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 *) Change the various "***_up()" functions, that increment reference
16 counts on the various structure types, to "***_up_ref()". The naming
17 is more consistent with existing code and generally a better choice.
20 *) Alter all existing ENGINE implementations (except "openssl" and
21 "openbsd") to dynamically instantiate their own error strings. This
22 makes them more flexible to be built both as statically-linked ENGINEs
23 and self-contained shared-libraries loadable via the "dynamic" ENGINE.
24 Also, add stub code to each that makes building them as self-contained
25 shared-libraries easier (see README.ENGINE).
28 *) Add a "dynamic" ENGINE that provides a mechanism for binding ENGINE
29 implementations into applications that are completely implemented in
30 self-contained shared-libraries. The "dynamic" ENGINE exposes control
31 commands that can be used to configure what shared-library to load and
32 to control aspects of the way it is handled. Also, made an update to
33 the README.ENGINE file that brings its information up-to-date and
34 provides some information and instructions on the "dynamic" ENGINE
35 (ie. how to use it, how to build "dynamic"-loadable ENGINEs, etc).
39 [Ulf Möller, reported by Jim Ellis <jim.ellis@cavium.com>]
41 *) Make it possible to unload ranges of ERR strings with a new
42 "ERR_unload_strings" function.
45 *) Rabin-Miller test analyses assume uniformly distributed witnesses,
46 so use BN_pseudo_rand_range() instead of using BN_pseudo_rand()
47 followed by modular reduction.
48 [Bodo Moeller; pointed out by Adam Young <AYoung1@NCSUS.JNJ.COM>]
50 *) Add BN_pseudo_rand_range() with obvious functionality: BN_rand_range()
51 requivalent based on BN_pseudo_rand() instead of BN_rand().
54 +) Add a copy() function to EVP_MD.
57 +) Make EVP_MD routines take a context pointer instead of just the
61 +) Add flags to EVP_MD and EVP_MD_CTX. EVP_MD_FLAG_ONESHOT indicates
62 that the digest can only process a single chunk of data
63 (typically because it is provided by a piece of
64 hardware). EVP_MD_CTX_FLAG_ONESHOT indicates that the application
65 is only going to provide a single chunk of data, and hence the
66 framework needn't accumulate the data for oneshot drivers.
69 +) As with "ERR", make it possible to replace the underlying "ex_data"
70 functions. This change also alters the storage and management of global
71 ex_data state - it's now all inside ex_data.c and all "class" code (eg.
72 RSA, BIO, SSL_CTX, etc) no longer stores its own STACKS and per-class
73 index counters. The API functions that use this state have been changed
74 to take a "class_index" rather than pointers to the class's local STACK
75 and counter, and there is now an API function to dynamically create new
76 classes. This centralisation allows us to (a) plug a lot of the
77 thread-safety problems that existed, and (b) makes it possible to clean
78 up all allocated state using "CRYPTO_cleanup_all_ex_data()". W.r.t. (b)
79 such data would previously have always leaked in application code and
80 workarounds were in place to make the memory debugging turn a blind eye
81 to it. Application code that doesn't use this new function will still
82 leak as before, but their memory debugging output will announce it now
83 rather than letting it slide.
85 Besides the addition of CRYPTO_cleanup_all_ex_data(), another API change
86 induced by the "ex_data" overhaul is that X509_STORE_CTX_init() now
87 has a return value to indicate success or failure.
90 +) Make it possible to replace the underlying "ERR" functions such that the
91 global state (2 LHASH tables and 2 locks) is only used by the "default"
92 implementation. This change also adds two functions to "get" and "set"
93 the implementation prior to it being automatically set the first time
94 any other ERR function takes place. Ie. an application can call "get",
95 pass the return value to a module it has just loaded, and that module
96 can call its own "set" function using that value. This means the
97 module's "ERR" operations will use (and modify) the error state in the
98 application and not in its own statically linked copy of OpenSSL code.
101 +) Give DH, DSA, and RSA types their own "**_up()" function to increment
102 reference counts. This performs normal REF_PRINT/REF_CHECK macros on
103 the operation, and provides a more encapsulated way for external code
104 (crypto/evp/ and ssl/) to do this. Also changed the evp and ssl code
105 to use these functions rather than manually incrementing the counts.
108 *) s3_srvr.c: allow sending of large client certificate lists (> 16 kB).
109 This function was broken, as the check for a new client hello message
110 to handle SGC did not allow these large messages.
111 (Tracked down by "Douglas E. Engert" <deengert@anl.gov>.)
114 *) Add alert descriptions for TLSv1 to SSL_alert_desc_string[_long]().
117 +) Add EVP test program.
120 +) Add symmetric cipher support to ENGINE. Expect the API to change!
123 +) New CRL functions: X509_CRL_set_version(), X509_CRL_set_issuer_name()
124 X509_CRL_set_lastUpdate(), X509_CRL_set_nextUpdate(), X509_CRL_sort(),
125 X509_REVOKED_set_serialNumber(), and X509_REVOKED_set_revocationDate().
126 These allow a CRL to be built without having to access X509_CRL fields
127 directly. Modify 'ca' application to use new functions.
130 *) Fix buggy behaviour of BIO_get_num_renegotiates() and BIO_ctrl()
131 for BIO_C_GET_WRITE_BUF_SIZE ("Stephen Hinton" <shinton@netopia.com>).
134 *) Rework the configuration and shared library support for Tru64 Unix.
135 The configuration part makes use of modern compiler features and
136 still retains old compiler behavior for those that run older versions
137 of the OS. The shared library support part includes a variant that
138 uses the RPATH feature, and is available through the speciel
139 configuration target "alpha-cc-rpath", which will never be selected
141 [Tim Mooney <mooney@dogbert.cc.ndsu.NoDak.edu> via Richard Levitte]
143 *) In ssl3_get_key_exchange (ssl/s3_clnt.c), call ssl3_get_message()
144 with the same message size as in ssl3_get_certificate_request().
145 Otherwise, if no ServerKeyExchange message occurs, CertificateRequest
146 messages might inadvertently be reject as too long.
147 [Petr Lampa <lampa@fee.vutbr.cz>]
149 +) Move SSL_OP_TLS_ROLLBACK_BUG out of the SSL_OP_ALL list of recommended
150 bug workarounds. Rollback attack detection is a security feature.
151 The problem will only arise on OpenSSL servers when TLSv1 is not
152 available (sslv3_server_method() or SSL_OP_NO_TLSv1).
153 Software authors not wanting to support TLSv1 will have special reasons
154 for their choice and can explicitly enable this option.
155 [Bodo Moeller, Lutz Jaenicke]
157 +) Rationalise EVP so it can be extended: don't include a union of
158 cipher/digest structures, add init/cleanup functions. This also reduces
159 the number of header dependencies.
162 +) Make DES key schedule conform to the usual scheme, as well as
163 correcting its structure. This means that calls to DES functions
164 now have to pass a pointer to a des_key_schedule instead of a
165 plain des_key_schedule (which was actually always a pointer
169 +) Enhanced support for IA-64 Unix platforms (well, Linux and HP-UX).
172 *) Modified SSL library such that the verify_callback that has been set
173 specificly for an SSL object with SSL_set_verify() is actually being
174 used. Before the change, a verify_callback set with this function was
175 ignored and the verify_callback() set in the SSL_CTX at the time of
176 the call was used. New function X509_STORE_CTX_set_verify_cb() introduced
177 to allow the necessary settings.
180 +) Initial reduction of linker bloat: the use of some functions, such as
181 PEM causes large amounts of unused functions to be linked in due to
182 poor organisation. For example pem_all.c contains every PEM function
183 which has a knock on effect of linking in large amounts of (unused)
184 ASN1 code. Grouping together similar functions and splitting unrelated
185 functions prevents this.
188 *) Initialize static variable in crypto/dsa/dsa_lib.c and crypto/dh/dh_lib.c
189 explicitely to NULL, as at least on Solaris 8 this seems not always to be
190 done automatically (in contradiction to the requirements of the C
191 standard). This made problems when used from OpenSSH.
194 *) In OpenSSL 0.9.6a and 0.9.6b, crypto/dh/dh_key.c ignored
195 dh->length and always used
197 BN_rand_range(priv_key, dh->p).
199 BN_rand_range() is not necessary for Diffie-Hellman, and this
200 specific range makes Diffie-Hellman unnecessarily inefficient if
201 dh->length (recommended exponent length) is much smaller than the
202 length of dh->p. We could use BN_rand_range() if the order of
203 the subgroup was stored in the DH structure, but we only have
208 BN_rand(priv_key, l, ...)
210 where 'l' is dh->length if this is defined, or BN_num_bits(dh->p)-1
216 RSA_eay_public_encrypt
217 RSA_eay_private_decrypt
218 RSA_eay_private_encrypt (signing)
219 RSA_eay_public_decrypt (signature verification)
221 (default implementations for RSA_public_encrypt,
222 RSA_private_decrypt, RSA_private_encrypt, RSA_public_decrypt),
223 always reject numbers >= n.
226 *) In crypto/rand/md_rand.c, use a new short-time lock CRYPTO_LOCK_RAND2
227 to synchronize access to 'locking_thread'. This is necessary on
228 systems where access to 'locking_thread' (an 'unsigned long'
229 variable) is not atomic.
232 *) In crypto/rand/md_rand.c, set 'locking_thread' to current thread's ID
233 *before* setting the 'crypto_lock_rand' flag. The previous code had
234 a race condition if 0 is a valid thread ID.
235 [Travis Vitek <vitek@roguewave.com>]
237 +) Cleanup of EVP macros.
240 +) Change historical references to {NID,SN,LN}_des_ede and ede3 to add the
244 +) Add initial OCSP responder support to ocsp application. The
245 revocation information is handled using the text based index
246 use by the ca application. The responder can either handle
247 requests generated internally, supplied in files (for example
248 via a CGI script) or using an internal minimal server.
251 +) Add configuration choices to get zlib compression for TLS.
254 +) Changes to Kerberos SSL for RFC 2712 compliance:
255 1. Implemented real KerberosWrapper, instead of just using
256 KRB5 AP_REQ message. [Thanks to Simon Wilkinson <sxw@sxw.org.uk>]
257 2. Implemented optional authenticator field of KerberosWrapper.
259 Added openssl-style ASN.1 macros for Kerberos ticket, ap_req,
260 and authenticator structs; see crypto/krb5/.
262 Generalized Kerberos calls to support multiple Kerberos libraries.
263 [Vern Staats <staatsvr@asc.hpc.mil>,
264 Jeffrey Altman <jaltman@columbia.edu>
267 +) Cause 'openssl speed' to use fully hard-coded DSA keys as it
268 already does with RSA. testdsa.h now has 'priv_key/pub_key'
269 values for each of the key sizes rather than having just
270 parameters (and 'speed' generating keys each time).
273 -) OpenSSL 0.9.6b released [9 July 2001]
275 *) Change ssleay_rand_bytes (crypto/rand/md_rand.c)
276 to avoid a SSLeay/OpenSSL PRNG weakness pointed out by
277 Markku-Juhani O. Saarinen <markku-juhani.saarinen@nokia.com>:
278 PRNG state recovery was possible based on the output of
279 one PRNG request appropriately sized to gain knowledge on
280 'md' followed by enough consecutive 1-byte PRNG requests
281 to traverse all of 'state'.
283 1. When updating 'md_local' (the current thread's copy of 'md')
284 during PRNG output generation, hash all of the previous
285 'md_local' value, not just the half used for PRNG output.
287 2. Make the number of bytes from 'state' included into the hash
288 independent from the number of PRNG bytes requested.
290 The first measure alone would be sufficient to avoid
291 Markku-Juhani's attack. (Actually it had never occurred
292 to me that the half of 'md_local' used for chaining was the
293 half from which PRNG output bytes were taken -- I had always
294 assumed that the secret half would be used.) The second
295 measure makes sure that additional data from 'state' is never
296 mixed into 'md_local' in small portions; this heuristically
297 further strengthens the PRNG.
300 +) Speed up EVP routines.
303 type 8 bytes 64 bytes 256 bytes 1024 bytes 8192 bytes
304 des-cbc 4408.85k 5560.51k 5778.46k 5862.20k 5825.16k
305 des-cbc 4389.55k 5571.17k 5792.23k 5846.91k 5832.11k
306 des-cbc 4394.32k 5575.92k 5807.44k 5848.37k 5841.30k
308 des-cbc 3482.66k 5069.49k 5496.39k 5614.16k 5639.28k
309 des-cbc 3480.74k 5068.76k 5510.34k 5609.87k 5635.52k
310 des-cbc 3483.72k 5067.62k 5504.60k 5708.01k 5724.80k
313 des-cbc 4660.16k 5650.19k 5807.19k 5827.13k 5783.32k
315 des-cbc 3624.96k 5258.21k 5530.91k 5624.30k 5628.26k
318 *) Fix crypto/bn/asm/mips3.s.
321 *) When only the key is given to "enc", the IV is undefined. Print out
322 an error message in this case.
325 +) Added the OS2-EMX target.
326 ["Brian Havard" <brianh@kheldar.apana.org.au> and Richard Levitte]
328 +) Rewrite apps to use NCONF routines instead of the old CONF. New functions
329 to support NCONF routines in extension code. New function CONF_set_nconf()
330 to allow functions which take an NCONF to also handle the old LHASH
331 structure: this means that the old CONF compatible routines can be
332 retained (in particular wrt extensions) without having to duplicate the
333 code. New function X509V3_add_ext_nconf_sk to add extensions to a stack.
336 *) Handle special case when X509_NAME is empty in X509 printing routines.
339 *) In dsa_do_verify (crypto/dsa/dsa_ossl.c), verify that r and s are
340 positive and less than q.
343 +) Enhance the general user interface with mechanisms for inner control
344 and with pssibilities to have yes/no kind of prompts.
347 +) Change all calls to low level digest routines in the library and
348 applications to use EVP. Add missing calls to HMAC_cleanup() and
349 don't assume HMAC_CTX can be copied using memcpy().
350 [Verdon Walker <VWalker@novell.com>, Steve Henson]
352 +) Add the possibility to control engines through control names but with
353 arbitrary arguments instead of just a string.
354 Change the key loaders to take a UI_METHOD instead of a callback
355 function pointer. NOTE: this breaks binary compatibility with earlier
356 versions of OpenSSL [engine].
357 Addapt the nCipher code for these new conditions and add a card insertion
361 +) Enhance the general user interface with mechanisms to better support
362 dialog box interfaces, application-defined prompts, the possibility
363 to use defaults (for example default passwords from somewhere else)
364 and interrupts/cancelations.
367 *) Don't change *pointer in CRYPTO_add_lock() is add_lock_callback is
368 used: it isn't thread safe and the add_lock_callback should handle
370 [Paul Rose <Paul.Rose@bridge.com>]
372 *) Verify that incoming data obeys the block size in
373 ssl3_enc (ssl/s3_enc.c) and tls1_enc (ssl/t1_enc.c).
376 +) Tidy up PKCS#12 attribute handling. Add support for the CSP name
377 attribute in PKCS#12 files, add new -CSP option to pkcs12 utility.
381 [Ulf Möller, Bodo Möller]
383 *) The countermeasure against Bleichbacher's attack on PKCS #1 v1.5
384 RSA encryption was accidentily removed in s3_srvr.c in OpenSSL 0.9.5
385 when fixing the server behaviour for backwards-compatible 'client
386 hello' messages. (Note that the attack is impractical against
387 SSL 3.0 and TLS 1.0 anyway because length and version checking
388 means that the probability of guessing a valid ciphertext is
389 around 2^-40; see section 5 in Bleichenbacher's CRYPTO '98
392 Before 0.9.5, the countermeasure (hide the error by generating a
393 random 'decryption result') did not work properly because
394 ERR_clear_error() was missing, meaning that SSL_get_error() would
395 detect the supposedly ignored error.
397 Both problems are now fixed.
400 *) In crypto/bio/bf_buff.c, increase DEFAULT_BUFFER_SIZE to 4096
401 (previously it was 1024).
404 +) Fix a memory leak in 'sk_dup()' in the case reallocation fails. (Also
405 tidy up some unecessarily weird code in 'sk_new()').
406 [Geoff, reported by Diego Tartara <dtartara@novamens.com>]
408 +) Change the key loading routines for ENGINEs to use the same kind
409 callback (pem_password_cb) as all other routines that need this
413 *) Fix for compatibility mode trust settings: ignore trust settings
414 unless some valid trust or reject settings are present.
417 *) Fix for blowfish EVP: its a variable length cipher.
420 +) Increase ENTROPY_NEEDED to 32 bytes, as Rijndael can operate with
421 256 bit (=32 byte) keys. Of course seeding with more entropy bytes
422 than this minimum value is recommended.
425 +) New random seeder for OpenVMS, using the system process statistics
426 that are easily reachable.
429 +) Windows apparently can't transparently handle global
430 variables defined in DLLs. Initialisations such as:
432 const ASN1_ITEM *it = &ASN1_INTEGER_it;
434 wont compile. This is used by the any applications that need to
435 delcare their own ASN1 modules. This was fixed by adding the option
436 EXPORT_VAR_AS_FN to all Win32 platforms, although this isn't strictly
437 needed for static libraries under Win32.
440 +) New functions X509_PURPOSE_set() and X509_TRUST_set() to handle
441 setting of purpose and trust fields. New X509_STORE trust and
442 purpose functions and tidy up setting in other SSL functions.
445 +) Add copies of X509_STORE_CTX fields and callbacks to X509_STORE
446 structure. These are inherited by X509_STORE_CTX when it is
447 initialised. This allows various defaults to be set in the
448 X509_STORE structure (such as flags for CRL checking and custom
449 purpose or trust settings) for functions which only use X509_STORE_CTX
450 internally such as S/MIME.
452 Modify X509_STORE_CTX_purpose_inherit() so it only sets purposes and
453 trust settings if they are not set in X509_STORE. This allows X509_STORE
454 purposes and trust (in S/MIME for example) to override any set by default.
456 Add command line options for CRL checking to smime, s_client and s_server
460 +) Initial CRL based revocation checking. If the CRL checking flag(s)
461 are set then the CRL is looked up in the X509_STORE structure and
462 its validity and signature checked, then if the certificate is found
463 in the CRL the verify fails with a revoked error.
