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
16 [Ulf Möller, reported by Jim Ellis <jim.ellis@cavium.com>]
18 *) Make it possible to unload ranges of ERR strings with a new
19 "ERR_unload_strings" function.
22 *) Rabin-Miller test analyses assume uniformly distributed witnesses,
23 so use BN_pseudo_rand_range() instead of using BN_pseudo_rand()
24 followed by modular reduction.
25 [Bodo Moeller; pointed out by Adam Young <AYoung1@NCSUS.JNJ.COM>]
27 *) Add BN_pseudo_rand_range() with obvious functionality: BN_rand_range()
28 requivalent based on BN_pseudo_rand() instead of BN_rand().
31 +) Add a copy() function to EVP_MD.
34 +) Make EVP_MD routines take a context pointer instead of just the
38 +) Add flags to EVP_MD and EVP_MD_CTX. EVP_MD_FLAG_ONESHOT indicates
39 that the digest can only process a single chunk of data
40 (typically because it is provided by a piece of
41 hardware). EVP_MD_CTX_FLAG_ONESHOT indicates that the application
42 is only going to provide a single chunk of data, and hence the
43 framework needn't accumulate the data for oneshot drivers.
46 +) As with "ERR", make it possible to replace the underlying "ex_data"
47 functions. This change also alters the storage and management of global
48 ex_data state - it's now all inside ex_data.c and all "class" code (eg.
49 RSA, BIO, SSL_CTX, etc) no longer stores its own STACKS and per-class
50 index counters. The API functions that use this state have been changed
51 to take a "class_index" rather than pointers to the class's local STACK
52 and counter, and there is now an API function to dynamically create new
53 classes. This centralisation allows us to (a) plug a lot of the
54 thread-safety problems that existed, and (b) makes it possible to clean
55 up all allocated state using "CRYPTO_cleanup_all_ex_data()". W.r.t. (b)
56 such data would previously have always leaked in application code and
57 workarounds were in place to make the memory debugging turn a blind eye
58 to it. Application code that doesn't use this new function will still
59 leak as before, but their memory debugging output will announce it now
60 rather than letting it slide.
62 Besides the addition of CRYPTO_cleanup_all_ex_data(), another API change
63 induced by the "ex_data" overhaul is that X509_STORE_CTX_init() now
64 has a return value to indicate success or failure.
67 +) Make it possible to replace the underlying "ERR" functions such that the
68 global state (2 LHASH tables and 2 locks) is only used by the "default"
69 implementation. This change also adds two functions to "get" and "set"
70 the implementation prior to it being automatically set the first time
71 any other ERR function takes place. Ie. an application can call "get",
72 pass the return value to a module it has just loaded, and that module
73 can call its own "set" function using that value. This means the
74 module's "ERR" operations will use (and modify) the error state in the
75 application and not in its own statically linked copy of OpenSSL code.
78 +) Give DH, DSA, and RSA types their own "**_up()" function to increment
79 reference counts. This performs normal REF_PRINT/REF_CHECK macros on
80 the operation, and provides a more encapsulated way for external code
81 (crypto/evp/ and ssl/) to do this. Also changed the evp and ssl code
82 to use these functions rather than manually incrementing the counts.
85 *) s3_srvr.c: allow sending of large client certificate lists (> 16 kB).
86 This function was broken, as the check for a new client hello message
87 to handle SGC did not allow these large messages.
88 (Tracked down by "Douglas E. Engert" <deengert@anl.gov>.)
91 *) Add alert descriptions for TLSv1 to SSL_alert_desc_string[_long]().
94 +) Add EVP test program.
97 +) Add symmetric cipher support to ENGINE. Expect the API to change!
100 +) New CRL functions: X509_CRL_set_version(), X509_CRL_set_issuer_name()
101 X509_CRL_set_lastUpdate(), X509_CRL_set_nextUpdate(), X509_CRL_sort(),
102 X509_REVOKED_set_serialNumber(), and X509_REVOKED_set_revocationDate().
103 These allow a CRL to be built without having to access X509_CRL fields
104 directly. Modify 'ca' application to use new functions.
107 *) Fix buggy behaviour of BIO_get_num_renegotiates() and BIO_ctrl()
108 for BIO_C_GET_WRITE_BUF_SIZE ("Stephen Hinton" <shinton@netopia.com>).
111 *) Rework the configuration and shared library support for Tru64 Unix.
112 The configuration part makes use of modern compiler features and
113 still retains old compiler behavior for those that run older versions
114 of the OS. The shared library support part includes a variant that
115 uses the RPATH feature, and is available through the speciel
116 configuration target "alpha-cc-rpath", which will never be selected
118 [Tim Mooney <mooney@dogbert.cc.ndsu.NoDak.edu> via Richard Levitte]
120 *) In ssl3_get_key_exchange (ssl/s3_clnt.c), call ssl3_get_message()
121 with the same message size as in ssl3_get_certificate_request().
122 Otherwise, if no ServerKeyExchange message occurs, CertificateRequest
123 messages might inadvertently be reject as too long.
124 [Petr Lampa <lampa@fee.vutbr.cz>]
126 +) Move SSL_OP_TLS_ROLLBACK_BUG out of the SSL_OP_ALL list of recommended
127 bug workarounds. Rollback attack detection is a security feature.
128 The problem will only arise on OpenSSL servers when TLSv1 is not
129 available (sslv3_server_method() or SSL_OP_NO_TLSv1).
130 Software authors not wanting to support TLSv1 will have special reasons
131 for their choice and can explicitly enable this option.
132 [Bodo Moeller, Lutz Jaenicke]
134 +) Rationalise EVP so it can be extended: don't include a union of
135 cipher/digest structures, add init/cleanup functions. This also reduces
136 the number of header dependencies.
139 +) Make DES key schedule conform to the usual scheme, as well as
140 correcting its structure. This means that calls to DES functions
141 now have to pass a pointer to a des_key_schedule instead of a
142 plain des_key_schedule (which was actually always a pointer
146 +) Enhanced support for IA-64 Unix platforms (well, Linux and HP-UX).
149 *) Modified SSL library such that the verify_callback that has been set
150 specificly for an SSL object with SSL_set_verify() is actually being
151 used. Before the change, a verify_callback set with this function was
152 ignored and the verify_callback() set in the SSL_CTX at the time of
153 the call was used. New function X509_STORE_CTX_set_verify_cb() introduced
154 to allow the necessary settings.
157 +) Initial reduction of linker bloat: the use of some functions, such as
158 PEM causes large amounts of unused functions to be linked in due to
159 poor organisation. For example pem_all.c contains every PEM function
160 which has a knock on effect of linking in large amounts of (unused)
161 ASN1 code. Grouping together similar functions and splitting unrelated
162 functions prevents this.
165 *) Initialize static variable in crypto/dsa/dsa_lib.c and crypto/dh/dh_lib.c
166 explicitely to NULL, as at least on Solaris 8 this seems not always to be
167 done automatically (in contradiction to the requirements of the C
168 standard). This made problems when used from OpenSSH.
171 *) In OpenSSL 0.9.6a and 0.9.6b, crypto/dh/dh_key.c ignored
172 dh->length and always used
174 BN_rand_range(priv_key, dh->p).
176 BN_rand_range() is not necessary for Diffie-Hellman, and this
177 specific range makes Diffie-Hellman unnecessarily inefficient if
178 dh->length (recommended exponent length) is much smaller than the
179 length of dh->p. We could use BN_rand_range() if the order of
180 the subgroup was stored in the DH structure, but we only have
185 BN_rand(priv_key, l, ...)
187 where 'l' is dh->length if this is defined, or BN_num_bits(dh->p)-1
193 RSA_eay_public_encrypt
194 RSA_eay_private_decrypt
195 RSA_eay_private_encrypt (signing)
196 RSA_eay_public_decrypt (signature verification)
198 (default implementations for RSA_public_encrypt,
199 RSA_private_decrypt, RSA_private_encrypt, RSA_public_decrypt),
200 always reject numbers >= n.
203 *) In crypto/rand/md_rand.c, use a new short-time lock CRYPTO_LOCK_RAND2
204 to synchronize access to 'locking_thread'. This is necessary on
205 systems where access to 'locking_thread' (an 'unsigned long'
206 variable) is not atomic.
209 *) In crypto/rand/md_rand.c, set 'locking_thread' to current thread's ID
210 *before* setting the 'crypto_lock_rand' flag. The previous code had
211 a race condition if 0 is a valid thread ID.
212 [Travis Vitek <vitek@roguewave.com>]
214 +) Cleanup of EVP macros.
217 +) Change historical references to {NID,SN,LN}_des_ede and ede3 to add the
221 +) Add initial OCSP responder support to ocsp application. The
222 revocation information is handled using the text based index
223 use by the ca application. The responder can either handle
224 requests generated internally, supplied in files (for example
225 via a CGI script) or using an internal minimal server.
228 +) Add configuration choices to get zlib compression for TLS.
231 +) Changes to Kerberos SSL for RFC 2712 compliance:
232 1. Implemented real KerberosWrapper, instead of just using
233 KRB5 AP_REQ message. [Thanks to Simon Wilkinson <sxw@sxw.org.uk>]
234 2. Implemented optional authenticator field of KerberosWrapper.
236 Added openssl-style ASN.1 macros for Kerberos ticket, ap_req,
237 and authenticator structs; see crypto/krb5/.
239 Generalized Kerberos calls to support multiple Kerberos libraries.
240 [Vern Staats <staatsvr@asc.hpc.mil>,
241 Jeffrey Altman <jaltman@columbia.edu>
244 +) Cause 'openssl speed' to use fully hard-coded DSA keys as it
245 already does with RSA. testdsa.h now has 'priv_key/pub_key'
246 values for each of the key sizes rather than having just
247 parameters (and 'speed' generating keys each time).
250 -) OpenSSL 0.9.6b released [9 July 2001]
252 *) Change ssleay_rand_bytes (crypto/rand/md_rand.c)
253 to avoid a SSLeay/OpenSSL PRNG weakness pointed out by
254 Markku-Juhani O. Saarinen <markku-juhani.saarinen@nokia.com>:
255 PRNG state recovery was possible based on the output of
256 one PRNG request appropriately sized to gain knowledge on
257 'md' followed by enough consecutive 1-byte PRNG requests
258 to traverse all of 'state'.
260 1. When updating 'md_local' (the current thread's copy of 'md')
261 during PRNG output generation, hash all of the previous
262 'md_local' value, not just the half used for PRNG output.
264 2. Make the number of bytes from 'state' included into the hash
265 independent from the number of PRNG bytes requested.
267 The first measure alone would be sufficient to avoid
268 Markku-Juhani's attack. (Actually it had never occurred
269 to me that the half of 'md_local' used for chaining was the
270 half from which PRNG output bytes were taken -- I had always
271 assumed that the secret half would be used.) The second
272 measure makes sure that additional data from 'state' is never
273 mixed into 'md_local' in small portions; this heuristically
274 further strengthens the PRNG.
277 +) Speed up EVP routines.
280 type 8 bytes 64 bytes 256 bytes 1024 bytes 8192 bytes
281 des-cbc 4408.85k 5560.51k 5778.46k 5862.20k 5825.16k
282 des-cbc 4389.55k 5571.17k 5792.23k 5846.91k 5832.11k
283 des-cbc 4394.32k 5575.92k 5807.44k 5848.37k 5841.30k
285 des-cbc 3482.66k 5069.49k 5496.39k 5614.16k 5639.28k
286 des-cbc 3480.74k 5068.76k 5510.34k 5609.87k 5635.52k
287 des-cbc 3483.72k 5067.62k 5504.60k 5708.01k 5724.80k
290 des-cbc 4660.16k 5650.19k 5807.19k 5827.13k 5783.32k
292 des-cbc 3624.96k 5258.21k 5530.91k 5624.30k 5628.26k
295 *) Fix crypto/bn/asm/mips3.s.
298 *) When only the key is given to "enc", the IV is undefined. Print out
299 an error message in this case.
302 +) Added the OS2-EMX target.
303 ["Brian Havard" <brianh@kheldar.apana.org.au> and Richard Levitte]
305 +) Rewrite apps to use NCONF routines instead of the old CONF. New functions
306 to support NCONF routines in extension code. New function CONF_set_nconf()
307 to allow functions which take an NCONF to also handle the old LHASH
308 structure: this means that the old CONF compatible routines can be
309 retained (in particular wrt extensions) without having to duplicate the
310 code. New function X509V3_add_ext_nconf_sk to add extensions to a stack.
313 *) Handle special case when X509_NAME is empty in X509 printing routines.
316 *) In dsa_do_verify (crypto/dsa/dsa_ossl.c), verify that r and s are
317 positive and less than q.
320 +) Enhance the general user interface with mechanisms for inner control
321 and with pssibilities to have yes/no kind of prompts.
324 +) Change all calls to low level digest routines in the library and
325 applications to use EVP. Add missing calls to HMAC_cleanup() and
326 don't assume HMAC_CTX can be copied using memcpy().
327 [Verdon Walker <VWalker@novell.com>, Steve Henson]
329 +) Add the possibility to control engines through control names but with
330 arbitrary arguments instead of just a string.
331 Change the key loaders to take a UI_METHOD instead of a callback
332 function pointer. NOTE: this breaks binary compatibility with earlier
333 versions of OpenSSL [engine].
334 Addapt the nCipher code for these new conditions and add a card insertion
338 +) Enhance the general user interface with mechanisms to better support
339 dialog box interfaces, application-defined prompts, the possibility
340 to use defaults (for example default passwords from somewhere else)
341 and interrupts/cancelations.
344 *) Don't change *pointer in CRYPTO_add_lock() is add_lock_callback is
345 used: it isn't thread safe and the add_lock_callback should handle
347 [Paul Rose <Paul.Rose@bridge.com>]
349 *) Verify that incoming data obeys the block size in
350 ssl3_enc (ssl/s3_enc.c) and tls1_enc (ssl/t1_enc.c).
353 +) Tidy up PKCS#12 attribute handling. Add support for the CSP name
354 attribute in PKCS#12 files, add new -CSP option to pkcs12 utility.
358 [Ulf Möller, Bodo Möller]
360 *) The countermeasure against Bleichbacher's attack on PKCS #1 v1.5
361 RSA encryption was accidentily removed in s3_srvr.c in OpenSSL 0.9.5
362 when fixing the server behaviour for backwards-compatible 'client
363 hello' messages. (Note that the attack is impractical against
364 SSL 3.0 and TLS 1.0 anyway because length and version checking
365 means that the probability of guessing a valid ciphertext is
366 around 2^-40; see section 5 in Bleichenbacher's CRYPTO '98
369 Before 0.9.5, the countermeasure (hide the error by generating a
370 random 'decryption result') did not work properly because
371 ERR_clear_error() was missing, meaning that SSL_get_error() would
372 detect the supposedly ignored error.
374 Both problems are now fixed.
377 *) In crypto/bio/bf_buff.c, increase DEFAULT_BUFFER_SIZE to 4096
378 (previously it was 1024).
381 +) Fix a memory leak in 'sk_dup()' in the case reallocation fails. (Also
382 tidy up some unecessarily weird code in 'sk_new()').
383 [Geoff, reported by Diego Tartara <dtartara@novamens.com>]
385 +) Change the key loading routines for ENGINEs to use the same kind
386 callback (pem_password_cb) as all other routines that need this
390 *) Fix for compatibility mode trust settings: ignore trust settings
391 unless some valid trust or reject settings are present.
394 *) Fix for blowfish EVP: its a variable length cipher.
397 +) Increase ENTROPY_NEEDED to 32 bytes, as Rijndael can operate with
398 256 bit (=32 byte) keys. Of course seeding with more entropy bytes
399 than this minimum value is recommended.
402 +) New random seeder for OpenVMS, using the system process statistics
403 that are easily reachable.
406 +) Windows apparently can't transparently handle global
407 variables defined in DLLs. Initialisations such as:
409 const ASN1_ITEM *it = &ASN1_INTEGER_it;
411 wont compile. This is used by the any applications that need to
412 delcare their own ASN1 modules. This was fixed by adding the option
413 EXPORT_VAR_AS_FN to all Win32 platforms, although this isn't strictly
414 needed for static libraries under Win32.
417 +) New functions X509_PURPOSE_set() and X509_TRUST_set() to handle
418 setting of purpose and trust fields. New X509_STORE trust and
419 purpose functions and tidy up setting in other SSL functions.
422 +) Add copies of X509_STORE_CTX fields and callbacks to X509_STORE
423 structure. These are inherited by X509_STORE_CTX when it is
424 initialised. This allows various defaults to be set in the
425 X509_STORE structure (such as flags for CRL checking and custom
426 purpose or trust settings) for functions which only use X509_STORE_CTX
427 internally such as S/MIME.
429 Modify X509_STORE_CTX_purpose_inherit() so it only sets purposes and
430 trust settings if they are not set in X509_STORE. This allows X509_STORE
431 purposes and trust (in S/MIME for example) to override any set by default.
433 Add command line options for CRL checking to smime, s_client and s_server
437 +) Initial CRL based revocation checking. If the CRL checking flag(s)
438 are set then the CRL is looked up in the X509_STORE structure and
439 its validity and signature checked, then if the certificate is found
440 in the CRL the verify fails with a revoked error.
442 Various new CRL related callbacks added to X509_STORE_CTX structure.
444 Command line options added to 'verify' application to support this.
446 This needs some additional work, such as being able to handle multiple
447 CRLs with different times, extension based lookup (rather than just
448 by subject name) and ultimately more complete V2 CRL extension
452 +) Add a general user interface API (crypto/ui/). This is designed
453 to replace things like des_read_password and friends (backward
454 compatibility functions using this new API are provided).
455 The purpose is to remove prompting functions from the DES code
456 section as well as provide for prompting through dialog boxes in
457 a window system and the like.
460 *) Fix various bugs related to DSA S/MIME verification. Handle missing
461 parameters in DSA public key structures and return an error in the
462 DSA routines if parameters are absent.
465 *) In versions up to 0.9.6, RAND_file_name() resorted to file ".rnd"
466 in the current directory if neither $RANDFILE nor $HOME was set.
467 RAND_file_name() in 0.9.6a returned NULL in this case. This has
468 caused some confusion to Windows users who haven't defined $HOME.
469 Thus RAND_file_name() is changed again: e_os.h can define a
470 DEFAULT_HOME, which will be used if $HOME is not set.
471 For Windows, we use "C:"; on other platforms, we still require
472 environment variables.
474 +) Add "ex_data" support to ENGINE so implementations can add state at a
475 per-structure level rather than having to store it globally.
478 +) Make it possible for ENGINE structures to be copied when retrieved by
479 ENGINE_by_id() if the ENGINE specifies a new flag: ENGINE_FLAGS_BY_ID_COPY.
480 This causes the "original" ENGINE structure to act like a template,
481 analogous to the RSA vs. RSA_METHOD type of separation. Because of this
482 operational state can be localised to each ENGINE structure, despite the
483 fact they all share the same "methods". New ENGINE structures returned in
484 this case have no functional references and the return value is the single
485 structural reference. This matches the single structural reference returned
486 by ENGINE_by_id() normally, when it is incremented on the pre-existing
490 +) Fix ASN1 decoder when decoding type ANY and V_ASN1_OTHER: since this
491 needs to match any other type at all we need to manually clear the
495 +) Changes to the "openssl engine" utility to include;
496 - verbosity levels ('-v', '-vv', and '-vvv') that provide information
497 about an ENGINE's available control commands.
498 - executing control commands from command line arguments using the
499 '-pre' and '-post' switches. '-post' is only used if '-t' is
500 specified and the ENGINE is successfully initialised. The syntax for
501 the individual commands are colon-separated, for example;
502 openssl engine chil -pre FORK_CHECK:0 -pre SO_PATH:/lib/test.so
505 +) New dynamic control command support for ENGINEs. ENGINEs can now
506 declare their own commands (numbers), names (strings), descriptions,
507 and input types for run-time discovery by calling applications. A
508 subset of these commands are implicitly classed as "executable"
509 depending on their input type, and only these can be invoked through
510 the new string-based API function ENGINE_ctrl_cmd_string(). (Eg. this
511 can be based on user input, config files, etc). The distinction is
512 that "executable" commands cannot return anything other than a boolean
513 result and can only support numeric or string input, whereas some
514 discoverable commands may only be for direct use through
515 ENGINE_ctrl(), eg. supporting the exchange of binary data, function
516 pointers, or other custom uses. The "executable" commands are to
517 support parameterisations of ENGINE behaviour that can be
518 unambiguously defined by ENGINEs and used consistently across any
519 OpenSSL-based application. Commands have been added to all the
520 existing hardware-supporting ENGINEs, noticeably "SO_PATH" to allow
521 control over shared-library paths without source code alterations.
524 +) Changed all ENGINE implementations to dynamically allocate their
525 ENGINEs rather than declaring them statically. Apart from this being
526 necessary with the removal of the ENGINE_FLAGS_MALLOCED distinction,
527 this also allows the implementations to compile without using the
528 internal engine_int.h header.
