5 Changes between 0.9.6 and 0.9.7 [xx XXX 2001]
7 OpenSSL 0.9.6a/0.9.6b (bugfix releases, 5 Apr 2001 and 9 July 2001)
8 and OpenSSL 0.9.7 were developped in parallel, based on OpenSSL 0.9.6.
10 Change log entries are tagged as follows:
11 -) applies to 0.9.6a/0.9.6b/0.9.6c only
12 *) applies to 0.9.6a/0.9.6b/0.9.6c and 0.9.7
13 +) applies to 0.9.7 only
15 *) s3_srvr.c: allow sending of large client certificate lists (> 16 kB).
16 This function was broken, as the check for a new client hello message
17 to handle SGC did not allow these large messages.
18 (Tracked down by "Douglas E. Engert" <deengert@anl.gov>.)
21 *) Add alert descriptions for TLSv1 to SSL_alert_desc_string[_long]().
24 +) Add EVP test program.
27 +) Add symmetric cipher support to ENGINE. Expect the API to change!
30 +) New CRL functions: X509_CRL_set_version(), X509_CRL_set_issuer_name()
31 X509_CRL_set_lastUpdate(), X509_CRL_set_nextUpdate(), X509_CRL_sort(),
32 X509_REVOKED_set_serialNumber(), and X509_REVOKED_set_revocationDate().
33 These allow a CRL to be built without having to access X509_CRL fields
34 directly. Modify 'ca' application to use new functions.
37 *) Fix buggy behaviour of BIO_get_num_renegotiates() and BIO_ctrl()
38 for BIO_C_GET_WRITE_BUF_SIZE ("Stephen Hinton" <shinton@netopia.com>).
41 *) Rework the configuration and shared library support for Tru64 Unix.
42 The configuration part makes use of modern compiler features and
43 still retains old compiler behavior for those that run older versions
44 of the OS. The shared library support part includes a variant that
45 uses the RPATH feature, and is available through the speciel
46 configuration target "alpha-cc-rpath", which will never be selected
48 [Tim Mooney <mooney@dogbert.cc.ndsu.NoDak.edu> via Richard Levitte]
50 *) In ssl3_get_key_exchange (ssl/s3_clnt.c), call ssl3_get_message()
51 with the same message size as in ssl3_get_certificate_request().
52 Otherwise, if no ServerKeyExchange message occurs, CertificateRequest
53 messages might inadvertently be reject as too long.
54 [Petr Lampa <lampa@fee.vutbr.cz>]
56 +) Move SSL_OP_TLS_ROLLBACK_BUG out of the SSL_OP_ALL list of recommended
57 bug workarounds. Rollback attack detection is a security feature.
58 The problem will only arise on OpenSSL servers when TLSv1 is not
59 available (sslv3_server_method() or SSL_OP_NO_TLSv1).
60 Software authors not wanting to support TLSv1 will have special reasons
61 for their choice and can explicitly enable this option.
62 [Bodo Moeller, Lutz Jaenicke]
64 +) Rationalise EVP so it can be extended: don't include a union of
65 cipher/digest structures, add init/cleanup functions. This also reduces
66 the number of header dependencies.
69 +) Make DES key schedule conform to the usual scheme, as well as
70 correcting its structure. This means that calls to DES functions
71 now have to pass a pointer to a des_key_schedule instead of a
72 plain des_key_schedule (which was actually always a pointer
76 +) Enhanced support for IA-64 Unix platforms (well, Linux and HP-UX).
79 *) Modified SSL library such that the verify_callback that has been set
80 specificly for an SSL object with SSL_set_verify() is actually being
81 used. Before the change, a verify_callback set with this function was
82 ignored and the verify_callback() set in the SSL_CTX at the time of
83 the call was used. New function X509_STORE_CTX_set_verify_cb() introduced
84 to allow the necessary settings.
87 +) Initial reduction of linker bloat: the use of some functions, such as
88 PEM causes large amounts of unused functions to be linked in due to
89 poor organisation. For example pem_all.c contains every PEM function
90 which has a knock on effect of linking in large amounts of (unused)
91 ASN1 code. Grouping together similar functions and splitting unrelated
92 functions prevents this.
95 *) Initialize static variable in crypto/dsa/dsa_lib.c and crypto/dh/dh_lib.c
96 explicitely to NULL, as at least on Solaris 8 this seems not always to be
97 done automatically (in contradiction to the requirements of the C
98 standard). This made problems when used from OpenSSH.
101 *) In OpenSSL 0.9.6a and 0.9.6b, crypto/dh/dh_key.c ignored
102 dh->length and always used
104 BN_rand_range(priv_key, dh->p).
106 BN_rand_range() is not necessary for Diffie-Hellman, and this
107 specific range makes Diffie-Hellman unnecessarily inefficient if
108 dh->length (recommended exponent length) is much smaller than the
109 length of dh->p. We could use BN_rand_range() if the order of
110 the subgroup was stored in the DH structure, but we only have
115 BN_rand(priv_key, l, ...)
117 where 'l' is dh->length if this is defined, or BN_num_bits(dh->p)-1
123 RSA_eay_public_encrypt
124 RSA_eay_private_decrypt
125 RSA_eay_private_encrypt (signing)
126 RSA_eay_public_decrypt (signature verification)
128 (default implementations for RSA_public_encrypt,
129 RSA_private_decrypt, RSA_private_encrypt, RSA_public_decrypt),
130 always reject numbers >= n.
133 *) In crypto/rand/md_rand.c, use a new short-time lock CRYPTO_LOCK_RAND2
134 to synchronize access to 'locking_thread'. This is necessary on
135 systems where access to 'locking_thread' (an 'unsigned long'
136 variable) is not atomic.
139 *) In crypto/rand/md_rand.c, set 'locking_thread' to current thread's ID
140 *before* setting the 'crypto_lock_rand' flag. The previous code had
141 a race condition if 0 is a valid thread ID.
142 [Travis Vitek <vitek@roguewave.com>]
144 +) Cleanup of EVP macros.
147 +) Change historical references to {NID,SN,LN}_des_ede and ede3 to add the
151 +) Add initial OCSP responder support to ocsp application. The
152 revocation information is handled using the text based index
153 use by the ca application. The responder can either handle
154 requests generated internally, supplied in files (for example
155 via a CGI script) or using an internal minimal server.
158 +) Add configuration choices to get zlib compression for TLS.
161 +) Changes to Kerberos SSL for RFC 2712 compliance:
162 1. Implemented real KerberosWrapper, instead of just using
163 KRB5 AP_REQ message. [Thanks to Simon Wilkinson <sxw@sxw.org.uk>]
164 2. Implemented optional authenticator field of KerberosWrapper.
166 Added openssl-style ASN.1 macros for Kerberos ticket, ap_req,
167 and authenticator structs; see crypto/krb5/.
169 Generalized Kerberos calls to support multiple Kerberos libraries.
170 [Vern Staats <staatsvr@asc.hpc.mil>,
171 Jeffrey Altman <jaltman@columbia.edu>
174 +) Cause 'openssl speed' to use fully hard-coded DSA keys as it
175 already does with RSA. testdsa.h now has 'priv_key/pub_key'
176 values for each of the key sizes rather than having just
177 parameters (and 'speed' generating keys each time).
180 -) OpenSSL 0.9.6b released [9 July 2001]
182 *) Change ssleay_rand_bytes (crypto/rand/md_rand.c)
183 to avoid a SSLeay/OpenSSL PRNG weakness pointed out by
184 Markku-Juhani O. Saarinen <markku-juhani.saarinen@nokia.com>:
185 PRNG state recovery was possible based on the output of
186 one PRNG request appropriately sized to gain knowledge on
187 'md' followed by enough consecutive 1-byte PRNG requests
188 to traverse all of 'state'.
190 1. When updating 'md_local' (the current thread's copy of 'md')
191 during PRNG output generation, hash all of the previous
192 'md_local' value, not just the half used for PRNG output.
194 2. Make the number of bytes from 'state' included into the hash
195 independent from the number of PRNG bytes requested.
197 The first measure alone would be sufficient to avoid
198 Markku-Juhani's attack. (Actually it had never occurred
199 to me that the half of 'md_local' used for chaining was the
200 half from which PRNG output bytes were taken -- I had always
201 assumed that the secret half would be used.) The second
202 measure makes sure that additional data from 'state' is never
203 mixed into 'md_local' in small portions; this heuristically
204 further strengthens the PRNG.
207 +) Speed up EVP routines.
210 type 8 bytes 64 bytes 256 bytes 1024 bytes 8192 bytes
211 des-cbc 4408.85k 5560.51k 5778.46k 5862.20k 5825.16k
212 des-cbc 4389.55k 5571.17k 5792.23k 5846.91k 5832.11k
213 des-cbc 4394.32k 5575.92k 5807.44k 5848.37k 5841.30k
215 des-cbc 3482.66k 5069.49k 5496.39k 5614.16k 5639.28k
216 des-cbc 3480.74k 5068.76k 5510.34k 5609.87k 5635.52k
217 des-cbc 3483.72k 5067.62k 5504.60k 5708.01k 5724.80k
220 des-cbc 4660.16k 5650.19k 5807.19k 5827.13k 5783.32k
222 des-cbc 3624.96k 5258.21k 5530.91k 5624.30k 5628.26k
225 *) Fix crypto/bn/asm/mips3.s.
228 *) When only the key is given to "enc", the IV is undefined. Print out
229 an error message in this case.
232 +) Added the OS2-EMX target.
233 ["Brian Havard" <brianh@kheldar.apana.org.au> and Richard Levitte]
235 +) Rewrite apps to use NCONF routines instead of the old CONF. New functions
236 to support NCONF routines in extension code. New function CONF_set_nconf()
237 to allow functions which take an NCONF to also handle the old LHASH
238 structure: this means that the old CONF compatible routines can be
239 retained (in particular wrt extensions) without having to duplicate the
240 code. New function X509V3_add_ext_nconf_sk to add extensions to a stack.
243 *) Handle special case when X509_NAME is empty in X509 printing routines.
246 *) In dsa_do_verify (crypto/dsa/dsa_ossl.c), verify that r and s are
247 positive and less than q.
250 +) Enhance the general user interface with mechanisms for inner control
251 and with pssibilities to have yes/no kind of prompts.
254 +) Change all calls to low level digest routines in the library and
255 applications to use EVP. Add missing calls to HMAC_cleanup() and
256 don't assume HMAC_CTX can be copied using memcpy().
257 [Verdon Walker <VWalker@novell.com>, Steve Henson]
259 +) Add the possibility to control engines through control names but with
260 arbitrary arguments instead of just a string.
261 Change the key loaders to take a UI_METHOD instead of a callback
262 function pointer. NOTE: this breaks binary compatibility with earlier
263 versions of OpenSSL [engine].
264 Addapt the nCipher code for these new conditions and add a card insertion
268 +) Enhance the general user interface with mechanisms to better support
269 dialog box interfaces, application-defined prompts, the possibility
270 to use defaults (for example default passwords from somewhere else)
271 and interrupts/cancelations.
274 *) Don't change *pointer in CRYPTO_add_lock() is add_lock_callback is
275 used: it isn't thread safe and the add_lock_callback should handle
277 [Paul Rose <Paul.Rose@bridge.com>]
279 *) Verify that incoming data obeys the block size in
280 ssl3_enc (ssl/s3_enc.c) and tls1_enc (ssl/t1_enc.c).
283 +) Tidy up PKCS#12 attribute handling. Add support for the CSP name
284 attribute in PKCS#12 files, add new -CSP option to pkcs12 utility.
288 [Ulf Möller, Bodo Möller]
290 *) The countermeasure against Bleichbacher's attack on PKCS #1 v1.5
291 RSA encryption was accidentily removed in s3_srvr.c in OpenSSL 0.9.5
292 when fixing the server behaviour for backwards-compatible 'client
293 hello' messages. (Note that the attack is impractical against
294 SSL 3.0 and TLS 1.0 anyway because length and version checking
295 means that the probability of guessing a valid ciphertext is
296 around 2^-40; see section 5 in Bleichenbacher's CRYPTO '98
299 Before 0.9.5, the countermeasure (hide the error by generating a
300 random 'decryption result') did not work properly because
301 ERR_clear_error() was missing, meaning that SSL_get_error() would
302 detect the supposedly ignored error.
304 Both problems are now fixed.
307 *) In crypto/bio/bf_buff.c, increase DEFAULT_BUFFER_SIZE to 4096
308 (previously it was 1024).
311 +) Fix a memory leak in 'sk_dup()' in the case reallocation fails. (Also
312 tidy up some unecessarily weird code in 'sk_new()').
313 [Geoff, reported by Diego Tartara <dtartara@novamens.com>]
315 +) Change the key loading routines for ENGINEs to use the same kind
316 callback (pem_password_cb) as all other routines that need this
320 *) Fix for compatibility mode trust settings: ignore trust settings
321 unless some valid trust or reject settings are present.
324 *) Fix for blowfish EVP: its a variable length cipher.
327 +) Increase ENTROPY_NEEDED to 32 bytes, as Rijndael can operate with
328 256 bit (=32 byte) keys. Of course seeding with more entropy bytes
329 than this minimum value is recommended.
332 +) New random seeder for OpenVMS, using the system process statistics
333 that are easily reachable.
336 +) Windows apparently can't transparently handle global
337 variables defined in DLLs. Initialisations such as:
339 const ASN1_ITEM *it = &ASN1_INTEGER_it;
341 wont compile. This is used by the any applications that need to
342 delcare their own ASN1 modules. This was fixed by adding the option
343 EXPORT_VAR_AS_FN to all Win32 platforms, although this isn't strictly
344 needed for static libraries under Win32.
347 +) New functions X509_PURPOSE_set() and X509_TRUST_set() to handle
348 setting of purpose and trust fields. New X509_STORE trust and
349 purpose functions and tidy up setting in other SSL functions.
352 +) Add copies of X509_STORE_CTX fields and callbacks to X509_STORE
353 structure. These are inherited by X509_STORE_CTX when it is
354 initialised. This allows various defaults to be set in the
355 X509_STORE structure (such as flags for CRL checking and custom
356 purpose or trust settings) for functions which only use X509_STORE_CTX
357 internally such as S/MIME.
359 Modify X509_STORE_CTX_purpose_inherit() so it only sets purposes and
360 trust settings if they are not set in X509_STORE. This allows X509_STORE
361 purposes and trust (in S/MIME for example) to override any set by default.
363 Add command line options for CRL checking to smime, s_client and s_server
367 +) Initial CRL based revocation checking. If the CRL checking flag(s)
368 are set then the CRL is looked up in the X509_STORE structure and
369 its validity and signature checked, then if the certificate is found
370 in the CRL the verify fails with a revoked error.
372 Various new CRL related callbacks added to X509_STORE_CTX structure.
374 Command line options added to 'verify' application to support this.
376 This needs some additional work, such as being able to handle multiple
377 CRLs with different times, extension based lookup (rather than just
378 by subject name) and ultimately more complete V2 CRL extension
382 +) Add a general user interface API (crypto/ui/). This is designed
383 to replace things like des_read_password and friends (backward
384 compatibility functions using this new API are provided).
385 The purpose is to remove prompting functions from the DES code
386 section as well as provide for prompting through dialog boxes in
387 a window system and the like.
390 *) Fix various bugs related to DSA S/MIME verification. Handle missing
391 parameters in DSA public key structures and return an error in the
392 DSA routines if parameters are absent.
395 *) In versions up to 0.9.6, RAND_file_name() resorted to file ".rnd"
396 in the current directory if neither $RANDFILE nor $HOME was set.
397 RAND_file_name() in 0.9.6a returned NULL in this case. This has
398 caused some confusion to Windows users who haven't defined $HOME.
399 Thus RAND_file_name() is changed again: e_os.h can define a
400 DEFAULT_HOME, which will be used if $HOME is not set.
401 For Windows, we use "C:"; on other platforms, we still require
402 environment variables.
404 +) Add "ex_data" support to ENGINE so implementations can add state at a
405 per-structure level rather than having to store it globally.
408 +) Make it possible for ENGINE structures to be copied when retrieved by
409 ENGINE_by_id() if the ENGINE specifies a new flag: ENGINE_FLAGS_BY_ID_COPY.
410 This causes the "original" ENGINE structure to act like a template,
411 analogous to the RSA vs. RSA_METHOD type of separation. Because of this
412 operational state can be localised to each ENGINE structure, despite the
413 fact they all share the same "methods". New ENGINE structures returned in
414 this case have no functional references and the return value is the single
415 structural reference. This matches the single structural reference returned
416 by ENGINE_by_id() normally, when it is incremented on the pre-existing
420 +) Fix ASN1 decoder when decoding type ANY and V_ASN1_OTHER: since this
421 needs to match any other type at all we need to manually clear the
425 +) Changes to the "openssl engine" utility to include;
426 - verbosity levels ('-v', '-vv', and '-vvv') that provide information
427 about an ENGINE's available control commands.
428 - executing control commands from command line arguments using the
429 '-pre' and '-post' switches. '-post' is only used if '-t' is
430 specified and the ENGINE is successfully initialised. The syntax for
431 the individual commands are colon-separated, for example;
432 openssl engine chil -pre FORK_CHECK:0 -pre SO_PATH:/lib/test.so
435 +) New dynamic control command support for ENGINEs. ENGINEs can now
436 declare their own commands (numbers), names (strings), descriptions,
437 and input types for run-time discovery by calling applications. A
438 subset of these commands are implicitly classed as "executable"
439 depending on their input type, and only these can be invoked through
440 the new string-based API function ENGINE_ctrl_cmd_string(). (Eg. this
441 can be based on user input, config files, etc). The distinction is
442 that "executable" commands cannot return anything other than a boolean
443 result and can only support numeric or string input, whereas some
444 discoverable commands may only be for direct use through
445 ENGINE_ctrl(), eg. supporting the exchange of binary data, function
446 pointers, or other custom uses. The "executable" commands are to
447 support parameterisations of ENGINE behaviour that can be
448 unambiguously defined by ENGINEs and used consistently across any
449 OpenSSL-based application. Commands have been added to all the
450 existing hardware-supporting ENGINEs, noticeably "SO_PATH" to allow
451 control over shared-library paths without source code alterations.
454 +) Changed all ENGINE implementations to dynamically allocate their
455 ENGINEs rather than declaring them statically. Apart from this being
456 necessary with the removal of the ENGINE_FLAGS_MALLOCED distinction,
457 this also allows the implementations to compile without using the
458 internal engine_int.h header.