465 Various new CRL related callbacks added to X509_STORE_CTX structure.
467 Command line options added to 'verify' application to support this.
469 This needs some additional work, such as being able to handle multiple
470 CRLs with different times, extension based lookup (rather than just
471 by subject name) and ultimately more complete V2 CRL extension
475 +) Add a general user interface API (crypto/ui/). This is designed
476 to replace things like des_read_password and friends (backward
477 compatibility functions using this new API are provided).
478 The purpose is to remove prompting functions from the DES code
479 section as well as provide for prompting through dialog boxes in
480 a window system and the like.
483 *) Fix various bugs related to DSA S/MIME verification. Handle missing
484 parameters in DSA public key structures and return an error in the
485 DSA routines if parameters are absent.
488 *) In versions up to 0.9.6, RAND_file_name() resorted to file ".rnd"
489 in the current directory if neither $RANDFILE nor $HOME was set.
490 RAND_file_name() in 0.9.6a returned NULL in this case. This has
491 caused some confusion to Windows users who haven't defined $HOME.
492 Thus RAND_file_name() is changed again: e_os.h can define a
493 DEFAULT_HOME, which will be used if $HOME is not set.
494 For Windows, we use "C:"; on other platforms, we still require
495 environment variables.
497 +) Add "ex_data" support to ENGINE so implementations can add state at a
498 per-structure level rather than having to store it globally.
501 +) Make it possible for ENGINE structures to be copied when retrieved by
502 ENGINE_by_id() if the ENGINE specifies a new flag: ENGINE_FLAGS_BY_ID_COPY.
503 This causes the "original" ENGINE structure to act like a template,
504 analogous to the RSA vs. RSA_METHOD type of separation. Because of this
505 operational state can be localised to each ENGINE structure, despite the
506 fact they all share the same "methods". New ENGINE structures returned in
507 this case have no functional references and the return value is the single
508 structural reference. This matches the single structural reference returned
509 by ENGINE_by_id() normally, when it is incremented on the pre-existing
513 +) Fix ASN1 decoder when decoding type ANY and V_ASN1_OTHER: since this
514 needs to match any other type at all we need to manually clear the
518 +) Changes to the "openssl engine" utility to include;
519 - verbosity levels ('-v', '-vv', and '-vvv') that provide information
520 about an ENGINE's available control commands.
521 - executing control commands from command line arguments using the
522 '-pre' and '-post' switches. '-post' is only used if '-t' is
523 specified and the ENGINE is successfully initialised. The syntax for
524 the individual commands are colon-separated, for example;
525 openssl engine chil -pre FORK_CHECK:0 -pre SO_PATH:/lib/test.so
528 +) New dynamic control command support for ENGINEs. ENGINEs can now
529 declare their own commands (numbers), names (strings), descriptions,
530 and input types for run-time discovery by calling applications. A
531 subset of these commands are implicitly classed as "executable"
532 depending on their input type, and only these can be invoked through
533 the new string-based API function ENGINE_ctrl_cmd_string(). (Eg. this
534 can be based on user input, config files, etc). The distinction is
535 that "executable" commands cannot return anything other than a boolean
536 result and can only support numeric or string input, whereas some
537 discoverable commands may only be for direct use through
538 ENGINE_ctrl(), eg. supporting the exchange of binary data, function
539 pointers, or other custom uses. The "executable" commands are to
540 support parameterisations of ENGINE behaviour that can be
541 unambiguously defined by ENGINEs and used consistently across any
542 OpenSSL-based application. Commands have been added to all the
543 existing hardware-supporting ENGINEs, noticeably "SO_PATH" to allow
544 control over shared-library paths without source code alterations.
547 +) Changed all ENGINE implementations to dynamically allocate their
548 ENGINEs rather than declaring them statically. Apart from this being
549 necessary with the removal of the ENGINE_FLAGS_MALLOCED distinction,
550 this also allows the implementations to compile without using the
551 internal engine_int.h header.
554 +) Minor adjustment to "rand" code. RAND_get_rand_method() now returns a
555 'const' value. Any code that should be able to modify a RAND_METHOD
556 should already have non-const pointers to it (ie. they should only
557 modify their own ones).
560 +) Made a variety of little tweaks to the ENGINE code.
561 - "atalla" and "ubsec" string definitions were moved from header files
562 to C code. "nuron" string definitions were placed in variables
563 rather than hard-coded - allowing parameterisation of these values
564 later on via ctrl() commands.
565 - Removed unused "#if 0"'d code.
566 - Fixed engine list iteration code so it uses ENGINE_free() to release
567 structural references.
568 - Constified the RAND_METHOD element of ENGINE structures.
569 - Constified various get/set functions as appropriate and added
570 missing functions (including a catch-all ENGINE_cpy that duplicates
571 all ENGINE values onto a new ENGINE except reference counts/state).
572 - Removed NULL parameter checks in get/set functions. Setting a method
573 or function to NULL is a way of cancelling out a previously set
574 value. Passing a NULL ENGINE parameter is just plain stupid anyway
575 and doesn't justify the extra error symbols and code.
576 - Deprecate the ENGINE_FLAGS_MALLOCED define and move the area for
577 flags from engine_int.h to engine.h.
578 - Changed prototypes for ENGINE handler functions (init(), finish(),
579 ctrl(), key-load functions, etc) to take an (ENGINE*) parameter.
582 *) Move 'if (!initialized) RAND_poll()' into regions protected by
583 CRYPTO_LOCK_RAND. This is not strictly necessary, but avoids
584 having multiple threads call RAND_poll() concurrently.
587 *) In crypto/rand/md_rand.c, replace 'add_do_not_lock' flag by a
588 combination of a flag and a thread ID variable.
589 Otherwise while one thread is in ssleay_rand_bytes (which sets the
590 flag), *other* threads can enter ssleay_add_bytes without obeying
591 the CRYPTO_LOCK_RAND lock (and may even illegaly release the lock
592 that they do not hold after the first thread unsets add_do_not_lock).
595 +) Implement binary inversion algorithm for BN_mod_inverse in addition
596 to the algorithm using long divison. The binary algorithm can be
597 used only if the modulus is odd. On 32-bit systems, it is faster
598 only for relatively small moduli (roughly 20-30% for 128-bit moduli,
599 roughly 5-15% for 256-bit moduli), so we use it only for moduli
600 up to 450 bits. In 64-bit environments, the binary algorithm
601 appears to be advantageous for much longer moduli; here we use it
602 for moduli up to 2048 bits.
605 *) Change bctest again: '-x' expressions are not available in all
609 -) OpenSSL 0.9.6a released [5 Apr 2001]
611 *) Fix a couple of memory leaks in PKCS7_dataDecode()
612 [Steve Henson, reported by Heyun Zheng <hzheng@atdsprint.com>]
614 *) Change Configure and Makefiles to provide EXE_EXT, which will contain
615 the default extension for executables, if any. Also, make the perl
616 scripts that use symlink() to test if it really exists and use "cp"
617 if it doesn't. All this made OpenSSL compilable and installable in
621 +) Rewrite CHOICE field setting in ASN1_item_ex_d2i(). The old code
622 could not support the combine flag in choice fields.
625 -) Fix for asn1_GetSequence() for indefinite length constructed data.
626 If SEQUENCE is length is indefinite just set c->slen to the total
627 amount of data available.
628 [Steve Henson, reported by shige@FreeBSD.org]
629 [This change does not apply to 0.9.7.]
631 *) Change bctest to avoid here-documents inside command substitution
632 (workaround for FreeBSD /bin/sh bug).
633 For compatibility with Ultrix, avoid shell functions (introduced
634 in the bctest version that searches along $PATH).
637 *) Rename 'des_encrypt' to 'des_encrypt1'. This avoids the clashes
638 with des_encrypt() defined on some operating systems, like Solaris
642 *) Check the result of RSA-CRT (see D. Boneh, R. DeMillo, R. Lipton:
643 On the Importance of Eliminating Errors in Cryptographic
644 Computations, J. Cryptology 14 (2001) 2, 101-119,
645 http://theory.stanford.edu/~dabo/papers/faults.ps.gz).
648 *) MIPS assembler BIGNUM division bug fix.
651 *) Disabled incorrect Alpha assembler code.
654 -) Fix PKCS#7 decode routines so they correctly update the length
655 after reading an EOC for the EXPLICIT tag.
657 [This change does not apply to 0.9.7.]
659 *) Fix bug in PKCS#12 key generation routines. This was triggered
660 if a 3DES key was generated with a 0 initial byte. Include
661 PKCS12_BROKEN_KEYGEN compilation option to retain the old
662 (but broken) behaviour.
665 *) Enhance bctest to search for a working bc along $PATH and print
667 [Tim Rice <tim@multitalents.net> via Richard Levitte]
669 +) Add a 'copy_extensions' option to the 'ca' utility. This copies
670 extensions from a certificate request to the certificate.
673 +) Allow multiple 'certopt' and 'nameopt' options to be separated
674 by commas. Add 'namopt' and 'certopt' options to the 'ca' config
675 file: this allows the display of the certificate about to be
676 signed to be customised, to allow certain fields to be included
677 or excluded and extension details. The old system didn't display
678 multicharacter strings properly, omitted fields not in the policy
679 and couldn't display additional details such as extensions.
682 *) Fix memory leaks in err.c: free err_data string if necessary;
683 don't write to the wrong index in ERR_set_error_data.
686 +) Function EC_POINTs_mul for simultaneous scalar multiplication
687 of an arbitrary number of elliptic curve points, optionally
688 including the generator defined for the EC_GROUP.
689 EC_POINT_mul is a simple wrapper function for the typical case
690 that the point list has just one item (besides the optional
694 +) First EC_METHODs for curves over GF(p):
696 EC_GFp_simple_method() uses the basic BN_mod_mul and BN_mod_sqr
697 operations and provides various method functions that can also
698 operate with faster implementations of modular arithmetic.
700 EC_GFp_mont_method() reuses most functions that are part of
701 EC_GFp_simple_method, but uses Montgomery arithmetic.
703 [Bodo Moeller; point addition and point doubling
704 implementation directly derived from source code provided by
705 Lenka Fibikova <fibikova@exp-math.uni-essen.de>]
707 +) Framework for elliptic curves (crypto/ec/ec.h, crypto/ec/ec_lcl.h,
710 Curves are EC_GROUP objects (with an optional group generator)
711 based on EC_METHODs that are built into the library.
713 Points are EC_POINT objects based on EC_GROUP objects.
715 Most of the framework would be able to handle curves over arbitrary
716 finite fields, but as there are no obvious types for fields other
717 than GF(p), some functions are limited to that for now.
720 +) Add the -HTTP option to s_server. It is similar to -WWW, but requires
721 that the file contains a complete HTTP response.
724 +) Add the ec directory to mkdef.pl and mkfiles.pl. In mkdef.pl
725 change the def and num file printf format specifier from "%-40sXXX"
726 to "%-39s XXX". The latter will always guarantee a space after the
727 field while the former will cause them to run together if the field
728 is 40 of more characters long.
731 +) Constify the cipher and digest 'method' functions and structures
732 and modify related functions to take constant EVP_MD and EVP_CIPHER
736 *) Implement ssl23_peek (analogous to ssl23_read), which previously
740 *) Replace rdtsc with _emit statements for VC++ version 5.
741 [Jeremy Cooper <jeremy@baymoo.org>]
743 +) Hide BN_CTX structure details in bn_lcl.h instead of publishing them
744 in <openssl/bn.h>. Also further increase BN_CTX_NUM to 32.
747 +) Modify EVP_Digest*() routines so they now return values. Although the
748 internal software routines can never fail additional hardware versions
752 +) Clean up crypto/err/err.h and change some error codes to avoid conflicts:
754 Previously ERR_R_FATAL was too small and coincided with ERR_LIB_PKCS7
755 (= ERR_R_PKCS7_LIB); it is now 64 instead of 32.
758 ERR_R_NESTED_ASN1_ERROR
760 ERR_R_MISSING_ASN1_EOS
761 were 4 .. 9, conflicting with
762 ERR_LIB_RSA (= ERR_R_RSA_LIB)
764 ERR_LIB_PEM (= ERR_R_PEM_LIB).
765 They are now 58 .. 63 (i.e., just below ERR_R_FATAL).
767 Add new error code 'ERR_R_INTERNAL_ERROR'.
770 +) Don't overuse locks in crypto/err/err.c: For data retrieval, CRYPTO_r_lock
774 +) New option '-subj arg' for 'openssl req' and 'openssl ca'. This
775 sets the subject name for a new request or supersedes the
776 subject name in a given request. Formats that can be parsed are
777 'CN=Some Name, OU=myOU, C=IT'
779 'CN=Some Name/OU=myOU/C=IT'.
781 Add options '-batch' and '-verbose' to 'openssl req'.
782 [Massimiliano Pala <madwolf@hackmasters.net>]
784 *) Make it possible to reuse SSLv2 sessions.
787 +) Introduce the possibility to access global variables through
788 functions on platform were that's the best way to handle exporting
789 global variables in shared libraries. To enable this functionality,
790 one must configure with "EXPORT_VAR_AS_FN" or defined the C macro
791 "OPENSSL_EXPORT_VAR_AS_FUNCTION" in crypto/opensslconf.h (the latter
792 is normally done by Configure or something similar).
794 To implement a global variable, use the macro OPENSSL_IMPLEMENT_GLOBAL
795 in the source file (foo.c) like this:
797 OPENSSL_IMPLEMENT_GLOBAL(int,foo)=1;
798 OPENSSL_IMPLEMENT_GLOBAL(double,bar);
800 To declare a global variable, use the macros OPENSSL_DECLARE_GLOBAL
801 and OPENSSL_GLOBAL_REF in the header file (foo.h) like this:
803 OPENSSL_DECLARE_GLOBAL(int,foo);
804 #define foo OPENSSL_GLOBAL_REF(foo)
805 OPENSSL_DECLARE_GLOBAL(double,bar);
806 #define bar OPENSSL_GLOBAL_REF(bar)
808 The #defines are very important, and therefore so is including the
809 header file everywere where the defined globals are used.
811 The macro OPENSSL_EXPORT_VAR_AS_FUNCTION also affects the definition
812 of ASN.1 items, but that structure is a bt different.
814 The largest change is in util/mkdef.pl which has been enhanced with
815 better and easier to understand logic to choose which symbols should
816 go into the Windows .def files as well as a number of fixes and code
817 cleanup (among others, algorithm keywords are now sorted
818 lexicographically to avoid constant rewrites).
821 *) In copy_email() check for >= 0 as a return value for
822 X509_NAME_get_index_by_NID() since 0 is a valid index.
823 [Steve Henson reported by Massimiliano Pala <madwolf@opensca.org>]
825 +) In BN_div() keep a copy of the sign of 'num' before writing the
826 result to 'rm' because if rm==num the value will be overwritten
827 and produce the wrong result if 'num' is negative: this caused
828 problems with BN_mod() and BN_nnmod().
831 +) Function OCSP_request_verify(). This checks the signature on an
832 OCSP request and verifies the signer certificate. The signer
833 certificate is just checked for a generic purpose and OCSP request
837 +) Add OCSP_check_validity() function to check the validity of OCSP
838 responses. OCSP responses are prepared in real time and may only
839 be a few seconds old. Simply checking that the current time lies
840 between thisUpdate and nextUpdate max reject otherwise valid responses
841 caused by either OCSP responder or client clock innacuracy. Instead
842 we allow thisUpdate and nextUpdate to fall within a certain period of
843 the current time. The age of the response can also optionally be
844 checked. Two new options -validity_period and -status_age added to
848 +) If signature or public key algorithm is unrecognized print out its
849 OID rather that just UNKOWN.
852 *) Avoid coredump with unsupported or invalid public keys by checking if
853 X509_get_pubkey() fails in PKCS7_verify(). Fix memory leak when
854 PKCS7_verify() fails with non detached data.
857 +) Change OCSP_cert_to_id() to tolerate a NULL subject certificate and
858 OCSP_cert_id_new() a NULL serialNumber. This allows a partial certificate
859 ID to be generated from the issuer certificate alone which can then be
860 passed to OCSP_id_issuer_cmp().
863 +) New compilation option ASN1_ITEM_FUNCTIONS. This causes the new
864 ASN1 modules to export functions returning ASN1_ITEM pointers
865 instead of the ASN1_ITEM structures themselves. This adds several
866 new macros which allow the underlying ASN1 function/structure to
867 be accessed transparently. As a result code should not use ASN1_ITEM
868 references directly (such as &X509_it) but instead use the relevant
869 macros (such as ASN1_ITEM_rptr(X509)). This option is to allow
870 use of the new ASN1 code on platforms where exporting structures
871 is problematical (for example in shared libraries) but exporting
872 functions returning pointers to structures is not.
875 +) Add support for overriding the generation of SSL/TLS session IDs.
876 These callbacks can be registered either in an SSL_CTX or per SSL.
877 The purpose of this is to allow applications to control, if they wish,
878 the arbitrary values chosen for use as session IDs, particularly as it
879 can be useful for session caching in multiple-server environments. A
880 command-line switch for testing this (and any client code that wishes
881 to use such a feature) has been added to "s_server".
882 [Geoff Thorpe, Lutz Jaenicke]
884 +) Modify mkdef.pl to recognise and parse prprocessor conditionals
885 of the form '#if defined(...) || defined(...) || ...' and
886 '#if !defined(...) && !defined(...) && ...'. This also avoids
887 the growing number of special cases it was previously handling.
890 *) Don't use getenv in library functions when run as setuid/setgid.
891 New function OPENSSL_issetugid().
894 +) Make all configuration macros available for application by making
895 sure they are available in opensslconf.h, by giving them names starting
896 with "OPENSSL_" to avoid conflicts with other packages and by making
897 sure e_os2.h will cover all platform-specific cases together with
899 Additionally, it is now possible to define configuration/platform-
900 specific names (called "system identities"). In the C code, these
901 are prefixed with "OPENSSL_SYSNAME_". e_os2.h will create another
902 macro with the name beginning with "OPENSSL_SYS_", which is determined
903 from "OPENSSL_SYSNAME_*" or compiler-specific macros depending on
907 +) New option -set_serial to 'req' and 'x509' this allows the serial
908 number to use to be specified on the command line. Previously self
909 signed certificates were hard coded with serial number 0 and the
910 CA options of 'x509' had to use a serial number in a file which was
914 *) Avoid false positives in memory leak detection code (crypto/mem_dbg.c)
915 due to incorrect handling of multi-threading:
917 1. Fix timing glitch in the MemCheck_off() portion of CRYPTO_mem_ctrl().
919 2. Fix logical glitch in is_MemCheck_on() aka CRYPTO_is_mem_check_on().