531 +) Minor adjustment to "rand" code. RAND_get_rand_method() now returns a
532 'const' value. Any code that should be able to modify a RAND_METHOD
533 should already have non-const pointers to it (ie. they should only
534 modify their own ones).
537 +) Made a variety of little tweaks to the ENGINE code.
538 - "atalla" and "ubsec" string definitions were moved from header files
539 to C code. "nuron" string definitions were placed in variables
540 rather than hard-coded - allowing parameterisation of these values
541 later on via ctrl() commands.
542 - Removed unused "#if 0"'d code.
543 - Fixed engine list iteration code so it uses ENGINE_free() to release
544 structural references.
545 - Constified the RAND_METHOD element of ENGINE structures.
546 - Constified various get/set functions as appropriate and added
547 missing functions (including a catch-all ENGINE_cpy that duplicates
548 all ENGINE values onto a new ENGINE except reference counts/state).
549 - Removed NULL parameter checks in get/set functions. Setting a method
550 or function to NULL is a way of cancelling out a previously set
551 value. Passing a NULL ENGINE parameter is just plain stupid anyway
552 and doesn't justify the extra error symbols and code.
553 - Deprecate the ENGINE_FLAGS_MALLOCED define and move the area for
554 flags from engine_int.h to engine.h.
555 - Changed prototypes for ENGINE handler functions (init(), finish(),
556 ctrl(), key-load functions, etc) to take an (ENGINE*) parameter.
559 *) Move 'if (!initialized) RAND_poll()' into regions protected by
560 CRYPTO_LOCK_RAND. This is not strictly necessary, but avoids
561 having multiple threads call RAND_poll() concurrently.
564 *) In crypto/rand/md_rand.c, replace 'add_do_not_lock' flag by a
565 combination of a flag and a thread ID variable.
566 Otherwise while one thread is in ssleay_rand_bytes (which sets the
567 flag), *other* threads can enter ssleay_add_bytes without obeying
568 the CRYPTO_LOCK_RAND lock (and may even illegaly release the lock
569 that they do not hold after the first thread unsets add_do_not_lock).
572 +) Implement binary inversion algorithm for BN_mod_inverse in addition
573 to the algorithm using long divison. The binary algorithm can be
574 used only if the modulus is odd. On 32-bit systems, it is faster
575 only for relatively small moduli (roughly 20-30% for 128-bit moduli,
576 roughly 5-15% for 256-bit moduli), so we use it only for moduli
577 up to 450 bits. In 64-bit environments, the binary algorithm
578 appears to be advantageous for much longer moduli; here we use it
579 for moduli up to 2048 bits.
582 *) Change bctest again: '-x' expressions are not available in all
586 -) OpenSSL 0.9.6a released [5 Apr 2001]
588 *) Fix a couple of memory leaks in PKCS7_dataDecode()
589 [Steve Henson, reported by Heyun Zheng <hzheng@atdsprint.com>]
591 *) Change Configure and Makefiles to provide EXE_EXT, which will contain
592 the default extension for executables, if any. Also, make the perl
593 scripts that use symlink() to test if it really exists and use "cp"
594 if it doesn't. All this made OpenSSL compilable and installable in
598 +) Rewrite CHOICE field setting in ASN1_item_ex_d2i(). The old code
599 could not support the combine flag in choice fields.
602 -) Fix for asn1_GetSequence() for indefinite length constructed data.
603 If SEQUENCE is length is indefinite just set c->slen to the total
604 amount of data available.
605 [Steve Henson, reported by shige@FreeBSD.org]
606 [This change does not apply to 0.9.7.]
608 *) Change bctest to avoid here-documents inside command substitution
609 (workaround for FreeBSD /bin/sh bug).
610 For compatibility with Ultrix, avoid shell functions (introduced
611 in the bctest version that searches along $PATH).
614 *) Rename 'des_encrypt' to 'des_encrypt1'. This avoids the clashes
615 with des_encrypt() defined on some operating systems, like Solaris
619 *) Check the result of RSA-CRT (see D. Boneh, R. DeMillo, R. Lipton:
620 On the Importance of Eliminating Errors in Cryptographic
621 Computations, J. Cryptology 14 (2001) 2, 101-119,
622 http://theory.stanford.edu/~dabo/papers/faults.ps.gz).
625 *) MIPS assembler BIGNUM division bug fix.
628 *) Disabled incorrect Alpha assembler code.
631 -) Fix PKCS#7 decode routines so they correctly update the length
632 after reading an EOC for the EXPLICIT tag.
634 [This change does not apply to 0.9.7.]
636 *) Fix bug in PKCS#12 key generation routines. This was triggered
637 if a 3DES key was generated with a 0 initial byte. Include
638 PKCS12_BROKEN_KEYGEN compilation option to retain the old
639 (but broken) behaviour.
642 *) Enhance bctest to search for a working bc along $PATH and print
644 [Tim Rice <tim@multitalents.net> via Richard Levitte]
646 +) Add a 'copy_extensions' option to the 'ca' utility. This copies
647 extensions from a certificate request to the certificate.
650 +) Allow multiple 'certopt' and 'nameopt' options to be separated
651 by commas. Add 'namopt' and 'certopt' options to the 'ca' config
652 file: this allows the display of the certificate about to be
653 signed to be customised, to allow certain fields to be included
654 or excluded and extension details. The old system didn't display
655 multicharacter strings properly, omitted fields not in the policy
656 and couldn't display additional details such as extensions.
659 *) Fix memory leaks in err.c: free err_data string if necessary;
660 don't write to the wrong index in ERR_set_error_data.
663 +) Function EC_POINTs_mul for simultaneous scalar multiplication
664 of an arbitrary number of elliptic curve points, optionally
665 including the generator defined for the EC_GROUP.
666 EC_POINT_mul is a simple wrapper function for the typical case
667 that the point list has just one item (besides the optional
671 +) First EC_METHODs for curves over GF(p):
673 EC_GFp_simple_method() uses the basic BN_mod_mul and BN_mod_sqr
674 operations and provides various method functions that can also
675 operate with faster implementations of modular arithmetic.
677 EC_GFp_mont_method() reuses most functions that are part of
678 EC_GFp_simple_method, but uses Montgomery arithmetic.
680 [Bodo Moeller; point addition and point doubling
681 implementation directly derived from source code provided by
682 Lenka Fibikova <fibikova@exp-math.uni-essen.de>]
684 +) Framework for elliptic curves (crypto/ec/ec.h, crypto/ec/ec_lcl.h,
687 Curves are EC_GROUP objects (with an optional group generator)
688 based on EC_METHODs that are built into the library.
690 Points are EC_POINT objects based on EC_GROUP objects.
692 Most of the framework would be able to handle curves over arbitrary
693 finite fields, but as there are no obvious types for fields other
694 than GF(p), some functions are limited to that for now.
697 +) Add the -HTTP option to s_server. It is similar to -WWW, but requires
698 that the file contains a complete HTTP response.
701 +) Add the ec directory to mkdef.pl and mkfiles.pl. In mkdef.pl
702 change the def and num file printf format specifier from "%-40sXXX"
703 to "%-39s XXX". The latter will always guarantee a space after the
704 field while the former will cause them to run together if the field
705 is 40 of more characters long.
708 +) Constify the cipher and digest 'method' functions and structures
709 and modify related functions to take constant EVP_MD and EVP_CIPHER
713 *) Implement ssl23_peek (analogous to ssl23_read), which previously
717 *) Replace rdtsc with _emit statements for VC++ version 5.
718 [Jeremy Cooper <jeremy@baymoo.org>]
720 +) Hide BN_CTX structure details in bn_lcl.h instead of publishing them
721 in <openssl/bn.h>. Also further increase BN_CTX_NUM to 32.
724 +) Modify EVP_Digest*() routines so they now return values. Although the
725 internal software routines can never fail additional hardware versions
729 +) Clean up crypto/err/err.h and change some error codes to avoid conflicts:
731 Previously ERR_R_FATAL was too small and coincided with ERR_LIB_PKCS7
732 (= ERR_R_PKCS7_LIB); it is now 64 instead of 32.
735 ERR_R_NESTED_ASN1_ERROR
737 ERR_R_MISSING_ASN1_EOS
738 were 4 .. 9, conflicting with
739 ERR_LIB_RSA (= ERR_R_RSA_LIB)
741 ERR_LIB_PEM (= ERR_R_PEM_LIB).
742 They are now 58 .. 63 (i.e., just below ERR_R_FATAL).
744 Add new error code 'ERR_R_INTERNAL_ERROR'.
747 +) Don't overuse locks in crypto/err/err.c: For data retrieval, CRYPTO_r_lock
751 +) New option '-subj arg' for 'openssl req' and 'openssl ca'. This
752 sets the subject name for a new request or supersedes the
753 subject name in a given request. Formats that can be parsed are
754 'CN=Some Name, OU=myOU, C=IT'
756 'CN=Some Name/OU=myOU/C=IT'.
758 Add options '-batch' and '-verbose' to 'openssl req'.
759 [Massimiliano Pala <madwolf@hackmasters.net>]
761 *) Make it possible to reuse SSLv2 sessions.
764 +) Introduce the possibility to access global variables through
765 functions on platform were that's the best way to handle exporting
766 global variables in shared libraries. To enable this functionality,
767 one must configure with "EXPORT_VAR_AS_FN" or defined the C macro
768 "OPENSSL_EXPORT_VAR_AS_FUNCTION" in crypto/opensslconf.h (the latter
769 is normally done by Configure or something similar).
771 To implement a global variable, use the macro OPENSSL_IMPLEMENT_GLOBAL
772 in the source file (foo.c) like this:
774 OPENSSL_IMPLEMENT_GLOBAL(int,foo)=1;
775 OPENSSL_IMPLEMENT_GLOBAL(double,bar);
777 To declare a global variable, use the macros OPENSSL_DECLARE_GLOBAL
778 and OPENSSL_GLOBAL_REF in the header file (foo.h) like this:
780 OPENSSL_DECLARE_GLOBAL(int,foo);
781 #define foo OPENSSL_GLOBAL_REF(foo)
782 OPENSSL_DECLARE_GLOBAL(double,bar);
783 #define bar OPENSSL_GLOBAL_REF(bar)
785 The #defines are very important, and therefore so is including the
786 header file everywere where the defined globals are used.
788 The macro OPENSSL_EXPORT_VAR_AS_FUNCTION also affects the definition
789 of ASN.1 items, but that structure is a bt different.
791 The largest change is in util/mkdef.pl which has been enhanced with
792 better and easier to understand logic to choose which symbols should
793 go into the Windows .def files as well as a number of fixes and code
794 cleanup (among others, algorithm keywords are now sorted
795 lexicographically to avoid constant rewrites).
798 *) In copy_email() check for >= 0 as a return value for
799 X509_NAME_get_index_by_NID() since 0 is a valid index.
800 [Steve Henson reported by Massimiliano Pala <madwolf@opensca.org>]
802 +) In BN_div() keep a copy of the sign of 'num' before writing the
803 result to 'rm' because if rm==num the value will be overwritten
804 and produce the wrong result if 'num' is negative: this caused
805 problems with BN_mod() and BN_nnmod().
808 +) Function OCSP_request_verify(). This checks the signature on an
809 OCSP request and verifies the signer certificate. The signer
810 certificate is just checked for a generic purpose and OCSP request
814 +) Add OCSP_check_validity() function to check the validity of OCSP
815 responses. OCSP responses are prepared in real time and may only
816 be a few seconds old. Simply checking that the current time lies
817 between thisUpdate and nextUpdate max reject otherwise valid responses
818 caused by either OCSP responder or client clock innacuracy. Instead
819 we allow thisUpdate and nextUpdate to fall within a certain period of
820 the current time. The age of the response can also optionally be
821 checked. Two new options -validity_period and -status_age added to
825 +) If signature or public key algorithm is unrecognized print out its
826 OID rather that just UNKOWN.
829 *) Avoid coredump with unsupported or invalid public keys by checking if
830 X509_get_pubkey() fails in PKCS7_verify(). Fix memory leak when
831 PKCS7_verify() fails with non detached data.
834 +) Change OCSP_cert_to_id() to tolerate a NULL subject certificate and
835 OCSP_cert_id_new() a NULL serialNumber. This allows a partial certificate
836 ID to be generated from the issuer certificate alone which can then be
837 passed to OCSP_id_issuer_cmp().
840 +) New compilation option ASN1_ITEM_FUNCTIONS. This causes the new
841 ASN1 modules to export functions returning ASN1_ITEM pointers
842 instead of the ASN1_ITEM structures themselves. This adds several
843 new macros which allow the underlying ASN1 function/structure to
844 be accessed transparently. As a result code should not use ASN1_ITEM
845 references directly (such as &X509_it) but instead use the relevant
846 macros (such as ASN1_ITEM_rptr(X509)). This option is to allow
847 use of the new ASN1 code on platforms where exporting structures
848 is problematical (for example in shared libraries) but exporting
849 functions returning pointers to structures is not.
852 +) Add support for overriding the generation of SSL/TLS session IDs.
853 These callbacks can be registered either in an SSL_CTX or per SSL.
854 The purpose of this is to allow applications to control, if they wish,
855 the arbitrary values chosen for use as session IDs, particularly as it
856 can be useful for session caching in multiple-server environments. A
857 command-line switch for testing this (and any client code that wishes
858 to use such a feature) has been added to "s_server".
859 [Geoff Thorpe, Lutz Jaenicke]
861 +) Modify mkdef.pl to recognise and parse prprocessor conditionals
862 of the form '#if defined(...) || defined(...) || ...' and
863 '#if !defined(...) && !defined(...) && ...'. This also avoids
864 the growing number of special cases it was previously handling.
867 *) Don't use getenv in library functions when run as setuid/setgid.
868 New function OPENSSL_issetugid().
871 +) Make all configuration macros available for application by making
872 sure they are available in opensslconf.h, by giving them names starting
873 with "OPENSSL_" to avoid conflicts with other packages and by making
874 sure e_os2.h will cover all platform-specific cases together with
876 Additionally, it is now possible to define configuration/platform-
877 specific names (called "system identities"). In the C code, these
878 are prefixed with "OPENSSL_SYSNAME_". e_os2.h will create another
879 macro with the name beginning with "OPENSSL_SYS_", which is determined
880 from "OPENSSL_SYSNAME_*" or compiler-specific macros depending on
884 +) New option -set_serial to 'req' and 'x509' this allows the serial
885 number to use to be specified on the command line. Previously self
886 signed certificates were hard coded with serial number 0 and the
887 CA options of 'x509' had to use a serial number in a file which was
891 *) Avoid false positives in memory leak detection code (crypto/mem_dbg.c)
892 due to incorrect handling of multi-threading:
894 1. Fix timing glitch in the MemCheck_off() portion of CRYPTO_mem_ctrl().
896 2. Fix logical glitch in is_MemCheck_on() aka CRYPTO_is_mem_check_on().
898 3. Count how many times MemCheck_off() has been called so that
899 nested use can be treated correctly. This also avoids
900 inband-signalling in the previous code (which relied on the
901 assumption that thread ID 0 is impossible).
904 +) New options to 'ca' utility to support V2 CRL entry extensions.
905 Currently CRL reason, invalidity date and hold instruction are
906 supported. Add new CRL extensions to V3 code and some new objects.
909 *) Add "-rand" option also to s_client and s_server.
912 +) New function EVP_CIPHER_CTX_set_padding() this is used to
913 disable standard block padding (aka PKCS#5 padding) in the EVP
914 API, which was previously mandatory. This means that the data is
915 not padded in any way and so the total length much be a multiple
916 of the block size, otherwise an error occurs.
919 +) Initial (incomplete) OCSP SSL support.
922 *) Fix CPU detection on Irix 6.x.
923 [Kurt Hockenbury <khockenb@stevens-tech.edu> and
924 "Bruce W. Forsberg" <bruce.forsberg@baesystems.com>]
926 +) New function OCSP_parse_url(). This splits up a URL into its host,
927 port and path components: primarily to parse OCSP URLs. New -url
928 option to ocsp utility.
931 +) New nonce behavior. The return value of OCSP_check_nonce() now
932 reflects the various checks performed. Applications can decide
933 whether to tolerate certain situations such as an absent nonce
934 in a response when one was present in a request: the ocsp application
935 just prints out a warning. New function OCSP_add1_basic_nonce()
936 this is to allow responders to include a nonce in a response even if
937 the request is nonce-less.
940 *) Fix X509_NAME bug which produced incorrect encoding if X509_NAME
943 [This change does not apply to 0.9.7.]
945 *) Use the cached encoding of an X509_NAME structure rather than
946 copying it. This is apparently the reason for the libsafe "errors"
947 but the code is actually correct.
950 +) Disable stdin buffering in load_cert (apps/apps.c) so that no certs are
951 skipped when using openssl x509 multiple times on a single input file,
952 e.g. "(openssl x509 -out cert1; openssl x509 -out cert2) <certs".
955 +) Make ASN1_UTCTIME_set_string() and ASN1_GENERALIZEDTIME_set_string()
956 set string type: to handle setting ASN1_TIME structures. Fix ca
957 utility to correctly initialize revocation date of CRLs.
960 +) New option SSL_OP_CIPHER_SERVER_PREFERENCE allows the server to override
961 the clients preferred ciphersuites and rather use its own preferences.
962 Should help to work around M$ SGC (Server Gated Cryptography) bug in
963 Internet Explorer by ensuring unchanged hash method during stepup.
964 (Also replaces the broken/deactivated SSL_OP_NON_EXPORT_FIRST option.)
967 +) Make mkdef.pl recognise all DECLARE_ASN1 macros, change rijndael
968 to aes and add a new 'exist' option to print out symbols that don't
972 +) Additional options to ocsp utility to allow flags to be set and
973 additional certificates supplied.
976 +) Add the option -VAfile to 'openssl ocsp', so the user can give the
977 OCSP client a number of certificate to only verify the response
981 *) Add new function BN_rand_range(), and fix DSA_sign_setup() to prevent
982 Bleichenbacher's DSA attack.
983 Extend BN_[pseudo_]rand: As before, top=1 forces the highest two bits
984 to be set and top=0 forces the highest bit to be set; top=-1 is new
985 and leaves the highest bit random.
986 [Ulf Moeller, Bodo Moeller]
988 +) Update Rijndael code to version 3.0 and change EVP AES ciphers to
989 handle the new API. Currently only ECB, CBC modes supported. Add new
990 AES OIDs. Add TLS AES ciphersuites as described in the "AES Ciphersuites
991 for TLS" draft-ietf-tls-ciphersuite-03.txt.
992 [Ben Laurie, Steve Henson]
994 *) In the NCONF_...-based implementations for CONF_... queries
995 (crypto/conf/conf_lib.c), if the input LHASH is NULL, avoid using
996 a temporary CONF structure with the data component set to NULL
997 (which gives segmentation faults in lh_retrieve).
998 Instead, use NULL for the CONF pointer in CONF_get_string and
999 CONF_get_number (which may use environment variables) and directly
1000 return NULL from CONF_get_section.
1003 *) Fix potential buffer overrun for EBCDIC.
1006 +) New function OCSP_copy_nonce() to copy nonce value (if present) from
1007 request to response.
1010 +) Functions for OCSP responders. OCSP_request_onereq_count(),
1011 OCSP_request_onereq_get0(), OCSP_onereq_get0_id() and OCSP_id_get0_info()
1012 extract information from a certificate request. OCSP_response_create()
1013 creates a response and optionally adds a basic response structure.
1014 OCSP_basic_add1_status() adds a complete single response to a basic
1015 reponse and returns the OCSP_SINGLERESP structure just added (to allow
1016 extensions to be included for example). OCSP_basic_add1_cert() adds a
1017 certificate to a basic response and OCSP_basic_sign() signs a basic
1018 response with various flags. New helper functions ASN1_TIME_check()
1019 (checks validity of ASN1_TIME structure) and ASN1_TIME_to_generalizedtime()
1020 (converts ASN1_TIME to GeneralizedTime).
1023 +) Various new functions. EVP_Digest() combines EVP_Digest{Init,Update,Final}()
1024 in a single operation. X509_get0_pubkey_bitstr() extracts the public_key
1025 structure from a certificate. X509_pubkey_digest() digests tha public_key
1026 contents: this is used in various key identifiers.
1029 *) Tolerate nonRepudiation as being valid for S/MIME signing and certSign
1030 keyUsage if basicConstraints absent for a CA.
1033 *) Make SMIME_write_PKCS7() write mail header values with a format that
1034 is more generally accepted (no spaces before the semicolon), since
1035 some programs can't parse those values properly otherwise. Also make
1036 sure BIO's that break lines after each write do not create invalid
1040 +) Make sk_sort() tolerate a NULL argument.
1041 [Steve Henson reported by Massimiliano Pala <madwolf@comune.modena.it>]
1043 +) New OCSP verify flag OCSP_TRUSTOTHER. When set the "other" certificates
1044 passed by the function are trusted implicitly. If any of them signed the
1045 reponse then it is assumed to be valid and is not verified.
1048 -) Make the CRL encoding routines work with empty SEQUENCE OF. The
1049 macros previously used would not encode an empty SEQUENCE OF
1050 and break the signature.
1052 [This change does not apply to 0.9.7.]