461 +) Minor adjustment to "rand" code. RAND_get_rand_method() now returns a
462 'const' value. Any code that should be able to modify a RAND_METHOD
463 should already have non-const pointers to it (ie. they should only
464 modify their own ones).
467 +) Made a variety of little tweaks to the ENGINE code.
468 - "atalla" and "ubsec" string definitions were moved from header files
469 to C code. "nuron" string definitions were placed in variables
470 rather than hard-coded - allowing parameterisation of these values
471 later on via ctrl() commands.
472 - Removed unused "#if 0"'d code.
473 - Fixed engine list iteration code so it uses ENGINE_free() to release
474 structural references.
475 - Constified the RAND_METHOD element of ENGINE structures.
476 - Constified various get/set functions as appropriate and added
477 missing functions (including a catch-all ENGINE_cpy that duplicates
478 all ENGINE values onto a new ENGINE except reference counts/state).
479 - Removed NULL parameter checks in get/set functions. Setting a method
480 or function to NULL is a way of cancelling out a previously set
481 value. Passing a NULL ENGINE parameter is just plain stupid anyway
482 and doesn't justify the extra error symbols and code.
483 - Deprecate the ENGINE_FLAGS_MALLOCED define and move the area for
484 flags from engine_int.h to engine.h.
485 - Changed prototypes for ENGINE handler functions (init(), finish(),
486 ctrl(), key-load functions, etc) to take an (ENGINE*) parameter.
489 *) Move 'if (!initialized) RAND_poll()' into regions protected by
490 CRYPTO_LOCK_RAND. This is not strictly necessary, but avoids
491 having multiple threads call RAND_poll() concurrently.
494 *) In crypto/rand/md_rand.c, replace 'add_do_not_lock' flag by a
495 combination of a flag and a thread ID variable.
496 Otherwise while one thread is in ssleay_rand_bytes (which sets the
497 flag), *other* threads can enter ssleay_add_bytes without obeying
498 the CRYPTO_LOCK_RAND lock (and may even illegaly release the lock
499 that they do not hold after the first thread unsets add_do_not_lock).
502 +) Implement binary inversion algorithm for BN_mod_inverse in addition
503 to the algorithm using long divison. The binary algorithm can be
504 used only if the modulus is odd. On 32-bit systems, it is faster
505 only for relatively small moduli (roughly 20-30% for 128-bit moduli,
506 roughly 5-15% for 256-bit moduli), so we use it only for moduli
507 up to 450 bits. In 64-bit environments, the binary algorithm
508 appears to be advantageous for much longer moduli; here we use it
509 for moduli up to 2048 bits.
512 *) Change bctest again: '-x' expressions are not available in all
516 -) OpenSSL 0.9.6a released [5 Apr 2001]
518 *) Fix a couple of memory leaks in PKCS7_dataDecode()
519 [Steve Henson, reported by Heyun Zheng <hzheng@atdsprint.com>]
521 *) Change Configure and Makefiles to provide EXE_EXT, which will contain
522 the default extension for executables, if any. Also, make the perl
523 scripts that use symlink() to test if it really exists and use "cp"
524 if it doesn't. All this made OpenSSL compilable and installable in
528 +) Rewrite CHOICE field setting in ASN1_item_ex_d2i(). The old code
529 could not support the combine flag in choice fields.
532 -) Fix for asn1_GetSequence() for indefinite length constructed data.
533 If SEQUENCE is length is indefinite just set c->slen to the total
534 amount of data available.
535 [Steve Henson, reported by shige@FreeBSD.org]
536 [This change does not apply to 0.9.7.]
538 *) Change bctest to avoid here-documents inside command substitution
539 (workaround for FreeBSD /bin/sh bug).
540 For compatibility with Ultrix, avoid shell functions (introduced
541 in the bctest version that searches along $PATH).
544 *) Rename 'des_encrypt' to 'des_encrypt1'. This avoids the clashes
545 with des_encrypt() defined on some operating systems, like Solaris
549 *) Check the result of RSA-CRT (see D. Boneh, R. DeMillo, R. Lipton:
550 On the Importance of Eliminating Errors in Cryptographic
551 Computations, J. Cryptology 14 (2001) 2, 101-119,
552 http://theory.stanford.edu/~dabo/papers/faults.ps.gz).
555 *) MIPS assembler BIGNUM division bug fix.
558 *) Disabled incorrect Alpha assembler code.
561 -) Fix PKCS#7 decode routines so they correctly update the length
562 after reading an EOC for the EXPLICIT tag.
564 [This change does not apply to 0.9.7.]
566 *) Fix bug in PKCS#12 key generation routines. This was triggered
567 if a 3DES key was generated with a 0 initial byte. Include
568 PKCS12_BROKEN_KEYGEN compilation option to retain the old
569 (but broken) behaviour.
572 *) Enhance bctest to search for a working bc along $PATH and print
574 [Tim Rice <tim@multitalents.net> via Richard Levitte]
576 +) Add a 'copy_extensions' option to the 'ca' utility. This copies
577 extensions from a certificate request to the certificate.
580 +) Allow multiple 'certopt' and 'nameopt' options to be separated
581 by commas. Add 'namopt' and 'certopt' options to the 'ca' config
582 file: this allows the display of the certificate about to be
583 signed to be customised, to allow certain fields to be included
584 or excluded and extension details. The old system didn't display
585 multicharacter strings properly, omitted fields not in the policy
586 and couldn't display additional details such as extensions.
589 *) Fix memory leaks in err.c: free err_data string if necessary;
590 don't write to the wrong index in ERR_set_error_data.
593 +) Function EC_POINTs_mul for simultaneous scalar multiplication
594 of an arbitrary number of elliptic curve points, optionally
595 including the generator defined for the EC_GROUP.
596 EC_POINT_mul is a simple wrapper function for the typical case
597 that the point list has just one item (besides the optional
601 +) First EC_METHODs for curves over GF(p):
603 EC_GFp_simple_method() uses the basic BN_mod_mul and BN_mod_sqr
604 operations and provides various method functions that can also
605 operate with faster implementations of modular arithmetic.
607 EC_GFp_mont_method() reuses most functions that are part of
608 EC_GFp_simple_method, but uses Montgomery arithmetic.
610 [Bodo Moeller; point addition and point doubling
611 implementation directly derived from source code provided by
612 Lenka Fibikova <fibikova@exp-math.uni-essen.de>]
614 +) Framework for elliptic curves (crypto/ec/ec.h, crypto/ec/ec_lcl.h,
617 Curves are EC_GROUP objects (with an optional group generator)
618 based on EC_METHODs that are built into the library.
620 Points are EC_POINT objects based on EC_GROUP objects.
622 Most of the framework would be able to handle curves over arbitrary
623 finite fields, but as there are no obvious types for fields other
624 than GF(p), some functions are limited to that for now.
627 +) Add the -HTTP option to s_server. It is similar to -WWW, but requires
628 that the file contains a complete HTTP response.
631 +) Add the ec directory to mkdef.pl and mkfiles.pl. In mkdef.pl
632 change the def and num file printf format specifier from "%-40sXXX"
633 to "%-39s XXX". The latter will always guarantee a space after the
634 field while the former will cause them to run together if the field
635 is 40 of more characters long.
638 +) Constify the cipher and digest 'method' functions and structures
639 and modify related functions to take constant EVP_MD and EVP_CIPHER
643 *) Implement ssl23_peek (analogous to ssl23_read), which previously
647 *) Replace rdtsc with _emit statements for VC++ version 5.
648 [Jeremy Cooper <jeremy@baymoo.org>]
650 +) Hide BN_CTX structure details in bn_lcl.h instead of publishing them
651 in <openssl/bn.h>. Also further increase BN_CTX_NUM to 32.
654 +) Modify EVP_Digest*() routines so they now return values. Although the
655 internal software routines can never fail additional hardware versions
659 +) Clean up crypto/err/err.h and change some error codes to avoid conflicts:
661 Previously ERR_R_FATAL was too small and coincided with ERR_LIB_PKCS7
662 (= ERR_R_PKCS7_LIB); it is now 64 instead of 32.
665 ERR_R_NESTED_ASN1_ERROR
667 ERR_R_MISSING_ASN1_EOS
668 were 4 .. 9, conflicting with
669 ERR_LIB_RSA (= ERR_R_RSA_LIB)
671 ERR_LIB_PEM (= ERR_R_PEM_LIB).
672 They are now 58 .. 63 (i.e., just below ERR_R_FATAL).
674 Add new error code 'ERR_R_INTERNAL_ERROR'.
677 +) Don't overuse locks in crypto/err/err.c: For data retrieval, CRYPTO_r_lock
681 +) New option '-subj arg' for 'openssl req' and 'openssl ca'. This
682 sets the subject name for a new request or supersedes the
683 subject name in a given request. Formats that can be parsed are
684 'CN=Some Name, OU=myOU, C=IT'
686 'CN=Some Name/OU=myOU/C=IT'.
688 Add options '-batch' and '-verbose' to 'openssl req'.
689 [Massimiliano Pala <madwolf@hackmasters.net>]
691 *) Make it possible to reuse SSLv2 sessions.
694 +) Introduce the possibility to access global variables through
695 functions on platform were that's the best way to handle exporting
696 global variables in shared libraries. To enable this functionality,
697 one must configure with "EXPORT_VAR_AS_FN" or defined the C macro
698 "OPENSSL_EXPORT_VAR_AS_FUNCTION" in crypto/opensslconf.h (the latter
699 is normally done by Configure or something similar).
701 To implement a global variable, use the macro OPENSSL_IMPLEMENT_GLOBAL
702 in the source file (foo.c) like this:
704 OPENSSL_IMPLEMENT_GLOBAL(int,foo)=1;
705 OPENSSL_IMPLEMENT_GLOBAL(double,bar);
707 To declare a global variable, use the macros OPENSSL_DECLARE_GLOBAL
708 and OPENSSL_GLOBAL_REF in the header file (foo.h) like this:
710 OPENSSL_DECLARE_GLOBAL(int,foo);
711 #define foo OPENSSL_GLOBAL_REF(foo)
712 OPENSSL_DECLARE_GLOBAL(double,bar);
713 #define bar OPENSSL_GLOBAL_REF(bar)
715 The #defines are very important, and therefore so is including the
716 header file everywere where the defined globals are used.
718 The macro OPENSSL_EXPORT_VAR_AS_FUNCTION also affects the definition
719 of ASN.1 items, but that structure is a bt different.
721 The largest change is in util/mkdef.pl which has been enhanced with
722 better and easier to understand logic to choose which symbols should
723 go into the Windows .def files as well as a number of fixes and code
724 cleanup (among others, algorithm keywords are now sorted
725 lexicographically to avoid constant rewrites).
728 *) In copy_email() check for >= 0 as a return value for
729 X509_NAME_get_index_by_NID() since 0 is a valid index.
730 [Steve Henson reported by Massimiliano Pala <madwolf@opensca.org>]
732 +) In BN_div() keep a copy of the sign of 'num' before writing the
733 result to 'rm' because if rm==num the value will be overwritten
734 and produce the wrong result if 'num' is negative: this caused
735 problems with BN_mod() and BN_nnmod().
738 +) Function OCSP_request_verify(). This checks the signature on an
739 OCSP request and verifies the signer certificate. The signer
740 certificate is just checked for a generic purpose and OCSP request
744 +) Add OCSP_check_validity() function to check the validity of OCSP
745 responses. OCSP responses are prepared in real time and may only
746 be a few seconds old. Simply checking that the current time lies
747 between thisUpdate and nextUpdate max reject otherwise valid responses
748 caused by either OCSP responder or client clock innacuracy. Instead
749 we allow thisUpdate and nextUpdate to fall within a certain period of
750 the current time. The age of the response can also optionally be
751 checked. Two new options -validity_period and -status_age added to
755 +) If signature or public key algorithm is unrecognized print out its
756 OID rather that just UNKOWN.
759 *) Avoid coredump with unsupported or invalid public keys by checking if
760 X509_get_pubkey() fails in PKCS7_verify(). Fix memory leak when
761 PKCS7_verify() fails with non detached data.
764 +) Change OCSP_cert_to_id() to tolerate a NULL subject certificate and
765 OCSP_cert_id_new() a NULL serialNumber. This allows a partial certificate
766 ID to be generated from the issuer certificate alone which can then be
767 passed to OCSP_id_issuer_cmp().
770 +) New compilation option ASN1_ITEM_FUNCTIONS. This causes the new
771 ASN1 modules to export functions returning ASN1_ITEM pointers
772 instead of the ASN1_ITEM structures themselves. This adds several
773 new macros which allow the underlying ASN1 function/structure to
774 be accessed transparently. As a result code should not use ASN1_ITEM
775 references directly (such as &X509_it) but instead use the relevant
776 macros (such as ASN1_ITEM_rptr(X509)). This option is to allow
777 use of the new ASN1 code on platforms where exporting structures
778 is problematical (for example in shared libraries) but exporting
779 functions returning pointers to structures is not.
782 +) Add support for overriding the generation of SSL/TLS session IDs.
783 These callbacks can be registered either in an SSL_CTX or per SSL.
784 The purpose of this is to allow applications to control, if they wish,
785 the arbitrary values chosen for use as session IDs, particularly as it
786 can be useful for session caching in multiple-server environments. A
787 command-line switch for testing this (and any client code that wishes
788 to use such a feature) has been added to "s_server".
789 [Geoff Thorpe, Lutz Jaenicke]
791 +) Modify mkdef.pl to recognise and parse prprocessor conditionals
792 of the form '#if defined(...) || defined(...) || ...' and
793 '#if !defined(...) && !defined(...) && ...'. This also avoids
794 the growing number of special cases it was previously handling.
797 *) Don't use getenv in library functions when run as setuid/setgid.
798 New function OPENSSL_issetugid().
801 +) Make all configuration macros available for application by making
802 sure they are available in opensslconf.h, by giving them names starting
803 with "OPENSSL_" to avoid conflicts with other packages and by making
804 sure e_os2.h will cover all platform-specific cases together with
806 Additionally, it is now possible to define configuration/platform-
807 specific names (called "system identities"). In the C code, these
808 are prefixed with "OPENSSL_SYSNAME_". e_os2.h will create another
809 macro with the name beginning with "OPENSSL_SYS_", which is determined
810 from "OPENSSL_SYSNAME_*" or compiler-specific macros depending on
814 +) New option -set_serial to 'req' and 'x509' this allows the serial
815 number to use to be specified on the command line. Previously self
816 signed certificates were hard coded with serial number 0 and the
817 CA options of 'x509' had to use a serial number in a file which was
821 *) Avoid false positives in memory leak detection code (crypto/mem_dbg.c)
822 due to incorrect handling of multi-threading:
824 1. Fix timing glitch in the MemCheck_off() portion of CRYPTO_mem_ctrl().
826 2. Fix logical glitch in is_MemCheck_on() aka CRYPTO_is_mem_check_on().
828 3. Count how many times MemCheck_off() has been called so that
829 nested use can be treated correctly. This also avoids
830 inband-signalling in the previous code (which relied on the
831 assumption that thread ID 0 is impossible).
834 +) New options to 'ca' utility to support V2 CRL entry extensions.
835 Currently CRL reason, invalidity date and hold instruction are
836 supported. Add new CRL extensions to V3 code and some new objects.
839 *) Add "-rand" option also to s_client and s_server.
842 +) New function EVP_CIPHER_CTX_set_padding() this is used to
843 disable standard block padding (aka PKCS#5 padding) in the EVP
844 API, which was previously mandatory. This means that the data is
845 not padded in any way and so the total length much be a multiple
846 of the block size, otherwise an error occurs.
849 +) Initial (incomplete) OCSP SSL support.
852 *) Fix CPU detection on Irix 6.x.
853 [Kurt Hockenbury <khockenb@stevens-tech.edu> and
854 "Bruce W. Forsberg" <bruce.forsberg@baesystems.com>]
856 +) New function OCSP_parse_url(). This splits up a URL into its host,
857 port and path components: primarily to parse OCSP URLs. New -url
858 option to ocsp utility.
861 +) New nonce behavior. The return value of OCSP_check_nonce() now
862 reflects the various checks performed. Applications can decide
863 whether to tolerate certain situations such as an absent nonce
864 in a response when one was present in a request: the ocsp application
865 just prints out a warning. New function OCSP_add1_basic_nonce()
866 this is to allow responders to include a nonce in a response even if
867 the request is nonce-less.
870 *) Fix X509_NAME bug which produced incorrect encoding if X509_NAME
873 [This change does not apply to 0.9.7.]
875 *) Use the cached encoding of an X509_NAME structure rather than
876 copying it. This is apparently the reason for the libsafe "errors"
877 but the code is actually correct.
880 +) Disable stdin buffering in load_cert (apps/apps.c) so that no certs are
881 skipped when using openssl x509 multiple times on a single input file,
882 e.g. "(openssl x509 -out cert1; openssl x509 -out cert2) <certs".
885 +) Make ASN1_UTCTIME_set_string() and ASN1_GENERALIZEDTIME_set_string()
886 set string type: to handle setting ASN1_TIME structures. Fix ca
887 utility to correctly initialize revocation date of CRLs.
890 +) New option SSL_OP_CIPHER_SERVER_PREFERENCE allows the server to override
891 the clients preferred ciphersuites and rather use its own preferences.
892 Should help to work around M$ SGC (Server Gated Cryptography) bug in
893 Internet Explorer by ensuring unchanged hash method during stepup.
894 (Also replaces the broken/deactivated SSL_OP_NON_EXPORT_FIRST option.)
897 +) Make mkdef.pl recognise all DECLARE_ASN1 macros, change rijndael
898 to aes and add a new 'exist' option to print out symbols that don't
902 +) Additional options to ocsp utility to allow flags to be set and
903 additional certificates supplied.
906 +) Add the option -VAfile to 'openssl ocsp', so the user can give the
907 OCSP client a number of certificate to only verify the response
911 *) Add new function BN_rand_range(), and fix DSA_sign_setup() to prevent
912 Bleichenbacher's DSA attack.
913 Extend BN_[pseudo_]rand: As before, top=1 forces the highest two bits
914 to be set and top=0 forces the highest bit to be set; top=-1 is new
915 and leaves the highest bit random.
916 [Ulf Moeller, Bodo Moeller]
918 +) Update Rijndael code to version 3.0 and change EVP AES ciphers to
919 handle the new API. Currently only ECB, CBC modes supported. Add new
920 AES OIDs. Add TLS AES ciphersuites as described in the "AES Ciphersuites
921 for TLS" draft-ietf-tls-ciphersuite-03.txt.