921 3. Count how many times MemCheck_off() has been called so that
922 nested use can be treated correctly. This also avoids
923 inband-signalling in the previous code (which relied on the
924 assumption that thread ID 0 is impossible).
927 +) New options to 'ca' utility to support V2 CRL entry extensions.
928 Currently CRL reason, invalidity date and hold instruction are
929 supported. Add new CRL extensions to V3 code and some new objects.
932 *) Add "-rand" option also to s_client and s_server.
935 +) New function EVP_CIPHER_CTX_set_padding() this is used to
936 disable standard block padding (aka PKCS#5 padding) in the EVP
937 API, which was previously mandatory. This means that the data is
938 not padded in any way and so the total length much be a multiple
939 of the block size, otherwise an error occurs.
942 +) Initial (incomplete) OCSP SSL support.
945 *) Fix CPU detection on Irix 6.x.
946 [Kurt Hockenbury <khockenb@stevens-tech.edu> and
947 "Bruce W. Forsberg" <bruce.forsberg@baesystems.com>]
949 +) New function OCSP_parse_url(). This splits up a URL into its host,
950 port and path components: primarily to parse OCSP URLs. New -url
951 option to ocsp utility.
954 +) New nonce behavior. The return value of OCSP_check_nonce() now
955 reflects the various checks performed. Applications can decide
956 whether to tolerate certain situations such as an absent nonce
957 in a response when one was present in a request: the ocsp application
958 just prints out a warning. New function OCSP_add1_basic_nonce()
959 this is to allow responders to include a nonce in a response even if
960 the request is nonce-less.
963 *) Fix X509_NAME bug which produced incorrect encoding if X509_NAME
966 [This change does not apply to 0.9.7.]
968 *) Use the cached encoding of an X509_NAME structure rather than
969 copying it. This is apparently the reason for the libsafe "errors"
970 but the code is actually correct.
973 +) Disable stdin buffering in load_cert (apps/apps.c) so that no certs are
974 skipped when using openssl x509 multiple times on a single input file,
975 e.g. "(openssl x509 -out cert1; openssl x509 -out cert2) <certs".
978 +) Make ASN1_UTCTIME_set_string() and ASN1_GENERALIZEDTIME_set_string()
979 set string type: to handle setting ASN1_TIME structures. Fix ca
980 utility to correctly initialize revocation date of CRLs.
983 +) New option SSL_OP_CIPHER_SERVER_PREFERENCE allows the server to override
984 the clients preferred ciphersuites and rather use its own preferences.
985 Should help to work around M$ SGC (Server Gated Cryptography) bug in
986 Internet Explorer by ensuring unchanged hash method during stepup.
987 (Also replaces the broken/deactivated SSL_OP_NON_EXPORT_FIRST option.)
990 +) Make mkdef.pl recognise all DECLARE_ASN1 macros, change rijndael
991 to aes and add a new 'exist' option to print out symbols that don't
995 +) Additional options to ocsp utility to allow flags to be set and
996 additional certificates supplied.
999 +) Add the option -VAfile to 'openssl ocsp', so the user can give the
1000 OCSP client a number of certificate to only verify the response
1004 *) Add new function BN_rand_range(), and fix DSA_sign_setup() to prevent
1005 Bleichenbacher's DSA attack.
1006 Extend BN_[pseudo_]rand: As before, top=1 forces the highest two bits
1007 to be set and top=0 forces the highest bit to be set; top=-1 is new
1008 and leaves the highest bit random.
1009 [Ulf Moeller, Bodo Moeller]
1011 +) Update Rijndael code to version 3.0 and change EVP AES ciphers to
1012 handle the new API. Currently only ECB, CBC modes supported. Add new
1013 AES OIDs. Add TLS AES ciphersuites as described in the "AES Ciphersuites
1014 for TLS" draft-ietf-tls-ciphersuite-03.txt.
1015 [Ben Laurie, Steve Henson]
1017 *) In the NCONF_...-based implementations for CONF_... queries
1018 (crypto/conf/conf_lib.c), if the input LHASH is NULL, avoid using
1019 a temporary CONF structure with the data component set to NULL
1020 (which gives segmentation faults in lh_retrieve).
1021 Instead, use NULL for the CONF pointer in CONF_get_string and
1022 CONF_get_number (which may use environment variables) and directly
1023 return NULL from CONF_get_section.
1026 *) Fix potential buffer overrun for EBCDIC.
1029 +) New function OCSP_copy_nonce() to copy nonce value (if present) from
1030 request to response.
1033 +) Functions for OCSP responders. OCSP_request_onereq_count(),
1034 OCSP_request_onereq_get0(), OCSP_onereq_get0_id() and OCSP_id_get0_info()
1035 extract information from a certificate request. OCSP_response_create()
1036 creates a response and optionally adds a basic response structure.
1037 OCSP_basic_add1_status() adds a complete single response to a basic
1038 reponse and returns the OCSP_SINGLERESP structure just added (to allow
1039 extensions to be included for example). OCSP_basic_add1_cert() adds a
1040 certificate to a basic response and OCSP_basic_sign() signs a basic
1041 response with various flags. New helper functions ASN1_TIME_check()
1042 (checks validity of ASN1_TIME structure) and ASN1_TIME_to_generalizedtime()
1043 (converts ASN1_TIME to GeneralizedTime).
1046 +) Various new functions. EVP_Digest() combines EVP_Digest{Init,Update,Final}()
1047 in a single operation. X509_get0_pubkey_bitstr() extracts the public_key
1048 structure from a certificate. X509_pubkey_digest() digests tha public_key
1049 contents: this is used in various key identifiers.
1052 *) Tolerate nonRepudiation as being valid for S/MIME signing and certSign
1053 keyUsage if basicConstraints absent for a CA.
1056 *) Make SMIME_write_PKCS7() write mail header values with a format that
1057 is more generally accepted (no spaces before the semicolon), since
1058 some programs can't parse those values properly otherwise. Also make
1059 sure BIO's that break lines after each write do not create invalid
1063 +) Make sk_sort() tolerate a NULL argument.
1064 [Steve Henson reported by Massimiliano Pala <madwolf@comune.modena.it>]
1066 +) New OCSP verify flag OCSP_TRUSTOTHER. When set the "other" certificates
1067 passed by the function are trusted implicitly. If any of them signed the
1068 reponse then it is assumed to be valid and is not verified.
1071 -) Make the CRL encoding routines work with empty SEQUENCE OF. The
1072 macros previously used would not encode an empty SEQUENCE OF
1073 and break the signature.
1075 [This change does not apply to 0.9.7.]
1077 *) Zero the premaster secret after deriving the master secret in
1081 +) In PKCS7_set_type() initialise content_type in PKCS7_ENC_CONTENT
1082 to data. This was previously part of the PKCS7 ASN1 code. This
1083 was causing problems with OpenSSL created PKCS#12 and PKCS#7 structures.
1084 [Steve Henson, reported by Kenneth R. Robinette
1085 <support@securenetterm.com>]
1087 +) Add CRYPTO_push_info() and CRYPTO_pop_info() calls to new ASN1
1088 routines: without these tracing memory leaks is very painful.
1089 Fix leaks in PKCS12 and PKCS7 routines.
1092 *) Add some EVP_add_digest_alias registrations (as found in
1093 OpenSSL_add_all_digests()) to SSL_library_init()
1094 aka OpenSSL_add_ssl_algorithms(). This provides improved
1095 compatibility with peers using X.509 certificates
1096 with unconventional AlgorithmIdentifier OIDs.
1099 *) Fix for Irix with NO_ASM.
1100 ["Bruce W. Forsberg" <bruce.forsberg@baesystems.com>]
1102 *) ./config script fixes.
1103 [Ulf Moeller, Richard Levitte]
1105 +) Make X509_time_adj() cope with the new behaviour of ASN1_TIME_new().
1106 Previously it initialised the 'type' argument to V_ASN1_UTCTIME which
1107 effectively meant GeneralizedTime would never be used. Now it
1108 is initialised to -1 but X509_time_adj() now has to check the value
1109 and use ASN1_TIME_set() if the value is not V_ASN1_UTCTIME or
1110 V_ASN1_GENERALIZEDTIME, without this it always uses GeneralizedTime.
1111 [Steve Henson, reported by Kenneth R. Robinette
1112 <support@securenetterm.com>]
1114 +) Fixes to BN_to_ASN1_INTEGER when bn is zero. This would previously
1115 result in a zero length in the ASN1_INTEGER structure which was
1116 not consistent with the structure when d2i_ASN1_INTEGER() was used
1117 and would cause ASN1_INTEGER_cmp() to fail. Enhance s2i_ASN1_INTEGER()
1118 to cope with hex and negative integers. Fix bug in i2a_ASN1_INTEGER()
1119 where it did not print out a minus for negative ASN1_INTEGER.
1122 *) Fix 'openssl passwd -1'.
1125 +) Add summary printout to ocsp utility. The various functions which
1126 convert status values to strings have been renamed to:
1127 OCSP_response_status_str(), OCSP_cert_status_str() and
1128 OCSP_crl_reason_str() and are no longer static. New options
1129 to verify nonce values and to disable verification. OCSP response
1130 printout format cleaned up.
1133 +) Add additional OCSP certificate checks. These are those specified
1134 in RFC2560. This consists of two separate checks: the CA of the
1135 certificate being checked must either be the OCSP signer certificate
1136 or the issuer of the OCSP signer certificate. In the latter case the
1137 OCSP signer certificate must contain the OCSP signing extended key
1138 usage. This check is performed by attempting to match the OCSP
1139 signer or the OCSP signer CA to the issuerNameHash and issuerKeyHash
1140 in the OCSP_CERTID structures of the response.
1143 +) Initial OCSP certificate verification added to OCSP_basic_verify()
1144 and related routines. This uses the standard OpenSSL certificate
1145 verify routines to perform initial checks (just CA validity) and
1146 to obtain the certificate chain. Then additional checks will be
1147 performed on the chain. Currently the root CA is checked to see
1148 if it is explicitly trusted for OCSP signing. This is used to set
1149 a root CA as a global signing root: that is any certificate that
1150 chains to that CA is an acceptable OCSP signing certificate.
1153 +) New '-extfile ...' option to 'openssl ca' for reading X.509v3
1154 extensions from a separate configuration file.
1155 As when reading extensions from the main configuration file,
1156 the '-extensions ...' option may be used for specifying the
1158 [Massimiliano Pala <madwolf@comune.modena.it>]
1160 *) Change PKCS12_key_gen_asc() so it can cope with non null
1161 terminated strings whose length is passed in the passlen
1162 parameter, for example from PEM callbacks. This was done
1163 by adding an extra length parameter to asc2uni().
1164 [Steve Henson, reported by <oddissey@samsung.co.kr>]
1166 +) New OCSP utility. Allows OCSP requests to be generated or
1167 read. The request can be sent to a responder and the output
1168 parsed, outputed or printed in text form. Not complete yet:
1169 still needs to check the OCSP response validity.
1172 +) New subcommands for 'openssl ca':
1173 'openssl ca -status <serial>' prints the status of the cert with
1174 the given serial number (according to the index file).
1175 'openssl ca -updatedb' updates the expiry status of certificates
1177 [Massimiliano Pala <madwolf@comune.modena.it>]
1179 +) New '-newreq-nodes' command option to CA.pl. This is like
1180 '-newreq', but calls 'openssl req' with the '-nodes' option
1181 so that the resulting key is not encrypted.
1182 [Damien Miller <djm@mindrot.org>]
1184 +) New configuration for the GNU Hurd.
1185 [Jonathan Bartlett <johnnyb@wolfram.com> via Richard Levitte]
1187 +) Initial code to implement OCSP basic response verify. This
1188 is currently incomplete. Currently just finds the signer's
1189 certificate and verifies the signature on the response.
1192 +) New SSLeay_version code SSLEAY_DIR to determine the compiled-in
1193 value of OPENSSLDIR. This is available via the new '-d' option
1194 to 'openssl version', and is also included in 'openssl version -a'.
1197 *) Fix C code generated by 'openssl dsaparam -C': If a BN_bin2bn
1198 call failed, free the DSA structure.
1201 +) Allowing defining memory allocation callbacks that will be given
1202 file name and line number information in additional arguments
1203 (a const char* and an int). The basic functionality remains, as
1204 well as the original possibility to just replace malloc(),
1205 realloc() and free() by functions that do not know about these
1206 additional arguments. To register and find out the current
1207 settings for extended allocation functions, the following
1208 functions are provided:
1210 CRYPTO_set_mem_ex_functions
1211 CRYPTO_set_locked_mem_ex_functions
1212 CRYPTO_get_mem_ex_functions
1213 CRYPTO_get_locked_mem_ex_functions
1215 These work the same way as CRYPTO_set_mem_functions and friends.
1216 CRYPTO_get_[locked_]mem_functions now writes 0 where such an
1217 extended allocation function is enabled.
1218 Similarly, CRYPTO_get_[locked_]mem_ex_functions writes 0 where
1219 a conventional allocation function is enabled.
1220 [Richard Levitte, Bodo Moeller]
1222 *) Fix to uni2asc() to cope with zero length Unicode strings.
1223 These are present in some PKCS#12 files.
1226 +) Finish off removing the remaining LHASH function pointer casts.
1227 There should no longer be any prototype-casting required when using
1228 the LHASH abstraction, and any casts that remain are "bugs". See
1229 the callback types and macros at the head of lhash.h for details
1230 (and "OBJ_cleanup" in crypto/objects/obj_dat.c as an example).
1233 +) Add automatic query of EGD sockets in RAND_poll() for the unix variant.
1234 If /dev/[u]random devices are not available or do not return enough
1235 entropy, EGD style sockets (served by EGD or PRNGD) will automatically
1237 The locations /var/run/egd-pool, /dev/egd-pool, /etc/egd-pool, and
1238 /etc/entropy will be queried once each in this sequence, quering stops
1239 when enough entropy was collected without querying more sockets.
1242 +) Change the Unix RAND_poll() variant to be able to poll several
1243 random devices, as specified by DEVRANDOM, until a sufficient amount
1244 of data has been collected. We spend at most 10 ms on each file
1245 (select timeout) and read in non-blocking mode. DEVRANDOM now
1246 defaults to the list "/dev/urandom", "/dev/random", "/dev/srandom"
1247 (previously it was just the string "/dev/urandom"), so on typical
1248 platforms the 10 ms delay will never occur.
1249 Also separate out the Unix variant to its own file, rand_unix.c.
1250 For VMS, there's a currently-empty rand_vms.c.
1253 +) Move OCSP client related routines to ocsp_cl.c. These
1254 provide utility functions which an application needing
1255 to issue a request to an OCSP responder and analyse the
1256 response will typically need: as opposed to those which an
1257 OCSP responder itself would need which will be added later.
1259 OCSP_request_sign() signs an OCSP request with an API similar
1260 to PKCS7_sign(). OCSP_response_status() returns status of OCSP
1261 response. OCSP_response_get1_basic() extracts basic response
1262 from response. OCSP_resp_find_status(): finds and extracts status
1263 information from an OCSP_CERTID structure (which will be created
1264 when the request structure is built). These are built from lower
1265 level functions which work on OCSP_SINGLERESP structures but
1266 wont normally be used unless the application wishes to examine
1267 extensions in the OCSP response for example.
1269 Replace nonce routines with a pair of functions.
1270 OCSP_request_add1_nonce() adds a nonce value and optionally
1271 generates a random value. OCSP_check_nonce() checks the
1272 validity of the nonce in an OCSP response.
1275 +) Change function OCSP_request_add() to OCSP_request_add0_id().
1276 This doesn't copy the supplied OCSP_CERTID and avoids the
1277 need to free up the newly created id. Change return type
1278 to OCSP_ONEREQ to return the internal OCSP_ONEREQ structure.
1279 This can then be used to add extensions to the request.
1280 Deleted OCSP_request_new(), since most of its functionality
1281 is now in OCSP_REQUEST_new() (and the case insensitive name
1282 clash) apart from the ability to set the request name which
1283 will be added elsewhere.
1286 +) Update OCSP API. Remove obsolete extensions argument from
1287 various functions. Extensions are now handled using the new
1288 OCSP extension code. New simple OCSP HTTP function which
1289 can be used to send requests and parse the response.
1292 +) Fix the PKCS#7 (S/MIME) code to work with new ASN1. Two new
1293 ASN1_ITEM structures help with sign and verify. PKCS7_ATTR_SIGN
1294 uses the special reorder version of SET OF to sort the attributes
1295 and reorder them to match the encoded order. This resolves a long
1296 standing problem: a verify on a PKCS7 structure just after signing
1297 it used to fail because the attribute order did not match the
1298 encoded order. PKCS7_ATTR_VERIFY does not reorder the attributes:
1299 it uses the received order. This is necessary to tolerate some broken
1300 software that does not order SET OF. This is handled by encoding
1301 as a SEQUENCE OF but using implicit tagging (with UNIVERSAL class)
1302 to produce the required SET OF.
1305 +) Have mk1mf.pl generate the macros OPENSSL_BUILD_SHLIBCRYPTO and
1306 OPENSSL_BUILD_SHLIBSSL and use them appropriately in the header
1307 files to get correct declarations of the ASN.1 item variables.
1310 +) Rewrite of PKCS#12 code to use new ASN1 functionality. Replace many
1311 PKCS#12 macros with real functions. Fix two unrelated ASN1 bugs:
1312 asn1_check_tlen() would sometimes attempt to use 'ctx' when it was
1313 NULL and ASN1_TYPE was not dereferenced properly in asn1_ex_c2i().
1314 New ASN1 macro: DECLARE_ASN1_ITEM() which just declares the relevant
1315 ASN1_ITEM and no wrapper functions.
1318 +) New functions or ASN1_item_d2i_fp() and ASN1_item_d2i_bio(). These
1319 replace the old function pointer based I/O routines. Change most of
1320 the *_d2i_bio() and *_d2i_fp() functions to use these.
1323 +) Enhance mkdef.pl to be more accepting about spacing in C preprocessor
1324 lines, recognice more "algorithms" that can be deselected, and make
1325 it complain about algorithm deselection that isn't recognised.
1328 +) New ASN1 functions to handle dup, sign, verify, digest, pack and
1329 unpack operations in terms of ASN1_ITEM. Modify existing wrappers
1330 to use new functions. Add NO_ASN1_OLD which can be set to remove
1331 some old style ASN1 functions: this can be used to determine if old
1332 code will still work when these eventually go away.
