xref: /freebsd-13-stable/crypto/openssl/CHANGES (revision cf3d446e18bf2ddb6739fa4495e9f5ad6538aa54)
1
2 OpenSSL CHANGES
3 _______________
4
5 This is a high-level summary of the most important changes.
6 For a full list of changes, see the git commit log; for example,
7 https://github.com/openssl/openssl/commits/ and pick the appropriate
8 release branch.
9
10 Changes between 1.1.1v and 1.1.1w [11 Sep 2023]
11
12 *) Fix POLY1305 MAC implementation corrupting XMM registers on Windows.
13
14    The POLY1305 MAC (message authentication code) implementation in OpenSSL
15    does not save the contents of non-volatile XMM registers on Windows 64
16    platform when calculating the MAC of data larger than 64 bytes. Before
17    returning to the caller all the XMM registers are set to zero rather than
18    restoring their previous content. The vulnerable code is used only on newer
19    x86_64 processors supporting the AVX512-IFMA instructions.
20
21    The consequences of this kind of internal application state corruption can
22    be various - from no consequences, if the calling application does not
23    depend on the contents of non-volatile XMM registers at all, to the worst
24    consequences, where the attacker could get complete control of the
25    application process. However given the contents of the registers are just
26    zeroized so the attacker cannot put arbitrary values inside, the most likely
27    consequence, if any, would be an incorrect result of some application
28    dependent calculations or a crash leading to a denial of service.
29
30    (CVE-2023-4807)
31    [Bernd Edlinger]
32
33
34 Changes between 1.1.1u and 1.1.1v [1 Aug 2023]
35
36 *) Fix excessive time spent checking DH q parameter value.
37
38    The function DH_check() performs various checks on DH parameters. After
39    fixing CVE-2023-3446 it was discovered that a large q parameter value can
40    also trigger an overly long computation during some of these checks.
41    A correct q value, if present, cannot be larger than the modulus p
42    parameter, thus it is unnecessary to perform these checks if q is larger
43    than p.
44
45    If DH_check() is called with such q parameter value,
46    DH_CHECK_INVALID_Q_VALUE return flag is set and the computationally
47    intensive checks are skipped.
48
49    (CVE-2023-3817)
50    [Tomáš Mráz]
51
52 *) Fix DH_check() excessive time with over sized modulus
53
54    The function DH_check() performs various checks on DH parameters. One of
55    those checks confirms that the modulus ("p" parameter) is not too large.
56    Trying to use a very large modulus is slow and OpenSSL will not normally use
57    a modulus which is over 10,000 bits in length.
58
59    However the DH_check() function checks numerous aspects of the key or
60    parameters that have been supplied. Some of those checks use the supplied
61    modulus value even if it has already been found to be too large.
62
63    A new limit has been added to DH_check of 32,768 bits. Supplying a
64    key/parameters with a modulus over this size will simply cause DH_check()
65    to fail.
66    (CVE-2023-3446)
67    [Matt Caswell]
68
69 Changes between 1.1.1t and 1.1.1u [30 May 2023]
70
71  *) Mitigate for the time it takes for `OBJ_obj2txt` to translate gigantic
72     OBJECT IDENTIFIER sub-identifiers to canonical numeric text form.
73
74     OBJ_obj2txt() would translate any size OBJECT IDENTIFIER to canonical
75     numeric text form.  For gigantic sub-identifiers, this would take a very
76     long time, the time complexity being O(n^2) where n is the size of that
77     sub-identifier.  (CVE-2023-2650)
78
79     To mitigitate this, `OBJ_obj2txt()` will only translate an OBJECT
80     IDENTIFIER to canonical numeric text form if the size of that OBJECT
81     IDENTIFIER is 586 bytes or less, and fail otherwise.
82
83     The basis for this restriction is RFC 2578 (STD 58), section 3.5. OBJECT
84     IDENTIFIER values, which stipulates that OBJECT IDENTIFIERS may have at
85     most 128 sub-identifiers, and that the maximum value that each sub-
86     identifier may have is 2^32-1 (4294967295 decimal).
87
88     For each byte of every sub-identifier, only the 7 lower bits are part of
89     the value, so the maximum amount of bytes that an OBJECT IDENTIFIER with
90     these restrictions may occupy is 32 * 128 / 7, which is approximately 586
91     bytes.
92
93     Ref: https://datatracker.ietf.org/doc/html/rfc2578#section-3.5
94
95     [Richard Levitte]
96
97  *) Reworked the Fix for the Timing Oracle in RSA Decryption (CVE-2022-4304).
98     The previous fix for this timing side channel turned out to cause
99     a severe 2-3x performance regression in the typical use case
100     compared to 1.1.1s. The new fix uses existing constant time
101     code paths, and restores the previous performance level while
102     fully eliminating all existing timing side channels.
103     The fix was developed by Bernd Edlinger with testing support
104     by Hubert Kario.
105     [Bernd Edlinger]
106
107  *) Corrected documentation of X509_VERIFY_PARAM_add0_policy() to mention
108     that it does not enable policy checking. Thanks to
109     David Benjamin for discovering this issue. (CVE-2023-0466)
110     [Tomas Mraz]
111
112  *) Fixed an issue where invalid certificate policies in leaf certificates are
113     silently ignored by OpenSSL and other certificate policy checks are skipped
114     for that certificate. A malicious CA could use this to deliberately assert
115     invalid certificate policies in order to circumvent policy checking on the
116     certificate altogether. (CVE-2023-0465)
117     [Matt Caswell]
118
119  *) Limited the number of nodes created in a policy tree to mitigate
120     against CVE-2023-0464.  The default limit is set to 1000 nodes, which
121     should be sufficient for most installations.  If required, the limit
122     can be adjusted by setting the OPENSSL_POLICY_TREE_NODES_MAX build
123     time define to a desired maximum number of nodes or zero to allow
124     unlimited growth. (CVE-2023-0464)
125     [Paul Dale]
126
127 Changes between 1.1.1s and 1.1.1t [7 Feb 2023]
128
129  *) Fixed X.400 address type confusion in X.509 GeneralName.
130
131     There is a type confusion vulnerability relating to X.400 address processing
132     inside an X.509 GeneralName. X.400 addresses were parsed as an ASN1_STRING
133     but subsequently interpreted by GENERAL_NAME_cmp as an ASN1_TYPE. This
134     vulnerability may allow an attacker who can provide a certificate chain and
135     CRL (neither of which need have a valid signature) to pass arbitrary
136     pointers to a memcmp call, creating a possible read primitive, subject to
137     some constraints. Refer to the advisory for more information. Thanks to
138     David Benjamin for discovering this issue. (CVE-2023-0286)
139
140     This issue has been fixed by changing the public header file definition of
141     GENERAL_NAME so that x400Address reflects the implementation. It was not
142     possible for any existing application to successfully use the existing
143     definition; however, if any application references the x400Address field
144     (e.g. in dead code), note that the type of this field has changed. There is
145     no ABI change.
146     [Hugo Landau]
147
148  *) Fixed Use-after-free following BIO_new_NDEF.
149
150     The public API function BIO_new_NDEF is a helper function used for
151     streaming ASN.1 data via a BIO. It is primarily used internally to OpenSSL
152     to support the SMIME, CMS and PKCS7 streaming capabilities, but may also
153     be called directly by end user applications.
154
155     The function receives a BIO from the caller, prepends a new BIO_f_asn1
156     filter BIO onto the front of it to form a BIO chain, and then returns
157     the new head of the BIO chain to the caller. Under certain conditions,
158     for example if a CMS recipient public key is invalid, the new filter BIO
159     is freed and the function returns a NULL result indicating a failure.
160     However, in this case, the BIO chain is not properly cleaned up and the
161     BIO passed by the caller still retains internal pointers to the previously
162     freed filter BIO. If the caller then goes on to call BIO_pop() on the BIO
163     then a use-after-free will occur. This will most likely result in a crash.
164     (CVE-2023-0215)
165     [Viktor Dukhovni, Matt Caswell]
166
167  *) Fixed Double free after calling PEM_read_bio_ex.
168
169     The function PEM_read_bio_ex() reads a PEM file from a BIO and parses and
170     decodes the "name" (e.g. "CERTIFICATE"), any header data and the payload
171     data. If the function succeeds then the "name_out", "header" and "data"
172     arguments are populated with pointers to buffers containing the relevant
173     decoded data. The caller is responsible for freeing those buffers. It is
174     possible to construct a PEM file that results in 0 bytes of payload data.
175     In this case PEM_read_bio_ex() will return a failure code but will populate
176     the header argument with a pointer to a buffer that has already been freed.
177     If the caller also frees this buffer then a double free will occur. This
178     will most likely lead to a crash.
179
180     The functions PEM_read_bio() and PEM_read() are simple wrappers around
181     PEM_read_bio_ex() and therefore these functions are also directly affected.
182
183     These functions are also called indirectly by a number of other OpenSSL
184     functions including PEM_X509_INFO_read_bio_ex() and
185     SSL_CTX_use_serverinfo_file() which are also vulnerable. Some OpenSSL
186     internal uses of these functions are not vulnerable because the caller does
187     not free the header argument if PEM_read_bio_ex() returns a failure code.
188     (CVE-2022-4450)
189     [Kurt Roeckx, Matt Caswell]
190
191  *) Fixed Timing Oracle in RSA Decryption.
192
193     A timing based side channel exists in the OpenSSL RSA Decryption
194     implementation which could be sufficient to recover a plaintext across
195     a network in a Bleichenbacher style attack. To achieve a successful
196     decryption an attacker would have to be able to send a very large number
197     of trial messages for decryption. The vulnerability affects all RSA padding
198     modes: PKCS#1 v1.5, RSA-OEAP and RSASVE.
199     (CVE-2022-4304)
200     [Dmitry Belyavsky, Hubert Kario]
201
202 Changes between 1.1.1r and 1.1.1s [1 Nov 2022]
203
204  *) Fixed a regression introduced in 1.1.1r version not refreshing the
205     certificate data to be signed before signing the certificate.
206     [Gibeom Gwon]
207
208 Changes between 1.1.1q and 1.1.1r [11 Oct 2022]
209
210  *) Fixed the linux-mips64 Configure target which was missing the
211     SIXTY_FOUR_BIT bn_ops flag. This was causing heap corruption on that
212     platform.
213     [Adam Joseph]
214
215  *) Fixed a strict aliasing problem in bn_nist. Clang-14 optimisation was
216     causing incorrect results in some cases as a result.
217     [Paul Dale]
218
219  *) Fixed SSL_pending() and SSL_has_pending() with DTLS which were failing to
220     report correct results in some cases
221     [Matt Caswell]
222
223  *) Fixed a regression introduced in 1.1.1o for re-signing certificates with
224     different key sizes
225     [Todd Short]
226
227  *) Added the loongarch64 target
228     [Shi Pujin]
229
230  *) Fixed a DRBG seed propagation thread safety issue
231     [Bernd Edlinger]
232
233  *) Fixed a memory leak in tls13_generate_secret
234     [Bernd Edlinger]
235
236  *) Fixed reported performance degradation on aarch64. Restored the
237     implementation prior to commit 2621751 ("aes/asm/aesv8-armx.pl: avoid
238     32-bit lane assignment in CTR mode") for 64bit targets only, since it is
239     reportedly 2-17% slower and the silicon errata only affects 32bit targets.
240     The new algorithm is still used for 32 bit targets.
241     [Bernd Edlinger]
242
243  *) Added a missing header for memcmp that caused compilation failure on some
244     platforms
245     [Gregor Jasny]
246
247 Changes between 1.1.1p and 1.1.1q [5 Jul 2022]
248
249  *) AES OCB mode for 32-bit x86 platforms using the AES-NI assembly optimised
250     implementation would not encrypt the entirety of the data under some
251     circumstances.  This could reveal sixteen bytes of data that was
252     preexisting in the memory that wasn't written.  In the special case of
253     "in place" encryption, sixteen bytes of the plaintext would be revealed.
254
255     Since OpenSSL does not support OCB based cipher suites for TLS and DTLS,
256     they are both unaffected.
257     (CVE-2022-2097)
258     [Alex Chernyakhovsky, David Benjamin, Alejandro Sedeño]
259
260 Changes between 1.1.1o and 1.1.1p [21 Jun 2022]
261
262  *) In addition to the c_rehash shell command injection identified in
263     CVE-2022-1292, further bugs where the c_rehash script does not
264     properly sanitise shell metacharacters to prevent command injection have been
265     fixed.
266
267     When the CVE-2022-1292 was fixed it was not discovered that there
268     are other places in the script where the file names of certificates
269     being hashed were possibly passed to a command executed through the shell.
270
271     This script is distributed by some operating systems in a manner where
272     it is automatically executed.  On such operating systems, an attacker
273     could execute arbitrary commands with the privileges of the script.
274
275     Use of the c_rehash script is considered obsolete and should be replaced
276     by the OpenSSL rehash command line tool.
277     (CVE-2022-2068)
278     [Daniel Fiala, Tomáš Mráz]
279
280  *) When OpenSSL TLS client is connecting without any supported elliptic
281     curves and TLS-1.3 protocol is disabled the connection will no longer fail
282     if a ciphersuite that does not use a key exchange based on elliptic
283     curves can be negotiated.
284     [Tomáš Mráz]
285
286 Changes between 1.1.1n and 1.1.1o [3 May 2022]
287
288  *) Fixed a bug in the c_rehash script which was not properly sanitising shell
289     metacharacters to prevent command injection.  This script is distributed
290     by some operating systems in a manner where it is automatically executed.
291     On such operating systems, an attacker could execute arbitrary commands
292     with the privileges of the script.
293
294     Use of the c_rehash script is considered obsolete and should be replaced
295     by the OpenSSL rehash command line tool.
296     (CVE-2022-1292)
297     [Tomáš Mráz]
298
299 Changes between 1.1.1m and 1.1.1n [15 Mar 2022]
300
301  *) Fixed a bug in the BN_mod_sqrt() function that can cause it to loop forever
302     for non-prime moduli.
303
304     Internally this function is used when parsing certificates that contain
305     elliptic curve public keys in compressed form or explicit elliptic curve
306     parameters with a base point encoded in compressed form.
307
308     It is possible to trigger the infinite loop by crafting a certificate that
309     has invalid explicit curve parameters.
310
311     Since certificate parsing happens prior to verification of the certificate
312     signature, any process that parses an externally supplied certificate may
313     thus be subject to a denial of service attack. The infinite loop can also
314     be reached when parsing crafted private keys as they can contain explicit
315     elliptic curve parameters.
316
317     Thus vulnerable situations include:
318
319      - TLS clients consuming server certificates
320      - TLS servers consuming client certificates
321      - Hosting providers taking certificates or private keys from customers
322      - Certificate authorities parsing certification requests from subscribers
323      - Anything else which parses ASN.1 elliptic curve parameters
324
325     Also any other applications that use the BN_mod_sqrt() where the attacker
326     can control the parameter values are vulnerable to this DoS issue.
327     (CVE-2022-0778)
328     [Tomáš Mráz]
329
330  *) Add ciphersuites based on DHE_PSK (RFC 4279) and ECDHE_PSK (RFC 5489)
331     to the list of ciphersuites providing Perfect Forward Secrecy as
332     required by SECLEVEL >= 3.
333
334     [Dmitry Belyavskiy, Nicola Tuveri]
335
336 Changes between 1.1.1l and 1.1.1m [14 Dec 2021]
337
338  *) Avoid loading of a dynamic engine twice.
339
340     [Bernd Edlinger]
341
342  *) Fixed building on Debian with kfreebsd kernels
343
344     [Mattias Ellert]
345
346  *) Prioritise DANE TLSA issuer certs over peer certs
347
348     [Viktor Dukhovni]
349
350  *) Fixed random API for MacOS prior to 10.12
351
352     These MacOS versions don't support the CommonCrypto APIs
353
354     [Lenny Primak]
355
356 Changes between 1.1.1k and 1.1.1l [24 Aug 2021]
357
358  *) Fixed an SM2 Decryption Buffer Overflow.
359
360     In order to decrypt SM2 encrypted data an application is expected to call the
361     API function EVP_PKEY_decrypt(). Typically an application will call this
362     function twice. The first time, on entry, the "out" parameter can be NULL and,
363     on exit, the "outlen" parameter is populated with the buffer size required to
364     hold the decrypted plaintext. The application can then allocate a sufficiently
365     sized buffer and call EVP_PKEY_decrypt() again, but this time passing a non-NULL
366     value for the "out" parameter.
367
368     A bug in the implementation of the SM2 decryption code means that the
369     calculation of the buffer size required to hold the plaintext returned by the
370     first call to EVP_PKEY_decrypt() can be smaller than the actual size required by
371     the second call. This can lead to a buffer overflow when EVP_PKEY_decrypt() is
372     called by the application a second time with a buffer that is too small.
373
374     A malicious attacker who is able present SM2 content for decryption to an
375     application could cause attacker chosen data to overflow the buffer by up to a
376     maximum of 62 bytes altering the contents of other data held after the
377     buffer, possibly changing application behaviour or causing the application to
378     crash. The location of the buffer is application dependent but is typically
379     heap allocated.
380     (CVE-2021-3711)
381     [Matt Caswell]
382
383  *) Fixed various read buffer overruns processing ASN.1 strings
384
385     ASN.1 strings are represented internally within OpenSSL as an ASN1_STRING
386     structure which contains a buffer holding the string data and a field holding
387     the buffer length. This contrasts with normal C strings which are repesented as
388     a buffer for the string data which is terminated with a NUL (0) byte.
389
390     Although not a strict requirement, ASN.1 strings that are parsed using OpenSSL's
391     own "d2i" functions (and other similar parsing functions) as well as any string
392     whose value has been set with the ASN1_STRING_set() function will additionally
393     NUL terminate the byte array in the ASN1_STRING structure.
394
395     However, it is possible for applications to directly construct valid ASN1_STRING
396     structures which do not NUL terminate the byte array by directly setting the
397     "data" and "length" fields in the ASN1_STRING array. This can also happen by
398     using the ASN1_STRING_set0() function.
399
400     Numerous OpenSSL functions that print ASN.1 data have been found to assume that
401     the ASN1_STRING byte array will be NUL terminated, even though this is not
402     guaranteed for strings that have been directly constructed. Where an application
403     requests an ASN.1 structure to be printed, and where that ASN.1 structure
404     contains ASN1_STRINGs that have been directly constructed by the application
405     without NUL terminating the "data" field, then a read buffer overrun can occur.
406
407     The same thing can also occur during name constraints processing of certificates
408     (for example if a certificate has been directly constructed by the application
409     instead of loading it via the OpenSSL parsing functions, and the certificate
410     contains non NUL terminated ASN1_STRING structures). It can also occur in the
411     X509_get1_email(), X509_REQ_get1_email() and X509_get1_ocsp() functions.
412
413     If a malicious actor can cause an application to directly construct an
414     ASN1_STRING and then process it through one of the affected OpenSSL functions
415     then this issue could be hit. This might result in a crash (causing a Denial of
416     Service attack). It could also result in the disclosure of private memory
417     contents (such as private keys, or sensitive plaintext).
418     (CVE-2021-3712)
419     [Matt Caswell]
420
421 Changes between 1.1.1j and 1.1.1k [25 Mar 2021]
422
423  *) Fixed a problem with verifying a certificate chain when using the
424     X509_V_FLAG_X509_STRICT flag. This flag enables additional security checks
425     of the certificates present in a certificate chain. It is not set by
426     default.
427
428     Starting from OpenSSL version 1.1.1h a check to disallow certificates in
429     the chain that have explicitly encoded elliptic curve parameters was added
430     as an additional strict check.
431
432     An error in the implementation of this check meant that the result of a
433     previous check to confirm that certificates in the chain are valid CA
434     certificates was overwritten. This effectively bypasses the check
435     that non-CA certificates must not be able to issue other certificates.
436
437     If a "purpose" has been configured then there is a subsequent opportunity
438     for checks that the certificate is a valid CA.  All of the named "purpose"
439     values implemented in libcrypto perform this check.  Therefore, where
440     a purpose is set the certificate chain will still be rejected even when the
441     strict flag has been used. A purpose is set by default in libssl client and
442     server certificate verification routines, but it can be overridden or
443     removed by an application.
444
445     In order to be affected, an application must explicitly set the
446     X509_V_FLAG_X509_STRICT verification flag and either not set a purpose
447     for the certificate verification or, in the case of TLS client or server
448     applications, override the default purpose.
449     (CVE-2021-3450)
450     [Tomáš Mráz]
451
452  *) Fixed an issue where an OpenSSL TLS server may crash if sent a maliciously
453     crafted renegotiation ClientHello message from a client. If a TLSv1.2
454     renegotiation ClientHello omits the signature_algorithms extension (where
455     it was present in the initial ClientHello), but includes a
456     signature_algorithms_cert extension then a NULL pointer dereference will
457     result, leading to a crash and a denial of service attack.
458
459     A server is only vulnerable if it has TLSv1.2 and renegotiation enabled
460     (which is the default configuration). OpenSSL TLS clients are not impacted
461     by this issue.
462     (CVE-2021-3449)
463     [Peter Kästle and Samuel Sapalski]
464
465 Changes between 1.1.1i and 1.1.1j [16 Feb 2021]
466
467  *) Fixed the X509_issuer_and_serial_hash() function. It attempts to
468     create a unique hash value based on the issuer and serial number data
469     contained within an X509 certificate. However it was failing to correctly
470     handle any errors that may occur while parsing the issuer field (which might
471     occur if the issuer field is maliciously constructed). This may subsequently
472     result in a NULL pointer deref and a crash leading to a potential denial of
473     service attack.
474     (CVE-2021-23841)
475     [Matt Caswell]
476
477  *) Fixed the RSA_padding_check_SSLv23() function and the RSA_SSLV23_PADDING
478     padding mode to correctly check for rollback attacks. This is considered a
479     bug in OpenSSL 1.1.1 because it does not support SSLv2. In 1.0.2 this is
480     CVE-2021-23839.
481     [Matt Caswell]
482
483  *) Fixed the EVP_CipherUpdate, EVP_EncryptUpdate and EVP_DecryptUpdate
484     functions. Previously they could overflow the output length argument in some
485     cases where the input length is close to the maximum permissable length for
486     an integer on the platform. In such cases the return value from the function
487     call would be 1 (indicating success), but the output length value would be
488     negative. This could cause applications to behave incorrectly or crash.
489     (CVE-2021-23840)
490     [Matt Caswell]
491
492  *) Fixed SRP_Calc_client_key so that it runs in constant time. The previous
493     implementation called BN_mod_exp without setting BN_FLG_CONSTTIME. This
494     could be exploited in a side channel attack to recover the password. Since
495     the attack is local host only this is outside of the current OpenSSL
496     threat model and therefore no CVE is assigned.
497
498     Thanks to Mohammed Sabt and Daniel De Almeida Braga for reporting this
499     issue.
500     [Matt Caswell]
501
502 Changes between 1.1.1h and 1.1.1i [8 Dec 2020]
503
504  *) Fixed NULL pointer deref in the GENERAL_NAME_cmp function
505     This function could crash if both GENERAL_NAMEs contain an EDIPARTYNAME.
506     If an attacker can control both items being compared  then this could lead
507     to a possible denial of service attack. OpenSSL itself uses the
508     GENERAL_NAME_cmp function for two purposes:
509     1) Comparing CRL distribution point names between an available CRL and a
510        CRL distribution point embedded in an X509 certificate
511     2) When verifying that a timestamp response token signer matches the
512        timestamp authority name (exposed via the API functions
513        TS_RESP_verify_response and TS_RESP_verify_token)
514     (CVE-2020-1971)
515     [Matt Caswell]
516
517  *) Add support for Apple Silicon M1 Macs with the darwin64-arm64-cc target.
518     [Stuart Carnie]
519
520  *) The security callback, which can be customised by application code, supports
521     the security operation SSL_SECOP_TMP_DH. This is defined to take an EVP_PKEY
522     in the "other" parameter. In most places this is what is passed. All these
523     places occur server side. However there was one client side call of this
524     security operation and it passed a DH object instead. This is incorrect
525     according to the definition of SSL_SECOP_TMP_DH, and is inconsistent with all
526     of the other locations. Therefore this client side call has been changed to
527     pass an EVP_PKEY instead.
528     [Matt Caswell]
529
530  *) In 1.1.1h, an expired trusted (root) certificate was not anymore rejected
531     when validating a certificate path. This check is restored in 1.1.1i.
532     [David von Oheimb]
533
534 Changes between 1.1.1g and 1.1.1h [22 Sep 2020]
535
536  *) Certificates with explicit curve parameters are now disallowed in
537     verification chains if the X509_V_FLAG_X509_STRICT flag is used.
538     [Tomas Mraz]
539
540  *) The 'MinProtocol' and 'MaxProtocol' configuration commands now silently
541     ignore TLS protocol version bounds when configuring DTLS-based contexts, and
542     conversely, silently ignore DTLS protocol version bounds when configuring
543     TLS-based contexts.  The commands can be repeated to set bounds of both
544     types.  The same applies with the corresponding "min_protocol" and
545     "max_protocol" command-line switches, in case some application uses both TLS
546     and DTLS.
547
548     SSL_CTX instances that are created for a fixed protocol version (e.g.
549     TLSv1_server_method()) also silently ignore version bounds.  Previously
550     attempts to apply bounds to these protocol versions would result in an
551     error.  Now only the "version-flexible" SSL_CTX instances are subject to
552     limits in configuration files in command-line options.
553     [Viktor Dukhovni]
554
555  *) Handshake now fails if Extended Master Secret extension is dropped
556     on renegotiation.
557     [Tomas Mraz]
558
559  *) Accidentally, an expired trusted (root) certificate is not anymore rejected
560     when validating a certificate path.
561     [David von Oheimb]
562
563  *) The Oracle Developer Studio compiler will start reporting deprecated APIs
564
565 Changes between 1.1.1f and 1.1.1g [21 Apr 2020]
566
567  *) Fixed segmentation fault in SSL_check_chain()
568     Server or client applications that call the SSL_check_chain() function
569     during or after a TLS 1.3 handshake may crash due to a NULL pointer
570     dereference as a result of incorrect handling of the
571     "signature_algorithms_cert" TLS extension. The crash occurs if an invalid
572     or unrecognised signature algorithm is received from the peer. This could
573     be exploited by a malicious peer in a Denial of Service attack.
574     (CVE-2020-1967)
575     [Benjamin Kaduk]
576
577  *) Added AES consttime code for no-asm configurations
578     an optional constant time support for AES was added
579     when building openssl for no-asm.
580     Enable with: ./config no-asm -DOPENSSL_AES_CONST_TIME
581     Disable with: ./config no-asm -DOPENSSL_NO_AES_CONST_TIME
582     At this time this feature is by default disabled.
583     It will be enabled by default in 3.0.
584     [Bernd Edlinger]
585
586 Changes between 1.1.1e and 1.1.1f [31 Mar 2020]
587
588  *) Revert the change of EOF detection while reading in libssl to avoid
589     regressions in applications depending on the current way of reporting
590     the EOF. As the existing method is not fully accurate the change to
591     reporting the EOF via SSL_ERROR_SSL is kept on the current development
592     branch and will be present in the 3.0 release.
593     [Tomas Mraz]
594
595  *) Revised BN_generate_prime_ex to not avoid factors 3..17863 in p-1
596     when primes for RSA keys are computed.
597     Since we previously always generated primes == 2 (mod 3) for RSA keys,
598     the 2-prime and 3-prime RSA modules were easy to distinguish, since
599     N = p*q = 1 (mod 3), but N = p*q*r = 2 (mod 3). Therefore fingerprinting
600     2-prime vs. 3-prime RSA keys was possible by computing N mod 3.
601     This avoids possible fingerprinting of newly generated RSA modules.
602     [Bernd Edlinger]
603
604 Changes between 1.1.1d and 1.1.1e [17 Mar 2020]
605  *) Properly detect EOF while reading in libssl. Previously if we hit an EOF
606     while reading in libssl then we would report an error back to the
607     application (SSL_ERROR_SYSCALL) but errno would be 0. We now add
608     an error to the stack (which means we instead return SSL_ERROR_SSL) and
609     therefore give a hint as to what went wrong.
610     [Matt Caswell]
611
612  *) Check that ed25519 and ed448 are allowed by the security level. Previously
613     signature algorithms not using an MD were not being checked that they were
614     allowed by the security level.
615     [Kurt Roeckx]
616
617  *) Fixed SSL_get_servername() behaviour. The behaviour of SSL_get_servername()
618     was not quite right. The behaviour was not consistent between resumption
619     and normal handshakes, and also not quite consistent with historical
620     behaviour. The behaviour in various scenarios has been clarified and
621     it has been updated to make it match historical behaviour as closely as
622     possible.
623     [Matt Caswell]
624
625  *) [VMS only] The header files that the VMS compilers include automatically,
626     __DECC_INCLUDE_PROLOGUE.H and __DECC_INCLUDE_EPILOGUE.H, use pragmas that
627     the C++ compiler doesn't understand.  This is a shortcoming in the
628     compiler, but can be worked around with __cplusplus guards.
629
630     C++ applications that use OpenSSL libraries must be compiled using the
631     qualifier '/NAMES=(AS_IS,SHORTENED)' to be able to use all the OpenSSL
632     functions.  Otherwise, only functions with symbols of less than 31
633     characters can be used, as the linker will not be able to successfully
634     resolve symbols with longer names.
635     [Richard Levitte]
636
637  *) Corrected the documentation of the return values from the EVP_DigestSign*
638     set of functions.  The documentation mentioned negative values for some
639     errors, but this was never the case, so the mention of negative values
640     was removed.
641
642     Code that followed the documentation and thereby check with something
643     like 'EVP_DigestSignInit(...) <= 0' will continue to work undisturbed.
644     [Richard Levitte]
645
646  *) Fixed an an overflow bug in the x64_64 Montgomery squaring procedure
647     used in exponentiation with 512-bit moduli. No EC algorithms are
648     affected. Analysis suggests that attacks against 2-prime RSA1024,
649     3-prime RSA1536, and DSA1024 as a result of this defect would be very
650     difficult to perform and are not believed likely. Attacks against DH512
651     are considered just feasible. However, for an attack the target would
652     have to re-use the DH512 private key, which is not recommended anyway.
653     Also applications directly using the low level API BN_mod_exp may be
654     affected if they use BN_FLG_CONSTTIME.
655     (CVE-2019-1551)
656     [Andy Polyakov]
657
658  *) Added a new method to gather entropy on VMS, based on SYS$GET_ENTROPY.
659     The presence of this system service is determined at run-time.
660     [Richard Levitte]
661
662  *) Added newline escaping functionality to a filename when using openssl dgst.
663     This output format is to replicate the output format found in the '*sum'
664     checksum programs. This aims to preserve backward compatibility.
665     [Matt Eaton, Richard Levitte, and Paul Dale]
666
667  *) Print all values for a PKCS#12 attribute with 'openssl pkcs12', not just
668     the first value.
669     [Jon Spillett]
670
671 Changes between 1.1.1c and 1.1.1d [10 Sep 2019]
672
673  *) Fixed a fork protection issue. OpenSSL 1.1.1 introduced a rewritten random
674     number generator (RNG). This was intended to include protection in the
675     event of a fork() system call in order to ensure that the parent and child
676     processes did not share the same RNG state. However this protection was not
677     being used in the default case.
678
679     A partial mitigation for this issue is that the output from a high
680     precision timer is mixed into the RNG state so the likelihood of a parent
681     and child process sharing state is significantly reduced.
682
683     If an application already calls OPENSSL_init_crypto() explicitly using
684     OPENSSL_INIT_ATFORK then this problem does not occur at all.
685     (CVE-2019-1549)
686     [Matthias St. Pierre]
687
688  *) For built-in EC curves, ensure an EC_GROUP built from the curve name is
689     used even when parsing explicit parameters, when loading a serialized key
690     or calling `EC_GROUP_new_from_ecpkparameters()`/
691     `EC_GROUP_new_from_ecparameters()`.
692     This prevents bypass of security hardening and performance gains,
693     especially for curves with specialized EC_METHODs.
694     By default, if a key encoded with explicit parameters is loaded and later
695     serialized, the output is still encoded with explicit parameters, even if
696     internally a "named" EC_GROUP is used for computation.
697     [Nicola Tuveri]
698
699  *) Compute ECC cofactors if not provided during EC_GROUP construction. Before
700     this change, EC_GROUP_set_generator would accept order and/or cofactor as
701     NULL. After this change, only the cofactor parameter can be NULL. It also
702     does some minimal sanity checks on the passed order.
703     (CVE-2019-1547)
704     [Billy Bob Brumley]
705
706  *) Fixed a padding oracle in PKCS7_dataDecode and CMS_decrypt_set1_pkey.
707     An attack is simple, if the first CMS_recipientInfo is valid but the
708     second CMS_recipientInfo is chosen ciphertext. If the second
709     recipientInfo decodes to PKCS #1 v1.5 form plaintext, the correct
710     encryption key will be replaced by garbage, and the message cannot be
711     decoded, but if the RSA decryption fails, the correct encryption key is
712     used and the recipient will not notice the attack.
713     As a work around for this potential attack the length of the decrypted
714     key must be equal to the cipher default key length, in case the
715     certifiate is not given and all recipientInfo are tried out.
716     The old behaviour can be re-enabled in the CMS code by setting the
717     CMS_DEBUG_DECRYPT flag.
718     (CVE-2019-1563)
719     [Bernd Edlinger]
720
721  *) Early start up entropy quality from the DEVRANDOM seed source has been
722     improved for older Linux systems.  The RAND subsystem will wait for
723     /dev/random to be producing output before seeding from /dev/urandom.
724     The seeded state is stored for future library initialisations using
725     a system global shared memory segment.  The shared memory identifier
726     can be configured by defining OPENSSL_RAND_SEED_DEVRANDOM_SHM_ID to
727     the desired value.  The default identifier is 114.
728     [Paul Dale]
729
730  *) Correct the extended master secret constant on EBCDIC systems. Without this
731     fix TLS connections between an EBCDIC system and a non-EBCDIC system that
732     negotiate EMS will fail. Unfortunately this also means that TLS connections
733     between EBCDIC systems with this fix, and EBCDIC systems without this
734     fix will fail if they negotiate EMS.
735     [Matt Caswell]
736
737  *) Use Windows installation paths in the mingw builds
738
739     Mingw isn't a POSIX environment per se, which means that Windows
740     paths should be used for installation.
741     (CVE-2019-1552)
742     [Richard Levitte]
743
744  *) Changed DH_check to accept parameters with order q and 2q subgroups.
745     With order 2q subgroups the bit 0 of the private key is not secret
746     but DH_generate_key works around that by clearing bit 0 of the
747     private key for those. This avoids leaking bit 0 of the private key.
748     [Bernd Edlinger]
749
750  *) Significantly reduce secure memory usage by the randomness pools.
751     [Paul Dale]
752
753  *) Revert the DEVRANDOM_WAIT feature for Linux systems
754
755     The DEVRANDOM_WAIT feature added a select() call to wait for the
756     /dev/random device to become readable before reading from the
757     /dev/urandom device.
758
759     It turned out that this change had negative side effects on
760     performance which were not acceptable. After some discussion it
761     was decided to revert this feature and leave it up to the OS
762     resp. the platform maintainer to ensure a proper initialization
763     during early boot time.
764     [Matthias St. Pierre]
765
766 Changes between 1.1.1b and 1.1.1c [28 May 2019]
767
768  *) Add build tests for C++.  These are generated files that only do one
769     thing, to include one public OpenSSL head file each.  This tests that
770     the public header files can be usefully included in a C++ application.
771
772     This test isn't enabled by default.  It can be enabled with the option
773     'enable-buildtest-c++'.
774     [Richard Levitte]
775
776  *) Enable SHA3 pre-hashing for ECDSA and DSA.
777     [Patrick Steuer]
778
779  *) Change the default RSA, DSA and DH size to 2048 bit instead of 1024.
780     This changes the size when using the genpkey app when no size is given. It
781     fixes an omission in earlier changes that changed all RSA, DSA and DH
782     generation apps to use 2048 bits by default.
783     [Kurt Roeckx]
784
785  *) Reorganize the manual pages to consistently have RETURN VALUES,
786     EXAMPLES, SEE ALSO and HISTORY come in that order, and adjust
787     util/fix-doc-nits accordingly.
788     [Paul Yang, Joshua Lock]
789
790  *) Add the missing accessor EVP_PKEY_get0_engine()
791     [Matt Caswell]
792
793  *) Have apps like 's_client' and 's_server' output the signature scheme
794     along with other cipher suite parameters when debugging.
795     [Lorinczy Zsigmond]
796
797  *) Make OPENSSL_config() error agnostic again.
798     [Richard Levitte]
799
800  *) Do the error handling in RSA decryption constant time.
801     [Bernd Edlinger]
802
803  *) Prevent over long nonces in ChaCha20-Poly1305.
804
805     ChaCha20-Poly1305 is an AEAD cipher, and requires a unique nonce input
806     for every encryption operation. RFC 7539 specifies that the nonce value
807     (IV) should be 96 bits (12 bytes). OpenSSL allows a variable nonce length
808     and front pads the nonce with 0 bytes if it is less than 12
809     bytes. However it also incorrectly allows a nonce to be set of up to 16
810     bytes. In this case only the last 12 bytes are significant and any
811     additional leading bytes are ignored.
812
813     It is a requirement of using this cipher that nonce values are
814     unique. Messages encrypted using a reused nonce value are susceptible to
815     serious confidentiality and integrity attacks. If an application changes
816     the default nonce length to be longer than 12 bytes and then makes a
817     change to the leading bytes of the nonce expecting the new value to be a
818     new unique nonce then such an application could inadvertently encrypt
819     messages with a reused nonce.
820
821     Additionally the ignored bytes in a long nonce are not covered by the
822     integrity guarantee of this cipher. Any application that relies on the
823     integrity of these ignored leading bytes of a long nonce may be further
824     affected. Any OpenSSL internal use of this cipher, including in SSL/TLS,
825     is safe because no such use sets such a long nonce value. However user
826     applications that use this cipher directly and set a non-default nonce
827     length to be longer than 12 bytes may be vulnerable.
828
829     This issue was reported to OpenSSL on 16th of March 2019 by Joran Dirk
830     Greef of Ronomon.
831     (CVE-2019-1543)
832     [Matt Caswell]
833
834  *) Add DEVRANDOM_WAIT feature for Linux systems
835
836     On older Linux systems where the getrandom() system call is not available,
837     OpenSSL normally uses the /dev/urandom device for seeding its CSPRNG.
838     Contrary to getrandom(), the /dev/urandom device will not block during
839     early boot when the kernel CSPRNG has not been seeded yet.
840
841     To mitigate this known weakness, use select() to wait for /dev/random to
842     become readable before reading from /dev/urandom.
843
844  *) Ensure that SM2 only uses SM3 as digest algorithm
845     [Paul Yang]
846
847 Changes between 1.1.1a and 1.1.1b [26 Feb 2019]
848
849  *) Added SCA hardening for modular field inversion in EC_GROUP through
850     a new dedicated field_inv() pointer in EC_METHOD.
851     This also addresses a leakage affecting conversions from projective
852     to affine coordinates.
853     [Billy Bob Brumley, Nicola Tuveri]
854
855  *) Change the info callback signals for the start and end of a post-handshake
856     message exchange in TLSv1.3. In 1.1.1/1.1.1a we used SSL_CB_HANDSHAKE_START
857     and SSL_CB_HANDSHAKE_DONE. Experience has shown that many applications get
858     confused by this and assume that a TLSv1.2 renegotiation has started. This
859     can break KeyUpdate handling. Instead we no longer signal the start and end
860     of a post handshake message exchange (although the messages themselves are
861     still signalled). This could break some applications that were expecting
862     the old signals. However without this KeyUpdate is not usable for many
863     applications.
864     [Matt Caswell]
865
866  *) Fix a bug in the computation of the endpoint-pair shared secret used
867     by DTLS over SCTP. This breaks interoperability with older versions
868     of OpenSSL like OpenSSL 1.1.0 and OpenSSL 1.0.2. There is a runtime
869     switch SSL_MODE_DTLS_SCTP_LABEL_LENGTH_BUG (off by default) enabling
870     interoperability with such broken implementations. However, enabling
871     this switch breaks interoperability with correct implementations.
872
873  *) Fix a use after free bug in d2i_X509_PUBKEY when overwriting a
874     re-used X509_PUBKEY object if the second PUBKEY is malformed.
875     [Bernd Edlinger]
876
877  *) Move strictness check from EVP_PKEY_asn1_new() to EVP_PKEY_asn1_add0().
878     [Richard Levitte]
879
880  *) Remove the 'dist' target and add a tarball building script.  The
881     'dist' target has fallen out of use, and it shouldn't be
882     necessary to configure just to create a source distribution.
883     [Richard Levitte]
884
885  *) Added support for Linux Kernel TLS data-path. The Linux Kernel data-path
886     improves application performance by removing data copies and providing
887     applications with zero-copy system calls such as sendfile and splice.
888     [Boris Pismenny]
889
890 Changes between 1.1.1 and 1.1.1a [20 Nov 2018]
891
892  *) Timing vulnerability in DSA signature generation
893
894     The OpenSSL DSA signature algorithm has been shown to be vulnerable to a
895     timing side channel attack. An attacker could use variations in the signing
896     algorithm to recover the private key.
897
898     This issue was reported to OpenSSL on 16th October 2018 by Samuel Weiser.
899     (CVE-2018-0734)
900     [Paul Dale]
901
902  *) Timing vulnerability in ECDSA signature generation
903
904     The OpenSSL ECDSA signature algorithm has been shown to be vulnerable to a
905     timing side channel attack. An attacker could use variations in the signing
906     algorithm to recover the private key.
907
908     This issue was reported to OpenSSL on 25th October 2018 by Samuel Weiser.
909     (CVE-2018-0735)
910     [Paul Dale]
911
912  *) Added EVP_PKEY_ECDH_KDF_X9_63 and ecdh_KDF_X9_63() as replacements for
913     the EVP_PKEY_ECDH_KDF_X9_62 KDF type and ECDH_KDF_X9_62(). The old names
914     are retained for backwards compatibility.
915     [Antoine Salon]
916
917  *) Fixed the issue that RAND_add()/RAND_seed() silently discards random input
918     if its length exceeds 4096 bytes. The limit has been raised to a buffer size
919     of two gigabytes and the error handling improved.
920
921     This issue was reported to OpenSSL by Dr. Falko Strenzke. It has been
922     categorized as a normal bug, not a security issue, because the DRBG reseeds
923     automatically and is fully functional even without additional randomness
924     provided by the application.
925
926 Changes between 1.1.0i and 1.1.1 [11 Sep 2018]
927
928  *) Add a new ClientHello callback. Provides a callback interface that gives
929     the application the ability to adjust the nascent SSL object at the
930     earliest stage of ClientHello processing, immediately after extensions have
931     been collected but before they have been processed. In particular, this
932     callback can adjust the supported TLS versions in response to the contents
933     of the ClientHello
934     [Benjamin Kaduk]
935
936  *) Add SM2 base algorithm support.
937     [Jack Lloyd]
938
939  *) s390x assembly pack: add (improved) hardware-support for the following
940     cryptographic primitives: sha3, shake, aes-gcm, aes-ccm, aes-ctr, aes-ofb,
941     aes-cfb/cfb8, aes-ecb.
942     [Patrick Steuer]
943
944  *) Make EVP_PKEY_asn1_new() a bit stricter about its input.  A NULL pem_str
945     parameter is no longer accepted, as it leads to a corrupt table.  NULL
946     pem_str is reserved for alias entries only.
947     [Richard Levitte]
948
949  *) Use the new ec_scalar_mul_ladder scaffold to implement a specialized ladder
950     step for prime curves. The new implementation is based on formulae from
951     differential addition-and-doubling in homogeneous projective coordinates
952     from Izu-Takagi "A fast parallel elliptic curve multiplication resistant
953     against side channel attacks" and Brier-Joye "Weierstrass Elliptic Curves
954     and Side-Channel Attacks" Eq. (8) for y-coordinate recovery, modified
955     to work in projective coordinates.
956     [Billy Bob Brumley, Nicola Tuveri]
957
958  *) Change generating and checking of primes so that the error rate of not
959     being prime depends on the intended use based on the size of the input.
960     For larger primes this will result in more rounds of Miller-Rabin.
961     The maximal error rate for primes with more than 1080 bits is lowered
962     to 2^-128.
963     [Kurt Roeckx, Annie Yousar]
964
965  *) Increase the number of Miller-Rabin rounds for DSA key generating to 64.
966     [Kurt Roeckx]
967
968  *) The 'tsget' script is renamed to 'tsget.pl', to avoid confusion when
969     moving between systems, and to avoid confusion when a Windows build is
970     done with mingw vs with MSVC.  For POSIX installs, there's still a
971     symlink or copy named 'tsget' to avoid that confusion as well.
972     [Richard Levitte]
973
974  *) Revert blinding in ECDSA sign and instead make problematic addition
975     length-invariant. Switch even to fixed-length Montgomery multiplication.
976     [Andy Polyakov]
977
978  *) Use the new ec_scalar_mul_ladder scaffold to implement a specialized ladder
979     step for binary curves. The new implementation is based on formulae from
980     differential addition-and-doubling in mixed Lopez-Dahab projective
981     coordinates, modified to independently blind the operands.
982     [Billy Bob Brumley, Sohaib ul Hassan, Nicola Tuveri]
983
984  *) Add a scaffold to optionally enhance the Montgomery ladder implementation
985     for `ec_scalar_mul_ladder` (formerly `ec_mul_consttime`) allowing
986     EC_METHODs to implement their own specialized "ladder step", to take
987     advantage of more favorable coordinate systems or more efficient
988     differential addition-and-doubling algorithms.
989     [Billy Bob Brumley, Sohaib ul Hassan, Nicola Tuveri]
990
991  *) Modified the random device based seed sources to keep the relevant
992     file descriptors open rather than reopening them on each access.
993     This allows such sources to operate in a chroot() jail without
994     the associated device nodes being available. This behaviour can be
995     controlled using RAND_keep_random_devices_open().
996     [Paul Dale]
997
998  *) Numerous side-channel attack mitigations have been applied. This may have
999     performance impacts for some algorithms for the benefit of improved
1000     security. Specific changes are noted in this change log by their respective
1001     authors.
1002     [Matt Caswell]
1003
1004  *) AIX shared library support overhaul. Switch to AIX "natural" way of
1005     handling shared libraries, which means collecting shared objects of
1006     different versions and bitnesses in one common archive. This allows to
1007     mitigate conflict between 1.0 and 1.1 side-by-side installations. It
1008     doesn't affect the way 3rd party applications are linked, only how
1009     multi-version installation is managed.
1010     [Andy Polyakov]
1011
1012  *) Make ec_group_do_inverse_ord() more robust and available to other
1013     EC cryptosystems, so that irrespective of BN_FLG_CONSTTIME, SCA
1014     mitigations are applied to the fallback BN_mod_inverse().
1015     When using this function rather than BN_mod_inverse() directly, new
1016     EC cryptosystem implementations are then safer-by-default.
1017     [Billy Bob Brumley]
1018
1019  *) Add coordinate blinding for EC_POINT and implement projective
1020     coordinate blinding for generic prime curves as a countermeasure to
1021     chosen point SCA attacks.
1022     [Sohaib ul Hassan, Nicola Tuveri, Billy Bob Brumley]
1023
1024  *) Add blinding to ECDSA and DSA signatures to protect against side channel
1025     attacks discovered by Keegan Ryan (NCC Group).
1026     [Matt Caswell]
1027
1028  *) Enforce checking in the pkeyutl command line app to ensure that the input
1029     length does not exceed the maximum supported digest length when performing
1030     a sign, verify or verifyrecover operation.
1031     [Matt Caswell]
1032
1033  *) SSL_MODE_AUTO_RETRY is enabled by default. Applications that use blocking
1034     I/O in combination with something like select() or poll() will hang. This
1035     can be turned off again using SSL_CTX_clear_mode().
1036     Many applications do not properly handle non-application data records, and
1037     TLS 1.3 sends more of such records. Setting SSL_MODE_AUTO_RETRY works
1038     around the problems in those applications, but can also break some.
1039     It's recommended to read the manpages about SSL_read(), SSL_write(),
1040     SSL_get_error(), SSL_shutdown(), SSL_CTX_set_mode() and
1041     SSL_CTX_set_read_ahead() again.
1042     [Kurt Roeckx]
1043
1044  *) When unlocking a pass phrase protected PEM file or PKCS#8 container, we
1045     now allow empty (zero character) pass phrases.
1046     [Richard Levitte]
1047
1048  *) Apply blinding to binary field modular inversion and remove patent
1049     pending (OPENSSL_SUN_GF2M_DIV) BN_GF2m_mod_div implementation.
1050     [Billy Bob Brumley]
1051
1052  *) Deprecate ec2_mult.c and unify scalar multiplication code paths for
1053     binary and prime elliptic curves.
1054     [Billy Bob Brumley]
1055
1056  *) Remove ECDSA nonce padding: EC_POINT_mul is now responsible for
1057     constant time fixed point multiplication.
1058     [Billy Bob Brumley]
1059
1060  *) Revise elliptic curve scalar multiplication with timing attack
1061     defenses: ec_wNAF_mul redirects to a constant time implementation
1062     when computing fixed point and variable point multiplication (which
1063     in OpenSSL are mostly used with secret scalars in keygen, sign,
1064     ECDH derive operations).
1065     [Billy Bob Brumley, Nicola Tuveri, Cesar Pereida García,
1066      Sohaib ul Hassan]
1067
1068  *) Updated CONTRIBUTING
1069     [Rich Salz]
1070
1071  *) Updated DRBG / RAND to request nonce and additional low entropy
1072     randomness from the system.
1073     [Matthias St. Pierre]
1074
1075  *) Updated 'openssl rehash' to use OpenSSL consistent default.
1076     [Richard Levitte]
1077
1078  *) Moved the load of the ssl_conf module to libcrypto, which helps
1079     loading engines that libssl uses before libssl is initialised.
1080     [Matt Caswell]
1081
1082  *) Added EVP_PKEY_sign() and EVP_PKEY_verify() for EdDSA
1083     [Matt Caswell]
1084
1085  *) Fixed X509_NAME_ENTRY_set to get multi-valued RDNs right in all cases.
1086     [Ingo Schwarze, Rich Salz]
1087
1088  *) Added output of accepting IP address and port for 'openssl s_server'
1089     [Richard Levitte]
1090
1091  *) Added a new API for TLSv1.3 ciphersuites:
1092        SSL_CTX_set_ciphersuites()
1093        SSL_set_ciphersuites()
1094     [Matt Caswell]
1095
1096  *) Memory allocation failures consistently add an error to the error
1097     stack.
1098     [Rich Salz]
1099
1100  *) Don't use OPENSSL_ENGINES and OPENSSL_CONF environment values
1101     in libcrypto when run as setuid/setgid.
1102     [Bernd Edlinger]
1103
1104  *) Load any config file by default when libssl is used.
1105     [Matt Caswell]
1106
1107  *) Added new public header file <openssl/rand_drbg.h> and documentation
1108     for the RAND_DRBG API. See manual page RAND_DRBG(7) for an overview.
1109     [Matthias St. Pierre]
1110
1111  *) QNX support removed (cannot find contributors to get their approval
1112     for the license change).
1113     [Rich Salz]
1114
1115  *) TLSv1.3 replay protection for early data has been implemented. See the
1116     SSL_read_early_data() man page for further details.
1117     [Matt Caswell]
1118
1119  *) Separated TLSv1.3 ciphersuite configuration out from TLSv1.2 ciphersuite
1120     configuration. TLSv1.3 ciphersuites are not compatible with TLSv1.2 and
1121     below. Similarly TLSv1.2 ciphersuites are not compatible with TLSv1.3.
1122     In order to avoid issues where legacy TLSv1.2 ciphersuite configuration
1123     would otherwise inadvertently disable all TLSv1.3 ciphersuites the
1124     configuration has been separated out. See the ciphers man page or the
1125     SSL_CTX_set_ciphersuites() man page for more information.
1126     [Matt Caswell]
1127
1128  *) On POSIX (BSD, Linux, ...) systems the ocsp(1) command running
1129     in responder mode now supports the new "-multi" option, which
1130     spawns the specified number of child processes to handle OCSP
1131     requests.  The "-timeout" option now also limits the OCSP
1132     responder's patience to wait to receive the full client request
1133     on a newly accepted connection. Child processes are respawned
1134     as needed, and the CA index file is automatically reloaded
1135     when changed.  This makes it possible to run the "ocsp" responder
1136     as a long-running service, making the OpenSSL CA somewhat more
1137     feature-complete.  In this mode, most diagnostic messages logged
1138     after entering the event loop are logged via syslog(3) rather than
1139     written to stderr.
1140     [Viktor Dukhovni]
1141
1142  *) Added support for X448 and Ed448. Heavily based on original work by
1143     Mike Hamburg.
1144     [Matt Caswell]
1145
1146  *) Extend OSSL_STORE with capabilities to search and to narrow the set of
1147     objects loaded.  This adds the functions OSSL_STORE_expect() and
1148     OSSL_STORE_find() as well as needed tools to construct searches and
1149     get the search data out of them.
1150     [Richard Levitte]
1151
1152  *) Support for TLSv1.3 added. Note that users upgrading from an earlier
1153     version of OpenSSL should review their configuration settings to ensure
1154     that they are still appropriate for TLSv1.3. For further information see:
1155     https://wiki.openssl.org/index.php/TLS1.3
1156     [Matt Caswell]
1157
1158  *) Grand redesign of the OpenSSL random generator
1159
1160     The default RAND method now utilizes an AES-CTR DRBG according to
1161     NIST standard SP 800-90Ar1. The new random generator is essentially
1162     a port of the default random generator from the OpenSSL FIPS 2.0
1163     object module. It is a hybrid deterministic random bit generator
1164     using an AES-CTR bit stream and which seeds and reseeds itself
1165     automatically using trusted system entropy sources.
1166
1167     Some of its new features are:
1168      o Support for multiple DRBG instances with seed chaining.
1169      o The default RAND method makes use of a DRBG.
1170      o There is a public and private DRBG instance.
1171      o The DRBG instances are fork-safe.
1172      o Keep all global DRBG instances on the secure heap if it is enabled.
1173      o The public and private DRBG instance are per thread for lock free
1174        operation
1175     [Paul Dale, Benjamin Kaduk, Kurt Roeckx, Rich Salz, Matthias St. Pierre]
1176
1177  *) Changed Configure so it only says what it does and doesn't dump
1178     so much data.  Instead, ./configdata.pm should be used as a script
1179     to display all sorts of configuration data.
1180     [Richard Levitte]
1181
1182  *) Added processing of "make variables" to Configure.
1183     [Richard Levitte]
1184
1185  *) Added SHA512/224 and SHA512/256 algorithm support.
1186     [Paul Dale]
1187
1188  *) The last traces of Netware support, first removed in 1.1.0, have
1189     now been removed.
1190     [Rich Salz]
1191
1192  *) Get rid of Makefile.shared, and in the process, make the processing
1193     of certain files (rc.obj, or the .def/.map/.opt files produced from
1194     the ordinal files) more visible and hopefully easier to trace and
1195     debug (or make silent).
1196     [Richard Levitte]
1197
1198  *) Make it possible to have environment variable assignments as
1199     arguments to config / Configure.
1200     [Richard Levitte]
1201
1202  *) Add multi-prime RSA (RFC 8017) support.
1203     [Paul Yang]
1204
1205  *) Add SM3 implemented according to GB/T 32905-2016
1206     [ Jack Lloyd <jack.lloyd@ribose.com>,
1207       Ronald Tse <ronald.tse@ribose.com>,
1208       Erick Borsboom <erick.borsboom@ribose.com> ]
1209
1210  *) Add 'Maximum Fragment Length' TLS extension negotiation and support
1211     as documented in RFC6066.
1212     Based on a patch from Tomasz Moń
1213     [Filipe Raimundo da Silva]
1214
1215  *) Add SM4 implemented according to GB/T 32907-2016.
1216     [ Jack Lloyd <jack.lloyd@ribose.com>,
1217       Ronald Tse <ronald.tse@ribose.com>,
1218       Erick Borsboom <erick.borsboom@ribose.com> ]
1219
1220  *) Reimplement -newreq-nodes and ERR_error_string_n; the
1221     original author does not agree with the license change.
1222     [Rich Salz]
1223
1224  *) Add ARIA AEAD TLS support.
1225     [Jon Spillett]
1226
1227  *) Some macro definitions to support VS6 have been removed.  Visual
1228     Studio 6 has not worked since 1.1.0
1229     [Rich Salz]
1230
1231  *) Add ERR_clear_last_mark(), to allow callers to clear the last mark
1232     without clearing the errors.
1233     [Richard Levitte]
1234
1235  *) Add "atfork" functions.  If building on a system that without
1236     pthreads, see doc/man3/OPENSSL_fork_prepare.pod for application
1237     requirements.  The RAND facility now uses/requires this.
1238     [Rich Salz]
1239
1240  *) Add SHA3.
1241     [Andy Polyakov]
1242
1243  *) The UI API becomes a permanent and integral part of libcrypto, i.e.
1244     not possible to disable entirely.  However, it's still possible to
1245     disable the console reading UI method, UI_OpenSSL() (use UI_null()
1246     as a fallback).
1247
1248     To disable, configure with 'no-ui-console'.  'no-ui' is still
1249     possible to use as an alias.  Check at compile time with the
1250     macro OPENSSL_NO_UI_CONSOLE.  The macro OPENSSL_NO_UI is still
1251     possible to check and is an alias for OPENSSL_NO_UI_CONSOLE.
1252     [Richard Levitte]
1253
1254  *) Add a STORE module, which implements a uniform and URI based reader of
1255     stores that can contain keys, certificates, CRLs and numerous other
1256     objects.  The main API is loosely based on a few stdio functions,
1257     and includes OSSL_STORE_open, OSSL_STORE_load, OSSL_STORE_eof,
1258     OSSL_STORE_error and OSSL_STORE_close.
1259     The implementation uses backends called "loaders" to implement arbitrary
1260     URI schemes.  There is one built in "loader" for the 'file' scheme.
1261     [Richard Levitte]
1262
1263  *) Add devcrypto engine.  This has been implemented against cryptodev-linux,
1264     then adjusted to work on FreeBSD 8.4 as well.
1265     Enable by configuring with 'enable-devcryptoeng'.  This is done by default
1266     on BSD implementations, as cryptodev.h is assumed to exist on all of them.
1267     [Richard Levitte]
1268
1269  *) Module names can prefixed with OSSL_ or OPENSSL_.  This affects
1270     util/mkerr.pl, which is adapted to allow those prefixes, leading to
1271     error code calls like this:
1272
1273         OSSL_FOOerr(OSSL_FOO_F_SOMETHING, OSSL_FOO_R_WHATEVER);
1274
1275     With this change, we claim the namespaces OSSL and OPENSSL in a manner
1276     that can be encoded in C.  For the foreseeable future, this will only
1277     affect new modules.
1278     [Richard Levitte and Tim Hudson]
1279
1280  *) Removed BSD cryptodev engine.
1281     [Rich Salz]
1282
1283  *) Add a build target 'build_all_generated', to build all generated files
1284     and only that.  This can be used to prepare everything that requires
1285     things like perl for a system that lacks perl and then move everything
1286     to that system and do the rest of the build there.
1287     [Richard Levitte]
1288
1289  *) In the UI interface, make it possible to duplicate the user data.  This
1290     can be used by engines that need to retain the data for a longer time
1291     than just the call where this user data is passed.
1292     [Richard Levitte]
1293
1294  *) Ignore the '-named_curve auto' value for compatibility of applications
1295     with OpenSSL 1.0.2.
1296     [Tomas Mraz <tmraz@fedoraproject.org>]
1297
1298  *) Fragmented SSL/TLS alerts are no longer accepted. An alert message is 2
1299     bytes long. In theory it is permissible in SSLv3 - TLSv1.2 to fragment such
1300     alerts across multiple records (some of which could be empty). In practice
1301     it make no sense to send an empty alert record, or to fragment one. TLSv1.3
1302     prohibits this altogether and other libraries (BoringSSL, NSS) do not
1303     support this at all. Supporting it adds significant complexity to the
1304     record layer, and its removal is unlikely to cause interoperability
1305     issues.
1306     [Matt Caswell]
1307
1308  *) Add the ASN.1 types INT32, UINT32, INT64, UINT64 and variants prefixed
1309     with Z.  These are meant to replace LONG and ZLONG and to be size safe.
1310     The use of LONG and ZLONG is discouraged and scheduled for deprecation
1311     in OpenSSL 1.2.0.
1312     [Richard Levitte]
1313
1314  *) Add the 'z' and 'j' modifiers to BIO_printf() et al formatting string,
1315     'z' is to be used for [s]size_t, and 'j' - with [u]int64_t.
1316     [Richard Levitte, Andy Polyakov]
1317
1318  *) Add EC_KEY_get0_engine(), which does for EC_KEY what RSA_get0_engine()
1319     does for RSA, etc.
1320     [Richard Levitte]
1321
1322  *) Have 'config' recognise 64-bit mingw and choose 'mingw64' as the target
1323     platform rather than 'mingw'.
1324     [Richard Levitte]
1325
1326  *) The functions X509_STORE_add_cert and X509_STORE_add_crl return
1327     success if they are asked to add an object which already exists
1328     in the store. This change cascades to other functions which load
1329     certificates and CRLs.
1330     [Paul Dale]
1331
1332  *) x86_64 assembly pack: annotate code with DWARF CFI directives to
1333     facilitate stack unwinding even from assembly subroutines.
1334     [Andy Polyakov]
1335
1336  *) Remove VAX C specific definitions of OPENSSL_EXPORT, OPENSSL_EXTERN.
1337     Also remove OPENSSL_GLOBAL entirely, as it became a no-op.
1338     [Richard Levitte]
1339
1340  *) Remove the VMS-specific reimplementation of gmtime from crypto/o_times.c.
1341     VMS C's RTL has a fully up to date gmtime() and gmtime_r() since V7.1,
1342     which is the minimum version we support.
1343     [Richard Levitte]
1344
1345  *) Certificate time validation (X509_cmp_time) enforces stricter
1346     compliance with RFC 5280. Fractional seconds and timezone offsets
1347     are no longer allowed.
1348     [Emilia Käsper]
1349
1350  *) Add support for ARIA
1351     [Paul Dale]
1352
1353  *) s_client will now send the Server Name Indication (SNI) extension by
1354     default unless the new "-noservername" option is used. The server name is
1355     based on the host provided to the "-connect" option unless overridden by
1356     using "-servername".
1357     [Matt Caswell]
1358
1359  *) Add support for SipHash
1360     [Todd Short]
1361
1362  *) OpenSSL now fails if it receives an unrecognised record type in TLS1.0
1363     or TLS1.1. Previously this only happened in SSLv3 and TLS1.2. This is to
1364     prevent issues where no progress is being made and the peer continually
1365     sends unrecognised record types, using up resources processing them.
1366     [Matt Caswell]
1367
1368  *) 'openssl passwd' can now produce SHA256 and SHA512 based output,
1369     using the algorithm defined in
1370     https://www.akkadia.org/drepper/SHA-crypt.txt
1371     [Richard Levitte]
1372
1373  *) Heartbeat support has been removed; the ABI is changed for now.
1374     [Richard Levitte, Rich Salz]
1375
1376  *) Support for SSL_OP_NO_ENCRYPT_THEN_MAC in SSL_CONF_cmd.
1377     [Emilia Käsper]
1378
1379  *) The RSA "null" method, which was partially supported to avoid patent
1380     issues, has been replaced to always returns NULL.
1381     [Rich Salz]
1382
1383
1384 Changes between 1.1.0h and 1.1.0i [xx XXX xxxx]
1385
1386  *) Client DoS due to large DH parameter
1387
1388     During key agreement in a TLS handshake using a DH(E) based ciphersuite a
1389     malicious server can send a very large prime value to the client. This will
1390     cause the client to spend an unreasonably long period of time generating a
1391     key for this prime resulting in a hang until the client has finished. This
1392     could be exploited in a Denial Of Service attack.
1393
1394     This issue was reported to OpenSSL on 5th June 2018 by Guido Vranken
1395     (CVE-2018-0732)
1396     [Guido Vranken]
1397
1398  *) Cache timing vulnerability in RSA Key Generation
1399
1400     The OpenSSL RSA Key generation algorithm has been shown to be vulnerable to
1401     a cache timing side channel attack. An attacker with sufficient access to
1402     mount cache timing attacks during the RSA key generation process could
1403     recover the private key.
1404
1405     This issue was reported to OpenSSL on 4th April 2018 by Alejandro Cabrera
1406     Aldaya, Billy Brumley, Cesar Pereida Garcia and Luis Manuel Alvarez Tapia.
1407     (CVE-2018-0737)
1408     [Billy Brumley]
1409
1410  *) Make EVP_PKEY_asn1_new() a bit stricter about its input.  A NULL pem_str
1411     parameter is no longer accepted, as it leads to a corrupt table.  NULL
1412     pem_str is reserved for alias entries only.
1413     [Richard Levitte]
1414
1415  *) Revert blinding in ECDSA sign and instead make problematic addition
1416     length-invariant. Switch even to fixed-length Montgomery multiplication.
1417     [Andy Polyakov]
1418
1419  *) Change generating and checking of primes so that the error rate of not
1420     being prime depends on the intended use based on the size of the input.
1421     For larger primes this will result in more rounds of Miller-Rabin.
1422     The maximal error rate for primes with more than 1080 bits is lowered
1423     to 2^-128.
1424     [Kurt Roeckx, Annie Yousar]
1425
1426  *) Increase the number of Miller-Rabin rounds for DSA key generating to 64.
1427     [Kurt Roeckx]
1428
1429  *) Add blinding to ECDSA and DSA signatures to protect against side channel
1430     attacks discovered by Keegan Ryan (NCC Group).
1431     [Matt Caswell]
1432
1433  *) When unlocking a pass phrase protected PEM file or PKCS#8 container, we
1434     now allow empty (zero character) pass phrases.
1435     [Richard Levitte]
1436
1437  *) Certificate time validation (X509_cmp_time) enforces stricter
1438     compliance with RFC 5280. Fractional seconds and timezone offsets
1439     are no longer allowed.
1440     [Emilia Käsper]
1441
1442  *) Fixed a text canonicalisation bug in CMS
1443
1444     Where a CMS detached signature is used with text content the text goes
1445     through a canonicalisation process first prior to signing or verifying a
1446     signature. This process strips trailing space at the end of lines, converts
1447     line terminators to CRLF and removes additional trailing line terminators
1448     at the end of a file. A bug in the canonicalisation process meant that
1449     some characters, such as form-feed, were incorrectly treated as whitespace
1450     and removed. This is contrary to the specification (RFC5485). This fix
1451     could mean that detached text data signed with an earlier version of
1452     OpenSSL 1.1.0 may fail to verify using the fixed version, or text data
1453     signed with a fixed OpenSSL may fail to verify with an earlier version of
1454     OpenSSL 1.1.0. A workaround is to only verify the canonicalised text data
1455     and use the "-binary" flag (for the "cms" command line application) or set
1456     the SMIME_BINARY/PKCS7_BINARY/CMS_BINARY flags (if using CMS_verify()).
1457     [Matt Caswell]
1458
1459 Changes between 1.1.0g and 1.1.0h [27 Mar 2018]
1460
1461  *) Constructed ASN.1 types with a recursive definition could exceed the stack
1462
1463     Constructed ASN.1 types with a recursive definition (such as can be found
1464     in PKCS7) could eventually exceed the stack given malicious input with
1465     excessive recursion. This could result in a Denial Of Service attack. There
1466     are no such structures used within SSL/TLS that come from untrusted sources
1467     so this is considered safe.
1468
1469     This issue was reported to OpenSSL on 4th January 2018 by the OSS-fuzz
1470     project.
1471     (CVE-2018-0739)
1472     [Matt Caswell]
1473
1474  *) Incorrect CRYPTO_memcmp on HP-UX PA-RISC
1475
1476     Because of an implementation bug the PA-RISC CRYPTO_memcmp function is
1477     effectively reduced to only comparing the least significant bit of each
1478     byte. This allows an attacker to forge messages that would be considered as
1479     authenticated in an amount of tries lower than that guaranteed by the
1480     security claims of the scheme. The module can only be compiled by the
1481     HP-UX assembler, so that only HP-UX PA-RISC targets are affected.
1482
1483     This issue was reported to OpenSSL on 2nd March 2018 by Peter Waltenberg
1484     (IBM).
1485     (CVE-2018-0733)
1486     [Andy Polyakov]
1487
1488  *) Add a build target 'build_all_generated', to build all generated files
1489     and only that.  This can be used to prepare everything that requires
1490     things like perl for a system that lacks perl and then move everything
1491     to that system and do the rest of the build there.
1492     [Richard Levitte]
1493
1494  *) Backport SSL_OP_NO_RENGOTIATION
1495
1496     OpenSSL 1.0.2 and below had the ability to disable renegotiation using the
1497     (undocumented) SSL3_FLAGS_NO_RENEGOTIATE_CIPHERS flag. Due to the opacity
1498     changes this is no longer possible in 1.1.0. Therefore the new
1499     SSL_OP_NO_RENEGOTIATION option from 1.1.1-dev has been backported to
1500     1.1.0 to provide equivalent functionality.
1501
1502     Note that if an application built against 1.1.0h headers (or above) is run
1503     using an older version of 1.1.0 (prior to 1.1.0h) then the option will be
1504     accepted but nothing will happen, i.e. renegotiation will not be prevented.
1505     [Matt Caswell]
1506
1507  *) Removed the OS390-Unix config target.  It relied on a script that doesn't
1508     exist.
1509     [Rich Salz]
1510
1511  *) rsaz_1024_mul_avx2 overflow bug on x86_64
1512
1513     There is an overflow bug in the AVX2 Montgomery multiplication procedure
1514     used in exponentiation with 1024-bit moduli. No EC algorithms are affected.
1515     Analysis suggests that attacks against RSA and DSA as a result of this
1516     defect would be very difficult to perform and are not believed likely.
1517     Attacks against DH1024 are considered just feasible, because most of the
1518     work necessary to deduce information about a private key may be performed
1519     offline. The amount of resources required for such an attack would be
1520     significant. However, for an attack on TLS to be meaningful, the server
1521     would have to share the DH1024 private key among multiple clients, which is
1522     no longer an option since CVE-2016-0701.
1523
1524     This only affects processors that support the AVX2 but not ADX extensions
1525     like Intel Haswell (4th generation).
1526
1527     This issue was reported to OpenSSL by David Benjamin (Google). The issue
1528     was originally found via the OSS-Fuzz project.
1529     (CVE-2017-3738)
1530     [Andy Polyakov]
1531
1532 Changes between 1.1.0f and 1.1.0g [2 Nov 2017]
1533
1534  *) bn_sqrx8x_internal carry bug on x86_64
1535
1536     There is a carry propagating bug in the x86_64 Montgomery squaring
1537     procedure. No EC algorithms are affected. Analysis suggests that attacks
1538     against RSA and DSA as a result of this defect would be very difficult to
1539     perform and are not believed likely. Attacks against DH are considered just
1540     feasible (although very difficult) because most of the work necessary to
1541     deduce information about a private key may be performed offline. The amount
1542     of resources required for such an attack would be very significant and
1543     likely only accessible to a limited number of attackers. An attacker would
1544     additionally need online access to an unpatched system using the target
1545     private key in a scenario with persistent DH parameters and a private
1546     key that is shared between multiple clients.
1547
1548     This only affects processors that support the BMI1, BMI2 and ADX extensions
1549     like Intel Broadwell (5th generation) and later or AMD Ryzen.
1550
1551     This issue was reported to OpenSSL by the OSS-Fuzz project.
1552     (CVE-2017-3736)
1553     [Andy Polyakov]
1554
1555  *) Malformed X.509 IPAddressFamily could cause OOB read
1556
1557     If an X.509 certificate has a malformed IPAddressFamily extension,
1558     OpenSSL could do a one-byte buffer overread. The most likely result
1559     would be an erroneous display of the certificate in text format.
1560
1561     This issue was reported to OpenSSL by the OSS-Fuzz project.
1562     (CVE-2017-3735)
1563     [Rich Salz]
1564
1565 Changes between 1.1.0e and 1.1.0f [25 May 2017]
1566
1567  *) Have 'config' recognise 64-bit mingw and choose 'mingw64' as the target
1568     platform rather than 'mingw'.
1569     [Richard Levitte]
1570
1571  *) Remove the VMS-specific reimplementation of gmtime from crypto/o_times.c.
1572     VMS C's RTL has a fully up to date gmtime() and gmtime_r() since V7.1,
1573     which is the minimum version we support.
1574     [Richard Levitte]
1575
1576 Changes between 1.1.0d and 1.1.0e [16 Feb 2017]
1577
1578  *) Encrypt-Then-Mac renegotiation crash
1579
1580     During a renegotiation handshake if the Encrypt-Then-Mac extension is
1581     negotiated where it was not in the original handshake (or vice-versa) then
1582     this can cause OpenSSL to crash (dependant on ciphersuite). Both clients
1583     and servers are affected.
1584
1585     This issue was reported to OpenSSL by Joe Orton (Red Hat).
1586     (CVE-2017-3733)
1587     [Matt Caswell]
1588
1589 Changes between 1.1.0c and 1.1.0d [26 Jan 2017]
1590
1591  *) Truncated packet could crash via OOB read
1592
1593     If one side of an SSL/TLS path is running on a 32-bit host and a specific
1594     cipher is being used, then a truncated packet can cause that host to
1595     perform an out-of-bounds read, usually resulting in a crash.
1596
1597     This issue was reported to OpenSSL by Robert Święcki of Google.
1598     (CVE-2017-3731)
1599     [Andy Polyakov]
1600
1601  *) Bad (EC)DHE parameters cause a client crash
1602
1603     If a malicious server supplies bad parameters for a DHE or ECDHE key
1604     exchange then this can result in the client attempting to dereference a
1605     NULL pointer leading to a client crash. This could be exploited in a Denial
1606     of Service attack.
1607
1608     This issue was reported to OpenSSL by Guido Vranken.
1609     (CVE-2017-3730)
1610     [Matt Caswell]
1611
1612  *) BN_mod_exp may produce incorrect results on x86_64
1613
1614     There is a carry propagating bug in the x86_64 Montgomery squaring
1615     procedure. No EC algorithms are affected. Analysis suggests that attacks
1616     against RSA and DSA as a result of this defect would be very difficult to
1617     perform and are not believed likely. Attacks against DH are considered just
1618     feasible (although very difficult) because most of the work necessary to
1619     deduce information about a private key may be performed offline. The amount
1620     of resources required for such an attack would be very significant and
1621     likely only accessible to a limited number of attackers. An attacker would
1622     additionally need online access to an unpatched system using the target
1623     private key in a scenario with persistent DH parameters and a private
1624     key that is shared between multiple clients. For example this can occur by
1625     default in OpenSSL DHE based SSL/TLS ciphersuites. Note: This issue is very
1626     similar to CVE-2015-3193 but must be treated as a separate problem.
1627
1628     This issue was reported to OpenSSL by the OSS-Fuzz project.
1629     (CVE-2017-3732)
1630     [Andy Polyakov]
1631
1632 Changes between 1.1.0b and 1.1.0c [10 Nov 2016]
1633
1634  *) ChaCha20/Poly1305 heap-buffer-overflow
1635
1636     TLS connections using *-CHACHA20-POLY1305 ciphersuites are susceptible to
1637     a DoS attack by corrupting larger payloads. This can result in an OpenSSL
1638     crash. This issue is not considered to be exploitable beyond a DoS.
1639
1640     This issue was reported to OpenSSL by Robert Święcki (Google Security Team)
1641     (CVE-2016-7054)
1642     [Richard Levitte]
1643
1644  *) CMS Null dereference
1645
1646     Applications parsing invalid CMS structures can crash with a NULL pointer
1647     dereference. This is caused by a bug in the handling of the ASN.1 CHOICE
1648     type in OpenSSL 1.1.0 which can result in a NULL value being passed to the
1649     structure callback if an attempt is made to free certain invalid encodings.
1650     Only CHOICE structures using a callback which do not handle NULL value are
1651     affected.
1652
1653     This issue was reported to OpenSSL by Tyler Nighswander of ForAllSecure.
1654     (CVE-2016-7053)
1655     [Stephen Henson]
1656
1657  *) Montgomery multiplication may produce incorrect results
1658
1659     There is a carry propagating bug in the Broadwell-specific Montgomery
1660     multiplication procedure that handles input lengths divisible by, but
1661     longer than 256 bits. Analysis suggests that attacks against RSA, DSA
1662     and DH private keys are impossible. This is because the subroutine in
1663     question is not used in operations with the private key itself and an input
1664     of the attacker's direct choice. Otherwise the bug can manifest itself as
1665     transient authentication and key negotiation failures or reproducible
1666     erroneous outcome of public-key operations with specially crafted input.
1667     Among EC algorithms only Brainpool P-512 curves are affected and one
1668     presumably can attack ECDH key negotiation. Impact was not analyzed in
1669     detail, because pre-requisites for attack are considered unlikely. Namely
1670     multiple clients have to choose the curve in question and the server has to
1671     share the private key among them, neither of which is default behaviour.
1672     Even then only clients that chose the curve will be affected.
1673
1674     This issue was publicly reported as transient failures and was not
1675     initially recognized as a security issue. Thanks to Richard Morgan for
1676     providing reproducible case.
1677     (CVE-2016-7055)
1678     [Andy Polyakov]
1679
1680  *) Removed automatic addition of RPATH in shared libraries and executables,
1681     as this was a remainder from OpenSSL 1.0.x and isn't needed any more.
1682     [Richard Levitte]
1683
1684 Changes between 1.1.0a and 1.1.0b [26 Sep 2016]
1685
1686  *) Fix Use After Free for large message sizes
1687
1688     The patch applied to address CVE-2016-6307 resulted in an issue where if a
1689     message larger than approx 16k is received then the underlying buffer to
1690     store the incoming message is reallocated and moved. Unfortunately a
1691     dangling pointer to the old location is left which results in an attempt to
1692     write to the previously freed location. This is likely to result in a
1693     crash, however it could potentially lead to execution of arbitrary code.
1694
1695     This issue only affects OpenSSL 1.1.0a.
1696
1697     This issue was reported to OpenSSL by Robert Święcki.
1698     (CVE-2016-6309)
1699     [Matt Caswell]
1700
1701 Changes between 1.1.0 and 1.1.0a [22 Sep 2016]
1702
1703  *) OCSP Status Request extension unbounded memory growth
1704
1705     A malicious client can send an excessively large OCSP Status Request
1706     extension. If that client continually requests renegotiation, sending a
1707     large OCSP Status Request extension each time, then there will be unbounded
1708     memory growth on the server. This will eventually lead to a Denial Of
1709     Service attack through memory exhaustion. Servers with a default
1710     configuration are vulnerable even if they do not support OCSP. Builds using
1711     the "no-ocsp" build time option are not affected.
1712
1713     This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.)
1714     (CVE-2016-6304)
1715     [Matt Caswell]
1716
1717  *) SSL_peek() hang on empty record
1718
1719     OpenSSL 1.1.0 SSL/TLS will hang during a call to SSL_peek() if the peer
1720     sends an empty record. This could be exploited by a malicious peer in a
1721     Denial Of Service attack.
1722
1723     This issue was reported to OpenSSL by Alex Gaynor.
1724     (CVE-2016-6305)
1725     [Matt Caswell]
1726
1727  *) Excessive allocation of memory in tls_get_message_header() and
1728     dtls1_preprocess_fragment()
1729
1730     A (D)TLS message includes 3 bytes for its length in the header for the
1731     message. This would allow for messages up to 16Mb in length. Messages of
1732     this length are excessive and OpenSSL includes a check to ensure that a
1733     peer is sending reasonably sized messages in order to avoid too much memory
1734     being consumed to service a connection. A flaw in the logic of version
1735     1.1.0 means that memory for the message is allocated too early, prior to
1736     the excessive message length check. Due to way memory is allocated in
1737     OpenSSL this could mean an attacker could force up to 21Mb to be allocated
1738     to service a connection. This could lead to a Denial of Service through
1739     memory exhaustion. However, the excessive message length check still takes
1740     place, and this would cause the connection to immediately fail. Assuming
1741     that the application calls SSL_free() on the failed connection in a timely
1742     manner then the 21Mb of allocated memory will then be immediately freed
1743     again. Therefore the excessive memory allocation will be transitory in
1744     nature. This then means that there is only a security impact if:
1745
1746     1) The application does not call SSL_free() in a timely manner in the event
1747     that the connection fails
1748     or
1749     2) The application is working in a constrained environment where there is
1750     very little free memory
1751     or
1752     3) The attacker initiates multiple connection attempts such that there are
1753     multiple connections in a state where memory has been allocated for the
1754     connection; SSL_free() has not yet been called; and there is insufficient
1755     memory to service the multiple requests.
1756
1757     Except in the instance of (1) above any Denial Of Service is likely to be
1758     transitory because as soon as the connection fails the memory is
1759     subsequently freed again in the SSL_free() call. However there is an
1760     increased risk during this period of application crashes due to the lack of
1761     memory - which would then mean a more serious Denial of Service.