922 [Ben Laurie, Steve Henson]
924 *) In the NCONF_...-based implementations for CONF_... queries
925 (crypto/conf/conf_lib.c), if the input LHASH is NULL, avoid using
926 a temporary CONF structure with the data component set to NULL
927 (which gives segmentation faults in lh_retrieve).
928 Instead, use NULL for the CONF pointer in CONF_get_string and
929 CONF_get_number (which may use environment variables) and directly
930 return NULL from CONF_get_section.
933 *) Fix potential buffer overrun for EBCDIC.
936 +) New function OCSP_copy_nonce() to copy nonce value (if present) from
940 +) Functions for OCSP responders. OCSP_request_onereq_count(),
941 OCSP_request_onereq_get0(), OCSP_onereq_get0_id() and OCSP_id_get0_info()
942 extract information from a certificate request. OCSP_response_create()
943 creates a response and optionally adds a basic response structure.
944 OCSP_basic_add1_status() adds a complete single response to a basic
945 reponse and returns the OCSP_SINGLERESP structure just added (to allow
946 extensions to be included for example). OCSP_basic_add1_cert() adds a
947 certificate to a basic response and OCSP_basic_sign() signs a basic
948 response with various flags. New helper functions ASN1_TIME_check()
949 (checks validity of ASN1_TIME structure) and ASN1_TIME_to_generalizedtime()
950 (converts ASN1_TIME to GeneralizedTime).
953 +) Various new functions. EVP_Digest() combines EVP_Digest{Init,Update,Final}()
954 in a single operation. X509_get0_pubkey_bitstr() extracts the public_key
955 structure from a certificate. X509_pubkey_digest() digests tha public_key
956 contents: this is used in various key identifiers.
959 *) Tolerate nonRepudiation as being valid for S/MIME signing and certSign
960 keyUsage if basicConstraints absent for a CA.
963 *) Make SMIME_write_PKCS7() write mail header values with a format that
964 is more generally accepted (no spaces before the semicolon), since
965 some programs can't parse those values properly otherwise. Also make
966 sure BIO's that break lines after each write do not create invalid
970 +) Make sk_sort() tolerate a NULL argument.
971 [Steve Henson reported by Massimiliano Pala <madwolf@comune.modena.it>]
973 +) New OCSP verify flag OCSP_TRUSTOTHER. When set the "other" certificates
974 passed by the function are trusted implicitly. If any of them signed the
975 reponse then it is assumed to be valid and is not verified.
978 -) Make the CRL encoding routines work with empty SEQUENCE OF. The
979 macros previously used would not encode an empty SEQUENCE OF
980 and break the signature.
982 [This change does not apply to 0.9.7.]
984 *) Zero the premaster secret after deriving the master secret in
988 +) In PKCS7_set_type() initialise content_type in PKCS7_ENC_CONTENT
989 to data. This was previously part of the PKCS7 ASN1 code. This
990 was causing problems with OpenSSL created PKCS#12 and PKCS#7 structures.
991 [Steve Henson, reported by Kenneth R. Robinette
992 <support@securenetterm.com>]
994 +) Add CRYPTO_push_info() and CRYPTO_pop_info() calls to new ASN1
995 routines: without these tracing memory leaks is very painful.
996 Fix leaks in PKCS12 and PKCS7 routines.
999 *) Add some EVP_add_digest_alias registrations (as found in
1000 OpenSSL_add_all_digests()) to SSL_library_init()
1001 aka OpenSSL_add_ssl_algorithms(). This provides improved
1002 compatibility with peers using X.509 certificates
1003 with unconventional AlgorithmIdentifier OIDs.
1006 *) Fix for Irix with NO_ASM.
1007 ["Bruce W. Forsberg" <bruce.forsberg@baesystems.com>]
1009 *) ./config script fixes.
1010 [Ulf Moeller, Richard Levitte]
1012 +) Make X509_time_adj() cope with the new behaviour of ASN1_TIME_new().
1013 Previously it initialised the 'type' argument to V_ASN1_UTCTIME which
1014 effectively meant GeneralizedTime would never be used. Now it
1015 is initialised to -1 but X509_time_adj() now has to check the value
1016 and use ASN1_TIME_set() if the value is not V_ASN1_UTCTIME or
1017 V_ASN1_GENERALIZEDTIME, without this it always uses GeneralizedTime.
1018 [Steve Henson, reported by Kenneth R. Robinette
1019 <support@securenetterm.com>]
1021 +) Fixes to BN_to_ASN1_INTEGER when bn is zero. This would previously
1022 result in a zero length in the ASN1_INTEGER structure which was
1023 not consistent with the structure when d2i_ASN1_INTEGER() was used
1024 and would cause ASN1_INTEGER_cmp() to fail. Enhance s2i_ASN1_INTEGER()
1025 to cope with hex and negative integers. Fix bug in i2a_ASN1_INTEGER()
1026 where it did not print out a minus for negative ASN1_INTEGER.
1029 *) Fix 'openssl passwd -1'.
1032 +) Add summary printout to ocsp utility. The various functions which
1033 convert status values to strings have been renamed to:
1034 OCSP_response_status_str(), OCSP_cert_status_str() and
1035 OCSP_crl_reason_str() and are no longer static. New options
1036 to verify nonce values and to disable verification. OCSP response
1037 printout format cleaned up.
1040 +) Add additional OCSP certificate checks. These are those specified
1041 in RFC2560. This consists of two separate checks: the CA of the
1042 certificate being checked must either be the OCSP signer certificate
1043 or the issuer of the OCSP signer certificate. In the latter case the
1044 OCSP signer certificate must contain the OCSP signing extended key
1045 usage. This check is performed by attempting to match the OCSP
1046 signer or the OCSP signer CA to the issuerNameHash and issuerKeyHash
1047 in the OCSP_CERTID structures of the response.
1050 +) Initial OCSP certificate verification added to OCSP_basic_verify()
1051 and related routines. This uses the standard OpenSSL certificate
1052 verify routines to perform initial checks (just CA validity) and
1053 to obtain the certificate chain. Then additional checks will be
1054 performed on the chain. Currently the root CA is checked to see
1055 if it is explicitly trusted for OCSP signing. This is used to set
1056 a root CA as a global signing root: that is any certificate that
1057 chains to that CA is an acceptable OCSP signing certificate.
1060 +) New '-extfile ...' option to 'openssl ca' for reading X.509v3
1061 extensions from a separate configuration file.
1062 As when reading extensions from the main configuration file,
1063 the '-extensions ...' option may be used for specifying the
1065 [Massimiliano Pala <madwolf@comune.modena.it>]
1067 *) Change PKCS12_key_gen_asc() so it can cope with non null
1068 terminated strings whose length is passed in the passlen
1069 parameter, for example from PEM callbacks. This was done
1070 by adding an extra length parameter to asc2uni().
1071 [Steve Henson, reported by <oddissey@samsung.co.kr>]
1073 +) New OCSP utility. Allows OCSP requests to be generated or
1074 read. The request can be sent to a responder and the output
1075 parsed, outputed or printed in text form. Not complete yet:
1076 still needs to check the OCSP response validity.
1079 +) New subcommands for 'openssl ca':
1080 'openssl ca -status <serial>' prints the status of the cert with
1081 the given serial number (according to the index file).
1082 'openssl ca -updatedb' updates the expiry status of certificates
1084 [Massimiliano Pala <madwolf@comune.modena.it>]
1086 +) New '-newreq-nodes' command option to CA.pl. This is like
1087 '-newreq', but calls 'openssl req' with the '-nodes' option
1088 so that the resulting key is not encrypted.
1089 [Damien Miller <djm@mindrot.org>]
1091 +) New configuration for the GNU Hurd.
1092 [Jonathan Bartlett <johnnyb@wolfram.com> via Richard Levitte]
1094 +) Initial code to implement OCSP basic response verify. This
1095 is currently incomplete. Currently just finds the signer's
1096 certificate and verifies the signature on the response.
1099 +) New SSLeay_version code SSLEAY_DIR to determine the compiled-in
1100 value of OPENSSLDIR. This is available via the new '-d' option
1101 to 'openssl version', and is also included in 'openssl version -a'.
1104 *) Fix C code generated by 'openssl dsaparam -C': If a BN_bin2bn
1105 call failed, free the DSA structure.
1108 +) Allowing defining memory allocation callbacks that will be given
1109 file name and line number information in additional arguments
1110 (a const char* and an int). The basic functionality remains, as
1111 well as the original possibility to just replace malloc(),
1112 realloc() and free() by functions that do not know about these
1113 additional arguments. To register and find out the current
1114 settings for extended allocation functions, the following
1115 functions are provided:
1117 CRYPTO_set_mem_ex_functions
1118 CRYPTO_set_locked_mem_ex_functions
1119 CRYPTO_get_mem_ex_functions
1120 CRYPTO_get_locked_mem_ex_functions
1122 These work the same way as CRYPTO_set_mem_functions and friends.
1123 CRYPTO_get_[locked_]mem_functions now writes 0 where such an
1124 extended allocation function is enabled.
1125 Similarly, CRYPTO_get_[locked_]mem_ex_functions writes 0 where
1126 a conventional allocation function is enabled.
1127 [Richard Levitte, Bodo Moeller]
1129 *) Fix to uni2asc() to cope with zero length Unicode strings.
1130 These are present in some PKCS#12 files.
1133 +) Finish off removing the remaining LHASH function pointer casts.
1134 There should no longer be any prototype-casting required when using
1135 the LHASH abstraction, and any casts that remain are "bugs". See
1136 the callback types and macros at the head of lhash.h for details
1137 (and "OBJ_cleanup" in crypto/objects/obj_dat.c as an example).
1140 +) Add automatic query of EGD sockets in RAND_poll() for the unix variant.
1141 If /dev/[u]random devices are not available or do not return enough
1142 entropy, EGD style sockets (served by EGD or PRNGD) will automatically
1144 The locations /var/run/egd-pool, /dev/egd-pool, /etc/egd-pool, and
1145 /etc/entropy will be queried once each in this sequence, quering stops
1146 when enough entropy was collected without querying more sockets.
1149 +) Change the Unix RAND_poll() variant to be able to poll several
1150 random devices, as specified by DEVRANDOM, until a sufficient amount
1151 of data has been collected. We spend at most 10 ms on each file
1152 (select timeout) and read in non-blocking mode. DEVRANDOM now
1153 defaults to the list "/dev/urandom", "/dev/random", "/dev/srandom"
1154 (previously it was just the string "/dev/urandom"), so on typical
1155 platforms the 10 ms delay will never occur.
1156 Also separate out the Unix variant to its own file, rand_unix.c.
1157 For VMS, there's a currently-empty rand_vms.c.
1160 +) Move OCSP client related routines to ocsp_cl.c. These
1161 provide utility functions which an application needing
1162 to issue a request to an OCSP responder and analyse the
1163 response will typically need: as opposed to those which an
1164 OCSP responder itself would need which will be added later.
1166 OCSP_request_sign() signs an OCSP request with an API similar
1167 to PKCS7_sign(). OCSP_response_status() returns status of OCSP
1168 response. OCSP_response_get1_basic() extracts basic response
1169 from response. OCSP_resp_find_status(): finds and extracts status
1170 information from an OCSP_CERTID structure (which will be created
1171 when the request structure is built). These are built from lower
1172 level functions which work on OCSP_SINGLERESP structures but
1173 wont normally be used unless the application wishes to examine
1174 extensions in the OCSP response for example.
1176 Replace nonce routines with a pair of functions.
1177 OCSP_request_add1_nonce() adds a nonce value and optionally
1178 generates a random value. OCSP_check_nonce() checks the
1179 validity of the nonce in an OCSP response.
1182 +) Change function OCSP_request_add() to OCSP_request_add0_id().
1183 This doesn't copy the supplied OCSP_CERTID and avoids the
1184 need to free up the newly created id. Change return type
1185 to OCSP_ONEREQ to return the internal OCSP_ONEREQ structure.
1186 This can then be used to add extensions to the request.
1187 Deleted OCSP_request_new(), since most of its functionality
1188 is now in OCSP_REQUEST_new() (and the case insensitive name
1189 clash) apart from the ability to set the request name which
1190 will be added elsewhere.
1193 +) Update OCSP API. Remove obsolete extensions argument from
1194 various functions. Extensions are now handled using the new
1195 OCSP extension code. New simple OCSP HTTP function which
1196 can be used to send requests and parse the response.
1199 +) Fix the PKCS#7 (S/MIME) code to work with new ASN1. Two new
1200 ASN1_ITEM structures help with sign and verify. PKCS7_ATTR_SIGN
1201 uses the special reorder version of SET OF to sort the attributes
1202 and reorder them to match the encoded order. This resolves a long
1203 standing problem: a verify on a PKCS7 structure just after signing
1204 it used to fail because the attribute order did not match the
1205 encoded order. PKCS7_ATTR_VERIFY does not reorder the attributes:
1206 it uses the received order. This is necessary to tolerate some broken
1207 software that does not order SET OF. This is handled by encoding
1208 as a SEQUENCE OF but using implicit tagging (with UNIVERSAL class)
1209 to produce the required SET OF.
1212 +) Have mk1mf.pl generate the macros OPENSSL_BUILD_SHLIBCRYPTO and
1213 OPENSSL_BUILD_SHLIBSSL and use them appropriately in the header
1214 files to get correct declarations of the ASN.1 item variables.
1217 +) Rewrite of PKCS#12 code to use new ASN1 functionality. Replace many
1218 PKCS#12 macros with real functions. Fix two unrelated ASN1 bugs:
1219 asn1_check_tlen() would sometimes attempt to use 'ctx' when it was
1220 NULL and ASN1_TYPE was not dereferenced properly in asn1_ex_c2i().
1221 New ASN1 macro: DECLARE_ASN1_ITEM() which just declares the relevant
1222 ASN1_ITEM and no wrapper functions.
1225 +) New functions or ASN1_item_d2i_fp() and ASN1_item_d2i_bio(). These
1226 replace the old function pointer based I/O routines. Change most of
1227 the *_d2i_bio() and *_d2i_fp() functions to use these.
1230 +) Enhance mkdef.pl to be more accepting about spacing in C preprocessor
1231 lines, recognice more "algorithms" that can be deselected, and make
1232 it complain about algorithm deselection that isn't recognised.
1235 +) New ASN1 functions to handle dup, sign, verify, digest, pack and
1236 unpack operations in terms of ASN1_ITEM. Modify existing wrappers
1237 to use new functions. Add NO_ASN1_OLD which can be set to remove
1238 some old style ASN1 functions: this can be used to determine if old
1239 code will still work when these eventually go away.
1242 +) New extension functions for OCSP structures, these follow the
1243 same conventions as certificates and CRLs.
1246 +) New function X509V3_add1_i2d(). This automatically encodes and
1247 adds an extension. Its behaviour can be customised with various
1248 flags to append, replace or delete. Various wrappers added for
1249 certifcates and CRLs.
1252 +) Fix to avoid calling the underlying ASN1 print routine when
1253 an extension cannot be parsed. Correct a typo in the
1254 OCSP_SERVICELOC extension. Tidy up print OCSP format.
1257 *) Increase s2->wbuf allocation by one byte in ssl2_new (ssl/s2_lib.c).
1258 Otherwise do_ssl_write (ssl/s2_pkt.c) will write beyond buffer limits
1259 when writing a 32767 byte record.
1260 [Bodo Moeller; problem reported by Eric Day <eday@concentric.net>]
1262 *) In RSA_eay_public_{en,ed}crypt and RSA_eay_mod_exp (rsa_eay.c),
1263 obtain lock CRYPTO_LOCK_RSA before setting rsa->_method_mod_{n,p,q}.
1265 (RSA objects have a reference count access to which is protected
1266 by CRYPTO_LOCK_RSA [see rsa_lib.c, s3_srvr.c, ssl_cert.c, ssl_rsa.c],
1267 so they are meant to be shared between threads.)
1268 [Bodo Moeller, Geoff Thorpe; original patch submitted by
1269 "Reddie, Steven" <Steven.Reddie@ca.com>]
1271 +) Make mkdef.pl parse some of the ASN1 macros and add apropriate
1272 entries for variables.
1275 *) Fix a deadlock in CRYPTO_mem_leaks().
1278 +) Add functionality to apps/openssl.c for detecting locking
1279 problems: As the program is single-threaded, all we have
1280 to do is register a locking callback using an array for
1281 storing which locks are currently held by the program.
1284 +) Use a lock around the call to CRYPTO_get_ex_new_index() in
1285 SSL_get_ex_data_X509_STORE_idx(), which is used in
1286 ssl_verify_cert_chain() and thus can be called at any time
1287 during TLS/SSL handshakes so that thread-safety is essential.
1288 Unfortunately, the ex_data design is not at all suited
1289 for multi-threaded use, so it probably should be abolished.
1292 +) Added Broadcom "ubsec" ENGINE to OpenSSL.
1293 [Broadcom, tweaked and integrated by Geoff Thorpe]
1295 +) Move common extension printing code to new function
1296 X509V3_print_extensions(). Reorganise OCSP print routines and
1297 implement some needed OCSP ASN1 functions. Add OCSP extensions.
1300 +) New function X509_signature_print() to remove duplication in some
1304 +) Add a special meaning when SET OF and SEQUENCE OF flags are both
1305 set (this was treated exactly the same as SET OF previously). This
1306 is used to reorder the STACK representing the structure to match the
1307 encoding. This will be used to get round a problem where a PKCS7
1308 structure which was signed could not be verified because the STACK
1309 order did not reflect the encoded order.
1312 +) Reimplement the OCSP ASN1 module using the new code.
1315 +) Update the X509V3 code to permit the use of an ASN1_ITEM structure
1316 for its ASN1 operations. The old style function pointers still exist
1317 for now but they will eventually go away.
1320 +) Merge in replacement ASN1 code from the ASN1 branch. This almost
1321 completely replaces the old ASN1 functionality with a table driven
1322 encoder and decoder which interprets an ASN1_ITEM structure describing
1323 the ASN1 module. Compatibility with the existing ASN1 API (i2d,d2i) is
1324 largely maintained. Almost all of the old asn1_mac.h macro based ASN1
1325 has also been converted to the new form.
1328 +) Change BN_mod_exp_recp so that negative moduli are tolerated
1329 (the sign is ignored). Similarly, ignore the sign in BN_MONT_CTX_set
1330 so that BN_mod_exp_mont and BN_mod_exp_mont_word work
1331 for negative moduli.
1334 +) Fix BN_uadd and BN_usub: Always return non-negative results instead
1335 of not touching the result's sign bit.
