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 developed in parallel, based on OpenSSL 0.9.6.
10 Change log entries are tagged as follows:
11 -) applies to 0.9.6a/0.9.6b/0.9.6c only
12 *) applies to 0.9.6a/0.9.6b/0.9.6c and 0.9.7
13 +) applies to 0.9.7 only
15 +) Change mkdef.pl to sort symbols that get the same entry number,
16 and make sure the automatically generated functions ERR_load_*
17 become part of libeay.num as well.
20 *) Avoid infinite loop in ssl3_get_message (ssl/s3_both.c) if a
21 client receives HelloRequest while in a handshake.
22 [Bodo Moeller; bug noticed by Andy Schneider <andy.schneider@bjss.co.uk>]
24 +) New function SSL_renegotiate_pending(). This returns true once
25 renegotiation has been requested (either SSL_renegotiate() call
26 or HelloRequest/ClientHello receveived from the peer) and becomes
27 false once a handshake has been completed.
28 (For servers, SSL_renegotiate() followed by SSL_do_handshake()
29 sends a HelloRequest, but does not ensure that a handshake takes
30 place. SSL_renegotiate_pending() is useful for checking if the
31 client has followed the request.)
34 +) New SSL option SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION.
35 By default, clients may request session resumption even during
36 renegotiation (if session ID contexts permit); with this option,
37 session resumption is possible only in the first handshake.
40 *) Bugfix in ssl3_accept (ssl/s3_srvr.c): Case SSL3_ST_SW_HELLO_REQ_C
41 should end in 'break', not 'goto end' which circuments various
42 cleanups done in state SSL_ST_OK. But session related stuff
43 must be disabled for SSL_ST_OK in the case that we just sent a
46 Also avoid some overhead by not calling ssl_init_wbio_buffer()
47 before just sending a HelloRequest.
48 [Bodo Moeller, Eric Rescorla <ekr@rtfm.com>]
50 *) Fix ssl/s3_enc.c, ssl/t1_enc.c and ssl/s3_pkt.c so that we don't
51 reveal whether illegal block cipher padding was found or a MAC
52 verification error occured. (Neither SSLerr() codes nor alerts
53 are directly visible to potential attackers, but the information
54 may leak via logfiles.)
56 Similar changes are not required for the SSL 2.0 implementation
57 because the number of padding bytes is sent in clear for SSL 2.0,
58 and the extra bytes are just ignored. However ssl/s2_pkt.c
59 failed to verify that the purported number of padding bytes is in
63 +) Add some demos for certificate and certificate request creation.
66 +) Make maximum certificate chain size accepted from the peer application
67 settable (SSL*_get/set_max_cert_list()), as proposed by
68 "Douglas E. Engert" <deengert@anl.gov>.
71 +) Add support for shared libraries for Unixware-7 and support including
72 shared libraries for OpenUNIX-8 (Boyd Lynn Gerber <gerberb@zenez.com>).
75 *) Improve RSA_padding_check_PKCS1_OAEP() check again to avoid
76 'wristwatch attack' using huge encoding parameters (cf.
77 James H. Manger's CRYPTO 2001 paper). Note that the
78 RSA_PKCS1_OAEP_PADDING case of RSA_private_decrypt() does not use
79 encoding parameters and hence was not vulnerable.
82 +) Add a "destroy" handler to ENGINEs that allows structural cleanup to
83 be done prior to destruction. Use this to unload error strings from
84 ENGINEs that load their own error strings. NB: This adds two new API
85 functions to "get" and "set" this destroy handler in an ENGINE.
88 +) Alter all existing ENGINE implementations (except "openssl" and
89 "openbsd") to dynamically instantiate their own error strings. This
90 makes them more flexible to be built both as statically-linked ENGINEs
91 and self-contained shared-libraries loadable via the "dynamic" ENGINE.
92 Also, add stub code to each that makes building them as self-contained
93 shared-libraries easier (see README.ENGINE).
96 +) Add a "dynamic" ENGINE that provides a mechanism for binding ENGINE
97 implementations into applications that are completely implemented in
98 self-contained shared-libraries. The "dynamic" ENGINE exposes control
99 commands that can be used to configure what shared-library to load and
100 to control aspects of the way it is handled. Also, made an update to
101 the README.ENGINE file that brings its information up-to-date and
102 provides some information and instructions on the "dynamic" ENGINE
103 (ie. how to use it, how to build "dynamic"-loadable ENGINEs, etc).
107 [Ulf Möller, reported by Jim Ellis <jim.ellis@cavium.com>]
109 *) Make it possible to unload ranges of ERR strings with a new
110 "ERR_unload_strings" function.
113 *) Rabin-Miller test analyses assume uniformly distributed witnesses,
114 so use BN_pseudo_rand_range() instead of using BN_pseudo_rand()
115 followed by modular reduction.
116 [Bodo Moeller; pointed out by Adam Young <AYoung1@NCSUS.JNJ.COM>]
118 *) Add BN_pseudo_rand_range() with obvious functionality: BN_rand_range()
119 equivalent based on BN_pseudo_rand() instead of BN_rand().
122 +) Add a copy() function to EVP_MD.
125 +) Make EVP_MD routines take a context pointer instead of just the
126 md_data void pointer.
129 +) Add flags to EVP_MD and EVP_MD_CTX. EVP_MD_FLAG_ONESHOT indicates
130 that the digest can only process a single chunk of data
131 (typically because it is provided by a piece of
132 hardware). EVP_MD_CTX_FLAG_ONESHOT indicates that the application
133 is only going to provide a single chunk of data, and hence the
134 framework needn't accumulate the data for oneshot drivers.
137 +) As with "ERR", make it possible to replace the underlying "ex_data"
138 functions. This change also alters the storage and management of global
139 ex_data state - it's now all inside ex_data.c and all "class" code (eg.
140 RSA, BIO, SSL_CTX, etc) no longer stores its own STACKS and per-class
141 index counters. The API functions that use this state have been changed
142 to take a "class_index" rather than pointers to the class's local STACK
143 and counter, and there is now an API function to dynamically create new
144 classes. This centralisation allows us to (a) plug a lot of the
145 thread-safety problems that existed, and (b) makes it possible to clean
146 up all allocated state using "CRYPTO_cleanup_all_ex_data()". W.r.t. (b)
147 such data would previously have always leaked in application code and
148 workarounds were in place to make the memory debugging turn a blind eye
149 to it. Application code that doesn't use this new function will still
150 leak as before, but their memory debugging output will announce it now
151 rather than letting it slide.
153 Besides the addition of CRYPTO_cleanup_all_ex_data(), another API change
154 induced by the "ex_data" overhaul is that X509_STORE_CTX_init() now
155 has a return value to indicate success or failure.
158 +) Make it possible to replace the underlying "ERR" functions such that the
159 global state (2 LHASH tables and 2 locks) is only used by the "default"
160 implementation. This change also adds two functions to "get" and "set"
161 the implementation prior to it being automatically set the first time
162 any other ERR function takes place. Ie. an application can call "get",
163 pass the return value to a module it has just loaded, and that module
164 can call its own "set" function using that value. This means the
165 module's "ERR" operations will use (and modify) the error state in the
166 application and not in its own statically linked copy of OpenSSL code.
169 +) Give DH, DSA, and RSA types their own "**_up_ref()" function to increment
170 reference counts. This performs normal REF_PRINT/REF_CHECK macros on
171 the operation, and provides a more encapsulated way for external code
172 (crypto/evp/ and ssl/) to do this. Also changed the evp and ssl code
173 to use these functions rather than manually incrementing the counts.
175 Also rename "DSO_up()" function to more descriptive "DSO_up_ref()".
178 *) s3_srvr.c: allow sending of large client certificate lists (> 16 kB).
179 This function was broken, as the check for a new client hello message
180 to handle SGC did not allow these large messages.
181 (Tracked down by "Douglas E. Engert" <deengert@anl.gov>.)
184 *) Add alert descriptions for TLSv1 to SSL_alert_desc_string[_long]().
187 +) Add EVP test program.
190 +) Add symmetric cipher support to ENGINE. Expect the API to change!
193 +) New CRL functions: X509_CRL_set_version(), X509_CRL_set_issuer_name()
194 X509_CRL_set_lastUpdate(), X509_CRL_set_nextUpdate(), X509_CRL_sort(),
195 X509_REVOKED_set_serialNumber(), and X509_REVOKED_set_revocationDate().
196 These allow a CRL to be built without having to access X509_CRL fields
197 directly. Modify 'ca' application to use new functions.
200 *) Fix buggy behaviour of BIO_get_num_renegotiates() and BIO_ctrl()
201 for BIO_C_GET_WRITE_BUF_SIZE ("Stephen Hinton" <shinton@netopia.com>).
204 *) Rework the configuration and shared library support for Tru64 Unix.
205 The configuration part makes use of modern compiler features and
206 still retains old compiler behavior for those that run older versions
207 of the OS. The shared library support part includes a variant that
208 uses the RPATH feature, and is available through the special
209 configuration target "alpha-cc-rpath", which will never be selected
211 [Tim Mooney <mooney@dogbert.cc.ndsu.NoDak.edu> via Richard Levitte]
213 *) In ssl3_get_key_exchange (ssl/s3_clnt.c), call ssl3_get_message()
214 with the same message size as in ssl3_get_certificate_request().
215 Otherwise, if no ServerKeyExchange message occurs, CertificateRequest
216 messages might inadvertently be reject as too long.
217 [Petr Lampa <lampa@fee.vutbr.cz>]
219 +) Move SSL_OP_TLS_ROLLBACK_BUG out of the SSL_OP_ALL list of recommended
220 bug workarounds. Rollback attack detection is a security feature.
221 The problem will only arise on OpenSSL servers when TLSv1 is not
222 available (sslv3_server_method() or SSL_OP_NO_TLSv1).
223 Software authors not wanting to support TLSv1 will have special reasons
224 for their choice and can explicitly enable this option.
225 [Bodo Moeller, Lutz Jaenicke]
227 +) Rationalise EVP so it can be extended: don't include a union of
228 cipher/digest structures, add init/cleanup functions. This also reduces
229 the number of header dependencies.
232 +) Make DES key schedule conform to the usual scheme, as well as
233 correcting its structure. This means that calls to DES functions
234 now have to pass a pointer to a des_key_schedule instead of a
235 plain des_key_schedule (which was actually always a pointer
239 +) Enhanced support for IA-64 Unix platforms (well, Linux and HP-UX).
242 *) Modified SSL library such that the verify_callback that has been set
243 specificly for an SSL object with SSL_set_verify() is actually being
244 used. Before the change, a verify_callback set with this function was
245 ignored and the verify_callback() set in the SSL_CTX at the time of
246 the call was used. New function X509_STORE_CTX_set_verify_cb() introduced
247 to allow the necessary settings.
250 +) Initial reduction of linker bloat: the use of some functions, such as
251 PEM causes large amounts of unused functions to be linked in due to
252 poor organisation. For example pem_all.c contains every PEM function
253 which has a knock on effect of linking in large amounts of (unused)
254 ASN1 code. Grouping together similar functions and splitting unrelated
255 functions prevents this.
258 *) Initialize static variable in crypto/dsa/dsa_lib.c and crypto/dh/dh_lib.c
259 explicitly to NULL, as at least on Solaris 8 this seems not always to be
260 done automatically (in contradiction to the requirements of the C
261 standard). This made problems when used from OpenSSH.
264 *) In OpenSSL 0.9.6a and 0.9.6b, crypto/dh/dh_key.c ignored
265 dh->length and always used
267 BN_rand_range(priv_key, dh->p).
269 BN_rand_range() is not necessary for Diffie-Hellman, and this
270 specific range makes Diffie-Hellman unnecessarily inefficient if
271 dh->length (recommended exponent length) is much smaller than the
272 length of dh->p. We could use BN_rand_range() if the order of
273 the subgroup was stored in the DH structure, but we only have
278 BN_rand(priv_key, l, ...)
280 where 'l' is dh->length if this is defined, or BN_num_bits(dh->p)-1
286 RSA_eay_public_encrypt
287 RSA_eay_private_decrypt
288 RSA_eay_private_encrypt (signing)
289 RSA_eay_public_decrypt (signature verification)
291 (default implementations for RSA_public_encrypt,
292 RSA_private_decrypt, RSA_private_encrypt, RSA_public_decrypt),
293 always reject numbers >= n.
296 *) In crypto/rand/md_rand.c, use a new short-time lock CRYPTO_LOCK_RAND2
297 to synchronize access to 'locking_thread'. This is necessary on
298 systems where access to 'locking_thread' (an 'unsigned long'
299 variable) is not atomic.
302 *) In crypto/rand/md_rand.c, set 'locking_thread' to current thread's ID
303 *before* setting the 'crypto_lock_rand' flag. The previous code had
304 a race condition if 0 is a valid thread ID.
305 [Travis Vitek <vitek@roguewave.com>]
307 +) Cleanup of EVP macros.
310 +) Change historical references to {NID,SN,LN}_des_ede and ede3 to add the
314 +) Add initial OCSP responder support to ocsp application. The
315 revocation information is handled using the text based index
316 use by the ca application. The responder can either handle
317 requests generated internally, supplied in files (for example
318 via a CGI script) or using an internal minimal server.
321 +) Add configuration choices to get zlib compression for TLS.
324 +) Changes to Kerberos SSL for RFC 2712 compliance:
325 1. Implemented real KerberosWrapper, instead of just using
326 KRB5 AP_REQ message. [Thanks to Simon Wilkinson <sxw@sxw.org.uk>]
327 2. Implemented optional authenticator field of KerberosWrapper.
329 Added openssl-style ASN.1 macros for Kerberos ticket, ap_req,
330 and authenticator structs; see crypto/krb5/.
332 Generalized Kerberos calls to support multiple Kerberos libraries.
333 [Vern Staats <staatsvr@asc.hpc.mil>,
334 Jeffrey Altman <jaltman@columbia.edu>
337 +) Cause 'openssl speed' to use fully hard-coded DSA keys as it
338 already does with RSA. testdsa.h now has 'priv_key/pub_key'
339 values for each of the key sizes rather than having just
340 parameters (and 'speed' generating keys each time).
343 -) OpenSSL 0.9.6b released [9 July 2001]
345 *) Change ssleay_rand_bytes (crypto/rand/md_rand.c)
346 to avoid a SSLeay/OpenSSL PRNG weakness pointed out by
347 Markku-Juhani O. Saarinen <markku-juhani.saarinen@nokia.com>:
348 PRNG state recovery was possible based on the output of
349 one PRNG request appropriately sized to gain knowledge on
350 'md' followed by enough consecutive 1-byte PRNG requests
351 to traverse all of 'state'.
353 1. When updating 'md_local' (the current thread's copy of 'md')
354 during PRNG output generation, hash all of the previous
355 'md_local' value, not just the half used for PRNG output.
357 2. Make the number of bytes from 'state' included into the hash
358 independent from the number of PRNG bytes requested.
360 The first measure alone would be sufficient to avoid
361 Markku-Juhani's attack. (Actually it had never occurred
362 to me that the half of 'md_local' used for chaining was the
363 half from which PRNG output bytes were taken -- I had always
364 assumed that the secret half would be used.) The second
365 measure makes sure that additional data from 'state' is never
366 mixed into 'md_local' in small portions; this heuristically
367 further strengthens the PRNG.
370 +) Speed up EVP routines.
373 type 8 bytes 64 bytes 256 bytes 1024 bytes 8192 bytes
374 des-cbc 4408.85k 5560.51k 5778.46k 5862.20k 5825.16k
375 des-cbc 4389.55k 5571.17k 5792.23k 5846.91k 5832.11k
376 des-cbc 4394.32k 5575.92k 5807.44k 5848.37k 5841.30k
378 des-cbc 3482.66k 5069.49k 5496.39k 5614.16k 5639.28k
379 des-cbc 3480.74k 5068.76k 5510.34k 5609.87k 5635.52k
380 des-cbc 3483.72k 5067.62k 5504.60k 5708.01k 5724.80k
383 des-cbc 4660.16k 5650.19k 5807.19k 5827.13k 5783.32k
385 des-cbc 3624.96k 5258.21k 5530.91k 5624.30k 5628.26k
388 *) Fix crypto/bn/asm/mips3.s.
391 *) When only the key is given to "enc", the IV is undefined. Print out
392 an error message in this case.
395 +) Added the OS2-EMX target.
396 ["Brian Havard" <brianh@kheldar.apana.org.au> and Richard Levitte]
398 +) Rewrite apps to use NCONF routines instead of the old CONF. New functions
399 to support NCONF routines in extension code. New function CONF_set_nconf()
400 to allow functions which take an NCONF to also handle the old LHASH
401 structure: this means that the old CONF compatible routines can be
402 retained (in particular wrt extensions) without having to duplicate the
403 code. New function X509V3_add_ext_nconf_sk to add extensions to a stack.
406 *) Handle special case when X509_NAME is empty in X509 printing routines.
409 *) In dsa_do_verify (crypto/dsa/dsa_ossl.c), verify that r and s are
410 positive and less than q.
413 +) Enhance the general user interface with mechanisms for inner control
414 and with possibilities to have yes/no kind of prompts.
417 +) Change all calls to low level digest routines in the library and
418 applications to use EVP. Add missing calls to HMAC_cleanup() and
419 don't assume HMAC_CTX can be copied using memcpy().
420 [Verdon Walker <VWalker@novell.com>, Steve Henson]
422 +) Add the possibility to control engines through control names but with
423 arbitrary arguments instead of just a string.
424 Change the key loaders to take a UI_METHOD instead of a callback
425 function pointer. NOTE: this breaks binary compatibility with earlier
426 versions of OpenSSL [engine].
427 Adapt the nCipher code for these new conditions and add a card insertion
431 +) Enhance the general user interface with mechanisms to better support
432 dialog box interfaces, application-defined prompts, the possibility
433 to use defaults (for example default passwords from somewhere else)
434 and interrupts/cancellations.
437 *) Don't change *pointer in CRYPTO_add_lock() is add_lock_callback is
438 used: it isn't thread safe and the add_lock_callback should handle
440 [Paul Rose <Paul.Rose@bridge.com>]
442 *) Verify that incoming data obeys the block size in
443 ssl3_enc (ssl/s3_enc.c) and tls1_enc (ssl/t1_enc.c).
446 +) Tidy up PKCS#12 attribute handling. Add support for the CSP name
447 attribute in PKCS#12 files, add new -CSP option to pkcs12 utility.