1054 *) Zero the premaster secret after deriving the master secret in
1058 +) In PKCS7_set_type() initialise content_type in PKCS7_ENC_CONTENT
1059 to data. This was previously part of the PKCS7 ASN1 code. This
1060 was causing problems with OpenSSL created PKCS#12 and PKCS#7 structures.
1061 [Steve Henson, reported by Kenneth R. Robinette
1062 <support@securenetterm.com>]
1064 +) Add CRYPTO_push_info() and CRYPTO_pop_info() calls to new ASN1
1065 routines: without these tracing memory leaks is very painful.
1066 Fix leaks in PKCS12 and PKCS7 routines.
1069 *) Add some EVP_add_digest_alias registrations (as found in
1070 OpenSSL_add_all_digests()) to SSL_library_init()
1071 aka OpenSSL_add_ssl_algorithms(). This provides improved
1072 compatibility with peers using X.509 certificates
1073 with unconventional AlgorithmIdentifier OIDs.
1076 *) Fix for Irix with NO_ASM.
1077 ["Bruce W. Forsberg" <bruce.forsberg@baesystems.com>]
1079 *) ./config script fixes.
1080 [Ulf Moeller, Richard Levitte]
1082 +) Make X509_time_adj() cope with the new behaviour of ASN1_TIME_new().
1083 Previously it initialised the 'type' argument to V_ASN1_UTCTIME which
1084 effectively meant GeneralizedTime would never be used. Now it
1085 is initialised to -1 but X509_time_adj() now has to check the value
1086 and use ASN1_TIME_set() if the value is not V_ASN1_UTCTIME or
1087 V_ASN1_GENERALIZEDTIME, without this it always uses GeneralizedTime.
1088 [Steve Henson, reported by Kenneth R. Robinette
1089 <support@securenetterm.com>]
1091 +) Fixes to BN_to_ASN1_INTEGER when bn is zero. This would previously
1092 result in a zero length in the ASN1_INTEGER structure which was
1093 not consistent with the structure when d2i_ASN1_INTEGER() was used
1094 and would cause ASN1_INTEGER_cmp() to fail. Enhance s2i_ASN1_INTEGER()
1095 to cope with hex and negative integers. Fix bug in i2a_ASN1_INTEGER()
1096 where it did not print out a minus for negative ASN1_INTEGER.
1099 *) Fix 'openssl passwd -1'.
1102 +) Add summary printout to ocsp utility. The various functions which
1103 convert status values to strings have been renamed to:
1104 OCSP_response_status_str(), OCSP_cert_status_str() and
1105 OCSP_crl_reason_str() and are no longer static. New options
1106 to verify nonce values and to disable verification. OCSP response
1107 printout format cleaned up.
1110 +) Add additional OCSP certificate checks. These are those specified
1111 in RFC2560. This consists of two separate checks: the CA of the
1112 certificate being checked must either be the OCSP signer certificate
1113 or the issuer of the OCSP signer certificate. In the latter case the
1114 OCSP signer certificate must contain the OCSP signing extended key
1115 usage. This check is performed by attempting to match the OCSP
1116 signer or the OCSP signer CA to the issuerNameHash and issuerKeyHash
1117 in the OCSP_CERTID structures of the response.
1120 +) Initial OCSP certificate verification added to OCSP_basic_verify()
1121 and related routines. This uses the standard OpenSSL certificate
1122 verify routines to perform initial checks (just CA validity) and
1123 to obtain the certificate chain. Then additional checks will be
1124 performed on the chain. Currently the root CA is checked to see
1125 if it is explicitly trusted for OCSP signing. This is used to set
1126 a root CA as a global signing root: that is any certificate that
1127 chains to that CA is an acceptable OCSP signing certificate.
1130 +) New '-extfile ...' option to 'openssl ca' for reading X.509v3
1131 extensions from a separate configuration file.
1132 As when reading extensions from the main configuration file,
1133 the '-extensions ...' option may be used for specifying the
1135 [Massimiliano Pala <madwolf@comune.modena.it>]
1137 *) Change PKCS12_key_gen_asc() so it can cope with non null
1138 terminated strings whose length is passed in the passlen
1139 parameter, for example from PEM callbacks. This was done
1140 by adding an extra length parameter to asc2uni().
1141 [Steve Henson, reported by <oddissey@samsung.co.kr>]
1143 +) New OCSP utility. Allows OCSP requests to be generated or
1144 read. The request can be sent to a responder and the output
1145 parsed, outputed or printed in text form. Not complete yet:
1146 still needs to check the OCSP response validity.
1149 +) New subcommands for 'openssl ca':
1150 'openssl ca -status <serial>' prints the status of the cert with
1151 the given serial number (according to the index file).
1152 'openssl ca -updatedb' updates the expiry status of certificates
1154 [Massimiliano Pala <madwolf@comune.modena.it>]
1156 +) New '-newreq-nodes' command option to CA.pl. This is like
1157 '-newreq', but calls 'openssl req' with the '-nodes' option
1158 so that the resulting key is not encrypted.
1159 [Damien Miller <djm@mindrot.org>]
1161 +) New configuration for the GNU Hurd.
1162 [Jonathan Bartlett <johnnyb@wolfram.com> via Richard Levitte]
1164 +) Initial code to implement OCSP basic response verify. This
1165 is currently incomplete. Currently just finds the signer's
1166 certificate and verifies the signature on the response.
1169 +) New SSLeay_version code SSLEAY_DIR to determine the compiled-in
1170 value of OPENSSLDIR. This is available via the new '-d' option
1171 to 'openssl version', and is also included in 'openssl version -a'.
1174 *) Fix C code generated by 'openssl dsaparam -C': If a BN_bin2bn
1175 call failed, free the DSA structure.
1178 +) Allowing defining memory allocation callbacks that will be given
1179 file name and line number information in additional arguments
1180 (a const char* and an int). The basic functionality remains, as
1181 well as the original possibility to just replace malloc(),
1182 realloc() and free() by functions that do not know about these
1183 additional arguments. To register and find out the current
1184 settings for extended allocation functions, the following
1185 functions are provided:
1187 CRYPTO_set_mem_ex_functions
1188 CRYPTO_set_locked_mem_ex_functions
1189 CRYPTO_get_mem_ex_functions
1190 CRYPTO_get_locked_mem_ex_functions
1192 These work the same way as CRYPTO_set_mem_functions and friends.
1193 CRYPTO_get_[locked_]mem_functions now writes 0 where such an
1194 extended allocation function is enabled.
1195 Similarly, CRYPTO_get_[locked_]mem_ex_functions writes 0 where
1196 a conventional allocation function is enabled.
1197 [Richard Levitte, Bodo Moeller]
1199 *) Fix to uni2asc() to cope with zero length Unicode strings.
1200 These are present in some PKCS#12 files.
1203 +) Finish off removing the remaining LHASH function pointer casts.
1204 There should no longer be any prototype-casting required when using
1205 the LHASH abstraction, and any casts that remain are "bugs". See
1206 the callback types and macros at the head of lhash.h for details
1207 (and "OBJ_cleanup" in crypto/objects/obj_dat.c as an example).
1210 +) Add automatic query of EGD sockets in RAND_poll() for the unix variant.
1211 If /dev/[u]random devices are not available or do not return enough
1212 entropy, EGD style sockets (served by EGD or PRNGD) will automatically
1214 The locations /var/run/egd-pool, /dev/egd-pool, /etc/egd-pool, and
1215 /etc/entropy will be queried once each in this sequence, quering stops
1216 when enough entropy was collected without querying more sockets.
1219 +) Change the Unix RAND_poll() variant to be able to poll several
1220 random devices, as specified by DEVRANDOM, until a sufficient amount
1221 of data has been collected. We spend at most 10 ms on each file
1222 (select timeout) and read in non-blocking mode. DEVRANDOM now
1223 defaults to the list "/dev/urandom", "/dev/random", "/dev/srandom"
1224 (previously it was just the string "/dev/urandom"), so on typical
1225 platforms the 10 ms delay will never occur.
1226 Also separate out the Unix variant to its own file, rand_unix.c.
1227 For VMS, there's a currently-empty rand_vms.c.
1230 +) Move OCSP client related routines to ocsp_cl.c. These
1231 provide utility functions which an application needing
1232 to issue a request to an OCSP responder and analyse the
1233 response will typically need: as opposed to those which an
1234 OCSP responder itself would need which will be added later.
1236 OCSP_request_sign() signs an OCSP request with an API similar
1237 to PKCS7_sign(). OCSP_response_status() returns status of OCSP
1238 response. OCSP_response_get1_basic() extracts basic response
1239 from response. OCSP_resp_find_status(): finds and extracts status
1240 information from an OCSP_CERTID structure (which will be created
1241 when the request structure is built). These are built from lower
1242 level functions which work on OCSP_SINGLERESP structures but
1243 wont normally be used unless the application wishes to examine
1244 extensions in the OCSP response for example.
1246 Replace nonce routines with a pair of functions.
1247 OCSP_request_add1_nonce() adds a nonce value and optionally
1248 generates a random value. OCSP_check_nonce() checks the
1249 validity of the nonce in an OCSP response.
1252 +) Change function OCSP_request_add() to OCSP_request_add0_id().
1253 This doesn't copy the supplied OCSP_CERTID and avoids the
1254 need to free up the newly created id. Change return type
1255 to OCSP_ONEREQ to return the internal OCSP_ONEREQ structure.
1256 This can then be used to add extensions to the request.
1257 Deleted OCSP_request_new(), since most of its functionality
1258 is now in OCSP_REQUEST_new() (and the case insensitive name
1259 clash) apart from the ability to set the request name which
1260 will be added elsewhere.
1263 +) Update OCSP API. Remove obsolete extensions argument from
1264 various functions. Extensions are now handled using the new
1265 OCSP extension code. New simple OCSP HTTP function which
1266 can be used to send requests and parse the response.
1269 +) Fix the PKCS#7 (S/MIME) code to work with new ASN1. Two new
1270 ASN1_ITEM structures help with sign and verify. PKCS7_ATTR_SIGN
1271 uses the special reorder version of SET OF to sort the attributes
1272 and reorder them to match the encoded order. This resolves a long
1273 standing problem: a verify on a PKCS7 structure just after signing
1274 it used to fail because the attribute order did not match the
1275 encoded order. PKCS7_ATTR_VERIFY does not reorder the attributes:
1276 it uses the received order. This is necessary to tolerate some broken
1277 software that does not order SET OF. This is handled by encoding
1278 as a SEQUENCE OF but using implicit tagging (with UNIVERSAL class)
1279 to produce the required SET OF.
1282 +) Have mk1mf.pl generate the macros OPENSSL_BUILD_SHLIBCRYPTO and
1283 OPENSSL_BUILD_SHLIBSSL and use them appropriately in the header
1284 files to get correct declarations of the ASN.1 item variables.
1287 +) Rewrite of PKCS#12 code to use new ASN1 functionality. Replace many
1288 PKCS#12 macros with real functions. Fix two unrelated ASN1 bugs:
1289 asn1_check_tlen() would sometimes attempt to use 'ctx' when it was
1290 NULL and ASN1_TYPE was not dereferenced properly in asn1_ex_c2i().
1291 New ASN1 macro: DECLARE_ASN1_ITEM() which just declares the relevant
1292 ASN1_ITEM and no wrapper functions.
1295 +) New functions or ASN1_item_d2i_fp() and ASN1_item_d2i_bio(). These
1296 replace the old function pointer based I/O routines. Change most of
1297 the *_d2i_bio() and *_d2i_fp() functions to use these.
1300 +) Enhance mkdef.pl to be more accepting about spacing in C preprocessor
1301 lines, recognice more "algorithms" that can be deselected, and make
1302 it complain about algorithm deselection that isn't recognised.
1305 +) New ASN1 functions to handle dup, sign, verify, digest, pack and
1306 unpack operations in terms of ASN1_ITEM. Modify existing wrappers
1307 to use new functions. Add NO_ASN1_OLD which can be set to remove
1308 some old style ASN1 functions: this can be used to determine if old
1309 code will still work when these eventually go away.
1312 +) New extension functions for OCSP structures, these follow the
1313 same conventions as certificates and CRLs.
1316 +) New function X509V3_add1_i2d(). This automatically encodes and
1317 adds an extension. Its behaviour can be customised with various
1318 flags to append, replace or delete. Various wrappers added for
1319 certifcates and CRLs.
1322 +) Fix to avoid calling the underlying ASN1 print routine when
1323 an extension cannot be parsed. Correct a typo in the
1324 OCSP_SERVICELOC extension. Tidy up print OCSP format.
1327 *) Increase s2->wbuf allocation by one byte in ssl2_new (ssl/s2_lib.c).
1328 Otherwise do_ssl_write (ssl/s2_pkt.c) will write beyond buffer limits
1329 when writing a 32767 byte record.
1330 [Bodo Moeller; problem reported by Eric Day <eday@concentric.net>]
1332 *) In RSA_eay_public_{en,ed}crypt and RSA_eay_mod_exp (rsa_eay.c),
1333 obtain lock CRYPTO_LOCK_RSA before setting rsa->_method_mod_{n,p,q}.
1335 (RSA objects have a reference count access to which is protected
1336 by CRYPTO_LOCK_RSA [see rsa_lib.c, s3_srvr.c, ssl_cert.c, ssl_rsa.c],
1337 so they are meant to be shared between threads.)
1338 [Bodo Moeller, Geoff Thorpe; original patch submitted by
1339 "Reddie, Steven" <Steven.Reddie@ca.com>]
1341 +) Make mkdef.pl parse some of the ASN1 macros and add apropriate
1342 entries for variables.
1345 *) Fix a deadlock in CRYPTO_mem_leaks().
1348 +) Add functionality to apps/openssl.c for detecting locking
1349 problems: As the program is single-threaded, all we have
1350 to do is register a locking callback using an array for
1351 storing which locks are currently held by the program.
1354 +) Use a lock around the call to CRYPTO_get_ex_new_index() in
1355 SSL_get_ex_data_X509_STORE_idx(), which is used in
1356 ssl_verify_cert_chain() and thus can be called at any time
1357 during TLS/SSL handshakes so that thread-safety is essential.
1358 Unfortunately, the ex_data design is not at all suited
1359 for multi-threaded use, so it probably should be abolished.
1362 +) Added Broadcom "ubsec" ENGINE to OpenSSL.
1363 [Broadcom, tweaked and integrated by Geoff Thorpe]
1365 +) Move common extension printing code to new function
1366 X509V3_print_extensions(). Reorganise OCSP print routines and
1367 implement some needed OCSP ASN1 functions. Add OCSP extensions.
1370 +) New function X509_signature_print() to remove duplication in some
1374 +) Add a special meaning when SET OF and SEQUENCE OF flags are both
1375 set (this was treated exactly the same as SET OF previously). This
1376 is used to reorder the STACK representing the structure to match the
1377 encoding. This will be used to get round a problem where a PKCS7
1378 structure which was signed could not be verified because the STACK
1379 order did not reflect the encoded order.
1382 +) Reimplement the OCSP ASN1 module using the new code.
1385 +) Update the X509V3 code to permit the use of an ASN1_ITEM structure
1386 for its ASN1 operations. The old style function pointers still exist
1387 for now but they will eventually go away.
1390 +) Merge in replacement ASN1 code from the ASN1 branch. This almost
1391 completely replaces the old ASN1 functionality with a table driven
1392 encoder and decoder which interprets an ASN1_ITEM structure describing
1393 the ASN1 module. Compatibility with the existing ASN1 API (i2d,d2i) is
1394 largely maintained. Almost all of the old asn1_mac.h macro based ASN1
1395 has also been converted to the new form.
1398 +) Change BN_mod_exp_recp so that negative moduli are tolerated
1399 (the sign is ignored). Similarly, ignore the sign in BN_MONT_CTX_set
1400 so that BN_mod_exp_mont and BN_mod_exp_mont_word work
1401 for negative moduli.
1404 +) Fix BN_uadd and BN_usub: Always return non-negative results instead
1405 of not touching the result's sign bit.
1408 +) BN_div bugfix: If the result is 0, the sign (res->neg) must not be
1412 +) Changed the LHASH code to use prototypes for callbacks, and created
1413 macros to declare and implement thin (optionally static) functions
1414 that provide type-safety and avoid function pointer casting for the
1415 type-specific callbacks.
1418 *) Use better test patterns in bntest.
1421 +) Added Kerberos Cipher Suites to be used with TLS, as written in
1423 [Veers Staats <staatsvr@asc.hpc.mil>,
1424 Jeffrey Altman <jaltman@columbia.edu>, via Richard Levitte]
1426 *) rand_win.c fix for Borland C.
1429 *) BN_rshift bugfix for n == 0.
1432 +) Reformat the FAQ so the different questions and answers can be divided
1433 in sections depending on the subject.
1436 +) Have the zlib compression code load ZLIB.DLL dynamically under
1440 +) New function BN_mod_sqrt for computing square roots modulo a prime
1441 (using the probabilistic Tonelli-Shanks algorithm unless
1442 p == 3 (mod 4) or p == 5 (mod 8), which are cases that can
1443 be handled deterministically).
1444 [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller]
1446 *) Add a 'bctest' script that checks for some known 'bc' bugs
1447 so that 'make test' does not abort just because 'bc' is broken.
1450 *) Store verify_result within SSL_SESSION also for client side to
1451 avoid potential security hole. (Re-used sessions on the client side
1452 always resulted in verify_result==X509_V_OK, not using the original
1453 result of the server certificate verification.)
1456 +) Make BN_mod_inverse faster by explicitly handling small quotients
1457 in the Euclid loop. (Speed gain about 20% for small moduli [256 or
1458 512 bits], about 30% for larger ones [1024 or 2048 bits].)
1461 *) Fix ssl3_pending: If the record in s->s3->rrec is not of type
1462 SSL3_RT_APPLICATION_DATA, return 0.
1463 Similarly, change ssl2_pending to return 0 if SSL_in_init(s) is true.
1467 Both ssl2_peek and ssl3_peek, which were totally broken in earlier
1468 releases, have been re-implemented by renaming the previous
1469 implementations of ssl2_read and ssl3_read to ssl2_read_internal
1470 and ssl3_read_internal, respectively, and adding 'peek' parameters
1471 to them. The new ssl[23]_{read,peek} functions are calls to
1472 ssl[23]_read_internal with the 'peek' flag set appropriately.
1473 A 'peek' parameter has also been added to ssl3_read_bytes, which
1474 does the actual work for ssl3_read_internal.
1477 +) New function BN_kronecker.
1480 +) Fix BN_gcd so that it works on negative inputs; the result is
1481 positive unless both parameters are zero.
1482 Previously something reasonably close to an infinite loop was
1483 possible because numbers could be growing instead of shrinking
1484 in the implementation of Euclid's algorithm.
1487 +) Fix BN_is_word() and BN_is_one() macros to take into account the
1488 sign of the number in question.
1490 Fix BN_is_word(a,w) to work correctly for w == 0.
1492 The old BN_is_word(a,w) macro is now called BN_abs_is_word(a,w)
1493 because its test if the absolute value of 'a' equals 'w'.
1494 Note that BN_abs_is_word does *not* handle w == 0 reliably;
1495 it exists mostly for use in the implementations of BN_is_zero(),
1496 BN_is_one(), and BN_is_word().
1499 *) Initialise "ex_data" member of RSA/DSA/DH structures prior to calling
1500 the method-specific "init()" handler. Also clean up ex_data after
1501 calling the method-specific "finish()" handler. Previously, this was
1502 happening the other way round.
1505 +) New function BN_swap.
1508 +) Use BN_nnmod instead of BN_mod in crypto/bn/bn_exp.c so that
1509 the exponentiation functions are more likely to produce reasonable
1510 results on negative inputs.
1513 +) Change BN_mod_mul so that the result is always non-negative.
1514 Previously, it could be negative if one of the factors was negative;
1515 I don't think anyone really wanted that behaviour.
1518 +) Move BN_mod_... functions into new file crypto/bn/bn_mod.c
1519 (except for exponentiation, which stays in crypto/bn/bn_exp.c,
1520 and BN_mod_mul_reciprocal, which stays in crypto/bn/bn_recp.c)
1521 and add new functions:
1530 BN_mod_lshift1_quick
1534 These functions always generate non-negative results.
1536 BN_nnmod otherwise is like BN_mod (if BN_mod computes a remainder r
1537 such that |m| < r < 0, BN_nnmod will output rem + |m| instead).
1539 BN_mod_XXX_quick(r, a, [b,] m) generates the same result as
1540 BN_mod_XXX(r, a, [b,] m, ctx), but requires that a [and b]
1541 be reduced modulo m.
1542 [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller]
1544 +) Remove a few calls to bn_wexpand() in BN_sqr() (the one in there
1545 was actually never needed) and in BN_mul(). The removal in BN_mul()
1546 required a small change in bn_mul_part_recursive() and the addition
1547 of the functions bn_cmp_part_words(), bn_sub_part_words() and
1548 bn_add_part_words(), which do the same thing as bn_cmp_words(),
1549 bn_sub_words() and bn_add_words() except they take arrays with
1553 +) In 'openssl passwd', verify passwords read from the terminal
1554 unless the '-salt' option is used (which usually means that
1555 verification would just waste user's time since the resulting
1556 hash is going to be compared with some given password hash)
1557 or the new '-noverify' option is used.