1335 +) New extension functions for OCSP structures, these follow the
1336 same conventions as certificates and CRLs.
1339 +) New function X509V3_add1_i2d(). This automatically encodes and
1340 adds an extension. Its behaviour can be customised with various
1341 flags to append, replace or delete. Various wrappers added for
1342 certifcates and CRLs.
1345 +) Fix to avoid calling the underlying ASN1 print routine when
1346 an extension cannot be parsed. Correct a typo in the
1347 OCSP_SERVICELOC extension. Tidy up print OCSP format.
1350 *) Increase s2->wbuf allocation by one byte in ssl2_new (ssl/s2_lib.c).
1351 Otherwise do_ssl_write (ssl/s2_pkt.c) will write beyond buffer limits
1352 when writing a 32767 byte record.
1353 [Bodo Moeller; problem reported by Eric Day <eday@concentric.net>]
1355 *) In RSA_eay_public_{en,ed}crypt and RSA_eay_mod_exp (rsa_eay.c),
1356 obtain lock CRYPTO_LOCK_RSA before setting rsa->_method_mod_{n,p,q}.
1358 (RSA objects have a reference count access to which is protected
1359 by CRYPTO_LOCK_RSA [see rsa_lib.c, s3_srvr.c, ssl_cert.c, ssl_rsa.c],
1360 so they are meant to be shared between threads.)
1361 [Bodo Moeller, Geoff Thorpe; original patch submitted by
1362 "Reddie, Steven" <Steven.Reddie@ca.com>]
1364 +) Make mkdef.pl parse some of the ASN1 macros and add apropriate
1365 entries for variables.
1368 *) Fix a deadlock in CRYPTO_mem_leaks().
1371 +) Add functionality to apps/openssl.c for detecting locking
1372 problems: As the program is single-threaded, all we have
1373 to do is register a locking callback using an array for
1374 storing which locks are currently held by the program.
1377 +) Use a lock around the call to CRYPTO_get_ex_new_index() in
1378 SSL_get_ex_data_X509_STORE_idx(), which is used in
1379 ssl_verify_cert_chain() and thus can be called at any time
1380 during TLS/SSL handshakes so that thread-safety is essential.
1381 Unfortunately, the ex_data design is not at all suited
1382 for multi-threaded use, so it probably should be abolished.
1385 +) Added Broadcom "ubsec" ENGINE to OpenSSL.
1386 [Broadcom, tweaked and integrated by Geoff Thorpe]
1388 +) Move common extension printing code to new function
1389 X509V3_print_extensions(). Reorganise OCSP print routines and
1390 implement some needed OCSP ASN1 functions. Add OCSP extensions.
1393 +) New function X509_signature_print() to remove duplication in some
1397 +) Add a special meaning when SET OF and SEQUENCE OF flags are both
1398 set (this was treated exactly the same as SET OF previously). This
1399 is used to reorder the STACK representing the structure to match the
1400 encoding. This will be used to get round a problem where a PKCS7
1401 structure which was signed could not be verified because the STACK
1402 order did not reflect the encoded order.
1405 +) Reimplement the OCSP ASN1 module using the new code.
1408 +) Update the X509V3 code to permit the use of an ASN1_ITEM structure
1409 for its ASN1 operations. The old style function pointers still exist
1410 for now but they will eventually go away.
1413 +) Merge in replacement ASN1 code from the ASN1 branch. This almost
1414 completely replaces the old ASN1 functionality with a table driven
1415 encoder and decoder which interprets an ASN1_ITEM structure describing
1416 the ASN1 module. Compatibility with the existing ASN1 API (i2d,d2i) is
1417 largely maintained. Almost all of the old asn1_mac.h macro based ASN1
1418 has also been converted to the new form.
1421 +) Change BN_mod_exp_recp so that negative moduli are tolerated
1422 (the sign is ignored). Similarly, ignore the sign in BN_MONT_CTX_set
1423 so that BN_mod_exp_mont and BN_mod_exp_mont_word work
1424 for negative moduli.
1427 +) Fix BN_uadd and BN_usub: Always return non-negative results instead
1428 of not touching the result's sign bit.
1431 +) BN_div bugfix: If the result is 0, the sign (res->neg) must not be
1435 +) Changed the LHASH code to use prototypes for callbacks, and created
1436 macros to declare and implement thin (optionally static) functions
1437 that provide type-safety and avoid function pointer casting for the
1438 type-specific callbacks.
1441 *) Use better test patterns in bntest.
1444 +) Added Kerberos Cipher Suites to be used with TLS, as written in
1446 [Veers Staats <staatsvr@asc.hpc.mil>,
1447 Jeffrey Altman <jaltman@columbia.edu>, via Richard Levitte]
1449 *) rand_win.c fix for Borland C.
1452 *) BN_rshift bugfix for n == 0.
1455 +) Reformat the FAQ so the different questions and answers can be divided
1456 in sections depending on the subject.
1459 +) Have the zlib compression code load ZLIB.DLL dynamically under
1463 +) New function BN_mod_sqrt for computing square roots modulo a prime
1464 (using the probabilistic Tonelli-Shanks algorithm unless
1465 p == 3 (mod 4) or p == 5 (mod 8), which are cases that can
1466 be handled deterministically).
1467 [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller]
1469 *) Add a 'bctest' script that checks for some known 'bc' bugs
1470 so that 'make test' does not abort just because 'bc' is broken.
1473 *) Store verify_result within SSL_SESSION also for client side to
1474 avoid potential security hole. (Re-used sessions on the client side
1475 always resulted in verify_result==X509_V_OK, not using the original
1476 result of the server certificate verification.)
1479 +) Make BN_mod_inverse faster by explicitly handling small quotients
1480 in the Euclid loop. (Speed gain about 20% for small moduli [256 or
1481 512 bits], about 30% for larger ones [1024 or 2048 bits].)
1484 *) Fix ssl3_pending: If the record in s->s3->rrec is not of type
1485 SSL3_RT_APPLICATION_DATA, return 0.
1486 Similarly, change ssl2_pending to return 0 if SSL_in_init(s) is true.
1490 Both ssl2_peek and ssl3_peek, which were totally broken in earlier
1491 releases, have been re-implemented by renaming the previous
1492 implementations of ssl2_read and ssl3_read to ssl2_read_internal
1493 and ssl3_read_internal, respectively, and adding 'peek' parameters
1494 to them. The new ssl[23]_{read,peek} functions are calls to
1495 ssl[23]_read_internal with the 'peek' flag set appropriately.
1496 A 'peek' parameter has also been added to ssl3_read_bytes, which
1497 does the actual work for ssl3_read_internal.
1500 +) New function BN_kronecker.
1503 +) Fix BN_gcd so that it works on negative inputs; the result is
1504 positive unless both parameters are zero.
1505 Previously something reasonably close to an infinite loop was
1506 possible because numbers could be growing instead of shrinking
1507 in the implementation of Euclid's algorithm.
1510 +) Fix BN_is_word() and BN_is_one() macros to take into account the
1511 sign of the number in question.
1513 Fix BN_is_word(a,w) to work correctly for w == 0.
1515 The old BN_is_word(a,w) macro is now called BN_abs_is_word(a,w)
1516 because its test if the absolute value of 'a' equals 'w'.
1517 Note that BN_abs_is_word does *not* handle w == 0 reliably;
1518 it exists mostly for use in the implementations of BN_is_zero(),
1519 BN_is_one(), and BN_is_word().
1522 *) Initialise "ex_data" member of RSA/DSA/DH structures prior to calling
1523 the method-specific "init()" handler. Also clean up ex_data after
1524 calling the method-specific "finish()" handler. Previously, this was
1525 happening the other way round.
1528 +) New function BN_swap.
1531 +) Use BN_nnmod instead of BN_mod in crypto/bn/bn_exp.c so that
1532 the exponentiation functions are more likely to produce reasonable
1533 results on negative inputs.
1536 +) Change BN_mod_mul so that the result is always non-negative.
1537 Previously, it could be negative if one of the factors was negative;
1538 I don't think anyone really wanted that behaviour.
1541 +) Move BN_mod_... functions into new file crypto/bn/bn_mod.c
1542 (except for exponentiation, which stays in crypto/bn/bn_exp.c,
1543 and BN_mod_mul_reciprocal, which stays in crypto/bn/bn_recp.c)
1544 and add new functions:
1553 BN_mod_lshift1_quick
1557 These functions always generate non-negative results.
1559 BN_nnmod otherwise is like BN_mod (if BN_mod computes a remainder r
1560 such that |m| < r < 0, BN_nnmod will output rem + |m| instead).
1562 BN_mod_XXX_quick(r, a, [b,] m) generates the same result as
1563 BN_mod_XXX(r, a, [b,] m, ctx), but requires that a [and b]
1564 be reduced modulo m.
1565 [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller]
1567 +) Remove a few calls to bn_wexpand() in BN_sqr() (the one in there
1568 was actually never needed) and in BN_mul(). The removal in BN_mul()
1569 required a small change in bn_mul_part_recursive() and the addition
1570 of the functions bn_cmp_part_words(), bn_sub_part_words() and
1571 bn_add_part_words(), which do the same thing as bn_cmp_words(),
1572 bn_sub_words() and bn_add_words() except they take arrays with
1576 +) In 'openssl passwd', verify passwords read from the terminal
1577 unless the '-salt' option is used (which usually means that
1578 verification would just waste user's time since the resulting
1579 hash is going to be compared with some given password hash)
1580 or the new '-noverify' option is used.
1582 This is an incompatible change, but it does not affect
1583 non-interactive use of 'openssl passwd' (passwords on the command
1584 line, '-stdin' option, '-in ...' option) and thus should not
1588 +) Remove all references to RSAref, since there's no more need for it.
1591 *) Increase BN_CTX_NUM (the number of BIGNUMs in a BN_CTX) to 16.
1592 The previous value, 12, was not always sufficient for BN_mod_exp().
1595 +) Make DSO load along a path given through an environment variable
1596 (SHLIB_PATH) with shl_load().
1599 +) Constify the ENGINE code as a result of BIGNUM constification.
1600 Also constify the RSA code and most things related to it. In a
1601 few places, most notable in the depth of the ASN.1 code, ugly
1602 casts back to non-const were required (to be solved at a later
1606 +) Make it so the openssl application has all engines loaded by default.
1609 +) Constify the BIGNUM routines a little more.
1612 *) Make sure that shared libraries get the internal name engine with
1613 the full version number and not just 0. This should mark the
1614 shared libraries as not backward compatible. Of course, this should
1615 be changed again when we can guarantee backward binary compatibility.
1618 +) Add the following functions:
1620 ENGINE_load_cswift()
1622 ENGINE_load_atalla()
1624 ENGINE_load_builtin_engines()
1626 That way, an application can itself choose if external engines that
1627 are built-in in OpenSSL shall ever be used or not. The benefit is
1628 that applications won't have to be linked with libdl or other dso
1629 libraries unless it's really needed.
1631 Changed 'openssl engine' to load all engines on demand.
1632 Changed the engine header files to avoid the duplication of some
1633 declarations (they differed!).
1636 +) 'openssl engine' can now list capabilities.
1639 +) Better error reporting in 'openssl engine'.
1642 +) Never call load_dh_param(NULL) in s_server.
1645 +) Add engine application. It can currently list engines by name and
1646 identity, and test if they are actually available.
1649 +) Add support for shared libraries under Irix.
1650 [Albert Chin-A-Young <china@thewrittenword.com>]
1652 +) Improve RPM specification file by forcing symbolic linking and making
1653 sure the installed documentation is also owned by root.root.
1654 [Damien Miller <djm@mindrot.org>]
1656 +) Add configuration option to build on Linux on both big-endian and
1658 [Ralf Baechle <ralf@uni-koblenz.de>]
1660 +) Give the OpenSSL applications more possibilities to make use of
1661 keys (public as well as private) handled by engines.
1664 +) Add OCSP code that comes from CertCo.
1667 +) Add VMS support for the Rijndael code.
1670 +) Added untested support for Nuron crypto accelerator.
1673 +) Add support for external cryptographic devices. This code was
1674 previously distributed separately as the "engine" branch.
1675 [Geoff Thorpe, Richard Levitte]
1677 +) Rework the filename-translation in the DSO code. It is now possible to
1678 have far greater control over how a "name" is turned into a filename
1679 depending on the operating environment and any oddities about the
1680 different shared library filenames on each system.
1683 +) Support threads on FreeBSD-elf in Configure.
1686 +) Add the possibility to create shared libraries on HP-UX
1689 +) Fix for SHA1 assembly problem with MASM: it produces
1690 warnings about corrupt line number information when assembling
1691 with debugging information. This is caused by the overlapping
1693 [Bernd Matthes <mainbug@celocom.de>, Steve Henson]
1696 NCONF_get_number() has no error checking at all. As a replacement,
1697 NCONF_get_number_e() is defined (_e for "error checking") and is
1698 promoted strongly. The old NCONF_get_number is kept around for
1699 binary backward compatibility.
1700 Make it possible for methods to load from something other than a BIO,
1701 by providing a function pointer that is given a name instead of a BIO.
1702 For example, this could be used to load configuration data from an
1706 *) Fix typo in get_cert_by_subject() in by_dir.c
1707 [Jean-Marc Desperrier <jean-marc.desperrier@certplus.com>]
1709 *) Rework the system to generate shared libraries:
1711 - Make note of the expected extension for the shared libraries and
1712 if there is a need for symbolic links from for example libcrypto.so.0
1713 to libcrypto.so.0.9.7. There is extended info in Configure for
1716 - Make as few rebuilds of the shared libraries as possible.
1718 - Still avoid linking the OpenSSL programs with the shared libraries.
1720 - When installing, install the shared libraries separately from the
1724 +) Fix for non blocking accept BIOs. Added new I/O special reason
1725 BIO_RR_ACCEPT to cover this case. Previously use of accept BIOs
1726 with non blocking I/O was not possible because no retry code was
1727 implemented. Also added new SSL code SSL_WANT_ACCEPT to cover
1731 +) Added the beginnings of Rijndael support.
1734 +) Fix for bug in DirectoryString mask setting. Add support for
1735 X509_NAME_print_ex() in 'req' and X509_print_ex() function
1736 to allow certificate printing to more controllable, additional
1737 'certopt' option to 'x509' to allow new printing options to be
1741 +) Clean old EAY MD5 hack from e_os.h.
1744 *) Fix SSL_CTX_set_read_ahead macro to actually use its argument.
1746 Copy SSL_CTX's read_ahead flag to SSL object directly in SSL_new
1747 and not in SSL_clear because the latter is also used by the
1748 accept/connect functions; previously, the settings made by
1749 SSL_set_read_ahead would be lost during the handshake.
1750 [Bodo Moeller; problems reported by Anders Gertz <gertz@epact.se>]
1752 *) Correct util/mkdef.pl to be selective about disabled algorithms.
1753 Previously, it would create entries for disableed algorithms no
1757 *) Added several new manual pages for SSL_* function.
1760 Changes between 0.9.5a and 0.9.6 [24 Sep 2000]
1762 *) In ssl23_get_client_hello, generate an error message when faced
1763 with an initial SSL 3.0/TLS record that is too small to contain the
1764 first two bytes of the ClientHello message, i.e. client_version.
1765 (Note that this is a pathologic case that probably has never happened
1766 in real life.) The previous approach was to use the version number
1767 from the record header as a substitute; but our protocol choice
1768 should not depend on that one because it is not authenticated
1769 by the Finished messages.
1772 *) More robust randomness gathering functions for Windows.
1773 [Jeffrey Altman <jaltman@columbia.edu>]
1775 *) For compatibility reasons if the flag X509_V_FLAG_ISSUER_CHECK is
1776 not set then we don't setup the error code for issuer check errors
1777 to avoid possibly overwriting other errors which the callback does
1778 handle. If an application does set the flag then we assume it knows
1779 what it is doing and can handle the new informational codes
1783 *) Fix for a nasty bug in ASN1_TYPE handling. ASN1_TYPE is used for
1784 a general "ANY" type, as such it should be able to decode anything
1785 including tagged types. However it didn't check the class so it would
1786 wrongly interpret tagged types in the same way as their universal
1787 counterpart and unknown types were just rejected. Changed so that the
1788 tagged and unknown types are handled in the same way as a SEQUENCE:
1789 that is the encoding is stored intact. There is also a new type
1790 "V_ASN1_OTHER" which is used when the class is not universal, in this
1791 case we have no idea what the actual type is so we just lump them all
1795 *) On VMS, stdout may very well lead to a file that is written to
1796 in a record-oriented fashion. That means that every write() will
1797 write a separate record, which will be read separately by the
1798 programs trying to read from it. This can be very confusing.
1800 The solution is to put a BIO filter in the way that will buffer
1801 text until a linefeed is reached, and then write everything a
1802 line at a time, so every record written will be an actual line,
1803 not chunks of lines and not (usually doesn't happen, but I've
1804 seen it once) several lines in one record. BIO_f_linebuffer() is
1807 Currently, it's a VMS-only method, because that's where it has
1808 been tested well enough.
1811 *) Remove 'optimized' squaring variant in BN_mod_mul_montgomery,
1812 it can return incorrect results.
1813 (Note: The buggy variant was not enabled in OpenSSL 0.9.5a,
1814 but it was in 0.9.6-beta[12].)
1817 *) Disable the check for content being present when verifying detached
1818 signatures in pk7_smime.c. Some versions of Netscape (wrongly)
1819 include zero length content when signing messages.
1822 *) New BIO_shutdown_wr macro, which invokes the BIO_C_SHUTDOWN_WR
1823 BIO_ctrl (for BIO pairs).
1826 *) Add DSO method for VMS.
1829 *) Bug fix: Montgomery multiplication could produce results with the
1833 *) Add RPM specification openssl.spec and modify it to build three
1834 packages. The default package contains applications, application
1835 documentation and run-time libraries. The devel package contains
1836 include files, static libraries and function documentation. The
1837 doc package contains the contents of the doc directory. The original
1838 openssl.spec was provided by Damien Miller <djm@mindrot.org>.
1841 *) Add a large number of documentation files for many SSL routines.
1842 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>]
1844 *) Add a configuration entry for Sony News 4.
1845 [NAKAJI Hiroyuki <nakaji@tutrp.tut.ac.jp>]
1847 *) Don't set the two most significant bits to one when generating a
1848 random number < q in the DSA library.
1851 *) New SSL API mode 'SSL_MODE_AUTO_RETRY'. This disables the default
1852 behaviour that SSL_read may result in SSL_ERROR_WANT_READ (even if
1853 the underlying transport is blocking) if a handshake took place.