451 [Ulf Möller, Bodo Möller]
453 *) The countermeasure against Bleichbacher's attack on PKCS #1 v1.5
454 RSA encryption was accidentally removed in s3_srvr.c in OpenSSL 0.9.5
455 when fixing the server behaviour for backwards-compatible 'client
456 hello' messages. (Note that the attack is impractical against
457 SSL 3.0 and TLS 1.0 anyway because length and version checking
458 means that the probability of guessing a valid ciphertext is
459 around 2^-40; see section 5 in Bleichenbacher's CRYPTO '98
462 Before 0.9.5, the countermeasure (hide the error by generating a
463 random 'decryption result') did not work properly because
464 ERR_clear_error() was missing, meaning that SSL_get_error() would
465 detect the supposedly ignored error.
467 Both problems are now fixed.
470 *) In crypto/bio/bf_buff.c, increase DEFAULT_BUFFER_SIZE to 4096
471 (previously it was 1024).
474 +) Fix a memory leak in 'sk_dup()' in the case reallocation fails. (Also
475 tidy up some unnecessarily weird code in 'sk_new()').
476 [Geoff, reported by Diego Tartara <dtartara@novamens.com>]
478 +) Change the key loading routines for ENGINEs to use the same kind
479 callback (pem_password_cb) as all other routines that need this
483 *) Fix for compatibility mode trust settings: ignore trust settings
484 unless some valid trust or reject settings are present.
487 *) Fix for blowfish EVP: its a variable length cipher.
490 +) Increase ENTROPY_NEEDED to 32 bytes, as Rijndael can operate with
491 256 bit (=32 byte) keys. Of course seeding with more entropy bytes
492 than this minimum value is recommended.
495 +) New random seeder for OpenVMS, using the system process statistics
496 that are easily reachable.
499 +) Windows apparently can't transparently handle global
500 variables defined in DLLs. Initialisations such as:
502 const ASN1_ITEM *it = &ASN1_INTEGER_it;
504 wont compile. This is used by the any applications that need to
505 declare their own ASN1 modules. This was fixed by adding the option
506 EXPORT_VAR_AS_FN to all Win32 platforms, although this isn't strictly
507 needed for static libraries under Win32.
510 +) New functions X509_PURPOSE_set() and X509_TRUST_set() to handle
511 setting of purpose and trust fields. New X509_STORE trust and
512 purpose functions and tidy up setting in other SSL functions.
515 +) Add copies of X509_STORE_CTX fields and callbacks to X509_STORE
516 structure. These are inherited by X509_STORE_CTX when it is
517 initialised. This allows various defaults to be set in the
518 X509_STORE structure (such as flags for CRL checking and custom
519 purpose or trust settings) for functions which only use X509_STORE_CTX
520 internally such as S/MIME.
522 Modify X509_STORE_CTX_purpose_inherit() so it only sets purposes and
523 trust settings if they are not set in X509_STORE. This allows X509_STORE
524 purposes and trust (in S/MIME for example) to override any set by default.
526 Add command line options for CRL checking to smime, s_client and s_server
530 +) Initial CRL based revocation checking. If the CRL checking flag(s)
531 are set then the CRL is looked up in the X509_STORE structure and
532 its validity and signature checked, then if the certificate is found
533 in the CRL the verify fails with a revoked error.
535 Various new CRL related callbacks added to X509_STORE_CTX structure.
537 Command line options added to 'verify' application to support this.
539 This needs some additional work, such as being able to handle multiple
540 CRLs with different times, extension based lookup (rather than just
541 by subject name) and ultimately more complete V2 CRL extension
545 +) Add a general user interface API (crypto/ui/). This is designed
546 to replace things like des_read_password and friends (backward
547 compatibility functions using this new API are provided).
548 The purpose is to remove prompting functions from the DES code
549 section as well as provide for prompting through dialog boxes in
550 a window system and the like.
553 *) Fix various bugs related to DSA S/MIME verification. Handle missing
554 parameters in DSA public key structures and return an error in the
555 DSA routines if parameters are absent.
558 *) In versions up to 0.9.6, RAND_file_name() resorted to file ".rnd"
559 in the current directory if neither $RANDFILE nor $HOME was set.
560 RAND_file_name() in 0.9.6a returned NULL in this case. This has
561 caused some confusion to Windows users who haven't defined $HOME.
562 Thus RAND_file_name() is changed again: e_os.h can define a
563 DEFAULT_HOME, which will be used if $HOME is not set.
564 For Windows, we use "C:"; on other platforms, we still require
565 environment variables.
567 +) Add "ex_data" support to ENGINE so implementations can add state at a
568 per-structure level rather than having to store it globally.
571 +) Make it possible for ENGINE structures to be copied when retrieved by
572 ENGINE_by_id() if the ENGINE specifies a new flag: ENGINE_FLAGS_BY_ID_COPY.
573 This causes the "original" ENGINE structure to act like a template,
574 analogous to the RSA vs. RSA_METHOD type of separation. Because of this
575 operational state can be localised to each ENGINE structure, despite the
576 fact they all share the same "methods". New ENGINE structures returned in
577 this case have no functional references and the return value is the single
578 structural reference. This matches the single structural reference returned
579 by ENGINE_by_id() normally, when it is incremented on the pre-existing
583 +) Fix ASN1 decoder when decoding type ANY and V_ASN1_OTHER: since this
584 needs to match any other type at all we need to manually clear the
588 +) Changes to the "openssl engine" utility to include;
589 - verbosity levels ('-v', '-vv', and '-vvv') that provide information
590 about an ENGINE's available control commands.
591 - executing control commands from command line arguments using the
592 '-pre' and '-post' switches. '-post' is only used if '-t' is
593 specified and the ENGINE is successfully initialised. The syntax for
594 the individual commands are colon-separated, for example;
595 openssl engine chil -pre FORK_CHECK:0 -pre SO_PATH:/lib/test.so
598 +) New dynamic control command support for ENGINEs. ENGINEs can now
599 declare their own commands (numbers), names (strings), descriptions,
600 and input types for run-time discovery by calling applications. A
601 subset of these commands are implicitly classed as "executable"
602 depending on their input type, and only these can be invoked through
603 the new string-based API function ENGINE_ctrl_cmd_string(). (Eg. this
604 can be based on user input, config files, etc). The distinction is
605 that "executable" commands cannot return anything other than a boolean
606 result and can only support numeric or string input, whereas some
607 discoverable commands may only be for direct use through
608 ENGINE_ctrl(), eg. supporting the exchange of binary data, function
609 pointers, or other custom uses. The "executable" commands are to
610 support parameterisations of ENGINE behaviour that can be
611 unambiguously defined by ENGINEs and used consistently across any
612 OpenSSL-based application. Commands have been added to all the
613 existing hardware-supporting ENGINEs, noticeably "SO_PATH" to allow
614 control over shared-library paths without source code alterations.
617 +) Changed all ENGINE implementations to dynamically allocate their
618 ENGINEs rather than declaring them statically. Apart from this being
619 necessary with the removal of the ENGINE_FLAGS_MALLOCED distinction,
620 this also allows the implementations to compile without using the
621 internal engine_int.h header.
624 +) Minor adjustment to "rand" code. RAND_get_rand_method() now returns a
625 'const' value. Any code that should be able to modify a RAND_METHOD
626 should already have non-const pointers to it (ie. they should only
627 modify their own ones).
630 +) Made a variety of little tweaks to the ENGINE code.
631 - "atalla" and "ubsec" string definitions were moved from header files
632 to C code. "nuron" string definitions were placed in variables
633 rather than hard-coded - allowing parameterisation of these values
634 later on via ctrl() commands.
635 - Removed unused "#if 0"'d code.
636 - Fixed engine list iteration code so it uses ENGINE_free() to release
637 structural references.
638 - Constified the RAND_METHOD element of ENGINE structures.
639 - Constified various get/set functions as appropriate and added
640 missing functions (including a catch-all ENGINE_cpy that duplicates
641 all ENGINE values onto a new ENGINE except reference counts/state).
642 - Removed NULL parameter checks in get/set functions. Setting a method
643 or function to NULL is a way of cancelling out a previously set
644 value. Passing a NULL ENGINE parameter is just plain stupid anyway
645 and doesn't justify the extra error symbols and code.
646 - Deprecate the ENGINE_FLAGS_MALLOCED define and move the area for
647 flags from engine_int.h to engine.h.
648 - Changed prototypes for ENGINE handler functions (init(), finish(),
649 ctrl(), key-load functions, etc) to take an (ENGINE*) parameter.
652 *) Move 'if (!initialized) RAND_poll()' into regions protected by
653 CRYPTO_LOCK_RAND. This is not strictly necessary, but avoids
654 having multiple threads call RAND_poll() concurrently.
657 *) In crypto/rand/md_rand.c, replace 'add_do_not_lock' flag by a
658 combination of a flag and a thread ID variable.
659 Otherwise while one thread is in ssleay_rand_bytes (which sets the
660 flag), *other* threads can enter ssleay_add_bytes without obeying
661 the CRYPTO_LOCK_RAND lock (and may even illegally release the lock
662 that they do not hold after the first thread unsets add_do_not_lock).
665 +) Implement binary inversion algorithm for BN_mod_inverse in addition
666 to the algorithm using long division. The binary algorithm can be
667 used only if the modulus is odd. On 32-bit systems, it is faster
668 only for relatively small moduli (roughly 20-30% for 128-bit moduli,
669 roughly 5-15% for 256-bit moduli), so we use it only for moduli
670 up to 450 bits. In 64-bit environments, the binary algorithm
671 appears to be advantageous for much longer moduli; here we use it
672 for moduli up to 2048 bits.
675 *) Change bctest again: '-x' expressions are not available in all
679 -) OpenSSL 0.9.6a released [5 Apr 2001]
681 *) Fix a couple of memory leaks in PKCS7_dataDecode()
682 [Steve Henson, reported by Heyun Zheng <hzheng@atdsprint.com>]
684 *) Change Configure and Makefiles to provide EXE_EXT, which will contain
685 the default extension for executables, if any. Also, make the perl
686 scripts that use symlink() to test if it really exists and use "cp"
687 if it doesn't. All this made OpenSSL compilable and installable in
691 +) Rewrite CHOICE field setting in ASN1_item_ex_d2i(). The old code
692 could not support the combine flag in choice fields.
695 -) Fix for asn1_GetSequence() for indefinite length constructed data.
696 If SEQUENCE is length is indefinite just set c->slen to the total
697 amount of data available.
698 [Steve Henson, reported by shige@FreeBSD.org]
699 [This change does not apply to 0.9.7.]
701 *) Change bctest to avoid here-documents inside command substitution
702 (workaround for FreeBSD /bin/sh bug).
703 For compatibility with Ultrix, avoid shell functions (introduced
704 in the bctest version that searches along $PATH).
707 *) Rename 'des_encrypt' to 'des_encrypt1'. This avoids the clashes
708 with des_encrypt() defined on some operating systems, like Solaris
712 *) Check the result of RSA-CRT (see D. Boneh, R. DeMillo, R. Lipton:
713 On the Importance of Eliminating Errors in Cryptographic
714 Computations, J. Cryptology 14 (2001) 2, 101-119,
715 http://theory.stanford.edu/~dabo/papers/faults.ps.gz).
718 *) MIPS assembler BIGNUM division bug fix.
721 *) Disabled incorrect Alpha assembler code.
724 -) Fix PKCS#7 decode routines so they correctly update the length
725 after reading an EOC for the EXPLICIT tag.
727 [This change does not apply to 0.9.7.]
729 *) Fix bug in PKCS#12 key generation routines. This was triggered
730 if a 3DES key was generated with a 0 initial byte. Include
731 PKCS12_BROKEN_KEYGEN compilation option to retain the old
732 (but broken) behaviour.
735 *) Enhance bctest to search for a working bc along $PATH and print
737 [Tim Rice <tim@multitalents.net> via Richard Levitte]
739 +) Add a 'copy_extensions' option to the 'ca' utility. This copies
740 extensions from a certificate request to the certificate.
743 +) Allow multiple 'certopt' and 'nameopt' options to be separated
744 by commas. Add 'namopt' and 'certopt' options to the 'ca' config
745 file: this allows the display of the certificate about to be
746 signed to be customised, to allow certain fields to be included
747 or excluded and extension details. The old system didn't display
748 multicharacter strings properly, omitted fields not in the policy
749 and couldn't display additional details such as extensions.
752 *) Fix memory leaks in err.c: free err_data string if necessary;
753 don't write to the wrong index in ERR_set_error_data.
756 +) Function EC_POINTs_mul for simultaneous scalar multiplication
757 of an arbitrary number of elliptic curve points, optionally
758 including the generator defined for the EC_GROUP.
759 EC_POINT_mul is a simple wrapper function for the typical case
760 that the point list has just one item (besides the optional
764 +) First EC_METHODs for curves over GF(p):
766 EC_GFp_simple_method() uses the basic BN_mod_mul and BN_mod_sqr
767 operations and provides various method functions that can also
768 operate with faster implementations of modular arithmetic.
770 EC_GFp_mont_method() reuses most functions that are part of
771 EC_GFp_simple_method, but uses Montgomery arithmetic.
773 [Bodo Moeller; point addition and point doubling
774 implementation directly derived from source code provided by
775 Lenka Fibikova <fibikova@exp-math.uni-essen.de>]
777 +) Framework for elliptic curves (crypto/ec/ec.h, crypto/ec/ec_lcl.h,
780 Curves are EC_GROUP objects (with an optional group generator)
781 based on EC_METHODs that are built into the library.
783 Points are EC_POINT objects based on EC_GROUP objects.
785 Most of the framework would be able to handle curves over arbitrary
786 finite fields, but as there are no obvious types for fields other
787 than GF(p), some functions are limited to that for now.
790 +) Add the -HTTP option to s_server. It is similar to -WWW, but requires
791 that the file contains a complete HTTP response.
794 +) Add the ec directory to mkdef.pl and mkfiles.pl. In mkdef.pl
795 change the def and num file printf format specifier from "%-40sXXX"
796 to "%-39s XXX". The latter will always guarantee a space after the
797 field while the former will cause them to run together if the field
798 is 40 of more characters long.
801 +) Constify the cipher and digest 'method' functions and structures
802 and modify related functions to take constant EVP_MD and EVP_CIPHER
806 *) Implement ssl23_peek (analogous to ssl23_read), which previously
810 *) Replace rdtsc with _emit statements for VC++ version 5.
811 [Jeremy Cooper <jeremy@baymoo.org>]
813 +) Hide BN_CTX structure details in bn_lcl.h instead of publishing them
814 in <openssl/bn.h>. Also further increase BN_CTX_NUM to 32.
817 +) Modify EVP_Digest*() routines so they now return values. Although the
818 internal software routines can never fail additional hardware versions
822 +) Clean up crypto/err/err.h and change some error codes to avoid conflicts:
824 Previously ERR_R_FATAL was too small and coincided with ERR_LIB_PKCS7
825 (= ERR_R_PKCS7_LIB); it is now 64 instead of 32.
828 ERR_R_NESTED_ASN1_ERROR
830 ERR_R_MISSING_ASN1_EOS
831 were 4 .. 9, conflicting with
832 ERR_LIB_RSA (= ERR_R_RSA_LIB)
834 ERR_LIB_PEM (= ERR_R_PEM_LIB).
835 They are now 58 .. 63 (i.e., just below ERR_R_FATAL).
837 Add new error code 'ERR_R_INTERNAL_ERROR'.
840 +) Don't overuse locks in crypto/err/err.c: For data retrieval, CRYPTO_r_lock
844 +) New option '-subj arg' for 'openssl req' and 'openssl ca'. This
845 sets the subject name for a new request or supersedes the
846 subject name in a given request. Formats that can be parsed are
847 'CN=Some Name, OU=myOU, C=IT'
849 'CN=Some Name/OU=myOU/C=IT'.
851 Add options '-batch' and '-verbose' to 'openssl req'.
852 [Massimiliano Pala <madwolf@hackmasters.net>]
854 *) Make it possible to reuse SSLv2 sessions.
857 +) Introduce the possibility to access global variables through
858 functions on platform were that's the best way to handle exporting
859 global variables in shared libraries. To enable this functionality,
860 one must configure with "EXPORT_VAR_AS_FN" or defined the C macro
861 "OPENSSL_EXPORT_VAR_AS_FUNCTION" in crypto/opensslconf.h (the latter
862 is normally done by Configure or something similar).
864 To implement a global variable, use the macro OPENSSL_IMPLEMENT_GLOBAL
865 in the source file (foo.c) like this:
867 OPENSSL_IMPLEMENT_GLOBAL(int,foo)=1;
868 OPENSSL_IMPLEMENT_GLOBAL(double,bar);
870 To declare a global variable, use the macros OPENSSL_DECLARE_GLOBAL
871 and OPENSSL_GLOBAL_REF in the header file (foo.h) like this:
873 OPENSSL_DECLARE_GLOBAL(int,foo);
874 #define foo OPENSSL_GLOBAL_REF(foo)
875 OPENSSL_DECLARE_GLOBAL(double,bar);
876 #define bar OPENSSL_GLOBAL_REF(bar)
878 The #defines are very important, and therefore so is including the
879 header file everywhere where the defined globals are used.
881 The macro OPENSSL_EXPORT_VAR_AS_FUNCTION also affects the definition
882 of ASN.1 items, but that structure is a bit different.
884 The largest change is in util/mkdef.pl which has been enhanced with
885 better and easier to understand logic to choose which symbols should
886 go into the Windows .def files as well as a number of fixes and code
887 cleanup (among others, algorithm keywords are now sorted
888 lexicographically to avoid constant rewrites).
891 *) In copy_email() check for >= 0 as a return value for
892 X509_NAME_get_index_by_NID() since 0 is a valid index.
893 [Steve Henson reported by Massimiliano Pala <madwolf@opensca.org>]
895 +) In BN_div() keep a copy of the sign of 'num' before writing the
896 result to 'rm' because if rm==num the value will be overwritten
897 and produce the wrong result if 'num' is negative: this caused
898 problems with BN_mod() and BN_nnmod().