1338 +) BN_div bugfix: If the result is 0, the sign (res->neg) must not be
1342 +) Changed the LHASH code to use prototypes for callbacks, and created
1343 macros to declare and implement thin (optionally static) functions
1344 that provide type-safety and avoid function pointer casting for the
1345 type-specific callbacks.
1348 *) Use better test patterns in bntest.
1351 +) Added Kerberos Cipher Suites to be used with TLS, as written in
1353 [Veers Staats <staatsvr@asc.hpc.mil>,
1354 Jeffrey Altman <jaltman@columbia.edu>, via Richard Levitte]
1356 *) rand_win.c fix for Borland C.
1359 *) BN_rshift bugfix for n == 0.
1362 +) Reformat the FAQ so the different questions and answers can be divided
1363 in sections depending on the subject.
1366 +) Have the zlib compression code load ZLIB.DLL dynamically under
1370 +) New function BN_mod_sqrt for computing square roots modulo a prime
1371 (using the probabilistic Tonelli-Shanks algorithm unless
1372 p == 3 (mod 4) or p == 5 (mod 8), which are cases that can
1373 be handled deterministically).
1374 [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller]
1376 *) Add a 'bctest' script that checks for some known 'bc' bugs
1377 so that 'make test' does not abort just because 'bc' is broken.
1380 *) Store verify_result within SSL_SESSION also for client side to
1381 avoid potential security hole. (Re-used sessions on the client side
1382 always resulted in verify_result==X509_V_OK, not using the original
1383 result of the server certificate verification.)
1386 +) Make BN_mod_inverse faster by explicitly handling small quotients
1387 in the Euclid loop. (Speed gain about 20% for small moduli [256 or
1388 512 bits], about 30% for larger ones [1024 or 2048 bits].)
1391 *) Fix ssl3_pending: If the record in s->s3->rrec is not of type
1392 SSL3_RT_APPLICATION_DATA, return 0.
1393 Similarly, change ssl2_pending to return 0 if SSL_in_init(s) is true.
1397 Both ssl2_peek and ssl3_peek, which were totally broken in earlier
1398 releases, have been re-implemented by renaming the previous
1399 implementations of ssl2_read and ssl3_read to ssl2_read_internal
1400 and ssl3_read_internal, respectively, and adding 'peek' parameters
1401 to them. The new ssl[23]_{read,peek} functions are calls to
1402 ssl[23]_read_internal with the 'peek' flag set appropriately.
1403 A 'peek' parameter has also been added to ssl3_read_bytes, which
1404 does the actual work for ssl3_read_internal.
1407 +) New function BN_kronecker.
1410 +) Fix BN_gcd so that it works on negative inputs; the result is
1411 positive unless both parameters are zero.
1412 Previously something reasonably close to an infinite loop was
1413 possible because numbers could be growing instead of shrinking
1414 in the implementation of Euclid's algorithm.
1417 +) Fix BN_is_word() and BN_is_one() macros to take into account the
1418 sign of the number in question.
1420 Fix BN_is_word(a,w) to work correctly for w == 0.
1422 The old BN_is_word(a,w) macro is now called BN_abs_is_word(a,w)
1423 because its test if the absolute value of 'a' equals 'w'.
1424 Note that BN_abs_is_word does *not* handle w == 0 reliably;
1425 it exists mostly for use in the implementations of BN_is_zero(),
1426 BN_is_one(), and BN_is_word().
1429 *) Initialise "ex_data" member of RSA/DSA/DH structures prior to calling
1430 the method-specific "init()" handler. Also clean up ex_data after
1431 calling the method-specific "finish()" handler. Previously, this was
1432 happening the other way round.
1435 +) New function BN_swap.
1438 +) Use BN_nnmod instead of BN_mod in crypto/bn/bn_exp.c so that
1439 the exponentiation functions are more likely to produce reasonable
1440 results on negative inputs.
1443 +) Change BN_mod_mul so that the result is always non-negative.
1444 Previously, it could be negative if one of the factors was negative;
1445 I don't think anyone really wanted that behaviour.
1448 +) Move BN_mod_... functions into new file crypto/bn/bn_mod.c
1449 (except for exponentiation, which stays in crypto/bn/bn_exp.c,
1450 and BN_mod_mul_reciprocal, which stays in crypto/bn/bn_recp.c)
1451 and add new functions:
1460 BN_mod_lshift1_quick
1464 These functions always generate non-negative results.
1466 BN_nnmod otherwise is like BN_mod (if BN_mod computes a remainder r
1467 such that |m| < r < 0, BN_nnmod will output rem + |m| instead).
1469 BN_mod_XXX_quick(r, a, [b,] m) generates the same result as
1470 BN_mod_XXX(r, a, [b,] m, ctx), but requires that a [and b]
1471 be reduced modulo m.
1472 [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller]
1474 +) Remove a few calls to bn_wexpand() in BN_sqr() (the one in there
1475 was actually never needed) and in BN_mul(). The removal in BN_mul()
1476 required a small change in bn_mul_part_recursive() and the addition
1477 of the functions bn_cmp_part_words(), bn_sub_part_words() and
1478 bn_add_part_words(), which do the same thing as bn_cmp_words(),
1479 bn_sub_words() and bn_add_words() except they take arrays with
1483 +) In 'openssl passwd', verify passwords read from the terminal
1484 unless the '-salt' option is used (which usually means that
1485 verification would just waste user's time since the resulting
1486 hash is going to be compared with some given password hash)
1487 or the new '-noverify' option is used.
1489 This is an incompatible change, but it does not affect
1490 non-interactive use of 'openssl passwd' (passwords on the command
1491 line, '-stdin' option, '-in ...' option) and thus should not
1495 +) Remove all references to RSAref, since there's no more need for it.
1498 *) Increase BN_CTX_NUM (the number of BIGNUMs in a BN_CTX) to 16.
1499 The previous value, 12, was not always sufficient for BN_mod_exp().
1502 +) Make DSO load along a path given through an environment variable
1503 (SHLIB_PATH) with shl_load().
1506 +) Constify the ENGINE code as a result of BIGNUM constification.
1507 Also constify the RSA code and most things related to it. In a
1508 few places, most notable in the depth of the ASN.1 code, ugly
1509 casts back to non-const were required (to be solved at a later
1513 +) Make it so the openssl application has all engines loaded by default.
1516 +) Constify the BIGNUM routines a little more.
1519 *) Make sure that shared libraries get the internal name engine with
1520 the full version number and not just 0. This should mark the
1521 shared libraries as not backward compatible. Of course, this should
1522 be changed again when we can guarantee backward binary compatibility.
1525 +) Add the following functions:
1527 ENGINE_load_cswift()
1529 ENGINE_load_atalla()
1531 ENGINE_load_builtin_engines()
1533 That way, an application can itself choose if external engines that
1534 are built-in in OpenSSL shall ever be used or not. The benefit is
1535 that applications won't have to be linked with libdl or other dso
1536 libraries unless it's really needed.
1538 Changed 'openssl engine' to load all engines on demand.
1539 Changed the engine header files to avoid the duplication of some
1540 declarations (they differed!).
1543 +) 'openssl engine' can now list capabilities.
1546 +) Better error reporting in 'openssl engine'.
1549 +) Never call load_dh_param(NULL) in s_server.
1552 +) Add engine application. It can currently list engines by name and
1553 identity, and test if they are actually available.
1556 +) Add support for shared libraries under Irix.
1557 [Albert Chin-A-Young <china@thewrittenword.com>]
1559 +) Improve RPM specification file by forcing symbolic linking and making
1560 sure the installed documentation is also owned by root.root.
1561 [Damien Miller <djm@mindrot.org>]
1563 +) Add configuration option to build on Linux on both big-endian and
1565 [Ralf Baechle <ralf@uni-koblenz.de>]
1567 +) Give the OpenSSL applications more possibilities to make use of
1568 keys (public as well as private) handled by engines.
1571 +) Add OCSP code that comes from CertCo.
1574 +) Add VMS support for the Rijndael code.
1577 +) Added untested support for Nuron crypto accelerator.
1580 +) Add support for external cryptographic devices. This code was
1581 previously distributed separately as the "engine" branch.
1582 [Geoff Thorpe, Richard Levitte]
1584 +) Rework the filename-translation in the DSO code. It is now possible to
1585 have far greater control over how a "name" is turned into a filename
1586 depending on the operating environment and any oddities about the
1587 different shared library filenames on each system.
1590 +) Support threads on FreeBSD-elf in Configure.
1593 +) Add the possibility to create shared libraries on HP-UX
1596 +) Fix for SHA1 assembly problem with MASM: it produces
1597 warnings about corrupt line number information when assembling
1598 with debugging information. This is caused by the overlapping
1600 [Bernd Matthes <mainbug@celocom.de>, Steve Henson]
1603 NCONF_get_number() has no error checking at all. As a replacement,
1604 NCONF_get_number_e() is defined (_e for "error checking") and is
1605 promoted strongly. The old NCONF_get_number is kept around for
1606 binary backward compatibility.
1607 Make it possible for methods to load from something other than a BIO,
1608 by providing a function pointer that is given a name instead of a BIO.
1609 For example, this could be used to load configuration data from an
1613 *) Fix typo in get_cert_by_subject() in by_dir.c
1614 [Jean-Marc Desperrier <jean-marc.desperrier@certplus.com>]
1616 *) Rework the system to generate shared libraries:
1618 - Make note of the expected extension for the shared libraries and
1619 if there is a need for symbolic links from for example libcrypto.so.0
1620 to libcrypto.so.0.9.7. There is extended info in Configure for
1623 - Make as few rebuilds of the shared libraries as possible.
1625 - Still avoid linking the OpenSSL programs with the shared libraries.
1627 - When installing, install the shared libraries separately from the
1631 +) Fix for non blocking accept BIOs. Added new I/O special reason
1632 BIO_RR_ACCEPT to cover this case. Previously use of accept BIOs
1633 with non blocking I/O was not possible because no retry code was
1634 implemented. Also added new SSL code SSL_WANT_ACCEPT to cover
1638 +) Added the beginnings of Rijndael support.
1641 +) Fix for bug in DirectoryString mask setting. Add support for
1642 X509_NAME_print_ex() in 'req' and X509_print_ex() function
1643 to allow certificate printing to more controllable, additional
1644 'certopt' option to 'x509' to allow new printing options to be
1648 +) Clean old EAY MD5 hack from e_os.h.
1651 *) Fix SSL_CTX_set_read_ahead macro to actually use its argument.
1653 Copy SSL_CTX's read_ahead flag to SSL object directly in SSL_new
1654 and not in SSL_clear because the latter is also used by the
1655 accept/connect functions; previously, the settings made by
1656 SSL_set_read_ahead would be lost during the handshake.
1657 [Bodo Moeller; problems reported by Anders Gertz <gertz@epact.se>]
1659 *) Correct util/mkdef.pl to be selective about disabled algorithms.
1660 Previously, it would create entries for disableed algorithms no
1664 *) Added several new manual pages for SSL_* function.
1667 Changes between 0.9.5a and 0.9.6 [24 Sep 2000]
1669 *) In ssl23_get_client_hello, generate an error message when faced
1670 with an initial SSL 3.0/TLS record that is too small to contain the
1671 first two bytes of the ClientHello message, i.e. client_version.
1672 (Note that this is a pathologic case that probably has never happened
1673 in real life.) The previous approach was to use the version number
1674 from the record header as a substitute; but our protocol choice
1675 should not depend on that one because it is not authenticated
1676 by the Finished messages.
1679 *) More robust randomness gathering functions for Windows.
1680 [Jeffrey Altman <jaltman@columbia.edu>]
1682 *) For compatibility reasons if the flag X509_V_FLAG_ISSUER_CHECK is
1683 not set then we don't setup the error code for issuer check errors
1684 to avoid possibly overwriting other errors which the callback does
1685 handle. If an application does set the flag then we assume it knows
1686 what it is doing and can handle the new informational codes
1690 *) Fix for a nasty bug in ASN1_TYPE handling. ASN1_TYPE is used for
1691 a general "ANY" type, as such it should be able to decode anything
1692 including tagged types. However it didn't check the class so it would
1693 wrongly interpret tagged types in the same way as their universal
1694 counterpart and unknown types were just rejected. Changed so that the
1695 tagged and unknown types are handled in the same way as a SEQUENCE:
1696 that is the encoding is stored intact. There is also a new type
1697 "V_ASN1_OTHER" which is used when the class is not universal, in this
1698 case we have no idea what the actual type is so we just lump them all
1702 *) On VMS, stdout may very well lead to a file that is written to
1703 in a record-oriented fashion. That means that every write() will
1704 write a separate record, which will be read separately by the
1705 programs trying to read from it. This can be very confusing.
1707 The solution is to put a BIO filter in the way that will buffer
1708 text until a linefeed is reached, and then write everything a
1709 line at a time, so every record written will be an actual line,
1710 not chunks of lines and not (usually doesn't happen, but I've
1711 seen it once) several lines in one record. BIO_f_linebuffer() is
1714 Currently, it's a VMS-only method, because that's where it has
1715 been tested well enough.
1718 *) Remove 'optimized' squaring variant in BN_mod_mul_montgomery,
1719 it can return incorrect results.
1720 (Note: The buggy variant was not enabled in OpenSSL 0.9.5a,
1721 but it was in 0.9.6-beta[12].)
1724 *) Disable the check for content being present when verifying detached
1725 signatures in pk7_smime.c. Some versions of Netscape (wrongly)
1726 include zero length content when signing messages.
1729 *) New BIO_shutdown_wr macro, which invokes the BIO_C_SHUTDOWN_WR
1730 BIO_ctrl (for BIO pairs).
1733 *) Add DSO method for VMS.
1736 *) Bug fix: Montgomery multiplication could produce results with the
1740 *) Add RPM specification openssl.spec and modify it to build three
1741 packages. The default package contains applications, application
1742 documentation and run-time libraries. The devel package contains
1743 include files, static libraries and function documentation. The
1744 doc package contains the contents of the doc directory. The original
1745 openssl.spec was provided by Damien Miller <djm@mindrot.org>.
1748 *) Add a large number of documentation files for many SSL routines.
1749 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>]
1751 *) Add a configuration entry for Sony News 4.
1752 [NAKAJI Hiroyuki <nakaji@tutrp.tut.ac.jp>]
1754 *) Don't set the two most significant bits to one when generating a
1755 random number < q in the DSA library.
1758 *) New SSL API mode 'SSL_MODE_AUTO_RETRY'. This disables the default
1759 behaviour that SSL_read may result in SSL_ERROR_WANT_READ (even if
1760 the underlying transport is blocking) if a handshake took place.
1761 (The default behaviour is needed by applications such as s_client
1762 and s_server that use select() to determine when to use SSL_read;
1763 but for applications that know in advance when to expect data, it
1764 just makes things more complicated.)
1767 *) Add RAND_egd_bytes(), which gives control over the number of bytes read
1771 *) Add a few more EBCDIC conditionals that make `req' and `x509'
1772 work better on such systems.
1773 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>]
1775 *) Add two demo programs for PKCS12_parse() and PKCS12_create().
1776 Update PKCS12_parse() so it copies the friendlyName and the
1777 keyid to the certificates aux info.
1780 *) Fix bug in PKCS7_verify() which caused an infinite loop
1781 if there was more than one signature.
1782 [Sven Uszpelkat <su@celocom.de>]
1784 *) Major change in util/mkdef.pl to include extra information
1785 about each symbol, as well as presentig variables as well
1786 as functions. This change means that there's n more need
1787 to rebuild the .num files when some algorithms are excluded.
1790 *) Allow the verify time to be set by an application,
1791 rather than always using the current time.
1794 *) Phase 2 verify code reorganisation. The certificate
1795 verify code now looks up an issuer certificate by a
1796 number of criteria: subject name, authority key id
1797 and key usage. It also verifies self signed certificates
1798 by the same criteria. The main comparison function is
1799 X509_check_issued() which performs these checks.
1801 Lot of changes were necessary in order to support this
1802 without completely rewriting the lookup code.
1804 Authority and subject key identifier are now cached.
1806 The LHASH 'certs' is X509_STORE has now been replaced
1807 by a STACK_OF(X509_OBJECT). This is mainly because an
1808 LHASH can't store or retrieve multiple objects with
1809 the same hash value.
1811 As a result various functions (which were all internal
1812 use only) have changed to handle the new X509_STORE
1813 structure. This will break anything that messed round
1814 with X509_STORE internally.
1816 The functions X509_STORE_add_cert() now checks for an
1817 exact match, rather than just subject name.
1819 The X509_STORE API doesn't directly support the retrieval
1820 of multiple certificates matching a given criteria, however
1821 this can be worked round by performing a lookup first
1822 (which will fill the cache with candidate certificates)
1823 and then examining the cache for matches. This is probably
1824 the best we can do without throwing out X509_LOOKUP
1825 entirely (maybe later...).
1827 The X509_VERIFY_CTX structure has been enhanced considerably.
1829 All certificate lookup operations now go via a get_issuer()
1830 callback. Although this currently uses an X509_STORE it
1831 can be replaced by custom lookups. This is a simple way
1832 to bypass the X509_STORE hackery necessary to make this
1833 work and makes it possible to use more efficient techniques
1834 in future. A very simple version which uses a simple
1835 STACK for its trusted certificate store is also provided
1836 using X509_STORE_CTX_trusted_stack().
1838 The verify_cb() and verify() callbacks now have equivalents
1839 in the X509_STORE_CTX structure.
1841 X509_STORE_CTX also has a 'flags' field which can be used
1842 to customise the verify behaviour.
1845 *) Add new PKCS#7 signing option PKCS7_NOSMIMECAP which
1846 excludes S/MIME capabilities.
1849 *) When a certificate request is read in keep a copy of the
1850 original encoding of the signed data and use it when outputing
1851 again. Signatures then use the original encoding rather than
1852 a decoded, encoded version which may cause problems if the
1853 request is improperly encoded.
1856 *) For consistency with other BIO_puts implementations, call
1857 buffer_write(b, ...) directly in buffer_puts instead of calling
1860 In BIO_puts, increment b->num_write as in BIO_write.
1861 [Peter.Sylvester@EdelWeb.fr]
1863 *) Fix BN_mul_word for the case where the word is 0. (We have to use
1864 BN_zero, we may not return a BIGNUM with an array consisting of
1868 *) Avoid calling abort() from within the library when problems are
1869 detected, except if preprocessor symbols have been defined
1870 (such as REF_CHECK, BN_DEBUG etc.).