1559 This is an incompatible change, but it does not affect
1560 non-interactive use of 'openssl passwd' (passwords on the command
1561 line, '-stdin' option, '-in ...' option) and thus should not
1565 +) Remove all references to RSAref, since there's no more need for it.
1568 *) Increase BN_CTX_NUM (the number of BIGNUMs in a BN_CTX) to 16.
1569 The previous value, 12, was not always sufficient for BN_mod_exp().
1572 +) Make DSO load along a path given through an environment variable
1573 (SHLIB_PATH) with shl_load().
1576 +) Constify the ENGINE code as a result of BIGNUM constification.
1577 Also constify the RSA code and most things related to it. In a
1578 few places, most notable in the depth of the ASN.1 code, ugly
1579 casts back to non-const were required (to be solved at a later
1583 +) Make it so the openssl application has all engines loaded by default.
1586 +) Constify the BIGNUM routines a little more.
1589 *) Make sure that shared libraries get the internal name engine with
1590 the full version number and not just 0. This should mark the
1591 shared libraries as not backward compatible. Of course, this should
1592 be changed again when we can guarantee backward binary compatibility.
1595 +) Add the following functions:
1597 ENGINE_load_cswift()
1599 ENGINE_load_atalla()
1601 ENGINE_load_builtin_engines()
1603 That way, an application can itself choose if external engines that
1604 are built-in in OpenSSL shall ever be used or not. The benefit is
1605 that applications won't have to be linked with libdl or other dso
1606 libraries unless it's really needed.
1608 Changed 'openssl engine' to load all engines on demand.
1609 Changed the engine header files to avoid the duplication of some
1610 declarations (they differed!).
1613 +) 'openssl engine' can now list capabilities.
1616 +) Better error reporting in 'openssl engine'.
1619 +) Never call load_dh_param(NULL) in s_server.
1622 +) Add engine application. It can currently list engines by name and
1623 identity, and test if they are actually available.
1626 +) Add support for shared libraries under Irix.
1627 [Albert Chin-A-Young <china@thewrittenword.com>]
1629 +) Improve RPM specification file by forcing symbolic linking and making
1630 sure the installed documentation is also owned by root.root.
1631 [Damien Miller <djm@mindrot.org>]
1633 +) Add configuration option to build on Linux on both big-endian and
1635 [Ralf Baechle <ralf@uni-koblenz.de>]
1637 +) Give the OpenSSL applications more possibilities to make use of
1638 keys (public as well as private) handled by engines.
1641 +) Add OCSP code that comes from CertCo.
1644 +) Add VMS support for the Rijndael code.
1647 +) Added untested support for Nuron crypto accelerator.
1650 +) Add support for external cryptographic devices. This code was
1651 previously distributed separately as the "engine" branch.
1652 [Geoff Thorpe, Richard Levitte]
1654 +) Rework the filename-translation in the DSO code. It is now possible to
1655 have far greater control over how a "name" is turned into a filename
1656 depending on the operating environment and any oddities about the
1657 different shared library filenames on each system.
1660 +) Support threads on FreeBSD-elf in Configure.
1663 +) Add the possibility to create shared libraries on HP-UX
1666 +) Fix for SHA1 assembly problem with MASM: it produces
1667 warnings about corrupt line number information when assembling
1668 with debugging information. This is caused by the overlapping
1670 [Bernd Matthes <mainbug@celocom.de>, Steve Henson]
1673 NCONF_get_number() has no error checking at all. As a replacement,
1674 NCONF_get_number_e() is defined (_e for "error checking") and is
1675 promoted strongly. The old NCONF_get_number is kept around for
1676 binary backward compatibility.
1677 Make it possible for methods to load from something other than a BIO,
1678 by providing a function pointer that is given a name instead of a BIO.
1679 For example, this could be used to load configuration data from an
1683 *) Fix typo in get_cert_by_subject() in by_dir.c
1684 [Jean-Marc Desperrier <jean-marc.desperrier@certplus.com>]
1686 *) Rework the system to generate shared libraries:
1688 - Make note of the expected extension for the shared libraries and
1689 if there is a need for symbolic links from for example libcrypto.so.0
1690 to libcrypto.so.0.9.7. There is extended info in Configure for
1693 - Make as few rebuilds of the shared libraries as possible.
1695 - Still avoid linking the OpenSSL programs with the shared libraries.
1697 - When installing, install the shared libraries separately from the
1701 +) Fix for non blocking accept BIOs. Added new I/O special reason
1702 BIO_RR_ACCEPT to cover this case. Previously use of accept BIOs
1703 with non blocking I/O was not possible because no retry code was
1704 implemented. Also added new SSL code SSL_WANT_ACCEPT to cover
1708 +) Added the beginnings of Rijndael support.
1711 +) Fix for bug in DirectoryString mask setting. Add support for
1712 X509_NAME_print_ex() in 'req' and X509_print_ex() function
1713 to allow certificate printing to more controllable, additional
1714 'certopt' option to 'x509' to allow new printing options to be
1718 +) Clean old EAY MD5 hack from e_os.h.
1721 *) Fix SSL_CTX_set_read_ahead macro to actually use its argument.
1723 Copy SSL_CTX's read_ahead flag to SSL object directly in SSL_new
1724 and not in SSL_clear because the latter is also used by the
1725 accept/connect functions; previously, the settings made by
1726 SSL_set_read_ahead would be lost during the handshake.
1727 [Bodo Moeller; problems reported by Anders Gertz <gertz@epact.se>]
1729 *) Correct util/mkdef.pl to be selective about disabled algorithms.
1730 Previously, it would create entries for disableed algorithms no
1734 *) Added several new manual pages for SSL_* function.
1737 Changes between 0.9.5a and 0.9.6 [24 Sep 2000]
1739 *) In ssl23_get_client_hello, generate an error message when faced
1740 with an initial SSL 3.0/TLS record that is too small to contain the
1741 first two bytes of the ClientHello message, i.e. client_version.
1742 (Note that this is a pathologic case that probably has never happened
1743 in real life.) The previous approach was to use the version number
1744 from the record header as a substitute; but our protocol choice
1745 should not depend on that one because it is not authenticated
1746 by the Finished messages.
1749 *) More robust randomness gathering functions for Windows.
1750 [Jeffrey Altman <jaltman@columbia.edu>]
1752 *) For compatibility reasons if the flag X509_V_FLAG_ISSUER_CHECK is
1753 not set then we don't setup the error code for issuer check errors
1754 to avoid possibly overwriting other errors which the callback does
1755 handle. If an application does set the flag then we assume it knows
1756 what it is doing and can handle the new informational codes
1760 *) Fix for a nasty bug in ASN1_TYPE handling. ASN1_TYPE is used for
1761 a general "ANY" type, as such it should be able to decode anything
1762 including tagged types. However it didn't check the class so it would
1763 wrongly interpret tagged types in the same way as their universal
1764 counterpart and unknown types were just rejected. Changed so that the
1765 tagged and unknown types are handled in the same way as a SEQUENCE:
1766 that is the encoding is stored intact. There is also a new type
1767 "V_ASN1_OTHER" which is used when the class is not universal, in this
1768 case we have no idea what the actual type is so we just lump them all
1772 *) On VMS, stdout may very well lead to a file that is written to
1773 in a record-oriented fashion. That means that every write() will
1774 write a separate record, which will be read separately by the
1775 programs trying to read from it. This can be very confusing.
1777 The solution is to put a BIO filter in the way that will buffer
1778 text until a linefeed is reached, and then write everything a
1779 line at a time, so every record written will be an actual line,
1780 not chunks of lines and not (usually doesn't happen, but I've
1781 seen it once) several lines in one record. BIO_f_linebuffer() is
1784 Currently, it's a VMS-only method, because that's where it has
1785 been tested well enough.
1788 *) Remove 'optimized' squaring variant in BN_mod_mul_montgomery,
1789 it can return incorrect results.
1790 (Note: The buggy variant was not enabled in OpenSSL 0.9.5a,
1791 but it was in 0.9.6-beta[12].)
1794 *) Disable the check for content being present when verifying detached
1795 signatures in pk7_smime.c. Some versions of Netscape (wrongly)
1796 include zero length content when signing messages.
1799 *) New BIO_shutdown_wr macro, which invokes the BIO_C_SHUTDOWN_WR
1800 BIO_ctrl (for BIO pairs).
1803 *) Add DSO method for VMS.
1806 *) Bug fix: Montgomery multiplication could produce results with the
1810 *) Add RPM specification openssl.spec and modify it to build three
1811 packages. The default package contains applications, application
1812 documentation and run-time libraries. The devel package contains
1813 include files, static libraries and function documentation. The
1814 doc package contains the contents of the doc directory. The original
1815 openssl.spec was provided by Damien Miller <djm@mindrot.org>.
1818 *) Add a large number of documentation files for many SSL routines.
1819 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>]
1821 *) Add a configuration entry for Sony News 4.
1822 [NAKAJI Hiroyuki <nakaji@tutrp.tut.ac.jp>]
1824 *) Don't set the two most significant bits to one when generating a
1825 random number < q in the DSA library.
1828 *) New SSL API mode 'SSL_MODE_AUTO_RETRY'. This disables the default
1829 behaviour that SSL_read may result in SSL_ERROR_WANT_READ (even if
1830 the underlying transport is blocking) if a handshake took place.
1831 (The default behaviour is needed by applications such as s_client
1832 and s_server that use select() to determine when to use SSL_read;
1833 but for applications that know in advance when to expect data, it
1834 just makes things more complicated.)
1837 *) Add RAND_egd_bytes(), which gives control over the number of bytes read
1841 *) Add a few more EBCDIC conditionals that make `req' and `x509'
1842 work better on such systems.
1843 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>]
1845 *) Add two demo programs for PKCS12_parse() and PKCS12_create().
1846 Update PKCS12_parse() so it copies the friendlyName and the
1847 keyid to the certificates aux info.
1850 *) Fix bug in PKCS7_verify() which caused an infinite loop
1851 if there was more than one signature.
1852 [Sven Uszpelkat <su@celocom.de>]
1854 *) Major change in util/mkdef.pl to include extra information
1855 about each symbol, as well as presentig variables as well
1856 as functions. This change means that there's n more need
1857 to rebuild the .num files when some algorithms are excluded.
1860 *) Allow the verify time to be set by an application,
1861 rather than always using the current time.
1864 *) Phase 2 verify code reorganisation. The certificate
1865 verify code now looks up an issuer certificate by a
1866 number of criteria: subject name, authority key id
1867 and key usage. It also verifies self signed certificates
1868 by the same criteria. The main comparison function is
1869 X509_check_issued() which performs these checks.
1871 Lot of changes were necessary in order to support this
1872 without completely rewriting the lookup code.
1874 Authority and subject key identifier are now cached.
1876 The LHASH 'certs' is X509_STORE has now been replaced
1877 by a STACK_OF(X509_OBJECT). This is mainly because an
1878 LHASH can't store or retrieve multiple objects with
1879 the same hash value.
1881 As a result various functions (which were all internal
1882 use only) have changed to handle the new X509_STORE
1883 structure. This will break anything that messed round
1884 with X509_STORE internally.
1886 The functions X509_STORE_add_cert() now checks for an
1887 exact match, rather than just subject name.
1889 The X509_STORE API doesn't directly support the retrieval
1890 of multiple certificates matching a given criteria, however
1891 this can be worked round by performing a lookup first
1892 (which will fill the cache with candidate certificates)
1893 and then examining the cache for matches. This is probably
1894 the best we can do without throwing out X509_LOOKUP
1895 entirely (maybe later...).
1897 The X509_VERIFY_CTX structure has been enhanced considerably.
1899 All certificate lookup operations now go via a get_issuer()
1900 callback. Although this currently uses an X509_STORE it
1901 can be replaced by custom lookups. This is a simple way
1902 to bypass the X509_STORE hackery necessary to make this
1903 work and makes it possible to use more efficient techniques
1904 in future. A very simple version which uses a simple
1905 STACK for its trusted certificate store is also provided
1906 using X509_STORE_CTX_trusted_stack().
1908 The verify_cb() and verify() callbacks now have equivalents
1909 in the X509_STORE_CTX structure.
1911 X509_STORE_CTX also has a 'flags' field which can be used
1912 to customise the verify behaviour.
1915 *) Add new PKCS#7 signing option PKCS7_NOSMIMECAP which
1916 excludes S/MIME capabilities.
1919 *) When a certificate request is read in keep a copy of the
1920 original encoding of the signed data and use it when outputing
1921 again. Signatures then use the original encoding rather than
1922 a decoded, encoded version which may cause problems if the
1923 request is improperly encoded.
1926 *) For consistency with other BIO_puts implementations, call
1927 buffer_write(b, ...) directly in buffer_puts instead of calling
1930 In BIO_puts, increment b->num_write as in BIO_write.
1931 [Peter.Sylvester@EdelWeb.fr]
1933 *) Fix BN_mul_word for the case where the word is 0. (We have to use
1934 BN_zero, we may not return a BIGNUM with an array consisting of
1938 *) Avoid calling abort() from within the library when problems are
1939 detected, except if preprocessor symbols have been defined
1940 (such as REF_CHECK, BN_DEBUG etc.).
1943 *) New openssl application 'rsautl'. This utility can be
1944 used for low level RSA operations. DER public key
1945 BIO/fp routines also added.
1948 *) New Configure entry and patches for compiling on QNX 4.
1949 [Andreas Schneider <andreas@ds3.etech.fh-hamburg.de>]
1951 *) A demo state-machine implementation was sponsored by
1952 Nuron (http://www.nuron.com/) and is now available in
1953 demos/state_machine.
1956 *) New options added to the 'dgst' utility for signature
1957 generation and verification.
1960 *) Unrecognized PKCS#7 content types are now handled via a
1961 catch all ASN1_TYPE structure. This allows unsupported
1962 types to be stored as a "blob" and an application can
1963 encode and decode it manually.
1966 *) Fix various signed/unsigned issues to make a_strex.c
1968 [Oscar Jacobsson <oscar.jacobsson@celocom.com>]
1970 *) ASN1 fixes. i2d_ASN1_OBJECT was not returning the correct
1971 length if passed a buffer. ASN1_INTEGER_to_BN failed
1972 if passed a NULL BN and its argument was negative.
1973 [Steve Henson, pointed out by Sven Heiberg <sven@tartu.cyber.ee>]
1975 *) Modification to PKCS#7 encoding routines to output definite
1976 length encoding. Since currently the whole structures are in
1977 memory there's not real point in using indefinite length
1978 constructed encoding. However if OpenSSL is compiled with
1979 the flag PKCS7_INDEFINITE_ENCODING the old form is used.
1982 *) Added BIO_vprintf() and BIO_vsnprintf().
1985 *) Added more prefixes to parse for in the the strings written
1986 through a logging bio, to cover all the levels that are available
1987 through syslog. The prefixes are now:
1989 PANIC, EMERG, EMR => LOG_EMERG
1990 ALERT, ALR => LOG_ALERT
1991 CRIT, CRI => LOG_CRIT
1992 ERROR, ERR => LOG_ERR
1993 WARNING, WARN, WAR => LOG_WARNING
1994 NOTICE, NOTE, NOT => LOG_NOTICE
1995 INFO, INF => LOG_INFO
1996 DEBUG, DBG => LOG_DEBUG
1998 and as before, if none of those prefixes are present at the
1999 beginning of the string, LOG_ERR is chosen.
2001 On Win32, the LOG_* levels are mapped according to this:
2003 LOG_EMERG, LOG_ALERT, LOG_CRIT, LOG_ERR => EVENTLOG_ERROR_TYPE
2004 LOG_WARNING => EVENTLOG_WARNING_TYPE
2005 LOG_NOTICE, LOG_INFO, LOG_DEBUG => EVENTLOG_INFORMATION_TYPE
2009 *) Made it possible to reconfigure with just the configuration
2010 argument "reconf" or "reconfigure". The command line arguments
2011 are stored in Makefile.ssl in the variable CONFIGURE_ARGS,
2012 and are retrieved from there when reconfiguring.
2016 [Assar Westerlund <assar@sics.se>, Richard Levitte]
2018 *) Add the arguments -CAfile and -CApath to the pkcs12 utility.
2021 *) The obj_dat.pl script was messing up the sorting of object
2022 names. The reason was that it compared the quoted version
2023 of strings as a result "OCSP" > "OCSP Signing" because
2024 " > SPACE. Changed script to store unquoted versions of
2025 names and add quotes on output. It was also omitting some
2026 names from the lookup table if they were given a default
2027 value (that is if SN is missing it is given the same
2028 value as LN and vice versa), these are now added on the
2029 grounds that if an object has a name we should be able to
2030 look it up. Finally added warning output when duplicate
2031 short or long names are found.
2034 *) Changes needed for Tandem NSK.
2035 [Scott Uroff <scott@xypro.com>]
2037 *) Fix SSL 2.0 rollback checking: Due to an off-by-one error in
2038 RSA_padding_check_SSLv23(), special padding was never detected
2039 and thus the SSL 3.0/TLS 1.0 countermeasure against protocol
2040 version rollback attacks was not effective.
2042 In s23_clnt.c, don't use special rollback-attack detection padding
2043 (RSA_SSLV23_PADDING) if SSL 2.0 is the only protocol enabled in the
2044 client; similarly, in s23_srvr.c, don't do the rollback check if
2045 SSL 2.0 is the only protocol enabled in the server.
2048 *) Make it possible to get hexdumps of unprintable data with 'openssl
2049 asn1parse'. By implication, the functions ASN1_parse_dump() and
2050 BIO_dump_indent() are added.
2053 *) New functions ASN1_STRING_print_ex() and X509_NAME_print_ex()
2054 these print out strings and name structures based on various
2055 flags including RFC2253 support and proper handling of
2056 multibyte characters. Added options to the 'x509' utility
2057 to allow the various flags to be set.
2060 *) Various fixes to use ASN1_TIME instead of ASN1_UTCTIME.
2061 Also change the functions X509_cmp_current_time() and
2062 X509_gmtime_adj() work with an ASN1_TIME structure,
2063 this will enable certificates using GeneralizedTime in validity
2064 dates to be checked.
2067 *) Make the NEG_PUBKEY_BUG code (which tolerates invalid
2068 negative public key encodings) on by default,
2069 NO_NEG_PUBKEY_BUG can be set to disable it.
2072 *) New function c2i_ASN1_OBJECT() which acts on ASN1_OBJECT
2073 content octets. An i2c_ASN1_OBJECT is unnecessary because
2074 the encoding can be trivially obtained from the structure.
2077 *) crypto/err.c locking bugfix: Use write locks (CRYPTO_w_[un]lock),
2078 not read locks (CRYPTO_r_[un]lock).
2081 *) A first attempt at creating official support for shared
2082 libraries through configuration. I've kept it so the
2083 default is static libraries only, and the OpenSSL programs
2084 are always statically linked for now, but there are
2085 preparations for dynamic linking in place.
2086 This has been tested on Linux and Tru64.
2089 *) Randomness polling function for Win9x, as described in:
2090 Peter Gutmann, Software Generation of Practically Strong
2094 *) Fix so PRNG is seeded in req if using an already existing
2098 *) New options to smime application. -inform and -outform
2099 allow alternative formats for the S/MIME message including
2100 PEM and DER. The -content option allows the content to be
2101 specified separately. This should allow things like Netscape
2102 form signing output easier to verify.
2105 *) Fix the ASN1 encoding of tags using the 'long form'.
2108 *) New ASN1 functions, i2c_* and c2i_* for INTEGER and BIT
2109 STRING types. These convert content octets to and from the
2110 underlying type. The actual tag and length octets are
2111 already assumed to have been read in and checked. These
2112 are needed because all other string types have virtually
2113 identical handling apart from the tag. By having versions
2114 of the ASN1 functions that just operate on content octets
2115 IMPLICIT tagging can be handled properly. It also allows
2116 the ASN1_ENUMERATED code to be cut down because ASN1_ENUMERATED
2117 and ASN1_INTEGER are identical apart from the tag.
2120 *) Change the handling of OID objects as follows:
2122 - New object identifiers are inserted in objects.txt, following
2123 the syntax given in objects.README.
2124 - objects.pl is used to process obj_mac.num and create a new
2126 - obj_dat.pl is used to create a new obj_dat.h, using the data in
2129 This is currently kind of a hack, and the perl code in objects.pl
2130 isn't very elegant, but it works as I intended. The simplest way
2131 to check that it worked correctly is to look in obj_dat.h and
2132 check the array nid_objs and make sure the objects haven't moved
2133 around (this is important!). Additions are OK, as well as
2134 consistent name changes.
2137 *) Add BSD-style MD5-based passwords to 'openssl passwd' (option '-1').
2140 *) Addition of the command line parameter '-rand file' to 'openssl req'.
2141 The given file adds to whatever has already been seeded into the
2142 random pool through the RANDFILE configuration file option or
2143 environment variable, or the default random state file.