901 +) Function OCSP_request_verify(). This checks the signature on an
902 OCSP request and verifies the signer certificate. The signer
903 certificate is just checked for a generic purpose and OCSP request
907 +) Add OCSP_check_validity() function to check the validity of OCSP
908 responses. OCSP responses are prepared in real time and may only
909 be a few seconds old. Simply checking that the current time lies
910 between thisUpdate and nextUpdate max reject otherwise valid responses
911 caused by either OCSP responder or client clock inaccuracy. Instead
912 we allow thisUpdate and nextUpdate to fall within a certain period of
913 the current time. The age of the response can also optionally be
914 checked. Two new options -validity_period and -status_age added to
918 +) If signature or public key algorithm is unrecognized print out its
919 OID rather that just UNKNOWN.
922 *) Avoid coredump with unsupported or invalid public keys by checking if
923 X509_get_pubkey() fails in PKCS7_verify(). Fix memory leak when
924 PKCS7_verify() fails with non detached data.
927 +) Change OCSP_cert_to_id() to tolerate a NULL subject certificate and
928 OCSP_cert_id_new() a NULL serialNumber. This allows a partial certificate
929 ID to be generated from the issuer certificate alone which can then be
930 passed to OCSP_id_issuer_cmp().
933 +) New compilation option ASN1_ITEM_FUNCTIONS. This causes the new
934 ASN1 modules to export functions returning ASN1_ITEM pointers
935 instead of the ASN1_ITEM structures themselves. This adds several
936 new macros which allow the underlying ASN1 function/structure to
937 be accessed transparently. As a result code should not use ASN1_ITEM
938 references directly (such as &X509_it) but instead use the relevant
939 macros (such as ASN1_ITEM_rptr(X509)). This option is to allow
940 use of the new ASN1 code on platforms where exporting structures
941 is problematical (for example in shared libraries) but exporting
942 functions returning pointers to structures is not.
945 +) Add support for overriding the generation of SSL/TLS session IDs.
946 These callbacks can be registered either in an SSL_CTX or per SSL.
947 The purpose of this is to allow applications to control, if they wish,
948 the arbitrary values chosen for use as session IDs, particularly as it
949 can be useful for session caching in multiple-server environments. A
950 command-line switch for testing this (and any client code that wishes
951 to use such a feature) has been added to "s_server".
952 [Geoff Thorpe, Lutz Jaenicke]
954 +) Modify mkdef.pl to recognise and parse preprocessor conditionals
955 of the form '#if defined(...) || defined(...) || ...' and
956 '#if !defined(...) && !defined(...) && ...'. This also avoids
957 the growing number of special cases it was previously handling.
960 *) Don't use getenv in library functions when run as setuid/setgid.
961 New function OPENSSL_issetugid().
964 +) Make all configuration macros available for application by making
965 sure they are available in opensslconf.h, by giving them names starting
966 with "OPENSSL_" to avoid conflicts with other packages and by making
967 sure e_os2.h will cover all platform-specific cases together with
969 Additionally, it is now possible to define configuration/platform-
970 specific names (called "system identities"). In the C code, these
971 are prefixed with "OPENSSL_SYSNAME_". e_os2.h will create another
972 macro with the name beginning with "OPENSSL_SYS_", which is determined
973 from "OPENSSL_SYSNAME_*" or compiler-specific macros depending on
977 +) New option -set_serial to 'req' and 'x509' this allows the serial
978 number to use to be specified on the command line. Previously self
979 signed certificates were hard coded with serial number 0 and the
980 CA options of 'x509' had to use a serial number in a file which was
984 *) Avoid false positives in memory leak detection code (crypto/mem_dbg.c)
985 due to incorrect handling of multi-threading:
987 1. Fix timing glitch in the MemCheck_off() portion of CRYPTO_mem_ctrl().
989 2. Fix logical glitch in is_MemCheck_on() aka CRYPTO_is_mem_check_on().
991 3. Count how many times MemCheck_off() has been called so that
992 nested use can be treated correctly. This also avoids
993 inband-signalling in the previous code (which relied on the
994 assumption that thread ID 0 is impossible).
997 +) New options to 'ca' utility to support V2 CRL entry extensions.
998 Currently CRL reason, invalidity date and hold instruction are
999 supported. Add new CRL extensions to V3 code and some new objects.
1002 *) Add "-rand" option also to s_client and s_server.
1005 +) New function EVP_CIPHER_CTX_set_padding() this is used to
1006 disable standard block padding (aka PKCS#5 padding) in the EVP
1007 API, which was previously mandatory. This means that the data is
1008 not padded in any way and so the total length much be a multiple
1009 of the block size, otherwise an error occurs.
1012 +) Initial (incomplete) OCSP SSL support.
1015 *) Fix CPU detection on Irix 6.x.
1016 [Kurt Hockenbury <khockenb@stevens-tech.edu> and
1017 "Bruce W. Forsberg" <bruce.forsberg@baesystems.com>]
1019 +) New function OCSP_parse_url(). This splits up a URL into its host,
1020 port and path components: primarily to parse OCSP URLs. New -url
1021 option to ocsp utility.
1024 +) New nonce behavior. The return value of OCSP_check_nonce() now
1025 reflects the various checks performed. Applications can decide
1026 whether to tolerate certain situations such as an absent nonce
1027 in a response when one was present in a request: the ocsp application
1028 just prints out a warning. New function OCSP_add1_basic_nonce()
1029 this is to allow responders to include a nonce in a response even if
1030 the request is nonce-less.
1033 *) Fix X509_NAME bug which produced incorrect encoding if X509_NAME
1036 [This change does not apply to 0.9.7.]
1038 *) Use the cached encoding of an X509_NAME structure rather than
1039 copying it. This is apparently the reason for the libsafe "errors"
1040 but the code is actually correct.
1043 +) Disable stdin buffering in load_cert (apps/apps.c) so that no certs are
1044 skipped when using openssl x509 multiple times on a single input file,
1045 e.g. "(openssl x509 -out cert1; openssl x509 -out cert2) <certs".
1048 +) Make ASN1_UTCTIME_set_string() and ASN1_GENERALIZEDTIME_set_string()
1049 set string type: to handle setting ASN1_TIME structures. Fix ca
1050 utility to correctly initialize revocation date of CRLs.
1053 +) New option SSL_OP_CIPHER_SERVER_PREFERENCE allows the server to override
1054 the clients preferred ciphersuites and rather use its own preferences.
1055 Should help to work around M$ SGC (Server Gated Cryptography) bug in
1056 Internet Explorer by ensuring unchanged hash method during stepup.
1057 (Also replaces the broken/deactivated SSL_OP_NON_EXPORT_FIRST option.)
1060 +) Make mkdef.pl recognise all DECLARE_ASN1 macros, change rijndael
1061 to aes and add a new 'exist' option to print out symbols that don't
1065 +) Additional options to ocsp utility to allow flags to be set and
1066 additional certificates supplied.
1069 +) Add the option -VAfile to 'openssl ocsp', so the user can give the
1070 OCSP client a number of certificate to only verify the response
1074 *) Add new function BN_rand_range(), and fix DSA_sign_setup() to prevent
1075 Bleichenbacher's DSA attack.
1076 Extend BN_[pseudo_]rand: As before, top=1 forces the highest two bits
1077 to be set and top=0 forces the highest bit to be set; top=-1 is new
1078 and leaves the highest bit random.
1079 [Ulf Moeller, Bodo Moeller]
1081 +) Update Rijndael code to version 3.0 and change EVP AES ciphers to
1082 handle the new API. Currently only ECB, CBC modes supported. Add new
1083 AES OIDs. Add TLS AES ciphersuites as described in the "AES Ciphersuites
1084 for TLS" draft-ietf-tls-ciphersuite-03.txt.
1085 [Ben Laurie, Steve Henson]
1087 *) In the NCONF_...-based implementations for CONF_... queries
1088 (crypto/conf/conf_lib.c), if the input LHASH is NULL, avoid using
1089 a temporary CONF structure with the data component set to NULL
1090 (which gives segmentation faults in lh_retrieve).
1091 Instead, use NULL for the CONF pointer in CONF_get_string and
1092 CONF_get_number (which may use environment variables) and directly
1093 return NULL from CONF_get_section.
1096 *) Fix potential buffer overrun for EBCDIC.
1099 +) New function OCSP_copy_nonce() to copy nonce value (if present) from
1100 request to response.
1103 +) Functions for OCSP responders. OCSP_request_onereq_count(),
1104 OCSP_request_onereq_get0(), OCSP_onereq_get0_id() and OCSP_id_get0_info()
1105 extract information from a certificate request. OCSP_response_create()
1106 creates a response and optionally adds a basic response structure.
1107 OCSP_basic_add1_status() adds a complete single response to a basic
1108 response and returns the OCSP_SINGLERESP structure just added (to allow
1109 extensions to be included for example). OCSP_basic_add1_cert() adds a
1110 certificate to a basic response and OCSP_basic_sign() signs a basic
1111 response with various flags. New helper functions ASN1_TIME_check()
1112 (checks validity of ASN1_TIME structure) and ASN1_TIME_to_generalizedtime()
1113 (converts ASN1_TIME to GeneralizedTime).
1116 +) Various new functions. EVP_Digest() combines EVP_Digest{Init,Update,Final}()
1117 in a single operation. X509_get0_pubkey_bitstr() extracts the public_key
1118 structure from a certificate. X509_pubkey_digest() digests the public_key
1119 contents: this is used in various key identifiers.
1122 *) Tolerate nonRepudiation as being valid for S/MIME signing and certSign
1123 keyUsage if basicConstraints absent for a CA.
1126 *) Make SMIME_write_PKCS7() write mail header values with a format that
1127 is more generally accepted (no spaces before the semicolon), since
1128 some programs can't parse those values properly otherwise. Also make
1129 sure BIO's that break lines after each write do not create invalid
1133 +) Make sk_sort() tolerate a NULL argument.
1134 [Steve Henson reported by Massimiliano Pala <madwolf@comune.modena.it>]
1136 +) New OCSP verify flag OCSP_TRUSTOTHER. When set the "other" certificates
1137 passed by the function are trusted implicitly. If any of them signed the
1138 response then it is assumed to be valid and is not verified.
1141 -) Make the CRL encoding routines work with empty SEQUENCE OF. The
1142 macros previously used would not encode an empty SEQUENCE OF
1143 and break the signature.
1145 [This change does not apply to 0.9.7.]
1147 *) Zero the premaster secret after deriving the master secret in
1151 +) In PKCS7_set_type() initialise content_type in PKCS7_ENC_CONTENT
1152 to data. This was previously part of the PKCS7 ASN1 code. This
1153 was causing problems with OpenSSL created PKCS#12 and PKCS#7 structures.
1154 [Steve Henson, reported by Kenneth R. Robinette
1155 <support@securenetterm.com>]
1157 +) Add CRYPTO_push_info() and CRYPTO_pop_info() calls to new ASN1
1158 routines: without these tracing memory leaks is very painful.
1159 Fix leaks in PKCS12 and PKCS7 routines.
1162 *) Add some EVP_add_digest_alias registrations (as found in
1163 OpenSSL_add_all_digests()) to SSL_library_init()
1164 aka OpenSSL_add_ssl_algorithms(). This provides improved
1165 compatibility with peers using X.509 certificates
1166 with unconventional AlgorithmIdentifier OIDs.
1169 *) Fix for Irix with NO_ASM.
1170 ["Bruce W. Forsberg" <bruce.forsberg@baesystems.com>]
1172 *) ./config script fixes.
1173 [Ulf Moeller, Richard Levitte]
1175 +) Make X509_time_adj() cope with the new behaviour of ASN1_TIME_new().
1176 Previously it initialised the 'type' argument to V_ASN1_UTCTIME which
1177 effectively meant GeneralizedTime would never be used. Now it
1178 is initialised to -1 but X509_time_adj() now has to check the value
1179 and use ASN1_TIME_set() if the value is not V_ASN1_UTCTIME or
1180 V_ASN1_GENERALIZEDTIME, without this it always uses GeneralizedTime.
1181 [Steve Henson, reported by Kenneth R. Robinette
1182 <support@securenetterm.com>]
1184 +) Fixes to BN_to_ASN1_INTEGER when bn is zero. This would previously
1185 result in a zero length in the ASN1_INTEGER structure which was
1186 not consistent with the structure when d2i_ASN1_INTEGER() was used
1187 and would cause ASN1_INTEGER_cmp() to fail. Enhance s2i_ASN1_INTEGER()
1188 to cope with hex and negative integers. Fix bug in i2a_ASN1_INTEGER()
1189 where it did not print out a minus for negative ASN1_INTEGER.
1192 *) Fix 'openssl passwd -1'.
1195 +) Add summary printout to ocsp utility. The various functions which
1196 convert status values to strings have been renamed to:
1197 OCSP_response_status_str(), OCSP_cert_status_str() and
1198 OCSP_crl_reason_str() and are no longer static. New options
1199 to verify nonce values and to disable verification. OCSP response
1200 printout format cleaned up.
1203 +) Add additional OCSP certificate checks. These are those specified
1204 in RFC2560. This consists of two separate checks: the CA of the
1205 certificate being checked must either be the OCSP signer certificate
1206 or the issuer of the OCSP signer certificate. In the latter case the
1207 OCSP signer certificate must contain the OCSP signing extended key
1208 usage. This check is performed by attempting to match the OCSP
1209 signer or the OCSP signer CA to the issuerNameHash and issuerKeyHash
1210 in the OCSP_CERTID structures of the response.
1213 +) Initial OCSP certificate verification added to OCSP_basic_verify()
1214 and related routines. This uses the standard OpenSSL certificate
1215 verify routines to perform initial checks (just CA validity) and
1216 to obtain the certificate chain. Then additional checks will be
1217 performed on the chain. Currently the root CA is checked to see
1218 if it is explicitly trusted for OCSP signing. This is used to set
1219 a root CA as a global signing root: that is any certificate that
1220 chains to that CA is an acceptable OCSP signing certificate.
1223 +) New '-extfile ...' option to 'openssl ca' for reading X.509v3
1224 extensions from a separate configuration file.
1225 As when reading extensions from the main configuration file,
1226 the '-extensions ...' option may be used for specifying the
1228 [Massimiliano Pala <madwolf@comune.modena.it>]
1230 *) Change PKCS12_key_gen_asc() so it can cope with non null
1231 terminated strings whose length is passed in the passlen
1232 parameter, for example from PEM callbacks. This was done
1233 by adding an extra length parameter to asc2uni().
1234 [Steve Henson, reported by <oddissey@samsung.co.kr>]
1236 +) New OCSP utility. Allows OCSP requests to be generated or
1237 read. The request can be sent to a responder and the output
1238 parsed, outputed or printed in text form. Not complete yet:
1239 still needs to check the OCSP response validity.
1242 +) New subcommands for 'openssl ca':
1243 'openssl ca -status <serial>' prints the status of the cert with
1244 the given serial number (according to the index file).
1245 'openssl ca -updatedb' updates the expiry status of certificates
1247 [Massimiliano Pala <madwolf@comune.modena.it>]
1249 +) New '-newreq-nodes' command option to CA.pl. This is like
1250 '-newreq', but calls 'openssl req' with the '-nodes' option
1251 so that the resulting key is not encrypted.
1252 [Damien Miller <djm@mindrot.org>]
1254 +) New configuration for the GNU Hurd.
1255 [Jonathan Bartlett <johnnyb@wolfram.com> via Richard Levitte]
1257 +) Initial code to implement OCSP basic response verify. This
1258 is currently incomplete. Currently just finds the signer's
1259 certificate and verifies the signature on the response.
1262 +) New SSLeay_version code SSLEAY_DIR to determine the compiled-in
1263 value of OPENSSLDIR. This is available via the new '-d' option
1264 to 'openssl version', and is also included in 'openssl version -a'.
1267 *) Fix C code generated by 'openssl dsaparam -C': If a BN_bin2bn
1268 call failed, free the DSA structure.
1271 +) Allowing defining memory allocation callbacks that will be given
1272 file name and line number information in additional arguments
1273 (a const char* and an int). The basic functionality remains, as
1274 well as the original possibility to just replace malloc(),
1275 realloc() and free() by functions that do not know about these
1276 additional arguments. To register and find out the current
1277 settings for extended allocation functions, the following
1278 functions are provided:
1280 CRYPTO_set_mem_ex_functions
1281 CRYPTO_set_locked_mem_ex_functions
1282 CRYPTO_get_mem_ex_functions
1283 CRYPTO_get_locked_mem_ex_functions
1285 These work the same way as CRYPTO_set_mem_functions and friends.
1286 CRYPTO_get_[locked_]mem_functions now writes 0 where such an
1287 extended allocation function is enabled.
1288 Similarly, CRYPTO_get_[locked_]mem_ex_functions writes 0 where
1289 a conventional allocation function is enabled.
1290 [Richard Levitte, Bodo Moeller]
1292 *) Fix to uni2asc() to cope with zero length Unicode strings.
1293 These are present in some PKCS#12 files.
1296 +) Finish off removing the remaining LHASH function pointer casts.
1297 There should no longer be any prototype-casting required when using
1298 the LHASH abstraction, and any casts that remain are "bugs". See
1299 the callback types and macros at the head of lhash.h for details
1300 (and "OBJ_cleanup" in crypto/objects/obj_dat.c as an example).
1303 +) Add automatic query of EGD sockets in RAND_poll() for the unix variant.
1304 If /dev/[u]random devices are not available or do not return enough
1305 entropy, EGD style sockets (served by EGD or PRNGD) will automatically
1307 The locations /var/run/egd-pool, /dev/egd-pool, /etc/egd-pool, and
1308 /etc/entropy will be queried once each in this sequence, quering stops
1309 when enough entropy was collected without querying more sockets.
1312 +) Change the Unix RAND_poll() variant to be able to poll several
1313 random devices, as specified by DEVRANDOM, until a sufficient amount
1314 of data has been collected. We spend at most 10 ms on each file
1315 (select timeout) and read in non-blocking mode. DEVRANDOM now
1316 defaults to the list "/dev/urandom", "/dev/random", "/dev/srandom"
1317 (previously it was just the string "/dev/urandom"), so on typical
1318 platforms the 10 ms delay will never occur.
1319 Also separate out the Unix variant to its own file, rand_unix.c.
1320 For VMS, there's a currently-empty rand_vms.c.
1323 +) Move OCSP client related routines to ocsp_cl.c. These
1324 provide utility functions which an application needing
1325 to issue a request to an OCSP responder and analyse the
1326 response will typically need: as opposed to those which an
1327 OCSP responder itself would need which will be added later.
1329 OCSP_request_sign() signs an OCSP request with an API similar
1330 to PKCS7_sign(). OCSP_response_status() returns status of OCSP
1331 response. OCSP_response_get1_basic() extracts basic response
1332 from response. OCSP_resp_find_status(): finds and extracts status
1333 information from an OCSP_CERTID structure (which will be created
1334 when the request structure is built). These are built from lower
1335 level functions which work on OCSP_SINGLERESP structures but
1336 wont normally be used unless the application wishes to examine
1337 extensions in the OCSP response for example.