1762
1763     This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.)
1764     (CVE-2016-6307 and CVE-2016-6308)
1765     [Matt Caswell]
1766
1767  *) solaris-x86-cc, i.e. 32-bit configuration with vendor compiler,
1768     had to be removed. Primary reason is that vendor assembler can't
1769     assemble our modules with -KPIC flag. As result it, assembly
1770     support, was not even available as option. But its lack means
1771     lack of side-channel resistant code, which is incompatible with
1772     security by todays standards. Fortunately gcc is readily available
1773     prepackaged option, which we firmly point at...
1774     [Andy Polyakov]
1775
1776 Changes between 1.0.2h and 1.1.0  [25 Aug 2016]
1777
1778  *) Windows command-line tool supports UTF-8 opt-in option for arguments
1779     and console input. Setting OPENSSL_WIN32_UTF8 environment variable
1780     (to any value) allows Windows user to access PKCS#12 file generated
1781     with Windows CryptoAPI and protected with non-ASCII password, as well
1782     as files generated under UTF-8 locale on Linux also protected with
1783     non-ASCII password.
1784     [Andy Polyakov]
1785
1786  *) To mitigate the SWEET32 attack (CVE-2016-2183), 3DES cipher suites
1787     have been disabled by default and removed from DEFAULT, just like RC4.
1788     See the RC4 item below to re-enable both.
1789     [Rich Salz]
1790
1791  *) The method for finding the storage location for the Windows RAND seed file
1792     has changed. First we check %RANDFILE%. If that is not set then we check
1793     the directories %HOME%, %USERPROFILE% and %SYSTEMROOT% in that order. If
1794     all else fails we fall back to C:\.
1795     [Matt Caswell]
1796
1797  *) The EVP_EncryptUpdate() function has had its return type changed from void
1798     to int. A return of 0 indicates and error while a return of 1 indicates
1799     success.
1800     [Matt Caswell]
1801
1802  *) The flags RSA_FLAG_NO_CONSTTIME, DSA_FLAG_NO_EXP_CONSTTIME and
1803     DH_FLAG_NO_EXP_CONSTTIME which previously provided the ability to switch
1804     off the constant time implementation for RSA, DSA and DH have been made
1805     no-ops and deprecated.
1806     [Matt Caswell]
1807
1808  *) Windows RAND implementation was simplified to only get entropy by
1809     calling CryptGenRandom(). Various other RAND-related tickets
1810     were also closed.
1811     [Joseph Wylie Yandle, Rich Salz]
1812
1813  *) The stack and lhash API's were renamed to start with OPENSSL_SK_
1814     and OPENSSL_LH_, respectively.  The old names are available
1815     with API compatibility.  They new names are now completely documented.
1816     [Rich Salz]
1817
1818  *) Unify TYPE_up_ref(obj) methods signature.
1819     SSL_CTX_up_ref(), SSL_up_ref(), X509_up_ref(), EVP_PKEY_up_ref(),
1820     X509_CRL_up_ref(), X509_OBJECT_up_ref_count() methods are now returning an
1821     int (instead of void) like all others TYPE_up_ref() methods.
1822     So now these methods also check the return value of CRYPTO_atomic_add(),
1823     and the validity of object reference counter.
1824     [fdasilvayy@gmail.com]
1825
1826  *) With Windows Visual Studio builds, the .pdb files are installed
1827     alongside the installed libraries and executables.  For a static
1828     library installation, ossl_static.pdb is the associate compiler
1829     generated .pdb file to be used when linking programs.
1830     [Richard Levitte]
1831
1832  *) Remove openssl.spec.  Packaging files belong with the packagers.
1833     [Richard Levitte]
1834
1835  *) Automatic Darwin/OSX configuration has had a refresh, it will now
1836     recognise x86_64 architectures automatically.  You can still decide
1837     to build for a different bitness with the environment variable
1838     KERNEL_BITS (can be 32 or 64), for example:
1839
1840         KERNEL_BITS=32 ./config
1841
1842     [Richard Levitte]
1843
1844  *) Change default algorithms in pkcs8 utility to use PKCS#5 v2.0,
1845     256 bit AES and HMAC with SHA256.
1846     [Steve Henson]
1847
1848  *) Remove support for MIPS o32 ABI on IRIX (and IRIX only).
1849     [Andy Polyakov]
1850
1851  *) Triple-DES ciphers have been moved from HIGH to MEDIUM.
1852     [Rich Salz]
1853
1854  *) To enable users to have their own config files and build file templates,
1855     Configure looks in the directory indicated by the environment variable
1856     OPENSSL_LOCAL_CONFIG_DIR as well as the in-source Configurations/
1857     directory.  On VMS, OPENSSL_LOCAL_CONFIG_DIR is expected to be a logical
1858     name and is used as is.
1859     [Richard Levitte]
1860
1861  *) The following datatypes were made opaque: X509_OBJECT, X509_STORE_CTX,
1862     X509_STORE, X509_LOOKUP, and X509_LOOKUP_METHOD.  The unused type
1863     X509_CERT_FILE_CTX was removed.
1864     [Rich Salz]
1865
1866  *) "shared" builds are now the default. To create only static libraries use
1867     the "no-shared" Configure option.
1868     [Matt Caswell]
1869
1870  *) Remove the no-aes, no-hmac, no-rsa, no-sha and no-md5 Configure options.
1871     All of these option have not worked for some while and are fundamental
1872     algorithms.
1873     [Matt Caswell]
1874
1875  *) Make various cleanup routines no-ops and mark them as deprecated. Most
1876     global cleanup functions are no longer required because they are handled
1877     via auto-deinit (see OPENSSL_init_crypto and OPENSSL_init_ssl man pages).
1878     Explicitly de-initing can cause problems (e.g. where a library that uses
1879     OpenSSL de-inits, but an application is still using it). The affected
1880     functions are CONF_modules_free(), ENGINE_cleanup(), OBJ_cleanup(),
1881     EVP_cleanup(), BIO_sock_cleanup(), CRYPTO_cleanup_all_ex_data(),
1882     RAND_cleanup(), SSL_COMP_free_compression_methods(), ERR_free_strings() and
1883     COMP_zlib_cleanup().
1884     [Matt Caswell]
1885
1886  *) --strict-warnings no longer enables runtime debugging options
1887     such as REF_DEBUG. Instead, debug options are automatically
1888     enabled with '--debug' builds.
1889     [Andy Polyakov, Emilia Käsper]
1890
1891  *) Made DH and DH_METHOD opaque. The structures for managing DH objects
1892     have been moved out of the public header files. New functions for managing
1893     these have been added.
1894     [Matt Caswell]
1895
1896  *) Made RSA and RSA_METHOD opaque. The structures for managing RSA
1897     objects have been moved out of the public header files. New
1898     functions for managing these have been added.
1899     [Richard Levitte]
1900
1901  *) Made DSA and DSA_METHOD opaque. The structures for managing DSA objects
1902     have been moved out of the public header files. New functions for managing
1903     these have been added.
1904     [Matt Caswell]
1905
1906  *) Made BIO and BIO_METHOD opaque. The structures for managing BIOs have been
1907     moved out of the public header files. New functions for managing these
1908     have been added.
1909     [Matt Caswell]
1910
1911  *) Removed no-rijndael as a config option. Rijndael is an old name for AES.
1912     [Matt Caswell]
1913
1914  *) Removed the mk1mf build scripts.
1915     [Richard Levitte]
1916
1917  *) Headers are now wrapped, if necessary, with OPENSSL_NO_xxx, so
1918     it is always safe to #include a header now.
1919     [Rich Salz]
1920
1921  *) Removed the aged BC-32 config and all its supporting scripts
1922     [Richard Levitte]
1923
1924  *) Removed support for Ultrix, Netware, and OS/2.
1925     [Rich Salz]
1926
1927  *) Add support for HKDF.
1928     [Alessandro Ghedini]
1929
1930  *) Add support for blake2b and blake2s
1931     [Bill Cox]
1932
1933  *) Added support for "pipelining". Ciphers that have the
1934     EVP_CIPH_FLAG_PIPELINE flag set have a capability to process multiple
1935     encryptions/decryptions simultaneously. There are currently no built-in
1936     ciphers with this property but the expectation is that engines will be able
1937     to offer it to significantly improve throughput. Support has been extended
1938     into libssl so that multiple records for a single connection can be
1939     processed in one go (for >=TLS 1.1).
1940     [Matt Caswell]
1941
1942  *) Added the AFALG engine. This is an async capable engine which is able to
1943     offload work to the Linux kernel. In this initial version it only supports
1944     AES128-CBC. The kernel must be version 4.1.0 or greater.
1945     [Catriona Lucey]
1946
1947  *) OpenSSL now uses a new threading API. It is no longer necessary to
1948     set locking callbacks to use OpenSSL in a multi-threaded environment. There
1949     are two supported threading models: pthreads and windows threads. It is
1950     also possible to configure OpenSSL at compile time for "no-threads". The
1951     old threading API should no longer be used. The functions have been
1952     replaced with "no-op" compatibility macros.
1953     [Alessandro Ghedini, Matt Caswell]
1954
1955  *) Modify behavior of ALPN to invoke callback after SNI/servername
1956     callback, such that updates to the SSL_CTX affect ALPN.
1957     [Todd Short]
1958
1959  *) Add SSL_CIPHER queries for authentication and key-exchange.
1960     [Todd Short]
1961
1962  *) Changes to the DEFAULT cipherlist:
1963       - Prefer (EC)DHE handshakes over plain RSA.
1964       - Prefer AEAD ciphers over legacy ciphers.
1965       - Prefer ECDSA over RSA when both certificates are available.
1966       - Prefer TLSv1.2 ciphers/PRF.
1967       - Remove DSS, SEED, IDEA, CAMELLIA, and AES-CCM from the
1968         default cipherlist.
1969     [Emilia Käsper]
1970
1971  *) Change the ECC default curve list to be this, in order: x25519,
1972     secp256r1, secp521r1, secp384r1.
1973     [Rich Salz]
1974
1975  *) RC4 based libssl ciphersuites are now classed as "weak" ciphers and are
1976     disabled by default. They can be re-enabled using the
1977     enable-weak-ssl-ciphers option to Configure.
1978     [Matt Caswell]
1979
1980  *) If the server has ALPN configured, but supports no protocols that the
1981     client advertises, send a fatal "no_application_protocol" alert.
1982     This behaviour is SHALL in RFC 7301, though it isn't universally
1983     implemented by other servers.
1984     [Emilia Käsper]
1985
1986  *) Add X25519 support.
1987     Add ASN.1 and EVP_PKEY methods for X25519. This includes support
1988     for public and private key encoding using the format documented in
1989     draft-ietf-curdle-pkix-02. The corresponding EVP_PKEY method supports
1990     key generation and key derivation.
1991
1992     TLS support complies with draft-ietf-tls-rfc4492bis-08 and uses
1993     X25519(29).
1994     [Steve Henson]
1995
1996  *) Deprecate SRP_VBASE_get_by_user.
1997     SRP_VBASE_get_by_user had inconsistent memory management behaviour.
1998     In order to fix an unavoidable memory leak (CVE-2016-0798),
1999     SRP_VBASE_get_by_user was changed to ignore the "fake user" SRP
2000     seed, even if the seed is configured.
2001
2002     Users should use SRP_VBASE_get1_by_user instead. Note that in
2003     SRP_VBASE_get1_by_user, caller must free the returned value. Note
2004     also that even though configuring the SRP seed attempts to hide
2005     invalid usernames by continuing the handshake with fake
2006     credentials, this behaviour is not constant time and no strong
2007     guarantees are made that the handshake is indistinguishable from
2008     that of a valid user.
2009     [Emilia Käsper]
2010
2011  *) Configuration change; it's now possible to build dynamic engines
2012     without having to build shared libraries and vice versa.  This
2013     only applies to the engines in engines/, those in crypto/engine/
2014     will always be built into libcrypto (i.e. "static").
2015
2016     Building dynamic engines is enabled by default; to disable, use
2017     the configuration option "disable-dynamic-engine".
2018
2019     The only requirements for building dynamic engines are the
2020     presence of the DSO module and building with position independent
2021     code, so they will also automatically be disabled if configuring
2022     with "disable-dso" or "disable-pic".
2023
2024     The macros OPENSSL_NO_STATIC_ENGINE and OPENSSL_NO_DYNAMIC_ENGINE
2025     are also taken away from openssl/opensslconf.h, as they are
2026     irrelevant.
2027     [Richard Levitte]
2028
2029  *) Configuration change; if there is a known flag to compile
2030     position independent code, it will always be applied on the
2031     libcrypto and libssl object files, and never on the application
2032     object files.  This means other libraries that use routines from
2033     libcrypto / libssl can be made into shared libraries regardless
2034     of how OpenSSL was configured.
2035
2036     If this isn't desirable, the configuration options "disable-pic"
2037     or "no-pic" can be used to disable the use of PIC.  This will
2038     also disable building shared libraries and dynamic engines.
2039     [Richard Levitte]
2040
2041  *) Removed JPAKE code.  It was experimental and has no wide use.
2042     [Rich Salz]
2043
2044  *) The INSTALL_PREFIX Makefile variable has been renamed to
2045     DESTDIR.  That makes for less confusion on what this variable
2046     is for.  Also, the configuration option --install_prefix is
2047     removed.
2048     [Richard Levitte]
2049
2050  *) Heartbeat for TLS has been removed and is disabled by default
2051     for DTLS; configure with enable-heartbeats.  Code that uses the
2052     old #define's might need to be updated.
2053     [Emilia Käsper, Rich Salz]
2054
2055  *) Rename REF_CHECK to REF_DEBUG.
2056     [Rich Salz]
2057
2058  *) New "unified" build system
2059
2060     The "unified" build system is aimed to be a common system for all
2061     platforms we support.  With it comes new support for VMS.
2062
2063     This system builds supports building in a different directory tree
2064     than the source tree.  It produces one Makefile (for unix family
2065     or lookalikes), or one descrip.mms (for VMS).
2066
2067     The source of information to make the Makefile / descrip.mms is
2068     small files called 'build.info', holding the necessary
2069     information for each directory with source to compile, and a
2070     template in Configurations, like unix-Makefile.tmpl or
2071     descrip.mms.tmpl.
2072
2073     With this change, the library names were also renamed on Windows
2074     and on VMS.  They now have names that are closer to the standard
2075     on Unix, and include the major version number, and in certain
2076     cases, the architecture they are built for.  See "Notes on shared
2077     libraries" in INSTALL.
2078
2079     We rely heavily on the perl module Text::Template.
2080     [Richard Levitte]
2081
2082  *) Added support for auto-initialisation and de-initialisation of the library.
2083     OpenSSL no longer requires explicit init or deinit routines to be called,
2084     except in certain circumstances. See the OPENSSL_init_crypto() and
2085     OPENSSL_init_ssl() man pages for further information.
2086     [Matt Caswell]
2087
2088  *) The arguments to the DTLSv1_listen function have changed. Specifically the
2089     "peer" argument is now expected to be a BIO_ADDR object.
2090
2091  *) Rewrite of BIO networking library. The BIO library lacked consistent
2092     support of IPv6, and adding it required some more extensive
2093     modifications.  This introduces the BIO_ADDR and BIO_ADDRINFO types,
2094     which hold all types of addresses and chains of address information.
2095     It also introduces a new API, with functions like BIO_socket,
2096     BIO_connect, BIO_listen, BIO_lookup and a rewrite of BIO_accept.
2097     The source/sink BIOs BIO_s_connect, BIO_s_accept and BIO_s_datagram
2098     have been adapted accordingly.
2099     [Richard Levitte]
2100
2101  *) RSA_padding_check_PKCS1_type_1 now accepts inputs with and without
2102     the leading 0-byte.
2103     [Emilia Käsper]
2104
2105  *) CRIME protection: disable compression by default, even if OpenSSL is
2106     compiled with zlib enabled. Applications can still enable compression
2107     by calling SSL_CTX_clear_options(ctx, SSL_OP_NO_COMPRESSION), or by
2108     using the SSL_CONF library to configure compression.
2109     [Emilia Käsper]
2110
2111  *) The signature of the session callback configured with
2112     SSL_CTX_sess_set_get_cb was changed. The read-only input buffer
2113     was explicitly marked as 'const unsigned char*' instead of
2114     'unsigned char*'.
2115     [Emilia Käsper]
2116
2117  *) Always DPURIFY. Remove the use of uninitialized memory in the
2118     RNG, and other conditional uses of DPURIFY. This makes -DPURIFY a no-op.
2119     [Emilia Käsper]
2120
2121  *) Removed many obsolete configuration items, including
2122        DES_PTR, DES_RISC1, DES_RISC2, DES_INT
2123        MD2_CHAR, MD2_INT, MD2_LONG
2124        BF_PTR, BF_PTR2
2125        IDEA_SHORT, IDEA_LONG
2126        RC2_SHORT, RC2_LONG, RC4_LONG, RC4_CHUNK, RC4_INDEX
2127     [Rich Salz, with advice from Andy Polyakov]
2128
2129  *) Many BN internals have been moved to an internal header file.
2130     [Rich Salz with help from Andy Polyakov]
2131
2132  *) Configuration and writing out the results from it has changed.
2133     Files such as Makefile include/openssl/opensslconf.h and are now
2134     produced through general templates, such as Makefile.in and
2135     crypto/opensslconf.h.in and some help from the perl module
2136     Text::Template.
2137
2138     Also, the center of configuration information is no longer
2139     Makefile.  Instead, Configure produces a perl module in
2140     configdata.pm which holds most of the config data (in the hash
2141     table %config), the target data that comes from the target
2142     configuration in one of the Configurations/*.conf files (in
2143     %target).
2144     [Richard Levitte]
2145
2146  *) To clarify their intended purposes, the Configure options
2147     --prefix and --openssldir change their semantics, and become more
2148     straightforward and less interdependent.
2149
2150     --prefix shall be used exclusively to give the location INSTALLTOP
2151     where programs, scripts, libraries, include files and manuals are
2152     going to be installed.  The default is now /usr/local.
2153
2154     --openssldir shall be used exclusively to give the default
2155     location OPENSSLDIR where certificates, private keys, CRLs are
2156     managed.  This is also where the default openssl.cnf gets
2157     installed.
2158     If the directory given with this option is a relative path, the
2159     values of both the --prefix value and the --openssldir value will
2160     be combined to become OPENSSLDIR.
2161     The default for --openssldir is INSTALLTOP/ssl.
2162
2163     Anyone who uses --openssldir to specify where OpenSSL is to be
2164     installed MUST change to use --prefix instead.
2165     [Richard Levitte]
2166
2167  *) The GOST engine was out of date and therefore it has been removed. An up
2168     to date GOST engine is now being maintained in an external repository.
2169     See: https://wiki.openssl.org/index.php/Binaries. Libssl still retains
2170     support for GOST ciphersuites (these are only activated if a GOST engine
2171     is present).
2172     [Matt Caswell]
2173
2174  *) EGD is no longer supported by default; use enable-egd when
2175     configuring.
2176     [Ben Kaduk and Rich Salz]
2177
2178  *) The distribution now has Makefile.in files, which are used to
2179     create Makefile's when Configure is run.  *Configure must be run
2180     before trying to build now.*
2181     [Rich Salz]
2182
2183  *) The return value for SSL_CIPHER_description() for error conditions
2184     has changed.
2185     [Rich Salz]
2186
2187  *) Support for RFC6698/RFC7671 DANE TLSA peer authentication.
2188
2189     Obtaining and performing DNSSEC validation of TLSA records is
2190     the application's responsibility.  The application provides
2191     the TLSA records of its choice to OpenSSL, and these are then
2192     used to authenticate the peer.
2193
2194     The TLSA records need not even come from DNS.  They can, for
2195     example, be used to implement local end-entity certificate or
2196     trust-anchor "pinning", where the "pin" data takes the form
2197     of TLSA records, which can augment or replace verification
2198     based on the usual WebPKI public certification authorities.
2199     [Viktor Dukhovni]
2200
2201  *) Revert default OPENSSL_NO_DEPRECATED setting.  Instead OpenSSL
2202     continues to support deprecated interfaces in default builds.
2203     However, applications are strongly advised to compile their
2204     source files with -DOPENSSL_API_COMPAT=0x10100000L, which hides
2205     the declarations of all interfaces deprecated in 0.9.8, 1.0.0
2206     or the 1.1.0 releases.
2207
2208     In environments in which all applications have been ported to
2209     not use any deprecated interfaces OpenSSL's Configure script
2210     should be used with the --api=1.1.0 option to entirely remove
2211     support for the deprecated features from the library and
2212     unconditionally disable them in the installed headers.
2213     Essentially the same effect can be achieved with the "no-deprecated"
2214     argument to Configure, except that this will always restrict
2215     the build to just the latest API, rather than a fixed API
2216     version.
2217
2218     As applications are ported to future revisions of the API,
2219     they should update their compile-time OPENSSL_API_COMPAT define
2220     accordingly, but in most cases should be able to continue to
2221     compile with later releases.
2222
2223     The OPENSSL_API_COMPAT versions for 1.0.0, and 0.9.8 are
2224     0x10000000L and 0x00908000L, respectively.  However those
2225     versions did not support the OPENSSL_API_COMPAT feature, and
2226     so applications are not typically tested for explicit support
2227     of just the undeprecated features of either release.
2228     [Viktor Dukhovni]
2229
2230  *) Add support for setting the minimum and maximum supported protocol.
2231     It can bet set via the SSL_set_min_proto_version() and
2232     SSL_set_max_proto_version(), or via the SSL_CONF's MinProtocol and
2233     MaxProtocol.  It's recommended to use the new APIs to disable
2234     protocols instead of disabling individual protocols using
2235     SSL_set_options() or SSL_CONF's Protocol.  This change also
2236     removes support for disabling TLS 1.2 in the OpenSSL TLS
2237     client at compile time by defining OPENSSL_NO_TLS1_2_CLIENT.
2238     [Kurt Roeckx]
2239
2240  *) Support for ChaCha20 and Poly1305 added to libcrypto and libssl.
2241     [Andy Polyakov]
2242
2243  *) New EC_KEY_METHOD, this replaces the older ECDSA_METHOD and ECDH_METHOD
2244     and integrates ECDSA and ECDH functionality into EC. Implementations can
2245     now redirect key generation and no longer need to convert to or from
2246     ECDSA_SIG format.
2247
2248     Note: the ecdsa.h and ecdh.h headers are now no longer needed and just
2249     include the ec.h header file instead.
2250     [Steve Henson]
2251
2252  *) Remove support for all 40 and 56 bit ciphers.  This includes all the export
2253     ciphers who are no longer supported and drops support the ephemeral RSA key
2254     exchange. The LOW ciphers currently doesn't have any ciphers in it.
2255     [Kurt Roeckx]
2256
2257  *) Made EVP_MD_CTX, EVP_MD, EVP_CIPHER_CTX, EVP_CIPHER and HMAC_CTX
2258     opaque.  For HMAC_CTX, the following constructors and destructors
2259     were added:
2260
2261        HMAC_CTX *HMAC_CTX_new(void);
2262        void HMAC_CTX_free(HMAC_CTX *ctx);
2263
2264     For EVP_MD and EVP_CIPHER, complete APIs to create, fill and
2265     destroy such methods has been added.  See EVP_MD_meth_new(3) and
2266     EVP_CIPHER_meth_new(3) for documentation.
2267
2268     Additional changes:
2269     1) EVP_MD_CTX_cleanup(), EVP_CIPHER_CTX_cleanup() and
2270        HMAC_CTX_cleanup() were removed.  HMAC_CTX_reset() and
2271        EVP_MD_CTX_reset() should be called instead to reinitialise
2272        an already created structure.
2273     2) For consistency with the majority of our object creators and
2274        destructors, EVP_MD_CTX_(create|destroy) were renamed to
2275        EVP_MD_CTX_(new|free).  The old names are retained as macros
2276        for deprecated builds.
2277     [Richard Levitte]
2278
2279  *) Added ASYNC support. Libcrypto now includes the async sub-library to enable
2280     cryptographic operations to be performed asynchronously as long as an
2281     asynchronous capable engine is used. See the ASYNC_start_job() man page for
2282     further details. Libssl has also had this capability integrated with the
2283     introduction of the new mode SSL_MODE_ASYNC and associated error
2284     SSL_ERROR_WANT_ASYNC. See the SSL_CTX_set_mode() and SSL_get_error() man
2285     pages. This work was developed in partnership with Intel Corp.
2286     [Matt Caswell]
2287
2288  *) SSL_{CTX_}set_ecdh_auto() has been removed and ECDH is support is
2289     always enabled now.  If you want to disable the support you should
2290     exclude it using the list of supported ciphers. This also means that the
2291     "-no_ecdhe" option has been removed from s_server.
2292     [Kurt Roeckx]
2293
2294  *) SSL_{CTX}_set_tmp_ecdh() which can set 1 EC curve now internally calls
2295     SSL_{CTX_}set1_curves() which can set a list.
2296     [Kurt Roeckx]
2297
2298  *) Remove support for SSL_{CTX_}set_tmp_ecdh_callback().  You should set the
2299     curve you want to support using SSL_{CTX_}set1_curves().
2300     [Kurt Roeckx]
2301
2302  *) State machine rewrite. The state machine code has been significantly
2303     refactored in order to remove much duplication of code and solve issues
2304     with the old code (see ssl/statem/README for further details). This change
2305     does have some associated API changes. Notably the SSL_state() function
2306     has been removed and replaced by SSL_get_state which now returns an
2307     "OSSL_HANDSHAKE_STATE" instead of an int. SSL_set_state() has been removed
2308     altogether. The previous handshake states defined in ssl.h and ssl3.h have
2309     also been removed.
2310     [Matt Caswell]
2311
2312  *) All instances of the string "ssleay" in the public API were replaced
2313     with OpenSSL (case-matching; e.g., OPENSSL_VERSION for #define's)
2314     Some error codes related to internal RSA_eay API's were renamed.
2315     [Rich Salz]
2316
2317  *) The demo files in crypto/threads were moved to demo/threads.
2318     [Rich Salz]
2319
2320  *) Removed obsolete engines: 4758cca, aep, atalla, cswift, nuron, gmp,
2321     sureware and ubsec.
2322     [Matt Caswell, Rich Salz]
2323
2324  *) New ASN.1 embed macro.
2325
2326     New ASN.1 macro ASN1_EMBED. This is the same as ASN1_SIMPLE except the
2327     structure is not allocated: it is part of the parent. That is instead of
2328
2329     FOO *x;
2330
2331     it must be:
2332
2333     FOO x;
2334
2335     This reduces memory fragmentation and make it impossible to accidentally
2336     set a mandatory field to NULL.
2337
2338     This currently only works for some fields specifically a SEQUENCE, CHOICE,
2339     or ASN1_STRING type which is part of a parent SEQUENCE. Since it is
2340     equivalent to ASN1_SIMPLE it cannot be tagged, OPTIONAL, SET OF or
2341     SEQUENCE OF.
2342     [Steve Henson]
2343
2344  *) Remove EVP_CHECK_DES_KEY, a compile-time option that never compiled.
2345     [Emilia Käsper]
2346
2347  *) Removed DES and RC4 ciphersuites from DEFAULT. Also removed RC2 although
2348     in 1.0.2 EXPORT was already removed and the only RC2 ciphersuite is also
2349     an EXPORT one. COMPLEMENTOFDEFAULT has been updated accordingly to add
2350     DES and RC4 ciphersuites.
2351     [Matt Caswell]
2352
2353  *) Rewrite EVP_DecodeUpdate (base64 decoding) to fix several bugs.
2354     This changes the decoding behaviour for some invalid messages,
2355     though the change is mostly in the more lenient direction, and
2356     legacy behaviour is preserved as much as possible.
2357     [Emilia Käsper]
2358
2359  *) Fix no-stdio build.
2360    [ David Woodhouse <David.Woodhouse@intel.com> and also
2361      Ivan Nestlerode <ivan.nestlerode@sonos.com> ]
2362
2363  *) New testing framework
2364     The testing framework has been largely rewritten and is now using
2365     perl and the perl modules Test::Harness and an extended variant of
2366     Test::More called OpenSSL::Test to do its work.  All test scripts in
2367     test/ have been rewritten into test recipes, and all direct calls to
2368     executables in test/Makefile have become individual recipes using the
2369     simplified testing OpenSSL::Test::Simple.
2370
2371     For documentation on our testing modules, do:
2372
2373        perldoc test/testlib/OpenSSL/Test/Simple.pm
2374        perldoc test/testlib/OpenSSL/Test.pm
2375
2376     [Richard Levitte]
2377
2378  *) Revamped memory debug; only -DCRYPTO_MDEBUG and -DCRYPTO_MDEBUG_ABORT
2379     are used; the latter aborts on memory leaks (usually checked on exit).
2380     Some undocumented "set malloc, etc., hooks" functions were removed
2381     and others were changed.  All are now documented.
2382     [Rich Salz]
2383
2384  *) In DSA_generate_parameters_ex, if the provided seed is too short,
2385     return an error
2386     [Rich Salz and Ismo Puustinen <ismo.puustinen@intel.com>]
2387
2388  *) Rewrite PSK to support ECDHE_PSK, DHE_PSK and RSA_PSK. Add ciphersuites
2389     from RFC4279, RFC4785, RFC5487, RFC5489.
2390
2391     Thanks to Christian J. Dietrich and Giuseppe D'Angelo for the
2392     original RSA_PSK patch.
2393     [Steve Henson]
2394
2395  *) Dropped support for the SSL3_FLAGS_DELAY_CLIENT_FINISHED flag. This SSLeay
2396     era flag was never set throughout the codebase (only read). Also removed
2397     SSL3_FLAGS_POP_BUFFER which was only used if
2398     SSL3_FLAGS_DELAY_CLIENT_FINISHED was also set.
2399     [Matt Caswell]
2400
2401  *) Changed the default name options in the "ca", "crl", "req" and "x509"
2402     to be "oneline" instead of "compat".
2403     [Richard Levitte]
2404
2405  *) Remove SSL_OP_TLS_BLOCK_PADDING_BUG. This is SSLeay legacy, we're
2406     not aware of clients that still exhibit this bug, and the workaround
2407     hasn't been working properly for a while.
2408     [Emilia Käsper]
2409
2410  *) The return type of BIO_number_read() and BIO_number_written() as well as
2411     the corresponding num_read and num_write members in the BIO structure has
2412     changed from unsigned long to uint64_t. On platforms where an unsigned
2413     long is 32 bits (e.g. Windows) these counters could overflow if >4Gb is
2414     transferred.
2415     [Matt Caswell]
2416
2417  *) Given the pervasive nature of TLS extensions it is inadvisable to run
2418     OpenSSL without support for them. It also means that maintaining
2419     the OPENSSL_NO_TLSEXT option within the code is very invasive (and probably
2420     not well tested). Therefore the OPENSSL_NO_TLSEXT option has been removed.
2421     [Matt Caswell]
2422
2423  *) Removed support for the two export grade static DH ciphersuites
2424     EXP-DH-RSA-DES-CBC-SHA and EXP-DH-DSS-DES-CBC-SHA. These two ciphersuites
2425     were newly added (along with a number of other static DH ciphersuites) to
2426     1.0.2. However the two export ones have *never* worked since they were
2427     introduced. It seems strange in any case to be adding new export
2428     ciphersuites, and given "logjam" it also does not seem correct to fix them.
2429     [Matt Caswell]
2430
2431  *) Version negotiation has been rewritten. In particular SSLv23_method(),
2432     SSLv23_client_method() and SSLv23_server_method() have been deprecated,
2433     and turned into macros which simply call the new preferred function names
2434     TLS_method(), TLS_client_method() and TLS_server_method(). All new code
2435     should use the new names instead. Also as part of this change the ssl23.h
2436     header file has been removed.
2437     [Matt Caswell]
2438
2439  *) Support for Kerberos ciphersuites in TLS (RFC2712) has been removed. This
2440     code and the associated standard is no longer considered fit-for-purpose.
2441     [Matt Caswell]
2442
2443  *) RT2547 was closed.  When generating a private key, try to make the
2444     output file readable only by the owner.  This behavior change might
2445     be noticeable when interacting with other software.
2446
2447  *) Documented all exdata functions.  Added CRYPTO_free_ex_index.
2448     Added a test.
2449     [Rich Salz]
2450
2451  *) Added HTTP GET support to the ocsp command.
2452     [Rich Salz]
2453
2454  *) Changed default digest for the dgst and enc commands from MD5 to
2455     sha256
2456     [Rich Salz]
2457
2458  *) RAND_pseudo_bytes has been deprecated. Users should use RAND_bytes instead.
2459     [Matt Caswell]
2460
2461  *) Added support for TLS extended master secret from
2462     draft-ietf-tls-session-hash-03.txt. Thanks for Alfredo Pironti for an
2463     initial patch which was a great help during development.
2464     [Steve Henson]
2465
2466  *) All libssl internal structures have been removed from the public header
2467     files, and the OPENSSL_NO_SSL_INTERN option has been removed (since it is
2468     now redundant). Users should not attempt to access internal structures
2469     directly. Instead they should use the provided API functions.
2470     [Matt Caswell]
2471
2472  *) config has been changed so that by default OPENSSL_NO_DEPRECATED is used.
2473     Access to deprecated functions can be re-enabled by running config with
2474     "enable-deprecated". In addition applications wishing to use deprecated
2475     functions must define OPENSSL_USE_DEPRECATED. Note that this new behaviour
2476     will, by default, disable some transitive includes that previously existed
2477     in the header files (e.g. ec.h will no longer, by default, include bn.h)
2478     [Matt Caswell]
2479
2480  *) Added support for OCB mode. OpenSSL has been granted a patent license
2481     compatible with the OpenSSL license for use of OCB. Details are available
2482     at https://www.openssl.org/source/OCB-patent-grant-OpenSSL.pdf. Support
2483     for OCB can be removed by calling config with no-ocb.
2484     [Matt Caswell]
2485
2486  *) SSLv2 support has been removed.  It still supports receiving a SSLv2
2487     compatible client hello.
2488     [Kurt Roeckx]
2489
2490  *) Increased the minimal RSA keysize from 256 to 512 bits [Rich Salz],
2491     done while fixing the error code for the key-too-small case.
2492     [Annie Yousar <a.yousar@informatik.hu-berlin.de>]
2493
2494  *) CA.sh has been removed; use CA.pl instead.
2495     [Rich Salz]
2496
2497  *) Removed old DES API.
2498     [Rich Salz]
2499
2500  *) Remove various unsupported platforms:
2501        Sony NEWS4
2502        BEOS and BEOS_R5
2503        NeXT
2504        SUNOS
2505        MPE/iX
2506        Sinix/ReliantUNIX RM400
2507        DGUX
2508        NCR
2509        Tandem
2510        Cray
2511        16-bit platforms such as WIN16
2512     [Rich Salz]
2513
2514  *) Clean up OPENSSL_NO_xxx #define's
2515        Use setbuf() and remove OPENSSL_NO_SETVBUF_IONBF
2516        Rename OPENSSL_SYSNAME_xxx to OPENSSL_SYS_xxx
2517        OPENSSL_NO_EC{DH,DSA} merged into OPENSSL_NO_EC
2518        OPENSSL_NO_RIPEMD160, OPENSSL_NO_RIPEMD merged into OPENSSL_NO_RMD160
2519        OPENSSL_NO_FP_API merged into OPENSSL_NO_STDIO
2520        Remove OPENSSL_NO_BIO OPENSSL_NO_BUFFER OPENSSL_NO_CHAIN_VERIFY
2521        OPENSSL_NO_EVP OPENSSL_NO_FIPS_ERR OPENSSL_NO_HASH_COMP
2522        OPENSSL_NO_LHASH OPENSSL_NO_OBJECT OPENSSL_NO_SPEED OPENSSL_NO_STACK
2523        OPENSSL_NO_X509 OPENSSL_NO_X509_VERIFY
2524        Remove MS_STATIC; it's a relic from platforms <32 bits.
2525     [Rich Salz]
2526
2527  *) Cleaned up dead code
2528        Remove all but one '#ifdef undef' which is to be looked at.
2529     [Rich Salz]
2530
2531  *) Clean up calling of xxx_free routines.
2532        Just like free(), fix most of the xxx_free routines to accept
2533        NULL.  Remove the non-null checks from callers.  Save much code.
2534     [Rich Salz]
2535
2536  *) Add secure heap for storage of private keys (when possible).
2537     Add BIO_s_secmem(), CBIGNUM, etc.
2538     Contributed by Akamai Technologies under our Corporate CLA.
2539     [Rich Salz]
2540
2541  *) Experimental support for a new, fast, unbiased prime candidate generator,
2542     bn_probable_prime_dh_coprime(). Not currently used by any prime generator.
2543     [Felix Laurie von Massenbach <felix@erbridge.co.uk>]
2544
2545  *) New output format NSS in the sess_id command line tool. This allows
2546     exporting the session id and the master key in NSS keylog format.
2547     [Martin Kaiser <martin@kaiser.cx>]
2548
2549  *) Harmonize version and its documentation. -f flag is used to display
2550     compilation flags.
2551     [mancha <mancha1@zoho.com>]
2552
2553  *) Fix eckey_priv_encode so it immediately returns an error upon a failure
2554     in i2d_ECPrivateKey.  Thanks to Ted Unangst for feedback on this issue.
2555     [mancha <mancha1@zoho.com>]
2556
2557  *) Fix some double frees. These are not thought to be exploitable.
2558     [mancha <mancha1@zoho.com>]
2559
2560  *) A missing bounds check in the handling of the TLS heartbeat extension
2561     can be used to reveal up to 64k of memory to a connected client or
2562     server.
2563
2564     Thanks for Neel Mehta of Google Security for discovering this bug and to
2565     Adam Langley <agl@chromium.org> and Bodo Moeller <bmoeller@acm.org> for
2566     preparing the fix (CVE-2014-0160)
2567     [Adam Langley, Bodo Moeller]
2568
2569  *) Fix for the attack described in the paper "Recovering OpenSSL
2570     ECDSA Nonces Using the FLUSH+RELOAD Cache Side-channel Attack"
2571     by Yuval Yarom and Naomi Benger. Details can be obtained from:
2572     http://eprint.iacr.org/2014/140
2573
2574     Thanks to Yuval Yarom and Naomi Benger for discovering this
2575     flaw and to Yuval Yarom for supplying a fix (CVE-2014-0076)
2576     [Yuval Yarom and Naomi Benger]
2577
2578  *) Use algorithm specific chains in SSL_CTX_use_certificate_chain_file():
2579     this fixes a limitation in previous versions of OpenSSL.
2580     [Steve Henson]
2581
2582  *) Experimental encrypt-then-mac support.
2583
2584     Experimental support for encrypt then mac from
2585     draft-gutmann-tls-encrypt-then-mac-02.txt
2586
2587     To enable it set the appropriate extension number (0x42 for the test
2588     server) using e.g. -DTLSEXT_TYPE_encrypt_then_mac=0x42
2589
2590     For non-compliant peers (i.e. just about everything) this should have no
2591     effect.
2592
2593     WARNING: EXPERIMENTAL, SUBJECT TO CHANGE.
2594
2595     [Steve Henson]
2596
2597  *) Add EVP support for key wrapping algorithms, to avoid problems with
2598     existing code the flag EVP_CIPHER_CTX_WRAP_ALLOW has to be set in
2599     the EVP_CIPHER_CTX or an error is returned. Add AES and DES3 wrap
2600     algorithms and include tests cases.
2601     [Steve Henson]
2602
2603  *) Extend CMS code to support RSA-PSS signatures and RSA-OAEP for
2604     enveloped data.
2605     [Steve Henson]
2606
2607  *) Extended RSA OAEP support via EVP_PKEY API. Options to specify digest,
2608     MGF1 digest and OAEP label.
2609     [Steve Henson]
2610
2611  *) Make openssl verify return errors.
2612     [Chris Palmer <palmer@google.com> and Ben Laurie]
2613
2614  *) New function ASN1_TIME_diff to calculate the difference between two
2615     ASN1_TIME structures or one structure and the current time.
2616     [Steve Henson]
2617
2618  *) Update fips_test_suite to support multiple command line options. New
2619     test to induce all self test errors in sequence and check expected
2620     failures.
2621     [Steve Henson]
2622
2623  *) Add FIPS_{rsa,dsa,ecdsa}_{sign,verify} functions which digest and
2624     sign or verify all in one operation.
2625     [Steve Henson]
2626
2627  *) Add fips_algvs: a multicall fips utility incorporating all the algorithm
2628     test programs and fips_test_suite. Includes functionality to parse
2629     the minimal script output of fipsalgest.pl directly.
2630     [Steve Henson]
2631
2632  *) Add authorisation parameter to FIPS_module_mode_set().
2633     [Steve Henson]
2634
2635  *) Add FIPS selftest for ECDH algorithm using P-224 and B-233 curves.
2636     [Steve Henson]
2637
2638  *) Use separate DRBG fields for internal and external flags. New function
2639     FIPS_drbg_health_check() to perform on demand health checking. Add
2640     generation tests to fips_test_suite with reduced health check interval to
2641     demonstrate periodic health checking. Add "nodh" option to
2642     fips_test_suite to skip very slow DH test.
2643     [Steve Henson]
2644
2645  *) New function FIPS_get_cipherbynid() to lookup FIPS supported ciphers
2646     based on NID.
2647     [Steve Henson]
2648
2649  *) More extensive health check for DRBG checking many more failure modes.
2650     New function FIPS_selftest_drbg_all() to handle every possible DRBG
2651     combination: call this in fips_test_suite.
2652     [Steve Henson]
2653
2654  *) Add support for canonical generation of DSA parameter 'g'. See
2655     FIPS 186-3 A.2.3.
2656
2657  *) Add support for HMAC DRBG from SP800-90. Update DRBG algorithm test and
2658     POST to handle HMAC cases.
2659     [Steve Henson]
2660
2661  *) Add functions FIPS_module_version() and FIPS_module_version_text()
2662     to return numerical and string versions of the FIPS module number.
2663     [Steve Henson]
2664
2665  *) Rename FIPS_mode_set and FIPS_mode to FIPS_module_mode_set and
2666     FIPS_module_mode. FIPS_mode and FIPS_mode_set will be implemented
2667     outside the validated module in the FIPS capable OpenSSL.
2668     [Steve Henson]
2669
2670  *) Minor change to DRBG entropy callback semantics. In some cases
2671     there is no multiple of the block length between min_len and
2672     max_len. Allow the callback to return more than max_len bytes
2673     of entropy but discard any extra: it is the callback's responsibility
2674     to ensure that the extra data discarded does not impact the
2675     requested amount of entropy.
2676     [Steve Henson]
2677
2678  *) Add PRNG security strength checks to RSA, DSA and ECDSA using
2679     information in FIPS186-3, SP800-57 and SP800-131A.
2680     [Steve Henson]
2681
2682  *) CCM support via EVP. Interface is very similar to GCM case except we
2683     must supply all data in one chunk (i.e. no update, final) and the
2684     message length must be supplied if AAD is used. Add algorithm test
2685     support.
2686     [Steve Henson]
2687
2688  *) Initial version of POST overhaul. Add POST callback to allow the status
2689     of POST to be monitored and/or failures induced. Modify fips_test_suite
2690     to use callback. Always run all selftests even if one fails.
2691     [Steve Henson]
2692
2693  *) XTS support including algorithm test driver in the fips_gcmtest program.
2694     Note: this does increase the maximum key length from 32 to 64 bytes but
2695     there should be no binary compatibility issues as existing applications
2696     will never use XTS mode.
2697     [Steve Henson]
2698
2699  *) Extensive reorganisation of FIPS PRNG behaviour. Remove all dependencies
2700     to OpenSSL RAND code and replace with a tiny FIPS RAND API which also
2701     performs algorithm blocking for unapproved PRNG types. Also do not
2702     set PRNG type in FIPS_mode_set(): leave this to the application.
2703     Add default OpenSSL DRBG handling: sets up FIPS PRNG and seeds with
2704     the standard OpenSSL PRNG: set additional data to a date time vector.
2705     [Steve Henson]
2706
2707  *) Rename old X9.31 PRNG functions of the form FIPS_rand* to FIPS_x931*.
2708     This shouldn't present any incompatibility problems because applications
2709     shouldn't be using these directly and any that are will need to rethink
2710     anyway as the X9.31 PRNG is now deprecated by FIPS 140-2
2711     [Steve Henson]
2712
2713  *) Extensive self tests and health checking required by SP800-90 DRBG.
2714     Remove strength parameter from FIPS_drbg_instantiate and always
2715     instantiate at maximum supported strength.
2716     [Steve Henson]
2717
2718  *) Add ECDH code to fips module and fips_ecdhvs for primitives only testing.
2719     [Steve Henson]
2720
2721  *) New algorithm test program fips_dhvs to handle DH primitives only testing.
2722     [Steve Henson]
2723
2724  *) New function DH_compute_key_padded() to compute a DH key and pad with
2725     leading zeroes if needed: this complies with SP800-56A et al.
2726     [Steve Henson]
2727
2728  *) Initial implementation of SP800-90 DRBGs for Hash and CTR. Not used by
2729     anything, incomplete, subject to change and largely untested at present.
2730     [Steve Henson]
2731
2732  *) Modify fipscanisteronly build option to only build the necessary object
2733     files by filtering FIPS_EX_OBJ through a perl script in crypto/Makefile.
2734     [Steve Henson]
2735
2736  *) Add experimental option FIPSSYMS to give all symbols in
2737     fipscanister.o and FIPS or fips prefix. This will avoid
2738     conflicts with future versions of OpenSSL. Add perl script
2739     util/fipsas.pl to preprocess assembly language source files
2740     and rename any affected symbols.
2741     [Steve Henson]
2742
2743  *) Add selftest checks and algorithm block of non-fips algorithms in
2744     FIPS mode. Remove DES2 from selftests.
2745     [Steve Henson]
2746
2747  *) Add ECDSA code to fips module. Add tiny fips_ecdsa_check to just
2748     return internal method without any ENGINE dependencies. Add new
2749     tiny fips sign and verify functions.
2750     [Steve Henson]
2751
2752  *) New build option no-ec2m to disable characteristic 2 code.
2753     [Steve Henson]
2754
2755  *) New build option "fipscanisteronly". This only builds fipscanister.o
2756     and (currently) associated fips utilities. Uses the file Makefile.fips
2757     instead of Makefile.org as the prototype.
2758     [Steve Henson]
2759
2760  *) Add some FIPS mode restrictions to GCM. Add internal IV generator.
2761     Update fips_gcmtest to use IV generator.
2762     [Steve Henson]
2763
2764  *) Initial, experimental EVP support for AES-GCM. AAD can be input by
2765     setting output buffer to NULL. The *Final function must be
2766     called although it will not retrieve any additional data. The tag
2767     can be set or retrieved with a ctrl. The IV length is by default 12
2768     bytes (96 bits) but can be set to an alternative value. If the IV
2769     length exceeds the maximum IV length (currently 16 bytes) it cannot be
2770     set before the key.
2771     [Steve Henson]
2772
2773  *) New flag in ciphers: EVP_CIPH_FLAG_CUSTOM_CIPHER. This means the
2774     underlying do_cipher function handles all cipher semantics itself
2775     including padding and finalisation. This is useful if (for example)
2776     an ENGINE cipher handles block padding itself. The behaviour of
2777     do_cipher is subtly changed if this flag is set: the return value
2778     is the number of characters written to the output buffer (zero is
2779     no longer an error code) or a negative error code. Also if the
2780     input buffer is NULL and length 0 finalisation should be performed.
2781     [Steve Henson]
2782
2783  *) If a candidate issuer certificate is already part of the constructed
2784     path ignore it: new debug notification X509_V_ERR_PATH_LOOP for this case.
2785     [Steve Henson]
2786
2787  *) Improve forward-security support: add functions
2788
2789       void SSL_CTX_set_not_resumable_session_callback(SSL_CTX *ctx, int (*cb)(SSL *ssl, int is_forward_secure))
2790       void SSL_set_not_resumable_session_callback(SSL *ssl, int (*cb)(SSL *ssl, int is_forward_secure))
2791
2792     for use by SSL/TLS servers; the callback function will be called whenever a
2793     new session is created, and gets to decide whether the session may be
2794     cached to make it resumable (return 0) or not (return 1).  (As by the
2795     SSL/TLS protocol specifications, the session_id sent by the server will be
2796     empty to indicate that the session is not resumable; also, the server will
2797     not generate RFC 4507 (RFC 5077) session tickets.)
2798
2799     A simple reasonable callback implementation is to return is_forward_secure.
2800     This parameter will be set to 1 or 0 depending on the ciphersuite selected
2801     by the SSL/TLS server library, indicating whether it can provide forward
2802     security.
2803     [Emilia Käsper <emilia.kasper@esat.kuleuven.be> (Google)]
2804
2805  *) New -verify_name option in command line utilities to set verification
2806     parameters by name.
2807     [Steve Henson]
2808
2809  *) Initial CMAC implementation. WARNING: EXPERIMENTAL, API MAY CHANGE.
2810     Add CMAC pkey methods.
2811     [Steve Henson]
2812
2813  *) Experimental renegotiation in s_server -www mode. If the client
2814     browses /reneg connection is renegotiated. If /renegcert it is
2815     renegotiated requesting a certificate.
2816     [Steve Henson]
2817
2818  *) Add an "external" session cache for debugging purposes to s_server. This
2819     should help trace issues which normally are only apparent in deployed
2820     multi-process servers.
2821     [Steve Henson]
2822
2823  *) Extensive audit of libcrypto with DEBUG_UNUSED. Fix many cases where
2824     return value is ignored. NB. The functions RAND_add(), RAND_seed(),
2825     BIO_set_cipher() and some obscure PEM functions were changed so they
2826     can now return an error. The RAND changes required a change to the
2827     RAND_METHOD structure.
2828     [Steve Henson]
2829
2830  *) New macro __owur for "OpenSSL Warn Unused Result". This makes use of
2831     a gcc attribute to warn if the result of a function is ignored. This
2832     is enable if DEBUG_UNUSED is set. Add to several functions in evp.h
2833     whose return value is often ignored.
2834     [Steve Henson]
2835
2836  *) New -noct, -requestct, -requirect and -ctlogfile options for s_client.
2837     These allow SCTs (signed certificate timestamps) to be requested and
2838     validated when establishing a connection.
2839     [Rob Percival <robpercival@google.com>]
2840
2841 Changes between 1.0.2g and 1.0.2h [3 May 2016]
2842
2843  *) Prevent padding oracle in AES-NI CBC MAC check
2844
2845     A MITM attacker can use a padding oracle attack to decrypt traffic
2846     when the connection uses an AES CBC cipher and the server support
2847     AES-NI.
2848
2849     This issue was introduced as part of the fix for Lucky 13 padding
2850     attack (CVE-2013-0169). The padding check was rewritten to be in
2851     constant time by making sure that always the same bytes are read and
2852     compared against either the MAC or padding bytes. But it no longer
2853     checked that there was enough data to have both the MAC and padding
2854     bytes.
2855
2856     This issue was reported by Juraj Somorovsky using TLS-Attacker.
2857     (CVE-2016-2107)
2858     [Kurt Roeckx]
2859
2860  *) Fix EVP_EncodeUpdate overflow
2861
2862     An overflow can occur in the EVP_EncodeUpdate() function which is used for
2863     Base64 encoding of binary data. If an attacker is able to supply very large
2864     amounts of input data then a length check can overflow resulting in a heap
2865     corruption.
2866
2867     Internally to OpenSSL the EVP_EncodeUpdate() function is primarily used by
2868     the PEM_write_bio* family of functions. These are mainly used within the
2869     OpenSSL command line applications, so any application which processes data
2870     from an untrusted source and outputs it as a PEM file should be considered
2871     vulnerable to this issue. User applications that call these APIs directly
2872     with large amounts of untrusted data may also be vulnerable.
2873
2874     This issue was reported by Guido Vranken.
2875     (CVE-2016-2105)
2876     [Matt Caswell]
2877
2878  *) Fix EVP_EncryptUpdate overflow
2879
2880     An overflow can occur in the EVP_EncryptUpdate() function. If an attacker
2881     is able to supply very large amounts of input data after a previous call to
2882     EVP_EncryptUpdate() with a partial block then a length check can overflow
2883     resulting in a heap corruption. Following an analysis of all OpenSSL
2884     internal usage of the EVP_EncryptUpdate() function all usage is one of two
2885     forms. The first form is where the EVP_EncryptUpdate() call is known to be
2886     the first called function after an EVP_EncryptInit(), and therefore that
2887     specific call must be safe. The second form is where the length passed to
2888     EVP_EncryptUpdate() can be seen from the code to be some small value and
2889     therefore there is no possibility of an overflow. Since all instances are
2890     one of these two forms, it is believed that there can be no overflows in
2891     internal code due to this problem. It should be noted that
2892     EVP_DecryptUpdate() can call EVP_EncryptUpdate() in certain code paths.
2893     Also EVP_CipherUpdate() is a synonym for EVP_EncryptUpdate(). All instances
2894     of these calls have also been analysed too and it is believed there are no
2895     instances in internal usage where an overflow could occur.
2896
2897     This issue was reported by Guido Vranken.
2898     (CVE-2016-2106)
2899     [Matt Caswell]
2900
2901  *) Prevent ASN.1 BIO excessive memory allocation
2902
2903     When ASN.1 data is read from a BIO using functions such as d2i_CMS_bio()
2904     a short invalid encoding can cause allocation of large amounts of memory
2905     potentially consuming excessive resources or exhausting memory.
2906
2907     Any application parsing untrusted data through d2i BIO functions is
2908     affected. The memory based functions such as d2i_X509() are *not* affected.
2909     Since the memory based functions are used by the TLS library, TLS
2910     applications are not affected.
2911
2912     This issue was reported by Brian Carpenter.
2913     (CVE-2016-2109)
2914     [Stephen Henson]
2915
2916  *) EBCDIC overread
2917
2918     ASN1 Strings that are over 1024 bytes can cause an overread in applications
2919     using the X509_NAME_oneline() function on EBCDIC systems. This could result
2920     in arbitrary stack data being returned in the buffer.
2921
2922     This issue was reported by Guido Vranken.
2923     (CVE-2016-2176)
2924     [Matt Caswell]
2925
2926  *) Modify behavior of ALPN to invoke callback after SNI/servername
2927     callback, such that updates to the SSL_CTX affect ALPN.
2928     [Todd Short]
2929
2930  *) Remove LOW from the DEFAULT cipher list.  This removes singles DES from the
2931     default.
2932     [Kurt Roeckx]
2933
2934  *) Only remove the SSLv2 methods with the no-ssl2-method option. When the
2935     methods are enabled and ssl2 is disabled the methods return NULL.
2936     [Kurt Roeckx]
2937
2938 Changes between 1.0.2f and 1.0.2g [1 Mar 2016]
2939
2940  * Disable weak ciphers in SSLv3 and up in default builds of OpenSSL.
2941    Builds that are not configured with "enable-weak-ssl-ciphers" will not
2942    provide any "EXPORT" or "LOW" strength ciphers.
2943    [Viktor Dukhovni]
2944
2945  * Disable SSLv2 default build, default negotiation and weak ciphers.  SSLv2
2946    is by default disabled at build-time.  Builds that are not configured with
2947    "enable-ssl2" will not support SSLv2.  Even if "enable-ssl2" is used,
2948    users who want to negotiate SSLv2 via the version-flexible SSLv23_method()
2949    will need to explicitly call either of:
2950
2951        SSL_CTX_clear_options(ctx, SSL_OP_NO_SSLv2);
2952    or
2953        SSL_clear_options(ssl, SSL_OP_NO_SSLv2);
2954
2955    as appropriate.  Even if either of those is used, or the application
2956    explicitly uses the version-specific SSLv2_method() or its client and
2957    server variants, SSLv2 ciphers vulnerable to exhaustive search key
2958    recovery have been removed.  Specifically, the SSLv2 40-bit EXPORT
2959    ciphers, and SSLv2 56-bit DES are no longer available.
2960    (CVE-2016-0800)
2961    [Viktor Dukhovni]
2962
2963  *) Fix a double-free in DSA code
2964
2965     A double free bug was discovered when OpenSSL parses malformed DSA private
2966     keys and could lead to a DoS attack or memory corruption for applications
2967     that receive DSA private keys from untrusted sources.  This scenario is
2968     considered rare.
2969
2970     This issue was reported to OpenSSL by Adam Langley(Google/BoringSSL) using
2971     libFuzzer.
2972     (CVE-2016-0705)
2973     [Stephen Henson]
2974
2975  *) Disable SRP fake user seed to address a server memory leak.
2976
2977     Add a new method SRP_VBASE_get1_by_user that handles the seed properly.
2978
2979     SRP_VBASE_get_by_user had inconsistent memory management behaviour.
2980     In order to fix an unavoidable memory leak, SRP_VBASE_get_by_user
2981     was changed to ignore the "fake user" SRP seed, even if the seed
2982     is configured.
2983
2984     Users should use SRP_VBASE_get1_by_user instead. Note that in
2985     SRP_VBASE_get1_by_user, caller must free the returned value. Note
2986     also that even though configuring the SRP seed attempts to hide
2987     invalid usernames by continuing the handshake with fake
2988     credentials, this behaviour is not constant time and no strong
2989     guarantees are made that the handshake is indistinguishable from
2990     that of a valid user.
2991     (CVE-2016-0798)
2992     [Emilia Käsper]
2993
2994  *) Fix BN_hex2bn/BN_dec2bn NULL pointer deref/heap corruption
2995
2996     In the BN_hex2bn function the number of hex digits is calculated using an
2997     int value |i|. Later |bn_expand| is called with a value of |i * 4|. For
2998     large values of |i| this can result in |bn_expand| not allocating any
2999     memory because |i * 4| is negative. This can leave the internal BIGNUM data
3000     field as NULL leading to a subsequent NULL ptr deref. For very large values
3001     of |i|, the calculation |i * 4| could be a positive value smaller than |i|.
3002     In this case memory is allocated to the internal BIGNUM data field, but it
3003     is insufficiently sized leading to heap corruption. A similar issue exists
3004     in BN_dec2bn. This could have security consequences if BN_hex2bn/BN_dec2bn
3005     is ever called by user applications with very large untrusted hex/dec data.
3006     This is anticipated to be a rare occurrence.
3007
3008     All OpenSSL internal usage of these functions use data that is not expected
3009     to be untrusted, e.g. config file data or application command line
3010     arguments. If user developed applications generate config file data based
3011     on untrusted data then it is possible that this could also lead to security
3012     consequences. This is also anticipated to be rare.
3013
3014     This issue was reported to OpenSSL by Guido Vranken.
3015     (CVE-2016-0797)
3016     [Matt Caswell]
3017
3018  *) Fix memory issues in BIO_*printf functions
3019
3020     The internal |fmtstr| function used in processing a "%s" format string in
3021     the BIO_*printf functions could overflow while calculating the length of a
3022     string and cause an OOB read when printing very long strings.
3023
3024     Additionally the internal |doapr_outch| function can attempt to write to an
3025     OOB memory location (at an offset from the NULL pointer) in the event of a
3026     memory allocation failure. In 1.0.2 and below this could be caused where
3027     the size of a buffer to be allocated is greater than INT_MAX. E.g. this
3028     could be in processing a very long "%s" format string. Memory leaks can
3029     also occur.
3030
3031     The first issue may mask the second issue dependent on compiler behaviour.
3032     These problems could enable attacks where large amounts of untrusted data
3033     is passed to the BIO_*printf functions. If applications use these functions
3034     in this way then they could be vulnerable. OpenSSL itself uses these
3035     functions when printing out human-readable dumps of ASN.1 data. Therefore
3036     applications that print this data could be vulnerable if the data is from
3037     untrusted sources. OpenSSL command line applications could also be
3038     vulnerable where they print out ASN.1 data, or if untrusted data is passed
3039     as command line arguments.
3040
3041     Libssl is not considered directly vulnerable. Additionally certificates etc
3042     received via remote connections via libssl are also unlikely to be able to
3043     trigger these issues because of message size limits enforced within libssl.
3044
3045     This issue was reported to OpenSSL Guido Vranken.
3046     (CVE-2016-0799)
3047     [Matt Caswell]
3048
3049  *) Side channel attack on modular exponentiation
3050
3051     A side-channel attack was found which makes use of cache-bank conflicts on
3052     the Intel Sandy-Bridge microarchitecture which could lead to the recovery
3053     of RSA keys.  The ability to exploit this issue is limited as it relies on
3054     an attacker who has control of code in a thread running on the same
3055     hyper-threaded core as the victim thread which is performing decryptions.
3056
3057     This issue was reported to OpenSSL by Yuval Yarom, The University of
3058     Adelaide and NICTA, Daniel Genkin, Technion and Tel Aviv University, and
3059     Nadia Heninger, University of Pennsylvania with more information at
3060     http://cachebleed.info.
3061     (CVE-2016-0702)
3062     [Andy Polyakov]
3063
3064  *) Change the req app to generate a 2048-bit RSA/DSA key by default,
3065     if no keysize is specified with default_bits. This fixes an
3066     omission in an earlier change that changed all RSA/DSA key generation
3067     apps to use 2048 bits by default.
3068     [Emilia Käsper]
3069
3070 Changes between 1.0.2e and 1.0.2f [28 Jan 2016]
3071  *) DH small subgroups
3072
3073     Historically OpenSSL only ever generated DH parameters based on "safe"
3074     primes. More recently (in version 1.0.2) support was provided for
3075     generating X9.42 style parameter files such as those required for RFC 5114
3076     support. The primes used in such files may not be "safe". Where an
3077     application is using DH configured with parameters based on primes that are
3078     not "safe" then an attacker could use this fact to find a peer's private
3079     DH exponent. This attack requires that the attacker complete multiple
3080     handshakes in which the peer uses the same private DH exponent. For example
3081     this could be used to discover a TLS server's private DH exponent if it's
3082     reusing the private DH exponent or it's using a static DH ciphersuite.
3083
3084     OpenSSL provides the option SSL_OP_SINGLE_DH_USE for ephemeral DH (DHE) in
3085     TLS. It is not on by default. If the option is not set then the server
3086     reuses the same private DH exponent for the life of the server process and
3087     would be vulnerable to this attack. It is believed that many popular
3088     applications do set this option and would therefore not be at risk.
3089
3090     The fix for this issue adds an additional check where a "q" parameter is
3091     available (as is the case in X9.42 based parameters). This detects the
3092     only known attack, and is the only possible defense for static DH
3093     ciphersuites. This could have some performance impact.
3094
3095     Additionally the SSL_OP_SINGLE_DH_USE option has been switched on by
3096     default and cannot be disabled. This could have some performance impact.
3097
3098     This issue was reported to OpenSSL by Antonio Sanso (Adobe).
3099     (CVE-2016-0701)
3100     [Matt Caswell]
3101
3102  *) SSLv2 doesn't block disabled ciphers
3103
3104     A malicious client can negotiate SSLv2 ciphers that have been disabled on
3105     the server and complete SSLv2 handshakes even if all SSLv2 ciphers have
3106     been disabled, provided that the SSLv2 protocol was not also disabled via
3107     SSL_OP_NO_SSLv2.
3108
3109     This issue was reported to OpenSSL on 26th December 2015 by Nimrod Aviram
3110     and Sebastian Schinzel.
3111     (CVE-2015-3197)
3112     [Viktor Dukhovni]
3113
3114 Changes between 1.0.2d and 1.0.2e [3 Dec 2015]
3115
3116  *) BN_mod_exp may produce incorrect results on x86_64
3117
3118     There is a carry propagating bug in the x86_64 Montgomery squaring
3119     procedure. No EC algorithms are affected. Analysis suggests that attacks
3120     against RSA and DSA as a result of this defect would be very difficult to
3121     perform and are not believed likely. Attacks against DH are considered just
3122     feasible (although very difficult) because most of the work necessary to
3123     deduce information about a private key may be performed offline. The amount
3124     of resources required for such an attack would be very significant and
3125     likely only accessible to a limited number of attackers. An attacker would
3126     additionally need online access to an unpatched system using the target
3127     private key in a scenario with persistent DH parameters and a private
3128     key that is shared between multiple clients. For example this can occur by
3129     default in OpenSSL DHE based SSL/TLS ciphersuites.
3130
3131     This issue was reported to OpenSSL by Hanno Böck.
3132     (CVE-2015-3193)
3133     [Andy Polyakov]
3134
3135  *) Certificate verify crash with missing PSS parameter
3136
3137     The signature verification routines will crash with a NULL pointer
3138     dereference if presented with an ASN.1 signature using the RSA PSS
3139     algorithm and absent mask generation function parameter. Since these
3140     routines are used to verify certificate signature algorithms this can be
3141     used to crash any certificate verification operation and exploited in a
3142     DoS attack. Any application which performs certificate verification is
3143     vulnerable including OpenSSL clients and servers which enable client
3144     authentication.
3145
3146     This issue was reported to OpenSSL by Loïc Jonas Etienne (Qnective AG).
3147     (CVE-2015-3194)
3148     [Stephen Henson]
3149
3150  *) X509_ATTRIBUTE memory leak
3151
3152     When presented with a malformed X509_ATTRIBUTE structure OpenSSL will leak
3153     memory. This structure is used by the PKCS#7 and CMS routines so any
3154     application which reads PKCS#7 or CMS data from untrusted sources is
3155     affected. SSL/TLS is not affected.
3156
3157     This issue was reported to OpenSSL by Adam Langley (Google/BoringSSL) using
3158     libFuzzer.
3159     (CVE-2015-3195)
3160     [Stephen Henson]
3161
3162  *) Rewrite EVP_DecodeUpdate (base64 decoding) to fix several bugs.
3163     This changes the decoding behaviour for some invalid messages,
3164     though the change is mostly in the more lenient direction, and
3165     legacy behaviour is preserved as much as possible.
3166     [Emilia Käsper]
3167
3168  *) In DSA_generate_parameters_ex, if the provided seed is too short,
3169     return an error
3170     [Rich Salz and Ismo Puustinen <ismo.puustinen@intel.com>]
3171
3172 Changes between 1.0.2c and 1.0.2d [9 Jul 2015]
3173
3174  *) Alternate chains certificate forgery
3175
3176     During certificate verification, OpenSSL will attempt to find an
3177     alternative certificate chain if the first attempt to build such a chain
3178     fails. An error in the implementation of this logic can mean that an
3179     attacker could cause certain checks on untrusted certificates to be
3180     bypassed, such as the CA flag, enabling them to use a valid leaf
3181     certificate to act as a CA and "issue" an invalid certificate.
3182
3183     This issue was reported to OpenSSL by Adam Langley/David Benjamin
3184     (Google/BoringSSL).
3185     [Matt Caswell]
3186
3187 Changes between 1.0.2b and 1.0.2c [12 Jun 2015]
3188
3189  *) Fix HMAC ABI incompatibility. The previous version introduced an ABI
3190     incompatibility in the handling of HMAC. The previous ABI has now been
3191     restored.
3192     [Matt Caswell]
3193
3194 Changes between 1.0.2a and 1.0.2b [11 Jun 2015]
3195
3196  *) Malformed ECParameters causes infinite loop
3197
3198     When processing an ECParameters structure OpenSSL enters an infinite loop
3199     if the curve specified is over a specially malformed binary polynomial
3200     field.
3201
3202     This can be used to perform denial of service against any
3203     system which processes public keys, certificate requests or
3204     certificates.  This includes TLS clients and TLS servers with
3205     client authentication enabled.
3206
3207     This issue was reported to OpenSSL by Joseph Barr-Pixton.
3208     (CVE-2015-1788)
3209     [Andy Polyakov]
3210
3211  *) Exploitable out-of-bounds read in X509_cmp_time
3212
3213     X509_cmp_time does not properly check the length of the ASN1_TIME
3214     string and can read a few bytes out of bounds. In addition,
3215     X509_cmp_time accepts an arbitrary number of fractional seconds in the
3216     time string.
3217
3218     An attacker can use this to craft malformed certificates and CRLs of
3219     various sizes and potentially cause a segmentation fault, resulting in
3220     a DoS on applications that verify certificates or CRLs. TLS clients
3221     that verify CRLs are affected. TLS clients and servers with client
3222     authentication enabled may be affected if they use custom verification
3223     callbacks.
3224
3225     This issue was reported to OpenSSL by Robert Swiecki (Google), and
3226     independently by Hanno Böck.
3227     (CVE-2015-1789)
3228     [Emilia Käsper]
3229
3230  *) PKCS7 crash with missing EnvelopedContent
3231
3232     The PKCS#7 parsing code does not handle missing inner EncryptedContent
3233     correctly. An attacker can craft malformed ASN.1-encoded PKCS#7 blobs
3234     with missing content and trigger a NULL pointer dereference on parsing.
3235
3236     Applications that decrypt PKCS#7 data or otherwise parse PKCS#7
3237     structures from untrusted sources are affected. OpenSSL clients and
3238     servers are not affected.
3239
3240     This issue was reported to OpenSSL by Michal Zalewski (Google).
3241     (CVE-2015-1790)
3242     [Emilia Käsper]
3243
3244  *) CMS verify infinite loop with unknown hash function
3245
3246     When verifying a signedData message the CMS code can enter an infinite loop
3247     if presented with an unknown hash function OID. This can be used to perform
3248     denial of service against any system which verifies signedData messages using
3249     the CMS code.
3250     This issue was reported to OpenSSL by Johannes Bauer.
3251     (CVE-2015-1792)
3252     [Stephen Henson]
3253
3254  *) Race condition handling NewSessionTicket
3255
3256     If a NewSessionTicket is received by a multi-threaded client when attempting to
3257     reuse a previous ticket then a race condition can occur potentially leading to
3258     a double free of the ticket data.
3259     (CVE-2015-1791)
3260     [Matt Caswell]
3261
3262  *) Only support 256-bit or stronger elliptic curves with the
3263     'ecdh_auto' setting (server) or by default (client). Of supported
3264     curves, prefer P-256 (both).
3265     [Emilia Kasper]
3266
3267 Changes between 1.0.2 and 1.0.2a [19 Mar 2015]
3268
3269  *) ClientHello sigalgs DoS fix
3270
3271     If a client connects to an OpenSSL 1.0.2 server and renegotiates with an
3272     invalid signature algorithms extension a NULL pointer dereference will
3273     occur. This can be exploited in a DoS attack against the server.
3274
3275     This issue was was reported to OpenSSL by David Ramos of Stanford
3276     University.
3277     (CVE-2015-0291)
3278     [Stephen Henson and Matt Caswell]
3279
3280  *) Multiblock corrupted pointer fix
3281
3282     OpenSSL 1.0.2 introduced the "multiblock" performance improvement. This
3283     feature only applies on 64 bit x86 architecture platforms that support AES
3284     NI instructions. A defect in the implementation of "multiblock" can cause
3285     OpenSSL's internal write buffer to become incorrectly set to NULL when
3286     using non-blocking IO. Typically, when the user application is using a
3287     socket BIO for writing, this will only result in a failed connection.
3288     However if some other BIO is used then it is likely that a segmentation
3289     fault will be triggered, thus enabling a potential DoS attack.
3290
3291     This issue was reported to OpenSSL by Daniel Danner and Rainer Mueller.
3292     (CVE-2015-0290)
3293     [Matt Caswell]
3294
3295  *) Segmentation fault in DTLSv1_listen fix
3296
3297     The DTLSv1_listen function is intended to be stateless and processes the
3298     initial ClientHello from many peers. It is common for user code to loop
3299     over the call to DTLSv1_listen until a valid ClientHello is received with
3300     an associated cookie. A defect in the implementation of DTLSv1_listen means
3301     that state is preserved in the SSL object from one invocation to the next
3302     that can lead to a segmentation fault. Errors processing the initial
3303     ClientHello can trigger this scenario. An example of such an error could be
3304     that a DTLS1.0 only client is attempting to connect to a DTLS1.2 only
3305     server.
3306
3307     This issue was reported to OpenSSL by Per Allansson.
3308     (CVE-2015-0207)
3309     [Matt Caswell]
3310
3311  *) Segmentation fault in ASN1_TYPE_cmp fix
3312
3313     The function ASN1_TYPE_cmp will crash with an invalid read if an attempt is
3314     made to compare ASN.1 boolean types. Since ASN1_TYPE_cmp is used to check
3315     certificate signature algorithm consistency this can be used to crash any
3316     certificate verification operation and exploited in a DoS attack. Any
3317     application which performs certificate verification is vulnerable including
3318     OpenSSL clients and servers which enable client authentication.
3319     (CVE-2015-0286)
3320     [Stephen Henson]
3321
3322  *) Segmentation fault for invalid PSS parameters fix
3323
3324     The signature verification routines will crash with a NULL pointer
3325     dereference if presented with an ASN.1 signature using the RSA PSS
3326     algorithm and invalid parameters. Since these routines are used to verify
3327     certificate signature algorithms this can be used to crash any
3328     certificate verification operation and exploited in a DoS attack. Any
3329     application which performs certificate verification is vulnerable including
3330     OpenSSL clients and servers which enable client authentication.
3331
3332     This issue was was reported to OpenSSL by Brian Carpenter.
3333     (CVE-2015-0208)
3334     [Stephen Henson]
3335
3336  *) ASN.1 structure reuse memory corruption fix
3337
3338     Reusing a structure in ASN.1 parsing may allow an attacker to cause
3339     memory corruption via an invalid write. Such reuse is and has been
3340     strongly discouraged and is believed to be rare.
3341
3342     Applications that parse structures containing CHOICE or ANY DEFINED BY
3343     components may be affected. Certificate parsing (d2i_X509 and related
3344     functions) are however not affected. OpenSSL clients and servers are
3345     not affected.
3346     (CVE-2015-0287)
3347     [Stephen Henson]
3348
3349  *) PKCS7 NULL pointer dereferences fix
3350
3351     The PKCS#7 parsing code does not handle missing outer ContentInfo
3352     correctly. An attacker can craft malformed ASN.1-encoded PKCS#7 blobs with
3353     missing content and trigger a NULL pointer dereference on parsing.
3354
3355     Applications that verify PKCS#7 signatures, decrypt PKCS#7 data or
3356     otherwise parse PKCS#7 structures from untrusted sources are
3357     affected. OpenSSL clients and servers are not affected.
3358
3359     This issue was reported to OpenSSL by Michal Zalewski (Google).
3360     (CVE-2015-0289)
3361     [Emilia Käsper]
3362
3363  *) DoS via reachable assert in SSLv2 servers fix
3364
3365     A malicious client can trigger an OPENSSL_assert (i.e., an abort) in
3366     servers that both support SSLv2 and enable export cipher suites by sending
3367     a specially crafted SSLv2 CLIENT-MASTER-KEY message.
3368
3369     This issue was discovered by Sean Burford (Google) and Emilia Käsper
3370     (OpenSSL development team).
3371     (CVE-2015-0293)
3372     [Emilia Käsper]
3373
3374  *) Empty CKE with client auth and DHE fix
3375
3376     If client auth is used then a server can seg fault in the event of a DHE
3377     ciphersuite being selected and a zero length ClientKeyExchange message
3378     being sent by the client. This could be exploited in a DoS attack.
3379     (CVE-2015-1787)
3380     [Matt Caswell]
3381
3382  *) Handshake with unseeded PRNG fix
3383
3384     Under certain conditions an OpenSSL 1.0.2 client can complete a handshake
3385     with an unseeded PRNG. The conditions are:
3386     - The client is on a platform where the PRNG has not been seeded
3387     automatically, and the user has not seeded manually
3388     - A protocol specific client method version has been used (i.e. not
3389     SSL_client_methodv23)
3390     - A ciphersuite is used that does not require additional random data from
3391     the PRNG beyond the initial ClientHello client random (e.g. PSK-RC4-SHA).
3392
3393     If the handshake succeeds then the client random that has been used will
3394     have been generated from a PRNG with insufficient entropy and therefore the
3395     output may be predictable.
3396
3397     For example using the following command with an unseeded openssl will
3398     succeed on an unpatched platform:
3399
3400     openssl s_client -psk 1a2b3c4d -tls1_2 -cipher PSK-RC4-SHA
3401     (CVE-2015-0285)
3402     [Matt Caswell]
3403
3404  *) Use After Free following d2i_ECPrivatekey error fix
3405
3406     A malformed EC private key file consumed via the d2i_ECPrivateKey function
3407     could cause a use after free condition. This, in turn, could cause a double
3408     free in several private key parsing functions (such as d2i_PrivateKey
3409     or EVP_PKCS82PKEY) and could lead to a DoS attack or memory corruption
3410     for applications that receive EC private keys from untrusted
3411     sources. This scenario is considered rare.
3412
3413     This issue was discovered by the BoringSSL project and fixed in their
3414     commit 517073cd4b.
3415     (CVE-2015-0209)
3416     [Matt Caswell]
3417
3418  *) X509_to_X509_REQ NULL pointer deref fix
3419
3420     The function X509_to_X509_REQ will crash with a NULL pointer dereference if
3421     the certificate key is invalid. This function is rarely used in practice.
3422
3423     This issue was discovered by Brian Carpenter.
3424     (CVE-2015-0288)
3425     [Stephen Henson]
3426
3427  *) Removed the export ciphers from the DEFAULT ciphers
3428     [Kurt Roeckx]
3429
3430 Changes between 1.0.1l and 1.0.2 [22 Jan 2015]
3431
3432  *) Facilitate "universal" ARM builds targeting range of ARM ISAs, e.g.
3433     ARMv5 through ARMv8, as opposite to "locking" it to single one.
3434     So far those who have to target multiple platforms would compromise
3435     and argue that binary targeting say ARMv5 would still execute on
3436     ARMv8. "Universal" build resolves this compromise by providing
3437     near-optimal performance even on newer platforms.
3438     [Andy Polyakov]
3439
3440  *) Accelerated NIST P-256 elliptic curve implementation for x86_64
3441     (other platforms pending).
3442     [Shay Gueron & Vlad Krasnov (Intel Corp), Andy Polyakov]
3443
3444  *) Add support for the SignedCertificateTimestampList certificate and
3445     OCSP response extensions from RFC6962.
3446     [Rob Stradling]
3447
3448  *) Fix ec_GFp_simple_points_make_affine (thus, EC_POINTs_mul etc.)
3449     for corner cases. (Certain input points at infinity could lead to
3450     bogus results, with non-infinity inputs mapped to infinity too.)
3451     [Bodo Moeller]
3452
3453  *) Initial support for PowerISA 2.0.7, first implemented in POWER8.
3454     This covers AES, SHA256/512 and GHASH. "Initial" means that most
3455     common cases are optimized and there still is room for further
3456     improvements. Vector Permutation AES for Altivec is also added.
3457     [Andy Polyakov]
3458
3459  *) Add support for little-endian ppc64 Linux target.
3460     [Marcelo Cerri (IBM)]
3461
3462  *) Initial support for AMRv8 ISA crypto extensions. This covers AES,
3463     SHA1, SHA256 and GHASH. "Initial" means that most common cases
3464     are optimized and there still is room for further improvements.
3465     Both 32- and 64-bit modes are supported.
3466     [Andy Polyakov, Ard Biesheuvel (Linaro)]
3467
3468  *) Improved ARMv7 NEON support.
3469     [Andy Polyakov]
3470
3471  *) Support for SPARC Architecture 2011 crypto extensions, first
3472     implemented in SPARC T4. This covers AES, DES, Camellia, SHA1,
3473     SHA256/512, MD5, GHASH and modular exponentiation.
3474     [Andy Polyakov, David Miller]
3475
3476  *) Accelerated modular exponentiation for Intel processors, a.k.a.
3477     RSAZ.
3478     [Shay Gueron & Vlad Krasnov (Intel Corp)]
3479
3480  *) Support for new and upcoming Intel processors, including AVX2,
3481     BMI and SHA ISA extensions. This includes additional "stitched"
3482     implementations, AESNI-SHA256 and GCM, and multi-buffer support
3483     for TLS encrypt.
3484
3485     This work was sponsored by Intel Corp.
3486     [Andy Polyakov]
3487
3488  *) Support for DTLS 1.2. This adds two sets of DTLS methods: DTLS_*_method()
3489     supports both DTLS 1.2 and 1.0 and should use whatever version the peer
3490     supports and DTLSv1_2_*_method() which supports DTLS 1.2 only.
3491     [Steve Henson]
3492
3493  *) Use algorithm specific chains in SSL_CTX_use_certificate_chain_file():
3494     this fixes a limitation in previous versions of OpenSSL.
3495     [Steve Henson]
3496
3497  *) Extended RSA OAEP support via EVP_PKEY API. Options to specify digest,
3498     MGF1 digest and OAEP label.
3499     [Steve Henson]
3500
3501  *) Add EVP support for key wrapping algorithms, to avoid problems with
3502     existing code the flag EVP_CIPHER_CTX_WRAP_ALLOW has to be set in
3503     the EVP_CIPHER_CTX or an error is returned. Add AES and DES3 wrap
3504     algorithms and include tests cases.
3505     [Steve Henson]
3506
3507  *) Add functions to allocate and set the fields of an ECDSA_METHOD
3508     structure.
3509     [Douglas E. Engert, Steve Henson]
3510
3511  *) New functions OPENSSL_gmtime_diff and ASN1_TIME_diff to find the
3512     difference in days and seconds between two tm or ASN1_TIME structures.
3513     [Steve Henson]
3514
3515  *) Add -rev test option to s_server to just reverse order of characters
3516     received by client and send back to server. Also prints an abbreviated
3517     summary of the connection parameters.
3518     [Steve Henson]
3519
3520  *) New option -brief for s_client and s_server to print out a brief summary
3521     of connection parameters.
3522     [Steve Henson]
3523
3524  *) Add callbacks for arbitrary TLS extensions.