2146 *) mkstack.pl now sorts each macro group into lexical order.
2147 Previously the output order depended on the order the files
2148 appeared in the directory, resulting in needless rewriting
2152 *) Patches to make OpenSSL compile under Win32 again. Mostly
2153 work arounds for the VC++ problem that it treats func() as
2154 func(void). Also stripped out the parts of mkdef.pl that
2155 added extra typesafe functions: these no longer exist.
2158 *) Reorganisation of the stack code. The macros are now all
2159 collected in safestack.h . Each macro is defined in terms of
2160 a "stack macro" of the form SKM_<name>(type, a, b). The
2161 DEBUG_SAFESTACK is now handled in terms of function casts,
2162 this has the advantage of retaining type safety without the
2163 use of additional functions. If DEBUG_SAFESTACK is not defined
2164 then the non typesafe macros are used instead. Also modified the
2165 mkstack.pl script to handle the new form. Needs testing to see
2166 if which (if any) compilers it chokes and maybe make DEBUG_SAFESTACK
2167 the default if no major problems. Similar behaviour for ASN1_SET_OF
2168 and PKCS12_STACK_OF.
2171 *) When some versions of IIS use the 'NET' form of private key the
2172 key derivation algorithm is different. Normally MD5(password) is
2173 used as a 128 bit RC4 key. In the modified case
2174 MD5(MD5(password) + "SGCKEYSALT") is used insted. Added some
2175 new functions i2d_RSA_NET(), d2i_RSA_NET() etc which are the same
2176 as the old Netscape_RSA functions except they have an additional
2177 'sgckey' parameter which uses the modified algorithm. Also added
2178 an -sgckey command line option to the rsa utility. Thanks to
2179 Adrian Peck <bertie@ncipher.com> for posting details of the modified
2180 algorithm to openssl-dev.
2183 *) The evp_local.h macros were using 'c.##kname' which resulted in
2184 invalid expansion on some systems (SCO 5.0.5 for example).
2185 Corrected to 'c.kname'.
2186 [Phillip Porch <root@theporch.com>]
2188 *) New X509_get1_email() and X509_REQ_get1_email() functions that return
2189 a STACK of email addresses from a certificate or request, these look
2190 in the subject name and the subject alternative name extensions and
2191 omit any duplicate addresses.
2194 *) Re-implement BN_mod_exp2_mont using independent (and larger) windows.
2195 This makes DSA verification about 2 % faster.
2198 *) Increase maximum window size in BN_mod_exp_... to 6 bits instead of 5
2199 (meaning that now 2^5 values will be precomputed, which is only 4 KB
2200 plus overhead for 1024 bit moduli).
2201 This makes exponentiations about 0.5 % faster for 1024 bit
2202 exponents (as measured by "openssl speed rsa2048").
2205 *) Rename memory handling macros to avoid conflicts with other
2207 Malloc => OPENSSL_malloc
2208 Malloc_locked => OPENSSL_malloc_locked
2209 Realloc => OPENSSL_realloc
2210 Free => OPENSSL_free
2213 *) New function BN_mod_exp_mont_word for small bases (roughly 15%
2214 faster than BN_mod_exp_mont, i.e. 7% for a full DH exchange).
2217 *) CygWin32 support.
2218 [John Jarvie <jjarvie@newsguy.com>]
2220 *) The type-safe stack code has been rejigged. It is now only compiled
2221 in when OpenSSL is configured with the DEBUG_SAFESTACK option and
2222 by default all type-specific stack functions are "#define"d back to
2223 standard stack functions. This results in more streamlined output
2224 but retains the type-safety checking possibilities of the original
2228 *) The STACK code has been cleaned up, and certain type declarations
2229 that didn't make a lot of sense have been brought in line. This has
2230 also involved a cleanup of sorts in safestack.h to more correctly
2231 map type-safe stack functions onto their plain stack counterparts.
2232 This work has also resulted in a variety of "const"ifications of
2233 lots of the code, especially "_cmp" operations which should normally
2234 be prototyped with "const" parameters anyway.
2237 *) When generating bytes for the first time in md_rand.c, 'stir the pool'
2238 by seeding with STATE_SIZE dummy bytes (with zero entropy count).
2239 (The PRNG state consists of two parts, the large pool 'state' and 'md',
2240 where all of 'md' is used each time the PRNG is used, but 'state'
2241 is used only indexed by a cyclic counter. As entropy may not be
2242 well distributed from the beginning, 'md' is important as a
2243 chaining variable. However, the output function chains only half
2244 of 'md', i.e. 80 bits. ssleay_rand_add, on the other hand, chains
2245 all of 'md', and seeding with STATE_SIZE dummy bytes will result
2246 in all of 'state' being rewritten, with the new values depending
2247 on virtually all of 'md'. This overcomes the 80 bit limitation.)
2250 *) In ssl/s2_clnt.c and ssl/s3_clnt.c, call ERR_clear_error() when
2251 the handshake is continued after ssl_verify_cert_chain();
2252 otherwise, if SSL_VERIFY_NONE is set, remaining error codes
2253 can lead to 'unexplainable' connection aborts later.
2254 [Bodo Moeller; problem tracked down by Lutz Jaenicke]
2256 *) Major EVP API cipher revision.
2257 Add hooks for extra EVP features. This allows various cipher
2258 parameters to be set in the EVP interface. Support added for variable
2259 key length ciphers via the EVP_CIPHER_CTX_set_key_length() function and
2260 setting of RC2 and RC5 parameters.
2262 Modify EVP_OpenInit() and EVP_SealInit() to cope with variable key length
2265 Remove lots of duplicated code from the EVP library. For example *every*
2266 cipher init() function handles the 'iv' in the same way according to the
2267 cipher mode. They also all do nothing if the 'key' parameter is NULL and
2268 for CFB and OFB modes they zero ctx->num.
2270 New functionality allows removal of S/MIME code RC2 hack.
2272 Most of the routines have the same form and so can be declared in terms
2275 By shifting this to the top level EVP_CipherInit() it can be removed from
2276 all individual ciphers. If the cipher wants to handle IVs or keys
2277 differently it can set the EVP_CIPH_CUSTOM_IV or EVP_CIPH_ALWAYS_CALL_INIT
2280 Change lots of functions like EVP_EncryptUpdate() to now return a
2281 value: although software versions of the algorithms cannot fail
2282 any installed hardware versions can.
2285 *) Implement SSL_OP_TLS_ROLLBACK_BUG: In ssl3_get_client_key_exchange, if
2286 this option is set, tolerate broken clients that send the negotiated
2287 protocol version number instead of the requested protocol version
2291 *) Call dh_tmp_cb (set by ..._TMP_DH_CB) with correct 'is_export' flag;
2292 i.e. non-zero for export ciphersuites, zero otherwise.
2293 Previous versions had this flag inverted, inconsistent with
2294 rsa_tmp_cb (..._TMP_RSA_CB).
2295 [Bodo Moeller; problem reported by Amit Chopra]
2297 *) Add missing DSA library text string. Work around for some IIS
2298 key files with invalid SEQUENCE encoding.
2301 *) Add a document (doc/standards.txt) that list all kinds of standards
2302 and so on that are implemented in OpenSSL.
2305 *) Enhance c_rehash script. Old version would mishandle certificates
2306 with the same subject name hash and wouldn't handle CRLs at all.
2307 Added -fingerprint option to crl utility, to support new c_rehash
2311 *) Eliminate non-ANSI declarations in crypto.h and stack.h.
2314 *) Fix for SSL server purpose checking. Server checking was
2315 rejecting certificates which had extended key usage present
2316 but no ssl client purpose.
2317 [Steve Henson, reported by Rene Grosser <grosser@hisolutions.com>]
2319 *) Make PKCS#12 code work with no password. The PKCS#12 spec
2320 is a little unclear about how a blank password is handled.
2321 Since the password in encoded as a BMPString with terminating
2322 double NULL a zero length password would end up as just the
2323 double NULL. However no password at all is different and is
2324 handled differently in the PKCS#12 key generation code. NS
2325 treats a blank password as zero length. MSIE treats it as no
2326 password on export: but it will try both on import. We now do
2327 the same: PKCS12_parse() tries zero length and no password if
2328 the password is set to "" or NULL (NULL is now a valid password:
2329 it wasn't before) as does the pkcs12 application.
2332 *) Bugfixes in apps/x509.c: Avoid a memory leak; and don't use
2333 perror when PEM_read_bio_X509_REQ fails, the error message must
2334 be obtained from the error queue.
2337 *) Avoid 'thread_hash' memory leak in crypto/err/err.c by freeing
2338 it in ERR_remove_state if appropriate, and change ERR_get_state
2339 accordingly to avoid race conditions (this is necessary because
2340 thread_hash is no longer constant once set).
2343 *) Bugfix for linux-elf makefile.one.
2346 *) RSA_get_default_method() will now cause a default
2347 RSA_METHOD to be chosen if one doesn't exist already.
2348 Previously this was only set during a call to RSA_new()
2349 or RSA_new_method(NULL) meaning it was possible for
2350 RSA_get_default_method() to return NULL.
2353 *) Added native name translation to the existing DSO code
2354 that will convert (if the flag to do so is set) filenames
2355 that are sufficiently small and have no path information
2356 into a canonical native form. Eg. "blah" converted to
2357 "libblah.so" or "blah.dll" etc.
2360 *) New function ERR_error_string_n(e, buf, len) which is like
2361 ERR_error_string(e, buf), but writes at most 'len' bytes
2362 including the 0 terminator. For ERR_error_string_n, 'buf'
2364 [Damien Miller <djm@mindrot.org>, Bodo Moeller]
2366 *) CONF library reworked to become more general. A new CONF
2367 configuration file reader "class" is implemented as well as a
2368 new functions (NCONF_*, for "New CONF") to handle it. The now
2369 old CONF_* functions are still there, but are reimplemented to
2370 work in terms of the new functions. Also, a set of functions
2371 to handle the internal storage of the configuration data is
2372 provided to make it easier to write new configuration file
2373 reader "classes" (I can definitely see something reading a
2374 configuration file in XML format, for example), called _CONF_*,
2375 or "the configuration storage API"...
2377 The new configuration file reading functions are:
2379 NCONF_new, NCONF_free, NCONF_load, NCONF_load_fp, NCONF_load_bio,
2380 NCONF_get_section, NCONF_get_string, NCONF_get_numbre
2382 NCONF_default, NCONF_WIN32
2384 NCONF_dump_fp, NCONF_dump_bio
2386 NCONF_default and NCONF_WIN32 are method (or "class") choosers,
2387 NCONF_new creates a new CONF object. This works in the same way
2388 as other interfaces in OpenSSL, like the BIO interface.
2389 NCONF_dump_* dump the internal storage of the configuration file,
2390 which is useful for debugging. All other functions take the same
2391 arguments as the old CONF_* functions wth the exception of the
2392 first that must be a `CONF *' instead of a `LHASH *'.
2394 To make it easer to use the new classes with the old CONF_* functions,
2395 the function CONF_set_default_method is provided.
2398 *) Add '-tls1' option to 'openssl ciphers', which was already
2399 mentioned in the documentation but had not been implemented.
2400 (This option is not yet really useful because even the additional
2401 experimental TLS 1.0 ciphers are currently treated as SSL 3.0 ciphers.)
2404 *) Initial DSO code added into libcrypto for letting OpenSSL (and
2405 OpenSSL-based applications) load shared libraries and bind to
2406 them in a portable way.
2407 [Geoff Thorpe, with contributions from Richard Levitte]
2409 Changes between 0.9.5 and 0.9.5a [1 Apr 2000]
2411 *) Make sure _lrotl and _lrotr are only used with MSVC.
2413 *) Use lock CRYPTO_LOCK_RAND correctly in ssleay_rand_status
2414 (the default implementation of RAND_status).
2416 *) Rename openssl x509 option '-crlext', which was added in 0.9.5,
2417 to '-clrext' (= clear extensions), as intended and documented.
2418 [Bodo Moeller; inconsistency pointed out by Michael Attili
2421 *) Fix for HMAC. It wasn't zeroing the rest of the block if the key length
2422 was larger than the MD block size.
2423 [Steve Henson, pointed out by Yost William <YostW@tce.com>]
2425 *) Modernise PKCS12_parse() so it uses STACK_OF(X509) for its ca argument
2426 fix a leak when the ca argument was passed as NULL. Stop X509_PUBKEY_set()
2427 using the passed key: if the passed key was a private key the result
2428 of X509_print(), for example, would be to print out all the private key
2432 *) des_quad_cksum() byte order bug fix.
2433 [Ulf Möller, using the problem description in krb4-0.9.7, where
2434 the solution is attributed to Derrick J Brashear <shadow@DEMENTIA.ORG>]
2436 *) Fix so V_ASN1_APP_CHOOSE works again: however its use is strongly
2438 [Steve Henson, pointed out by Brian Korver <briank@cs.stanford.edu>]
2440 *) For easily testing in shell scripts whether some command
2441 'openssl XXX' exists, the new pseudo-command 'openssl no-XXX'
2442 returns with exit code 0 iff no command of the given name is available.
2443 'no-XXX' is printed in this case, 'XXX' otherwise. In both cases,
2444 the output goes to stdout and nothing is printed to stderr.
2445 Additional arguments are always ignored.
2447 Since for each cipher there is a command of the same name,
2448 the 'no-cipher' compilation switches can be tested this way.
2450 ('openssl no-XXX' is not able to detect pseudo-commands such
2451 as 'quit', 'list-XXX-commands', or 'no-XXX' itself.)
2454 *) Update test suite so that 'make test' succeeds in 'no-rsa' configuration.
2457 *) For SSL_[CTX_]set_tmp_dh, don't create a DH key if SSL_OP_SINGLE_DH_USE
2458 is set; it will be thrown away anyway because each handshake creates
2460 ssl_cert_dup, which is used by SSL_new, now copies DH keys in addition
2461 to parameters -- in previous versions (since OpenSSL 0.9.3) the
2462 'default key' from SSL_CTX_set_tmp_dh would always be lost, meanining
2463 you effectivly got SSL_OP_SINGLE_DH_USE when using this macro.
2466 *) New s_client option -ign_eof: EOF at stdin is ignored, and
2467 'Q' and 'R' lose their special meanings (quit/renegotiate).
2468 This is part of what -quiet does; unlike -quiet, -ign_eof
2469 does not suppress any output.
2472 *) Add compatibility options to the purpose and trust code. The
2473 purpose X509_PURPOSE_ANY is "any purpose" which automatically
2474 accepts a certificate or CA, this was the previous behaviour,
2475 with all the associated security issues.
2477 X509_TRUST_COMPAT is the old trust behaviour: only and
2478 automatically trust self signed roots in certificate store. A
2479 new trust setting X509_TRUST_DEFAULT is used to specify that
2480 a purpose has no associated trust setting and it should instead
2481 use the value in the default purpose.
2484 *) Fix the PKCS#8 DSA private key code so it decodes keys again
2485 and fix a memory leak.
2488 *) In util/mkerr.pl (which implements 'make errors'), preserve
2489 reason strings from the previous version of the .c file, as
2490 the default to have only downcase letters (and digits) in
2491 automatically generated reasons codes is not always appropriate.
2494 *) In ERR_load_ERR_strings(), build an ERR_LIB_SYS error reason table
2495 using strerror. Previously, ERR_reason_error_string() returned
2496 library names as reason strings for SYSerr; but SYSerr is a special
2497 case where small numbers are errno values, not library numbers.
2500 *) Add '-dsaparam' option to 'openssl dhparam' application. This
2501 converts DSA parameters into DH parameters. (When creating parameters,
2502 DSA_generate_parameters is used.)
2505 *) Include 'length' (recommended exponent length) in C code generated
2506 by 'openssl dhparam -C'.
2509 *) The second argument to set_label in perlasm was already being used
2510 so couldn't be used as a "file scope" flag. Moved to third argument
2514 *) In PEM_ASN1_write_bio and some other functions, use RAND_pseudo_bytes
2515 instead of RAND_bytes for encryption IVs and salts.
2518 *) Include RAND_status() into RAND_METHOD instead of implementing
2519 it only for md_rand.c Otherwise replacing the PRNG by calling
2520 RAND_set_rand_method would be impossible.
2523 *) Don't let DSA_generate_key() enter an infinite loop if the random
2524 number generation fails.
2527 *) New 'rand' application for creating pseudo-random output.
2530 *) Added configuration support for Linux/IA64
2531 [Rolf Haberrecker <rolf@suse.de>]
2533 *) Assembler module support for Mingw32.
2536 *) Shared library support for HPUX (in shlib/).
2537 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE> and Anonymous]
2539 *) Shared library support for Solaris gcc.
2540 [Lutz Behnke <behnke@trustcenter.de>]
2542 Changes between 0.9.4 and 0.9.5 [28 Feb 2000]
2544 *) PKCS7_encrypt() was adding text MIME headers twice because they
2545 were added manually and by SMIME_crlf_copy().
2548 *) In bntest.c don't call BN_rand with zero bits argument.
2549 [Steve Henson, pointed out by Andrew W. Gray <agray@iconsinc.com>]
2551 *) BN_mul bugfix: In bn_mul_part_recursion() only the a>a[n] && b>b[n]
2552 case was implemented. This caused BN_div_recp() to fail occasionally.
2555 *) Add an optional second argument to the set_label() in the perl
2556 assembly language builder. If this argument exists and is set
2557 to 1 it signals that the assembler should use a symbol whose
2558 scope is the entire file, not just the current function. This
2559 is needed with MASM which uses the format label:: for this scope.
2560 [Steve Henson, pointed out by Peter Runestig <peter@runestig.com>]
2562 *) Change the ASN1 types so they are typedefs by default. Before
2563 almost all types were #define'd to ASN1_STRING which was causing
2564 STACK_OF() problems: you couldn't declare STACK_OF(ASN1_UTF8STRING)
2568 *) Change names of new functions to the new get1/get0 naming
2569 convention: After 'get1', the caller owns a reference count
2570 and has to call ..._free; 'get0' returns a pointer to some
2571 data structure without incrementing reference counters.
2572 (Some of the existing 'get' functions increment a reference
2573 counter, some don't.)
2574 Similarly, 'set1' and 'add1' functions increase reference
2575 counters or duplicate objects.
2578 *) Allow for the possibility of temp RSA key generation failure:
2579 the code used to assume it always worked and crashed on failure.
2582 *) Fix potential buffer overrun problem in BIO_printf().
2583 [Ulf Möller, using public domain code by Patrick Powell; problem
2584 pointed out by David Sacerdote <das33@cornell.edu>]
2586 *) Support EGD <http://www.lothar.com/tech/crypto/>. New functions
2587 RAND_egd() and RAND_status(). In the command line application,
2588 the EGD socket can be specified like a seed file using RANDFILE
2592 *) Allow the string CERTIFICATE to be tolerated in PKCS#7 structures.
2593 Some CAs (e.g. Verisign) distribute certificates in this form.
2596 *) Remove the SSL_ALLOW_ADH compile option and set the default cipher
2597 list to exclude them. This means that no special compilation option
2598 is needed to use anonymous DH: it just needs to be included in the
2602 *) Change the EVP_MD_CTX_type macro so its meaning consistent with
2603 EVP_MD_type. The old functionality is available in a new macro called
2604 EVP_MD_md(). Change code that uses it and update docs.
2607 *) ..._ctrl functions now have corresponding ..._callback_ctrl functions
2608 where the 'void *' argument is replaced by a function pointer argument.
2609 Previously 'void *' was abused to point to functions, which works on
2610 many platforms, but is not correct. As these functions are usually
2611 called by macros defined in OpenSSL header files, most source code
2612 should work without changes.
2615 *) <openssl/opensslconf.h> (which is created by Configure) now contains
2616 sections with information on -D... compiler switches used for
2617 compiling the library so that applications can see them. To enable
2618 one of these sections, a pre-processor symbol OPENSSL_..._DEFINES
2619 must be defined. E.g.,
2620 #define OPENSSL_ALGORITHM_DEFINES
2621 #include <openssl/opensslconf.h>
2622 defines all pertinent NO_<algo> symbols, such as NO_IDEA, NO_RSA, etc.
2623 [Richard Levitte, Ulf and Bodo Möller]
2625 *) Bugfix: Tolerate fragmentation and interleaving in the SSL 3/TLS
2629 *) Change the 'other' type in certificate aux info to a STACK_OF
2630 X509_ALGOR. Although not an AlgorithmIdentifier as such it has
2631 the required ASN1 format: arbitrary types determined by an OID.
2634 *) Add some PEM_write_X509_REQ_NEW() functions and a command line
2635 argument to 'req'. This is not because the function is newer or
2636 better than others it just uses the work 'NEW' in the certificate
2637 request header lines. Some software needs this.
2640 *) Reorganise password command line arguments: now passwords can be
2641 obtained from various sources. Delete the PEM_cb function and make
2642 it the default behaviour: i.e. if the callback is NULL and the
2643 usrdata argument is not NULL interpret it as a null terminated pass
2644 phrase. If usrdata and the callback are NULL then the pass phrase
2645 is prompted for as usual.
