrapper functions should be used
2726 So we also now have some wrapper functions that call the X509at functions
2727 when passed certificate requests. (TO DO: similar things can be done with
2728 PKCS#7 signed and unsigned attributes, PKCS#12 attributes and a few other
2729 things. Some of these need some d2i or i2d and print functionality
2730 because they handle more complex structures.)
2733 *) Add missing #ifndefs that caused missing symbols when building libssl
2734 as a shared library without RSA. Use #ifndef NO_SSL2 instead of
2735 NO_RSA in ssl/s2*.c.
2736 [Kris Kennaway <kris@hub.freebsd.org>, modified by Ulf Möller]
2738 *) Precautions against using the PRNG uninitialized: RAND_bytes() now
2739 has a return value which indicates the quality of the random data
2740 (1 = ok, 0 = not seeded). Also an error is recorded on the thread's
2741 error queue. New function RAND_pseudo_bytes() generates output that is
2742 guaranteed to be unique but not unpredictable. RAND_add is like
2743 RAND_seed, but takes an extra argument for an entropy estimate
2744 (RAND_seed always assumes full entropy).
2747 *) Do more iterations of Rabin-Miller probable prime test (specifically,
2748 3 for 1024-bit primes, 6 for 512-bit primes, 12 for 256-bit primes
2749 instead of only 2 for all lengths; see BN_prime_checks_for_size definition
2750 in crypto/bn/bn_prime.c for the complete table). This guarantees a
2751 false-positive rate of at most 2^-80 for random input.
2754 *) Rewrite ssl3_read_n (ssl/s3_pkt.c) avoiding a couple of bugs.
2757 *) New function X509_CTX_rget_chain() (renamed to X509_CTX_get1_chain
2758 in the 0.9.5 release), this returns the chain
2759 from an X509_CTX structure with a dup of the stack and all
2760 the X509 reference counts upped: so the stack will exist
2761 after X509_CTX_cleanup() has been called. Modify pkcs12.c
2764 Also make SSL_SESSION_print() print out the verify return
2768 *) Add manpage for the pkcs12 command. Also change the default
2769 behaviour so MAC iteration counts are used unless the new
2770 -nomaciter option is used. This improves file security and
2771 only older versions of MSIE (4.0 for example) need it.
2774 *) Honor the no-xxx Configure options when creating .DEF files.
2777 *) Add PKCS#10 attributes to field table: challengePassword,
2778 unstructuredName and unstructuredAddress. These are taken from
2779 draft PKCS#9 v2.0 but are compatible with v1.2 provided no
2780 international characters are used.
2782 More changes to X509_ATTRIBUTE code: allow the setting of types
2783 based on strings. Remove the 'loc' parameter when adding
2784 attributes because these will be a SET OF encoding which is sorted
2788 *) Initial changes to the 'req' utility to allow request generation
2789 automation. This will allow an application to just generate a template
2790 file containing all the field values and have req construct the
2793 Initial support for X509_ATTRIBUTE handling. Stacks of these are
2794 used all over the place including certificate requests and PKCS#7
2795 structures. They are currently handled manually where necessary with
2796 some primitive wrappers for PKCS#7. The new functions behave in a
2797 manner analogous to the X509 extension functions: they allow
2798 attributes to be looked up by NID and added.
2800 Later something similar to the X509V3 code would be desirable to
2801 automatically handle the encoding, decoding and printing of the
2802 more complex types. The string types like challengePassword can
2803 be handled by the string table functions.
2805 Also modified the multi byte string table handling. Now there is
2806 a 'global mask' which masks out certain types. The table itself
2807 can use the flag STABLE_NO_MASK to ignore the mask setting: this
2808 is useful when for example there is only one permissible type
2809 (as in countryName) and using the mask might result in no valid
2813 *) Clean up 'Finished' handling, and add functions SSL_get_finished and
2814 SSL_get_peer_finished to allow applications to obtain the latest
2815 Finished messages sent to the peer or expected from the peer,
2816 respectively. (SSL_get_peer_finished is usually the Finished message
2817 actually received from the peer, otherwise the protocol will be aborted.)
2819 As the Finished message are message digests of the complete handshake
2820 (with a total of 192 bits for TLS 1.0 and more for SSL 3.0), they can
2821 be used for external authentication procedures when the authentication
2822 provided by SSL/TLS is not desired or is not enough.
2825 *) Enhanced support for Alpha Linux is added. Now ./config checks if
2826 the host supports BWX extension and if Compaq C is present on the
2827 $PATH. Just exploiting of the BWX extension results in 20-30%
2828 performance kick for some algorithms, e.g. DES and RC4 to mention
2829 a couple. Compaq C in turn generates ~20% faster code for MD5 and
2833 *) Add support for MS "fast SGC". This is arguably a violation of the
2834 SSL3/TLS protocol. Netscape SGC does two handshakes: the first with
2835 weak crypto and after checking the certificate is SGC a second one
2836 with strong crypto. MS SGC stops the first handshake after receiving
2837 the server certificate message and sends a second client hello. Since
2838 a server will typically do all the time consuming operations before
2839 expecting any further messages from the client (server key exchange
2840 is the most expensive) there is little difference between the two.