1854 (The default behaviour is needed by applications such as s_client
1855 and s_server that use select() to determine when to use SSL_read;
1856 but for applications that know in advance when to expect data, it
1857 just makes things more complicated.)
1860 *) Add RAND_egd_bytes(), which gives control over the number of bytes read
1864 *) Add a few more EBCDIC conditionals that make `req' and `x509'
1865 work better on such systems.
1866 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>]
1868 *) Add two demo programs for PKCS12_parse() and PKCS12_create().
1869 Update PKCS12_parse() so it copies the friendlyName and the
1870 keyid to the certificates aux info.
1873 *) Fix bug in PKCS7_verify() which caused an infinite loop
1874 if there was more than one signature.
1875 [Sven Uszpelkat <su@celocom.de>]
1877 *) Major change in util/mkdef.pl to include extra information
1878 about each symbol, as well as presentig variables as well
1879 as functions. This change means that there's n more need
1880 to rebuild the .num files when some algorithms are excluded.
1883 *) Allow the verify time to be set by an application,
1884 rather than always using the current time.
1887 *) Phase 2 verify code reorganisation. The certificate
1888 verify code now looks up an issuer certificate by a
1889 number of criteria: subject name, authority key id
1890 and key usage. It also verifies self signed certificates
1891 by the same criteria. The main comparison function is
1892 X509_check_issued() which performs these checks.
1894 Lot of changes were necessary in order to support this
1895 without completely rewriting the lookup code.
1897 Authority and subject key identifier are now cached.
1899 The LHASH 'certs' is X509_STORE has now been replaced
1900 by a STACK_OF(X509_OBJECT). This is mainly because an
1901 LHASH can't store or retrieve multiple objects with
1902 the same hash value.
1904 As a result various functions (which were all internal
1905 use only) have changed to handle the new X509_STORE
1906 structure. This will break anything that messed round
1907 with X509_STORE internally.
1909 The functions X509_STORE_add_cert() now checks for an
1910 exact match, rather than just subject name.
1912 The X509_STORE API doesn't directly support the retrieval
1913 of multiple certificates matching a given criteria, however
1914 this can be worked round by performing a lookup first
1915 (which will fill the cache with candidate certificates)
1916 and then examining the cache for matches. This is probably
1917 the best we can do without throwing out X509_LOOKUP
1918 entirely (maybe later...).
1920 The X509_VERIFY_CTX structure has been enhanced considerably.
1922 All certificate lookup operations now go via a get_issuer()
1923 callback. Although this currently uses an X509_STORE it
1924 can be replaced by custom lookups. This is a simple way
1925 to bypass the X509_STORE hackery necessary to make this
1926 work and makes it possible to use more efficient techniques
1927 in future. A very simple version which uses a simple
1928 STACK for its trusted certificate store is also provided
1929 using X509_STORE_CTX_trusted_stack().
1931 The verify_cb() and verify() callbacks now have equivalents
1932 in the X509_STORE_CTX structure.
1934 X509_STORE_CTX also has a 'flags' field which can be used
1935 to customise the verify behaviour.
1938 *) Add new PKCS#7 signing option PKCS7_NOSMIMECAP which
1939 excludes S/MIME capabilities.
1942 *) When a certificate request is read in keep a copy of the
1943 original encoding of the signed data and use it when outputing
1944 again. Signatures then use the original encoding rather than
1945 a decoded, encoded version which may cause problems if the
1946 request is improperly encoded.
1949 *) For consistency with other BIO_puts implementations, call
1950 buffer_write(b, ...) directly in buffer_puts instead of calling
1953 In BIO_puts, increment b->num_write as in BIO_write.
1954 [Peter.Sylvester@EdelWeb.fr]
1956 *) Fix BN_mul_word for the case where the word is 0. (We have to use
1957 BN_zero, we may not return a BIGNUM with an array consisting of
1961 *) Avoid calling abort() from within the library when problems are
1962 detected, except if preprocessor symbols have been defined
1963 (such as REF_CHECK, BN_DEBUG etc.).
1966 *) New openssl application 'rsautl'. This utility can be
1967 used for low level RSA operations. DER public key
1968 BIO/fp routines also added.
1971 *) New Configure entry and patches for compiling on QNX 4.
1972 [Andreas Schneider <andreas@ds3.etech.fh-hamburg.de>]
1974 *) A demo state-machine implementation was sponsored by
1975 Nuron (http://www.nuron.com/) and is now available in
1976 demos/state_machine.
1979 *) New options added to the 'dgst' utility for signature
1980 generation and verification.
1983 *) Unrecognized PKCS#7 content types are now handled via a
1984 catch all ASN1_TYPE structure. This allows unsupported
1985 types to be stored as a "blob" and an application can
1986 encode and decode it manually.
1989 *) Fix various signed/unsigned issues to make a_strex.c
1991 [Oscar Jacobsson <oscar.jacobsson@celocom.com>]
1993 *) ASN1 fixes. i2d_ASN1_OBJECT was not returning the correct
1994 length if passed a buffer. ASN1_INTEGER_to_BN failed
1995 if passed a NULL BN and its argument was negative.
1996 [Steve Henson, pointed out by Sven Heiberg <sven@tartu.cyber.ee>]
1998 *) Modification to PKCS#7 encoding routines to output definite
1999 length encoding. Since currently the whole structures are in
2000 memory there's not real point in using indefinite length
2001 constructed encoding. However if OpenSSL is compiled with
2002 the flag PKCS7_INDEFINITE_ENCODING the old form is used.
2005 *) Added BIO_vprintf() and BIO_vsnprintf().
2008 *) Added more prefixes to parse for in the the strings written
2009 through a logging bio, to cover all the levels that are available
2010 through syslog. The prefixes are now:
2012 PANIC, EMERG, EMR => LOG_EMERG
2013 ALERT, ALR => LOG_ALERT
2014 CRIT, CRI => LOG_CRIT
2015 ERROR, ERR => LOG_ERR
2016 WARNING, WARN, WAR => LOG_WARNING
2017 NOTICE, NOTE, NOT => LOG_NOTICE
2018 INFO, INF => LOG_INFO
2019 DEBUG, DBG => LOG_DEBUG
2021 and as before, if none of those prefixes are present at the
2022 beginning of the string, LOG_ERR is chosen.
2024 On Win32, the LOG_* levels are mapped according to this:
2026 LOG_EMERG, LOG_ALERT, LOG_CRIT, LOG_ERR => EVENTLOG_ERROR_TYPE
2027 LOG_WARNING => EVENTLOG_WARNING_TYPE
2028 LOG_NOTICE, LOG_INFO, LOG_DEBUG => EVENTLOG_INFORMATION_TYPE
2032 *) Made it possible to reconfigure with just the configuration
2033 argument "reconf" or "reconfigure". The command line arguments
2034 are stored in Makefile.ssl in the variable CONFIGURE_ARGS,
2035 and are retrieved from there when reconfiguring.
2039 [Assar Westerlund <assar@sics.se>, Richard Levitte]
2041 *) Add the arguments -CAfile and -CApath to the pkcs12 utility.
2044 *) The obj_dat.pl script was messing up the sorting of object
2045 names. The reason was that it compared the quoted version
2046 of strings as a result "OCSP" > "OCSP Signing" because
2047 " > SPACE. Changed script to store unquoted versions of
2048 names and add quotes on output. It was also omitting some
2049 names from the lookup table if they were given a default
2050 value (that is if SN is missing it is given the same
2051 value as LN and vice versa), these are now added on the
2052 grounds that if an object has a name we should be able to
2053 look it up. Finally added warning output when duplicate
2054 short or long names are found.
2057 *) Changes needed for Tandem NSK.
2058 [Scott Uroff <scott@xypro.com>]
2060 *) Fix SSL 2.0 rollback checking: Due to an off-by-one error in
2061 RSA_padding_check_SSLv23(), special padding was never detected
2062 and thus the SSL 3.0/TLS 1.0 countermeasure against protocol
2063 version rollback attacks was not effective.
2065 In s23_clnt.c, don't use special rollback-attack detection padding
2066 (RSA_SSLV23_PADDING) if SSL 2.0 is the only protocol enabled in the
2067 client; similarly, in s23_srvr.c, don't do the rollback check if
2068 SSL 2.0 is the only protocol enabled in the server.
2071 *) Make it possible to get hexdumps of unprintable data with 'openssl
2072 asn1parse'. By implication, the functions ASN1_parse_dump() and
2073 BIO_dump_indent() are added.
2076 *) New functions ASN1_STRING_print_ex() and X509_NAME_print_ex()
2077 these print out strings and name structures based on various
2078 flags including RFC2253 support and proper handling of
2079 multibyte characters. Added options to the 'x509' utility
2080 to allow the various flags to be set.
2083 *) Various fixes to use ASN1_TIME instead of ASN1_UTCTIME.
2084 Also change the functions X509_cmp_current_time() and
2085 X509_gmtime_adj() work with an ASN1_TIME structure,
2086 this will enable certificates using GeneralizedTime in validity
2087 dates to be checked.
2090 *) Make the NEG_PUBKEY_BUG code (which tolerates invalid
2091 negative public key encodings) on by default,
2092 NO_NEG_PUBKEY_BUG can be set to disable it.
2095 *) New function c2i_ASN1_OBJECT() which acts on ASN1_OBJECT
2096 content octets. An i2c_ASN1_OBJECT is unnecessary because
2097 the encoding can be trivially obtained from the structure.
2100 *) crypto/err.c locking bugfix: Use write locks (CRYPTO_w_[un]lock),
2101 not read locks (CRYPTO_r_[un]lock).
2104 *) A first attempt at creating official support for shared
2105 libraries through configuration. I've kept it so the
2106 default is static libraries only, and the OpenSSL programs
2107 are always statically linked for now, but there are
2108 preparations for dynamic linking in place.
2109 This has been tested on Linux and Tru64.
2112 *) Randomness polling function for Win9x, as described in:
2113 Peter Gutmann, Software Generation of Practically Strong
2117 *) Fix so PRNG is seeded in req if using an already existing
2121 *) New options to smime application. -inform and -outform
2122 allow alternative formats for the S/MIME message including
2123 PEM and DER. The -content option allows the content to be
2124 specified separately. This should allow things like Netscape
2125 form signing output easier to verify.
2128 *) Fix the ASN1 encoding of tags using the 'long form'.
2131 *) New ASN1 functions, i2c_* and c2i_* for INTEGER and BIT
2132 STRING types. These convert content octets to and from the
2133 underlying type. The actual tag and length octets are
2134 already assumed to have been read in and checked. These
2135 are needed because all other string types have virtually
2136 identical handling apart from the tag. By having versions
2137 of the ASN1 functions that just operate on content octets
2138 IMPLICIT tagging can be handled properly. It also allows
2139 the ASN1_ENUMERATED code to be cut down because ASN1_ENUMERATED
2140 and ASN1_INTEGER are identical apart from the tag.
2143 *) Change the handling of OID objects as follows:
2145 - New object identifiers are inserted in objects.txt, following
2146 the syntax given in objects.README.
2147 - objects.pl is used to process obj_mac.num and create a new
2149 - obj_dat.pl is used to create a new obj_dat.h, using the data in
2152 This is currently kind of a hack, and the perl code in objects.pl
2153 isn't very elegant, but it works as I intended. The simplest way
2154 to check that it worked correctly is to look in obj_dat.h and
2155 check the array nid_objs and make sure the objects haven't moved
2156 around (this is important!). Additions are OK, as well as
2157 consistent name changes.
2160 *) Add BSD-style MD5-based passwords to 'openssl passwd' (option '-1').
2163 *) Addition of the command line parameter '-rand file' to 'openssl req'.
2164 The given file adds to whatever has already been seeded into the
2165 random pool through the RANDFILE configuration file option or
2166 environment variable, or the default random state file.
2169 *) mkstack.pl now sorts each macro group into lexical order.
2170 Previously the output order depended on the order the files
2171 appeared in the directory, resulting in needless rewriting
2175 *) Patches to make OpenSSL compile under Win32 again. Mostly
2176 work arounds for the VC++ problem that it treats func() as
2177 func(void). Also stripped out the parts of mkdef.pl that
2178 added extra typesafe functions: these no longer exist.
2181 *) Reorganisation of the stack code. The macros are now all
2182 collected in safestack.h . Each macro is defined in terms of
2183 a "stack macro" of the form SKM_<name>(type, a, b). The
2184 DEBUG_SAFESTACK is now handled in terms of function casts,
2185 this has the advantage of retaining type safety without the
2186 use of additional functions. If DEBUG_SAFESTACK is not defined
2187 then the non typesafe macros are used instead. Also modified the
2188 mkstack.pl script to handle the new form. Needs testing to see
2189 if which (if any) compilers it chokes and maybe make DEBUG_SAFESTACK
2190 the default if no major problems. Similar behaviour for ASN1_SET_OF
2191 and PKCS12_STACK_OF.
2194 *) When some versions of IIS use the 'NET' form of private key the
2195 key derivation algorithm is different. Normally MD5(password) is
2196 used as a 128 bit RC4 key. In the modified case
2197 MD5(MD5(password) + "SGCKEYSALT") is used insted. Added some
2198 new functions i2d_RSA_NET(), d2i_RSA_NET() etc which are the same
2199 as the old Netscape_RSA functions except they have an additional
2200 'sgckey' parameter which uses the modified algorithm. Also added
2201 an -sgckey command line option to the rsa utility. Thanks to
2202 Adrian Peck <bertie@ncipher.com> for posting details of the modified
2203 algorithm to openssl-dev.
2206 *) The evp_local.h macros were using 'c.##kname' which resulted in
2207 invalid expansion on some systems (SCO 5.0.5 for example).
2208 Corrected to 'c.kname'.
2209 [Phillip Porch <root@theporch.com>]
2211 *) New X509_get1_email() and X509_REQ_get1_email() functions that return
2212 a STACK of email addresses from a certificate or request, these look
2213 in the subject name and the subject alternative name extensions and
2214 omit any duplicate addresses.
2217 *) Re-implement BN_mod_exp2_mont using independent (and larger) windows.
2218 This makes DSA verification about 2 % faster.
2221 *) Increase maximum window size in BN_mod_exp_... to 6 bits instead of 5
2222 (meaning that now 2^5 values will be precomputed, which is only 4 KB
2223 plus overhead for 1024 bit moduli).
2224 This makes exponentiations about 0.5 % faster for 1024 bit
2225 exponents (as measured by "openssl speed rsa2048").
2228 *) Rename memory handling macros to avoid conflicts with other
2230 Malloc => OPENSSL_malloc
2231 Malloc_locked => OPENSSL_malloc_locked
2232 Realloc => OPENSSL_realloc
2233 Free => OPENSSL_free
2236 *) New function BN_mod_exp_mont_word for small bases (roughly 15%
2237 faster than BN_mod_exp_mont, i.e. 7% for a full DH exchange).
2240 *) CygWin32 support.
2241 [John Jarvie <jjarvie@newsguy.com>]
2243 *) The type-safe stack code has been rejigged. It is now only compiled
2244 in when OpenSSL is configured with the DEBUG_SAFESTACK option and
2245 by default all type-specific stack functions are "#define"d back to
2246 standard stack functions. This results in more streamlined output
2247 but retains the type-safety checking possibilities of the original
2251 *) The STACK code has been cleaned up, and certain type declarations
2252 that didn't make a lot of sense have been brought in line. This has
2253 also involved a cleanup of sorts in safestack.h to more correctly
2254 map type-safe stack functions onto their plain stack counterparts.
2255 This work has also resulted in a variety of "const"ifications of
2256 lots of the code, especially "_cmp" operations which should normally
2257 be prototyped with "const" parameters anyway.
2260 *) When generating bytes for the first time in md_rand.c, 'stir the pool'
2261 by seeding with STATE_SIZE dummy bytes (with zero entropy count).
2262 (The PRNG state consists of two parts, the large pool 'state' and 'md',
2263 where all of 'md' is used each time the PRNG is used, but 'state'
2264 is used only indexed by a cyclic counter. As entropy may not be
2265 well distributed from the beginning, 'md' is important as a
2266 chaining variable. However, the output function chains only half
2267 of 'md', i.e. 80 bits. ssleay_rand_add, on the other hand, chains
2268 all of 'md', and seeding with STATE_SIZE dummy bytes will result
2269 in all of 'state' being rewritten, with the new values depending
2270 on virtually all of 'md'. This overcomes the 80 bit limitation.)
2273 *) In ssl/s2_clnt.c and ssl/s3_clnt.c, call ERR_clear_error() when
2274 the handshake is continued after ssl_verify_cert_chain();
2275 otherwise, if SSL_VERIFY_NONE is set, remaining error codes
2276 can lead to 'unexplainable' connection aborts later.
2277 [Bodo Moeller; problem tracked down by Lutz Jaenicke]
2279 *) Major EVP API cipher revision.
2280 Add hooks for extra EVP features. This allows various cipher
2281 parameters to be set in the EVP interface. Support added for variable
2282 key length ciphers via the EVP_CIPHER_CTX_set_key_length() function and
2283 setting of RC2 and RC5 parameters.
2285 Modify EVP_OpenInit() and EVP_SealInit() to cope with variable key length
2288 Remove lots of duplicated code from the EVP library. For example *every*
2289 cipher init() function handles the 'iv' in the same way according to the
2290 cipher mode. They also all do nothing if the 'key' parameter is NULL and
2291 for CFB and OFB modes they zero ctx->num.
2293 New functionality allows removal of S/MIME code RC2 hack.
2295 Most of the routines have the same form and so can be declared in terms
2298 By shifting this to the top level EVP_CipherInit() it can be removed from
2299 all individual ciphers. If the cipher wants to handle IVs or keys
2300 differently it can set the EVP_CIPH_CUSTOM_IV or EVP_CIPH_ALWAYS_CALL_INIT
2303 Change lots of functions like EVP_EncryptUpdate() to now return a
2304 value: although software versions of the algorithms cannot fail
2305 any installed hardware versions can.
2308 *) Implement SSL_OP_TLS_ROLLBACK_BUG: In ssl3_get_client_key_exchange, if
2309 this option is set, tolerate broken clients that send the negotiated
2310 protocol version number instead of the requested protocol version
2314 *) Call dh_tmp_cb (set by ..._TMP_DH_CB) with correct 'is_export' flag;
2315 i.e. non-zero for export ciphersuites, zero otherwise.