3525     [Trevor Perrin <trevp@trevp.net> and Ben Laurie]
3526
3527  *) New option -crl_download in several openssl utilities to download CRLs
3528     from CRLDP extension in certificates.
3529     [Steve Henson]
3530
3531  *) New options -CRL and -CRLform for s_client and s_server for CRLs.
3532     [Steve Henson]
3533
3534  *) New function X509_CRL_diff to generate a delta CRL from the difference
3535     of two full CRLs. Add support to "crl" utility.
3536     [Steve Henson]
3537
3538  *) New functions to set lookup_crls function and to retrieve
3539     X509_STORE from X509_STORE_CTX.
3540     [Steve Henson]
3541
3542  *) Print out deprecated issuer and subject unique ID fields in
3543     certificates.
3544     [Steve Henson]
3545
3546  *) Extend OCSP I/O functions so they can be used for simple general purpose
3547     HTTP as well as OCSP. New wrapper function which can be used to download
3548     CRLs using the OCSP API.
3549     [Steve Henson]
3550
3551  *) Delegate command line handling in s_client/s_server to SSL_CONF APIs.
3552     [Steve Henson]
3553
3554  *) SSL_CONF* functions. These provide a common framework for application
3555     configuration using configuration files or command lines.
3556     [Steve Henson]
3557
3558  *) SSL/TLS tracing code. This parses out SSL/TLS records using the
3559     message callback and prints the results. Needs compile time option
3560     "enable-ssl-trace". New options to s_client and s_server to enable
3561     tracing.
3562     [Steve Henson]
3563
3564  *) New ctrl and macro to retrieve supported points extensions.
3565     Print out extension in s_server and s_client.
3566     [Steve Henson]
3567
3568  *) New functions to retrieve certificate signature and signature
3569     OID NID.
3570     [Steve Henson]
3571
3572  *) Add functions to retrieve and manipulate the raw cipherlist sent by a
3573     client to OpenSSL.
3574     [Steve Henson]
3575
3576  *) New Suite B modes for TLS code. These use and enforce the requirements
3577     of RFC6460: restrict ciphersuites, only permit Suite B algorithms and
3578     only use Suite B curves. The Suite B modes can be set by using the
3579     strings "SUITEB128", "SUITEB192" or "SUITEB128ONLY" for the cipherstring.
3580     [Steve Henson]
3581
3582  *) New chain verification flags for Suite B levels of security. Check
3583     algorithms are acceptable when flags are set in X509_verify_cert.
3584     [Steve Henson]
3585
3586  *) Make tls1_check_chain return a set of flags indicating checks passed
3587     by a certificate chain. Add additional tests to handle client
3588     certificates: checks for matching certificate type and issuer name
3589     comparison.
3590     [Steve Henson]
3591
3592  *) If an attempt is made to use a signature algorithm not in the peer
3593     preference list abort the handshake. If client has no suitable
3594     signature algorithms in response to a certificate request do not
3595     use the certificate.
3596     [Steve Henson]
3597
3598  *) If server EC tmp key is not in client preference list abort handshake.
3599     [Steve Henson]
3600
3601  *) Add support for certificate stores in CERT structure. This makes it
3602     possible to have different stores per SSL structure or one store in
3603     the parent SSL_CTX. Include distinct stores for certificate chain
3604     verification and chain building. New ctrl SSL_CTRL_BUILD_CERT_CHAIN
3605     to build and store a certificate chain in CERT structure: returning
3606     an error if the chain cannot be built: this will allow applications
3607     to test if a chain is correctly configured.
3608
3609     Note: if the CERT based stores are not set then the parent SSL_CTX
3610     store is used to retain compatibility with existing behaviour.
3611
3612     [Steve Henson]
3613
3614  *) New function ssl_set_client_disabled to set a ciphersuite disabled
3615     mask based on the current session, check mask when sending client
3616     hello and checking the requested ciphersuite.
3617     [Steve Henson]
3618
3619  *) New ctrls to retrieve and set certificate types in a certificate
3620     request message. Print out received values in s_client. If certificate
3621     types is not set with custom values set sensible values based on
3622     supported signature algorithms.
3623     [Steve Henson]
3624
3625  *) Support for distinct client and server supported signature algorithms.
3626     [Steve Henson]
3627
3628  *) Add certificate callback. If set this is called whenever a certificate
3629     is required by client or server. An application can decide which
3630     certificate chain to present based on arbitrary criteria: for example
3631     supported signature algorithms. Add very simple example to s_server.
3632     This fixes many of the problems and restrictions of the existing client
3633     certificate callback: for example you can now clear an existing
3634     certificate and specify the whole chain.
3635     [Steve Henson]
3636
3637  *) Add new "valid_flags" field to CERT_PKEY structure which determines what
3638     the certificate can be used for (if anything). Set valid_flags field
3639     in new tls1_check_chain function. Simplify ssl_set_cert_masks which used
3640     to have similar checks in it.
3641
3642     Add new "cert_flags" field to CERT structure and include a "strict mode".
3643     This enforces some TLS certificate requirements (such as only permitting
3644     certificate signature algorithms contained in the supported algorithms
3645     extension) which some implementations ignore: this option should be used
3646     with caution as it could cause interoperability issues.
3647     [Steve Henson]
3648
3649  *) Update and tidy signature algorithm extension processing. Work out
3650     shared signature algorithms based on preferences and peer algorithms
3651     and print them out in s_client and s_server. Abort handshake if no
3652     shared signature algorithms.
3653     [Steve Henson]
3654
3655  *) Add new functions to allow customised supported signature algorithms
3656     for SSL and SSL_CTX structures. Add options to s_client and s_server
3657     to support them.
3658     [Steve Henson]
3659
3660  *) New function SSL_certs_clear() to delete all references to certificates
3661     from an SSL structure. Before this once a certificate had been added
3662     it couldn't be removed.
3663     [Steve Henson]
3664
3665  *) Integrate hostname, email address and IP address checking with certificate
3666     verification. New verify options supporting checking in openssl utility.
3667     [Steve Henson]
3668
3669  *) Fixes and wildcard matching support to hostname and email checking
3670     functions. Add manual page.
3671     [Florian Weimer (Red Hat Product Security Team)]
3672
3673  *) New functions to check a hostname email or IP address against a
3674     certificate. Add options x509 utility to print results of checks against
3675     a certificate.
3676     [Steve Henson]
3677
3678  *) Fix OCSP checking.
3679     [Rob Stradling <rob.stradling@comodo.com> and Ben Laurie]
3680
3681  *) Initial experimental support for explicitly trusted non-root CAs.
3682     OpenSSL still tries to build a complete chain to a root but if an
3683     intermediate CA has a trust setting included that is used. The first
3684     setting is used: whether to trust (e.g., -addtrust option to the x509
3685     utility) or reject.
3686     [Steve Henson]
3687
3688  *) Add -trusted_first option which attempts to find certificates in the
3689     trusted store even if an untrusted chain is also supplied.
3690     [Steve Henson]
3691
3692  *) MIPS assembly pack updates: support for MIPS32r2 and SmartMIPS ASE,
3693     platform support for Linux and Android.
3694     [Andy Polyakov]
3695
3696  *) Support for linux-x32, ILP32 environment in x86_64 framework.
3697     [Andy Polyakov]
3698
3699  *) Experimental multi-implementation support for FIPS capable OpenSSL.
3700     When in FIPS mode the approved implementations are used as normal,
3701     when not in FIPS mode the internal unapproved versions are used instead.
3702     This means that the FIPS capable OpenSSL isn't forced to use the
3703     (often lower performance) FIPS implementations outside FIPS mode.
3704     [Steve Henson]
3705
3706  *) Transparently support X9.42 DH parameters when calling
3707     PEM_read_bio_DHparameters. This means existing applications can handle
3708     the new parameter format automatically.
3709     [Steve Henson]
3710
3711  *) Initial experimental support for X9.42 DH parameter format: mainly
3712     to support use of 'q' parameter for RFC5114 parameters.
3713     [Steve Henson]
3714
3715  *) Add DH parameters from RFC5114 including test data to dhtest.
3716     [Steve Henson]
3717
3718  *) Support for automatic EC temporary key parameter selection. If enabled
3719     the most preferred EC parameters are automatically used instead of
3720     hardcoded fixed parameters. Now a server just has to call:
3721     SSL_CTX_set_ecdh_auto(ctx, 1) and the server will automatically
3722     support ECDH and use the most appropriate parameters.
3723     [Steve Henson]
3724
3725  *) Enhance and tidy EC curve and point format TLS extension code. Use
3726     static structures instead of allocation if default values are used.
3727     New ctrls to set curves we wish to support and to retrieve shared curves.
3728     Print out shared curves in s_server. New options to s_server and s_client
3729     to set list of supported curves.
3730     [Steve Henson]
3731
3732  *) New ctrls to retrieve supported signature algorithms and
3733     supported curve values as an array of NIDs. Extend openssl utility
3734     to print out received values.
3735     [Steve Henson]
3736
3737  *) Add new APIs EC_curve_nist2nid and EC_curve_nid2nist which convert
3738     between NIDs and the more common NIST names such as "P-256". Enhance
3739     ecparam utility and ECC method to recognise the NIST names for curves.
3740     [Steve Henson]
3741
3742  *) Enhance SSL/TLS certificate chain handling to support different
3743     chains for each certificate instead of one chain in the parent SSL_CTX.
3744     [Steve Henson]
3745
3746  *) Support for fixed DH ciphersuite client authentication: where both
3747     server and client use DH certificates with common parameters.
3748     [Steve Henson]
3749
3750  *) Support for fixed DH ciphersuites: those requiring DH server
3751     certificates.
3752     [Steve Henson]
3753
3754  *) New function i2d_re_X509_tbs for re-encoding the TBS portion of
3755     the certificate.
3756     Note: Related 1.0.2-beta specific macros X509_get_cert_info,
3757     X509_CINF_set_modified, X509_CINF_get_issuer, X509_CINF_get_extensions and
3758     X509_CINF_get_signature were reverted post internal team review.
3759
3760 Changes between 1.0.1k and 1.0.1l [15 Jan 2015]
3761
3762  *) Build fixes for the Windows and OpenVMS platforms
3763     [Matt Caswell and Richard Levitte]
3764
3765 Changes between 1.0.1j and 1.0.1k [8 Jan 2015]
3766
3767  *) Fix DTLS segmentation fault in dtls1_get_record. A carefully crafted DTLS
3768     message can cause a segmentation fault in OpenSSL due to a NULL pointer
3769     dereference. This could lead to a Denial Of Service attack. Thanks to
3770     Markus Stenberg of Cisco Systems, Inc. for reporting this issue.
3771     (CVE-2014-3571)
3772     [Steve Henson]
3773
3774  *) Fix DTLS memory leak in dtls1_buffer_record. A memory leak can occur in the
3775     dtls1_buffer_record function under certain conditions. In particular this
3776     could occur if an attacker sent repeated DTLS records with the same
3777     sequence number but for the next epoch. The memory leak could be exploited
3778     by an attacker in a Denial of Service attack through memory exhaustion.
3779     Thanks to Chris Mueller for reporting this issue.
3780     (CVE-2015-0206)
3781     [Matt Caswell]
3782
3783  *) Fix issue where no-ssl3 configuration sets method to NULL. When openssl is
3784     built with the no-ssl3 option and a SSL v3 ClientHello is received the ssl
3785     method would be set to NULL which could later result in a NULL pointer
3786     dereference. Thanks to Frank Schmirler for reporting this issue.
3787     (CVE-2014-3569)
3788     [Kurt Roeckx]
3789
3790  *) Abort handshake if server key exchange message is omitted for ephemeral
3791     ECDH ciphersuites.
3792
3793     Thanks to Karthikeyan Bhargavan of the PROSECCO team at INRIA for
3794     reporting this issue.
3795     (CVE-2014-3572)
3796     [Steve Henson]
3797
3798  *) Remove non-export ephemeral RSA code on client and server. This code
3799     violated the TLS standard by allowing the use of temporary RSA keys in
3800     non-export ciphersuites and could be used by a server to effectively
3801     downgrade the RSA key length used to a value smaller than the server
3802     certificate. Thanks for Karthikeyan Bhargavan of the PROSECCO team at
3803     INRIA or reporting this issue.
3804     (CVE-2015-0204)
3805     [Steve Henson]
3806
3807  *) Fixed issue where DH client certificates are accepted without verification.
3808     An OpenSSL server will accept a DH certificate for client authentication
3809     without the certificate verify message. This effectively allows a client to
3810     authenticate without the use of a private key. This only affects servers
3811     which trust a client certificate authority which issues certificates
3812     containing DH keys: these are extremely rare and hardly ever encountered.
3813     Thanks for Karthikeyan Bhargavan of the PROSECCO team at INRIA or reporting
3814     this issue.
3815     (CVE-2015-0205)
3816     [Steve Henson]
3817
3818  *) Ensure that the session ID context of an SSL is updated when its
3819     SSL_CTX is updated via SSL_set_SSL_CTX.
3820
3821     The session ID context is typically set from the parent SSL_CTX,
3822     and can vary with the CTX.
3823     [Adam Langley]
3824
3825  *) Fix various certificate fingerprint issues.
3826
3827     By using non-DER or invalid encodings outside the signed portion of a
3828     certificate the fingerprint can be changed without breaking the signature.
3829     Although no details of the signed portion of the certificate can be changed
3830     this can cause problems with some applications: e.g. those using the
3831     certificate fingerprint for blacklists.
3832
3833     1. Reject signatures with non zero unused bits.
3834
3835     If the BIT STRING containing the signature has non zero unused bits reject
3836     the signature. All current signature algorithms require zero unused bits.
3837
3838     2. Check certificate algorithm consistency.
3839
3840     Check the AlgorithmIdentifier inside TBS matches the one in the
3841     certificate signature. NB: this will result in signature failure
3842     errors for some broken certificates.
3843
3844     Thanks to Konrad Kraszewski from Google for reporting this issue.
3845
3846     3. Check DSA/ECDSA signatures use DER.
3847
3848     Re-encode DSA/ECDSA signatures and compare with the original received
3849     signature. Return an error if there is a mismatch.
3850
3851     This will reject various cases including garbage after signature
3852     (thanks to Antti Karjalainen and Tuomo Untinen from the Codenomicon CROSS
3853     program for discovering this case) and use of BER or invalid ASN.1 INTEGERs
3854     (negative or with leading zeroes).
3855
3856     Further analysis was conducted and fixes were developed by Stephen Henson
3857     of the OpenSSL core team.
3858
3859     (CVE-2014-8275)
3860     [Steve Henson]
3861
3862   *) Correct Bignum squaring. Bignum squaring (BN_sqr) may produce incorrect
3863      results on some platforms, including x86_64. This bug occurs at random
3864      with a very low probability, and is not known to be exploitable in any
3865      way, though its exact impact is difficult to determine. Thanks to Pieter
3866      Wuille (Blockstream) who reported this issue and also suggested an initial
3867      fix. Further analysis was conducted by the OpenSSL development team and
3868      Adam Langley of Google. The final fix was developed by Andy Polyakov of
3869      the OpenSSL core team.
3870      (CVE-2014-3570)
3871      [Andy Polyakov]
3872
3873   *) Do not resume sessions on the server if the negotiated protocol
3874      version does not match the session's version. Resuming with a different
3875      version, while not strictly forbidden by the RFC, is of questionable
3876      sanity and breaks all known clients.
3877      [David Benjamin, Emilia Käsper]
3878
3879   *) Tighten handling of the ChangeCipherSpec (CCS) message: reject
3880      early CCS messages during renegotiation. (Note that because
3881      renegotiation is encrypted, this early CCS was not exploitable.)
3882      [Emilia Käsper]
3883
3884   *) Tighten client-side session ticket handling during renegotiation:
3885      ensure that the client only accepts a session ticket if the server sends
3886      the extension anew in the ServerHello. Previously, a TLS client would
3887      reuse the old extension state and thus accept a session ticket if one was
3888      announced in the initial ServerHello.
3889
3890      Similarly, ensure that the client requires a session ticket if one
3891      was advertised in the ServerHello. Previously, a TLS client would
3892      ignore a missing NewSessionTicket message.
3893      [Emilia Käsper]
3894
3895 Changes between 1.0.1i and 1.0.1j [15 Oct 2014]
3896
3897  *) SRTP Memory Leak.
3898
3899     A flaw in the DTLS SRTP extension parsing code allows an attacker, who
3900     sends a carefully crafted handshake message, to cause OpenSSL to fail
3901     to free up to 64k of memory causing a memory leak. This could be
3902     exploited in a Denial Of Service attack. This issue affects OpenSSL
3903     1.0.1 server implementations for both SSL/TLS and DTLS regardless of
3904     whether SRTP is used or configured. Implementations of OpenSSL that
3905     have been compiled with OPENSSL_NO_SRTP defined are not affected.
3906
3907     The fix was developed by the OpenSSL team.
3908     (CVE-2014-3513)
3909     [OpenSSL team]
3910
3911  *) Session Ticket Memory Leak.
3912
3913     When an OpenSSL SSL/TLS/DTLS server receives a session ticket the
3914     integrity of that ticket is first verified. In the event of a session
3915     ticket integrity check failing, OpenSSL will fail to free memory
3916     causing a memory leak. By sending a large number of invalid session
3917     tickets an attacker could exploit this issue in a Denial Of Service
3918     attack.
3919     (CVE-2014-3567)
3920     [Steve Henson]
3921
3922  *) Build option no-ssl3 is incomplete.
3923
3924     When OpenSSL is configured with "no-ssl3" as a build option, servers
3925     could accept and complete a SSL 3.0 handshake, and clients could be
3926     configured to send them.
3927     (CVE-2014-3568)
3928     [Akamai and the OpenSSL team]
3929
3930  *) Add support for TLS_FALLBACK_SCSV.
3931     Client applications doing fallback retries should call
3932     SSL_set_mode(s, SSL_MODE_SEND_FALLBACK_SCSV).
3933     (CVE-2014-3566)
3934     [Adam Langley, Bodo Moeller]
3935
3936  *) Add additional DigestInfo checks.
3937
3938     Re-encode DigestInto in DER and check against the original when
3939     verifying RSA signature: this will reject any improperly encoded
3940     DigestInfo structures.
3941
3942     Note: this is a precautionary measure and no attacks are currently known.
3943
3944     [Steve Henson]
3945
3946 Changes between 1.0.1h and 1.0.1i [6 Aug 2014]
3947
3948  *) Fix SRP buffer overrun vulnerability. Invalid parameters passed to the
3949     SRP code can be overrun an internal buffer. Add sanity check that
3950     g, A, B < N to SRP code.
3951
3952     Thanks to Sean Devlin and Watson Ladd of Cryptography Services, NCC
3953     Group for discovering this issue.
3954     (CVE-2014-3512)
3955     [Steve Henson]
3956
3957  *) A flaw in the OpenSSL SSL/TLS server code causes the server to negotiate
3958     TLS 1.0 instead of higher protocol versions when the ClientHello message
3959     is badly fragmented. This allows a man-in-the-middle attacker to force a
3960     downgrade to TLS 1.0 even if both the server and the client support a
3961     higher protocol version, by modifying the client's TLS records.
3962
3963     Thanks to David Benjamin and Adam Langley (Google) for discovering and
3964     researching this issue.
3965     (CVE-2014-3511)
3966     [David Benjamin]
3967
3968  *) OpenSSL DTLS clients enabling anonymous (EC)DH ciphersuites are subject
3969     to a denial of service attack. A malicious server can crash the client
3970     with a null pointer dereference (read) by specifying an anonymous (EC)DH
3971     ciphersuite and sending carefully crafted handshake messages.
3972
3973     Thanks to Felix Gröbert (Google) for discovering and researching this
3974     issue.
3975     (CVE-2014-3510)
3976     [Emilia Käsper]
3977
3978  *) By sending carefully crafted DTLS packets an attacker could cause openssl
3979     to leak memory. This can be exploited through a Denial of Service attack.
3980     Thanks to Adam Langley for discovering and researching this issue.
3981     (CVE-2014-3507)
3982     [Adam Langley]
3983
3984  *) An attacker can force openssl to consume large amounts of memory whilst
3985     processing DTLS handshake messages. This can be exploited through a
3986     Denial of Service attack.
3987     Thanks to Adam Langley for discovering and researching this issue.
3988     (CVE-2014-3506)
3989     [Adam Langley]
3990
3991  *) An attacker can force an error condition which causes openssl to crash
3992     whilst processing DTLS packets due to memory being freed twice. This
3993     can be exploited through a Denial of Service attack.
3994     Thanks to Adam Langley and Wan-Teh Chang for discovering and researching
3995     this issue.
3996     (CVE-2014-3505)
3997     [Adam Langley]
3998
3999  *) If a multithreaded client connects to a malicious server using a resumed
4000     session and the server sends an ec point format extension it could write
4001     up to 255 bytes to freed memory.
4002
4003     Thanks to Gabor Tyukasz (LogMeIn Inc) for discovering and researching this
4004     issue.
4005     (CVE-2014-3509)
4006     [Gabor Tyukasz]
4007
4008  *) A malicious server can crash an OpenSSL client with a null pointer
4009     dereference (read) by specifying an SRP ciphersuite even though it was not
4010     properly negotiated with the client. This can be exploited through a
4011     Denial of Service attack.
4012
4013     Thanks to Joonas Kuorilehto and Riku Hietamäki (Codenomicon) for
4014     discovering and researching this issue.
4015     (CVE-2014-5139)
4016     [Steve Henson]
4017
4018  *) A flaw in OBJ_obj2txt may cause pretty printing functions such as
4019     X509_name_oneline, X509_name_print_ex et al. to leak some information
4020     from the stack. Applications may be affected if they echo pretty printing
4021     output to the attacker.
4022
4023     Thanks to Ivan Fratric (Google) for discovering this issue.
4024     (CVE-2014-3508)
4025     [Emilia Käsper, and Steve Henson]
4026
4027  *) Fix ec_GFp_simple_points_make_affine (thus, EC_POINTs_mul etc.)
4028     for corner cases. (Certain input points at infinity could lead to
4029     bogus results, with non-infinity inputs mapped to infinity too.)
4030     [Bodo Moeller]
4031
4032 Changes between 1.0.1g and 1.0.1h [5 Jun 2014]
4033
4034  *) Fix for SSL/TLS MITM flaw. An attacker using a carefully crafted
4035     handshake can force the use of weak keying material in OpenSSL
4036     SSL/TLS clients and servers.
4037
4038     Thanks to KIKUCHI Masashi (Lepidum Co. Ltd.) for discovering and
4039     researching this issue. (CVE-2014-0224)
4040     [KIKUCHI Masashi, Steve Henson]
4041
4042  *) Fix DTLS recursion flaw. By sending an invalid DTLS handshake to an
4043     OpenSSL DTLS client the code can be made to recurse eventually crashing
4044     in a DoS attack.
4045
4046     Thanks to Imre Rad (Search-Lab Ltd.) for discovering this issue.
4047     (CVE-2014-0221)
4048     [Imre Rad, Steve Henson]
4049
4050  *) Fix DTLS invalid fragment vulnerability. A buffer overrun attack can
4051     be triggered by sending invalid DTLS fragments to an OpenSSL DTLS
4052     client or server. This is potentially exploitable to run arbitrary
4053     code on a vulnerable client or server.
4054
4055     Thanks to Jüri Aedla for reporting this issue. (CVE-2014-0195)
4056     [Jüri Aedla, Steve Henson]
4057
4058  *) Fix bug in TLS code where clients enable anonymous ECDH ciphersuites
4059     are subject to a denial of service attack.
4060
4061     Thanks to Felix Gröbert and Ivan Fratric at Google for discovering
4062     this issue. (CVE-2014-3470)
4063     [Felix Gröbert, Ivan Fratric, Steve Henson]
4064
4065  *) Harmonize version and its documentation. -f flag is used to display
4066     compilation flags.
4067     [mancha <mancha1@zoho.com>]
4068
4069  *) Fix eckey_priv_encode so it immediately returns an error upon a failure
4070     in i2d_ECPrivateKey.
4071     [mancha <mancha1@zoho.com>]
4072
4073  *) Fix some double frees. These are not thought to be exploitable.
4074     [mancha <mancha1@zoho.com>]
4075
4076 Changes between 1.0.1f and 1.0.1g [7 Apr 2014]
4077
4078  *) A missing bounds check in the handling of the TLS heartbeat extension
4079     can be used to reveal up to 64k of memory to a connected client or
4080     server.
4081
4082     Thanks for Neel Mehta of Google Security for discovering this bug and to
4083     Adam Langley <agl@chromium.org> and Bodo Moeller <bmoeller@acm.org> for
4084     preparing the fix (CVE-2014-0160)
4085     [Adam Langley, Bodo Moeller]
4086
4087  *) Fix for the attack described in the paper "Recovering OpenSSL
4088     ECDSA Nonces Using the FLUSH+RELOAD Cache Side-channel Attack"
4089     by Yuval Yarom and Naomi Benger. Details can be obtained from:
4090     http://eprint.iacr.org/2014/140
4091
4092     Thanks to Yuval Yarom and Naomi Benger for discovering this
4093     flaw and to Yuval Yarom for supplying a fix (CVE-2014-0076)
4094     [Yuval Yarom and Naomi Benger]
4095
4096  *) TLS pad extension: draft-agl-tls-padding-03
4097
4098     Workaround for the "TLS hang bug" (see FAQ and PR#2771): if the
4099     TLS client Hello record length value would otherwise be > 255 and
4100     less that 512 pad with a dummy extension containing zeroes so it
4101     is at least 512 bytes long.
4102
4103     [Adam Langley, Steve Henson]
4104
4105 Changes between 1.0.1e and 1.0.1f [6 Jan 2014]
4106
4107  *) Fix for TLS record tampering bug. A carefully crafted invalid
4108     handshake could crash OpenSSL with a NULL pointer exception.
4109     Thanks to Anton Johansson for reporting this issues.
4110     (CVE-2013-4353)
4111
4112  *) Keep original DTLS digest and encryption contexts in retransmission
4113     structures so we can use the previous session parameters if they need
4114     to be resent. (CVE-2013-6450)
4115     [Steve Henson]
4116
4117  *) Add option SSL_OP_SAFARI_ECDHE_ECDSA_BUG (part of SSL_OP_ALL) which
4118     avoids preferring ECDHE-ECDSA ciphers when the client appears to be
4119     Safari on OS X.  Safari on OS X 10.8..10.8.3 advertises support for
4120     several ECDHE-ECDSA ciphers, but fails to negotiate them.  The bug
4121     is fixed in OS X 10.8.4, but Apple have ruled out both hot fixing
4122     10.8..10.8.3 and forcing users to upgrade to 10.8.4 or newer.
4123     [Rob Stradling, Adam Langley]
4124
4125 Changes between 1.0.1d and 1.0.1e [11 Feb 2013]
4126
4127  *) Correct fix for CVE-2013-0169. The original didn't work on AES-NI
4128     supporting platforms or when small records were transferred.
4129     [Andy Polyakov, Steve Henson]
4130
4131 Changes between 1.0.1c and 1.0.1d [5 Feb 2013]
4132
4133  *) Make the decoding of SSLv3, TLS and DTLS CBC records constant time.
4134
4135     This addresses the flaw in CBC record processing discovered by
4136     Nadhem Alfardan and Kenny Paterson. Details of this attack can be found
4137     at: http://www.isg.rhul.ac.uk/tls/
4138
4139     Thanks go to Nadhem Alfardan and Kenny Paterson of the Information
4140     Security Group at Royal Holloway, University of London
4141     (www.isg.rhul.ac.uk) for discovering this flaw and Adam Langley and
4142     Emilia Käsper for the initial patch.
4143     (CVE-2013-0169)
4144     [Emilia Käsper, Adam Langley, Ben Laurie, Andy Polyakov, Steve Henson]
4145
4146  *) Fix flaw in AESNI handling of TLS 1.2 and 1.1 records for CBC mode
4147     ciphersuites which can be exploited in a denial of service attack.
4148     Thanks go to and to Adam Langley <agl@chromium.org> for discovering
4149     and detecting this bug and to Wolfgang Ettlinger
4150     <wolfgang.ettlinger@gmail.com> for independently discovering this issue.
4151     (CVE-2012-2686)
4152     [Adam Langley]
4153
4154  *) Return an error when checking OCSP signatures when key is NULL.
4155     This fixes a DoS attack. (CVE-2013-0166)
4156     [Steve Henson]
4157
4158  *) Make openssl verify return errors.
4159     [Chris Palmer <palmer@google.com> and Ben Laurie]
4160
4161  *) Call OCSP Stapling callback after ciphersuite has been chosen, so
4162     the right response is stapled. Also change SSL_get_certificate()
4163     so it returns the certificate actually sent.
4164     See http://rt.openssl.org/Ticket/Display.html?id=2836.
4165     [Rob Stradling <rob.stradling@comodo.com>]
4166
4167  *) Fix possible deadlock when decoding public keys.
4168     [Steve Henson]
4169
4170  *) Don't use TLS 1.0 record version number in initial client hello
4171     if renegotiating.
4172     [Steve Henson]
4173
4174 Changes between 1.0.1b and 1.0.1c [10 May 2012]
4175
4176  *) Sanity check record length before skipping explicit IV in TLS
4177     1.2, 1.1 and DTLS to fix DoS attack.
4178
4179     Thanks to Codenomicon for discovering this issue using Fuzz-o-Matic
4180     fuzzing as a service testing platform.
4181     (CVE-2012-2333)
4182     [Steve Henson]
4183
4184  *) Initialise tkeylen properly when encrypting CMS messages.
4185     Thanks to Solar Designer of Openwall for reporting this issue.
4186     [Steve Henson]
4187
4188  *) In FIPS mode don't try to use composite ciphers as they are not
4189     approved.
4190     [Steve Henson]
4191
4192 Changes between 1.0.1a and 1.0.1b [26 Apr 2012]
4193
4194  *) OpenSSL 1.0.0 sets SSL_OP_ALL to 0x80000FFFL and OpenSSL 1.0.1 and
4195     1.0.1a set SSL_OP_NO_TLSv1_1 to 0x00000400L which would unfortunately
4196     mean any application compiled against OpenSSL 1.0.0 headers setting
4197     SSL_OP_ALL would also set SSL_OP_NO_TLSv1_1, unintentionally disabling
4198     TLS 1.1 also. Fix this by changing the value of SSL_OP_NO_TLSv1_1 to
4199     0x10000000L Any application which was previously compiled against
4200     OpenSSL 1.0.1 or 1.0.1a headers and which cares about SSL_OP_NO_TLSv1_1
4201     will need to be recompiled as a result. Letting be results in
4202     inability to disable specifically TLS 1.1 and in client context,
4203     in unlike event, limit maximum offered version to TLS 1.0 [see below].
4204     [Steve Henson]
4205
4206  *) In order to ensure interoperability SSL_OP_NO_protocolX does not
4207     disable just protocol X, but all protocols above X *if* there are
4208     protocols *below* X still enabled. In more practical terms it means
4209     that if application wants to disable TLS1.0 in favor of TLS1.1 and
4210     above, it's not sufficient to pass SSL_OP_NO_TLSv1, one has to pass
4211     SSL_OP_NO_TLSv1|SSL_OP_NO_SSLv3|SSL_OP_NO_SSLv2. This applies to
4212     client side.
4213     [Andy Polyakov]
4214
4215 Changes between 1.0.1 and 1.0.1a [19 Apr 2012]
4216
4217  *) Check for potentially exploitable overflows in asn1_d2i_read_bio
4218     BUF_mem_grow and BUF_mem_grow_clean. Refuse attempts to shrink buffer
4219     in CRYPTO_realloc_clean.
4220
4221     Thanks to Tavis Ormandy, Google Security Team, for discovering this
4222     issue and to Adam Langley <agl@chromium.org> for fixing it.
4223     (CVE-2012-2110)
4224     [Adam Langley (Google), Tavis Ormandy, Google Security Team]
4225
4226  *) Don't allow TLS 1.2 SHA-256 ciphersuites in TLS 1.0, 1.1 connections.
4227     [Adam Langley]
4228
4229  *) Workarounds for some broken servers that "hang" if a client hello
4230     record length exceeds 255 bytes.
4231
4232     1. Do not use record version number > TLS 1.0 in initial client
4233        hello: some (but not all) hanging servers will now work.
4234     2. If we set OPENSSL_MAX_TLS1_2_CIPHER_LENGTH this will truncate
4235        the number of ciphers sent in the client hello. This should be
4236        set to an even number, such as 50, for example by passing:
4237        -DOPENSSL_MAX_TLS1_2_CIPHER_LENGTH=50 to config or Configure.
4238        Most broken servers should now work.
4239     3. If all else fails setting OPENSSL_NO_TLS1_2_CLIENT will disable
4240        TLS 1.2 client support entirely.
4241     [Steve Henson]
4242
4243  *) Fix SEGV in Vector Permutation AES module observed in OpenSSH.
4244     [Andy Polyakov]
4245
4246 Changes between 1.0.0h and 1.0.1  [14 Mar 2012]
4247
4248  *) Add compatibility with old MDC2 signatures which use an ASN1 OCTET
4249     STRING form instead of a DigestInfo.
4250     [Steve Henson]
4251
4252  *) The format used for MDC2 RSA signatures is inconsistent between EVP
4253     and the RSA_sign/RSA_verify functions. This was made more apparent when
4254     OpenSSL used RSA_sign/RSA_verify for some RSA signatures in particular
4255     those which went through EVP_PKEY_METHOD in 1.0.0 and later. Detect
4256     the correct format in RSA_verify so both forms transparently work.
4257     [Steve Henson]
4258
4259  *) Some servers which support TLS 1.0 can choke if we initially indicate
4260     support for TLS 1.2 and later renegotiate using TLS 1.0 in the RSA
4261     encrypted premaster secret. As a workaround use the maximum permitted
4262     client version in client hello, this should keep such servers happy
4263     and still work with previous versions of OpenSSL.
4264     [Steve Henson]
4265
4266  *) Add support for TLS/DTLS heartbeats.
4267     [Robin Seggelmann <seggelmann@fh-muenster.de>]
4268
4269  *) Add support for SCTP.
4270     [Robin Seggelmann <seggelmann@fh-muenster.de>]
4271
4272  *) Improved PRNG seeding for VOS.
4273     [Paul Green <Paul.Green@stratus.com>]
4274
4275  *) Extensive assembler packs updates, most notably:
4276
4277        - x86[_64]:     AES-NI, PCLMULQDQ, RDRAND support;
4278        - x86[_64]:     SSSE3 support (SHA1, vector-permutation AES);
4279        - x86_64:       bit-sliced AES implementation;
4280        - ARM:          NEON support, contemporary platforms optimizations;
4281        - s390x:        z196 support;
4282        - *:            GHASH and GF(2^m) multiplication implementations;
4283
4284     [Andy Polyakov]
4285
4286  *) Make TLS-SRP code conformant with RFC 5054 API cleanup
4287     (removal of unnecessary code)
4288     [Peter Sylvester <peter.sylvester@edelweb.fr>]
4289
4290  *) Add TLS key material exporter from RFC 5705.
4291     [Eric Rescorla]
4292
4293  *) Add DTLS-SRTP negotiation from RFC 5764.
4294     [Eric Rescorla]
4295
4296  *) Add Next Protocol Negotiation,
4297     http://tools.ietf.org/html/draft-agl-tls-nextprotoneg-00. Can be
4298     disabled with a no-npn flag to config or Configure. Code donated
4299     by Google.
4300     [Adam Langley <agl@google.com> and Ben Laurie]
4301
4302  *) Add optional 64-bit optimized implementations of elliptic curves NIST-P224,
4303     NIST-P256, NIST-P521, with constant-time single point multiplication on
4304     typical inputs. Compiler support for the nonstandard type __uint128_t is
4305     required to use this (present in gcc 4.4 and later, for 64-bit builds).
4306     Code made available under Apache License version 2.0.
4307
4308     Specify "enable-ec_nistp_64_gcc_128" on the Configure (or config) command
4309     line to include this in your build of OpenSSL, and run "make depend" (or
4310     "make update"). This enables the following EC_METHODs:
4311
4312         EC_GFp_nistp224_method()
4313         EC_GFp_nistp256_method()
4314         EC_GFp_nistp521_method()
4315
4316     EC_GROUP_new_by_curve_name() will automatically use these (while
4317     EC_GROUP_new_curve_GFp() currently prefers the more flexible
4318     implementations).
4319     [Emilia Käsper, Adam Langley, Bodo Moeller (Google)]
4320
4321  *) Use type ossl_ssize_t instead of ssize_t which isn't available on
4322     all platforms. Move ssize_t definition from e_os.h to the public
4323     header file e_os2.h as it now appears in public header file cms.h
4324     [Steve Henson]
4325
4326  *) New -sigopt option to the ca, req and x509 utilities. Additional
4327     signature parameters can be passed using this option and in
4328     particular PSS.
4329     [Steve Henson]
4330
4331  *) Add RSA PSS signing function. This will generate and set the
4332     appropriate AlgorithmIdentifiers for PSS based on those in the
4333     corresponding EVP_MD_CTX structure. No application support yet.
4334     [Steve Henson]
4335
4336  *) Support for companion algorithm specific ASN1 signing routines.
4337     New function ASN1_item_sign_ctx() signs a pre-initialised
4338     EVP_MD_CTX structure and sets AlgorithmIdentifiers based on
4339     the appropriate parameters.
4340     [Steve Henson]
4341
4342  *) Add new algorithm specific ASN1 verification initialisation function
4343     to EVP_PKEY_ASN1_METHOD: this is not in EVP_PKEY_METHOD since the ASN1
4344     handling will be the same no matter what EVP_PKEY_METHOD is used.
4345     Add a PSS handler to support verification of PSS signatures: checked
4346     against a number of sample certificates.
4347     [Steve Henson]
4348
4349  *) Add signature printing for PSS. Add PSS OIDs.
4350     [Steve Henson, Martin Kaiser <lists@kaiser.cx>]
4351
4352  *) Add algorithm specific signature printing. An individual ASN1 method
4353     can now print out signatures instead of the standard hex dump.
4354
4355     More complex signatures (e.g. PSS) can print out more meaningful
4356     information. Include DSA version that prints out the signature
4357     parameters r, s.
4358     [Steve Henson]
4359
4360  *) Password based recipient info support for CMS library: implementing
4361     RFC3211.
4362     [Steve Henson]
4363
4364  *) Split password based encryption into PBES2 and PBKDF2 functions. This
4365     neatly separates the code into cipher and PBE sections and is required
4366     for some algorithms that split PBES2 into separate pieces (such as
4367     password based CMS).
4368     [Steve Henson]
4369
4370  *) Session-handling fixes:
4371     - Fix handling of connections that are resuming with a session ID,
4372       but also support Session Tickets.
4373     - Fix a bug that suppressed issuing of a new ticket if the client
4374       presented a ticket with an expired session.
4375     - Try to set the ticket lifetime hint to something reasonable.
4376     - Make tickets shorter by excluding irrelevant information.
4377     - On the client side, don't ignore renewed tickets.
4378     [Adam Langley, Bodo Moeller (Google)]
4379
4380  *) Fix PSK session representation.
4381     [Bodo Moeller]
4382
4383  *) Add RC4-MD5 and AESNI-SHA1 "stitched" implementations.
4384
4385     This work was sponsored by Intel.
4386     [Andy Polyakov]
4387
4388  *) Add GCM support to TLS library. Some custom code is needed to split
4389     the IV between the fixed (from PRF) and explicit (from TLS record)
4390     portions. This adds all GCM ciphersuites supported by RFC5288 and
4391     RFC5289. Generalise some AES* cipherstrings to include GCM and
4392     add a special AESGCM string for GCM only.
4393     [Steve Henson]
4394
4395  *) Expand range of ctrls for AES GCM. Permit setting invocation
4396     field on decrypt and retrieval of invocation field only on encrypt.
4397     [Steve Henson]
4398
4399  *) Add HMAC ECC ciphersuites from RFC5289. Include SHA384 PRF support.
4400     As required by RFC5289 these ciphersuites cannot be used if for
4401     versions of TLS earlier than 1.2.
4402     [Steve Henson]
4403
4404  *) For FIPS capable OpenSSL interpret a NULL default public key method
4405     as unset and return the appropriate default but do *not* set the default.
4406     This means we can return the appropriate method in applications that
4407     switch between FIPS and non-FIPS modes.
4408     [Steve Henson]
4409
4410  *) Redirect HMAC and CMAC operations to FIPS module in FIPS mode. If an
4411     ENGINE is used then we cannot handle that in the FIPS module so we
4412     keep original code iff non-FIPS operations are allowed.
4413     [Steve Henson]
4414
4415  *) Add -attime option to openssl utilities.
4416     [Peter Eckersley <pde@eff.org>, Ben Laurie and Steve Henson]
4417
4418  *) Redirect DSA and DH operations to FIPS module in FIPS mode.
4419     [Steve Henson]
4420
4421  *) Redirect ECDSA and ECDH operations to FIPS module in FIPS mode. Also use
4422     FIPS EC methods unconditionally for now.
4423     [Steve Henson]
4424
4425  *) New build option no-ec2m to disable characteristic 2 code.
4426     [Steve Henson]
4427
4428  *) Backport libcrypto audit of return value checking from 1.1.0-dev; not
4429     all cases can be covered as some introduce binary incompatibilities.
4430     [Steve Henson]
4431
4432  *) Redirect RSA operations to FIPS module including keygen,
4433     encrypt, decrypt, sign and verify. Block use of non FIPS RSA methods.
4434     [Steve Henson]
4435
4436  *) Add similar low level API blocking to ciphers.
4437     [Steve Henson]
4438
4439  *) Low level digest APIs are not approved in FIPS mode: any attempt
4440     to use these will cause a fatal error. Applications that *really* want
4441     to use them can use the private_* version instead.
4442     [Steve Henson]
4443
4444  *) Redirect cipher operations to FIPS module for FIPS builds.
4445     [Steve Henson]
4446
4447  *) Redirect digest operations to FIPS module for FIPS builds.
4448     [Steve Henson]
4449
4450  *) Update build system to add "fips" flag which will link in fipscanister.o
4451     for static and shared library builds embedding a signature if needed.
4452     [Steve Henson]
4453
4454  *) Output TLS supported curves in preference order instead of numerical
4455     order. This is currently hardcoded for the highest order curves first.
4456     This should be configurable so applications can judge speed vs strength.
4457     [Steve Henson]
4458
4459  *) Add TLS v1.2 server support for client authentication.
4460     [Steve Henson]
4461
4462  *) Add support for FIPS mode in ssl library: disable SSLv3, non-FIPS ciphers
4463     and enable MD5.
4464     [Steve Henson]
4465
4466  *) Functions FIPS_mode_set() and FIPS_mode() which call the underlying
4467     FIPS modules versions.
4468     [Steve Henson]
4469
4470  *) Add TLS v1.2 client side support for client authentication. Keep cache
4471     of handshake records longer as we don't know the hash algorithm to use
4472     until after the certificate request message is received.
4473     [Steve Henson]
4474
4475  *) Initial TLS v1.2 client support. Add a default signature algorithms
4476     extension including all the algorithms we support. Parse new signature
4477     format in client key exchange. Relax some ECC signing restrictions for
4478     TLS v1.2 as indicated in RFC5246.
4479     [Steve Henson]
4480
4481  *) Add server support for TLS v1.2 signature algorithms extension. Switch
4482     to new signature format when needed using client digest preference.
4483     All server ciphersuites should now work correctly in TLS v1.2. No client
4484     support yet and no support for client certificates.
4485     [Steve Henson]
4486
4487  *) Initial TLS v1.2 support. Add new SHA256 digest to ssl code, switch
4488     to SHA256 for PRF when using TLS v1.2 and later. Add new SHA256 based
4489     ciphersuites. At present only RSA key exchange ciphersuites work with
4490     TLS v1.2. Add new option for TLS v1.2 replacing the old and obsolete
4491     SSL_OP_PKCS1_CHECK flags with SSL_OP_NO_TLSv1_2. New TLSv1.2 methods
4492     and version checking.
4493     [Steve Henson]
4494
4495  *) New option OPENSSL_NO_SSL_INTERN. If an application can be compiled
4496     with this defined it will not be affected by any changes to ssl internal
4497     structures. Add several utility functions to allow openssl application
4498     to work with OPENSSL_NO_SSL_INTERN defined.
4499     [Steve Henson]
4500
4501  *) A long standing patch to add support for SRP from EdelWeb (Peter
4502     Sylvester and Christophe Renou) was integrated.
4503     [Christophe Renou <christophe.renou@edelweb.fr>, Peter Sylvester
4504     <peter.sylvester@edelweb.fr>, Tom Wu <tjw@cs.stanford.edu>, and
4505     Ben Laurie]
4506
4507  *) Add functions to copy EVP_PKEY_METHOD and retrieve flags and id.
4508     [Steve Henson]
4509
4510  *) Permit abbreviated handshakes when renegotiating using the function
4511     SSL_renegotiate_abbreviated().
4512     [Robin Seggelmann <seggelmann@fh-muenster.de>]
4513
4514  *) Add call to ENGINE_register_all_complete() to
4515     ENGINE_load_builtin_engines(), so some implementations get used
4516     automatically instead of needing explicit application support.
4517     [Steve Henson]
4518
4519  *) Add support for TLS key exporter as described in RFC5705.
4520     [Robin Seggelmann <seggelmann@fh-muenster.de>, Steve Henson]
4521
4522  *) Initial TLSv1.1 support. Since TLSv1.1 is very similar to TLS v1.0 only
4523     a few changes are required:
4524
4525       Add SSL_OP_NO_TLSv1_1 flag.
4526       Add TLSv1_1 methods.
4527       Update version checking logic to handle version 1.1.
4528       Add explicit IV handling (ported from DTLS code).
4529       Add command line options to s_client/s_server.
4530     [Steve Henson]
4531
4532 Changes between 1.0.0g and 1.0.0h [12 Mar 2012]
4533
4534  *) Fix MMA (Bleichenbacher's attack on PKCS #1 v1.5 RSA padding) weakness
4535     in CMS and PKCS7 code. When RSA decryption fails use a random key for
4536     content decryption and always return the same error. Note: this attack
4537     needs on average 2^20 messages so it only affects automated senders. The
4538     old behaviour can be re-enabled in the CMS code by setting the
4539     CMS_DEBUG_DECRYPT flag: this is useful for debugging and testing where
4540     an MMA defence is not necessary.
4541     Thanks to Ivan Nestlerode <inestlerode@us.ibm.com> for discovering
4542     this issue. (CVE-2012-0884)
4543     [Steve Henson]
4544
4545  *) Fix CVE-2011-4619: make sure we really are receiving a
4546     client hello before rejecting multiple SGC restarts. Thanks to
4547     Ivan Nestlerode <inestlerode@us.ibm.com> for discovering this bug.
4548     [Steve Henson]
4549
4550 Changes between 1.0.0f and 1.0.0g [18 Jan 2012]
4551
4552  *) Fix for DTLS DoS issue introduced by fix for CVE-2011-4109.
4553     Thanks to Antonio Martin, Enterprise Secure Access Research and
4554     Development, Cisco Systems, Inc. for discovering this bug and
4555     preparing a fix. (CVE-2012-0050)
4556     [Antonio Martin]
4557
4558 Changes between 1.0.0e and 1.0.0f [4 Jan 2012]
4559
4560  *) Nadhem Alfardan and Kenny Paterson have discovered an extension
4561     of the Vaudenay padding oracle attack on CBC mode encryption
4562     which enables an efficient plaintext recovery attack against
4563     the OpenSSL implementation of DTLS. Their attack exploits timing
4564     differences arising during decryption processing. A research
4565     paper describing this attack can be found at:
4566                  http://www.isg.rhul.ac.uk/~kp/dtls.pdf
4567     Thanks go to Nadhem Alfardan and Kenny Paterson of the Information
4568     Security Group at Royal Holloway, University of London
4569     (www.isg.rhul.ac.uk) for discovering this flaw and to Robin Seggelmann
4570     <seggelmann@fh-muenster.de> and Michael Tuexen <tuexen@fh-muenster.de>
4571     for preparing the fix. (CVE-2011-4108)
4572     [Robin Seggelmann, Michael Tuexen]
4573
4574  *) Clear bytes used for block padding of SSL 3.0 records.
4575     (CVE-2011-4576)
4576     [Adam Langley (Google)]
4577
4578  *) Only allow one SGC handshake restart for SSL/TLS. Thanks to George
4579     Kadianakis <desnacked@gmail.com> for discovering this issue and
4580     Adam Langley for preparing the fix. (CVE-2011-4619)
4581     [Adam Langley (Google)]
4582
4583  *) Check parameters are not NULL in GOST ENGINE. (CVE-2012-0027)
4584     [Andrey Kulikov <amdeich@gmail.com>]
4585
4586  *) Prevent malformed RFC3779 data triggering an assertion failure.
4587     Thanks to Andrew Chi, BBN Technologies, for discovering the flaw
4588     and Rob Austein <sra@hactrn.net> for fixing it. (CVE-2011-4577)
4589     [Rob Austein <sra@hactrn.net>]
4590
4591  *) Improved PRNG seeding for VOS.
4592     [Paul Green <Paul.Green@stratus.com>]
4593
4594  *) Fix ssl_ciph.c set-up race.
4595     [Adam Langley (Google)]
4596
4597  *) Fix spurious failures in ecdsatest.c.
4598     [Emilia Käsper (Google)]
4599
4600  *) Fix the BIO_f_buffer() implementation (which was mixing different
4601     interpretations of the '..._len' fields).
4602     [Adam Langley (Google)]
4603
4604  *) Fix handling of BN_BLINDING: now BN_BLINDING_invert_ex (rather than
4605     BN_BLINDING_invert_ex) calls BN_BLINDING_update, ensuring that concurrent
4606     threads won't reuse the same blinding coefficients.
4607
4608     This also avoids the need to obtain the CRYPTO_LOCK_RSA_BLINDING
4609     lock to call BN_BLINDING_invert_ex, and avoids one use of
4610     BN_BLINDING_update for each BN_BLINDING structure (previously,
4611     the last update always remained unused).
4612     [Emilia Käsper (Google)]
4613
4614  *) In ssl3_clear, preserve s3->init_extra along with s3->rbuf.
4615     [Bob Buckholz (Google)]
4616
4617 Changes between 1.0.0d and 1.0.0e [6 Sep 2011]
4618
4619  *) Fix bug where CRLs with nextUpdate in the past are sometimes accepted
4620     by initialising X509_STORE_CTX properly. (CVE-2011-3207)
4621     [Kaspar Brand <ossl@velox.ch>]
4622
4623  *) Fix SSL memory handling for (EC)DH ciphersuites, in particular
4624     for multi-threaded use of ECDH. (CVE-2011-3210)
4625     [Adam Langley (Google)]
4626
4627  *) Fix x509_name_ex_d2i memory leak on bad inputs.
4628     [Bodo Moeller]
4629
4630  *) Remove hard coded ecdsaWithSHA1 signature tests in ssl code and check
4631     signature public key algorithm by using OID xref utilities instead.
4632     Before this you could only use some ECC ciphersuites with SHA1 only.
4633     [Steve Henson]
4634
4635  *) Add protection against ECDSA timing attacks as mentioned in the paper
4636     by Billy Bob Brumley and Nicola Tuveri, see:
4637
4638        http://eprint.iacr.org/2011/232.pdf
4639
4640     [Billy Bob Brumley and Nicola Tuveri]
4641
4642 Changes between 1.0.0c and 1.0.0d [8 Feb 2011]
4643
4644  *) Fix parsing of OCSP stapling ClientHello extension. CVE-2011-0014
4645     [Neel Mehta, Adam Langley, Bodo Moeller (Google)]
4646
4647  *) Fix bug in string printing code: if *any* escaping is enabled we must
4648     escape the escape character (backslash) or the resulting string is
4649     ambiguous.
4650     [Steve Henson]
4651
4652 Changes between 1.0.0b and 1.0.0c  [2 Dec 2010]
4653
4654  *) Disable code workaround for ancient and obsolete Netscape browsers
4655     and servers: an attacker can use it in a ciphersuite downgrade attack.
4656     Thanks to Martin Rex for discovering this bug. CVE-2010-4180
4657     [Steve Henson]
4658
4659  *) Fixed J-PAKE implementation error, originally discovered by
4660     Sebastien Martini, further info and confirmation from Stefan
4661     Arentz and Feng Hao. Note that this fix is a security fix. CVE-2010-4252
4662     [Ben Laurie]
4663
4664 Changes between 1.0.0a and 1.0.0b  [16 Nov 2010]
4665
4666  *) Fix extension code to avoid race conditions which can result in a buffer
4667     overrun vulnerability: resumed sessions must not be modified as they can
4668     be shared by multiple threads. CVE-2010-3864
4669     [Steve Henson]
4670
4671  *) Fix WIN32 build system to correctly link an ENGINE directory into
4672     a DLL.
4673     [Steve Henson]
4674
4675 Changes between 1.0.0 and 1.0.0a  [01 Jun 2010]
4676
4677  *) Check return value of int_rsa_verify in pkey_rsa_verifyrecover
4678     (CVE-2010-1633)
4679     [Steve Henson, Peter-Michael Hager <hager@dortmund.net>]
4680
4681 Changes between 0.9.8n and 1.0.0  [29 Mar 2010]
4682
4683  *) Add "missing" function EVP_CIPHER_CTX_copy(). This copies a cipher
4684     context. The operation can be customised via the ctrl mechanism in
4685     case ENGINEs want to include additional functionality.
4686     [Steve Henson]
4687
4688  *) Tolerate yet another broken PKCS#8 key format: private key value negative.
4689     [Steve Henson]
4690
4691  *) Add new -subject_hash_old and -issuer_hash_old options to x509 utility to
4692     output hashes compatible with older versions of OpenSSL.
4693     [Willy Weisz <weisz@vcpc.univie.ac.at>]
4694
4695  *) Fix compression algorithm handling: if resuming a session use the
4696     compression algorithm of the resumed session instead of determining
4697     it from client hello again. Don't allow server to change algorithm.
4698     [Steve Henson]
4699
4700  *) Add load_crls() function to apps tidying load_certs() too. Add option
4701     to verify utility to allow additional CRLs to be included.
4702     [Steve Henson]
4703
4704  *) Update OCSP request code to permit adding custom headers to the request:
4705     some responders need this.
4706     [Steve Henson]
4707
4708  *) The function EVP_PKEY_sign() returns <=0 on error: check return code
4709     correctly.
4710     [Julia Lawall <julia@diku.dk>]
4711
4712  *) Update verify callback code in apps/s_cb.c and apps/verify.c, it
4713     needlessly dereferenced structures, used obsolete functions and
4714     didn't handle all updated verify codes correctly.
4715     [Steve Henson]
4716
4717  *) Disable MD2 in the default configuration.
4718     [Steve Henson]
4719
4720  *) In BIO_pop() and BIO_push() use the ctrl argument (which was NULL) to
4721     indicate the initial BIO being pushed or popped. This makes it possible
4722     to determine whether the BIO is the one explicitly called or as a result
4723     of the ctrl being passed down the chain. Fix BIO_pop() and SSL BIOs so
4724     it handles reference counts correctly and doesn't zero out the I/O bio
4725     when it is not being explicitly popped. WARNING: applications which
4726     included workarounds for the old buggy behaviour will need to be modified
4727     or they could free up already freed BIOs.
4728     [Steve Henson]
4729
4730  *) Extend the uni2asc/asc2uni => OPENSSL_uni2asc/OPENSSL_asc2uni
4731     renaming to all platforms (within the 0.9.8 branch, this was
4732     done conditionally on Netware platforms to avoid a name clash).
4733     [Guenter <lists@gknw.net>]
4734
4735  *) Add ECDHE and PSK support to DTLS.
4736     [Michael Tuexen <tuexen@fh-muenster.de>]
4737
4738  *) Add CHECKED_STACK_OF macro to safestack.h, otherwise safestack can't
4739     be used on C++.
4740     [Steve Henson]
4741
4742  *) Add "missing" function EVP_MD_flags() (without this the only way to
4743     retrieve a digest flags is by accessing the structure directly. Update
4744     EVP_MD_do_all*() and EVP_CIPHER_do_all*() to include the name a digest
4745     or cipher is registered as in the "from" argument. Print out all
4746     registered digests in the dgst usage message instead of manually
4747     attempting to work them out.
4748     [Steve Henson]
4749
4750  *) If no SSLv2 ciphers are used don't use an SSLv2 compatible client hello:
4751     this allows the use of compression and extensions. Change default cipher
4752     string to remove SSLv2 ciphersuites. This effectively avoids ancient SSLv2
4753     by default unless an application cipher string requests it.
4754     [Steve Henson]
4755
4756  *) Alter match criteria in PKCS12_parse(). It used to try to use local
4757     key ids to find matching certificates and keys but some PKCS#12 files
4758     don't follow the (somewhat unwritten) rules and this strategy fails.
4759     Now just gather all certificates together and the first private key
4760     then look for the first certificate that matches the key.
4761     [Steve Henson]
4762
4763  *) Support use of registered digest and cipher names for dgst and cipher
4764     commands instead of having to add each one as a special case. So now
4765     you can do:
4766
4767        openssl sha256 foo
4768
4769     as well as:
4770
4771        openssl dgst -sha256 foo
4772
4773     and this works for ENGINE based algorithms too.
4774
4775     [Steve Henson]
4776
4777  *) Update Gost ENGINE to support parameter files.
4778     [Victor B. Wagner <vitus@cryptocom.ru>]
4779
4780  *) Support GeneralizedTime in ca utility.
4781     [Oliver Martin <oliver@volatilevoid.net>, Steve Henson]
4782
4783  *) Enhance the hash format used for certificate directory links. The new
4784     form uses the canonical encoding (meaning equivalent names will work
4785     even if they aren't identical) and uses SHA1 instead of MD5. This form
4786     is incompatible with the older format and as a result c_rehash should
4787     be used to rebuild symbolic links.
4788     [Steve Henson]
4789
4790  *) Make PKCS#8 the default write format for private keys, replacing the
4791     traditional format. This form is standardised, more secure and doesn't
4792     include an implicit MD5 dependency.
4793     [Steve Henson]
4794
4795  *) Add a $gcc_devteam_warn option to Configure. The idea is that any code
4796     committed to OpenSSL should pass this lot as a minimum.
4797     [Steve Henson]
4798
4799  *) Add session ticket override functionality for use by EAP-FAST.
4800     [Jouni Malinen <j@w1.fi>]
4801
4802  *) Modify HMAC functions to return a value. Since these can be implemented
4803     in an ENGINE errors can occur.
4804     [Steve Henson]
4805
4806  *) Type-checked OBJ_bsearch_ex.
4807     [Ben Laurie]
4808
4809  *) Type-checked OBJ_bsearch. Also some constification necessitated
4810     by type-checking.  Still to come: TXT_DB, bsearch(?),
4811     OBJ_bsearch_ex, qsort, CRYPTO_EX_DATA, ASN1_VALUE, ASN1_STRING,
4812     CONF_VALUE.
4813     [Ben Laurie]
4814
4815  *) New function OPENSSL_gmtime_adj() to add a specific number of days and
4816     seconds to a tm structure directly, instead of going through OS
4817     specific date routines. This avoids any issues with OS routines such
4818     as the year 2038 bug. New *_adj() functions for ASN1 time structures
4819     and X509_time_adj_ex() to cover the extended range. The existing
4820     X509_time_adj() is still usable and will no longer have any date issues.
4821     [Steve Henson]
4822
4823  *) Delta CRL support. New use deltas option which will attempt to locate
4824     and search any appropriate delta CRLs available.
4825
4826     This work was sponsored by Google.
4827     [Steve Henson]
4828
4829  *) Support for CRLs partitioned by reason code. Reorganise CRL processing
4830     code and add additional score elements. Validate alternate CRL paths
4831     as part of the CRL checking and indicate a new error "CRL path validation
4832     error" in this case. Applications wanting additional details can use
4833     the verify callback and check the new "parent" field. If this is not
4834     NULL CRL path validation is taking place. Existing applications won't
4835     see this because it requires extended CRL support which is off by
4836     default.
4837
4838     This work was sponsored by Google.
4839     [Steve Henson]
4840
4841  *) Support for freshest CRL extension.
4842
4843     This work was sponsored by Google.
4844     [Steve Henson]
4845
4846  *) Initial indirect CRL support. Currently only supported in the CRLs
4847     passed directly and not via lookup. Process certificate issuer
4848     CRL entry extension and lookup CRL entries by bother issuer name
4849     and serial number. Check and process CRL issuer entry in IDP extension.
4850
4851     This work was sponsored by Google.
4852     [Steve Henson]
4853
4854  *) Add support for distinct certificate and CRL paths. The CRL issuer
4855     certificate is validated separately in this case. Only enabled if
4856     an extended CRL support flag is set: this flag will enable additional
4857     CRL functionality in future.
4858
4859     This work was sponsored by Google.
4860     [Steve Henson]
4861
4862  *) Add support for policy mappings extension.
4863
4864     This work was sponsored by Google.
4865     [Steve Henson]
4866
4867  *) Fixes to pathlength constraint, self issued certificate handling,
4868     policy processing to align with RFC3280 and PKITS tests.
4869
4870     This work was sponsored by Google.
4871     [Steve Henson]
4872
4873  *) Support for name constraints certificate extension. DN, email, DNS
4874     and URI types are currently supported.
4875
4876     This work was sponsored by Google.
4877     [Steve Henson]
4878
4879  *) To cater for systems that provide a pointer-based thread ID rather
4880     than numeric, deprecate the current numeric thread ID mechanism and
4881     replace it with a structure and associated callback type. This
4882     mechanism allows a numeric "hash" to be extracted from a thread ID in
4883     either case, and on platforms where pointers are larger than 'long',
4884     mixing is done to help ensure the numeric 'hash' is usable even if it
4885     can't be guaranteed unique. The default mechanism is to use "&errno"
4886     as a pointer-based thread ID to distinguish between threads.
4887
4888     Applications that want to provide their own thread IDs should now use
4889     CRYPTO_THREADID_set_callback() to register a callback that will call
4890     either CRYPTO_THREADID_set_numeric() or CRYPTO_THREADID_set_pointer().
4891
4892     Note that ERR_remove_state() is now deprecated, because it is tied
4893     to the assumption that thread IDs are numeric.  ERR_remove_state(0)
4894     to free the current thread's error state should be replaced by
4895     ERR_remove_thread_state(NULL).
4896
4897     (This new approach replaces the functions CRYPTO_set_idptr_callback(),
4898     CRYPTO_get_idptr_callback(), and CRYPTO_thread_idptr() that existed in
4899     OpenSSL 0.9.9-dev between June 2006 and August 2008. Also, if an
4900     application was previously providing a numeric thread callback that
4901     was inappropriate for distinguishing threads, then uniqueness might
4902     have been obtained with &errno that happened immediately in the
4903     intermediate development versions of OpenSSL; this is no longer the
4904     case, the numeric thread callback will now override the automatic use
4905     of &errno.)
4906     [Geoff Thorpe, with help from Bodo Moeller]
4907
4908  *) Initial support for different CRL issuing certificates. This covers a
4909     simple case where the self issued certificates in the chain exist and
4910     the real CRL issuer is higher in the existing chain.
4911
4912     This work was sponsored by Google.
4913     [Steve Henson]
4914
4915  *) Removed effectively defunct crypto/store from the build.
4916     [Ben Laurie]
4917
4918  *) Revamp of STACK to provide stronger type-checking. Still to come:
4919     TXT_DB, bsearch(?), OBJ_bsearch, qsort, CRYPTO_EX_DATA, ASN1_VALUE,
4920     ASN1_STRING, CONF_VALUE.
4921     [Ben Laurie]
4922
4923  *) Add a new SSL_MODE_RELEASE_BUFFERS mode flag to release unused buffer
4924     RAM on SSL connections.  This option can save about 34k per idle SSL.
4925     [Nick Mathewson]
4926
4927  *) Revamp of LHASH to provide stronger type-checking. Still to come:
4928     STACK, TXT_DB, bsearch, qsort.
4929     [Ben Laurie]
4930
4931  *) Initial support for Cryptographic Message Syntax (aka CMS) based
4932     on RFC3850, RFC3851 and RFC3852. New cms directory and cms utility,
4933     support for data, signedData, compressedData, digestedData and
4934     encryptedData, envelopedData types included. Scripts to check against
4935     RFC4134 examples draft and interop and consistency checks of many
4936     content types and variants.
4937     [Steve Henson]
4938
4939  *) Add options to enc utility to support use of zlib compression BIO.
4940     [Steve Henson]
4941
4942  *) Extend mk1mf to support importing of options and assembly language
4943     files from Configure script, currently only included in VC-WIN32.
4944     The assembly language rules can now optionally generate the source
4945     files from the associated perl scripts.
4946     [Steve Henson]
4947
4948  *) Implement remaining functionality needed to support GOST ciphersuites.
4949     Interop testing has been performed using CryptoPro implementations.
4950     [Victor B. Wagner <vitus@cryptocom.ru>]
4951
4952  *) s390x assembler pack.
4953     [Andy Polyakov]
4954
4955  *) ARMv4 assembler pack. ARMv4 refers to v4 and later ISA, not CPU
4956     "family."
4957     [Andy Polyakov]
4958
4959  *) Implement Opaque PRF Input TLS extension as specified in
4960     draft-rescorla-tls-opaque-prf-input-00.txt.  Since this is not an
4961     official specification yet and no extension type assignment by
4962     IANA exists, this extension (for now) will have to be explicitly
4963     enabled when building OpenSSL by providing the extension number
4964     to use.  For example, specify an option
4965
4966         -DTLSEXT_TYPE_opaque_prf_input=0x9527
4967
4968     to the "config" or "Configure" script to enable the extension,
4969     assuming extension number 0x9527 (which is a completely arbitrary
4970     and unofficial assignment based on the MD5 hash of the Internet
4971     Draft).  Note that by doing so, you potentially lose
4972     interoperability with other TLS implementations since these might
4973     be using the same extension number for other purposes.
4974
4975     SSL_set_tlsext_opaque_prf_input(ssl, src, len) is used to set the
4976     opaque PRF input value to use in the handshake.  This will create
4977     an internal copy of the length-'len' string at 'src', and will
4978     return non-zero for success.
4979
4980     To get more control and flexibility, provide a callback function
4981     by using
4982
4983          SSL_CTX_set_tlsext_opaque_prf_input_callback(ctx, cb)
4984          SSL_CTX_set_tlsext_opaque_prf_input_callback_arg(ctx, arg)
4985
4986     where
4987
4988          int (*cb)(SSL *, void *peerinput, size_t len, void *arg);
4989          void *arg;
4990
4991     Callback function 'cb' will be called in handshakes, and is
4992     expected to use SSL_set_tlsext_opaque_prf_input() as appropriate.
4993     Argument 'arg' is for application purposes (the value as given to
4994     SSL_CTX_set_tlsext_opaque_prf_input_callback_arg() will directly
4995     be provided to the callback function).  The callback function
4996     has to return non-zero to report success: usually 1 to use opaque
4997     PRF input just if possible, or 2 to enforce use of the opaque PRF
4998     input.  In the latter case, the library will abort the handshake
4999     if opaque PRF input is not successfully negotiated.
5000
5001     Arguments 'peerinput' and 'len' given to the callback function
5002     will always be NULL and 0 in the case of a client.  A server will
5003     see the client's opaque PRF input through these variables if
5004     available (NULL and 0 otherwise).  Note that if the server
5005     provides an opaque PRF input, the length must be the same as the
5006     length of the client's opaque PRF input.
5007
5008     Note that the callback function will only be called when creating
5009     a new session (session resumption can resume whatever was
5010     previously negotiated), and will not be called in SSL 2.0
5011     handshakes; thus, SSL_CTX_set_options(ctx, SSL_OP_NO_SSLv2) or
5012     SSL_set_options(ssl, SSL_OP_NO_SSLv2) is especially recommended
5013     for applications that need to enforce opaque PRF input.
5014
5015     [Bodo Moeller]
5016
5017  *) Update ssl code to support digests other than SHA1+MD5 for handshake
5018     MAC.
5019
5020     [Victor B. Wagner <vitus@cryptocom.ru>]
5021
5022  *) Add RFC4507 support to OpenSSL. This includes the corrections in
5023     RFC4507bis. The encrypted ticket format is an encrypted encoded
5024     SSL_SESSION structure, that way new session features are automatically
5025     supported.
5026
5027     If a client application caches session in an SSL_SESSION structure
5028     support is transparent because tickets are now stored in the encoded
5029     SSL_SESSION.
5030
5031     The SSL_CTX structure automatically generates keys for ticket
5032     protection in servers so again support should be possible
5033     with no application modification.
5034
5035     If a client or server wishes to disable RFC4507 support then the option
5036     SSL_OP_NO_TICKET can be set.
5037
5038     Add a TLS extension debugging callback to allow the contents of any client
5039     or server extensions to be examined.
5040
5041     This work was sponsored by Google.
5042     [Steve Henson]
5043
5044  *) Final changes to avoid use of pointer pointer casts in OpenSSL.
5045     OpenSSL should now compile cleanly on gcc 4.2
5046     [Peter Hartley <pdh@utter.chaos.org.uk>, Steve Henson]
5047
5048  *) Update SSL library to use new EVP_PKEY MAC API. Include generic MAC
5049     support including streaming MAC support: this is required for GOST
5050     ciphersuite support.
5051     [Victor B. Wagner <vitus@cryptocom.ru>, Steve Henson]
5052
5053  *) Add option -stream to use PKCS#7 streaming in smime utility. New
5054     function i2d_PKCS7_bio_stream() and PEM_write_PKCS7_bio_stream()
5055     to output in BER and PEM format.
5056     [Steve Henson]
5057
5058  *) Experimental support for use of HMAC via EVP_PKEY interface. This
5059     allows HMAC to be handled via the EVP_DigestSign*() interface. The
5060     EVP_PKEY "key" in this case is the HMAC key, potentially allowing
5061     ENGINE support for HMAC keys which are unextractable. New -mac and
5062     -macopt options to dgst utility.
5063     [Steve Henson]
5064
5065  *) New option -sigopt to dgst utility. Update dgst to use
5066     EVP_Digest{Sign,Verify}*. These two changes make it possible to use
5067     alternative signing parameters such as X9.31 or PSS in the dgst
5068     utility.
5069     [Steve Henson]
5070
5071  *) Change ssl_cipher_apply_rule(), the internal function that does
5072     the work each time a ciphersuite string requests enabling
5073     ("foo+bar"), moving ("+foo+bar"), disabling ("-foo+bar", or
5074     removing ("!foo+bar") a class of ciphersuites: Now it maintains
5075     the order of disabled ciphersuites such that those ciphersuites
5076     that most recently went from enabled to disabled not only stay
5077     in order with respect to each other, but also have higher priority
5078     than other disabled ciphersuites the next time ciphersuites are
5079     enabled again.
5080
5081     This means that you can now say, e.g., "PSK:-PSK:HIGH" to enable
5082     the same ciphersuites as with "HIGH" alone, but in a specific
5083     order where the PSK ciphersuites come first (since they are the
5084     most recently disabled ciphersuites when "HIGH" is parsed).
5085
5086     Also, change ssl_create_cipher_list() (using this new
5087     functionality) such that between otherwise identical
5088     ciphersuites, ephemeral ECDH is preferred over ephemeral DH in
5089     the default order.
5090     [Bodo Moeller]
5091
5092  *) Change ssl_create_cipher_list() so that it automatically
5093     arranges the ciphersuites in reasonable order before starting
5094     to process the rule string.  Thus, the definition for "DEFAULT"
5095     (SSL_DEFAULT_CIPHER_LIST) now is just "ALL:!aNULL:!eNULL", but
5096     remains equivalent to "AES:ALL:!aNULL:!eNULL:+aECDH:+kRSA:+RC4:@STRENGTH".
5097     This makes it much easier to arrive at a reasonable default order
5098     in applications for which anonymous ciphers are OK (meaning
5099     that you can't actually use DEFAULT).
5100     [Bodo Moeller; suggested by Victor Duchovni]
5101
5102  *) Split the SSL/TLS algorithm mask (as used for ciphersuite string
5103     processing) into multiple integers instead of setting
5104     "SSL_MKEY_MASK" bits, "SSL_AUTH_MASK" bits, "SSL_ENC_MASK",
5105     "SSL_MAC_MASK", and "SSL_SSL_MASK" bits all in a single integer.
5106     (These masks as well as the individual bit definitions are hidden
5107     away into the non-exported interface ssl/ssl_locl.h, so this
5108     change to the definition of the SSL_CIPHER structure shouldn't
5109     affect applications.)  This give us more bits for each of these
5110     categories, so there is no longer a need to coagulate AES128 and
5111     AES256 into a single algorithm bit, and to coagulate Camellia128
5112     and Camellia256 into a single algorithm bit, which has led to all
5113     kinds of kludges.
5114
5115     Thus, among other things, the kludge introduced in 0.9.7m and
5116     0.9.8e for masking out AES256 independently of AES128 or masking
5117     out Camellia256 independently of AES256 is not needed here in 0.9.9.
5118
5119     With the change, we also introduce new ciphersuite aliases that
5120     so far were missing: "AES128", "AES256", "CAMELLIA128", and
5121     "CAMELLIA256".
5122     [Bodo Moeller]
5123
5124  *) Add support for dsa-with-SHA224 and dsa-with-SHA256.
5125     Use the leftmost N bytes of the signature input if the input is
5126     larger than the prime q (with N being the size in bytes of q).
5127     [Nils Larsch]
5128
5129  *) Very *very* experimental PKCS#7 streaming encoder support. Nothing uses
5130     it yet and it is largely untested.
5131     [Steve Henson]
5132
5133  *) Add support for the ecdsa-with-SHA224/256/384/512 signature types.
5134     [Nils Larsch]
5135
5136  *) Initial incomplete changes to avoid need for function casts in OpenSSL
5137     some compilers (gcc 4.2 and later) reject their use. Safestack is
5138     reimplemented.  Update ASN1 to avoid use of legacy functions.
5139     [Steve Henson]
5140
5141  *) Win32/64 targets are linked with Winsock2.
5142     [Andy Polyakov]
5143
5144  *) Add an X509_CRL_METHOD structure to allow CRL processing to be redirected
5145     to external functions. This can be used to increase CRL handling
5146     efficiency especially when CRLs are very large by (for example) storing
5147     the CRL revoked certificates in a database.
5148     [Steve Henson]
5149
5150  *) Overhaul of by_dir code. Add support for dynamic loading of CRLs so
5151     new CRLs added to a directory can be used. New command line option
5152     -verify_return_error to s_client and s_server. This causes real errors
5153     to be returned by the verify callback instead of carrying on no matter
5154     what. This reflects the way a "real world" verify callback would behave.
5155     [Steve Henson]
5156
5157  *) GOST engine, supporting several GOST algorithms and public key formats.
5158     Kindly donated by Cryptocom.
5159     [Cryptocom]
5160
5161  *) Partial support for Issuing Distribution Point CRL extension. CRLs
5162     partitioned by DP are handled but no indirect CRL or reason partitioning
5163     (yet). Complete overhaul of CRL handling: now the most suitable CRL is
5164     selected via a scoring technique which handles IDP and AKID in CRLs.
5165     [Steve Henson]
5166
5167  *) New X509_STORE_CTX callbacks lookup_crls() and lookup_certs() which
5168     will ultimately be used for all verify operations: this will remove the
5169     X509_STORE dependency on certificate verification and allow alternative
5170     lookup methods.  X509_STORE based implementations of these two callbacks.
5171     [Steve Henson]
5172
5173  *) Allow multiple CRLs to exist in an X509_STORE with matching issuer names.
5174     Modify get_crl() to find a valid (unexpired) CRL if possible.
5175     [Steve Henson]
5176
5177  *) New function X509_CRL_match() to check if two CRLs are identical. Normally
5178     this would be called X509_CRL_cmp() but that name is already used by
5179     a function that just compares CRL issuer names. Cache several CRL
5180     extensions in X509_CRL structure and cache CRLDP in X509.
5181     [Steve Henson]
5182
5183  *) Store a "canonical" representation of X509_NAME structure (ASN1 Name)
5184     this maps equivalent X509_NAME structures into a consistent structure.
5185     Name comparison can then be performed rapidly using memcmp().
5186     [Steve Henson]
5187
5188  *) Non-blocking OCSP request processing. Add -timeout option to ocsp
5189     utility.
5190     [Steve Henson]
5191
5192  *) Allow digests to supply their own micalg string for S/MIME type using
5193     the ctrl EVP_MD_CTRL_MICALG.
5194     [Steve Henson]
5195
5196  *) During PKCS7 signing pass the PKCS7 SignerInfo structure to the
5197     EVP_PKEY_METHOD before and after signing via the EVP_PKEY_CTRL_PKCS7_SIGN
5198     ctrl. It can then customise the structure before and/or after signing
5199     if necessary.
5200     [Steve Henson]
5201
5202  *) New function OBJ_add_sigid() to allow application defined signature OIDs
5203     to be added to OpenSSLs internal tables. New function OBJ_sigid_free()
5204     to free up any added signature OIDs.
5205     [Steve Henson]
5206
5207  *) New functions EVP_CIPHER_do_all(), EVP_CIPHER_do_all_sorted(),
5208     EVP_MD_do_all() and EVP_MD_do_all_sorted() to enumerate internal
5209     digest and cipher tables. New options added to openssl utility:
5210     list-message-digest-algorithms and list-cipher-algorithms.
5211     [Steve Henson]
5212
5213  *) Change the array representation of binary polynomials: the list
5214     of degrees of non-zero coefficients is now terminated with -1.
5215     Previously it was terminated with 0, which was also part of the
5216     value; thus, the array representation was not applicable to
5217     polynomials where t^0 has coefficient zero.  This change makes
5218     the array representation useful in a more general context.
5219     [Douglas Stebila]
5220
5221  *) Various modifications and fixes to SSL/TLS cipher string
5222     handling.  For ECC, the code now distinguishes between fixed ECDH
5223     with RSA certificates on the one hand and with ECDSA certificates
5224     on the other hand, since these are separate ciphersuites.  The
5225     unused code for Fortezza ciphersuites has been removed.
5226
5227     For consistency with EDH, ephemeral ECDH is now called "EECDH"
5228     (not "ECDHE").  For consistency with the code for DH
5229     certificates, use of ECDH certificates is now considered ECDH
5230     authentication, not RSA or ECDSA authentication (the latter is
5231     merely the CA's signing algorithm and not actively used in the
5232     protocol).
5233
5234     The temporary ciphersuite alias "ECCdraft" is no longer
5235     available, and ECC ciphersuites are no longer excluded from "ALL"
5236     and "DEFAULT".  The following aliases now exist for RFC 4492
5237     ciphersuites, most of these by analogy with the DH case:
5238
5239         kECDHr   - ECDH cert, signed with RSA
5240         kECDHe   - ECDH cert, signed with ECDSA
5241         kECDH    - ECDH cert (signed with either RSA or ECDSA)
5242         kEECDH   - ephemeral ECDH
5243         ECDH     - ECDH cert or ephemeral ECDH
5244
5245         aECDH    - ECDH cert
5246         aECDSA   - ECDSA cert
5247         ECDSA    - ECDSA cert
5248
5249         AECDH    - anonymous ECDH
5250         EECDH    - non-anonymous ephemeral ECDH (equivalent to "kEECDH:-AECDH")
5251
5252     [Bodo Moeller]
5253
5254  *) Add additional S/MIME capabilities for AES and GOST ciphers if supported.
5255     Use correct micalg parameters depending on digest(s) in signed message.
5256     [Steve Henson]
5257
5258  *) Add engine support for EVP_PKEY_ASN1_METHOD. Add functions to process
5259     an ENGINE asn1 method. Support ENGINE lookups in the ASN1 code.
5260     [Steve Henson]
5261
5262  *) Initial engine support for EVP_PKEY_METHOD. New functions to permit
5263     an engine to register a method. Add ENGINE lookups for methods and
5264     functional reference processing.
5265     [Steve Henson]
5266
5267  *) New functions EVP_Digest{Sign,Verify)*. These are enhanced versions of
5268     EVP_{Sign,Verify}* which allow an application to customise the signature
5269     process.
5270     [Steve Henson]
5271
5272  *) New -resign option to smime utility. This adds one or more signers
5273     to an existing PKCS#7 signedData structure. Also -md option to use an
5274     alternative message digest algorithm for signing.
5275     [Steve Henson]
5276
5277  *) Tidy up PKCS#7 routines and add new functions to make it easier to
5278     create PKCS7 structures containing multiple signers. Update smime
5279     application to support multiple signers.
5280     [Steve Henson]
5281
5282  *) New -macalg option to pkcs12 utility to allow setting of an alternative
5283     digest MAC.
5284     [Steve Henson]
5285
5286  *) Initial support for PKCS#5 v2.0 PRFs other than default SHA1 HMAC.
5287     Reorganize PBE internals to lookup from a static table using NIDs,
5288     add support for HMAC PBE OID translation. Add a EVP_CIPHER ctrl:
5289     EVP_CTRL_PBE_PRF_NID this allows a cipher to specify an alternative
5290     PRF which will be automatically used with PBES2.
5291     [Steve Henson]
5292
5293  *) Replace the algorithm specific calls to generate keys in "req" with the
5294     new API.
5295     [Steve Henson]
5296
5297  *) Update PKCS#7 enveloped data routines to use new API. This is now
5298     supported by any public key method supporting the encrypt operation. A
5299     ctrl is added to allow the public key algorithm to examine or modify
5300     the PKCS#7 RecipientInfo structure if it needs to: for RSA this is
5301     a no op.
