xref: /freebsd/crypto/openssl/CHANGES.md (revision 63f537551380d2dab29fa402ad1269feae17e594)
1OpenSSL CHANGES
2===============
3
4This is a high-level summary of the most important changes.
5For a full list of changes, see the [git commit log][log] and
6pick the appropriate release branch.
7
8  [log]: https://github.com/openssl/openssl/commits/
9
10OpenSSL Releases
11----------------
12
13 - [OpenSSL 3.0](#openssl-30)
14 - [OpenSSL 1.1.1](#openssl-111)
15 - [OpenSSL 1.1.0](#openssl-110)
16 - [OpenSSL 1.0.2](#openssl-102)
17 - [OpenSSL 1.0.1](#openssl-101)
18 - [OpenSSL 1.0.0](#openssl-100)
19 - [OpenSSL 0.9.x](#openssl-09x)
20
21OpenSSL 3.0
22-----------
23
24For OpenSSL 3.0 a [Migration guide][] has been added, so the CHANGES entries
25listed here are only a brief description.
26The migration guide contains more detailed information related to new features,
27breaking changes, and mappings for the large list of deprecated functions.
28
29[Migration guide]: https://github.com/openssl/openssl/tree/master/doc/man7/migration_guide.pod
30
31### Changes between 3.0.10 and 3.0.11 [19 Sep 2023]
32
33 * Fix POLY1305 MAC implementation corrupting XMM registers on Windows.
34
35   The POLY1305 MAC (message authentication code) implementation in OpenSSL
36   does not save the contents of non-volatile XMM registers on Windows 64
37   platform when calculating the MAC of data larger than 64 bytes. Before
38   returning to the caller all the XMM registers are set to zero rather than
39   restoring their previous content. The vulnerable code is used only on newer
40   x86_64 processors supporting the AVX512-IFMA instructions.
41
42   The consequences of this kind of internal application state corruption can
43   be various - from no consequences, if the calling application does not
44   depend on the contents of non-volatile XMM registers at all, to the worst
45   consequences, where the attacker could get complete control of the
46   application process. However given the contents of the registers are just
47   zeroized so the attacker cannot put arbitrary values inside, the most likely
48   consequence, if any, would be an incorrect result of some application
49   dependent calculations or a crash leading to a denial of service.
50
51   ([CVE-2023-4807])
52
53   *Bernd Edlinger*
54
55### Changes between 3.0.9 and 3.0.10 [1 Aug 2023]
56
57 * Fix excessive time spent checking DH q parameter value.
58
59   The function DH_check() performs various checks on DH parameters. After
60   fixing CVE-2023-3446 it was discovered that a large q parameter value can
61   also trigger an overly long computation during some of these checks.
62   A correct q value, if present, cannot be larger than the modulus p
63   parameter, thus it is unnecessary to perform these checks if q is larger
64   than p.
65
66   If DH_check() is called with such q parameter value,
67   DH_CHECK_INVALID_Q_VALUE return flag is set and the computationally
68   intensive checks are skipped.
69
70   ([CVE-2023-3817])
71
72   *Tomáš Mráz*
73
74 * Fix DH_check() excessive time with over sized modulus.
75
76   The function DH_check() performs various checks on DH parameters. One of
77   those checks confirms that the modulus ("p" parameter) is not too large.
78   Trying to use a very large modulus is slow and OpenSSL will not normally use
79   a modulus which is over 10,000 bits in length.
80
81   However the DH_check() function checks numerous aspects of the key or
82   parameters that have been supplied. Some of those checks use the supplied
83   modulus value even if it has already been found to be too large.
84
85   A new limit has been added to DH_check of 32,768 bits. Supplying a
86   key/parameters with a modulus over this size will simply cause DH_check() to
87   fail.
88
89   ([CVE-2023-3446])
90
91   *Matt Caswell*
92
93 * Do not ignore empty associated data entries with AES-SIV.
94
95   The AES-SIV algorithm allows for authentication of multiple associated
96   data entries along with the encryption. To authenticate empty data the
97   application has to call `EVP_EncryptUpdate()` (or `EVP_CipherUpdate()`)
98   with NULL pointer as the output buffer and 0 as the input buffer length.
99   The AES-SIV implementation in OpenSSL just returns success for such call
100   instead of performing the associated data authentication operation.
101   The empty data thus will not be authenticated. ([CVE-2023-2975])
102
103   Thanks to Juerg Wullschleger (Google) for discovering the issue.
104
105   The fix changes the authentication tag value and the ciphertext for
106   applications that use empty associated data entries with AES-SIV.
107   To decrypt data encrypted with previous versions of OpenSSL the application
108   has to skip calls to `EVP_DecryptUpdate()` for empty associated data
109   entries.
110
111   *Tomáš Mráz*
112
113### Changes between 3.0.8 and 3.0.9 [30 May 2023]
114
115 * Mitigate for the time it takes for `OBJ_obj2txt` to translate gigantic
116   OBJECT IDENTIFIER sub-identifiers to canonical numeric text form.
117
118   OBJ_obj2txt() would translate any size OBJECT IDENTIFIER to canonical
119   numeric text form.  For gigantic sub-identifiers, this would take a very
120   long time, the time complexity being O(n^2) where n is the size of that
121   sub-identifier.  ([CVE-2023-2650])
122
123   To mitigitate this, `OBJ_obj2txt()` will only translate an OBJECT
124   IDENTIFIER to canonical numeric text form if the size of that OBJECT
125   IDENTIFIER is 586 bytes or less, and fail otherwise.
126
127   The basis for this restriction is [RFC 2578 (STD 58), section 3.5]. OBJECT
128   IDENTIFIER values, which stipulates that OBJECT IDENTIFIERS may have at
129   most 128 sub-identifiers, and that the maximum value that each sub-
130   identifier may have is 2^32-1 (4294967295 decimal).
131
132   For each byte of every sub-identifier, only the 7 lower bits are part of
133   the value, so the maximum amount of bytes that an OBJECT IDENTIFIER with
134   these restrictions may occupy is 32 * 128 / 7, which is approximately 586
135   bytes.
136
137   *Richard Levitte*
138
139 * Fixed buffer overread in AES-XTS decryption on ARM 64 bit platforms which
140   happens if the buffer size is 4 mod 5 in 16 byte AES blocks. This can
141   trigger a crash of an application using AES-XTS decryption if the memory
142   just after the buffer being decrypted is not mapped.
143   Thanks to Anton Romanov (Amazon) for discovering the issue.
144   ([CVE-2023-1255])
145
146   *Nevine Ebeid*
147
148 * Reworked the Fix for the Timing Oracle in RSA Decryption ([CVE-2022-4304]).
149   The previous fix for this timing side channel turned out to cause
150   a severe 2-3x performance regression in the typical use case
151   compared to 3.0.7. The new fix uses existing constant time
152   code paths, and restores the previous performance level while
153   fully eliminating all existing timing side channels.
154   The fix was developed by Bernd Edlinger with testing support
155   by Hubert Kario.
156
157   *Bernd Edlinger*
158
159 * Corrected documentation of X509_VERIFY_PARAM_add0_policy() to mention
160   that it does not enable policy checking. Thanks to David Benjamin for
161   discovering this issue.
162   ([CVE-2023-0466])
163
164   *Tomáš Mráz*
165
166 * Fixed an issue where invalid certificate policies in leaf certificates are
167   silently ignored by OpenSSL and other certificate policy checks are skipped
168   for that certificate. A malicious CA could use this to deliberately assert
169   invalid certificate policies in order to circumvent policy checking on the
170   certificate altogether.
171   ([CVE-2023-0465])
172
173   *Matt Caswell*
174
175 * Limited the number of nodes created in a policy tree to mitigate
176   against CVE-2023-0464.  The default limit is set to 1000 nodes, which
177   should be sufficient for most installations.  If required, the limit
178   can be adjusted by setting the OPENSSL_POLICY_TREE_NODES_MAX build
179   time define to a desired maximum number of nodes or zero to allow
180   unlimited growth.
181   ([CVE-2023-0464])
182
183   *Paul Dale*
184
185### Changes between 3.0.7 and 3.0.8 [7 Feb 2023]
186
187 * Fixed NULL dereference during PKCS7 data verification.
188
189   A NULL pointer can be dereferenced when signatures are being
190   verified on PKCS7 signed or signedAndEnveloped data. In case the hash
191   algorithm used for the signature is known to the OpenSSL library but
192   the implementation of the hash algorithm is not available the digest
193   initialization will fail. There is a missing check for the return
194   value from the initialization function which later leads to invalid
195   usage of the digest API most likely leading to a crash.
196   ([CVE-2023-0401])
197
198   PKCS7 data is processed by the SMIME library calls and also by the
199   time stamp (TS) library calls. The TLS implementation in OpenSSL does
200   not call these functions however third party applications would be
201   affected if they call these functions to verify signatures on untrusted
202   data.
203
204   *Tomáš Mráz*
205
206 * Fixed X.400 address type confusion in X.509 GeneralName.
207
208   There is a type confusion vulnerability relating to X.400 address processing
209   inside an X.509 GeneralName. X.400 addresses were parsed as an ASN1_STRING
210   but the public structure definition for GENERAL_NAME incorrectly specified
211   the type of the x400Address field as ASN1_TYPE. This field is subsequently
212   interpreted by the OpenSSL function GENERAL_NAME_cmp as an ASN1_TYPE rather
213   than an ASN1_STRING.
214
215   When CRL checking is enabled (i.e. the application sets the
216   X509_V_FLAG_CRL_CHECK flag), this vulnerability may allow an attacker to
217   pass arbitrary pointers to a memcmp call, enabling them to read memory
218   contents or enact a denial of service.
219   ([CVE-2023-0286])
220
221   *Hugo Landau*
222
223 * Fixed NULL dereference validating DSA public key.
224
225   An invalid pointer dereference on read can be triggered when an
226   application tries to check a malformed DSA public key by the
227   EVP_PKEY_public_check() function. This will most likely lead
228   to an application crash. This function can be called on public
229   keys supplied from untrusted sources which could allow an attacker
230   to cause a denial of service attack.
231
232   The TLS implementation in OpenSSL does not call this function
233   but applications might call the function if there are additional
234   security requirements imposed by standards such as FIPS 140-3.
235   ([CVE-2023-0217])
236
237   *Shane Lontis, Tomáš Mráz*
238
239 * Fixed Invalid pointer dereference in d2i_PKCS7 functions.
240
241   An invalid pointer dereference on read can be triggered when an
242   application tries to load malformed PKCS7 data with the
243   d2i_PKCS7(), d2i_PKCS7_bio() or d2i_PKCS7_fp() functions.
244
245   The result of the dereference is an application crash which could
246   lead to a denial of service attack. The TLS implementation in OpenSSL
247   does not call this function however third party applications might
248   call these functions on untrusted data.
249   ([CVE-2023-0216])
250
251   *Tomáš Mráz*
252
253 * Fixed Use-after-free following BIO_new_NDEF.
254
255   The public API function BIO_new_NDEF is a helper function used for
256   streaming ASN.1 data via a BIO. It is primarily used internally to OpenSSL
257   to support the SMIME, CMS and PKCS7 streaming capabilities, but may also
258   be called directly by end user applications.
259
260   The function receives a BIO from the caller, prepends a new BIO_f_asn1
261   filter BIO onto the front of it to form a BIO chain, and then returns
262   the new head of the BIO chain to the caller. Under certain conditions,
263   for example if a CMS recipient public key is invalid, the new filter BIO
264   is freed and the function returns a NULL result indicating a failure.
265   However, in this case, the BIO chain is not properly cleaned up and the
266   BIO passed by the caller still retains internal pointers to the previously
267   freed filter BIO. If the caller then goes on to call BIO_pop() on the BIO
268   then a use-after-free will occur. This will most likely result in a crash.
269   ([CVE-2023-0215])
270
271   *Viktor Dukhovni, Matt Caswell*
272
273 * Fixed Double free after calling PEM_read_bio_ex.
274
275   The function PEM_read_bio_ex() reads a PEM file from a BIO and parses and
276   decodes the "name" (e.g. "CERTIFICATE"), any header data and the payload
277   data. If the function succeeds then the "name_out", "header" and "data"
278   arguments are populated with pointers to buffers containing the relevant
279   decoded data. The caller is responsible for freeing those buffers. It is
280   possible to construct a PEM file that results in 0 bytes of payload data.
281   In this case PEM_read_bio_ex() will return a failure code but will populate
282   the header argument with a pointer to a buffer that has already been freed.
283   If the caller also frees this buffer then a double free will occur. This
284   will most likely lead to a crash.
285
286   The functions PEM_read_bio() and PEM_read() are simple wrappers around
287   PEM_read_bio_ex() and therefore these functions are also directly affected.
288
289   These functions are also called indirectly by a number of other OpenSSL
290   functions including PEM_X509_INFO_read_bio_ex() and
291   SSL_CTX_use_serverinfo_file() which are also vulnerable. Some OpenSSL
292   internal uses of these functions are not vulnerable because the caller does
293   not free the header argument if PEM_read_bio_ex() returns a failure code.
294   ([CVE-2022-4450])
295
296   *Kurt Roeckx, Matt Caswell*
297
298 * Fixed Timing Oracle in RSA Decryption.
299
300   A timing based side channel exists in the OpenSSL RSA Decryption
301   implementation which could be sufficient to recover a plaintext across
302   a network in a Bleichenbacher style attack. To achieve a successful
303   decryption an attacker would have to be able to send a very large number
304   of trial messages for decryption. The vulnerability affects all RSA padding
305   modes: PKCS#1 v1.5, RSA-OEAP and RSASVE.
306   ([CVE-2022-4304])
307
308   *Dmitry Belyavsky, Hubert Kario*
309
310 * Fixed X.509 Name Constraints Read Buffer Overflow.
311
312   A read buffer overrun can be triggered in X.509 certificate verification,
313   specifically in name constraint checking. The read buffer overrun might
314   result in a crash which could lead to a denial of service attack.
315   In a TLS client, this can be triggered by connecting to a malicious
316   server. In a TLS server, this can be triggered if the server requests
317   client authentication and a malicious client connects.
318   ([CVE-2022-4203])
319
320   *Viktor Dukhovni*
321
322 * Fixed X.509 Policy Constraints Double Locking security issue.
323
324   If an X.509 certificate contains a malformed policy constraint and
325   policy processing is enabled, then a write lock will be taken twice
326   recursively.  On some operating systems (most widely: Windows) this
327   results in a denial of service when the affected process hangs.  Policy
328   processing being enabled on a publicly facing server is not considered
329   to be a common setup.
330   ([CVE-2022-3996])
331
332   *Paul Dale*
333
334 * Our provider implementations of `OSSL_FUNC_KEYMGMT_EXPORT` and
335   `OSSL_FUNC_KEYMGMT_GET_PARAMS` for EC and SM2 keys now honor
336   `OSSL_PKEY_PARAM_EC_POINT_CONVERSION_FORMAT` as set (and
337   default to `POINT_CONVERSION_UNCOMPRESSED`) when exporting
338   `OSSL_PKEY_PARAM_PUB_KEY`, instead of unconditionally using
339   `POINT_CONVERSION_COMPRESSED` as in previous 3.x releases.
340   For symmetry, our implementation of `EVP_PKEY_ASN1_METHOD->export_to`
341   for legacy EC and SM2 keys is also changed similarly to honor the
342   equivalent conversion format flag as specified in the underlying
343   `EC_KEY` object being exported to a provider, when this function is
344   called through `EVP_PKEY_export()`.
345
346   *Nicola Tuveri*
347
348### Changes between 3.0.6 and 3.0.7 [1 Nov 2022]
349
350 * Fixed two buffer overflows in punycode decoding functions.
351
352   A buffer overrun can be triggered in X.509 certificate verification,
353   specifically in name constraint checking. Note that this occurs after
354   certificate chain signature verification and requires either a CA to
355   have signed the malicious certificate or for the application to continue
356   certificate verification despite failure to construct a path to a trusted
357   issuer.
358
359   In a TLS client, this can be triggered by connecting to a malicious
360   server.  In a TLS server, this can be triggered if the server requests
361   client authentication and a malicious client connects.
362
363   An attacker can craft a malicious email address to overflow
364   an arbitrary number of bytes containing the `.`  character (decimal 46)
365   on the stack.  This buffer overflow could result in a crash (causing a
366   denial of service).
367   ([CVE-2022-3786])
368
369   An attacker can craft a malicious email address to overflow four
370   attacker-controlled bytes on the stack.  This buffer overflow could
371   result in a crash (causing a denial of service) or potentially remote code
372   execution depending on stack layout for any given platform/compiler.
373   ([CVE-2022-3602])
374
375   *Paul Dale*
376
377 * Removed all references to invalid OSSL_PKEY_PARAM_RSA names for CRT
378   parameters in OpenSSL code.
379   Applications should not use the names OSSL_PKEY_PARAM_RSA_FACTOR,
380   OSSL_PKEY_PARAM_RSA_EXPONENT and OSSL_PKEY_PARAM_RSA_COEFFICIENT.
381   Use the numbered names such as OSSL_PKEY_PARAM_RSA_FACTOR1 instead.
382   Using these invalid names may cause algorithms to use slower methods
383   that ignore the CRT parameters.
384
385   *Shane Lontis*
386
387 * Fixed a regression introduced in 3.0.6 version raising errors on some stack
388   operations.
389
390   *Tomáš Mráz*
391
392 * Fixed a regression introduced in 3.0.6 version not refreshing the certificate
393   data to be signed before signing the certificate.
394
395   *Gibeom Gwon*
396
397 * Added RIPEMD160 to the default provider.
398
399   *Paul Dale*
400
401 * Ensured that the key share group sent or accepted for the key exchange
402   is allowed for the protocol version.
403
404   *Matt Caswell*
405
406### Changes between 3.0.5 and 3.0.6 [11 Oct 2022]
407
408 * OpenSSL supports creating a custom cipher via the legacy
409   EVP_CIPHER_meth_new() function and associated function calls. This function
410   was deprecated in OpenSSL 3.0 and application authors are instead encouraged
411   to use the new provider mechanism in order to implement custom ciphers.
412
413   OpenSSL versions 3.0.0 to 3.0.5 incorrectly handle legacy custom ciphers
414   passed to the EVP_EncryptInit_ex2(), EVP_DecryptInit_ex2() and
415   EVP_CipherInit_ex2() functions (as well as other similarly named encryption
416   and decryption initialisation functions). Instead of using the custom cipher
417   directly it incorrectly tries to fetch an equivalent cipher from the
418   available providers. An equivalent cipher is found based on the NID passed to
419   EVP_CIPHER_meth_new(). This NID is supposed to represent the unique NID for a
420   given cipher. However it is possible for an application to incorrectly pass
421   NID_undef as this value in the call to EVP_CIPHER_meth_new(). When NID_undef
422   is used in this way the OpenSSL encryption/decryption initialisation function
423   will match the NULL cipher as being equivalent and will fetch this from the
424   available providers. This will succeed if the default provider has been
425   loaded (or if a third party provider has been loaded that offers this
426   cipher). Using the NULL cipher means that the plaintext is emitted as the
427   ciphertext.
428
429   Applications are only affected by this issue if they call
430   EVP_CIPHER_meth_new() using NID_undef and subsequently use it in a call to an
431   encryption/decryption initialisation function. Applications that only use
432   SSL/TLS are not impacted by this issue.
433   ([CVE-2022-3358])
434
435   *Matt Caswell*
436
437 * Fix LLVM vs Apple LLVM version numbering confusion that caused build failures
438   on MacOS 10.11
439
440   *Richard Levitte*
441
442 * Fixed the linux-mips64 Configure target which was missing the
443   SIXTY_FOUR_BIT bn_ops flag. This was causing heap corruption on that
444   platform.
445
446   *Adam Joseph*
447
448 * Fix handling of a ticket key callback that returns 0 in TLSv1.3 to not send a
449   ticket
450
451   *Matt Caswell*
452
453 * Correctly handle a retransmitted ClientHello in DTLS
454
455   *Matt Caswell*
456
457 * Fixed detection of ktls support in cross-compile environment on Linux
458
459   *Tomas Mraz*
460
461 * Fixed some regressions and test failures when running the 3.0.0 FIPS provider
462   against 3.0.x
463
464   *Paul Dale*
465
466 * Fixed SSL_pending() and SSL_has_pending() with DTLS which were failing to
467   report correct results in some cases
468
469   *Matt Caswell*
470
471 * Fix UWP builds by defining VirtualLock
472
473   *Charles Milette*
474
475 * For known safe primes use the minimum key length according to RFC 7919.
476   Longer private key sizes unnecessarily raise the cycles needed to compute the
477   shared secret without any increase of the real security. This fixes a
478   regression from 1.1.1 where these shorter keys were generated for the known
479   safe primes.
480
481   *Tomas Mraz*
482
483 * Added the loongarch64 target
484
485   *Shi Pujin*
486
487 * Fixed EC ASM flag passing. Flags for ASM implementations of EC curves were
488   only passed to the FIPS provider and not to the default or legacy provider.
489
490   *Juergen Christ*
491
492 * Fixed reported performance degradation on aarch64. Restored the
493   implementation prior to commit 2621751 ("aes/asm/aesv8-armx.pl: avoid
494   32-bit lane assignment in CTR mode") for 64bit targets only, since it is
495   reportedly 2-17% slower and the silicon errata only affects 32bit targets.
496   The new algorithm is still used for 32 bit targets.
497
498   *Bernd Edlinger*
499
500 * Added a missing header for memcmp that caused compilation failure on some
501   platforms
502
503   *Gregor Jasny*
504
505### Changes between 3.0.4 and 3.0.5 [5 Jul 2022]
506
507 * The OpenSSL 3.0.4 release introduced a serious bug in the RSA
508   implementation for X86_64 CPUs supporting the AVX512IFMA instructions.
509   This issue makes the RSA implementation with 2048 bit private keys
510   incorrect on such machines and memory corruption will happen during
511   the computation. As a consequence of the memory corruption an attacker
512   may be able to trigger a remote code execution on the machine performing
513   the computation.
514
515   SSL/TLS servers or other servers using 2048 bit RSA private keys running
516   on machines supporting AVX512IFMA instructions of the X86_64 architecture
517   are affected by this issue.
518   ([CVE-2022-2274])
519
520   *Xi Ruoyao*
521
522 * AES OCB mode for 32-bit x86 platforms using the AES-NI assembly optimised
523   implementation would not encrypt the entirety of the data under some
524   circumstances.  This could reveal sixteen bytes of data that was
525   preexisting in the memory that wasn't written.  In the special case of
526   "in place" encryption, sixteen bytes of the plaintext would be revealed.
527
528   Since OpenSSL does not support OCB based cipher suites for TLS and DTLS,
529   they are both unaffected.
530   ([CVE-2022-2097])
531
532   *Alex Chernyakhovsky, David Benjamin, Alejandro Sedeño*
533
534### Changes between 3.0.3 and 3.0.4 [21 Jun 2022]
535
536 * In addition to the c_rehash shell command injection identified in
537   CVE-2022-1292, further bugs where the c_rehash script does not
538   properly sanitise shell metacharacters to prevent command injection have been
539   fixed.
540
541   When the CVE-2022-1292 was fixed it was not discovered that there
542   are other places in the script where the file names of certificates
543   being hashed were possibly passed to a command executed through the shell.
544
545   This script is distributed by some operating systems in a manner where
546   it is automatically executed.  On such operating systems, an attacker
547   could execute arbitrary commands with the privileges of the script.
548
549   Use of the c_rehash script is considered obsolete and should be replaced
550   by the OpenSSL rehash command line tool.
551   (CVE-2022-2068)
552
553   *Daniel Fiala, Tomáš Mráz*
554
555 * Case insensitive string comparison no longer uses locales.  It has instead
556   been directly implemented.
557
558   *Paul Dale*
559
560### Changes between 3.0.2 and 3.0.3 [3 May 2022]
561
562 * Case insensitive string comparison is reimplemented via new locale-agnostic
563   comparison functions OPENSSL_str[n]casecmp always using the POSIX locale for
564   comparison. The previous implementation had problems when the Turkish locale
565   was used.
566
567   *Dmitry Belyavskiy*
568
569 * Fixed a bug in the c_rehash script which was not properly sanitising shell
570   metacharacters to prevent command injection.  This script is distributed by
571   some operating systems in a manner where it is automatically executed.  On
572   such operating systems, an attacker could execute arbitrary commands with the
573   privileges of the script.
574
575   Use of the c_rehash script is considered obsolete and should be replaced
576   by the OpenSSL rehash command line tool.
577   (CVE-2022-1292)
578
579   *Tomáš Mráz*
580
581 * Fixed a bug in the function `OCSP_basic_verify` that verifies the signer
582   certificate on an OCSP response. The bug caused the function in the case
583   where the (non-default) flag OCSP_NOCHECKS is used to return a postivie
584   response (meaning a successful verification) even in the case where the
585   response signing certificate fails to verify.