2648 *) Add support for the Compaq Atalla crypto accelerator. If it is installed,
2649 the support is automatically enabled. The resulting binaries will
2650 autodetect the card and use it if present.
2651 [Ben Laurie and Compaq Inc.]
2653 *) Work around for Netscape hang bug. This sends certificate request
2654 and server done in one record. Since this is perfectly legal in the
2655 SSL/TLS protocol it isn't a "bug" option and is on by default. See
2656 the bugs/SSLv3 entry for more info.
2659 *) HP-UX tune-up: new unified configs, HP C compiler bug workaround.
2662 *) Add -rand argument to smime and pkcs12 applications and read/write
2666 *) New 'passwd' tool for crypt(3) and apr1 password hashes.
2669 *) Add command line password options to the remaining applications.
2672 *) Bug fix for BN_div_recp() for numerators with an even number of
2676 *) More tests in bntest.c, and changed test_bn output.
2679 *) ./config recognizes MacOS X now.
2682 *) Bug fix for BN_div() when the first words of num and divsor are
2683 equal (it gave wrong results if (rem=(n1-q*d0)&BN_MASK2) < d0).
2686 *) Add support for various broken PKCS#8 formats, and command line
2687 options to produce them.
2690 *) New functions BN_CTX_start(), BN_CTX_get() and BT_CTX_end() to
2691 get temporary BIGNUMs from a BN_CTX.
2694 *) Correct return values in BN_mod_exp_mont() and BN_mod_exp2_mont()
2698 *) Change the SSLeay_add_all_*() functions to OpenSSL_add_all_*() and
2699 include a #define from the old name to the new. The original intent
2700 was that statically linked binaries could for example just call
2701 SSLeay_add_all_ciphers() to just add ciphers to the table and not
2702 link with digests. This never worked becayse SSLeay_add_all_digests()
2703 and SSLeay_add_all_ciphers() were in the same source file so calling
2704 one would link with the other. They are now in separate source files.
2707 *) Add a new -notext option to 'ca' and a -pubkey option to 'spkac'.
2710 *) Use a less unusual form of the Miller-Rabin primality test (it used
2711 a binary algorithm for exponentiation integrated into the Miller-Rabin
2712 loop, our standard modexp algorithms are faster).
2715 *) Support for the EBCDIC character set completed.
2716 [Martin Kraemer <Martin.Kraemer@Mch.SNI.De>]
2718 *) Source code cleanups: use const where appropriate, eliminate casts,
2719 use void * instead of char * in lhash.
2722 *) Bugfix: ssl3_send_server_key_exchange was not restartable
2723 (the state was not changed to SSL3_ST_SW_KEY_EXCH_B, and because of
2724 this the server could overwrite ephemeral keys that the client
2728 *) Turn DSA_is_prime into a macro that calls BN_is_prime,
2729 using 50 iterations of the Rabin-Miller test.
2731 DSA_generate_parameters now uses BN_is_prime_fasttest (with 50
2732 iterations of the Rabin-Miller test as required by the appendix
2733 to FIPS PUB 186[-1]) instead of DSA_is_prime.
2734 As BN_is_prime_fasttest includes trial division, DSA parameter
2735 generation becomes much faster.
2737 This implies a change for the callback functions in DSA_is_prime
2738 and DSA_generate_parameters: The callback function is called once
2739 for each positive witness in the Rabin-Miller test, not just
2740 occasionally in the inner loop; and the parameters to the
2741 callback function now provide an iteration count for the outer
2742 loop rather than for the current invocation of the inner loop.
2743 DSA_generate_parameters additionally can call the callback
2744 function with an 'iteration count' of -1, meaning that a
2745 candidate has passed the trial division test (when q is generated
2746 from an application-provided seed, trial division is skipped).
2749 *) New function BN_is_prime_fasttest that optionally does trial
2750 division before starting the Rabin-Miller test and has
2751 an additional BN_CTX * argument (whereas BN_is_prime always
2752 has to allocate at least one BN_CTX).
2753 'callback(1, -1, cb_arg)' is called when a number has passed the
2754 trial division stage.
2757 *) Fix for bug in CRL encoding. The validity dates weren't being handled
2761 *) New -pkcs12 option to CA.pl script to write out a PKCS#12 file.
2764 *) New function BN_pseudo_rand().
2767 *) Clean up BN_mod_mul_montgomery(): replace the broken (and unreadable)
2768 bignum version of BN_from_montgomery() with the working code from
2769 SSLeay 0.9.0 (the word based version is faster anyway), and clean up
2773 *) Avoid a race condition in s2_clnt.c (function get_server_hello) that
2774 made it impossible to use the same SSL_SESSION data structure in
2775 SSL2 clients in multiple threads.
2778 *) The return value of RAND_load_file() no longer counts bytes obtained
2779 by stat(). RAND_load_file(..., -1) is new and uses the complete file
2780 to seed the PRNG (previously an explicit byte count was required).
2781 [Ulf Möller, Bodo Möller]
2783 *) Clean up CRYPTO_EX_DATA functions, some of these didn't have prototypes
2784 used (char *) instead of (void *) and had casts all over the place.
2787 *) Make BN_generate_prime() return NULL on error if ret!=NULL.
2790 *) Retain source code compatibility for BN_prime_checks macro:
2791 BN_is_prime(..., BN_prime_checks, ...) now uses
2792 BN_prime_checks_for_size to determine the appropriate number of
2793 Rabin-Miller iterations.
2796 *) Diffie-Hellman uses "safe" primes: DH_check() return code renamed to
2797 DH_CHECK_P_NOT_SAFE_PRIME.
2798 (Check if this is true? OpenPGP calls them "strong".)
2801 *) Merge the functionality of "dh" and "gendh" programs into a new program
2802 "dhparam". The old programs are retained for now but will handle DH keys
2803 (instead of parameters) in future.
2806 *) Make the ciphers, s_server and s_client programs check the return values
2807 when a new cipher list is set.
2810 *) Enhance the SSL/TLS cipher mechanism to correctly handle the TLS 56bit
2811 ciphers. Before when the 56bit ciphers were enabled the sorting was
2814 The syntax for the cipher sorting has been extended to support sorting by
2815 cipher-strength (using the strength_bits hard coded in the tables).
2816 The new command is "@STRENGTH" (see also doc/apps/ciphers.pod).
2818 Fix a bug in the cipher-command parser: when supplying a cipher command
2819 string with an "undefined" symbol (neither command nor alphanumeric
2820 [A-Za-z0-9], ssl_set_cipher_list used to hang in an endless loop. Now
2821 an error is flagged.
2823 Due to the strength-sorting extension, the code of the
2824 ssl_create_cipher_list() function was completely rearranged. I hope that
2825 the readability was also increased :-)
2826 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>]
2828 *) Minor change to 'x509' utility. The -CAcreateserial option now uses 1
2829 for the first serial number and places 2 in the serial number file. This
2830 avoids problems when the root CA is created with serial number zero and
2831 the first user certificate has the same issuer name and serial number
2835 *) Fixes to X509_ATTRIBUTE utilities, change the 'req' program so it uses
2836 the new code. Add documentation for this stuff.
2839 *) Changes to X509_ATTRIBUTE utilities. These have been renamed from
2840 X509_*() to X509at_*() on the grounds that they don't handle X509
2841 structures and behave in an analagous way to the X509v3 functions:
2842 they shouldn't be called directly but wrapper functions should be used
2845 So we also now have some wrapper functions that call the X509at functions
2846 when passed certificate requests. (TO DO: similar things can be done with
2847 PKCS#7 signed and unsigned attributes, PKCS#12 attributes and a few other
2848 things. Some of these need some d2i or i2d and print functionality
2849 because they handle more complex structures.)
2852 *) Add missing #ifndefs that caused missing symbols when building libssl
2853 as a shared library without RSA. Use #ifndef NO_SSL2 instead of
2854 NO_RSA in ssl/s2*.c.
2855 [Kris Kennaway <kris@hub.freebsd.org>, modified by Ulf Möller]
2857 *) Precautions against using the PRNG uninitialized: RAND_bytes() now
2858 has a return value which indicates the quality of the random data
2859 (1 = ok, 0 = not seeded). Also an error is recorded on the thread's
2860 error queue. New function RAND_pseudo_bytes() generates output that is
2861 guaranteed to be unique but not unpredictable. RAND_add is like
2862 RAND_seed, but takes an extra argument for an entropy estimate
2863 (RAND_seed always assumes full entropy).
2866 *) Do more iterations of Rabin-Miller probable prime test (specifically,
2867 3 for 1024-bit primes, 6 for 512-bit primes, 12 for 256-bit primes
2868 instead of only 2 for all lengths; see BN_prime_checks_for_size definition
2869 in crypto/bn/bn_prime.c for the complete table). This guarantees a
2870 false-positive rate of at most 2^-80 for random input.
2873 *) Rewrite ssl3_read_n (ssl/s3_pkt.c) avoiding a couple of bugs.
2876 *) New function X509_CTX_rget_chain() (renamed to X509_CTX_get1_chain
2877 in the 0.9.5 release), this returns the chain
2878 from an X509_CTX structure with a dup of the stack and all
2879 the X509 reference counts upped: so the stack will exist
2880 after X509_CTX_cleanup() has been called. Modify pkcs12.c
2883 Also make SSL_SESSION_print() print out the verify return
2887 *) Add manpage for the pkcs12 command. Also change the default
2888 behaviour so MAC iteration counts are used unless the new
2889 -nomaciter option is used. This improves file security and
2890 only older versions of MSIE (4.0 for example) need it.
2893 *) Honor the no-xxx Configure options when creating .DEF files.
2896 *) Add PKCS#10 attributes to field table: challengePassword,
2897 unstructuredName and unstructuredAddress. These are taken from
2898 draft PKCS#9 v2.0 but are compatible with v1.2 provided no
2899 international characters are used.
2901 More changes to X509_ATTRIBUTE code: allow the setting of types
2902 based on strings. Remove the 'loc' parameter when adding
2903 attributes because these will be a SET OF encoding which is sorted
2907 *) Initial changes to the 'req' utility to allow request generation
2908 automation. This will allow an application to just generate a template
2909 file containing all the field values and have req construct the
2912 Initial support for X509_ATTRIBUTE handling. Stacks of these are
2913 used all over the place including certificate requests and PKCS#7
2914 structures. They are currently handled manually where necessary with
2915 some primitive wrappers for PKCS#7. The new functions behave in a
2916 manner analogous to the X509 extension functions: they allow
2917 attributes to be looked up by NID and added.
2919 Later something similar to the X509V3 code would be desirable to
2920 automatically handle the encoding, decoding and printing of the
2921 more complex types. The string types like challengePassword can
2922 be handled by the string table functions.
2924 Also modified the multi byte string table handling. Now there is
2925 a 'global mask' which masks out certain types. The table itself
2926 can use the flag STABLE_NO_MASK to ignore the mask setting: this
2927 is useful when for example there is only one permissible type
2928 (as in countryName) and using the mask might result in no valid
2932 *) Clean up 'Finished' handling, and add functions SSL_get_finished and
2933 SSL_get_peer_finished to allow applications to obtain the latest
2934 Finished messages sent to the peer or expected from the peer,
2935 respectively. (SSL_get_peer_finished is usually the Finished message
2936 actually received from the peer, otherwise the protocol will be aborted.)
2938 As the Finished message are message digests of the complete handshake
2939 (with a total of 192 bits for TLS 1.0 and more for SSL 3.0), they can
2940 be used for external authentication procedures when the authentication
2941 provided by SSL/TLS is not desired or is not enough.
2944 *) Enhanced support for Alpha Linux is added. Now ./config checks if
2945 the host supports BWX extension and if Compaq C is present on the
2946 $PATH. Just exploiting of the BWX extension results in 20-30%
2947 performance kick for some algorithms, e.g. DES and RC4 to mention
2948 a couple. Compaq C in turn generates ~20% faster code for MD5 and
2952 *) Add support for MS "fast SGC". This is arguably a violation of the
2953 SSL3/TLS protocol. Netscape SGC does two handshakes: the first with
2954 weak crypto and after checking the certificate is SGC a second one
2955 with strong crypto. MS SGC stops the first handshake after receiving
2956 the server certificate message and sends a second client hello. Since
2957 a server will typically do all the time consuming operations before
2958 expecting any further messages from the client (server key exchange
2959 is the most expensive) there is little difference between the two.
2961 To get OpenSSL to support MS SGC we have to permit a second client
2962 hello message after we have sent server done. In addition we have to
2963 reset the MAC if we do get this second client hello.
2966 *) Add a function 'd2i_AutoPrivateKey()' this will automatically decide
2967 if a DER encoded private key is RSA or DSA traditional format. Changed
2968 d2i_PrivateKey_bio() to use it. This is only needed for the "traditional"
2969 format DER encoded private key. Newer code should use PKCS#8 format which
2970 has the key type encoded in the ASN1 structure. Added DER private key
2971 support to pkcs8 application.
2974 *) SSL 3/TLS 1 servers now don't request certificates when an anonymous
2975 ciphersuites has been selected (as required by the SSL 3/TLS 1
2976 specifications). Exception: When SSL_VERIFY_FAIL_IF_NO_PEER_CERT
2977 is set, we interpret this as a request to violate the specification
2978 (the worst that can happen is a handshake failure, and 'correct'
2979 behaviour would result in a handshake failure anyway).
2982 *) In SSL_CTX_add_session, take into account that there might be multiple
2983 SSL_SESSION structures with the same session ID (e.g. when two threads
2984 concurrently obtain them from an external cache).
2985 The internal cache can handle only one SSL_SESSION with a given ID,
2986 so if there's a conflict, we now throw out the old one to achieve
2990 *) Add OIDs for idea and blowfish in CBC mode. This will allow both
2991 to be used in PKCS#5 v2.0 and S/MIME. Also add checking to
2992 some routines that use cipher OIDs: some ciphers do not have OIDs
2993 defined and so they cannot be used for S/MIME and PKCS#5 v2.0 for
2997 *) Simplify the trust setting structure and code. Now we just have
2998 two sequences of OIDs for trusted and rejected settings. These will
2999 typically have values the same as the extended key usage extension
3000 and any application specific purposes.
3002 The trust checking code now has a default behaviour: it will just
3003 check for an object with the same NID as the passed id. Functions can
3004 be provided to override either the default behaviour or the behaviour
3005 for a given id. SSL client, server and email already have functions
3006 in place for compatibility: they check the NID and also return "trusted"
3007 if the certificate is self signed.
3010 *) Add d2i,i2d bio/fp functions for PrivateKey: these convert the
3011 traditional format into an EVP_PKEY structure.
3014 *) Add a password callback function PEM_cb() which either prompts for
3015 a password if usr_data is NULL or otherwise assumes it is a null
3016 terminated password. Allow passwords to be passed on command line
3017 environment or config files in a few more utilities.
3020 *) Add a bunch of DER and PEM functions to handle PKCS#8 format private
3021 keys. Add some short names for PKCS#8 PBE algorithms and allow them
3022 to be specified on the command line for the pkcs8 and pkcs12 utilities.
3023 Update documentation.
3026 *) Support for ASN1 "NULL" type. This could be handled before by using
3027 ASN1_TYPE but there wasn't any function that would try to read a NULL
3028 and produce an error if it couldn't. For compatibility we also have
3029 ASN1_NULL_new() and ASN1_NULL_free() functions but these are faked and
3030 don't allocate anything because they don't need to.
3033 *) Initial support for MacOS is now provided. Examine INSTALL.MacOS
3035 [Andy Polyakov, Roy Woods <roy@centicsystems.ca>]
3037 *) Rebuild of the memory allocation routines used by OpenSSL code and
3038 possibly others as well. The purpose is to make an interface that
3039 provide hooks so anyone can build a separate set of allocation and
3040 deallocation routines to be used by OpenSSL, for example memory
3041 pool implementations, or something else, which was previously hard
3042 since Malloc(), Realloc() and Free() were defined as macros having
3043 the values malloc, realloc and free, respectively (except for Win32
3044 compilations). The same is provided for memory debugging code.
3045 OpenSSL already comes with functionality to find memory leaks, but
3046 this gives people a chance to debug other memory problems.
3048 With these changes, a new set of functions and macros have appeared:
3050 CRYPTO_set_mem_debug_functions() [F]
3051 CRYPTO_get_mem_debug_functions() [F]
3052 CRYPTO_dbg_set_options() [F]
3053 CRYPTO_dbg_get_options() [F]
3054 CRYPTO_malloc_debug_init() [M]
3056 The memory debug functions are NULL by default, unless the library
3057 is compiled with CRYPTO_MDEBUG or friends is defined. If someone
3058 wants to debug memory anyway, CRYPTO_malloc_debug_init() (which
3059 gives the standard debugging functions that come with OpenSSL) or
3060 CRYPTO_set_mem_debug_functions() (tells OpenSSL to use functions
3061 provided by the library user) must be used. When the standard
3062 debugging functions are used, CRYPTO_dbg_set_options can be used to
3063 request additional information:
3064 CRYPTO_dbg_set_options(V_CYRPTO_MDEBUG_xxx) corresponds to setting
3065 the CRYPTO_MDEBUG_xxx macro when compiling the library.
3067 Also, things like CRYPTO_set_mem_functions will always give the
3068 expected result (the new set of functions is used for allocation
3069 and deallocation) at all times, regardless of platform and compiler
3072 To finish it up, some functions that were never use in any other
3073 way than through macros have a new API and new semantic:
3076 CRYPTO_dbg_realloc()
3079 All macros of value have retained their old syntax.
3080 [Richard Levitte and Bodo Moeller]
3082 *) Some S/MIME fixes. The OID for SMIMECapabilities was wrong, the
3083 ordering of SMIMECapabilities wasn't in "strength order" and there
3084 was a missing NULL in the AlgorithmIdentifier for the SHA1 signature
3088 *) Some ASN1 types with illegal zero length encoding (INTEGER,
3089 ENUMERATED and OBJECT IDENTIFIER) choked the ASN1 routines.
3090 [Frans Heymans <fheymans@isaserver.be>, modified by Steve Henson]
3092 *) Merge in my S/MIME library for OpenSSL. This provides a simple
3093 S/MIME API on top of the PKCS#7 code, a MIME parser (with enough
3094 functionality to handle multipart/signed properly) and a utility
3095 called 'smime' to call all this stuff. This is based on code I
3096 originally wrote for Celo who have kindly allowed it to be
3097 included in OpenSSL.
3100 *) Add variants des_set_key_checked and des_set_key_unchecked of
3101 des_set_key (aka des_key_sched). Global variable des_check_key
3102 decides which of these is called by des_set_key; this way
3103 des_check_key behaves as it always did, but applications and
3104 the library itself, which was buggy for des_check_key == 1,
3105 have a cleaner way to pick the version they need.
3108 *) New function PKCS12_newpass() which changes the password of a
3112 *) Modify X509_TRUST and X509_PURPOSE so it also uses a static and
3113 dynamic mix. In both cases the ids can be used as an index into the
3114 table. Also modified the X509_TRUST_add() and X509_PURPOSE_add()
3115 functions so they accept a list of the field values and the
3116 application doesn't need to directly manipulate the X509_TRUST
3120 *) Modify the ASN1_STRING_TABLE stuff so it also uses bsearch and doesn't
3124 *) Modify the way the V3 extension code looks up extensions. This now
3125 works in a similar way to the object code: we have some "standard"
3126 extensions in a static table which is searched with OBJ_bsearch()
3127 and the application can add dynamic ones if needed. The file
3128 crypto/x509v3/ext_dat.h now has the info: this file needs to be
3129 updated whenever a new extension is added to the core code and kept
3130 in ext_nid order. There is a simple program 'tabtest.c' which checks
3131 this. New extensions are not added too often so this file can readily
3132 be maintained manually.
3134 There are two big advantages in doing things this way. The extensions
3135 can be looked up immediately and no longer need to be "added" using
3136 X509V3_add_standard_extensions(): this function now does nothing.
3137 [Side note: I get *lots* of email saying the extension code doesn't
3138 work because people forget to call this function]
3139 Also no dynamic allocation is done unless new extensions are added:
3140 so if we don't add custom extensions there is no need to call
3141 X509V3_EXT_cleanup().
3144 *) Modify enc utility's salting as follows: make salting the default. Add a
3145 magic header, so unsalted files fail gracefully instead of just decrypting
3146 to garbage. This is because not salting is a big security hole, so people
3147 should be discouraged from doing it.
3150 *) Fixes and enhancements to the 'x509' utility. It allowed a message
3151 digest to be passed on the command line but it only used this
3152 parameter when signing a certificate. Modified so all relevant
3153 operations are affected by the digest parameter including the
3154 -fingerprint and -x509toreq options. Also -x509toreq choked if a
3155 DSA key was used because it didn't fix the digest.
3158 *) Initial certificate chain verify code. Currently tests the untrusted
3159 certificates for consistency with the verify purpose (which is set
3160 when the X509_STORE_CTX structure is set up) and checks the pathlength.
3162 There is a NO_CHAIN_VERIFY compilation option to keep the old behaviour:
3163 this is because it will reject chains with invalid extensions whereas
3164 every previous version of OpenSSL and SSLeay made no checks at all.