2316 Previous versions had this flag inverted, inconsistent with
2317 rsa_tmp_cb (..._TMP_RSA_CB).
2318 [Bodo Moeller; problem reported by Amit Chopra]
2320 *) Add missing DSA library text string. Work around for some IIS
2321 key files with invalid SEQUENCE encoding.
2324 *) Add a document (doc/standards.txt) that list all kinds of standards
2325 and so on that are implemented in OpenSSL.
2328 *) Enhance c_rehash script. Old version would mishandle certificates
2329 with the same subject name hash and wouldn't handle CRLs at all.
2330 Added -fingerprint option to crl utility, to support new c_rehash
2334 *) Eliminate non-ANSI declarations in crypto.h and stack.h.
2337 *) Fix for SSL server purpose checking. Server checking was
2338 rejecting certificates which had extended key usage present
2339 but no ssl client purpose.
2340 [Steve Henson, reported by Rene Grosser <grosser@hisolutions.com>]
2342 *) Make PKCS#12 code work with no password. The PKCS#12 spec
2343 is a little unclear about how a blank password is handled.
2344 Since the password in encoded as a BMPString with terminating
2345 double NULL a zero length password would end up as just the
2346 double NULL. However no password at all is different and is
2347 handled differently in the PKCS#12 key generation code. NS
2348 treats a blank password as zero length. MSIE treats it as no
2349 password on export: but it will try both on import. We now do
2350 the same: PKCS12_parse() tries zero length and no password if
2351 the password is set to "" or NULL (NULL is now a valid password:
2352 it wasn't before) as does the pkcs12 application.
2355 *) Bugfixes in apps/x509.c: Avoid a memory leak; and don't use
2356 perror when PEM_read_bio_X509_REQ fails, the error message must
2357 be obtained from the error queue.
2360 *) Avoid 'thread_hash' memory leak in crypto/err/err.c by freeing
2361 it in ERR_remove_state if appropriate, and change ERR_get_state
2362 accordingly to avoid race conditions (this is necessary because
2363 thread_hash is no longer constant once set).
2366 *) Bugfix for linux-elf makefile.one.
2369 *) RSA_get_default_method() will now cause a default
2370 RSA_METHOD to be chosen if one doesn't exist already.
2371 Previously this was only set during a call to RSA_new()
2372 or RSA_new_method(NULL) meaning it was possible for
2373 RSA_get_default_method() to return NULL.
2376 *) Added native name translation to the existing DSO code
2377 that will convert (if the flag to do so is set) filenames
2378 that are sufficiently small and have no path information
2379 into a canonical native form. Eg. "blah" converted to
2380 "libblah.so" or "blah.dll" etc.
2383 *) New function ERR_error_string_n(e, buf, len) which is like
2384 ERR_error_string(e, buf), but writes at most 'len' bytes
2385 including the 0 terminator. For ERR_error_string_n, 'buf'
2387 [Damien Miller <djm@mindrot.org>, Bodo Moeller]
2389 *) CONF library reworked to become more general. A new CONF
2390 configuration file reader "class" is implemented as well as a
2391 new functions (NCONF_*, for "New CONF") to handle it. The now
2392 old CONF_* functions are still there, but are reimplemented to
2393 work in terms of the new functions. Also, a set of functions
2394 to handle the internal storage of the configuration data is
2395 provided to make it easier to write new configuration file
2396 reader "classes" (I can definitely see something reading a
2397 configuration file in XML format, for example), called _CONF_*,
2398 or "the configuration storage API"...
2400 The new configuration file reading functions are:
2402 NCONF_new, NCONF_free, NCONF_load, NCONF_load_fp, NCONF_load_bio,
2403 NCONF_get_section, NCONF_get_string, NCONF_get_numbre
2405 NCONF_default, NCONF_WIN32
2407 NCONF_dump_fp, NCONF_dump_bio
2409 NCONF_default and NCONF_WIN32 are method (or "class") choosers,
2410 NCONF_new creates a new CONF object. This works in the same way
2411 as other interfaces in OpenSSL, like the BIO interface.
2412 NCONF_dump_* dump the internal storage of the configuration file,
2413 which is useful for debugging. All other functions take the same
2414 arguments as the old CONF_* functions wth the exception of the
2415 first that must be a `CONF *' instead of a `LHASH *'.
2417 To make it easer to use the new classes with the old CONF_* functions,
2418 the function CONF_set_default_method is provided.
2421 *) Add '-tls1' option to 'openssl ciphers', which was already
2422 mentioned in the documentation but had not been implemented.
2423 (This option is not yet really useful because even the additional
2424 experimental TLS 1.0 ciphers are currently treated as SSL 3.0 ciphers.)
2427 *) Initial DSO code added into libcrypto for letting OpenSSL (and
2428 OpenSSL-based applications) load shared libraries and bind to
2429 them in a portable way.
2430 [Geoff Thorpe, with contributions from Richard Levitte]
2432 Changes between 0.9.5 and 0.9.5a [1 Apr 2000]
2434 *) Make sure _lrotl and _lrotr are only used with MSVC.
2436 *) Use lock CRYPTO_LOCK_RAND correctly in ssleay_rand_status
2437 (the default implementation of RAND_status).
2439 *) Rename openssl x509 option '-crlext', which was added in 0.9.5,
2440 to '-clrext' (= clear extensions), as intended and documented.
2441 [Bodo Moeller; inconsistency pointed out by Michael Attili
2444 *) Fix for HMAC. It wasn't zeroing the rest of the block if the key length
2445 was larger than the MD block size.
2446 [Steve Henson, pointed out by Yost William <YostW@tce.com>]
2448 *) Modernise PKCS12_parse() so it uses STACK_OF(X509) for its ca argument
2449 fix a leak when the ca argument was passed as NULL. Stop X509_PUBKEY_set()
2450 using the passed key: if the passed key was a private key the result
2451 of X509_print(), for example, would be to print out all the private key
2455 *) des_quad_cksum() byte order bug fix.
2456 [Ulf Möller, using the problem description in krb4-0.9.7, where
2457 the solution is attributed to Derrick J Brashear <shadow@DEMENTIA.ORG>]
2459 *) Fix so V_ASN1_APP_CHOOSE works again: however its use is strongly
2461 [Steve Henson, pointed out by Brian Korver <briank@cs.stanford.edu>]
2463 *) For easily testing in shell scripts whether some command
2464 'openssl XXX' exists, the new pseudo-command 'openssl no-XXX'
2465 returns with exit code 0 iff no command of the given name is available.
2466 'no-XXX' is printed in this case, 'XXX' otherwise. In both cases,
2467 the output goes to stdout and nothing is printed to stderr.
2468 Additional arguments are always ignored.
2470 Since for each cipher there is a command of the same name,
2471 the 'no-cipher' compilation switches can be tested this way.
2473 ('openssl no-XXX' is not able to detect pseudo-commands such
2474 as 'quit', 'list-XXX-commands', or 'no-XXX' itself.)
2477 *) Update test suite so that 'make test' succeeds in 'no-rsa' configuration.
2480 *) For SSL_[CTX_]set_tmp_dh, don't create a DH key if SSL_OP_SINGLE_DH_USE
2481 is set; it will be thrown away anyway because each handshake creates
2483 ssl_cert_dup, which is used by SSL_new, now copies DH keys in addition
2484 to parameters -- in previous versions (since OpenSSL 0.9.3) the
2485 'default key' from SSL_CTX_set_tmp_dh would always be lost, meanining
2486 you effectivly got SSL_OP_SINGLE_DH_USE when using this macro.
2489 *) New s_client option -ign_eof: EOF at stdin is ignored, and
2490 'Q' and 'R' lose their special meanings (quit/renegotiate).
2491 This is part of what -quiet does; unlike -quiet, -ign_eof
2492 does not suppress any output.
2495 *) Add compatibility options to the purpose and trust code. The
2496 purpose X509_PURPOSE_ANY is "any purpose" which automatically
2497 accepts a certificate or CA, this was the previous behaviour,
2498 with all the associated security issues.
2500 X509_TRUST_COMPAT is the old trust behaviour: only and
2501 automatically trust self signed roots in certificate store. A
2502 new trust setting X509_TRUST_DEFAULT is used to specify that
2503 a purpose has no associated trust setting and it should instead
2504 use the value in the default purpose.
2507 *) Fix the PKCS#8 DSA private key code so it decodes keys again
2508 and fix a memory leak.
2511 *) In util/mkerr.pl (which implements 'make errors'), preserve
2512 reason strings from the previous version of the .c file, as
2513 the default to have only downcase letters (and digits) in
2514 automatically generated reasons codes is not always appropriate.
2517 *) In ERR_load_ERR_strings(), build an ERR_LIB_SYS error reason table
2518 using strerror. Previously, ERR_reason_error_string() returned
2519 library names as reason strings for SYSerr; but SYSerr is a special
2520 case where small numbers are errno values, not library numbers.
2523 *) Add '-dsaparam' option to 'openssl dhparam' application. This
2524 converts DSA parameters into DH parameters. (When creating parameters,
2525 DSA_generate_parameters is used.)
2528 *) Include 'length' (recommended exponent length) in C code generated
2529 by 'openssl dhparam -C'.
2532 *) The second argument to set_label in perlasm was already being used
2533 so couldn't be used as a "file scope" flag. Moved to third argument
2537 *) In PEM_ASN1_write_bio and some other functions, use RAND_pseudo_bytes
2538 instead of RAND_bytes for encryption IVs and salts.
2541 *) Include RAND_status() into RAND_METHOD instead of implementing
2542 it only for md_rand.c Otherwise replacing the PRNG by calling
2543 RAND_set_rand_method would be impossible.
2546 *) Don't let DSA_generate_key() enter an infinite loop if the random
2547 number generation fails.
2550 *) New 'rand' application for creating pseudo-random output.
2553 *) Added configuration support for Linux/IA64
2554 [Rolf Haberrecker <rolf@suse.de>]
2556 *) Assembler module support for Mingw32.
2559 *) Shared library support for HPUX (in shlib/).
2560 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE> and Anonymous]
2562 *) Shared library support for Solaris gcc.
2563 [Lutz Behnke <behnke@trustcenter.de>]
2565 Changes between 0.9.4 and 0.9.5 [28 Feb 2000]
2567 *) PKCS7_encrypt() was adding text MIME headers twice because they
2568 were added manually and by SMIME_crlf_copy().
2571 *) In bntest.c don't call BN_rand with zero bits argument.
2572 [Steve Henson, pointed out by Andrew W. Gray <agray@iconsinc.com>]
2574 *) BN_mul bugfix: In bn_mul_part_recursion() only the a>a[n] && b>b[n]
2575 case was implemented. This caused BN_div_recp() to fail occasionally.
2578 *) Add an optional second argument to the set_label() in the perl
2579 assembly language builder. If this argument exists and is set
2580 to 1 it signals that the assembler should use a symbol whose
2581 scope is the entire file, not just the current function. This
2582 is needed with MASM which uses the format label:: for this scope.
2583 [Steve Henson, pointed out by Peter Runestig <peter@runestig.com>]
2585 *) Change the ASN1 types so they are typedefs by default. Before
2586 almost all types were #define'd to ASN1_STRING which was causing
2587 STACK_OF() problems: you couldn't declare STACK_OF(ASN1_UTF8STRING)
2591 *) Change names of new functions to the new get1/get0 naming
2592 convention: After 'get1', the caller owns a reference count
2593 and has to call ..._free; 'get0' returns a pointer to some
2594 data structure without incrementing reference counters.
2595 (Some of the existing 'get' functions increment a reference
2596 counter, some don't.)
2597 Similarly, 'set1' and 'add1' functions increase reference
2598 counters or duplicate objects.
2601 *) Allow for the possibility of temp RSA key generation failure:
2602 the code used to assume it always worked and crashed on failure.
2605 *) Fix potential buffer overrun problem in BIO_printf().
2606 [Ulf Möller, using public domain code by Patrick Powell; problem
2607 pointed out by David Sacerdote <das33@cornell.edu>]
2609 *) Support EGD <http://www.lothar.com/tech/crypto/>. New functions
2610 RAND_egd() and RAND_status(). In the command line application,
2611 the EGD socket can be specified like a seed file using RANDFILE
2615 *) Allow the string CERTIFICATE to be tolerated in PKCS#7 structures.
2616 Some CAs (e.g. Verisign) distribute certificates in this form.
2619 *) Remove the SSL_ALLOW_ADH compile option and set the default cipher
2620 list to exclude them. This means that no special compilation option
2621 is needed to use anonymous DH: it just needs to be included in the
2625 *) Change the EVP_MD_CTX_type macro so its meaning consistent with
2626 EVP_MD_type. The old functionality is available in a new macro called
2627 EVP_MD_md(). Change code that uses it and update docs.
2630 *) ..._ctrl functions now have corresponding ..._callback_ctrl functions
2631 where the 'void *' argument is replaced by a function pointer argument.
2632 Previously 'void *' was abused to point to functions, which works on
2633 many platforms, but is not correct. As these functions are usually
2634 called by macros defined in OpenSSL header files, most source code
2635 should work without changes.
2638 *) <openssl/opensslconf.h> (which is created by Configure) now contains
2639 sections with information on -D... compiler switches used for
2640 compiling the library so that applications can see them. To enable
2641 one of these sections, a pre-processor symbol OPENSSL_..._DEFINES
2642 must be defined. E.g.,
2643 #define OPENSSL_ALGORITHM_DEFINES
2644 #include <openssl/opensslconf.h>
2645 defines all pertinent NO_<algo> symbols, such as NO_IDEA, NO_RSA, etc.
2646 [Richard Levitte, Ulf and Bodo Möller]
2648 *) Bugfix: Tolerate fragmentation and interleaving in the SSL 3/TLS
2652 *) Change the 'other' type in certificate aux info to a STACK_OF
2653 X509_ALGOR. Although not an AlgorithmIdentifier as such it has
2654 the required ASN1 format: arbitrary types determined by an OID.
2657 *) Add some PEM_write_X509_REQ_NEW() functions and a command line
2658 argument to 'req'. This is not because the function is newer or
2659 better than others it just uses the work 'NEW' in the certificate
2660 request header lines. Some software needs this.
2663 *) Reorganise password command line arguments: now passwords can be
2664 obtained from various sources. Delete the PEM_cb function and make
2665 it the default behaviour: i.e. if the callback is NULL and the
2666 usrdata argument is not NULL interpret it as a null terminated pass
2667 phrase. If usrdata and the callback are NULL then the pass phrase
2668 is prompted for as usual.
2671 *) Add support for the Compaq Atalla crypto accelerator. If it is installed,
2672 the support is automatically enabled. The resulting binaries will
2673 autodetect the card and use it if present.
2674 [Ben Laurie and Compaq Inc.]
2676 *) Work around for Netscape hang bug. This sends certificate request
2677 and server done in one record. Since this is perfectly legal in the
2678 SSL/TLS protocol it isn't a "bug" option and is on by default. See
2679 the bugs/SSLv3 entry for more info.
2682 *) HP-UX tune-up: new unified configs, HP C compiler bug workaround.
2685 *) Add -rand argument to smime and pkcs12 applications and read/write
2689 *) New 'passwd' tool for crypt(3) and apr1 password hashes.
2692 *) Add command line password options to the remaining applications.
2695 *) Bug fix for BN_div_recp() for numerators with an even number of
2699 *) More tests in bntest.c, and changed test_bn output.
2702 *) ./config recognizes MacOS X now.
2705 *) Bug fix for BN_div() when the first words of num and divsor are
2706 equal (it gave wrong results if (rem=(n1-q*d0)&BN_MASK2) < d0).
2709 *) Add support for various broken PKCS#8 formats, and command line
2710 options to produce them.
2713 *) New functions BN_CTX_start(), BN_CTX_get() and BT_CTX_end() to
2714 get temporary BIGNUMs from a BN_CTX.
2717 *) Correct return values in BN_mod_exp_mont() and BN_mod_exp2_mont()
2721 *) Change the SSLeay_add_all_*() functions to OpenSSL_add_all_*() and
2722 include a #define from the old name to the new. The original intent
2723 was that statically linked binaries could for example just call
2724 SSLeay_add_all_ciphers() to just add ciphers to the table and not
2725 link with digests. This never worked becayse SSLeay_add_all_digests()
2726 and SSLeay_add_all_ciphers() were in the same source file so calling
2727 one would link with the other. They are now in separate source files.
2730 *) Add a new -notext option to 'ca' and a -pubkey option to 'spkac'.
2733 *) Use a less unusual form of the Miller-Rabin primality test (it used
2734 a binary algorithm for exponentiation integrated into the Miller-Rabin
2735 loop, our standard modexp algorithms are faster).
2738 *) Support for the EBCDIC character set completed.
2739 [Martin Kraemer <Martin.Kraemer@Mch.SNI.De>]
2741 *) Source code cleanups: use const where appropriate, eliminate casts,
2742 use void * instead of char * in lhash.
2745 *) Bugfix: ssl3_send_server_key_exchange was not restartable
2746 (the state was not changed to SSL3_ST_SW_KEY_EXCH_B, and because of
2747 this the server could overwrite ephemeral keys that the client
2751 *) Turn DSA_is_prime into a macro that calls BN_is_prime,
2752 using 50 iterations of the Rabin-Miller test.
2754 DSA_generate_parameters now uses BN_is_prime_fasttest (with 50
2755 iterations of the Rabin-Miller test as required by the appendix
2756 to FIPS PUB 186[-1]) instead of DSA_is_prime.
2757 As BN_is_prime_fasttest includes trial division, DSA parameter
2758 generation becomes much faster.
2760 This implies a change for the callback functions in DSA_is_prime
2761 and DSA_generate_parameters: The callback function is called once
2762 for each positive witness in the Rabin-Miller test, not just
2763 occasionally in the inner loop; and the parameters to the
2764 callback function now provide an iteration count for the outer
2765 loop rather than for the current invocation of the inner loop.
2766 DSA_generate_parameters additionally can call the callback
2767 function with an 'iteration count' of -1, meaning that a
2768 candidate has passed the trial division test (when q is generated
2769 from an application-provided seed, trial division is skipped).
2772 *) New function BN_is_prime_fasttest that optionally does trial
2773 division before starting the Rabin-Miller test and has
2774 an additional BN_CTX * argument (whereas BN_is_prime always
2775 has to allocate at least one BN_CTX).
2776 'callback(1, -1, cb_arg)' is called when a number has passed the
2777 trial division stage.