586
587   It is anticipated that most users of `OCSP_basic_verify` will not use the
588   OCSP_NOCHECKS flag. In this case the `OCSP_basic_verify` function will return
589   a negative value (indicating a fatal error) in the case of a certificate
590   verification failure. The normal expected return value in this case would be
591   0.
592
593   This issue also impacts the command line OpenSSL "ocsp" application. When
594   verifying an ocsp response with the "-no_cert_checks" option the command line
595   application will report that the verification is successful even though it
596   has in fact failed. In this case the incorrect successful response will also
597   be accompanied by error messages showing the failure and contradicting the
598   apparently successful result.
599   ([CVE-2022-1343])
600
601   *Matt Caswell*
602
603 * Fixed a bug where the RC4-MD5 ciphersuite incorrectly used the
604   AAD data as the MAC key. This made the MAC key trivially predictable.
605
606   An attacker could exploit this issue by performing a man-in-the-middle attack
607   to modify data being sent from one endpoint to an OpenSSL 3.0 recipient such
608   that the modified data would still pass the MAC integrity check.
609
610   Note that data sent from an OpenSSL 3.0 endpoint to a non-OpenSSL 3.0
611   endpoint will always be rejected by the recipient and the connection will
612   fail at that point. Many application protocols require data to be sent from
613   the client to the server first. Therefore, in such a case, only an OpenSSL
614   3.0 server would be impacted when talking to a non-OpenSSL 3.0 client.
615
616   If both endpoints are OpenSSL 3.0 then the attacker could modify data being
617   sent in both directions. In this case both clients and servers could be
618   affected, regardless of the application protocol.
619
620   Note that in the absence of an attacker this bug means that an OpenSSL 3.0
621   endpoint communicating with a non-OpenSSL 3.0 endpoint will fail to complete
622   the handshake when using this ciphersuite.
623
624   The confidentiality of data is not impacted by this issue, i.e. an attacker
625   cannot decrypt data that has been encrypted using this ciphersuite - they can
626   only modify it.
627
628   In order for this attack to work both endpoints must legitimately negotiate
629   the RC4-MD5 ciphersuite. This ciphersuite is not compiled by default in
630   OpenSSL 3.0, and is not available within the default provider or the default
631   ciphersuite list. This ciphersuite will never be used if TLSv1.3 has been
632   negotiated. In order for an OpenSSL 3.0 endpoint to use this ciphersuite the
633   following must have occurred:
634
635   1) OpenSSL must have been compiled with the (non-default) compile time option
636      enable-weak-ssl-ciphers
637
638   2) OpenSSL must have had the legacy provider explicitly loaded (either
639      through application code or via configuration)
640
641   3) The ciphersuite must have been explicitly added to the ciphersuite list
642
643   4) The libssl security level must have been set to 0 (default is 1)
644
645   5) A version of SSL/TLS below TLSv1.3 must have been negotiated
646
647   6) Both endpoints must negotiate the RC4-MD5 ciphersuite in preference to any
648      others that both endpoints have in common
649   (CVE-2022-1434)
650
651   *Matt Caswell*
652
653 * Fix a bug in the OPENSSL_LH_flush() function that breaks reuse of the memory
654   occuppied by the removed hash table entries.
655
656   This function is used when decoding certificates or keys. If a long lived
657   process periodically decodes certificates or keys its memory usage will
658   expand without bounds and the process might be terminated by the operating
659   system causing a denial of service. Also traversing the empty hash table
660   entries will take increasingly more time.
661
662   Typically such long lived processes might be TLS clients or TLS servers
663   configured to accept client certificate authentication.
664   (CVE-2022-1473)
665
666   *Hugo Landau, Aliaksei Levin*
667
668 * The functions `OPENSSL_LH_stats` and `OPENSSL_LH_stats_bio` now only report
669   the `num_items`, `num_nodes` and `num_alloc_nodes` statistics. All other
670   statistics are no longer supported. For compatibility, these statistics are
671   still listed in the output but are now always reported as zero.
672
673   *Hugo Landau*
674
675### Changes between 3.0.1 and 3.0.2 [15 Mar 2022]
676
677 * Fixed a bug in the BN_mod_sqrt() function that can cause it to loop forever
678   for non-prime moduli.
679
680   Internally this function is used when parsing certificates that contain
681   elliptic curve public keys in compressed form or explicit elliptic curve
682   parameters with a base point encoded in compressed form.
683
684   It is possible to trigger the infinite loop by crafting a certificate that
685   has invalid explicit curve parameters.
686
687   Since certificate parsing happens prior to verification of the certificate
688   signature, any process that parses an externally supplied certificate may thus
689   be subject to a denial of service attack. The infinite loop can also be
690   reached when parsing crafted private keys as they can contain explicit
691   elliptic curve parameters.
692
693   Thus vulnerable situations include:
694
695    - TLS clients consuming server certificates
696    - TLS servers consuming client certificates
697    - Hosting providers taking certificates or private keys from customers
698    - Certificate authorities parsing certification requests from subscribers
699    - Anything else which parses ASN.1 elliptic curve parameters
700
701   Also any other applications that use the BN_mod_sqrt() where the attacker
702   can control the parameter values are vulnerable to this DoS issue.
703   ([CVE-2022-0778])
704
705   *Tomáš Mráz*
706
707 * Add ciphersuites based on DHE_PSK (RFC 4279) and ECDHE_PSK (RFC 5489)
708   to the list of ciphersuites providing Perfect Forward Secrecy as
709   required by SECLEVEL >= 3.
710
711   *Dmitry Belyavskiy, Nicola Tuveri*
712
713 * Made the AES constant time code for no-asm configurations
714   optional due to the resulting 95% performance degradation.
715   The AES constant time code can be enabled, for no assembly
716   builds, with: ./config no-asm -DOPENSSL_AES_CONST_TIME
717
718   *Paul Dale*
719
720 * Fixed PEM_write_bio_PKCS8PrivateKey() to make it possible to use empty
721   passphrase strings.
722
723   *Darshan Sen*
724
725 * The negative return value handling of the certificate verification callback
726   was reverted. The replacement is to set the verification retry state with
727   the SSL_set_retry_verify() function.
728
729   *Tomáš Mráz*
730
731### Changes between 3.0.0 and 3.0.1 [14 Dec 2021]
732
733 * Fixed invalid handling of X509_verify_cert() internal errors in libssl
734   Internally libssl in OpenSSL calls X509_verify_cert() on the client side to
735   verify a certificate supplied by a server. That function may return a
736   negative return value to indicate an internal error (for example out of
737   memory). Such a negative return value is mishandled by OpenSSL and will cause
738   an IO function (such as SSL_connect() or SSL_do_handshake()) to not indicate
739   success and a subsequent call to SSL_get_error() to return the value
740   SSL_ERROR_WANT_RETRY_VERIFY. This return value is only supposed to be
741   returned by OpenSSL if the application has previously called
742   SSL_CTX_set_cert_verify_callback(). Since most applications do not do this
743   the SSL_ERROR_WANT_RETRY_VERIFY return value from SSL_get_error() will be
744   totally unexpected and applications may not behave correctly as a result. The
745   exact behaviour will depend on the application but it could result in
746   crashes, infinite loops or other similar incorrect responses.
747
748   This issue is made more serious in combination with a separate bug in OpenSSL
749   3.0 that will cause X509_verify_cert() to indicate an internal error when
750   processing a certificate chain. This will occur where a certificate does not
751   include the Subject Alternative Name extension but where a Certificate
752   Authority has enforced name constraints. This issue can occur even with valid
753   chains.
754   ([CVE-2021-4044])
755
756   *Matt Caswell*
757
758 * Corrected a few file name and file reference bugs in the build,
759   installation and setup scripts, which lead to installation verification
760   failures.  Slightly enhanced the installation verification script.
761
762   *Richard Levitte*
763
764 * Fixed EVP_PKEY_eq() to make it possible to use it with strictly private
765   keys.
766
767   *Richard Levitte*
768
769 * Fixed PVK encoder to properly query for the passphrase.
770
771   *Tomáš Mráz*
772
773 * Multiple fixes in the OSSL_HTTP API functions.
774
775   *David von Oheimb*
776
777 * Allow sign extension in OSSL_PARAM_allocate_from_text() for the
778   OSSL_PARAM_INTEGER data type and return error on negative numbers
779   used with the OSSL_PARAM_UNSIGNED_INTEGER data type. Make
780   OSSL_PARAM_BLD_push_BN{,_pad}() return an error on negative numbers.
781
782   *Richard Levitte*
783
784 * Allow copying uninitialized digest contexts with EVP_MD_CTX_copy_ex.
785
786   *Tomáš Mráz*
787
788 * Fixed detection of ARMv7 and ARM64 CPU features on FreeBSD.
789
790   *Allan Jude*
791
792 * Multiple threading fixes.
793
794   *Matt Caswell*
795
796 * Added NULL digest implementation to keep compatibility with 1.1.1 version.
797
798   *Tomáš Mráz*
799
800 * Allow fetching an operation from the provider that owns an unexportable key
801   as a fallback if that is still allowed by the property query.
802
803   *Richard Levitte*
804
805### Changes between 1.1.1 and 3.0.0 [7 sep 2021]
806
807 * TLS_MAX_VERSION, DTLS_MAX_VERSION and DTLS_MIN_VERSION constants are now
808   deprecated.
809
810   *Matt Caswell*
811
812 * The `OPENSSL_s390xcap` environment variable can be used to set bits in the
813   S390X capability vector to zero. This simplifies testing of different code
814   paths on S390X architecture.
815
816   *Patrick Steuer*
817
818 * Encrypting more than 2^64 TLS records with AES-GCM is disallowed
819   as per FIPS 140-2 IG A.5 "Key/IV Pair Uniqueness Requirements from
820   SP 800-38D". The communication will fail at this point.
821
822   *Paul Dale*
823
824 * The EC_GROUP_clear_free() function is deprecated as there is nothing
825   confidential in EC_GROUP data.
826
827   *Nicola Tuveri*
828
829 * The byte order mark (BOM) character is ignored if encountered at the
830   beginning of a PEM-formatted file.
831
832   *Dmitry Belyavskiy*
833
834 * Added CMS support for the Russian GOST algorithms.
835
836   *Dmitry Belyavskiy*
837
838 * Due to move of the implementation of cryptographic operations
839   to the providers, validation of various operation parameters can
840   be postponed until the actual operation is executed where previously
841   it happened immediately when an operation parameter was set.
842
843   For example when setting an unsupported curve with
844   EVP_PKEY_CTX_set_ec_paramgen_curve_nid() this function call will not
845   fail but later keygen operations with the EVP_PKEY_CTX will fail.
846
847   *OpenSSL team members and many third party contributors*
848
849 * The EVP_get_cipherbyname() function will return NULL for algorithms such as
850   "AES-128-SIV", "AES-128-CBC-CTS" and "CAMELLIA-128-CBC-CTS" which were
851   previously only accessible via low level interfaces. Use EVP_CIPHER_fetch()
852   instead to retrieve these algorithms from a provider.
853
854   *Shane Lontis*
855
856 * On build targets where the multilib postfix is set in the build
857   configuration the libdir directory was changing based on whether
858   the lib directory with the multilib postfix exists on the system
859   or not. This unpredictable behavior was removed and eventual
860   multilib postfix is now always added to the default libdir. Use
861   `--libdir=lib` to override the libdir if adding the postfix is
862   undesirable.
863
864   *Jan Lána*
865
866 * The triple DES key wrap functionality now conforms to RFC 3217 but is
867   no longer interoperable with OpenSSL 1.1.1.
868
869   *Paul Dale*
870
871 * The ERR_GET_FUNC() function was removed.  With the loss of meaningful
872   function codes, this function can only cause problems for calling
873   applications.
874
875   *Paul Dale*
876
877 * Add a configurable flag to output date formats as ISO 8601. Does not
878   change the default date format.
879
880   *William Edmisten*
881
882 * Version of MSVC earlier than 1300 could get link warnings, which could
883   be suppressed if the undocumented -DI_CAN_LIVE_WITH_LNK4049 was set.
884   Support for this flag has been removed.
885
886   *Rich Salz*
887
888 * Rework and make DEBUG macros consistent. Remove unused -DCONF_DEBUG,
889   -DBN_CTX_DEBUG, and REF_PRINT. Add a new tracing category and use it for
890   printing reference counts. Rename -DDEBUG_UNUSED to -DUNUSED_RESULT_DEBUG
891   Fix BN_DEBUG_RAND so it compiles and, when set, force DEBUG_RAND to be set
892   also. Rename engine_debug_ref to be ENGINE_REF_PRINT also for consistency.
893
894   *Rich Salz*
895
896 * The signatures of the functions to get and set options on SSL and
897   SSL_CTX objects changed from "unsigned long" to "uint64_t" type.
898   Some source code changes may be required.
899
900   *Rich Salz*
901
902 * The public definitions of conf_method_st and conf_st have been
903   deprecated. They will be made opaque in a future release.
904
905   *Rich Salz and Tomáš Mráz*
906
907 * Client-initiated renegotiation is disabled by default. To allow it, use
908   the -client_renegotiation option, the SSL_OP_ALLOW_CLIENT_RENEGOTIATION
909   flag, or the "ClientRenegotiation" config parameter as appropriate.
910
911   *Rich Salz*
912
913 * Add "abspath" and "includedir" pragma's to config files, to prevent,
914   or modify relative pathname inclusion.
915
916   *Rich Salz*
917
918 * OpenSSL includes a cryptographic module that is intended to be FIPS 140-2
919   validated. Please consult the README-FIPS and
920   README-PROVIDERS files, as well as the migration guide.
921
922   *OpenSSL team members and many third party contributors*
923
924 * For the key types DH and DHX the allowed settable parameters are now different.
925
926   *Shane Lontis*
927
928 * The openssl commands that read keys, certificates, and CRLs now
929   automatically detect the PEM or DER format of the input files.
930
931   *David von Oheimb, Richard Levitte, and Tomáš Mráz*
932
933 * Added enhanced PKCS#12 APIs which accept a library context.
934
935   *Jon Spillett*
936
937 * The default manual page suffix ($MANSUFFIX) has been changed to "ossl"
938
939   *Matt Caswell*
940
941 * Added support for Kernel TLS (KTLS).
942
943   *Boris Pismenny, John Baldwin and Andrew Gallatin*
944
945 * Support for RFC 5746 secure renegotiation is now required by default for
946   SSL or TLS connections to succeed.
947
948   *Benjamin Kaduk*
949
950 * The signature of the `copy` functional parameter of the
951   EVP_PKEY_meth_set_copy() function has changed so its `src` argument is
952   now `const EVP_PKEY_CTX *` instead of `EVP_PKEY_CTX *`. Similarly
953   the signature of the `pub_decode` functional parameter of the
954   EVP_PKEY_asn1_set_public() function has changed so its `pub` argument is
955   now `const X509_PUBKEY *` instead of `X509_PUBKEY *`.
956
957   *David von Oheimb*
958
959 * The error return values from some control calls (ctrl) have changed.
960
961   *Paul Dale*
962
963 * A public key check is now performed during EVP_PKEY_derive_set_peer().
964
965   *Shane Lontis*
966
967 * Many functions in the EVP_ namespace that are getters of values from
968   implementations or contexts were renamed to include get or get0 in their
969   names. Old names are provided as macro aliases for compatibility and
970   are not deprecated.
971
972   *Tomáš Mráz*
973
974 * The EVP_PKEY_CTRL_PKCS7_ENCRYPT, EVP_PKEY_CTRL_PKCS7_DECRYPT,
975   EVP_PKEY_CTRL_PKCS7_SIGN, EVP_PKEY_CTRL_CMS_ENCRYPT,
976   EVP_PKEY_CTRL_CMS_DECRYPT, and EVP_PKEY_CTRL_CMS_SIGN control operations
977   are deprecated.
978
979   *Tomáš Mráz*
980
981 * The EVP_PKEY_public_check() and EVP_PKEY_param_check() functions now work for
982   more key types.
983
984 * The output from the command line applications may have minor
985   changes.
986
987   *Paul Dale*
988
989 * The output from numerous "printing" may have minor changes.
990
991   *David von Oheimb*
992
993 * Windows thread synchronization uses read/write primitives (SRWLock) when
994   supported by the OS, otherwise CriticalSection continues to be used.
995
996   *Vincent Drake*
997
998 * Add filter BIO BIO_f_readbuffer() that allows BIO_tell() and BIO_seek() to
999   work on read only BIO source/sinks that do not support these functions.
1000   This allows piping or redirection of a file BIO using stdin to be buffered
1001   into memory. This is used internally in OSSL_DECODER_from_bio().
1002
1003   *Shane Lontis*
1004
1005 * OSSL_STORE_INFO_get_type() may now return an additional value. In 1.1.1
1006   this function would return one of the values OSSL_STORE_INFO_NAME,
1007   OSSL_STORE_INFO_PKEY, OSSL_STORE_INFO_PARAMS, OSSL_STORE_INFO_CERT or
1008   OSSL_STORE_INFO_CRL. Decoded public keys would previously have been reported
1009   as type OSSL_STORE_INFO_PKEY in 1.1.1. In 3.0 decoded public keys are now
1010   reported as having the new type OSSL_STORE_INFO_PUBKEY. Applications
1011   using this function should be amended to handle the changed return value.
1012
1013   *Richard Levitte*
1014
1015 * Improved adherence to Enhanced Security Services (ESS, RFC 2634 and RFC 5035)
1016   for the TSP and CMS Advanced Electronic Signatures (CAdES) implementations.
1017   As required by RFC 5035 check both ESSCertID and ESSCertIDv2 if both present.
1018   Correct the semantics of checking the validation chain in case ESSCertID{,v2}
1019   contains more than one certificate identifier: This means that all
1020   certificates referenced there MUST be part of the validation chain.
1021
1022   *David von Oheimb*
1023
1024 * The implementation of older EVP ciphers related to CAST, IDEA, SEED, RC2, RC4,
1025   RC5, DESX and DES have been moved to the legacy provider.
1026
1027   *Matt Caswell*
1028
1029 * The implementation of the EVP digests MD2, MD4, MDC2, WHIRLPOOL and
1030   RIPEMD-160 have been moved to the legacy provider.
1031
1032   *Matt Caswell*
1033
1034 * The deprecated function EVP_PKEY_get0() now returns NULL being called for a
1035   provided key.
1036
1037   *Dmitry Belyavskiy*
1038
1039 * The deprecated functions EVP_PKEY_get0_RSA(),
1040   EVP_PKEY_get0_DSA(), EVP_PKEY_get0_EC_KEY(), EVP_PKEY_get0_DH(),
1041   EVP_PKEY_get0_hmac(), EVP_PKEY_get0_poly1305() and EVP_PKEY_get0_siphash() as
1042   well as the similarly named "get1" functions behave differently in
1043   OpenSSL 3.0.
1044
1045   *Matt Caswell*
1046
1047 * A number of functions handling low-level keys or engines were deprecated
1048   including EVP_PKEY_set1_engine(), EVP_PKEY_get0_engine(), EVP_PKEY_assign(),
1049   EVP_PKEY_get0(), EVP_PKEY_get0_hmac(), EVP_PKEY_get0_poly1305() and
1050   EVP_PKEY_get0_siphash().
1051
1052   *Matt Caswell*
1053
1054 * PKCS#5 PBKDF1 key derivation has been moved from PKCS5_PBE_keyivgen() into
1055   the legacy crypto provider as an EVP_KDF. Applications requiring this KDF
1056   will need to load the legacy crypto provider. This includes these PBE
1057   algorithms which use this KDF:
1058   - NID_pbeWithMD2AndDES_CBC
1059   - NID_pbeWithMD5AndDES_CBC
1060   - NID_pbeWithSHA1AndRC2_CBC
1061   - NID_pbeWithMD2AndRC2_CBC
1062   - NID_pbeWithMD5AndRC2_CBC
1063   - NID_pbeWithSHA1AndDES_CBC
1064
1065   *Jon Spillett*
1066
1067 * Deprecated obsolete BIO_set_callback(), BIO_get_callback(), and
1068   BIO_debug_callback() functions.
1069
1070   *Tomáš Mráz*
1071
1072 * Deprecated obsolete EVP_PKEY_CTX_get0_dh_kdf_ukm() and
1073   EVP_PKEY_CTX_get0_ecdh_kdf_ukm() functions.
1074
1075   *Tomáš Mráz*
1076
1077 * The RAND_METHOD APIs have been deprecated.
1078
1079   *Paul Dale*
1080
1081 * The SRP APIs have been deprecated.
1082
1083   *Matt Caswell*
1084
1085 * Add a compile time option to prevent the caching of provider fetched
1086   algorithms.  This is enabled by including the no-cached-fetch option
1087   at configuration time.
1088
1089   *Paul Dale*
1090
1091 * pkcs12 now uses defaults of PBKDF2, AES and SHA-256, with a MAC iteration
1092   count of PKCS12_DEFAULT_ITER.
1093
1094   *Tomáš Mráz and Sahana Prasad*
1095
1096 * The openssl speed command does not use low-level API calls anymore.
1097
1098   *Tomáš Mráz*
1099
1100 * Parallel dual-prime 1024-bit modular exponentiation for AVX512_IFMA
1101   capable processors.
1102
1103   *Ilya Albrekht, Sergey Kirillov, Andrey Matyukov (Intel Corp)*
1104
1105 * Combining the Configure options no-ec and no-dh no longer disables TLSv1.3.
1106
1107   *Matt Caswell*
1108
1109 * Implemented support for fully "pluggable" TLSv1.3 groups. This means that
1110   providers may supply their own group implementations (using either the "key
1111   exchange" or the "key encapsulation" methods) which will automatically be
1112   detected and used by libssl.
1113
1114   *Matt Caswell, Nicola Tuveri*
1115
1116 * The undocumented function X509_certificate_type() has been deprecated;
1117
1118   *Rich Salz*
1119
1120 * Deprecated the obsolete BN_pseudo_rand() and BN_pseudo_rand_range().
1121
1122   *Tomáš Mráz*
1123
1124 * Removed RSA padding mode for SSLv23 (which was only used for
1125   SSLv2). This includes the functions RSA_padding_check_SSLv23() and
1126   RSA_padding_add_SSLv23() and the `-ssl` option in the deprecated
1127   `rsautl` command.
1128
1129   *Rich Salz*
1130
1131 * Deprecated the obsolete X9.31 RSA key generation related functions.
1132
1133 * While a callback function set via `SSL_CTX_set_cert_verify_callback()`
1134   is not allowed to return a value > 1, this is no more taken as failure.
1135
1136   *Viktor Dukhovni and David von Oheimb*
1137
1138 * Deprecated the obsolete X9.31 RSA key generation related functions
1139   BN_X931_generate_Xpq(), BN_X931_derive_prime_ex(), and
1140   BN_X931_generate_prime_ex().
1141
1142   *Tomáš Mráz*
1143
1144 * The default key generation method for the regular 2-prime RSA keys was
1145   changed to the FIPS 186-4 B.3.6 method.
1146
1147   *Shane Lontis*
1148
1149 * Deprecated the BN_is_prime_ex() and BN_is_prime_fasttest_ex() functions.
1150
1151   *Kurt Roeckx*
1152
1153 * Deprecated EVP_MD_CTX_set_update_fn() and EVP_MD_CTX_update_fn().
1154
1155   *Rich Salz*
1156
1157 * Deprecated the type OCSP_REQ_CTX and the functions OCSP_REQ_CTX_*() and
1158   replaced with OSSL_HTTP_REQ_CTX and the functions OSSL_HTTP_REQ_CTX_*().
1159
1160   *Rich Salz, Richard Levitte, and David von Oheimb*
1161
1162 * Deprecated `X509_http_nbio()` and `X509_CRL_http_nbio()`.
1163
1164   *David von Oheimb*
1165
1166 * Deprecated `OCSP_parse_url()`.
1167
1168   *David von Oheimb*
1169
1170 * Validation of SM2 keys has been separated from the validation of regular EC
1171   keys.
1172
1173   *Nicola Tuveri*
1174
1175 * Behavior of the `pkey` app is changed, when using the `-check` or `-pubcheck`
1176   switches: a validation failure triggers an early exit, returning a failure
1177   exit status to the parent process.
1178
1179   *Nicola Tuveri*
1180
1181 * Changed behavior of SSL_CTX_set_ciphersuites() and SSL_set_ciphersuites()
1182   to ignore unknown ciphers.
1183
1184   *Otto Hollmann*
1185
1186 * The `-cipher-commands` and `-digest-commands` options
1187   of the command line utility `list` have been deprecated.
1188   Instead use the `-cipher-algorithms` and `-digest-algorithms` options.
1189
1190   *Dmitry Belyavskiy*
1191
1192 * Added convenience functions for generating asymmetric key pairs:
1193   The 'quick' one-shot (yet somewhat limited) function L<EVP_PKEY_Q_keygen(3)>
1194   and macros for the most common cases: <EVP_RSA_gen(3)> and L<EVP_EC_gen(3)>.
1195
1196   *David von Oheimb*
1197
1198 * All of the low level EC_KEY functions have been deprecated.