5302     [Steve Henson]
5303
5304  *) Add a ctrl to asn1 method to allow a public key algorithm to express
5305     a default digest type to use. In most cases this will be SHA1 but some
5306     algorithms (such as GOST) need to specify an alternative digest. The
5307     return value indicates how strong the preference is 1 means optional and
5308     2 is mandatory (that is it is the only supported type). Modify
5309     ASN1_item_sign() to accept a NULL digest argument to indicate it should
5310     use the default md. Update openssl utilities to use the default digest
5311     type for signing if it is not explicitly indicated.
5312     [Steve Henson]
5313
5314  *) Use OID cross reference table in ASN1_sign() and ASN1_verify(). New
5315     EVP_MD flag EVP_MD_FLAG_PKEY_METHOD_SIGNATURE. This uses the relevant
5316     signing method from the key type. This effectively removes the link
5317     between digests and public key types.
5318     [Steve Henson]
5319
5320  *) Add an OID cross reference table and utility functions. Its purpose is to
5321     translate between signature OIDs such as SHA1WithrsaEncryption and SHA1,
5322     rsaEncryption. This will allow some of the algorithm specific hackery
5323     needed to use the correct OID to be removed.
5324     [Steve Henson]
5325
5326  *) Remove algorithm specific dependencies when setting PKCS7_SIGNER_INFO
5327     structures for PKCS7_sign(). They are now set up by the relevant public
5328     key ASN1 method.
5329     [Steve Henson]
5330
5331  *) Add provisional EC pkey method with support for ECDSA and ECDH.
5332     [Steve Henson]
5333
5334  *) Add support for key derivation (agreement) in the API, DH method and
5335     pkeyutl.
5336     [Steve Henson]
5337
5338  *) Add DSA pkey method and DH pkey methods, extend DH ASN1 method to support
5339     public and private key formats. As a side effect these add additional
5340     command line functionality not previously available: DSA signatures can be
5341     generated and verified using pkeyutl and DH key support and generation in
5342     pkey, genpkey.
5343     [Steve Henson]
5344
5345  *) BeOS support.
5346     [Oliver Tappe <zooey@hirschkaefer.de>]
5347
5348  *) New make target "install_html_docs" installs HTML renditions of the
5349     manual pages.
5350     [Oliver Tappe <zooey@hirschkaefer.de>]
5351
5352  *) New utility "genpkey" this is analogous to "genrsa" etc except it can
5353     generate keys for any algorithm. Extend and update EVP_PKEY_METHOD to
5354     support key and parameter generation and add initial key generation
5355     functionality for RSA.
5356     [Steve Henson]
5357
5358  *) Add functions for main EVP_PKEY_method operations. The undocumented
5359     functions EVP_PKEY_{encrypt,decrypt} have been renamed to
5360     EVP_PKEY_{encrypt,decrypt}_old.
5361     [Steve Henson]
5362
5363  *) Initial definitions for EVP_PKEY_METHOD. This will be a high level public
5364     key API, doesn't do much yet.
5365     [Steve Henson]
5366
5367  *) New function EVP_PKEY_asn1_get0_info() to retrieve information about
5368     public key algorithms. New option to openssl utility:
5369     "list-public-key-algorithms" to print out info.
5370     [Steve Henson]
5371
5372  *) Implement the Supported Elliptic Curves Extension for
5373     ECC ciphersuites from draft-ietf-tls-ecc-12.txt.
5374     [Douglas Stebila]
5375
5376  *) Don't free up OIDs in OBJ_cleanup() if they are in use by EVP_MD or
5377     EVP_CIPHER structures to avoid later problems in EVP_cleanup().
5378     [Steve Henson]
5379
5380  *) New utilities pkey and pkeyparam. These are similar to algorithm specific
5381     utilities such as rsa, dsa, dsaparam etc except they process any key
5382     type.
5383     [Steve Henson]
5384
5385  *) Transfer public key printing routines to EVP_PKEY_ASN1_METHOD. New
5386     functions EVP_PKEY_print_public(), EVP_PKEY_print_private(),
5387     EVP_PKEY_print_param() to print public key data from an EVP_PKEY
5388     structure.
5389     [Steve Henson]
5390
5391  *) Initial support for pluggable public key ASN1.
5392     De-spaghettify the public key ASN1 handling. Move public and private
5393     key ASN1 handling to a new EVP_PKEY_ASN1_METHOD structure. Relocate
5394     algorithm specific handling to a single module within the relevant
5395     algorithm directory. Add functions to allow (near) opaque processing
5396     of public and private key structures.
5397     [Steve Henson]
5398
5399  *) Implement the Supported Point Formats Extension for
5400     ECC ciphersuites from draft-ietf-tls-ecc-12.txt.
5401     [Douglas Stebila]
5402
5403  *) Add initial support for RFC 4279 PSK TLS ciphersuites. Add members
5404     for the psk identity [hint] and the psk callback functions to the
5405     SSL_SESSION, SSL and SSL_CTX structure.
5406
5407     New ciphersuites:
5408         PSK-RC4-SHA, PSK-3DES-EDE-CBC-SHA, PSK-AES128-CBC-SHA,
5409         PSK-AES256-CBC-SHA
5410
5411     New functions:
5412         SSL_CTX_use_psk_identity_hint
5413         SSL_get_psk_identity_hint
5414         SSL_get_psk_identity
5415         SSL_use_psk_identity_hint
5416
5417     [Mika Kousa and Pasi Eronen of Nokia Corporation]
5418
5419  *) Add RFC 3161 compliant time stamp request creation, response generation
5420     and response verification functionality.
5421     [Zoltán Glózik <zglozik@opentsa.org>, The OpenTSA Project]
5422
5423  *) Add initial support for TLS extensions, specifically for the server_name
5424     extension so far.  The SSL_SESSION, SSL_CTX, and SSL data structures now
5425     have new members for a host name.  The SSL data structure has an
5426     additional member SSL_CTX *initial_ctx so that new sessions can be
5427     stored in that context to allow for session resumption, even after the
5428     SSL has been switched to a new SSL_CTX in reaction to a client's
5429     server_name extension.
5430
5431     New functions (subject to change):
5432
5433         SSL_get_servername()
5434         SSL_get_servername_type()
5435         SSL_set_SSL_CTX()
5436
5437     New CTRL codes and macros (subject to change):
5438
5439         SSL_CTRL_SET_TLSEXT_SERVERNAME_CB
5440                                 - SSL_CTX_set_tlsext_servername_callback()
5441         SSL_CTRL_SET_TLSEXT_SERVERNAME_ARG
5442                                      - SSL_CTX_set_tlsext_servername_arg()
5443         SSL_CTRL_SET_TLSEXT_HOSTNAME           - SSL_set_tlsext_host_name()
5444
5445     openssl s_client has a new '-servername ...' option.
5446
5447     openssl s_server has new options '-servername_host ...', '-cert2 ...',
5448     '-key2 ...', '-servername_fatal' (subject to change).  This allows
5449     testing the HostName extension for a specific single host name ('-cert'
5450     and '-key' remain fallbacks for handshakes without HostName
5451     negotiation).  If the unrecognized_name alert has to be sent, this by
5452     default is a warning; it becomes fatal with the '-servername_fatal'
5453     option.
5454
5455     [Peter Sylvester,  Remy Allais, Christophe Renou]
5456
5457  *) Whirlpool hash implementation is added.
5458     [Andy Polyakov]
5459
5460  *) BIGNUM code on 64-bit SPARCv9 targets is switched from bn(64,64) to
5461     bn(64,32). Because of instruction set limitations it doesn't have
5462     any negative impact on performance. This was done mostly in order
5463     to make it possible to share assembler modules, such as bn_mul_mont
5464     implementations, between 32- and 64-bit builds without hassle.
5465     [Andy Polyakov]
5466
5467  *) Move code previously exiled into file crypto/ec/ec2_smpt.c
5468     to ec2_smpl.c, and no longer require the OPENSSL_EC_BIN_PT_COMP
5469     macro.
5470     [Bodo Moeller]
5471
5472  *) New candidate for BIGNUM assembler implementation, bn_mul_mont,
5473     dedicated Montgomery multiplication procedure, is introduced.
5474     BN_MONT_CTX is modified to allow bn_mul_mont to reach for higher
5475     "64-bit" performance on certain 32-bit targets.
5476     [Andy Polyakov]
5477
5478  *) New option SSL_OP_NO_COMP to disable use of compression selectively
5479     in SSL structures. New SSL ctrl to set maximum send fragment size.
5480     Save memory by setting the I/O buffer sizes dynamically instead of
5481     using the maximum available value.
5482     [Steve Henson]
5483
5484  *) New option -V for 'openssl ciphers'. This prints the ciphersuite code
5485     in addition to the text details.
5486     [Bodo Moeller]
5487
5488  *) Very, very preliminary EXPERIMENTAL support for printing of general
5489     ASN1 structures. This currently produces rather ugly output and doesn't
5490     handle several customised structures at all.
5491     [Steve Henson]
5492
5493  *) Integrated support for PVK file format and some related formats such
5494     as MS PUBLICKEYBLOB and PRIVATEKEYBLOB. Command line switches to support
5495     these in the 'rsa' and 'dsa' utilities.
5496     [Steve Henson]
5497
5498  *) Support for PKCS#1 RSAPublicKey format on rsa utility command line.
5499     [Steve Henson]
5500
5501  *) Remove the ancient ASN1_METHOD code. This was only ever used in one
5502     place for the (very old) "NETSCAPE" format certificates which are now
5503     handled using new ASN1 code equivalents.
5504     [Steve Henson]
5505
5506  *) Let the TLSv1_method() etc. functions return a 'const' SSL_METHOD
5507     pointer and make the SSL_METHOD parameter in SSL_CTX_new,
5508     SSL_CTX_set_ssl_version and SSL_set_ssl_method 'const'.
5509     [Nils Larsch]
5510
5511  *) Modify CRL distribution points extension code to print out previously
5512     unsupported fields. Enhance extension setting code to allow setting of
5513     all fields.
5514     [Steve Henson]
5515
5516  *) Add print and set support for Issuing Distribution Point CRL extension.
5517     [Steve Henson]
5518
5519  *) Change 'Configure' script to enable Camellia by default.
5520     [NTT]
5521
5522 Changes between 0.9.8m and 0.9.8n [24 Mar 2010]
5523
5524  *) When rejecting SSL/TLS records due to an incorrect version number, never
5525     update s->server with a new major version number.  As of
5526     - OpenSSL 0.9.8m if 'short' is a 16-bit type,
5527     - OpenSSL 0.9.8f if 'short' is longer than 16 bits,
5528     the previous behavior could result in a read attempt at NULL when
5529     receiving specific incorrect SSL/TLS records once record payload
5530     protection is active.  (CVE-2010-0740)
5531     [Bodo Moeller, Adam Langley <agl@chromium.org>]
5532
5533  *) Fix for CVE-2010-0433 where some kerberos enabled versions of OpenSSL
5534     could be crashed if the relevant tables were not present (e.g. chrooted).
5535     [Tomas Hoger <thoger@redhat.com>]
5536
5537 Changes between 0.9.8l and 0.9.8m [25 Feb 2010]
5538
5539  *) Always check bn_wexpand() return values for failure.  (CVE-2009-3245)
5540     [Martin Olsson, Neel Mehta]
5541
5542  *) Fix X509_STORE locking: Every 'objs' access requires a lock (to
5543     accommodate for stack sorting, always a write lock!).
5544     [Bodo Moeller]
5545
5546  *) On some versions of WIN32 Heap32Next is very slow. This can cause
5547     excessive delays in the RAND_poll(): over a minute. As a workaround
5548     include a time check in the inner Heap32Next loop too.
5549     [Steve Henson]
5550
5551  *) The code that handled flushing of data in SSL/TLS originally used the
5552     BIO_CTRL_INFO ctrl to see if any data was pending first. This caused
5553     the problem outlined in PR#1949. The fix suggested there however can
5554     trigger problems with buggy BIO_CTRL_WPENDING (e.g. some versions
5555     of Apache). So instead simplify the code to flush unconditionally.
5556     This should be fine since flushing with no data to flush is a no op.
5557     [Steve Henson]
5558
5559  *) Handle TLS versions 2.0 and later properly and correctly use the
5560     highest version of TLS/SSL supported. Although TLS >= 2.0 is some way
5561     off ancient servers have a habit of sticking around for a while...
5562     [Steve Henson]
5563
5564  *) Modify compression code so it frees up structures without using the
5565     ex_data callbacks. This works around a problem where some applications
5566     call CRYPTO_cleanup_all_ex_data() before application exit (e.g. when
5567     restarting) then use compression (e.g. SSL with compression) later.
5568     This results in significant per-connection memory leaks and
5569     has caused some security issues including CVE-2008-1678 and
5570     CVE-2009-4355.
5571     [Steve Henson]
5572
5573  *) Constify crypto/cast (i.e., <openssl/cast.h>): a CAST_KEY doesn't
5574     change when encrypting or decrypting.
5575     [Bodo Moeller]
5576
5577  *) Add option SSL_OP_LEGACY_SERVER_CONNECT which will allow clients to
5578     connect and renegotiate with servers which do not support RI.
5579     Until RI is more widely deployed this option is enabled by default.
5580     [Steve Henson]
5581
5582  *) Add "missing" ssl ctrls to clear options and mode.
5583     [Steve Henson]
5584
5585  *) If client attempts to renegotiate and doesn't support RI respond with
5586     a no_renegotiation alert as required by RFC5746.  Some renegotiating
5587     TLS clients will continue a connection gracefully when they receive
5588     the alert. Unfortunately OpenSSL mishandled this alert and would hang
5589     waiting for a server hello which it will never receive. Now we treat a
5590     received no_renegotiation alert as a fatal error. This is because
5591     applications requesting a renegotiation might well expect it to succeed
5592     and would have no code in place to handle the server denying it so the
5593     only safe thing to do is to terminate the connection.
5594     [Steve Henson]
5595
5596  *) Add ctrl macro SSL_get_secure_renegotiation_support() which returns 1 if
5597     peer supports secure renegotiation and 0 otherwise. Print out peer
5598     renegotiation support in s_client/s_server.
5599     [Steve Henson]
5600
5601  *) Replace the highly broken and deprecated SPKAC certification method with
5602     the updated NID creation version. This should correctly handle UTF8.
5603     [Steve Henson]
5604
5605  *) Implement RFC5746. Re-enable renegotiation but require the extension
5606     as needed. Unfortunately, SSL3_FLAGS_ALLOW_UNSAFE_LEGACY_RENEGOTIATION
5607     turns out to be a bad idea. It has been replaced by
5608     SSL_OP_ALLOW_UNSAFE_LEGACY_RENEGOTIATION which can be set with
5609     SSL_CTX_set_options(). This is really not recommended unless you
5610     know what you are doing.
5611     [Eric Rescorla <ekr@networkresonance.com>, Ben Laurie, Steve Henson]
5612
5613  *) Fixes to stateless session resumption handling. Use initial_ctx when
5614     issuing and attempting to decrypt tickets in case it has changed during
5615     servername handling. Use a non-zero length session ID when attempting
5616     stateless session resumption: this makes it possible to determine if
5617     a resumption has occurred immediately after receiving server hello
5618     (several places in OpenSSL subtly assume this) instead of later in
5619     the handshake.
5620     [Steve Henson]
5621
5622  *) The functions ENGINE_ctrl(), OPENSSL_isservice(),
5623     CMS_get1_RecipientRequest() and RAND_bytes() can return <=0 on error
5624     fixes for a few places where the return code is not checked
5625     correctly.
5626     [Julia Lawall <julia@diku.dk>]
5627
5628  *) Add --strict-warnings option to Configure script to include devteam
5629     warnings in other configurations.
5630     [Steve Henson]
5631
5632  *) Add support for --libdir option and LIBDIR variable in makefiles. This
5633     makes it possible to install openssl libraries in locations which
5634     have names other than "lib", for example "/usr/lib64" which some
5635     systems need.
5636     [Steve Henson, based on patch from Jeremy Utley]
5637
5638  *) Don't allow the use of leading 0x80 in OIDs. This is a violation of
5639     X690 8.9.12 and can produce some misleading textual output of OIDs.
5640     [Steve Henson, reported by Dan Kaminsky]
5641
5642  *) Delete MD2 from algorithm tables. This follows the recommendation in
5643     several standards that it is not used in new applications due to
5644     several cryptographic weaknesses. For binary compatibility reasons
5645     the MD2 API is still compiled in by default.
5646     [Steve Henson]
5647
5648  *) Add compression id to {d2i,i2d}_SSL_SESSION so it is correctly saved
5649     and restored.
5650     [Steve Henson]
5651
5652  *) Rename uni2asc and asc2uni functions to OPENSSL_uni2asc and
5653     OPENSSL_asc2uni conditionally on Netware platforms to avoid a name
5654     clash.
5655     [Guenter <lists@gknw.net>]
5656
5657  *) Fix the server certificate chain building code to use X509_verify_cert(),
5658     it used to have an ad-hoc builder which was unable to cope with anything
5659     other than a simple chain.
5660     [David Woodhouse <dwmw2@infradead.org>, Steve Henson]
5661
5662  *) Don't check self signed certificate signatures in X509_verify_cert()
5663     by default (a flag can override this): it just wastes time without
5664     adding any security. As a useful side effect self signed root CAs
5665     with non-FIPS digests are now usable in FIPS mode.
5666     [Steve Henson]
5667
5668  *) In dtls1_process_out_of_seq_message() the check if the current message
5669     is already buffered was missing. For every new message was memory
5670     allocated, allowing an attacker to perform an denial of service attack
5671     with sending out of seq handshake messages until there is no memory
5672     left. Additionally every future message was buffered, even if the
5673     sequence number made no sense and would be part of another handshake.
5674     So only messages with sequence numbers less than 10 in advance will be
5675     buffered.  (CVE-2009-1378)
5676     [Robin Seggelmann, discovered by Daniel Mentz]
5677
5678  *) Records are buffered if they arrive with a future epoch to be
5679     processed after finishing the corresponding handshake. There is
5680     currently no limitation to this buffer allowing an attacker to perform
5681     a DOS attack with sending records with future epochs until there is no
5682     memory left. This patch adds the pqueue_size() function to determine
5683     the size of a buffer and limits the record buffer to 100 entries.
5684     (CVE-2009-1377)
5685     [Robin Seggelmann, discovered by Daniel Mentz]
5686
5687  *) Keep a copy of frag->msg_header.frag_len so it can be used after the
5688     parent structure is freed.  (CVE-2009-1379)
5689     [Daniel Mentz]
5690
5691  *) Handle non-blocking I/O properly in SSL_shutdown() call.
5692     [Darryl Miles <darryl-mailinglists@netbauds.net>]
5693
5694  *) Add 2.5.4.* OIDs
5695     [Ilya O. <vrghost@gmail.com>]
5696
5697 Changes between 0.9.8k and 0.9.8l  [5 Nov 2009]
5698
5699  *) Disable renegotiation completely - this fixes a severe security
5700     problem (CVE-2009-3555) at the cost of breaking all
5701     renegotiation. Renegotiation can be re-enabled by setting
5702     SSL3_FLAGS_ALLOW_UNSAFE_LEGACY_RENEGOTIATION in s3->flags at
5703     run-time. This is really not recommended unless you know what
5704     you're doing.
5705     [Ben Laurie]
5706
5707 Changes between 0.9.8j and 0.9.8k  [25 Mar 2009]
5708
5709  *) Don't set val to NULL when freeing up structures, it is freed up by
5710     underlying code. If sizeof(void *) > sizeof(long) this can result in
5711     zeroing past the valid field. (CVE-2009-0789)
5712     [Paolo Ganci <Paolo.Ganci@AdNovum.CH>]
5713
5714  *) Fix bug where return value of CMS_SignerInfo_verify_content() was not
5715     checked correctly. This would allow some invalid signed attributes to
5716     appear to verify correctly. (CVE-2009-0591)
5717     [Ivan Nestlerode <inestlerode@us.ibm.com>]
5718
5719  *) Reject UniversalString and BMPString types with invalid lengths. This
5720     prevents a crash in ASN1_STRING_print_ex() which assumes the strings have
5721     a legal length. (CVE-2009-0590)
5722     [Steve Henson]
5723
5724  *) Set S/MIME signing as the default purpose rather than setting it
5725     unconditionally. This allows applications to override it at the store
5726     level.
5727     [Steve Henson]
5728
5729  *) Permit restricted recursion of ASN1 strings. This is needed in practice
5730     to handle some structures.
5731     [Steve Henson]
5732
5733  *) Improve efficiency of mem_gets: don't search whole buffer each time
5734     for a '\n'
5735     [Jeremy Shapiro <jnshapir@us.ibm.com>]
5736
5737  *) New -hex option for openssl rand.
5738     [Matthieu Herrb]
5739
5740  *) Print out UTF8String and NumericString when parsing ASN1.
5741     [Steve Henson]
5742
5743  *) Support NumericString type for name components.
5744     [Steve Henson]
5745
5746  *) Allow CC in the environment to override the automatically chosen
5747     compiler. Note that nothing is done to ensure flags work with the
5748     chosen compiler.
5749     [Ben Laurie]
5750
5751 Changes between 0.9.8i and 0.9.8j  [07 Jan 2009]
5752
5753  *) Properly check EVP_VerifyFinal() and similar return values
5754     (CVE-2008-5077).
5755     [Ben Laurie, Bodo Moeller, Google Security Team]
5756
5757  *) Enable TLS extensions by default.
5758     [Ben Laurie]
5759
5760  *) Allow the CHIL engine to be loaded, whether the application is
5761     multithreaded or not. (This does not release the developer from the
5762     obligation to set up the dynamic locking callbacks.)
5763     [Sander Temme <sander@temme.net>]
5764
5765  *) Use correct exit code if there is an error in dgst command.
5766     [Steve Henson; problem pointed out by Roland Dirlewanger]
5767
5768  *) Tweak Configure so that you need to say "experimental-jpake" to enable
5769     JPAKE, and need to use -DOPENSSL_EXPERIMENTAL_JPAKE in applications.
5770     [Bodo Moeller]
5771
5772  *) Add experimental JPAKE support, including demo authentication in
5773     s_client and s_server.
5774     [Ben Laurie]
5775
5776  *) Set the comparison function in v3_addr_canonize().
5777     [Rob Austein <sra@hactrn.net>]
5778
5779  *) Add support for XMPP STARTTLS in s_client.
5780     [Philip Paeps <philip@freebsd.org>]
5781
5782  *) Change the server-side SSL_OP_NETSCAPE_REUSE_CIPHER_CHANGE_BUG behavior
5783     to ensure that even with this option, only ciphersuites in the
5784     server's preference list will be accepted.  (Note that the option
5785     applies only when resuming a session, so the earlier behavior was
5786     just about the algorithm choice for symmetric cryptography.)
5787     [Bodo Moeller]
5788
5789 Changes between 0.9.8h and 0.9.8i  [15 Sep 2008]
5790
5791  *) Fix NULL pointer dereference if a DTLS server received
5792     ChangeCipherSpec as first record (CVE-2009-1386).
5793     [PR #1679]
5794
5795  *) Fix a state transition in s3_srvr.c and d1_srvr.c
5796     (was using SSL3_ST_CW_CLNT_HELLO_B, should be ..._ST_SW_SRVR_...).
5797     [Nagendra Modadugu]
5798
5799  *) The fix in 0.9.8c that supposedly got rid of unsafe
5800     double-checked locking was incomplete for RSA blinding,
5801     addressing just one layer of what turns out to have been
5802     doubly unsafe triple-checked locking.
5803
5804     So now fix this for real by retiring the MONT_HELPER macro
5805     in crypto/rsa/rsa_eay.c.
5806
5807     [Bodo Moeller; problem pointed out by Marius Schilder]
5808
5809  *) Various precautionary measures:
5810
5811     - Avoid size_t integer overflow in HASH_UPDATE (md32_common.h).
5812
5813     - Avoid a buffer overflow in d2i_SSL_SESSION() (ssl_asn1.c).
5814       (NB: This would require knowledge of the secret session ticket key
5815       to exploit, in which case you'd be SOL either way.)
5816
5817     - Change bn_nist.c so that it will properly handle input BIGNUMs
5818       outside the expected range.
5819
5820     - Enforce the 'num' check in BN_div() (bn_div.c) for non-BN_DEBUG
5821       builds.
5822
5823     [Neel Mehta, Bodo Moeller]
5824
5825  *) Allow engines to be "soft loaded" - i.e. optionally don't die if
5826     the load fails. Useful for distros.
5827     [Ben Laurie and the FreeBSD team]
5828
5829  *) Add support for Local Machine Keyset attribute in PKCS#12 files.
5830     [Steve Henson]
5831
5832  *) Fix BN_GF2m_mod_arr() top-bit cleanup code.
5833     [Huang Ying]
5834
5835  *) Expand ENGINE to support engine supplied SSL client certificate functions.
5836
5837     This work was sponsored by Logica.
5838     [Steve Henson]
5839
5840  *) Add CryptoAPI ENGINE to support use of RSA and DSA keys held in Windows
5841     keystores. Support for SSL/TLS client authentication too.
5842     Not compiled unless enable-capieng specified to Configure.
5843
5844     This work was sponsored by Logica.
5845     [Steve Henson]
5846
5847  *) Fix bug in X509_ATTRIBUTE creation: don't set attribute using
5848     ASN1_TYPE_set1 if MBSTRING flag set. This bug would crash certain
5849     attribute creation routines such as certificate requests and PKCS#12
5850     files.
5851     [Steve Henson]
5852
5853 Changes between 0.9.8g and 0.9.8h  [28 May 2008]
5854
5855  *) Fix flaw if 'Server Key exchange message' is omitted from a TLS
5856     handshake which could lead to a client crash as found using the
5857     Codenomicon TLS test suite (CVE-2008-1672)
5858     [Steve Henson, Mark Cox]
5859
5860  *) Fix double free in TLS server name extensions which could lead to
5861     a remote crash found by Codenomicon TLS test suite (CVE-2008-0891)
5862     [Joe Orton]
5863
5864  *) Clear error queue in SSL_CTX_use_certificate_chain_file()
5865
5866     Clear the error queue to ensure that error entries left from
5867     older function calls do not interfere with the correct operation.
5868     [Lutz Jaenicke, Erik de Castro Lopo]
5869
5870  *) Remove root CA certificates of commercial CAs:
5871
5872     The OpenSSL project does not recommend any specific CA and does not
5873     have any policy with respect to including or excluding any CA.
5874     Therefore it does not make any sense to ship an arbitrary selection
5875     of root CA certificates with the OpenSSL software.
5876     [Lutz Jaenicke]
5877
5878  *) RSA OAEP patches to fix two separate invalid memory reads.
5879     The first one involves inputs when 'lzero' is greater than
5880     'SHA_DIGEST_LENGTH' (it would read about SHA_DIGEST_LENGTH bytes
5881     before the beginning of from). The second one involves inputs where
5882     the 'db' section contains nothing but zeroes (there is a one-byte
5883     invalid read after the end of 'db').
5884     [Ivan Nestlerode <inestlerode@us.ibm.com>]
5885
5886  *) Partial backport from 0.9.9-dev:
5887
5888     Introduce bn_mul_mont (dedicated Montgomery multiplication
5889     procedure) as a candidate for BIGNUM assembler implementation.
5890     While 0.9.9-dev uses assembler for various architectures, only
5891     x86_64 is available by default here in the 0.9.8 branch, and
5892     32-bit x86 is available through a compile-time setting.
5893
5894     To try the 32-bit x86 assembler implementation, use Configure
5895     option "enable-montasm" (which exists only for this backport).
5896
5897     As "enable-montasm" for 32-bit x86 disclaims code stability
5898     anyway, in this constellation we activate additional code
5899     backported from 0.9.9-dev for further performance improvements,
5900     namely BN_from_montgomery_word.  (To enable this otherwise,
5901     e.g. x86_64, try "-DMONT_FROM_WORD___NON_DEFAULT_0_9_8_BUILD".)
5902
5903     [Andy Polyakov (backport partially by Bodo Moeller)]
5904
5905  *) Add TLS session ticket callback. This allows an application to set
5906     TLS ticket cipher and HMAC keys rather than relying on hardcoded fixed
5907     values. This is useful for key rollover for example where several key
5908     sets may exist with different names.
5909     [Steve Henson]
5910
5911  *) Reverse ENGINE-internal logic for caching default ENGINE handles.
5912     This was broken until now in 0.9.8 releases, such that the only way
5913     a registered ENGINE could be used (assuming it initialises
5914     successfully on the host) was to explicitly set it as the default
5915     for the relevant algorithms. This is in contradiction with 0.9.7
5916     behaviour and the documentation. With this fix, when an ENGINE is
5917     registered into a given algorithm's table of implementations, the
5918     'uptodate' flag is reset so that auto-discovery will be used next
5919     time a new context for that algorithm attempts to select an
5920     implementation.
5921     [Ian Lister (tweaked by Geoff Thorpe)]
5922
5923  *) Backport of CMS code to OpenSSL 0.9.8. This differs from the 0.9.9
5924     implementation in the following ways:
5925
5926     Lack of EVP_PKEY_ASN1_METHOD means algorithm parameters have to be
5927     hard coded.
5928
5929     Lack of BER streaming support means one pass streaming processing is
5930     only supported if data is detached: setting the streaming flag is
5931     ignored for embedded content.
5932
5933     CMS support is disabled by default and must be explicitly enabled
5934     with the enable-cms configuration option.
5935     [Steve Henson]
5936
5937  *) Update the GMP engine glue to do direct copies between BIGNUM and
5938     mpz_t when openssl and GMP use the same limb size. Otherwise the
5939     existing "conversion via a text string export" trick is still used.
5940     [Paul Sheer <paulsheer@gmail.com>]
5941
5942  *) Zlib compression BIO. This is a filter BIO which compressed and
5943     uncompresses any data passed through it.
5944     [Steve Henson]
5945
5946  *) Add AES_wrap_key() and AES_unwrap_key() functions to implement
5947     RFC3394 compatible AES key wrapping.
5948     [Steve Henson]
5949
5950  *) Add utility functions to handle ASN1 structures. ASN1_STRING_set0():
5951     sets string data without copying. X509_ALGOR_set0() and
5952     X509_ALGOR_get0(): set and retrieve X509_ALGOR (AlgorithmIdentifier)
5953     data. Attribute function X509at_get0_data_by_OBJ(): retrieves data
5954     from an X509_ATTRIBUTE structure optionally checking it occurs only
5955     once. ASN1_TYPE_set1(): set and ASN1_TYPE structure copying supplied
5956     data.
5957     [Steve Henson]
5958
5959  *) Fix BN flag handling in RSA_eay_mod_exp() and BN_MONT_CTX_set()
5960     to get the expected BN_FLG_CONSTTIME behavior.
5961     [Bodo Moeller (Google)]
5962
5963  *) Netware support:
5964
5965     - fixed wrong usage of ioctlsocket() when build for LIBC BSD sockets
5966     - fixed do_tests.pl to run the test suite with CLIB builds too (CLIB_OPT)
5967     - added some more tests to do_tests.pl
5968     - fixed RunningProcess usage so that it works with newer LIBC NDKs too
5969     - removed usage of BN_LLONG for CLIB builds to avoid runtime dependency
5970     - added new Configure targets netware-clib-bsdsock, netware-clib-gcc,
5971       netware-clib-bsdsock-gcc, netware-libc-bsdsock-gcc
5972     - various changes to netware.pl to enable gcc-cross builds on Win32
5973       platform
5974     - changed crypto/bio/b_sock.c to work with macro functions (CLIB BSD)
5975     - various changes to fix missing prototype warnings
5976     - fixed x86nasm.pl to create correct asm files for NASM COFF output
5977     - added AES, WHIRLPOOL and CPUID assembler code to build files
5978     - added missing AES assembler make rules to mk1mf.pl
5979     - fixed order of includes in apps/ocsp.c so that e_os.h settings apply
5980     [Guenter Knauf <eflash@gmx.net>]
5981
5982  *) Implement certificate status request TLS extension defined in RFC3546.
5983     A client can set the appropriate parameters and receive the encoded
5984     OCSP response via a callback. A server can query the supplied parameters
5985     and set the encoded OCSP response in the callback. Add simplified examples
5986     to s_client and s_server.
5987     [Steve Henson]
5988
5989 Changes between 0.9.8f and 0.9.8g  [19 Oct 2007]
5990
5991  *) Fix various bugs:
5992     + Binary incompatibility of ssl_ctx_st structure
5993     + DTLS interoperation with non-compliant servers
5994     + Don't call get_session_cb() without proposed session
5995     + Fix ia64 assembler code
5996     [Andy Polyakov, Steve Henson]
5997
5998 Changes between 0.9.8e and 0.9.8f  [11 Oct 2007]
5999
6000  *) DTLS Handshake overhaul. There were longstanding issues with
6001     OpenSSL DTLS implementation, which were making it impossible for
6002     RFC 4347 compliant client to communicate with OpenSSL server.
6003     Unfortunately just fixing these incompatibilities would "cut off"
6004     pre-0.9.8f clients. To allow for hassle free upgrade post-0.9.8e
6005     server keeps tolerating non RFC compliant syntax. The opposite is
6006     not true, 0.9.8f client can not communicate with earlier server.
6007     This update even addresses CVE-2007-4995.
6008     [Andy Polyakov]
6009
6010  *) Changes to avoid need for function casts in OpenSSL: some compilers
6011     (gcc 4.2 and later) reject their use.
6012     [Kurt Roeckx <kurt@roeckx.be>, Peter Hartley <pdh@utter.chaos.org.uk>,
6013      Steve Henson]
6014
6015  *) Add RFC4507 support to OpenSSL. This includes the corrections in
6016     RFC4507bis. The encrypted ticket format is an encrypted encoded
6017     SSL_SESSION structure, that way new session features are automatically
6018     supported.
6019
6020     If a client application caches session in an SSL_SESSION structure
6021     support is transparent because tickets are now stored in the encoded
6022     SSL_SESSION.
6023
6024     The SSL_CTX structure automatically generates keys for ticket
6025     protection in servers so again support should be possible
6026     with no application modification.
6027
6028     If a client or server wishes to disable RFC4507 support then the option
6029     SSL_OP_NO_TICKET can be set.
6030
6031     Add a TLS extension debugging callback to allow the contents of any client
6032     or server extensions to be examined.
6033
6034     This work was sponsored by Google.
6035     [Steve Henson]
6036
6037  *) Add initial support for TLS extensions, specifically for the server_name
6038     extension so far.  The SSL_SESSION, SSL_CTX, and SSL data structures now
6039     have new members for a host name.  The SSL data structure has an
6040     additional member SSL_CTX *initial_ctx so that new sessions can be
6041     stored in that context to allow for session resumption, even after the
6042     SSL has been switched to a new SSL_CTX in reaction to a client's
6043     server_name extension.
6044
6045     New functions (subject to change):
6046
6047         SSL_get_servername()
6048         SSL_get_servername_type()
6049         SSL_set_SSL_CTX()
6050
6051     New CTRL codes and macros (subject to change):
6052
6053         SSL_CTRL_SET_TLSEXT_SERVERNAME_CB
6054                                 - SSL_CTX_set_tlsext_servername_callback()
6055         SSL_CTRL_SET_TLSEXT_SERVERNAME_ARG
6056                                      - SSL_CTX_set_tlsext_servername_arg()
6057         SSL_CTRL_SET_TLSEXT_HOSTNAME           - SSL_set_tlsext_host_name()
6058
6059     openssl s_client has a new '-servername ...' option.
6060
6061     openssl s_server has new options '-servername_host ...', '-cert2 ...',
6062     '-key2 ...', '-servername_fatal' (subject to change).  This allows
6063     testing the HostName extension for a specific single host name ('-cert'
6064     and '-key' remain fallbacks for handshakes without HostName
6065     negotiation).  If the unrecognized_name alert has to be sent, this by
6066     default is a warning; it becomes fatal with the '-servername_fatal'
6067     option.
6068
6069     [Peter Sylvester,  Remy Allais, Christophe Renou, Steve Henson]
6070
6071  *) Add AES and SSE2 assembly language support to VC++ build.
6072     [Steve Henson]
6073
6074  *) Mitigate attack on final subtraction in Montgomery reduction.
6075     [Andy Polyakov]
6076
6077  *) Fix crypto/ec/ec_mult.c to work properly with scalars of value 0
6078     (which previously caused an internal error).
6079     [Bodo Moeller]
6080
6081  *) Squeeze another 10% out of IGE mode when in != out.
6082     [Ben Laurie]
6083
6084  *) AES IGE mode speedup.
6085     [Dean Gaudet (Google)]
6086
6087  *) Add the Korean symmetric 128-bit cipher SEED (see
6088     http://www.kisa.or.kr/kisa/seed/jsp/seed_eng.jsp) and
6089     add SEED ciphersuites from RFC 4162:
6090
6091        TLS_RSA_WITH_SEED_CBC_SHA      =  "SEED-SHA"
6092        TLS_DHE_DSS_WITH_SEED_CBC_SHA  =  "DHE-DSS-SEED-SHA"
6093        TLS_DHE_RSA_WITH_SEED_CBC_SHA  =  "DHE-RSA-SEED-SHA"
6094        TLS_DH_anon_WITH_SEED_CBC_SHA  =  "ADH-SEED-SHA"
6095
6096     To minimize changes between patchlevels in the OpenSSL 0.9.8
6097     series, SEED remains excluded from compilation unless OpenSSL
6098     is configured with 'enable-seed'.
6099     [KISA, Bodo Moeller]
6100
6101  *) Mitigate branch prediction attacks, which can be practical if a
6102     single processor is shared, allowing a spy process to extract
6103     information.  For detailed background information, see
6104     http://eprint.iacr.org/2007/039 (O. Aciicmez, S. Gueron,
6105     J.-P. Seifert, "New Branch Prediction Vulnerabilities in OpenSSL
6106     and Necessary Software Countermeasures").  The core of the change
6107     are new versions BN_div_no_branch() and
6108     BN_mod_inverse_no_branch() of BN_div() and BN_mod_inverse(),
6109     respectively, which are slower, but avoid the security-relevant
6110     conditional branches.  These are automatically called by BN_div()
6111     and BN_mod_inverse() if the flag BN_FLG_CONSTTIME is set for one
6112     of the input BIGNUMs.  Also, BN_is_bit_set() has been changed to
6113     remove a conditional branch.
6114
6115     BN_FLG_CONSTTIME is the new name for the previous
6116     BN_FLG_EXP_CONSTTIME flag, since it now affects more than just
6117     modular exponentiation.  (Since OpenSSL 0.9.7h, setting this flag
6118     in the exponent causes BN_mod_exp_mont() to use the alternative
6119     implementation in BN_mod_exp_mont_consttime().)  The old name
6120     remains as a deprecated alias.
6121
6122     Similarly, RSA_FLAG_NO_EXP_CONSTTIME is replaced by a more general
6123     RSA_FLAG_NO_CONSTTIME flag since the RSA implementation now uses
6124     constant-time implementations for more than just exponentiation.
6125     Here too the old name is kept as a deprecated alias.
6126
6127     BN_BLINDING_new() will now use BN_dup() for the modulus so that
6128     the BN_BLINDING structure gets an independent copy of the
6129     modulus.  This means that the previous "BIGNUM *m" argument to
6130     BN_BLINDING_new() and to BN_BLINDING_create_param() now
6131     essentially becomes "const BIGNUM *m", although we can't actually
6132     change this in the header file before 0.9.9.  It allows
6133     RSA_setup_blinding() to use BN_with_flags() on the modulus to
6134     enable BN_FLG_CONSTTIME.
6135
6136     [Matthew D Wood (Intel Corp)]
6137
6138  *) In the SSL/TLS server implementation, be strict about session ID
6139     context matching (which matters if an application uses a single
6140     external cache for different purposes).  Previously,
6141     out-of-context reuse was forbidden only if SSL_VERIFY_PEER was
6142     set.  This did ensure strict client verification, but meant that,
6143     with applications using a single external cache for quite
6144     different requirements, clients could circumvent ciphersuite
6145     restrictions for a given session ID context by starting a session
6146     in a different context.
6147     [Bodo Moeller]
6148
6149  *) Include "!eNULL" in SSL_DEFAULT_CIPHER_LIST to make sure that
6150     a ciphersuite string such as "DEFAULT:RSA" cannot enable
6151     authentication-only ciphersuites.
6152     [Bodo Moeller]
6153
6154  *) Update the SSL_get_shared_ciphers() fix CVE-2006-3738 which was
6155     not complete and could lead to a possible single byte overflow
6156     (CVE-2007-5135) [Ben Laurie]
6157
6158 Changes between 0.9.8d and 0.9.8e  [23 Feb 2007]
6159
6160  *) Since AES128 and AES256 (and similarly Camellia128 and
6161     Camellia256) share a single mask bit in the logic of
6162     ssl/ssl_ciph.c, the code for masking out disabled ciphers needs a
6163     kludge to work properly if AES128 is available and AES256 isn't
6164     (or if Camellia128 is available and Camellia256 isn't).
6165     [Victor Duchovni]
6166
6167  *) Fix the BIT STRING encoding generated by crypto/ec/ec_asn1.c
6168     (within i2d_ECPrivateKey, i2d_ECPKParameters, i2d_ECParameters):
6169     When a point or a seed is encoded in a BIT STRING, we need to
6170     prevent the removal of trailing zero bits to get the proper DER
6171     encoding.  (By default, crypto/asn1/a_bitstr.c assumes the case
6172     of a NamedBitList, for which trailing 0 bits need to be removed.)
6173     [Bodo Moeller]
6174
6175  *) Have SSL/TLS server implementation tolerate "mismatched" record
6176     protocol version while receiving ClientHello even if the
6177     ClientHello is fragmented.  (The server can't insist on the
6178     particular protocol version it has chosen before the ServerHello
6179     message has informed the client about his choice.)
6180     [Bodo Moeller]
6181
6182  *) Add RFC 3779 support.
6183     [Rob Austein for ARIN, Ben Laurie]
6184
6185  *) Load error codes if they are not already present instead of using a
6186     static variable. This allows them to be cleanly unloaded and reloaded.
6187     Improve header file function name parsing.
6188     [Steve Henson]
6189
6190  *) extend SMTP and IMAP protocol emulation in s_client to use EHLO
6191     or CAPABILITY handshake as required by RFCs.
6192     [Goetz Babin-Ebell]
6193
6194 Changes between 0.9.8c and 0.9.8d  [28 Sep 2006]
6195
6196  *) Introduce limits to prevent malicious keys being able to
6197     cause a denial of service.  (CVE-2006-2940)
6198     [Steve Henson, Bodo Moeller]
6199
6200  *) Fix ASN.1 parsing of certain invalid structures that can result
6201     in a denial of service.  (CVE-2006-2937)  [Steve Henson]
6202
6203  *) Fix buffer overflow in SSL_get_shared_ciphers() function.
6204     (CVE-2006-3738) [Tavis Ormandy and Will Drewry, Google Security Team]
6205
6206  *) Fix SSL client code which could crash if connecting to a
6207     malicious SSLv2 server.  (CVE-2006-4343)
6208     [Tavis Ormandy and Will Drewry, Google Security Team]
6209
6210  *) Since 0.9.8b, ciphersuite strings naming explicit ciphersuites
6211     match only those.  Before that, "AES256-SHA" would be interpreted
6212     as a pattern and match "AES128-SHA" too (since AES128-SHA got
6213     the same strength classification in 0.9.7h) as we currently only
6214     have a single AES bit in the ciphersuite description bitmap.
6215     That change, however, also applied to ciphersuite strings such as
6216     "RC4-MD5" that intentionally matched multiple ciphersuites --
6217     namely, SSL 2.0 ciphersuites in addition to the more common ones
6218     from SSL 3.0/TLS 1.0.
6219
6220     So we change the selection algorithm again: Naming an explicit
6221     ciphersuite selects this one ciphersuite, and any other similar
6222     ciphersuite (same bitmap) from *other* protocol versions.
6223     Thus, "RC4-MD5" again will properly select both the SSL 2.0
6224     ciphersuite and the SSL 3.0/TLS 1.0 ciphersuite.
6225
6226     Since SSL 2.0 does not have any ciphersuites for which the
6227     128/256 bit distinction would be relevant, this works for now.
6228     The proper fix will be to use different bits for AES128 and
6229     AES256, which would have avoided the problems from the beginning;
6230     however, bits are scarce, so we can only do this in a new release
6231     (not just a patchlevel) when we can change the SSL_CIPHER
6232     definition to split the single 'unsigned long mask' bitmap into
6233     multiple values to extend the available space.
6234
6235     [Bodo Moeller]
6236
6237 Changes between 0.9.8b and 0.9.8c  [05 Sep 2006]
6238
6239  *) Avoid PKCS #1 v1.5 signature attack discovered by Daniel Bleichenbacher
6240     (CVE-2006-4339)  [Ben Laurie and Google Security Team]
6241
6242  *) Add AES IGE and biIGE modes.
6243     [Ben Laurie]
6244
6245  *) Change the Unix randomness entropy gathering to use poll() when
6246     possible instead of select(), since the latter has some
6247     undesirable limitations.
6248     [Darryl Miles via Richard Levitte and Bodo Moeller]
6249
6250  *) Disable "ECCdraft" ciphersuites more thoroughly.  Now special
6251     treatment in ssl/ssl_ciph.s makes sure that these ciphersuites
6252     cannot be implicitly activated as part of, e.g., the "AES" alias.
6253     However, please upgrade to OpenSSL 0.9.9[-dev] for
6254     non-experimental use of the ECC ciphersuites to get TLS extension
6255     support, which is required for curve and point format negotiation
6256     to avoid potential handshake problems.
6257     [Bodo Moeller]
6258
6259  *) Disable rogue ciphersuites:
6260
6261      - SSLv2 0x08 0x00 0x80 ("RC4-64-MD5")
6262      - SSLv3/TLSv1 0x00 0x61 ("EXP1024-RC2-CBC-MD5")
6263      - SSLv3/TLSv1 0x00 0x60 ("EXP1024-RC4-MD5")
6264
6265     The latter two were purportedly from
6266     draft-ietf-tls-56-bit-ciphersuites-0[01].txt, but do not really
6267     appear there.
6268
6269     Also deactivate the remaining ciphersuites from
6270     draft-ietf-tls-56-bit-ciphersuites-01.txt.  These are just as
6271     unofficial, and the ID has long expired.
6272     [Bodo Moeller]
6273
6274  *) Fix RSA blinding Heisenbug (problems sometimes occurred on
6275     dual-core machines) and other potential thread-safety issues.
6276     [Bodo Moeller]
6277
6278  *) Add the symmetric cipher Camellia (128-bit, 192-bit, 256-bit key
6279     versions), which is now available for royalty-free use
6280     (see http://info.isl.ntt.co.jp/crypt/eng/info/chiteki.html).
6281     Also, add Camellia TLS ciphersuites from RFC 4132.
6282
6283     To minimize changes between patchlevels in the OpenSSL 0.9.8
6284     series, Camellia remains excluded from compilation unless OpenSSL
6285     is configured with 'enable-camellia'.
6286     [NTT]
6287
6288  *) Disable the padding bug check when compression is in use. The padding
6289     bug check assumes the first packet is of even length, this is not
6290     necessarily true if compression is enabled and can result in false
6291     positives causing handshake failure. The actual bug test is ancient
6292     code so it is hoped that implementations will either have fixed it by
6293     now or any which still have the bug do not support compression.
6294     [Steve Henson]
6295
6296 Changes between 0.9.8a and 0.9.8b  [04 May 2006]
6297
6298  *) When applying a cipher rule check to see if string match is an explicit
6299     cipher suite and only match that one cipher suite if it is.
6300     [Steve Henson]
6301
6302  *) Link in manifests for VC++ if needed.
6303     [Austin Ziegler <halostatue@gmail.com>]
6304
6305  *) Update support for ECC-based TLS ciphersuites according to
6306     draft-ietf-tls-ecc-12.txt with proposed changes (but without
6307     TLS extensions, which are supported starting with the 0.9.9
6308     branch, not in the OpenSSL 0.9.8 branch).
6309     [Douglas Stebila]
6310
6311  *) New functions EVP_CIPHER_CTX_new() and EVP_CIPHER_CTX_free() to support
6312     opaque EVP_CIPHER_CTX handling.
6313     [Steve Henson]
6314
6315  *) Fixes and enhancements to zlib compression code. We now only use
6316     "zlib1.dll" and use the default __cdecl calling convention on Win32
6317     to conform with the standards mentioned here:
6318           http://www.zlib.net/DLL_FAQ.txt
6319     Static zlib linking now works on Windows and the new --with-zlib-include
6320     --with-zlib-lib options to Configure can be used to supply the location
6321     of the headers and library. Gracefully handle case where zlib library
6322     can't be loaded.
6323     [Steve Henson]
6324
6325  *) Several fixes and enhancements to the OID generation code. The old code
6326     sometimes allowed invalid OIDs (1.X for X >= 40 for example), couldn't
6327     handle numbers larger than ULONG_MAX, truncated printing and had a
6328     non standard OBJ_obj2txt() behaviour.
6329     [Steve Henson]
6330
6331  *) Add support for building of engines under engine/ as shared libraries
6332     under VC++ build system.
6333     [Steve Henson]
6334
6335  *) Corrected the numerous bugs in the Win32 path splitter in DSO.
6336     Hopefully, we will not see any false combination of paths any more.
6337     [Richard Levitte]
6338
6339 Changes between 0.9.8 and 0.9.8a  [11 Oct 2005]
6340
6341  *) Remove the functionality of SSL_OP_MSIE_SSLV2_RSA_PADDING
6342     (part of SSL_OP_ALL).  This option used to disable the
6343     countermeasure against man-in-the-middle protocol-version
6344     rollback in the SSL 2.0 server implementation, which is a bad
6345     idea.  (CVE-2005-2969)
6346
6347     [Bodo Moeller; problem pointed out by Yutaka Oiwa (Research Center
6348     for Information Security, National Institute of Advanced Industrial
6349     Science and Technology [AIST], Japan)]
6350
6351  *) Add two function to clear and return the verify parameter flags.
6352     [Steve Henson]
6353
6354  *) Keep cipherlists sorted in the source instead of sorting them at
6355     runtime, thus removing the need for a lock.
6356     [Nils Larsch]
6357
6358  *) Avoid some small subgroup attacks in Diffie-Hellman.
6359     [Nick Mathewson and Ben Laurie]
6360
6361  *) Add functions for well-known primes.
6362     [Nick Mathewson]
6363
6364  *) Extended Windows CE support.
6365     [Satoshi Nakamura and Andy Polyakov]
6366
6367  *) Initialize SSL_METHOD structures at compile time instead of during
6368     runtime, thus removing the need for a lock.
6369     [Steve Henson]
6370
6371  *) Make PKCS7_decrypt() work even if no certificate is supplied by
6372     attempting to decrypt each encrypted key in turn. Add support to
6373     smime utility.
6374     [Steve Henson]
6375
6376 Changes between 0.9.7h and 0.9.8  [05 Jul 2005]
6377
6378  [NB: OpenSSL 0.9.7i and later 0.9.7 patch levels were released after
6379  OpenSSL 0.9.8.]
6380
6381  *) Add libcrypto.pc and libssl.pc for those who feel they need them.
6382     [Richard Levitte]
6383
6384  *) Change CA.sh and CA.pl so they don't bundle the CSR and the private
6385     key into the same file any more.
6386     [Richard Levitte]
6387
6388  *) Add initial support for Win64, both IA64 and AMD64/x64 flavors.
6389     [Andy Polyakov]
6390
6391  *) Add -utf8 command line and config file option to 'ca'.
6392     [Stefan <stf@udoma.org]
6393
6394  *) Removed the macro des_crypt(), as it seems to conflict with some
6395     libraries.  Use DES_crypt().
6396     [Richard Levitte]
6397
6398  *) Correct naming of the 'chil' and '4758cca' ENGINEs. This
6399     involves renaming the source and generated shared-libs for
6400     both. The engines will accept the corrected or legacy ids
6401     ('ncipher' and '4758_cca' respectively) when binding. NB,
6402     this only applies when building 'shared'.
6403     [Corinna Vinschen <vinschen@redhat.com> and Geoff Thorpe]
6404
6405  *) Add attribute functions to EVP_PKEY structure. Modify
6406     PKCS12_create() to recognize a CSP name attribute and
6407     use it. Make -CSP option work again in pkcs12 utility.
6408     [Steve Henson]
6409
6410  *) Add new functionality to the bn blinding code:
6411     - automatic re-creation of the BN_BLINDING parameters after
6412       a fixed number of uses (currently 32)
6413     - add new function for parameter creation
6414     - introduce flags to control the update behaviour of the
6415       BN_BLINDING parameters
6416     - hide BN_BLINDING structure
6417     Add a second BN_BLINDING slot to the RSA structure to improve
6418     performance when a single RSA object is shared among several
6419     threads.
6420     [Nils Larsch]
6421
6422  *) Add support for DTLS.
6423     [Nagendra Modadugu <nagendra@cs.stanford.edu> and Ben Laurie]
6424
6425  *) Add support for DER encoded private keys (SSL_FILETYPE_ASN1)
6426     to SSL_CTX_use_PrivateKey_file() and SSL_use_PrivateKey_file()
6427     [Walter Goulet]
6428
6429  *) Remove buggy and incomplete DH cert support from
6430     ssl/ssl_rsa.c and ssl/s3_both.c
6431     [Nils Larsch]
6432
6433  *) Use SHA-1 instead of MD5 as the default digest algorithm for
6434     the apps/openssl applications.
6435     [Nils Larsch]
6436
6437  *) Compile clean with "-Wall -Wmissing-prototypes
6438     -Wstrict-prototypes -Wmissing-declarations -Werror". Currently
6439     DEBUG_SAFESTACK must also be set.
6440     [Ben Laurie]
6441
6442  *) Change ./Configure so that certain algorithms can be disabled by default.
6443     The new counterpiece to "no-xxx" is "enable-xxx".
6444
6445     The patented RC5 and MDC2 algorithms will now be disabled unless
6446     "enable-rc5" and "enable-mdc2", respectively, are specified.
6447
6448     (IDEA remains enabled despite being patented.  This is because IDEA
6449     is frequently required for interoperability, and there is no license
6450     fee for non-commercial use.  As before, "no-idea" can be used to
6451     avoid this algorithm.)
6452
6453     [Bodo Moeller]
6454
6455  *) Add processing of proxy certificates (see RFC 3820).  This work was
6456     sponsored by KTH (The Royal Institute of Technology in Stockholm) and
6457     EGEE (Enabling Grids for E-science in Europe).
6458     [Richard Levitte]
6459
6460  *) RC4 performance overhaul on modern architectures/implementations, such
6461     as Intel P4, IA-64 and AMD64.
6462     [Andy Polyakov]
6463
6464  *) New utility extract-section.pl. This can be used specify an alternative
6465     section number in a pod file instead of having to treat each file as
6466     a separate case in Makefile. This can be done by adding two lines to the
6467     pod file:
6468
6469     =for comment openssl_section:XXX
6470
6471     The blank line is mandatory.
6472
6473     [Steve Henson]
6474
6475  *) New arguments -certform, -keyform and -pass for s_client and s_server
6476     to allow alternative format key and certificate files and passphrase
6477     sources.
6478     [Steve Henson]
6479
6480  *) New structure X509_VERIFY_PARAM which combines current verify parameters,
6481     update associated structures and add various utility functions.
6482
6483     Add new policy related verify parameters, include policy checking in
6484     standard verify code. Enhance 'smime' application with extra parameters
6485     to support policy checking and print out.
6486     [Steve Henson]
6487
6488  *) Add a new engine to support VIA PadLock ACE extensions in the VIA C3
6489     Nehemiah processors. These extensions support AES encryption in hardware
6490     as well as RNG (though RNG support is currently disabled).
6491     [Michal Ludvig <michal@logix.cz>, with help from Andy Polyakov]
6492
6493  *) Deprecate BN_[get|set]_params() functions (they were ignored internally).
6494     [Geoff Thorpe]
6495
6496  *) New FIPS 180-2 algorithms, SHA-224/-256/-384/-512 are implemented.
6497     [Andy Polyakov and a number of other people]
6498
6499  *) Improved PowerPC platform support. Most notably BIGNUM assembler
6500     implementation contributed by IBM.
6501     [Suresh Chari, Peter Waltenberg, Andy Polyakov]
6502
6503  *) The new 'RSA_generate_key_ex' function now takes a BIGNUM for the public
6504     exponent rather than 'unsigned long'. There is a corresponding change to
6505     the new 'rsa_keygen' element of the RSA_METHOD structure.
6506     [Jelte Jansen, Geoff Thorpe]
6507
6508  *) Functionality for creating the initial serial number file is now
6509     moved from CA.pl to the 'ca' utility with a new option -create_serial.
6510
6511     (Before OpenSSL 0.9.7e, CA.pl used to initialize the serial
6512     number file to 1, which is bound to cause problems.  To avoid
6513     the problems while respecting compatibility between different 0.9.7
6514     patchlevels, 0.9.7e  employed 'openssl x509 -next_serial' in
6515     CA.pl for serial number initialization.  With the new release 0.9.8,
6516     we can fix the problem directly in the 'ca' utility.)
6517     [Steve Henson]
6518
6519  *) Reduced header interdependencies by declaring more opaque objects in
6520     ossl_typ.h. As a consequence, including some headers (eg. engine.h) will
6521     give fewer recursive includes, which could break lazy source code - so
6522     this change is covered by the OPENSSL_NO_DEPRECATED symbol. As always,
6523     developers should define this symbol when building and using openssl to
6524     ensure they track the recommended behaviour, interfaces, [etc], but
6525     backwards-compatible behaviour prevails when this isn't defined.
6526     [Geoff Thorpe]
6527
6528  *) New function X509_POLICY_NODE_print() which prints out policy nodes.
6529     [Steve Henson]
6530
6531  *) Add new EVP function EVP_CIPHER_CTX_rand_key and associated functionality.
6532     This will generate a random key of the appropriate length based on the
6533     cipher context. The EVP_CIPHER can provide its own random key generation
6534     routine to support keys of a specific form. This is used in the des and
6535     3des routines to generate a key of the correct parity. Update S/MIME
6536     code to use new functions and hence generate correct parity DES keys.
6537     Add EVP_CHECK_DES_KEY #define to return an error if the key is not
6538     valid (weak or incorrect parity).
6539     [Steve Henson]
6540
6541  *) Add a local set of CRLs that can be used by X509_verify_cert() as well
6542     as looking them up. This is useful when the verified structure may contain
6543     CRLs, for example PKCS#7 signedData. Modify PKCS7_verify() to use any CRLs
6544     present unless the new PKCS7_NO_CRL flag is asserted.
6545     [Steve Henson]
6546
6547  *) Extend ASN1 oid configuration module. It now additionally accepts the
6548     syntax:
6549
6550     shortName = some long name, 1.2.3.4
6551     [Steve Henson]
6552
6553  *) Reimplemented the BN_CTX implementation. There is now no more static
6554     limitation on the number of variables it can handle nor the depth of the
6555     "stack" handling for BN_CTX_start()/BN_CTX_end() pairs. The stack
6556     information can now expand as required, and rather than having a single
6557     static array of bignums, BN_CTX now uses a linked-list of such arrays
6558     allowing it to expand on demand whilst maintaining the usefulness of
6559     BN_CTX's "bundling".
6560     [Geoff Thorpe]
6561
6562  *) Add a missing BN_CTX parameter to the 'rsa_mod_exp' callback in RSA_METHOD
6563     to allow all RSA operations to function using a single BN_CTX.
6564     [Geoff Thorpe]
6565
6566  *) Preliminary support for certificate policy evaluation and checking. This
6567     is initially intended to pass the tests outlined in "Conformance Testing
6568     of Relying Party Client Certificate Path Processing Logic" v1.07.
6569     [Steve Henson]
6570
6571  *) bn_dup_expand() has been deprecated, it was introduced in 0.9.7 and
6572     remained unused and not that useful. A variety of other little bignum
6573     tweaks and fixes have also been made continuing on from the audit (see
6574     below).
6575     [Geoff Thorpe]
6576
6577  *) Constify all or almost all d2i, c2i, s2i and r2i functions, along with
6578     associated ASN1, EVP and SSL functions and old ASN1 macros.
6579     [Richard Levitte]
6580
6581  *) BN_zero() only needs to set 'top' and 'neg' to zero for correct results,
6582     and this should never fail. So the return value from the use of
6583     BN_set_word() (which can fail due to needless expansion) is now deprecated;
6584     if OPENSSL_NO_DEPRECATED is defined, BN_zero() is a void macro.
6585     [Geoff Thorpe]
6586
6587  *) BN_CTX_get() should return zero-valued bignums, providing the same
6588     initialised value as BN_new().
6589     [Geoff Thorpe, suggested by Ulf Möller]
6590
6591  *) Support for inhibitAnyPolicy certificate extension.
6592     [Steve Henson]
6593
6594  *) An audit of the BIGNUM code is underway, for which debugging code is
6595     enabled when BN_DEBUG is defined. This makes stricter enforcements on what
6596     is considered valid when processing BIGNUMs, and causes execution to
6597     assert() when a problem is discovered. If BN_DEBUG_RAND is defined,
6598     further steps are taken to deliberately pollute unused data in BIGNUM
6599     structures to try and expose faulty code further on. For now, openssl will
6600     (in its default mode of operation) continue to tolerate the inconsistent
6601     forms that it has tolerated in the past, but authors and packagers should
6602     consider trying openssl and their own applications when compiled with
6603     these debugging symbols defined. It will help highlight potential bugs in
6604     their own code, and will improve the test coverage for OpenSSL itself. At
6605     some point, these tighter rules will become openssl's default to improve
6606     maintainability, though the assert()s and other overheads will remain only
6607     in debugging configurations. See bn.h for more details.
6608     [Geoff Thorpe, Nils Larsch, Ulf Möller]
6609
6610  *) BN_CTX_init() has been deprecated, as BN_CTX is an opaque structure
6611     that can only be obtained through BN_CTX_new() (which implicitly
6612     initialises it). The presence of this function only made it possible
6613     to overwrite an existing structure (and cause memory leaks).
6614     [Geoff Thorpe]
6615
6616  *) Because of the callback-based approach for implementing LHASH as a
6617     template type, lh_insert() adds opaque objects to hash-tables and
6618     lh_doall() or lh_doall_arg() are typically used with a destructor callback
6619     to clean up those corresponding objects before destroying the hash table
6620     (and losing the object pointers). So some over-zealous constifications in
6621     LHASH have been relaxed so that lh_insert() does not take (nor store) the
6622     objects as "const" and the lh_doall[_arg] callback wrappers are not
6623     prototyped to have "const" restrictions on the object pointers they are
6624     given (and so aren't required to cast them away any more).
6625     [Geoff Thorpe]
6626
6627  *) The tmdiff.h API was so ugly and minimal that our own timing utility
6628     (speed) prefers to use its own implementation. The two implementations
6629     haven't been consolidated as yet (volunteers?) but the tmdiff API has had
6630     its object type properly exposed (MS_TM) instead of casting to/from "char
6631     *". This may still change yet if someone realises MS_TM and "ms_time_***"
6632     aren't necessarily the greatest nomenclatures - but this is what was used
6633     internally to the implementation so I've used that for now.
6634     [Geoff Thorpe]
6635
6636  *) Ensure that deprecated functions do not get compiled when
6637     OPENSSL_NO_DEPRECATED is defined. Some "openssl" subcommands and a few of
6638     the self-tests were still using deprecated key-generation functions so
6639     these have been updated also.
6640     [Geoff Thorpe]
6641
6642  *) Reorganise PKCS#7 code to separate the digest location functionality
6643     into PKCS7_find_digest(), digest addition into PKCS7_bio_add_digest().
6644     New function PKCS7_set_digest() to set the digest type for PKCS#7
6645     digestedData type. Add additional code to correctly generate the
6646     digestedData type and add support for this type in PKCS7 initialization
6647     functions.
6648     [Steve Henson]
6649
6650  *) New function PKCS7_set0_type_other() this initializes a PKCS7
6651     structure of type "other".
6652     [Steve Henson]
6653
6654  *) Fix prime generation loop in crypto/bn/bn_prime.pl by making
6655     sure the loop does correctly stop and breaking ("division by zero")
6656     modulus operations are not performed. The (pre-generated) prime
6657     table crypto/bn/bn_prime.h was already correct, but it could not be
6658     re-generated on some platforms because of the "division by zero"
6659     situation in the script.
6660     [Ralf S. Engelschall]
6661
6662  *) Update support for ECC-based TLS ciphersuites according to
6663     draft-ietf-tls-ecc-03.txt: the KDF1 key derivation function with
6664     SHA-1 now is only used for "small" curves (where the
6665     representation of a field element takes up to 24 bytes); for
6666     larger curves, the field element resulting from ECDH is directly
6667     used as premaster secret.
6668     [Douglas Stebila (Sun Microsystems Laboratories)]
6669
6670  *) Add code for kP+lQ timings to crypto/ec/ectest.c, and add SEC2
6671     curve secp160r1 to the tests.
6672     [Douglas Stebila (Sun Microsystems Laboratories)]
6673
6674  *) Add the possibility to load symbols globally with DSO.
6675     [Götz Babin-Ebell <babin-ebell@trustcenter.de> via Richard Levitte]
6676
6677  *) Add the functions ERR_set_mark() and ERR_pop_to_mark() for better
6678     control of the error stack.
6679     [Richard Levitte]
6680
6681  *) Add support for STORE in ENGINE.
6682     [Richard Levitte]
6683
6684  *) Add the STORE type.  The intention is to provide a common interface
6685     to certificate and key stores, be they simple file-based stores, or
6686     HSM-type store, or LDAP stores, or...
6687     NOTE: The code is currently UNTESTED and isn't really used anywhere.
6688     [Richard Levitte]
6689
6690  *) Add a generic structure called OPENSSL_ITEM.  This can be used to
6691     pass a list of arguments to any function as well as provide a way
6692     for a function to pass data back to the caller.
6693     [Richard Levitte]
6694
6695  *) Add the functions BUF_strndup() and BUF_memdup().  BUF_strndup()
6696     works like BUF_strdup() but can be used to duplicate a portion of
6697     a string.  The copy gets NUL-terminated.  BUF_memdup() duplicates
6698     a memory area.
6699     [Richard Levitte]
6700
6701  *) Add the function sk_find_ex() which works like sk_find(), but will
6702     return an index to an element even if an exact match couldn't be
6703     found.  The index is guaranteed to point at the element where the
6704     searched-for key would be inserted to preserve sorting order.
6705     [Richard Levitte]
6706
6707  *) Add the function OBJ_bsearch_ex() which works like OBJ_bsearch() but
6708     takes an extra flags argument for optional functionality.  Currently,
6709     the following flags are defined:
6710
6711        OBJ_BSEARCH_VALUE_ON_NOMATCH
6712        This one gets OBJ_bsearch_ex() to return a pointer to the first
6713        element where the comparing function returns a negative or zero
6714        number.
6715
6716        OBJ_BSEARCH_FIRST_VALUE_ON_MATCH
6717        This one gets OBJ_bsearch_ex() to return a pointer to the first
6718        element where the comparing function returns zero.  This is useful
6719        if there are more than one element where the comparing function
6720        returns zero.
6721     [Richard Levitte]
6722
6723  *) Make it possible to create self-signed certificates with 'openssl ca'
6724     in such a way that the self-signed certificate becomes part of the
6725     CA database and uses the same mechanisms for serial number generation
6726     as all other certificate signing.  The new flag '-selfsign' enables
6727     this functionality.  Adapt CA.sh and CA.pl.in.
6728     [Richard Levitte]
6729
6730  *) Add functionality to check the public key of a certificate request
6731     against a given private.  This is useful to check that a certificate
6732     request can be signed by that key (self-signing).
6733     [Richard Levitte]
6734
6735  *) Make it possible to have multiple active certificates with the same
6736     subject in the CA index file.  This is done only if the keyword
6737     'unique_subject' is set to 'no' in the main CA section (default
6738     if 'CA_default') of the configuration file.  The value is saved
6739     with the database itself in a separate index attribute file,
6740     named like the index file with '.attr' appended to the name.
6741     [Richard Levitte]
6742
6743  *) Generate multi-valued AVAs using '+' notation in config files for
6744     req and dirName.
6745     [Steve Henson]
6746
6747  *) Support for nameConstraints certificate extension.
6748     [Steve Henson]
6749
6750  *) Support for policyConstraints certificate extension.
6751     [Steve Henson]
6752
6753  *) Support for policyMappings certificate extension.
6754     [Steve Henson]
6755
6756  *) Make sure the default DSA_METHOD implementation only uses its
6757     dsa_mod_exp() and/or bn_mod_exp() handlers if they are non-NULL,
6758     and change its own handlers to be NULL so as to remove unnecessary
6759     indirection. This lets alternative implementations fallback to the
6760     default implementation more easily.
6761     [Geoff Thorpe]
6762
6763  *) Support for directoryName in GeneralName related extensions
6764     in config files.
6765     [Steve Henson]
6766
6767  *) Make it possible to link applications using Makefile.shared.
6768     Make that possible even when linking against static libraries!
6769     [Richard Levitte]
6770
6771  *) Support for single pass processing for S/MIME signing. This now
6772     means that S/MIME signing can be done from a pipe, in addition
6773     cleartext signing (multipart/signed type) is effectively streaming
6774     and the signed data does not need to be all held in memory.
6775
6776     This is done with a new flag PKCS7_STREAM. When this flag is set
6777     PKCS7_sign() only initializes the PKCS7 structure and the actual signing
6778     is done after the data is output (and digests calculated) in
6779     SMIME_write_PKCS7().
6780     [Steve Henson]
6781
6782  *) Add full support for -rpath/-R, both in shared libraries and
6783     applications, at least on the platforms where it's known how
6784     to do it.
6785     [Richard Levitte]
6786
6787  *) In crypto/ec/ec_mult.c, implement fast point multiplication with
6788     precomputation, based on wNAF splitting: EC_GROUP_precompute_mult()
6789     will now compute a table of multiples of the generator that
6790     makes subsequent invocations of EC_POINTs_mul() or EC_POINT_mul()
6791     faster (notably in the case of a single point multiplication,
6792     scalar * generator).
6793     [Nils Larsch, Bodo Moeller]
6794
6795  *) IPv6 support for certificate extensions. The various extensions
6796     which use the IP:a.b.c.d can now take IPv6 addresses using the
6797     formats of RFC1884 2.2 . IPv6 addresses are now also displayed
6798     correctly.
6799     [Steve Henson]
6800
6801  *) Added an ENGINE that implements RSA by performing private key
6802     exponentiations with the GMP library. The conversions to and from
6803     GMP's mpz_t format aren't optimised nor are any montgomery forms
6804     cached, and on x86 it appears OpenSSL's own performance has caught up.
6805     However there are likely to be other architectures where GMP could
6806     provide a boost. This ENGINE is not built in by default, but it can be
6807     specified at Configure time and should be accompanied by the necessary
6808     linker additions, eg;
6809         ./config -DOPENSSL_USE_GMP -lgmp
6810     [Geoff Thorpe]
6811
6812  *) "openssl engine" will not display ENGINE/DSO load failure errors when
6813     testing availability of engines with "-t" - the old behaviour is
6814     produced by increasing the feature's verbosity with "-tt".
6815     [Geoff Thorpe]
6816
6817  *) ECDSA routines: under certain error conditions uninitialized BN objects
6818     could be freed. Solution: make sure initialization is performed early
6819     enough. (Reported and fix supplied by Nils Larsch <nla@trustcenter.de>
6820     via PR#459)
6821     [Lutz Jaenicke]
6822
6823  *) Key-generation can now be implemented in RSA_METHOD, DSA_METHOD
6824     and DH_METHOD (eg. by ENGINE implementations) to override the normal
6825     software implementations. For DSA and DH, parameter generation can
6826     also be overridden by providing the appropriate method callbacks.
6827     [Geoff Thorpe]
6828
6829  *) Change the "progress" mechanism used in key-generation and
6830     primality testing to functions that take a new BN_GENCB pointer in
6831     place of callback/argument pairs. The new API functions have "_ex"
6832     postfixes and the older functions are reimplemented as wrappers for
6833     the new ones. The OPENSSL_NO_DEPRECATED symbol can be used to hide
6834     declarations of the old functions to help (graceful) attempts to
6835     migrate to the new functions. Also, the new key-generation API
6836     functions operate on a caller-supplied key-structure and return
6837     success/failure rather than returning a key or NULL - this is to
6838     help make "keygen" another member function of RSA_METHOD etc.
6839
6840     Example for using the new callback interface:
6841
6842          int (*my_callback)(int a, int b, BN_GENCB *cb) = ...;
6843          void *my_arg = ...;
6844          BN_GENCB my_cb;
6845
6846          BN_GENCB_set(&my_cb, my_callback, my_arg);
6847
6848          return BN_is_prime_ex(some_bignum, BN_prime_checks, NULL, &cb);
6849          /* For the meaning of a, b in calls to my_callback(), see the
6850           * documentation of the function that calls the callback.
6851           * cb will point to my_cb; my_arg can be retrieved as cb->arg.
6852           * my_callback should return 1 if it wants BN_is_prime_ex()
6853           * to continue, or 0 to stop.
6854           */
6855
6856     [Geoff Thorpe]
6857
6858  *) Change the ZLIB compression method to be stateful, and make it
6859     available to TLS with the number defined in
6860     draft-ietf-tls-compression-04.txt.
6861     [Richard Levitte]
6862
6863  *) Add the ASN.1 structures and functions for CertificatePair, which
6864     is defined as follows (according to X.509_4thEditionDraftV6.pdf):
6865
6866     CertificatePair ::= SEQUENCE {
6867        forward         [0]     Certificate OPTIONAL,
6868        reverse         [1]     Certificate OPTIONAL,
6869        -- at least one of the pair shall be present -- }
6870
6871     Also implement the PEM functions to read and write certificate
6872     pairs, and defined the PEM tag as "CERTIFICATE PAIR".
6873
6874     This needed to be defined, mostly for the sake of the LDAP
6875     attribute crossCertificatePair, but may prove useful elsewhere as
6876     well.
6877     [Richard Levitte]
6878
6879  *) Make it possible to inhibit symlinking of shared libraries in
6880     Makefile.shared, for Cygwin's sake.
6881     [Richard Levitte]
6882
6883  *) Extend the BIGNUM API by creating a function
6884          void BN_set_negative(BIGNUM *a, int neg);
6885     and a macro that behave like
6886          int  BN_is_negative(const BIGNUM *a);
6887
6888     to avoid the need to access 'a->neg' directly in applications.
6889     [Nils Larsch]
6890
6891  *) Implement fast modular reduction for pseudo-Mersenne primes
6892     used in NIST curves (crypto/bn/bn_nist.c, crypto/ec/ecp_nist.c).
6893     EC_GROUP_new_curve_GFp() will now automatically use this
6894     if applicable.
6895     [Nils Larsch <nla@trustcenter.de>]
6896
6897  *) Add new lock type (CRYPTO_LOCK_BN).
6898     [Bodo Moeller]
6899
6900  *) Change the ENGINE framework to automatically load engines
6901     dynamically from specific directories unless they could be
6902     found to already be built in or loaded.  Move all the
6903     current engines except for the cryptodev one to a new
6904     directory engines/.
6905     The engines in engines/ are built as shared libraries if
6906     the "shared" options was given to ./Configure or ./config.
6907     Otherwise, they are inserted in libcrypto.a.
6908     /usr/local/ssl/engines is the default directory for dynamic
6909     engines, but that can be overridden at configure time through
6910     the usual use of --prefix and/or --openssldir, and at run
6911     time with the environment variable OPENSSL_ENGINES.
6912     [Geoff Thorpe and Richard Levitte]
6913
6914  *) Add Makefile.shared, a helper makefile to build shared
6915     libraries.  Adapt Makefile.org.
6916     [Richard Levitte]
6917
6918  *) Add version info to Win32 DLLs.
6919     [Peter 'Luna' Runestig" <peter@runestig.com>]
6920
6921  *) Add new 'medium level' PKCS#12 API. Certificates and keys
6922     can be added using this API to created arbitrary PKCS#12
6923     files while avoiding the low level API.
6924
6925     New options to PKCS12_create(), key or cert can be NULL and
6926     will then be omitted from the output file. The encryption
6927     algorithm NIDs can be set to -1 for no encryption, the mac
6928     iteration count can be set to 0 to omit the mac.
6929
6930     Enhance pkcs12 utility by making the -nokeys and -nocerts
6931     options work when creating a PKCS#12 file. New option -nomac
6932     to omit the mac, NONE can be set for an encryption algorithm.
6933     New code is modified to use the enhanced PKCS12_create()
6934     instead of the low level API.
6935     [Steve Henson]
6936
6937  *) Extend ASN1 encoder to support indefinite length constructed
6938     encoding. This can output sequences tags and octet strings in
6939     this form. Modify pk7_asn1.c to support indefinite length
6940     encoding. This is experimental and needs additional code to
6941     be useful, such as an ASN1 bio and some enhanced streaming
6942     PKCS#7 code.
6943
6944     Extend template encode functionality so that tagging is passed
6945     down to the template encoder.
6946     [Steve Henson]
6947
6948  *) Let 'openssl req' fail if an argument to '-newkey' is not
6949     recognized instead of using RSA as a default.
6950     [Bodo Moeller]
6951
6952  *) Add support for ECC-based ciphersuites from draft-ietf-tls-ecc-01.txt.
6953     As these are not official, they are not included in "ALL";
6954     the "ECCdraft" ciphersuite group alias can be used to select them.
6955     [Vipul Gupta and Sumit Gupta (Sun Microsystems Laboratories)]
6956
6957  *) Add ECDH engine support.
6958     [Nils Gura and Douglas Stebila (Sun Microsystems Laboratories)]
6959
6960  *) Add ECDH in new directory crypto/ecdh/.
6961     [Douglas Stebila (Sun Microsystems Laboratories)]
6962
6963  *) Let BN_rand_range() abort with an error after 100 iterations
6964     without success (which indicates a broken PRNG).
6965     [Bodo Moeller]
6966
6967  *) Change BN_mod_sqrt() so that it verifies that the input value
6968     is really the square of the return value.  (Previously,
6969     BN_mod_sqrt would show GIGO behaviour.)
6970     [Bodo Moeller]
6971
6972  *) Add named elliptic curves over binary fields from X9.62, SECG,
6973     and WAP/WTLS; add OIDs that were still missing.
6974
6975     [Sheueling Chang Shantz and Douglas Stebila
6976     (Sun Microsystems Laboratories)]
6977
6978  *) Extend the EC library for elliptic curves over binary fields
6979     (new files ec2_smpl.c, ec2_smpt.c, ec2_mult.c in crypto/ec/).
6980     New EC_METHOD:
6981
6982          EC_GF2m_simple_method
6983
6984     New API functions:
6985
6986          EC_GROUP_new_curve_GF2m
6987          EC_GROUP_set_curve_GF2m
6988          EC_GROUP_get_curve_GF2m
6989          EC_POINT_set_affine_coordinates_GF2m
6990          EC_POINT_get_affine_coordinates_GF2m
6991          EC_POINT_set_compressed_coordinates_GF2m
6992
6993     Point compression for binary fields is disabled by default for
6994     patent reasons (compile with OPENSSL_EC_BIN_PT_COMP defined to
6995     enable it).
6996
6997     As binary polynomials are represented as BIGNUMs, various members
6998     of the EC_GROUP and EC_POINT data structures can be shared
6999     between the implementations for prime fields and binary fields;
7000     the above ..._GF2m functions (except for EX_GROUP_new_curve_GF2m)
7001     are essentially identical to their ..._GFp counterparts.
7002     (For simplicity, the '..._GFp' prefix has been dropped from
7003     various internal method names.)
7004
7005     An internal 'field_div' method (similar to 'field_mul' and
7006     'field_sqr') has been added; this is used only for binary fields.
7007
7008     [Sheueling Chang Shantz and Douglas Stebila
7009     (Sun Microsystems Laboratories)]
7010
7011  *) Optionally dispatch EC_POINT_mul(), EC_POINT_precompute_mult()
7012     through methods ('mul', 'precompute_mult').
7013
7014     The generic implementations (now internally called 'ec_wNAF_mul'
7015     and 'ec_wNAF_precomputed_mult') remain the default if these
7016     methods are undefined.
7017
7018     [Sheueling Chang Shantz and Douglas Stebila
7019     (Sun Microsystems Laboratories)]
7020
7021  *) New function EC_GROUP_get_degree, which is defined through
7022     EC_METHOD.  For curves over prime fields, this returns the bit
7023     length of the modulus.
7024
7025     [Sheueling Chang Shantz and Douglas Stebila
7026     (Sun Microsystems Laboratories)]
7027
7028  *) New functions EC_GROUP_dup, EC_POINT_dup.
7029     (These simply call ..._new  and ..._copy).
7030
7031     [Sheueling Chang Shantz and Douglas Stebila
7032     (Sun Microsystems Laboratories)]
7033
7034  *) Add binary polynomial arithmetic software in crypto/bn/bn_gf2m.c.