3166 Trust code: checks the root CA for the relevant trust settings. Trust
3167 settings have an initial value consistent with the verify purpose: e.g.
3168 if the verify purpose is for SSL client use it expects the CA to be
3169 trusted for SSL client use. However the default value can be changed to
3170 permit custom trust settings: one example of this would be to only trust
3171 certificates from a specific "secure" set of CAs.
3173 Also added X509_STORE_CTX_new() and X509_STORE_CTX_free() functions
3174 which should be used for version portability: especially since the
3175 verify structure is likely to change more often now.
3177 SSL integration. Add purpose and trust to SSL_CTX and SSL and functions
3178 to set them. If not set then assume SSL clients will verify SSL servers
3181 Two new options to the verify program: -untrusted allows a set of
3182 untrusted certificates to be passed in and -purpose which sets the
3183 intended purpose of the certificate. If a purpose is set then the
3184 new chain verify code is used to check extension consistency.
3187 *) Support for the authority information access extension.
3190 *) Modify RSA and DSA PEM read routines to transparently handle
3191 PKCS#8 format private keys. New *_PUBKEY_* functions that handle
3192 public keys in a format compatible with certificate
3193 SubjectPublicKeyInfo structures. Unfortunately there were already
3194 functions called *_PublicKey_* which used various odd formats so
3195 these are retained for compatibility: however the DSA variants were
3196 never in a public release so they have been deleted. Changed dsa/rsa
3197 utilities to handle the new format: note no releases ever handled public
3198 keys so we should be OK.
3200 The primary motivation for this change is to avoid the same fiasco
3201 that dogs private keys: there are several incompatible private key
3202 formats some of which are standard and some OpenSSL specific and
3203 require various evil hacks to allow partial transparent handling and
3204 even then it doesn't work with DER formats. Given the option anything
3205 other than PKCS#8 should be dumped: but the other formats have to
3206 stay in the name of compatibility.
3208 With public keys and the benefit of hindsight one standard format
3209 is used which works with EVP_PKEY, RSA or DSA structures: though
3210 it clearly returns an error if you try to read the wrong kind of key.
3212 Added a -pubkey option to the 'x509' utility to output the public key.
3213 Also rename the EVP_PKEY_get_*() to EVP_PKEY_rget_*()
3214 (renamed to EVP_PKEY_get1_*() in the OpenSSL 0.9.5 release) and add
3215 EVP_PKEY_rset_*() functions (renamed to EVP_PKEY_set1_*())
3216 that do the same as the EVP_PKEY_assign_*() except they up the
3217 reference count of the added key (they don't "swallow" the
3221 *) Fixes to crypto/x509/by_file.c the code to read in certificates and
3222 CRLs would fail if the file contained no certificates or no CRLs:
3223 added a new function to read in both types and return the number
3224 read: this means that if none are read it will be an error. The
3225 DER versions of the certificate and CRL reader would always fail
3226 because it isn't possible to mix certificates and CRLs in DER format
3227 without choking one or the other routine. Changed this to just read
3228 a certificate: this is the best we can do. Also modified the code
3229 in apps/verify.c to take notice of return codes: it was previously
3230 attempting to read in certificates from NULL pointers and ignoring
3231 any errors: this is one reason why the cert and CRL reader seemed
3232 to work. It doesn't check return codes from the default certificate
3233 routines: these may well fail if the certificates aren't installed.
3236 *) Code to support otherName option in GeneralName.
3239 *) First update to verify code. Change the verify utility
3240 so it warns if it is passed a self signed certificate:
3241 for consistency with the normal behaviour. X509_verify
3242 has been modified to it will now verify a self signed
3243 certificate if *exactly* the same certificate appears
3244 in the store: it was previously impossible to trust a
3245 single self signed certificate. This means that:
3246 openssl verify ss.pem
3247 now gives a warning about a self signed certificate but
3248 openssl verify -CAfile ss.pem ss.pem
3252 *) For servers, store verify_result in SSL_SESSION data structure
3253 (and add it to external session representation).
3254 This is needed when client certificate verifications fails,
3255 but an application-provided verification callback (set by
3256 SSL_CTX_set_cert_verify_callback) allows accepting the session
3257 anyway (i.e. leaves x509_store_ctx->error != X509_V_OK
3258 but returns 1): When the session is reused, we have to set
3259 ssl->verify_result to the appropriate error code to avoid
3261 [Bodo Moeller, problem pointed out by Lutz Jaenicke]
3263 *) Fix a bug in the new PKCS#7 code: it didn't consider the
3264 case in PKCS7_dataInit() where the signed PKCS7 structure
3265 didn't contain any existing data because it was being created.
3266 [Po-Cheng Chen <pocheng@nst.com.tw>, slightly modified by Steve Henson]
3268 *) Add a salt to the key derivation routines in enc.c. This
3269 forms the first 8 bytes of the encrypted file. Also add a
3270 -S option to allow a salt to be input on the command line.
3273 *) New function X509_cmp(). Oddly enough there wasn't a function
3274 to compare two certificates. We do this by working out the SHA1
3275 hash and comparing that. X509_cmp() will be needed by the trust
3279 *) SSL_get1_session() is like SSL_get_session(), but increments
3280 the reference count in the SSL_SESSION returned.
3281 [Geoff Thorpe <geoff@eu.c2.net>]
3283 *) Fix for 'req': it was adding a null to request attributes.
3284 Also change the X509_LOOKUP and X509_INFO code to handle
3285 certificate auxiliary information.
3288 *) Add support for 40 and 64 bit RC2 and RC4 algorithms: document
3292 *) Add the possibility to add extra information to the memory leak
3293 detecting output, to form tracebacks, showing from where each
3294 allocation was originated: CRYPTO_push_info("constant string") adds
3295 the string plus current file name and line number to a per-thread
3296 stack, CRYPTO_pop_info() does the obvious, CRYPTO_remove_all_info()
3297 is like calling CYRPTO_pop_info() until the stack is empty.
3298 Also updated memory leak detection code to be multi-thread-safe.
3301 *) Add options -text and -noout to pkcs7 utility and delete the
3302 encryption options which never did anything. Update docs.
3305 *) Add options to some of the utilities to allow the pass phrase
3306 to be included on either the command line (not recommended on
3307 OSes like Unix) or read from the environment. Update the
3308 manpages and fix a few bugs.
3311 *) Add a few manpages for some of the openssl commands.
3314 *) Fix the -revoke option in ca. It was freeing up memory twice,
3315 leaking and not finding already revoked certificates.
3318 *) Extensive changes to support certificate auxiliary information.
3319 This involves the use of X509_CERT_AUX structure and X509_AUX
3320 functions. An X509_AUX function such as PEM_read_X509_AUX()
3321 can still read in a certificate file in the usual way but it
3322 will also read in any additional "auxiliary information". By
3323 doing things this way a fair degree of compatibility can be
3324 retained: existing certificates can have this information added
3325 using the new 'x509' options.
3327 Current auxiliary information includes an "alias" and some trust
3328 settings. The trust settings will ultimately be used in enhanced
3329 certificate chain verification routines: currently a certificate
3330 can only be trusted if it is self signed and then it is trusted
3334 *) Fix assembler for Alpha (tested only on DEC OSF not Linux or *BSD).
3335 The problem was that one of the replacement routines had not been working
3336 since SSLeay releases. For now the offending routine has been replaced
3337 with non-optimised assembler. Even so, this now gives around 95%
3338 performance improvement for 1024 bit RSA signs.
3341 *) Hack to fix PKCS#7 decryption when used with some unorthodox RC2
3342 handling. Most clients have the effective key size in bits equal to
3343 the key length in bits: so a 40 bit RC2 key uses a 40 bit (5 byte) key.
3344 A few however don't do this and instead use the size of the decrypted key
3345 to determine the RC2 key length and the AlgorithmIdentifier to determine
3346 the effective key length. In this case the effective key length can still
3347 be 40 bits but the key length can be 168 bits for example. This is fixed
3348 by manually forcing an RC2 key into the EVP_PKEY structure because the
3349 EVP code can't currently handle unusual RC2 key sizes: it always assumes
3350 the key length and effective key length are equal.
3353 *) Add a bunch of functions that should simplify the creation of
3354 X509_NAME structures. Now you should be able to do:
3355 X509_NAME_add_entry_by_txt(nm, "CN", MBSTRING_ASC, "Steve", -1, -1, 0);
3356 and have it automatically work out the correct field type and fill in
3357 the structures. The more adventurous can try:
3358 X509_NAME_add_entry_by_txt(nm, field, MBSTRING_UTF8, str, -1, -1, 0);
3359 and it will (hopefully) work out the correct multibyte encoding.
3362 *) Change the 'req' utility to use the new field handling and multibyte
3363 copy routines. Before the DN field creation was handled in an ad hoc
3364 way in req, ca, and x509 which was rather broken and didn't support
3365 BMPStrings or UTF8Strings. Since some software doesn't implement
3366 BMPStrings or UTF8Strings yet, they can be enabled using the config file
3367 using the dirstring_type option. See the new comment in the default
3368 openssl.cnf for more info.
3371 *) Make crypto/rand/md_rand.c more robust:
3372 - Assure unique random numbers after fork().
3373 - Make sure that concurrent threads access the global counter and
3374 md serializably so that we never lose entropy in them
3375 or use exactly the same state in multiple threads.
3376 Access to the large state is not always serializable because
3377 the additional locking could be a performance killer, and
3378 md should be large enough anyway.
3381 *) New file apps/app_rand.c with commonly needed functionality
3382 for handling the random seed file.
3384 Use the random seed file in some applications that previously did not:
3386 dsaparam -genkey (which also ignored its '-rand' option),
3389 x509 (when signing).
3390 Except on systems with /dev/urandom, it is crucial to have a random
3391 seed file at least for key creation, DSA signing, and for DH exchanges;
3392 for RSA signatures we could do without one.
3394 gendh and gendsa (unlike genrsa) used to read only the first byte
3395 of each file listed in the '-rand' option. The function as previously
3396 found in genrsa is now in app_rand.c and is used by all programs
3397 that support '-rand'.
3400 *) In RAND_write_file, use mode 0600 for creating files;
3401 don't just chmod when it may be too late.
3404 *) Report an error from X509_STORE_load_locations
3405 when X509_LOOKUP_load_file or X509_LOOKUP_add_dir failed.
3408 *) New function ASN1_mbstring_copy() this copies a string in either
3409 ASCII, Unicode, Universal (4 bytes per character) or UTF8 format
3410 into an ASN1_STRING type. A mask of permissible types is passed
3411 and it chooses the "minimal" type to use or an error if not type
3415 *) Add function equivalents to the various macros in asn1.h. The old
3416 macros are retained with an M_ prefix. Code inside the library can
3417 use the M_ macros. External code (including the openssl utility)
3418 should *NOT* in order to be "shared library friendly".
3421 *) Add various functions that can check a certificate's extensions
3422 to see if it usable for various purposes such as SSL client,
3423 server or S/MIME and CAs of these types. This is currently
3424 VERY EXPERIMENTAL but will ultimately be used for certificate chain
3425 verification. Also added a -purpose flag to x509 utility to
3426 print out all the purposes.
3429 *) Add a CRYPTO_EX_DATA to X509 certificate structure and associated
3433 *) New X509V3_{X509,CRL,REVOKED}_get_d2i() functions. These will search
3434 for, obtain and decode and extension and obtain its critical flag.
3435 This allows all the necessary extension code to be handled in a
3436 single function call.
3439 *) RC4 tune-up featuring 30-40% performance improvement on most RISC
3440 platforms. See crypto/rc4/rc4_enc.c for further details.
3443 *) New -noout option to asn1parse. This causes no output to be produced
3444 its main use is when combined with -strparse and -out to extract data
3445 from a file (which may not be in ASN.1 format).
3448 *) Fix for pkcs12 program. It was hashing an invalid certificate pointer
3449 when producing the local key id.
3450 [Richard Levitte <levitte@stacken.kth.se>]
3452 *) New option -dhparam in s_server. This allows a DH parameter file to be
3453 stated explicitly. If it is not stated then it tries the first server
3454 certificate file. The previous behaviour hard coded the filename
3458 *) Add -pubin and -pubout options to the rsa and dsa commands. These allow
3459 a public key to be input or output. For example:
3460 openssl rsa -in key.pem -pubout -out pubkey.pem
3461 Also added necessary DSA public key functions to handle this.
3464 *) Fix so PKCS7_dataVerify() doesn't crash if no certificates are contained
3465 in the message. This was handled by allowing
3466 X509_find_by_issuer_and_serial() to tolerate a NULL passed to it.
3467 [Steve Henson, reported by Sampo Kellomaki <sampo@mail.neuronio.pt>]
3469 *) Fix for bug in d2i_ASN1_bytes(): other ASN1 functions add an extra null
3470 to the end of the strings whereas this didn't. This would cause problems
3471 if strings read with d2i_ASN1_bytes() were later modified.
3472 [Steve Henson, reported by Arne Ansper <arne@ats.cyber.ee>]
3474 *) Fix for base64 decode bug. When a base64 bio reads only one line of
3475 data and it contains EOF it will end up returning an error. This is
3476 caused by input 46 bytes long. The cause is due to the way base64
3477 BIOs find the start of base64 encoded data. They do this by trying a
3478 trial decode on each line until they find one that works. When they
3479 do a flag is set and it starts again knowing it can pass all the
3480 data directly through the decoder. Unfortunately it doesn't reset
3481 the context it uses. This means that if EOF is reached an attempt
3482 is made to pass two EOFs through the context and this causes the
3483 resulting error. This can also cause other problems as well. As is
3484 usual with these problems it takes *ages* to find and the fix is
3485 trivial: move one line.
3486 [Steve Henson, reported by ian@uns.ns.ac.yu (Ivan Nejgebauer) ]
3488 *) Ugly workaround to get s_client and s_server working under Windows. The
3489 old code wouldn't work because it needed to select() on sockets and the
3490 tty (for keypresses and to see if data could be written). Win32 only
3491 supports select() on sockets so we select() with a 1s timeout on the
3492 sockets and then see if any characters are waiting to be read, if none
3493 are present then we retry, we also assume we can always write data to
3494 the tty. This isn't nice because the code then blocks until we've
3495 received a complete line of data and it is effectively polling the
3496 keyboard at 1s intervals: however it's quite a bit better than not
3497 working at all :-) A dedicated Windows application might handle this
3498 with an event loop for example.
3501 *) Enhance RSA_METHOD structure. Now there are two extra methods, rsa_sign
3502 and rsa_verify. When the RSA_FLAGS_SIGN_VER option is set these functions
3503 will be called when RSA_sign() and RSA_verify() are used. This is useful
3504 if rsa_pub_dec() and rsa_priv_enc() equivalents are not available.
3505 For this to work properly RSA_public_decrypt() and RSA_private_encrypt()
3506 should *not* be used: RSA_sign() and RSA_verify() must be used instead.
3507 This necessitated the support of an extra signature type NID_md5_sha1
3508 for SSL signatures and modifications to the SSL library to use it instead
3509 of calling RSA_public_decrypt() and RSA_private_encrypt().
3512 *) Add new -verify -CAfile and -CApath options to the crl program, these
3513 will lookup a CRL issuers certificate and verify the signature in a
3514 similar way to the verify program. Tidy up the crl program so it
3515 no longer accesses structures directly. Make the ASN1 CRL parsing a bit
3516 less strict. It will now permit CRL extensions even if it is not
3517 a V2 CRL: this will allow it to tolerate some broken CRLs.
3520 *) Initialize all non-automatic variables each time one of the openssl
3521 sub-programs is started (this is necessary as they may be started
3522 multiple times from the "OpenSSL>" prompt).
3523 [Lennart Bang, Bodo Moeller]
3525 *) Preliminary compilation option RSA_NULL which disables RSA crypto without
3526 removing all other RSA functionality (this is what NO_RSA does). This
3527 is so (for example) those in the US can disable those operations covered
3528 by the RSA patent while allowing storage and parsing of RSA keys and RSA
3532 *) Non-copying interface to BIO pairs.
3533 (still largely untested)
3536 *) New function ANS1_tag2str() to convert an ASN1 tag to a descriptive
3537 ASCII string. This was handled independently in various places before.
3540 *) New functions UTF8_getc() and UTF8_putc() that parse and generate
3541 UTF8 strings a character at a time.
3544 *) Use client_version from client hello to select the protocol
3545 (s23_srvr.c) and for RSA client key exchange verification
3546 (s3_srvr.c), as required by the SSL 3.0/TLS 1.0 specifications.
3549 *) Add various utility functions to handle SPKACs, these were previously
3550 handled by poking round in the structure internals. Added new function
3551 NETSCAPE_SPKI_print() to print out SPKAC and a new utility 'spkac' to
3552 print, verify and generate SPKACs. Based on an original idea from
3553 Massimiliano Pala <madwolf@comune.modena.it> but extensively modified.
3556 *) RIPEMD160 is operational on all platforms and is back in 'make test'.
3559 *) Allow the config file extension section to be overwritten on the
3560 command line. Based on an original idea from Massimiliano Pala
3561 <madwolf@comune.modena.it>. The new option is called -extensions
3562 and can be applied to ca, req and x509. Also -reqexts to override
3563 the request extensions in req and -crlexts to override the crl extensions
3567 *) Add new feature to the SPKAC handling in ca. Now you can include
3568 the same field multiple times by preceding it by "XXXX." for example:
3571 this is the same syntax as used in the req config file.
3574 *) Allow certificate extensions to be added to certificate requests. These
3575 are specified in a 'req_extensions' option of the req section of the
3576 config file. They can be printed out with the -text option to req but
3577 are otherwise ignored at present.
3580 *) Fix a horrible bug in enc_read() in crypto/evp/bio_enc.c: if the first
3581 data read consists of only the final block it would not decrypted because
3582 EVP_CipherUpdate() would correctly report zero bytes had been decrypted.
3583 A misplaced 'break' also meant the decrypted final block might not be
3584 copied until the next read.
3587 *) Initial support for DH_METHOD. Again based on RSA_METHOD. Also added
3588 a few extra parameters to the DH structure: these will be useful if
3589 for example we want the value of 'q' or implement X9.42 DH.
3592 *) Initial support for DSA_METHOD. This is based on the RSA_METHOD and
3593 provides hooks that allow the default DSA functions or functions on a
3594 "per key" basis to be replaced. This allows hardware acceleration and
3595 hardware key storage to be handled without major modification to the
3596 library. Also added low level modexp hooks and CRYPTO_EX structure and
3597 associated functions.
3600 *) Add a new flag to memory BIOs, BIO_FLAG_MEM_RDONLY. This marks the BIO
3601 as "read only": it can't be written to and the buffer it points to will
3602 not be freed. Reading from a read only BIO is much more efficient than
3603 a normal memory BIO. This was added because there are several times when
3604 an area of memory needs to be read from a BIO. The previous method was
3605 to create a memory BIO and write the data to it, this results in two
3606 copies of the data and an O(n^2) reading algorithm. There is a new
3607 function BIO_new_mem_buf() which creates a read only memory BIO from
3608 an area of memory. Also modified the PKCS#7 routines to use read only
3612 *) Bugfix: ssl23_get_client_hello did not work properly when called in
3613 state SSL23_ST_SR_CLNT_HELLO_B, i.e. when the first 7 bytes of
3614 a SSLv2-compatible client hello for SSLv3 or TLSv1 could be read,
3615 but a retry condition occured while trying to read the rest.
3618 *) The PKCS7_ENC_CONTENT_new() function was setting the content type as
3619 NID_pkcs7_encrypted by default: this was wrong since this should almost
3620 always be NID_pkcs7_data. Also modified the PKCS7_set_type() to handle
3621 the encrypted data type: this is a more sensible place to put it and it
3622 allows the PKCS#12 code to be tidied up that duplicated this
3626 *) Changed obj_dat.pl script so it takes its input and output files on
3627 the command line. This should avoid shell escape redirection problems
3631 *) Initial support for certificate extension requests, these are included
3632 in things like Xenroll certificate requests. Included functions to allow
3633 extensions to be obtained and added.
3636 *) -crlf option to s_client and s_server for sending newlines as
3637 CRLF (as required by many protocols).
3640 Changes between 0.9.3a and 0.9.4 [09 Aug 1999]
3642 *) Install libRSAglue.a when OpenSSL is built with RSAref.
3643 [Ralf S. Engelschall]
3645 *) A few more ``#ifndef NO_FP_API / #endif'' pairs for consistency.
3646 [Andrija Antonijevic <TheAntony2@bigfoot.com>]
3648 *) Fix -startdate and -enddate (which was missing) arguments to 'ca'
3652 *) New function DSA_dup_DH, which duplicates DSA parameters/keys as
3653 DH parameters/keys (q is lost during that conversion, but the resulting
3654 DH parameters contain its length).
3656 For 1024-bit p, DSA_generate_parameters followed by DSA_dup_DH is
3657 much faster than DH_generate_parameters (which creates parameters
3658 where p = 2*q + 1), and also the smaller q makes DH computations
3659 much more efficient (160-bit exponentiation instead of 1024-bit
3660 exponentiation); so this provides a convenient way to support DHE
3661 ciphersuites in SSL/TLS servers (see ssl/ssltest.c). It is of
3662 utter importance to use
3663 SSL_CTX_set_options(s_ctx, SSL_OP_SINGLE_DH_USE);
3665 SSL_set_options(s_ctx, SSL_OP_SINGLE_DH_USE);
3666 when such DH parameters are used, because otherwise small subgroup
3667 attacks may become possible!