1199
1200   *Shane Lontis, Paul Dale, Richard Levitte, and Tomáš Mráz*
1201
1202 * Deprecated all the libcrypto and libssl error string loading
1203   functions.
1204
1205   *Richard Levitte*
1206
1207 * The functions SSL_CTX_set_tmp_dh_callback and SSL_set_tmp_dh_callback, as
1208   well as the macros SSL_CTX_set_tmp_dh() and SSL_set_tmp_dh() have been
1209   deprecated.
1210
1211   *Matt Caswell*
1212
1213 * The `-crypt` option to the `passwd` command line tool has been removed.
1214
1215   *Paul Dale*
1216
1217 * The -C option to the `x509`, `dhparam`, `dsaparam`, and `ecparam` commands
1218   were removed.
1219
1220   *Rich Salz*
1221
1222 * Add support for AES Key Wrap inverse ciphers to the EVP layer.
1223
1224   *Shane Lontis*
1225
1226 * Deprecated EVP_PKEY_set1_tls_encodedpoint() and
1227   EVP_PKEY_get1_tls_encodedpoint().
1228
1229   *Matt Caswell*
1230
1231 * The security callback, which can be customised by application code, supports
1232   the security operation SSL_SECOP_TMP_DH. One location of the "other" parameter
1233   was incorrectly passing a DH object. It now passed an EVP_PKEY in all cases.
1234
1235   *Matt Caswell*
1236
1237 * Add PKCS7_get_octet_string() and PKCS7_type_is_other() to the public
1238   interface. Their functionality remains unchanged.
1239
1240   *Jordan Montgomery*
1241
1242 * Added new option for 'openssl list', '-providers', which will display the
1243   list of loaded providers, their names, version and status.  It optionally
1244   displays their gettable parameters.
1245
1246   *Paul Dale*
1247
1248 * Removed EVP_PKEY_set_alias_type().
1249
1250   *Richard Levitte*
1251
1252 * Deprecated `EVP_PKEY_CTX_set_rsa_keygen_pubexp()` and introduced
1253   `EVP_PKEY_CTX_set1_rsa_keygen_pubexp()`, which is now preferred.
1254
1255   *Jeremy Walch*
1256
1257 * Changed all "STACK" functions to be macros instead of inline functions. Macro
1258   parameters are still checked for type safety at compile time via helper
1259   inline functions.
1260
1261   *Matt Caswell*
1262
1263 * Remove the RAND_DRBG API
1264
1265   *Paul Dale and Matthias St. Pierre*
1266
1267 * Allow `SSL_set1_host()` and `SSL_add1_host()` to take IP literal addresses
1268   as well as actual hostnames.
1269
1270   *David Woodhouse*
1271
1272 * The 'MinProtocol' and 'MaxProtocol' configuration commands now silently
1273   ignore TLS protocol version bounds when configuring DTLS-based contexts, and
1274   conversely, silently ignore DTLS protocol version bounds when configuring
1275   TLS-based contexts.  The commands can be repeated to set bounds of both
1276   types.  The same applies with the corresponding "min_protocol" and
1277   "max_protocol" command-line switches, in case some application uses both TLS
1278   and DTLS.
1279
1280   SSL_CTX instances that are created for a fixed protocol version (e.g.
1281   `TLSv1_server_method()`) also silently ignore version bounds.  Previously
1282   attempts to apply bounds to these protocol versions would result in an
1283   error.  Now only the "version-flexible" SSL_CTX instances are subject to
1284   limits in configuration files in command-line options.
1285
1286   *Viktor Dukhovni*
1287
1288 * Deprecated the `ENGINE` API.  Engines should be replaced with providers
1289   going forward.
1290
1291   *Paul Dale*
1292
1293 * Reworked the recorded ERR codes to make better space for system errors.
1294   To distinguish them, the macro `ERR_SYSTEM_ERROR()` indicates if the
1295   given code is a system error (true) or an OpenSSL error (false).
1296
1297   *Richard Levitte*
1298
1299 * Reworked the test perl framework to better allow parallel testing.
1300
1301   *Nicola Tuveri and David von Oheimb*
1302
1303 * Added ciphertext stealing algorithms AES-128-CBC-CTS, AES-192-CBC-CTS and
1304   AES-256-CBC-CTS to the providers. CS1, CS2 and CS3 variants are supported.
1305
1306   *Shane Lontis*
1307
1308 * 'Configure' has been changed to figure out the configuration target if
1309   none is given on the command line.  Consequently, the 'config' script is
1310   now only a mere wrapper.  All documentation is changed to only mention
1311   'Configure'.
1312
1313   *Rich Salz and Richard Levitte*
1314
1315 * Added a library context `OSSL_LIB_CTX` that applications as well as
1316   other libraries can use to form a separate context within which
1317   libcrypto operations are performed.
1318
1319   *Richard Levitte*
1320
1321 * Added various `_ex` functions to the OpenSSL API that support using
1322   a non-default `OSSL_LIB_CTX`.
1323
1324   *OpenSSL team*
1325
1326 * Handshake now fails if Extended Master Secret extension is dropped
1327   on renegotiation.
1328
1329   *Tomáš Mráz*
1330
1331 * Dropped interactive mode from the `openssl` program.
1332
1333   *Richard Levitte*
1334
1335 * Deprecated `EVP_PKEY_cmp()` and `EVP_PKEY_cmp_parameters()`.
1336
1337   *David von Oheimb and Shane Lontis*
1338
1339 * Deprecated `EC_METHOD_get_field_type()`.
1340
1341   *Billy Bob Brumley*
1342
1343 * Deprecated EC_GFp_simple_method(), EC_GFp_mont_method(),
1344   EC_GF2m_simple_method(), EC_GFp_nist_method(), EC_GFp_nistp224_method()
1345   EC_GFp_nistp256_method(), and EC_GFp_nistp521_method().
1346
1347   *Billy Bob Brumley*
1348
1349 * Deprecated EC_GROUP_new(), EC_GROUP_method_of(), and EC_POINT_method_of().
1350
1351   *Billy Bob Brumley*
1352
1353 * Add CAdES-BES signature verification support, mostly derived
1354   from ESSCertIDv2 TS (RFC 5816) contribution by Marek Klein.
1355
1356   *Filipe Raimundo da Silva*
1357
1358 * Add CAdES-BES signature scheme and attributes support (RFC 5126) to CMS API.
1359
1360   *Antonio Iacono*
1361
1362 * Added the AuthEnvelopedData content type structure (RFC 5083) with AES-GCM
1363   parameter (RFC 5084) for the Cryptographic Message Syntax (CMS).
1364
1365   *Jakub Zelenka*
1366
1367 * Deprecated EC_POINT_make_affine() and EC_POINTs_make_affine().
1368
1369   *Billy Bob Brumley*
1370
1371 * Deprecated EC_GROUP_precompute_mult(), EC_GROUP_have_precompute_mult(), and
1372   EC_KEY_precompute_mult().
1373
1374   *Billy Bob Brumley*
1375
1376 * Deprecated EC_POINTs_mul().
1377
1378   *Billy Bob Brumley*
1379
1380 * Removed FIPS_mode() and FIPS_mode_set().
1381
1382   *Shane Lontis*
1383
1384 * The SSL option SSL_OP_IGNORE_UNEXPECTED_EOF is introduced.
1385
1386   *Dmitry Belyavskiy*
1387
1388 * Deprecated EC_POINT_set_Jprojective_coordinates_GFp() and
1389   EC_POINT_get_Jprojective_coordinates_GFp().
1390
1391   *Billy Bob Brumley*
1392
1393 * Added OSSL_PARAM_BLD to the public interface.  This allows OSSL_PARAM
1394   arrays to be more easily constructed via a series of utility functions.
1395   Create a parameter builder using OSSL_PARAM_BLD_new(), add parameters using
1396   the various push functions and finally convert to a passable OSSL_PARAM
1397   array using OSSL_PARAM_BLD_to_param().
1398
1399   *Paul Dale*
1400
1401 * The security strength of SHA1 and MD5 based signatures in TLS has been
1402   reduced.
1403
1404   *Kurt Roeckx*
1405
1406 * Added EVP_PKEY_set_type_by_keymgmt(), to initialise an EVP_PKEY to
1407   contain a provider side internal key.
1408
1409   *Richard Levitte*
1410
1411 * ASN1_verify(), ASN1_digest() and ASN1_sign() have been deprecated.
1412
1413   *Richard Levitte*
1414
1415 * Project text documents not yet having a proper file name extension
1416   (`HACKING`, `LICENSE`, `NOTES*`, `README*`, `VERSION`) have been renamed to
1417   `*.md` as far as reasonable, else `*.txt`, for better use with file managers.
1418
1419   *David von Oheimb*
1420
1421 * The main project documents (README, NEWS, CHANGES, INSTALL, SUPPORT)
1422   have been converted to Markdown with the goal to produce documents
1423   which not only look pretty when viewed online in the browser, but
1424   remain well readable inside a plain text editor.
1425
1426   To achieve this goal, a 'minimalistic' Markdown style has been applied
1427   which avoids formatting elements that interfere too much with the
1428   reading flow in the text file. For example, it
1429
1430   * avoids [ATX headings][] and uses [setext headings][] instead
1431     (which works for `<h1>` and `<h2>` headings only).
1432   * avoids [inline links][] and uses [reference links][] instead.
1433   * avoids [fenced code blocks][] and uses [indented code blocks][] instead.
1434
1435     [ATX headings]:         https://github.github.com/gfm/#atx-headings
1436     [setext headings]:      https://github.github.com/gfm/#setext-headings
1437     [inline links]:         https://github.github.com/gfm/#inline-link
1438     [reference links]:      https://github.github.com/gfm/#reference-link
1439     [fenced code blocks]:   https://github.github.com/gfm/#fenced-code-blocks
1440     [indented code blocks]: https://github.github.com/gfm/#indented-code-blocks
1441
1442   *Matthias St. Pierre*
1443
1444 * The test suite is changed to preserve results of each test recipe.
1445   A new directory test-runs/ with subdirectories named like the
1446   test recipes are created in the build tree for this purpose.
1447
1448   *Richard Levitte*
1449
1450 * Added an implementation of CMP and CRMF (RFC 4210, RFC 4211 RFC 6712).
1451   This adds `crypto/cmp/`, `crpyto/crmf/`, `apps/cmp.c`, and `test/cmp_*`.
1452   See L<openssl-cmp(1)> and L<OSSL_CMP_exec_IR_ses(3)> as starting points.
1453
1454   *David von Oheimb, Martin Peylo*
1455
1456 * Generalized the HTTP client code from `crypto/ocsp/` into `crpyto/http/`.
1457   It supports arbitrary request and response content types, GET redirection,
1458   TLS, connections via HTTP(S) proxies, connections and exchange via
1459   user-defined BIOs (allowing implicit connections), persistent connections,
1460   and timeout checks.  See L<OSSL_HTTP_transfer(3)> etc. for details.
1461   The legacy OCSP-focused (and only partly documented) API
1462   is retained for backward compatibility, while most of it is deprecated.
1463
1464   *David von Oheimb*
1465
1466 * Added `util/check-format.pl`, a tool for checking adherence to the
1467   OpenSSL coding style <https://www.openssl.org/policies/codingstyle.html>.
1468   The checks performed are incomplete and yield some false positives.
1469   Still the tool should be useful for detecting most typical glitches.
1470
1471   *David von Oheimb*
1472
1473 * `BIO_do_connect()` and `BIO_do_handshake()` have been extended:
1474   If domain name resolution yields multiple IP addresses all of them are tried
1475   after `connect()` failures.
1476
1477   *David von Oheimb*
1478
1479 * All of the low level RSA functions have been deprecated.
1480
1481   *Paul Dale*
1482
1483 * X509 certificates signed using SHA1 are no longer allowed at security
1484   level 1 and above.
1485
1486   *Kurt Roeckx*
1487
1488 * The command line utilities dhparam, dsa, gendsa and dsaparam have been
1489   modified to use PKEY APIs.  These commands are now in maintenance mode
1490   and no new features will be added to them.
1491
1492   *Paul Dale*
1493
1494 * The command line utility rsautl has been deprecated.
1495
1496   *Paul Dale*
1497
1498 * The command line utilities genrsa and rsa have been modified to use PKEY
1499   APIs. They now write PKCS#8 keys by default. These commands are now in
1500   maintenance mode and no new features will be added to them.
1501
1502   *Paul Dale*
1503
1504 * All of the low level DH functions have been deprecated.
1505
1506   *Paul Dale and Matt Caswell*
1507
1508 * All of the low level DSA functions have been deprecated.
1509
1510   *Paul Dale*
1511
1512 * Reworked the treatment of EC EVP_PKEYs with the SM2 curve to
1513   automatically become EVP_PKEY_SM2 rather than EVP_PKEY_EC.
1514
1515   *Richard Levitte*
1516
1517 * Deprecated low level ECDH and ECDSA functions.
1518
1519   *Paul Dale*
1520
1521 * Deprecated EVP_PKEY_decrypt_old() and EVP_PKEY_encrypt_old().
1522
1523   *Richard Levitte*
1524
1525 * Enhanced the documentation of EVP_PKEY_get_size(), EVP_PKEY_get_bits()
1526   and EVP_PKEY_get_security_bits().  Especially EVP_PKEY_get_size() needed
1527   a new formulation to include all the things it can be used for,
1528   as well as words of caution.
1529
1530   *Richard Levitte*
1531
1532 * The SSL_CTX_set_tlsext_ticket_key_cb(3) function has been deprecated.
1533
1534   *Paul Dale*
1535
1536 * All of the low level HMAC functions have been deprecated.
1537
1538   *Paul Dale and David von Oheimb*
1539
1540 * Over two thousand fixes were made to the documentation, including:
1541   - Common options (such as -rand/-writerand, TLS version control, etc)
1542     were refactored and point to newly-enhanced descriptions in openssl.pod.
1543   - Added style conformance for all options (with help from Richard Levitte),
1544     documented all reported missing options, added a CI build to check
1545     that all options are documented and that no unimplemented options
1546     are documented.
1547   - Documented some internals, such as all use of environment variables.
1548   - Addressed all internal broken L<> references.
1549
1550   *Rich Salz*
1551
1552 * All of the low level CMAC functions have been deprecated.
1553
1554   *Paul Dale*
1555
1556 * The low-level MD2, MD4, MD5, MDC2, RIPEMD160 and Whirlpool digest
1557   functions have been deprecated.
1558
1559   *Paul Dale and David von Oheimb*
1560
1561 * Corrected the documentation of the return values from the `EVP_DigestSign*`
1562   set of functions.  The documentation mentioned negative values for some
1563   errors, but this was never the case, so the mention of negative values
1564   was removed.
1565
1566   Code that followed the documentation and thereby check with something
1567   like `EVP_DigestSignInit(...) <= 0` will continue to work undisturbed.
1568
1569   *Richard Levitte*
1570
1571 * All of the low level cipher functions have been deprecated.
1572
1573   *Matt Caswell and Paul Dale*
1574
1575 * Removed include/openssl/opensslconf.h.in and replaced it with
1576   include/openssl/configuration.h.in, which differs in not including
1577   <openssl/macros.h>.  A short header include/openssl/opensslconf.h
1578   was added to include both.
1579
1580   This allows internal hacks where one might need to modify the set
1581   of configured macros, for example this if deprecated symbols are
1582   still supposed to be available internally:
1583
1584       #include <openssl/configuration.h>
1585
1586       #undef OPENSSL_NO_DEPRECATED
1587       #define OPENSSL_SUPPRESS_DEPRECATED
1588
1589       #include <openssl/macros.h>
1590
1591   This should not be used by applications that use the exported
1592   symbols, as that will lead to linking errors.
1593
1594   *Richard Levitte*
1595
1596 * Fixed an overflow bug in the x64_64 Montgomery squaring procedure
1597   used in exponentiation with 512-bit moduli. No EC algorithms are
1598   affected. Analysis suggests that attacks against 2-prime RSA1024,
1599   3-prime RSA1536, and DSA1024 as a result of this defect would be very
1600   difficult to perform and are not believed likely. Attacks against DH512
1601   are considered just feasible. However, for an attack the target would
1602   have to re-use the DH512 private key, which is not recommended anyway.
1603   Also applications directly using the low-level API BN_mod_exp may be
1604   affected if they use BN_FLG_CONSTTIME.
1605   ([CVE-2019-1551])
1606
1607   *Andy Polyakov*
1608
1609 * Most memory-debug features have been deprecated, and the functionality
1610   replaced with no-ops.
1611
1612   *Rich Salz*
1613
1614 * Added documentation for the STACK API.
1615
1616   *Rich Salz*
1617
1618 * Introduced a new method type and API, OSSL_ENCODER, to represent
1619   generic encoders.  These do the same sort of job that PEM writers
1620   and d2i functions do, but with support for methods supplied by
1621   providers, and the possibility for providers to support other
1622   formats as well.
1623
1624   *Richard Levitte*
1625
1626 * Introduced a new method type and API, OSSL_DECODER, to represent
1627   generic decoders.  These do the same sort of job that PEM readers
1628   and i2d functions do, but with support for methods supplied by
1629   providers, and the possibility for providers to support other
1630   formats as well.
1631
1632   *Richard Levitte*
1633
1634 * Added a .pragma directive to the syntax of configuration files, to
1635   allow varying behavior in a supported and predictable manner.
1636   Currently added pragma:
1637
1638           .pragma dollarid:on
1639
1640   This allows dollar signs to be a keyword character unless it's
1641   followed by a opening brace or parenthesis.  This is useful for
1642   platforms where dollar signs are commonly used in names, such as
1643   volume names and system directory names on VMS.
1644
1645   *Richard Levitte*
1646
1647 * Added functionality to create an EVP_PKEY from user data.
1648
1649   *Richard Levitte*
1650
1651 * Change the interpretation of the '--api' configuration option to
1652   mean that this is a desired API compatibility level with no
1653   further meaning.  The previous interpretation, that this would
1654   also mean to remove all deprecated symbols up to and including
1655   the given version, no requires that 'no-deprecated' is also used
1656   in the configuration.
1657
1658   When building applications, the desired API compatibility level
1659   can be set with the OPENSSL_API_COMPAT macro like before.  For
1660   API compatibility version below 3.0, the old style numerical
1661   value is valid as before, such as -DOPENSSL_API_COMPAT=0x10100000L.
1662   For version 3.0 and on, the value is expected to be the decimal
1663   value calculated from the major and minor version like this:
1664
1665           MAJOR * 10000 + MINOR * 100
1666
1667   Examples:
1668
1669           -DOPENSSL_API_COMPAT=30000             For 3.0
1670           -DOPENSSL_API_COMPAT=30200             For 3.2
1671
1672   To hide declarations that are deprecated up to and including the
1673   given API compatibility level, -DOPENSSL_NO_DEPRECATED must be
1674   given when building the application as well.
1675
1676   *Richard Levitte*
1677
1678 * Added the X509_LOOKUP_METHOD called X509_LOOKUP_store, to allow
1679   access to certificate and CRL stores via URIs and OSSL_STORE
1680   loaders.
1681
1682   This adds the following functions:
1683
1684   - X509_LOOKUP_store()
1685   - X509_STORE_load_file()
1686   - X509_STORE_load_path()
1687   - X509_STORE_load_store()
1688   - SSL_add_store_cert_subjects_to_stack()
1689   - SSL_CTX_set_default_verify_store()
1690   - SSL_CTX_load_verify_file()
1691   - SSL_CTX_load_verify_dir()
1692   - SSL_CTX_load_verify_store()
1693
1694   *Richard Levitte*
1695
1696 * Added a new method to gather entropy on VMS, based on SYS$GET_ENTROPY.
1697   The presence of this system service is determined at run-time.
1698
1699   *Richard Levitte*
1700
1701 * Added functionality to create an EVP_PKEY context based on data
1702   for methods from providers.  This takes an algorithm name and a
1703   property query string and simply stores them, with the intent
1704   that any operation that uses this context will use those strings
1705   to fetch the needed methods implicitly, thereby making the port
1706   of application written for pre-3.0 OpenSSL easier.
1707
1708   *Richard Levitte*
1709
1710 * The undocumented function NCONF_WIN32() has been deprecated; for
1711   conversion details see the HISTORY section of doc/man5/config.pod
1712
1713   *Rich Salz*
1714
1715 * Introduced the new functions EVP_DigestSignInit_ex() and
1716   EVP_DigestVerifyInit_ex(). The macros EVP_DigestSignUpdate() and
1717   EVP_DigestVerifyUpdate() have been converted to functions. See the man
1718   pages for further details.
1719
1720   *Matt Caswell*
1721
1722 * Over two thousand fixes were made to the documentation, including:
1723   adding missing command flags, better style conformance, documentation
1724   of internals, etc.
1725
1726   *Rich Salz, Richard Levitte*
1727
1728 * s390x assembly pack: add hardware-support for P-256, P-384, P-521,
1729   X25519, X448, Ed25519 and Ed448.
1730
1731   *Patrick Steuer*
1732
1733 * Print all values for a PKCS#12 attribute with 'openssl pkcs12', not just
1734   the first value.
1735
1736   *Jon Spillett*
1737
1738 * Deprecated the public definition of `ERR_STATE` as well as the function
1739   `ERR_get_state()`.  This is done in preparation of making `ERR_STATE` an
1740   opaque type.
1741
1742   *Richard Levitte*
1743
1744 * Added ERR functionality to give callers access to the stored function
1745   names that have replaced the older function code based functions.
1746
1747   New functions are ERR_peek_error_func(), ERR_peek_last_error_func(),
1748   ERR_peek_error_data(), ERR_peek_last_error_data(), ERR_get_error_all(),
1749   ERR_peek_error_all() and ERR_peek_last_error_all().
1750
1751   Deprecate ERR functions ERR_get_error_line(), ERR_get_error_line_data(),
1752   ERR_peek_error_line_data(), ERR_peek_last_error_line_data() and
1753   ERR_func_error_string().
1754
1755   *Richard Levitte*
1756
1757 * Extended testing to be verbose for failing tests only.  The make variables
1758   VERBOSE_FAILURE or VF can be used to enable this:
1759
1760           $ make VF=1 test                           # Unix
1761           $ mms /macro=(VF=1) test                   ! OpenVMS
1762           $ nmake VF=1 test                          # Windows
1763
1764   *Richard Levitte*
1765
1766 * Added the `-copy_extensions` option to the `x509` command for use with
1767   `-req` and `-x509toreq`. When given with the `copy` or `copyall` argument,
1768   all extensions in the request are copied to the certificate or vice versa.
1769
1770   *David von Oheimb*, *Kirill Stefanenkov <kirill_stefanenkov@rambler.ru>*
1771
1772 * Added the `-copy_extensions` option to the `req` command for use with
1773   `-x509`. When given with the `copy` or `copyall` argument,
1774   all extensions in the certification request are copied to the certificate.
1775
1776   *David von Oheimb*
1777
1778 * The `x509`, `req`, and `ca` commands now make sure that X.509v3 certificates
1779   they generate are by default RFC 5280 compliant in the following sense:
1780   There is a subjectKeyIdentifier extension with a hash value of the public key
1781   and for not self-signed certs there is an authorityKeyIdentifier extension
1782   with a keyIdentifier field or issuer information identifying the signing key.
1783   This is done unless some configuration overrides the new default behavior,
1784   such as `subjectKeyIdentifier = none` and `authorityKeyIdentifier = none`.
1785
1786   *David von Oheimb*
1787
1788 * Added several checks to `X509_verify_cert()` according to requirements in
1789   RFC 5280 in case `X509_V_FLAG_X509_STRICT` is set
1790   (which may be done by using the CLI option `-x509_strict`):
1791   * The basicConstraints of CA certificates must be marked critical.
1792   * CA certificates must explicitly include the keyUsage extension.
1793   * If a pathlenConstraint is given the key usage keyCertSign must be allowed.
1794   * The issuer name of any certificate must not be empty.
1795   * The subject name of CA certs, certs with keyUsage crlSign,
1796     and certs without subjectAlternativeName must not be empty.
1797   * If a subjectAlternativeName extension is given it must not be empty.
1798   * The signatureAlgorithm field and the cert signature must be consistent.
1799   * Any given authorityKeyIdentifier and any given subjectKeyIdentifier
1800     must not be marked critical.
1801   * The authorityKeyIdentifier must be given for X.509v3 certs
1802     unless they are self-signed.
1803   * The subjectKeyIdentifier must be given for all X.509v3 CA certs.
1804
1805   *David von Oheimb*
1806
1807 * Certificate verification using `X509_verify_cert()` meanwhile rejects EC keys
1808   with explicit curve parameters (specifiedCurve) as required by RFC 5480.
1809
1810   *Tomáš Mráz*
1811
1812 * For built-in EC curves, ensure an EC_GROUP built from the curve name is
1813   used even when parsing explicit parameters, when loading a encoded key
1814   or calling `EC_GROUP_new_from_ecpkparameters()`/
1815   `EC_GROUP_new_from_ecparameters()`.