1339 Replace nonce routines with a pair of functions.
1340 OCSP_request_add1_nonce() adds a nonce value and optionally
1341 generates a random value. OCSP_check_nonce() checks the
1342 validity of the nonce in an OCSP response.
1345 +) Change function OCSP_request_add() to OCSP_request_add0_id().
1346 This doesn't copy the supplied OCSP_CERTID and avoids the
1347 need to free up the newly created id. Change return type
1348 to OCSP_ONEREQ to return the internal OCSP_ONEREQ structure.
1349 This can then be used to add extensions to the request.
1350 Deleted OCSP_request_new(), since most of its functionality
1351 is now in OCSP_REQUEST_new() (and the case insensitive name
1352 clash) apart from the ability to set the request name which
1353 will be added elsewhere.
1356 +) Update OCSP API. Remove obsolete extensions argument from
1357 various functions. Extensions are now handled using the new
1358 OCSP extension code. New simple OCSP HTTP function which
1359 can be used to send requests and parse the response.
1362 +) Fix the PKCS#7 (S/MIME) code to work with new ASN1. Two new
1363 ASN1_ITEM structures help with sign and verify. PKCS7_ATTR_SIGN
1364 uses the special reorder version of SET OF to sort the attributes
1365 and reorder them to match the encoded order. This resolves a long
1366 standing problem: a verify on a PKCS7 structure just after signing
1367 it used to fail because the attribute order did not match the
1368 encoded order. PKCS7_ATTR_VERIFY does not reorder the attributes:
1369 it uses the received order. This is necessary to tolerate some broken
1370 software that does not order SET OF. This is handled by encoding
1371 as a SEQUENCE OF but using implicit tagging (with UNIVERSAL class)
1372 to produce the required SET OF.
1375 +) Have mk1mf.pl generate the macros OPENSSL_BUILD_SHLIBCRYPTO and
1376 OPENSSL_BUILD_SHLIBSSL and use them appropriately in the header
1377 files to get correct declarations of the ASN.1 item variables.
1380 +) Rewrite of PKCS#12 code to use new ASN1 functionality. Replace many
1381 PKCS#12 macros with real functions. Fix two unrelated ASN1 bugs:
1382 asn1_check_tlen() would sometimes attempt to use 'ctx' when it was
1383 NULL and ASN1_TYPE was not dereferenced properly in asn1_ex_c2i().
1384 New ASN1 macro: DECLARE_ASN1_ITEM() which just declares the relevant
1385 ASN1_ITEM and no wrapper functions.
1388 +) New functions or ASN1_item_d2i_fp() and ASN1_item_d2i_bio(). These
1389 replace the old function pointer based I/O routines. Change most of
1390 the *_d2i_bio() and *_d2i_fp() functions to use these.
1393 +) Enhance mkdef.pl to be more accepting about spacing in C preprocessor
1394 lines, recognice more "algorithms" that can be deselected, and make
1395 it complain about algorithm deselection that isn't recognised.
1398 +) New ASN1 functions to handle dup, sign, verify, digest, pack and
1399 unpack operations in terms of ASN1_ITEM. Modify existing wrappers
1400 to use new functions. Add NO_ASN1_OLD which can be set to remove
1401 some old style ASN1 functions: this can be used to determine if old
1402 code will still work when these eventually go away.
1405 +) New extension functions for OCSP structures, these follow the
1406 same conventions as certificates and CRLs.
1409 +) New function X509V3_add1_i2d(). This automatically encodes and
1410 adds an extension. Its behaviour can be customised with various
1411 flags to append, replace or delete. Various wrappers added for
1412 certifcates and CRLs.
1415 +) Fix to avoid calling the underlying ASN1 print routine when
1416 an extension cannot be parsed. Correct a typo in the
1417 OCSP_SERVICELOC extension. Tidy up print OCSP format.
1420 *) Increase s2->wbuf allocation by one byte in ssl2_new (ssl/s2_lib.c).
1421 Otherwise do_ssl_write (ssl/s2_pkt.c) will write beyond buffer limits
1422 when writing a 32767 byte record.
1423 [Bodo Moeller; problem reported by Eric Day <eday@concentric.net>]
1425 *) In RSA_eay_public_{en,ed}crypt and RSA_eay_mod_exp (rsa_eay.c),
1426 obtain lock CRYPTO_LOCK_RSA before setting rsa->_method_mod_{n,p,q}.
1428 (RSA objects have a reference count access to which is protected
1429 by CRYPTO_LOCK_RSA [see rsa_lib.c, s3_srvr.c, ssl_cert.c, ssl_rsa.c],
1430 so they are meant to be shared between threads.)
1431 [Bodo Moeller, Geoff Thorpe; original patch submitted by
1432 "Reddie, Steven" <Steven.Reddie@ca.com>]
1434 +) Make mkdef.pl parse some of the ASN1 macros and add apropriate
1435 entries for variables.
1438 *) Fix a deadlock in CRYPTO_mem_leaks().
1441 +) Add functionality to apps/openssl.c for detecting locking
1442 problems: As the program is single-threaded, all we have
1443 to do is register a locking callback using an array for
1444 storing which locks are currently held by the program.
1447 +) Use a lock around the call to CRYPTO_get_ex_new_index() in
1448 SSL_get_ex_data_X509_STORE_idx(), which is used in
1449 ssl_verify_cert_chain() and thus can be called at any time
1450 during TLS/SSL handshakes so that thread-safety is essential.
1451 Unfortunately, the ex_data design is not at all suited
1452 for multi-threaded use, so it probably should be abolished.
1455 +) Added Broadcom "ubsec" ENGINE to OpenSSL.
1456 [Broadcom, tweaked and integrated by Geoff Thorpe]
1458 +) Move common extension printing code to new function
1459 X509V3_print_extensions(). Reorganise OCSP print routines and
1460 implement some needed OCSP ASN1 functions. Add OCSP extensions.
1463 +) New function X509_signature_print() to remove duplication in some
1467 +) Add a special meaning when SET OF and SEQUENCE OF flags are both
1468 set (this was treated exactly the same as SET OF previously). This
1469 is used to reorder the STACK representing the structure to match the
1470 encoding. This will be used to get round a problem where a PKCS7
1471 structure which was signed could not be verified because the STACK
1472 order did not reflect the encoded order.
1475 +) Reimplement the OCSP ASN1 module using the new code.
1478 +) Update the X509V3 code to permit the use of an ASN1_ITEM structure
1479 for its ASN1 operations. The old style function pointers still exist
1480 for now but they will eventually go away.
1483 +) Merge in replacement ASN1 code from the ASN1 branch. This almost
1484 completely replaces the old ASN1 functionality with a table driven
1485 encoder and decoder which interprets an ASN1_ITEM structure describing
1486 the ASN1 module. Compatibility with the existing ASN1 API (i2d,d2i) is
1487 largely maintained. Almost all of the old asn1_mac.h macro based ASN1
1488 has also been converted to the new form.
1491 +) Change BN_mod_exp_recp so that negative moduli are tolerated
1492 (the sign is ignored). Similarly, ignore the sign in BN_MONT_CTX_set
1493 so that BN_mod_exp_mont and BN_mod_exp_mont_word work
1494 for negative moduli.
1497 +) Fix BN_uadd and BN_usub: Always return non-negative results instead
1498 of not touching the result's sign bit.
1501 +) BN_div bugfix: If the result is 0, the sign (res->neg) must not be
1505 +) Changed the LHASH code to use prototypes for callbacks, and created
1506 macros to declare and implement thin (optionally static) functions
1507 that provide type-safety and avoid function pointer casting for the
1508 type-specific callbacks.
1511 *) Use better test patterns in bntest.
1514 +) Added Kerberos Cipher Suites to be used with TLS, as written in
1516 [Veers Staats <staatsvr@asc.hpc.mil>,
1517 Jeffrey Altman <jaltman@columbia.edu>, via Richard Levitte]
1519 *) rand_win.c fix for Borland C.
1522 *) BN_rshift bugfix for n == 0.
1525 +) Reformat the FAQ so the different questions and answers can be divided
1526 in sections depending on the subject.
1529 +) Have the zlib compression code load ZLIB.DLL dynamically under
1533 +) New function BN_mod_sqrt for computing square roots modulo a prime
1534 (using the probabilistic Tonelli-Shanks algorithm unless
1535 p == 3 (mod 4) or p == 5 (mod 8), which are cases that can
1536 be handled deterministically).
1537 [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller]
1539 *) Add a 'bctest' script that checks for some known 'bc' bugs
1540 so that 'make test' does not abort just because 'bc' is broken.
1543 *) Store verify_result within SSL_SESSION also for client side to
1544 avoid potential security hole. (Re-used sessions on the client side
1545 always resulted in verify_result==X509_V_OK, not using the original
1546 result of the server certificate verification.)
1549 +) Make BN_mod_inverse faster by explicitly handling small quotients
1550 in the Euclid loop. (Speed gain about 20% for small moduli [256 or
1551 512 bits], about 30% for larger ones [1024 or 2048 bits].)
1554 *) Fix ssl3_pending: If the record in s->s3->rrec is not of type
1555 SSL3_RT_APPLICATION_DATA, return 0.
1556 Similarly, change ssl2_pending to return 0 if SSL_in_init(s) is true.
1560 Both ssl2_peek and ssl3_peek, which were totally broken in earlier
1561 releases, have been re-implemented by renaming the previous
1562 implementations of ssl2_read and ssl3_read to ssl2_read_internal
1563 and ssl3_read_internal, respectively, and adding 'peek' parameters
1564 to them. The new ssl[23]_{read,peek} functions are calls to
1565 ssl[23]_read_internal with the 'peek' flag set appropriately.
1566 A 'peek' parameter has also been added to ssl3_read_bytes, which
1567 does the actual work for ssl3_read_internal.
1570 +) New function BN_kronecker.
1573 +) Fix BN_gcd so that it works on negative inputs; the result is
1574 positive unless both parameters are zero.
1575 Previously something reasonably close to an infinite loop was
1576 possible because numbers could be growing instead of shrinking
1577 in the implementation of Euclid's algorithm.
1580 +) Fix BN_is_word() and BN_is_one() macros to take into account the
1581 sign of the number in question.
1583 Fix BN_is_word(a,w) to work correctly for w == 0.
1585 The old BN_is_word(a,w) macro is now called BN_abs_is_word(a,w)
1586 because its test if the absolute value of 'a' equals 'w'.
1587 Note that BN_abs_is_word does *not* handle w == 0 reliably;
1588 it exists mostly for use in the implementations of BN_is_zero(),
1589 BN_is_one(), and BN_is_word().
1592 *) Initialise "ex_data" member of RSA/DSA/DH structures prior to calling
1593 the method-specific "init()" handler. Also clean up ex_data after
1594 calling the method-specific "finish()" handler. Previously, this was
1595 happening the other way round.
1598 +) New function BN_swap.
1601 +) Use BN_nnmod instead of BN_mod in crypto/bn/bn_exp.c so that
1602 the exponentiation functions are more likely to produce reasonable
1603 results on negative inputs.
1606 +) Change BN_mod_mul so that the result is always non-negative.
1607 Previously, it could be negative if one of the factors was negative;
1608 I don't think anyone really wanted that behaviour.
1611 +) Move BN_mod_... functions into new file crypto/bn/bn_mod.c
1612 (except for exponentiation, which stays in crypto/bn/bn_exp.c,
1613 and BN_mod_mul_reciprocal, which stays in crypto/bn/bn_recp.c)
1614 and add new functions:
1623 BN_mod_lshift1_quick
1627 These functions always generate non-negative results.
1629 BN_nnmod otherwise is like BN_mod (if BN_mod computes a remainder r
1630 such that |m| < r < 0, BN_nnmod will output rem + |m| instead).
1632 BN_mod_XXX_quick(r, a, [b,] m) generates the same result as
1633 BN_mod_XXX(r, a, [b,] m, ctx), but requires that a [and b]
1634 be reduced modulo m.
1635 [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller]
1637 +) Remove a few calls to bn_wexpand() in BN_sqr() (the one in there
1638 was actually never needed) and in BN_mul(). The removal in BN_mul()
1639 required a small change in bn_mul_part_recursive() and the addition
1640 of the functions bn_cmp_part_words(), bn_sub_part_words() and
1641 bn_add_part_words(), which do the same thing as bn_cmp_words(),
1642 bn_sub_words() and bn_add_words() except they take arrays with
1646 +) In 'openssl passwd', verify passwords read from the terminal
1647 unless the '-salt' option is used (which usually means that
1648 verification would just waste user's time since the resulting
1649 hash is going to be compared with some given password hash)
1650 or the new '-noverify' option is used.
1652 This is an incompatible change, but it does not affect
1653 non-interactive use of 'openssl passwd' (passwords on the command
1654 line, '-stdin' option, '-in ...' option) and thus should not
1658 +) Remove all references to RSAref, since there's no more need for it.
1661 *) Increase BN_CTX_NUM (the number of BIGNUMs in a BN_CTX) to 16.
1662 The previous value, 12, was not always sufficient for BN_mod_exp().
1665 +) Make DSO load along a path given through an environment variable
1666 (SHLIB_PATH) with shl_load().
1669 +) Constify the ENGINE code as a result of BIGNUM constification.
1670 Also constify the RSA code and most things related to it. In a
1671 few places, most notable in the depth of the ASN.1 code, ugly
1672 casts back to non-const were required (to be solved at a later
1676 +) Make it so the openssl application has all engines loaded by default.
1679 +) Constify the BIGNUM routines a little more.
1682 *) Make sure that shared libraries get the internal name engine with
1683 the full version number and not just 0. This should mark the
1684 shared libraries as not backward compatible. Of course, this should
1685 be changed again when we can guarantee backward binary compatibility.
1688 +) Add the following functions:
1690 ENGINE_load_cswift()
1692 ENGINE_load_atalla()
1694 ENGINE_load_builtin_engines()
1696 That way, an application can itself choose if external engines that
1697 are built-in in OpenSSL shall ever be used or not. The benefit is
1698 that applications won't have to be linked with libdl or other dso
1699 libraries unless it's really needed.
1701 Changed 'openssl engine' to load all engines on demand.
1702 Changed the engine header files to avoid the duplication of some
1703 declarations (they differed!).
1706 +) 'openssl engine' can now list capabilities.
1709 +) Better error reporting in 'openssl engine'.
1712 +) Never call load_dh_param(NULL) in s_server.
1715 +) Add engine application. It can currently list engines by name and
1716 identity, and test if they are actually available.
1719 +) Add support for shared libraries under Irix.
1720 [Albert Chin-A-Young <china@thewrittenword.com>]
1722 +) Improve RPM specification file by forcing symbolic linking and making
1723 sure the installed documentation is also owned by root.root.
1724 [Damien Miller <djm@mindrot.org>]
1726 +) Add configuration option to build on Linux on both big-endian and
1728 [Ralf Baechle <ralf@uni-koblenz.de>]
1730 +) Give the OpenSSL applications more possibilities to make use of
1731 keys (public as well as private) handled by engines.
1734 +) Add OCSP code that comes from CertCo.
1737 +) Add VMS support for the Rijndael code.
1740 +) Added untested support for Nuron crypto accelerator.
1743 +) Add support for external cryptographic devices. This code was
1744 previously distributed separately as the "engine" branch.
1745 [Geoff Thorpe, Richard Levitte]
1747 +) Rework the filename-translation in the DSO code. It is now possible to
1748 have far greater control over how a "name" is turned into a filename
1749 depending on the operating environment and any oddities about the
1750 different shared library filenames on each system.
1753 +) Support threads on FreeBSD-elf in Configure.
1756 +) Add the possibility to create shared libraries on HP-UX
1759 +) Fix for SHA1 assembly problem with MASM: it produces
1760 warnings about corrupt line number information when assembling
1761 with debugging information. This is caused by the overlapping
1763 [Bernd Matthes <mainbug@celocom.de>, Steve Henson]
1766 NCONF_get_number() has no error checking at all. As a replacement,
1767 NCONF_get_number_e() is defined (_e for "error checking") and is
1768 promoted strongly. The old NCONF_get_number is kept around for
1769 binary backward compatibility.
1770 Make it possible for methods to load from something other than a BIO,
1771 by providing a function pointer that is given a name instead of a BIO.
1772 For example, this could be used to load configuration data from an
1776 *) Fix typo in get_cert_by_subject() in by_dir.c
1777 [Jean-Marc Desperrier <jean-marc.desperrier@certplus.com>]
1779 *) Rework the system to generate shared libraries:
1781 - Make note of the expected extension for the shared libraries and
1782 if there is a need for symbolic links from for example libcrypto.so.0
1783 to libcrypto.so.0.9.7. There is extended info in Configure for
1786 - Make as few rebuilds of the shared libraries as possible.
1788 - Still avoid linking the OpenSSL programs with the shared libraries.
1790 - When installing, install the shared libraries separately from the
1794 +) Fix for non blocking accept BIOs. Added new I/O special reason
1795 BIO_RR_ACCEPT to cover this case. Previously use of accept BIOs
1796 with non blocking I/O was not possible because no retry code was
1797 implemented. Also added new SSL code SSL_WANT_ACCEPT to cover
1801 +) Added the beginnings of Rijndael support.
1804 +) Fix for bug in DirectoryString mask setting. Add support for
1805 X509_NAME_print_ex() in 'req' and X509_print_ex() function
1806 to allow certificate printing to more controllable, additional
1807 'certopt' option to 'x509' to allow new printing options to be
1811 +) Clean old EAY MD5 hack from e_os.h.
1814 *) Fix SSL_CTX_set_read_ahead macro to actually use its argument.
1816 Copy SSL_CTX's read_ahead flag to SSL object directly in SSL_new
1817 and not in SSL_clear because the latter is also used by the
1818 accept/connect functions; previously, the settings made by
1819 SSL_set_read_ahead would be lost during the handshake.
1820 [Bodo Moeller; problems reported by Anders Gertz <gertz@epact.se>]
1822 *) Correct util/mkdef.pl to be selective about disabled algorithms.
1823 Previously, it would create entries for disableed algorithms no
1827 *) Added several new manual pages for SSL_* function.
1830 Changes between 0.9.5a and 0.9.6 [24 Sep 2000]
1832 *) In ssl23_get_client_hello, generate an error message when faced
1833 with an initial SSL 3.0/TLS record that is too small to contain the
1834 first two bytes of the ClientHello message, i.e. client_version.
1835 (Note that this is a pathologic case that probably has never happened
1836 in real life.) The previous approach was to use the version number
1837 from the record header as a substitute; but our protocol choice
1838 should not depend on that one because it is not authenticated
1839 by the Finished messages.