3670 *) Avoid memory leak in i2d_DHparams.
3673 *) Allow the -k option to be used more than once in the enc program:
3674 this allows the same encrypted message to be read by multiple recipients.
3677 *) New function OBJ_obj2txt(buf, buf_len, a, no_name), this converts
3678 an ASN1_OBJECT to a text string. If the "no_name" parameter is set then
3679 it will always use the numerical form of the OID, even if it has a short
3683 *) Added an extra RSA flag: RSA_FLAG_EXT_PKEY. Previously the rsa_mod_exp
3684 method only got called if p,q,dmp1,dmq1,iqmp components were present,
3685 otherwise bn_mod_exp was called. In the case of hardware keys for example
3686 no private key components need be present and it might store extra data
3687 in the RSA structure, which cannot be accessed from bn_mod_exp.
3688 By setting RSA_FLAG_EXT_PKEY rsa_mod_exp will always be called for
3689 private key operations.
3692 *) Added support for SPARC Linux.
3695 *) pem_password_cb function type incompatibly changed from
3696 typedef int pem_password_cb(char *buf, int size, int rwflag);
3698 ....(char *buf, int size, int rwflag, void *userdata);
3699 so that applications can pass data to their callbacks:
3700 The PEM[_ASN1]_{read,write}... functions and macros now take an
3701 additional void * argument, which is just handed through whenever
3702 the password callback is called.
3703 [Damien Miller <dmiller@ilogic.com.au>; tiny changes by Bodo Moeller]
3705 New function SSL_CTX_set_default_passwd_cb_userdata.
3707 Compatibility note: As many C implementations push function arguments
3708 onto the stack in reverse order, the new library version is likely to
3709 interoperate with programs that have been compiled with the old
3710 pem_password_cb definition (PEM_whatever takes some data that
3711 happens to be on the stack as its last argument, and the callback
3712 just ignores this garbage); but there is no guarantee whatsoever that
3715 *) The -DPLATFORM="\"$(PLATFORM)\"" definition and the similar -DCFLAGS=...
3716 (both in crypto/Makefile.ssl for use by crypto/cversion.c) caused
3717 problems not only on Windows, but also on some Unix platforms.
3718 To avoid problematic command lines, these definitions are now in an
3719 auto-generated file crypto/buildinf.h (created by crypto/Makefile.ssl
3720 for standard "make" builds, by util/mk1mf.pl for "mk1mf" builds).
3723 *) MIPS III/IV assembler module is reimplemented.
3726 *) More DES library cleanups: remove references to srand/rand and
3727 delete an unused file.
3730 *) Add support for the the free Netwide assembler (NASM) under Win32,
3731 since not many people have MASM (ml) and it can be hard to obtain.
3732 This is currently experimental but it seems to work OK and pass all
3733 the tests. Check out INSTALL.W32 for info.
3736 *) Fix memory leaks in s3_clnt.c: All non-anonymous SSL3/TLS1 connections
3737 without temporary keys kept an extra copy of the server key,
3738 and connections with temporary keys did not free everything in case
3742 *) New function RSA_check_key and new openssl rsa option -check
3743 for verifying the consistency of RSA keys.
3744 [Ulf Moeller, Bodo Moeller]
3746 *) Various changes to make Win32 compile work:
3747 1. Casts to avoid "loss of data" warnings in p5_crpt2.c
3748 2. Change unsigned int to int in b_dump.c to avoid "signed/unsigned
3749 comparison" warnings.
3750 3. Add sk_<TYPE>_sort to DEF file generator and do make update.
3753 *) Add a debugging option to PKCS#5 v2 key generation function: when
3754 you #define DEBUG_PKCS5V2 passwords, salts, iteration counts and
3755 derived keys are printed to stderr.
3758 *) Copy the flags in ASN1_STRING_dup().
3759 [Roman E. Pavlov <pre@mo.msk.ru>]
3761 *) The x509 application mishandled signing requests containing DSA
3762 keys when the signing key was also DSA and the parameters didn't match.
3764 It was supposed to omit the parameters when they matched the signing key:
3765 the verifying software was then supposed to automatically use the CA's
3766 parameters if they were absent from the end user certificate.
3768 Omitting parameters is no longer recommended. The test was also
3769 the wrong way round! This was probably due to unusual behaviour in
3770 EVP_cmp_parameters() which returns 1 if the parameters match.
3771 This meant that parameters were omitted when they *didn't* match and
3772 the certificate was useless. Certificates signed with 'ca' didn't have
3774 [Steve Henson, reported by Doug Erickson <Doug.Erickson@Part.NET>]
3776 *) Memory leak checking (-DCRYPTO_MDEBUG) had some problems.
3777 The interface is as follows:
3778 Applications can use
3779 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_ON) aka MemCheck_start(),
3780 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_OFF) aka MemCheck_stop();
3781 "off" is now the default.
3782 The library internally uses
3783 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_DISABLE) aka MemCheck_off(),
3784 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_ENABLE) aka MemCheck_on()
3785 to disable memory-checking temporarily.
3787 Some inconsistent states that previously were possible (and were
3788 even the default) are now avoided.
3790 -DCRYPTO_MDEBUG_TIME is new and additionally stores the current time
3791 with each memory chunk allocated; this is occasionally more helpful
3792 than just having a counter.
3794 -DCRYPTO_MDEBUG_THREAD is also new and adds the thread ID.
3796 -DCRYPTO_MDEBUG_ALL enables all of the above, plus any future
3800 *) Introduce "mode" for SSL structures (with defaults in SSL_CTX),
3801 which largely parallels "options", but is for changing API behaviour,
3802 whereas "options" are about protocol behaviour.
3803 Initial "mode" flags are:
3805 SSL_MODE_ENABLE_PARTIAL_WRITE Allow SSL_write to report success when
3806 a single record has been written.
3807 SSL_MODE_ACCEPT_MOVING_WRITE_BUFFER Don't insist that SSL_write
3808 retries use the same buffer location.
3809 (But all of the contents must be
3813 *) Bugfix: SSL_set_options ignored its parameter, only SSL_CTX_set_options
3816 *) Fix problems with no-hmac etc.
3817 [Ulf Möller, pointed out by Brian Wellington <bwelling@tislabs.com>]
3819 *) New functions RSA_get_default_method(), RSA_set_method() and
3820 RSA_get_method(). These allows replacement of RSA_METHODs without having
3821 to mess around with the internals of an RSA structure.
3824 *) Fix memory leaks in DSA_do_sign and DSA_is_prime.
3825 Also really enable memory leak checks in openssl.c and in some
3827 [Chad C. Mulligan, Bodo Moeller]
3829 *) Fix a bug in d2i_ASN1_INTEGER() and i2d_ASN1_INTEGER() which can mess
3830 up the length of negative integers. This has now been simplified to just
3831 store the length when it is first determined and use it later, rather
3832 than trying to keep track of where data is copied and updating it to
3834 [Steve Henson, reported by Brien Wheeler
3835 <bwheeler@authentica-security.com>]
3837 *) Add a new function PKCS7_signatureVerify. This allows the verification
3838 of a PKCS#7 signature but with the signing certificate passed to the
3839 function itself. This contrasts with PKCS7_dataVerify which assumes the
3840 certificate is present in the PKCS#7 structure. This isn't always the
3841 case: certificates can be omitted from a PKCS#7 structure and be
3842 distributed by "out of band" means (such as a certificate database).
3845 *) Complete the PEM_* macros with DECLARE_PEM versions to replace the
3846 function prototypes in pem.h, also change util/mkdef.pl to add the
3847 necessary function names.
3850 *) mk1mf.pl (used by Windows builds) did not properly read the
3851 options set by Configure in the top level Makefile, and Configure
3852 was not even able to write more than one option correctly.
3853 Fixed, now "no-idea no-rc5 -DCRYPTO_MDEBUG" etc. works as intended.
3856 *) New functions CONF_load_bio() and CONF_load_fp() to allow a config
3857 file to be loaded from a BIO or FILE pointer. The BIO version will
3858 for example allow memory BIOs to contain config info.
3861 *) New function "CRYPTO_num_locks" that returns CRYPTO_NUM_LOCKS.
3862 Whoever hopes to achieve shared-library compatibility across versions
3863 must use this, not the compile-time macro.
3864 (Exercise 0.9.4: Which is the minimum library version required by
3866 Note: All this applies only to multi-threaded programs, others don't
3870 *) Add missing case to s3_clnt.c state machine -- one of the new SSL tests
3871 through a BIO pair triggered the default case, i.e.
3872 SSLerr(...,SSL_R_UNKNOWN_STATE).
3875 *) New "BIO pair" concept (crypto/bio/bss_bio.c) so that applications
3876 can use the SSL library even if none of the specific BIOs is
3880 *) Fix a bug in i2d_DSAPublicKey() which meant it returned the wrong value
3881 for the encoded length.
3882 [Jeon KyoungHo <khjeon@sds.samsung.co.kr>]
3884 *) Add initial documentation of the X509V3 functions.
3887 *) Add a new pair of functions PEM_write_PKCS8PrivateKey() and
3888 PEM_write_bio_PKCS8PrivateKey() that are equivalent to
3889 PEM_write_PrivateKey() and PEM_write_bio_PrivateKey() but use the more
3890 secure PKCS#8 private key format with a high iteration count.
3893 *) Fix determination of Perl interpreter: A perl or perl5
3894 _directory_ in $PATH was also accepted as the interpreter.
3895 [Ralf S. Engelschall]
3897 *) Fix demos/sign/sign.c: well there wasn't anything strictly speaking
3898 wrong with it but it was very old and did things like calling
3899 PEM_ASN1_read() directly and used MD5 for the hash not to mention some
3903 *) Fix demos/selfsign.c: it used obsolete and deleted functions, changed
3904 to use the new extension code.
3907 *) Implement the PEM_read/PEM_write functions in crypto/pem/pem_all.c
3908 with macros. This should make it easier to change their form, add extra
3909 arguments etc. Fix a few PEM prototypes which didn't have cipher as a
3913 *) Add to configuration table a new entry that can specify an alternative
3914 name for unistd.h (for pre-POSIX systems); we need this for NeXTstep,
3915 according to Mark Crispin <MRC@Panda.COM>.
3919 *) DES CBC did not update the IV. Weird.
3922 des_cbc_encrypt does not update the IV, but des_ncbc_encrypt does.
3923 Changing the behaviour of the former might break existing programs --
3924 where IV updating is needed, des_ncbc_encrypt can be used.
3927 *) When bntest is run from "make test" it drives bc to check its
3928 calculations, as well as internally checking them. If an internal check
3929 fails, it needs to cause bc to give a non-zero result or make test carries
3930 on without noticing the failure. Fixed.
3933 *) DES library cleanups.
3936 *) Add support for PKCS#5 v2.0 PBE algorithms. This will permit PKCS#8 to be
3937 used with any cipher unlike PKCS#5 v1.5 which can at most handle 64 bit
3938 ciphers. NOTE: although the key derivation function has been verified
3939 against some published test vectors it has not been extensively tested
3940 yet. Added a -v2 "cipher" option to pkcs8 application to allow the use
3944 *) Instead of "mkdir -p", which is not fully portable, use new
3945 Perl script "util/mkdir-p.pl".
3948 *) Rewrite the way password based encryption (PBE) is handled. It used to
3949 assume that the ASN1 AlgorithmIdentifier parameter was a PBEParameter
3950 structure. This was true for the PKCS#5 v1.5 and PKCS#12 PBE algorithms
3951 but doesn't apply to PKCS#5 v2.0 where it can be something else. Now
3952 the 'parameter' field of the AlgorithmIdentifier is passed to the
3953 underlying key generation function so it must do its own ASN1 parsing.
3954 This has also changed the EVP_PBE_CipherInit() function which now has a
3955 'parameter' argument instead of literal salt and iteration count values
3956 and the function EVP_PBE_ALGOR_CipherInit() has been deleted.
3959 *) Support for PKCS#5 v1.5 compatible password based encryption algorithms
3960 and PKCS#8 functionality. New 'pkcs8' application linked to openssl.
3961 Needed to change the PEM_STRING_EVP_PKEY value which was just "PRIVATE
3962 KEY" because this clashed with PKCS#8 unencrypted string. Since this
3963 value was just used as a "magic string" and not used directly its
3964 value doesn't matter.
3967 *) Introduce some semblance of const correctness to BN. Shame C doesn't
3971 *) "linux-sparc64" configuration (ultrapenguin).
3972 [Ray Miller <ray.miller@oucs.ox.ac.uk>]
3973 "linux-sparc" configuration.
3974 [Christian Forster <fo@hawo.stw.uni-erlangen.de>]
3976 *) config now generates no-xxx options for missing ciphers.
3979 *) Support the EBCDIC character set (work in progress).
3980 File ebcdic.c not yet included because it has a different license.
3981 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>]
3983 *) Support BS2000/OSD-POSIX.
3984 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>]
3986 *) Make callbacks for key generation use void * instead of char *.
3989 *) Make S/MIME samples compile (not yet tested).
3992 *) Additional typesafe stacks.
3995 *) New configuration variants "bsdi-elf-gcc" (BSD/OS 4.x).
3999 Changes between 0.9.3 and 0.9.3a [29 May 1999]
4001 *) New configuration variant "sco5-gcc".
4003 *) Updated some demos.
4004 [Sean O Riordain, Wade Scholine]
4006 *) Add missing BIO_free at exit of pkcs12 application.
4009 *) Fix memory leak in conf.c.
4012 *) Updates for Win32 to assembler version of MD5.
4015 *) Set #! path to perl in apps/der_chop to where we found it
4016 instead of using a fixed path.
4019 *) SHA library changes for irix64-mips4-cc.
4022 *) Improvements for VMS support.
4026 Changes between 0.9.2b and 0.9.3 [24 May 1999]
4028 *) Bignum library bug fix. IRIX 6 passes "make test" now!
4029 This also avoids the problems with SC4.2 and unpatched SC5.
4030 [Andy Polyakov <appro@fy.chalmers.se>]
4032 *) New functions sk_num, sk_value and sk_set to replace the previous macros.
4033 These are required because of the typesafe stack would otherwise break
4034 existing code. If old code used a structure member which used to be STACK
4035 and is now STACK_OF (for example cert in a PKCS7_SIGNED structure) with
4036 sk_num or sk_value it would produce an error because the num, data members
4037 are not present in STACK_OF. Now it just produces a warning. sk_set
4038 replaces the old method of assigning a value to sk_value
4039 (e.g. sk_value(x, i) = y) which the library used in a few cases. Any code
4040 that does this will no longer work (and should use sk_set instead) but
4041 this could be regarded as a "questionable" behaviour anyway.
4044 *) Fix most of the other PKCS#7 bugs. The "experimental" code can now
4045 correctly handle encrypted S/MIME data.
4048 *) Change type of various DES function arguments from des_cblock
4049 (which means, in function argument declarations, pointer to char)
4050 to des_cblock * (meaning pointer to array with 8 char elements),
4051 which allows the compiler to do more typechecking; it was like
4052 that back in SSLeay, but with lots of ugly casts.
4054 Introduce new type const_des_cblock.
4057 *) Reorganise the PKCS#7 library and get rid of some of the more obvious
4058 problems: find RecipientInfo structure that matches recipient certificate
4059 and initialise the ASN1 structures properly based on passed cipher.
4062 *) Belatedly make the BN tests actually check the results.
4065 *) Fix the encoding and decoding of negative ASN1 INTEGERS and conversion
4066 to and from BNs: it was completely broken. New compilation option
4067 NEG_PUBKEY_BUG to allow for some broken certificates that encode public
4068 key elements as negative integers.
4071 *) Reorganize and speed up MD5.
4072 [Andy Polyakov <appro@fy.chalmers.se>]
4075 [Richard Levitte <richard@levitte.org>]
4077 *) New option -out to asn1parse to allow the parsed structure to be
4078 output to a file. This is most useful when combined with the -strparse
4079 option to examine the output of things like OCTET STRINGS.
4082 *) Make SSL library a little more fool-proof by not requiring any longer
4083 that SSL_set_{accept,connect}_state be called before
4084 SSL_{accept,connect} may be used (SSL_set_..._state is omitted
4085 in many applications because usually everything *appeared* to work as
4086 intended anyway -- now it really works as intended).
4089 *) Move openssl.cnf out of lib/.
4092 *) Fix various things to let OpenSSL even pass ``egcc -pipe -O2 -Wall
4093 -Wshadow -Wpointer-arith -Wcast-align -Wmissing-prototypes
4094 -Wmissing-declarations -Wnested-externs -Winline'' with EGCS 1.1.2+
4095 [Ralf S. Engelschall]
4097 *) Various fixes to the EVP and PKCS#7 code. It may now be able to
4098 handle PKCS#7 enveloped data properly.
4099 [Sebastian Akerman <sak@parallelconsulting.com>, modified by Steve]
4101 *) Create a duplicate of the SSL_CTX's CERT in SSL_new instead of
4102 copying pointers. The cert_st handling is changed by this in
4103 various ways (and thus what used to be known as ctx->default_cert
4104 is now called ctx->cert, since we don't resort to s->ctx->[default_]cert
4105 any longer when s->cert does not give us what we need).
4106 ssl_cert_instantiate becomes obsolete by this change.
4107 As soon as we've got the new code right (possibly it already is?),
4108 we have solved a couple of bugs of the earlier code where s->cert
4109 was used as if it could not have been shared with other SSL structures.
4111 Note that using the SSL API in certain dirty ways now will result
4112 in different behaviour than observed with earlier library versions:
4113 Changing settings for an SSL_CTX *ctx after having done s = SSL_new(ctx)
4114 does not influence s as it used to.
4116 In order to clean up things more thoroughly, inside SSL_SESSION
4117 we don't use CERT any longer, but a new structure SESS_CERT
4118 that holds per-session data (if available); currently, this is
4119 the peer's certificate chain and, for clients, the server's certificate
4120 and temporary key. CERT holds only those values that can have
4121 meaningful defaults in an SSL_CTX.
4124 *) New function X509V3_EXT_i2d() to create an X509_EXTENSION structure
4125 from the internal representation. Various PKCS#7 fixes: remove some
4126 evil casts and set the enc_dig_alg field properly based on the signing
4130 *) Allow PKCS#12 password to be set from the command line or the
4131 environment. Let 'ca' get its config file name from the environment
4132 variables "OPENSSL_CONF" or "SSLEAY_CONF" (for consistency with 'req'
4136 *) Allow certificate policies extension to use an IA5STRING for the
4137 organization field. This is contrary to the PKIX definition but
4138 VeriSign uses it and IE5 only recognises this form. Document 'x509'
4142 *) Add PEDANTIC compiler flag to allow compilation with gcc -pedantic,
4143 without disallowing inline assembler and the like for non-pedantic builds.
4146 *) Support Borland C++ builder.
4147 [Janez Jere <jj@void.si>, modified by Ulf Möller]
4152 *) SHA-1 cleanups and performance enhancements.
4153 [Andy Polyakov <appro@fy.chalmers.se>]
4155 *) Sparc v8plus assembler for the bignum library.
4156 [Andy Polyakov <appro@fy.chalmers.se>]
4158 *) Accept any -xxx and +xxx compiler options in Configure.
4161 *) Update HPUX configuration.
4164 *) Add missing sk_<type>_unshift() function to safestack.h
4165 [Ralf S. Engelschall]
4167 *) New function SSL_CTX_use_certificate_chain_file that sets the
4168 "extra_cert"s in addition to the certificate. (This makes sense
4169 only for "PEM" format files, as chains as a whole are not
4173 *) Support verify_depth from the SSL API.
4174 x509_vfy.c had what can be considered an off-by-one-error:
4175 Its depth (which was not part of the external interface)
4176 was actually counting the number of certificates in a chain;
4177 now it really counts the depth.
4180 *) Bugfix in crypto/x509/x509_cmp.c: The SSLerr macro was used
4181 instead of X509err, which often resulted in confusing error
4182 messages since the error codes are not globally unique
4183 (e.g. an alleged error in ssl3_accept when a certificate
4184 didn't match the private key).
4186 *) New function SSL_CTX_set_session_id_context that allows to set a default
4187 value (so that you don't need SSL_set_session_id_context for each
4188 connection using the SSL_CTX).
4191 *) OAEP decoding bug fix.
4194 *) Support INSTALL_PREFIX for package builders, as proposed by
4198 *) New Configure options "threads" and "no-threads". For systems
4199 where the proper compiler options are known (currently Solaris
4200 and Linux), "threads" is the default.
4203 *) New script util/mklink.pl as a faster substitute for util/mklink.sh.
4206 *) Install various scripts to $(OPENSSLDIR)/misc, not to
projects / openssl.git / blob