2842 To get OpenSSL to support MS SGC we have to permit a second client
2843 hello message after we have sent server done. In addition we have to
2844 reset the MAC if we do get this second client hello.
2847 *) Add a function 'd2i_AutoPrivateKey()' this will automatically decide
2848 if a DER encoded private key is RSA or DSA traditional format. Changed
2849 d2i_PrivateKey_bio() to use it. This is only needed for the "traditional"
2850 format DER encoded private key. Newer code should use PKCS#8 format which
2851 has the key type encoded in the ASN1 structure. Added DER private key
2852 support to pkcs8 application.
2855 *) SSL 3/TLS 1 servers now don't request certificates when an anonymous
2856 ciphersuites has been selected (as required by the SSL 3/TLS 1
2857 specifications). Exception: When SSL_VERIFY_FAIL_IF_NO_PEER_CERT
2858 is set, we interpret this as a request to violate the specification
2859 (the worst that can happen is a handshake failure, and 'correct'
2860 behaviour would result in a handshake failure anyway).
2863 *) In SSL_CTX_add_session, take into account that there might be multiple
2864 SSL_SESSION structures with the same session ID (e.g. when two threads
2865 concurrently obtain them from an external cache).
2866 The internal cache can handle only one SSL_SESSION with a given ID,
2867 so if there's a conflict, we now throw out the old one to achieve
2871 *) Add OIDs for idea and blowfish in CBC mode. This will allow both
2872 to be used in PKCS#5 v2.0 and S/MIME. Also add checking to
2873 some routines that use cipher OIDs: some ciphers do not have OIDs
2874 defined and so they cannot be used for S/MIME and PKCS#5 v2.0 for
2878 *) Simplify the trust setting structure and code. Now we just have
2879 two sequences of OIDs for trusted and rejected settings. These will
2880 typically have values the same as the extended key usage extension
2881 and any application specific purposes.
2883 The trust checking code now has a default behaviour: it will just
2884 check for an object with the same NID as the passed id. Functions can
2885 be provided to override either the default behaviour or the behaviour
2886 for a given id. SSL client, server and email already have functions
2887 in place for compatibility: they check the NID and also return "trusted"
2888 if the certificate is self signed.
2891 *) Add d2i,i2d bio/fp functions for PrivateKey: these convert the
2892 traditional format into an EVP_PKEY structure.
2895 *) Add a password callback function PEM_cb() which either prompts for
2896 a password if usr_data is NULL or otherwise assumes it is a null
2897 terminated password. Allow passwords to be passed on command line
2898 environment or config files in a few more utilities.
2901 *) Add a bunch of DER and PEM functions to handle PKCS#8 format private
2902 keys. Add some short names for PKCS#8 PBE algorithms and allow them
2903 to be specified on the command line for the pkcs8 and pkcs12 utilities.
2904 Update documentation.
2907 *) Support for ASN1 "NULL" type. This could be handled before by using
2908 ASN1_TYPE but there wasn't any function that would try to read a NULL
2909 and produce an error if it couldn't. For compatibility we also have
2910 ASN1_NULL_new() and ASN1_NULL_free() functions but these are faked and
2911 don't allocate anything because they don't need to.
2914 *) Initial support for MacOS is now provided. Examine INSTALL.MacOS
2916 [Andy Polyakov, Roy Woods <roy@centicsystems.ca>]
2918 *) Rebuild of the memory allocation routines used by OpenSSL code and
2919 possibly others as well. The purpose is to make an interface that
2920 provide hooks so anyone can build a separate set of allocation and
2921 deallocation routines to be used by OpenSSL, for example memory
2922 pool implementations, or something else, which was previously hard
2923 since Malloc(), Realloc() and Free() were defined as macros having
2924 the values malloc, realloc and free, respectively (except for Win32
2925 compilations). The same is provided for memory debugging code.
2926 OpenSSL already comes with functionality to find memory leaks, but
2927 this gives people a chance to debug other memory problems.
2929 With these changes, a new set of functions and macros have appeared:
2931 CRYPTO_set_mem_debug_functions() [F]
2932 CRYPTO_get_mem_debug_functions() [F]
2933 CRYPTO_dbg_set_options() [F]
2934 CRYPTO_dbg_get_options() [F]
2935 CRYPTO_malloc_debug_init() [M]
2937 The memory debug functions are NULL by default, unless the library
2938 is compiled with CRYPTO_MDEBUG or friends is defined. If someone
2939 wants to debug memory anyway, CRYPTO_malloc_debug_init() (which
2940 gives the standard debugging functions that come with OpenSSL) or
2941 CRYPTO_set_mem_debug_functions() (tells OpenSSL to use functions
2942 provided by the library user) must be used. When the standard
2943 debugging functions are used, CRYPTO_dbg_set_options can be used to
2944 request additional information:
2945 CRYPTO_dbg_set_options(V_CYRPTO_MDEBUG_xxx) corresponds to setting
2946 the CRYPTO_MDEBUG_xxx macro when compiling the library.