7035     Polynomials are represented as BIGNUMs (where the sign bit is not
7036     used) in the following functions [macros]:
7037
7038          BN_GF2m_add
7039          BN_GF2m_sub             [= BN_GF2m_add]
7040          BN_GF2m_mod             [wrapper for BN_GF2m_mod_arr]
7041          BN_GF2m_mod_mul         [wrapper for BN_GF2m_mod_mul_arr]
7042          BN_GF2m_mod_sqr         [wrapper for BN_GF2m_mod_sqr_arr]
7043          BN_GF2m_mod_inv
7044          BN_GF2m_mod_exp         [wrapper for BN_GF2m_mod_exp_arr]
7045          BN_GF2m_mod_sqrt        [wrapper for BN_GF2m_mod_sqrt_arr]
7046          BN_GF2m_mod_solve_quad  [wrapper for BN_GF2m_mod_solve_quad_arr]
7047          BN_GF2m_cmp             [= BN_ucmp]
7048
7049     (Note that only the 'mod' functions are actually for fields GF(2^m).
7050     BN_GF2m_add() is misnomer, but this is for the sake of consistency.)
7051
7052     For some functions, an the irreducible polynomial defining a
7053     field can be given as an 'unsigned int[]' with strictly
7054     decreasing elements giving the indices of those bits that are set;
7055     i.e., p[] represents the polynomial
7056          f(t) = t^p[0] + t^p[1] + ... + t^p[k]
7057     where
7058          p[0] > p[1] > ... > p[k] = 0.
7059     This applies to the following functions:
7060
7061          BN_GF2m_mod_arr
7062          BN_GF2m_mod_mul_arr
7063          BN_GF2m_mod_sqr_arr
7064          BN_GF2m_mod_inv_arr        [wrapper for BN_GF2m_mod_inv]
7065          BN_GF2m_mod_div_arr        [wrapper for BN_GF2m_mod_div]
7066          BN_GF2m_mod_exp_arr
7067          BN_GF2m_mod_sqrt_arr
7068          BN_GF2m_mod_solve_quad_arr
7069          BN_GF2m_poly2arr
7070          BN_GF2m_arr2poly
7071
7072     Conversion can be performed by the following functions:
7073
7074          BN_GF2m_poly2arr
7075          BN_GF2m_arr2poly
7076
7077     bntest.c has additional tests for binary polynomial arithmetic.
7078
7079     Two implementations for BN_GF2m_mod_div() are available.
7080     The default algorithm simply uses BN_GF2m_mod_inv() and
7081     BN_GF2m_mod_mul().  The alternative algorithm is compiled in only
7082     if OPENSSL_SUN_GF2M_DIV is defined (patent pending; read the
7083     copyright notice in crypto/bn/bn_gf2m.c before enabling it).
7084
7085     [Sheueling Chang Shantz and Douglas Stebila
7086     (Sun Microsystems Laboratories)]
7087
7088  *) Add new error code 'ERR_R_DISABLED' that can be used when some
7089     functionality is disabled at compile-time.
7090     [Douglas Stebila <douglas.stebila@sun.com>]
7091
7092  *) Change default behaviour of 'openssl asn1parse' so that more
7093     information is visible when viewing, e.g., a certificate:
7094
7095     Modify asn1_parse2 (crypto/asn1/asn1_par.c) so that in non-'dump'
7096     mode the content of non-printable OCTET STRINGs is output in a
7097     style similar to INTEGERs, but with '[HEX DUMP]' prepended to
7098     avoid the appearance of a printable string.
7099     [Nils Larsch <nla@trustcenter.de>]
7100
7101  *) Add 'asn1_flag' and 'asn1_form' member to EC_GROUP with access
7102     functions
7103          EC_GROUP_set_asn1_flag()
7104          EC_GROUP_get_asn1_flag()
7105          EC_GROUP_set_point_conversion_form()
7106          EC_GROUP_get_point_conversion_form()
7107     These control ASN1 encoding details:
7108     - Curves (i.e., groups) are encoded explicitly unless asn1_flag
7109       has been set to OPENSSL_EC_NAMED_CURVE.
7110     - Points are encoded in uncompressed form by default; options for
7111       asn1_for are as for point2oct, namely
7112          POINT_CONVERSION_COMPRESSED
7113          POINT_CONVERSION_UNCOMPRESSED
7114          POINT_CONVERSION_HYBRID
7115
7116     Also add 'seed' and 'seed_len' members to EC_GROUP with access
7117     functions
7118          EC_GROUP_set_seed()
7119          EC_GROUP_get0_seed()
7120          EC_GROUP_get_seed_len()
7121     This is used only for ASN1 purposes (so far).
7122     [Nils Larsch <nla@trustcenter.de>]
7123
7124  *) Add 'field_type' member to EC_METHOD, which holds the NID
7125     of the appropriate field type OID.  The new function
7126     EC_METHOD_get_field_type() returns this value.
7127     [Nils Larsch <nla@trustcenter.de>]
7128
7129  *) Add functions
7130          EC_POINT_point2bn()
7131          EC_POINT_bn2point()
7132          EC_POINT_point2hex()
7133          EC_POINT_hex2point()
7134     providing useful interfaces to EC_POINT_point2oct() and
7135     EC_POINT_oct2point().
7136     [Nils Larsch <nla@trustcenter.de>]
7137
7138  *) Change internals of the EC library so that the functions
7139          EC_GROUP_set_generator()
7140          EC_GROUP_get_generator()
7141          EC_GROUP_get_order()
7142          EC_GROUP_get_cofactor()
7143     are implemented directly in crypto/ec/ec_lib.c and not dispatched
7144     to methods, which would lead to unnecessary code duplication when
7145     adding different types of curves.
7146     [Nils Larsch <nla@trustcenter.de> with input by Bodo Moeller]
7147
7148  *) Implement compute_wNAF (crypto/ec/ec_mult.c) without BIGNUM
7149     arithmetic, and such that modified wNAFs are generated
7150     (which avoid length expansion in many cases).
7151     [Bodo Moeller]
7152
7153  *) Add a function EC_GROUP_check_discriminant() (defined via
7154     EC_METHOD) that verifies that the curve discriminant is non-zero.
7155
7156     Add a function EC_GROUP_check() that makes some sanity tests
7157     on a EC_GROUP, its generator and order.  This includes
7158     EC_GROUP_check_discriminant().
7159     [Nils Larsch <nla@trustcenter.de>]
7160
7161  *) Add ECDSA in new directory crypto/ecdsa/.
7162
7163     Add applications 'openssl ecparam' and 'openssl ecdsa'
7164     (these are based on 'openssl dsaparam' and 'openssl dsa').
7165
7166     ECDSA support is also included in various other files across the
7167     library.  Most notably,
7168     - 'openssl req' now has a '-newkey ecdsa:file' option;
7169     - EVP_PKCS82PKEY (crypto/evp/evp_pkey.c) now can handle ECDSA;
7170     - X509_PUBKEY_get (crypto/asn1/x_pubkey.c) and
7171       d2i_PublicKey (crypto/asn1/d2i_pu.c) have been modified to make
7172       them suitable for ECDSA where domain parameters must be
7173       extracted before the specific public key;
7174     - ECDSA engine support has been added.
7175     [Nils Larsch <nla@trustcenter.de>]
7176
7177  *) Include some named elliptic curves, and add OIDs from X9.62,
7178     SECG, and WAP/WTLS.  Each curve can be obtained from the new
7179     function
7180          EC_GROUP_new_by_curve_name(),
7181     and the list of available named curves can be obtained with
7182          EC_get_builtin_curves().
7183     Also add a 'curve_name' member to EC_GROUP objects, which can be
7184     accessed via
7185         EC_GROUP_set_curve_name()
7186         EC_GROUP_get_curve_name()
7187     [Nils Larsch <larsch@trustcenter.de, Bodo Moeller]
7188
7189  *) Remove a few calls to bn_wexpand() in BN_sqr() (the one in there
7190     was actually never needed) and in BN_mul().  The removal in BN_mul()
7191     required a small change in bn_mul_part_recursive() and the addition
7192     of the functions bn_cmp_part_words(), bn_sub_part_words() and
7193     bn_add_part_words(), which do the same thing as bn_cmp_words(),
7194     bn_sub_words() and bn_add_words() except they take arrays with
7195     differing sizes.
7196     [Richard Levitte]
7197
7198 Changes between 0.9.7l and 0.9.7m  [23 Feb 2007]
7199
7200  *) Cleanse PEM buffers before freeing them since they may contain
7201     sensitive data.
7202     [Benjamin Bennett <ben@psc.edu>]
7203
7204  *) Include "!eNULL" in SSL_DEFAULT_CIPHER_LIST to make sure that
7205     a ciphersuite string such as "DEFAULT:RSA" cannot enable
7206     authentication-only ciphersuites.
7207     [Bodo Moeller]
7208
7209  *) Since AES128 and AES256 share a single mask bit in the logic of
7210     ssl/ssl_ciph.c, the code for masking out disabled ciphers needs a
7211     kludge to work properly if AES128 is available and AES256 isn't.
7212     [Victor Duchovni]
7213
7214  *) Expand security boundary to match 1.1.1 module.
7215     [Steve Henson]
7216
7217  *) Remove redundant features: hash file source, editing of test vectors
7218     modify fipsld to use external fips_premain.c signature.
7219     [Steve Henson]
7220
7221  *) New perl script mkfipsscr.pl to create shell scripts or batch files to
7222     run algorithm test programs.
7223     [Steve Henson]
7224
7225  *) Make algorithm test programs more tolerant of whitespace.
7226     [Steve Henson]
7227
7228  *) Have SSL/TLS server implementation tolerate "mismatched" record
7229     protocol version while receiving ClientHello even if the
7230     ClientHello is fragmented.  (The server can't insist on the
7231     particular protocol version it has chosen before the ServerHello
7232     message has informed the client about his choice.)
7233     [Bodo Moeller]
7234
7235  *) Load error codes if they are not already present instead of using a
7236     static variable. This allows them to be cleanly unloaded and reloaded.
7237     [Steve Henson]
7238
7239 Changes between 0.9.7k and 0.9.7l  [28 Sep 2006]
7240
7241  *) Introduce limits to prevent malicious keys being able to
7242     cause a denial of service.  (CVE-2006-2940)
7243     [Steve Henson, Bodo Moeller]
7244
7245  *) Fix ASN.1 parsing of certain invalid structures that can result
7246     in a denial of service.  (CVE-2006-2937)  [Steve Henson]
7247
7248  *) Fix buffer overflow in SSL_get_shared_ciphers() function.
7249     (CVE-2006-3738) [Tavis Ormandy and Will Drewry, Google Security Team]
7250
7251  *) Fix SSL client code which could crash if connecting to a
7252     malicious SSLv2 server.  (CVE-2006-4343)
7253     [Tavis Ormandy and Will Drewry, Google Security Team]
7254
7255  *) Change ciphersuite string processing so that an explicit
7256     ciphersuite selects this one ciphersuite (so that "AES256-SHA"
7257     will no longer include "AES128-SHA"), and any other similar
7258     ciphersuite (same bitmap) from *other* protocol versions (so that
7259     "RC4-MD5" will still include both the SSL 2.0 ciphersuite and the
7260     SSL 3.0/TLS 1.0 ciphersuite).  This is a backport combining
7261     changes from 0.9.8b and 0.9.8d.
7262     [Bodo Moeller]
7263
7264 Changes between 0.9.7j and 0.9.7k  [05 Sep 2006]
7265
7266  *) Avoid PKCS #1 v1.5 signature attack discovered by Daniel Bleichenbacher
7267     (CVE-2006-4339)  [Ben Laurie and Google Security Team]
7268
7269  *) Change the Unix randomness entropy gathering to use poll() when
7270     possible instead of select(), since the latter has some
7271     undesirable limitations.
7272     [Darryl Miles via Richard Levitte and Bodo Moeller]
7273
7274  *) Disable rogue ciphersuites:
7275
7276      - SSLv2 0x08 0x00 0x80 ("RC4-64-MD5")
7277      - SSLv3/TLSv1 0x00 0x61 ("EXP1024-RC2-CBC-MD5")
7278      - SSLv3/TLSv1 0x00 0x60 ("EXP1024-RC4-MD5")
7279
7280     The latter two were purportedly from
7281     draft-ietf-tls-56-bit-ciphersuites-0[01].txt, but do not really
7282     appear there.
7283
7284     Also deactivate the remaining ciphersuites from
7285     draft-ietf-tls-56-bit-ciphersuites-01.txt.  These are just as
7286     unofficial, and the ID has long expired.
7287     [Bodo Moeller]
7288
7289  *) Fix RSA blinding Heisenbug (problems sometimes occurred on
7290     dual-core machines) and other potential thread-safety issues.
7291     [Bodo Moeller]
7292
7293 Changes between 0.9.7i and 0.9.7j  [04 May 2006]
7294
7295  *) Adapt fipsld and the build system to link against the validated FIPS
7296     module in FIPS mode.
7297     [Steve Henson]
7298
7299  *) Fixes for VC++ 2005 build under Windows.
7300     [Steve Henson]
7301
7302  *) Add new Windows build target VC-32-GMAKE for VC++. This uses GNU make
7303     from a Windows bash shell such as MSYS. It is autodetected from the
7304     "config" script when run from a VC++ environment. Modify standard VC++
7305     build to use fipscanister.o from the GNU make build.
7306     [Steve Henson]
7307
7308 Changes between 0.9.7h and 0.9.7i  [14 Oct 2005]
7309
7310  *) Wrapped the definition of EVP_MAX_MD_SIZE in a #ifdef OPENSSL_FIPS.
7311     The value now differs depending on if you build for FIPS or not.
7312     BEWARE!  A program linked with a shared FIPSed libcrypto can't be
7313     safely run with a non-FIPSed libcrypto, as it may crash because of
7314     the difference induced by this change.
7315     [Andy Polyakov]
7316
7317 Changes between 0.9.7g and 0.9.7h  [11 Oct 2005]
7318
7319  *) Remove the functionality of SSL_OP_MSIE_SSLV2_RSA_PADDING
7320     (part of SSL_OP_ALL).  This option used to disable the
7321     countermeasure against man-in-the-middle protocol-version
7322     rollback in the SSL 2.0 server implementation, which is a bad
7323     idea.  (CVE-2005-2969)
7324
7325     [Bodo Moeller; problem pointed out by Yutaka Oiwa (Research Center
7326     for Information Security, National Institute of Advanced Industrial
7327     Science and Technology [AIST], Japan)]
7328
7329  *) Minimal support for X9.31 signatures and PSS padding modes. This is
7330     mainly for FIPS compliance and not fully integrated at this stage.
7331     [Steve Henson]
7332
7333  *) For DSA signing, unless DSA_FLAG_NO_EXP_CONSTTIME is set, perform
7334     the exponentiation using a fixed-length exponent.  (Otherwise,
7335     the information leaked through timing could expose the secret key
7336     after many signatures; cf. Bleichenbacher's attack on DSA with
7337     biased k.)
7338     [Bodo Moeller]
7339
7340  *) Make a new fixed-window mod_exp implementation the default for
7341     RSA, DSA, and DH private-key operations so that the sequence of
7342     squares and multiplies and the memory access pattern are
7343     independent of the particular secret key.  This will mitigate
7344     cache-timing and potential related attacks.
7345
7346     BN_mod_exp_mont_consttime() is the new exponentiation implementation,
7347     and this is automatically used by BN_mod_exp_mont() if the new flag
7348     BN_FLG_EXP_CONSTTIME is set for the exponent.  RSA, DSA, and DH
7349     will use this BN flag for private exponents unless the flag
7350     RSA_FLAG_NO_EXP_CONSTTIME, DSA_FLAG_NO_EXP_CONSTTIME, or
7351     DH_FLAG_NO_EXP_CONSTTIME, respectively, is set.
7352
7353     [Matthew D Wood (Intel Corp), with some changes by Bodo Moeller]
7354
7355  *) Change the client implementation for SSLv23_method() and
7356     SSLv23_client_method() so that is uses the SSL 3.0/TLS 1.0
7357     Client Hello message format if the SSL_OP_NO_SSLv2 option is set.
7358     (Previously, the SSL 2.0 backwards compatible Client Hello
7359     message format would be used even with SSL_OP_NO_SSLv2.)
7360     [Bodo Moeller]
7361
7362  *) Add support for smime-type MIME parameter in S/MIME messages which some
7363     clients need.
7364     [Steve Henson]
7365
7366  *) New function BN_MONT_CTX_set_locked() to set montgomery parameters in
7367     a threadsafe manner. Modify rsa code to use new function and add calls
7368     to dsa and dh code (which had race conditions before).
7369     [Steve Henson]
7370
7371  *) Include the fixed error library code in the C error file definitions
7372     instead of fixing them up at runtime. This keeps the error code
7373     structures constant.
7374     [Steve Henson]
7375
7376 Changes between 0.9.7f and 0.9.7g  [11 Apr 2005]
7377
7378  [NB: OpenSSL 0.9.7h and later 0.9.7 patch levels were released after
7379  OpenSSL 0.9.8.]
7380
7381  *) Fixes for newer kerberos headers. NB: the casts are needed because
7382     the 'length' field is signed on one version and unsigned on another
7383     with no (?) obvious way to tell the difference, without these VC++
7384     complains. Also the "definition" of FAR (blank) is no longer included
7385     nor is the error ENOMEM. KRB5_PRIVATE has to be set to 1 to pick up
7386     some needed definitions.
7387     [Steve Henson]
7388
7389  *) Undo Cygwin change.
7390     [Ulf Möller]
7391
7392  *) Added support for proxy certificates according to RFC 3820.
7393     Because they may be a security thread to unaware applications,
7394     they must be explicitly allowed in run-time.  See
7395     docs/HOWTO/proxy_certificates.txt for further information.
7396     [Richard Levitte]
7397
7398 Changes between 0.9.7e and 0.9.7f  [22 Mar 2005]
7399
7400  *) Use (SSL_RANDOM_VALUE - 4) bytes of pseudo random data when generating
7401     server and client random values. Previously
7402     (SSL_RANDOM_VALUE - sizeof(time_t)) would be used which would result in
7403     less random data when sizeof(time_t) > 4 (some 64 bit platforms).
7404
7405     This change has negligible security impact because:
7406
7407     1. Server and client random values still have 24 bytes of pseudo random
7408        data.
7409
7410     2. Server and client random values are sent in the clear in the initial
7411        handshake.
7412
7413     3. The master secret is derived using the premaster secret (48 bytes in
7414        size for static RSA ciphersuites) as well as client server and random
7415        values.
7416
7417     The OpenSSL team would like to thank the UK NISCC for bringing this issue
7418     to our attention.
7419
7420     [Stephen Henson, reported by UK NISCC]
7421
7422  *) Use Windows randomness collection on Cygwin.
7423     [Ulf Möller]
7424
7425  *) Fix hang in EGD/PRNGD query when communication socket is closed
7426     prematurely by EGD/PRNGD.
7427     [Darren Tucker <dtucker@zip.com.au> via Lutz Jänicke, resolves #1014]
7428
7429  *) Prompt for pass phrases when appropriate for PKCS12 input format.
7430     [Steve Henson]
7431
7432  *) Back-port of selected performance improvements from development
7433     branch, as well as improved support for PowerPC platforms.
7434     [Andy Polyakov]
7435
7436  *) Add lots of checks for memory allocation failure, error codes to indicate
7437     failure and freeing up memory if a failure occurs.
7438     [Nauticus Networks SSL Team <openssl@nauticusnet.com>, Steve Henson]
7439
7440  *) Add new -passin argument to dgst.
7441     [Steve Henson]
7442
7443  *) Perform some character comparisons of different types in X509_NAME_cmp:
7444     this is needed for some certificates that re-encode DNs into UTF8Strings
7445     (in violation of RFC3280) and can't or won't issue name rollover
7446     certificates.
7447     [Steve Henson]
7448
7449  *) Make an explicit check during certificate validation to see that
7450     the CA setting in each certificate on the chain is correct.  As a
7451     side effect always do the following basic checks on extensions,
7452     not just when there's an associated purpose to the check:
7453
7454      - if there is an unhandled critical extension (unless the user
7455        has chosen to ignore this fault)
7456      - if the path length has been exceeded (if one is set at all)
7457      - that certain extensions fit the associated purpose (if one has
7458        been given)
7459     [Richard Levitte]
7460
7461 Changes between 0.9.7d and 0.9.7e  [25 Oct 2004]
7462
7463  *) Avoid a race condition when CRLs are checked in a multi threaded
7464     environment. This would happen due to the reordering of the revoked
7465     entries during signature checking and serial number lookup. Now the
7466     encoding is cached and the serial number sort performed under a lock.
7467     Add new STACK function sk_is_sorted().
7468     [Steve Henson]
7469
7470  *) Add Delta CRL to the extension code.
7471     [Steve Henson]
7472
7473  *) Various fixes to s3_pkt.c so alerts are sent properly.
7474     [David Holmes <d.holmes@f5.com>]
7475
7476  *) Reduce the chances of duplicate issuer name and serial numbers (in
7477     violation of RFC3280) using the OpenSSL certificate creation utilities.
7478     This is done by creating a random 64 bit value for the initial serial
7479     number when a serial number file is created or when a self signed
7480     certificate is created using 'openssl req -x509'. The initial serial
7481     number file is created using 'openssl x509 -next_serial' in CA.pl
7482     rather than being initialized to 1.
7483     [Steve Henson]
7484
7485 Changes between 0.9.7c and 0.9.7d  [17 Mar 2004]
7486
7487  *) Fix null-pointer assignment in do_change_cipher_spec() revealed
7488     by using the Codenomicon TLS Test Tool (CVE-2004-0079)
7489     [Joe Orton, Steve Henson]
7490
7491  *) Fix flaw in SSL/TLS handshaking when using Kerberos ciphersuites
7492     (CVE-2004-0112)
7493     [Joe Orton, Steve Henson]
7494
7495  *) Make it possible to have multiple active certificates with the same
7496     subject in the CA index file.  This is done only if the keyword
7497     'unique_subject' is set to 'no' in the main CA section (default
7498     if 'CA_default') of the configuration file.  The value is saved
7499     with the database itself in a separate index attribute file,
7500     named like the index file with '.attr' appended to the name.
7501     [Richard Levitte]
7502
7503  *) X509 verify fixes. Disable broken certificate workarounds when
7504     X509_V_FLAGS_X509_STRICT is set. Check CRL issuer has cRLSign set if
7505     keyUsage extension present. Don't accept CRLs with unhandled critical
7506     extensions: since verify currently doesn't process CRL extensions this
7507     rejects a CRL with *any* critical extensions. Add new verify error codes
7508     for these cases.
7509     [Steve Henson]
7510
7511  *) When creating an OCSP nonce use an OCTET STRING inside the extnValue.
7512     A clarification of RFC2560 will require the use of OCTET STRINGs and
7513     some implementations cannot handle the current raw format. Since OpenSSL
7514     copies and compares OCSP nonces as opaque blobs without any attempt at
7515     parsing them this should not create any compatibility issues.
7516     [Steve Henson]
7517
7518  *) New md flag EVP_MD_CTX_FLAG_REUSE this allows md_data to be reused when
7519     calling EVP_MD_CTX_copy_ex() to avoid calling OPENSSL_malloc(). Without
7520     this HMAC (and other) operations are several times slower than OpenSSL
7521     < 0.9.7.
7522     [Steve Henson]
7523
7524  *) Print out GeneralizedTime and UTCTime in ASN1_STRING_print_ex().
7525     [Peter Sylvester <Peter.Sylvester@EdelWeb.fr>]
7526
7527  *) Use the correct content when signing type "other".
7528     [Steve Henson]
7529
7530 Changes between 0.9.7b and 0.9.7c  [30 Sep 2003]
7531
7532  *) Fix various bugs revealed by running the NISCC test suite:
7533
7534     Stop out of bounds reads in the ASN1 code when presented with
7535     invalid tags (CVE-2003-0543 and CVE-2003-0544).
7536
7537     Free up ASN1_TYPE correctly if ANY type is invalid (CVE-2003-0545).
7538
7539     If verify callback ignores invalid public key errors don't try to check
7540     certificate signature with the NULL public key.
7541
7542     [Steve Henson]
7543
7544  *) New -ignore_err option in ocsp application to stop the server
7545     exiting on the first error in a request.
7546     [Steve Henson]
7547
7548  *) In ssl3_accept() (ssl/s3_srvr.c) only accept a client certificate
7549     if the server requested one: as stated in TLS 1.0 and SSL 3.0
7550     specifications.
7551     [Steve Henson]
7552
7553  *) In ssl3_get_client_hello() (ssl/s3_srvr.c), tolerate additional
7554     extra data after the compression methods not only for TLS 1.0
7555     but also for SSL 3.0 (as required by the specification).
7556     [Bodo Moeller; problem pointed out by Matthias Loepfe]
7557
7558  *) Change X509_certificate_type() to mark the key as exported/exportable
7559     when it's 512 *bits* long, not 512 bytes.
7560     [Richard Levitte]
7561
7562  *) Change AES_cbc_encrypt() so it outputs exact multiple of
7563     blocks during encryption.
7564     [Richard Levitte]
7565
7566  *) Various fixes to base64 BIO and non blocking I/O. On write
7567     flushes were not handled properly if the BIO retried. On read
7568     data was not being buffered properly and had various logic bugs.
7569     This also affects blocking I/O when the data being decoded is a
7570     certain size.
7571     [Steve Henson]
7572
7573  *) Various S/MIME bugfixes and compatibility changes:
7574     output correct application/pkcs7 MIME type if
7575     PKCS7_NOOLDMIMETYPE is set. Tolerate some broken signatures.
7576     Output CR+LF for EOL if PKCS7_CRLFEOL is set (this makes opening
7577     of files as .eml work). Correctly handle very long lines in MIME
7578     parser.
7579     [Steve Henson]
7580
7581 Changes between 0.9.7a and 0.9.7b  [10 Apr 2003]
7582
7583  *) Countermeasure against the Klima-Pokorny-Rosa extension of
7584     Bleichbacher's attack on PKCS #1 v1.5 padding: treat
7585     a protocol version number mismatch like a decryption error
7586     in ssl3_get_client_key_exchange (ssl/s3_srvr.c).
7587     [Bodo Moeller]
7588
7589  *) Turn on RSA blinding by default in the default implementation
7590     to avoid a timing attack. Applications that don't want it can call
7591     RSA_blinding_off() or use the new flag RSA_FLAG_NO_BLINDING.
7592     They would be ill-advised to do so in most cases.
7593     [Ben Laurie, Steve Henson, Geoff Thorpe, Bodo Moeller]
7594
7595  *) Change RSA blinding code so that it works when the PRNG is not
7596     seeded (in this case, the secret RSA exponent is abused as
7597     an unpredictable seed -- if it is not unpredictable, there
7598     is no point in blinding anyway).  Make RSA blinding thread-safe
7599     by remembering the creator's thread ID in rsa->blinding and
7600     having all other threads use local one-time blinding factors
7601     (this requires more computation than sharing rsa->blinding, but
7602     avoids excessive locking; and if an RSA object is not shared
7603     between threads, blinding will still be very fast).
7604     [Bodo Moeller]
7605
7606  *) Fixed a typo bug that would cause ENGINE_set_default() to set an
7607     ENGINE as defaults for all supported algorithms irrespective of
7608     the 'flags' parameter. 'flags' is now honoured, so applications
7609     should make sure they are passing it correctly.
7610     [Geoff Thorpe]
7611
7612  *) Target "mingw" now allows native Windows code to be generated in
7613     the Cygwin environment as well as with the MinGW compiler.
7614     [Ulf Moeller]
7615
7616 Changes between 0.9.7 and 0.9.7a  [19 Feb 2003]
7617
7618  *) In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked
7619     via timing by performing a MAC computation even if incorrect
7620     block cipher padding has been found.  This is a countermeasure
7621     against active attacks where the attacker has to distinguish
7622     between bad padding and a MAC verification error. (CVE-2003-0078)
7623
7624     [Bodo Moeller; problem pointed out by Brice Canvel (EPFL),
7625     Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and
7626     Martin Vuagnoux (EPFL, Ilion)]
7627
7628  *) Make the no-err option work as intended.  The intention with no-err
7629     is not to have the whole error stack handling routines removed from
7630     libcrypto, it's only intended to remove all the function name and
7631     reason texts, thereby removing some of the footprint that may not
7632     be interesting if those errors aren't displayed anyway.
7633
7634     NOTE: it's still possible for any application or module to have its
7635     own set of error texts inserted.  The routines are there, just not
7636     used by default when no-err is given.
7637     [Richard Levitte]
7638
7639  *) Add support for FreeBSD on IA64.
7640     [dirk.meyer@dinoex.sub.org via Richard Levitte, resolves #454]
7641
7642  *) Adjust DES_cbc_cksum() so it returns the same value as the MIT
7643     Kerberos function mit_des_cbc_cksum().  Before this change,
7644     the value returned by DES_cbc_cksum() was like the one from
7645     mit_des_cbc_cksum(), except the bytes were swapped.
7646     [Kevin Greaney <Kevin.Greaney@hp.com> and Richard Levitte]
7647
7648  *) Allow an application to disable the automatic SSL chain building.
7649     Before this a rather primitive chain build was always performed in
7650     ssl3_output_cert_chain(): an application had no way to send the
7651     correct chain if the automatic operation produced an incorrect result.
7652
7653     Now the chain builder is disabled if either:
7654
7655     1. Extra certificates are added via SSL_CTX_add_extra_chain_cert().
7656
7657     2. The mode flag SSL_MODE_NO_AUTO_CHAIN is set.
7658
7659     The reasoning behind this is that an application would not want the
7660     auto chain building to take place if extra chain certificates are
7661     present and it might also want a means of sending no additional
7662     certificates (for example the chain has two certificates and the
7663     root is omitted).
7664     [Steve Henson]
7665
7666  *) Add the possibility to build without the ENGINE framework.
7667     [Steven Reddie <smr@essemer.com.au> via Richard Levitte]
7668
7669  *) Under Win32 gmtime() can return NULL: check return value in
7670     OPENSSL_gmtime(). Add error code for case where gmtime() fails.
7671     [Steve Henson]
7672
7673  *) DSA routines: under certain error conditions uninitialized BN objects
7674     could be freed. Solution: make sure initialization is performed early
7675     enough. (Reported and fix supplied by Ivan D Nestlerode <nestler@MIT.EDU>,
7676     Nils Larsch <nla@trustcenter.de> via PR#459)
7677     [Lutz Jaenicke]
7678
7679  *) Another fix for SSLv2 session ID handling: the session ID was incorrectly
7680     checked on reconnect on the client side, therefore session resumption
7681     could still fail with a "ssl session id is different" error. This
7682     behaviour is masked when SSL_OP_ALL is used due to
7683     SSL_OP_MICROSOFT_SESS_ID_BUG being set.
7684     Behaviour observed by Crispin Flowerday <crispin@flowerday.cx> as
7685     followup to PR #377.
7686     [Lutz Jaenicke]
7687
7688  *) IA-32 assembler support enhancements: unified ELF targets, support
7689     for SCO/Caldera platforms, fix for Cygwin shared build.
7690     [Andy Polyakov]
7691
7692  *) Add support for FreeBSD on sparc64.  As a consequence, support for
7693     FreeBSD on non-x86 processors is separate from x86 processors on
7694     the config script, much like the NetBSD support.
7695     [Richard Levitte & Kris Kennaway <kris@obsecurity.org>]
7696
7697 Changes between 0.9.6h and 0.9.7  [31 Dec 2002]
7698
7699  [NB: OpenSSL 0.9.6i and later 0.9.6 patch levels were released after
7700  OpenSSL 0.9.7.]
7701
7702  *) Fix session ID handling in SSLv2 client code: the SERVER FINISHED
7703     code (06) was taken as the first octet of the session ID and the last
7704     octet was ignored consequently. As a result SSLv2 client side session
7705     caching could not have worked due to the session ID mismatch between
7706     client and server.
7707     Behaviour observed by Crispin Flowerday <crispin@flowerday.cx> as
7708     PR #377.
7709     [Lutz Jaenicke]
7710
7711  *) Change the declaration of needed Kerberos libraries to use EX_LIBS
7712     instead of the special (and badly supported) LIBKRB5.  LIBKRB5 is
7713     removed entirely.
7714     [Richard Levitte]
7715
7716  *) The hw_ncipher.c engine requires dynamic locks.  Unfortunately, it
7717     seems that in spite of existing for more than a year, many application
7718     author have done nothing to provide the necessary callbacks, which
7719     means that this particular engine will not work properly anywhere.
7720     This is a very unfortunate situation which forces us, in the name
7721     of usability, to give the hw_ncipher.c a static lock, which is part
7722     of libcrypto.
7723     NOTE: This is for the 0.9.7 series ONLY.  This hack will never
7724     appear in 0.9.8 or later.  We EXPECT application authors to have
7725     dealt properly with this when 0.9.8 is released (unless we actually
7726     make such changes in the libcrypto locking code that changes will
7727     have to be made anyway).
7728     [Richard Levitte]
7729
7730  *) In asn1_d2i_read_bio() repeatedly call BIO_read() until all content
7731     octets have been read, EOF or an error occurs. Without this change
7732     some truncated ASN1 structures will not produce an error.
7733     [Steve Henson]
7734
7735  *) Disable Heimdal support, since it hasn't been fully implemented.
7736     Still give the possibility to force the use of Heimdal, but with
7737     warnings and a request that patches get sent to openssl-dev.
7738     [Richard Levitte]
7739
7740  *) Add the VC-CE target, introduce the WINCE sysname, and add
7741     INSTALL.WCE and appropriate conditionals to make it build.
7742     [Steven Reddie <smr@essemer.com.au> via Richard Levitte]
7743
7744  *) Change the DLL names for Cygwin to cygcrypto-x.y.z.dll and
7745     cygssl-x.y.z.dll, where x, y and z are the major, minor and
7746     edit numbers of the version.
7747     [Corinna Vinschen <vinschen@redhat.com> and Richard Levitte]
7748
7749  *) Introduce safe string copy and catenation functions
7750     (BUF_strlcpy() and BUF_strlcat()).
7751     [Ben Laurie (CHATS) and Richard Levitte]
7752
7753  *) Avoid using fixed-size buffers for one-line DNs.
7754     [Ben Laurie (CHATS)]
7755
7756  *) Add BUF_MEM_grow_clean() to avoid information leakage when
7757     resizing buffers containing secrets, and use where appropriate.
7758     [Ben Laurie (CHATS)]
7759
7760  *) Avoid using fixed size buffers for configuration file location.
7761     [Ben Laurie (CHATS)]
7762
7763  *) Avoid filename truncation for various CA files.
7764     [Ben Laurie (CHATS)]
7765
7766  *) Use sizeof in preference to magic numbers.
7767     [Ben Laurie (CHATS)]
7768
7769  *) Avoid filename truncation in cert requests.
7770     [Ben Laurie (CHATS)]
7771
7772  *) Add assertions to check for (supposedly impossible) buffer
7773     overflows.
7774     [Ben Laurie (CHATS)]
7775
7776  *) Don't cache truncated DNS entries in the local cache (this could
7777     potentially lead to a spoofing attack).
7778     [Ben Laurie (CHATS)]
7779
7780  *) Fix various buffers to be large enough for hex/decimal
7781     representations in a platform independent manner.
7782     [Ben Laurie (CHATS)]
7783
7784  *) Add CRYPTO_realloc_clean() to avoid information leakage when
7785     resizing buffers containing secrets, and use where appropriate.
7786     [Ben Laurie (CHATS)]
7787
7788  *) Add BIO_indent() to avoid much slightly worrying code to do
7789     indents.
7790     [Ben Laurie (CHATS)]
7791
7792  *) Convert sprintf()/BIO_puts() to BIO_printf().
7793     [Ben Laurie (CHATS)]
7794
7795  *) buffer_gets() could terminate with the buffer only half
7796     full. Fixed.
7797     [Ben Laurie (CHATS)]
7798
7799  *) Add assertions to prevent user-supplied crypto functions from
7800     overflowing internal buffers by having large block sizes, etc.
7801     [Ben Laurie (CHATS)]
7802
7803  *) New OPENSSL_assert() macro (similar to assert(), but enabled
7804     unconditionally).
7805     [Ben Laurie (CHATS)]
7806
7807  *) Eliminate unused copy of key in RC4.
7808     [Ben Laurie (CHATS)]
7809
7810  *) Eliminate unused and incorrectly sized buffers for IV in pem.h.
7811     [Ben Laurie (CHATS)]
7812
7813  *) Fix off-by-one error in EGD path.
7814     [Ben Laurie (CHATS)]
7815
7816  *) If RANDFILE path is too long, ignore instead of truncating.
7817     [Ben Laurie (CHATS)]
7818
7819  *) Eliminate unused and incorrectly sized X.509 structure
7820     CBCParameter.
7821     [Ben Laurie (CHATS)]
7822
7823  *) Eliminate unused and dangerous function knumber().
7824     [Ben Laurie (CHATS)]
7825
7826  *) Eliminate unused and dangerous structure, KSSL_ERR.
7827     [Ben Laurie (CHATS)]
7828
7829  *) Protect against overlong session ID context length in an encoded
7830     session object. Since these are local, this does not appear to be
7831     exploitable.
7832     [Ben Laurie (CHATS)]
7833
7834  *) Change from security patch (see 0.9.6e below) that did not affect
7835     the 0.9.6 release series:
7836
7837     Remote buffer overflow in SSL3 protocol - an attacker could
7838     supply an oversized master key in Kerberos-enabled versions.
7839     (CVE-2002-0657)
7840     [Ben Laurie (CHATS)]
7841
7842  *) Change the SSL kerb5 codes to match RFC 2712.
7843     [Richard Levitte]
7844
7845  *) Make -nameopt work fully for req and add -reqopt switch.
7846     [Michael Bell <michael.bell@rz.hu-berlin.de>, Steve Henson]
7847
7848  *) The "block size" for block ciphers in CFB and OFB mode should be 1.
7849     [Steve Henson, reported by Yngve Nysaeter Pettersen <yngve@opera.com>]
7850
7851  *) Make sure tests can be performed even if the corresponding algorithms
7852     have been removed entirely.  This was also the last step to make
7853     OpenSSL compilable with DJGPP under all reasonable conditions.
7854     [Richard Levitte, Doug Kaufman <dkaufman@rahul.net>]
7855
7856  *) Add cipher selection rules COMPLEMENTOFALL and COMPLEMENTOFDEFAULT
7857     to allow version independent disabling of normally unselected ciphers,
7858     which may be activated as a side-effect of selecting a single cipher.
7859
7860     (E.g., cipher list string "RSA" enables ciphersuites that are left
7861     out of "ALL" because they do not provide symmetric encryption.
7862     "RSA:!COMPLEMEMENTOFALL" avoids these unsafe ciphersuites.)
7863     [Lutz Jaenicke, Bodo Moeller]
7864
7865  *) Add appropriate support for separate platform-dependent build
7866     directories.  The recommended way to make a platform-dependent
7867     build directory is the following (tested on Linux), maybe with
7868     some local tweaks:
7869
7870        # Place yourself outside of the OpenSSL source tree.  In
7871        # this example, the environment variable OPENSSL_SOURCE
7872        # is assumed to contain the absolute OpenSSL source directory.
7873        mkdir -p objtree/"`uname -s`-`uname -r`-`uname -m`"
7874        cd objtree/"`uname -s`-`uname -r`-`uname -m`"
7875        (cd $OPENSSL_SOURCE; find . -type f) | while read F; do
7876                mkdir -p `dirname $F`
7877                ln -s $OPENSSL_SOURCE/$F $F
7878        done
7879
7880     To be absolutely sure not to disturb the source tree, a "make clean"
7881     is a good thing.  If it isn't successful, don't worry about it,
7882     it probably means the source directory is very clean.
7883     [Richard Levitte]
7884
7885  *) Make sure any ENGINE control commands make local copies of string
7886     pointers passed to them whenever necessary. Otherwise it is possible
7887     the caller may have overwritten (or deallocated) the original string
7888     data when a later ENGINE operation tries to use the stored values.
7889     [Götz Babin-Ebell <babinebell@trustcenter.de>]
7890
7891  *) Improve diagnostics in file reading and command-line digests.
7892     [Ben Laurie aided and abetted by Solar Designer <solar@openwall.com>]
7893
7894  *) Add AES modes CFB and OFB to the object database.  Correct an
7895     error in AES-CFB decryption.
7896     [Richard Levitte]
7897
7898  *) Remove most calls to EVP_CIPHER_CTX_cleanup() in evp_enc.c, this
7899     allows existing EVP_CIPHER_CTX structures to be reused after
7900     calling EVP_*Final(). This behaviour is used by encryption
7901     BIOs and some applications. This has the side effect that
7902     applications must explicitly clean up cipher contexts with
7903     EVP_CIPHER_CTX_cleanup() or they will leak memory.
7904     [Steve Henson]
7905
7906  *) Check the values of dna and dnb in bn_mul_recursive before calling
7907     bn_mul_comba (a non zero value means the a or b arrays do not contain
7908     n2 elements) and fallback to bn_mul_normal if either is not zero.
7909     [Steve Henson]
7910
7911  *) Fix escaping of non-ASCII characters when using the -subj option
7912     of the "openssl req" command line tool. (Robert Joop <joop@fokus.gmd.de>)
7913     [Lutz Jaenicke]
7914
7915  *) Make object definitions compliant to LDAP (RFC2256): SN is the short
7916     form for "surname", serialNumber has no short form.
7917     Use "mail" as the short name for "rfc822Mailbox" according to RFC2798;
7918     therefore remove "mail" short name for "internet 7".
7919     The OID for unique identifiers in X509 certificates is
7920     x500UniqueIdentifier, not uniqueIdentifier.
7921     Some more OID additions. (Michael Bell <michael.bell@rz.hu-berlin.de>)
7922     [Lutz Jaenicke]
7923
7924  *) Add an "init" command to the ENGINE config module and auto initialize
7925     ENGINEs. Without any "init" command the ENGINE will be initialized
7926     after all ctrl commands have been executed on it. If init=1 the
7927     ENGINE is initialized at that point (ctrls before that point are run
7928     on the uninitialized ENGINE and after on the initialized one). If
7929     init=0 then the ENGINE will not be initialized at all.
7930     [Steve Henson]
7931
7932  *) Fix the 'app_verify_callback' interface so that the user-defined
7933     argument is actually passed to the callback: In the
7934     SSL_CTX_set_cert_verify_callback() prototype, the callback
7935     declaration has been changed from
7936          int (*cb)()
7937     into
7938          int (*cb)(X509_STORE_CTX *,void *);
7939     in ssl_verify_cert_chain (ssl/ssl_cert.c), the call
7940          i=s->ctx->app_verify_callback(&ctx)
7941     has been changed into
7942          i=s->ctx->app_verify_callback(&ctx, s->ctx->app_verify_arg).
7943
7944     To update applications using SSL_CTX_set_cert_verify_callback(),
7945     a dummy argument can be added to their callback functions.
7946     [D. K. Smetters <smetters@parc.xerox.com>]
7947
7948  *) Added the '4758cca' ENGINE to support IBM 4758 cards.
7949     [Maurice Gittens <maurice@gittens.nl>, touchups by Geoff Thorpe]
7950
7951  *) Add and OPENSSL_LOAD_CONF define which will cause
7952     OpenSSL_add_all_algorithms() to load the openssl.cnf config file.
7953     This allows older applications to transparently support certain
7954     OpenSSL features: such as crypto acceleration and dynamic ENGINE loading.
7955     Two new functions OPENSSL_add_all_algorithms_noconf() which will never
7956     load the config file and OPENSSL_add_all_algorithms_conf() which will
7957     always load it have also been added.
7958     [Steve Henson]
7959
7960  *) Add the OFB, CFB and CTR (all with 128 bit feedback) to AES.
7961     Adjust NIDs and EVP layer.
7962     [Stephen Sprunk <stephen@sprunk.org> and Richard Levitte]
7963
7964  *) Config modules support in openssl utility.
7965
7966     Most commands now load modules from the config file,
7967     though in a few (such as version) this isn't done
7968     because it couldn't be used for anything.
7969
7970     In the case of ca and req the config file used is
7971     the same as the utility itself: that is the -config
7972     command line option can be used to specify an
7973     alternative file.
7974     [Steve Henson]
7975
7976  *) Move default behaviour from OPENSSL_config(). If appname is NULL
7977     use "openssl_conf" if filename is NULL use default openssl config file.
7978     [Steve Henson]
7979
7980  *) Add an argument to OPENSSL_config() to allow the use of an alternative
7981     config section name. Add a new flag to tolerate a missing config file
7982     and move code to CONF_modules_load_file().
7983     [Steve Henson]
7984
7985  *) Support for crypto accelerator cards from Accelerated Encryption
7986     Processing, www.aep.ie.  (Use engine 'aep')
7987     The support was copied from 0.9.6c [engine] and adapted/corrected
7988     to work with the new engine framework.
7989     [AEP Inc. and Richard Levitte]
7990
7991  *) Support for SureWare crypto accelerator cards from Baltimore
7992     Technologies.  (Use engine 'sureware')
7993     The support was copied from 0.9.6c [engine] and adapted
7994     to work with the new engine framework.
7995     [Richard Levitte]
7996
7997  *) Have the CHIL engine fork-safe (as defined by nCipher) and actually
7998     make the newer ENGINE framework commands for the CHIL engine work.
7999     [Toomas Kiisk <vix@cyber.ee> and Richard Levitte]
8000
8001  *) Make it possible to produce shared libraries on ReliantUNIX.
8002     [Robert Dahlem <Robert.Dahlem@ffm2.siemens.de> via Richard Levitte]
8003
8004  *) Add the configuration target debug-linux-ppro.
8005     Make 'openssl rsa' use the general key loading routines
8006     implemented in apps.c, and make those routines able to
8007     handle the key format FORMAT_NETSCAPE and the variant
8008     FORMAT_IISSGC.
8009     [Toomas Kiisk <vix@cyber.ee> via Richard Levitte]
8010
8011 *) Fix a crashbug and a logic bug in hwcrhk_load_pubkey().
8012     [Toomas Kiisk <vix@cyber.ee> via Richard Levitte]
8013
8014  *) Add -keyform to rsautl, and document -engine.
8015     [Richard Levitte, inspired by Toomas Kiisk <vix@cyber.ee>]
8016
8017  *) Change BIO_new_file (crypto/bio/bss_file.c) to use new
8018     BIO_R_NO_SUCH_FILE error code rather than the generic
8019     ERR_R_SYS_LIB error code if fopen() fails with ENOENT.
8020     [Ben Laurie]
8021
8022  *) Add new functions
8023          ERR_peek_last_error
8024          ERR_peek_last_error_line
8025          ERR_peek_last_error_line_data.
8026     These are similar to
8027          ERR_peek_error
8028          ERR_peek_error_line
8029          ERR_peek_error_line_data,
8030     but report on the latest error recorded rather than the first one
8031     still in the error queue.
8032     [Ben Laurie, Bodo Moeller]
8033
8034  *) default_algorithms option in ENGINE config module. This allows things
8035     like:
8036     default_algorithms = ALL
8037     default_algorithms = RSA, DSA, RAND, CIPHERS, DIGESTS
8038     [Steve Henson]
8039
8040  *) Preliminary ENGINE config module.
8041     [Steve Henson]
8042
8043  *) New experimental application configuration code.
8044     [Steve Henson]
8045
8046  *) Change the AES code to follow the same name structure as all other
8047     symmetric ciphers, and behave the same way.  Move everything to
8048     the directory crypto/aes, thereby obsoleting crypto/rijndael.
8049     [Stephen Sprunk <stephen@sprunk.org> and Richard Levitte]
8050
8051  *) SECURITY: remove unsafe setjmp/signal interaction from ui_openssl.c.
8052     [Ben Laurie and Theo de Raadt]
8053
8054  *) Add option to output public keys in req command.
8055     [Massimiliano Pala madwolf@openca.org]
8056
8057  *) Use wNAFs in EC_POINTs_mul() for improved efficiency
8058     (up to about 10% better than before for P-192 and P-224).
8059     [Bodo Moeller]
8060
8061  *) New functions/macros
8062
8063          SSL_CTX_set_msg_callback(ctx, cb)
8064          SSL_CTX_set_msg_callback_arg(ctx, arg)
8065          SSL_set_msg_callback(ssl, cb)
8066          SSL_set_msg_callback_arg(ssl, arg)
8067
8068     to request calling a callback function
8069
8070          void cb(int write_p, int version, int content_type,
8071                  const void *buf, size_t len, SSL *ssl, void *arg)
8072
8073     whenever a protocol message has been completely received
8074     (write_p == 0) or sent (write_p == 1).  Here 'version' is the
8075     protocol version  according to which the SSL library interprets
8076     the current protocol message (SSL2_VERSION, SSL3_VERSION, or
8077     TLS1_VERSION).  'content_type' is 0 in the case of SSL 2.0, or
8078     the content type as defined in the SSL 3.0/TLS 1.0 protocol
8079     specification (change_cipher_spec(20), alert(21), handshake(22)).
8080     'buf' and 'len' point to the actual message, 'ssl' to the
8081     SSL object, and 'arg' is the application-defined value set by
8082     SSL[_CTX]_set_msg_callback_arg().
8083
8084     'openssl s_client' and 'openssl s_server' have new '-msg' options
8085     to enable a callback that displays all protocol messages.
8086     [Bodo Moeller]
8087
8088  *) Change the shared library support so shared libraries are built as
8089     soon as the corresponding static library is finished, and thereby get
8090     openssl and the test programs linked against the shared library.
8091     This still only happens when the keyword "shard" has been given to
8092     the configuration scripts.
8093
8094     NOTE: shared library support is still an experimental thing, and
8095     backward binary compatibility is still not guaranteed.
8096     ["Maciej W. Rozycki" <macro@ds2.pg.gda.pl> and Richard Levitte]
8097
8098  *) Add support for Subject Information Access extension.
8099     [Peter Sylvester <Peter.Sylvester@EdelWeb.fr>]
8100
8101  *) Make BUF_MEM_grow() behaviour more consistent: Initialise to zero
8102     additional bytes when new memory had to be allocated, not just
8103     when reusing an existing buffer.
8104     [Bodo Moeller]
8105
8106  *) New command line and configuration option 'utf8' for the req command.
8107     This allows field values to be specified as UTF8 strings.
8108     [Steve Henson]
8109
8110  *) Add -multi and -mr options to "openssl speed" - giving multiple parallel
8111     runs for the former and machine-readable output for the latter.
8112     [Ben Laurie]
8113
8114  *) Add '-noemailDN' option to 'openssl ca'.  This prevents inclusion
8115     of the e-mail address in the DN (i.e., it will go into a certificate
8116     extension only).  The new configuration file option 'email_in_dn = no'
8117     has the same effect.
8118     [Massimiliano Pala madwolf@openca.org]
8119
8120  *) Change all functions with names starting with des_ to be starting
8121     with DES_ instead.  Add wrappers that are compatible with libdes,
8122     but are named _ossl_old_des_*.  Finally, add macros that map the
8123     des_* symbols to the corresponding _ossl_old_des_* if libdes
8124     compatibility is desired.  If OpenSSL 0.9.6c compatibility is
8125     desired, the des_* symbols will be mapped to DES_*, with one
8126     exception.
8127
8128     Since we provide two compatibility mappings, the user needs to
8129     define the macro OPENSSL_DES_LIBDES_COMPATIBILITY if libdes
8130     compatibility is desired.  The default (i.e., when that macro
8131     isn't defined) is OpenSSL 0.9.6c compatibility.
8132
8133     There are also macros that enable and disable the support of old
8134     des functions altogether.  Those are OPENSSL_ENABLE_OLD_DES_SUPPORT
8135     and OPENSSL_DISABLE_OLD_DES_SUPPORT.  If none or both of those
8136     are defined, the default will apply: to support the old des routines.
8137
8138     In either case, one must include openssl/des.h to get the correct
8139     definitions.  Do not try to just include openssl/des_old.h, that
8140     won't work.
8141
8142     NOTE: This is a major break of an old API into a new one.  Software
8143     authors are encouraged to switch to the DES_ style functions.  Some
8144     time in the future, des_old.h and the libdes compatibility functions
8145     will be disable (i.e. OPENSSL_DISABLE_OLD_DES_SUPPORT will be the
8146     default), and then completely removed.
8147     [Richard Levitte]
8148
8149  *) Test for certificates which contain unsupported critical extensions.
8150     If such a certificate is found during a verify operation it is
8151     rejected by default: this behaviour can be overridden by either
8152     handling the new error X509_V_ERR_UNHANDLED_CRITICAL_EXTENSION or
8153     by setting the verify flag X509_V_FLAG_IGNORE_CRITICAL. A new function
8154     X509_supported_extension() has also been added which returns 1 if a
8155     particular extension is supported.
8156     [Steve Henson]
8157
8158  *) Modify the behaviour of EVP cipher functions in similar way to digests
8159     to retain compatibility with existing code.
8160     [Steve Henson]
8161
8162  *) Modify the behaviour of EVP_DigestInit() and EVP_DigestFinal() to retain
8163     compatibility with existing code. In particular the 'ctx' parameter does
8164     not have to be to be initialized before the call to EVP_DigestInit() and
8165     it is tidied up after a call to EVP_DigestFinal(). New function
8166     EVP_DigestFinal_ex() which does not tidy up the ctx. Similarly function
8167     EVP_MD_CTX_copy() changed to not require the destination to be
8168     initialized valid and new function EVP_MD_CTX_copy_ex() added which
8169     requires the destination to be valid.
8170
8171     Modify all the OpenSSL digest calls to use EVP_DigestInit_ex(),
8172     EVP_DigestFinal_ex() and EVP_MD_CTX_copy_ex().
8173     [Steve Henson]
8174
8175  *) Change ssl3_get_message (ssl/s3_both.c) and the functions using it
8176     so that complete 'Handshake' protocol structures are kept in memory
8177     instead of overwriting 'msg_type' and 'length' with 'body' data.
8178     [Bodo Moeller]
8179
8180  *) Add an implementation of SSL_add_dir_cert_subjects_to_stack for Win32.
8181     [Massimo Santin via Richard Levitte]
8182
8183  *) Major restructuring to the underlying ENGINE code. This includes
8184     reduction of linker bloat, separation of pure "ENGINE" manipulation
8185     (initialisation, etc) from functionality dealing with implementations
8186     of specific crypto interfaces. This change also introduces integrated
8187     support for symmetric ciphers and digest implementations - so ENGINEs
8188     can now accelerate these by providing EVP_CIPHER and EVP_MD
8189     implementations of their own. This is detailed in crypto/engine/README
8190     as it couldn't be adequately described here. However, there are a few
8191     API changes worth noting - some RSA, DSA, DH, and RAND functions that
8192     were changed in the original introduction of ENGINE code have now
8193     reverted back - the hooking from this code to ENGINE is now a good
8194     deal more passive and at run-time, operations deal directly with
8195     RSA_METHODs, DSA_METHODs (etc) as they did before, rather than
8196     dereferencing through an ENGINE pointer any more. Also, the ENGINE
8197     functions dealing with BN_MOD_EXP[_CRT] handlers have been removed -
8198     they were not being used by the framework as there is no concept of a
8199     BIGNUM_METHOD and they could not be generalised to the new
8200     'ENGINE_TABLE' mechanism that underlies the new code. Similarly,
8201     ENGINE_cpy() has been removed as it cannot be consistently defined in
8202     the new code.
8203     [Geoff Thorpe]
8204
8205  *) Change ASN1_GENERALIZEDTIME_check() to allow fractional seconds.
8206     [Steve Henson]
8207
8208  *) Change mkdef.pl to sort symbols that get the same entry number,
8209     and make sure the automatically generated functions ERR_load_*
8210     become part of libeay.num as well.
8211     [Richard Levitte]
8212
8213  *) New function SSL_renegotiate_pending().  This returns true once
8214     renegotiation has been requested (either SSL_renegotiate() call
8215     or HelloRequest/ClientHello received from the peer) and becomes
8216     false once a handshake has been completed.
8217     (For servers, SSL_renegotiate() followed by SSL_do_handshake()
8218     sends a HelloRequest, but does not ensure that a handshake takes
8219     place.  SSL_renegotiate_pending() is useful for checking if the
8220     client has followed the request.)
8221     [Bodo Moeller]
8222
8223  *) New SSL option SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION.
8224     By default, clients may request session resumption even during
8225     renegotiation (if session ID contexts permit); with this option,
8226     session resumption is possible only in the first handshake.
8227
8228     SSL_OP_ALL is now 0x00000FFFL instead of 0x000FFFFFL.  This makes
8229     more bits available for options that should not be part of
8230     SSL_OP_ALL (such as SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION).
8231     [Bodo Moeller]
8232
8233  *) Add some demos for certificate and certificate request creation.
8234     [Steve Henson]
8235
8236  *) Make maximum certificate chain size accepted from the peer application
8237     settable (SSL*_get/set_max_cert_list()), as proposed by
8238     "Douglas E. Engert" <deengert@anl.gov>.
8239     [Lutz Jaenicke]
8240
8241  *) Add support for shared libraries for Unixware-7
8242     (Boyd Lynn Gerber <gerberb@zenez.com>).
8243     [Lutz Jaenicke]
8244
8245  *) Add a "destroy" handler to ENGINEs that allows structural cleanup to
8246     be done prior to destruction. Use this to unload error strings from
8247     ENGINEs that load their own error strings. NB: This adds two new API
8248     functions to "get" and "set" this destroy handler in an ENGINE.
8249     [Geoff Thorpe]
8250
8251  *) Alter all existing ENGINE implementations (except "openssl" and
8252     "openbsd") to dynamically instantiate their own error strings. This
8253     makes them more flexible to be built both as statically-linked ENGINEs
8254     and self-contained shared-libraries loadable via the "dynamic" ENGINE.
8255     Also, add stub code to each that makes building them as self-contained
8256     shared-libraries easier (see README.ENGINE).
8257     [Geoff Thorpe]
8258
8259  *) Add a "dynamic" ENGINE that provides a mechanism for binding ENGINE
8260     implementations into applications that are completely implemented in
8261     self-contained shared-libraries. The "dynamic" ENGINE exposes control
8262     commands that can be used to configure what shared-library to load and
8263     to control aspects of the way it is handled. Also, made an update to
8264     the README.ENGINE file that brings its information up-to-date and
8265     provides some information and instructions on the "dynamic" ENGINE
8266     (ie. how to use it, how to build "dynamic"-loadable ENGINEs, etc).
8267     [Geoff Thorpe]
8268
8269  *) Make it possible to unload ranges of ERR strings with a new
8270     "ERR_unload_strings" function.
8271     [Geoff Thorpe]
8272
8273  *) Add a copy() function to EVP_MD.
8274     [Ben Laurie]
8275
8276  *) Make EVP_MD routines take a context pointer instead of just the
8277     md_data void pointer.
8278     [Ben Laurie]
8279
8280  *) Add flags to EVP_MD and EVP_MD_CTX. EVP_MD_FLAG_ONESHOT indicates
8281     that the digest can only process a single chunk of data
8282     (typically because it is provided by a piece of
8283     hardware). EVP_MD_CTX_FLAG_ONESHOT indicates that the application
8284     is only going to provide a single chunk of data, and hence the
8285     framework needn't accumulate the data for oneshot drivers.
8286     [Ben Laurie]
8287
8288  *) As with "ERR", make it possible to replace the underlying "ex_data"
8289     functions. This change also alters the storage and management of global
8290     ex_data state - it's now all inside ex_data.c and all "class" code (eg.
8291     RSA, BIO, SSL_CTX, etc) no longer stores its own STACKS and per-class
8292     index counters. The API functions that use this state have been changed
8293     to take a "class_index" rather than pointers to the class's local STACK
8294     and counter, and there is now an API function to dynamically create new
8295     classes. This centralisation allows us to (a) plug a lot of the
8296     thread-safety problems that existed, and (b) makes it possible to clean
8297     up all allocated state using "CRYPTO_cleanup_all_ex_data()". W.r.t. (b)
8298     such data would previously have always leaked in application code and
8299     workarounds were in place to make the memory debugging turn a blind eye
8300     to it. Application code that doesn't use this new function will still
8301     leak as before, but their memory debugging output will announce it now
8302     rather than letting it slide.
8303
8304     Besides the addition of CRYPTO_cleanup_all_ex_data(), another API change
8305     induced by the "ex_data" overhaul is that X509_STORE_CTX_init() now
8306     has a return value to indicate success or failure.
8307     [Geoff Thorpe]
8308
8309  *) Make it possible to replace the underlying "ERR" functions such that the
8310     global state (2 LHASH tables and 2 locks) is only used by the "default"
8311     implementation. This change also adds two functions to "get" and "set"
8312     the implementation prior to it being automatically set the first time
8313     any other ERR function takes place. Ie. an application can call "get",
8314     pass the return value to a module it has just loaded, and that module
8315     can call its own "set" function using that value. This means the
8316     module's "ERR" operations will use (and modify) the error state in the
8317     application and not in its own statically linked copy of OpenSSL code.
8318     [Geoff Thorpe]
8319
8320  *) Give DH, DSA, and RSA types their own "**_up_ref()" function to increment
8321     reference counts. This performs normal REF_PRINT/REF_CHECK macros on
8322     the operation, and provides a more encapsulated way for external code
8323     (crypto/evp/ and ssl/) to do this. Also changed the evp and ssl code
8324     to use these functions rather than manually incrementing the counts.
8325
8326     Also rename "DSO_up()" function to more descriptive "DSO_up_ref()".
8327     [Geoff Thorpe]
8328
8329  *) Add EVP test program.
8330     [Ben Laurie]
8331
8332  *) Add symmetric cipher support to ENGINE. Expect the API to change!
8333     [Ben Laurie]
8334
8335  *) New CRL functions: X509_CRL_set_version(), X509_CRL_set_issuer_name()
8336     X509_CRL_set_lastUpdate(), X509_CRL_set_nextUpdate(), X509_CRL_sort(),
8337     X509_REVOKED_set_serialNumber(), and X509_REVOKED_set_revocationDate().
8338     These allow a CRL to be built without having to access X509_CRL fields
8339     directly. Modify 'ca' application to use new functions.
8340     [Steve Henson]
8341
8342  *) Move SSL_OP_TLS_ROLLBACK_BUG out of the SSL_OP_ALL list of recommended
8343     bug workarounds. Rollback attack detection is a security feature.
8344     The problem will only arise on OpenSSL servers when TLSv1 is not
8345     available (sslv3_server_method() or SSL_OP_NO_TLSv1).
8346     Software authors not wanting to support TLSv1 will have special reasons
8347     for their choice and can explicitly enable this option.
8348     [Bodo Moeller, Lutz Jaenicke]
8349
8350  *) Rationalise EVP so it can be extended: don't include a union of
8351     cipher/digest structures, add init/cleanup functions for EVP_MD_CTX
8352     (similar to those existing for EVP_CIPHER_CTX).
8353     Usage example:
8354
8355         EVP_MD_CTX md;
8356
8357         EVP_MD_CTX_init(&md);             /* new function call */
8358         EVP_DigestInit(&md, EVP_sha1());
8359         EVP_DigestUpdate(&md, in, len);
8360         EVP_DigestFinal(&md, out, NULL);
8361         EVP_MD_CTX_cleanup(&md);          /* new function call */
8362
8363     [Ben Laurie]
8364
8365  *) Make DES key schedule conform to the usual scheme, as well as
8366     correcting its structure. This means that calls to DES functions
8367     now have to pass a pointer to a des_key_schedule instead of a
8368     plain des_key_schedule (which was actually always a pointer
8369     anyway): E.g.,
8370
8371         des_key_schedule ks;
8372
8373         des_set_key_checked(..., &ks);
8374         des_ncbc_encrypt(..., &ks, ...);
8375
8376     (Note that a later change renames 'des_...' into 'DES_...'.)
8377     [Ben Laurie]
8378
8379  *) Initial reduction of linker bloat: the use of some functions, such as
8380     PEM causes large amounts of unused functions to be linked in due to
8381     poor organisation. For example pem_all.c contains every PEM function
8382     which has a knock on effect of linking in large amounts of (unused)
8383     ASN1 code. Grouping together similar functions and splitting unrelated
8384     functions prevents this.
8385     [Steve Henson]
8386
8387  *) Cleanup of EVP macros.
8388     [Ben Laurie]
8389
8390  *) Change historical references to {NID,SN,LN}_des_ede and ede3 to add the
8391     correct _ecb suffix.
8392     [Ben Laurie]
8393
8394  *) Add initial OCSP responder support to ocsp application. The
8395     revocation information is handled using the text based index
8396     use by the ca application. The responder can either handle
8397     requests generated internally, supplied in files (for example
8398     via a CGI script) or using an internal minimal server.
8399     [Steve Henson]
8400
8401  *) Add configuration choices to get zlib compression for TLS.
8402     [Richard Levitte]
8403
8404  *) Changes to Kerberos SSL for RFC 2712 compliance:
8405     1.  Implemented real KerberosWrapper, instead of just using
8406         KRB5 AP_REQ message.  [Thanks to Simon Wilkinson <sxw@sxw.org.uk>]
8407     2.  Implemented optional authenticator field of KerberosWrapper.
8408
8409     Added openssl-style ASN.1 macros for Kerberos ticket, ap_req,
8410     and authenticator structs; see crypto/krb5/.
8411
8412     Generalized Kerberos calls to support multiple Kerberos libraries.
8413     [Vern Staats <staatsvr@asc.hpc.mil>,
8414      Jeffrey Altman <jaltman@columbia.edu>
8415      via Richard Levitte]
8416
8417  *) Cause 'openssl speed' to use fully hard-coded DSA keys as it
8418     already does with RSA. testdsa.h now has 'priv_key/pub_key'
8419     values for each of the key sizes rather than having just
8420     parameters (and 'speed' generating keys each time).
8421     [Geoff Thorpe]
8422
8423  *) Speed up EVP routines.
8424     Before:
8425encrypt
8426type              8 bytes     64 bytes    256 bytes   1024 bytes   8192 bytes
8427des-cbc           4408.85k     5560.51k     5778.46k     5862.20k     5825.16k
8428des-cbc           4389.55k     5571.17k     5792.23k     5846.91k     5832.11k
8429des-cbc           4394.32k     5575.92k     5807.44k     5848.37k     5841.30k
8430decrypt
8431des-cbc           3482.66k     5069.49k     5496.39k     5614.16k     5639.28k
8432des-cbc           3480.74k     5068.76k     5510.34k     5609.87k     5635.52k
8433des-cbc           3483.72k     5067.62k     5504.60k     5708.01k     5724.80k
8434     After:
8435encrypt
8436des-cbc           4660.16k     5650.19k     5807.19k     5827.13k     5783.32k
8437decrypt
8438des-cbc           3624.96k     5258.21k     5530.91k     5624.30k     5628.26k
8439     [Ben Laurie]
8440
8441  *) Added the OS2-EMX target.
8442     ["Brian Havard" <brianh@kheldar.apana.org.au> and Richard Levitte]
8443
8444  *) Rewrite apps to use NCONF routines instead of the old CONF. New functions
8445     to support NCONF routines in extension code. New function CONF_set_nconf()
8446     to allow functions which take an NCONF to also handle the old LHASH
8447     structure: this means that the old CONF compatible routines can be
8448     retained (in particular wrt extensions) without having to duplicate the
8449     code. New function X509V3_add_ext_nconf_sk to add extensions to a stack.
8450     [Steve Henson]
8451
8452  *) Enhance the general user interface with mechanisms for inner control
8453     and with possibilities to have yes/no kind of prompts.
8454     [Richard Levitte]
8455
8456  *) Change all calls to low level digest routines in the library and
8457     applications to use EVP. Add missing calls to HMAC_cleanup() and
8458     don't assume HMAC_CTX can be copied using memcpy().
8459     [Verdon Walker <VWalker@novell.com>, Steve Henson]
8460
8461  *) Add the possibility to control engines through control names but with
8462     arbitrary arguments instead of just a string.
8463     Change the key loaders to take a UI_METHOD instead of a callback
8464     function pointer.  NOTE: this breaks binary compatibility with earlier
8465     versions of OpenSSL [engine].
8466     Adapt the nCipher code for these new conditions and add a card insertion
8467     callback.
8468     [Richard Levitte]
8469
8470  *) Enhance the general user interface with mechanisms to better support
8471     dialog box interfaces, application-defined prompts, the possibility
8472     to use defaults (for example default passwords from somewhere else)
8473     and interrupts/cancellations.
8474     [Richard Levitte]
8475
8476  *) Tidy up PKCS#12 attribute handling. Add support for the CSP name
8477     attribute in PKCS#12 files, add new -CSP option to pkcs12 utility.
8478     [Steve Henson]
8479
8480  *) Fix a memory leak in 'sk_dup()' in the case reallocation fails. (Also
8481     tidy up some unnecessarily weird code in 'sk_new()').
8482     [Geoff, reported by Diego Tartara <dtartara@novamens.com>]
8483
8484  *) Change the key loading routines for ENGINEs to use the same kind
8485     callback (pem_password_cb) as all other routines that need this
8486     kind of callback.
8487     [Richard Levitte]
8488
8489  *) Increase ENTROPY_NEEDED to 32 bytes, as Rijndael can operate with
8490     256 bit (=32 byte) keys. Of course seeding with more entropy bytes
8491     than this minimum value is recommended.
8492     [Lutz Jaenicke]
8493
8494  *) New random seeder for OpenVMS, using the system process statistics
8495     that are easily reachable.
8496     [Richard Levitte]
8497
8498  *) Windows apparently can't transparently handle global
8499     variables defined in DLLs. Initialisations such as:
8500
8501        const ASN1_ITEM *it = &ASN1_INTEGER_it;
8502
8503     won't compile. This is used by the any applications that need to
8504     declare their own ASN1 modules. This was fixed by adding the option
8505     EXPORT_VAR_AS_FN to all Win32 platforms, although this isn't strictly
8506     needed for static libraries under Win32.
8507     [Steve Henson]
8508
8509  *) New functions X509_PURPOSE_set() and X509_TRUST_set() to handle
8510     setting of purpose and trust fields. New X509_STORE trust and
8511     purpose functions and tidy up setting in other SSL functions.
8512     [Steve Henson]
8513
8514  *) Add copies of X509_STORE_CTX fields and callbacks to X509_STORE
8515     structure. These are inherited by X509_STORE_CTX when it is
8516     initialised. This allows various defaults to be set in the
8517     X509_STORE structure (such as flags for CRL checking and custom
8518     purpose or trust settings) for functions which only use X509_STORE_CTX
8519     internally such as S/MIME.
8520
8521     Modify X509_STORE_CTX_purpose_inherit() so it only sets purposes and
8522     trust settings if they are not set in X509_STORE. This allows X509_STORE
8523     purposes and trust (in S/MIME for example) to override any set by default.
8524
8525     Add command line options for CRL checking to smime, s_client and s_server
8526     applications.
8527     [Steve Henson]
8528
8529  *) Initial CRL based revocation checking. If the CRL checking flag(s)
8530     are set then the CRL is looked up in the X509_STORE structure and
8531     its validity and signature checked, then if the certificate is found
8532     in the CRL the verify fails with a revoked error.
8533
8534     Various new CRL related callbacks added to X509_STORE_CTX structure.
8535
8536     Command line options added to 'verify' application to support this.
8537
8538     This needs some additional work, such as being able to handle multiple
8539     CRLs with different times, extension based lookup (rather than just
8540     by subject name) and ultimately more complete V2 CRL extension
8541     handling.
8542     [Steve Henson]
8543
8544  *) Add a general user interface API (crypto/ui/).  This is designed
8545     to replace things like des_read_password and friends (backward
8546     compatibility functions using this new API are provided).
8547     The purpose is to remove prompting functions from the DES code
8548     section as well as provide for prompting through dialog boxes in
8549     a window system and the like.
8550     [Richard Levitte]
8551
8552  *) Add "ex_data" support to ENGINE so implementations can add state at a
8553     per-structure level rather than having to store it globally.
8554     [Geoff]
8555
8556  *) Make it possible for ENGINE structures to be copied when retrieved by
8557     ENGINE_by_id() if the ENGINE specifies a new flag: ENGINE_FLAGS_BY_ID_COPY.
8558     This causes the "original" ENGINE structure to act like a template,
8559     analogous to the RSA vs. RSA_METHOD type of separation. Because of this
8560     operational state can be localised to each ENGINE structure, despite the
8561     fact they all share the same "methods". New ENGINE structures returned in
8562     this case have no functional references and the return value is the single
8563     structural reference. This matches the single structural reference returned
8564     by ENGINE_by_id() normally, when it is incremented on the pre-existing
8565     ENGINE structure.
8566     [Geoff]
8567
8568  *) Fix ASN1 decoder when decoding type ANY and V_ASN1_OTHER: since this
8569     needs to match any other type at all we need to manually clear the
8570     tag cache.
8571     [Steve Henson]
8572
8573  *) Changes to the "openssl engine" utility to include;
8574     - verbosity levels ('-v', '-vv', and '-vvv') that provide information
8575       about an ENGINE's available control commands.
8576     - executing control commands from command line arguments using the
8577       '-pre' and '-post' switches. '-post' is only used if '-t' is
8578       specified and the ENGINE is successfully initialised. The syntax for
8579       the individual commands are colon-separated, for example;
8580         openssl engine chil -pre FORK_CHECK:0 -pre SO_PATH:/lib/test.so
8581     [Geoff]
8582
8583  *) New dynamic control command support for ENGINEs. ENGINEs can now
8584     declare their own commands (numbers), names (strings), descriptions,
8585     and input types for run-time discovery by calling applications. A
8586     subset of these commands are implicitly classed as "executable"
8587     depending on their input type, and only these can be invoked through
8588     the new string-based API function ENGINE_ctrl_cmd_string(). (Eg. this
8589     can be based on user input, config files, etc). The distinction is
8590     that "executable" commands cannot return anything other than a boolean
8591     result and can only support numeric or string input, whereas some
8592     discoverable commands may only be for direct use through
8593     ENGINE_ctrl(), eg. supporting the exchange of binary data, function
8594     pointers, or other custom uses. The "executable" commands are to
8595     support parameterisations of ENGINE behaviour that can be
8596     unambiguously defined by ENGINEs and used consistently across any
8597     OpenSSL-based application. Commands have been added to all the
8598     existing hardware-supporting ENGINEs, noticeably "SO_PATH" to allow
8599     control over shared-library paths without source code alterations.
8600     [Geoff]
8601
8602  *) Changed all ENGINE implementations to dynamically allocate their
8603     ENGINEs rather than declaring them statically. Apart from this being
8604     necessary with the removal of the ENGINE_FLAGS_MALLOCED distinction,
8605     this also allows the implementations to compile without using the
8606     internal engine_int.h header.
8607     [Geoff]
8608
8609  *) Minor adjustment to "rand" code. RAND_get_rand_method() now returns a
8610     'const' value. Any code that should be able to modify a RAND_METHOD
8611     should already have non-const pointers to it (ie. they should only
8612     modify their own ones).
8613     [Geoff]
8614
8615  *) Made a variety of little tweaks to the ENGINE code.
8616     - "atalla" and "ubsec" string definitions were moved from header files
8617       to C code. "nuron" string definitions were placed in variables
8618       rather than hard-coded - allowing parameterisation of these values
8619       later on via ctrl() commands.
8620     - Removed unused "#if 0"'d code.
8621     - Fixed engine list iteration code so it uses ENGINE_free() to release
8622       structural references.
8623     - Constified the RAND_METHOD element of ENGINE structures.
8624     - Constified various get/set functions as appropriate and added
8625       missing functions (including a catch-all ENGINE_cpy that duplicates
8626       all ENGINE values onto a new ENGINE except reference counts/state).
8627     - Removed NULL parameter checks in get/set functions. Setting a method
8628       or function to NULL is a way of cancelling out a previously set
8629       value.  Passing a NULL ENGINE parameter is just plain stupid anyway
8630       and doesn't justify the extra error symbols and code.
8631     - Deprecate the ENGINE_FLAGS_MALLOCED define and move the area for
8632       flags from engine_int.h to engine.h.
8633     - Changed prototypes for ENGINE handler functions (init(), finish(),
8634       ctrl(), key-load functions, etc) to take an (ENGINE*) parameter.
8635     [Geoff]
8636
8637  *) Implement binary inversion algorithm for BN_mod_inverse in addition
8638     to the algorithm using long division.  The binary algorithm can be
8639     used only if the modulus is odd.  On 32-bit systems, it is faster
8640     only for relatively small moduli (roughly 20-30% for 128-bit moduli,
8641     roughly 5-15% for 256-bit moduli), so we use it only for moduli
8642     up to 450 bits.  In 64-bit environments, the binary algorithm
8643     appears to be advantageous for much longer moduli; here we use it
8644     for moduli up to 2048 bits.
8645     [Bodo Moeller]
8646
8647  *) Rewrite CHOICE field setting in ASN1_item_ex_d2i(). The old code
8648     could not support the combine flag in choice fields.
8649     [Steve Henson]
8650
8651  *) Add a 'copy_extensions' option to the 'ca' utility. This copies
8652     extensions from a certificate request to the certificate.
8653     [Steve Henson]
8654
8655  *) Allow multiple 'certopt' and 'nameopt' options to be separated
8656     by commas. Add 'namopt' and 'certopt' options to the 'ca' config
8657     file: this allows the display of the certificate about to be
8658     signed to be customised, to allow certain fields to be included
8659     or excluded and extension details. The old system didn't display
8660     multicharacter strings properly, omitted fields not in the policy
8661     and couldn't display additional details such as extensions.
8662     [Steve Henson]
8663
8664  *) Function EC_POINTs_mul for multiple scalar multiplication
8665     of an arbitrary number of elliptic curve points
8666          \sum scalars[i]*points[i],
8667     optionally including the generator defined for the EC_GROUP:
8668          scalar*generator +  \sum scalars[i]*points[i].
8669
8670     EC_POINT_mul is a simple wrapper function for the typical case
8671     that the point list has just one item (besides the optional
8672     generator).
8673     [Bodo Moeller]
8674
8675  *) First EC_METHODs for curves over GF(p):
8676
8677     EC_GFp_simple_method() uses the basic BN_mod_mul and BN_mod_sqr
8678     operations and provides various method functions that can also
8679     operate with faster implementations of modular arithmetic.
8680
8681     EC_GFp_mont_method() reuses most functions that are part of
8682     EC_GFp_simple_method, but uses Montgomery arithmetic.
8683
8684     [Bodo Moeller; point addition and point doubling
8685     implementation directly derived from source code provided by
8686     Lenka Fibikova <fibikova@exp-math.uni-essen.de>]
8687
8688  *) Framework for elliptic curves (crypto/ec/ec.h, crypto/ec/ec_lcl.h,
8689     crypto/ec/ec_lib.c):
8690
8691     Curves are EC_GROUP objects (with an optional group generator)
8692     based on EC_METHODs that are built into the library.
8693
8694     Points are EC_POINT objects based on EC_GROUP objects.
8695
8696     Most of the framework would be able to handle curves over arbitrary
8697     finite fields, but as there are no obvious types for fields other
8698     than GF(p), some functions are limited to that for now.
8699     [Bodo Moeller]
8700
8701  *) Add the -HTTP option to s_server.  It is similar to -WWW, but requires
8702     that the file contains a complete HTTP response.
8703     [Richard Levitte]
8704
8705  *) Add the ec directory to mkdef.pl and mkfiles.pl. In mkdef.pl
8706     change the def and num file printf format specifier from "%-40sXXX"
8707     to "%-39s XXX". The latter will always guarantee a space after the
8708     field while the former will cause them to run together if the field
8709     is 40 of more characters long.
8710     [Steve Henson]
8711
8712  *) Constify the cipher and digest 'method' functions and structures
8713     and modify related functions to take constant EVP_MD and EVP_CIPHER
8714     pointers.
8715     [Steve Henson]
8716
8717  *) Hide BN_CTX structure details in bn_lcl.h instead of publishing them
8718     in <openssl/bn.h>.  Also further increase BN_CTX_NUM to 32.
8719     [Bodo Moeller]
8720
8721  *) Modify EVP_Digest*() routines so they now return values. Although the
8722     internal software routines can never fail additional hardware versions
8723     might.
8724     [Steve Henson]
8725
8726  *) Clean up crypto/err/err.h and change some error codes to avoid conflicts:
8727
8728     Previously ERR_R_FATAL was too small and coincided with ERR_LIB_PKCS7
8729     (= ERR_R_PKCS7_LIB); it is now 64 instead of 32.
8730
8731     ASN1 error codes
8732          ERR_R_NESTED_ASN1_ERROR
8733          ...
8734          ERR_R_MISSING_ASN1_EOS
8735     were 4 .. 9, conflicting with
8736          ERR_LIB_RSA (= ERR_R_RSA_LIB)
8737          ...
8738          ERR_LIB_PEM (= ERR_R_PEM_LIB).
8739     They are now 58 .. 63 (i.e., just below ERR_R_FATAL).
8740
8741     Add new error code 'ERR_R_INTERNAL_ERROR'.
8742     [Bodo Moeller]
8743
8744  *) Don't overuse locks in crypto/err/err.c: For data retrieval, CRYPTO_r_lock
8745     suffices.
8746     [Bodo Moeller]
8747
8748  *) New option '-subj arg' for 'openssl req' and 'openssl ca'.  This
8749     sets the subject name for a new request or supersedes the
8750     subject name in a given request. Formats that can be parsed are
8751          'CN=Some Name, OU=myOU, C=IT'
8752     and
8753          'CN=Some Name/OU=myOU/C=IT'.