1842 *) More robust randomness gathering functions for Windows.
1843 [Jeffrey Altman <jaltman@columbia.edu>]
1845 *) For compatibility reasons if the flag X509_V_FLAG_ISSUER_CHECK is
1846 not set then we don't setup the error code for issuer check errors
1847 to avoid possibly overwriting other errors which the callback does
1848 handle. If an application does set the flag then we assume it knows
1849 what it is doing and can handle the new informational codes
1853 *) Fix for a nasty bug in ASN1_TYPE handling. ASN1_TYPE is used for
1854 a general "ANY" type, as such it should be able to decode anything
1855 including tagged types. However it didn't check the class so it would
1856 wrongly interpret tagged types in the same way as their universal
1857 counterpart and unknown types were just rejected. Changed so that the
1858 tagged and unknown types are handled in the same way as a SEQUENCE:
1859 that is the encoding is stored intact. There is also a new type
1860 "V_ASN1_OTHER" which is used when the class is not universal, in this
1861 case we have no idea what the actual type is so we just lump them all
1865 *) On VMS, stdout may very well lead to a file that is written to
1866 in a record-oriented fashion. That means that every write() will
1867 write a separate record, which will be read separately by the
1868 programs trying to read from it. This can be very confusing.
1870 The solution is to put a BIO filter in the way that will buffer
1871 text until a linefeed is reached, and then write everything a
1872 line at a time, so every record written will be an actual line,
1873 not chunks of lines and not (usually doesn't happen, but I've
1874 seen it once) several lines in one record. BIO_f_linebuffer() is
1877 Currently, it's a VMS-only method, because that's where it has
1878 been tested well enough.
1881 *) Remove 'optimized' squaring variant in BN_mod_mul_montgomery,
1882 it can return incorrect results.
1883 (Note: The buggy variant was not enabled in OpenSSL 0.9.5a,
1884 but it was in 0.9.6-beta[12].)
1887 *) Disable the check for content being present when verifying detached
1888 signatures in pk7_smime.c. Some versions of Netscape (wrongly)
1889 include zero length content when signing messages.
1892 *) New BIO_shutdown_wr macro, which invokes the BIO_C_SHUTDOWN_WR
1893 BIO_ctrl (for BIO pairs).
1896 *) Add DSO method for VMS.
1899 *) Bug fix: Montgomery multiplication could produce results with the
1903 *) Add RPM specification openssl.spec and modify it to build three
1904 packages. The default package contains applications, application
1905 documentation and run-time libraries. The devel package contains
1906 include files, static libraries and function documentation. The
1907 doc package contains the contents of the doc directory. The original
1908 openssl.spec was provided by Damien Miller <djm@mindrot.org>.
1911 *) Add a large number of documentation files for many SSL routines.
1912 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>]
1914 *) Add a configuration entry for Sony News 4.
1915 [NAKAJI Hiroyuki <nakaji@tutrp.tut.ac.jp>]
1917 *) Don't set the two most significant bits to one when generating a
1918 random number < q in the DSA library.
1921 *) New SSL API mode 'SSL_MODE_AUTO_RETRY'. This disables the default
1922 behaviour that SSL_read may result in SSL_ERROR_WANT_READ (even if
1923 the underlying transport is blocking) if a handshake took place.
1924 (The default behaviour is needed by applications such as s_client
1925 and s_server that use select() to determine when to use SSL_read;
1926 but for applications that know in advance when to expect data, it
1927 just makes things more complicated.)
1930 *) Add RAND_egd_bytes(), which gives control over the number of bytes read
1934 *) Add a few more EBCDIC conditionals that make `req' and `x509'
1935 work better on such systems.
1936 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>]
1938 *) Add two demo programs for PKCS12_parse() and PKCS12_create().
1939 Update PKCS12_parse() so it copies the friendlyName and the
1940 keyid to the certificates aux info.
1943 *) Fix bug in PKCS7_verify() which caused an infinite loop
1944 if there was more than one signature.
1945 [Sven Uszpelkat <su@celocom.de>]
1947 *) Major change in util/mkdef.pl to include extra information
1948 about each symbol, as well as presentig variables as well
1949 as functions. This change means that there's n more need
1950 to rebuild the .num files when some algorithms are excluded.
1953 *) Allow the verify time to be set by an application,
1954 rather than always using the current time.
1957 *) Phase 2 verify code reorganisation. The certificate
1958 verify code now looks up an issuer certificate by a
1959 number of criteria: subject name, authority key id
1960 and key usage. It also verifies self signed certificates
1961 by the same criteria. The main comparison function is
1962 X509_check_issued() which performs these checks.
1964 Lot of changes were necessary in order to support this
1965 without completely rewriting the lookup code.
1967 Authority and subject key identifier are now cached.
1969 The LHASH 'certs' is X509_STORE has now been replaced
1970 by a STACK_OF(X509_OBJECT). This is mainly because an
1971 LHASH can't store or retrieve multiple objects with
1972 the same hash value.
1974 As a result various functions (which were all internal
1975 use only) have changed to handle the new X509_STORE
1976 structure. This will break anything that messed round
1977 with X509_STORE internally.
1979 The functions X509_STORE_add_cert() now checks for an
1980 exact match, rather than just subject name.
1982 The X509_STORE API doesn't directly support the retrieval
1983 of multiple certificates matching a given criteria, however
1984 this can be worked round by performing a lookup first
1985 (which will fill the cache with candidate certificates)
1986 and then examining the cache for matches. This is probably
1987 the best we can do without throwing out X509_LOOKUP
1988 entirely (maybe later...).
1990 The X509_VERIFY_CTX structure has been enhanced considerably.
1992 All certificate lookup operations now go via a get_issuer()
1993 callback. Although this currently uses an X509_STORE it
1994 can be replaced by custom lookups. This is a simple way
1995 to bypass the X509_STORE hackery necessary to make this
1996 work and makes it possible to use more efficient techniques
1997 in future. A very simple version which uses a simple
1998 STACK for its trusted certificate store is also provided
1999 using X509_STORE_CTX_trusted_stack().
2001 The verify_cb() and verify() callbacks now have equivalents
2002 in the X509_STORE_CTX structure.
2004 X509_STORE_CTX also has a 'flags' field which can be used
2005 to customise the verify behaviour.
2008 *) Add new PKCS#7 signing option PKCS7_NOSMIMECAP which
2009 excludes S/MIME capabilities.
2012 *) When a certificate request is read in keep a copy of the
2013 original encoding of the signed data and use it when outputing
2014 again. Signatures then use the original encoding rather than
2015 a decoded, encoded version which may cause problems if the
2016 request is improperly encoded.
2019 *) For consistency with other BIO_puts implementations, call
2020 buffer_write(b, ...) directly in buffer_puts instead of calling
2023 In BIO_puts, increment b->num_write as in BIO_write.
2024 [Peter.Sylvester@EdelWeb.fr]
2026 *) Fix BN_mul_word for the case where the word is 0. (We have to use
2027 BN_zero, we may not return a BIGNUM with an array consisting of
2031 *) Avoid calling abort() from within the library when problems are
2032 detected, except if preprocessor symbols have been defined
2033 (such as REF_CHECK, BN_DEBUG etc.).
2036 *) New openssl application 'rsautl'. This utility can be
2037 used for low level RSA operations. DER public key
2038 BIO/fp routines also added.
2041 *) New Configure entry and patches for compiling on QNX 4.
2042 [Andreas Schneider <andreas@ds3.etech.fh-hamburg.de>]
2044 *) A demo state-machine implementation was sponsored by
2045 Nuron (http://www.nuron.com/) and is now available in
2046 demos/state_machine.
2049 *) New options added to the 'dgst' utility for signature
2050 generation and verification.
2053 *) Unrecognized PKCS#7 content types are now handled via a
2054 catch all ASN1_TYPE structure. This allows unsupported
2055 types to be stored as a "blob" and an application can
2056 encode and decode it manually.
2059 *) Fix various signed/unsigned issues to make a_strex.c
2061 [Oscar Jacobsson <oscar.jacobsson@celocom.com>]
2063 *) ASN1 fixes. i2d_ASN1_OBJECT was not returning the correct
2064 length if passed a buffer. ASN1_INTEGER_to_BN failed
2065 if passed a NULL BN and its argument was negative.
2066 [Steve Henson, pointed out by Sven Heiberg <sven@tartu.cyber.ee>]
2068 *) Modification to PKCS#7 encoding routines to output definite
2069 length encoding. Since currently the whole structures are in
2070 memory there's not real point in using indefinite length
2071 constructed encoding. However if OpenSSL is compiled with
2072 the flag PKCS7_INDEFINITE_ENCODING the old form is used.
2075 *) Added BIO_vprintf() and BIO_vsnprintf().
2078 *) Added more prefixes to parse for in the the strings written
2079 through a logging bio, to cover all the levels that are available
2080 through syslog. The prefixes are now:
2082 PANIC, EMERG, EMR => LOG_EMERG
2083 ALERT, ALR => LOG_ALERT
2084 CRIT, CRI => LOG_CRIT
2085 ERROR, ERR => LOG_ERR
2086 WARNING, WARN, WAR => LOG_WARNING
2087 NOTICE, NOTE, NOT => LOG_NOTICE
2088 INFO, INF => LOG_INFO
2089 DEBUG, DBG => LOG_DEBUG
2091 and as before, if none of those prefixes are present at the
2092 beginning of the string, LOG_ERR is chosen.
2094 On Win32, the LOG_* levels are mapped according to this:
2096 LOG_EMERG, LOG_ALERT, LOG_CRIT, LOG_ERR => EVENTLOG_ERROR_TYPE
2097 LOG_WARNING => EVENTLOG_WARNING_TYPE
2098 LOG_NOTICE, LOG_INFO, LOG_DEBUG => EVENTLOG_INFORMATION_TYPE
2102 *) Made it possible to reconfigure with just the configuration
2103 argument "reconf" or "reconfigure". The command line arguments
2104 are stored in Makefile.ssl in the variable CONFIGURE_ARGS,
2105 and are retrieved from there when reconfiguring.
2109 [Assar Westerlund <assar@sics.se>, Richard Levitte]
2111 *) Add the arguments -CAfile and -CApath to the pkcs12 utility.
2114 *) The obj_dat.pl script was messing up the sorting of object
2115 names. The reason was that it compared the quoted version
2116 of strings as a result "OCSP" > "OCSP Signing" because
2117 " > SPACE. Changed script to store unquoted versions of
2118 names and add quotes on output. It was also omitting some
2119 names from the lookup table if they were given a default
2120 value (that is if SN is missing it is given the same
2121 value as LN and vice versa), these are now added on the
2122 grounds that if an object has a name we should be able to
2123 look it up. Finally added warning output when duplicate
2124 short or long names are found.
2127 *) Changes needed for Tandem NSK.
2128 [Scott Uroff <scott@xypro.com>]
2130 *) Fix SSL 2.0 rollback checking: Due to an off-by-one error in
2131 RSA_padding_check_SSLv23(), special padding was never detected
2132 and thus the SSL 3.0/TLS 1.0 countermeasure against protocol
2133 version rollback attacks was not effective.
2135 In s23_clnt.c, don't use special rollback-attack detection padding
2136 (RSA_SSLV23_PADDING) if SSL 2.0 is the only protocol enabled in the
2137 client; similarly, in s23_srvr.c, don't do the rollback check if
2138 SSL 2.0 is the only protocol enabled in the server.
2141 *) Make it possible to get hexdumps of unprintable data with 'openssl
2142 asn1parse'. By implication, the functions ASN1_parse_dump() and
2143 BIO_dump_indent() are added.
2146 *) New functions ASN1_STRING_print_ex() and X509_NAME_print_ex()
2147 these print out strings and name structures based on various
2148 flags including RFC2253 support and proper handling of
2149 multibyte characters. Added options to the 'x509' utility
2150 to allow the various flags to be set.
2153 *) Various fixes to use ASN1_TIME instead of ASN1_UTCTIME.
2154 Also change the functions X509_cmp_current_time() and
2155 X509_gmtime_adj() work with an ASN1_TIME structure,
2156 this will enable certificates using GeneralizedTime in validity
2157 dates to be checked.
2160 *) Make the NEG_PUBKEY_BUG code (which tolerates invalid
2161 negative public key encodings) on by default,
2162 NO_NEG_PUBKEY_BUG can be set to disable it.
2165 *) New function c2i_ASN1_OBJECT() which acts on ASN1_OBJECT
2166 content octets. An i2c_ASN1_OBJECT is unnecessary because
2167 the encoding can be trivially obtained from the structure.
2170 *) crypto/err.c locking bugfix: Use write locks (CRYPTO_w_[un]lock),
2171 not read locks (CRYPTO_r_[un]lock).
2174 *) A first attempt at creating official support for shared
2175 libraries through configuration. I've kept it so the
2176 default is static libraries only, and the OpenSSL programs
2177 are always statically linked for now, but there are
2178 preparations for dynamic linking in place.
2179 This has been tested on Linux and Tru64.
2182 *) Randomness polling function for Win9x, as described in:
2183 Peter Gutmann, Software Generation of Practically Strong
2187 *) Fix so PRNG is seeded in req if using an already existing
2191 *) New options to smime application. -inform and -outform
2192 allow alternative formats for the S/MIME message including
2193 PEM and DER. The -content option allows the content to be
2194 specified separately. This should allow things like Netscape
2195 form signing output easier to verify.
2198 *) Fix the ASN1 encoding of tags using the 'long form'.
2201 *) New ASN1 functions, i2c_* and c2i_* for INTEGER and BIT
2202 STRING types. These convert content octets to and from the
2203 underlying type. The actual tag and length octets are
2204 already assumed to have been read in and checked. These
2205 are needed because all other string types have virtually
2206 identical handling apart from the tag. By having versions
2207 of the ASN1 functions that just operate on content octets
2208 IMPLICIT tagging can be handled properly. It also allows
2209 the ASN1_ENUMERATED code to be cut down because ASN1_ENUMERATED
2210 and ASN1_INTEGER are identical apart from the tag.
2213 *) Change the handling of OID objects as follows:
2215 - New object identifiers are inserted in objects.txt, following
2216 the syntax given in objects.README.
2217 - objects.pl is used to process obj_mac.num and create a new
2219 - obj_dat.pl is used to create a new obj_dat.h, using the data in
2222 This is currently kind of a hack, and the perl code in objects.pl
2223 isn't very elegant, but it works as I intended. The simplest way
2224 to check that it worked correctly is to look in obj_dat.h and
2225 check the array nid_objs and make sure the objects haven't moved
2226 around (this is important!). Additions are OK, as well as
2227 consistent name changes.
2230 *) Add BSD-style MD5-based passwords to 'openssl passwd' (option '-1').
2233 *) Addition of the command line parameter '-rand file' to 'openssl req'.
2234 The given file adds to whatever has already been seeded into the
2235 random pool through the RANDFILE configuration file option or
2236 environment variable, or the default random state file.
2239 *) mkstack.pl now sorts each macro group into lexical order.
2240 Previously the output order depended on the order the files
2241 appeared in the directory, resulting in needless rewriting
2245 *) Patches to make OpenSSL compile under Win32 again. Mostly
2246 work arounds for the VC++ problem that it treats func() as
2247 func(void). Also stripped out the parts of mkdef.pl that
2248 added extra typesafe functions: these no longer exist.
2251 *) Reorganisation of the stack code. The macros are now all
2252 collected in safestack.h . Each macro is defined in terms of
2253 a "stack macro" of the form SKM_<name>(type, a, b). The
2254 DEBUG_SAFESTACK is now handled in terms of function casts,
2255 this has the advantage of retaining type safety without the
2256 use of additional functions. If DEBUG_SAFESTACK is not defined
2257 then the non typesafe macros are used instead. Also modified the
2258 mkstack.pl script to handle the new form. Needs testing to see
2259 if which (if any) compilers it chokes and maybe make DEBUG_SAFESTACK
2260 the default if no major problems. Similar behaviour for ASN1_SET_OF
2261 and PKCS12_STACK_OF.
2264 *) When some versions of IIS use the 'NET' form of private key the
2265 key derivation algorithm is different. Normally MD5(password) is
2266 used as a 128 bit RC4 key. In the modified case
2267 MD5(MD5(password) + "SGCKEYSALT") is used insted. Added some
2268 new functions i2d_RSA_NET(), d2i_RSA_NET() etc which are the same
2269 as the old Netscape_RSA functions except they have an additional
2270 'sgckey' parameter which uses the modified algorithm. Also added
2271 an -sgckey command line option to the rsa utility. Thanks to
2272 Adrian Peck <bertie@ncipher.com> for posting details of the modified
2273 algorithm to openssl-dev.
2276 *) The evp_local.h macros were using 'c.##kname' which resulted in
2277 invalid expansion on some systems (SCO 5.0.5 for example).
2278 Corrected to 'c.kname'.
2279 [Phillip Porch <root@theporch.com>]
2281 *) New X509_get1_email() and X509_REQ_get1_email() functions that return
2282 a STACK of email addresses from a certificate or request, these look
2283 in the subject name and the subject alternative name extensions and
2284 omit any duplicate addresses.
2287 *) Re-implement BN_mod_exp2_mont using independent (and larger) windows.
2288 This makes DSA verification about 2 % faster.
2291 *) Increase maximum window size in BN_mod_exp_... to 6 bits instead of 5
2292 (meaning that now 2^5 values will be precomputed, which is only 4 KB
2293 plus overhead for 1024 bit moduli).
2294 This makes exponentiations about 0.5 % faster for 1024 bit
2295 exponents (as measured by "openssl speed rsa2048").
2298 *) Rename memory handling macros to avoid conflicts with other
2300 Malloc => OPENSSL_malloc
2301 Malloc_locked => OPENSSL_malloc_locked
2302 Realloc => OPENSSL_realloc
2303 Free => OPENSSL_free
2306 *) New function BN_mod_exp_mont_word for small bases (roughly 15%
2307 faster than BN_mod_exp_mont, i.e. 7% for a full DH exchange).
2310 *) CygWin32 support.
2311 [John Jarvie <jjarvie@newsguy.com>]
2313 *) The type-safe stack code has been rejigged. It is now only compiled
2314 in when OpenSSL is configured with the DEBUG_SAFESTACK option and
2315 by default all type-specific stack functions are "#define"d back to
2316 standard stack functions. This results in more streamlined output
2317 but retains the type-safety checking possibilities of the original
2321 *) The STACK code has been cleaned up, and certain type declarations
2322 that didn't make a lot of sense have been brought in line. This has
2323 also involved a cleanup of sorts in safestack.h to more correctly
2324 map type-safe stack functions onto their plain stack counterparts.