1816   This prevents bypass of security hardening and performance gains,
1817   especially for curves with specialized EC_METHODs.
1818   By default, if a key encoded with explicit parameters is loaded and later
1819   encoded, the output is still encoded with explicit parameters, even if
1820   internally a "named" EC_GROUP is used for computation.
1821
1822   *Nicola Tuveri*
1823
1824 * Compute ECC cofactors if not provided during EC_GROUP construction. Before
1825   this change, EC_GROUP_set_generator would accept order and/or cofactor as
1826   NULL. After this change, only the cofactor parameter can be NULL. It also
1827   does some minimal sanity checks on the passed order.
1828   ([CVE-2019-1547])
1829
1830   *Billy Bob Brumley*
1831
1832 * Fixed a padding oracle in PKCS7_dataDecode and CMS_decrypt_set1_pkey.
1833   An attack is simple, if the first CMS_recipientInfo is valid but the
1834   second CMS_recipientInfo is chosen ciphertext. If the second
1835   recipientInfo decodes to PKCS #1 v1.5 form plaintext, the correct
1836   encryption key will be replaced by garbage, and the message cannot be
1837   decoded, but if the RSA decryption fails, the correct encryption key is
1838   used and the recipient will not notice the attack.
1839   As a work around for this potential attack the length of the decrypted
1840   key must be equal to the cipher default key length, in case the
1841   certifiate is not given and all recipientInfo are tried out.
1842   The old behaviour can be re-enabled in the CMS code by setting the
1843   CMS_DEBUG_DECRYPT flag.
1844
1845   *Bernd Edlinger*
1846
1847 * Early start up entropy quality from the DEVRANDOM seed source has been
1848   improved for older Linux systems.  The RAND subsystem will wait for
1849   /dev/random to be producing output before seeding from /dev/urandom.
1850   The seeded state is stored for future library initialisations using
1851   a system global shared memory segment.  The shared memory identifier
1852   can be configured by defining OPENSSL_RAND_SEED_DEVRANDOM_SHM_ID to
1853   the desired value.  The default identifier is 114.
1854
1855   *Paul Dale*
1856
1857 * Revised BN_generate_prime_ex to not avoid factors 2..17863 in p-1
1858   when primes for RSA keys are computed.
1859   Since we previously always generated primes == 2 (mod 3) for RSA keys,
1860   the 2-prime and 3-prime RSA modules were easy to distinguish, since
1861   `N = p*q = 1 (mod 3)`, but `N = p*q*r = 2 (mod 3)`. Therefore fingerprinting
1862   2-prime vs. 3-prime RSA keys was possible by computing N mod 3.
1863   This avoids possible fingerprinting of newly generated RSA modules.
1864
1865   *Bernd Edlinger*
1866
1867 * Correct the extended master secret constant on EBCDIC systems. Without this
1868   fix TLS connections between an EBCDIC system and a non-EBCDIC system that
1869   negotiate EMS will fail. Unfortunately this also means that TLS connections
1870   between EBCDIC systems with this fix, and EBCDIC systems without this
1871   fix will fail if they negotiate EMS.
1872
1873   *Matt Caswell*
1874
1875 * Changed the library initialisation so that the config file is now loaded
1876   by default. This was already the case for libssl. It now occurs for both
1877   libcrypto and libssl. Use the OPENSSL_INIT_NO_LOAD_CONFIG option to
1878   `OPENSSL_init_crypto()` to suppress automatic loading of a config file.
1879
1880   *Matt Caswell*
1881
1882 * Introduced new error raising macros, `ERR_raise()` and `ERR_raise_data()`,
1883   where the former acts as a replacement for `ERR_put_error()`, and the
1884   latter replaces the combination `ERR_put_error()` + `ERR_add_error_data()`.
1885   `ERR_raise_data()` adds more flexibility by taking a format string and
1886   an arbitrary number of arguments following it, to be processed with
1887   `BIO_snprintf()`.
1888
1889   *Richard Levitte*
1890
1891 * Introduced a new function, `OSSL_PROVIDER_available()`, which can be used
1892   to check if a named provider is loaded and available.  When called, it
1893   will also activate all fallback providers if such are still present.
1894
1895   *Richard Levitte*
1896
1897 * Enforce a minimum DH modulus size of 512 bits.
1898
1899   *Bernd Edlinger*
1900
1901 * Changed DH parameters to generate the order q subgroup instead of 2q.
1902   Previously generated DH parameters are still accepted by DH_check
1903   but DH_generate_key works around that by clearing bit 0 of the
1904   private key for those. This avoids leaking bit 0 of the private key.
1905
1906   *Bernd Edlinger*
1907
1908 * Significantly reduce secure memory usage by the randomness pools.
1909
1910   *Paul Dale*
1911
1912 * `{CRYPTO,OPENSSL}_mem_debug_{push,pop}` are now no-ops and have been
1913   deprecated.
1914
1915   *Rich Salz*
1916
1917 * A new type, EVP_KEYEXCH, has been introduced to represent key exchange
1918   algorithms. An implementation of a key exchange algorithm can be obtained
1919   by using the function EVP_KEYEXCH_fetch(). An EVP_KEYEXCH algorithm can be
1920   used in a call to EVP_PKEY_derive_init_ex() which works in a similar way to
1921   the older EVP_PKEY_derive_init() function. See the man pages for the new
1922   functions for further details.
1923
1924   *Matt Caswell*
1925
1926 * The EVP_PKEY_CTX_set_dh_pad() macro has now been converted to a function.
1927
1928   *Matt Caswell*
1929
1930 * Removed the function names from error messages and deprecated the
1931   xxx_F_xxx define's.
1932
1933   *Richard Levitte*
1934
1935 * Removed NextStep support and the macro OPENSSL_UNISTD
1936
1937   *Rich Salz*
1938
1939 * Removed DES_check_key.  Also removed OPENSSL_IMPLEMENT_GLOBAL,
1940   OPENSSL_GLOBAL_REF, OPENSSL_DECLARE_GLOBAL.
1941   Also removed "export var as function" capability; we do not export
1942   variables, only functions.
1943
1944   *Rich Salz*
1945
1946 * RC5_32_set_key has been changed to return an int type, with 0 indicating
1947   an error and 1 indicating success. In previous versions of OpenSSL this
1948   was a void type. If a key was set longer than the maximum possible this
1949   would crash.
1950
1951   *Matt Caswell*
1952
1953 * Support SM2 signing and verification schemes with X509 certificate.
1954
1955   *Paul Yang*
1956
1957 * Use SHA256 as the default digest for TS query in the `ts` app.
1958
1959   *Tomáš Mráz*
1960
1961 * Change PBKDF2 to conform to SP800-132 instead of the older PKCS5 RFC2898.
1962
1963   *Shane Lontis*
1964
1965 * Default cipher lists/suites are now available via a function, the
1966   #defines are deprecated.
1967
1968   *Todd Short*
1969
1970 * Add target VC-WIN32-UWP, VC-WIN64A-UWP, VC-WIN32-ARM-UWP and
1971   VC-WIN64-ARM-UWP in Windows OneCore target for making building libraries
1972   for Windows Store apps easier. Also, the "no-uplink" option has been added.
1973
1974   *Kenji Mouri*
1975
1976 * Join the directories crypto/x509 and crypto/x509v3
1977
1978   *Richard Levitte*
1979
1980 * Added command 'openssl kdf' that uses the EVP_KDF API.
1981
1982   *Shane Lontis*
1983
1984 * Added command 'openssl mac' that uses the EVP_MAC API.
1985
1986   *Shane Lontis*
1987
1988 * Added OPENSSL_info() to get diverse built-in OpenSSL data, such
1989   as default directories.  Also added the command 'openssl info'
1990   for scripting purposes.
1991
1992   *Richard Levitte*
1993
1994 * The functions AES_ige_encrypt() and AES_bi_ige_encrypt() have been
1995   deprecated.
1996
1997   *Matt Caswell*
1998
1999 * Add prediction resistance to the DRBG reseeding process.
2000
2001   *Paul Dale*
2002
2003 * Limit the number of blocks in a data unit for AES-XTS to 2^20 as
2004   mandated by IEEE Std 1619-2018.
2005
2006   *Paul Dale*
2007
2008 * Added newline escaping functionality to a filename when using openssl dgst.
2009   This output format is to replicate the output format found in the `*sum`
2010   checksum programs. This aims to preserve backward compatibility.
2011
2012   *Matt Eaton, Richard Levitte, and Paul Dale*
2013
2014 * Removed the heartbeat message in DTLS feature, as it has very
2015   little usage and doesn't seem to fulfill a valuable purpose.
2016   The configuration option is now deprecated.
2017
2018   *Richard Levitte*
2019
2020 * Changed the output of 'openssl {digestname} < file' to display the
2021   digest name in its output.
2022
2023   *Richard Levitte*
2024
2025 * Added a new generic trace API which provides support for enabling
2026   instrumentation through trace output.
2027
2028   *Richard Levitte & Matthias St. Pierre*
2029
2030 * Added build tests for C++.  These are generated files that only do one
2031   thing, to include one public OpenSSL head file each.  This tests that
2032   the public header files can be usefully included in a C++ application.
2033
2034   This test isn't enabled by default.  It can be enabled with the option
2035   'enable-buildtest-c++'.
2036
2037   *Richard Levitte*
2038
2039 * Added KB KDF (EVP_KDF_KB) to EVP_KDF.
2040
2041   *Robbie Harwood*
2042
2043 * Added SSH KDF (EVP_KDF_SSHKDF) and KRB5 KDF (EVP_KDF_KRB5KDF) to EVP_KDF.
2044
2045   *Simo Sorce*
2046
2047 * Added Single Step KDF (EVP_KDF_SS), X963 KDF, and X942 KDF to EVP_KDF.
2048
2049   *Shane Lontis*
2050
2051 * Added KMAC to EVP_MAC.
2052
2053   *Shane Lontis*
2054
2055 * Added property based algorithm implementation selection framework to
2056   the core.
2057
2058   *Paul Dale*
2059
2060 * Added SCA hardening for modular field inversion in EC_GROUP through
2061   a new dedicated field_inv() pointer in EC_METHOD.
2062   This also addresses a leakage affecting conversions from projective
2063   to affine coordinates.
2064
2065   *Billy Bob Brumley, Nicola Tuveri*
2066
2067 * Added EVP_KDF, an EVP layer KDF API, to simplify adding KDF and PRF
2068   implementations.  This includes an EVP_PKEY to EVP_KDF bridge for
2069   those algorithms that were already supported through the EVP_PKEY API
2070   (scrypt, TLS1 PRF and HKDF).  The low-level KDF functions for PBKDF2
2071   and scrypt are now wrappers that call EVP_KDF.
2072
2073   *David Makepeace*
2074
2075 * Build devcrypto engine as a dynamic engine.
2076
2077   *Eneas U de Queiroz*
2078
2079 * Add keyed BLAKE2 to EVP_MAC.
2080
2081   *Antoine Salon*
2082
2083 * Fix a bug in the computation of the endpoint-pair shared secret used
2084   by DTLS over SCTP. This breaks interoperability with older versions
2085   of OpenSSL like OpenSSL 1.1.0 and OpenSSL 1.0.2. There is a runtime
2086   switch SSL_MODE_DTLS_SCTP_LABEL_LENGTH_BUG (off by default) enabling
2087   interoperability with such broken implementations. However, enabling
2088   this switch breaks interoperability with correct implementations.
2089
2090 * Fix a use after free bug in d2i_X509_PUBKEY when overwriting a
2091   re-used X509_PUBKEY object if the second PUBKEY is malformed.
2092
2093   *Bernd Edlinger*
2094
2095 * Move strictness check from EVP_PKEY_asn1_new() to EVP_PKEY_asn1_add0().
2096
2097   *Richard Levitte*
2098
2099 * Changed the license to the Apache License v2.0.
2100
2101   *Richard Levitte*
2102
2103 * Switch to a new version scheme using three numbers MAJOR.MINOR.PATCH.
2104
2105   - Major releases (indicated by incrementing the MAJOR release number)
2106     may introduce incompatible API/ABI changes.
2107   - Minor releases (indicated by incrementing the MINOR release number)
2108     may introduce new features but retain API/ABI compatibility.
2109   - Patch releases (indicated by incrementing the PATCH number)
2110     are intended for bug fixes and other improvements of existing
2111     features only (like improving performance or adding documentation)
2112     and retain API/ABI compatibility.
2113
2114   *Richard Levitte*
2115
2116 * Add support for RFC5297 SIV mode (siv128), including AES-SIV.
2117
2118   *Todd Short*
2119
2120 * Remove the 'dist' target and add a tarball building script.  The
2121   'dist' target has fallen out of use, and it shouldn't be
2122   necessary to configure just to create a source distribution.
2123
2124   *Richard Levitte*
2125
2126 * Recreate the OS390-Unix config target.  It no longer relies on a
2127   special script like it did for OpenSSL pre-1.1.0.
2128
2129   *Richard Levitte*
2130
2131 * Instead of having the source directories listed in Configure, add
2132   a 'build.info' keyword SUBDIRS to indicate what sub-directories to
2133   look into.
2134
2135   *Richard Levitte*
2136
2137 * Add GMAC to EVP_MAC.
2138
2139   *Paul Dale*
2140
2141 * Ported the HMAC, CMAC and SipHash EVP_PKEY_METHODs to EVP_MAC.
2142
2143   *Richard Levitte*
2144
2145 * Added EVP_MAC, an EVP layer MAC API, to simplify adding MAC
2146   implementations.  This includes a generic EVP_PKEY to EVP_MAC bridge,
2147   to facilitate the continued use of MACs through raw private keys in
2148   functionality such as `EVP_DigestSign*` and `EVP_DigestVerify*`.
2149
2150   *Richard Levitte*
2151
2152 * Deprecate ECDH_KDF_X9_62().
2153
2154   *Antoine Salon*
2155
2156 * Added EVP_PKEY_ECDH_KDF_X9_63 and ecdh_KDF_X9_63() as replacements for
2157   the EVP_PKEY_ECDH_KDF_X9_62 KDF type and ECDH_KDF_X9_62(). The old names
2158   are retained for backwards compatibility.
2159
2160   *Antoine Salon*
2161
2162 * AES-XTS mode now enforces that its two keys are different to mitigate
2163   the attacked described in "Efficient Instantiations of Tweakable
2164   Blockciphers and Refinements to Modes OCB and PMAC" by Phillip Rogaway.
2165   Details of this attack can be obtained from:
2166   <http://web.cs.ucdavis.edu/%7Erogaway/papers/offsets.pdf>
2167
2168   *Paul Dale*
2169
2170 * Rename the object files, i.e. give them other names than in previous
2171   versions.  Their names now include the name of the final product, as
2172   well as its type mnemonic (bin, lib, shlib).
2173
2174   *Richard Levitte*
2175
2176 * Added new option for 'openssl list', '-objects', which will display the
2177   list of built in objects, i.e. OIDs with names.
2178
2179   *Richard Levitte*
2180
2181 * Added the options `-crl_lastupdate` and `-crl_nextupdate` to `openssl ca`,
2182   allowing the `lastUpdate` and `nextUpdate` fields in the generated CRL to
2183   be set explicitly.
2184
2185   *Chris Novakovic*
2186
2187 * Added support for Linux Kernel TLS data-path. The Linux Kernel data-path
2188   improves application performance by removing data copies and providing
2189   applications with zero-copy system calls such as sendfile and splice.
2190
2191   *Boris Pismenny*
2192
2193 * The SSL option SSL_OP_CLEANSE_PLAINTEXT is introduced.
2194
2195   *Martin Elshuber*
2196
2197 * `PKCS12_parse` now maintains the order of the parsed certificates
2198   when outputting them via `*ca` (rather than reversing it).
2199
2200   *David von Oheimb*
2201
2202 * Deprecated pthread fork support methods.
2203
2204   *Randall S. Becker*
2205
2206 * Added support for FFDHE key exchange in TLS 1.3.
2207
2208   *Raja Ashok*
2209
2210 * Added a new concept for OpenSSL plugability: providers.  This
2211   functionality is designed to replace the ENGINE API and ENGINE
2212   implementations, and to be much more dynamic, allowing provider
2213   authors to introduce new algorithms among other things, as long as
2214   there's an API that supports the algorithm type.
2215
2216   With this concept comes a new core API for interaction between
2217   libcrypto and provider implementations.  Public libcrypto functions
2218   that want to use providers do so through this core API.
2219
2220   The main documentation for this core API is found in
2221   doc/man7/provider.pod, doc/man7/provider-base.pod, and they in turn
2222   refer to other manuals describing the API specific for supported
2223   algorithm types (also called operations).
2224
2225   *The OpenSSL team*
2226
2227OpenSSL 1.1.1
2228-------------
2229
2230### Changes between 1.1.1l and 1.1.1m [xx XXX xxxx]
2231
2232 * Avoid loading of a dynamic engine twice.
2233
2234   *Bernd Edlinger*
2235
2236 * Prioritise DANE TLSA issuer certs over peer certs
2237
2238   *Viktor Dukhovni*
2239
2240 * Fixed random API for MacOS prior to 10.12
2241
2242   These MacOS versions don't support the CommonCrypto APIs
2243
2244   *Lenny Primak*
2245
2246### Changes between 1.1.1k and 1.1.1l [24 Aug 2021]
2247
2248 * Fixed an SM2 Decryption Buffer Overflow.
2249
2250   In order to decrypt SM2 encrypted data an application is expected to
2251   call the API function EVP_PKEY_decrypt(). Typically an application will
2252   call this function twice. The first time, on entry, the "out" parameter
2253   can be NULL and, on exit, the "outlen" parameter is populated with the
2254   buffer size required to hold the decrypted plaintext. The application
2255   can then allocate a sufficiently sized buffer and call EVP_PKEY_decrypt()
2256   again, but this time passing a non-NULL value for the "out" parameter.
2257
2258   A bug in the implementation of the SM2 decryption code means that the
2259   calculation of the buffer size required to hold the plaintext returned
2260   by the first call to EVP_PKEY_decrypt() can be smaller than the actual
2261   size required by the second call. This can lead to a buffer overflow
2262   when EVP_PKEY_decrypt() is called by the application a second time with
2263   a buffer that is too small.
2264
2265   A malicious attacker who is able present SM2 content for decryption to
2266   an application could cause attacker chosen data to overflow the buffer
2267   by up to a maximum of 62 bytes altering the contents of other data held
2268   after the buffer, possibly changing application behaviour or causing
2269   the application to crash. The location of the buffer is application
2270   dependent but is typically heap allocated.
2271   ([CVE-2021-3711])
2272
2273   *Matt Caswell*
2274
2275 * Fixed various read buffer overruns processing ASN.1 strings
2276
2277   ASN.1 strings are represented internally within OpenSSL as an ASN1_STRING
2278   structure which contains a buffer holding the string data and a field
2279   holding the buffer length. This contrasts with normal C strings which
2280   are repesented as a buffer for the string data which is terminated
2281   with a NUL (0) byte.
2282
2283   Although not a strict requirement, ASN.1 strings that are parsed using
2284   OpenSSL's own "d2i" functions (and other similar parsing functions) as
2285   well as any string whose value has been set with the ASN1_STRING_set()
2286   function will additionally NUL terminate the byte array in the
2287   ASN1_STRING structure.
2288
2289   However, it is possible for applications to directly construct valid
2290   ASN1_STRING structures which do not NUL terminate the byte array by
2291   directly setting the "data" and "length" fields in the ASN1_STRING
2292   array. This can also happen by using the ASN1_STRING_set0() function.
2293
2294   Numerous OpenSSL functions that print ASN.1 data have been found to
2295   assume that the ASN1_STRING byte array will be NUL terminated, even
2296   though this is not guaranteed for strings that have been directly
2297   constructed. Where an application requests an ASN.1 structure to be
2298   printed, and where that ASN.1 structure contains ASN1_STRINGs that have
2299   been directly constructed by the application without NUL terminating
2300   the "data" field, then a read buffer overrun can occur.
2301
2302   The same thing can also occur during name constraints processing
2303   of certificates (for example if a certificate has been directly
2304   constructed by the application instead of loading it via the OpenSSL
2305   parsing functions, and the certificate contains non NUL terminated
2306   ASN1_STRING structures). It can also occur in the X509_get1_email(),
2307   X509_REQ_get1_email() and X509_get1_ocsp() functions.
2308
2309   If a malicious actor can cause an application to directly construct an
2310   ASN1_STRING and then process it through one of the affected OpenSSL
2311   functions then this issue could be hit. This might result in a crash
2312   (causing a Denial of Service attack). It could also result in the
2313   disclosure of private memory contents (such as private keys, or
2314   sensitive plaintext).
2315   ([CVE-2021-3712])
2316
2317   *Matt Caswell*
2318
2319### Changes between 1.1.1j and 1.1.1k [25 Mar 2021]
2320
2321 * Fixed a problem with verifying a certificate chain when using the
2322   X509_V_FLAG_X509_STRICT flag. This flag enables additional security checks of
2323   the certificates present in a certificate chain. It is not set by default.
2324
2325   Starting from OpenSSL version 1.1.1h a check to disallow certificates in
2326   the chain that have explicitly encoded elliptic curve parameters was added
2327   as an additional strict check.
2328
2329   An error in the implementation of this check meant that the result of a
2330   previous check to confirm that certificates in the chain are valid CA
2331   certificates was overwritten. This effectively bypasses the check
2332   that non-CA certificates must not be able to issue other certificates.
2333
2334   If a "purpose" has been configured then there is a subsequent opportunity
2335   for checks that the certificate is a valid CA.  All of the named "purpose"
2336   values implemented in libcrypto perform this check.  Therefore, where
2337   a purpose is set the certificate chain will still be rejected even when the
2338   strict flag has been used. A purpose is set by default in libssl client and
2339   server certificate verification routines, but it can be overridden or
2340   removed by an application.
2341
2342   In order to be affected, an application must explicitly set the
2343   X509_V_FLAG_X509_STRICT verification flag and either not set a purpose
2344   for the certificate verification or, in the case of TLS client or server
2345   applications, override the default purpose.
2346   ([CVE-2021-3450])
2347
2348   *Tomáš Mráz*
2349
2350 * Fixed an issue where an OpenSSL TLS server may crash if sent a maliciously
2351   crafted renegotiation ClientHello message from a client. If a TLSv1.2
2352   renegotiation ClientHello omits the signature_algorithms extension (where it
2353   was present in the initial ClientHello), but includes a
2354   signature_algorithms_cert extension then a NULL pointer dereference will
2355   result, leading to a crash and a denial of service attack.
2356
2357   A server is only vulnerable if it has TLSv1.2 and renegotiation enabled
2358   (which is the default configuration). OpenSSL TLS clients are not impacted by
2359   this issue.
2360   ([CVE-2021-3449])
2361
2362   *Peter Kästle and Samuel Sapalski*
2363
2364### Changes between 1.1.1i and 1.1.1j [16 Feb 2021]
2365
2366 * Fixed the X509_issuer_and_serial_hash() function. It attempts to
2367   create a unique hash value based on the issuer and serial number data
2368   contained within an X509 certificate. However it was failing to correctly
2369   handle any errors that may occur while parsing the issuer field (which might
2370   occur if the issuer field is maliciously constructed). This may subsequently
2371   result in a NULL pointer deref and a crash leading to a potential denial of
2372   service attack.
2373   ([CVE-2021-23841])
2374
2375   *Matt Caswell*
2376
2377 * Fixed the RSA_padding_check_SSLv23() function and the RSA_SSLV23_PADDING
2378   padding mode to correctly check for rollback attacks. This is considered a
2379   bug in OpenSSL 1.1.1 because it does not support SSLv2. In 1.0.2 this is
2380   CVE-2021-23839.
2381
2382   *Matt Caswell*
2383
2384   Fixed the EVP_CipherUpdate, EVP_EncryptUpdate and EVP_DecryptUpdate
2385   functions. Previously they could overflow the output length argument in some
2386   cases where the input length is close to the maximum permissable length for
2387   an integer on the platform. In such cases the return value from the function
2388   call would be 1 (indicating success), but the output length value would be
2389   negative. This could cause applications to behave incorrectly or crash.
2390   ([CVE-2021-23840])
2391
2392   *Matt Caswell*
2393
2394 * Fixed SRP_Calc_client_key so that it runs in constant time. The previous
2395   implementation called BN_mod_exp without setting BN_FLG_CONSTTIME. This
2396   could be exploited in a side channel attack to recover the password. Since
2397   the attack is local host only this is outside of the current OpenSSL
2398   threat model and therefore no CVE is assigned.
2399
2400   Thanks to Mohammed Sabt and Daniel De Almeida Braga for reporting this
2401   issue.