2780 *) Fix for bug in CRL encoding. The validity dates weren't being handled
2784 *) New -pkcs12 option to CA.pl script to write out a PKCS#12 file.
2787 *) New function BN_pseudo_rand().
2790 *) Clean up BN_mod_mul_montgomery(): replace the broken (and unreadable)
2791 bignum version of BN_from_montgomery() with the working code from
2792 SSLeay 0.9.0 (the word based version is faster anyway), and clean up
2796 *) Avoid a race condition in s2_clnt.c (function get_server_hello) that
2797 made it impossible to use the same SSL_SESSION data structure in
2798 SSL2 clients in multiple threads.
2801 *) The return value of RAND_load_file() no longer counts bytes obtained
2802 by stat(). RAND_load_file(..., -1) is new and uses the complete file
2803 to seed the PRNG (previously an explicit byte count was required).
2804 [Ulf Möller, Bodo Möller]
2806 *) Clean up CRYPTO_EX_DATA functions, some of these didn't have prototypes
2807 used (char *) instead of (void *) and had casts all over the place.
2810 *) Make BN_generate_prime() return NULL on error if ret!=NULL.
2813 *) Retain source code compatibility for BN_prime_checks macro:
2814 BN_is_prime(..., BN_prime_checks, ...) now uses
2815 BN_prime_checks_for_size to determine the appropriate number of
2816 Rabin-Miller iterations.
2819 *) Diffie-Hellman uses "safe" primes: DH_check() return code renamed to
2820 DH_CHECK_P_NOT_SAFE_PRIME.
2821 (Check if this is true? OpenPGP calls them "strong".)
2824 *) Merge the functionality of "dh" and "gendh" programs into a new program
2825 "dhparam". The old programs are retained for now but will handle DH keys
2826 (instead of parameters) in future.
2829 *) Make the ciphers, s_server and s_client programs check the return values
2830 when a new cipher list is set.
2833 *) Enhance the SSL/TLS cipher mechanism to correctly handle the TLS 56bit
2834 ciphers. Before when the 56bit ciphers were enabled the sorting was
2837 The syntax for the cipher sorting has been extended to support sorting by
2838 cipher-strength (using the strength_bits hard coded in the tables).
2839 The new command is "@STRENGTH" (see also doc/apps/ciphers.pod).
2841 Fix a bug in the cipher-command parser: when supplying a cipher command
2842 string with an "undefined" symbol (neither command nor alphanumeric
2843 [A-Za-z0-9], ssl_set_cipher_list used to hang in an endless loop. Now
2844 an error is flagged.
2846 Due to the strength-sorting extension, the code of the
2847 ssl_create_cipher_list() function was completely rearranged. I hope that
2848 the readability was also increased :-)
2849 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>]
2851 *) Minor change to 'x509' utility. The -CAcreateserial option now uses 1
2852 for the first serial number and places 2 in the serial number file. This
2853 avoids problems when the root CA is created with serial number zero and
2854 the first user certificate has the same issuer name and serial number
2858 *) Fixes to X509_ATTRIBUTE utilities, change the 'req' program so it uses
2859 the new code. Add documentation for this stuff.
2862 *) Changes to X509_ATTRIBUTE utilities. These have been renamed from
2863 X509_*() to X509at_*() on the grounds that they don't handle X509
2864 structures and behave in an analagous way to the X509v3 functions:
2865 they shouldn't be called directly but wrapper functions should be used
2868 So we also now have some wrapper functions that call the X509at functions
2869 when passed certificate requests. (TO DO: similar things can be done with
2870 PKCS#7 signed and unsigned attributes, PKCS#12 attributes and a few other
2871 things. Some of these need some d2i or i2d and print functionality
2872 because they handle more complex structures.)
2875 *) Add missing #ifndefs that caused missing symbols when building libssl
2876 as a shared library without RSA. Use #ifndef NO_SSL2 instead of
2877 NO_RSA in ssl/s2*.c.
2878 [Kris Kennaway <kris@hub.freebsd.org>, modified by Ulf Möller]
2880 *) Precautions against using the PRNG uninitialized: RAND_bytes() now
2881 has a return value which indicates the quality of the random data
2882 (1 = ok, 0 = not seeded). Also an error is recorded on the thread's
2883 error queue. New function RAND_pseudo_bytes() generates output that is
2884 guaranteed to be unique but not unpredictable. RAND_add is like
2885 RAND_seed, but takes an extra argument for an entropy estimate
2886 (RAND_seed always assumes full entropy).
2889 *) Do more iterations of Rabin-Miller probable prime test (specifically,
2890 3 for 1024-bit primes, 6 for 512-bit primes, 12 for 256-bit primes
2891 instead of only 2 for all lengths; see BN_prime_checks_for_size definition
2892 in crypto/bn/bn_prime.c for the complete table). This guarantees a
2893 false-positive rate of at most 2^-80 for random input.
2896 *) Rewrite ssl3_read_n (ssl/s3_pkt.c) avoiding a couple of bugs.
2899 *) New function X509_CTX_rget_chain() (renamed to X509_CTX_get1_chain
2900 in the 0.9.5 release), this returns the chain
2901 from an X509_CTX structure with a dup of the stack and all
2902 the X509 reference counts upped: so the stack will exist
2903 after X509_CTX_cleanup() has been called. Modify pkcs12.c
2906 Also make SSL_SESSION_print() print out the verify return
2910 *) Add manpage for the pkcs12 command. Also change the default
2911 behaviour so MAC iteration counts are used unless the new
2912 -nomaciter option is used. This improves file security and
2913 only older versions of MSIE (4.0 for example) need it.
2916 *) Honor the no-xxx Configure options when creating .DEF files.
2919 *) Add PKCS#10 attributes to field table: challengePassword,
2920 unstructuredName and unstructuredAddress. These are taken from
2921 draft PKCS#9 v2.0 but are compatible with v1.2 provided no
2922 international characters are used.
2924 More changes to X509_ATTRIBUTE code: allow the setting of types
2925 based on strings. Remove the 'loc' parameter when adding
2926 attributes because these will be a SET OF encoding which is sorted
2930 *) Initial changes to the 'req' utility to allow request generation
2931 automation. This will allow an application to just generate a template
2932 file containing all the field values and have req construct the
2935 Initial support for X509_ATTRIBUTE handling. Stacks of these are
2936 used all over the place including certificate requests and PKCS#7
2937 structures. They are currently handled manually where necessary with
2938 some primitive wrappers for PKCS#7. The new functions behave in a
2939 manner analogous to the X509 extension functions: they allow
2940 attributes to be looked up by NID and added.
2942 Later something similar to the X509V3 code would be desirable to
2943 automatically handle the encoding, decoding and printing of the
2944 more complex types. The string types like challengePassword can
2945 be handled by the string table functions.
2947 Also modified the multi byte string table handling. Now there is
2948 a 'global mask' which masks out certain types. The table itself
2949 can use the flag STABLE_NO_MASK to ignore the mask setting: this
2950 is useful when for example there is only one permissible type
2951 (as in countryName) and using the mask might result in no valid
2955 *) Clean up 'Finished' handling, and add functions SSL_get_finished and
2956 SSL_get_peer_finished to allow applications to obtain the latest
2957 Finished messages sent to the peer or expected from the peer,
2958 respectively. (SSL_get_peer_finished is usually the Finished message
2959 actually received from the peer, otherwise the protocol will be aborted.)
2961 As the Finished message are message digests of the complete handshake
2962 (with a total of 192 bits for TLS 1.0 and more for SSL 3.0), they can
2963 be used for external authentication procedures when the authentication
2964 provided by SSL/TLS is not desired or is not enough.
2967 *) Enhanced support for Alpha Linux is added. Now ./config checks if
2968 the host supports BWX extension and if Compaq C is present on the
2969 $PATH. Just exploiting of the BWX extension results in 20-30%
2970 performance kick for some algorithms, e.g. DES and RC4 to mention
2971 a couple. Compaq C in turn generates ~20% faster code for MD5 and
2975 *) Add support for MS "fast SGC". This is arguably a violation of the
2976 SSL3/TLS protocol. Netscape SGC does two handshakes: the first with
2977 weak crypto and after checking the certificate is SGC a second one
2978 with strong crypto. MS SGC stops the first handshake after receiving
2979 the server certificate message and sends a second client hello. Since
2980 a server will typically do all the time consuming operations before
2981 expecting any further messages from the client (server key exchange
2982 is the most expensive) there is little difference between the two.
2984 To get OpenSSL to support MS SGC we have to permit a second client
2985 hello message after we have sent server done. In addition we have to
2986 reset the MAC if we do get this second client hello.
2989 *) Add a function 'd2i_AutoPrivateKey()' this will automatically decide
2990 if a DER encoded private key is RSA or DSA traditional format. Changed
2991 d2i_PrivateKey_bio() to use it. This is only needed for the "traditional"
2992 format DER encoded private key. Newer code should use PKCS#8 format which
2993 has the key type encoded in the ASN1 structure. Added DER private key
2994 support to pkcs8 application.
2997 *) SSL 3/TLS 1 servers now don't request certificates when an anonymous
2998 ciphersuites has been selected (as required by the SSL 3/TLS 1
2999 specifications). Exception: When SSL_VERIFY_FAIL_IF_NO_PEER_CERT
3000 is set, we interpret this as a request to violate the specification
3001 (the worst that can happen is a handshake failure, and 'correct'
3002 behaviour would result in a handshake failure anyway).
3005 *) In SSL_CTX_add_session, take into account that there might be multiple
3006 SSL_SESSION structures with the same session ID (e.g. when two threads
3007 concurrently obtain them from an external cache).
3008 The internal cache can handle only one SSL_SESSION with a given ID,
3009 so if there's a conflict, we now throw out the old one to achieve
3013 *) Add OIDs for idea and blowfish in CBC mode. This will allow both
3014 to be used in PKCS#5 v2.0 and S/MIME. Also add checking to
3015 some routines that use cipher OIDs: some ciphers do not have OIDs
3016 defined and so they cannot be used for S/MIME and PKCS#5 v2.0 for
3020 *) Simplify the trust setting structure and code. Now we just have
3021 two sequences of OIDs for trusted and rejected settings. These will
3022 typically have values the same as the extended key usage extension
3023 and any application specific purposes.
3025 The trust checking code now has a default behaviour: it will just
3026 check for an object with the same NID as the passed id. Functions can
3027 be provided to override either the default behaviour or the behaviour
3028 for a given id. SSL client, server and email already have functions
3029 in place for compatibility: they check the NID and also return "trusted"
3030 if the certificate is self signed.
3033 *) Add d2i,i2d bio/fp functions for PrivateKey: these convert the
3034 traditional format into an EVP_PKEY structure.
3037 *) Add a password callback function PEM_cb() which either prompts for
3038 a password if usr_data is NULL or otherwise assumes it is a null
3039 terminated password. Allow passwords to be passed on command line
3040 environment or config files in a few more utilities.
3043 *) Add a bunch of DER and PEM functions to handle PKCS#8 format private
3044 keys. Add some short names for PKCS#8 PBE algorithms and allow them
3045 to be specified on the command line for the pkcs8 and pkcs12 utilities.
3046 Update documentation.
3049 *) Support for ASN1 "NULL" type. This could be handled before by using
3050 ASN1_TYPE but there wasn't any function that would try to read a NULL
3051 and produce an error if it couldn't. For compatibility we also have
3052 ASN1_NULL_new() and ASN1_NULL_free() functions but these are faked and
3053 don't allocate anything because they don't need to.
3056 *) Initial support for MacOS is now provided. Examine INSTALL.MacOS
3058 [Andy Polyakov, Roy Woods <roy@centicsystems.ca>]
3060 *) Rebuild of the memory allocation routines used by OpenSSL code and
3061 possibly others as well. The purpose is to make an interface that
3062 provide hooks so anyone can build a separate set of allocation and
3063 deallocation routines to be used by OpenSSL, for example memory
3064 pool implementations, or something else, which was previously hard
3065 since Malloc(), Realloc() and Free() were defined as macros having
3066 the values malloc, realloc and free, respectively (except for Win32
3067 compilations). The same is provided for memory debugging code.
3068 OpenSSL already comes with functionality to find memory leaks, but
3069 this gives people a chance to debug other memory problems.
3071 With these changes, a new set of functions and macros have appeared:
3073 CRYPTO_set_mem_debug_functions() [F]
3074 CRYPTO_get_mem_debug_functions() [F]
3075 CRYPTO_dbg_set_options() [F]
3076 CRYPTO_dbg_get_options() [F]
3077 CRYPTO_malloc_debug_init() [M]
3079 The memory debug functions are NULL by default, unless the library
3080 is compiled with CRYPTO_MDEBUG or friends is defined. If someone
3081 wants to debug memory anyway, CRYPTO_malloc_debug_init() (which
3082 gives the standard debugging functions that come with OpenSSL) or
3083 CRYPTO_set_mem_debug_functions() (tells OpenSSL to use functions
3084 provided by the library user) must be used. When the standard
3085 debugging functions are used, CRYPTO_dbg_set_options can be used to
3086 request additional information:
3087 CRYPTO_dbg_set_options(V_CYRPTO_MDEBUG_xxx) corresponds to setting
3088 the CRYPTO_MDEBUG_xxx macro when compiling the library.
3090 Also, things like CRYPTO_set_mem_functions will always give the
3091 expected result (the new set of functions is used for allocation
3092 and deallocation) at all times, regardless of platform and compiler
3095 To finish it up, some functions that were never use in any other
3096 way than through macros have a new API and new semantic:
3099 CRYPTO_dbg_realloc()
3102 All macros of value have retained their old syntax.
3103 [Richard Levitte and Bodo Moeller]
3105 *) Some S/MIME fixes. The OID for SMIMECapabilities was wrong, the
3106 ordering of SMIMECapabilities wasn't in "strength order" and there
3107 was a missing NULL in the AlgorithmIdentifier for the SHA1 signature
3111 *) Some ASN1 types with illegal zero length encoding (INTEGER,
3112 ENUMERATED and OBJECT IDENTIFIER) choked the ASN1 routines.
3113 [Frans Heymans <fheymans@isaserver.be>, modified by Steve Henson]
3115 *) Merge in my S/MIME library for OpenSSL. This provides a simple
3116 S/MIME API on top of the PKCS#7 code, a MIME parser (with enough
3117 functionality to handle multipart/signed properly) and a utility
3118 called 'smime' to call all this stuff. This is based on code I
3119 originally wrote for Celo who have kindly allowed it to be
3120 included in OpenSSL.
3123 *) Add variants des_set_key_checked and des_set_key_unchecked of
3124 des_set_key (aka des_key_sched). Global variable des_check_key
3125 decides which of these is called by des_set_key; this way
3126 des_check_key behaves as it always did, but applications and
3127 the library itself, which was buggy for des_check_key == 1,
3128 have a cleaner way to pick the version they need.
3131 *) New function PKCS12_newpass() which changes the password of a
3135 *) Modify X509_TRUST and X509_PURPOSE so it also uses a static and
3136 dynamic mix. In both cases the ids can be used as an index into the
3137 table. Also modified the X509_TRUST_add() and X509_PURPOSE_add()
3138 functions so they accept a list of the field values and the
3139 application doesn't need to directly manipulate the X509_TRUST
3143 *) Modify the ASN1_STRING_TABLE stuff so it also uses bsearch and doesn't
3147 *) Modify the way the V3 extension code looks up extensions. This now
3148 works in a similar way to the object code: we have some "standard"
3149 extensions in a static table which is searched with OBJ_bsearch()
3150 and the application can add dynamic ones if needed. The file
3151 crypto/x509v3/ext_dat.h now has the info: this file needs to be
3152 updated whenever a new extension is added to the core code and kept
3153 in ext_nid order. There is a simple program 'tabtest.c' which checks
3154 this. New extensions are not added too often so this file can readily
3155 be maintained manually.
3157 There are two big advantages in doing things this way. The extensions
3158 can be looked up immediately and no longer need to be "added" using
3159 X509V3_add_standard_extensions(): this function now does nothing.
3160 [Side note: I get *lots* of email saying the extension code doesn't
3161 work because people forget to call this function]
3162 Also no dynamic allocation is done unless new extensions are added:
3163 so if we don't add custom extensions there is no need to call
3164 X509V3_EXT_cleanup().
3167 *) Modify enc utility's salting as follows: make salting the default. Add a
3168 magic header, so unsalted files fail gracefully instead of just decrypting
3169 to garbage. This is because not salting is a big security hole, so people
3170 should be discouraged from doing it.
3173 *) Fixes and enhancements to the 'x509' utility. It allowed a message
3174 digest to be passed on the command line but it only used this
3175 parameter when signing a certificate. Modified so all relevant
3176 operations are affected by the digest parameter including the
3177 -fingerprint and -x509toreq options. Also -x509toreq choked if a
3178 DSA key was used because it didn't fix the digest.
3181 *) Initial certificate chain verify code. Currently tests the untrusted
3182 certificates for consistency with the verify purpose (which is set
3183 when the X509_STORE_CTX structure is set up) and checks the pathlength.
3185 There is a NO_CHAIN_VERIFY compilation option to keep the old behaviour:
3186 this is because it will reject chains with invalid extensions whereas
3187 every previous version of OpenSSL and SSLeay made no checks at all.
3189 Trust code: checks the root CA for the relevant trust settings. Trust
3190 settings have an initial value consistent with the verify purpose: e.g.
3191 if the verify purpose is for SSL client use it expects the CA to be
3192 trusted for SSL client use. However the default value can be changed to
3193 permit custom trust settings: one example of this would be to only trust
3194 certificates from a specific "secure" set of CAs.
3196 Also added X509_STORE_CTX_new() and X509_STORE_CTX_free() functions
3197 which should be used for version portability: especially since the
3198 verify structure is likely to change more often now.
3200 SSL integration. Add purpose and trust to SSL_CTX and SSL and functions
3201 to set them. If not set then assume SSL clients will verify SSL servers
3204 Two new options to the verify program: -untrusted allows a set of
3205 untrusted certificates to be passed in and -purpose which sets the
3206 intended purpose of the certificate. If a purpose is set then the
3207 new chain verify code is used to check extension consistency.
3210 *) Support for the authority information access extension.
3213 *) Modify RSA and DSA PEM read routines to transparently handle
3214 PKCS#8 format private keys. New *_PUBKEY_* functions that handle
3215 public keys in a format compatible with certificate
3216 SubjectPublicKeyInfo structures. Unfortunately there were already
3217 functions called *_PublicKey_* which used various odd formats so
3218 these are retained for compatibility: however the DSA variants were
3219 never in a public release so they have been deleted. Changed dsa/rsa
3220 utilities to handle the new format: note no releases ever handled public
3221 keys so we should be OK.