1873 *) New openssl application 'rsautl'. This utility can be
1874 used for low level RSA operations. DER public key
1875 BIO/fp routines also added.
1878 *) New Configure entry and patches for compiling on QNX 4.
1879 [Andreas Schneider <andreas@ds3.etech.fh-hamburg.de>]
1881 *) A demo state-machine implementation was sponsored by
1882 Nuron (http://www.nuron.com/) and is now available in
1883 demos/state_machine.
1886 *) New options added to the 'dgst' utility for signature
1887 generation and verification.
1890 *) Unrecognized PKCS#7 content types are now handled via a
1891 catch all ASN1_TYPE structure. This allows unsupported
1892 types to be stored as a "blob" and an application can
1893 encode and decode it manually.
1896 *) Fix various signed/unsigned issues to make a_strex.c
1898 [Oscar Jacobsson <oscar.jacobsson@celocom.com>]
1900 *) ASN1 fixes. i2d_ASN1_OBJECT was not returning the correct
1901 length if passed a buffer. ASN1_INTEGER_to_BN failed
1902 if passed a NULL BN and its argument was negative.
1903 [Steve Henson, pointed out by Sven Heiberg <sven@tartu.cyber.ee>]
1905 *) Modification to PKCS#7 encoding routines to output definite
1906 length encoding. Since currently the whole structures are in
1907 memory there's not real point in using indefinite length
1908 constructed encoding. However if OpenSSL is compiled with
1909 the flag PKCS7_INDEFINITE_ENCODING the old form is used.
1912 *) Added BIO_vprintf() and BIO_vsnprintf().
1915 *) Added more prefixes to parse for in the the strings written
1916 through a logging bio, to cover all the levels that are available
1917 through syslog. The prefixes are now:
1919 PANIC, EMERG, EMR => LOG_EMERG
1920 ALERT, ALR => LOG_ALERT
1921 CRIT, CRI => LOG_CRIT
1922 ERROR, ERR => LOG_ERR
1923 WARNING, WARN, WAR => LOG_WARNING
1924 NOTICE, NOTE, NOT => LOG_NOTICE
1925 INFO, INF => LOG_INFO
1926 DEBUG, DBG => LOG_DEBUG
1928 and as before, if none of those prefixes are present at the
1929 beginning of the string, LOG_ERR is chosen.
1931 On Win32, the LOG_* levels are mapped according to this:
1933 LOG_EMERG, LOG_ALERT, LOG_CRIT, LOG_ERR => EVENTLOG_ERROR_TYPE
1934 LOG_WARNING => EVENTLOG_WARNING_TYPE
1935 LOG_NOTICE, LOG_INFO, LOG_DEBUG => EVENTLOG_INFORMATION_TYPE
1939 *) Made it possible to reconfigure with just the configuration
1940 argument "reconf" or "reconfigure". The command line arguments
1941 are stored in Makefile.ssl in the variable CONFIGURE_ARGS,
1942 and are retrieved from there when reconfiguring.
1946 [Assar Westerlund <assar@sics.se>, Richard Levitte]
1948 *) Add the arguments -CAfile and -CApath to the pkcs12 utility.
1951 *) The obj_dat.pl script was messing up the sorting of object
1952 names. The reason was that it compared the quoted version
1953 of strings as a result "OCSP" > "OCSP Signing" because
1954 " > SPACE. Changed script to store unquoted versions of
1955 names and add quotes on output. It was also omitting some
1956 names from the lookup table if they were given a default
1957 value (that is if SN is missing it is given the same
1958 value as LN and vice versa), these are now added on the
1959 grounds that if an object has a name we should be able to
1960 look it up. Finally added warning output when duplicate
1961 short or long names are found.
1964 *) Changes needed for Tandem NSK.
1965 [Scott Uroff <scott@xypro.com>]
1967 *) Fix SSL 2.0 rollback checking: Due to an off-by-one error in
1968 RSA_padding_check_SSLv23(), special padding was never detected
1969 and thus the SSL 3.0/TLS 1.0 countermeasure against protocol
1970 version rollback attacks was not effective.
1972 In s23_clnt.c, don't use special rollback-attack detection padding
1973 (RSA_SSLV23_PADDING) if SSL 2.0 is the only protocol enabled in the
1974 client; similarly, in s23_srvr.c, don't do the rollback check if
1975 SSL 2.0 is the only protocol enabled in the server.
1978 *) Make it possible to get hexdumps of unprintable data with 'openssl
1979 asn1parse'. By implication, the functions ASN1_parse_dump() and
1980 BIO_dump_indent() are added.
1983 *) New functions ASN1_STRING_print_ex() and X509_NAME_print_ex()
1984 these print out strings and name structures based on various
1985 flags including RFC2253 support and proper handling of
1986 multibyte characters. Added options to the 'x509' utility
1987 to allow the various flags to be set.
1990 *) Various fixes to use ASN1_TIME instead of ASN1_UTCTIME.
1991 Also change the functions X509_cmp_current_time() and
1992 X509_gmtime_adj() work with an ASN1_TIME structure,
1993 this will enable certificates using GeneralizedTime in validity
1994 dates to be checked.
1997 *) Make the NEG_PUBKEY_BUG code (which tolerates invalid
1998 negative public key encodings) on by default,
1999 NO_NEG_PUBKEY_BUG can be set to disable it.
2002 *) New function c2i_ASN1_OBJECT() which acts on ASN1_OBJECT
2003 content octets. An i2c_ASN1_OBJECT is unnecessary because
2004 the encoding can be trivially obtained from the structure.
2007 *) crypto/err.c locking bugfix: Use write locks (CRYPTO_w_[un]lock),
2008 not read locks (CRYPTO_r_[un]lock).
2011 *) A first attempt at creating official support for shared
2012 libraries through configuration. I've kept it so the
2013 default is static libraries only, and the OpenSSL programs
2014 are always statically linked for now, but there are
2015 preparations for dynamic linking in place.
2016 This has been tested on Linux and Tru64.
2019 *) Randomness polling function for Win9x, as described in:
2020 Peter Gutmann, Software Generation of Practically Strong
2024 *) Fix so PRNG is seeded in req if using an already existing
2028 *) New options to smime application. -inform and -outform
2029 allow alternative formats for the S/MIME message including
2030 PEM and DER. The -content option allows the content to be
2031 specified separately. This should allow things like Netscape
2032 form signing output easier to verify.
2035 *) Fix the ASN1 encoding of tags using the 'long form'.
2038 *) New ASN1 functions, i2c_* and c2i_* for INTEGER and BIT
2039 STRING types. These convert content octets to and from the
2040 underlying type. The actual tag and length octets are
2041 already assumed to have been read in and checked. These
2042 are needed because all other string types have virtually
2043 identical handling apart from the tag. By having versions
2044 of the ASN1 functions that just operate on content octets
2045 IMPLICIT tagging can be handled properly. It also allows
2046 the ASN1_ENUMERATED code to be cut down because ASN1_ENUMERATED
2047 and ASN1_INTEGER are identical apart from the tag.
2050 *) Change the handling of OID objects as follows:
2052 - New object identifiers are inserted in objects.txt, following
2053 the syntax given in objects.README.
2054 - objects.pl is used to process obj_mac.num and create a new
2056 - obj_dat.pl is used to create a new obj_dat.h, using the data in
2059 This is currently kind of a hack, and the perl code in objects.pl
2060 isn't very elegant, but it works as I intended. The simplest way
2061 to check that it worked correctly is to look in obj_dat.h and
2062 check the array nid_objs and make sure the objects haven't moved
2063 around (this is important!). Additions are OK, as well as
2064 consistent name changes.
2067 *) Add BSD-style MD5-based passwords to 'openssl passwd' (option '-1').
2070 *) Addition of the command line parameter '-rand file' to 'openssl req'.
2071 The given file adds to whatever has already been seeded into the
2072 random pool through the RANDFILE configuration file option or
2073 environment variable, or the default random state file.
2076 *) mkstack.pl now sorts each macro group into lexical order.
2077 Previously the output order depended on the order the files
2078 appeared in the directory, resulting in needless rewriting
2082 *) Patches to make OpenSSL compile under Win32 again. Mostly
2083 work arounds for the VC++ problem that it treats func() as
2084 func(void). Also stripped out the parts of mkdef.pl that
2085 added extra typesafe functions: these no longer exist.
2088 *) Reorganisation of the stack code. The macros are now all
2089 collected in safestack.h . Each macro is defined in terms of
2090 a "stack macro" of the form SKM_<name>(type, a, b). The
2091 DEBUG_SAFESTACK is now handled in terms of function casts,
2092 this has the advantage of retaining type safety without the
2093 use of additional functions. If DEBUG_SAFESTACK is not defined
2094 then the non typesafe macros are used instead. Also modified the
2095 mkstack.pl script to handle the new form. Needs testing to see
2096 if which (if any) compilers it chokes and maybe make DEBUG_SAFESTACK
2097 the default if no major problems. Similar behaviour for ASN1_SET_OF
2098 and PKCS12_STACK_OF.
2101 *) When some versions of IIS use the 'NET' form of private key the
2102 key derivation algorithm is different. Normally MD5(password) is
2103 used as a 128 bit RC4 key. In the modified case
2104 MD5(MD5(password) + "SGCKEYSALT") is used insted. Added some
2105 new functions i2d_RSA_NET(), d2i_RSA_NET() etc which are the same
2106 as the old Netscape_RSA functions except they have an additional
2107 'sgckey' parameter which uses the modified algorithm. Also added
2108 an -sgckey command line option to the rsa utility. Thanks to
2109 Adrian Peck <bertie@ncipher.com> for posting details of the modified
2110 algorithm to openssl-dev.
2113 *) The evp_local.h macros were using 'c.##kname' which resulted in
2114 invalid expansion on some systems (SCO 5.0.5 for example).
2115 Corrected to 'c.kname'.
2116 [Phillip Porch <root@theporch.com>]
2118 *) New X509_get1_email() and X509_REQ_get1_email() functions that return
2119 a STACK of email addresses from a certificate or request, these look
2120 in the subject name and the subject alternative name extensions and
2121 omit any duplicate addresses.
2124 *) Re-implement BN_mod_exp2_mont using independent (and larger) windows.
2125 This makes DSA verification about 2 % faster.
2128 *) Increase maximum window size in BN_mod_exp_... to 6 bits instead of 5
2129 (meaning that now 2^5 values will be precomputed, which is only 4 KB
2130 plus overhead for 1024 bit moduli).
2131 This makes exponentiations about 0.5 % faster for 1024 bit
2132 exponents (as measured by "openssl speed rsa2048").
2135 *) Rename memory handling macros to avoid conflicts with other
2137 Malloc => OPENSSL_malloc
2138 Malloc_locked => OPENSSL_malloc_locked
2139 Realloc => OPENSSL_realloc
2140 Free => OPENSSL_free
2143 *) New function BN_mod_exp_mont_word for small bases (roughly 15%
2144 faster than BN_mod_exp_mont, i.e. 7% for a full DH exchange).
2147 *) CygWin32 support.
2148 [John Jarvie <jjarvie@newsguy.com>]
2150 *) The type-safe stack code has been rejigged. It is now only compiled
2151 in when OpenSSL is configured with the DEBUG_SAFESTACK option and
2152 by default all type-specific stack functions are "#define"d back to
2153 standard stack functions. This results in more streamlined output
2154 but retains the type-safety checking possibilities of the original
2158 *) The STACK code has been cleaned up, and certain type declarations
2159 that didn't make a lot of sense have been brought in line. This has
2160 also involved a cleanup of sorts in safestack.h to more correctly
2161 map type-safe stack functions onto their plain stack counterparts.
2162 This work has also resulted in a variety of "const"ifications of
2163 lots of the code, especially "_cmp" operations which should normally
2164 be prototyped with "const" parameters anyway.
2167 *) When generating bytes for the first time in md_rand.c, 'stir the pool'
2168 by seeding with STATE_SIZE dummy bytes (with zero entropy count).
2169 (The PRNG state consists of two parts, the large pool 'state' and 'md',
2170 where all of 'md' is used each time the PRNG is used, but 'state'
2171 is used only indexed by a cyclic counter. As entropy may not be
2172 well distributed from the beginning, 'md' is important as a
2173 chaining variable. However, the output function chains only half
2174 of 'md', i.e. 80 bits. ssleay_rand_add, on the other hand, chains
2175 all of 'md', and seeding with STATE_SIZE dummy bytes will result
2176 in all of 'state' being rewritten, with the new values depending
2177 on virtually all of 'md'. This overcomes the 80 bit limitation.)
2180 *) In ssl/s2_clnt.c and ssl/s3_clnt.c, call ERR_clear_error() when
2181 the handshake is continued after ssl_verify_cert_chain();
2182 otherwise, if SSL_VERIFY_NONE is set, remaining error codes
2183 can lead to 'unexplainable' connection aborts later.
2184 [Bodo Moeller; problem tracked down by Lutz Jaenicke]
2186 *) Major EVP API cipher revision.
2187 Add hooks for extra EVP features. This allows various cipher
2188 parameters to be set in the EVP interface. Support added for variable
2189 key length ciphers via the EVP_CIPHER_CTX_set_key_length() function and
2190 setting of RC2 and RC5 parameters.
2192 Modify EVP_OpenInit() and EVP_SealInit() to cope with variable key length
2195 Remove lots of duplicated code from the EVP library. For example *every*
2196 cipher init() function handles the 'iv' in the same way according to the
2197 cipher mode. They also all do nothing if the 'key' parameter is NULL and
2198 for CFB and OFB modes they zero ctx->num.
2200 New functionality allows removal of S/MIME code RC2 hack.
2202 Most of the routines have the same form and so can be declared in terms
2205 By shifting this to the top level EVP_CipherInit() it can be removed from
2206 all individual ciphers. If the cipher wants to handle IVs or keys
2207 differently it can set the EVP_CIPH_CUSTOM_IV or EVP_CIPH_ALWAYS_CALL_INIT
2210 Change lots of functions like EVP_EncryptUpdate() to now return a
2211 value: although software versions of the algorithms cannot fail
2212 any installed hardware versions can.
2215 *) Implement SSL_OP_TLS_ROLLBACK_BUG: In ssl3_get_client_key_exchange, if
2216 this option is set, tolerate broken clients that send the negotiated
2217 protocol version number instead of the requested protocol version
2221 *) Call dh_tmp_cb (set by ..._TMP_DH_CB) with correct 'is_export' flag;
2222 i.e. non-zero for export ciphersuites, zero otherwise.
2223 Previous versions had this flag inverted, inconsistent with
2224 rsa_tmp_cb (..._TMP_RSA_CB).
2225 [Bodo Moeller; problem reported by Amit Chopra]
2227 *) Add missing DSA library text string. Work around for some IIS
2228 key files with invalid SEQUENCE encoding.
2231 *) Add a document (doc/standards.txt) that list all kinds of standards
2232 and so on that are implemented in OpenSSL.
2235 *) Enhance c_rehash script. Old version would mishandle certificates
2236 with the same subject name hash and wouldn't handle CRLs at all.
2237 Added -fingerprint option to crl utility, to support new c_rehash
2241 *) Eliminate non-ANSI declarations in crypto.h and stack.h.
2244 *) Fix for SSL server purpose checking. Server checking was
2245 rejecting certificates which had extended key usage present
2246 but no ssl client purpose.
2247 [Steve Henson, reported by Rene Grosser <grosser@hisolutions.com>]
2249 *) Make PKCS#12 code work with no password. The PKCS#12 spec
2250 is a little unclear about how a blank password is handled.
2251 Since the password in encoded as a BMPString with terminating
2252 double NULL a zero length password would end up as just the
2253 double NULL. However no password at all is different and is
2254 handled differently in the PKCS#12 key generation code. NS
2255 treats a blank password as zero length. MSIE treats it as no
2256 password on export: but it will try both on import. We now do
2257 the same: PKCS12_parse() tries zero length and no password if
2258 the password is set to "" or NULL (NULL is now a valid password:
2259 it wasn't before) as does the pkcs12 application.
2262 *) Bugfixes in apps/x509.c: Avoid a memory leak; and don't use
2263 perror when PEM_read_bio_X509_REQ fails, the error message must
2264 be obtained from the error queue.
2267 *) Avoid 'thread_hash' memory leak in crypto/err/err.c by freeing
2268 it in ERR_remove_state if appropriate, and change ERR_get_state
2269 accordingly to avoid race conditions (this is necessary because
2270 thread_hash is no longer constant once set).
2273 *) Bugfix for linux-elf makefile.one.
2276 *) RSA_get_default_method() will now cause a default
2277 RSA_METHOD to be chosen if one doesn't exist already.
2278 Previously this was only set during a call to RSA_new()
2279 or RSA_new_method(NULL) meaning it was possible for
2280 RSA_get_default_method() to return NULL.
2283 *) Added native name translation to the existing DSO code
2284 that will convert (if the flag to do so is set) filenames
2285 that are sufficiently small and have no path information
2286 into a canonical native form. Eg. "blah" converted to
2287 "libblah.so" or "blah.dll" etc.
2290 *) New function ERR_error_string_n(e, buf, len) which is like
2291 ERR_error_string(e, buf), but writes at most 'len' bytes
2292 including the 0 terminator. For ERR_error_string_n, 'buf'
2294 [Damien Miller <djm@mindrot.org>, Bodo Moeller]
2296 *) CONF library reworked to become more general. A new CONF
2297 configuration file reader "class" is implemented as well as a
2298 new functions (NCONF_*, for "New CONF") to handle it. The now
2299 old CONF_* functions are still there, but are reimplemented to
2300 work in terms of the new functions. Also, a set of functions
2301 to handle the internal storage of the configuration data is
2302 provided to make it easier to write new configuration file
2303 reader "classes" (I can definitely see something reading a
2304 configuration file in XML format, for example), called _CONF_*,
2305 or "the configuration storage API"...
2307 The new configuration file reading functions are:
2309 NCONF_new, NCONF_free, NCONF_load, NCONF_load_fp, NCONF_load_bio,
2310 NCONF_get_section, NCONF_get_string, NCONF_get_numbre
2312 NCONF_default, NCONF_WIN32
2314 NCONF_dump_fp, NCONF_dump_bio
2316 NCONF_default and NCONF_WIN32 are method (or "class") choosers,
2317 NCONF_new creates a new CONF object. This works in the same way
2318 as other interfaces in OpenSSL, like the BIO interface.