2948 Also, things like CRYPTO_set_mem_functions will always give the
2949 expected result (the new set of functions is used for allocation
2950 and deallocation) at all times, regardless of platform and compiler
2953 To finish it up, some functions that were never use in any other
2954 way than through macros have a new API and new semantic:
2957 CRYPTO_dbg_realloc()
2960 All macros of value have retained their old syntax.
2961 [Richard Levitte and Bodo Moeller]
2963 *) Some S/MIME fixes. The OID for SMIMECapabilities was wrong, the
2964 ordering of SMIMECapabilities wasn't in "strength order" and there
2965 was a missing NULL in the AlgorithmIdentifier for the SHA1 signature
2969 *) Some ASN1 types with illegal zero length encoding (INTEGER,
2970 ENUMERATED and OBJECT IDENTIFIER) choked the ASN1 routines.
2971 [Frans Heymans <fheymans@isaserver.be>, modified by Steve Henson]
2973 *) Merge in my S/MIME library for OpenSSL. This provides a simple
2974 S/MIME API on top of the PKCS#7 code, a MIME parser (with enough
2975 functionality to handle multipart/signed properly) and a utility
2976 called 'smime' to call all this stuff. This is based on code I
2977 originally wrote for Celo who have kindly allowed it to be
2978 included in OpenSSL.
2981 *) Add variants des_set_key_checked and des_set_key_unchecked of
2982 des_set_key (aka des_key_sched). Global variable des_check_key
2983 decides which of these is called by des_set_key; this way
2984 des_check_key behaves as it always did, but applications and
2985 the library itself, which was buggy for des_check_key == 1,
2986 have a cleaner way to pick the version they need.
2989 *) New function PKCS12_newpass() which changes the password of a
2993 *) Modify X509_TRUST and X509_PURPOSE so it also uses a static and
2994 dynamic mix. In both cases the ids can be used as an index into the
2995 table. Also modified the X509_TRUST_add() and X509_PURPOSE_add()
2996 functions so they accept a list of the field values and the
2997 application doesn't need to directly manipulate the X509_TRUST
3001 *) Modify the ASN1_STRING_TABLE stuff so it also uses bsearch and doesn't
3005 *) Modify the way the V3 extension code looks up extensions. This now
3006 works in a similar way to the object code: we have some "standard"
3007 extensions in a static table which is searched with OBJ_bsearch()
3008 and the application can add dynamic ones if needed. The file
3009 crypto/x509v3/ext_dat.h now has the info: this file needs to be
3010 updated whenever a new extension is added to the core code and kept
3011 in ext_nid order. There is a simple program 'tabtest.c' which checks
3012 this. New extensions are not added too often so this file can readily
3013 be maintained manually.
3015 There are two big advantages in doing things this way. The extensions
3016 can be looked up immediately and no longer need to be "added" using
3017 X509V3_add_standard_extensions(): this function now does nothing.
3018 [Side note: I get *lots* of email saying the extension code doesn't
3019 work because people forget to call this function]
3020 Also no dynamic allocation is done unless new extensions are added:
3021 so if we don't add custom extensions there is no need to call
3022 X509V3_EXT_cleanup().
3025 *) Modify enc utility's salting as follows: make salting the default. Add a
3026 magic header, so unsalted files fail gracefully instead of just decrypting
3027 to garbage. This is because not salting is a big security hole, so people
3028 should be discouraged from doing it.
3031 *) Fixes and enhancements to the 'x509' utility. It allowed a message
3032 digest to be passed on the command line but it only used this
3033 parameter when signing a certificate. Modified so all relevant
3034 operations are affected by the digest parameter including the
3035 -fingerprint and -x509toreq options. Also -x509toreq choked if a
3036 DSA key was used because it didn't fix the digest.
3039 *) Initial certificate chain verify code. Currently tests the untrusted
3040 certificates for consistency with the verify purpose (which is set
3041 when the X509_STORE_CTX structure is set up) and checks the pathlength.
3043 There is a NO_CHAIN_VERIFY compilation option to keep the old behaviour:
3044 this is because it will reject chains with invalid extensions whereas
3045 every previous version of OpenSSL and SSLeay made no checks at all.
3047 Trust code: checks the root CA for the relevant trust settings. Trust
3048 settings have an initial value consistent with the verify purpose: e.g.
3049 if the verify purpose is for SSL client use it expects the CA to be
3050 trusted for SSL client use. However the default value can be changed to
3051 permit custom trust settings: one example of this would be to only trust
3052 certificates from a specific "secure" set of CAs.
3054 Also added X509_STORE_CTX_new() and X509_STORE_CTX_free() functions
3055 which should be used for version portability: especially since the
3056 verify structure is likely to change more often now.
3058 SSL integration. Add purpose and trust to SSL_CTX and SSL and functions
3059 to set them. If not set then assume SSL clients will verify SSL servers
3062 Two new options to the verify program: -untrusted allows a set of
3063 untrusted certificates to be passed in and -purpose which sets the
3064 intended purpose of the certificate. If a purpose is set then the
3065 new chain verify code is used to check extension consistency.
3068 *) Support for the authority information access extension.