2402
2403   *Matt Caswell*
2404
2405### Changes between 1.1.1h and 1.1.1i [8 Dec 2020]
2406
2407 * Fixed NULL pointer deref in the GENERAL_NAME_cmp function
2408   This function could crash if both GENERAL_NAMEs contain an EDIPARTYNAME.
2409    If an attacker can control both items being compared  then this could lead
2410    to a possible denial of service attack. OpenSSL itself uses the
2411    GENERAL_NAME_cmp function for two purposes:
2412    1) Comparing CRL distribution point names between an available CRL and a
2413       CRL distribution point embedded in an X509 certificate
2414    2) When verifying that a timestamp response token signer matches the
2415       timestamp authority name (exposed via the API functions
2416       TS_RESP_verify_response and TS_RESP_verify_token)
2417   ([CVE-2020-1971])
2418
2419   *Matt Caswell*
2420
2421### Changes between 1.1.1g and 1.1.1h [22 Sep 2020]
2422
2423 * Certificates with explicit curve parameters are now disallowed in
2424   verification chains if the X509_V_FLAG_X509_STRICT flag is used.
2425
2426   *Tomáš Mráz*
2427
2428 * The 'MinProtocol' and 'MaxProtocol' configuration commands now silently
2429   ignore TLS protocol version bounds when configuring DTLS-based contexts, and
2430   conversely, silently ignore DTLS protocol version bounds when configuring
2431   TLS-based contexts.  The commands can be repeated to set bounds of both
2432   types.  The same applies with the corresponding "min_protocol" and
2433   "max_protocol" command-line switches, in case some application uses both TLS
2434   and DTLS.
2435
2436   SSL_CTX instances that are created for a fixed protocol version (e.g.
2437   TLSv1_server_method()) also silently ignore version bounds.  Previously
2438   attempts to apply bounds to these protocol versions would result in an
2439   error.  Now only the "version-flexible" SSL_CTX instances are subject to
2440   limits in configuration files in command-line options.
2441
2442   *Viktor Dukhovni*
2443
2444 * Handshake now fails if Extended Master Secret extension is dropped
2445   on renegotiation.
2446
2447   *Tomáš Mráz*
2448
2449 * The Oracle Developer Studio compiler will start reporting deprecated APIs
2450
2451### Changes between 1.1.1f and 1.1.1g [21 Apr 2020]
2452
2453 * Fixed segmentation fault in SSL_check_chain()
2454   Server or client applications that call the SSL_check_chain() function
2455   during or after a TLS 1.3 handshake may crash due to a NULL pointer
2456   dereference as a result of incorrect handling of the
2457   "signature_algorithms_cert" TLS extension. The crash occurs if an invalid
2458   or unrecognised signature algorithm is received from the peer. This could
2459   be exploited by a malicious peer in a Denial of Service attack.
2460   ([CVE-2020-1967])
2461
2462   *Benjamin Kaduk*
2463
2464 * Added AES consttime code for no-asm configurations
2465   an optional constant time support for AES was added
2466   when building openssl for no-asm.
2467   Enable with: ./config no-asm -DOPENSSL_AES_CONST_TIME
2468   Disable with: ./config no-asm -DOPENSSL_NO_AES_CONST_TIME
2469   At this time this feature is by default disabled.
2470   It will be enabled by default in 3.0.
2471
2472   *Bernd Edlinger*
2473
2474### Changes between 1.1.1e and 1.1.1f [31 Mar 2020]
2475
2476 * Revert the change of EOF detection while reading in libssl to avoid
2477   regressions in applications depending on the current way of reporting
2478   the EOF. As the existing method is not fully accurate the change to
2479   reporting the EOF via SSL_ERROR_SSL is kept on the current development
2480   branch and will be present in the 3.0 release.
2481
2482   *Tomáš Mráz*
2483
2484 * Revised BN_generate_prime_ex to not avoid factors 3..17863 in p-1
2485   when primes for RSA keys are computed.
2486   Since we previously always generated primes == 2 (mod 3) for RSA keys,
2487   the 2-prime and 3-prime RSA modules were easy to distinguish, since
2488   N = p*q = 1 (mod 3), but N = p*q*r = 2 (mod 3). Therefore fingerprinting
2489   2-prime vs. 3-prime RSA keys was possible by computing N mod 3.
2490   This avoids possible fingerprinting of newly generated RSA modules.
2491
2492   *Bernd Edlinger*
2493
2494### Changes between 1.1.1d and 1.1.1e [17 Mar 2020]
2495
2496 * Properly detect EOF while reading in libssl. Previously if we hit an EOF
2497   while reading in libssl then we would report an error back to the
2498   application (SSL_ERROR_SYSCALL) but errno would be 0. We now add
2499   an error to the stack (which means we instead return SSL_ERROR_SSL) and
2500   therefore give a hint as to what went wrong.
2501
2502   *Matt Caswell*
2503
2504 * Check that ed25519 and ed448 are allowed by the security level. Previously
2505   signature algorithms not using an MD were not being checked that they were
2506   allowed by the security level.
2507
2508   *Kurt Roeckx*
2509
2510 * Fixed SSL_get_servername() behaviour. The behaviour of SSL_get_servername()
2511   was not quite right. The behaviour was not consistent between resumption
2512   and normal handshakes, and also not quite consistent with historical
2513   behaviour. The behaviour in various scenarios has been clarified and
2514   it has been updated to make it match historical behaviour as closely as
2515   possible.
2516
2517   *Matt Caswell*
2518
2519 * *[VMS only]* The header files that the VMS compilers include automatically,
2520   `__DECC_INCLUDE_PROLOGUE.H` and `__DECC_INCLUDE_EPILOGUE.H`, use pragmas
2521   that the C++ compiler doesn't understand.  This is a shortcoming in the
2522   compiler, but can be worked around with `__cplusplus` guards.
2523
2524   C++ applications that use OpenSSL libraries must be compiled using the
2525   qualifier `/NAMES=(AS_IS,SHORTENED)` to be able to use all the OpenSSL
2526   functions.  Otherwise, only functions with symbols of less than 31
2527   characters can be used, as the linker will not be able to successfully
2528   resolve symbols with longer names.
2529
2530   *Richard Levitte*
2531
2532 * Added a new method to gather entropy on VMS, based on SYS$GET_ENTROPY.
2533   The presence of this system service is determined at run-time.
2534
2535   *Richard Levitte*
2536
2537 * Print all values for a PKCS#12 attribute with 'openssl pkcs12', not just
2538   the first value.
2539
2540   *Jon Spillett*
2541
2542### Changes between 1.1.1c and 1.1.1d [10 Sep 2019]
2543
2544 * Fixed a fork protection issue. OpenSSL 1.1.1 introduced a rewritten random
2545   number generator (RNG). This was intended to include protection in the
2546   event of a fork() system call in order to ensure that the parent and child
2547   processes did not share the same RNG state. However this protection was not
2548   being used in the default case.
2549
2550   A partial mitigation for this issue is that the output from a high
2551   precision timer is mixed into the RNG state so the likelihood of a parent
2552   and child process sharing state is significantly reduced.
2553
2554   If an application already calls OPENSSL_init_crypto() explicitly using
2555   OPENSSL_INIT_ATFORK then this problem does not occur at all.
2556   ([CVE-2019-1549])
2557
2558   *Matthias St. Pierre*
2559
2560 * For built-in EC curves, ensure an EC_GROUP built from the curve name is
2561   used even when parsing explicit parameters, when loading a encoded key
2562   or calling `EC_GROUP_new_from_ecpkparameters()`/
2563   `EC_GROUP_new_from_ecparameters()`.
2564   This prevents bypass of security hardening and performance gains,
2565   especially for curves with specialized EC_METHODs.
2566   By default, if a key encoded with explicit parameters is loaded and later
2567   encoded, the output is still encoded with explicit parameters, even if
2568   internally a "named" EC_GROUP is used for computation.
2569
2570   *Nicola Tuveri*
2571
2572 * Compute ECC cofactors if not provided during EC_GROUP construction. Before
2573   this change, EC_GROUP_set_generator would accept order and/or cofactor as
2574   NULL. After this change, only the cofactor parameter can be NULL. It also
2575   does some minimal sanity checks on the passed order.
2576   ([CVE-2019-1547])
2577
2578   *Billy Bob Brumley*
2579
2580 * Fixed a padding oracle in PKCS7_dataDecode and CMS_decrypt_set1_pkey.
2581   An attack is simple, if the first CMS_recipientInfo is valid but the
2582   second CMS_recipientInfo is chosen ciphertext. If the second
2583   recipientInfo decodes to PKCS #1 v1.5 form plaintext, the correct
2584   encryption key will be replaced by garbage, and the message cannot be
2585   decoded, but if the RSA decryption fails, the correct encryption key is
2586   used and the recipient will not notice the attack.
2587   As a work around for this potential attack the length of the decrypted
2588   key must be equal to the cipher default key length, in case the
2589   certifiate is not given and all recipientInfo are tried out.
2590   The old behaviour can be re-enabled in the CMS code by setting the
2591   CMS_DEBUG_DECRYPT flag.
2592   ([CVE-2019-1563])
2593
2594   *Bernd Edlinger*
2595
2596 * Early start up entropy quality from the DEVRANDOM seed source has been
2597   improved for older Linux systems.  The RAND subsystem will wait for
2598   /dev/random to be producing output before seeding from /dev/urandom.
2599   The seeded state is stored for future library initialisations using
2600   a system global shared memory segment.  The shared memory identifier
2601   can be configured by defining OPENSSL_RAND_SEED_DEVRANDOM_SHM_ID to
2602   the desired value.  The default identifier is 114.
2603
2604   *Paul Dale*
2605
2606 * Correct the extended master secret constant on EBCDIC systems. Without this
2607   fix TLS connections between an EBCDIC system and a non-EBCDIC system that
2608   negotiate EMS will fail. Unfortunately this also means that TLS connections
2609   between EBCDIC systems with this fix, and EBCDIC systems without this
2610   fix will fail if they negotiate EMS.
2611
2612   *Matt Caswell*
2613
2614 * Use Windows installation paths in the mingw builds
2615
2616   Mingw isn't a POSIX environment per se, which means that Windows
2617   paths should be used for installation.
2618   ([CVE-2019-1552])
2619
2620   *Richard Levitte*
2621
2622 * Changed DH_check to accept parameters with order q and 2q subgroups.
2623   With order 2q subgroups the bit 0 of the private key is not secret
2624   but DH_generate_key works around that by clearing bit 0 of the
2625   private key for those. This avoids leaking bit 0 of the private key.
2626
2627   *Bernd Edlinger*
2628
2629 * Significantly reduce secure memory usage by the randomness pools.
2630
2631   *Paul Dale*
2632
2633 * Revert the DEVRANDOM_WAIT feature for Linux systems
2634
2635   The DEVRANDOM_WAIT feature added a select() call to wait for the
2636   /dev/random device to become readable before reading from the
2637   /dev/urandom device.
2638
2639   It turned out that this change had negative side effects on
2640   performance which were not acceptable. After some discussion it
2641   was decided to revert this feature and leave it up to the OS
2642   resp. the platform maintainer to ensure a proper initialization
2643   during early boot time.
2644
2645   *Matthias St. Pierre*
2646
2647### Changes between 1.1.1b and 1.1.1c [28 May 2019]
2648
2649 * Add build tests for C++.  These are generated files that only do one
2650   thing, to include one public OpenSSL head file each.  This tests that
2651   the public header files can be usefully included in a C++ application.
2652
2653   This test isn't enabled by default.  It can be enabled with the option
2654   'enable-buildtest-c++'.
2655
2656   *Richard Levitte*
2657
2658 * Enable SHA3 pre-hashing for ECDSA and DSA.
2659
2660   *Patrick Steuer*
2661
2662 * Change the default RSA, DSA and DH size to 2048 bit instead of 1024.
2663   This changes the size when using the `genpkey` command when no size is given.
2664   It fixes an omission in earlier changes that changed all RSA, DSA and DH
2665   generation commands to use 2048 bits by default.
2666
2667   *Kurt Roeckx*
2668
2669 * Reorganize the manual pages to consistently have RETURN VALUES,
2670   EXAMPLES, SEE ALSO and HISTORY come in that order, and adjust
2671   util/fix-doc-nits accordingly.
2672
2673   *Paul Yang, Joshua Lock*
2674
2675 * Add the missing accessor EVP_PKEY_get0_engine()
2676
2677   *Matt Caswell*
2678
2679 * Have commands like `s_client` and `s_server` output the signature scheme
2680   along with other cipher suite parameters when debugging.
2681
2682   *Lorinczy Zsigmond*
2683
2684 * Make OPENSSL_config() error agnostic again.
2685
2686   *Richard Levitte*
2687
2688 * Do the error handling in RSA decryption constant time.
2689
2690   *Bernd Edlinger*
2691
2692 * Prevent over long nonces in ChaCha20-Poly1305.
2693
2694   ChaCha20-Poly1305 is an AEAD cipher, and requires a unique nonce input
2695   for every encryption operation. RFC 7539 specifies that the nonce value
2696   (IV) should be 96 bits (12 bytes). OpenSSL allows a variable nonce length
2697   and front pads the nonce with 0 bytes if it is less than 12
2698   bytes. However it also incorrectly allows a nonce to be set of up to 16
2699   bytes. In this case only the last 12 bytes are significant and any
2700   additional leading bytes are ignored.
2701
2702   It is a requirement of using this cipher that nonce values are
2703   unique. Messages encrypted using a reused nonce value are susceptible to
2704   serious confidentiality and integrity attacks. If an application changes
2705   the default nonce length to be longer than 12 bytes and then makes a
2706   change to the leading bytes of the nonce expecting the new value to be a
2707   new unique nonce then such an application could inadvertently encrypt
2708   messages with a reused nonce.
2709
2710   Additionally the ignored bytes in a long nonce are not covered by the
2711   integrity guarantee of this cipher. Any application that relies on the
2712   integrity of these ignored leading bytes of a long nonce may be further
2713   affected. Any OpenSSL internal use of this cipher, including in SSL/TLS,
2714   is safe because no such use sets such a long nonce value. However user
2715   applications that use this cipher directly and set a non-default nonce
2716   length to be longer than 12 bytes may be vulnerable.
2717
2718   This issue was reported to OpenSSL on 16th of March 2019 by Joran Dirk
2719   Greef of Ronomon.
2720   ([CVE-2019-1543])
2721
2722   *Matt Caswell*
2723
2724 * Add DEVRANDOM_WAIT feature for Linux systems
2725
2726   On older Linux systems where the getrandom() system call is not available,
2727   OpenSSL normally uses the /dev/urandom device for seeding its CSPRNG.
2728   Contrary to getrandom(), the /dev/urandom device will not block during
2729   early boot when the kernel CSPRNG has not been seeded yet.
2730
2731   To mitigate this known weakness, use select() to wait for /dev/random to
2732   become readable before reading from /dev/urandom.
2733
2734 * Ensure that SM2 only uses SM3 as digest algorithm
2735
2736   *Paul Yang*
2737
2738### Changes between 1.1.1a and 1.1.1b [26 Feb 2019]
2739
2740 * Change the info callback signals for the start and end of a post-handshake
2741   message exchange in TLSv1.3. In 1.1.1/1.1.1a we used SSL_CB_HANDSHAKE_START
2742   and SSL_CB_HANDSHAKE_DONE. Experience has shown that many applications get
2743   confused by this and assume that a TLSv1.2 renegotiation has started. This
2744   can break KeyUpdate handling. Instead we no longer signal the start and end
2745   of a post handshake message exchange (although the messages themselves are
2746   still signalled). This could break some applications that were expecting
2747   the old signals. However without this KeyUpdate is not usable for many
2748   applications.
2749
2750   *Matt Caswell*
2751
2752### Changes between 1.1.1 and 1.1.1a [20 Nov 2018]
2753
2754 * Timing vulnerability in DSA signature generation
2755
2756   The OpenSSL DSA signature algorithm has been shown to be vulnerable to a
2757   timing side channel attack. An attacker could use variations in the signing
2758   algorithm to recover the private key.
2759
2760   This issue was reported to OpenSSL on 16th October 2018 by Samuel Weiser.
2761   ([CVE-2018-0734])
2762
2763   *Paul Dale*
2764
2765 * Timing vulnerability in ECDSA signature generation
2766
2767   The OpenSSL ECDSA signature algorithm has been shown to be vulnerable to a
2768   timing side channel attack. An attacker could use variations in the signing
2769   algorithm to recover the private key.
2770
2771   This issue was reported to OpenSSL on 25th October 2018 by Samuel Weiser.
2772   ([CVE-2018-0735])
2773
2774   *Paul Dale*
2775
2776 * Fixed the issue that RAND_add()/RAND_seed() silently discards random input
2777   if its length exceeds 4096 bytes. The limit has been raised to a buffer size
2778   of two gigabytes and the error handling improved.
2779
2780   This issue was reported to OpenSSL by Dr. Falko Strenzke. It has been
2781   categorized as a normal bug, not a security issue, because the DRBG reseeds
2782   automatically and is fully functional even without additional randomness
2783   provided by the application.
2784
2785### Changes between 1.1.0i and 1.1.1 [11 Sep 2018]
2786
2787 * Add a new ClientHello callback. Provides a callback interface that gives
2788   the application the ability to adjust the nascent SSL object at the
2789   earliest stage of ClientHello processing, immediately after extensions have
2790   been collected but before they have been processed. In particular, this
2791   callback can adjust the supported TLS versions in response to the contents
2792   of the ClientHello
2793
2794   *Benjamin Kaduk*
2795
2796 * Add SM2 base algorithm support.
2797
2798   *Jack Lloyd*
2799
2800 * s390x assembly pack: add (improved) hardware-support for the following
2801   cryptographic primitives: sha3, shake, aes-gcm, aes-ccm, aes-ctr, aes-ofb,
2802   aes-cfb/cfb8, aes-ecb.
2803
2804   *Patrick Steuer*
2805
2806 * Make EVP_PKEY_asn1_new() a bit stricter about its input.  A NULL pem_str
2807   parameter is no longer accepted, as it leads to a corrupt table.  NULL
2808   pem_str is reserved for alias entries only.
2809
2810   *Richard Levitte*
2811
2812 * Use the new ec_scalar_mul_ladder scaffold to implement a specialized ladder
2813   step for prime curves. The new implementation is based on formulae from
2814   differential addition-and-doubling in homogeneous projective coordinates
2815   from Izu-Takagi "A fast parallel elliptic curve multiplication resistant
2816   against side channel attacks" and Brier-Joye "Weierstrass Elliptic Curves
2817   and Side-Channel Attacks" Eq. (8) for y-coordinate recovery, modified
2818   to work in projective coordinates.
2819
2820   *Billy Bob Brumley, Nicola Tuveri*
2821
2822 * Change generating and checking of primes so that the error rate of not
2823   being prime depends on the intended use based on the size of the input.
2824   For larger primes this will result in more rounds of Miller-Rabin.
2825   The maximal error rate for primes with more than 1080 bits is lowered
2826   to 2^-128.
2827
2828   *Kurt Roeckx, Annie Yousar*
2829
2830 * Increase the number of Miller-Rabin rounds for DSA key generating to 64.
2831
2832   *Kurt Roeckx*
2833
2834 * The 'tsget' script is renamed to 'tsget.pl', to avoid confusion when
2835   moving between systems, and to avoid confusion when a Windows build is
2836   done with mingw vs with MSVC.  For POSIX installs, there's still a
2837   symlink or copy named 'tsget' to avoid that confusion as well.
2838
2839   *Richard Levitte*
2840
2841 * Revert blinding in ECDSA sign and instead make problematic addition
2842   length-invariant. Switch even to fixed-length Montgomery multiplication.
2843
2844   *Andy Polyakov*
2845
2846 * Use the new ec_scalar_mul_ladder scaffold to implement a specialized ladder
2847   step for binary curves. The new implementation is based on formulae from
2848   differential addition-and-doubling in mixed Lopez-Dahab projective
2849   coordinates, modified to independently blind the operands.
2850
2851   *Billy Bob Brumley, Sohaib ul Hassan, Nicola Tuveri*
2852
2853 * Add a scaffold to optionally enhance the Montgomery ladder implementation
2854   for `ec_scalar_mul_ladder` (formerly `ec_mul_consttime`) allowing
2855   EC_METHODs to implement their own specialized "ladder step", to take
2856   advantage of more favorable coordinate systems or more efficient
2857   differential addition-and-doubling algorithms.
2858
2859   *Billy Bob Brumley, Sohaib ul Hassan, Nicola Tuveri*
2860
2861 * Modified the random device based seed sources to keep the relevant
2862   file descriptors open rather than reopening them on each access.
2863   This allows such sources to operate in a chroot() jail without
2864   the associated device nodes being available. This behaviour can be
2865   controlled using RAND_keep_random_devices_open().
2866
2867   *Paul Dale*
2868
2869 * Numerous side-channel attack mitigations have been applied. This may have
2870   performance impacts for some algorithms for the benefit of improved
2871   security. Specific changes are noted in this change log by their respective
2872   authors.
2873
2874   *Matt Caswell*
2875
2876 * AIX shared library support overhaul. Switch to AIX "natural" way of
2877   handling shared libraries, which means collecting shared objects of
2878   different versions and bitnesses in one common archive. This allows to
2879   mitigate conflict between 1.0 and 1.1 side-by-side installations. It
2880   doesn't affect the way 3rd party applications are linked, only how
2881   multi-version installation is managed.
2882
2883   *Andy Polyakov*
2884
2885 * Make ec_group_do_inverse_ord() more robust and available to other
2886   EC cryptosystems, so that irrespective of BN_FLG_CONSTTIME, SCA
2887   mitigations are applied to the fallback BN_mod_inverse().
2888   When using this function rather than BN_mod_inverse() directly, new
2889   EC cryptosystem implementations are then safer-by-default.
2890
2891   *Billy Bob Brumley*
2892
2893 * Add coordinate blinding for EC_POINT and implement projective
2894   coordinate blinding for generic prime curves as a countermeasure to
2895   chosen point SCA attacks.
2896
2897   *Sohaib ul Hassan, Nicola Tuveri, Billy Bob Brumley*
2898
2899 * Add blinding to ECDSA and DSA signatures to protect against side channel
2900   attacks discovered by Keegan Ryan (NCC Group).
2901
2902   *Matt Caswell*
2903
2904 * Enforce checking in the `pkeyutl` command to ensure that the input
2905   length does not exceed the maximum supported digest length when performing
2906   a sign, verify or verifyrecover operation.
2907
2908   *Matt Caswell*
2909
2910 * SSL_MODE_AUTO_RETRY is enabled by default. Applications that use blocking
2911   I/O in combination with something like select() or poll() will hang. This
2912   can be turned off again using SSL_CTX_clear_mode().
2913   Many applications do not properly handle non-application data records, and
2914   TLS 1.3 sends more of such records. Setting SSL_MODE_AUTO_RETRY works
2915   around the problems in those applications, but can also break some.
2916   It's recommended to read the manpages about SSL_read(), SSL_write(),
2917   SSL_get_error(), SSL_shutdown(), SSL_CTX_set_mode() and
2918   SSL_CTX_set_read_ahead() again.
2919
2920   *Kurt Roeckx*
2921
2922 * When unlocking a pass phrase protected PEM file or PKCS#8 container, we
2923   now allow empty (zero character) pass phrases.
2924
2925   *Richard Levitte*
2926
2927 * Apply blinding to binary field modular inversion and remove patent
2928   pending (OPENSSL_SUN_GF2M_DIV) BN_GF2m_mod_div implementation.
2929
2930   *Billy Bob Brumley*
2931
2932 * Deprecate ec2_mult.c and unify scalar multiplication code paths for
2933   binary and prime elliptic curves.
2934
2935   *Billy Bob Brumley*
2936
2937 * Remove ECDSA nonce padding: EC_POINT_mul is now responsible for
2938   constant time fixed point multiplication.
2939
2940   *Billy Bob Brumley*
2941
2942 * Revise elliptic curve scalar multiplication with timing attack
2943   defenses: ec_wNAF_mul redirects to a constant time implementation
2944   when computing fixed point and variable point multiplication (which
2945   in OpenSSL are mostly used with secret scalars in keygen, sign,
2946   ECDH derive operations).
2947   *Billy Bob Brumley, Nicola Tuveri, Cesar Pereida García,
2948    Sohaib ul Hassan*
2949
2950 * Updated CONTRIBUTING
2951
2952   *Rich Salz*
2953
2954 * Updated DRBG / RAND to request nonce and additional low entropy
2955   randomness from the system.
2956
2957   *Matthias St. Pierre*
2958
2959 * Updated 'openssl rehash' to use OpenSSL consistent default.
2960
2961   *Richard Levitte*
2962
2963 * Moved the load of the ssl_conf module to libcrypto, which helps
2964   loading engines that libssl uses before libssl is initialised.
2965
2966   *Matt Caswell*
2967
2968 * Added EVP_PKEY_sign() and EVP_PKEY_verify() for EdDSA
2969
2970   *Matt Caswell*
2971
2972 * Fixed X509_NAME_ENTRY_set to get multi-valued RDNs right in all cases.