8754
8755     Add options '-batch' and '-verbose' to 'openssl req'.
8756     [Massimiliano Pala <madwolf@hackmasters.net>]
8757
8758  *) Introduce the possibility to access global variables through
8759     functions on platform were that's the best way to handle exporting
8760     global variables in shared libraries.  To enable this functionality,
8761     one must configure with "EXPORT_VAR_AS_FN" or defined the C macro
8762     "OPENSSL_EXPORT_VAR_AS_FUNCTION" in crypto/opensslconf.h (the latter
8763     is normally done by Configure or something similar).
8764
8765     To implement a global variable, use the macro OPENSSL_IMPLEMENT_GLOBAL
8766     in the source file (foo.c) like this:
8767
8768        OPENSSL_IMPLEMENT_GLOBAL(int,foo)=1;
8769        OPENSSL_IMPLEMENT_GLOBAL(double,bar);
8770
8771     To declare a global variable, use the macros OPENSSL_DECLARE_GLOBAL
8772     and OPENSSL_GLOBAL_REF in the header file (foo.h) like this:
8773
8774        OPENSSL_DECLARE_GLOBAL(int,foo);
8775        #define foo OPENSSL_GLOBAL_REF(foo)
8776        OPENSSL_DECLARE_GLOBAL(double,bar);
8777        #define bar OPENSSL_GLOBAL_REF(bar)
8778
8779     The #defines are very important, and therefore so is including the
8780     header file everywhere where the defined globals are used.
8781
8782     The macro OPENSSL_EXPORT_VAR_AS_FUNCTION also affects the definition
8783     of ASN.1 items, but that structure is a bit different.
8784
8785     The largest change is in util/mkdef.pl which has been enhanced with
8786     better and easier to understand logic to choose which symbols should
8787     go into the Windows .def files as well as a number of fixes and code
8788     cleanup (among others, algorithm keywords are now sorted
8789     lexicographically to avoid constant rewrites).
8790     [Richard Levitte]
8791
8792  *) In BN_div() keep a copy of the sign of 'num' before writing the
8793     result to 'rm' because if rm==num the value will be overwritten
8794     and produce the wrong result if 'num' is negative: this caused
8795     problems with BN_mod() and BN_nnmod().
8796     [Steve Henson]
8797
8798  *) Function OCSP_request_verify(). This checks the signature on an
8799     OCSP request and verifies the signer certificate. The signer
8800     certificate is just checked for a generic purpose and OCSP request
8801     trust settings.
8802     [Steve Henson]
8803
8804  *) Add OCSP_check_validity() function to check the validity of OCSP
8805     responses. OCSP responses are prepared in real time and may only
8806     be a few seconds old. Simply checking that the current time lies
8807     between thisUpdate and nextUpdate max reject otherwise valid responses
8808     caused by either OCSP responder or client clock inaccuracy. Instead
8809     we allow thisUpdate and nextUpdate to fall within a certain period of
8810     the current time. The age of the response can also optionally be
8811     checked. Two new options -validity_period and -status_age added to
8812     ocsp utility.
8813     [Steve Henson]
8814
8815  *) If signature or public key algorithm is unrecognized print out its
8816     OID rather that just UNKNOWN.
8817     [Steve Henson]
8818
8819  *) Change OCSP_cert_to_id() to tolerate a NULL subject certificate and
8820     OCSP_cert_id_new() a NULL serialNumber. This allows a partial certificate
8821     ID to be generated from the issuer certificate alone which can then be
8822     passed to OCSP_id_issuer_cmp().
8823     [Steve Henson]
8824
8825  *) New compilation option ASN1_ITEM_FUNCTIONS. This causes the new
8826     ASN1 modules to export functions returning ASN1_ITEM pointers
8827     instead of the ASN1_ITEM structures themselves. This adds several
8828     new macros which allow the underlying ASN1 function/structure to
8829     be accessed transparently. As a result code should not use ASN1_ITEM
8830     references directly (such as &X509_it) but instead use the relevant
8831     macros (such as ASN1_ITEM_rptr(X509)). This option is to allow
8832     use of the new ASN1 code on platforms where exporting structures
8833     is problematical (for example in shared libraries) but exporting
8834     functions returning pointers to structures is not.
8835     [Steve Henson]
8836
8837  *) Add support for overriding the generation of SSL/TLS session IDs.
8838     These callbacks can be registered either in an SSL_CTX or per SSL.
8839     The purpose of this is to allow applications to control, if they wish,
8840     the arbitrary values chosen for use as session IDs, particularly as it
8841     can be useful for session caching in multiple-server environments. A
8842     command-line switch for testing this (and any client code that wishes
8843     to use such a feature) has been added to "s_server".
8844     [Geoff Thorpe, Lutz Jaenicke]
8845
8846  *) Modify mkdef.pl to recognise and parse preprocessor conditionals
8847     of the form '#if defined(...) || defined(...) || ...' and
8848     '#if !defined(...) && !defined(...) && ...'.  This also avoids
8849     the growing number of special cases it was previously handling.
8850     [Richard Levitte]
8851
8852  *) Make all configuration macros available for application by making
8853     sure they are available in opensslconf.h, by giving them names starting
8854     with "OPENSSL_" to avoid conflicts with other packages and by making
8855     sure e_os2.h will cover all platform-specific cases together with
8856     opensslconf.h.
8857     Additionally, it is now possible to define configuration/platform-
8858     specific names (called "system identities").  In the C code, these
8859     are prefixed with "OPENSSL_SYSNAME_".  e_os2.h will create another
8860     macro with the name beginning with "OPENSSL_SYS_", which is determined
8861     from "OPENSSL_SYSNAME_*" or compiler-specific macros depending on
8862     what is available.
8863     [Richard Levitte]
8864
8865  *) New option -set_serial to 'req' and 'x509' this allows the serial
8866     number to use to be specified on the command line. Previously self
8867     signed certificates were hard coded with serial number 0 and the
8868     CA options of 'x509' had to use a serial number in a file which was
8869     auto incremented.
8870     [Steve Henson]
8871
8872  *) New options to 'ca' utility to support V2 CRL entry extensions.
8873     Currently CRL reason, invalidity date and hold instruction are
8874     supported. Add new CRL extensions to V3 code and some new objects.
8875     [Steve Henson]
8876
8877  *) New function EVP_CIPHER_CTX_set_padding() this is used to
8878     disable standard block padding (aka PKCS#5 padding) in the EVP
8879     API, which was previously mandatory. This means that the data is
8880     not padded in any way and so the total length much be a multiple
8881     of the block size, otherwise an error occurs.
8882     [Steve Henson]
8883
8884  *) Initial (incomplete) OCSP SSL support.
8885     [Steve Henson]
8886
8887  *) New function OCSP_parse_url(). This splits up a URL into its host,
8888     port and path components: primarily to parse OCSP URLs. New -url
8889     option to ocsp utility.
8890     [Steve Henson]
8891
8892  *) New nonce behavior. The return value of OCSP_check_nonce() now
8893     reflects the various checks performed. Applications can decide
8894     whether to tolerate certain situations such as an absent nonce
8895     in a response when one was present in a request: the ocsp application
8896     just prints out a warning. New function OCSP_add1_basic_nonce()
8897     this is to allow responders to include a nonce in a response even if
8898     the request is nonce-less.
8899     [Steve Henson]
8900
8901  *) Disable stdin buffering in load_cert (apps/apps.c) so that no certs are
8902     skipped when using openssl x509 multiple times on a single input file,
8903     e.g. "(openssl x509 -out cert1; openssl x509 -out cert2) <certs".
8904     [Bodo Moeller]
8905
8906  *) Make ASN1_UTCTIME_set_string() and ASN1_GENERALIZEDTIME_set_string()
8907     set string type: to handle setting ASN1_TIME structures. Fix ca
8908     utility to correctly initialize revocation date of CRLs.
8909     [Steve Henson]
8910
8911  *) New option SSL_OP_CIPHER_SERVER_PREFERENCE allows the server to override
8912     the clients preferred ciphersuites and rather use its own preferences.
8913     Should help to work around M$ SGC (Server Gated Cryptography) bug in
8914     Internet Explorer by ensuring unchanged hash method during stepup.
8915     (Also replaces the broken/deactivated SSL_OP_NON_EXPORT_FIRST option.)
8916     [Lutz Jaenicke]
8917
8918  *) Make mkdef.pl recognise all DECLARE_ASN1 macros, change rijndael
8919     to aes and add a new 'exist' option to print out symbols that don't
8920     appear to exist.
8921     [Steve Henson]
8922
8923  *) Additional options to ocsp utility to allow flags to be set and
8924     additional certificates supplied.
8925     [Steve Henson]
8926
8927  *) Add the option -VAfile to 'openssl ocsp', so the user can give the
8928     OCSP client a number of certificate to only verify the response
8929     signature against.
8930     [Richard Levitte]
8931
8932  *) Update Rijndael code to version 3.0 and change EVP AES ciphers to
8933     handle the new API. Currently only ECB, CBC modes supported. Add new
8934     AES OIDs.
8935
8936     Add TLS AES ciphersuites as described in RFC3268, "Advanced
8937     Encryption Standard (AES) Ciphersuites for Transport Layer
8938     Security (TLS)".  (In beta versions of OpenSSL 0.9.7, these were
8939     not enabled by default and were not part of the "ALL" ciphersuite
8940     alias because they were not yet official; they could be
8941     explicitly requested by specifying the "AESdraft" ciphersuite
8942     group alias.  In the final release of OpenSSL 0.9.7, the group
8943     alias is called "AES" and is part of "ALL".)
8944     [Ben Laurie, Steve  Henson, Bodo Moeller]
8945
8946  *) New function OCSP_copy_nonce() to copy nonce value (if present) from
8947     request to response.
8948     [Steve Henson]
8949
8950  *) Functions for OCSP responders. OCSP_request_onereq_count(),
8951     OCSP_request_onereq_get0(), OCSP_onereq_get0_id() and OCSP_id_get0_info()
8952     extract information from a certificate request. OCSP_response_create()
8953     creates a response and optionally adds a basic response structure.
8954     OCSP_basic_add1_status() adds a complete single response to a basic
8955     response and returns the OCSP_SINGLERESP structure just added (to allow
8956     extensions to be included for example). OCSP_basic_add1_cert() adds a
8957     certificate to a basic response and OCSP_basic_sign() signs a basic
8958     response with various flags. New helper functions ASN1_TIME_check()
8959     (checks validity of ASN1_TIME structure) and ASN1_TIME_to_generalizedtime()
8960     (converts ASN1_TIME to GeneralizedTime).
8961     [Steve Henson]
8962
8963  *) Various new functions. EVP_Digest() combines EVP_Digest{Init,Update,Final}()
8964     in a single operation. X509_get0_pubkey_bitstr() extracts the public_key
8965     structure from a certificate. X509_pubkey_digest() digests the public_key
8966     contents: this is used in various key identifiers.
8967     [Steve Henson]
8968
8969  *) Make sk_sort() tolerate a NULL argument.
8970     [Steve Henson reported by Massimiliano Pala <madwolf@comune.modena.it>]
8971
8972  *) New OCSP verify flag OCSP_TRUSTOTHER. When set the "other" certificates
8973     passed by the function are trusted implicitly. If any of them signed the
8974     response then it is assumed to be valid and is not verified.
8975     [Steve Henson]
8976
8977  *) In PKCS7_set_type() initialise content_type in PKCS7_ENC_CONTENT
8978     to data. This was previously part of the PKCS7 ASN1 code. This
8979     was causing problems with OpenSSL created PKCS#12 and PKCS#7 structures.
8980     [Steve Henson, reported by Kenneth R. Robinette
8981                                <support@securenetterm.com>]
8982
8983  *) Add CRYPTO_push_info() and CRYPTO_pop_info() calls to new ASN1
8984     routines: without these tracing memory leaks is very painful.
8985     Fix leaks in PKCS12 and PKCS7 routines.
8986     [Steve Henson]
8987
8988  *) Make X509_time_adj() cope with the new behaviour of ASN1_TIME_new().
8989     Previously it initialised the 'type' argument to V_ASN1_UTCTIME which
8990     effectively meant GeneralizedTime would never be used. Now it
8991     is initialised to -1 but X509_time_adj() now has to check the value
8992     and use ASN1_TIME_set() if the value is not V_ASN1_UTCTIME or
8993     V_ASN1_GENERALIZEDTIME, without this it always uses GeneralizedTime.
8994     [Steve Henson, reported by Kenneth R. Robinette
8995                                <support@securenetterm.com>]
8996
8997  *) Fixes to BN_to_ASN1_INTEGER when bn is zero. This would previously
8998     result in a zero length in the ASN1_INTEGER structure which was
8999     not consistent with the structure when d2i_ASN1_INTEGER() was used
9000     and would cause ASN1_INTEGER_cmp() to fail. Enhance s2i_ASN1_INTEGER()
9001     to cope with hex and negative integers. Fix bug in i2a_ASN1_INTEGER()
9002     where it did not print out a minus for negative ASN1_INTEGER.
9003     [Steve Henson]
9004
9005  *) Add summary printout to ocsp utility. The various functions which
9006     convert status values to strings have been renamed to:
9007     OCSP_response_status_str(), OCSP_cert_status_str() and
9008     OCSP_crl_reason_str() and are no longer static. New options
9009     to verify nonce values and to disable verification. OCSP response
9010     printout format cleaned up.
9011     [Steve Henson]
9012
9013  *) Add additional OCSP certificate checks. These are those specified
9014     in RFC2560. This consists of two separate checks: the CA of the
9015     certificate being checked must either be the OCSP signer certificate
9016     or the issuer of the OCSP signer certificate. In the latter case the
9017     OCSP signer certificate must contain the OCSP signing extended key
9018     usage. This check is performed by attempting to match the OCSP
9019     signer or the OCSP signer CA to the issuerNameHash and issuerKeyHash
9020     in the OCSP_CERTID structures of the response.
9021     [Steve Henson]
9022
9023  *) Initial OCSP certificate verification added to OCSP_basic_verify()
9024     and related routines. This uses the standard OpenSSL certificate
9025     verify routines to perform initial checks (just CA validity) and
9026     to obtain the certificate chain. Then additional checks will be
9027     performed on the chain. Currently the root CA is checked to see
9028     if it is explicitly trusted for OCSP signing. This is used to set
9029     a root CA as a global signing root: that is any certificate that
9030     chains to that CA is an acceptable OCSP signing certificate.
9031     [Steve Henson]
9032
9033  *) New '-extfile ...' option to 'openssl ca' for reading X.509v3
9034     extensions from a separate configuration file.
9035     As when reading extensions from the main configuration file,
9036     the '-extensions ...' option may be used for specifying the
9037     section to use.
9038     [Massimiliano Pala <madwolf@comune.modena.it>]
9039
9040  *) New OCSP utility. Allows OCSP requests to be generated or
9041     read. The request can be sent to a responder and the output
9042     parsed, outputted or printed in text form. Not complete yet:
9043     still needs to check the OCSP response validity.
9044     [Steve Henson]
9045
9046  *) New subcommands for 'openssl ca':
9047     'openssl ca -status <serial>' prints the status of the cert with
9048     the given serial number (according to the index file).
9049     'openssl ca -updatedb' updates the expiry status of certificates
9050     in the index file.
9051     [Massimiliano Pala <madwolf@comune.modena.it>]
9052
9053  *) New '-newreq-nodes' command option to CA.pl.  This is like
9054     '-newreq', but calls 'openssl req' with the '-nodes' option
9055     so that the resulting key is not encrypted.
9056     [Damien Miller <djm@mindrot.org>]
9057
9058  *) New configuration for the GNU Hurd.
9059     [Jonathan Bartlett <johnnyb@wolfram.com> via Richard Levitte]
9060
9061  *) Initial code to implement OCSP basic response verify. This
9062     is currently incomplete. Currently just finds the signer's
9063     certificate and verifies the signature on the response.
9064     [Steve Henson]
9065
9066  *) New SSLeay_version code SSLEAY_DIR to determine the compiled-in
9067     value of OPENSSLDIR.  This is available via the new '-d' option
9068     to 'openssl version', and is also included in 'openssl version -a'.
9069     [Bodo Moeller]
9070
9071  *) Allowing defining memory allocation callbacks that will be given
9072     file name and line number information in additional arguments
9073     (a const char* and an int).  The basic functionality remains, as
9074     well as the original possibility to just replace malloc(),
9075     realloc() and free() by functions that do not know about these
9076     additional arguments.  To register and find out the current
9077     settings for extended allocation functions, the following
9078     functions are provided:
9079
9080        CRYPTO_set_mem_ex_functions
9081        CRYPTO_set_locked_mem_ex_functions
9082        CRYPTO_get_mem_ex_functions
9083        CRYPTO_get_locked_mem_ex_functions
9084
9085     These work the same way as CRYPTO_set_mem_functions and friends.
9086     CRYPTO_get_[locked_]mem_functions now writes 0 where such an
9087     extended allocation function is enabled.
9088     Similarly, CRYPTO_get_[locked_]mem_ex_functions writes 0 where
9089     a conventional allocation function is enabled.
9090     [Richard Levitte, Bodo Moeller]
9091
9092  *) Finish off removing the remaining LHASH function pointer casts.
9093     There should no longer be any prototype-casting required when using
9094     the LHASH abstraction, and any casts that remain are "bugs". See
9095     the callback types and macros at the head of lhash.h for details
9096     (and "OBJ_cleanup" in crypto/objects/obj_dat.c as an example).
9097     [Geoff Thorpe]
9098
9099  *) Add automatic query of EGD sockets in RAND_poll() for the unix variant.
9100     If /dev/[u]random devices are not available or do not return enough
9101     entropy, EGD style sockets (served by EGD or PRNGD) will automatically
9102     be queried.
9103     The locations /var/run/egd-pool, /dev/egd-pool, /etc/egd-pool, and
9104     /etc/entropy will be queried once each in this sequence, querying stops
9105     when enough entropy was collected without querying more sockets.
9106     [Lutz Jaenicke]
9107
9108  *) Change the Unix RAND_poll() variant to be able to poll several
9109     random devices, as specified by DEVRANDOM, until a sufficient amount
9110     of data has been collected.   We spend at most 10 ms on each file
9111     (select timeout) and read in non-blocking mode.  DEVRANDOM now
9112     defaults to the list "/dev/urandom", "/dev/random", "/dev/srandom"
9113     (previously it was just the string "/dev/urandom"), so on typical
9114     platforms the 10 ms delay will never occur.
9115     Also separate out the Unix variant to its own file, rand_unix.c.
9116     For VMS, there's a currently-empty rand_vms.c.
9117     [Richard Levitte]
9118
9119  *) Move OCSP client related routines to ocsp_cl.c. These
9120     provide utility functions which an application needing
9121     to issue a request to an OCSP responder and analyse the
9122     response will typically need: as opposed to those which an
9123     OCSP responder itself would need which will be added later.
9124
9125     OCSP_request_sign() signs an OCSP request with an API similar
9126     to PKCS7_sign(). OCSP_response_status() returns status of OCSP
9127     response. OCSP_response_get1_basic() extracts basic response
9128     from response. OCSP_resp_find_status(): finds and extracts status
9129     information from an OCSP_CERTID structure (which will be created
9130     when the request structure is built). These are built from lower
9131     level functions which work on OCSP_SINGLERESP structures but
9132     won't normally be used unless the application wishes to examine
9133     extensions in the OCSP response for example.
9134
9135     Replace nonce routines with a pair of functions.
9136     OCSP_request_add1_nonce() adds a nonce value and optionally
9137     generates a random value. OCSP_check_nonce() checks the
9138     validity of the nonce in an OCSP response.
9139     [Steve Henson]
9140
9141  *) Change function OCSP_request_add() to OCSP_request_add0_id().
9142     This doesn't copy the supplied OCSP_CERTID and avoids the
9143     need to free up the newly created id. Change return type
9144     to OCSP_ONEREQ to return the internal OCSP_ONEREQ structure.
9145     This can then be used to add extensions to the request.
9146     Deleted OCSP_request_new(), since most of its functionality
9147     is now in OCSP_REQUEST_new() (and the case insensitive name
9148     clash) apart from the ability to set the request name which
9149     will be added elsewhere.
9150     [Steve Henson]
9151
9152  *) Update OCSP API. Remove obsolete extensions argument from
9153     various functions. Extensions are now handled using the new
9154     OCSP extension code. New simple OCSP HTTP function which
9155     can be used to send requests and parse the response.
9156     [Steve Henson]
9157
9158  *) Fix the PKCS#7 (S/MIME) code to work with new ASN1. Two new
9159     ASN1_ITEM structures help with sign and verify. PKCS7_ATTR_SIGN
9160     uses the special reorder version of SET OF to sort the attributes
9161     and reorder them to match the encoded order. This resolves a long
9162     standing problem: a verify on a PKCS7 structure just after signing
9163     it used to fail because the attribute order did not match the
9164     encoded order. PKCS7_ATTR_VERIFY does not reorder the attributes:
9165     it uses the received order. This is necessary to tolerate some broken
9166     software that does not order SET OF. This is handled by encoding
9167     as a SEQUENCE OF but using implicit tagging (with UNIVERSAL class)
9168     to produce the required SET OF.
9169     [Steve Henson]
9170
9171  *) Have mk1mf.pl generate the macros OPENSSL_BUILD_SHLIBCRYPTO and
9172     OPENSSL_BUILD_SHLIBSSL and use them appropriately in the header
9173     files to get correct declarations of the ASN.1 item variables.
9174     [Richard Levitte]
9175
9176  *) Rewrite of PKCS#12 code to use new ASN1 functionality. Replace many
9177     PKCS#12 macros with real functions. Fix two unrelated ASN1 bugs:
9178     asn1_check_tlen() would sometimes attempt to use 'ctx' when it was
9179     NULL and ASN1_TYPE was not dereferenced properly in asn1_ex_c2i().
9180     New ASN1 macro: DECLARE_ASN1_ITEM() which just declares the relevant
9181     ASN1_ITEM and no wrapper functions.
9182     [Steve Henson]
9183
9184  *) New functions or ASN1_item_d2i_fp() and ASN1_item_d2i_bio(). These
9185     replace the old function pointer based I/O routines. Change most of
9186     the *_d2i_bio() and *_d2i_fp() functions to use these.
9187     [Steve Henson]
9188
9189  *) Enhance mkdef.pl to be more accepting about spacing in C preprocessor
9190     lines, recognize more "algorithms" that can be deselected, and make
9191     it complain about algorithm deselection that isn't recognised.
9192     [Richard Levitte]
9193
9194  *) New ASN1 functions to handle dup, sign, verify, digest, pack and
9195     unpack operations in terms of ASN1_ITEM. Modify existing wrappers
9196     to use new functions. Add NO_ASN1_OLD which can be set to remove
9197     some old style ASN1 functions: this can be used to determine if old
9198     code will still work when these eventually go away.
9199     [Steve Henson]
9200
9201  *) New extension functions for OCSP structures, these follow the
9202     same conventions as certificates and CRLs.
9203     [Steve Henson]
9204
9205  *) New function X509V3_add1_i2d(). This automatically encodes and
9206     adds an extension. Its behaviour can be customised with various
9207     flags to append, replace or delete. Various wrappers added for
9208     certificates and CRLs.
9209     [Steve Henson]
9210
9211  *) Fix to avoid calling the underlying ASN1 print routine when
9212     an extension cannot be parsed. Correct a typo in the
9213     OCSP_SERVICELOC extension. Tidy up print OCSP format.
9214     [Steve Henson]
9215
9216  *) Make mkdef.pl parse some of the ASN1 macros and add appropriate
9217     entries for variables.
9218     [Steve Henson]
9219
9220  *) Add functionality to apps/openssl.c for detecting locking
9221     problems: As the program is single-threaded, all we have
9222     to do is register a locking callback using an array for
9223     storing which locks are currently held by the program.
9224     [Bodo Moeller]
9225
9226  *) Use a lock around the call to CRYPTO_get_ex_new_index() in
9227     SSL_get_ex_data_X509_STORE_idx(), which is used in
9228     ssl_verify_cert_chain() and thus can be called at any time
9229     during TLS/SSL handshakes so that thread-safety is essential.
9230     Unfortunately, the ex_data design is not at all suited
9231     for multi-threaded use, so it probably should be abolished.
9232     [Bodo Moeller]
9233
9234  *) Added Broadcom "ubsec" ENGINE to OpenSSL.
9235     [Broadcom, tweaked and integrated by Geoff Thorpe]
9236
9237  *) Move common extension printing code to new function
9238     X509V3_print_extensions(). Reorganise OCSP print routines and
9239     implement some needed OCSP ASN1 functions. Add OCSP extensions.
9240     [Steve Henson]
9241
9242  *) New function X509_signature_print() to remove duplication in some
9243     print routines.
9244     [Steve Henson]
9245
9246  *) Add a special meaning when SET OF and SEQUENCE OF flags are both
9247     set (this was treated exactly the same as SET OF previously). This
9248     is used to reorder the STACK representing the structure to match the
9249     encoding. This will be used to get round a problem where a PKCS7
9250     structure which was signed could not be verified because the STACK
9251     order did not reflect the encoded order.
9252     [Steve Henson]
9253
9254  *) Reimplement the OCSP ASN1 module using the new code.
9255     [Steve Henson]
9256
9257  *) Update the X509V3 code to permit the use of an ASN1_ITEM structure
9258     for its ASN1 operations. The old style function pointers still exist
9259     for now but they will eventually go away.
9260     [Steve Henson]
9261
9262  *) Merge in replacement ASN1 code from the ASN1 branch. This almost
9263     completely replaces the old ASN1 functionality with a table driven
9264     encoder and decoder which interprets an ASN1_ITEM structure describing
9265     the ASN1 module. Compatibility with the existing ASN1 API (i2d,d2i) is
9266     largely maintained. Almost all of the old asn1_mac.h macro based ASN1
9267     has also been converted to the new form.
9268     [Steve Henson]
9269
9270  *) Change BN_mod_exp_recp so that negative moduli are tolerated
9271     (the sign is ignored).  Similarly, ignore the sign in BN_MONT_CTX_set
9272     so that BN_mod_exp_mont and BN_mod_exp_mont_word work
9273     for negative moduli.
9274     [Bodo Moeller]
9275
9276  *) Fix BN_uadd and BN_usub: Always return non-negative results instead
9277     of not touching the result's sign bit.
9278     [Bodo Moeller]
9279
9280  *) BN_div bugfix: If the result is 0, the sign (res->neg) must not be
9281     set.
9282     [Bodo Moeller]
9283
9284  *) Changed the LHASH code to use prototypes for callbacks, and created
9285     macros to declare and implement thin (optionally static) functions
9286     that provide type-safety and avoid function pointer casting for the
9287     type-specific callbacks.
9288     [Geoff Thorpe]
9289
9290  *) Added Kerberos Cipher Suites to be used with TLS, as written in
9291     RFC 2712.
9292     [Veers Staats <staatsvr@asc.hpc.mil>,
9293      Jeffrey Altman <jaltman@columbia.edu>, via Richard Levitte]
9294
9295  *) Reformat the FAQ so the different questions and answers can be divided
9296     in sections depending on the subject.
9297     [Richard Levitte]
9298
9299  *) Have the zlib compression code load ZLIB.DLL dynamically under
9300     Windows.
9301     [Richard Levitte]
9302
9303  *) New function BN_mod_sqrt for computing square roots modulo a prime
9304     (using the probabilistic Tonelli-Shanks algorithm unless
9305     p == 3 (mod 4)  or  p == 5 (mod 8),  which are cases that can
9306     be handled deterministically).
9307     [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller]
9308
9309  *) Make BN_mod_inverse faster by explicitly handling small quotients
9310     in the Euclid loop. (Speed gain about 20% for small moduli [256 or
9311     512 bits], about 30% for larger ones [1024 or 2048 bits].)
9312     [Bodo Moeller]
9313
9314  *) New function BN_kronecker.
9315     [Bodo Moeller]
9316
9317  *) Fix BN_gcd so that it works on negative inputs; the result is
9318     positive unless both parameters are zero.
9319     Previously something reasonably close to an infinite loop was
9320     possible because numbers could be growing instead of shrinking
9321     in the implementation of Euclid's algorithm.
9322     [Bodo Moeller]
9323
9324  *) Fix BN_is_word() and BN_is_one() macros to take into account the
9325     sign of the number in question.
9326
9327     Fix BN_is_word(a,w) to work correctly for w == 0.
9328
9329     The old BN_is_word(a,w) macro is now called BN_abs_is_word(a,w)
9330     because its test if the absolute value of 'a' equals 'w'.
9331     Note that BN_abs_is_word does *not* handle w == 0 reliably;
9332     it exists mostly for use in the implementations of BN_is_zero(),
9333     BN_is_one(), and BN_is_word().
9334     [Bodo Moeller]
9335
9336  *) New function BN_swap.
9337     [Bodo Moeller]
9338
9339  *) Use BN_nnmod instead of BN_mod in crypto/bn/bn_exp.c so that
9340     the exponentiation functions are more likely to produce reasonable
9341     results on negative inputs.
9342     [Bodo Moeller]
9343
9344  *) Change BN_mod_mul so that the result is always non-negative.
9345     Previously, it could be negative if one of the factors was negative;
9346     I don't think anyone really wanted that behaviour.
9347     [Bodo Moeller]
9348
9349  *) Move BN_mod_... functions into new file crypto/bn/bn_mod.c
9350     (except for exponentiation, which stays in crypto/bn/bn_exp.c,
9351     and BN_mod_mul_reciprocal, which stays in crypto/bn/bn_recp.c)
9352     and add new functions:
9353
9354          BN_nnmod
9355          BN_mod_sqr
9356          BN_mod_add
9357          BN_mod_add_quick
9358          BN_mod_sub
9359          BN_mod_sub_quick
9360          BN_mod_lshift1
9361          BN_mod_lshift1_quick
9362          BN_mod_lshift
9363          BN_mod_lshift_quick
9364
9365     These functions always generate non-negative results.
9366
9367     BN_nnmod otherwise is like BN_mod (if BN_mod computes a remainder  r
9368     such that  |m| < r < 0,  BN_nnmod will output  rem + |m|  instead).
9369
9370     BN_mod_XXX_quick(r, a, [b,] m) generates the same result as
9371     BN_mod_XXX(r, a, [b,] m, ctx), but requires that  a  [and  b]
9372     be reduced modulo  m.
9373     [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller]
9374
9375#if 0
9376     The following entry accidentally appeared in the CHANGES file
9377     distributed with OpenSSL 0.9.7.  The modifications described in
9378     it do *not* apply to OpenSSL 0.9.7.
9379
9380  *) Remove a few calls to bn_wexpand() in BN_sqr() (the one in there
9381     was actually never needed) and in BN_mul().  The removal in BN_mul()
9382     required a small change in bn_mul_part_recursive() and the addition
9383     of the functions bn_cmp_part_words(), bn_sub_part_words() and
9384     bn_add_part_words(), which do the same thing as bn_cmp_words(),
9385     bn_sub_words() and bn_add_words() except they take arrays with
9386     differing sizes.
9387     [Richard Levitte]
9388#endif
9389
9390  *) In 'openssl passwd', verify passwords read from the terminal
9391     unless the '-salt' option is used (which usually means that
9392     verification would just waste user's time since the resulting
9393     hash is going to be compared with some given password hash)
9394     or the new '-noverify' option is used.
9395
9396     This is an incompatible change, but it does not affect
9397     non-interactive use of 'openssl passwd' (passwords on the command
9398     line, '-stdin' option, '-in ...' option) and thus should not
9399     cause any problems.
9400     [Bodo Moeller]
9401
9402  *) Remove all references to RSAref, since there's no more need for it.
9403     [Richard Levitte]
9404
9405  *) Make DSO load along a path given through an environment variable
9406     (SHLIB_PATH) with shl_load().
9407     [Richard Levitte]
9408
9409  *) Constify the ENGINE code as a result of BIGNUM constification.
9410     Also constify the RSA code and most things related to it.  In a
9411     few places, most notable in the depth of the ASN.1 code, ugly
9412     casts back to non-const were required (to be solved at a later
9413     time)
9414     [Richard Levitte]
9415
9416  *) Make it so the openssl application has all engines loaded by default.
9417     [Richard Levitte]
9418
9419  *) Constify the BIGNUM routines a little more.
9420     [Richard Levitte]
9421
9422  *) Add the following functions:
9423
9424        ENGINE_load_cswift()
9425        ENGINE_load_chil()
9426        ENGINE_load_atalla()
9427        ENGINE_load_nuron()
9428        ENGINE_load_builtin_engines()
9429
9430     That way, an application can itself choose if external engines that
9431     are built-in in OpenSSL shall ever be used or not.  The benefit is
9432     that applications won't have to be linked with libdl or other dso
9433     libraries unless it's really needed.
9434
9435     Changed 'openssl engine' to load all engines on demand.
9436     Changed the engine header files to avoid the duplication of some
9437     declarations (they differed!).
9438     [Richard Levitte]
9439
9440  *) 'openssl engine' can now list capabilities.
9441     [Richard Levitte]
9442
9443  *) Better error reporting in 'openssl engine'.
9444     [Richard Levitte]
9445
9446  *) Never call load_dh_param(NULL) in s_server.
9447     [Bodo Moeller]
9448
9449  *) Add engine application.  It can currently list engines by name and
9450     identity, and test if they are actually available.
9451     [Richard Levitte]
9452
9453  *) Improve RPM specification file by forcing symbolic linking and making
9454     sure the installed documentation is also owned by root.root.
9455     [Damien Miller <djm@mindrot.org>]
9456
9457  *) Give the OpenSSL applications more possibilities to make use of
9458     keys (public as well as private) handled by engines.
9459     [Richard Levitte]
9460
9461  *) Add OCSP code that comes from CertCo.
9462     [Richard Levitte]
9463
9464  *) Add VMS support for the Rijndael code.
9465     [Richard Levitte]
9466
9467  *) Added untested support for Nuron crypto accelerator.
9468     [Ben Laurie]
9469
9470  *) Add support for external cryptographic devices.  This code was
9471     previously distributed separately as the "engine" branch.
9472     [Geoff Thorpe, Richard Levitte]
9473
9474  *) Rework the filename-translation in the DSO code. It is now possible to
9475     have far greater control over how a "name" is turned into a filename
9476     depending on the operating environment and any oddities about the
9477     different shared library filenames on each system.
9478     [Geoff Thorpe]
9479
9480  *) Support threads on FreeBSD-elf in Configure.
9481     [Richard Levitte]
9482
9483  *) Fix for SHA1 assembly problem with MASM: it produces
9484     warnings about corrupt line number information when assembling
9485     with debugging information. This is caused by the overlapping
9486     of two sections.
9487     [Bernd Matthes <mainbug@celocom.de>, Steve Henson]
9488
9489  *) NCONF changes.
9490     NCONF_get_number() has no error checking at all.  As a replacement,
9491     NCONF_get_number_e() is defined (_e for "error checking") and is
9492     promoted strongly.  The old NCONF_get_number is kept around for
9493     binary backward compatibility.
9494     Make it possible for methods to load from something other than a BIO,
9495     by providing a function pointer that is given a name instead of a BIO.
9496     For example, this could be used to load configuration data from an
9497     LDAP server.
9498     [Richard Levitte]
9499
9500  *) Fix for non blocking accept BIOs. Added new I/O special reason
9501     BIO_RR_ACCEPT to cover this case. Previously use of accept BIOs
9502     with non blocking I/O was not possible because no retry code was
9503     implemented. Also added new SSL code SSL_WANT_ACCEPT to cover
9504     this case.
9505     [Steve Henson]
9506
9507  *) Added the beginnings of Rijndael support.
9508     [Ben Laurie]
9509
9510  *) Fix for bug in DirectoryString mask setting. Add support for
9511     X509_NAME_print_ex() in 'req' and X509_print_ex() function
9512     to allow certificate printing to more controllable, additional
9513     'certopt' option to 'x509' to allow new printing options to be
9514     set.
9515     [Steve Henson]
9516
9517  *) Clean old EAY MD5 hack from e_os.h.
9518     [Richard Levitte]
9519
9520 Changes between 0.9.6l and 0.9.6m  [17 Mar 2004]
9521
9522  *) Fix null-pointer assignment in do_change_cipher_spec() revealed
9523     by using the Codenomicon TLS Test Tool (CVE-2004-0079)
9524     [Joe Orton, Steve Henson]
9525
9526 Changes between 0.9.6k and 0.9.6l  [04 Nov 2003]
9527
9528  *) Fix additional bug revealed by the NISCC test suite:
9529
9530     Stop bug triggering large recursion when presented with
9531     certain ASN.1 tags (CVE-2003-0851)
9532     [Steve Henson]
9533
9534 Changes between 0.9.6j and 0.9.6k  [30 Sep 2003]
9535
9536  *) Fix various bugs revealed by running the NISCC test suite:
9537
9538     Stop out of bounds reads in the ASN1 code when presented with
9539     invalid tags (CVE-2003-0543 and CVE-2003-0544).
9540
9541     If verify callback ignores invalid public key errors don't try to check
9542     certificate signature with the NULL public key.
9543
9544     [Steve Henson]
9545
9546  *) In ssl3_accept() (ssl/s3_srvr.c) only accept a client certificate
9547     if the server requested one: as stated in TLS 1.0 and SSL 3.0
9548     specifications.
9549     [Steve Henson]
9550
9551  *) In ssl3_get_client_hello() (ssl/s3_srvr.c), tolerate additional
9552     extra data after the compression methods not only for TLS 1.0
9553     but also for SSL 3.0 (as required by the specification).
9554     [Bodo Moeller; problem pointed out by Matthias Loepfe]
9555
9556  *) Change X509_certificate_type() to mark the key as exported/exportable
9557     when it's 512 *bits* long, not 512 bytes.
9558     [Richard Levitte]
9559
9560 Changes between 0.9.6i and 0.9.6j  [10 Apr 2003]
9561
9562  *) Countermeasure against the Klima-Pokorny-Rosa extension of
9563     Bleichbacher's attack on PKCS #1 v1.5 padding: treat
9564     a protocol version number mismatch like a decryption error
9565     in ssl3_get_client_key_exchange (ssl/s3_srvr.c).
9566     [Bodo Moeller]
9567
9568  *) Turn on RSA blinding by default in the default implementation
9569     to avoid a timing attack. Applications that don't want it can call
9570     RSA_blinding_off() or use the new flag RSA_FLAG_NO_BLINDING.
9571     They would be ill-advised to do so in most cases.
9572     [Ben Laurie, Steve Henson, Geoff Thorpe, Bodo Moeller]
9573
9574  *) Change RSA blinding code so that it works when the PRNG is not
9575     seeded (in this case, the secret RSA exponent is abused as
9576     an unpredictable seed -- if it is not unpredictable, there
9577     is no point in blinding anyway).  Make RSA blinding thread-safe
9578     by remembering the creator's thread ID in rsa->blinding and
9579     having all other threads use local one-time blinding factors
9580     (this requires more computation than sharing rsa->blinding, but
9581     avoids excessive locking; and if an RSA object is not shared
9582     between threads, blinding will still be very fast).
9583     [Bodo Moeller]
9584
9585 Changes between 0.9.6h and 0.9.6i  [19 Feb 2003]
9586
9587  *) In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked
9588     via timing by performing a MAC computation even if incorrect
9589     block cipher padding has been found.  This is a countermeasure
9590     against active attacks where the attacker has to distinguish
9591     between bad padding and a MAC verification error. (CVE-2003-0078)
9592
9593     [Bodo Moeller; problem pointed out by Brice Canvel (EPFL),
9594     Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and
9595     Martin Vuagnoux (EPFL, Ilion)]
9596
9597 Changes between 0.9.6g and 0.9.6h  [5 Dec 2002]
9598
9599  *) New function OPENSSL_cleanse(), which is used to cleanse a section of
9600     memory from its contents.  This is done with a counter that will
9601     place alternating values in each byte.  This can be used to solve
9602     two issues: 1) the removal of calls to memset() by highly optimizing
9603     compilers, and 2) cleansing with other values than 0, since those can
9604     be read through on certain media, for example a swap space on disk.
9605     [Geoff Thorpe]
9606
9607  *) Bugfix: client side session caching did not work with external caching,
9608     because the session->cipher setting was not restored when reloading
9609     from the external cache. This problem was masked, when
9610     SSL_OP_NETSCAPE_REUSE_CIPHER_CHANGE_BUG (part of SSL_OP_ALL) was set.
9611     (Found by Steve Haslam <steve@araqnid.ddts.net>.)
9612     [Lutz Jaenicke]
9613
9614  *) Fix client_certificate (ssl/s2_clnt.c): The permissible total
9615     length of the REQUEST-CERTIFICATE message is 18 .. 34, not 17 .. 33.
9616     [Zeev Lieber <zeev-l@yahoo.com>]
9617
9618  *) Undo an undocumented change introduced in 0.9.6e which caused
9619     repeated calls to OpenSSL_add_all_ciphers() and
9620     OpenSSL_add_all_digests() to be ignored, even after calling
9621     EVP_cleanup().
9622     [Richard Levitte]
9623
9624  *) Change the default configuration reader to deal with last line not
9625     being properly terminated.
9626     [Richard Levitte]
9627
9628  *) Change X509_NAME_cmp() so it applies the special rules on handling
9629     DN values that are of type PrintableString, as well as RDNs of type
9630     emailAddress where the value has the type ia5String.
9631     [stefank@valicert.com via Richard Levitte]
9632
9633  *) Add a SSL_SESS_CACHE_NO_INTERNAL_STORE flag to take over half
9634     the job SSL_SESS_CACHE_NO_INTERNAL_LOOKUP was inconsistently
9635     doing, define a new flag (SSL_SESS_CACHE_NO_INTERNAL) to be
9636     the bitwise-OR of the two for use by the majority of applications
9637     wanting this behaviour, and update the docs. The documented
9638     behaviour and actual behaviour were inconsistent and had been
9639     changing anyway, so this is more a bug-fix than a behavioural
9640     change.
9641     [Geoff Thorpe, diagnosed by Nadav Har'El]
9642
9643  *) Don't impose a 16-byte length minimum on session IDs in ssl/s3_clnt.c
9644     (the SSL 3.0 and TLS 1.0 specifications allow any length up to 32 bytes).
9645     [Bodo Moeller]
9646
9647  *) Fix initialization code race conditions in
9648        SSLv23_method(),  SSLv23_client_method(),   SSLv23_server_method(),
9649        SSLv2_method(),   SSLv2_client_method(),    SSLv2_server_method(),
9650        SSLv3_method(),   SSLv3_client_method(),    SSLv3_server_method(),
9651        TLSv1_method(),   TLSv1_client_method(),    TLSv1_server_method(),
9652        ssl2_get_cipher_by_char(),
9653        ssl3_get_cipher_by_char().
9654     [Patrick McCormick <patrick@tellme.com>, Bodo Moeller]
9655
9656  *) Reorder cleanup sequence in SSL_CTX_free(): only remove the ex_data after
9657     the cached sessions are flushed, as the remove_cb() might use ex_data
9658     contents. Bug found by Sam Varshavchik <mrsam@courier-mta.com>
9659     (see [openssl.org #212]).
9660     [Geoff Thorpe, Lutz Jaenicke]
9661
9662  *) Fix typo in OBJ_txt2obj which incorrectly passed the content
9663     length, instead of the encoding length to d2i_ASN1_OBJECT.
9664     [Steve Henson]
9665
9666 Changes between 0.9.6f and 0.9.6g  [9 Aug 2002]
9667
9668  *) [In 0.9.6g-engine release:]
9669     Fix crypto/engine/vendor_defns/cswift.h for WIN32 (use '_stdcall').
9670     [Lynn Gazis <lgazis@rainbow.com>]
9671
9672 Changes between 0.9.6e and 0.9.6f  [8 Aug 2002]
9673
9674  *) Fix ASN1 checks. Check for overflow by comparing with LONG_MAX
9675     and get fix the header length calculation.
9676     [Florian Weimer <Weimer@CERT.Uni-Stuttgart.DE>,
9677        Alon Kantor <alonk@checkpoint.com> (and others),
9678        Steve Henson]
9679
9680  *) Use proper error handling instead of 'assertions' in buffer
9681     overflow checks added in 0.9.6e.  This prevents DoS (the
9682     assertions could call abort()).
9683     [Arne Ansper <arne@ats.cyber.ee>, Bodo Moeller]
9684
9685 Changes between 0.9.6d and 0.9.6e  [30 Jul 2002]
9686
9687  *) Add various sanity checks to asn1_get_length() to reject
9688     the ASN1 length bytes if they exceed sizeof(long), will appear
9689     negative or the content length exceeds the length of the
9690     supplied buffer.
9691     [Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>]
9692
9693  *) Fix cipher selection routines: ciphers without encryption had no flags
9694     for the cipher strength set and where therefore not handled correctly
9695     by the selection routines (PR #130).
9696     [Lutz Jaenicke]
9697
9698  *) Fix EVP_dsa_sha macro.
9699     [Nils Larsch]
9700
9701  *) New option
9702          SSL_OP_DONT_INSERT_EMPTY_FRAGMENTS
9703     for disabling the SSL 3.0/TLS 1.0 CBC vulnerability countermeasure
9704     that was added in OpenSSL 0.9.6d.
9705
9706     As the countermeasure turned out to be incompatible with some
9707     broken SSL implementations, the new option is part of SSL_OP_ALL.
9708     SSL_OP_ALL is usually employed when compatibility with weird SSL
9709     implementations is desired (e.g. '-bugs' option to 's_client' and
9710     's_server'), so the new option is automatically set in many
9711     applications.
9712     [Bodo Moeller]
9713
9714  *) Changes in security patch:
9715
9716     Changes marked "(CHATS)" were sponsored by the Defense Advanced
9717     Research Projects Agency (DARPA) and Air Force Research Laboratory,
9718     Air Force Materiel Command, USAF, under agreement number
9719     F30602-01-2-0537.
9720
9721  *) Add various sanity checks to asn1_get_length() to reject
9722     the ASN1 length bytes if they exceed sizeof(long), will appear
9723     negative or the content length exceeds the length of the
9724     supplied buffer. (CVE-2002-0659)
9725     [Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>]
9726
9727  *) Assertions for various potential buffer overflows, not known to
9728     happen in practice.
9729     [Ben Laurie (CHATS)]
9730
9731  *) Various temporary buffers to hold ASCII versions of integers were
9732     too small for 64 bit platforms. (CVE-2002-0655)
9733     [Matthew Byng-Maddick <mbm@aldigital.co.uk> and Ben Laurie (CHATS)>
9734
9735  *) Remote buffer overflow in SSL3 protocol - an attacker could
9736     supply an oversized session ID to a client. (CVE-2002-0656)
9737     [Ben Laurie (CHATS)]
9738
9739  *) Remote buffer overflow in SSL2 protocol - an attacker could
9740     supply an oversized client master key. (CVE-2002-0656)
9741     [Ben Laurie (CHATS)]
9742
9743 Changes between 0.9.6c and 0.9.6d  [9 May 2002]
9744
9745  *) Fix crypto/asn1/a_sign.c so that 'parameters' is omitted (not
9746     encoded as NULL) with id-dsa-with-sha1.
9747     [Nils Larsch <nla@trustcenter.de>; problem pointed out by Bodo Moeller]
9748
9749  *) Check various X509_...() return values in apps/req.c.
9750     [Nils Larsch <nla@trustcenter.de>]
9751
9752  *) Fix BASE64 decode (EVP_DecodeUpdate) for data with CR/LF ended lines:
9753     an end-of-file condition would erroneously be flagged, when the CRLF
9754     was just at the end of a processed block. The bug was discovered when
9755     processing data through a buffering memory BIO handing the data to a
9756     BASE64-decoding BIO. Bug fund and patch submitted by Pavel Tsekov
9757     <ptsekov@syntrex.com> and Nedelcho Stanev.
9758     [Lutz Jaenicke]
9759
9760  *) Implement a countermeasure against a vulnerability recently found
9761     in CBC ciphersuites in SSL 3.0/TLS 1.0: Send an empty fragment
9762     before application data chunks to avoid the use of known IVs
9763     with data potentially chosen by the attacker.
9764     [Bodo Moeller]
9765
9766  *) Fix length checks in ssl3_get_client_hello().
9767     [Bodo Moeller]
9768
9769  *) TLS/SSL library bugfix: use s->s3->in_read_app_data differently
9770     to prevent ssl3_read_internal() from incorrectly assuming that
9771     ssl3_read_bytes() found application data while handshake
9772     processing was enabled when in fact s->s3->in_read_app_data was
9773     merely automatically cleared during the initial handshake.
9774     [Bodo Moeller; problem pointed out by Arne Ansper <arne@ats.cyber.ee>]
9775
9776  *) Fix object definitions for Private and Enterprise: they were not
9777     recognized in their shortname (=lowercase) representation. Extend
9778     obj_dat.pl to issue an error when using undefined keywords instead
9779     of silently ignoring the problem (Svenning Sorensen
9780     <sss@sss.dnsalias.net>).
9781     [Lutz Jaenicke]
9782
9783  *) Fix DH_generate_parameters() so that it works for 'non-standard'
9784     generators, i.e. generators other than 2 and 5.  (Previously, the
9785     code did not properly initialise the 'add' and 'rem' values to
9786     BN_generate_prime().)
9787
9788     In the new general case, we do not insist that 'generator' is
9789     actually a primitive root: This requirement is rather pointless;
9790     a generator of the order-q subgroup is just as good, if not
9791     better.
9792     [Bodo Moeller]
9793
9794  *) Map new X509 verification errors to alerts. Discovered and submitted by
9795     Tom Wu <tom@arcot.com>.
9796     [Lutz Jaenicke]
9797
9798  *) Fix ssl3_pending() (ssl/s3_lib.c) to prevent SSL_pending() from
9799     returning non-zero before the data has been completely received
9800     when using non-blocking I/O.
9801     [Bodo Moeller; problem pointed out by John Hughes]
9802
9803  *) Some of the ciphers missed the strength entry (SSL_LOW etc).
9804     [Ben Laurie, Lutz Jaenicke]
9805
9806  *) Fix bug in SSL_clear(): bad sessions were not removed (found by
9807     Yoram Zahavi <YoramZ@gilian.com>).
9808     [Lutz Jaenicke]
9809
9810  *) Add information about CygWin 1.3 and on, and preserve proper
9811     configuration for the versions before that.
9812     [Corinna Vinschen <vinschen@redhat.com> and Richard Levitte]
9813
9814  *) Make removal from session cache (SSL_CTX_remove_session()) more robust:
9815     check whether we deal with a copy of a session and do not delete from
9816     the cache in this case. Problem reported by "Izhar Shoshani Levi"
9817     <izhar@checkpoint.com>.
9818     [Lutz Jaenicke]
9819
9820  *) Do not store session data into the internal session cache, if it
9821     is never intended to be looked up (SSL_SESS_CACHE_NO_INTERNAL_LOOKUP
9822     flag is set). Proposed by Aslam <aslam@funk.com>.
9823     [Lutz Jaenicke]
9824
9825  *) Have ASN1_BIT_STRING_set_bit() really clear a bit when the requested
9826     value is 0.
9827     [Richard Levitte]
9828
9829  *) [In 0.9.6d-engine release:]
9830     Fix a crashbug and a logic bug in hwcrhk_load_pubkey().
9831     [Toomas Kiisk <vix@cyber.ee> via Richard Levitte]
9832
9833  *) Add the configuration target linux-s390x.
9834     [Neale Ferguson <Neale.Ferguson@SoftwareAG-USA.com> via Richard Levitte]
9835
9836  *) The earlier bugfix for the SSL3_ST_SW_HELLO_REQ_C case of
9837     ssl3_accept (ssl/s3_srvr.c) incorrectly used a local flag
9838     variable as an indication that a ClientHello message has been
9839     received.  As the flag value will be lost between multiple
9840     invocations of ssl3_accept when using non-blocking I/O, the
9841     function may not be aware that a handshake has actually taken
9842     place, thus preventing a new session from being added to the
9843     session cache.
9844
9845     To avoid this problem, we now set s->new_session to 2 instead of
9846     using a local variable.
9847     [Lutz Jaenicke, Bodo Moeller]
9848
9849  *) Bugfix: Return -1 from ssl3_get_server_done (ssl3/s3_clnt.c)
9850     if the SSL_R_LENGTH_MISMATCH error is detected.
9851     [Geoff Thorpe, Bodo Moeller]
9852
9853  *) New 'shared_ldflag' column in Configure platform table.
9854     [Richard Levitte]
9855
9856  *) Fix EVP_CIPHER_mode macro.
9857     ["Dan S. Camper" <dan@bti.net>]
9858
9859  *) Fix ssl3_read_bytes (ssl/s3_pkt.c): To ignore messages of unknown
9860     type, we must throw them away by setting rr->length to 0.
9861     [D P Chang <dpc@qualys.com>]
9862
9863 Changes between 0.9.6b and 0.9.6c  [21 dec 2001]
9864
9865  *) Fix BN_rand_range bug pointed out by Dominikus Scherkl
9866     <Dominikus.Scherkl@biodata.com>.  (The previous implementation
9867     worked incorrectly for those cases where  range = 10..._2  and
9868     3*range  is two bits longer than  range.)
9869     [Bodo Moeller]
9870
9871  *) Only add signing time to PKCS7 structures if it is not already
9872     present.
9873     [Steve Henson]
9874
9875  *) Fix crypto/objects/objects.h: "ld-ce" should be "id-ce",
9876     OBJ_ld_ce should be OBJ_id_ce.
9877     Also some ip-pda OIDs in crypto/objects/objects.txt were
9878     incorrect (cf. RFC 3039).
9879     [Matt Cooper, Frederic Giudicelli, Bodo Moeller]
9880
9881  *) Release CRYPTO_LOCK_DYNLOCK when CRYPTO_destroy_dynlockid()
9882     returns early because it has nothing to do.
9883     [Andy Schneider <andy.schneider@bjss.co.uk>]
9884
9885  *) [In 0.9.6c-engine release:]
9886     Fix mutex callback return values in crypto/engine/hw_ncipher.c.
9887     [Andy Schneider <andy.schneider@bjss.co.uk>]
9888
9889  *) [In 0.9.6c-engine release:]
9890     Add support for Cryptographic Appliance's keyserver technology.
9891     (Use engine 'keyclient')
9892     [Cryptographic Appliances and Geoff Thorpe]
9893
9894  *) Add a configuration entry for OS/390 Unix.  The C compiler 'c89'
9895     is called via tools/c89.sh because arguments have to be
9896     rearranged (all '-L' options must appear before the first object
9897     modules).
9898     [Richard Shapiro <rshapiro@abinitio.com>]
9899
9900  *) [In 0.9.6c-engine release:]
9901     Add support for Broadcom crypto accelerator cards, backported
9902     from 0.9.7.
9903     [Broadcom, Nalin Dahyabhai <nalin@redhat.com>, Mark Cox]
9904
9905  *) [In 0.9.6c-engine release:]
9906     Add support for SureWare crypto accelerator cards from
9907     Baltimore Technologies.  (Use engine 'sureware')
9908     [Baltimore Technologies and Mark Cox]
9909
9910  *) [In 0.9.6c-engine release:]
9911     Add support for crypto accelerator cards from Accelerated
9912     Encryption Processing, www.aep.ie.  (Use engine 'aep')
9913     [AEP Inc. and Mark Cox]
9914
9915  *) Add a configuration entry for gcc on UnixWare.
9916     [Gary Benson <gbenson@redhat.com>]
9917
9918  *) Change ssl/s2_clnt.c and ssl/s2_srvr.c so that received handshake
9919     messages are stored in a single piece (fixed-length part and
9920     variable-length part combined) and fix various bugs found on the way.
9921     [Bodo Moeller]
9922
9923  *) Disable caching in BIO_gethostbyname(), directly use gethostbyname()
9924     instead.  BIO_gethostbyname() does not know what timeouts are
9925     appropriate, so entries would stay in cache even when they have
9926     become invalid.
9927     [Bodo Moeller; problem pointed out by Rich Salz <rsalz@zolera.com>
9928
9929  *) Change ssl23_get_client_hello (ssl/s23_srvr.c) behaviour when
9930     faced with a pathologically small ClientHello fragment that does
9931     not contain client_version: Instead of aborting with an error,
9932     simply choose the highest available protocol version (i.e.,
9933     TLS 1.0 unless it is disabled).  In practice, ClientHello
9934     messages are never sent like this, but this change gives us
9935     strictly correct behaviour at least for TLS.
9936     [Bodo Moeller]
9937
9938  *) Fix SSL handshake functions and SSL_clear() such that SSL_clear()
9939     never resets s->method to s->ctx->method when called from within
9940     one of the SSL handshake functions.
9941     [Bodo Moeller; problem pointed out by Niko Baric]
9942
9943  *) In ssl3_get_client_hello (ssl/s3_srvr.c), generate a fatal alert
9944     (sent using the client's version number) if client_version is
9945     smaller than the protocol version in use.  Also change
9946     ssl23_get_client_hello (ssl/s23_srvr.c) to select TLS 1.0 if
9947     the client demanded SSL 3.0 but only TLS 1.0 is enabled; then
9948     the client will at least see that alert.
9949     [Bodo Moeller]
9950
9951  *) Fix ssl3_get_message (ssl/s3_both.c) to handle message fragmentation
9952     correctly.
9953     [Bodo Moeller]
9954
9955  *) Avoid infinite loop in ssl3_get_message (ssl/s3_both.c) if a
9956     client receives HelloRequest while in a handshake.
9957     [Bodo Moeller; bug noticed by Andy Schneider <andy.schneider@bjss.co.uk>]
9958
9959  *) Bugfix in ssl3_accept (ssl/s3_srvr.c): Case SSL3_ST_SW_HELLO_REQ_C
9960     should end in 'break', not 'goto end' which circumvents various
9961     cleanups done in state SSL_ST_OK.   But session related stuff
9962     must be disabled for SSL_ST_OK in the case that we just sent a
9963     HelloRequest.
9964
9965     Also avoid some overhead by not calling ssl_init_wbio_buffer()
9966     before just sending a HelloRequest.
9967     [Bodo Moeller, Eric Rescorla <ekr@rtfm.com>]
9968
9969  *) Fix ssl/s3_enc.c, ssl/t1_enc.c and ssl/s3_pkt.c so that we don't
9970     reveal whether illegal block cipher padding was found or a MAC
9971     verification error occurred.  (Neither SSLerr() codes nor alerts
9972     are directly visible to potential attackers, but the information
9973     may leak via logfiles.)
9974
9975     Similar changes are not required for the SSL 2.0 implementation
9976     because the number of padding bytes is sent in clear for SSL 2.0,
9977     and the extra bytes are just ignored.  However ssl/s2_pkt.c
9978     failed to verify that the purported number of padding bytes is in
9979     the legal range.
9980     [Bodo Moeller]
9981
9982  *) Add OpenUNIX-8 support including shared libraries
9983     (Boyd Lynn Gerber <gerberb@zenez.com>).
9984     [Lutz Jaenicke]
9985
9986  *) Improve RSA_padding_check_PKCS1_OAEP() check again to avoid
9987     'wristwatch attack' using huge encoding parameters (cf.
9988     James H. Manger's CRYPTO 2001 paper).  Note that the
9989     RSA_PKCS1_OAEP_PADDING case of RSA_private_decrypt() does not use
9990     encoding parameters and hence was not vulnerable.
9991     [Bodo Moeller]
9992
9993  *) BN_sqr() bug fix.
9994     [Ulf Möller, reported by Jim Ellis <jim.ellis@cavium.com>]
9995
9996  *) Rabin-Miller test analyses assume uniformly distributed witnesses,
9997     so use BN_pseudo_rand_range() instead of using BN_pseudo_rand()
9998     followed by modular reduction.
9999     [Bodo Moeller; pointed out by Adam Young <AYoung1@NCSUS.JNJ.COM>]
10000
10001  *) Add BN_pseudo_rand_range() with obvious functionality: BN_rand_range()
10002     equivalent based on BN_pseudo_rand() instead of BN_rand().
10003     [Bodo Moeller]
10004
10005  *) s3_srvr.c: allow sending of large client certificate lists (> 16 kB).
10006     This function was broken, as the check for a new client hello message
10007     to handle SGC did not allow these large messages.
10008     (Tracked down by "Douglas E. Engert" <deengert@anl.gov>.)
10009     [Lutz Jaenicke]
10010
10011  *) Add alert descriptions for TLSv1 to SSL_alert_desc_string[_long]().
10012     [Lutz Jaenicke]
10013
10014  *) Fix buggy behaviour of BIO_get_num_renegotiates() and BIO_ctrl()
10015     for BIO_C_GET_WRITE_BUF_SIZE ("Stephen Hinton" <shinton@netopia.com>).
10016     [Lutz Jaenicke]
10017
10018  *) Rework the configuration and shared library support for Tru64 Unix.
10019     The configuration part makes use of modern compiler features and
10020     still retains old compiler behavior for those that run older versions
10021     of the OS.  The shared library support part includes a variant that
10022     uses the RPATH feature, and is available through the special
10023     configuration target "alpha-cc-rpath", which will never be selected
10024     automatically.
10025     [Tim Mooney <mooney@dogbert.cc.ndsu.NoDak.edu> via Richard Levitte]
10026
10027  *) In ssl3_get_key_exchange (ssl/s3_clnt.c), call ssl3_get_message()
10028     with the same message size as in ssl3_get_certificate_request().
10029     Otherwise, if no ServerKeyExchange message occurs, CertificateRequest
10030     messages might inadvertently be reject as too long.
10031     [Petr Lampa <lampa@fee.vutbr.cz>]
10032
10033  *) Enhanced support for IA-64 Unix platforms (well, Linux and HP-UX).
10034     [Andy Polyakov]
10035
10036  *) Modified SSL library such that the verify_callback that has been set
10037     specifically for an SSL object with SSL_set_verify() is actually being
10038     used. Before the change, a verify_callback set with this function was
10039     ignored and the verify_callback() set in the SSL_CTX at the time of
10040     the call was used. New function X509_STORE_CTX_set_verify_cb() introduced
10041     to allow the necessary settings.
10042     [Lutz Jaenicke]
10043
10044  *) Initialize static variable in crypto/dsa/dsa_lib.c and crypto/dh/dh_lib.c
10045     explicitly to NULL, as at least on Solaris 8 this seems not always to be
10046     done automatically (in contradiction to the requirements of the C
10047     standard). This made problems when used from OpenSSH.
10048     [Lutz Jaenicke]
10049
10050  *) In OpenSSL 0.9.6a and 0.9.6b, crypto/dh/dh_key.c ignored
10051     dh->length and always used
10052
10053          BN_rand_range(priv_key, dh->p).
10054
10055     BN_rand_range() is not necessary for Diffie-Hellman, and this
10056     specific range makes Diffie-Hellman unnecessarily inefficient if
10057     dh->length (recommended exponent length) is much smaller than the
10058     length of dh->p.  We could use BN_rand_range() if the order of
10059     the subgroup was stored in the DH structure, but we only have
10060     dh->length.
10061
10062     So switch back to
10063
10064          BN_rand(priv_key, l, ...)
10065
10066     where 'l' is dh->length if this is defined, or BN_num_bits(dh->p)-1
10067     otherwise.
10068     [Bodo Moeller]
10069
10070  *) In
10071
10072          RSA_eay_public_encrypt
10073          RSA_eay_private_decrypt
10074          RSA_eay_private_encrypt (signing)
10075          RSA_eay_public_decrypt (signature verification)
10076
10077     (default implementations for RSA_public_encrypt,
10078     RSA_private_decrypt, RSA_private_encrypt, RSA_public_decrypt),
10079     always reject numbers >= n.
10080     [Bodo Moeller]
10081
10082  *) In crypto/rand/md_rand.c, use a new short-time lock CRYPTO_LOCK_RAND2
10083     to synchronize access to 'locking_thread'.  This is necessary on
10084     systems where access to 'locking_thread' (an 'unsigned long'
10085     variable) is not atomic.
10086     [Bodo Moeller]
10087
10088  *) In crypto/rand/md_rand.c, set 'locking_thread' to current thread's ID
10089     *before* setting the 'crypto_lock_rand' flag.  The previous code had
10090     a race condition if 0 is a valid thread ID.
10091     [Travis Vitek <vitek@roguewave.com>]
10092
10093  *) Add support for shared libraries under Irix.
10094     [Albert Chin-A-Young <china@thewrittenword.com>]
10095
10096  *) Add configuration option to build on Linux on both big-endian and
10097     little-endian MIPS.
10098     [Ralf Baechle <ralf@uni-koblenz.de>]
10099
10100  *) Add the possibility to create shared libraries on HP-UX.
10101     [Richard Levitte]
10102
10103 Changes between 0.9.6a and 0.9.6b  [9 Jul 2001]
10104
10105  *) Change ssleay_rand_bytes (crypto/rand/md_rand.c)
10106     to avoid a SSLeay/OpenSSL PRNG weakness pointed out by
10107     Markku-Juhani O. Saarinen <markku-juhani.saarinen@nokia.com>:
10108     PRNG state recovery was possible based on the output of
10109     one PRNG request appropriately sized to gain knowledge on
10110     'md' followed by enough consecutive 1-byte PRNG requests
10111     to traverse all of 'state'.
10112
10113     1. When updating 'md_local' (the current thread's copy of 'md')
10114        during PRNG output generation, hash all of the previous
10115        'md_local' value, not just the half used for PRNG output.
10116
10117     2. Make the number of bytes from 'state' included into the hash
10118        independent from the number of PRNG bytes requested.
10119
10120     The first measure alone would be sufficient to avoid
10121     Markku-Juhani's attack.  (Actually it had never occurred
10122     to me that the half of 'md_local' used for chaining was the
10123     half from which PRNG output bytes were taken -- I had always
10124     assumed that the secret half would be used.)  The second
10125     measure makes sure that additional data from 'state' is never
10126     mixed into 'md_local' in small portions; this heuristically
10127     further strengthens the PRNG.
10128     [Bodo Moeller]
10129
10130  *) Fix crypto/bn/asm/mips3.s.
10131     [Andy Polyakov]
10132
10133  *) When only the key is given to "enc", the IV is undefined. Print out
10134     an error message in this case.
10135     [Lutz Jaenicke]
10136
10137  *) Handle special case when X509_NAME is empty in X509 printing routines.
10138     [Steve Henson]
10139
10140  *) In dsa_do_verify (crypto/dsa/dsa_ossl.c), verify that r and s are
10141     positive and less than q.
10142     [Bodo Moeller]
10143
10144  *) Don't change *pointer in CRYPTO_add_lock() is add_lock_callback is
10145     used: it isn't thread safe and the add_lock_callback should handle
10146     that itself.
10147     [Paul Rose <Paul.Rose@bridge.com>]
10148
10149  *) Verify that incoming data obeys the block size in
10150     ssl3_enc (ssl/s3_enc.c) and tls1_enc (ssl/t1_enc.c).
10151     [Bodo Moeller]
10152
10153  *) Fix OAEP check.
10154     [Ulf Möller, Bodo Möller]
10155
10156  *) The countermeasure against Bleichbacher's attack on PKCS #1 v1.5
10157     RSA encryption was accidentally removed in s3_srvr.c in OpenSSL 0.9.5
10158     when fixing the server behaviour for backwards-compatible 'client
10159     hello' messages.  (Note that the attack is impractical against
10160     SSL 3.0 and TLS 1.0 anyway because length and version checking
10161     means that the probability of guessing a valid ciphertext is
10162     around 2^-40; see section 5 in Bleichenbacher's CRYPTO '98
10163     paper.)
10164
10165     Before 0.9.5, the countermeasure (hide the error by generating a
10166     random 'decryption result') did not work properly because
10167     ERR_clear_error() was missing, meaning that SSL_get_error() would
10168     detect the supposedly ignored error.
10169
10170     Both problems are now fixed.
10171     [Bodo Moeller]
10172
10173  *) In crypto/bio/bf_buff.c, increase DEFAULT_BUFFER_SIZE to 4096
10174     (previously it was 1024).
10175     [Bodo Moeller]
10176
10177  *) Fix for compatibility mode trust settings: ignore trust settings
10178     unless some valid trust or reject settings are present.
10179     [Steve Henson]
10180
10181  *) Fix for blowfish EVP: its a variable length cipher.
10182     [Steve Henson]
10183
10184  *) Fix various bugs related to DSA S/MIME verification. Handle missing
10185     parameters in DSA public key structures and return an error in the
10186     DSA routines if parameters are absent.
10187     [Steve Henson]
10188
10189  *) In versions up to 0.9.6, RAND_file_name() resorted to file ".rnd"
10190     in the current directory if neither $RANDFILE nor $HOME was set.
10191     RAND_file_name() in 0.9.6a returned NULL in this case.  This has
10192     caused some confusion to Windows users who haven't defined $HOME.
10193     Thus RAND_file_name() is changed again: e_os.h can define a
10194     DEFAULT_HOME, which will be used if $HOME is not set.
10195     For Windows, we use "C:"; on other platforms, we still require
10196     environment variables.
10197
10198  *) Move 'if (!initialized) RAND_poll()' into regions protected by
10199     CRYPTO_LOCK_RAND.  This is not strictly necessary, but avoids
10200     having multiple threads call RAND_poll() concurrently.
10201     [Bodo Moeller]
10202
10203  *) In crypto/rand/md_rand.c, replace 'add_do_not_lock' flag by a
10204     combination of a flag and a thread ID variable.
10205     Otherwise while one thread is in ssleay_rand_bytes (which sets the
10206     flag), *other* threads can enter ssleay_add_bytes without obeying
10207     the CRYPTO_LOCK_RAND lock (and may even illegally release the lock
10208     that they do not hold after the first thread unsets add_do_not_lock).
10209     [Bodo Moeller]
10210
10211  *) Change bctest again: '-x' expressions are not available in all
10212     versions of 'test'.
10213     [Bodo Moeller]
10214
10215 Changes between 0.9.6 and 0.9.6a  [5 Apr 2001]
10216
10217  *) Fix a couple of memory leaks in PKCS7_dataDecode()
10218     [Steve Henson, reported by Heyun Zheng <hzheng@atdsprint.com>]
10219
10220  *) Change Configure and Makefiles to provide EXE_EXT, which will contain
10221     the default extension for executables, if any.  Also, make the perl
10222     scripts that use symlink() to test if it really exists and use "cp"
10223     if it doesn't.  All this made OpenSSL compilable and installable in
10224     CygWin.
10225     [Richard Levitte]
10226
10227  *) Fix for asn1_GetSequence() for indefinite length constructed data.
10228     If SEQUENCE is length is indefinite just set c->slen to the total
10229     amount of data available.
10230     [Steve Henson, reported by shige@FreeBSD.org]
10231     [This change does not apply to 0.9.7.]
10232
10233  *) Change bctest to avoid here-documents inside command substitution
10234     (workaround for FreeBSD /bin/sh bug).
10235     For compatibility with Ultrix, avoid shell functions (introduced
10236     in the bctest version that searches along $PATH).
10237     [Bodo Moeller]
10238
10239  *) Rename 'des_encrypt' to 'des_encrypt1'.  This avoids the clashes
10240     with des_encrypt() defined on some operating systems, like Solaris
10241     and UnixWare.
10242     [Richard Levitte]
10243
10244  *) Check the result of RSA-CRT (see D. Boneh, R. DeMillo, R. Lipton:
10245     On the Importance of Eliminating Errors in Cryptographic
10246     Computations, J. Cryptology 14 (2001) 2, 101-119,
10247     http://theory.stanford.edu/~dabo/papers/faults.ps.gz).
10248     [Ulf Moeller]
10249
10250  *) MIPS assembler BIGNUM division bug fix.
10251     [Andy Polyakov]
10252
10253  *) Disabled incorrect Alpha assembler code.
10254     [Richard Levitte]
10255
10256  *) Fix PKCS#7 decode routines so they correctly update the length
10257     after reading an EOC for the EXPLICIT tag.
10258     [Steve Henson]
10259     [This change does not apply to 0.9.7.]
10260
10261  *) Fix bug in PKCS#12 key generation routines. This was triggered
10262     if a 3DES key was generated with a 0 initial byte. Include
10263     PKCS12_BROKEN_KEYGEN compilation option to retain the old
10264     (but broken) behaviour.
10265     [Steve Henson]
10266
10267  *) Enhance bctest to search for a working bc along $PATH and print
10268     it when found.
10269     [Tim Rice <tim@multitalents.net> via Richard Levitte]
10270
10271  *) Fix memory leaks in err.c: free err_data string if necessary;
10272     don't write to the wrong index in ERR_set_error_data.
10273     [Bodo Moeller]
10274
10275  *) Implement ssl23_peek (analogous to ssl23_read), which previously
10276     did not exist.
10277     [Bodo Moeller]
10278
10279  *) Replace rdtsc with _emit statements for VC++ version 5.
10280     [Jeremy Cooper <jeremy@baymoo.org>]
10281
10282  *) Make it possible to reuse SSLv2 sessions.
10283     [Richard Levitte]
10284
10285  *) In copy_email() check for >= 0 as a return value for
10286     X509_NAME_get_index_by_NID() since 0 is a valid index.
10287     [Steve Henson reported by Massimiliano Pala <madwolf@opensca.org>]
10288
10289  *) Avoid coredump with unsupported or invalid public keys by checking if
10290     X509_get_pubkey() fails in PKCS7_verify(). Fix memory leak when
10291     PKCS7_verify() fails with non detached data.
10292     [Steve Henson]
10293
10294  *) Don't use getenv in library functions when run as setuid/setgid.
10295     New function OPENSSL_issetugid().
10296     [Ulf Moeller]
10297
10298  *) Avoid false positives in memory leak detection code (crypto/mem_dbg.c)
10299     due to incorrect handling of multi-threading:
10300
10301     1. Fix timing glitch in the MemCheck_off() portion of CRYPTO_mem_ctrl().
10302
10303     2. Fix logical glitch in is_MemCheck_on() aka CRYPTO_is_mem_check_on().
10304
10305     3. Count how many times MemCheck_off() has been called so that
10306        nested use can be treated correctly.  This also avoids
10307        inband-signalling in the previous code (which relied on the
10308        assumption that thread ID 0 is impossible).
10309     [Bodo Moeller]
10310
10311  *) Add "-rand" option also to s_client and s_server.
10312     [Lutz Jaenicke]
10313
10314  *) Fix CPU detection on Irix 6.x.
10315     [Kurt Hockenbury <khockenb@stevens-tech.edu> and
10316      "Bruce W. Forsberg" <bruce.forsberg@baesystems.com>]
10317
10318  *) Fix X509_NAME bug which produced incorrect encoding if X509_NAME
10319     was empty.
10320     [Steve Henson]
10321     [This change does not apply to 0.9.7.]
10322
10323  *) Use the cached encoding of an X509_NAME structure rather than
10324     copying it. This is apparently the reason for the libsafe "errors"
10325     but the code is actually correct.
10326     [Steve Henson]
10327
10328  *) Add new function BN_rand_range(), and fix DSA_sign_setup() to prevent
10329     Bleichenbacher's DSA attack.
10330     Extend BN_[pseudo_]rand: As before, top=1 forces the highest two bits
10331     to be set and top=0 forces the highest bit to be set; top=-1 is new
10332     and leaves the highest bit random.
10333     [Ulf Moeller, Bodo Moeller]
10334
10335  *) In the NCONF_...-based implementations for CONF_... queries
10336     (crypto/conf/conf_lib.c), if the input LHASH is NULL, avoid using
10337     a temporary CONF structure with the data component set to NULL
10338     (which gives segmentation faults in lh_retrieve).
10339     Instead, use NULL for the CONF pointer in CONF_get_string and
10340     CONF_get_number (which may use environment variables) and directly
10341     return NULL from CONF_get_section.
10342     [Bodo Moeller]
10343
10344  *) Fix potential buffer overrun for EBCDIC.
10345     [Ulf Moeller]
10346
10347  *) Tolerate nonRepudiation as being valid for S/MIME signing and certSign
10348     keyUsage if basicConstraints absent for a CA.
10349     [Steve Henson]
10350
10351  *) Make SMIME_write_PKCS7() write mail header values with a format that
10352     is more generally accepted (no spaces before the semicolon), since
10353     some programs can't parse those values properly otherwise.  Also make
10354     sure BIO's that break lines after each write do not create invalid
10355     headers.
10356     [Richard Levitte]
10357
10358  *) Make the CRL encoding routines work with empty SEQUENCE OF. The
10359     macros previously used would not encode an empty SEQUENCE OF
10360     and break the signature.
10361     [Steve Henson]
10362     [This change does not apply to 0.9.7.]
10363
10364  *) Zero the premaster secret after deriving the master secret in
10365     DH ciphersuites.
10366     [Steve Henson]
10367
10368  *) Add some EVP_add_digest_alias registrations (as found in
10369     OpenSSL_add_all_digests()) to SSL_library_init()
10370     aka OpenSSL_add_ssl_algorithms().  This provides improved
10371     compatibility with peers using X.509 certificates
10372     with unconventional AlgorithmIdentifier OIDs.
10373     [Bodo Moeller]
10374
10375  *) Fix for Irix with NO_ASM.
10376     ["Bruce W. Forsberg" <bruce.forsberg@baesystems.com>]
10377
10378  *) ./config script fixes.
10379     [Ulf Moeller, Richard Levitte]
10380
10381  *) Fix 'openssl passwd -1'.
10382     [Bodo Moeller]
10383
10384  *) Change PKCS12_key_gen_asc() so it can cope with non null
10385     terminated strings whose length is passed in the passlen
10386     parameter, for example from PEM callbacks. This was done
10387     by adding an extra length parameter to asc2uni().
10388     [Steve Henson, reported by <oddissey@samsung.co.kr>]
10389
10390  *) Fix C code generated by 'openssl dsaparam -C': If a BN_bin2bn
10391     call failed, free the DSA structure.
10392     [Bodo Moeller]
10393
10394  *) Fix to uni2asc() to cope with zero length Unicode strings.
10395     These are present in some PKCS#12 files.
10396     [Steve Henson]
10397
10398  *) Increase s2->wbuf allocation by one byte in ssl2_new (ssl/s2_lib.c).
10399     Otherwise do_ssl_write (ssl/s2_pkt.c) will write beyond buffer limits
10400     when writing a 32767 byte record.
10401     [Bodo Moeller; problem reported by Eric Day <eday@concentric.net>]
10402
10403  *) In RSA_eay_public_{en,ed}crypt and RSA_eay_mod_exp (rsa_eay.c),
10404     obtain lock CRYPTO_LOCK_RSA before setting rsa->_method_mod_{n,p,q}.
10405
10406     (RSA objects have a reference count access to which is protected
10407     by CRYPTO_LOCK_RSA [see rsa_lib.c, s3_srvr.c, ssl_cert.c, ssl_rsa.c],
10408     so they are meant to be shared between threads.)
10409     [Bodo Moeller, Geoff Thorpe; original patch submitted by
10410     "Reddie, Steven" <Steven.Reddie@ca.com>]
10411
10412  *) Fix a deadlock in CRYPTO_mem_leaks().
10413     [Bodo Moeller]
10414
10415  *) Use better test patterns in bntest.
10416     [Ulf Möller]
10417
10418  *) rand_win.c fix for Borland C.
10419     [Ulf Möller]
10420
10421  *) BN_rshift bugfix for n == 0.
10422     [Bodo Moeller]
10423
10424  *) Add a 'bctest' script that checks for some known 'bc' bugs
10425     so that 'make test' does not abort just because 'bc' is broken.
10426     [Bodo Moeller]
10427
10428  *) Store verify_result within SSL_SESSION also for client side to
10429     avoid potential security hole. (Re-used sessions on the client side
10430     always resulted in verify_result==X509_V_OK, not using the original
10431     result of the server certificate verification.)
10432     [Lutz Jaenicke]
10433
10434  *) Fix ssl3_pending: If the record in s->s3->rrec is not of type
10435     SSL3_RT_APPLICATION_DATA, return 0.
10436     Similarly, change ssl2_pending to return 0 if SSL_in_init(s) is true.
10437     [Bodo Moeller]
10438
10439  *) Fix SSL_peek:
10440     Both ssl2_peek and ssl3_peek, which were totally broken in earlier
10441     releases, have been re-implemented by renaming the previous
10442     implementations of ssl2_read and ssl3_read to ssl2_read_internal
10443     and ssl3_read_internal, respectively, and adding 'peek' parameters
10444     to them.  The new ssl[23]_{read,peek} functions are calls to
10445     ssl[23]_read_internal with the 'peek' flag set appropriately.
10446     A 'peek' parameter has also been added to ssl3_read_bytes, which
10447     does the actual work for ssl3_read_internal.
10448     [Bodo Moeller]
10449
10450  *) Initialise "ex_data" member of RSA/DSA/DH structures prior to calling
10451     the method-specific "init()" handler. Also clean up ex_data after
10452     calling the method-specific "finish()" handler. Previously, this was
10453     happening the other way round.
10454     [Geoff Thorpe]
10455
10456  *) Increase BN_CTX_NUM (the number of BIGNUMs in a BN_CTX) to 16.
10457     The previous value, 12, was not always sufficient for BN_mod_exp().
10458     [Bodo Moeller]
10459
10460  *) Make sure that shared libraries get the internal name engine with
10461     the full version number and not just 0.  This should mark the
10462     shared libraries as not backward compatible.  Of course, this should
10463     be changed again when we can guarantee backward binary compatibility.
10464     [Richard Levitte]
10465
10466  *) Fix typo in get_cert_by_subject() in by_dir.c
10467     [Jean-Marc Desperrier <jean-marc.desperrier@certplus.com>]
10468
10469  *) Rework the system to generate shared libraries:
10470
10471     - Make note of the expected extension for the shared libraries and
10472       if there is a need for symbolic links from for example libcrypto.so.0
10473       to libcrypto.so.0.9.7.  There is extended info in Configure for
10474       that.