2325 This work has also resulted in a variety of "const"ifications of
2326 lots of the code, especially "_cmp" operations which should normally
2327 be prototyped with "const" parameters anyway.
2330 *) When generating bytes for the first time in md_rand.c, 'stir the pool'
2331 by seeding with STATE_SIZE dummy bytes (with zero entropy count).
2332 (The PRNG state consists of two parts, the large pool 'state' and 'md',
2333 where all of 'md' is used each time the PRNG is used, but 'state'
2334 is used only indexed by a cyclic counter. As entropy may not be
2335 well distributed from the beginning, 'md' is important as a
2336 chaining variable. However, the output function chains only half
2337 of 'md', i.e. 80 bits. ssleay_rand_add, on the other hand, chains
2338 all of 'md', and seeding with STATE_SIZE dummy bytes will result
2339 in all of 'state' being rewritten, with the new values depending
2340 on virtually all of 'md'. This overcomes the 80 bit limitation.)
2343 *) In ssl/s2_clnt.c and ssl/s3_clnt.c, call ERR_clear_error() when
2344 the handshake is continued after ssl_verify_cert_chain();
2345 otherwise, if SSL_VERIFY_NONE is set, remaining error codes
2346 can lead to 'unexplainable' connection aborts later.
2347 [Bodo Moeller; problem tracked down by Lutz Jaenicke]
2349 *) Major EVP API cipher revision.
2350 Add hooks for extra EVP features. This allows various cipher
2351 parameters to be set in the EVP interface. Support added for variable
2352 key length ciphers via the EVP_CIPHER_CTX_set_key_length() function and
2353 setting of RC2 and RC5 parameters.
2355 Modify EVP_OpenInit() and EVP_SealInit() to cope with variable key length
2358 Remove lots of duplicated code from the EVP library. For example *every*
2359 cipher init() function handles the 'iv' in the same way according to the
2360 cipher mode. They also all do nothing if the 'key' parameter is NULL and
2361 for CFB and OFB modes they zero ctx->num.
2363 New functionality allows removal of S/MIME code RC2 hack.
2365 Most of the routines have the same form and so can be declared in terms
2368 By shifting this to the top level EVP_CipherInit() it can be removed from
2369 all individual ciphers. If the cipher wants to handle IVs or keys
2370 differently it can set the EVP_CIPH_CUSTOM_IV or EVP_CIPH_ALWAYS_CALL_INIT
2373 Change lots of functions like EVP_EncryptUpdate() to now return a
2374 value: although software versions of the algorithms cannot fail
2375 any installed hardware versions can.
2378 *) Implement SSL_OP_TLS_ROLLBACK_BUG: In ssl3_get_client_key_exchange, if
2379 this option is set, tolerate broken clients that send the negotiated
2380 protocol version number instead of the requested protocol version
2384 *) Call dh_tmp_cb (set by ..._TMP_DH_CB) with correct 'is_export' flag;
2385 i.e. non-zero for export ciphersuites, zero otherwise.
2386 Previous versions had this flag inverted, inconsistent with
2387 rsa_tmp_cb (..._TMP_RSA_CB).
2388 [Bodo Moeller; problem reported by Amit Chopra]
2390 *) Add missing DSA library text string. Work around for some IIS
2391 key files with invalid SEQUENCE encoding.
2394 *) Add a document (doc/standards.txt) that list all kinds of standards
2395 and so on that are implemented in OpenSSL.
2398 *) Enhance c_rehash script. Old version would mishandle certificates
2399 with the same subject name hash and wouldn't handle CRLs at all.
2400 Added -fingerprint option to crl utility, to support new c_rehash
2404 *) Eliminate non-ANSI declarations in crypto.h and stack.h.
2407 *) Fix for SSL server purpose checking. Server checking was
2408 rejecting certificates which had extended key usage present
2409 but no ssl client purpose.
2410 [Steve Henson, reported by Rene Grosser <grosser@hisolutions.com>]
2412 *) Make PKCS#12 code work with no password. The PKCS#12 spec
2413 is a little unclear about how a blank password is handled.
2414 Since the password in encoded as a BMPString with terminating
2415 double NULL a zero length password would end up as just the
2416 double NULL. However no password at all is different and is
2417 handled differently in the PKCS#12 key generation code. NS
2418 treats a blank password as zero length. MSIE treats it as no
2419 password on export: but it will try both on import. We now do
2420 the same: PKCS12_parse() tries zero length and no password if
2421 the password is set to "" or NULL (NULL is now a valid password:
2422 it wasn't before) as does the pkcs12 application.
2425 *) Bugfixes in apps/x509.c: Avoid a memory leak; and don't use
2426 perror when PEM_read_bio_X509_REQ fails, the error message must
2427 be obtained from the error queue.
2430 *) Avoid 'thread_hash' memory leak in crypto/err/err.c by freeing
2431 it in ERR_remove_state if appropriate, and change ERR_get_state
2432 accordingly to avoid race conditions (this is necessary because
2433 thread_hash is no longer constant once set).
2436 *) Bugfix for linux-elf makefile.one.
2439 *) RSA_get_default_method() will now cause a default
2440 RSA_METHOD to be chosen if one doesn't exist already.
2441 Previously this was only set during a call to RSA_new()
2442 or RSA_new_method(NULL) meaning it was possible for
2443 RSA_get_default_method() to return NULL.
2446 *) Added native name translation to the existing DSO code
2447 that will convert (if the flag to do so is set) filenames
2448 that are sufficiently small and have no path information
2449 into a canonical native form. Eg. "blah" converted to
2450 "libblah.so" or "blah.dll" etc.
2453 *) New function ERR_error_string_n(e, buf, len) which is like
2454 ERR_error_string(e, buf), but writes at most 'len' bytes
2455 including the 0 terminator. For ERR_error_string_n, 'buf'
2457 [Damien Miller <djm@mindrot.org>, Bodo Moeller]
2459 *) CONF library reworked to become more general. A new CONF
2460 configuration file reader "class" is implemented as well as a
2461 new functions (NCONF_*, for "New CONF") to handle it. The now
2462 old CONF_* functions are still there, but are reimplemented to
2463 work in terms of the new functions. Also, a set of functions
2464 to handle the internal storage of the configuration data is
2465 provided to make it easier to write new configuration file
2466 reader "classes" (I can definitely see something reading a
2467 configuration file in XML format, for example), called _CONF_*,
2468 or "the configuration storage API"...
2470 The new configuration file reading functions are:
2472 NCONF_new, NCONF_free, NCONF_load, NCONF_load_fp, NCONF_load_bio,
2473 NCONF_get_section, NCONF_get_string, NCONF_get_numbre
2475 NCONF_default, NCONF_WIN32
2477 NCONF_dump_fp, NCONF_dump_bio
2479 NCONF_default and NCONF_WIN32 are method (or "class") choosers,
2480 NCONF_new creates a new CONF object. This works in the same way
2481 as other interfaces in OpenSSL, like the BIO interface.
2482 NCONF_dump_* dump the internal storage of the configuration file,
2483 which is useful for debugging. All other functions take the same
2484 arguments as the old CONF_* functions wth the exception of the
2485 first that must be a `CONF *' instead of a `LHASH *'.
2487 To make it easer to use the new classes with the old CONF_* functions,
2488 the function CONF_set_default_method is provided.
2491 *) Add '-tls1' option to 'openssl ciphers', which was already
2492 mentioned in the documentation but had not been implemented.
2493 (This option is not yet really useful because even the additional
2494 experimental TLS 1.0 ciphers are currently treated as SSL 3.0 ciphers.)
2497 *) Initial DSO code added into libcrypto for letting OpenSSL (and
2498 OpenSSL-based applications) load shared libraries and bind to
2499 them in a portable way.
2500 [Geoff Thorpe, with contributions from Richard Levitte]
2502 Changes between 0.9.5 and 0.9.5a [1 Apr 2000]
2504 *) Make sure _lrotl and _lrotr are only used with MSVC.
2506 *) Use lock CRYPTO_LOCK_RAND correctly in ssleay_rand_status
2507 (the default implementation of RAND_status).
2509 *) Rename openssl x509 option '-crlext', which was added in 0.9.5,
2510 to '-clrext' (= clear extensions), as intended and documented.
2511 [Bodo Moeller; inconsistency pointed out by Michael Attili
2514 *) Fix for HMAC. It wasn't zeroing the rest of the block if the key length
2515 was larger than the MD block size.
2516 [Steve Henson, pointed out by Yost William <YostW@tce.com>]
2518 *) Modernise PKCS12_parse() so it uses STACK_OF(X509) for its ca argument
2519 fix a leak when the ca argument was passed as NULL. Stop X509_PUBKEY_set()
2520 using the passed key: if the passed key was a private key the result
2521 of X509_print(), for example, would be to print out all the private key
2525 *) des_quad_cksum() byte order bug fix.
2526 [Ulf Möller, using the problem description in krb4-0.9.7, where
2527 the solution is attributed to Derrick J Brashear <shadow@DEMENTIA.ORG>]
2529 *) Fix so V_ASN1_APP_CHOOSE works again: however its use is strongly
2531 [Steve Henson, pointed out by Brian Korver <briank@cs.stanford.edu>]
2533 *) For easily testing in shell scripts whether some command
2534 'openssl XXX' exists, the new pseudo-command 'openssl no-XXX'
2535 returns with exit code 0 iff no command of the given name is available.
2536 'no-XXX' is printed in this case, 'XXX' otherwise. In both cases,
2537 the output goes to stdout and nothing is printed to stderr.
2538 Additional arguments are always ignored.
2540 Since for each cipher there is a command of the same name,
2541 the 'no-cipher' compilation switches can be tested this way.
2543 ('openssl no-XXX' is not able to detect pseudo-commands such
2544 as 'quit', 'list-XXX-commands', or 'no-XXX' itself.)
2547 *) Update test suite so that 'make test' succeeds in 'no-rsa' configuration.
2550 *) For SSL_[CTX_]set_tmp_dh, don't create a DH key if SSL_OP_SINGLE_DH_USE
2551 is set; it will be thrown away anyway because each handshake creates
2553 ssl_cert_dup, which is used by SSL_new, now copies DH keys in addition
2554 to parameters -- in previous versions (since OpenSSL 0.9.3) the
2555 'default key' from SSL_CTX_set_tmp_dh would always be lost, meanining
2556 you effectivly got SSL_OP_SINGLE_DH_USE when using this macro.
2559 *) New s_client option -ign_eof: EOF at stdin is ignored, and
2560 'Q' and 'R' lose their special meanings (quit/renegotiate).
2561 This is part of what -quiet does; unlike -quiet, -ign_eof
2562 does not suppress any output.
2565 *) Add compatibility options to the purpose and trust code. The
2566 purpose X509_PURPOSE_ANY is "any purpose" which automatically
2567 accepts a certificate or CA, this was the previous behaviour,
2568 with all the associated security issues.
2570 X509_TRUST_COMPAT is the old trust behaviour: only and
2571 automatically trust self signed roots in certificate store. A
2572 new trust setting X509_TRUST_DEFAULT is used to specify that
2573 a purpose has no associated trust setting and it should instead
2574 use the value in the default purpose.
2577 *) Fix the PKCS#8 DSA private key code so it decodes keys again
2578 and fix a memory leak.
2581 *) In util/mkerr.pl (which implements 'make errors'), preserve
2582 reason strings from the previous version of the .c file, as
2583 the default to have only downcase letters (and digits) in
2584 automatically generated reasons codes is not always appropriate.
2587 *) In ERR_load_ERR_strings(), build an ERR_LIB_SYS error reason table
2588 using strerror. Previously, ERR_reason_error_string() returned
2589 library names as reason strings for SYSerr; but SYSerr is a special
2590 case where small numbers are errno values, not library numbers.
2593 *) Add '-dsaparam' option to 'openssl dhparam' application. This
2594 converts DSA parameters into DH parameters. (When creating parameters,
2595 DSA_generate_parameters is used.)
2598 *) Include 'length' (recommended exponent length) in C code generated
2599 by 'openssl dhparam -C'.
2602 *) The second argument to set_label in perlasm was already being used
2603 so couldn't be used as a "file scope" flag. Moved to third argument
2607 *) In PEM_ASN1_write_bio and some other functions, use RAND_pseudo_bytes
2608 instead of RAND_bytes for encryption IVs and salts.
2611 *) Include RAND_status() into RAND_METHOD instead of implementing
2612 it only for md_rand.c Otherwise replacing the PRNG by calling
2613 RAND_set_rand_method would be impossible.
2616 *) Don't let DSA_generate_key() enter an infinite loop if the random
2617 number generation fails.
2620 *) New 'rand' application for creating pseudo-random output.
2623 *) Added configuration support for Linux/IA64
2624 [Rolf Haberrecker <rolf@suse.de>]
2626 *) Assembler module support for Mingw32.
2629 *) Shared library support for HPUX (in shlib/).
2630 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE> and Anonymous]
2632 *) Shared library support for Solaris gcc.
2633 [Lutz Behnke <behnke@trustcenter.de>]
2635 Changes between 0.9.4 and 0.9.5 [28 Feb 2000]
2637 *) PKCS7_encrypt() was adding text MIME headers twice because they
2638 were added manually and by SMIME_crlf_copy().
2641 *) In bntest.c don't call BN_rand with zero bits argument.
2642 [Steve Henson, pointed out by Andrew W. Gray <agray@iconsinc.com>]
2644 *) BN_mul bugfix: In bn_mul_part_recursion() only the a>a[n] && b>b[n]
2645 case was implemented. This caused BN_div_recp() to fail occasionally.
2648 *) Add an optional second argument to the set_label() in the perl
2649 assembly language builder. If this argument exists and is set
2650 to 1 it signals that the assembler should use a symbol whose
2651 scope is the entire file, not just the current function. This
2652 is needed with MASM which uses the format label:: for this scope.
2653 [Steve Henson, pointed out by Peter Runestig <peter@runestig.com>]
2655 *) Change the ASN1 types so they are typedefs by default. Before
2656 almost all types were #define'd to ASN1_STRING which was causing
2657 STACK_OF() problems: you couldn't declare STACK_OF(ASN1_UTF8STRING)
2661 *) Change names of new functions to the new get1/get0 naming
2662 convention: After 'get1', the caller owns a reference count
2663 and has to call ..._free; 'get0' returns a pointer to some
2664 data structure without incrementing reference counters.
2665 (Some of the existing 'get' functions increment a reference
2666 counter, some don't.)
2667 Similarly, 'set1' and 'add1' functions increase reference
2668 counters or duplicate objects.
2671 *) Allow for the possibility of temp RSA key generation failure:
2672 the code used to assume it always worked and crashed on failure.
2675 *) Fix potential buffer overrun problem in BIO_printf().
2676 [Ulf Möller, using public domain code by Patrick Powell; problem
2677 pointed out by David Sacerdote <das33@cornell.edu>]
2679 *) Support EGD <http://www.lothar.com/tech/crypto/>. New functions
2680 RAND_egd() and RAND_status(). In the command line application,
2681 the EGD socket can be specified like a seed file using RANDFILE
2685 *) Allow the string CERTIFICATE to be tolerated in PKCS#7 structures.
2686 Some CAs (e.g. Verisign) distribute certificates in this form.
2689 *) Remove the SSL_ALLOW_ADH compile option and set the default cipher
2690 list to exclude them. This means that no special compilation option
2691 is needed to use anonymous DH: it just needs to be included in the
2695 *) Change the EVP_MD_CTX_type macro so its meaning consistent with
2696 EVP_MD_type. The old functionality is available in a new macro called
2697 EVP_MD_md(). Change code that uses it and update docs.
2700 *) ..._ctrl functions now have corresponding ..._callback_ctrl functions
2701 where the 'void *' argument is replaced by a function pointer argument.
2702 Previously 'void *' was abused to point to functions, which works on
2703 many platforms, but is not correct. As these functions are usually
2704 called by macros defined in OpenSSL header files, most source code
2705 should work without changes.
2708 *) <openssl/opensslconf.h> (which is created by Configure) now contains
2709 sections with information on -D... compiler switches used for
2710 compiling the library so that applications can see them. To enable
2711 one of these sections, a pre-processor symbol OPENSSL_..._DEFINES
2712 must be defined. E.g.,
2713 #define OPENSSL_ALGORITHM_DEFINES
2714 #include <openssl/opensslconf.h>
2715 defines all pertinent NO_<algo> symbols, such as NO_IDEA, NO_RSA, etc.
2716 [Richard Levitte, Ulf and Bodo Möller]
2718 *) Bugfix: Tolerate fragmentation and interleaving in the SSL 3/TLS
2722 *) Change the 'other' type in certificate aux info to a STACK_OF
2723 X509_ALGOR. Although not an AlgorithmIdentifier as such it has
2724 the required ASN1 format: arbitrary types determined by an OID.
2727 *) Add some PEM_write_X509_REQ_NEW() functions and a command line
2728 argument to 'req'. This is not because the function is newer or
2729 better than others it just uses the work 'NEW' in the certificate
2730 request header lines. Some software needs this.
2733 *) Reorganise password command line arguments: now passwords can be
2734 obtained from various sources. Delete the PEM_cb function and make
2735 it the default behaviour: i.e. if the callback is NULL and the
2736 usrdata argument is not NULL interpret it as a null terminated pass
2737 phrase. If usrdata and the callback are NULL then the pass phrase
2738 is prompted for as usual.
2741 *) Add support for the Compaq Atalla crypto accelerator. If it is installed,
2742 the support is automatically enabled. The resulting binaries will
2743 autodetect the card and use it if present.
2744 [Ben Laurie and Compaq Inc.]
2746 *) Work around for Netscape hang bug. This sends certificate request
2747 and server done in one record. Since this is perfectly legal in the
2748 SSL/TLS protocol it isn't a "bug" option and is on by default. See
2749 the bugs/SSLv3 entry for more info.
2752 *) HP-UX tune-up: new unified configs, HP C compiler bug workaround.
2755 *) Add -rand argument to smime and pkcs12 applications and read/write
2759 *) New 'passwd' tool for crypt(3) and apr1 password hashes.
2762 *) Add command line password options to the remaining applications.
2765 *) Bug fix for BN_div_recp() for numerators with an even number of
2769 *) More tests in bntest.c, and changed test_bn output.
2772 *) ./config recognizes MacOS X now.