2319 NCONF_dump_* dump the internal storage of the configuration file,
2320 which is useful for debugging. All other functions take the same
2321 arguments as the old CONF_* functions wth the exception of the
2322 first that must be a `CONF *' instead of a `LHASH *'.
2324 To make it easer to use the new classes with the old CONF_* functions,
2325 the function CONF_set_default_method is provided.
2328 *) Add '-tls1' option to 'openssl ciphers', which was already
2329 mentioned in the documentation but had not been implemented.
2330 (This option is not yet really useful because even the additional
2331 experimental TLS 1.0 ciphers are currently treated as SSL 3.0 ciphers.)
2334 *) Initial DSO code added into libcrypto for letting OpenSSL (and
2335 OpenSSL-based applications) load shared libraries and bind to
2336 them in a portable way.
2337 [Geoff Thorpe, with contributions from Richard Levitte]
2339 Changes between 0.9.5 and 0.9.5a [1 Apr 2000]
2341 *) Make sure _lrotl and _lrotr are only used with MSVC.
2343 *) Use lock CRYPTO_LOCK_RAND correctly in ssleay_rand_status
2344 (the default implementation of RAND_status).
2346 *) Rename openssl x509 option '-crlext', which was added in 0.9.5,
2347 to '-clrext' (= clear extensions), as intended and documented.
2348 [Bodo Moeller; inconsistency pointed out by Michael Attili
2351 *) Fix for HMAC. It wasn't zeroing the rest of the block if the key length
2352 was larger than the MD block size.
2353 [Steve Henson, pointed out by Yost William <YostW@tce.com>]
2355 *) Modernise PKCS12_parse() so it uses STACK_OF(X509) for its ca argument
2356 fix a leak when the ca argument was passed as NULL. Stop X509_PUBKEY_set()
2357 using the passed key: if the passed key was a private key the result
2358 of X509_print(), for example, would be to print out all the private key
2362 *) des_quad_cksum() byte order bug fix.
2363 [Ulf Möller, using the problem description in krb4-0.9.7, where
2364 the solution is attributed to Derrick J Brashear <shadow@DEMENTIA.ORG>]
2366 *) Fix so V_ASN1_APP_CHOOSE works again: however its use is strongly
2368 [Steve Henson, pointed out by Brian Korver <briank@cs.stanford.edu>]
2370 *) For easily testing in shell scripts whether some command
2371 'openssl XXX' exists, the new pseudo-command 'openssl no-XXX'
2372 returns with exit code 0 iff no command of the given name is available.
2373 'no-XXX' is printed in this case, 'XXX' otherwise. In both cases,
2374 the output goes to stdout and nothing is printed to stderr.
2375 Additional arguments are always ignored.
2377 Since for each cipher there is a command of the same name,
2378 the 'no-cipher' compilation switches can be tested this way.
2380 ('openssl no-XXX' is not able to detect pseudo-commands such
2381 as 'quit', 'list-XXX-commands', or 'no-XXX' itself.)
2384 *) Update test suite so that 'make test' succeeds in 'no-rsa' configuration.
2387 *) For SSL_[CTX_]set_tmp_dh, don't create a DH key if SSL_OP_SINGLE_DH_USE
2388 is set; it will be thrown away anyway because each handshake creates
2390 ssl_cert_dup, which is used by SSL_new, now copies DH keys in addition
2391 to parameters -- in previous versions (since OpenSSL 0.9.3) the
2392 'default key' from SSL_CTX_set_tmp_dh would always be lost, meanining
2393 you effectivly got SSL_OP_SINGLE_DH_USE when using this macro.
2396 *) New s_client option -ign_eof: EOF at stdin is ignored, and
2397 'Q' and 'R' lose their special meanings (quit/renegotiate).
2398 This is part of what -quiet does; unlike -quiet, -ign_eof
2399 does not suppress any output.
2402 *) Add compatibility options to the purpose and trust code. The
2403 purpose X509_PURPOSE_ANY is "any purpose" which automatically
2404 accepts a certificate or CA, this was the previous behaviour,
2405 with all the associated security issues.
2407 X509_TRUST_COMPAT is the old trust behaviour: only and
2408 automatically trust self signed roots in certificate store. A
2409 new trust setting X509_TRUST_DEFAULT is used to specify that
2410 a purpose has no associated trust setting and it should instead
2411 use the value in the default purpose.
2414 *) Fix the PKCS#8 DSA private key code so it decodes keys again
2415 and fix a memory leak.
2418 *) In util/mkerr.pl (which implements 'make errors'), preserve
2419 reason strings from the previous version of the .c file, as
2420 the default to have only downcase letters (and digits) in
2421 automatically generated reasons codes is not always appropriate.
2424 *) In ERR_load_ERR_strings(), build an ERR_LIB_SYS error reason table
2425 using strerror. Previously, ERR_reason_error_string() returned
2426 library names as reason strings for SYSerr; but SYSerr is a special
2427 case where small numbers are errno values, not library numbers.
2430 *) Add '-dsaparam' option to 'openssl dhparam' application. This
2431 converts DSA parameters into DH parameters. (When creating parameters,
2432 DSA_generate_parameters is used.)
2435 *) Include 'length' (recommended exponent length) in C code generated
2436 by 'openssl dhparam -C'.
2439 *) The second argument to set_label in perlasm was already being used
2440 so couldn't be used as a "file scope" flag. Moved to third argument
2444 *) In PEM_ASN1_write_bio and some other functions, use RAND_pseudo_bytes
2445 instead of RAND_bytes for encryption IVs and salts.
2448 *) Include RAND_status() into RAND_METHOD instead of implementing
2449 it only for md_rand.c Otherwise replacing the PRNG by calling
2450 RAND_set_rand_method would be impossible.
2453 *) Don't let DSA_generate_key() enter an infinite loop if the random
2454 number generation fails.
2457 *) New 'rand' application for creating pseudo-random output.
2460 *) Added configuration support for Linux/IA64
2461 [Rolf Haberrecker <rolf@suse.de>]
2463 *) Assembler module support for Mingw32.
2466 *) Shared library support for HPUX (in shlib/).
2467 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE> and Anonymous]
2469 *) Shared library support for Solaris gcc.
2470 [Lutz Behnke <behnke@trustcenter.de>]
2472 Changes between 0.9.4 and 0.9.5 [28 Feb 2000]
2474 *) PKCS7_encrypt() was adding text MIME headers twice because they
2475 were added manually and by SMIME_crlf_copy().
2478 *) In bntest.c don't call BN_rand with zero bits argument.
2479 [Steve Henson, pointed out by Andrew W. Gray <agray@iconsinc.com>]
2481 *) BN_mul bugfix: In bn_mul_part_recursion() only the a>a[n] && b>b[n]
2482 case was implemented. This caused BN_div_recp() to fail occasionally.
2485 *) Add an optional second argument to the set_label() in the perl
2486 assembly language builder. If this argument exists and is set
2487 to 1 it signals that the assembler should use a symbol whose
2488 scope is the entire file, not just the current function. This
2489 is needed with MASM which uses the format label:: for this scope.
2490 [Steve Henson, pointed out by Peter Runestig <peter@runestig.com>]
2492 *) Change the ASN1 types so they are typedefs by default. Before
2493 almost all types were #define'd to ASN1_STRING which was causing
2494 STACK_OF() problems: you couldn't declare STACK_OF(ASN1_UTF8STRING)
2498 *) Change names of new functions to the new get1/get0 naming
2499 convention: After 'get1', the caller owns a reference count
2500 and has to call ..._free; 'get0' returns a pointer to some
2501 data structure without incrementing reference counters.
2502 (Some of the existing 'get' functions increment a reference
2503 counter, some don't.)
2504 Similarly, 'set1' and 'add1' functions increase reference
2505 counters or duplicate objects.
2508 *) Allow for the possibility of temp RSA key generation failure:
2509 the code used to assume it always worked and crashed on failure.
2512 *) Fix potential buffer overrun problem in BIO_printf().
2513 [Ulf Möller, using public domain code by Patrick Powell; problem
2514 pointed out by David Sacerdote <das33@cornell.edu>]
2516 *) Support EGD <http://www.lothar.com/tech/crypto/>. New functions
2517 RAND_egd() and RAND_status(). In the command line application,
2518 the EGD socket can be specified like a seed file using RANDFILE
2522 *) Allow the string CERTIFICATE to be tolerated in PKCS#7 structures.
2523 Some CAs (e.g. Verisign) distribute certificates in this form.
2526 *) Remove the SSL_ALLOW_ADH compile option and set the default cipher
2527 list to exclude them. This means that no special compilation option
2528 is needed to use anonymous DH: it just needs to be included in the
2532 *) Change the EVP_MD_CTX_type macro so its meaning consistent with
2533 EVP_MD_type. The old functionality is available in a new macro called
2534 EVP_MD_md(). Change code that uses it and update docs.
2537 *) ..._ctrl functions now have corresponding ..._callback_ctrl functions
2538 where the 'void *' argument is replaced by a function pointer argument.
2539 Previously 'void *' was abused to point to functions, which works on
2540 many platforms, but is not correct. As these functions are usually
2541 called by macros defined in OpenSSL header files, most source code
2542 should work without changes.
2545 *) <openssl/opensslconf.h> (which is created by Configure) now contains
2546 sections with information on -D... compiler switches used for
2547 compiling the library so that applications can see them. To enable
2548 one of these sections, a pre-processor symbol OPENSSL_..._DEFINES
2549 must be defined. E.g.,
2550 #define OPENSSL_ALGORITHM_DEFINES
2551 #include <openssl/opensslconf.h>
2552 defines all pertinent NO_<algo> symbols, such as NO_IDEA, NO_RSA, etc.
2553 [Richard Levitte, Ulf and Bodo Möller]
2555 *) Bugfix: Tolerate fragmentation and interleaving in the SSL 3/TLS
2559 *) Change the 'other' type in certificate aux info to a STACK_OF
2560 X509_ALGOR. Although not an AlgorithmIdentifier as such it has
2561 the required ASN1 format: arbitrary types determined by an OID.
2564 *) Add some PEM_write_X509_REQ_NEW() functions and a command line
2565 argument to 'req'. This is not because the function is newer or
2566 better than others it just uses the work 'NEW' in the certificate
2567 request header lines. Some software needs this.
2570 *) Reorganise password command line arguments: now passwords can be
2571 obtained from various sources. Delete the PEM_cb function and make
2572 it the default behaviour: i.e. if the callback is NULL and the
2573 usrdata argument is not NULL interpret it as a null terminated pass
2574 phrase. If usrdata and the callback are NULL then the pass phrase
2575 is prompted for as usual.
2578 *) Add support for the Compaq Atalla crypto accelerator. If it is installed,
2579 the support is automatically enabled. The resulting binaries will
2580 autodetect the card and use it if present.
2581 [Ben Laurie and Compaq Inc.]
2583 *) Work around for Netscape hang bug. This sends certificate request
2584 and server done in one record. Since this is perfectly legal in the
2585 SSL/TLS protocol it isn't a "bug" option and is on by default. See
2586 the bugs/SSLv3 entry for more info.
2589 *) HP-UX tune-up: new unified configs, HP C compiler bug workaround.
2592 *) Add -rand argument to smime and pkcs12 applications and read/write
2596 *) New 'passwd' tool for crypt(3) and apr1 password hashes.
2599 *) Add command line password options to the remaining applications.
2602 *) Bug fix for BN_div_recp() for numerators with an even number of
2606 *) More tests in bntest.c, and changed test_bn output.
2609 *) ./config recognizes MacOS X now.
2612 *) Bug fix for BN_div() when the first words of num and divsor are
2613 equal (it gave wrong results if (rem=(n1-q*d0)&BN_MASK2) < d0).
2616 *) Add support for various broken PKCS#8 formats, and command line
2617 options to produce them.
2620 *) New functions BN_CTX_start(), BN_CTX_get() and BT_CTX_end() to
2621 get temporary BIGNUMs from a BN_CTX.
2624 *) Correct return values in BN_mod_exp_mont() and BN_mod_exp2_mont()
2628 *) Change the SSLeay_add_all_*() functions to OpenSSL_add_all_*() and
2629 include a #define from the old name to the new. The original intent
2630 was that statically linked binaries could for example just call
2631 SSLeay_add_all_ciphers() to just add ciphers to the table and not
2632 link with digests. This never worked becayse SSLeay_add_all_digests()
2633 and SSLeay_add_all_ciphers() were in the same source file so calling
2634 one would link with the other. They are now in separate source files.
2637 *) Add a new -notext option to 'ca' and a -pubkey option to 'spkac'.
2640 *) Use a less unusual form of the Miller-Rabin primality test (it used
2641 a binary algorithm for exponentiation integrated into the Miller-Rabin
2642 loop, our standard modexp algorithms are faster).
2645 *) Support for the EBCDIC character set completed.
2646 [Martin Kraemer <Martin.Kraemer@Mch.SNI.De>]
2648 *) Source code cleanups: use const where appropriate, eliminate casts,
2649 use void * instead of char * in lhash.
2652 *) Bugfix: ssl3_send_server_key_exchange was not restartable
2653 (the state was not changed to SSL3_ST_SW_KEY_EXCH_B, and because of
2654 this the server could overwrite ephemeral keys that the client
2658 *) Turn DSA_is_prime into a macro that calls BN_is_prime,
2659 using 50 iterations of the Rabin-Miller test.
2661 DSA_generate_parameters now uses BN_is_prime_fasttest (with 50
2662 iterations of the Rabin-Miller test as required by the appendix
2663 to FIPS PUB 186[-1]) instead of DSA_is_prime.
2664 As BN_is_prime_fasttest includes trial division, DSA parameter
2665 generation becomes much faster.
2667 This implies a change for the callback functions in DSA_is_prime
2668 and DSA_generate_parameters: The callback function is called once
2669 for each positive witness in the Rabin-Miller test, not just
2670 occasionally in the inner loop; and the parameters to the
2671 callback function now provide an iteration count for the outer
2672 loop rather than for the current invocation of the inner loop.
2673 DSA_generate_parameters additionally can call the callback
2674 function with an 'iteration count' of -1, meaning that a
2675 candidate has passed the trial division test (when q is generated
2676 from an application-provided seed, trial division is skipped).
2679 *) New function BN_is_prime_fasttest that optionally does trial
2680 division before starting the Rabin-Miller test and has
2681 an additional BN_CTX * argument (whereas BN_is_prime always
2682 has to allocate at least one BN_CTX).
2683 'callback(1, -1, cb_arg)' is called when a number has passed the
2684 trial division stage.
2687 *) Fix for bug in CRL encoding. The validity dates weren't being handled
2691 *) New -pkcs12 option to CA.pl script to write out a PKCS#12 file.
2694 *) New function BN_pseudo_rand().
2697 *) Clean up BN_mod_mul_montgomery(): replace the broken (and unreadable)
2698 bignum version of BN_from_montgomery() with the working code from
2699 SSLeay 0.9.0 (the word based version is faster anyway), and clean up
2703 *) Avoid a race condition in s2_clnt.c (function get_server_hello) that
2704 made it impossible to use the same SSL_SESSION data structure in
2705 SSL2 clients in multiple threads.
2708 *) The return value of RAND_load_file() no longer counts bytes obtained
2709 by stat(). RAND_load_file(..., -1) is new and uses the complete file
2710 to seed the PRNG (previously an explicit byte count was required).
2711 [Ulf Möller, Bodo Möller]
2713 *) Clean up CRYPTO_EX_DATA functions, some of these didn't have prototypes
2714 used (char *) instead of (void *) and had casts all over the place.
2717 *) Make BN_generate_prime() return NULL on error if ret!=NULL.
2720 *) Retain source code compatibility for BN_prime_checks macro:
2721 BN_is_prime(..., BN_prime_checks, ...) now uses
2722 BN_prime_checks_for_size to determine the appropriate number of
2723 Rabin-Miller iterations.
2726 *) Diffie-Hellman uses "safe" primes: DH_check() return code renamed to
2727 DH_CHECK_P_NOT_SAFE_PRIME.
2728 (Check if this is true? OpenPGP calls them "strong".)
2731 *) Merge the functionality of "dh" and "gendh" programs into a new program
2732 "dhparam". The old programs are retained for now but will handle DH keys
2733 (instead of parameters) in future.
2736 *) Make the ciphers, s_server and s_client programs check the return values
2737 when a new cipher list is set.
2740 *) Enhance the SSL/TLS cipher mechanism to correctly handle the TLS 56bit
2741 ciphers. Before when the 56bit ciphers were enabled the sorting was
2744 The syntax for the cipher sorting has been extended to support sorting by
2745 cipher-strength (using the strength_bits hard coded in the tables).
2746 The new command is "@STRENGTH" (see also doc/apps/ciphers.pod).
2748 Fix a bug in the cipher-command parser: when supplying a cipher command
2749 string with an "undefined" symbol (neither command nor alphanumeric
2750 [A-Za-z0-9], ssl_set_cipher_list used to hang in an endless loop. Now
2751 an error is flagged.
2753 Due to the strength-sorting extension, the code of the
2754 ssl_create_cipher_list() function was completely rearranged. I hope that
2755 the readability was also increased :-)
2756 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>]
2758 *) Minor change to 'x509' utility. The -CAcreateserial option now uses 1
2759 for the first serial number and places 2 in the serial number file. This
2760 avoids problems when the root CA is created with serial number zero and
2761 the first user certificate has the same issuer name and serial number
2765 *) Fixes to X509_ATTRIBUTE utilities, change the 'req' program so it uses
2766 the new code. Add documentation for this stuff.
2769 *) Changes to X509_ATTRIBUTE utilities. These have been renamed from
2770 X509_*() to X509at_*() on the grounds that they don't handle X509
2771 structures and behave in an analagous way to the X509v3 functions:
2772 they shouldn't be called directly but wrapper functions should be used
2775 So we also now have some wrapper functions that call the X509at functions
2776 when passed certificate requests. (TO DO: similar things can be done with
2777 PKCS#7 signed and unsigned attributes, PKCS#12 attributes and a few other
2778 things. Some of these need some d2i or i2d and print functionality
2779 because they handle more complex structures.)
2782 *) Add missing #ifndefs that caused missing symbols when building libssl
2783 as a shared library without RSA. Use #ifndef NO_SSL2 instead of
2784 NO_RSA in ssl/s2*.c.
2785 [Kris Kennaway <kris@hub.freebsd.org>, modified by Ulf Möller]
2787 *) Precautions against using the PRNG uninitialized: RAND_bytes() now
2788 has a return value which indicates the quality of the random data
2789 (1 = ok, 0 = not seeded). Also an error is recorded on the thread's
2790 error queue. New function RAND_pseudo_bytes() generates output that is
2791 guaranteed to be unique but not unpredictable. RAND_add is like
2792 RAND_seed, but takes an extra argument for an entropy estimate
2793 (RAND_seed always assumes full entropy).