3071 *) Modify RSA and DSA PEM read routines to transparently handle
3072 PKCS#8 format private keys. New *_PUBKEY_* functions that handle
3073 public keys in a format compatible with certificate
3074 SubjectPublicKeyInfo structures. Unfortunately there were already
3075 functions called *_PublicKey_* which used various odd formats so
3076 these are retained for compatibility: however the DSA variants were
3077 never in a public release so they have been deleted. Changed dsa/rsa
3078 utilities to handle the new format: note no releases ever handled public
3079 keys so we should be OK.
3081 The primary motivation for this change is to avoid the same fiasco
3082 that dogs private keys: there are several incompatible private key
3083 formats some of which are standard and some OpenSSL specific and
3084 require various evil hacks to allow partial transparent handling and
3085 even then it doesn't work with DER formats. Given the option anything
3086 other than PKCS#8 should be dumped: but the other formats have to
3087 stay in the name of compatibility.
3089 With public keys and the benefit of hindsight one standard format
3090 is used which works with EVP_PKEY, RSA or DSA structures: though
3091 it clearly returns an error if you try to read the wrong kind of key.
3093 Added a -pubkey option to the 'x509' utility to output the public key.
3094 Also rename the EVP_PKEY_get_*() to EVP_PKEY_rget_*()
3095 (renamed to EVP_PKEY_get1_*() in the OpenSSL 0.9.5 release) and add
3096 EVP_PKEY_rset_*() functions (renamed to EVP_PKEY_set1_*())
3097 that do the same as the EVP_PKEY_assign_*() except they up the
3098 reference count of the added key (they don't "swallow" the
3102 *) Fixes to crypto/x509/by_file.c the code to read in certificates and
3103 CRLs would fail if the file contained no certificates or no CRLs:
3104 added a new function to read in both types and return the number
3105 read: this means that if none are read it will be an error. The
3106 DER versions of the certificate and CRL reader would always fail
3107 because it isn't possible to mix certificates and CRLs in DER format
3108 without choking one or the other routine. Changed this to just read
3109 a certificate: this is the best we can do. Also modified the code
3110 in apps/verify.c to take notice of return codes: it was previously
3111 attempting to read in certificates from NULL pointers and ignoring
3112 any errors: this is one reason why the cert and CRL reader seemed
3113 to work. It doesn't check return codes from the default certificate
3114 routines: these may well fail if the certificates aren't installed.
3117 *) Code to support otherName option in GeneralName.
3120 *) First update to verify code. Change the verify utility
3121 so it warns if it is passed a self signed certificate:
3122 for consistency with the normal behaviour. X509_verify
3123 has been modified to it will now verify a self signed
3124 certificate if *exactly* the same certificate appears
3125 in the store: it was previously impossible to trust a
3126 single self signed certificate. This means that:
3127 openssl verify ss.pem
3128 now gives a warning about a self signed certificate but
3129 openssl verify -CAfile ss.pem ss.pem
3133 *) For servers, store verify_result in SSL_SESSION data structure
3134 (and add it to external session representation).
3135 This is needed when client certificate verifications fails,
3136 but an application-provided verification callback (set by
3137 SSL_CTX_set_cert_verify_callback) allows accepting the session
3138 anyway (i.e. leaves x509_store_ctx->error != X509_V_OK
3139 but returns 1): When the session is reused, we have to set
3140 ssl->verify_result to the appropriate error code to avoid
3142 [Bodo Moeller, problem pointed out by Lutz Jaenicke]
3144 *) Fix a bug in the new PKCS#7 code: it didn't consider the
3145 case in PKCS7_dataInit() where the signed PKCS7 structure
3146 didn't contain any existing data because it was being created.
3147 [Po-Cheng Chen <pocheng@nst.com.tw>, slightly modified by Steve Henson]
3149 *) Add a salt to the key derivation routines in enc.c. This
3150 forms the first 8 bytes of the encrypted file. Also add a
3151 -S option to allow a salt to be input on the command line.
3154 *) New function X509_cmp(). Oddly enough there wasn't a function
3155 to compare two certificates. We do this by working out the SHA1
3156 hash and comparing that. X509_cmp() will be needed by the trust
3160 *) SSL_get1_session() is like SSL_get_session(), but increments
3161 the reference count in the SSL_SESSION returned.
3162 [Geoff Thorpe <geoff@eu.c2.net>]
3164 *) Fix for 'req': it was adding a null to request attributes.
3165 Also change the X509_LOOKUP and X509_INFO code to handle
3166 certificate auxiliary information.
3169 *) Add support for 40 and 64 bit RC2 and RC4 algorithms: document
3173 *) Add the possibility to add extra information to the memory leak
3174 detecting output, to form tracebacks, showing from where each
3175 allocation was originated: CRYPTO_push_info("constant string") adds
3176 the string plus current file name and line number to a per-thread
3177 stack, CRYPTO_pop_info() does the obvious, CRYPTO_remove_all_info()
3178 is like calling CYRPTO_pop_info() until the stack is empty.
3179 Also updated memory leak detection code to be multi-thread-safe.
3182 *) Add options -text and -noout to pkcs7 utility and delete the
3183 encryption options which never did anything. Update docs.