2775 *) Bug fix for BN_div() when the first words of num and divsor are
2776 equal (it gave wrong results if (rem=(n1-q*d0)&BN_MASK2) < d0).
2779 *) Add support for various broken PKCS#8 formats, and command line
2780 options to produce them.
2783 *) New functions BN_CTX_start(), BN_CTX_get() and BT_CTX_end() to
2784 get temporary BIGNUMs from a BN_CTX.
2787 *) Correct return values in BN_mod_exp_mont() and BN_mod_exp2_mont()
2791 *) Change the SSLeay_add_all_*() functions to OpenSSL_add_all_*() and
2792 include a #define from the old name to the new. The original intent
2793 was that statically linked binaries could for example just call
2794 SSLeay_add_all_ciphers() to just add ciphers to the table and not
2795 link with digests. This never worked becayse SSLeay_add_all_digests()
2796 and SSLeay_add_all_ciphers() were in the same source file so calling
2797 one would link with the other. They are now in separate source files.
2800 *) Add a new -notext option to 'ca' and a -pubkey option to 'spkac'.
2803 *) Use a less unusual form of the Miller-Rabin primality test (it used
2804 a binary algorithm for exponentiation integrated into the Miller-Rabin
2805 loop, our standard modexp algorithms are faster).
2808 *) Support for the EBCDIC character set completed.
2809 [Martin Kraemer <Martin.Kraemer@Mch.SNI.De>]
2811 *) Source code cleanups: use const where appropriate, eliminate casts,
2812 use void * instead of char * in lhash.
2815 *) Bugfix: ssl3_send_server_key_exchange was not restartable
2816 (the state was not changed to SSL3_ST_SW_KEY_EXCH_B, and because of
2817 this the server could overwrite ephemeral keys that the client
2821 *) Turn DSA_is_prime into a macro that calls BN_is_prime,
2822 using 50 iterations of the Rabin-Miller test.
2824 DSA_generate_parameters now uses BN_is_prime_fasttest (with 50
2825 iterations of the Rabin-Miller test as required by the appendix
2826 to FIPS PUB 186[-1]) instead of DSA_is_prime.
2827 As BN_is_prime_fasttest includes trial division, DSA parameter
2828 generation becomes much faster.
2830 This implies a change for the callback functions in DSA_is_prime
2831 and DSA_generate_parameters: The callback function is called once
2832 for each positive witness in the Rabin-Miller test, not just
2833 occasionally in the inner loop; and the parameters to the
2834 callback function now provide an iteration count for the outer
2835 loop rather than for the current invocation of the inner loop.
2836 DSA_generate_parameters additionally can call the callback
2837 function with an 'iteration count' of -1, meaning that a
2838 candidate has passed the trial division test (when q is generated
2839 from an application-provided seed, trial division is skipped).
2842 *) New function BN_is_prime_fasttest that optionally does trial
2843 division before starting the Rabin-Miller test and has
2844 an additional BN_CTX * argument (whereas BN_is_prime always
2845 has to allocate at least one BN_CTX).
2846 'callback(1, -1, cb_arg)' is called when a number has passed the
2847 trial division stage.
2850 *) Fix for bug in CRL encoding. The validity dates weren't being handled
2854 *) New -pkcs12 option to CA.pl script to write out a PKCS#12 file.
2857 *) New function BN_pseudo_rand().
2860 *) Clean up BN_mod_mul_montgomery(): replace the broken (and unreadable)
2861 bignum version of BN_from_montgomery() with the working code from
2862 SSLeay 0.9.0 (the word based version is faster anyway), and clean up
2866 *) Avoid a race condition in s2_clnt.c (function get_server_hello) that
2867 made it impossible to use the same SSL_SESSION data structure in
2868 SSL2 clients in multiple threads.
2871 *) The return value of RAND_load_file() no longer counts bytes obtained
2872 by stat(). RAND_load_file(..., -1) is new and uses the complete file
2873 to seed the PRNG (previously an explicit byte count was required).
2874 [Ulf Möller, Bodo Möller]
2876 *) Clean up CRYPTO_EX_DATA functions, some of these didn't have prototypes
2877 used (char *) instead of (void *) and had casts all over the place.
2880 *) Make BN_generate_prime() return NULL on error if ret!=NULL.
2883 *) Retain source code compatibility for BN_prime_checks macro:
2884 BN_is_prime(..., BN_prime_checks, ...) now uses
2885 BN_prime_checks_for_size to determine the appropriate number of
2886 Rabin-Miller iterations.
2889 *) Diffie-Hellman uses "safe" primes: DH_check() return code renamed to
2890 DH_CHECK_P_NOT_SAFE_PRIME.
2891 (Check if this is true? OpenPGP calls them "strong".)
2894 *) Merge the functionality of "dh" and "gendh" programs into a new program
2895 "dhparam". The old programs are retained for now but will handle DH keys
2896 (instead of parameters) in future.
2899 *) Make the ciphers, s_server and s_client programs check the return values
2900 when a new cipher list is set.
2903 *) Enhance the SSL/TLS cipher mechanism to correctly handle the TLS 56bit
2904 ciphers. Before when the 56bit ciphers were enabled the sorting was
2907 The syntax for the cipher sorting has been extended to support sorting by
2908 cipher-strength (using the strength_bits hard coded in the tables).
2909 The new command is "@STRENGTH" (see also doc/apps/ciphers.pod).
2911 Fix a bug in the cipher-command parser: when supplying a cipher command
2912 string with an "undefined" symbol (neither command nor alphanumeric
2913 [A-Za-z0-9], ssl_set_cipher_list used to hang in an endless loop. Now
2914 an error is flagged.
2916 Due to the strength-sorting extension, the code of the
2917 ssl_create_cipher_list() function was completely rearranged. I hope that
2918 the readability was also increased :-)
2919 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>]
2921 *) Minor change to 'x509' utility. The -CAcreateserial option now uses 1
2922 for the first serial number and places 2 in the serial number file. This
2923 avoids problems when the root CA is created with serial number zero and
2924 the first user certificate has the same issuer name and serial number
2928 *) Fixes to X509_ATTRIBUTE utilities, change the 'req' program so it uses
2929 the new code. Add documentation for this stuff.
2932 *) Changes to X509_ATTRIBUTE utilities. These have been renamed from
2933 X509_*() to X509at_*() on the grounds that they don't handle X509
2934 structures and behave in an analagous way to the X509v3 functions:
2935 they shouldn't be called directly but wrapper functions should be used
2938 So we also now have some wrapper functions that call the X509at functions
2939 when passed certificate requests. (TO DO: similar things can be done with
2940 PKCS#7 signed and unsigned attributes, PKCS#12 attributes and a few other
2941 things. Some of these need some d2i or i2d and print functionality
2942 because they handle more complex structures.)
2945 *) Add missing #ifndefs that caused missing symbols when building libssl
2946 as a shared library without RSA. Use #ifndef NO_SSL2 instead of
2947 NO_RSA in ssl/s2*.c.
2948 [Kris Kennaway <kris@hub.freebsd.org>, modified by Ulf Möller]
2950 *) Precautions against using the PRNG uninitialized: RAND_bytes() now
2951 has a return value which indicates the quality of the random data
2952 (1 = ok, 0 = not seeded). Also an error is recorded on the thread's
2953 error queue. New function RAND_pseudo_bytes() generates output that is
2954 guaranteed to be unique but not unpredictable. RAND_add is like
2955 RAND_seed, but takes an extra argument for an entropy estimate
2956 (RAND_seed always assumes full entropy).
2959 *) Do more iterations of Rabin-Miller probable prime test (specifically,
2960 3 for 1024-bit primes, 6 for 512-bit primes, 12 for 256-bit primes
2961 instead of only 2 for all lengths; see BN_prime_checks_for_size definition
2962 in crypto/bn/bn_prime.c for the complete table). This guarantees a
2963 false-positive rate of at most 2^-80 for random input.
2966 *) Rewrite ssl3_read_n (ssl/s3_pkt.c) avoiding a couple of bugs.
2969 *) New function X509_CTX_rget_chain() (renamed to X509_CTX_get1_chain
2970 in the 0.9.5 release), this returns the chain
2971 from an X509_CTX structure with a dup of the stack and all
2972 the X509 reference counts upped: so the stack will exist
2973 after X509_CTX_cleanup() has been called. Modify pkcs12.c
2976 Also make SSL_SESSION_print() print out the verify return
2980 *) Add manpage for the pkcs12 command. Also change the default
2981 behaviour so MAC iteration counts are used unless the new
2982 -nomaciter option is used. This improves file security and
2983 only older versions of MSIE (4.0 for example) need it.
2986 *) Honor the no-xxx Configure options when creating .DEF files.
2989 *) Add PKCS#10 attributes to field table: challengePassword,
2990 unstructuredName and unstructuredAddress. These are taken from
2991 draft PKCS#9 v2.0 but are compatible with v1.2 provided no
2992 international characters are used.
2994 More changes to X509_ATTRIBUTE code: allow the setting of types
2995 based on strings. Remove the 'loc' parameter when adding
2996 attributes because these will be a SET OF encoding which is sorted
3000 *) Initial changes to the 'req' utility to allow request generation
3001 automation. This will allow an application to just generate a template
3002 file containing all the field values and have req construct the
3005 Initial support for X509_ATTRIBUTE handling. Stacks of these are
3006 used all over the place including certificate requests and PKCS#7
3007 structures. They are currently handled manually where necessary with
3008 some primitive wrappers for PKCS#7. The new functions behave in a
3009 manner analogous to the X509 extension functions: they allow
3010 attributes to be looked up by NID and added.
3012 Later something similar to the X509V3 code would be desirable to
3013 automatically handle the encoding, decoding and printing of the
3014 more complex types. The string types like challengePassword can
3015 be handled by the string table functions.
3017 Also modified the multi byte string table handling. Now there is
3018 a 'global mask' which masks out certain types. The table itself
3019 can use the flag STABLE_NO_MASK to ignore the mask setting: this
3020 is useful when for example there is only one permissible type
3021 (as in countryName) and using the mask might result in no valid
3025 *) Clean up 'Finished' handling, and add functions SSL_get_finished and
3026 SSL_get_peer_finished to allow applications to obtain the latest
3027 Finished messages sent to the peer or expected from the peer,
3028 respectively. (SSL_get_peer_finished is usually the Finished message
3029 actually received from the peer, otherwise the protocol will be aborted.)
3031 As the Finished message are message digests of the complete handshake
3032 (with a total of 192 bits for TLS 1.0 and more for SSL 3.0), they can
3033 be used for external authentication procedures when the authentication
3034 provided by SSL/TLS is not desired or is not enough.
3037 *) Enhanced support for Alpha Linux is added. Now ./config checks if
3038 the host supports BWX extension and if Compaq C is present on the
3039 $PATH. Just exploiting of the BWX extension results in 20-30%
3040 performance kick for some algorithms, e.g. DES and RC4 to mention
3041 a couple. Compaq C in turn generates ~20% faster code for MD5 and
3045 *) Add support for MS "fast SGC". This is arguably a violation of the
3046 SSL3/TLS protocol. Netscape SGC does two handshakes: the first with
3047 weak crypto and after checking the certificate is SGC a second one
3048 with strong crypto. MS SGC stops the first handshake after receiving
3049 the server certificate message and sends a second client hello. Since
3050 a server will typically do all the time consuming operations before
3051 expecting any further messages from the client (server key exchange
3052 is the most expensive) there is little difference between the two.
3054 To get OpenSSL to support MS SGC we have to permit a second client
3055 hello message after we have sent server done. In addition we have to
3056 reset the MAC if we do get this second client hello.
3059 *) Add a function 'd2i_AutoPrivateKey()' this will automatically decide
3060 if a DER encoded private key is RSA or DSA traditional format. Changed
3061 d2i_PrivateKey_bio() to use it. This is only needed for the "traditional"
3062 format DER encoded private key. Newer code should use PKCS#8 format which
3063 has the key type encoded in the ASN1 structure. Added DER private key
3064 support to pkcs8 application.
3067 *) SSL 3/TLS 1 servers now don't request certificates when an anonymous
3068 ciphersuites has been selected (as required by the SSL 3/TLS 1
3069 specifications). Exception: When SSL_VERIFY_FAIL_IF_NO_PEER_CERT
3070 is set, we interpret this as a request to violate the specification
3071 (the worst that can happen is a handshake failure, and 'correct'
3072 behaviour would result in a handshake failure anyway).
3075 *) In SSL_CTX_add_session, take into account that there might be multiple
3076 SSL_SESSION structures with the same session ID (e.g. when two threads
3077 concurrently obtain them from an external cache).
3078 The internal cache can handle only one SSL_SESSION with a given ID,
3079 so if there's a conflict, we now throw out the old one to achieve
3083 *) Add OIDs for idea and blowfish in CBC mode. This will allow both
3084 to be used in PKCS#5 v2.0 and S/MIME. Also add checking to
3085 some routines that use cipher OIDs: some ciphers do not have OIDs
3086 defined and so they cannot be used for S/MIME and PKCS#5 v2.0 for
3090 *) Simplify the trust setting structure and code. Now we just have
3091 two sequences of OIDs for trusted and rejected settings. These will
3092 typically have values the same as the extended key usage extension
3093 and any application specific purposes.
3095 The trust checking code now has a default behaviour: it will just
3096 check for an object with the same NID as the passed id. Functions can
3097 be provided to override either the default behaviour or the behaviour
3098 for a given id. SSL client, server and email already have functions
3099 in place for compatibility: they check the NID and also return "trusted"
3100 if the certificate is self signed.
3103 *) Add d2i,i2d bio/fp functions for PrivateKey: these convert the
3104 traditional format into an EVP_PKEY structure.
3107 *) Add a password callback function PEM_cb() which either prompts for
3108 a password if usr_data is NULL or otherwise assumes it is a null
3109 terminated password. Allow passwords to be passed on command line
3110 environment or config files in a few more utilities.
3113 *) Add a bunch of DER and PEM functions to handle PKCS#8 format private
3114 keys. Add some short names for PKCS#8 PBE algorithms and allow them
3115 to be specified on the command line for the pkcs8 and pkcs12 utilities.
3116 Update documentation.
3119 *) Support for ASN1 "NULL" type. This could be handled before by using
3120 ASN1_TYPE but there wasn't any function that would try to read a NULL
3121 and produce an error if it couldn't. For compatibility we also have
3122 ASN1_NULL_new() and ASN1_NULL_free() functions but these are faked and
3123 don't allocate anything because they don't need to.
3126 *) Initial support for MacOS is now provided. Examine INSTALL.MacOS
3128 [Andy Polyakov, Roy Woods <roy@centicsystems.ca>]
3130 *) Rebuild of the memory allocation routines used by OpenSSL code and
3131 possibly others as well. The purpose is to make an interface that
3132 provide hooks so anyone can build a separate set of allocation and
3133 deallocation routines to be used by OpenSSL, for example memory
3134 pool implementations, or something else, which was previously hard
3135 since Malloc(), Realloc() and Free() were defined as macros having
3136 the values malloc, realloc and free, respectively (except for Win32
3137 compilations). The same is provided for memory debugging code.
3138 OpenSSL already comes with functionality to find memory leaks, but
3139 this gives people a chance to debug other memory problems.
3141 With these changes, a new set of functions and macros have appeared:
3143 CRYPTO_set_mem_debug_functions() [F]
3144 CRYPTO_get_mem_debug_functions() [F]
3145 CRYPTO_dbg_set_options() [F]
3146 CRYPTO_dbg_get_options() [F]
3147 CRYPTO_malloc_debug_init() [M]
3149 The memory debug functions are NULL by default, unless the library
3150 is compiled with CRYPTO_MDEBUG or friends is defined. If someone
3151 wants to debug memory anyway, CRYPTO_malloc_debug_init() (which
3152 gives the standard debugging functions that come with OpenSSL) or
3153 CRYPTO_set_mem_debug_functions() (tells OpenSSL to use functions
3154 provided by the library user) must be used. When the standard
3155 debugging functions are used, CRYPTO_dbg_set_options can be used to
3156 request additional information:
3157 CRYPTO_dbg_set_options(V_CYRPTO_MDEBUG_xxx) corresponds to setting
3158 the CRYPTO_MDEBUG_xxx macro when compiling the library.
3160 Also, things like CRYPTO_set_mem_functions will always give the
3161 expected result (the new set of functions is used for allocation
3162 and deallocation) at all times, regardless of platform and compiler
3165 To finish it up, some functions that were never use in any other
3166 way than through macros have a new API and new semantic:
3169 CRYPTO_dbg_realloc()
3172 All macros of value have retained their old syntax.
3173 [Richard Levitte and Bodo Moeller]
3175 *) Some S/MIME fixes. The OID for SMIMECapabilities was wrong, the
3176 ordering of SMIMECapabilities wasn't in "strength order" and there
3177 was a missing NULL in the AlgorithmIdentifier for the SHA1 signature
3181 *) Some ASN1 types with illegal zero length encoding (INTEGER,
3182 ENUMERATED and OBJECT IDENTIFIER) choked the ASN1 routines.
3183 [Frans Heymans <fheymans@isaserver.be>, modified by Steve Henson]
3185 *) Merge in my S/MIME library for OpenSSL. This provides a simple
3186 S/MIME API on top of the PKCS#7 code, a MIME parser (with enough
3187 functionality to handle multipart/signed properly) and a utility
3188 called 'smime' to call all this stuff. This is based on code I
3189 originally wrote for Celo who have kindly allowed it to be
3190 included in OpenSSL.
3193 *) Add variants des_set_key_checked and des_set_key_unchecked of
3194 des_set_key (aka des_key_sched). Global variable des_check_key
3195 decides which of these is called by des_set_key; this way
3196 des_check_key behaves as it always did, but applications and
3197 the library itself, which was buggy for des_check_key == 1,
3198 have a cleaner way to pick the version they need.
3201 *) New function PKCS12_newpass() which changes the password of a
3205 *) Modify X509_TRUST and X509_PURPOSE so it also uses a static and
3206 dynamic mix. In both cases the ids can be used as an index into the
3207 table. Also modified the X509_TRUST_add() and X509_PURPOSE_add()
3208 functions so they accept a list of the field values and the
3209 application doesn't need to directly manipulate the X509_TRUST
3213 *) Modify the ASN1_STRING_TABLE stuff so it also uses bsearch and doesn't
3217 *) Modify the way the V3 extension code looks up extensions. This now
3218 works in a similar way to the object code: we have some "standard"
3219 extensions in a static table which is searched with OBJ_bsearch()
3220 and the application can add dynamic ones if needed. The file
3221 crypto/x509v3/ext_dat.h now has the info: this file needs to be
3222 updated whenever a new extension is added to the core code and kept
3223 in ext_nid order. There is a simple program 'tabtest.c' which checks
3224 this. New extensions are not added too often so this file can readily
3225 be maintained manually.