2973
2974   *Ingo Schwarze, Rich Salz*
2975
2976 * Added output of accepting IP address and port for 'openssl s_server'
2977
2978   *Richard Levitte*
2979
2980 * Added a new API for TLSv1.3 ciphersuites:
2981      SSL_CTX_set_ciphersuites()
2982      SSL_set_ciphersuites()
2983
2984   *Matt Caswell*
2985
2986 * Memory allocation failures consistently add an error to the error
2987   stack.
2988
2989   *Rich Salz*
2990
2991 * Don't use OPENSSL_ENGINES and OPENSSL_CONF environment values
2992   in libcrypto when run as setuid/setgid.
2993
2994   *Bernd Edlinger*
2995
2996 * Load any config file by default when libssl is used.
2997
2998   *Matt Caswell*
2999
3000 * Added new public header file <openssl/rand_drbg.h> and documentation
3001   for the RAND_DRBG API. See manual page RAND_DRBG(7) for an overview.
3002
3003   *Matthias St. Pierre*
3004
3005 * QNX support removed (cannot find contributors to get their approval
3006   for the license change).
3007
3008   *Rich Salz*
3009
3010 * TLSv1.3 replay protection for early data has been implemented. See the
3011   SSL_read_early_data() man page for further details.
3012
3013   *Matt Caswell*
3014
3015 * Separated TLSv1.3 ciphersuite configuration out from TLSv1.2 ciphersuite
3016   configuration. TLSv1.3 ciphersuites are not compatible with TLSv1.2 and
3017   below. Similarly TLSv1.2 ciphersuites are not compatible with TLSv1.3.
3018   In order to avoid issues where legacy TLSv1.2 ciphersuite configuration
3019   would otherwise inadvertently disable all TLSv1.3 ciphersuites the
3020   configuration has been separated out. See the ciphers man page or the
3021   SSL_CTX_set_ciphersuites() man page for more information.
3022
3023   *Matt Caswell*
3024
3025 * On POSIX (BSD, Linux, ...) systems the ocsp(1) command running
3026   in responder mode now supports the new "-multi" option, which
3027   spawns the specified number of child processes to handle OCSP
3028   requests.  The "-timeout" option now also limits the OCSP
3029   responder's patience to wait to receive the full client request
3030   on a newly accepted connection. Child processes are respawned
3031   as needed, and the CA index file is automatically reloaded
3032   when changed.  This makes it possible to run the "ocsp" responder
3033   as a long-running service, making the OpenSSL CA somewhat more
3034   feature-complete.  In this mode, most diagnostic messages logged
3035   after entering the event loop are logged via syslog(3) rather than
3036   written to stderr.
3037
3038   *Viktor Dukhovni*
3039
3040 * Added support for X448 and Ed448. Heavily based on original work by
3041   Mike Hamburg.
3042
3043   *Matt Caswell*
3044
3045 * Extend OSSL_STORE with capabilities to search and to narrow the set of
3046   objects loaded.  This adds the functions OSSL_STORE_expect() and
3047   OSSL_STORE_find() as well as needed tools to construct searches and
3048   get the search data out of them.
3049
3050   *Richard Levitte*
3051
3052 * Support for TLSv1.3 added. Note that users upgrading from an earlier
3053   version of OpenSSL should review their configuration settings to ensure
3054   that they are still appropriate for TLSv1.3. For further information see:
3055   <https://wiki.openssl.org/index.php/TLS1.3>
3056
3057   *Matt Caswell*
3058
3059 * Grand redesign of the OpenSSL random generator
3060
3061   The default RAND method now utilizes an AES-CTR DRBG according to
3062   NIST standard SP 800-90Ar1. The new random generator is essentially
3063   a port of the default random generator from the OpenSSL FIPS 2.0
3064   object module. It is a hybrid deterministic random bit generator
3065   using an AES-CTR bit stream and which seeds and reseeds itself
3066   automatically using trusted system entropy sources.
3067
3068   Some of its new features are:
3069    - Support for multiple DRBG instances with seed chaining.
3070    - The default RAND method makes use of a DRBG.
3071    - There is a public and private DRBG instance.
3072    - The DRBG instances are fork-safe.
3073    - Keep all global DRBG instances on the secure heap if it is enabled.
3074    - The public and private DRBG instance are per thread for lock free
3075      operation
3076
3077   *Paul Dale, Benjamin Kaduk, Kurt Roeckx, Rich Salz, Matthias St. Pierre*
3078
3079 * Changed Configure so it only says what it does and doesn't dump
3080   so much data.  Instead, ./configdata.pm should be used as a script
3081   to display all sorts of configuration data.
3082
3083   *Richard Levitte*
3084
3085 * Added processing of "make variables" to Configure.
3086
3087   *Richard Levitte*
3088
3089 * Added SHA512/224 and SHA512/256 algorithm support.
3090
3091   *Paul Dale*
3092
3093 * The last traces of Netware support, first removed in 1.1.0, have
3094   now been removed.
3095
3096   *Rich Salz*
3097
3098 * Get rid of Makefile.shared, and in the process, make the processing
3099   of certain files (rc.obj, or the .def/.map/.opt files produced from
3100   the ordinal files) more visible and hopefully easier to trace and
3101   debug (or make silent).
3102
3103   *Richard Levitte*
3104
3105 * Make it possible to have environment variable assignments as
3106   arguments to config / Configure.
3107
3108   *Richard Levitte*
3109
3110 * Add multi-prime RSA (RFC 8017) support.
3111
3112   *Paul Yang*
3113
3114 * Add SM3 implemented according to GB/T 32905-2016
3115   *Jack Lloyd <jack.lloyd@ribose.com>,*
3116   *Ronald Tse <ronald.tse@ribose.com>,*
3117   *Erick Borsboom <erick.borsboom@ribose.com>*
3118
3119 * Add 'Maximum Fragment Length' TLS extension negotiation and support
3120   as documented in RFC6066.
3121   Based on a patch from Tomasz Moń
3122
3123   *Filipe Raimundo da Silva*
3124
3125 * Add SM4 implemented according to GB/T 32907-2016.
3126   *Jack Lloyd <jack.lloyd@ribose.com>,*
3127   *Ronald Tse <ronald.tse@ribose.com>,*
3128   *Erick Borsboom <erick.borsboom@ribose.com>*
3129
3130 * Reimplement -newreq-nodes and ERR_error_string_n; the
3131   original author does not agree with the license change.
3132
3133   *Rich Salz*
3134
3135 * Add ARIA AEAD TLS support.
3136
3137   *Jon Spillett*
3138
3139 * Some macro definitions to support VS6 have been removed.  Visual
3140   Studio 6 has not worked since 1.1.0
3141
3142   *Rich Salz*
3143
3144 * Add ERR_clear_last_mark(), to allow callers to clear the last mark
3145   without clearing the errors.
3146
3147   *Richard Levitte*
3148
3149 * Add "atfork" functions.  If building on a system that without
3150   pthreads, see doc/man3/OPENSSL_fork_prepare.pod for application
3151   requirements.  The RAND facility now uses/requires this.
3152
3153   *Rich Salz*
3154
3155 * Add SHA3.
3156
3157   *Andy Polyakov*
3158
3159 * The UI API becomes a permanent and integral part of libcrypto, i.e.
3160   not possible to disable entirely.  However, it's still possible to
3161   disable the console reading UI method, UI_OpenSSL() (use UI_null()
3162   as a fallback).
3163
3164   To disable, configure with 'no-ui-console'.  'no-ui' is still
3165   possible to use as an alias.  Check at compile time with the
3166   macro OPENSSL_NO_UI_CONSOLE.  The macro OPENSSL_NO_UI is still
3167   possible to check and is an alias for OPENSSL_NO_UI_CONSOLE.
3168
3169   *Richard Levitte*
3170
3171 * Add a STORE module, which implements a uniform and URI based reader of
3172   stores that can contain keys, certificates, CRLs and numerous other
3173   objects.  The main API is loosely based on a few stdio functions,
3174   and includes OSSL_STORE_open, OSSL_STORE_load, OSSL_STORE_eof,
3175   OSSL_STORE_error and OSSL_STORE_close.
3176   The implementation uses backends called "loaders" to implement arbitrary
3177   URI schemes.  There is one built in "loader" for the 'file' scheme.
3178
3179   *Richard Levitte*
3180
3181 * Add devcrypto engine.  This has been implemented against cryptodev-linux,
3182   then adjusted to work on FreeBSD 8.4 as well.
3183   Enable by configuring with 'enable-devcryptoeng'.  This is done by default
3184   on BSD implementations, as cryptodev.h is assumed to exist on all of them.
3185
3186   *Richard Levitte*
3187
3188 * Module names can prefixed with OSSL_ or OPENSSL_.  This affects
3189   util/mkerr.pl, which is adapted to allow those prefixes, leading to
3190   error code calls like this:
3191
3192           OSSL_FOOerr(OSSL_FOO_F_SOMETHING, OSSL_FOO_R_WHATEVER);
3193
3194   With this change, we claim the namespaces OSSL and OPENSSL in a manner
3195   that can be encoded in C.  For the foreseeable future, this will only
3196   affect new modules.
3197
3198   *Richard Levitte and Tim Hudson*
3199
3200 * Removed BSD cryptodev engine.
3201
3202   *Rich Salz*
3203
3204 * Add a build target 'build_all_generated', to build all generated files
3205   and only that.  This can be used to prepare everything that requires
3206   things like perl for a system that lacks perl and then move everything
3207   to that system and do the rest of the build there.
3208
3209   *Richard Levitte*
3210
3211 * In the UI interface, make it possible to duplicate the user data.  This
3212   can be used by engines that need to retain the data for a longer time
3213   than just the call where this user data is passed.
3214
3215   *Richard Levitte*
3216
3217 * Ignore the '-named_curve auto' value for compatibility of applications
3218   with OpenSSL 1.0.2.
3219
3220   *Tomáš Mráz <tmraz@fedoraproject.org>*
3221
3222 * Fragmented SSL/TLS alerts are no longer accepted. An alert message is 2
3223   bytes long. In theory it is permissible in SSLv3 - TLSv1.2 to fragment such
3224   alerts across multiple records (some of which could be empty). In practice
3225   it make no sense to send an empty alert record, or to fragment one. TLSv1.3
3226   prohibits this altogether and other libraries (BoringSSL, NSS) do not
3227   support this at all. Supporting it adds significant complexity to the
3228   record layer, and its removal is unlikely to cause interoperability
3229   issues.
3230
3231   *Matt Caswell*
3232
3233 * Add the ASN.1 types INT32, UINT32, INT64, UINT64 and variants prefixed
3234   with Z.  These are meant to replace LONG and ZLONG and to be size safe.
3235   The use of LONG and ZLONG is discouraged and scheduled for deprecation
3236   in OpenSSL 1.2.0.
3237
3238   *Richard Levitte*
3239
3240 * Add the 'z' and 'j' modifiers to BIO_printf() et al formatting string,
3241   'z' is to be used for [s]size_t, and 'j' - with [u]int64_t.
3242
3243   *Richard Levitte, Andy Polyakov*
3244
3245 * Add EC_KEY_get0_engine(), which does for EC_KEY what RSA_get0_engine()
3246   does for RSA, etc.
3247
3248   *Richard Levitte*
3249
3250 * Have 'config' recognise 64-bit mingw and choose 'mingw64' as the target
3251   platform rather than 'mingw'.
3252
3253   *Richard Levitte*
3254
3255 * The functions X509_STORE_add_cert and X509_STORE_add_crl return
3256   success if they are asked to add an object which already exists
3257   in the store. This change cascades to other functions which load
3258   certificates and CRLs.
3259
3260   *Paul Dale*
3261
3262 * x86_64 assembly pack: annotate code with DWARF CFI directives to
3263   facilitate stack unwinding even from assembly subroutines.
3264
3265   *Andy Polyakov*
3266
3267 * Remove VAX C specific definitions of OPENSSL_EXPORT, OPENSSL_EXTERN.
3268   Also remove OPENSSL_GLOBAL entirely, as it became a no-op.
3269
3270   *Richard Levitte*
3271
3272 * Remove the VMS-specific reimplementation of gmtime from crypto/o_times.c.
3273   VMS C's RTL has a fully up to date gmtime() and gmtime_r() since V7.1,
3274   which is the minimum version we support.
3275
3276   *Richard Levitte*
3277
3278 * Certificate time validation (X509_cmp_time) enforces stricter
3279   compliance with RFC 5280. Fractional seconds and timezone offsets
3280   are no longer allowed.
3281
3282   *Emilia Käsper*
3283
3284 * Add support for ARIA
3285
3286   *Paul Dale*
3287
3288 * s_client will now send the Server Name Indication (SNI) extension by
3289   default unless the new "-noservername" option is used. The server name is
3290   based on the host provided to the "-connect" option unless overridden by
3291   using "-servername".
3292
3293   *Matt Caswell*
3294
3295 * Add support for SipHash
3296
3297   *Todd Short*
3298
3299 * OpenSSL now fails if it receives an unrecognised record type in TLS1.0
3300   or TLS1.1. Previously this only happened in SSLv3 and TLS1.2. This is to
3301   prevent issues where no progress is being made and the peer continually
3302   sends unrecognised record types, using up resources processing them.
3303
3304   *Matt Caswell*
3305
3306 * 'openssl passwd' can now produce SHA256 and SHA512 based output,
3307   using the algorithm defined in
3308   <https://www.akkadia.org/drepper/SHA-crypt.txt>
3309
3310   *Richard Levitte*
3311
3312 * Heartbeat support has been removed; the ABI is changed for now.
3313
3314   *Richard Levitte, Rich Salz*
3315
3316 * Support for SSL_OP_NO_ENCRYPT_THEN_MAC in SSL_CONF_cmd.
3317
3318   *Emilia Käsper*
3319
3320 * The RSA "null" method, which was partially supported to avoid patent
3321   issues, has been replaced to always returns NULL.
3322
3323   *Rich Salz*
3324
3325OpenSSL 1.1.0
3326-------------
3327
3328### Changes between 1.1.0k and 1.1.0l [10 Sep 2019]
3329
3330 * For built-in EC curves, ensure an EC_GROUP built from the curve name is
3331   used even when parsing explicit parameters, when loading a encoded key
3332   or calling `EC_GROUP_new_from_ecpkparameters()`/
3333   `EC_GROUP_new_from_ecparameters()`.
3334   This prevents bypass of security hardening and performance gains,
3335   especially for curves with specialized EC_METHODs.
3336   By default, if a key encoded with explicit parameters is loaded and later
3337   encoded, the output is still encoded with explicit parameters, even if
3338   internally a "named" EC_GROUP is used for computation.
3339
3340   *Nicola Tuveri*
3341
3342 * Compute ECC cofactors if not provided during EC_GROUP construction. Before
3343   this change, EC_GROUP_set_generator would accept order and/or cofactor as
3344   NULL. After this change, only the cofactor parameter can be NULL. It also
3345   does some minimal sanity checks on the passed order.
3346   ([CVE-2019-1547])
3347
3348   *Billy Bob Brumley*
3349
3350 * Fixed a padding oracle in PKCS7_dataDecode and CMS_decrypt_set1_pkey.
3351   An attack is simple, if the first CMS_recipientInfo is valid but the
3352   second CMS_recipientInfo is chosen ciphertext. If the second
3353   recipientInfo decodes to PKCS #1 v1.5 form plaintext, the correct
3354   encryption key will be replaced by garbage, and the message cannot be
3355   decoded, but if the RSA decryption fails, the correct encryption key is
3356   used and the recipient will not notice the attack.
3357   As a work around for this potential attack the length of the decrypted
3358   key must be equal to the cipher default key length, in case the
3359   certifiate is not given and all recipientInfo are tried out.
3360   The old behaviour can be re-enabled in the CMS code by setting the
3361   CMS_DEBUG_DECRYPT flag.
3362   ([CVE-2019-1563])
3363
3364   *Bernd Edlinger*
3365
3366 * Use Windows installation paths in the mingw builds
3367
3368   Mingw isn't a POSIX environment per se, which means that Windows
3369   paths should be used for installation.
3370   ([CVE-2019-1552])
3371
3372   *Richard Levitte*
3373
3374### Changes between 1.1.0j and 1.1.0k [28 May 2019]
3375
3376 * Change the default RSA, DSA and DH size to 2048 bit instead of 1024.
3377   This changes the size when using the `genpkey` command when no size is given.
3378   It fixes an omission in earlier changes that changed all RSA, DSA and DH
3379   generation commands to use 2048 bits by default.
3380
3381   *Kurt Roeckx*
3382
3383 * Prevent over long nonces in ChaCha20-Poly1305.
3384
3385   ChaCha20-Poly1305 is an AEAD cipher, and requires a unique nonce input
3386   for every encryption operation. RFC 7539 specifies that the nonce value
3387   (IV) should be 96 bits (12 bytes). OpenSSL allows a variable nonce length
3388   and front pads the nonce with 0 bytes if it is less than 12
3389   bytes. However it also incorrectly allows a nonce to be set of up to 16
3390   bytes. In this case only the last 12 bytes are significant and any
3391   additional leading bytes are ignored.
3392
3393   It is a requirement of using this cipher that nonce values are
3394   unique. Messages encrypted using a reused nonce value are susceptible to
3395   serious confidentiality and integrity attacks. If an application changes
3396   the default nonce length to be longer than 12 bytes and then makes a
3397   change to the leading bytes of the nonce expecting the new value to be a
3398   new unique nonce then such an application could inadvertently encrypt
3399   messages with a reused nonce.
3400
3401   Additionally the ignored bytes in a long nonce are not covered by the
3402   integrity guarantee of this cipher. Any application that relies on the
3403   integrity of these ignored leading bytes of a long nonce may be further
3404   affected. Any OpenSSL internal use of this cipher, including in SSL/TLS,
3405   is safe because no such use sets such a long nonce value. However user
3406   applications that use this cipher directly and set a non-default nonce
3407   length to be longer than 12 bytes may be vulnerable.
3408
3409   This issue was reported to OpenSSL on 16th of March 2019 by Joran Dirk
3410   Greef of Ronomon.
3411   ([CVE-2019-1543])
3412
3413   *Matt Caswell*
3414
3415 * Added SCA hardening for modular field inversion in EC_GROUP through
3416   a new dedicated field_inv() pointer in EC_METHOD.
3417   This also addresses a leakage affecting conversions from projective
3418   to affine coordinates.
3419
3420   *Billy Bob Brumley, Nicola Tuveri*
3421
3422 * Fix a use after free bug in d2i_X509_PUBKEY when overwriting a
3423   re-used X509_PUBKEY object if the second PUBKEY is malformed.
3424
3425   *Bernd Edlinger*
3426
3427 * Move strictness check from EVP_PKEY_asn1_new() to EVP_PKEY_asn1_add0().
3428
3429   *Richard Levitte*
3430
3431 * Remove the 'dist' target and add a tarball building script.  The
3432   'dist' target has fallen out of use, and it shouldn't be
3433   necessary to configure just to create a source distribution.
3434
3435   *Richard Levitte*
3436
3437### Changes between 1.1.0i and 1.1.0j [20 Nov 2018]
3438
3439 * Timing vulnerability in DSA signature generation
3440
3441   The OpenSSL DSA signature algorithm has been shown to be vulnerable to a
3442   timing side channel attack. An attacker could use variations in the signing
3443   algorithm to recover the private key.
3444
3445   This issue was reported to OpenSSL on 16th October 2018 by Samuel Weiser.
3446   ([CVE-2018-0734])
3447
3448   *Paul Dale*
3449
3450 * Timing vulnerability in ECDSA signature generation
3451
3452   The OpenSSL ECDSA signature algorithm has been shown to be vulnerable to a
3453   timing side channel attack. An attacker could use variations in the signing
3454   algorithm to recover the private key.
3455
3456   This issue was reported to OpenSSL on 25th October 2018 by Samuel Weiser.
3457   ([CVE-2018-0735])
3458
3459   *Paul Dale*
3460
3461 * Add coordinate blinding for EC_POINT and implement projective
3462   coordinate blinding for generic prime curves as a countermeasure to
3463   chosen point SCA attacks.
3464
3465   *Sohaib ul Hassan, Nicola Tuveri, Billy Bob Brumley*
3466
3467### Changes between 1.1.0h and 1.1.0i [14 Aug 2018]
3468
3469 * Client DoS due to large DH parameter
3470
3471   During key agreement in a TLS handshake using a DH(E) based ciphersuite a
3472   malicious server can send a very large prime value to the client. This will
3473   cause the client to spend an unreasonably long period of time generating a
3474   key for this prime resulting in a hang until the client has finished. This
3475   could be exploited in a Denial Of Service attack.
3476
3477   This issue was reported to OpenSSL on 5th June 2018 by Guido Vranken
3478   ([CVE-2018-0732])
3479
3480   *Guido Vranken*
3481
3482 * Cache timing vulnerability in RSA Key Generation
3483
3484   The OpenSSL RSA Key generation algorithm has been shown to be vulnerable to
3485   a cache timing side channel attack. An attacker with sufficient access to
3486   mount cache timing attacks during the RSA key generation process could
3487   recover the private key.
3488
3489   This issue was reported to OpenSSL on 4th April 2018 by Alejandro Cabrera
3490   Aldaya, Billy Brumley, Cesar Pereida Garcia and Luis Manuel Alvarez Tapia.
3491   ([CVE-2018-0737])
3492
3493   *Billy Brumley*
3494
3495 * Make EVP_PKEY_asn1_new() a bit stricter about its input.  A NULL pem_str
3496   parameter is no longer accepted, as it leads to a corrupt table.  NULL
3497   pem_str is reserved for alias entries only.
3498
3499   *Richard Levitte*
3500
3501 * Revert blinding in ECDSA sign and instead make problematic addition
3502   length-invariant. Switch even to fixed-length Montgomery multiplication.
3503
3504   *Andy Polyakov*
3505
3506 * Change generating and checking of primes so that the error rate of not
3507   being prime depends on the intended use based on the size of the input.
3508   For larger primes this will result in more rounds of Miller-Rabin.
3509   The maximal error rate for primes with more than 1080 bits is lowered
3510   to 2^-128.
3511
3512   *Kurt Roeckx, Annie Yousar*
3513
3514 * Increase the number of Miller-Rabin rounds for DSA key generating to 64.
3515
3516   *Kurt Roeckx*
3517
3518 * Add blinding to ECDSA and DSA signatures to protect against side channel
3519   attacks discovered by Keegan Ryan (NCC Group).
3520
3521   *Matt Caswell*
3522
3523 * When unlocking a pass phrase protected PEM file or PKCS#8 container, we
3524   now allow empty (zero character) pass phrases.
3525
3526   *Richard Levitte*
3527
3528 * Certificate time validation (X509_cmp_time) enforces stricter
3529   compliance with RFC 5280. Fractional seconds and timezone offsets
3530   are no longer allowed.
3531
3532   *Emilia Käsper*
3533
3534 * Fixed a text canonicalisation bug in CMS
3535
3536   Where a CMS detached signature is used with text content the text goes
3537   through a canonicalisation process first prior to signing or verifying a
3538   signature. This process strips trailing space at the end of lines, converts
3539   line terminators to CRLF and removes additional trailing line terminators
3540   at the end of a file. A bug in the canonicalisation process meant that
3541   some characters, such as form-feed, were incorrectly treated as whitespace
3542   and removed. This is contrary to the specification (RFC5485). This fix
3543   could mean that detached text data signed with an earlier version of
3544   OpenSSL 1.1.0 may fail to verify using the fixed version, or text data
3545   signed with a fixed OpenSSL may fail to verify with an earlier version of
3546   OpenSSL 1.1.0. A workaround is to only verify the canonicalised text data
3547   and use the "-binary" flag (for the "cms" command line application) or set
3548   the SMIME_BINARY/PKCS7_BINARY/CMS_BINARY flags (if using CMS_verify()).
3549
3550   *Matt Caswell*
3551
3552### Changes between 1.1.0g and 1.1.0h [27 Mar 2018]
3553
3554 * Constructed ASN.1 types with a recursive definition could exceed the stack
3555
3556   Constructed ASN.1 types with a recursive definition (such as can be found
3557   in PKCS7) could eventually exceed the stack given malicious input with
3558   excessive recursion. This could result in a Denial Of Service attack. There
3559   are no such structures used within SSL/TLS that come from untrusted sources
3560   so this is considered safe.
3561
3562   This issue was reported to OpenSSL on 4th January 2018 by the OSS-fuzz
3563   project.
3564   ([CVE-2018-0739])
3565
3566   *Matt Caswell*
3567
3568 * Incorrect CRYPTO_memcmp on HP-UX PA-RISC
3569
3570   Because of an implementation bug the PA-RISC CRYPTO_memcmp function is
3571   effectively reduced to only comparing the least significant bit of each
3572   byte. This allows an attacker to forge messages that would be considered as
3573   authenticated in an amount of tries lower than that guaranteed by the
3574   security claims of the scheme. The module can only be compiled by the
3575   HP-UX assembler, so that only HP-UX PA-RISC targets are affected.
3576
3577   This issue was reported to OpenSSL on 2nd March 2018 by Peter Waltenberg
3578   (IBM).