10475
10476     - Make as few rebuilds of the shared libraries as possible.
10477
10478     - Still avoid linking the OpenSSL programs with the shared libraries.
10479
10480     - When installing, install the shared libraries separately from the
10481       static ones.
10482     [Richard Levitte]
10483
10484  *) Fix SSL_CTX_set_read_ahead macro to actually use its argument.
10485
10486     Copy SSL_CTX's read_ahead flag to SSL object directly in SSL_new
10487     and not in SSL_clear because the latter is also used by the
10488     accept/connect functions; previously, the settings made by
10489     SSL_set_read_ahead would be lost during the handshake.
10490     [Bodo Moeller; problems reported by Anders Gertz <gertz@epact.se>]
10491
10492  *) Correct util/mkdef.pl to be selective about disabled algorithms.
10493     Previously, it would create entries for disabled algorithms no
10494     matter what.
10495     [Richard Levitte]
10496
10497  *) Added several new manual pages for SSL_* function.
10498     [Lutz Jaenicke]
10499
10500 Changes between 0.9.5a and 0.9.6  [24 Sep 2000]
10501
10502  *) In ssl23_get_client_hello, generate an error message when faced
10503     with an initial SSL 3.0/TLS record that is too small to contain the
10504     first two bytes of the ClientHello message, i.e. client_version.
10505     (Note that this is a pathologic case that probably has never happened
10506     in real life.)  The previous approach was to use the version number
10507     from the record header as a substitute; but our protocol choice
10508     should not depend on that one because it is not authenticated
10509     by the Finished messages.
10510     [Bodo Moeller]
10511
10512  *) More robust randomness gathering functions for Windows.
10513     [Jeffrey Altman <jaltman@columbia.edu>]
10514
10515  *) For compatibility reasons if the flag X509_V_FLAG_ISSUER_CHECK is
10516     not set then we don't setup the error code for issuer check errors
10517     to avoid possibly overwriting other errors which the callback does
10518     handle. If an application does set the flag then we assume it knows
10519     what it is doing and can handle the new informational codes
10520     appropriately.
10521     [Steve Henson]
10522
10523  *) Fix for a nasty bug in ASN1_TYPE handling. ASN1_TYPE is used for
10524     a general "ANY" type, as such it should be able to decode anything
10525     including tagged types. However it didn't check the class so it would
10526     wrongly interpret tagged types in the same way as their universal
10527     counterpart and unknown types were just rejected. Changed so that the
10528     tagged and unknown types are handled in the same way as a SEQUENCE:
10529     that is the encoding is stored intact. There is also a new type
10530     "V_ASN1_OTHER" which is used when the class is not universal, in this
10531     case we have no idea what the actual type is so we just lump them all
10532     together.
10533     [Steve Henson]
10534
10535  *) On VMS, stdout may very well lead to a file that is written to
10536     in a record-oriented fashion.  That means that every write() will
10537     write a separate record, which will be read separately by the
10538     programs trying to read from it.  This can be very confusing.
10539
10540     The solution is to put a BIO filter in the way that will buffer
10541     text until a linefeed is reached, and then write everything a
10542     line at a time, so every record written will be an actual line,
10543     not chunks of lines and not (usually doesn't happen, but I've
10544     seen it once) several lines in one record.  BIO_f_linebuffer() is
10545     the answer.
10546
10547     Currently, it's a VMS-only method, because that's where it has
10548     been tested well enough.
10549     [Richard Levitte]
10550
10551  *) Remove 'optimized' squaring variant in BN_mod_mul_montgomery,
10552     it can return incorrect results.
10553     (Note: The buggy variant was not enabled in OpenSSL 0.9.5a,
10554     but it was in 0.9.6-beta[12].)
10555     [Bodo Moeller]
10556
10557  *) Disable the check for content being present when verifying detached
10558     signatures in pk7_smime.c. Some versions of Netscape (wrongly)
10559     include zero length content when signing messages.
10560     [Steve Henson]
10561
10562  *) New BIO_shutdown_wr macro, which invokes the BIO_C_SHUTDOWN_WR
10563     BIO_ctrl (for BIO pairs).
10564     [Bodo Möller]
10565
10566  *) Add DSO method for VMS.
10567     [Richard Levitte]
10568
10569  *) Bug fix: Montgomery multiplication could produce results with the
10570     wrong sign.
10571     [Ulf Möller]
10572
10573  *) Add RPM specification openssl.spec and modify it to build three
10574     packages.  The default package contains applications, application
10575     documentation and run-time libraries.  The devel package contains
10576     include files, static libraries and function documentation.  The
10577     doc package contains the contents of the doc directory.  The original
10578     openssl.spec was provided by Damien Miller <djm@mindrot.org>.
10579     [Richard Levitte]
10580
10581  *) Add a large number of documentation files for many SSL routines.
10582     [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>]
10583
10584  *) Add a configuration entry for Sony News 4.
10585     [NAKAJI Hiroyuki <nakaji@tutrp.tut.ac.jp>]
10586
10587  *) Don't set the two most significant bits to one when generating a
10588     random number < q in the DSA library.
10589     [Ulf Möller]
10590
10591  *) New SSL API mode 'SSL_MODE_AUTO_RETRY'.  This disables the default
10592     behaviour that SSL_read may result in SSL_ERROR_WANT_READ (even if
10593     the underlying transport is blocking) if a handshake took place.
10594     (The default behaviour is needed by applications such as s_client
10595     and s_server that use select() to determine when to use SSL_read;
10596     but for applications that know in advance when to expect data, it
10597     just makes things more complicated.)
10598     [Bodo Moeller]
10599
10600  *) Add RAND_egd_bytes(), which gives control over the number of bytes read
10601     from EGD.
10602     [Ben Laurie]
10603
10604  *) Add a few more EBCDIC conditionals that make `req' and `x509'
10605     work better on such systems.
10606     [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>]
10607
10608  *) Add two demo programs for PKCS12_parse() and PKCS12_create().
10609     Update PKCS12_parse() so it copies the friendlyName and the
10610     keyid to the certificates aux info.
10611     [Steve Henson]
10612
10613  *) Fix bug in PKCS7_verify() which caused an infinite loop
10614     if there was more than one signature.
10615     [Sven Uszpelkat <su@celocom.de>]
10616
10617  *) Major change in util/mkdef.pl to include extra information
10618     about each symbol, as well as presenting variables as well
10619     as functions.  This change means that there's n more need
10620     to rebuild the .num files when some algorithms are excluded.
10621     [Richard Levitte]
10622
10623  *) Allow the verify time to be set by an application,
10624     rather than always using the current time.
10625     [Steve Henson]
10626
10627  *) Phase 2 verify code reorganisation. The certificate
10628     verify code now looks up an issuer certificate by a
10629     number of criteria: subject name, authority key id
10630     and key usage. It also verifies self signed certificates
10631     by the same criteria. The main comparison function is
10632     X509_check_issued() which performs these checks.
10633
10634     Lot of changes were necessary in order to support this
10635     without completely rewriting the lookup code.
10636
10637     Authority and subject key identifier are now cached.
10638
10639     The LHASH 'certs' is X509_STORE has now been replaced
10640     by a STACK_OF(X509_OBJECT). This is mainly because an
10641     LHASH can't store or retrieve multiple objects with
10642     the same hash value.
10643
10644     As a result various functions (which were all internal
10645     use only) have changed to handle the new X509_STORE
10646     structure. This will break anything that messed round
10647     with X509_STORE internally.
10648
10649     The functions X509_STORE_add_cert() now checks for an
10650     exact match, rather than just subject name.
10651
10652     The X509_STORE API doesn't directly support the retrieval
10653     of multiple certificates matching a given criteria, however
10654     this can be worked round by performing a lookup first
10655     (which will fill the cache with candidate certificates)
10656     and then examining the cache for matches. This is probably
10657     the best we can do without throwing out X509_LOOKUP
10658     entirely (maybe later...).
10659
10660     The X509_VERIFY_CTX structure has been enhanced considerably.
10661
10662     All certificate lookup operations now go via a get_issuer()
10663     callback. Although this currently uses an X509_STORE it
10664     can be replaced by custom lookups. This is a simple way
10665     to bypass the X509_STORE hackery necessary to make this
10666     work and makes it possible to use more efficient techniques
10667     in future. A very simple version which uses a simple
10668     STACK for its trusted certificate store is also provided
10669     using X509_STORE_CTX_trusted_stack().
10670
10671     The verify_cb() and verify() callbacks now have equivalents
10672     in the X509_STORE_CTX structure.
10673
10674     X509_STORE_CTX also has a 'flags' field which can be used
10675     to customise the verify behaviour.
10676     [Steve Henson]
10677
10678  *) Add new PKCS#7 signing option PKCS7_NOSMIMECAP which
10679     excludes S/MIME capabilities.
10680     [Steve Henson]
10681
10682  *) When a certificate request is read in keep a copy of the
10683     original encoding of the signed data and use it when outputting
10684     again. Signatures then use the original encoding rather than
10685     a decoded, encoded version which may cause problems if the
10686     request is improperly encoded.
10687     [Steve Henson]
10688
10689  *) For consistency with other BIO_puts implementations, call
10690     buffer_write(b, ...) directly in buffer_puts instead of calling
10691     BIO_write(b, ...).
10692
10693     In BIO_puts, increment b->num_write as in BIO_write.
10694     [Peter.Sylvester@EdelWeb.fr]
10695
10696  *) Fix BN_mul_word for the case where the word is 0. (We have to use
10697     BN_zero, we may not return a BIGNUM with an array consisting of
10698     words set to zero.)
10699     [Bodo Moeller]
10700
10701  *) Avoid calling abort() from within the library when problems are
10702     detected, except if preprocessor symbols have been defined
10703     (such as REF_CHECK, BN_DEBUG etc.).
10704     [Bodo Moeller]
10705
10706  *) New openssl application 'rsautl'. This utility can be
10707     used for low level RSA operations. DER public key
10708     BIO/fp routines also added.
10709     [Steve Henson]
10710
10711  *) New Configure entry and patches for compiling on QNX 4.
10712     [Andreas Schneider <andreas@ds3.etech.fh-hamburg.de>]
10713
10714  *) A demo state-machine implementation was sponsored by
10715     Nuron (http://www.nuron.com/) and is now available in
10716     demos/state_machine.
10717     [Ben Laurie]
10718
10719  *) New options added to the 'dgst' utility for signature
10720     generation and verification.
10721     [Steve Henson]
10722
10723  *) Unrecognized PKCS#7 content types are now handled via a
10724     catch all ASN1_TYPE structure. This allows unsupported
10725     types to be stored as a "blob" and an application can
10726     encode and decode it manually.
10727     [Steve Henson]
10728
10729  *) Fix various signed/unsigned issues to make a_strex.c
10730     compile under VC++.
10731     [Oscar Jacobsson <oscar.jacobsson@celocom.com>]
10732
10733  *) ASN1 fixes. i2d_ASN1_OBJECT was not returning the correct
10734     length if passed a buffer. ASN1_INTEGER_to_BN failed
10735     if passed a NULL BN and its argument was negative.
10736     [Steve Henson, pointed out by Sven Heiberg <sven@tartu.cyber.ee>]
10737
10738  *) Modification to PKCS#7 encoding routines to output definite
10739     length encoding. Since currently the whole structures are in
10740     memory there's not real point in using indefinite length
10741     constructed encoding. However if OpenSSL is compiled with
10742     the flag PKCS7_INDEFINITE_ENCODING the old form is used.
10743     [Steve Henson]
10744
10745  *) Added BIO_vprintf() and BIO_vsnprintf().
10746     [Richard Levitte]
10747
10748  *) Added more prefixes to parse for in the strings written
10749     through a logging bio, to cover all the levels that are available
10750     through syslog.  The prefixes are now:
10751
10752        PANIC, EMERG, EMR       =>      LOG_EMERG
10753        ALERT, ALR              =>      LOG_ALERT
10754        CRIT, CRI               =>      LOG_CRIT
10755        ERROR, ERR              =>      LOG_ERR
10756        WARNING, WARN, WAR      =>      LOG_WARNING
10757        NOTICE, NOTE, NOT       =>      LOG_NOTICE
10758        INFO, INF               =>      LOG_INFO
10759        DEBUG, DBG              =>      LOG_DEBUG
10760
10761     and as before, if none of those prefixes are present at the
10762     beginning of the string, LOG_ERR is chosen.
10763
10764     On Win32, the LOG_* levels are mapped according to this:
10765
10766        LOG_EMERG, LOG_ALERT, LOG_CRIT, LOG_ERR => EVENTLOG_ERROR_TYPE
10767        LOG_WARNING                             => EVENTLOG_WARNING_TYPE
10768        LOG_NOTICE, LOG_INFO, LOG_DEBUG         => EVENTLOG_INFORMATION_TYPE
10769
10770     [Richard Levitte]
10771
10772  *) Made it possible to reconfigure with just the configuration
10773     argument "reconf" or "reconfigure".  The command line arguments
10774     are stored in Makefile.ssl in the variable CONFIGURE_ARGS,
10775     and are retrieved from there when reconfiguring.
10776     [Richard Levitte]
10777
10778  *) MD4 implemented.
10779     [Assar Westerlund <assar@sics.se>, Richard Levitte]
10780
10781  *) Add the arguments -CAfile and -CApath to the pkcs12 utility.
10782     [Richard Levitte]
10783
10784  *) The obj_dat.pl script was messing up the sorting of object
10785     names. The reason was that it compared the quoted version
10786     of strings as a result "OCSP" > "OCSP Signing" because
10787     " > SPACE. Changed script to store unquoted versions of
10788     names and add quotes on output. It was also omitting some
10789     names from the lookup table if they were given a default
10790     value (that is if SN is missing it is given the same
10791     value as LN and vice versa), these are now added on the
10792     grounds that if an object has a name we should be able to
10793     look it up. Finally added warning output when duplicate
10794     short or long names are found.
10795     [Steve Henson]
10796
10797  *) Changes needed for Tandem NSK.
10798     [Scott Uroff <scott@xypro.com>]
10799
10800  *) Fix SSL 2.0 rollback checking: Due to an off-by-one error in
10801     RSA_padding_check_SSLv23(), special padding was never detected
10802     and thus the SSL 3.0/TLS 1.0 countermeasure against protocol
10803     version rollback attacks was not effective.
10804
10805     In s23_clnt.c, don't use special rollback-attack detection padding
10806     (RSA_SSLV23_PADDING) if SSL 2.0 is the only protocol enabled in the
10807     client; similarly, in s23_srvr.c, don't do the rollback check if
10808     SSL 2.0 is the only protocol enabled in the server.
10809     [Bodo Moeller]
10810
10811  *) Make it possible to get hexdumps of unprintable data with 'openssl
10812     asn1parse'.  By implication, the functions ASN1_parse_dump() and
10813     BIO_dump_indent() are added.
10814     [Richard Levitte]
10815
10816  *) New functions ASN1_STRING_print_ex() and X509_NAME_print_ex()
10817     these print out strings and name structures based on various
10818     flags including RFC2253 support and proper handling of
10819     multibyte characters. Added options to the 'x509' utility
10820     to allow the various flags to be set.
10821     [Steve Henson]
10822
10823  *) Various fixes to use ASN1_TIME instead of ASN1_UTCTIME.
10824     Also change the functions X509_cmp_current_time() and
10825     X509_gmtime_adj() work with an ASN1_TIME structure,
10826     this will enable certificates using GeneralizedTime in validity
10827     dates to be checked.
10828     [Steve Henson]
10829
10830  *) Make the NEG_PUBKEY_BUG code (which tolerates invalid
10831     negative public key encodings) on by default,
10832     NO_NEG_PUBKEY_BUG can be set to disable it.
10833     [Steve Henson]
10834
10835  *) New function c2i_ASN1_OBJECT() which acts on ASN1_OBJECT
10836     content octets. An i2c_ASN1_OBJECT is unnecessary because
10837     the encoding can be trivially obtained from the structure.
10838     [Steve Henson]
10839
10840  *) crypto/err.c locking bugfix: Use write locks (CRYPTO_w_[un]lock),
10841     not read locks (CRYPTO_r_[un]lock).
10842     [Bodo Moeller]
10843
10844  *) A first attempt at creating official support for shared
10845     libraries through configuration.  I've kept it so the
10846     default is static libraries only, and the OpenSSL programs
10847     are always statically linked for now, but there are
10848     preparations for dynamic linking in place.
10849     This has been tested on Linux and Tru64.
10850     [Richard Levitte]
10851
10852  *) Randomness polling function for Win9x, as described in:
10853     Peter Gutmann, Software Generation of Practically Strong
10854     Random Numbers.
10855     [Ulf Möller]
10856
10857  *) Fix so PRNG is seeded in req if using an already existing
10858     DSA key.
10859     [Steve Henson]
10860
10861  *) New options to smime application. -inform and -outform
10862     allow alternative formats for the S/MIME message including
10863     PEM and DER. The -content option allows the content to be
10864     specified separately. This should allow things like Netscape
10865     form signing output easier to verify.
10866     [Steve Henson]
10867
10868  *) Fix the ASN1 encoding of tags using the 'long form'.
10869     [Steve Henson]
10870
10871  *) New ASN1 functions, i2c_* and c2i_* for INTEGER and BIT
10872     STRING types. These convert content octets to and from the
10873     underlying type. The actual tag and length octets are
10874     already assumed to have been read in and checked. These
10875     are needed because all other string types have virtually
10876     identical handling apart from the tag. By having versions
10877     of the ASN1 functions that just operate on content octets
10878     IMPLICIT tagging can be handled properly. It also allows
10879     the ASN1_ENUMERATED code to be cut down because ASN1_ENUMERATED
10880     and ASN1_INTEGER are identical apart from the tag.
10881     [Steve Henson]
10882
10883  *) Change the handling of OID objects as follows:
10884
10885     - New object identifiers are inserted in objects.txt, following
10886       the syntax given in objects.README.
10887     - objects.pl is used to process obj_mac.num and create a new
10888       obj_mac.h.
10889     - obj_dat.pl is used to create a new obj_dat.h, using the data in
10890       obj_mac.h.
10891
10892     This is currently kind of a hack, and the perl code in objects.pl
10893     isn't very elegant, but it works as I intended.  The simplest way
10894     to check that it worked correctly is to look in obj_dat.h and
10895     check the array nid_objs and make sure the objects haven't moved
10896     around (this is important!).  Additions are OK, as well as
10897     consistent name changes.
10898     [Richard Levitte]
10899
10900  *) Add BSD-style MD5-based passwords to 'openssl passwd' (option '-1').
10901     [Bodo Moeller]
10902
10903  *) Addition of the command line parameter '-rand file' to 'openssl req'.
10904     The given file adds to whatever has already been seeded into the
10905     random pool through the RANDFILE configuration file option or
10906     environment variable, or the default random state file.
10907     [Richard Levitte]
10908
10909  *) mkstack.pl now sorts each macro group into lexical order.
10910     Previously the output order depended on the order the files
10911     appeared in the directory, resulting in needless rewriting
10912     of safestack.h .
10913     [Steve Henson]
10914
10915  *) Patches to make OpenSSL compile under Win32 again. Mostly
10916     work arounds for the VC++ problem that it treats func() as
10917     func(void). Also stripped out the parts of mkdef.pl that
10918     added extra typesafe functions: these no longer exist.
10919     [Steve Henson]
10920
10921  *) Reorganisation of the stack code. The macros are now all
10922     collected in safestack.h . Each macro is defined in terms of
10923     a "stack macro" of the form SKM_<name>(type, a, b). The
10924     DEBUG_SAFESTACK is now handled in terms of function casts,
10925     this has the advantage of retaining type safety without the
10926     use of additional functions. If DEBUG_SAFESTACK is not defined
10927     then the non typesafe macros are used instead. Also modified the
10928     mkstack.pl script to handle the new form. Needs testing to see
10929     if which (if any) compilers it chokes and maybe make DEBUG_SAFESTACK
10930     the default if no major problems. Similar behaviour for ASN1_SET_OF
10931     and PKCS12_STACK_OF.
10932     [Steve Henson]
10933
10934  *) When some versions of IIS use the 'NET' form of private key the
10935     key derivation algorithm is different. Normally MD5(password) is
10936     used as a 128 bit RC4 key. In the modified case
10937     MD5(MD5(password) + "SGCKEYSALT")  is used instead. Added some
10938     new functions i2d_RSA_NET(), d2i_RSA_NET() etc which are the same
10939     as the old Netscape_RSA functions except they have an additional
10940     'sgckey' parameter which uses the modified algorithm. Also added
10941     an -sgckey command line option to the rsa utility. Thanks to
10942     Adrian Peck <bertie@ncipher.com> for posting details of the modified
10943     algorithm to openssl-dev.
10944     [Steve Henson]
10945
10946  *) The evp_local.h macros were using 'c.##kname' which resulted in
10947     invalid expansion on some systems (SCO 5.0.5 for example).
10948     Corrected to 'c.kname'.
10949     [Phillip Porch <root@theporch.com>]
10950
10951  *) New X509_get1_email() and X509_REQ_get1_email() functions that return
10952     a STACK of email addresses from a certificate or request, these look
10953     in the subject name and the subject alternative name extensions and
10954     omit any duplicate addresses.
10955     [Steve Henson]
10956
10957  *) Re-implement BN_mod_exp2_mont using independent (and larger) windows.
10958     This makes DSA verification about 2 % faster.
10959     [Bodo Moeller]
10960
10961  *) Increase maximum window size in BN_mod_exp_... to 6 bits instead of 5
10962     (meaning that now 2^5 values will be precomputed, which is only 4 KB
10963     plus overhead for 1024 bit moduli).
10964     This makes exponentiations about 0.5 % faster for 1024 bit
10965     exponents (as measured by "openssl speed rsa2048").
10966     [Bodo Moeller]
10967
10968  *) Rename memory handling macros to avoid conflicts with other
10969     software:
10970          Malloc         =>  OPENSSL_malloc
10971          Malloc_locked  =>  OPENSSL_malloc_locked
10972          Realloc        =>  OPENSSL_realloc
10973          Free           =>  OPENSSL_free
10974     [Richard Levitte]
10975
10976  *) New function BN_mod_exp_mont_word for small bases (roughly 15%
10977     faster than BN_mod_exp_mont, i.e. 7% for a full DH exchange).
10978     [Bodo Moeller]
10979
10980  *) CygWin32 support.
10981     [John Jarvie <jjarvie@newsguy.com>]
10982
10983  *) The type-safe stack code has been rejigged. It is now only compiled
10984     in when OpenSSL is configured with the DEBUG_SAFESTACK option and
10985     by default all type-specific stack functions are "#define"d back to
10986     standard stack functions. This results in more streamlined output
10987     but retains the type-safety checking possibilities of the original
10988     approach.
10989     [Geoff Thorpe]
10990
10991  *) The STACK code has been cleaned up, and certain type declarations
10992     that didn't make a lot of sense have been brought in line. This has
10993     also involved a cleanup of sorts in safestack.h to more correctly
10994     map type-safe stack functions onto their plain stack counterparts.
10995     This work has also resulted in a variety of "const"ifications of
10996     lots of the code, especially "_cmp" operations which should normally
10997     be prototyped with "const" parameters anyway.
10998     [Geoff Thorpe]
10999
11000  *) When generating bytes for the first time in md_rand.c, 'stir the pool'
11001     by seeding with STATE_SIZE dummy bytes (with zero entropy count).
11002     (The PRNG state consists of two parts, the large pool 'state' and 'md',
11003     where all of 'md' is used each time the PRNG is used, but 'state'
11004     is used only indexed by a cyclic counter. As entropy may not be
11005     well distributed from the beginning, 'md' is important as a
11006     chaining variable. However, the output function chains only half
11007     of 'md', i.e. 80 bits.  ssleay_rand_add, on the other hand, chains
11008     all of 'md', and seeding with STATE_SIZE dummy bytes will result
11009     in all of 'state' being rewritten, with the new values depending
11010     on virtually all of 'md'.  This overcomes the 80 bit limitation.)
11011     [Bodo Moeller]
11012
11013  *) In ssl/s2_clnt.c and ssl/s3_clnt.c, call ERR_clear_error() when
11014     the handshake is continued after ssl_verify_cert_chain();
11015     otherwise, if SSL_VERIFY_NONE is set, remaining error codes
11016     can lead to 'unexplainable' connection aborts later.
11017     [Bodo Moeller; problem tracked down by Lutz Jaenicke]
11018
11019  *) Major EVP API cipher revision.
11020     Add hooks for extra EVP features. This allows various cipher
11021     parameters to be set in the EVP interface. Support added for variable
11022     key length ciphers via the EVP_CIPHER_CTX_set_key_length() function and
11023     setting of RC2 and RC5 parameters.
11024
11025     Modify EVP_OpenInit() and EVP_SealInit() to cope with variable key length
11026     ciphers.
11027
11028     Remove lots of duplicated code from the EVP library. For example *every*
11029     cipher init() function handles the 'iv' in the same way according to the
11030     cipher mode. They also all do nothing if the 'key' parameter is NULL and
11031     for CFB and OFB modes they zero ctx->num.
11032
11033     New functionality allows removal of S/MIME code RC2 hack.
11034
11035     Most of the routines have the same form and so can be declared in terms
11036     of macros.
11037
11038     By shifting this to the top level EVP_CipherInit() it can be removed from
11039     all individual ciphers. If the cipher wants to handle IVs or keys
11040     differently it can set the EVP_CIPH_CUSTOM_IV or EVP_CIPH_ALWAYS_CALL_INIT
11041     flags.
11042
11043     Change lots of functions like EVP_EncryptUpdate() to now return a
11044     value: although software versions of the algorithms cannot fail
11045     any installed hardware versions can.
11046     [Steve Henson]
11047
11048  *) Implement SSL_OP_TLS_ROLLBACK_BUG: In ssl3_get_client_key_exchange, if
11049     this option is set, tolerate broken clients that send the negotiated
11050     protocol version number instead of the requested protocol version
11051     number.
11052     [Bodo Moeller]
11053
11054  *) Call dh_tmp_cb (set by ..._TMP_DH_CB) with correct 'is_export' flag;
11055     i.e. non-zero for export ciphersuites, zero otherwise.
11056     Previous versions had this flag inverted, inconsistent with
11057     rsa_tmp_cb (..._TMP_RSA_CB).
11058     [Bodo Moeller; problem reported by Amit Chopra]
11059
11060  *) Add missing DSA library text string. Work around for some IIS
11061     key files with invalid SEQUENCE encoding.
11062     [Steve Henson]
11063
11064  *) Add a document (doc/standards.txt) that list all kinds of standards
11065     and so on that are implemented in OpenSSL.
11066     [Richard Levitte]
11067
11068  *) Enhance c_rehash script. Old version would mishandle certificates
11069     with the same subject name hash and wouldn't handle CRLs at all.
11070     Added -fingerprint option to crl utility, to support new c_rehash
11071     features.
11072     [Steve Henson]
11073
11074  *) Eliminate non-ANSI declarations in crypto.h and stack.h.
11075     [Ulf Möller]
11076
11077  *) Fix for SSL server purpose checking. Server checking was
11078     rejecting certificates which had extended key usage present
11079     but no ssl client purpose.
11080     [Steve Henson, reported by Rene Grosser <grosser@hisolutions.com>]
11081
11082  *) Make PKCS#12 code work with no password. The PKCS#12 spec
11083     is a little unclear about how a blank password is handled.
11084     Since the password in encoded as a BMPString with terminating
11085     double NULL a zero length password would end up as just the
11086     double NULL. However no password at all is different and is
11087     handled differently in the PKCS#12 key generation code. NS
11088     treats a blank password as zero length. MSIE treats it as no
11089     password on export: but it will try both on import. We now do
11090     the same: PKCS12_parse() tries zero length and no password if
11091     the password is set to "" or NULL (NULL is now a valid password:
11092     it wasn't before) as does the pkcs12 application.
11093     [Steve Henson]
11094
11095  *) Bugfixes in apps/x509.c: Avoid a memory leak; and don't use
11096     perror when PEM_read_bio_X509_REQ fails, the error message must
11097     be obtained from the error queue.
11098     [Bodo Moeller]
11099
11100  *) Avoid 'thread_hash' memory leak in crypto/err/err.c by freeing
11101     it in ERR_remove_state if appropriate, and change ERR_get_state
11102     accordingly to avoid race conditions (this is necessary because
11103     thread_hash is no longer constant once set).
11104     [Bodo Moeller]
11105
11106  *) Bugfix for linux-elf makefile.one.
11107     [Ulf Möller]
11108
11109  *) RSA_get_default_method() will now cause a default
11110     RSA_METHOD to be chosen if one doesn't exist already.
11111     Previously this was only set during a call to RSA_new()
11112     or RSA_new_method(NULL) meaning it was possible for
11113     RSA_get_default_method() to return NULL.
11114     [Geoff Thorpe]
11115
11116  *) Added native name translation to the existing DSO code
11117     that will convert (if the flag to do so is set) filenames
11118     that are sufficiently small and have no path information
11119     into a canonical native form. Eg. "blah" converted to
11120     "libblah.so" or "blah.dll" etc.
11121     [Geoff Thorpe]
11122
11123  *) New function ERR_error_string_n(e, buf, len) which is like
11124     ERR_error_string(e, buf), but writes at most 'len' bytes
11125     including the 0 terminator.  For ERR_error_string_n, 'buf'
11126     may not be NULL.
11127     [Damien Miller <djm@mindrot.org>, Bodo Moeller]
11128
11129  *) CONF library reworked to become more general.  A new CONF
11130     configuration file reader "class" is implemented as well as a
11131     new functions (NCONF_*, for "New CONF") to handle it.  The now
11132     old CONF_* functions are still there, but are reimplemented to
11133     work in terms of the new functions.  Also, a set of functions
11134     to handle the internal storage of the configuration data is
11135     provided to make it easier to write new configuration file
11136     reader "classes" (I can definitely see something reading a
11137     configuration file in XML format, for example), called _CONF_*,
11138     or "the configuration storage API"...
11139
11140     The new configuration file reading functions are:
11141
11142        NCONF_new, NCONF_free, NCONF_load, NCONF_load_fp, NCONF_load_bio,
11143        NCONF_get_section, NCONF_get_string, NCONF_get_numbre
11144
11145        NCONF_default, NCONF_WIN32
11146
11147        NCONF_dump_fp, NCONF_dump_bio
11148
11149     NCONF_default and NCONF_WIN32 are method (or "class") choosers,
11150     NCONF_new creates a new CONF object.  This works in the same way
11151     as other interfaces in OpenSSL, like the BIO interface.
11152     NCONF_dump_* dump the internal storage of the configuration file,
11153     which is useful for debugging.  All other functions take the same
11154     arguments as the old CONF_* functions with the exception of the
11155     first that must be a `CONF *' instead of a `LHASH *'.
11156
11157     To make it easier to use the new classes with the old CONF_* functions,
11158     the function CONF_set_default_method is provided.
11159     [Richard Levitte]
11160
11161  *) Add '-tls1' option to 'openssl ciphers', which was already
11162     mentioned in the documentation but had not been implemented.
11163     (This option is not yet really useful because even the additional
11164     experimental TLS 1.0 ciphers are currently treated as SSL 3.0 ciphers.)
11165     [Bodo Moeller]
11166
11167  *) Initial DSO code added into libcrypto for letting OpenSSL (and
11168     OpenSSL-based applications) load shared libraries and bind to
11169     them in a portable way.
11170     [Geoff Thorpe, with contributions from Richard Levitte]
11171
11172 Changes between 0.9.5 and 0.9.5a  [1 Apr 2000]
11173
11174  *) Make sure _lrotl and _lrotr are only used with MSVC.
11175
11176  *) Use lock CRYPTO_LOCK_RAND correctly in ssleay_rand_status
11177     (the default implementation of RAND_status).
11178
11179  *) Rename openssl x509 option '-crlext', which was added in 0.9.5,
11180     to '-clrext' (= clear extensions), as intended and documented.
11181     [Bodo Moeller; inconsistency pointed out by Michael Attili
11182     <attili@amaxo.com>]
11183
11184  *) Fix for HMAC. It wasn't zeroing the rest of the block if the key length
11185     was larger than the MD block size.
11186     [Steve Henson, pointed out by Yost William <YostW@tce.com>]
11187
11188  *) Modernise PKCS12_parse() so it uses STACK_OF(X509) for its ca argument
11189     fix a leak when the ca argument was passed as NULL. Stop X509_PUBKEY_set()
11190     using the passed key: if the passed key was a private key the result
11191     of X509_print(), for example, would be to print out all the private key
11192     components.
11193     [Steve Henson]
11194
11195  *) des_quad_cksum() byte order bug fix.
11196     [Ulf Möller, using the problem description in krb4-0.9.7, where
11197      the solution is attributed to Derrick J Brashear <shadow@DEMENTIA.ORG>]
11198
11199  *) Fix so V_ASN1_APP_CHOOSE works again: however its use is strongly
11200     discouraged.
11201     [Steve Henson, pointed out by Brian Korver <briank@cs.stanford.edu>]
11202
11203  *) For easily testing in shell scripts whether some command
11204     'openssl XXX' exists, the new pseudo-command 'openssl no-XXX'
11205     returns with exit code 0 iff no command of the given name is available.
11206     'no-XXX' is printed in this case, 'XXX' otherwise.  In both cases,
11207     the output goes to stdout and nothing is printed to stderr.
11208     Additional arguments are always ignored.
11209
11210     Since for each cipher there is a command of the same name,
11211     the 'no-cipher' compilation switches can be tested this way.
11212
11213     ('openssl no-XXX' is not able to detect pseudo-commands such
11214     as 'quit', 'list-XXX-commands', or 'no-XXX' itself.)
11215     [Bodo Moeller]
11216
11217  *) Update test suite so that 'make test' succeeds in 'no-rsa' configuration.
11218     [Bodo Moeller]
11219
11220  *) For SSL_[CTX_]set_tmp_dh, don't create a DH key if SSL_OP_SINGLE_DH_USE
11221     is set; it will be thrown away anyway because each handshake creates
11222     its own key.
11223     ssl_cert_dup, which is used by SSL_new, now copies DH keys in addition
11224     to parameters -- in previous versions (since OpenSSL 0.9.3) the
11225     'default key' from SSL_CTX_set_tmp_dh would always be lost, meaning
11226     you effectively got SSL_OP_SINGLE_DH_USE when using this macro.
11227     [Bodo Moeller]
11228
11229  *) New s_client option -ign_eof: EOF at stdin is ignored, and
11230     'Q' and 'R' lose their special meanings (quit/renegotiate).
11231     This is part of what -quiet does; unlike -quiet, -ign_eof
11232     does not suppress any output.
11233     [Richard Levitte]
11234
11235  *) Add compatibility options to the purpose and trust code. The
11236     purpose X509_PURPOSE_ANY is "any purpose" which automatically
11237     accepts a certificate or CA, this was the previous behaviour,
11238     with all the associated security issues.
11239
11240     X509_TRUST_COMPAT is the old trust behaviour: only and
11241     automatically trust self signed roots in certificate store. A
11242     new trust setting X509_TRUST_DEFAULT is used to specify that
11243     a purpose has no associated trust setting and it should instead
11244     use the value in the default purpose.
11245     [Steve Henson]
11246
11247  *) Fix the PKCS#8 DSA private key code so it decodes keys again
11248     and fix a memory leak.
11249     [Steve Henson]
11250
11251  *) In util/mkerr.pl (which implements 'make errors'), preserve
11252     reason strings from the previous version of the .c file, as
11253     the default to have only downcase letters (and digits) in
11254     automatically generated reasons codes is not always appropriate.
11255     [Bodo Moeller]
11256
11257  *) In ERR_load_ERR_strings(), build an ERR_LIB_SYS error reason table
11258     using strerror.  Previously, ERR_reason_error_string() returned
11259     library names as reason strings for SYSerr; but SYSerr is a special
11260     case where small numbers are errno values, not library numbers.
11261     [Bodo Moeller]
11262
11263  *) Add '-dsaparam' option to 'openssl dhparam' application.  This
11264     converts DSA parameters into DH parameters. (When creating parameters,
11265     DSA_generate_parameters is used.)
11266     [Bodo Moeller]
11267
11268  *) Include 'length' (recommended exponent length) in C code generated
11269     by 'openssl dhparam -C'.
11270     [Bodo Moeller]
11271
11272  *) The second argument to set_label in perlasm was already being used
11273     so couldn't be used as a "file scope" flag. Moved to third argument
11274     which was free.
11275     [Steve Henson]
11276
11277  *) In PEM_ASN1_write_bio and some other functions, use RAND_pseudo_bytes
11278     instead of RAND_bytes for encryption IVs and salts.
11279     [Bodo Moeller]
11280
11281  *) Include RAND_status() into RAND_METHOD instead of implementing
11282     it only for md_rand.c  Otherwise replacing the PRNG by calling
11283     RAND_set_rand_method would be impossible.
11284     [Bodo Moeller]
11285
11286  *) Don't let DSA_generate_key() enter an infinite loop if the random
11287     number generation fails.
11288     [Bodo Moeller]
11289
11290  *) New 'rand' application for creating pseudo-random output.
11291     [Bodo Moeller]
11292
11293  *) Added configuration support for Linux/IA64
11294     [Rolf Haberrecker <rolf@suse.de>]
11295
11296  *) Assembler module support for Mingw32.
11297     [Ulf Möller]
11298
11299  *) Shared library support for HPUX (in shlib/).
11300     [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE> and Anonymous]
11301
11302  *) Shared library support for Solaris gcc.
11303     [Lutz Behnke <behnke@trustcenter.de>]
11304
11305 Changes between 0.9.4 and 0.9.5  [28 Feb 2000]
11306
11307  *) PKCS7_encrypt() was adding text MIME headers twice because they
11308     were added manually and by SMIME_crlf_copy().
11309     [Steve Henson]
11310
11311  *) In bntest.c don't call BN_rand with zero bits argument.
11312     [Steve Henson, pointed out by Andrew W. Gray <agray@iconsinc.com>]
11313
11314  *) BN_mul bugfix: In bn_mul_part_recursion() only the a>a[n] && b>b[n]
11315     case was implemented. This caused BN_div_recp() to fail occasionally.
11316     [Ulf Möller]
11317
11318  *) Add an optional second argument to the set_label() in the perl
11319     assembly language builder. If this argument exists and is set
11320     to 1 it signals that the assembler should use a symbol whose
11321     scope is the entire file, not just the current function. This
11322     is needed with MASM which uses the format label:: for this scope.
11323     [Steve Henson, pointed out by Peter Runestig <peter@runestig.com>]
11324
11325  *) Change the ASN1 types so they are typedefs by default. Before
11326     almost all types were #define'd to ASN1_STRING which was causing
11327     STACK_OF() problems: you couldn't declare STACK_OF(ASN1_UTF8STRING)
11328     for example.
11329     [Steve Henson]
11330
11331  *) Change names of new functions to the new get1/get0 naming
11332     convention: After 'get1', the caller owns a reference count
11333     and has to call ..._free; 'get0' returns a pointer to some
11334     data structure without incrementing reference counters.
11335     (Some of the existing 'get' functions increment a reference
11336     counter, some don't.)
11337     Similarly, 'set1' and 'add1' functions increase reference
11338     counters or duplicate objects.
11339     [Steve Henson]
11340
11341  *) Allow for the possibility of temp RSA key generation failure:
11342     the code used to assume it always worked and crashed on failure.
11343     [Steve Henson]
11344
11345  *) Fix potential buffer overrun problem in BIO_printf().
11346     [Ulf Möller, using public domain code by Patrick Powell; problem
11347      pointed out by David Sacerdote <das33@cornell.edu>]
11348
11349  *) Support EGD <http://www.lothar.com/tech/crypto/>.  New functions
11350     RAND_egd() and RAND_status().  In the command line application,
11351     the EGD socket can be specified like a seed file using RANDFILE
11352     or -rand.
11353     [Ulf Möller]
11354
11355  *) Allow the string CERTIFICATE to be tolerated in PKCS#7 structures.
11356     Some CAs (e.g. Verisign) distribute certificates in this form.
11357     [Steve Henson]
11358
11359  *) Remove the SSL_ALLOW_ADH compile option and set the default cipher
11360     list to exclude them. This means that no special compilation option
11361     is needed to use anonymous DH: it just needs to be included in the
11362     cipher list.
11363     [Steve Henson]
11364
11365  *) Change the EVP_MD_CTX_type macro so its meaning consistent with
11366     EVP_MD_type. The old functionality is available in a new macro called
11367     EVP_MD_md(). Change code that uses it and update docs.
11368     [Steve Henson]
11369
11370  *) ..._ctrl functions now have corresponding ..._callback_ctrl functions
11371     where the 'void *' argument is replaced by a function pointer argument.
11372     Previously 'void *' was abused to point to functions, which works on
11373     many platforms, but is not correct.  As these functions are usually
11374     called by macros defined in OpenSSL header files, most source code
11375     should work without changes.
11376     [Richard Levitte]
11377
11378  *) <openssl/opensslconf.h> (which is created by Configure) now contains
11379     sections with information on -D... compiler switches used for
11380     compiling the library so that applications can see them.  To enable
11381     one of these sections, a pre-processor symbol OPENSSL_..._DEFINES
11382     must be defined.  E.g.,
11383        #define OPENSSL_ALGORITHM_DEFINES
11384        #include <openssl/opensslconf.h>
11385     defines all pertinent NO_<algo> symbols, such as NO_IDEA, NO_RSA, etc.
11386     [Richard Levitte, Ulf and Bodo Möller]
11387
11388  *) Bugfix: Tolerate fragmentation and interleaving in the SSL 3/TLS
11389     record layer.
11390     [Bodo Moeller]
11391
11392  *) Change the 'other' type in certificate aux info to a STACK_OF
11393     X509_ALGOR. Although not an AlgorithmIdentifier as such it has
11394     the required ASN1 format: arbitrary types determined by an OID.
11395     [Steve Henson]
11396
11397  *) Add some PEM_write_X509_REQ_NEW() functions and a command line
11398     argument to 'req'. This is not because the function is newer or
11399     better than others it just uses the work 'NEW' in the certificate
11400     request header lines. Some software needs this.
11401     [Steve Henson]
11402
11403  *) Reorganise password command line arguments: now passwords can be
11404     obtained from various sources. Delete the PEM_cb function and make
11405     it the default behaviour: i.e. if the callback is NULL and the
11406     usrdata argument is not NULL interpret it as a null terminated pass
11407     phrase. If usrdata and the callback are NULL then the pass phrase
11408     is prompted for as usual.
11409     [Steve Henson]
11410
11411  *) Add support for the Compaq Atalla crypto accelerator. If it is installed,
11412     the support is automatically enabled. The resulting binaries will
11413     autodetect the card and use it if present.
11414     [Ben Laurie and Compaq Inc.]
11415
11416  *) Work around for Netscape hang bug. This sends certificate request
11417     and server done in one record. Since this is perfectly legal in the
11418     SSL/TLS protocol it isn't a "bug" option and is on by default. See
11419     the bugs/SSLv3 entry for more info.
11420     [Steve Henson]
11421
11422  *) HP-UX tune-up: new unified configs, HP C compiler bug workaround.
11423     [Andy Polyakov]
11424
11425  *) Add -rand argument to smime and pkcs12 applications and read/write
11426     of seed file.
11427     [Steve Henson]
11428
11429  *) New 'passwd' tool for crypt(3) and apr1 password hashes.
11430     [Bodo Moeller]
11431
11432  *) Add command line password options to the remaining applications.
11433     [Steve Henson]
11434
11435  *) Bug fix for BN_div_recp() for numerators with an even number of
11436     bits.
11437     [Ulf Möller]
11438
11439  *) More tests in bntest.c, and changed test_bn output.
11440     [Ulf Möller]
11441
11442  *) ./config recognizes MacOS X now.
11443     [Andy Polyakov]
11444
11445  *) Bug fix for BN_div() when the first words of num and divisor are
11446     equal (it gave wrong results if (rem=(n1-q*d0)&BN_MASK2) < d0).
11447     [Ulf Möller]
11448
11449  *) Add support for various broken PKCS#8 formats, and command line
11450     options to produce them.
11451     [Steve Henson]
11452
11453  *) New functions BN_CTX_start(), BN_CTX_get() and BT_CTX_end() to
11454     get temporary BIGNUMs from a BN_CTX.
11455     [Ulf Möller]
11456
11457  *) Correct return values in BN_mod_exp_mont() and BN_mod_exp2_mont()
11458     for p == 0.
11459     [Ulf Möller]
11460
11461  *) Change the SSLeay_add_all_*() functions to OpenSSL_add_all_*() and
11462     include a #define from the old name to the new. The original intent
11463     was that statically linked binaries could for example just call
11464     SSLeay_add_all_ciphers() to just add ciphers to the table and not
11465     link with digests. This never worked because SSLeay_add_all_digests()
11466     and SSLeay_add_all_ciphers() were in the same source file so calling
11467     one would link with the other. They are now in separate source files.
11468     [Steve Henson]
11469
11470  *) Add a new -notext option to 'ca' and a -pubkey option to 'spkac'.
11471     [Steve Henson]
11472
11473  *) Use a less unusual form of the Miller-Rabin primality test (it used
11474     a binary algorithm for exponentiation integrated into the Miller-Rabin
11475     loop, our standard modexp algorithms are faster).
11476     [Bodo Moeller]
11477
11478  *) Support for the EBCDIC character set completed.
11479     [Martin Kraemer <Martin.Kraemer@Mch.SNI.De>]
11480
11481  *) Source code cleanups: use const where appropriate, eliminate casts,
11482     use void * instead of char * in lhash.
11483     [Ulf Möller]
11484
11485  *) Bugfix: ssl3_send_server_key_exchange was not restartable
11486     (the state was not changed to SSL3_ST_SW_KEY_EXCH_B, and because of
11487     this the server could overwrite ephemeral keys that the client
11488     has already seen).
11489     [Bodo Moeller]
11490
11491  *) Turn DSA_is_prime into a macro that calls BN_is_prime,
11492     using 50 iterations of the Rabin-Miller test.
11493
11494     DSA_generate_parameters now uses BN_is_prime_fasttest (with 50
11495     iterations of the Rabin-Miller test as required by the appendix
11496     to FIPS PUB 186[-1]) instead of DSA_is_prime.
11497     As BN_is_prime_fasttest includes trial division, DSA parameter
11498     generation becomes much faster.
11499
11500     This implies a change for the callback functions in DSA_is_prime
11501     and DSA_generate_parameters: The callback function is called once
11502     for each positive witness in the Rabin-Miller test, not just
11503     occasionally in the inner loop; and the parameters to the
11504     callback function now provide an iteration count for the outer
11505     loop rather than for the current invocation of the inner loop.
11506     DSA_generate_parameters additionally can call the callback
11507     function with an 'iteration count' of -1, meaning that a
11508     candidate has passed the trial division test (when q is generated
11509     from an application-provided seed, trial division is skipped).
11510     [Bodo Moeller]
11511
11512  *) New function BN_is_prime_fasttest that optionally does trial
11513     division before starting the Rabin-Miller test and has
11514     an additional BN_CTX * argument (whereas BN_is_prime always
11515     has to allocate at least one BN_CTX).
11516     'callback(1, -1, cb_arg)' is called when a number has passed the
11517     trial division stage.
11518     [Bodo Moeller]
11519
11520  *) Fix for bug in CRL encoding. The validity dates weren't being handled
11521     as ASN1_TIME.
11522     [Steve Henson]
11523
11524  *) New -pkcs12 option to CA.pl script to write out a PKCS#12 file.
11525     [Steve Henson]
11526
11527  *) New function BN_pseudo_rand().
11528     [Ulf Möller]
11529
11530  *) Clean up BN_mod_mul_montgomery(): replace the broken (and unreadable)
11531     bignum version of BN_from_montgomery() with the working code from
11532     SSLeay 0.9.0 (the word based version is faster anyway), and clean up
11533     the comments.
11534     [Ulf Möller]
11535
11536  *) Avoid a race condition in s2_clnt.c (function get_server_hello) that
11537     made it impossible to use the same SSL_SESSION data structure in
11538     SSL2 clients in multiple threads.
11539     [Bodo Moeller]
11540
11541  *) The return value of RAND_load_file() no longer counts bytes obtained
11542     by stat().  RAND_load_file(..., -1) is new and uses the complete file
11543     to seed the PRNG (previously an explicit byte count was required).
11544     [Ulf Möller, Bodo Möller]
11545
11546  *) Clean up CRYPTO_EX_DATA functions, some of these didn't have prototypes
11547     used (char *) instead of (void *) and had casts all over the place.
11548     [Steve Henson]
11549
11550  *) Make BN_generate_prime() return NULL on error if ret!=NULL.
11551     [Ulf Möller]
11552
11553  *) Retain source code compatibility for BN_prime_checks macro:
11554     BN_is_prime(..., BN_prime_checks, ...) now uses
11555     BN_prime_checks_for_size to determine the appropriate number of
11556     Rabin-Miller iterations.
11557     [Ulf Möller]
11558
11559  *) Diffie-Hellman uses "safe" primes: DH_check() return code renamed to
11560     DH_CHECK_P_NOT_SAFE_PRIME.
11561     (Check if this is true? OpenPGP calls them "strong".)
11562     [Ulf Möller]
11563
11564  *) Merge the functionality of "dh" and "gendh" programs into a new program
11565     "dhparam". The old programs are retained for now but will handle DH keys
11566     (instead of parameters) in future.
11567     [Steve Henson]
11568
11569  *) Make the ciphers, s_server and s_client programs check the return values
11570     when a new cipher list is set.
11571     [Steve Henson]
11572
11573  *) Enhance the SSL/TLS cipher mechanism to correctly handle the TLS 56bit
11574     ciphers. Before when the 56bit ciphers were enabled the sorting was
11575     wrong.
11576
11577     The syntax for the cipher sorting has been extended to support sorting by
11578     cipher-strength (using the strength_bits hard coded in the tables).
11579     The new command is "@STRENGTH" (see also doc/apps/ciphers.pod).
11580
11581     Fix a bug in the cipher-command parser: when supplying a cipher command
11582     string with an "undefined" symbol (neither command nor alphanumeric
11583     [A-Za-z0-9], ssl_set_cipher_list used to hang in an endless loop. Now
11584     an error is flagged.
11585
11586     Due to the strength-sorting extension, the code of the
11587     ssl_create_cipher_list() function was completely rearranged. I hope that
11588     the readability was also increased :-)
11589     [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>]
11590
11591  *) Minor change to 'x509' utility. The -CAcreateserial option now uses 1
11592     for the first serial number and places 2 in the serial number file. This
11593     avoids problems when the root CA is created with serial number zero and
11594     the first user certificate has the same issuer name and serial number
11595     as the root CA.
11596     [Steve Henson]
11597
11598  *) Fixes to X509_ATTRIBUTE utilities, change the 'req' program so it uses
11599     the new code. Add documentation for this stuff.
11600     [Steve Henson]
11601
11602  *) Changes to X509_ATTRIBUTE utilities. These have been renamed from
11603     X509_*() to X509at_*() on the grounds that they don't handle X509
11604     structures and behave in an analogous way to the X509v3 functions:
11605     they shouldn't be called directly but wrapper functions should be used
11606     instead.
11607
11608     So we also now have some wrapper functions that call the X509at functions
11609     when passed certificate requests. (TO DO: similar things can be done with
11610     PKCS#7 signed and unsigned attributes, PKCS#12 attributes and a few other
11611     things. Some of these need some d2i or i2d and print functionality
11612     because they handle more complex structures.)
11613     [Steve Henson]
11614
11615  *) Add missing #ifndefs that caused missing symbols when building libssl
11616     as a shared library without RSA.  Use #ifndef NO_SSL2 instead of
11617     NO_RSA in ssl/s2*.c.
11618     [Kris Kennaway <kris@hub.freebsd.org>, modified by Ulf Möller]
11619
11620  *) Precautions against using the PRNG uninitialized: RAND_bytes() now
11621     has a return value which indicates the quality of the random data
11622     (1 = ok, 0 = not seeded).  Also an error is recorded on the thread's
11623     error queue. New function RAND_pseudo_bytes() generates output that is
11624     guaranteed to be unique but not unpredictable. RAND_add is like
11625     RAND_seed, but takes an extra argument for an entropy estimate
11626     (RAND_seed always assumes full entropy).
11627     [Ulf Möller]
11628
11629  *) Do more iterations of Rabin-Miller probable prime test (specifically,
11630     3 for 1024-bit primes, 6 for 512-bit primes, 12 for 256-bit primes
11631     instead of only 2 for all lengths; see BN_prime_checks_for_size definition
11632     in crypto/bn/bn_prime.c for the complete table).  This guarantees a
11633     false-positive rate of at most 2^-80 for random input.
11634     [Bodo Moeller]
11635
11636  *) Rewrite ssl3_read_n (ssl/s3_pkt.c) avoiding a couple of bugs.
11637     [Bodo Moeller]
11638
11639  *) New function X509_CTX_rget_chain() (renamed to X509_CTX_get1_chain
11640     in the 0.9.5 release), this returns the chain
11641     from an X509_CTX structure with a dup of the stack and all
11642     the X509 reference counts upped: so the stack will exist
11643     after X509_CTX_cleanup() has been called. Modify pkcs12.c
11644     to use this.
11645
11646     Also make SSL_SESSION_print() print out the verify return
11647     code.
11648     [Steve Henson]
11649
11650  *) Add manpage for the pkcs12 command. Also change the default
11651     behaviour so MAC iteration counts are used unless the new
11652     -nomaciter option is used. This improves file security and
11653     only older versions of MSIE (4.0 for example) need it.
11654     [Steve Henson]
11655
11656  *) Honor the no-xxx Configure options when creating .DEF files.
11657     [Ulf Möller]
11658
11659  *) Add PKCS#10 attributes to field table: challengePassword,
11660     unstructuredName and unstructuredAddress. These are taken from
11661     draft PKCS#9 v2.0 but are compatible with v1.2 provided no
11662     international characters are used.
11663
11664     More changes to X509_ATTRIBUTE code: allow the setting of types
11665     based on strings. Remove the 'loc' parameter when adding
11666     attributes because these will be a SET OF encoding which is sorted
11667     in ASN1 order.
11668     [Steve Henson]
11669
11670  *) Initial changes to the 'req' utility to allow request generation
11671     automation. This will allow an application to just generate a template
11672     file containing all the field values and have req construct the
11673     request.
11674
11675     Initial support for X509_ATTRIBUTE handling. Stacks of these are
11676     used all over the place including certificate requests and PKCS#7
11677     structures. They are currently handled manually where necessary with
11678     some primitive wrappers for PKCS#7. The new functions behave in a
11679     manner analogous to the X509 extension functions: they allow
11680     attributes to be looked up by NID and added.
11681
11682     Later something similar to the X509V3 code would be desirable to
11683     automatically handle the encoding, decoding and printing of the
11684     more complex types. The string types like challengePassword can
11685     be handled by the string table functions.
11686
11687     Also modified the multi byte string table handling. Now there is
11688     a 'global mask' which masks out certain types. The table itself
11689     can use the flag STABLE_NO_MASK to ignore the mask setting: this
11690     is useful when for example there is only one permissible type
11691     (as in countryName) and using the mask might result in no valid
11692     types at all.
11693     [Steve Henson]
11694
11695  *) Clean up 'Finished' handling, and add functions SSL_get_finished and
11696     SSL_get_peer_finished to allow applications to obtain the latest
11697     Finished messages sent to the peer or expected from the peer,
11698     respectively.  (SSL_get_peer_finished is usually the Finished message
11699     actually received from the peer, otherwise the protocol will be aborted.)
11700
11701     As the Finished message are message digests of the complete handshake
11702     (with a total of 192 bits for TLS 1.0 and more for SSL 3.0), they can
11703     be used for external authentication procedures when the authentication
11704     provided by SSL/TLS is not desired or is not enough.
11705     [Bodo Moeller]
11706
11707  *) Enhanced support for Alpha Linux is added. Now ./config checks if
11708     the host supports BWX extension and if Compaq C is present on the
11709     $PATH. Just exploiting of the BWX extension results in 20-30%
11710     performance kick for some algorithms, e.g. DES and RC4 to mention
11711     a couple. Compaq C in turn generates ~20% faster code for MD5 and
11712     SHA1.
11713     [Andy Polyakov]
11714
11715  *) Add support for MS "fast SGC". This is arguably a violation of the
11716     SSL3/TLS protocol. Netscape SGC does two handshakes: the first with
11717     weak crypto and after checking the certificate is SGC a second one
11718     with strong crypto. MS SGC stops the first handshake after receiving
11719     the server certificate message and sends a second client hello. Since
11720     a server will typically do all the time consuming operations before
11721     expecting any further messages from the client (server key exchange
11722     is the most expensive) there is little difference between the two.
11723
11724     To get OpenSSL to support MS SGC we have to permit a second client
11725     hello message after we have sent server done. In addition we have to
11726     reset the MAC if we do get this second client hello.
11727     [Steve Henson]
11728
11729  *) Add a function 'd2i_AutoPrivateKey()' this will automatically decide
11730     if a DER encoded private key is RSA or DSA traditional format. Changed
11731     d2i_PrivateKey_bio() to use it. This is only needed for the "traditional"
11732     format DER encoded private key. Newer code should use PKCS#8 format which
11733     has the key type encoded in the ASN1 structure. Added DER private key
11734     support to pkcs8 application.
11735     [Steve Henson]
11736
11737  *) SSL 3/TLS 1 servers now don't request certificates when an anonymous
11738     ciphersuites has been selected (as required by the SSL 3/TLS 1
11739     specifications).  Exception: When SSL_VERIFY_FAIL_IF_NO_PEER_CERT
11740     is set, we interpret this as a request to violate the specification
11741     (the worst that can happen is a handshake failure, and 'correct'
11742     behaviour would result in a handshake failure anyway).
11743     [Bodo Moeller]
11744
11745  *) In SSL_CTX_add_session, take into account that there might be multiple
11746     SSL_SESSION structures with the same session ID (e.g. when two threads
11747     concurrently obtain them from an external cache).
11748     The internal cache can handle only one SSL_SESSION with a given ID,
11749     so if there's a conflict, we now throw out the old one to achieve
11750     consistency.
11751     [Bodo Moeller]
11752
11753  *) Add OIDs for idea and blowfish in CBC mode. This will allow both
11754     to be used in PKCS#5 v2.0 and S/MIME.  Also add checking to
11755     some routines that use cipher OIDs: some ciphers do not have OIDs
11756     defined and so they cannot be used for S/MIME and PKCS#5 v2.0 for
11757     example.
11758     [Steve Henson]
11759
11760  *) Simplify the trust setting structure and code. Now we just have
11761     two sequences of OIDs for trusted and rejected settings. These will
11762     typically have values the same as the extended key usage extension
11763     and any application specific purposes.
11764
11765     The trust checking code now has a default behaviour: it will just
11766     check for an object with the same NID as the passed id. Functions can
11767     be provided to override either the default behaviour or the behaviour
11768     for a given id. SSL client, server and email already have functions
11769     in place for compatibility: they check the NID and also return "trusted"
11770     if the certificate is self signed.
11771     [Steve Henson]
11772
11773  *) Add d2i,i2d bio/fp functions for PrivateKey: these convert the
11774     traditional format into an EVP_PKEY structure.
11775     [Steve Henson]
11776
11777  *) Add a password callback function PEM_cb() which either prompts for
11778     a password if usr_data is NULL or otherwise assumes it is a null
11779     terminated password. Allow passwords to be passed on command line
11780     environment or config files in a few more utilities.
11781     [Steve Henson]
11782
11783  *) Add a bunch of DER and PEM functions to handle PKCS#8 format private
11784     keys. Add some short names for PKCS#8 PBE algorithms and allow them
11785     to be specified on the command line for the pkcs8 and pkcs12 utilities.
11786     Update documentation.
11787     [Steve Henson]
11788
11789  *) Support for ASN1 "NULL" type. This could be handled before by using
11790     ASN1_TYPE but there wasn't any function that would try to read a NULL
11791     and produce an error if it couldn't. For compatibility we also have
11792     ASN1_NULL_new() and ASN1_NULL_free() functions but these are faked and
11793     don't allocate anything because they don't need to.
11794     [Steve Henson]
11795
11796  *) Initial support for MacOS is now provided. Examine INSTALL.MacOS
11797     for details.
11798     [Andy Polyakov, Roy Woods <roy@centicsystems.ca>]
11799
11800  *) Rebuild of the memory allocation routines used by OpenSSL code and
11801     possibly others as well.  The purpose is to make an interface that
11802     provide hooks so anyone can build a separate set of allocation and
11803     deallocation routines to be used by OpenSSL, for example memory
11804     pool implementations, or something else, which was previously hard
11805     since Malloc(), Realloc() and Free() were defined as macros having
11806     the values malloc, realloc and free, respectively (except for Win32
11807     compilations).  The same is provided for memory debugging code.
11808     OpenSSL already comes with functionality to find memory leaks, but
11809     this gives people a chance to debug other memory problems.
11810
11811     With these changes, a new set of functions and macros have appeared:
11812
11813       CRYPTO_set_mem_debug_functions()         [F]
11814       CRYPTO_get_mem_debug_functions()         [F]
11815       CRYPTO_dbg_set_options()                 [F]
11816       CRYPTO_dbg_get_options()                 [F]
11817       CRYPTO_malloc_debug_init()               [M]
11818
11819     The memory debug functions are NULL by default, unless the library
11820     is compiled with CRYPTO_MDEBUG or friends is defined.  If someone
11821     wants to debug memory anyway, CRYPTO_malloc_debug_init() (which
11822     gives the standard debugging functions that come with OpenSSL) or
11823     CRYPTO_set_mem_debug_functions() (tells OpenSSL to use functions
11824     provided by the library user) must be used.  When the standard
11825     debugging functions are used, CRYPTO_dbg_set_options can be used to
11826     request additional information:
11827     CRYPTO_dbg_set_options(V_CYRPTO_MDEBUG_xxx) corresponds to setting
11828     the CRYPTO_MDEBUG_xxx macro when compiling the library.
11829
11830     Also, things like CRYPTO_set_mem_functions will always give the
11831     expected result (the new set of functions is used for allocation
11832     and deallocation) at all times, regardless of platform and compiler
11833     options.
11834
11835     To finish it up, some functions that were never use in any other
11836     way than through macros have a new API and new semantic:
11837
11838       CRYPTO_dbg_malloc()
11839       CRYPTO_dbg_realloc()
11840       CRYPTO_dbg_free()
11841
11842     All macros of value have retained their old syntax.
11843     [Richard Levitte and Bodo Moeller]
11844
11845  *) Some S/MIME fixes. The OID for SMIMECapabilities was wrong, the
11846     ordering of SMIMECapabilities wasn't in "strength order" and there
11847     was a missing NULL in the AlgorithmIdentifier for the SHA1 signature
11848     algorithm.
11849     [Steve Henson]
11850
11851  *) Some ASN1 types with illegal zero length encoding (INTEGER,
11852     ENUMERATED and OBJECT IDENTIFIER) choked the ASN1 routines.
11853     [Frans Heymans <fheymans@isaserver.be>, modified by Steve Henson]
11854
11855  *) Merge in my S/MIME library for OpenSSL. This provides a simple
11856     S/MIME API on top of the PKCS#7 code, a MIME parser (with enough
11857     functionality to handle multipart/signed properly) and a utility
11858     called 'smime' to call all this stuff. This is based on code I
11859     originally wrote for Celo who have kindly allowed it to be
11860     included in OpenSSL.
11861     [Steve Henson]
11862
11863  *) Add variants des_set_key_checked and des_set_key_unchecked of
11864     des_set_key (aka des_key_sched).  Global variable des_check_key
11865     decides which of these is called by des_set_key; this way
11866     des_check_key behaves as it always did, but applications and
11867     the library itself, which was buggy for des_check_key == 1,
11868     have a cleaner way to pick the version they need.
11869     [Bodo Moeller]
11870
11871  *) New function PKCS12_newpass() which changes the password of a
11872     PKCS12 structure.
11873     [Steve Henson]
11874
11875  *) Modify X509_TRUST and X509_PURPOSE so it also uses a static and
11876     dynamic mix. In both cases the ids can be used as an index into the
11877     table. Also modified the X509_TRUST_add() and X509_PURPOSE_add()
11878     functions so they accept a list of the field values and the
11879     application doesn't need to directly manipulate the X509_TRUST
11880     structure.
11881     [Steve Henson]
11882
11883  *) Modify the ASN1_STRING_TABLE stuff so it also uses bsearch and doesn't
11884     need initialising.
11885     [Steve Henson]
11886
11887  *) Modify the way the V3 extension code looks up extensions. This now
11888     works in a similar way to the object code: we have some "standard"
11889     extensions in a static table which is searched with OBJ_bsearch()
11890     and the application can add dynamic ones if needed. The file
11891     crypto/x509v3/ext_dat.h now has the info: this file needs to be
11892     updated whenever a new extension is added to the core code and kept
11893     in ext_nid order. There is a simple program 'tabtest.c' which checks
11894     this. New extensions are not added too often so this file can readily
11895     be maintained manually.
11896
11897     There are two big advantages in doing things this way. The extensions
11898     can be looked up immediately and no longer need to be "added" using
11899     X509V3_add_standard_extensions(): this function now does nothing.
11900     [Side note: I get *lots* of email saying the extension code doesn't
11901      work because people forget to call this function]
11902     Also no dynamic allocation is done unless new extensions are added:
11903     so if we don't add custom extensions there is no need to call
11904     X509V3_EXT_cleanup().
11905     [Steve Henson]
11906
11907  *) Modify enc utility's salting as follows: make salting the default. Add a
11908     magic header, so unsalted files fail gracefully instead of just decrypting
11909     to garbage. This is because not salting is a big security hole, so people
11910     should be discouraged from doing it.
11911     [Ben Laurie]
11912
11913  *) Fixes and enhancements to the 'x509' utility. It allowed a message
11914     digest to be passed on the command line but it only used this
11915     parameter when signing a certificate. Modified so all relevant
11916     operations are affected by the digest parameter including the
11917     -fingerprint and -x509toreq options. Also -x509toreq choked if a
11918     DSA key was used because it didn't fix the digest.
11919     [Steve Henson]
11920
11921  *) Initial certificate chain verify code. Currently tests the untrusted
11922     certificates for consistency with the verify purpose (which is set
11923     when the X509_STORE_CTX structure is set up) and checks the pathlength.
11924
11925     There is a NO_CHAIN_VERIFY compilation option to keep the old behaviour:
11926     this is because it will reject chains with invalid extensions whereas
11927     every previous version of OpenSSL and SSLeay made no checks at all.
11928
11929     Trust code: checks the root CA for the relevant trust settings. Trust
11930     settings have an initial value consistent with the verify purpose: e.g.
11931     if the verify purpose is for SSL client use it expects the CA to be
11932     trusted for SSL client use. However the default value can be changed to
11933     permit custom trust settings: one example of this would be to only trust
11934     certificates from a specific "secure" set of CAs.
11935
11936     Also added X509_STORE_CTX_new() and X509_STORE_CTX_free() functions
11937     which should be used for version portability: especially since the
11938     verify structure is likely to change more often now.
11939
11940     SSL integration. Add purpose and trust to SSL_CTX and SSL and functions
11941     to set them. If not set then assume SSL clients will verify SSL servers
11942     and vice versa.
11943
11944     Two new options to the verify program: -untrusted allows a set of
11945     untrusted certificates to be passed in and -purpose which sets the
11946     intended purpose of the certificate. If a purpose is set then the
11947     new chain verify code is used to check extension consistency.
11948     [Steve Henson]
11949
11950  *) Support for the authority information access extension.
11951     [Steve Henson]
11952
11953  *) Modify RSA and DSA PEM read routines to transparently handle
11954     PKCS#8 format private keys. New *_PUBKEY_* functions that handle
11955     public keys in a format compatible with certificate
11956     SubjectPublicKeyInfo structures. Unfortunately there were already
11957     functions called *_PublicKey_* which used various odd formats so
11958     these are retained for compatibility: however the DSA variants were
11959     never in a public release so they have been deleted. Changed dsa/rsa
11960     utilities to handle the new format: note no releases ever handled public
11961     keys so we should be OK.
11962
11963     The primary motivation for this change is to avoid the same fiasco
11964     that dogs private keys: there are several incompatible private key
11965     formats some of which are standard and some OpenSSL specific and
11966     require various evil hacks to allow partial transparent handling and
11967     even then it doesn't work with DER formats. Given the option anything
11968     other than PKCS#8 should be dumped: but the other formats have to
11969     stay in the name of compatibility.
11970
11971     With public keys and the benefit of hindsight one standard format
11972     is used which works with EVP_PKEY, RSA or DSA structures: though
11973     it clearly returns an error if you try to read the wrong kind of key.
11974
11975     Added a -pubkey option to the 'x509' utility to output the public key.
11976     Also rename the EVP_PKEY_get_*() to EVP_PKEY_rget_*()
11977     (renamed to EVP_PKEY_get1_*() in the OpenSSL 0.9.5 release) and add
11978     EVP_PKEY_rset_*() functions (renamed to EVP_PKEY_set1_*())
11979     that do the same as the EVP_PKEY_assign_*() except they up the
11980     reference count of the added key (they don't "swallow" the
11981     supplied key).
11982     [Steve Henson]
11983
11984  *) Fixes to crypto/x509/by_file.c the code to read in certificates and
11985     CRLs would fail if the file contained no certificates or no CRLs:
11986     added a new function to read in both types and return the number
11987     read: this means that if none are read it will be an error. The
11988     DER versions of the certificate and CRL reader would always fail
11989     because it isn't possible to mix certificates and CRLs in DER format
11990     without choking one or the other routine. Changed this to just read
11991     a certificate: this is the best we can do. Also modified the code
11992     in apps/verify.c to take notice of return codes: it was previously
11993     attempting to read in certificates from NULL pointers and ignoring
11994     any errors: this is one reason why the cert and CRL reader seemed
11995     to work. It doesn't check return codes from the default certificate
11996     routines: these may well fail if the certificates aren't installed.
11997     [Steve Henson]
11998
11999  *) Code to support otherName option in GeneralName.
12000     [Steve Henson]
12001
12002  *) First update to verify code. Change the verify utility
12003     so it warns if it is passed a self signed certificate:
12004     for consistency with the normal behaviour. X509_verify
12005     has been modified to it will now verify a self signed
12006     certificate if *exactly* the same certificate appears
12007     in the store: it was previously impossible to trust a
12008     single self signed certificate. This means that:
12009     openssl verify ss.pem
12010     now gives a warning about a self signed certificate but
12011     openssl verify -CAfile ss.pem ss.pem
12012     is OK.
12013     [Steve Henson]
12014
12015  *) For servers, store verify_result in SSL_SESSION data structure
12016     (and add it to external session representation).
12017     This is needed when client certificate verifications fails,
12018     but an application-provided verification callback (set by
12019     SSL_CTX_set_cert_verify_callback) allows accepting the session
12020     anyway (i.e. leaves x509_store_ctx->error != X509_V_OK
12021     but returns 1): When the session is reused, we have to set
12022     ssl->verify_result to the appropriate error code to avoid
12023     security holes.
12024     [Bodo Moeller, problem pointed out by Lutz Jaenicke]
12025
12026  *) Fix a bug in the new PKCS#7 code: it didn't consider the
12027     case in PKCS7_dataInit() where the signed PKCS7 structure
12028     didn't contain any existing data because it was being created.
12029     [Po-Cheng Chen <pocheng@nst.com.tw>, slightly modified by Steve Henson]
12030
12031  *) Add a salt to the key derivation routines in enc.c. This
12032     forms the first 8 bytes of the encrypted file. Also add a
12033     -S option to allow a salt to be input on the command line.
12034     [Steve Henson]
12035
12036  *) New function X509_cmp(). Oddly enough there wasn't a function
12037     to compare two certificates. We do this by working out the SHA1
12038     hash and comparing that. X509_cmp() will be needed by the trust
12039     code.
12040     [Steve Henson]
12041
12042  *) SSL_get1_session() is like SSL_get_session(), but increments
12043     the reference count in the SSL_SESSION returned.
12044     [Geoff Thorpe <geoff@eu.c2.net>]
12045
12046  *) Fix for 'req': it was adding a null to request attributes.
12047     Also change the X509_LOOKUP and X509_INFO code to handle
12048     certificate auxiliary information.
12049     [Steve Henson]
12050
12051  *) Add support for 40 and 64 bit RC2 and RC4 algorithms: document
12052     the 'enc' command.
12053     [Steve Henson]
12054
12055  *) Add the possibility to add extra information to the memory leak
12056     detecting output, to form tracebacks, showing from where each
12057     allocation was originated: CRYPTO_push_info("constant string") adds
12058     the string plus current file name and line number to a per-thread
12059     stack, CRYPTO_pop_info() does the obvious, CRYPTO_remove_all_info()
12060     is like calling CYRPTO_pop_info() until the stack is empty.
12061     Also updated memory leak detection code to be multi-thread-safe.
12062     [Richard Levitte]
12063
12064  *) Add options -text and -noout to pkcs7 utility and delete the
12065     encryption options which never did anything. Update docs.
12066     [Steve Henson]
12067
12068  *) Add options to some of the utilities to allow the pass phrase
12069     to be included on either the command line (not recommended on
12070     OSes like Unix) or read from the environment. Update the
12071     manpages and fix a few bugs.
12072     [Steve Henson]
12073
12074  *) Add a few manpages for some of the openssl commands.
12075     [Steve Henson]
12076
12077  *) Fix the -revoke option in ca. It was freeing up memory twice,
12078     leaking and not finding already revoked certificates.
12079     [Steve Henson]
12080
12081  *) Extensive changes to support certificate auxiliary information.
12082     This involves the use of X509_CERT_AUX structure and X509_AUX
12083     functions. An X509_AUX function such as PEM_read_X509_AUX()
12084     can still read in a certificate file in the usual way but it
12085     will also read in any additional "auxiliary information". By
12086     doing things this way a fair degree of compatibility can be
12087     retained: existing certificates can have this information added
12088     using the new 'x509' options.
12089
12090     Current auxiliary information includes an "alias" and some trust
12091     settings. The trust settings will ultimately be used in enhanced
12092     certificate chain verification routines: currently a certificate
12093     can only be trusted if it is self signed and then it is trusted
12094     for all purposes.
12095     [Steve Henson]
12096
12097  *) Fix assembler for Alpha (tested only on DEC OSF not Linux or *BSD).
12098     The problem was that one of the replacement routines had not been working
12099     since SSLeay releases.  For now the offending routine has been replaced
12100     with non-optimised assembler.  Even so, this now gives around 95%
12101     performance improvement for 1024 bit RSA signs.
12102     [Mark Cox]
12103
12104  *) Hack to fix PKCS#7 decryption when used with some unorthodox RC2
12105     handling. Most clients have the effective key size in bits equal to
12106     the key length in bits: so a 40 bit RC2 key uses a 40 bit (5 byte) key.
12107     A few however don't do this and instead use the size of the decrypted key
12108     to determine the RC2 key length and the AlgorithmIdentifier to determine
12109     the effective key length. In this case the effective key length can still
12110     be 40 bits but the key length can be 168 bits for example. This is fixed
12111     by manually forcing an RC2 key into the EVP_PKEY structure because the
12112     EVP code can't currently handle unusual RC2 key sizes: it always assumes
12113     the key length and effective key length are equal.
12114     [Steve Henson]
12115
12116  *) Add a bunch of functions that should simplify the creation of
12117     X509_NAME structures. Now you should be able to do:
12118     X509_NAME_add_entry_by_txt(nm, "CN", MBSTRING_ASC, "Steve", -1, -1, 0);
12119     and have it automatically work out the correct field type and fill in
12120     the structures. The more adventurous can try:
12121     X509_NAME_add_entry_by_txt(nm, field, MBSTRING_UTF8, str, -1, -1, 0);
12122     and it will (hopefully) work out the correct multibyte encoding.
12123     [Steve Henson]
12124
12125  *) Change the 'req' utility to use the new field handling and multibyte
12126     copy routines. Before the DN field creation was handled in an ad hoc
12127     way in req, ca, and x509 which was rather broken and didn't support
12128     BMPStrings or UTF8Strings. Since some software doesn't implement
12129     BMPStrings or UTF8Strings yet, they can be enabled using the config file
12130     using the dirstring_type option. See the new comment in the default
12131     openssl.cnf for more info.
12132     [Steve Henson]
12133
12134  *) Make crypto/rand/md_rand.c more robust:
12135     - Assure unique random numbers after fork().
12136     - Make sure that concurrent threads access the global counter and
12137       md serializably so that we never lose entropy in them
12138       or use exactly the same state in multiple threads.
12139       Access to the large state is not always serializable because
12140       the additional locking could be a performance killer, and
12141       md should be large enough anyway.
12142     [Bodo Moeller]
12143
12144  *) New file apps/app_rand.c with commonly needed functionality
12145     for handling the random seed file.
12146
12147     Use the random seed file in some applications that previously did not:
12148          ca,
12149          dsaparam -genkey (which also ignored its '-rand' option),
12150          s_client,
12151          s_server,
12152          x509 (when signing).
12153     Except on systems with /dev/urandom, it is crucial to have a random
12154     seed file at least for key creation, DSA signing, and for DH exchanges;
12155     for RSA signatures we could do without one.
12156
12157     gendh and gendsa (unlike genrsa) used to read only the first byte
12158     of each file listed in the '-rand' option.  The function as previously
12159     found in genrsa is now in app_rand.c and is used by all programs
12160     that support '-rand'.
12161     [Bodo Moeller]
12162
12163  *) In RAND_write_file, use mode 0600 for creating files;
12164     don't just chmod when it may be too late.
12165     [Bodo Moeller]
12166
12167  *) Report an error from X509_STORE_load_locations
12168     when X509_LOOKUP_load_file or X509_LOOKUP_add_dir failed.
12169     [Bill Perry]
12170
12171  *) New function ASN1_mbstring_copy() this copies a string in either
12172     ASCII, Unicode, Universal (4 bytes per character) or UTF8 format
12173     into an ASN1_STRING type. A mask of permissible types is passed
12174     and it chooses the "minimal" type to use or an error if not type
12175     is suitable.
12176     [Steve Henson]
12177
12178  *) Add function equivalents to the various macros in asn1.h. The old
12179     macros are retained with an M_ prefix. Code inside the library can
12180     use the M_ macros. External code (including the openssl utility)
12181     should *NOT* in order to be "shared library friendly".
12182     [Steve Henson]
12183
12184  *) Add various functions that can check a certificate's extensions
12185     to see if it usable for various purposes such as SSL client,
12186     server or S/MIME and CAs of these types. This is currently
12187     VERY EXPERIMENTAL but will ultimately be used for certificate chain
12188     verification. Also added a -purpose flag to x509 utility to
12189     print out all the purposes.
12190     [Steve Henson]
12191
12192  *) Add a CRYPTO_EX_DATA to X509 certificate structure and associated
12193     functions.
12194     [Steve Henson]
12195
12196  *) New X509V3_{X509,CRL,REVOKED}_get_d2i() functions. These will search
12197     for, obtain and decode and extension and obtain its critical flag.
12198     This allows all the necessary extension code to be handled in a
12199     single function call.
12200     [Steve Henson]
12201
12202  *) RC4 tune-up featuring 30-40% performance improvement on most RISC
12203     platforms. See crypto/rc4/rc4_enc.c for further details.
12204     [Andy Polyakov]
12205
12206  *) New -noout option to asn1parse. This causes no output to be produced
12207     its main use is when combined with -strparse and -out to extract data
12208     from a file (which may not be in ASN.1 format).
12209     [Steve Henson]
12210
12211  *) Fix for pkcs12 program. It was hashing an invalid certificate pointer
12212     when producing the local key id.
12213     [Richard Levitte <levitte@stacken.kth.se>]
12214
12215  *) New option -dhparam in s_server. This allows a DH parameter file to be
12216     stated explicitly. If it is not stated then it tries the first server
12217     certificate file. The previous behaviour hard coded the filename
12218     "server.pem".
12219     [Steve Henson]
12220
12221  *) Add -pubin and -pubout options to the rsa and dsa commands. These allow
12222     a public key to be input or output. For example:
12223     openssl rsa -in key.pem -pubout -out pubkey.pem
12224     Also added necessary DSA public key functions to handle this.
12225     [Steve Henson]
12226
12227  *) Fix so PKCS7_dataVerify() doesn't crash if no certificates are contained
12228     in the message. This was handled by allowing
12229     X509_find_by_issuer_and_serial() to tolerate a NULL passed to it.
12230     [Steve Henson, reported by Sampo Kellomaki <sampo@mail.neuronio.pt>]
12231
12232  *) Fix for bug in d2i_ASN1_bytes(): other ASN1 functions add an extra null
12233     to the end of the strings whereas this didn't. This would cause problems
12234     if strings read with d2i_ASN1_bytes() were later modified.
12235     [Steve Henson, reported by Arne Ansper <arne@ats.cyber.ee>]
12236
12237  *) Fix for base64 decode bug. When a base64 bio reads only one line of
12238     data and it contains EOF it will end up returning an error. This is
12239     caused by input 46 bytes long. The cause is due to the way base64
12240     BIOs find the start of base64 encoded data. They do this by trying a
12241     trial decode on each line until they find one that works. When they
12242     do a flag is set and it starts again knowing it can pass all the
12243     data directly through the decoder. Unfortunately it doesn't reset
12244     the context it uses. This means that if EOF is reached an attempt
12245     is made to pass two EOFs through the context and this causes the
12246     resulting error. This can also cause other problems as well. As is
12247     usual with these problems it takes *ages* to find and the fix is
12248     trivial: move one line.