3227 There are two big advantages in doing things this way. The extensions
3228 can be looked up immediately and no longer need to be "added" using
3229 X509V3_add_standard_extensions(): this function now does nothing.
3230 [Side note: I get *lots* of email saying the extension code doesn't
3231 work because people forget to call this function]
3232 Also no dynamic allocation is done unless new extensions are added:
3233 so if we don't add custom extensions there is no need to call
3234 X509V3_EXT_cleanup().
3237 *) Modify enc utility's salting as follows: make salting the default. Add a
3238 magic header, so unsalted files fail gracefully instead of just decrypting
3239 to garbage. This is because not salting is a big security hole, so people
3240 should be discouraged from doing it.
3243 *) Fixes and enhancements to the 'x509' utility. It allowed a message
3244 digest to be passed on the command line but it only used this
3245 parameter when signing a certificate. Modified so all relevant
3246 operations are affected by the digest parameter including the
3247 -fingerprint and -x509toreq options. Also -x509toreq choked if a
3248 DSA key was used because it didn't fix the digest.
3251 *) Initial certificate chain verify code. Currently tests the untrusted
3252 certificates for consistency with the verify purpose (which is set
3253 when the X509_STORE_CTX structure is set up) and checks the pathlength.
3255 There is a NO_CHAIN_VERIFY compilation option to keep the old behaviour:
3256 this is because it will reject chains with invalid extensions whereas
3257 every previous version of OpenSSL and SSLeay made no checks at all.
3259 Trust code: checks the root CA for the relevant trust settings. Trust
3260 settings have an initial value consistent with the verify purpose: e.g.
3261 if the verify purpose is for SSL client use it expects the CA to be
3262 trusted for SSL client use. However the default value can be changed to
3263 permit custom trust settings: one example of this would be to only trust
3264 certificates from a specific "secure" set of CAs.
3266 Also added X509_STORE_CTX_new() and X509_STORE_CTX_free() functions
3267 which should be used for version portability: especially since the
3268 verify structure is likely to change more often now.
3270 SSL integration. Add purpose and trust to SSL_CTX and SSL and functions
3271 to set them. If not set then assume SSL clients will verify SSL servers
3274 Two new options to the verify program: -untrusted allows a set of
3275 untrusted certificates to be passed in and -purpose which sets the
3276 intended purpose of the certificate. If a purpose is set then the
3277 new chain verify code is used to check extension consistency.
3280 *) Support for the authority information access extension.
3283 *) Modify RSA and DSA PEM read routines to transparently handle
3284 PKCS#8 format private keys. New *_PUBKEY_* functions that handle
3285 public keys in a format compatible with certificate
3286 SubjectPublicKeyInfo structures. Unfortunately there were already
3287 functions called *_PublicKey_* which used various odd formats so
3288 these are retained for compatibility: however the DSA variants were
3289 never in a public release so they have been deleted. Changed dsa/rsa
3290 utilities to handle the new format: note no releases ever handled public
3291 keys so we should be OK.
3293 The primary motivation for this change is to avoid the same fiasco
3294 that dogs private keys: there are several incompatible private key
3295 formats some of which are standard and some OpenSSL specific and
3296 require various evil hacks to allow partial transparent handling and
3297 even then it doesn't work with DER formats. Given the option anything
3298 other than PKCS#8 should be dumped: but the other formats have to
3299 stay in the name of compatibility.
3301 With public keys and the benefit of hindsight one standard format
3302 is used which works with EVP_PKEY, RSA or DSA structures: though
3303 it clearly returns an error if you try to read the wrong kind of key.
3305 Added a -pubkey option to the 'x509' utility to output the public key.
3306 Also rename the EVP_PKEY_get_*() to EVP_PKEY_rget_*()
3307 (renamed to EVP_PKEY_get1_*() in the OpenSSL 0.9.5 release) and add
3308 EVP_PKEY_rset_*() functions (renamed to EVP_PKEY_set1_*())
3309 that do the same as the EVP_PKEY_assign_*() except they up the
3310 reference count of the added key (they don't "swallow" the
3314 *) Fixes to crypto/x509/by_file.c the code to read in certificates and
3315 CRLs would fail if the file contained no certificates or no CRLs:
3316 added a new function to read in both types and return the number
3317 read: this means that if none are read it will be an error. The
3318 DER versions of the certificate and CRL reader would always fail
3319 because it isn't possible to mix certificates and CRLs in DER format
3320 without choking one or the other routine. Changed this to just read
3321 a certificate: this is the best we can do. Also modified the code
3322 in apps/verify.c to take notice of return codes: it was previously
3323 attempting to read in certificates from NULL pointers and ignoring
3324 any errors: this is one reason why the cert and CRL reader seemed
3325 to work. It doesn't check return codes from the default certificate
3326 routines: these may well fail if the certificates aren't installed.
3329 *) Code to support otherName option in GeneralName.
3332 *) First update to verify code. Change the verify utility
3333 so it warns if it is passed a self signed certificate:
3334 for consistency with the normal behaviour. X509_verify
3335 has been modified to it will now verify a self signed
3336 certificate if *exactly* the same certificate appears
3337 in the store: it was previously impossible to trust a
3338 single self signed certificate. This means that:
3339 openssl verify ss.pem
3340 now gives a warning about a self signed certificate but
3341 openssl verify -CAfile ss.pem ss.pem
3345 *) For servers, store verify_result in SSL_SESSION data structure
3346 (and add it to external session representation).
3347 This is needed when client certificate verifications fails,
3348 but an application-provided verification callback (set by
3349 SSL_CTX_set_cert_verify_callback) allows accepting the session
3350 anyway (i.e. leaves x509_store_ctx->error != X509_V_OK
3351 but returns 1): When the session is reused, we have to set
3352 ssl->verify_result to the appropriate error code to avoid
3354 [Bodo Moeller, problem pointed out by Lutz Jaenicke]
3356 *) Fix a bug in the new PKCS#7 code: it didn't consider the
3357 case in PKCS7_dataInit() where the signed PKCS7 structure
3358 didn't contain any existing data because it was being created.
3359 [Po-Cheng Chen <pocheng@nst.com.tw>, slightly modified by Steve Henson]
3361 *) Add a salt to the key derivation routines in enc.c. This
3362 forms the first 8 bytes of the encrypted file. Also add a
3363 -S option to allow a salt to be input on the command line.
3366 *) New function X509_cmp(). Oddly enough there wasn't a function
3367 to compare two certificates. We do this by working out the SHA1
3368 hash and comparing that. X509_cmp() will be needed by the trust
3372 *) SSL_get1_session() is like SSL_get_session(), but increments
3373 the reference count in the SSL_SESSION returned.
3374 [Geoff Thorpe <geoff@eu.c2.net>]
3376 *) Fix for 'req': it was adding a null to request attributes.
3377 Also change the X509_LOOKUP and X509_INFO code to handle
3378 certificate auxiliary information.
3381 *) Add support for 40 and 64 bit RC2 and RC4 algorithms: document
3385 *) Add the possibility to add extra information to the memory leak
3386 detecting output, to form tracebacks, showing from where each
3387 allocation was originated: CRYPTO_push_info("constant string") adds
3388 the string plus current file name and line number to a per-thread
3389 stack, CRYPTO_pop_info() does the obvious, CRYPTO_remove_all_info()
3390 is like calling CYRPTO_pop_info() until the stack is empty.
3391 Also updated memory leak detection code to be multi-thread-safe.
3394 *) Add options -text and -noout to pkcs7 utility and delete the
3395 encryption options which never did anything. Update docs.
3398 *) Add options to some of the utilities to allow the pass phrase
3399 to be included on either the command line (not recommended on
3400 OSes like Unix) or read from the environment. Update the
3401 manpages and fix a few bugs.
3404 *) Add a few manpages for some of the openssl commands.
3407 *) Fix the -revoke option in ca. It was freeing up memory twice,
3408 leaking and not finding already revoked certificates.
3411 *) Extensive changes to support certificate auxiliary information.
3412 This involves the use of X509_CERT_AUX structure and X509_AUX
3413 functions. An X509_AUX function such as PEM_read_X509_AUX()
3414 can still read in a certificate file in the usual way but it
3415 will also read in any additional "auxiliary information". By
3416 doing things this way a fair degree of compatibility can be
3417 retained: existing certificates can have this information added
3418 using the new 'x509' options.
3420 Current auxiliary information includes an "alias" and some trust
3421 settings. The trust settings will ultimately be used in enhanced
3422 certificate chain verification routines: currently a certificate
3423 can only be trusted if it is self signed and then it is trusted
3427 *) Fix assembler for Alpha (tested only on DEC OSF not Linux or *BSD).
3428 The problem was that one of the replacement routines had not been working
3429 since SSLeay releases. For now the offending routine has been replaced
3430 with non-optimised assembler. Even so, this now gives around 95%
3431 performance improvement for 1024 bit RSA signs.
3434 *) Hack to fix PKCS#7 decryption when used with some unorthodox RC2
3435 handling. Most clients have the effective key size in bits equal to
3436 the key length in bits: so a 40 bit RC2 key uses a 40 bit (5 byte) key.
3437 A few however don't do this and instead use the size of the decrypted key
3438 to determine the RC2 key length and the AlgorithmIdentifier to determine
3439 the effective key length. In this case the effective key length can still
3440 be 40 bits but the key length can be 168 bits for example. This is fixed
3441 by manually forcing an RC2 key into the EVP_PKEY structure because the
3442 EVP code can't currently handle unusual RC2 key sizes: it always assumes
3443 the key length and effective key length are equal.
3446 *) Add a bunch of functions that should simplify the creation of
3447 X509_NAME structures. Now you should be able to do:
3448 X509_NAME_add_entry_by_txt(nm, "CN", MBSTRING_ASC, "Steve", -1, -1, 0);
3449 and have it automatically work out the correct field type and fill in
3450 the structures. The more adventurous can try:
3451 X509_NAME_add_entry_by_txt(nm, field, MBSTRING_UTF8, str, -1, -1, 0);
3452 and it will (hopefully) work out the correct multibyte encoding.
3455 *) Change the 'req' utility to use the new field handling and multibyte
3456 copy routines. Before the DN field creation was handled in an ad hoc
3457 way in req, ca, and x509 which was rather broken and didn't support
3458 BMPStrings or UTF8Strings. Since some software doesn't implement
3459 BMPStrings or UTF8Strings yet, they can be enabled using the config file
3460 using the dirstring_type option. See the new comment in the default
3461 openssl.cnf for more info.
3464 *) Make crypto/rand/md_rand.c more robust:
3465 - Assure unique random numbers after fork().
3466 - Make sure that concurrent threads access the global counter and
3467 md serializably so that we never lose entropy in them
3468 or use exactly the same state in multiple threads.
3469 Access to the large state is not always serializable because
3470 the additional locking could be a performance killer, and
3471 md should be large enough anyway.
3474 *) New file apps/app_rand.c with commonly needed functionality
3475 for handling the random seed file.
3477 Use the random seed file in some applications that previously did not:
3479 dsaparam -genkey (which also ignored its '-rand' option),
3482 x509 (when signing).
3483 Except on systems with /dev/urandom, it is crucial to have a random
3484 seed file at least for key creation, DSA signing, and for DH exchanges;
3485 for RSA signatures we could do without one.
3487 gendh and gendsa (unlike genrsa) used to read only the first byte
3488 of each file listed in the '-rand' option. The function as previously
3489 found in genrsa is now in app_rand.c and is used by all programs
3490 that support '-rand'.
3493 *) In RAND_write_file, use mode 0600 for creating files;
3494 don't just chmod when it may be too late.
3497 *) Report an error from X509_STORE_load_locations
3498 when X509_LOOKUP_load_file or X509_LOOKUP_add_dir failed.
3501 *) New function ASN1_mbstring_copy() this copies a string in either
3502 ASCII, Unicode, Universal (4 bytes per character) or UTF8 format
3503 into an ASN1_STRING type. A mask of permissible types is passed
3504 and it chooses the "minimal" type to use or an error if not type
3508 *) Add function equivalents to the various macros in asn1.h. The old
3509 macros are retained with an M_ prefix. Code inside the library can
3510 use the M_ macros. External code (including the openssl utility)
3511 should *NOT* in order to be "shared library friendly".
3514 *) Add various functions that can check a certificate's extensions
3515 to see if it usable for various purposes such as SSL client,
3516 server or S/MIME and CAs of these types. This is currently
3517 VERY EXPERIMENTAL but will ultimately be used for certificate chain
3518 verification. Also added a -purpose flag to x509 utility to
3519 print out all the purposes.
3522 *) Add a CRYPTO_EX_DATA to X509 certificate structure and associated
3526 *) New X509V3_{X509,CRL,REVOKED}_get_d2i() functions. These will search
3527 for, obtain and decode and extension and obtain its critical flag.
3528 This allows all the necessary extension code to be handled in a
3529 single function call.
3532 *) RC4 tune-up featuring 30-40% performance improvement on most RISC
3533 platforms. See crypto/rc4/rc4_enc.c for further details.
3536 *) New -noout option to asn1parse. This causes no output to be produced
3537 its main use is when combined with -strparse and -out to extract data
3538 from a file (which may not be in ASN.1 format).
3541 *) Fix for pkcs12 program. It was hashing an invalid certificate pointer
3542 when producing the local key id.
3543 [Richard Levitte <levitte@stacken.kth.se>]
3545 *) New option -dhparam in s_server. This allows a DH parameter file to be
3546 stated explicitly. If it is not stated then it tries the first server
3547 certificate file. The previous behaviour hard coded the filename
3551 *) Add -pubin and -pubout options to the rsa and dsa commands. These allow
3552 a public key to be input or output. For example:
3553 openssl rsa -in key.pem -pubout -out pubkey.pem
3554 Also added necessary DSA public key functions to handle this.
3557 *) Fix so PKCS7_dataVerify() doesn't crash if no certificates are contained
3558 in the message. This was handled by allowing
3559 X509_find_by_issuer_and_serial() to tolerate a NULL passed to it.
3560 [Steve Henson, reported by Sampo Kellomaki <sampo@mail.neuronio.pt>]
3562 *) Fix for bug in d2i_ASN1_bytes(): other ASN1 functions add an extra null
3563 to the end of the strings whereas this didn't. This would cause problems
3564 if strings read with d2i_ASN1_bytes() were later modified.
3565 [Steve Henson, reported by Arne Ansper <arne@ats.cyber.ee>]
3567 *) Fix for base64 decode bug. When a base64 bio reads only one line of
3568 data and it contains EOF it will end up returning an error. This is
3569 caused by input 46 bytes long. The cause is due to the way base64
3570 BIOs find the start of base64 encoded data. They do this by trying a
3571 trial decode on each line until they find one that works. When they
3572 do a flag is set and it starts again knowing it can pass all the
3573 data directly through the decoder. Unfortunately it doesn't reset
3574 the context it uses. This means that if EOF is reached an attempt
3575 is made to pass two EOFs through the context and this causes the
3576 resulting error. This can also cause other problems as well. As is
3577 usual with these problems it takes *ages* to find and the fix is
3578 trivial: move one line.
3579 [Steve Henson, reported by ian@uns.ns.ac.yu (Ivan Nejgebauer) ]
3581 *) Ugly workaround to get s_client and s_server working under Windows. The
3582 old code wouldn't work because it needed to select() on sockets and the
3583 tty (for keypresses and to see if data could be written). Win32 only
3584 supports select() on sockets so we select() with a 1s timeout on the
3585 sockets and then see if any characters are waiting to be read, if none
3586 are present then we retry, we also assume we can always write data to
3587 the tty. This isn't nice because the code then blocks until we've
3588 received a complete line of data and it is effectively polling the
3589 keyboard at 1s intervals: however it's quite a bit better than not
3590 working at all :-) A dedicated Windows application might handle this
3591 with an event loop for example.
3594 *) Enhance RSA_METHOD structure. Now there are two extra methods, rsa_sign
3595 and rsa_verify. When the RSA_FLAGS_SIGN_VER option is set these functions
3596 will be called when RSA_sign() and RSA_verify() are used. This is useful
3597 if rsa_pub_dec() and rsa_priv_enc() equivalents are not available.
3598 For this to work properly RSA_public_decrypt() and RSA_private_encrypt()
3599 should *not* be used: RSA_sign() and RSA_verify() must be used instead.
3600 This necessitated the support of an extra signature type NID_md5_sha1
3601 for SSL signatures and modifications to the SSL library to use it instead
3602 of calling RSA_public_decrypt() and RSA_private_encrypt().
3605 *) Add new -verify -CAfile and -CApath options to the crl program, these
3606 will lookup a CRL issuers certificate and verify the signature in a
3607 similar way to the verify program. Tidy up the crl program so it
3608 no longer accesses structures directly. Make the ASN1 CRL parsing a bit
3609 less strict. It will now permit CRL extensions even if it is not
3610 a V2 CRL: this will allow it to tolerate some broken CRLs.
3613 *) Initialize all non-automatic variables each time one of the openssl
3614 sub-programs is started (this is necessary as they may be started
3615 multiple times from the "OpenSSL>" prompt).
3616 [Lennart Bang, Bodo Moeller]
3618 *) Preliminary compilation option RSA_NULL which disables RSA crypto without
3619 removing all other RSA functionality (this is what NO_RSA does). This
3620 is so (for example) those in the US can disable those operations covered
3621 by the RSA patent while allowing storage and parsing of RSA keys and RSA
3625 *) Non-copying interface to BIO pairs.
3626 (still largely untested)
3629 *) New function ANS1_tag2str() to convert an ASN1 tag to a descriptive
3630 ASCII string. This was handled independently in various places before.
3633 *) New functions UTF8_getc() and UTF8_putc() that parse and generate
3634 UTF8 strings a character at a time.
3637 *) Use client_version from client hello to select the protocol
3638 (s23_srvr.c) and for RSA client key exchange verification
3639 (s3_srvr.c), as required by the SSL 3.0/TLS 1.0 specifications.
3642 *) Add various utility functions to handle SPKACs, these were previously
3643 handled by poking round in the structure internals. Added new function
3644 NETSCAPE_SPKI_print() to print out SPKAC and a new utility 'spkac' to
3645 print, verify and generate SPKACs. Based on an original idea from
3646 Massimiliano Pala <madwolf@comune.modena.it> but extensively modified.
projects / openssl.git / blob