3579   ([CVE-2018-0733])
3580
3581   *Andy Polyakov*
3582
3583 * Add a build target 'build_all_generated', to build all generated files
3584   and only that.  This can be used to prepare everything that requires
3585   things like perl for a system that lacks perl and then move everything
3586   to that system and do the rest of the build there.
3587
3588   *Richard Levitte*
3589
3590 * Backport SSL_OP_NO_RENGOTIATION
3591
3592   OpenSSL 1.0.2 and below had the ability to disable renegotiation using the
3593   (undocumented) SSL3_FLAGS_NO_RENEGOTIATE_CIPHERS flag. Due to the opacity
3594   changes this is no longer possible in 1.1.0. Therefore the new
3595   SSL_OP_NO_RENEGOTIATION option from 1.1.1-dev has been backported to
3596   1.1.0 to provide equivalent functionality.
3597
3598   Note that if an application built against 1.1.0h headers (or above) is run
3599   using an older version of 1.1.0 (prior to 1.1.0h) then the option will be
3600   accepted but nothing will happen, i.e. renegotiation will not be prevented.
3601
3602   *Matt Caswell*
3603
3604 * Removed the OS390-Unix config target.  It relied on a script that doesn't
3605   exist.
3606
3607   *Rich Salz*
3608
3609 * rsaz_1024_mul_avx2 overflow bug on x86_64
3610
3611   There is an overflow bug in the AVX2 Montgomery multiplication procedure
3612   used in exponentiation with 1024-bit moduli. No EC algorithms are affected.
3613   Analysis suggests that attacks against RSA and DSA as a result of this
3614   defect would be very difficult to perform and are not believed likely.
3615   Attacks against DH1024 are considered just feasible, because most of the
3616   work necessary to deduce information about a private key may be performed
3617   offline. The amount of resources required for such an attack would be
3618   significant. However, for an attack on TLS to be meaningful, the server
3619   would have to share the DH1024 private key among multiple clients, which is
3620   no longer an option since CVE-2016-0701.
3621
3622   This only affects processors that support the AVX2 but not ADX extensions
3623   like Intel Haswell (4th generation).
3624
3625   This issue was reported to OpenSSL by David Benjamin (Google). The issue
3626   was originally found via the OSS-Fuzz project.
3627   ([CVE-2017-3738])
3628
3629   *Andy Polyakov*
3630
3631### Changes between 1.1.0f and 1.1.0g [2 Nov 2017]
3632
3633 * bn_sqrx8x_internal carry bug on x86_64
3634
3635   There is a carry propagating bug in the x86_64 Montgomery squaring
3636   procedure. No EC algorithms are affected. Analysis suggests that attacks
3637   against RSA and DSA as a result of this defect would be very difficult to
3638   perform and are not believed likely. Attacks against DH are considered just
3639   feasible (although very difficult) because most of the work necessary to
3640   deduce information about a private key may be performed offline. The amount
3641   of resources required for such an attack would be very significant and
3642   likely only accessible to a limited number of attackers. An attacker would
3643   additionally need online access to an unpatched system using the target
3644   private key in a scenario with persistent DH parameters and a private
3645   key that is shared between multiple clients.
3646
3647   This only affects processors that support the BMI1, BMI2 and ADX extensions
3648   like Intel Broadwell (5th generation) and later or AMD Ryzen.
3649
3650   This issue was reported to OpenSSL by the OSS-Fuzz project.
3651   ([CVE-2017-3736])
3652
3653   *Andy Polyakov*
3654
3655 * Malformed X.509 IPAddressFamily could cause OOB read
3656
3657   If an X.509 certificate has a malformed IPAddressFamily extension,
3658   OpenSSL could do a one-byte buffer overread. The most likely result
3659   would be an erroneous display of the certificate in text format.
3660
3661   This issue was reported to OpenSSL by the OSS-Fuzz project.
3662   ([CVE-2017-3735])
3663
3664   *Rich Salz*
3665
3666### Changes between 1.1.0e and 1.1.0f [25 May 2017]
3667
3668 * Have 'config' recognise 64-bit mingw and choose 'mingw64' as the target
3669   platform rather than 'mingw'.
3670
3671   *Richard Levitte*
3672
3673 * Remove the VMS-specific reimplementation of gmtime from crypto/o_times.c.
3674   VMS C's RTL has a fully up to date gmtime() and gmtime_r() since V7.1,
3675   which is the minimum version we support.
3676
3677   *Richard Levitte*
3678
3679### Changes between 1.1.0d and 1.1.0e [16 Feb 2017]
3680
3681 * Encrypt-Then-Mac renegotiation crash
3682
3683   During a renegotiation handshake if the Encrypt-Then-Mac extension is
3684   negotiated where it was not in the original handshake (or vice-versa) then
3685   this can cause OpenSSL to crash (dependant on ciphersuite). Both clients
3686   and servers are affected.
3687
3688   This issue was reported to OpenSSL by Joe Orton (Red Hat).
3689   ([CVE-2017-3733])
3690
3691   *Matt Caswell*
3692
3693### Changes between 1.1.0c and 1.1.0d [26 Jan 2017]
3694
3695 * Truncated packet could crash via OOB read
3696
3697   If one side of an SSL/TLS path is running on a 32-bit host and a specific
3698   cipher is being used, then a truncated packet can cause that host to
3699   perform an out-of-bounds read, usually resulting in a crash.
3700
3701   This issue was reported to OpenSSL by Robert Święcki of Google.
3702   ([CVE-2017-3731])
3703
3704   *Andy Polyakov*
3705
3706 * Bad (EC)DHE parameters cause a client crash
3707
3708   If a malicious server supplies bad parameters for a DHE or ECDHE key
3709   exchange then this can result in the client attempting to dereference a
3710   NULL pointer leading to a client crash. This could be exploited in a Denial
3711   of Service attack.
3712
3713   This issue was reported to OpenSSL by Guido Vranken.
3714   ([CVE-2017-3730])
3715
3716   *Matt Caswell*
3717
3718 * BN_mod_exp may produce incorrect results on x86_64
3719
3720   There is a carry propagating bug in the x86_64 Montgomery squaring
3721   procedure. No EC algorithms are affected. Analysis suggests that attacks
3722   against RSA and DSA as a result of this defect would be very difficult to
3723   perform and are not believed likely. Attacks against DH are considered just
3724   feasible (although very difficult) because most of the work necessary to
3725   deduce information about a private key may be performed offline. The amount
3726   of resources required for such an attack would be very significant and
3727   likely only accessible to a limited number of attackers. An attacker would
3728   additionally need online access to an unpatched system using the target
3729   private key in a scenario with persistent DH parameters and a private
3730   key that is shared between multiple clients. For example this can occur by
3731   default in OpenSSL DHE based SSL/TLS ciphersuites. Note: This issue is very
3732   similar to CVE-2015-3193 but must be treated as a separate problem.
3733
3734   This issue was reported to OpenSSL by the OSS-Fuzz project.
3735   ([CVE-2017-3732])
3736
3737   *Andy Polyakov*
3738
3739### Changes between 1.1.0b and 1.1.0c [10 Nov 2016]
3740
3741 * ChaCha20/Poly1305 heap-buffer-overflow
3742
3743   TLS connections using `*-CHACHA20-POLY1305` ciphersuites are susceptible to
3744   a DoS attack by corrupting larger payloads. This can result in an OpenSSL
3745   crash. This issue is not considered to be exploitable beyond a DoS.
3746
3747   This issue was reported to OpenSSL by Robert Święcki (Google Security Team)
3748   ([CVE-2016-7054])
3749
3750   *Richard Levitte*
3751
3752 * CMS Null dereference
3753
3754   Applications parsing invalid CMS structures can crash with a NULL pointer
3755   dereference. This is caused by a bug in the handling of the ASN.1 CHOICE
3756   type in OpenSSL 1.1.0 which can result in a NULL value being passed to the
3757   structure callback if an attempt is made to free certain invalid encodings.
3758   Only CHOICE structures using a callback which do not handle NULL value are
3759   affected.
3760
3761   This issue was reported to OpenSSL by Tyler Nighswander of ForAllSecure.
3762   ([CVE-2016-7053])
3763
3764   *Stephen Henson*
3765
3766 * Montgomery multiplication may produce incorrect results
3767
3768   There is a carry propagating bug in the Broadwell-specific Montgomery
3769   multiplication procedure that handles input lengths divisible by, but
3770   longer than 256 bits. Analysis suggests that attacks against RSA, DSA
3771   and DH private keys are impossible. This is because the subroutine in
3772   question is not used in operations with the private key itself and an input
3773   of the attacker's direct choice. Otherwise the bug can manifest itself as
3774   transient authentication and key negotiation failures or reproducible
3775   erroneous outcome of public-key operations with specially crafted input.
3776   Among EC algorithms only Brainpool P-512 curves are affected and one
3777   presumably can attack ECDH key negotiation. Impact was not analyzed in
3778   detail, because pre-requisites for attack are considered unlikely. Namely
3779   multiple clients have to choose the curve in question and the server has to
3780   share the private key among them, neither of which is default behaviour.
3781   Even then only clients that chose the curve will be affected.
3782
3783   This issue was publicly reported as transient failures and was not
3784   initially recognized as a security issue. Thanks to Richard Morgan for
3785   providing reproducible case.
3786   ([CVE-2016-7055])
3787
3788   *Andy Polyakov*
3789
3790 * Removed automatic addition of RPATH in shared libraries and executables,
3791   as this was a remainder from OpenSSL 1.0.x and isn't needed any more.
3792
3793   *Richard Levitte*
3794
3795### Changes between 1.1.0a and 1.1.0b [26 Sep 2016]
3796
3797 * Fix Use After Free for large message sizes
3798
3799   The patch applied to address CVE-2016-6307 resulted in an issue where if a
3800   message larger than approx 16k is received then the underlying buffer to
3801   store the incoming message is reallocated and moved. Unfortunately a
3802   dangling pointer to the old location is left which results in an attempt to
3803   write to the previously freed location. This is likely to result in a
3804   crash, however it could potentially lead to execution of arbitrary code.
3805
3806   This issue only affects OpenSSL 1.1.0a.
3807
3808   This issue was reported to OpenSSL by Robert Święcki.
3809   ([CVE-2016-6309])
3810
3811   *Matt Caswell*
3812
3813### Changes between 1.1.0 and 1.1.0a [22 Sep 2016]
3814
3815 * OCSP Status Request extension unbounded memory growth
3816
3817   A malicious client can send an excessively large OCSP Status Request
3818   extension. If that client continually requests renegotiation, sending a
3819   large OCSP Status Request extension each time, then there will be unbounded
3820   memory growth on the server. This will eventually lead to a Denial Of
3821   Service attack through memory exhaustion. Servers with a default
3822   configuration are vulnerable even if they do not support OCSP. Builds using
3823   the "no-ocsp" build time option are not affected.
3824
3825   This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.)
3826   ([CVE-2016-6304])
3827
3828   *Matt Caswell*
3829
3830 * SSL_peek() hang on empty record
3831
3832   OpenSSL 1.1.0 SSL/TLS will hang during a call to SSL_peek() if the peer
3833   sends an empty record. This could be exploited by a malicious peer in a
3834   Denial Of Service attack.
3835
3836   This issue was reported to OpenSSL by Alex Gaynor.
3837   ([CVE-2016-6305])
3838
3839   *Matt Caswell*
3840
3841 * Excessive allocation of memory in tls_get_message_header() and
3842   dtls1_preprocess_fragment()
3843
3844   A (D)TLS message includes 3 bytes for its length in the header for the
3845   message. This would allow for messages up to 16Mb in length. Messages of
3846   this length are excessive and OpenSSL includes a check to ensure that a
3847   peer is sending reasonably sized messages in order to avoid too much memory
3848   being consumed to service a connection. A flaw in the logic of version
3849   1.1.0 means that memory for the message is allocated too early, prior to
3850   the excessive message length check. Due to way memory is allocated in
3851   OpenSSL this could mean an attacker could force up to 21Mb to be allocated
3852   to service a connection. This could lead to a Denial of Service through
3853   memory exhaustion. However, the excessive message length check still takes
3854   place, and this would cause the connection to immediately fail. Assuming
3855   that the application calls SSL_free() on the failed connection in a timely
3856   manner then the 21Mb of allocated memory will then be immediately freed
3857   again. Therefore the excessive memory allocation will be transitory in
3858   nature. This then means that there is only a security impact if:
3859
3860   1) The application does not call SSL_free() in a timely manner in the event
3861   that the connection fails
3862   or
3863   2) The application is working in a constrained environment where there is
3864   very little free memory
3865   or
3866   3) The attacker initiates multiple connection attempts such that there are
3867   multiple connections in a state where memory has been allocated for the
3868   connection; SSL_free() has not yet been called; and there is insufficient
3869   memory to service the multiple requests.
3870
3871   Except in the instance of (1) above any Denial Of Service is likely to be
3872   transitory because as soon as the connection fails the memory is
3873   subsequently freed again in the SSL_free() call. However there is an
3874   increased risk during this period of application crashes due to the lack of
3875   memory - which would then mean a more serious Denial of Service.
3876
3877   This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.)
3878   (CVE-2016-6307 and CVE-2016-6308)
3879
3880   *Matt Caswell*
3881
3882 * solaris-x86-cc, i.e. 32-bit configuration with vendor compiler,
3883   had to be removed. Primary reason is that vendor assembler can't
3884   assemble our modules with -KPIC flag. As result it, assembly
3885   support, was not even available as option. But its lack means
3886   lack of side-channel resistant code, which is incompatible with
3887   security by todays standards. Fortunately gcc is readily available
3888   prepackaged option, which we firmly point at...
3889
3890   *Andy Polyakov*
3891
3892### Changes between 1.0.2h and 1.1.0  [25 Aug 2016]
3893
3894 * Windows command-line tool supports UTF-8 opt-in option for arguments
3895   and console input. Setting OPENSSL_WIN32_UTF8 environment variable
3896   (to any value) allows Windows user to access PKCS#12 file generated
3897   with Windows CryptoAPI and protected with non-ASCII password, as well
3898   as files generated under UTF-8 locale on Linux also protected with
3899   non-ASCII password.
3900
3901   *Andy Polyakov*
3902
3903 * To mitigate the SWEET32 attack ([CVE-2016-2183]), 3DES cipher suites
3904   have been disabled by default and removed from DEFAULT, just like RC4.
3905   See the RC4 item below to re-enable both.
3906
3907   *Rich Salz*
3908
3909 * The method for finding the storage location for the Windows RAND seed file
3910   has changed. First we check %RANDFILE%. If that is not set then we check
3911   the directories %HOME%, %USERPROFILE% and %SYSTEMROOT% in that order. If
3912   all else fails we fall back to C:\.
3913
3914   *Matt Caswell*
3915
3916 * The EVP_EncryptUpdate() function has had its return type changed from void
3917   to int. A return of 0 indicates and error while a return of 1 indicates
3918   success.
3919
3920   *Matt Caswell*
3921
3922 * The flags RSA_FLAG_NO_CONSTTIME, DSA_FLAG_NO_EXP_CONSTTIME and
3923   DH_FLAG_NO_EXP_CONSTTIME which previously provided the ability to switch
3924   off the constant time implementation for RSA, DSA and DH have been made
3925   no-ops and deprecated.
3926
3927   *Matt Caswell*
3928
3929 * Windows RAND implementation was simplified to only get entropy by
3930   calling CryptGenRandom(). Various other RAND-related tickets
3931   were also closed.
3932
3933   *Joseph Wylie Yandle, Rich Salz*
3934
3935 * The stack and lhash API's were renamed to start with `OPENSSL_SK_`
3936   and `OPENSSL_LH_`, respectively.  The old names are available
3937   with API compatibility.  They new names are now completely documented.
3938
3939   *Rich Salz*
3940
3941 * Unify TYPE_up_ref(obj) methods signature.
3942   SSL_CTX_up_ref(), SSL_up_ref(), X509_up_ref(), EVP_PKEY_up_ref(),
3943   X509_CRL_up_ref(), X509_OBJECT_up_ref_count() methods are now returning an
3944   int (instead of void) like all others TYPE_up_ref() methods.
3945   So now these methods also check the return value of CRYPTO_atomic_add(),
3946   and the validity of object reference counter.
3947
3948   *fdasilvayy@gmail.com*
3949
3950 * With Windows Visual Studio builds, the .pdb files are installed
3951   alongside the installed libraries and executables.  For a static
3952   library installation, ossl_static.pdb is the associate compiler
3953   generated .pdb file to be used when linking programs.
3954
3955   *Richard Levitte*
3956
3957 * Remove openssl.spec.  Packaging files belong with the packagers.
3958
3959   *Richard Levitte*
3960
3961 * Automatic Darwin/OSX configuration has had a refresh, it will now
3962   recognise x86_64 architectures automatically.  You can still decide
3963   to build for a different bitness with the environment variable
3964   KERNEL_BITS (can be 32 or 64), for example:
3965
3966           KERNEL_BITS=32 ./config
3967
3968   *Richard Levitte*
3969
3970 * Change default algorithms in pkcs8 utility to use PKCS#5 v2.0,
3971   256 bit AES and HMAC with SHA256.
3972
3973   *Steve Henson*
3974
3975 * Remove support for MIPS o32 ABI on IRIX (and IRIX only).
3976
3977   *Andy Polyakov*
3978
3979 * Triple-DES ciphers have been moved from HIGH to MEDIUM.
3980
3981   *Rich Salz*
3982
3983 * To enable users to have their own config files and build file templates,
3984   Configure looks in the directory indicated by the environment variable
3985   OPENSSL_LOCAL_CONFIG_DIR as well as the in-source Configurations/
3986   directory.  On VMS, OPENSSL_LOCAL_CONFIG_DIR is expected to be a logical
3987   name and is used as is.
3988
3989   *Richard Levitte*
3990
3991 * The following datatypes were made opaque: X509_OBJECT, X509_STORE_CTX,
3992   X509_STORE, X509_LOOKUP, and X509_LOOKUP_METHOD.  The unused type
3993   X509_CERT_FILE_CTX was removed.
3994
3995   *Rich Salz*
3996
3997 * "shared" builds are now the default. To create only static libraries use
3998   the "no-shared" Configure option.
3999
4000   *Matt Caswell*
4001
4002 * Remove the no-aes, no-hmac, no-rsa, no-sha and no-md5 Configure options.
4003   All of these option have not worked for some while and are fundamental
4004   algorithms.
4005
4006   *Matt Caswell*
4007
4008 * Make various cleanup routines no-ops and mark them as deprecated. Most
4009   global cleanup functions are no longer required because they are handled
4010   via auto-deinit (see OPENSSL_init_crypto and OPENSSL_init_ssl man pages).
4011   Explicitly de-initing can cause problems (e.g. where a library that uses
4012   OpenSSL de-inits, but an application is still using it). The affected
4013   functions are CONF_modules_free(), ENGINE_cleanup(), OBJ_cleanup(),
4014   EVP_cleanup(), BIO_sock_cleanup(), CRYPTO_cleanup_all_ex_data(),
4015   RAND_cleanup(), SSL_COMP_free_compression_methods(), ERR_free_strings() and
4016   COMP_zlib_cleanup().
4017
4018   *Matt Caswell*
4019
4020 * --strict-warnings no longer enables runtime debugging options
4021   such as REF_DEBUG. Instead, debug options are automatically
4022   enabled with '--debug' builds.
4023
4024   *Andy Polyakov, Emilia Käsper*
4025
4026 * Made DH and DH_METHOD opaque. The structures for managing DH objects
4027   have been moved out of the public header files. New functions for managing
4028   these have been added.
4029
4030   *Matt Caswell*
4031
4032 * Made RSA and RSA_METHOD opaque. The structures for managing RSA
4033   objects have been moved out of the public header files. New
4034   functions for managing these have been added.
4035
4036   *Richard Levitte*
4037
4038 * Made DSA and DSA_METHOD opaque. The structures for managing DSA objects
4039   have been moved out of the public header files. New functions for managing
4040   these have been added.
4041
4042   *Matt Caswell*
4043
4044 * Made BIO and BIO_METHOD opaque. The structures for managing BIOs have been
4045   moved out of the public header files. New functions for managing these
4046   have been added.
4047
4048   *Matt Caswell*
4049
4050 * Removed no-rijndael as a config option. Rijndael is an old name for AES.
4051
4052   *Matt Caswell*
4053
4054 * Removed the mk1mf build scripts.
4055
4056   *Richard Levitte*
4057
4058 * Headers are now wrapped, if necessary, with OPENSSL_NO_xxx, so
4059   it is always safe to #include a header now.
4060
4061   *Rich Salz*
4062
4063 * Removed the aged BC-32 config and all its supporting scripts
4064
4065   *Richard Levitte*
4066
4067 * Removed support for Ultrix, Netware, and OS/2.
4068
4069   *Rich Salz*
4070
4071 * Add support for HKDF.
4072
4073   *Alessandro Ghedini*
4074
4075 * Add support for blake2b and blake2s
4076
4077   *Bill Cox*
4078
4079 * Added support for "pipelining". Ciphers that have the
4080   EVP_CIPH_FLAG_PIPELINE flag set have a capability to process multiple
4081   encryptions/decryptions simultaneously. There are currently no built-in
4082   ciphers with this property but the expectation is that engines will be able
4083   to offer it to significantly improve throughput. Support has been extended
4084   into libssl so that multiple records for a single connection can be
4085   processed in one go (for >=TLS 1.1).
4086
4087   *Matt Caswell*
4088
4089 * Added the AFALG engine. This is an async capable engine which is able to
4090   offload work to the Linux kernel. In this initial version it only supports
4091   AES128-CBC. The kernel must be version 4.1.0 or greater.
4092
4093   *Catriona Lucey*
4094
4095 * OpenSSL now uses a new threading API. It is no longer necessary to
4096   set locking callbacks to use OpenSSL in a multi-threaded environment. There
4097   are two supported threading models: pthreads and windows threads. It is
4098   also possible to configure OpenSSL at compile time for "no-threads". The
4099   old threading API should no longer be used. The functions have been
4100   replaced with "no-op" compatibility macros.
4101
4102   *Alessandro Ghedini, Matt Caswell*
4103
4104 * Modify behavior of ALPN to invoke callback after SNI/servername
4105   callback, such that updates to the SSL_CTX affect ALPN.
4106
4107   *Todd Short*
4108
4109 * Add SSL_CIPHER queries for authentication and key-exchange.
4110
4111   *Todd Short*
4112
4113 * Changes to the DEFAULT cipherlist:
4114   - Prefer (EC)DHE handshakes over plain RSA.
4115   - Prefer AEAD ciphers over legacy ciphers.
4116   - Prefer ECDSA over RSA when both certificates are available.
4117   - Prefer TLSv1.2 ciphers/PRF.
4118   - Remove DSS, SEED, IDEA, CAMELLIA, and AES-CCM from the
4119     default cipherlist.
4120
4121   *Emilia Käsper*
4122
4123 * Change the ECC default curve list to be this, in order: x25519,
4124   secp256r1, secp521r1, secp384r1.
4125
4126   *Rich Salz*
4127
4128 * RC4 based libssl ciphersuites are now classed as "weak" ciphers and are
4129   disabled by default. They can be re-enabled using the
4130   enable-weak-ssl-ciphers option to Configure.
4131
4132   *Matt Caswell*
4133
4134 * If the server has ALPN configured, but supports no protocols that the
4135   client advertises, send a fatal "no_application_protocol" alert.
4136   This behaviour is SHALL in RFC 7301, though it isn't universally
4137   implemented by other servers.
4138
4139   *Emilia Käsper*
4140
4141 * Add X25519 support.
4142   Add ASN.1 and EVP_PKEY methods for X25519. This includes support
4143   for public and private key encoding using the format documented in
4144   draft-ietf-curdle-pkix-02. The corresponding EVP_PKEY method supports
4145   key generation and key derivation.
4146
4147   TLS support complies with draft-ietf-tls-rfc4492bis-08 and uses
4148   X25519(29).
4149
4150   *Steve Henson*
4151
4152 * Deprecate SRP_VBASE_get_by_user.
4153   SRP_VBASE_get_by_user had inconsistent memory management behaviour.
4154   In order to fix an unavoidable memory leak ([CVE-2016-0798]),
4155   SRP_VBASE_get_by_user was changed to ignore the "fake user" SRP
4156   seed, even if the seed is configured.
4157
4158   Users should use SRP_VBASE_get1_by_user instead. Note that in
4159   SRP_VBASE_get1_by_user, caller must free the returned value. Note
4160   also that even though configuring the SRP seed attempts to hide
4161   invalid usernames by continuing the handshake with fake
4162   credentials, this behaviour is not constant time and no strong
4163   guarantees are made that the handshake is indistinguishable from
4164   that of a valid user.
4165
4166   *Emilia Käsper*
4167
4168 * Configuration change; it's now possible to build dynamic engines
4169   without having to build shared libraries and vice versa.  This
4170   only applies to the engines in `engines/`, those in `crypto/engine/`
4171   will always be built into libcrypto (i.e. "static").