3186 *) Add options to some of the utilities to allow the pass phrase
3187 to be included on either the command line (not recommended on
3188 OSes like Unix) or read from the environment. Update the
3189 manpages and fix a few bugs.
3192 *) Add a few manpages for some of the openssl commands.
3195 *) Fix the -revoke option in ca. It was freeing up memory twice,
3196 leaking and not finding already revoked certificates.
3199 *) Extensive changes to support certificate auxiliary information.
3200 This involves the use of X509_CERT_AUX structure and X509_AUX
3201 functions. An X509_AUX function such as PEM_read_X509_AUX()
3202 can still read in a certificate file in the usual way but it
3203 will also read in any additional "auxiliary information". By
3204 doing things this way a fair degree of compatibility can be
3205 retained: existing certificates can have this information added
3206 using the new 'x509' options.
3208 Current auxiliary information includes an "alias" and some trust
3209 settings. The trust settings will ultimately be used in enhanced
3210 certificate chain verification routines: currently a certificate
3211 can only be trusted if it is self signed and then it is trusted
3215 *) Fix assembler for Alpha (tested only on DEC OSF not Linux or *BSD).
3216 The problem was that one of the replacement routines had not been working
3217 since SSLeay releases. For now the offending routine has been replaced
3218 with non-optimised assembler. Even so, this now gives around 95%
3219 performance improvement for 1024 bit RSA signs.
3222 *) Hack to fix PKCS#7 decryption when used with some unorthodox RC2
3223 handling. Most clients have the effective key size in bits equal to
3224 the key length in bits: so a 40 bit RC2 key uses a 40 bit (5 byte) key.
3225 A few however don't do this and instead use the size of the decrypted key
3226 to determine the RC2 key length and the AlgorithmIdentifier to determine
3227 the effective key length. In this case the effective key length can still
3228 be 40 bits but the key length can be 168 bits for example. This is fixed
3229 by manually forcing an RC2 key into the EVP_PKEY structure because the
3230 EVP code can't currently handle unusual RC2 key sizes: it always assumes
3231 the key length and effective key length are equal.
3234 *) Add a bunch of functions that should simplify the creation of
3235 X509_NAME structures. Now you should be able to do:
3236 X509_NAME_add_entry_by_txt(nm, "CN", MBSTRING_ASC, "Steve", -1, -1, 0);
3237 and have it automatically work out the correct field type and fill in
3238 the structures. The more adventurous can try:
3239 X509_NAME_add_entry_by_txt(nm, field, MBSTRING_UTF8, str, -1, -1, 0);
3240 and it will (hopefully) work out the correct multibyte encoding.
3243 *) Change the 'req' utility to use the new field handling and multibyte
3244 copy routines. Before the DN field creation was handled in an ad hoc
3245 way in req, ca, and x509 which was rather broken and didn't support
3246 BMPStrings or UTF8Strings. Since some software doesn't implement
3247 BMPStrings or UTF8Strings yet, they can be enabled using the config file
3248 using the dirstring_type option. See the new comment in the default
3249 openssl.cnf for more info.
3252 *) Make crypto/rand/md_rand.c more robust:
3253 - Assure unique random numbers after fork().
3254 - Make sure that concurrent threads access the global counter and
3255 md serializably so that we never lose entropy in them
3256 or use exactly the same state in multiple threads.
3257 Access to the large state is not always serializable because
3258 the additional locking could be a performance killer, and
3259 md should be large enough anyway.
3262 *) New file apps/app_rand.c with commonly needed functionality
3263 for handling the random seed file.
3265 Use the random seed file in some applications that previously did not:
3267 dsaparam -genkey (which also ignored its '-rand' option),
3270 x509 (when signing).
3271 Except on systems with /dev/urandom, it is crucial to have a random
3272 seed file at least for key creation, DSA signing, and for DH exchanges;
3273 for RSA signatures we could do without one.
3275 gendh and gendsa (unlike genrsa) used to read only the first byte
3276 of each file listed in the '-rand' option. The function as previously
3277 found in genrsa is now in app_rand.c and is used by all programs
3278 that support '-rand'.
3281 *) In RAND_write_file, use mode 0600 for creating files;
3282 don't just chmod when it may be too late.
3285 *) Report an error from X509_STORE_load_locations
3286 when X509_LOOKUP_load_file or X509_LOOKUP_add_dir failed.
3289 *) New function ASN1_mbstring_copy() this copies a string in either
3290 ASCII, Unicode, Universal (4 bytes per character) or UTF8 format
3291 into an ASN1_STRING type. A mask of permissible types is passed
3292 and it chooses the "minimal" type to use or an error if not type
3296 *) Add function equivalents to the various macros in asn1.h. The old
3297 macros are retained with an M_ prefix. Code inside the library can
3298 use the M_ macros. External code (including the openssl utility)
3299 should *NOT* in order to be "shared library friendly".