3223 The primary motivation for this change is to avoid the same fiasco
3224 that dogs private keys: there are several incompatible private key
3225 formats some of which are standard and some OpenSSL specific and
3226 require various evil hacks to allow partial transparent handling and
3227 even then it doesn't work with DER formats. Given the option anything
3228 other than PKCS#8 should be dumped: but the other formats have to
3229 stay in the name of compatibility.
3231 With public keys and the benefit of hindsight one standard format
3232 is used which works with EVP_PKEY, RSA or DSA structures: though
3233 it clearly returns an error if you try to read the wrong kind of key.
3235 Added a -pubkey option to the 'x509' utility to output the public key.
3236 Also rename the EVP_PKEY_get_*() to EVP_PKEY_rget_*()
3237 (renamed to EVP_PKEY_get1_*() in the OpenSSL 0.9.5 release) and add
3238 EVP_PKEY_rset_*() functions (renamed to EVP_PKEY_set1_*())
3239 that do the same as the EVP_PKEY_assign_*() except they up the
3240 reference count of the added key (they don't "swallow" the
3244 *) Fixes to crypto/x509/by_file.c the code to read in certificates and
3245 CRLs would fail if the file contained no certificates or no CRLs:
3246 added a new function to read in both types and return the number
3247 read: this means that if none are read it will be an error. The
3248 DER versions of the certificate and CRL reader would always fail
3249 because it isn't possible to mix certificates and CRLs in DER format
3250 without choking one or the other routine. Changed this to just read
3251 a certificate: this is the best we can do. Also modified the code
3252 in apps/verify.c to take notice of return codes: it was previously
3253 attempting to read in certificates from NULL pointers and ignoring
3254 any errors: this is one reason why the cert and CRL reader seemed
3255 to work. It doesn't check return codes from the default certificate
3256 routines: these may well fail if the certificates aren't installed.
3259 *) Code to support otherName option in GeneralName.
3262 *) First update to verify code. Change the verify utility
3263 so it warns if it is passed a self signed certificate:
3264 for consistency with the normal behaviour. X509_verify
3265 has been modified to it will now verify a self signed
3266 certificate if *exactly* the same certificate appears
3267 in the store: it was previously impossible to trust a
3268 single self signed certificate. This means that:
3269 openssl verify ss.pem
3270 now gives a warning about a self signed certificate but
3271 openssl verify -CAfile ss.pem ss.pem
3275 *) For servers, store verify_result in SSL_SESSION data structure
3276 (and add it to external session representation).
3277 This is needed when client certificate verifications fails,
3278 but an application-provided verification callback (set by
3279 SSL_CTX_set_cert_verify_callback) allows accepting the session
3280 anyway (i.e. leaves x509_store_ctx->error != X509_V_OK
3281 but returns 1): When the session is reused, we have to set
3282 ssl->verify_result to the appropriate error code to avoid
3284 [Bodo Moeller, problem pointed out by Lutz Jaenicke]
3286 *) Fix a bug in the new PKCS#7 code: it didn't consider the
3287 case in PKCS7_dataInit() where the signed PKCS7 structure
3288 didn't contain any existing data because it was being created.
3289 [Po-Cheng Chen <pocheng@nst.com.tw>, slightly modified by Steve Henson]
3291 *) Add a salt to the key derivation routines in enc.c. This
3292 forms the first 8 bytes of the encrypted file. Also add a
3293 -S option to allow a salt to be input on the command line.
3296 *) New function X509_cmp(). Oddly enough there wasn't a function
3297 to compare two certificates. We do this by working out the SHA1
3298 hash and comparing that. X509_cmp() will be needed by the trust
3302 *) SSL_get1_session() is like SSL_get_session(), but increments
3303 the reference count in the SSL_SESSION returned.
3304 [Geoff Thorpe <geoff@eu.c2.net>]
3306 *) Fix for 'req': it was adding a null to request attributes.
3307 Also change the X509_LOOKUP and X509_INFO code to handle
3308 certificate auxiliary information.
3311 *) Add support for 40 and 64 bit RC2 and RC4 algorithms: document
3315 *) Add the possibility to add extra information to the memory leak
3316 detecting output, to form tracebacks, showing from where each
3317 allocation was originated: CRYPTO_push_info("constant string") adds
3318 the string plus current file name and line number to a per-thread
3319 stack, CRYPTO_pop_info() does the obvious, CRYPTO_remove_all_info()
3320 is like calling CYRPTO_pop_info() until the stack is empty.
3321 Also updated memory leak detection code to be multi-thread-safe.
3324 *) Add options -text and -noout to pkcs7 utility and delete the
3325 encryption options which never did anything. Update docs.
3328 *) Add options to some of the utilities to allow the pass phrase
3329 to be included on either the command line (not recommended on
3330 OSes like Unix) or read from the environment. Update the
3331 manpages and fix a few bugs.
3334 *) Add a few manpages for some of the openssl commands.
3337 *) Fix the -revoke option in ca. It was freeing up memory twice,
3338 leaking and not finding already revoked certificates.
3341 *) Extensive changes to support certificate auxiliary information.
3342 This involves the use of X509_CERT_AUX structure and X509_AUX
3343 functions. An X509_AUX function such as PEM_read_X509_AUX()
3344 can still read in a certificate file in the usual way but it
3345 will also read in any additional "auxiliary information". By
3346 doing things this way a fair degree of compatibility can be
3347 retained: existing certificates can have this information added
3348 using the new 'x509' options.
3350 Current auxiliary information includes an "alias" and some trust
3351 settings. The trust settings will ultimately be used in enhanced
3352 certificate chain verification routines: currently a certificate
3353 can only be trusted if it is self signed and then it is trusted
3357 *) Fix assembler for Alpha (tested only on DEC OSF not Linux or *BSD).
3358 The problem was that one of the replacement routines had not been working
3359 since SSLeay releases. For now the offending routine has been replaced
3360 with non-optimised assembler. Even so, this now gives around 95%
3361 performance improvement for 1024 bit RSA signs.
3364 *) Hack to fix PKCS#7 decryption when used with some unorthodox RC2
3365 handling. Most clients have the effective key size in bits equal to
3366 the key length in bits: so a 40 bit RC2 key uses a 40 bit (5 byte) key.
3367 A few however don't do this and instead use the size of the decrypted key
3368 to determine the RC2 key length and the AlgorithmIdentifier to determine
3369 the effective key length. In this case the effective key length can still
3370 be 40 bits but the key length can be 168 bits for example. This is fixed
3371 by manually forcing an RC2 key into the EVP_PKEY structure because the
3372 EVP code can't currently handle unusual RC2 key sizes: it always assumes
3373 the key length and effective key length are equal.
3376 *) Add a bunch of functions that should simplify the creation of
3377 X509_NAME structures. Now you should be able to do:
3378 X509_NAME_add_entry_by_txt(nm, "CN", MBSTRING_ASC, "Steve", -1, -1, 0);
3379 and have it automatically work out the correct field type and fill in
3380 the structures. The more adventurous can try:
3381 X509_NAME_add_entry_by_txt(nm, field, MBSTRING_UTF8, str, -1, -1, 0);
3382 and it will (hopefully) work out the correct multibyte encoding.
3385 *) Change the 'req' utility to use the new field handling and multibyte
3386 copy routines. Before the DN field creation was handled in an ad hoc
3387 way in req, ca, and x509 which was rather broken and didn't support
3388 BMPStrings or UTF8Strings. Since some software doesn't implement
3389 BMPStrings or UTF8Strings yet, they can be enabled using the config file
3390 using the dirstring_type option. See the new comment in the default
3391 openssl.cnf for more info.
3394 *) Make crypto/rand/md_rand.c more robust:
3395 - Assure unique random numbers after fork().
3396 - Make sure that concurrent threads access the global counter and
3397 md serializably so that we never lose entropy in them
3398 or use exactly the same state in multiple threads.
3399 Access to the large state is not always serializable because
3400 the additional locking could be a performance killer, and
3401 md should be large enough anyway.
3404 *) New file apps/app_rand.c with commonly needed functionality
3405 for handling the random seed file.
3407 Use the random seed file in some applications that previously did not:
3409 dsaparam -genkey (which also ignored its '-rand' option),
3412 x509 (when signing).
3413 Except on systems with /dev/urandom, it is crucial to have a random
3414 seed file at least for key creation, DSA signing, and for DH exchanges;
3415 for RSA signatures we could do without one.
3417 gendh and gendsa (unlike genrsa) used to read only the first byte
3418 of each file listed in the '-rand' option. The function as previously
3419 found in genrsa is now in app_rand.c and is used by all programs
3420 that support '-rand'.
3423 *) In RAND_write_file, use mode 0600 for creating files;
3424 don't just chmod when it may be too late.
3427 *) Report an error from X509_STORE_load_locations
3428 when X509_LOOKUP_load_file or X509_LOOKUP_add_dir failed.
3431 *) New function ASN1_mbstring_copy() this copies a string in either
3432 ASCII, Unicode, Universal (4 bytes per character) or UTF8 format
3433 into an ASN1_STRING type. A mask of permissible types is passed
3434 and it chooses the "minimal" type to use or an error if not type
3438 *) Add function equivalents to the various macros in asn1.h. The old
3439 macros are retained with an M_ prefix. Code inside the library can
3440 use the M_ macros. External code (including the openssl utility)
3441 should *NOT* in order to be "shared library friendly".
3444 *) Add various functions that can check a certificate's extensions
3445 to see if it usable for various purposes such as SSL client,
3446 server or S/MIME and CAs of these types. This is currently
3447 VERY EXPERIMENTAL but will ultimately be used for certificate chain
3448 verification. Also added a -purpose flag to x509 utility to
3449 print out all the purposes.
3452 *) Add a CRYPTO_EX_DATA to X509 certificate structure and associated
3456 *) New X509V3_{X509,CRL,REVOKED}_get_d2i() functions. These will search
3457 for, obtain and decode and extension and obtain its critical flag.
3458 This allows all the necessary extension code to be handled in a
3459 single function call.
3462 *) RC4 tune-up featuring 30-40% performance improvement on most RISC
3463 platforms. See crypto/rc4/rc4_enc.c for further details.
3466 *) New -noout option to asn1parse. This causes no output to be produced
3467 its main use is when combined with -strparse and -out to extract data
3468 from a file (which may not be in ASN.1 format).
3471 *) Fix for pkcs12 program. It was hashing an invalid certificate pointer
3472 when producing the local key id.
3473 [Richard Levitte <levitte@stacken.kth.se>]
3475 *) New option -dhparam in s_server. This allows a DH parameter file to be
3476 stated explicitly. If it is not stated then it tries the first server
3477 certificate file. The previous behaviour hard coded the filename
3481 *) Add -pubin and -pubout options to the rsa and dsa commands. These allow
3482 a public key to be input or output. For example:
3483 openssl rsa -in key.pem -pubout -out pubkey.pem
3484 Also added necessary DSA public key functions to handle this.
3487 *) Fix so PKCS7_dataVerify() doesn't crash if no certificates are contained
3488 in the message. This was handled by allowing
3489 X509_find_by_issuer_and_serial() to tolerate a NULL passed to it.
3490 [Steve Henson, reported by Sampo Kellomaki <sampo@mail.neuronio.pt>]
3492 *) Fix for bug in d2i_ASN1_bytes(): other ASN1 functions add an extra null
3493 to the end of the strings whereas this didn't. This would cause problems
3494 if strings read with d2i_ASN1_bytes() were later modified.
3495 [Steve Henson, reported by Arne Ansper <arne@ats.cyber.ee>]
3497 *) Fix for base64 decode bug. When a base64 bio reads only one line of
3498 data and it contains EOF it will end up returning an error. This is
3499 caused by input 46 bytes long. The cause is due to the way base64
3500 BIOs find the start of base64 encoded data. They do this by trying a
3501 trial decode on each line until they find one that works. When they
3502 do a flag is set and it starts again knowing it can pass all the
3503 data directly through the decoder. Unfortunately it doesn't reset
3504 the context it uses. This means that if EOF is reached an attempt
3505 is made to pass two EOFs through the context and this causes the
3506 resulting error. This can also cause other problems as well. As is
3507 usual with these problems it takes *ages* to find and the fix is
3508 trivial: move one line.
3509 [Steve Henson, reported by ian@uns.ns.ac.yu (Ivan Nejgebauer) ]
3511 *) Ugly workaround to get s_client and s_server working under Windows. The
3512 old code wouldn't work because it needed to select() on sockets and the
3513 tty (for keypresses and to see if data could be written). Win32 only
3514 supports select() on sockets so we select() with a 1s timeout on the
3515 sockets and then see if any characters are waiting to be read, if none
3516 are present then we retry, we also assume we can always write data to
3517 the tty. This isn't nice because the code then blocks until we've
3518 received a complete line of data and it is effectively polling the
3519 keyboard at 1s intervals: however it's quite a bit better than not
3520 working at all :-) A dedicated Windows application might handle this
3521 with an event loop for example.
3524 *) Enhance RSA_METHOD structure. Now there are two extra methods, rsa_sign
3525 and rsa_verify. When the RSA_FLAGS_SIGN_VER option is set these functions
3526 will be called when RSA_sign() and RSA_verify() are used. This is useful
3527 if rsa_pub_dec() and rsa_priv_enc() equivalents are not available.
3528 For this to work properly RSA_public_decrypt() and RSA_private_encrypt()
3529 should *not* be used: RSA_sign() and RSA_verify() must be used instead.
3530 This necessitated the support of an extra signature type NID_md5_sha1
3531 for SSL signatures and modifications to the SSL library to use it instead
3532 of calling RSA_public_decrypt() and RSA_private_encrypt().
3535 *) Add new -verify -CAfile and -CApath options to the crl program, these
3536 will lookup a CRL issuers certificate and verify the signature in a
3537 similar way to the verify program. Tidy up the crl program so it
3538 no longer accesses structures directly. Make the ASN1 CRL parsing a bit
3539 less strict. It will now permit CRL extensions even if it is not
3540 a V2 CRL: this will allow it to tolerate some broken CRLs.
3543 *) Initialize all non-automatic variables each time one of the openssl
3544 sub-programs is started (this is necessary as they may be started
3545 multiple times from the "OpenSSL>" prompt).
3546 [Lennart Bang, Bodo Moeller]
3548 *) Preliminary compilation option RSA_NULL which disables RSA crypto without
3549 removing all other RSA functionality (this is what NO_RSA does). This
3550 is so (for example) those in the US can disable those operations covered
3551 by the RSA patent while allowing storage and parsing of RSA keys and RSA
3555 *) Non-copying interface to BIO pairs.
3556 (still largely untested)
3559 *) New function ANS1_tag2str() to convert an ASN1 tag to a descriptive
3560 ASCII string. This was handled independently in various places before.
3563 *) New functions UTF8_getc() and UTF8_putc() that parse and generate
3564 UTF8 strings a character at a time.
3567 *) Use client_version from client hello to select the protocol
3568 (s23_srvr.c) and for RSA client key exchange verification
3569 (s3_srvr.c), as required by the SSL 3.0/TLS 1.0 specifications.
3572 *) Add various utility functions to handle SPKACs, these were previously
3573 handled by poking round in the structure internals. Added new function
3574 NETSCAPE_SPKI_print() to print out SPKAC and a new utility 'spkac' to
3575 print, verify and generate SPKACs. Based on an original idea from
3576 Massimiliano Pala <madwolf@comune.modena.it> but extensively modified.
3579 *) RIPEMD160 is operational on all platforms and is back in 'make test'.
3582 *) Allow the config file extension section to be overwritten on the
3583 command line. Based on an original idea from Massimiliano Pala
3584 <madwolf@comune.modena.it>. The new option is called -extensions
3585 and can be applied to ca, req and x509. Also -reqexts to override
3586 the request extensions in req and -crlexts to override the crl extensions
3590 *) Add new feature to the SPKAC handling in ca. Now you can include
3591 the same field multiple times by preceding it by "XXXX." for example:
3594 this is the same syntax as used in the req config file.
3597 *) Allow certificate extensions to be added to certificate requests. These
3598 are specified in a 'req_extensions' option of the req section of the
3599 config file. They can be printed out with the -text option to req but
3600 are otherwise ignored at present.
3603 *) Fix a horrible bug in enc_read() in crypto/evp/bio_enc.c: if the first
3604 data read consists of only the final block it would not decrypted because
3605 EVP_CipherUpdate() would correctly report zero bytes had been decrypted.
3606 A misplaced 'break' also meant the decrypted final block might not be
3607 copied until the next read.
3610 *) Initial support for DH_METHOD. Again based on RSA_METHOD. Also added
3611 a few extra parameters to the DH structure: these will be useful if
3612 for example we want the value of 'q' or implement X9.42 DH.
3615 *) Initial support for DSA_METHOD. This is based on the RSA_METHOD and
3616 provides hooks that allow the default DSA functions or functions on a
3617 "per key" basis to be replaced. This allows hardware acceleration and
3618 hardware key storage to be handled without major modification to the
3619 library. Also added low level modexp hooks and CRYPTO_EX structure and
3620 associated functions.
3623 *) Add a new flag to memory BIOs, BIO_FLAG_MEM_RDONLY. This marks the BIO
3624 as "read only": it can't be written to and the buffer it points to will
3625 not be freed. Reading from a read only BIO is much more efficient than
3626 a normal memory BIO. This was added because there are several times when
3627 an area of memory needs to be read from a BIO. The previous method was
3628 to create a memory BIO and write the data to it, this results in two
3629 copies of the data and an O(n^2) reading algorithm. There is a new
3630 function BIO_new_mem_buf() which creates a read only memory BIO from
3631 an area of memory. Also modified the PKCS#7 routines to use read only
3635 *) Bugfix: ssl23_get_client_hello did not work properly when called in
3636 state SSL23_ST_SR_CLNT_HELLO_B, i.e. when the first 7 bytes of
3637 a SSLv2-compatible client hello for SSLv3 or TLSv1 could be read,
3638 but a retry condition occured while trying to read the rest.
3641 *) The PKCS7_ENC_CONTENT_new() function was setting the content type as
3642 NID_pkcs7_encrypted by default: this was wrong since this should almost
3643 always be NID_pkcs7_data. Also modified the PKCS7_set_type() to handle
3644 the encrypted data type: this is a more sensible place to put it and it
projects / openssl.git / blob