2796 *) Do more iterations of Rabin-Miller probable prime test (specifically,
2797 3 for 1024-bit primes, 6 for 512-bit primes, 12 for 256-bit primes
2798 instead of only 2 for all lengths; see BN_prime_checks_for_size definition
2799 in crypto/bn/bn_prime.c for the complete table). This guarantees a
2800 false-positive rate of at most 2^-80 for random input.
2803 *) Rewrite ssl3_read_n (ssl/s3_pkt.c) avoiding a couple of bugs.
2806 *) New function X509_CTX_rget_chain() (renamed to X509_CTX_get1_chain
2807 in the 0.9.5 release), this returns the chain
2808 from an X509_CTX structure with a dup of the stack and all
2809 the X509 reference counts upped: so the stack will exist
2810 after X509_CTX_cleanup() has been called. Modify pkcs12.c
2813 Also make SSL_SESSION_print() print out the verify return
2817 *) Add manpage for the pkcs12 command. Also change the default
2818 behaviour so MAC iteration counts are used unless the new
2819 -nomaciter option is used. This improves file security and
2820 only older versions of MSIE (4.0 for example) need it.
2823 *) Honor the no-xxx Configure options when creating .DEF files.
2826 *) Add PKCS#10 attributes to field table: challengePassword,
2827 unstructuredName and unstructuredAddress. These are taken from
2828 draft PKCS#9 v2.0 but are compatible with v1.2 provided no
2829 international characters are used.
2831 More changes to X509_ATTRIBUTE code: allow the setting of types
2832 based on strings. Remove the 'loc' parameter when adding
2833 attributes because these will be a SET OF encoding which is sorted
2837 *) Initial changes to the 'req' utility to allow request generation
2838 automation. This will allow an application to just generate a template
2839 file containing all the field values and have req construct the
2842 Initial support for X509_ATTRIBUTE handling. Stacks of these are
2843 used all over the place including certificate requests and PKCS#7
2844 structures. They are currently handled manually where necessary with
2845 some primitive wrappers for PKCS#7. The new functions behave in a
2846 manner analogous to the X509 extension functions: they allow
2847 attributes to be looked up by NID and added.
2849 Later something similar to the X509V3 code would be desirable to
2850 automatically handle the encoding, decoding and printing of the
2851 more complex types. The string types like challengePassword can
2852 be handled by the string table functions.
2854 Also modified the multi byte string table handling. Now there is
2855 a 'global mask' which masks out certain types. The table itself
2856 can use the flag STABLE_NO_MASK to ignore the mask setting: this
2857 is useful when for example there is only one permissible type
2858 (as in countryName) and using the mask might result in no valid
2862 *) Clean up 'Finished' handling, and add functions SSL_get_finished and
2863 SSL_get_peer_finished to allow applications to obtain the latest
2864 Finished messages sent to the peer or expected from the peer,
2865 respectively. (SSL_get_peer_finished is usually the Finished message
2866 actually received from the peer, otherwise the protocol will be aborted.)
2868 As the Finished message are message digests of the complete handshake
2869 (with a total of 192 bits for TLS 1.0 and more for SSL 3.0), they can
2870 be used for external authentication procedures when the authentication
2871 provided by SSL/TLS is not desired or is not enough.
2874 *) Enhanced support for Alpha Linux is added. Now ./config checks if
2875 the host supports BWX extension and if Compaq C is present on the
2876 $PATH. Just exploiting of the BWX extension results in 20-30%
2877 performance kick for some algorithms, e.g. DES and RC4 to mention
2878 a couple. Compaq C in turn generates ~20% faster code for MD5 and
2882 *) Add support for MS "fast SGC". This is arguably a violation of the
2883 SSL3/TLS protocol. Netscape SGC does two handshakes: the first with
2884 weak crypto and after checking the certificate is SGC a second one
2885 with strong crypto. MS SGC stops the first handshake after receiving
2886 the server certificate message and sends a second client hello. Since
2887 a server will typically do all the time consuming operations before
2888 expecting any further messages from the client (server key exchange
2889 is the most expensive) there is little difference between the two.
2891 To get OpenSSL to support MS SGC we have to permit a second client
2892 hello message after we have sent server done. In addition we have to
2893 reset the MAC if we do get this second client hello.
2896 *) Add a function 'd2i_AutoPrivateKey()' this will automatically decide
2897 if a DER encoded private key is RSA or DSA traditional format. Changed
2898 d2i_PrivateKey_bio() to use it. This is only needed for the "traditional"
2899 format DER encoded private key. Newer code should use PKCS#8 format which
2900 has the key type encoded in the ASN1 structure. Added DER private key
2901 support to pkcs8 application.
2904 *) SSL 3/TLS 1 servers now don't request certificates when an anonymous
2905 ciphersuites has been selected (as required by the SSL 3/TLS 1
2906 specifications). Exception: When SSL_VERIFY_FAIL_IF_NO_PEER_CERT
2907 is set, we interpret this as a request to violate the specification
2908 (the worst that can happen is a handshake failure, and 'correct'
2909 behaviour would result in a handshake failure anyway).
2912 *) In SSL_CTX_add_session, take into account that there might be multiple
2913 SSL_SESSION structures with the same session ID (e.g. when two threads
2914 concurrently obtain them from an external cache).
2915 The internal cache can handle only one SSL_SESSION with a given ID,
2916 so if there's a conflict, we now throw out the old one to achieve
2920 *) Add OIDs for idea and blowfish in CBC mode. This will allow both
2921 to be used in PKCS#5 v2.0 and S/MIME. Also add checking to
2922 some routines that use cipher OIDs: some ciphers do not have OIDs
2923 defined and so they cannot be used for S/MIME and PKCS#5 v2.0 for
2927 *) Simplify the trust setting structure and code. Now we just have
2928 two sequences of OIDs for trusted and rejected settings. These will
2929 typically have values the same as the extended key usage extension
2930 and any application specific purposes.
2932 The trust checking code now has a default behaviour: it will just
2933 check for an object with the same NID as the passed id. Functions can
2934 be provided to override either the default behaviour or the behaviour
2935 for a given id. SSL client, server and email already have functions
2936 in place for compatibility: they check the NID and also return "trusted"
2937 if the certificate is self signed.
2940 *) Add d2i,i2d bio/fp functions for PrivateKey: these convert the
2941 traditional format into an EVP_PKEY structure.
2944 *) Add a password callback function PEM_cb() which either prompts for
2945 a password if usr_data is NULL or otherwise assumes it is a null
2946 terminated password. Allow passwords to be passed on command line
2947 environment or config files in a few more utilities.
2950 *) Add a bunch of DER and PEM functions to handle PKCS#8 format private
2951 keys. Add some short names for PKCS#8 PBE algorithms and allow them
2952 to be specified on the command line for the pkcs8 and pkcs12 utilities.
2953 Update documentation.
2956 *) Support for ASN1 "NULL" type. This could be handled before by using
2957 ASN1_TYPE but there wasn't any function that would try to read a NULL
2958 and produce an error if it couldn't. For compatibility we also have
2959 ASN1_NULL_new() and ASN1_NULL_free() functions but these are faked and
2960 don't allocate anything because they don't need to.
2963 *) Initial support for MacOS is now provided. Examine INSTALL.MacOS
2965 [Andy Polyakov, Roy Woods <roy@centicsystems.ca>]
2967 *) Rebuild of the memory allocation routines used by OpenSSL code and
2968 possibly others as well. The purpose is to make an interface that
2969 provide hooks so anyone can build a separate set of allocation and
2970 deallocation routines to be used by OpenSSL, for example memory
2971 pool implementations, or something else, which was previously hard
2972 since Malloc(), Realloc() and Free() were defined as macros having
2973 the values malloc, realloc and free, respectively (except for Win32
2974 compilations). The same is provided for memory debugging code.
2975 OpenSSL already comes with functionality to find memory leaks, but
2976 this gives people a chance to debug other memory problems.
2978 With these changes, a new set of functions and macros have appeared:
2980 CRYPTO_set_mem_debug_functions() [F]
2981 CRYPTO_get_mem_debug_functions() [F]
2982 CRYPTO_dbg_set_options() [F]
2983 CRYPTO_dbg_get_options() [F]
2984 CRYPTO_malloc_debug_init() [M]
2986 The memory debug functions are NULL by default, unless the library
2987 is compiled with CRYPTO_MDEBUG or friends is defined. If someone
2988 wants to debug memory anyway, CRYPTO_malloc_debug_init() (which
2989 gives the standard debugging functions that come with OpenSSL) or
2990 CRYPTO_set_mem_debug_functions() (tells OpenSSL to use functions
2991 provided by the library user) must be used. When the standard
2992 debugging functions are used, CRYPTO_dbg_set_options can be used to
2993 request additional information:
2994 CRYPTO_dbg_set_options(V_CYRPTO_MDEBUG_xxx) corresponds to setting
2995 the CRYPTO_MDEBUG_xxx macro when compiling the library.
2997 Also, things like CRYPTO_set_mem_functions will always give the
2998 expected result (the new set of functions is used for allocation
2999 and deallocation) at all times, regardless of platform and compiler
3002 To finish it up, some functions that were never use in any other
3003 way than through macros have a new API and new semantic:
3006 CRYPTO_dbg_realloc()
3009 All macros of value have retained their old syntax.
3010 [Richard Levitte and Bodo Moeller]
3012 *) Some S/MIME fixes. The OID for SMIMECapabilities was wrong, the
3013 ordering of SMIMECapabilities wasn't in "strength order" and there
3014 was a missing NULL in the AlgorithmIdentifier for the SHA1 signature
3018 *) Some ASN1 types with illegal zero length encoding (INTEGER,
3019 ENUMERATED and OBJECT IDENTIFIER) choked the ASN1 routines.
3020 [Frans Heymans <fheymans@isaserver.be>, modified by Steve Henson]
3022 *) Merge in my S/MIME library for OpenSSL. This provides a simple
3023 S/MIME API on top of the PKCS#7 code, a MIME parser (with enough
3024 functionality to handle multipart/signed properly) and a utility
3025 called 'smime' to call all this stuff. This is based on code I
3026 originally wrote for Celo who have kindly allowed it to be
3027 included in OpenSSL.
3030 *) Add variants des_set_key_checked and des_set_key_unchecked of
3031 des_set_key (aka des_key_sched). Global variable des_check_key
3032 decides which of these is called by des_set_key; this way
3033 des_check_key behaves as it always did, but applications and
3034 the library itself, which was buggy for des_check_key == 1,
3035 have a cleaner way to pick the version they need.
3038 *) New function PKCS12_newpass() which changes the password of a
3042 *) Modify X509_TRUST and X509_PURPOSE so it also uses a static and
3043 dynamic mix. In both cases the ids can be used as an index into the
3044 table. Also modified the X509_TRUST_add() and X509_PURPOSE_add()
3045 functions so they accept a list of the field values and the
3046 application doesn't need to directly manipulate the X509_TRUST
3050 *) Modify the ASN1_STRING_TABLE stuff so it also uses bsearch and doesn't
3054 *) Modify the way the V3 extension code looks up extensions. This now
3055 works in a similar way to the object code: we have some "standard"
3056 extensions in a static table which is searched with OBJ_bsearch()
3057 and the application can add dynamic ones if needed. The file
3058 crypto/x509v3/ext_dat.h now has the info: this file needs to be
3059 updated whenever a new extension is added to the core code and kept
3060 in ext_nid order. There is a simple program 'tabtest.c' which checks
3061 this. New extensions are not added too often so this file can readily
3062 be maintained manually.
3064 There are two big advantages in doing things this way. The extensions
3065 can be looked up immediately and no longer need to be "added" using
3066 X509V3_add_standard_extensions(): this function now does nothing.
3067 [Side note: I get *lots* of email saying the extension code doesn't
3068 work because people forget to call this function]
3069 Also no dynamic allocation is done unless new extensions are added:
3070 so if we don't add custom extensions there is no need to call
3071 X509V3_EXT_cleanup().
3074 *) Modify enc utility's salting as follows: make salting the default. Add a
3075 magic header, so unsalted files fail gracefully instead of just decrypting
3076 to garbage. This is because not salting is a big security hole, so people
3077 should be discouraged from doing it.
3080 *) Fixes and enhancements to the 'x509' utility. It allowed a message
3081 digest to be passed on the command line but it only used this
3082 parameter when signing a certificate. Modified so all relevant
3083 operations are affected by the digest parameter including the
3084 -fingerprint and -x509toreq options. Also -x509toreq choked if a
3085 DSA key was used because it didn't fix the digest.
3088 *) Initial certificate chain verify code. Currently tests the untrusted
3089 certificates for consistency with the verify purpose (which is set
3090 when the X509_STORE_CTX structure is set up) and checks the pathlength.
3092 There is a NO_CHAIN_VERIFY compilation option to keep the old behaviour:
3093 this is because it will reject chains with invalid extensions whereas
3094 every previous version of OpenSSL and SSLeay made no checks at all.
3096 Trust code: checks the root CA for the relevant trust settings. Trust
3097 settings have an initial value consistent with the verify purpose: e.g.
3098 if the verify purpose is for SSL client use it expects the CA to be
3099 trusted for SSL client use. However the default value can be changed to
3100 permit custom trust settings: one example of this would be to only trust
3101 certificates from a specific "secure" set of CAs.
3103 Also added X509_STORE_CTX_new() and X509_STORE_CTX_free() functions
3104 which should be used for version portability: especially since the
3105 verify structure is likely to change more often now.
3107 SSL integration. Add purpose and trust to SSL_CTX and SSL and functions
3108 to set them. If not set then assume SSL clients will verify SSL servers
3111 Two new options to the verify program: -untrusted allows a set of
3112 untrusted certificates to be passed in and -purpose which sets the
3113 intended purpose of the certificate. If a purpose is set then the
3114 new chain verify code is used to check extension consistency.
3117 *) Support for the authority information access extension.
3120 *) Modify RSA and DSA PEM read routines to transparently handle
3121 PKCS#8 format private keys. New *_PUBKEY_* functions that handle
3122 public keys in a format compatible with certificate
3123 SubjectPublicKeyInfo structures. Unfortunately there were already
3124 functions called *_PublicKey_* which used various odd formats so
3125 these are retained for compatibility: however the DSA variants were
3126 never in a public release so they have been deleted. Changed dsa/rsa
3127 utilities to handle the new format: note no releases ever handled public
3128 keys so we should be OK.
3130 The primary motivation for this change is to avoid the same fiasco
3131 that dogs private keys: there are several incompatible private key
3132 formats some of which are standard and some OpenSSL specific and
3133 require various evil hacks to allow partial transparent handling and
3134 even then it doesn't work with DER formats. Given the option anything
3135 other than PKCS#8 should be dumped: but the other formats have to
3136 stay in the name of compatibility.
3138 With public keys and the benefit of hindsight one standard format
3139 is used which works with EVP_PKEY, RSA or DSA structures: though
3140 it clearly returns an error if you try to read the wrong kind of key.
3142 Added a -pubkey option to the 'x509' utility to output the public key.
3143 Also rename the EVP_PKEY_get_*() to EVP_PKEY_rget_*()
3144 (renamed to EVP_PKEY_get1_*() in the OpenSSL 0.9.5 release) and add
3145 EVP_PKEY_rset_*() functions (renamed to EVP_PKEY_set1_*())
3146 that do the same as the EVP_PKEY_assign_*() except they up the
3147 reference count of the added key (they don't "swallow" the
3151 *) Fixes to crypto/x509/by_file.c the code to read in certificates and
3152 CRLs would fail if the file contained no certificates or no CRLs:
3153 added a new function to read in both types and return the number
3154 read: this means that if none are read it will be an error. The
3155 DER versions of the certificate and CRL reader would always fail
3156 because it isn't possible to mix certificates and CRLs in DER format
3157 without choking one or the other routine. Changed this to just read
3158 a certificate: this is the best we can do. Also modified the code
3159 in apps/verify.c to take notice of return codes: it was previously
3160 attempting to read in certificates from NULL pointers and ignoring
3161 any errors: this is one reason why the cert and CRL reader seemed
3162 to work. It doesn't check return codes from the default certificate
3163 routines: these may well fail if the certificates aren't installed.
3166 *) Code to support otherName option in GeneralName.
3169 *) First update to verify code. Change the verify utility
3170 so it warns if it is passed a self signed certificate:
3171 for consistency with the normal behaviour. X509_verify
3172 has been modified to it will now verify a self signed
3173 certificate if *exactly* the same certificate appears
3174 in the store: it was previously impossible to trust a
3175 single self signed certificate. This means that:
3176 openssl verify ss.pem
3177 now gives a warning about a self signed certificate but
3178 openssl verify -CAfile ss.pem ss.pem
3182 *) For servers, store verify_result in SSL_SESSION data structure
3183 (and add it to external session representation).
3184 This is needed when client certificate verifications fails,
3185 but an application-provided verification callback (set by
3186 SSL_CTX_set_cert_verify_callback) allows accepting the session
3187 anyway (i.e. leaves x509_store_ctx->error != X509_V_OK
3188 but returns 1): When the session is reused, we have to set
3189 ssl->verify_result to the appropriate error code to avoid
3191 [Bodo Moeller, problem pointed out by Lutz Jaenicke]
3193 *) Fix a bug in the new PKCS#7 code: it didn't consider the
3194 case in PKCS7_dataInit() where the signed PKCS7 structure
3195 didn't contain any existing data because it was being created.
3196 [Po-Cheng Chen <pocheng@nst.com.tw>, slightly modified by Steve Henson]
3198 *) Add a salt to the key derivation routines in enc.c. This
3199 forms the first 8 bytes of the encrypted file. Also add a
3200 -S option to allow a salt to be input on the command line.
3203 *) New function X509_cmp(). Oddly enough there wasn't a function
3204 to compare two certificates. We do this by working out the SHA1
3205 hash and comparing that. X509_cmp() will be needed by the trust
3209 *) SSL_get1_session() is like SSL_get_session(), but increments
3210 the reference count in the SSL_SESSION returned.
3211 [Geoff Thorpe <geoff@eu.c2.net>]
3213 *) Fix for 'req': it was adding a null to request attributes.
3214 Also change the X509_LOOKUP and X509_INFO code to handle
3215 certificate auxiliary information.