3302 *) Add various functions that can check a certificate's extensions
3303 to see if it usable for various purposes such as SSL client,
3304 server or S/MIME and CAs of these types. This is currently
3305 VERY EXPERIMENTAL but will ultimately be used for certificate chain
3306 verification. Also added a -purpose flag to x509 utility to
3307 print out all the purposes.
3310 *) Add a CRYPTO_EX_DATA to X509 certificate structure and associated
3314 *) New X509V3_{X509,CRL,REVOKED}_get_d2i() functions. These will search
3315 for, obtain and decode and extension and obtain its critical flag.
3316 This allows all the necessary extension code to be handled in a
3317 single function call.
3320 *) RC4 tune-up featuring 30-40% performance improvement on most RISC
3321 platforms. See crypto/rc4/rc4_enc.c for further details.
3324 *) New -noout option to asn1parse. This causes no output to be produced
3325 its main use is when combined with -strparse and -out to extract data
3326 from a file (which may not be in ASN.1 format).
3329 *) Fix for pkcs12 program. It was hashing an invalid certificate pointer
3330 when producing the local key id.
3331 [Richard Levitte <levitte@stacken.kth.se>]
3333 *) New option -dhparam in s_server. This allows a DH parameter file to be
3334 stated explicitly. If it is not stated then it tries the first server
3335 certificate file. The previous behaviour hard coded the filename
3339 *) Add -pubin and -pubout options to the rsa and dsa commands. These allow
3340 a public key to be input or output. For example:
3341 openssl rsa -in key.pem -pubout -out pubkey.pem
3342 Also added necessary DSA public key functions to handle this.
3345 *) Fix so PKCS7_dataVerify() doesn't crash if no certificates are contained
3346 in the message. This was handled by allowing
3347 X509_find_by_issuer_and_serial() to tolerate a NULL passed to it.
3348 [Steve Henson, reported by Sampo Kellomaki <sampo@mail.neuronio.pt>]
3350 *) Fix for bug in d2i_ASN1_bytes(): other ASN1 functions add an extra null
3351 to the end of the strings whereas this didn't. This would cause problems
3352 if strings read with d2i_ASN1_bytes() were later modified.
3353 [Steve Henson, reported by Arne Ansper <arne@ats.cyber.ee>]
3355 *) Fix for base64 decode bug. When a base64 bio reads only one line of
3356 data and it contains EOF it will end up returning an error. This is
3357 caused by input 46 bytes long. The cause is due to the way base64
3358 BIOs find the start of base64 encoded data. They do this by trying a
3359 trial decode on each line until they find one that works. When they
3360 do a flag is set and it starts again knowing it can pass all the
3361 data directly through the decoder. Unfortunately it doesn't reset
3362 the context it uses. This means that if EOF is reached an attempt
3363 is made to pass two EOFs through the context and this causes the
3364 resulting error. This can also cause other problems as well. As is
3365 usual with these problems it takes *ages* to find and the fix is
3366 trivial: move one line.
3367 [Steve Henson, reported by ian@uns.ns.ac.yu (Ivan Nejgebauer) ]
3369 *) Ugly workaround to get s_client and s_server working under Windows. The
3370 old code wouldn't work because it needed to select() on sockets and the
3371 tty (for keypresses and to see if data could be written). Win32 only
3372 supports select() on sockets so we select() with a 1s timeout on the
3373 sockets and then see if any characters are waiting to be read, if none
3374 are present then we retry, we also assume we can always write data to
3375 the tty. This isn't nice because the code then blocks until we've
3376 received a complete line of data and it is effectively polling the
3377 keyboard at 1s intervals: however it's quite a bit better than not
3378 working at all :-) A dedicated Windows application might handle this
3379 with an event loop for example.
3382 *) Enhance RSA_METHOD structure. Now there are two extra methods, rsa_sign
3383 and rsa_verify. When the RSA_FLAGS_SIGN_VER option is set these functions
3384 will be called when RSA_sign() and RSA_verify() are used. This is useful
3385 if rsa_pub_dec() and rsa_priv_enc() equivalents are not available.
3386 For this to work properly RSA_public_decrypt() and RSA_private_encrypt()
3387 should *not* be used: RSA_sign() and RSA_verify() must be used instead.
3388 This necessitated the support of an extra signature type NID_md5_sha1
3389 for SSL signatures and modifications to the SSL library to use it instead
3390 of calling RSA_public_decrypt() and RSA_private_encrypt().
3393 *) Add new -verify -CAfile and -CApath options to the crl program, these
3394 will lookup a CRL issuers certificate and verify the signature in a
3395 similar way to the verify program. Tidy up the crl program so it
3396 no longer accesses structures directly. Make the ASN1 CRL parsing a bit
3397 less strict. It will now permit CRL extensions even if it is not
3398 a V2 CRL: this will allow it to tolerate some broken CRLs.
3401 *) Initialize all non-automatic variables each time one of the openssl
3402 sub-programs is started (this is necessary as they may be started
3403 multiple times from the "OpenSSL>" prompt).
3404 [Lennart Bang, Bodo Moeller]
3406 *) Preliminary compilation option RSA_NULL which disables RSA crypto without
3407 removing all other RSA functionality (this is what NO_RSA does). This
3408 is so (for example) those in the US can disable those operations covered
3409 by the RSA patent while allowing storage and parsing of RSA keys and RSA
3413 *) Non-copying interface to BIO pairs.