12249     [Steve Henson, reported by ian@uns.ns.ac.yu (Ivan Nejgebauer) ]
12250
12251  *) Ugly workaround to get s_client and s_server working under Windows. The
12252     old code wouldn't work because it needed to select() on sockets and the
12253     tty (for keypresses and to see if data could be written). Win32 only
12254     supports select() on sockets so we select() with a 1s timeout on the
12255     sockets and then see if any characters are waiting to be read, if none
12256     are present then we retry, we also assume we can always write data to
12257     the tty. This isn't nice because the code then blocks until we've
12258     received a complete line of data and it is effectively polling the
12259     keyboard at 1s intervals: however it's quite a bit better than not
12260     working at all :-) A dedicated Windows application might handle this
12261     with an event loop for example.
12262     [Steve Henson]
12263
12264  *) Enhance RSA_METHOD structure. Now there are two extra methods, rsa_sign
12265     and rsa_verify. When the RSA_FLAGS_SIGN_VER option is set these functions
12266     will be called when RSA_sign() and RSA_verify() are used. This is useful
12267     if rsa_pub_dec() and rsa_priv_enc() equivalents are not available.
12268     For this to work properly RSA_public_decrypt() and RSA_private_encrypt()
12269     should *not* be used: RSA_sign() and RSA_verify() must be used instead.
12270     This necessitated the support of an extra signature type NID_md5_sha1
12271     for SSL signatures and modifications to the SSL library to use it instead
12272     of calling RSA_public_decrypt() and RSA_private_encrypt().
12273     [Steve Henson]
12274
12275  *) Add new -verify -CAfile and -CApath options to the crl program, these
12276     will lookup a CRL issuers certificate and verify the signature in a
12277     similar way to the verify program. Tidy up the crl program so it
12278     no longer accesses structures directly. Make the ASN1 CRL parsing a bit
12279     less strict. It will now permit CRL extensions even if it is not
12280     a V2 CRL: this will allow it to tolerate some broken CRLs.
12281     [Steve Henson]
12282
12283  *) Initialize all non-automatic variables each time one of the openssl
12284     sub-programs is started (this is necessary as they may be started
12285     multiple times from the "OpenSSL>" prompt).
12286     [Lennart Bang, Bodo Moeller]
12287
12288  *) Preliminary compilation option RSA_NULL which disables RSA crypto without
12289     removing all other RSA functionality (this is what NO_RSA does). This
12290     is so (for example) those in the US can disable those operations covered
12291     by the RSA patent while allowing storage and parsing of RSA keys and RSA
12292     key generation.
12293     [Steve Henson]
12294
12295  *) Non-copying interface to BIO pairs.
12296     (still largely untested)
12297     [Bodo Moeller]
12298
12299  *) New function ANS1_tag2str() to convert an ASN1 tag to a descriptive
12300     ASCII string. This was handled independently in various places before.
12301     [Steve Henson]
12302
12303  *) New functions UTF8_getc() and UTF8_putc() that parse and generate
12304     UTF8 strings a character at a time.
12305     [Steve Henson]
12306
12307  *) Use client_version from client hello to select the protocol
12308     (s23_srvr.c) and for RSA client key exchange verification
12309     (s3_srvr.c), as required by the SSL 3.0/TLS 1.0 specifications.
12310     [Bodo Moeller]
12311
12312  *) Add various utility functions to handle SPKACs, these were previously
12313     handled by poking round in the structure internals. Added new function
12314     NETSCAPE_SPKI_print() to print out SPKAC and a new utility 'spkac' to
12315     print, verify and generate SPKACs. Based on an original idea from
12316     Massimiliano Pala <madwolf@comune.modena.it> but extensively modified.
12317     [Steve Henson]
12318
12319  *) RIPEMD160 is operational on all platforms and is back in 'make test'.
12320     [Andy Polyakov]
12321
12322  *) Allow the config file extension section to be overwritten on the
12323     command line. Based on an original idea from Massimiliano Pala
12324     <madwolf@comune.modena.it>. The new option is called -extensions
12325     and can be applied to ca, req and x509. Also -reqexts to override
12326     the request extensions in req and -crlexts to override the crl extensions
12327     in ca.
12328     [Steve Henson]
12329
12330  *) Add new feature to the SPKAC handling in ca.  Now you can include
12331     the same field multiple times by preceding it by "XXXX." for example:
12332     1.OU="Unit name 1"
12333     2.OU="Unit name 2"
12334     this is the same syntax as used in the req config file.
12335     [Steve Henson]
12336
12337  *) Allow certificate extensions to be added to certificate requests. These
12338     are specified in a 'req_extensions' option of the req section of the
12339     config file. They can be printed out with the -text option to req but
12340     are otherwise ignored at present.
12341     [Steve Henson]
12342
12343  *) Fix a horrible bug in enc_read() in crypto/evp/bio_enc.c: if the first
12344     data read consists of only the final block it would not decrypted because
12345     EVP_CipherUpdate() would correctly report zero bytes had been decrypted.
12346     A misplaced 'break' also meant the decrypted final block might not be
12347     copied until the next read.
12348     [Steve Henson]
12349
12350  *) Initial support for DH_METHOD. Again based on RSA_METHOD. Also added
12351     a few extra parameters to the DH structure: these will be useful if
12352     for example we want the value of 'q' or implement X9.42 DH.
12353     [Steve Henson]
12354
12355  *) Initial support for DSA_METHOD. This is based on the RSA_METHOD and
12356     provides hooks that allow the default DSA functions or functions on a
12357     "per key" basis to be replaced. This allows hardware acceleration and
12358     hardware key storage to be handled without major modification to the
12359     library. Also added low level modexp hooks and CRYPTO_EX structure and
12360     associated functions.
12361     [Steve Henson]
12362
12363  *) Add a new flag to memory BIOs, BIO_FLAG_MEM_RDONLY. This marks the BIO
12364     as "read only": it can't be written to and the buffer it points to will
12365     not be freed. Reading from a read only BIO is much more efficient than
12366     a normal memory BIO. This was added because there are several times when
12367     an area of memory needs to be read from a BIO. The previous method was
12368     to create a memory BIO and write the data to it, this results in two
12369     copies of the data and an O(n^2) reading algorithm. There is a new
12370     function BIO_new_mem_buf() which creates a read only memory BIO from
12371     an area of memory. Also modified the PKCS#7 routines to use read only
12372     memory BIOs.
12373     [Steve Henson]
12374
12375  *) Bugfix: ssl23_get_client_hello did not work properly when called in
12376     state SSL23_ST_SR_CLNT_HELLO_B, i.e. when the first 7 bytes of
12377     a SSLv2-compatible client hello for SSLv3 or TLSv1 could be read,
12378     but a retry condition occurred while trying to read the rest.
12379     [Bodo Moeller]
12380
12381  *) The PKCS7_ENC_CONTENT_new() function was setting the content type as
12382     NID_pkcs7_encrypted by default: this was wrong since this should almost
12383     always be NID_pkcs7_data. Also modified the PKCS7_set_type() to handle
12384     the encrypted data type: this is a more sensible place to put it and it
12385     allows the PKCS#12 code to be tidied up that duplicated this
12386     functionality.
12387     [Steve Henson]
12388
12389  *) Changed obj_dat.pl script so it takes its input and output files on
12390     the command line. This should avoid shell escape redirection problems
12391     under Win32.
12392     [Steve Henson]
12393
12394  *) Initial support for certificate extension requests, these are included
12395     in things like Xenroll certificate requests. Included functions to allow
12396     extensions to be obtained and added.
12397     [Steve Henson]
12398
12399  *) -crlf option to s_client and s_server for sending newlines as
12400     CRLF (as required by many protocols).
12401     [Bodo Moeller]
12402
12403 Changes between 0.9.3a and 0.9.4  [09 Aug 1999]
12404
12405  *) Install libRSAglue.a when OpenSSL is built with RSAref.
12406     [Ralf S. Engelschall]
12407
12408  *) A few more ``#ifndef NO_FP_API / #endif'' pairs for consistency.
12409     [Andrija Antonijevic <TheAntony2@bigfoot.com>]
12410
12411  *) Fix -startdate and -enddate (which was missing) arguments to 'ca'
12412     program.
12413     [Steve Henson]
12414
12415  *) New function DSA_dup_DH, which duplicates DSA parameters/keys as
12416     DH parameters/keys (q is lost during that conversion, but the resulting
12417     DH parameters contain its length).
12418
12419     For 1024-bit p, DSA_generate_parameters followed by DSA_dup_DH is
12420     much faster than DH_generate_parameters (which creates parameters
12421     where p = 2*q + 1), and also the smaller q makes DH computations
12422     much more efficient (160-bit exponentiation instead of 1024-bit
12423     exponentiation); so this provides a convenient way to support DHE
12424     ciphersuites in SSL/TLS servers (see ssl/ssltest.c).  It is of
12425     utter importance to use
12426         SSL_CTX_set_options(s_ctx, SSL_OP_SINGLE_DH_USE);
12427     or
12428         SSL_set_options(s_ctx, SSL_OP_SINGLE_DH_USE);
12429     when such DH parameters are used, because otherwise small subgroup
12430     attacks may become possible!
12431     [Bodo Moeller]
12432
12433  *) Avoid memory leak in i2d_DHparams.
12434     [Bodo Moeller]
12435
12436  *) Allow the -k option to be used more than once in the enc program:
12437     this allows the same encrypted message to be read by multiple recipients.
12438     [Steve Henson]
12439
12440  *) New function OBJ_obj2txt(buf, buf_len, a, no_name), this converts
12441     an ASN1_OBJECT to a text string. If the "no_name" parameter is set then
12442     it will always use the numerical form of the OID, even if it has a short
12443     or long name.
12444     [Steve Henson]
12445
12446  *) Added an extra RSA flag: RSA_FLAG_EXT_PKEY. Previously the rsa_mod_exp
12447     method only got called if p,q,dmp1,dmq1,iqmp components were present,
12448     otherwise bn_mod_exp was called. In the case of hardware keys for example
12449     no private key components need be present and it might store extra data
12450     in the RSA structure, which cannot be accessed from bn_mod_exp.
12451     By setting RSA_FLAG_EXT_PKEY rsa_mod_exp will always be called for
12452     private key operations.
12453     [Steve Henson]
12454
12455  *) Added support for SPARC Linux.
12456     [Andy Polyakov]
12457
12458  *) pem_password_cb function type incompatibly changed from
12459          typedef int pem_password_cb(char *buf, int size, int rwflag);
12460     to
12461          ....(char *buf, int size, int rwflag, void *userdata);
12462     so that applications can pass data to their callbacks:
12463     The PEM[_ASN1]_{read,write}... functions and macros now take an
12464     additional void * argument, which is just handed through whenever
12465     the password callback is called.
12466     [Damien Miller <dmiller@ilogic.com.au>; tiny changes by Bodo Moeller]
12467
12468     New function SSL_CTX_set_default_passwd_cb_userdata.
12469
12470     Compatibility note: As many C implementations push function arguments
12471     onto the stack in reverse order, the new library version is likely to
12472     interoperate with programs that have been compiled with the old
12473     pem_password_cb definition (PEM_whatever takes some data that
12474     happens to be on the stack as its last argument, and the callback
12475     just ignores this garbage); but there is no guarantee whatsoever that
12476     this will work.
12477
12478  *) The -DPLATFORM="\"$(PLATFORM)\"" definition and the similar -DCFLAGS=...
12479     (both in crypto/Makefile.ssl for use by crypto/cversion.c) caused
12480     problems not only on Windows, but also on some Unix platforms.
12481     To avoid problematic command lines, these definitions are now in an
12482     auto-generated file crypto/buildinf.h (created by crypto/Makefile.ssl
12483     for standard "make" builds, by util/mk1mf.pl for "mk1mf" builds).
12484     [Bodo Moeller]
12485
12486  *) MIPS III/IV assembler module is reimplemented.
12487     [Andy Polyakov]
12488
12489  *) More DES library cleanups: remove references to srand/rand and
12490     delete an unused file.
12491     [Ulf Möller]
12492
12493  *) Add support for the free Netwide assembler (NASM) under Win32,
12494     since not many people have MASM (ml) and it can be hard to obtain.
12495     This is currently experimental but it seems to work OK and pass all
12496     the tests. Check out INSTALL.W32 for info.
12497     [Steve Henson]
12498
12499  *) Fix memory leaks in s3_clnt.c: All non-anonymous SSL3/TLS1 connections
12500     without temporary keys kept an extra copy of the server key,
12501     and connections with temporary keys did not free everything in case
12502     of an error.
12503     [Bodo Moeller]
12504
12505  *) New function RSA_check_key and new openssl rsa option -check
12506     for verifying the consistency of RSA keys.
12507     [Ulf Moeller, Bodo Moeller]
12508
12509  *) Various changes to make Win32 compile work:
12510     1. Casts to avoid "loss of data" warnings in p5_crpt2.c
12511     2. Change unsigned int to int in b_dump.c to avoid "signed/unsigned
12512        comparison" warnings.
12513     3. Add sk_<TYPE>_sort to DEF file generator and do make update.
12514     [Steve Henson]
12515
12516  *) Add a debugging option to PKCS#5 v2 key generation function: when
12517     you #define DEBUG_PKCS5V2 passwords, salts, iteration counts and
12518     derived keys are printed to stderr.
12519     [Steve Henson]
12520
12521  *) Copy the flags in ASN1_STRING_dup().
12522     [Roman E. Pavlov <pre@mo.msk.ru>]
12523
12524  *) The x509 application mishandled signing requests containing DSA
12525     keys when the signing key was also DSA and the parameters didn't match.
12526
12527     It was supposed to omit the parameters when they matched the signing key:
12528     the verifying software was then supposed to automatically use the CA's
12529     parameters if they were absent from the end user certificate.
12530
12531     Omitting parameters is no longer recommended. The test was also
12532     the wrong way round! This was probably due to unusual behaviour in
12533     EVP_cmp_parameters() which returns 1 if the parameters match.
12534     This meant that parameters were omitted when they *didn't* match and
12535     the certificate was useless. Certificates signed with 'ca' didn't have
12536     this bug.
12537     [Steve Henson, reported by Doug Erickson <Doug.Erickson@Part.NET>]
12538
12539  *) Memory leak checking (-DCRYPTO_MDEBUG) had some problems.
12540     The interface is as follows:
12541     Applications can use
12542         CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_ON) aka MemCheck_start(),
12543         CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_OFF) aka MemCheck_stop();
12544     "off" is now the default.
12545     The library internally uses
12546         CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_DISABLE) aka MemCheck_off(),
12547         CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_ENABLE) aka MemCheck_on()
12548     to disable memory-checking temporarily.
12549
12550     Some inconsistent states that previously were possible (and were
12551     even the default) are now avoided.
12552
12553     -DCRYPTO_MDEBUG_TIME is new and additionally stores the current time
12554     with each memory chunk allocated; this is occasionally more helpful
12555     than just having a counter.
12556
12557     -DCRYPTO_MDEBUG_THREAD is also new and adds the thread ID.
12558
12559     -DCRYPTO_MDEBUG_ALL enables all of the above, plus any future
12560     extensions.
12561     [Bodo Moeller]
12562
12563  *) Introduce "mode" for SSL structures (with defaults in SSL_CTX),
12564     which largely parallels "options", but is for changing API behaviour,
12565     whereas "options" are about protocol behaviour.
12566     Initial "mode" flags are:
12567
12568     SSL_MODE_ENABLE_PARTIAL_WRITE   Allow SSL_write to report success when
12569                                     a single record has been written.
12570     SSL_MODE_ACCEPT_MOVING_WRITE_BUFFER  Don't insist that SSL_write
12571                                     retries use the same buffer location.
12572                                     (But all of the contents must be
12573                                     copied!)
12574     [Bodo Moeller]
12575
12576  *) Bugfix: SSL_set_options ignored its parameter, only SSL_CTX_set_options
12577     worked.
12578
12579  *) Fix problems with no-hmac etc.
12580     [Ulf Möller, pointed out by Brian Wellington <bwelling@tislabs.com>]
12581
12582  *) New functions RSA_get_default_method(), RSA_set_method() and
12583     RSA_get_method(). These allows replacement of RSA_METHODs without having
12584     to mess around with the internals of an RSA structure.
12585     [Steve Henson]
12586
12587  *) Fix memory leaks in DSA_do_sign and DSA_is_prime.
12588     Also really enable memory leak checks in openssl.c and in some
12589     test programs.
12590     [Chad C. Mulligan, Bodo Moeller]
12591
12592  *) Fix a bug in d2i_ASN1_INTEGER() and i2d_ASN1_INTEGER() which can mess
12593     up the length of negative integers. This has now been simplified to just
12594     store the length when it is first determined and use it later, rather
12595     than trying to keep track of where data is copied and updating it to
12596     point to the end.
12597     [Steve Henson, reported by Brien Wheeler
12598      <bwheeler@authentica-security.com>]
12599
12600  *) Add a new function PKCS7_signatureVerify. This allows the verification
12601     of a PKCS#7 signature but with the signing certificate passed to the
12602     function itself. This contrasts with PKCS7_dataVerify which assumes the
12603     certificate is present in the PKCS#7 structure. This isn't always the
12604     case: certificates can be omitted from a PKCS#7 structure and be
12605     distributed by "out of band" means (such as a certificate database).
12606     [Steve Henson]
12607
12608  *) Complete the PEM_* macros with DECLARE_PEM versions to replace the
12609     function prototypes in pem.h, also change util/mkdef.pl to add the
12610     necessary function names.
12611     [Steve Henson]
12612
12613  *) mk1mf.pl (used by Windows builds) did not properly read the
12614     options set by Configure in the top level Makefile, and Configure
12615     was not even able to write more than one option correctly.
12616     Fixed, now "no-idea no-rc5 -DCRYPTO_MDEBUG" etc. works as intended.
12617     [Bodo Moeller]
12618
12619  *) New functions CONF_load_bio() and CONF_load_fp() to allow a config
12620     file to be loaded from a BIO or FILE pointer. The BIO version will
12621     for example allow memory BIOs to contain config info.
12622     [Steve Henson]
12623
12624  *) New function "CRYPTO_num_locks" that returns CRYPTO_NUM_LOCKS.
12625     Whoever hopes to achieve shared-library compatibility across versions
12626     must use this, not the compile-time macro.
12627     (Exercise 0.9.4: Which is the minimum library version required by
12628     such programs?)
12629     Note: All this applies only to multi-threaded programs, others don't
12630     need locks.
12631     [Bodo Moeller]
12632
12633  *) Add missing case to s3_clnt.c state machine -- one of the new SSL tests
12634     through a BIO pair triggered the default case, i.e.
12635     SSLerr(...,SSL_R_UNKNOWN_STATE).
12636     [Bodo Moeller]
12637
12638  *) New "BIO pair" concept (crypto/bio/bss_bio.c) so that applications
12639     can use the SSL library even if none of the specific BIOs is
12640     appropriate.
12641     [Bodo Moeller]
12642
12643  *) Fix a bug in i2d_DSAPublicKey() which meant it returned the wrong value
12644     for the encoded length.
12645     [Jeon KyoungHo <khjeon@sds.samsung.co.kr>]
12646
12647  *) Add initial documentation of the X509V3 functions.
12648     [Steve Henson]
12649
12650  *) Add a new pair of functions PEM_write_PKCS8PrivateKey() and
12651     PEM_write_bio_PKCS8PrivateKey() that are equivalent to
12652     PEM_write_PrivateKey() and PEM_write_bio_PrivateKey() but use the more
12653     secure PKCS#8 private key format with a high iteration count.
12654     [Steve Henson]
12655
12656  *) Fix determination of Perl interpreter: A perl or perl5
12657     _directory_ in $PATH was also accepted as the interpreter.
12658     [Ralf S. Engelschall]
12659
12660  *) Fix demos/sign/sign.c: well there wasn't anything strictly speaking
12661     wrong with it but it was very old and did things like calling
12662     PEM_ASN1_read() directly and used MD5 for the hash not to mention some
12663     unusual formatting.
12664     [Steve Henson]
12665
12666  *) Fix demos/selfsign.c: it used obsolete and deleted functions, changed
12667     to use the new extension code.
12668     [Steve Henson]
12669
12670  *) Implement the PEM_read/PEM_write functions in crypto/pem/pem_all.c
12671     with macros. This should make it easier to change their form, add extra
12672     arguments etc. Fix a few PEM prototypes which didn't have cipher as a
12673     constant.
12674     [Steve Henson]
12675
12676  *) Add to configuration table a new entry that can specify an alternative
12677     name for unistd.h (for pre-POSIX systems); we need this for NeXTstep,
12678     according to Mark Crispin <MRC@Panda.COM>.
12679     [Bodo Moeller]
12680
12681#if 0
12682  *) DES CBC did not update the IV. Weird.
12683     [Ben Laurie]
12684#else
12685     des_cbc_encrypt does not update the IV, but des_ncbc_encrypt does.
12686     Changing the behaviour of the former might break existing programs --
12687     where IV updating is needed, des_ncbc_encrypt can be used.
12688#endif
12689
12690  *) When bntest is run from "make test" it drives bc to check its
12691     calculations, as well as internally checking them. If an internal check
12692     fails, it needs to cause bc to give a non-zero result or make test carries
12693     on without noticing the failure. Fixed.
12694     [Ben Laurie]
12695
12696  *) DES library cleanups.
12697     [Ulf Möller]
12698
12699  *) Add support for PKCS#5 v2.0 PBE algorithms. This will permit PKCS#8 to be
12700     used with any cipher unlike PKCS#5 v1.5 which can at most handle 64 bit
12701     ciphers. NOTE: although the key derivation function has been verified
12702     against some published test vectors it has not been extensively tested
12703     yet. Added a -v2 "cipher" option to pkcs8 application to allow the use
12704     of v2.0.
12705     [Steve Henson]
12706
12707  *) Instead of "mkdir -p", which is not fully portable, use new
12708     Perl script "util/mkdir-p.pl".
12709     [Bodo Moeller]
12710
12711  *) Rewrite the way password based encryption (PBE) is handled. It used to
12712     assume that the ASN1 AlgorithmIdentifier parameter was a PBEParameter
12713     structure. This was true for the PKCS#5 v1.5 and PKCS#12 PBE algorithms
12714     but doesn't apply to PKCS#5 v2.0 where it can be something else. Now
12715     the 'parameter' field of the AlgorithmIdentifier is passed to the
12716     underlying key generation function so it must do its own ASN1 parsing.
12717     This has also changed the EVP_PBE_CipherInit() function which now has a
12718     'parameter' argument instead of literal salt and iteration count values
12719     and the function EVP_PBE_ALGOR_CipherInit() has been deleted.
12720     [Steve Henson]
12721
12722  *) Support for PKCS#5 v1.5 compatible password based encryption algorithms
12723     and PKCS#8 functionality. New 'pkcs8' application linked to openssl.
12724     Needed to change the PEM_STRING_EVP_PKEY value which was just "PRIVATE
12725     KEY" because this clashed with PKCS#8 unencrypted string. Since this
12726     value was just used as a "magic string" and not used directly its
12727     value doesn't matter.
12728     [Steve Henson]
12729
12730  *) Introduce some semblance of const correctness to BN. Shame C doesn't
12731     support mutable.
12732     [Ben Laurie]
12733
12734  *) "linux-sparc64" configuration (ultrapenguin).
12735     [Ray Miller <ray.miller@oucs.ox.ac.uk>]
12736     "linux-sparc" configuration.
12737     [Christian Forster <fo@hawo.stw.uni-erlangen.de>]
12738
12739  *) config now generates no-xxx options for missing ciphers.
12740     [Ulf Möller]
12741
12742  *) Support the EBCDIC character set (work in progress).
12743     File ebcdic.c not yet included because it has a different license.
12744     [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>]
12745
12746  *) Support BS2000/OSD-POSIX.
12747     [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>]
12748
12749  *) Make callbacks for key generation use void * instead of char *.
12750     [Ben Laurie]
12751
12752  *) Make S/MIME samples compile (not yet tested).
12753     [Ben Laurie]
12754
12755  *) Additional typesafe stacks.
12756     [Ben Laurie]
12757
12758  *) New configuration variants "bsdi-elf-gcc" (BSD/OS 4.x).
12759     [Bodo Moeller]
12760
12761
12762 Changes between 0.9.3 and 0.9.3a  [29 May 1999]
12763
12764  *) New configuration variant "sco5-gcc".
12765
12766  *) Updated some demos.
12767     [Sean O Riordain, Wade Scholine]
12768
12769  *) Add missing BIO_free at exit of pkcs12 application.
12770     [Wu Zhigang]
12771
12772  *) Fix memory leak in conf.c.
12773     [Steve Henson]
12774
12775  *) Updates for Win32 to assembler version of MD5.
12776     [Steve Henson]
12777
12778  *) Set #! path to perl in apps/der_chop to where we found it
12779     instead of using a fixed path.
12780     [Bodo Moeller]
12781
12782  *) SHA library changes for irix64-mips4-cc.
12783     [Andy Polyakov]
12784
12785  *) Improvements for VMS support.
12786     [Richard Levitte]
12787
12788
12789 Changes between 0.9.2b and 0.9.3  [24 May 1999]
12790
12791  *) Bignum library bug fix. IRIX 6 passes "make test" now!
12792     This also avoids the problems with SC4.2 and unpatched SC5.
12793     [Andy Polyakov <appro@fy.chalmers.se>]
12794
12795  *) New functions sk_num, sk_value and sk_set to replace the previous macros.
12796     These are required because of the typesafe stack would otherwise break
12797     existing code. If old code used a structure member which used to be STACK
12798     and is now STACK_OF (for example cert in a PKCS7_SIGNED structure) with
12799     sk_num or sk_value it would produce an error because the num, data members
12800     are not present in STACK_OF. Now it just produces a warning. sk_set
12801     replaces the old method of assigning a value to sk_value
12802     (e.g. sk_value(x, i) = y) which the library used in a few cases. Any code
12803     that does this will no longer work (and should use sk_set instead) but
12804     this could be regarded as a "questionable" behaviour anyway.
12805     [Steve Henson]
12806
12807  *) Fix most of the other PKCS#7 bugs. The "experimental" code can now
12808     correctly handle encrypted S/MIME data.
12809     [Steve Henson]
12810
12811  *) Change type of various DES function arguments from des_cblock
12812     (which means, in function argument declarations, pointer to char)
12813     to des_cblock * (meaning pointer to array with 8 char elements),
12814     which allows the compiler to do more typechecking; it was like
12815     that back in SSLeay, but with lots of ugly casts.
12816
12817     Introduce new type const_des_cblock.
12818     [Bodo Moeller]
12819
12820  *) Reorganise the PKCS#7 library and get rid of some of the more obvious
12821     problems: find RecipientInfo structure that matches recipient certificate
12822     and initialise the ASN1 structures properly based on passed cipher.
12823     [Steve Henson]
12824
12825  *) Belatedly make the BN tests actually check the results.
12826     [Ben Laurie]
12827
12828  *) Fix the encoding and decoding of negative ASN1 INTEGERS and conversion
12829     to and from BNs: it was completely broken. New compilation option
12830     NEG_PUBKEY_BUG to allow for some broken certificates that encode public
12831     key elements as negative integers.
12832     [Steve Henson]
12833
12834  *) Reorganize and speed up MD5.
12835     [Andy Polyakov <appro@fy.chalmers.se>]
12836
12837  *) VMS support.
12838     [Richard Levitte <richard@levitte.org>]
12839
12840  *) New option -out to asn1parse to allow the parsed structure to be
12841     output to a file. This is most useful when combined with the -strparse
12842     option to examine the output of things like OCTET STRINGS.
12843     [Steve Henson]
12844
12845  *) Make SSL library a little more fool-proof by not requiring any longer
12846     that SSL_set_{accept,connect}_state be called before
12847     SSL_{accept,connect} may be used (SSL_set_..._state is omitted
12848     in many applications because usually everything *appeared* to work as
12849     intended anyway -- now it really works as intended).
12850     [Bodo Moeller]
12851
12852  *) Move openssl.cnf out of lib/.
12853     [Ulf Möller]
12854
12855  *) Fix various things to let OpenSSL even pass ``egcc -pipe -O2 -Wall
12856     -Wshadow -Wpointer-arith -Wcast-align -Wmissing-prototypes
12857     -Wmissing-declarations -Wnested-externs -Winline'' with EGCS 1.1.2+
12858     [Ralf S. Engelschall]
12859
12860  *) Various fixes to the EVP and PKCS#7 code. It may now be able to
12861     handle PKCS#7 enveloped data properly.
12862     [Sebastian Akerman <sak@parallelconsulting.com>, modified by Steve]
12863
12864  *) Create a duplicate of the SSL_CTX's CERT in SSL_new instead of
12865     copying pointers.  The cert_st handling is changed by this in
12866     various ways (and thus what used to be known as ctx->default_cert
12867     is now called ctx->cert, since we don't resort to s->ctx->[default_]cert
12868     any longer when s->cert does not give us what we need).
12869     ssl_cert_instantiate becomes obsolete by this change.
12870     As soon as we've got the new code right (possibly it already is?),
12871     we have solved a couple of bugs of the earlier code where s->cert
12872     was used as if it could not have been shared with other SSL structures.
12873
12874     Note that using the SSL API in certain dirty ways now will result
12875     in different behaviour than observed with earlier library versions:
12876     Changing settings for an SSL_CTX *ctx after having done s = SSL_new(ctx)
12877     does not influence s as it used to.
12878
12879     In order to clean up things more thoroughly, inside SSL_SESSION
12880     we don't use CERT any longer, but a new structure SESS_CERT
12881     that holds per-session data (if available); currently, this is
12882     the peer's certificate chain and, for clients, the server's certificate
12883     and temporary key.  CERT holds only those values that can have
12884     meaningful defaults in an SSL_CTX.
12885     [Bodo Moeller]
12886
12887  *) New function X509V3_EXT_i2d() to create an X509_EXTENSION structure
12888     from the internal representation. Various PKCS#7 fixes: remove some
12889     evil casts and set the enc_dig_alg field properly based on the signing
12890     key type.
12891     [Steve Henson]
12892
12893  *) Allow PKCS#12 password to be set from the command line or the
12894     environment. Let 'ca' get its config file name from the environment
12895     variables "OPENSSL_CONF" or "SSLEAY_CONF" (for consistency with 'req'
12896     and 'x509').
12897     [Steve Henson]
12898
12899  *) Allow certificate policies extension to use an IA5STRING for the
12900     organization field. This is contrary to the PKIX definition but
12901     VeriSign uses it and IE5 only recognises this form. Document 'x509'
12902     extension option.
12903     [Steve Henson]
12904
12905  *) Add PEDANTIC compiler flag to allow compilation with gcc -pedantic,
12906     without disallowing inline assembler and the like for non-pedantic builds.
12907     [Ben Laurie]
12908
12909  *) Support Borland C++ builder.
12910     [Janez Jere <jj@void.si>, modified by Ulf Möller]
12911
12912  *) Support Mingw32.
12913     [Ulf Möller]
12914
12915  *) SHA-1 cleanups and performance enhancements.
12916     [Andy Polyakov <appro@fy.chalmers.se>]
12917
12918  *) Sparc v8plus assembler for the bignum library.
12919     [Andy Polyakov <appro@fy.chalmers.se>]
12920
12921  *) Accept any -xxx and +xxx compiler options in Configure.
12922     [Ulf Möller]
12923
12924  *) Update HPUX configuration.
12925     [Anonymous]
12926
12927  *) Add missing sk_<type>_unshift() function to safestack.h
12928     [Ralf S. Engelschall]
12929
12930  *) New function SSL_CTX_use_certificate_chain_file that sets the
12931     "extra_cert"s in addition to the certificate.  (This makes sense
12932     only for "PEM" format files, as chains as a whole are not
12933     DER-encoded.)
12934     [Bodo Moeller]
12935
12936  *) Support verify_depth from the SSL API.
12937     x509_vfy.c had what can be considered an off-by-one-error:
12938     Its depth (which was not part of the external interface)
12939     was actually counting the number of certificates in a chain;
12940     now it really counts the depth.
12941     [Bodo Moeller]
12942
12943  *) Bugfix in crypto/x509/x509_cmp.c: The SSLerr macro was used
12944     instead of X509err, which often resulted in confusing error
12945     messages since the error codes are not globally unique
12946     (e.g. an alleged error in ssl3_accept when a certificate
12947     didn't match the private key).
12948
12949  *) New function SSL_CTX_set_session_id_context that allows to set a default
12950     value (so that you don't need SSL_set_session_id_context for each
12951     connection using the SSL_CTX).
12952     [Bodo Moeller]
12953
12954  *) OAEP decoding bug fix.
12955     [Ulf Möller]
12956
12957  *) Support INSTALL_PREFIX for package builders, as proposed by
12958     David Harris.
12959     [Bodo Moeller]
12960
12961  *) New Configure options "threads" and "no-threads".  For systems
12962     where the proper compiler options are known (currently Solaris
12963     and Linux), "threads" is the default.
12964     [Bodo Moeller]
12965
12966  *) New script util/mklink.pl as a faster substitute for util/mklink.sh.
12967     [Bodo Moeller]
12968
12969  *) Install various scripts to $(OPENSSLDIR)/misc, not to
12970     $(INSTALLTOP)/bin -- they shouldn't clutter directories
12971     such as /usr/local/bin.
12972     [Bodo Moeller]
12973
12974  *) "make linux-shared" to build shared libraries.
12975     [Niels Poppe <niels@netbox.org>]
12976
12977  *) New Configure option no-<cipher> (rsa, idea, rc5, ...).
12978     [Ulf Möller]
12979
12980  *) Add the PKCS#12 API documentation to openssl.txt. Preliminary support for
12981     extension adding in x509 utility.
12982     [Steve Henson]
12983
12984  *) Remove NOPROTO sections and error code comments.
12985     [Ulf Möller]
12986
12987  *) Partial rewrite of the DEF file generator to now parse the ANSI
12988     prototypes.
12989     [Steve Henson]
12990
12991  *) New Configure options --prefix=DIR and --openssldir=DIR.
12992     [Ulf Möller]
12993
12994  *) Complete rewrite of the error code script(s). It is all now handled
12995     by one script at the top level which handles error code gathering,
12996     header rewriting and C source file generation. It should be much better
12997     than the old method: it now uses a modified version of Ulf's parser to
12998     read the ANSI prototypes in all header files (thus the old K&R definitions
12999     aren't needed for error creation any more) and do a better job of
13000     translating function codes into names. The old 'ASN1 error code embedded
13001     in a comment' is no longer necessary and it doesn't use .err files which
13002     have now been deleted. Also the error code call doesn't have to appear all
13003     on one line (which resulted in some large lines...).
13004     [Steve Henson]
13005
13006  *) Change #include filenames from <foo.h> to <openssl/foo.h>.
13007     [Bodo Moeller]
13008
13009  *) Change behaviour of ssl2_read when facing length-0 packets: Don't return
13010     0 (which usually indicates a closed connection), but continue reading.
13011     [Bodo Moeller]
13012
13013  *) Fix some race conditions.
13014     [Bodo Moeller]
13015
13016  *) Add support for CRL distribution points extension. Add Certificate
13017     Policies and CRL distribution points documentation.
13018     [Steve Henson]
13019
13020  *) Move the autogenerated header file parts to crypto/opensslconf.h.
13021     [Ulf Möller]
13022
13023  *) Fix new 56-bit DES export ciphersuites: they were using 7 bytes instead of
13024     8 of keying material. Merlin has also confirmed interop with this fix
13025     between OpenSSL and Baltimore C/SSL 2.0 and J/SSL 2.0.
13026     [Merlin Hughes <merlin@baltimore.ie>]
13027
13028  *) Fix lots of warnings.
13029     [Richard Levitte <levitte@stacken.kth.se>]
13030
13031  *) In add_cert_dir() in crypto/x509/by_dir.c, break out of the loop if
13032     the directory spec didn't end with a LIST_SEPARATOR_CHAR.
13033     [Richard Levitte <levitte@stacken.kth.se>]
13034
13035  *) Fix problems with sizeof(long) == 8.
13036     [Andy Polyakov <appro@fy.chalmers.se>]
13037
13038  *) Change functions to ANSI C.
13039     [Ulf Möller]
13040
13041  *) Fix typos in error codes.
13042     [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>, Ulf Möller]
13043
13044  *) Remove defunct assembler files from Configure.
13045     [Ulf Möller]
13046
13047  *) SPARC v8 assembler BIGNUM implementation.
13048     [Andy Polyakov <appro@fy.chalmers.se>]
13049
13050  *) Support for Certificate Policies extension: both print and set.
13051     Various additions to support the r2i method this uses.
13052     [Steve Henson]
13053
13054  *) A lot of constification, and fix a bug in X509_NAME_oneline() that could
13055     return a const string when you are expecting an allocated buffer.
13056     [Ben Laurie]
13057
13058  *) Add support for ASN1 types UTF8String and VISIBLESTRING, also the CHOICE
13059     types DirectoryString and DisplayText.
13060     [Steve Henson]
13061
13062  *) Add code to allow r2i extensions to access the configuration database,
13063     add an LHASH database driver and add several ctx helper functions.
13064     [Steve Henson]
13065
13066  *) Fix an evil bug in bn_expand2() which caused various BN functions to
13067     fail when they extended the size of a BIGNUM.
13068     [Steve Henson]
13069
13070  *) Various utility functions to handle SXNet extension. Modify mkdef.pl to
13071     support typesafe stack.
13072     [Steve Henson]
13073
13074  *) Fix typo in SSL_[gs]et_options().
13075     [Nils Frostberg <nils@medcom.se>]
13076
13077  *) Delete various functions and files that belonged to the (now obsolete)
13078     old X509V3 handling code.
13079     [Steve Henson]
13080
13081  *) New Configure option "rsaref".
13082     [Ulf Möller]
13083
13084  *) Don't auto-generate pem.h.
13085     [Bodo Moeller]
13086
13087  *) Introduce type-safe ASN.1 SETs.
13088     [Ben Laurie]
13089
13090  *) Convert various additional casted stacks to type-safe STACK_OF() variants.
13091     [Ben Laurie, Ralf S. Engelschall, Steve Henson]
13092
13093  *) Introduce type-safe STACKs. This will almost certainly break lots of code
13094     that links with OpenSSL (well at least cause lots of warnings), but fear
13095     not: the conversion is trivial, and it eliminates loads of evil casts. A
13096     few STACKed things have been converted already. Feel free to convert more.
13097     In the fullness of time, I'll do away with the STACK type altogether.
13098     [Ben Laurie]
13099
13100  *) Add `openssl ca -revoke <certfile>' facility which revokes a certificate
13101     specified in <certfile> by updating the entry in the index.txt file.
13102     This way one no longer has to edit the index.txt file manually for
13103     revoking a certificate. The -revoke option does the gory details now.
13104     [Massimiliano Pala <madwolf@openca.org>, Ralf S. Engelschall]
13105
13106  *) Fix `openssl crl -noout -text' combination where `-noout' killed the
13107     `-text' option at all and this way the `-noout -text' combination was
13108     inconsistent in `openssl crl' with the friends in `openssl x509|rsa|dsa'.
13109     [Ralf S. Engelschall]
13110
13111  *) Make sure a corresponding plain text error message exists for the
13112     X509_V_ERR_CERT_REVOKED/23 error number which can occur when a
13113     verify callback function determined that a certificate was revoked.
13114     [Ralf S. Engelschall]
13115
13116  *) Bugfix: In test/testenc, don't test "openssl <cipher>" for
13117     ciphers that were excluded, e.g. by -DNO_IDEA.  Also, test
13118     all available ciphers including rc5, which was forgotten until now.
13119     In order to let the testing shell script know which algorithms
13120     are available, a new (up to now undocumented) command
13121     "openssl list-cipher-commands" is used.
13122     [Bodo Moeller]
13123
13124  *) Bugfix: s_client occasionally would sleep in select() when
13125     it should have checked SSL_pending() first.
13126     [Bodo Moeller]
13127
13128  *) New functions DSA_do_sign and DSA_do_verify to provide access to
13129     the raw DSA values prior to ASN.1 encoding.
13130     [Ulf Möller]
13131
13132  *) Tweaks to Configure
13133     [Niels Poppe <niels@netbox.org>]
13134
13135  *) Add support for PKCS#5 v2.0 ASN1 PBES2 structures. No other support,
13136     yet...
13137     [Steve Henson]
13138
13139  *) New variables $(RANLIB) and $(PERL) in the Makefiles.
13140     [Ulf Möller]
13141
13142  *) New config option to avoid instructions that are illegal on the 80386.
13143     The default code is faster, but requires at least a 486.
13144     [Ulf Möller]
13145
13146  *) Got rid of old SSL2_CLIENT_VERSION (inconsistently used) and
13147     SSL2_SERVER_VERSION (not used at all) macros, which are now the
13148     same as SSL2_VERSION anyway.
13149     [Bodo Moeller]
13150
13151  *) New "-showcerts" option for s_client.
13152     [Bodo Moeller]
13153
13154  *) Still more PKCS#12 integration. Add pkcs12 application to openssl
13155     application. Various cleanups and fixes.
13156     [Steve Henson]
13157
13158  *) More PKCS#12 integration. Add new pkcs12 directory with Makefile.ssl and
13159     modify error routines to work internally. Add error codes and PBE init
13160     to library startup routines.
13161     [Steve Henson]
13162
13163  *) Further PKCS#12 integration. Added password based encryption, PKCS#8 and
13164     packing functions to asn1 and evp. Changed function names and error
13165     codes along the way.
13166     [Steve Henson]
13167
13168  *) PKCS12 integration: and so it begins... First of several patches to
13169     slowly integrate PKCS#12 functionality into OpenSSL. Add PKCS#12
13170     objects to objects.h
13171     [Steve Henson]
13172
13173  *) Add a new 'indent' option to some X509V3 extension code. Initial ASN1
13174     and display support for Thawte strong extranet extension.
13175     [Steve Henson]
13176
13177  *) Add LinuxPPC support.
13178     [Jeff Dubrule <igor@pobox.org>]
13179
13180  *) Get rid of redundant BN file bn_mulw.c, and rename bn_div64 to
13181     bn_div_words in alpha.s.
13182     [Hannes Reinecke <H.Reinecke@hw.ac.uk> and Ben Laurie]
13183
13184  *) Make sure the RSA OAEP test is skipped under -DRSAref because
13185     OAEP isn't supported when OpenSSL is built with RSAref.
13186     [Ulf Moeller <ulf@fitug.de>]
13187
13188  *) Move definitions of IS_SET/IS_SEQUENCE inside crypto/asn1/asn1.h
13189     so they no longer are missing under -DNOPROTO.
13190     [Soren S. Jorvang <soren@t.dk>]
13191
13192
13193 Changes between 0.9.1c and 0.9.2b  [22 Mar 1999]
13194
13195  *) Make SSL_get_peer_cert_chain() work in servers. Unfortunately, it still
13196     doesn't work when the session is reused. Coming soon!
13197     [Ben Laurie]
13198
13199  *) Fix a security hole, that allows sessions to be reused in the wrong
13200     context thus bypassing client cert protection! All software that uses
13201     client certs and session caches in multiple contexts NEEDS PATCHING to
13202     allow session reuse! A fuller solution is in the works.
13203     [Ben Laurie, problem pointed out by Holger Reif, Bodo Moeller (and ???)]
13204
13205  *) Some more source tree cleanups (removed obsolete files
13206     crypto/bf/asm/bf586.pl, test/test.txt and crypto/sha/asm/f.s; changed
13207     permission on "config" script to be executable) and a fix for the INSTALL
13208     document.
13209     [Ulf Moeller <ulf@fitug.de>]
13210
13211  *) Remove some legacy and erroneous uses of malloc, free instead of
13212     Malloc, Free.
13213     [Lennart Bang <lob@netstream.se>, with minor changes by Steve]
13214
13215  *) Make rsa_oaep_test return non-zero on error.
13216     [Ulf Moeller <ulf@fitug.de>]
13217
13218  *) Add support for native Solaris shared libraries. Configure
13219     solaris-sparc-sc4-pic, make, then run shlib/solaris-sc4.sh. It'd be nice
13220     if someone would make that last step automatic.
13221     [Matthias Loepfe <Matthias.Loepfe@AdNovum.CH>]
13222
13223  *) ctx_size was not built with the right compiler during "make links". Fixed.
13224     [Ben Laurie]
13225
13226  *) Change the meaning of 'ALL' in the cipher list. It now means "everything
13227     except NULL ciphers". This means the default cipher list will no longer
13228     enable NULL ciphers. They need to be specifically enabled e.g. with
13229     the string "DEFAULT:eNULL".
13230     [Steve Henson]
13231
13232  *) Fix to RSA private encryption routines: if p < q then it would
13233     occasionally produce an invalid result. This will only happen with
13234     externally generated keys because OpenSSL (and SSLeay) ensure p > q.
13235     [Steve Henson]
13236
13237  *) Be less restrictive and allow also `perl util/perlpath.pl
13238     /path/to/bin/perl' in addition to `perl util/perlpath.pl /path/to/bin',
13239     because this way one can also use an interpreter named `perl5' (which is
13240     usually the name of Perl 5.xxx on platforms where an Perl 4.x is still
13241     installed as `perl').
13242     [Matthias Loepfe <Matthias.Loepfe@adnovum.ch>]
13243
13244  *) Let util/clean-depend.pl work also with older Perl 5.00x versions.
13245     [Matthias Loepfe <Matthias.Loepfe@adnovum.ch>]
13246
13247  *) Fix Makefile.org so CC,CFLAG etc are passed to 'make links' add
13248     advapi32.lib to Win32 build and change the pem test comparison
13249     to fc.exe (thanks to Ulrich Kroener <kroneru@yahoo.com> for the
13250     suggestion). Fix misplaced ASNI prototypes and declarations in evp.h
13251     and crypto/des/ede_cbcm_enc.c.
13252     [Steve Henson]
13253
13254  *) DES quad checksum was broken on big-endian architectures. Fixed.
13255     [Ben Laurie]
13256
13257  *) Comment out two functions in bio.h that aren't implemented. Fix up the
13258     Win32 test batch file so it (might) work again. The Win32 test batch file
13259     is horrible: I feel ill....
13260     [Steve Henson]
13261
13262  *) Move various #ifdefs around so NO_SYSLOG, NO_DIRENT etc are now selected
13263     in e_os.h. Audit of header files to check ANSI and non ANSI
13264     sections: 10 functions were absent from non ANSI section and not exported
13265     from Windows DLLs. Fixed up libeay.num for new functions.
13266     [Steve Henson]
13267
13268  *) Make `openssl version' output lines consistent.
13269     [Ralf S. Engelschall]
13270
13271  *) Fix Win32 symbol export lists for BIO functions: Added
13272     BIO_get_ex_new_index, BIO_get_ex_num, BIO_get_ex_data and BIO_set_ex_data
13273     to ms/libeay{16,32}.def.
13274     [Ralf S. Engelschall]
13275
13276  *) Second round of fixing the OpenSSL perl/ stuff. It now at least compiled
13277     fine under Unix and passes some trivial tests I've now added. But the
13278     whole stuff is horribly incomplete, so a README.1ST with a disclaimer was
13279     added to make sure no one expects that this stuff really works in the
13280     OpenSSL 0.9.2 release.  Additionally I've started to clean the XS sources
13281     up and fixed a few little bugs and inconsistencies in OpenSSL.{pm,xs} and
13282     openssl_bio.xs.
13283     [Ralf S. Engelschall]
13284
13285  *) Fix the generation of two part addresses in perl.
13286     [Kenji Miyake <kenji@miyake.org>, integrated by Ben Laurie]
13287
13288  *) Add config entry for Linux on MIPS.
13289     [John Tobey <jtobey@channel1.com>]
13290
13291  *) Make links whenever Configure is run, unless we are on Windoze.
13292     [Ben Laurie]
13293
13294  *) Permit extensions to be added to CRLs using crl_section in openssl.cnf.
13295     Currently only issuerAltName and AuthorityKeyIdentifier make any sense
13296     in CRLs.
13297     [Steve Henson]
13298
13299  *) Add a useful kludge to allow package maintainers to specify compiler and
13300     other platforms details on the command line without having to patch the
13301     Configure script every time: One now can use ``perl Configure
13302     <id>:<details>'', i.e. platform ids are allowed to have details appended
13303     to them (separated by colons). This is treated as there would be a static
13304     pre-configured entry in Configure's %table under key <id> with value
13305     <details> and ``perl Configure <id>'' is called.  So, when you want to
13306     perform a quick test-compile under FreeBSD 3.1 with pgcc and without
13307     assembler stuff you can use ``perl Configure "FreeBSD-elf:pgcc:-O6:::"''
13308     now, which overrides the FreeBSD-elf entry on-the-fly.
13309     [Ralf S. Engelschall]
13310
13311  *) Disable new TLS1 ciphersuites by default: they aren't official yet.
13312     [Ben Laurie]
13313
13314  *) Allow DSO flags like -fpic, -fPIC, -KPIC etc. to be specified
13315     on the `perl Configure ...' command line. This way one can compile
13316     OpenSSL libraries with Position Independent Code (PIC) which is needed
13317     for linking it into DSOs.
13318     [Ralf S. Engelschall]
13319
13320  *) Remarkably, export ciphers were totally broken and no-one had noticed!
13321     Fixed.
13322     [Ben Laurie]
13323
13324  *) Cleaned up the LICENSE document: The official contact for any license
13325     questions now is the OpenSSL core team under openssl-core@openssl.org.
13326     And add a paragraph about the dual-license situation to make sure people
13327     recognize that _BOTH_ the OpenSSL license _AND_ the SSLeay license apply
13328     to the OpenSSL toolkit.
13329     [Ralf S. Engelschall]
13330
13331  *) General source tree makefile cleanups: Made `making xxx in yyy...'
13332     display consistent in the source tree and replaced `/bin/rm' by `rm'.
13333     Additionally cleaned up the `make links' target: Remove unnecessary
13334     semicolons, subsequent redundant removes, inline point.sh into mklink.sh
13335     to speed processing and no longer clutter the display with confusing
13336     stuff. Instead only the actually done links are displayed.
13337     [Ralf S. Engelschall]
13338
13339  *) Permit null encryption ciphersuites, used for authentication only. It used
13340     to be necessary to set the preprocessor define SSL_ALLOW_ENULL to do this.
13341     It is now necessary to set SSL_FORBID_ENULL to prevent the use of null
13342     encryption.
13343     [Ben Laurie]
13344
13345  *) Add a bunch of fixes to the PKCS#7 stuff. It used to sometimes reorder
13346     signed attributes when verifying signatures (this would break them),
13347     the detached data encoding was wrong and public keys obtained using
13348     X509_get_pubkey() weren't freed.
13349     [Steve Henson]
13350
13351  *) Add text documentation for the BUFFER functions. Also added a work around
13352     to a Win95 console bug. This was triggered by the password read stuff: the
13353     last character typed gets carried over to the next fread(). If you were
13354     generating a new cert request using 'req' for example then the last
13355     character of the passphrase would be CR which would then enter the first
13356     field as blank.
13357     [Steve Henson]
13358
13359  *) Added the new `Includes OpenSSL Cryptography Software' button as
13360     doc/openssl_button.{gif,html} which is similar in style to the old SSLeay
13361     button and can be used by applications based on OpenSSL to show the
13362     relationship to the OpenSSL project.
13363     [Ralf S. Engelschall]
13364
13365  *) Remove confusing variables in function signatures in files
13366     ssl/ssl_lib.c and ssl/ssl.h.
13367     [Lennart Bong <lob@kulthea.stacken.kth.se>]
13368
13369  *) Don't install bss_file.c under PREFIX/include/
13370     [Lennart Bong <lob@kulthea.stacken.kth.se>]
13371
13372  *) Get the Win32 compile working again. Modify mkdef.pl so it can handle
13373     functions that return function pointers and has support for NT specific
13374     stuff. Fix mk1mf.pl and VC-32.pl to support NT differences also. Various
13375     #ifdef WIN32 and WINNTs sprinkled about the place and some changes from
13376     unsigned to signed types: this was killing the Win32 compile.
13377     [Steve Henson]
13378
13379  *) Add new certificate file to stack functions,
13380     SSL_add_dir_cert_subjects_to_stack() and
13381     SSL_add_file_cert_subjects_to_stack().  These largely supplant
13382     SSL_load_client_CA_file(), and can be used to add multiple certs easily
13383     to a stack (usually this is then handed to SSL_CTX_set_client_CA_list()).
13384     This means that Apache-SSL and similar packages don't have to mess around
13385     to add as many CAs as they want to the preferred list.
13386     [Ben Laurie]
13387
13388  *) Experiment with doxygen documentation. Currently only partially applied to
13389     ssl/ssl_lib.c.
13390     See http://www.stack.nl/~dimitri/doxygen/index.html, and run doxygen with
13391     openssl.doxy as the configuration file.
13392     [Ben Laurie]
13393
13394  *) Get rid of remaining C++-style comments which strict C compilers hate.
13395     [Ralf S. Engelschall, pointed out by Carlos Amengual]
13396
13397  *) Changed BN_RECURSION in bn_mont.c to BN_RECURSION_MONT so it is not
13398     compiled in by default: it has problems with large keys.
13399     [Steve Henson]
13400
13401  *) Add a bunch of SSL_xxx() functions for configuring the temporary RSA and
13402     DH private keys and/or callback functions which directly correspond to
13403     their SSL_CTX_xxx() counterparts but work on a per-connection basis. This
13404     is needed for applications which have to configure certificates on a
13405     per-connection basis (e.g. Apache+mod_ssl) instead of a per-context basis
13406     (e.g. s_server).
13407        For the RSA certificate situation is makes no difference, but
13408     for the DSA certificate situation this fixes the "no shared cipher"
13409     problem where the OpenSSL cipher selection procedure failed because the
13410     temporary keys were not overtaken from the context and the API provided
13411     no way to reconfigure them.
13412        The new functions now let applications reconfigure the stuff and they
13413     are in detail: SSL_need_tmp_RSA, SSL_set_tmp_rsa, SSL_set_tmp_dh,
13414     SSL_set_tmp_rsa_callback and SSL_set_tmp_dh_callback.  Additionally a new
13415     non-public-API function ssl_cert_instantiate() is used as a helper
13416     function and also to reduce code redundancy inside ssl_rsa.c.
13417     [Ralf S. Engelschall]
13418
13419  *) Move s_server -dcert and -dkey options out of the undocumented feature
13420     area because they are useful for the DSA situation and should be
13421     recognized by the users.
13422     [Ralf S. Engelschall]
13423
13424  *) Fix the cipher decision scheme for export ciphers: the export bits are
13425     *not* within SSL_MKEY_MASK or SSL_AUTH_MASK, they are within
13426     SSL_EXP_MASK.  So, the original variable has to be used instead of the
13427     already masked variable.
13428     [Richard Levitte <levitte@stacken.kth.se>]
13429
13430  *) Fix 'port' variable from `int' to `unsigned int' in crypto/bio/b_sock.c
13431     [Richard Levitte <levitte@stacken.kth.se>]
13432
13433  *) Change type of another md_len variable in pk7_doit.c:PKCS7_dataFinal()
13434     from `int' to `unsigned int' because it's a length and initialized by
13435     EVP_DigestFinal() which expects an `unsigned int *'.
13436     [Richard Levitte <levitte@stacken.kth.se>]
13437
13438  *) Don't hard-code path to Perl interpreter on shebang line of Configure
13439     script. Instead use the usual Shell->Perl transition trick.
13440     [Ralf S. Engelschall]
13441
13442  *) Make `openssl x509 -noout -modulus' functional also for DSA certificates
13443     (in addition to RSA certificates) to match the behaviour of `openssl dsa
13444     -noout -modulus' as it's already the case for `openssl rsa -noout
13445     -modulus'.  For RSA the -modulus is the real "modulus" while for DSA
13446     currently the public key is printed (a decision which was already done by
13447     `openssl dsa -modulus' in the past) which serves a similar purpose.
13448     Additionally the NO_RSA no longer completely removes the whole -modulus
13449     option; it now only avoids using the RSA stuff. Same applies to NO_DSA
13450     now, too.
13451     [Ralf S.  Engelschall]
13452
13453  *) Add Arne Ansper's reliable BIO - this is an encrypted, block-digested
13454     BIO. See the source (crypto/evp/bio_ok.c) for more info.
13455     [Arne Ansper <arne@ats.cyber.ee>]
13456
13457  *) Dump the old yucky req code that tried (and failed) to allow raw OIDs
13458     to be added. Now both 'req' and 'ca' can use new objects defined in the
13459     config file.
13460     [Steve Henson]
13461
13462  *) Add cool BIO that does syslog (or event log on NT).
13463     [Arne Ansper <arne@ats.cyber.ee>, integrated by Ben Laurie]
13464
13465  *) Add support for new TLS ciphersuites, TLS_RSA_EXPORT56_WITH_RC4_56_MD5,
13466     TLS_RSA_EXPORT56_WITH_RC2_CBC_56_MD5 and
13467     TLS_RSA_EXPORT56_WITH_DES_CBC_SHA, as specified in "56-bit Export Cipher
13468     Suites For TLS", draft-ietf-tls-56-bit-ciphersuites-00.txt.
13469     [Ben Laurie]
13470
13471  *) Add preliminary config info for new extension code.
13472     [Steve Henson]
13473
13474  *) Make RSA_NO_PADDING really use no padding.
13475     [Ulf Moeller <ulf@fitug.de>]
13476
13477  *) Generate errors when private/public key check is done.
13478     [Ben Laurie]
13479
13480  *) Overhaul for 'crl' utility. New function X509_CRL_print. Partial support
13481     for some CRL extensions and new objects added.
13482     [Steve Henson]
13483
13484  *) Really fix the ASN1 IMPLICIT bug this time... Partial support for private
13485     key usage extension and fuller support for authority key id.
13486     [Steve Henson]
13487
13488  *) Add OAEP encryption for the OpenSSL crypto library. OAEP is the improved
13489     padding method for RSA, which is recommended for new applications in PKCS
13490     #1 v2.0 (RFC 2437, October 1998).
13491     OAEP (Optimal Asymmetric Encryption Padding) has better theoretical
13492     foundations than the ad-hoc padding used in PKCS #1 v1.5. It is secure
13493     against Bleichbacher's attack on RSA.
13494     [Ulf Moeller <ulf@fitug.de>, reformatted, corrected and integrated by
13495      Ben Laurie]
13496
13497  *) Updates to the new SSL compression code
13498     [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)]
13499
13500  *) Fix so that the version number in the master secret, when passed
13501     via RSA, checks that if TLS was proposed, but we roll back to SSLv3
13502     (because the server will not accept higher), that the version number
13503     is 0x03,0x01, not 0x03,0x00
13504     [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)]
13505
13506  *) Run extensive memory leak checks on SSL apps. Fixed *lots* of memory
13507     leaks in ssl/ relating to new X509_get_pubkey() behaviour. Also fixes
13508     in apps/ and an unrelated leak in crypto/dsa/dsa_vrf.c
13509     [Steve Henson]
13510
13511  *) Support for RAW extensions where an arbitrary extension can be
13512     created by including its DER encoding. See apps/openssl.cnf for
13513     an example.
13514     [Steve Henson]
13515
13516  *) Make sure latest Perl versions don't interpret some generated C array
13517     code as Perl array code in the crypto/err/err_genc.pl script.
13518     [Lars Weber <3weber@informatik.uni-hamburg.de>]
13519
13520  *) Modify ms/do_ms.bat to not generate assembly language makefiles since
13521     not many people have the assembler. Various Win32 compilation fixes and
13522     update to the INSTALL.W32 file with (hopefully) more accurate Win32
13523     build instructions.
13524     [Steve Henson]
13525
13526  *) Modify configure script 'Configure' to automatically create crypto/date.h
13527     file under Win32 and also build pem.h from pem.org. New script
13528     util/mkfiles.pl to create the MINFO file on environments that can't do a
13529     'make files': perl util/mkfiles.pl >MINFO should work.
13530     [Steve Henson]
13531
13532  *) Major rework of DES function declarations, in the pursuit of correctness
13533     and purity. As a result, many evil casts evaporated, and some weirdness,
13534     too. You may find this causes warnings in your code. Zapping your evil
13535     casts will probably fix them. Mostly.
13536     [Ben Laurie]
13537
13538  *) Fix for a typo in asn1.h. Bug fix to object creation script
13539     obj_dat.pl. It considered a zero in an object definition to mean
13540     "end of object": none of the objects in objects.h have any zeros
13541     so it wasn't spotted.
13542     [Steve Henson, reported by Erwann ABALEA <eabalea@certplus.com>]
13543
13544  *) Add support for Triple DES Cipher Block Chaining with Output Feedback
13545     Masking (CBCM). In the absence of test vectors, the best I have been able
13546     to do is check that the decrypt undoes the encrypt, so far. Send me test
13547     vectors if you have them.
13548     [Ben Laurie]
13549
13550  *) Correct calculation of key length for export ciphers (too much space was
13551     allocated for null ciphers). This has not been tested!
13552     [Ben Laurie]
13553
13554  *) Modifications to the mkdef.pl for Win32 DEF file creation. The usage
13555     message is now correct (it understands "crypto" and "ssl" on its
13556     command line). There is also now an "update" option. This will update
13557     the util/ssleay.num and util/libeay.num files with any new functions.
13558     If you do a:
13559     perl util/mkdef.pl crypto ssl update
13560     it will update them.
13561     [Steve Henson]
13562
13563  *) Overhauled the Perl interface (perl/*):
13564     - ported BN stuff to OpenSSL's different BN library
13565     - made the perl/ source tree CVS-aware
13566     - renamed the package from SSLeay to OpenSSL (the files still contain
13567       their history because I've copied them in the repository)
13568     - removed obsolete files (the test scripts will be replaced
13569       by better Test::Harness variants in the future)
13570     [Ralf S. Engelschall]
13571
13572  *) First cut for a very conservative source tree cleanup:
13573     1. merge various obsolete readme texts into doc/ssleay.txt
13574     where we collect the old documents and readme texts.
13575     2. remove the first part of files where I'm already sure that we no
13576     longer need them because of three reasons: either they are just temporary
13577     files which were left by Eric or they are preserved original files where
13578     I've verified that the diff is also available in the CVS via "cvs diff
13579     -rSSLeay_0_8_1b" or they were renamed (as it was definitely the case for
13580     the crypto/md/ stuff).
13581     [Ralf S. Engelschall]
13582
13583  *) More extension code. Incomplete support for subject and issuer alt
13584     name, issuer and authority key id. Change the i2v function parameters
13585     and add an extra 'crl' parameter in the X509V3_CTX structure: guess
13586     what that's for :-) Fix to ASN1 macro which messed up
13587     IMPLICIT tag and add f_enum.c which adds a2i, i2a for ENUMERATED.
13588     [Steve Henson]
13589
13590  *) Preliminary support for ENUMERATED type. This is largely copied from the
13591     INTEGER code.
13592     [Steve Henson]
13593
13594  *) Add new function, EVP_MD_CTX_copy() to replace frequent use of memcpy.
13595     [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)]
13596
13597  *) Make sure `make rehash' target really finds the `openssl' program.
13598     [Ralf S. Engelschall, Matthias Loepfe <Matthias.Loepfe@adnovum.ch>]
13599
13600  *) Squeeze another 7% of speed out of MD5 assembler, at least on a P2. I'd
13601     like to hear about it if this slows down other processors.
13602     [Ben Laurie]
13603
13604  *) Add CygWin32 platform information to Configure script.
13605     [Alan Batie <batie@aahz.jf.intel.com>]
13606
13607  *) Fixed ms/32all.bat script: `no_asm' -> `no-asm'
13608     [Rainer W. Gerling <gerling@mpg-gv.mpg.de>]
13609
13610  *) New program nseq to manipulate netscape certificate sequences
13611     [Steve Henson]
13612
13613  *) Modify crl2pkcs7 so it supports multiple -certfile arguments. Fix a
13614     few typos.
13615     [Steve Henson]
13616
13617  *) Fixes to BN code.  Previously the default was to define BN_RECURSION
13618     but the BN code had some problems that would cause failures when
13619     doing certificate verification and some other functions.
13620     [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)]
13621
13622  *) Add ASN1 and PEM code to support netscape certificate sequences.
13623     [Steve Henson]
13624
13625  *) Add ASN1 and PEM code to support netscape certificate sequences.
13626     [Steve Henson]
13627
13628  *) Add several PKIX and private extended key usage OIDs.
13629     [Steve Henson]
13630
13631  *) Modify the 'ca' program to handle the new extension code. Modify
13632     openssl.cnf for new extension format, add comments.
13633     [Steve Henson]
13634
13635  *) More X509 V3 changes. Fix typo in v3_bitstr.c. Add support to 'req'
13636     and add a sample to openssl.cnf so req -x509 now adds appropriate
13637     CA extensions.
13638     [Steve Henson]
13639
13640  *) Continued X509 V3 changes. Add to other makefiles, integrate with the
13641     error code, add initial support to X509_print() and x509 application.
13642     [Steve Henson]
13643
13644  *) Takes a deep breath and start adding X509 V3 extension support code. Add
13645     files in crypto/x509v3. Move original stuff to crypto/x509v3/old. All this
13646     stuff is currently isolated and isn't even compiled yet.
13647     [Steve Henson]
13648
13649  *) Continuing patches for GeneralizedTime. Fix up certificate and CRL
13650     ASN1 to use ASN1_TIME and modify print routines to use ASN1_TIME_print.
13651     Removed the versions check from X509 routines when loading extensions:
13652     this allows certain broken certificates that don't set the version
13653     properly to be processed.
13654     [Steve Henson]
13655
13656  *) Deal with irritating shit to do with dependencies, in YAAHW (Yet Another
13657     Ad Hoc Way) - Makefile.ssls now all contain local dependencies, which
13658     can still be regenerated with "make depend".
13659     [Ben Laurie]
13660
13661  *) Spelling mistake in C version of CAST-128.
13662     [Ben Laurie, reported by Jeremy Hylton <jeremy@cnri.reston.va.us>]
13663
13664  *) Changes to the error generation code. The perl script err-code.pl
13665     now reads in the old error codes and retains the old numbers, only
13666     adding new ones if necessary. It also only changes the .err files if new
13667     codes are added. The makefiles have been modified to only insert errors
13668     when needed (to avoid needlessly modifying header files). This is done
13669     by only inserting errors if the .err file is newer than the auto generated
13670     C file. To rebuild all the error codes from scratch (the old behaviour)
13671     either modify crypto/Makefile.ssl to pass the -regen flag to err_code.pl
13672     or delete all the .err files.
13673     [Steve Henson]
13674
13675  *) CAST-128 was incorrectly implemented for short keys. The C version has
13676     been fixed, but is untested. The assembler versions are also fixed, but
13677     new assembler HAS NOT BEEN GENERATED FOR WIN32 - the Makefile needs fixing
13678     to regenerate it if needed.
13679     [Ben Laurie, reported (with fix for C version) by Jun-ichiro itojun
13680      Hagino <itojun@kame.net>]
13681
13682  *) File was opened incorrectly in randfile.c.
13683     [Ulf Möller <ulf@fitug.de>]
13684
13685  *) Beginning of support for GeneralizedTime. d2i, i2d, check and print
13686     functions. Also ASN1_TIME suite which is a CHOICE of UTCTime or
13687     GeneralizedTime. ASN1_TIME is the proper type used in certificates et
13688     al: it's just almost always a UTCTime. Note this patch adds new error
13689     codes so do a "make errors" if there are problems.
13690     [Steve Henson]
13691
13692  *) Correct Linux 1 recognition in config.
13693     [Ulf Möller <ulf@fitug.de>]
13694
13695  *) Remove pointless MD5 hash when using DSA keys in ca.
13696     [Anonymous <nobody@replay.com>]
13697
13698  *) Generate an error if given an empty string as a cert directory. Also
13699     generate an error if handed NULL (previously returned 0 to indicate an
13700     error, but didn't set one).
13701     [Ben Laurie, reported by Anonymous <nobody@replay.com>]
13702
13703  *) Add prototypes to SSL methods. Make SSL_write's buffer const, at last.
13704     [Ben Laurie]
13705
13706  *) Fix the dummy function BN_ref_mod_exp() in rsaref.c to have the correct
13707     parameters. This was causing a warning which killed off the Win32 compile.
13708     [Steve Henson]
13709
13710  *) Remove C++ style comments from crypto/bn/bn_local.h.
13711     [Neil Costigan <neil.costigan@celocom.com>]
13712
13713  *) The function OBJ_txt2nid was broken. It was supposed to return a nid
13714     based on a text string, looking up short and long names and finally
13715     "dot" format. The "dot" format stuff didn't work. Added new function
13716     OBJ_txt2obj to do the same but return an ASN1_OBJECT and rewrote
13717     OBJ_txt2nid to use it. OBJ_txt2obj can also return objects even if the
13718     OID is not part of the table.
13719     [Steve Henson]
13720
13721  *) Add prototypes to X509 lookup/verify methods, fixing a bug in
13722     X509_LOOKUP_by_alias().
13723     [Ben Laurie]
13724
13725  *) Sort openssl functions by name.
13726     [Ben Laurie]
13727
13728  *) Get the gendsa program working (hopefully) and add it to app list. Remove
13729     encryption from sample DSA keys (in case anyone is interested the password
13730     was "1234").
13731     [Steve Henson]
13732
13733  *) Make _all_ *_free functions accept a NULL pointer.
13734     [Frans Heymans <fheymans@isaserver.be>]
13735
13736  *) If a DH key is generated in s3_srvr.c, don't blow it by trying to use
13737     NULL pointers.
13738     [Anonymous <nobody@replay.com>]
13739
13740  *) s_server should send the CAfile as acceptable CAs, not its own cert.
13741     [Bodo Moeller <3moeller@informatik.uni-hamburg.de>]
13742
13743  *) Don't blow it for numeric -newkey arguments to apps/req.
13744     [Bodo Moeller <3moeller@informatik.uni-hamburg.de>]
13745
13746  *) Temp key "for export" tests were wrong in s3_srvr.c.
13747     [Anonymous <nobody@replay.com>]
13748
13749  *) Add prototype for temp key callback functions
13750     SSL_CTX_set_tmp_{rsa,dh}_callback().
13751     [Ben Laurie]
13752
13753  *) Make DH_free() tolerate being passed a NULL pointer (like RSA_free() and
13754     DSA_free()). Make X509_PUBKEY_set() check for errors in d2i_PublicKey().
13755     [Steve Henson]
13756
13757  *) X509_name_add_entry() freed the wrong thing after an error.
13758     [Arne Ansper <arne@ats.cyber.ee>]
13759
13760  *) rsa_eay.c would attempt to free a NULL context.
13761     [Arne Ansper <arne@ats.cyber.ee>]
13762
13763  *) BIO_s_socket() had a broken should_retry() on Windoze.
13764     [Arne Ansper <arne@ats.cyber.ee>]
13765
13766  *) BIO_f_buffer() didn't pass on BIO_CTRL_FLUSH.
13767     [Arne Ansper <arne@ats.cyber.ee>]
13768
13769  *) Make sure the already existing X509_STORE->depth variable is initialized
13770     in X509_STORE_new(), but document the fact that this variable is still
13771     unused in the certificate verification process.
13772     [Ralf S. Engelschall]
13773
13774  *) Fix the various library and apps files to free up pkeys obtained from
13775     X509_PUBKEY_get() et al. Also allow x509.c to handle netscape extensions.
13776     [Steve Henson]
13777
13778  *) Fix reference counting in X509_PUBKEY_get(). This makes
13779     demos/maurice/example2.c work, amongst others, probably.
13780     [Steve Henson and Ben Laurie]
13781
13782  *) First cut of a cleanup for apps/. First the `ssleay' program is now named
13783     `openssl' and second, the shortcut symlinks for the `openssl <command>'
13784     are no longer created. This way we have a single and consistent command
13785     line interface `openssl <command>', similar to `cvs <command>'.
13786     [Ralf S. Engelschall, Paul Sutton and Ben Laurie]
13787
13788  *) ca.c: move test for DSA keys inside #ifndef NO_DSA. Make pubkey
13789     BIT STRING wrapper always have zero unused bits.
13790     [Steve Henson]
13791
13792  *) Add CA.pl, perl version of CA.sh, add extended key usage OID.
13793     [Steve Henson]
13794
13795  *) Make the top-level INSTALL documentation easier to understand.
13796     [Paul Sutton]
13797
13798  *) Makefiles updated to exit if an error occurs in a sub-directory
13799     make (including if user presses ^C) [Paul Sutton]
13800
13801  *) Make Montgomery context stuff explicit in RSA data structure.
13802     [Ben Laurie]
13803
13804  *) Fix build order of pem and err to allow for generated pem.h.
13805     [Ben Laurie]
13806
13807  *) Fix renumbering bug in X509_NAME_delete_entry().
13808     [Ben Laurie]
13809
13810  *) Enhanced the err-ins.pl script so it makes the error library number
13811     global and can add a library name. This is needed for external ASN1 and
13812     other error libraries.
13813     [Steve Henson]
13814
13815  *) Fixed sk_insert which never worked properly.
13816     [Steve Henson]
13817
13818  *) Fix ASN1 macros so they can handle indefinite length constructed
13819     EXPLICIT tags. Some non standard certificates use these: they can now
13820     be read in.
13821     [Steve Henson]
13822
13823  *) Merged the various old/obsolete SSLeay documentation files (doc/xxx.doc)
13824     into a single doc/ssleay.txt bundle. This way the information is still
13825     preserved but no longer messes up this directory. Now it's new room for
13826     the new set of documentation files.
13827     [Ralf S. Engelschall]
13828
13829  *) SETs were incorrectly DER encoded. This was a major pain, because they
13830     shared code with SEQUENCEs, which aren't coded the same. This means that
13831     almost everything to do with SETs or SEQUENCEs has either changed name or
13832     number of arguments.
13833     [Ben Laurie, based on a partial fix by GP Jayan <gp@nsj.co.jp>]
13834
13835  *) Fix test data to work with the above.
13836     [Ben Laurie]
13837
13838  *) Fix the RSA header declarations that hid a bug I fixed in 0.9.0b but
13839     was already fixed by Eric for 0.9.1 it seems.
13840     [Ben Laurie - pointed out by Ulf Möller <ulf@fitug.de>]
13841
13842  *) Autodetect FreeBSD3.
13843     [Ben Laurie]
13844
13845  *) Fix various bugs in Configure. This affects the following platforms:
13846     nextstep
13847     ncr-scde
13848     unixware-2.0
13849     unixware-2.0-pentium
13850     sco5-cc.
13851     [Ben Laurie]
13852
13853  *) Eliminate generated files from CVS. Reorder tests to regenerate files
13854     before they are needed.
13855     [Ben Laurie]
13856
13857  *) Generate Makefile.ssl from Makefile.org (to keep CVS happy).
13858     [Ben Laurie]
13859
13860
13861 Changes between 0.9.1b and 0.9.1c  [23-Dec-1998]
13862
13863  *) Added OPENSSL_VERSION_NUMBER to crypto/crypto.h and
13864     changed SSLeay to OpenSSL in version strings.
13865     [Ralf S. Engelschall]
13866
13867  *) Some fixups to the top-level documents.
13868     [Paul Sutton]
13869
13870  *) Fixed the nasty bug where rsaref.h was not found under compile-time
13871     because the symlink to include/ was missing.
13872     [Ralf S. Engelschall]
13873
13874  *) Incorporated the popular no-RSA/DSA-only patches
13875     which allow to compile a RSA-free SSLeay.
13876     [Andrew Cooke / Interrader Ldt., Ralf S. Engelschall]
13877
13878  *) Fixed nasty rehash problem under `make -f Makefile.ssl links'
13879     when "ssleay" is still not found.
13880     [Ralf S. Engelschall]
13881
13882  *) Added more platforms to Configure: Cray T3E, HPUX 11,
13883     [Ralf S. Engelschall, Beckmann <beckman@acl.lanl.gov>]
13884
13885  *) Updated the README file.
13886     [Ralf S. Engelschall]
13887
13888  *) Added various .cvsignore files in the CVS repository subdirs
13889     to make a "cvs update" really silent.
13890     [Ralf S. Engelschall]
13891
13892  *) Recompiled the error-definition header files and added
13893     missing symbols to the Win32 linker tables.
13894     [Ralf S. Engelschall]
13895
13896  *) Cleaned up the top-level documents;
13897     o new files: CHANGES and LICENSE
13898     o merged VERSION, HISTORY* and README* files a CHANGES.SSLeay
13899     o merged COPYRIGHT into LICENSE
13900     o removed obsolete TODO file
13901     o renamed MICROSOFT to INSTALL.W32
13902     [Ralf S. Engelschall]
13903
13904  *) Removed dummy files from the 0.9.1b source tree:
13905     crypto/asn1/x crypto/bio/cd crypto/bio/fg crypto/bio/grep crypto/bio/vi
13906     crypto/bn/asm/......add.c crypto/bn/asm/a.out crypto/dsa/f crypto/md5/f
13907     crypto/pem/gmon.out crypto/perlasm/f crypto/pkcs7/build crypto/rsa/f
13908     crypto/sha/asm/f crypto/threads/f ms/zzz ssl/f ssl/f.mak test/f
13909     util/f.mak util/pl/f util/pl/f.mak crypto/bf/bf_locl.old apps/f
13910     [Ralf S. Engelschall]
13911
13912  *) Added various platform portability fixes.
13913     [Mark J. Cox]
13914
13915  *) The Genesis of the OpenSSL rpject:
13916     We start with the latest (unreleased) SSLeay version 0.9.1b which Eric A.
13917     Young and Tim J. Hudson created while they were working for C2Net until
13918     summer 1998.
13919     [The OpenSSL Project]
13920
13921
13922 Changes between 0.9.0b and 0.9.1b  [not released]
13923
13924  *) Updated a few CA certificates under certs/
13925     [Eric A. Young]
13926
13927  *) Changed some BIGNUM api stuff.
13928     [Eric A. Young]
13929
13930  *) Various platform ports: OpenBSD, Ultrix, IRIX 64bit, NetBSD,
13931     DGUX x86, Linux Alpha, etc.
13932     [Eric A. Young]
13933
13934  *) New COMP library [crypto/comp/] for SSL Record Layer Compression:
13935     RLE (dummy implemented) and ZLIB (really implemented when ZLIB is
13936     available).
13937     [Eric A. Young]
13938
13939  *) Add -strparse option to asn1pars program which parses nested
13940     binary structures
13941     [Dr Stephen Henson <shenson@bigfoot.com>]
13942
13943  *) Added "oid_file" to ssleay.cnf for "ca" and "req" programs.
13944     [Eric A. Young]
13945
13946  *) DSA fix for "ca" program.
13947     [Eric A. Young]
13948
13949  *) Added "-genkey" option to "dsaparam" program.
13950     [Eric A. Young]
13951
13952  *) Added RIPE MD160 (rmd160) message digest.
13953     [Eric A. Young]
13954
13955  *) Added -a (all) option to "ssleay version" command.
13956     [Eric A. Young]
13957
13958  *) Added PLATFORM define which is the id given to Configure.
13959     [Eric A. Young]
13960
13961  *) Added MemCheck_XXXX functions to crypto/mem.c for memory checking.
13962     [Eric A. Young]
13963
13964  *) Extended the ASN.1 parser routines.
13965     [Eric A. Young]
13966
13967  *) Extended BIO routines to support REUSEADDR, seek, tell, etc.
13968     [Eric A. Young]
13969
13970  *) Added a BN_CTX to the BN library.
13971     [Eric A. Young]
13972
13973  *) Fixed the weak key values in DES library
13974     [Eric A. Young]
13975
13976  *) Changed API in EVP library for cipher aliases.
13977     [Eric A. Young]
13978
13979  *) Added support for RC2/64bit cipher.
13980     [Eric A. Young]
13981
13982  *) Converted the lhash library to the crypto/mem.c functions.
13983     [Eric A. Young]
13984
13985  *) Added more recognized ASN.1 object ids.
13986     [Eric A. Young]
13987
13988  *) Added more RSA padding checks for SSL/TLS.
13989     [Eric A. Young]
13990
13991  *) Added BIO proxy/filter functionality.
13992     [Eric A. Young]
13993
13994  *) Added extra_certs to SSL_CTX which can be used
13995     send extra CA certificates to the client in the CA cert chain sending
13996     process. It can be configured with SSL_CTX_add_extra_chain_cert().
13997     [Eric A. Young]
13998
13999  *) Now Fortezza is denied in the authentication phase because
14000     this is key exchange mechanism is not supported by SSLeay at all.
14001     [Eric A. Young]
14002
14003  *) Additional PKCS1 checks.
14004     [Eric A. Young]
14005
14006  *) Support the string "TLSv1" for all TLS v1 ciphers.
14007     [Eric A. Young]
14008
14009  *) Added function SSL_get_ex_data_X509_STORE_CTX_idx() which gives the
14010     ex_data index of the SSL context in the X509_STORE_CTX ex_data.
14011     [Eric A. Young]
14012
14013  *) Fixed a few memory leaks.
14014     [Eric A. Young]
14015
14016  *) Fixed various code and comment typos.
14017     [Eric A. Young]
14018
14019  *) A minor bug in ssl/s3_clnt.c where there would always be 4 0
14020     bytes sent in the client random.
14021     [Edward Bishop <ebishop@spyglass.com>]
14022