3218 *) Add support for 40 and 64 bit RC2 and RC4 algorithms: document
3222 *) Add the possibility to add extra information to the memory leak
3223 detecting output, to form tracebacks, showing from where each
3224 allocation was originated: CRYPTO_push_info("constant string") adds
3225 the string plus current file name and line number to a per-thread
3226 stack, CRYPTO_pop_info() does the obvious, CRYPTO_remove_all_info()
3227 is like calling CYRPTO_pop_info() until the stack is empty.
3228 Also updated memory leak detection code to be multi-thread-safe.
3231 *) Add options -text and -noout to pkcs7 utility and delete the
3232 encryption options which never did anything. Update docs.
3235 *) Add options to some of the utilities to allow the pass phrase
3236 to be included on either the command line (not recommended on
3237 OSes like Unix) or read from the environment. Update the
3238 manpages and fix a few bugs.
3241 *) Add a few manpages for some of the openssl commands.
3244 *) Fix the -revoke option in ca. It was freeing up memory twice,
3245 leaking and not finding already revoked certificates.
3248 *) Extensive changes to support certificate auxiliary information.
3249 This involves the use of X509_CERT_AUX structure and X509_AUX
3250 functions. An X509_AUX function such as PEM_read_X509_AUX()
3251 can still read in a certificate file in the usual way but it
3252 will also read in any additional "auxiliary information". By
3253 doing things this way a fair degree of compatibility can be
3254 retained: existing certificates can have this information added
3255 using the new 'x509' options.
3257 Current auxiliary information includes an "alias" and some trust
3258 settings. The trust settings will ultimately be used in enhanced
3259 certificate chain verification routines: currently a certificate
3260 can only be trusted if it is self signed and then it is trusted
3264 *) Fix assembler for Alpha (tested only on DEC OSF not Linux or *BSD).
3265 The problem was that one of the replacement routines had not been working
3266 since SSLeay releases. For now the offending routine has been replaced
3267 with non-optimised assembler. Even so, this now gives around 95%
3268 performance improvement for 1024 bit RSA signs.
3271 *) Hack to fix PKCS#7 decryption when used with some unorthodox RC2
3272 handling. Most clients have the effective key size in bits equal to
3273 the key length in bits: so a 40 bit RC2 key uses a 40 bit (5 byte) key.
3274 A few however don't do this and instead use the size of the decrypted key
3275 to determine the RC2 key length and the AlgorithmIdentifier to determine
3276 the effective key length. In this case the effective key length can still
3277 be 40 bits but the key length can be 168 bits for example. This is fixed
3278 by manually forcing an RC2 key into the EVP_PKEY structure because the
3279 EVP code can't currently handle unusual RC2 key sizes: it always assumes
3280 the key length and effective key length are equal.
3283 *) Add a bunch of functions that should simplify the creation of
3284 X509_NAME structures. Now you should be able to do:
3285 X509_NAME_add_entry_by_txt(nm, "CN", MBSTRING_ASC, "Steve", -1, -1, 0);
3286 and have it automatically work out the correct field type and fill in
3287 the structures. The more adventurous can try:
3288 X509_NAME_add_entry_by_txt(nm, field, MBSTRING_UTF8, str, -1, -1, 0);
3289 and it will (hopefully) work out the correct multibyte encoding.
3292 *) Change the 'req' utility to use the new field handling and multibyte
3293 copy routines. Before the DN field creation was handled in an ad hoc
3294 way in req, ca, and x509 which was rather broken and didn't support
3295 BMPStrings or UTF8Strings. Since some software doesn't implement
3296 BMPStrings or UTF8Strings yet, they can be enabled using the config file
3297 using the dirstring_type option. See the new comment in the default
3298 openssl.cnf for more info.
3301 *) Make crypto/rand/md_rand.c more robust:
3302 - Assure unique random numbers after fork().
3303 - Make sure that concurrent threads access the global counter and
3304 md serializably so that we never lose entropy in them
3305 or use exactly the same state in multiple threads.
3306 Access to the large state is not always serializable because
3307 the additional locking could be a performance killer, and
3308 md should be large enough anyway.
3311 *) New file apps/app_rand.c with commonly needed functionality
3312 for handling the random seed file.
3314 Use the random seed file in some applications that previously did not:
3316 dsaparam -genkey (which also ignored its '-rand' option),
3319 x509 (when signing).
3320 Except on systems with /dev/urandom, it is crucial to have a random
3321 seed file at least for key creation, DSA signing, and for DH exchanges;
3322 for RSA signatures we could do without one.
3324 gendh and gendsa (unlike genrsa) used to read only the first byte
3325 of each file listed in the '-rand' option. The function as previously
3326 found in genrsa is now in app_rand.c and is used by all programs
3327 that support '-rand'.
3330 *) In RAND_write_file, use mode 0600 for creating files;
3331 don't just chmod when it may be too late.
3334 *) Report an error from X509_STORE_load_locations
3335 when X509_LOOKUP_load_file or X509_LOOKUP_add_dir failed.
3338 *) New function ASN1_mbstring_copy() this copies a string in either
3339 ASCII, Unicode, Universal (4 bytes per character) or UTF8 format
3340 into an ASN1_STRING type. A mask of permissible types is passed
3341 and it chooses the "minimal" type to use or an error if not type
3345 *) Add function equivalents to the various macros in asn1.h. The old
3346 macros are retained with an M_ prefix. Code inside the library can
3347 use the M_ macros. External code (including the openssl utility)
3348 should *NOT* in order to be "shared library friendly".
3351 *) Add various functions that can check a certificate's extensions
3352 to see if it usable for various purposes such as SSL client,
3353 server or S/MIME and CAs of these types. This is currently
3354 VERY EXPERIMENTAL but will ultimately be used for certificate chain
3355 verification. Also added a -purpose flag to x509 utility to
3356 print out all the purposes.
3359 *) Add a CRYPTO_EX_DATA to X509 certificate structure and associated
3363 *) New X509V3_{X509,CRL,REVOKED}_get_d2i() functions. These will search
3364 for, obtain and decode and extension and obtain its critical flag.
3365 This allows all the necessary extension code to be handled in a
3366 single function call.
3369 *) RC4 tune-up featuring 30-40% performance improvement on most RISC
3370 platforms. See crypto/rc4/rc4_enc.c for further details.
3373 *) New -noout option to asn1parse. This causes no output to be produced
3374 its main use is when combined with -strparse and -out to extract data
3375 from a file (which may not be in ASN.1 format).
3378 *) Fix for pkcs12 program. It was hashing an invalid certificate pointer
3379 when producing the local key id.
3380 [Richard Levitte <levitte@stacken.kth.se>]
3382 *) New option -dhparam in s_server. This allows a DH parameter file to be
3383 stated explicitly. If it is not stated then it tries the first server
3384 certificate file. The previous behaviour hard coded the filename
3388 *) Add -pubin and -pubout options to the rsa and dsa commands. These allow
3389 a public key to be input or output. For example:
3390 openssl rsa -in key.pem -pubout -out pubkey.pem
3391 Also added necessary DSA public key functions to handle this.
3394 *) Fix so PKCS7_dataVerify() doesn't crash if no certificates are contained
3395 in the message. This was handled by allowing
3396 X509_find_by_issuer_and_serial() to tolerate a NULL passed to it.
3397 [Steve Henson, reported by Sampo Kellomaki <sampo@mail.neuronio.pt>]
3399 *) Fix for bug in d2i_ASN1_bytes(): other ASN1 functions add an extra null
3400 to the end of the strings whereas this didn't. This would cause problems
3401 if strings read with d2i_ASN1_bytes() were later modified.
3402 [Steve Henson, reported by Arne Ansper <arne@ats.cyber.ee>]
3404 *) Fix for base64 decode bug. When a base64 bio reads only one line of
3405 data and it contains EOF it will end up returning an error. This is
3406 caused by input 46 bytes long. The cause is due to the way base64
3407 BIOs find the start of base64 encoded data. They do this by trying a
3408 trial decode on each line until they find one that works. When they
3409 do a flag is set and it starts again knowing it can pass all the
3410 data directly through the decoder. Unfortunately it doesn't reset
3411 the context it uses. This means that if EOF is reached an attempt
3412 is made to pass two EOFs through the context and this causes the
3413 resulting error. This can also cause other problems as well. As is
3414 usual with these problems it takes *ages* to find and the fix is
3415 trivial: move one line.
3416 [Steve Henson, reported by ian@uns.ns.ac.yu (Ivan Nejgebauer) ]
3418 *) Ugly workaround to get s_client and s_server working under Windows. The
3419 old code wouldn't work because it needed to select() on sockets and the
3420 tty (for keypresses and to see if data could be written). Win32 only
3421 supports select() on sockets so we select() with a 1s timeout on the
3422 sockets and then see if any characters are waiting to be read, if none
3423 are present then we retry, we also assume we can always write data to
3424 the tty. This isn't nice because the code then blocks until we've
3425 received a complete line of data and it is effectively polling the
3426 keyboard at 1s intervals: however it's quite a bit better than not
3427 working at all :-) A dedicated Windows application might handle this
3428 with an event loop for example.
3431 *) Enhance RSA_METHOD structure. Now there are two extra methods, rsa_sign
3432 and rsa_verify. When the RSA_FLAGS_SIGN_VER option is set these functions
3433 will be called when RSA_sign() and RSA_verify() are used. This is useful
3434 if rsa_pub_dec() and rsa_priv_enc() equivalents are not available.
3435 For this to work properly RSA_public_decrypt() and RSA_private_encrypt()
3436 should *not* be used: RSA_sign() and RSA_verify() must be used instead.
3437 This necessitated the support of an extra signature type NID_md5_sha1
3438 for SSL signatures and modifications to the SSL library to use it instead
3439 of calling RSA_public_decrypt() and RSA_private_encrypt().
3442 *) Add new -verify -CAfile and -CApath options to the crl program, these
3443 will lookup a CRL issuers certificate and verify the signature in a
3444 similar way to the verify program. Tidy up the crl program so it
3445 no longer accesses structures directly. Make the ASN1 CRL parsing a bit
3446 less strict. It will now permit CRL extensions even if it is not
3447 a V2 CRL: this will allow it to tolerate some broken CRLs.
3450 *) Initialize all non-automatic variables each time one of the openssl
3451 sub-programs is started (this is necessary as they may be started
3452 multiple times from the "OpenSSL>" prompt).
3453 [Lennart Bang, Bodo Moeller]
3455 *) Preliminary compilation option RSA_NULL which disables RSA crypto without
3456 removing all other RSA functionality (this is what NO_RSA does). This
3457 is so (for example) those in the US can disable those operations covered
3458 by the RSA patent while allowing storage and parsing of RSA keys and RSA
3462 *) Non-copying interface to BIO pairs.
3463 (still largely untested)
3466 *) New function ANS1_tag2str() to convert an ASN1 tag to a descriptive
3467 ASCII string. This was handled independently in various places before.
3470 *) New functions UTF8_getc() and UTF8_putc() that parse and generate
3471 UTF8 strings a character at a time.
3474 *) Use client_version from client hello to select the protocol
3475 (s23_srvr.c) and for RSA client key exchange verification
3476 (s3_srvr.c), as required by the SSL 3.0/TLS 1.0 specifications.
3479 *) Add various utility functions to handle SPKACs, these were previously
3480 handled by poking round in the structure internals. Added new function
3481 NETSCAPE_SPKI_print() to print out SPKAC and a new utility 'spkac' to
3482 print, verify and generate SPKACs. Based on an original idea from
3483 Massimiliano Pala <madwolf@comune.modena.it> but extensively modified.
3486 *) RIPEMD160 is operational on all platforms and is back in 'make test'.
3489 *) Allow the config file extension section to be overwritten on the
3490 command line. Based on an original idea from Massimiliano Pala
3491 <madwolf@comune.modena.it>. The new option is called -extensions
3492 and can be applied to ca, req and x509. Also -reqexts to override
3493 the request extensions in req and -crlexts to override the crl extensions
3497 *) Add new feature to the SPKAC handling in ca. Now you can include
3498 the same field multiple times by preceding it by "XXXX." for example:
3501 this is the same syntax as used in the req config file.
3504 *) Allow certificate extensions to be added to certificate requests. These
3505 are specified in a 'req_extensions' option of the req section of the
3506 config file. They can be printed out with the -text option to req but
3507 are otherwise ignored at present.
3510 *) Fix a horrible bug in enc_read() in crypto/evp/bio_enc.c: if the first
3511 data read consists of only the final block it would not decrypted because
3512 EVP_CipherUpdate() would correctly report zero bytes had been decrypted.
3513 A misplaced 'break' also meant the decrypted final block might not be
3514 copied until the next read.
3517 *) Initial support for DH_METHOD. Again based on RSA_METHOD. Also added
3518 a few extra parameters to the DH structure: these will be useful if
3519 for example we want the value of 'q' or implement X9.42 DH.
3522 *) Initial support for DSA_METHOD. This is based on the RSA_METHOD and
3523 provides hooks that allow the default DSA functions or functions on a
3524 "per key" basis to be replaced. This allows hardware acceleration and
3525 hardware key storage to be handled without major modification to the
3526 library. Also added low level modexp hooks and CRYPTO_EX structure and
3527 associated functions.
3530 *) Add a new flag to memory BIOs, BIO_FLAG_MEM_RDONLY. This marks the BIO
3531 as "read only": it can't be written to and the buffer it points to will
3532 not be freed. Reading from a read only BIO is much more efficient than
3533 a normal memory BIO. This was added because there are several times when
3534 an area of memory needs to be read from a BIO. The previous method was
3535 to create a memory BIO and write the data to it, this results in two
3536 copies of the data and an O(n^2) reading algorithm. There is a new
3537 function BIO_new_mem_buf() which creates a read only memory BIO from
3538 an area of memory. Also modified the PKCS#7 routines to use read only
3542 *) Bugfix: ssl23_get_client_hello did not work properly when called in
3543 state SSL23_ST_SR_CLNT_HELLO_B, i.e. when the first 7 bytes of
3544 a SSLv2-compatible client hello for SSLv3 or TLSv1 could be read,
3545 but a retry condition occured while trying to read the rest.
3548 *) The PKCS7_ENC_CONTENT_new() function was setting the content type as
3549 NID_pkcs7_encrypted by default: this was wrong since this should almost
3550 always be NID_pkcs7_data. Also modified the PKCS7_set_type() to handle
3551 the encrypted data type: this is a more sensible place to put it and it
3552 allows the PKCS#12 code to be tidied up that duplicated this
3556 *) Changed obj_dat.pl script so it takes its input and output files on
3557 the command line. This should avoid shell escape redirection problems
3561 *) Initial support for certificate extension requests, these are included
3562 in things like Xenroll certificate requests. Included functions to allow
3563 extensions to be obtained and added.
3566 *) -crlf option to s_client and s_server for sending newlines as
3567 CRLF (as required by many protocols).
3570 Changes between 0.9.3a and 0.9.4 [09 Aug 1999]
3572 *) Install libRSAglue.a when OpenSSL is built with RSAref.
3573 [Ralf S. Engelschall]
3575 *) A few more ``#ifndef NO_FP_API / #endif'' pairs for consistency.
3576 [Andrija Antonijevic <TheAntony2@bigfoot.com>]
3578 *) Fix -startdate and -enddate (which was missing) arguments to 'ca'
3582 *) New function DSA_dup_DH, which duplicates DSA parameters/keys as
3583 DH parameters/keys (q is lost during that conversion, but the resulting
3584 DH parameters contain its length).
3586 For 1024-bit p, DSA_generate_parameters followed by DSA_dup_DH is
3587 much faster than DH_generate_parameters (which creates parameters
3588 where p = 2*q + 1), and also the smaller q makes DH computations
3589 much more efficient (160-bit exponentiation instead of 1024-bit
3590 exponentiation); so this provides a convenient way to support DHE
3591 ciphersuites in SSL/TLS servers (see ssl/ssltest.c). It is of
3592 utter importance to use
3593 SSL_CTX_set_options(s_ctx, SSL_OP_SINGLE_DH_USE);
3595 SSL_set_options(s_ctx, SSL_OP_SINGLE_DH_USE);
3596 when such DH parameters are used, because otherwise small subgroup
3597 attacks may become possible!
3600 *) Avoid memory leak in i2d_DHparams.
3603 *) Allow the -k option to be used more than once in the enc program:
3604 this allows the same encrypted message to be read by multiple recipients.
3607 *) New function OBJ_obj2txt(buf, buf_len, a, no_name), this converts
3608 an ASN1_OBJECT to a text string. If the "no_name" parameter is set then
3609 it will always use the numerical form of the OID, even if it has a short
3613 *) Added an extra RSA flag: RSA_FLAG_EXT_PKEY. Previously the rsa_mod_exp
3614 method only got called if p,q,dmp1,dmq1,iqmp components were present,
3615 otherwise bn_mod_exp was called. In the case of hardware keys for example
3616 no private key components need be present and it might store extra data
3617 in the RSA structure, which cannot be accessed from bn_mod_exp.
3618 By setting RSA_FLAG_EXT_PKEY rsa_mod_exp will always be called for
3619 private key operations.
3622 *) Added support for SPARC Linux.
3625 *) pem_password_cb function type incompatibly changed from
3626 typedef int pem_password_cb(char *buf, int size, int rwflag);
3628 ....(char *buf, int size, int rwflag, void *userdata);
3629 so that applications can pass data to their callbacks:
3630 The PEM[_ASN1]_{read,write}... functions and macros now take an
3631 additional void * argument, which is just handed through whenever
3632 the password callback is called.
3633 [Damien Miller <dmiller@ilogic.com.au>; tiny changes by Bodo Moeller]
3635 New function SSL_CTX_set_default_passwd_cb_userdata.
3637 Compatibility note: As many C implementations push function arguments
3638 onto the stack in reverse order, the new library version is likely to
3639 interoperate with programs that have been compiled with the old
3640 pem_password_cb definition (PEM_whatever takes some data that
3641 happens to be on the stack as its last argument, and the callback
3642 just ignores this garbage); but there is no guarantee whatsoever that
3645 *) The -DPLATFORM="\"$(PLATFORM)\"" definition and the similar -DCFLAGS=...
3646 (both in crypto/Makefile.ssl for use by crypto/cversion.c) caused
3647 problems not only on Windows, but also on some Unix platforms.
3648 To avoid problematic command lines, these definitions are now in an
3649 auto-generated file crypto/buildinf.h (created by crypto/Makefile.ssl
3650 for standard "make" builds, by util/mk1mf.pl for "mk1mf" builds).
projects / openssl.git / blob