3414 (still largely untested)
3417 *) New function ANS1_tag2str() to convert an ASN1 tag to a descriptive
3418 ASCII string. This was handled independently in various places before.
3421 *) New functions UTF8_getc() and UTF8_putc() that parse and generate
3422 UTF8 strings a character at a time.
3425 *) Use client_version from client hello to select the protocol
3426 (s23_srvr.c) and for RSA client key exchange verification
3427 (s3_srvr.c), as required by the SSL 3.0/TLS 1.0 specifications.
3430 *) Add various utility functions to handle SPKACs, these were previously
3431 handled by poking round in the structure internals. Added new function
3432 NETSCAPE_SPKI_print() to print out SPKAC and a new utility 'spkac' to
3433 print, verify and generate SPKACs. Based on an original idea from
3434 Massimiliano Pala <madwolf@comune.modena.it> but extensively modified.
3437 *) RIPEMD160 is operational on all platforms and is back in 'make test'.
3440 *) Allow the config file extension section to be overwritten on the
3441 command line. Based on an original idea from Massimiliano Pala
3442 <madwolf@comune.modena.it>. The new option is called -extensions
3443 and can be applied to ca, req and x509. Also -reqexts to override
3444 the request extensions in req and -crlexts to override the crl extensions
3448 *) Add new feature to the SPKAC handling in ca. Now you can include
3449 the same field multiple times by preceding it by "XXXX." for example:
3452 this is the same syntax as used in the req config file.
3455 *) Allow certificate extensions to be added to certificate requests. These
3456 are specified in a 'req_extensions' option of the req section of the
3457 config file. They can be printed out with the -text option to req but
3458 are otherwise ignored at present.
3461 *) Fix a horrible bug in enc_read() in crypto/evp/bio_enc.c: if the first
3462 data read consists of only the final block it would not decrypted because
3463 EVP_CipherUpdate() would correctly report zero bytes had been decrypted.
3464 A misplaced 'break' also meant the decrypted final block might not be
3465 copied until the next read.
3468 *) Initial support for DH_METHOD. Again based on RSA_METHOD. Also added
3469 a few extra parameters to the DH structure: these will be useful if
3470 for example we want the value of 'q' or implement X9.42 DH.
3473 *) Initial support for DSA_METHOD. This is based on the RSA_METHOD and
3474 provides hooks that allow the default DSA functions or functions on a
3475 "per key" basis to be replaced. This allows hardware acceleration and
3476 hardware key storage to be handled without major modification to the
3477 library. Also added low level modexp hooks and CRYPTO_EX structure and
3478 associated functions.
3481 *) Add a new flag to memory BIOs, BIO_FLAG_MEM_RDONLY. This marks the BIO
3482 as "read only": it can't be written to and the buffer it points to will
3483 not be freed. Reading from a read only BIO is much more efficient than
3484 a normal memory BIO. This was added because there are several times when
3485 an area of memory needs to be read from a BIO. The previous method was
3486 to create a memory BIO and write the data to it, this results in two
3487 copies of the data and an O(n^2) reading algorithm. There is a new
3488 function BIO_new_mem_buf() which creates a read only memory BIO from
3489 an area of memory. Also modified the PKCS#7 routines to use read only
3493 *) Bugfix: ssl23_get_client_hello did not work properly when called in
3494 state SSL23_ST_SR_CLNT_HELLO_B, i.e. when the first 7 bytes of
3495 a SSLv2-compatible client hello for SSLv3 or TLSv1 could be read,
3496 but a retry condition occured while trying to read the rest.
3499 *) The PKCS7_ENC_CONTENT_new() function was setting the content type as
3500 NID_pkcs7_encrypted by default: this was wrong since this should almost
3501 always be NID_pkcs7_data. Also modified the PKCS7_set_type() to handle
3502 the encrypted data type: this is a more sensible place to put it and it
3503 allows the PKCS#12 code to be tidied up that duplicated this
3507 *) Changed obj_dat.pl script so it takes its input and output files on
3508 the command line. This should avoid shell escape redirection problems
3512 *) Initial support for certificate extension requests, these are included
3513 in things like Xenroll certificate requests. Included functions to allow
3514 extensions to be obtained and added.
3517 *) -crlf option to s_client and s_server for sending newlines as
3518 CRLF (as required by many protocols).
3521 Changes between 0.9.3a and 0.9.4 [09 Aug 1999]
3523 *) Install libRSAglue.a when OpenSSL is built with RSAref.
3524 [Ralf S. Engelschall]
3526 *) A few more ``#ifndef NO_FP_API / #endif'' pairs for consistency.
3527 [Andrija Antonijevic <TheAntony2@bigfoot.com>]
3529 *) Fix -startdate and -enddate (which was missing) arguments to 'ca'
3533 *) New function DSA_dup_DH, which duplicates DSA parameters/keys as
3534 DH parameters/keys (q is lost during that conversion, but the resulting
3535 DH parameters contain its length).
3537 For 1024-bit p, DSA_generate_parameters followed by DSA_dup_DH is
3538 much faster than DH_generate_parameters (which creates parameters
3539 where p = 2*q + 1), and also the smaller q makes DH computations
3540 much more efficient (160-bit exponentiation instead of 1024-bit
3541 exponentiation); so this provides a convenient way to support DHE
3542 ciphersuites in SSL/TLS servers (see ssl/ssltest.c). It is of
3543 utter importance to use
3544 SSL_CTX_set_options(s_ctx, SSL_OP_SINGLE_DH_USE);
3546 SSL_set_options(s_ctx, SSL_OP_SINGLE_DH_USE);
3547 when such DH parameters are used, because otherwise small subgroup
3548 attacks may become possible!
3551 *) Avoid memory leak in i2d_DHparams.
3554 *) Allow the -k option to be used more than once in the enc program:
3555 this allows the same encrypted message to be read by multiple recipients.
3558 *) New function OBJ_obj2txt(buf, buf_len, a, no_name), this converts
3559 an ASN1_OBJECT to a text string. If the "no_name" parameter is set then
3560 it will always use the numerical form of the OID, even if it has a short
3564 *) Added an extra RSA flag: RSA_FLAG_EXT_PKEY. Previously the rsa_mod_exp
3565 method only got called if p,q,dmp1,dmq1,iqmp components were present,
3566 otherwise bn_mod_exp was called. In the case of hardware keys for example
3567 no private key components need be present and it might store extra data
3568 in the RSA structure, which cannot be accessed from bn_mod_exp.
3569 By setting RSA_FLAG_EXT_PKEY rsa_mod_exp will always be called for
3570 private key operations.
3573 *) Added support for SPARC Linux.
3576 *) pem_password_cb function type incompatibly changed from
3577 typedef int pem_password_cb(char *buf, int size, int rwflag);
3579 ....(char *buf, int size, int rwflag, void *userdata);
3580 so that applications can pass data to their callbacks:
3581 The PEM[_ASN1]_{read,write}... functions and macros now take an
3582 additional void * argument, which is just handed through whenever
3583 the password callback is called.
3584 [Damien Miller <dmiller@ilogic.com.au>; tiny changes by Bodo Moeller]
3586 New function SSL_CTX_set_default_passwd_cb_userdata.
3588 Compatibility note: As many C implementations push function arguments
3589 onto the stack in reverse order, the new library version is likely to
3590 interoperate with programs that have been compiled with the old
3591 pem_password_cb definition (PEM_whatever takes some data that
3592 happens to be on the stack as its last argument, and the callback
3593 just ignores this garbage); but there is no guarantee whatsoever that
3596 *) The -DPLATFORM="\"$(PLATFORM)\"" definition and the similar -DCFLAGS=...
3597 (both in crypto/Makefile.ssl for use by crypto/cversion.c) caused
3598 problems not only on Windows, but also on some Unix platforms.
3599 To avoid problematic command lines, these definitions are now in an
3600 auto-generated file crypto/buildinf.h (created by crypto/Makefile.ssl
3601 for standard "make" builds, by util/mk1mf.pl for "mk1mf" builds).
3604 *) MIPS III/IV assembler module is reimplemented.
3607 *) More DES library cleanups: remove references to srand/rand and
3608 delete an unused file.
3611 *) Add support for the the free Netwide assembler (NASM) under Win32,
3612 since not many people have MASM (ml) and it can be hard to obtain.
3613 This is currently experimental but it seems to work OK and pass all
3614 the tests. Check out INSTALL.W32 for info.
3617 *) Fix memory leaks in s3_clnt.c: All non-anonymous SSL3/TLS1 connections
3618 without temporary keys kept an extra copy of the server key,
3619 and connections with temporary keys did not free everything in case
3623 *) New function RSA_check_key and new openssl rsa option -check
3624 for verifying the consistency of RSA keys.
3625 [Ulf Moeller, Bodo Moeller]
3627 *) Various changes to make Win32 compile work:
3628 1. Casts to avoid "loss of data" warnings in p5_crpt2.c
3629 2. Change unsigned int to int in b_dump.c to avoid "signed/unsigned
3630 comparison" warnings.
3631 3. Add sk_<TYPE>_sort to DEF file generator and do make update.
3634 *) Add a debugging option to PKCS#5 v2 key generation function: when
3635 you #define DEBUG_PKCS5V2 passwords, salts, iteration counts and
3636 derived keys are printed to stderr.
3639 *) Copy the flags in ASN1_STRING_dup().
3640 [Roman E. Pavlov <pre@mo.msk.ru>]
3642 *) The x509 application mishandled signing requests containing DSA
3643 keys when the signing key was also DSA and the parameters didn't match.
3645 It was supposed to omit the parameters when they matched the signing key:
3646 the verifying software was then supposed to automatically use the CA's
3647 parameters if they were absent from the end user certificate.
3649 Omitting parameters is no longer recommended. The test was also
3650 the wrong way round! This was probably due to unusual behaviour in
3651 EVP_cmp_parameters() which returns 1 if the parameters match.
projects / openssl.git / blob