4172
4173   Building dynamic engines is enabled by default; to disable, use
4174   the configuration option "disable-dynamic-engine".
4175
4176   The only requirements for building dynamic engines are the
4177   presence of the DSO module and building with position independent
4178   code, so they will also automatically be disabled if configuring
4179   with "disable-dso" or "disable-pic".
4180
4181   The macros OPENSSL_NO_STATIC_ENGINE and OPENSSL_NO_DYNAMIC_ENGINE
4182   are also taken away from openssl/opensslconf.h, as they are
4183   irrelevant.
4184
4185   *Richard Levitte*
4186
4187 * Configuration change; if there is a known flag to compile
4188   position independent code, it will always be applied on the
4189   libcrypto and libssl object files, and never on the application
4190   object files.  This means other libraries that use routines from
4191   libcrypto / libssl can be made into shared libraries regardless
4192   of how OpenSSL was configured.
4193
4194   If this isn't desirable, the configuration options "disable-pic"
4195   or "no-pic" can be used to disable the use of PIC.  This will
4196   also disable building shared libraries and dynamic engines.
4197
4198   *Richard Levitte*
4199
4200 * Removed JPAKE code.  It was experimental and has no wide use.
4201
4202   *Rich Salz*
4203
4204 * The INSTALL_PREFIX Makefile variable has been renamed to
4205   DESTDIR.  That makes for less confusion on what this variable
4206   is for.  Also, the configuration option --install_prefix is
4207   removed.
4208
4209   *Richard Levitte*
4210
4211 * Heartbeat for TLS has been removed and is disabled by default
4212   for DTLS; configure with enable-heartbeats.  Code that uses the
4213   old #define's might need to be updated.
4214
4215   *Emilia Käsper, Rich Salz*
4216
4217 * Rename REF_CHECK to REF_DEBUG.
4218
4219   *Rich Salz*
4220
4221 * New "unified" build system
4222
4223   The "unified" build system is aimed to be a common system for all
4224   platforms we support.  With it comes new support for VMS.
4225
4226   This system builds supports building in a different directory tree
4227   than the source tree.  It produces one Makefile (for unix family
4228   or lookalikes), or one descrip.mms (for VMS).
4229
4230   The source of information to make the Makefile / descrip.mms is
4231   small files called 'build.info', holding the necessary
4232   information for each directory with source to compile, and a
4233   template in Configurations, like unix-Makefile.tmpl or
4234   descrip.mms.tmpl.
4235
4236   With this change, the library names were also renamed on Windows
4237   and on VMS.  They now have names that are closer to the standard
4238   on Unix, and include the major version number, and in certain
4239   cases, the architecture they are built for.  See "Notes on shared
4240   libraries" in INSTALL.
4241
4242   We rely heavily on the perl module Text::Template.
4243
4244   *Richard Levitte*
4245
4246 * Added support for auto-initialisation and de-initialisation of the library.
4247   OpenSSL no longer requires explicit init or deinit routines to be called,
4248   except in certain circumstances. See the OPENSSL_init_crypto() and
4249   OPENSSL_init_ssl() man pages for further information.
4250
4251   *Matt Caswell*
4252
4253 * The arguments to the DTLSv1_listen function have changed. Specifically the
4254   "peer" argument is now expected to be a BIO_ADDR object.
4255
4256 * Rewrite of BIO networking library. The BIO library lacked consistent
4257   support of IPv6, and adding it required some more extensive
4258   modifications.  This introduces the BIO_ADDR and BIO_ADDRINFO types,
4259   which hold all types of addresses and chains of address information.
4260   It also introduces a new API, with functions like BIO_socket,
4261   BIO_connect, BIO_listen, BIO_lookup and a rewrite of BIO_accept.
4262   The source/sink BIOs BIO_s_connect, BIO_s_accept and BIO_s_datagram
4263   have been adapted accordingly.
4264
4265   *Richard Levitte*
4266
4267 * RSA_padding_check_PKCS1_type_1 now accepts inputs with and without
4268   the leading 0-byte.
4269
4270   *Emilia Käsper*
4271
4272 * CRIME protection: disable compression by default, even if OpenSSL is
4273   compiled with zlib enabled. Applications can still enable compression
4274   by calling SSL_CTX_clear_options(ctx, SSL_OP_NO_COMPRESSION), or by
4275   using the SSL_CONF library to configure compression.
4276
4277   *Emilia Käsper*
4278
4279 * The signature of the session callback configured with
4280   SSL_CTX_sess_set_get_cb was changed. The read-only input buffer
4281   was explicitly marked as `const unsigned char*` instead of
4282   `unsigned char*`.
4283
4284   *Emilia Käsper*
4285
4286 * Always DPURIFY. Remove the use of uninitialized memory in the
4287   RNG, and other conditional uses of DPURIFY. This makes -DPURIFY a no-op.
4288
4289   *Emilia Käsper*
4290
4291 * Removed many obsolete configuration items, including
4292      DES_PTR, DES_RISC1, DES_RISC2, DES_INT
4293      MD2_CHAR, MD2_INT, MD2_LONG
4294      BF_PTR, BF_PTR2
4295      IDEA_SHORT, IDEA_LONG
4296      RC2_SHORT, RC2_LONG, RC4_LONG, RC4_CHUNK, RC4_INDEX
4297
4298   *Rich Salz, with advice from Andy Polyakov*
4299
4300 * Many BN internals have been moved to an internal header file.
4301
4302   *Rich Salz with help from Andy Polyakov*
4303
4304 * Configuration and writing out the results from it has changed.
4305   Files such as Makefile include/openssl/opensslconf.h and are now
4306   produced through general templates, such as Makefile.in and
4307   crypto/opensslconf.h.in and some help from the perl module
4308   Text::Template.
4309
4310   Also, the center of configuration information is no longer
4311   Makefile.  Instead, Configure produces a perl module in
4312   configdata.pm which holds most of the config data (in the hash
4313   table %config), the target data that comes from the target
4314   configuration in one of the `Configurations/*.conf` files (in
4315   %target).
4316
4317   *Richard Levitte*
4318
4319 * To clarify their intended purposes, the Configure options
4320   --prefix and --openssldir change their semantics, and become more
4321   straightforward and less interdependent.
4322
4323   --prefix shall be used exclusively to give the location INSTALLTOP
4324   where programs, scripts, libraries, include files and manuals are
4325   going to be installed.  The default is now /usr/local.
4326
4327   --openssldir shall be used exclusively to give the default
4328   location OPENSSLDIR where certificates, private keys, CRLs are
4329   managed.  This is also where the default openssl.cnf gets
4330   installed.
4331   If the directory given with this option is a relative path, the
4332   values of both the --prefix value and the --openssldir value will
4333   be combined to become OPENSSLDIR.
4334   The default for --openssldir is INSTALLTOP/ssl.
4335
4336   Anyone who uses --openssldir to specify where OpenSSL is to be
4337   installed MUST change to use --prefix instead.
4338
4339   *Richard Levitte*
4340
4341 * The GOST engine was out of date and therefore it has been removed. An up
4342   to date GOST engine is now being maintained in an external repository.
4343   See: <https://wiki.openssl.org/index.php/Binaries>. Libssl still retains
4344   support for GOST ciphersuites (these are only activated if a GOST engine
4345   is present).
4346
4347   *Matt Caswell*
4348
4349 * EGD is no longer supported by default; use enable-egd when
4350   configuring.
4351
4352   *Ben Kaduk and Rich Salz*
4353
4354 * The distribution now has Makefile.in files, which are used to
4355   create Makefile's when Configure is run.  *Configure must be run
4356   before trying to build now.*
4357
4358   *Rich Salz*
4359
4360 * The return value for SSL_CIPHER_description() for error conditions
4361   has changed.
4362
4363   *Rich Salz*
4364
4365 * Support for RFC6698/RFC7671 DANE TLSA peer authentication.
4366
4367   Obtaining and performing DNSSEC validation of TLSA records is
4368   the application's responsibility.  The application provides
4369   the TLSA records of its choice to OpenSSL, and these are then
4370   used to authenticate the peer.
4371
4372   The TLSA records need not even come from DNS.  They can, for
4373   example, be used to implement local end-entity certificate or
4374   trust-anchor "pinning", where the "pin" data takes the form
4375   of TLSA records, which can augment or replace verification
4376   based on the usual WebPKI public certification authorities.
4377
4378   *Viktor Dukhovni*
4379
4380 * Revert default OPENSSL_NO_DEPRECATED setting.  Instead OpenSSL
4381   continues to support deprecated interfaces in default builds.
4382   However, applications are strongly advised to compile their
4383   source files with -DOPENSSL_API_COMPAT=0x10100000L, which hides
4384   the declarations of all interfaces deprecated in 0.9.8, 1.0.0
4385   or the 1.1.0 releases.
4386
4387   In environments in which all applications have been ported to
4388   not use any deprecated interfaces OpenSSL's Configure script
4389   should be used with the --api=1.1.0 option to entirely remove
4390   support for the deprecated features from the library and
4391   unconditionally disable them in the installed headers.
4392   Essentially the same effect can be achieved with the "no-deprecated"
4393   argument to Configure, except that this will always restrict
4394   the build to just the latest API, rather than a fixed API
4395   version.
4396
4397   As applications are ported to future revisions of the API,
4398   they should update their compile-time OPENSSL_API_COMPAT define
4399   accordingly, but in most cases should be able to continue to
4400   compile with later releases.
4401
4402   The OPENSSL_API_COMPAT versions for 1.0.0, and 0.9.8 are
4403   0x10000000L and 0x00908000L, respectively.  However those
4404   versions did not support the OPENSSL_API_COMPAT feature, and
4405   so applications are not typically tested for explicit support
4406   of just the undeprecated features of either release.
4407
4408   *Viktor Dukhovni*
4409
4410 * Add support for setting the minimum and maximum supported protocol.
4411   It can bet set via the SSL_set_min_proto_version() and
4412   SSL_set_max_proto_version(), or via the SSL_CONF's MinProtocol and
4413   MaxProtocol.  It's recommended to use the new APIs to disable
4414   protocols instead of disabling individual protocols using
4415   SSL_set_options() or SSL_CONF's Protocol.  This change also
4416   removes support for disabling TLS 1.2 in the OpenSSL TLS
4417   client at compile time by defining OPENSSL_NO_TLS1_2_CLIENT.
4418
4419   *Kurt Roeckx*
4420
4421 * Support for ChaCha20 and Poly1305 added to libcrypto and libssl.
4422
4423   *Andy Polyakov*
4424
4425 * New EC_KEY_METHOD, this replaces the older ECDSA_METHOD and ECDH_METHOD
4426   and integrates ECDSA and ECDH functionality into EC. Implementations can
4427   now redirect key generation and no longer need to convert to or from
4428   ECDSA_SIG format.
4429
4430   Note: the ecdsa.h and ecdh.h headers are now no longer needed and just
4431   include the ec.h header file instead.
4432
4433   *Steve Henson*
4434
4435 * Remove support for all 40 and 56 bit ciphers.  This includes all the export
4436   ciphers who are no longer supported and drops support the ephemeral RSA key
4437   exchange. The LOW ciphers currently doesn't have any ciphers in it.
4438
4439   *Kurt Roeckx*
4440
4441 * Made EVP_MD_CTX, EVP_MD, EVP_CIPHER_CTX, EVP_CIPHER and HMAC_CTX
4442   opaque.  For HMAC_CTX, the following constructors and destructors
4443   were added:
4444
4445       HMAC_CTX *HMAC_CTX_new(void);
4446       void HMAC_CTX_free(HMAC_CTX *ctx);
4447
4448   For EVP_MD and EVP_CIPHER, complete APIs to create, fill and
4449   destroy such methods has been added.  See EVP_MD_meth_new(3) and
4450   EVP_CIPHER_meth_new(3) for documentation.
4451
4452   Additional changes:
4453   1) `EVP_MD_CTX_cleanup()`, `EVP_CIPHER_CTX_cleanup()` and
4454      `HMAC_CTX_cleanup()` were removed. `HMAC_CTX_reset()` and
4455      `EVP_MD_CTX_reset()` should be called instead to reinitialise
4456      an already created structure.
4457   2) For consistency with the majority of our object creators and
4458      destructors, `EVP_MD_CTX_(create|destroy)` were renamed to
4459      `EVP_MD_CTX_(new|free)`.  The old names are retained as macros
4460      for deprecated builds.
4461
4462   *Richard Levitte*
4463
4464 * Added ASYNC support. Libcrypto now includes the async sub-library to enable
4465   cryptographic operations to be performed asynchronously as long as an
4466   asynchronous capable engine is used. See the ASYNC_start_job() man page for
4467   further details. Libssl has also had this capability integrated with the
4468   introduction of the new mode SSL_MODE_ASYNC and associated error
4469   SSL_ERROR_WANT_ASYNC. See the SSL_CTX_set_mode() and SSL_get_error() man
4470   pages. This work was developed in partnership with Intel Corp.
4471
4472   *Matt Caswell*
4473
4474 * SSL_{CTX_}set_ecdh_auto() has been removed and ECDH is support is
4475   always enabled now.  If you want to disable the support you should
4476   exclude it using the list of supported ciphers. This also means that the
4477   "-no_ecdhe" option has been removed from s_server.
4478
4479   *Kurt Roeckx*
4480
4481 * SSL_{CTX}_set_tmp_ecdh() which can set 1 EC curve now internally calls
4482   SSL_{CTX_}set1_curves() which can set a list.
4483
4484   *Kurt Roeckx*
4485
4486 * Remove support for SSL_{CTX_}set_tmp_ecdh_callback().  You should set the
4487   curve you want to support using SSL_{CTX_}set1_curves().
4488
4489   *Kurt Roeckx*
4490
4491 * State machine rewrite. The state machine code has been significantly
4492   refactored in order to remove much duplication of code and solve issues
4493   with the old code (see [ssl/statem/README.md](ssl/statem/README.md) for
4494   further details). This change does have some associated API changes.
4495   Notably the SSL_state() function has been removed and replaced by
4496   SSL_get_state which now returns an "OSSL_HANDSHAKE_STATE" instead of an int.
4497   SSL_set_state() has been removed altogether. The previous handshake states
4498   defined in ssl.h and ssl3.h have also been removed.
4499
4500   *Matt Caswell*
4501
4502 * All instances of the string "ssleay" in the public API were replaced
4503   with OpenSSL (case-matching; e.g., OPENSSL_VERSION for #define's)
4504   Some error codes related to internal RSA_eay API's were renamed.
4505
4506   *Rich Salz*
4507
4508 * The demo files in crypto/threads were moved to demo/threads.
4509
4510   *Rich Salz*
4511
4512 * Removed obsolete engines: 4758cca, aep, atalla, cswift, nuron, gmp,
4513   sureware and ubsec.
4514
4515   *Matt Caswell, Rich Salz*
4516
4517 * New ASN.1 embed macro.
4518
4519   New ASN.1 macro ASN1_EMBED. This is the same as ASN1_SIMPLE except the
4520   structure is not allocated: it is part of the parent. That is instead of
4521
4522           FOO *x;
4523
4524   it must be:
4525
4526           FOO x;
4527
4528   This reduces memory fragmentation and make it impossible to accidentally
4529   set a mandatory field to NULL.
4530
4531   This currently only works for some fields specifically a SEQUENCE, CHOICE,
4532   or ASN1_STRING type which is part of a parent SEQUENCE. Since it is
4533   equivalent to ASN1_SIMPLE it cannot be tagged, OPTIONAL, SET OF or
4534   SEQUENCE OF.
4535
4536   *Steve Henson*
4537
4538 * Remove EVP_CHECK_DES_KEY, a compile-time option that never compiled.
4539
4540   *Emilia Käsper*
4541
4542 * Removed DES and RC4 ciphersuites from DEFAULT. Also removed RC2 although
4543   in 1.0.2 EXPORT was already removed and the only RC2 ciphersuite is also
4544   an EXPORT one. COMPLEMENTOFDEFAULT has been updated accordingly to add
4545   DES and RC4 ciphersuites.
4546
4547   *Matt Caswell*
4548
4549 * Rewrite EVP_DecodeUpdate (base64 decoding) to fix several bugs.
4550   This changes the decoding behaviour for some invalid messages,
4551   though the change is mostly in the more lenient direction, and
4552   legacy behaviour is preserved as much as possible.
4553
4554   *Emilia Käsper*
4555
4556 * Fix no-stdio build.
4557   *David Woodhouse <David.Woodhouse@intel.com> and also*
4558   *Ivan Nestlerode <ivan.nestlerode@sonos.com>*
4559
4560 * New testing framework
4561   The testing framework has been largely rewritten and is now using
4562   perl and the perl modules Test::Harness and an extended variant of
4563   Test::More called OpenSSL::Test to do its work.  All test scripts in
4564   test/ have been rewritten into test recipes, and all direct calls to
4565   executables in test/Makefile have become individual recipes using the
4566   simplified testing OpenSSL::Test::Simple.
4567
4568   For documentation on our testing modules, do:
4569
4570           perldoc test/testlib/OpenSSL/Test/Simple.pm
4571           perldoc test/testlib/OpenSSL/Test.pm
4572
4573   *Richard Levitte*
4574
4575 * Revamped memory debug; only -DCRYPTO_MDEBUG and -DCRYPTO_MDEBUG_ABORT
4576   are used; the latter aborts on memory leaks (usually checked on exit).
4577   Some undocumented "set malloc, etc., hooks" functions were removed
4578   and others were changed.  All are now documented.
4579
4580   *Rich Salz*
4581
4582 * In DSA_generate_parameters_ex, if the provided seed is too short,
4583   return an error
4584
4585   *Rich Salz and Ismo Puustinen <ismo.puustinen@intel.com>*
4586
4587 * Rewrite PSK to support ECDHE_PSK, DHE_PSK and RSA_PSK. Add ciphersuites
4588   from RFC4279, RFC4785, RFC5487, RFC5489.
4589
4590   Thanks to Christian J. Dietrich and Giuseppe D'Angelo for the
4591   original RSA_PSK patch.
4592
4593   *Steve Henson*
4594
4595 * Dropped support for the SSL3_FLAGS_DELAY_CLIENT_FINISHED flag. This SSLeay
4596   era flag was never set throughout the codebase (only read). Also removed
4597   SSL3_FLAGS_POP_BUFFER which was only used if
4598   SSL3_FLAGS_DELAY_CLIENT_FINISHED was also set.
4599
4600   *Matt Caswell*
4601
4602 * Changed the default name options in the "ca", "crl", "req" and "x509"
4603   to be "oneline" instead of "compat".
4604
4605   *Richard Levitte*
4606
4607 * Remove SSL_OP_TLS_BLOCK_PADDING_BUG. This is SSLeay legacy, we're
4608   not aware of clients that still exhibit this bug, and the workaround
4609   hasn't been working properly for a while.
4610
4611   *Emilia Käsper*
4612
4613 * The return type of BIO_number_read() and BIO_number_written() as well as
4614   the corresponding num_read and num_write members in the BIO structure has
4615   changed from unsigned long to uint64_t. On platforms where an unsigned
4616   long is 32 bits (e.g. Windows) these counters could overflow if >4Gb is
4617   transferred.
4618
4619   *Matt Caswell*
4620
4621 * Given the pervasive nature of TLS extensions it is inadvisable to run
4622   OpenSSL without support for them. It also means that maintaining
4623   the OPENSSL_NO_TLSEXT option within the code is very invasive (and probably
4624   not well tested). Therefore the OPENSSL_NO_TLSEXT option has been removed.
4625
4626   *Matt Caswell*
4627
4628 * Removed support for the two export grade static DH ciphersuites
4629   EXP-DH-RSA-DES-CBC-SHA and EXP-DH-DSS-DES-CBC-SHA. These two ciphersuites
4630   were newly added (along with a number of other static DH ciphersuites) to
4631   1.0.2. However the two export ones have *never* worked since they were
4632   introduced. It seems strange in any case to be adding new export
4633   ciphersuites, and given "logjam" it also does not seem correct to fix them.
4634
4635   *Matt Caswell*
4636
4637 * Version negotiation has been rewritten. In particular SSLv23_method(),
4638   SSLv23_client_method() and SSLv23_server_method() have been deprecated,
4639   and turned into macros which simply call the new preferred function names
4640   TLS_method(), TLS_client_method() and TLS_server_method(). All new code
4641   should use the new names instead. Also as part of this change the ssl23.h
4642   header file has been removed.
4643
4644   *Matt Caswell*
4645
4646 * Support for Kerberos ciphersuites in TLS (RFC2712) has been removed. This
4647   code and the associated standard is no longer considered fit-for-purpose.
4648
4649   *Matt Caswell*
4650
4651 * RT2547 was closed.  When generating a private key, try to make the
4652   output file readable only by the owner.  This behavior change might
4653   be noticeable when interacting with other software.
4654
4655 * Documented all exdata functions.  Added CRYPTO_free_ex_index.
4656   Added a test.
4657
4658   *Rich Salz*
4659
4660 * Added HTTP GET support to the ocsp command.
4661
4662   *Rich Salz*
4663
4664 * Changed default digest for the dgst and enc commands from MD5 to
4665   sha256
4666
4667   *Rich Salz*
4668
4669 * RAND_pseudo_bytes has been deprecated. Users should use RAND_bytes instead.
4670
4671   *Matt Caswell*
4672
4673 * Added support for TLS extended master secret from
4674   draft-ietf-tls-session-hash-03.txt. Thanks for Alfredo Pironti for an
4675   initial patch which was a great help during development.
4676
4677   *Steve Henson*
4678
4679 * All libssl internal structures have been removed from the public header
4680   files, and the OPENSSL_NO_SSL_INTERN option has been removed (since it is
4681   now redundant). Users should not attempt to access internal structures
4682   directly. Instead they should use the provided API functions.
4683
4684   *Matt Caswell*
4685
4686 * config has been changed so that by default OPENSSL_NO_DEPRECATED is used.
4687   Access to deprecated functions can be re-enabled by running config with
4688   "enable-deprecated". In addition applications wishing to use deprecated
4689   functions must define OPENSSL_USE_DEPRECATED. Note that this new behaviour
4690   will, by default, disable some transitive includes that previously existed
4691   in the header files (e.g. ec.h will no longer, by default, include bn.h)
4692
4693   *Matt Caswell*
4694
4695 * Added support for OCB mode. OpenSSL has been granted a patent license
4696   compatible with the OpenSSL license for use of OCB. Details are available
4697   at <https://www.openssl.org/source/OCB-patent-grant-OpenSSL.pdf>. Support
4698   for OCB can be removed by calling config with no-ocb.
4699
4700   *Matt Caswell*
4701
4702 * SSLv2 support has been removed.  It still supports receiving a SSLv2
4703   compatible client hello.
4704
4705   *Kurt Roeckx*
4706
4707 * Increased the minimal RSA keysize from 256 to 512 bits [Rich Salz],
4708   done while fixing the error code for the key-too-small case.
4709
4710   *Annie Yousar <a.yousar@informatik.hu-berlin.de>*
4711
4712 * CA.sh has been removed; use CA.pl instead.
4713
4714   *Rich Salz*
4715
4716 * Removed old DES API.
4717
4718   *Rich Salz*
4719
4720 * Remove various unsupported platforms:
4721      Sony NEWS4
4722      BEOS and BEOS_R5
4723      NeXT
4724      SUNOS
4725      MPE/iX
4726      Sinix/ReliantUNIX RM400
4727      DGUX
4728      NCR
4729      Tandem
4730      Cray
4731      16-bit platforms such as WIN16
4732
4733   *Rich Salz*
4734
4735 * Clean up OPENSSL_NO_xxx #define's
4736   - Use setbuf() and remove OPENSSL_NO_SETVBUF_IONBF
4737   - Rename OPENSSL_SYSNAME_xxx to OPENSSL_SYS_xxx
4738   - OPENSSL_NO_EC{DH,DSA} merged into OPENSSL_NO_EC
4739   - OPENSSL_NO_RIPEMD160, OPENSSL_NO_RIPEMD merged into OPENSSL_NO_RMD160
4740   - OPENSSL_NO_FP_API merged into OPENSSL_NO_STDIO
4741   - Remove OPENSSL_NO_BIO OPENSSL_NO_BUFFER OPENSSL_NO_CHAIN_VERIFY
4742     OPENSSL_NO_EVP OPENSSL_NO_FIPS_ERR OPENSSL_NO_HASH_COMP
4743     OPENSSL_NO_LHASH OPENSSL_NO_OBJECT OPENSSL_NO_SPEED OPENSSL_NO_STACK
4744     OPENSSL_NO_X509 OPENSSL_NO_X509_VERIFY
4745   - Remove MS_STATIC; it's a relic from platforms <32 bits.
4746
4747   *Rich Salz*
4748
4749 * Cleaned up dead code
4750     Remove all but one '#ifdef undef' which is to be looked at.
4751
4752   *Rich Salz*
4753
4754 * Clean up calling of xxx_free routines.
4755      Just like free(), fix most of the xxx_free routines to accept
4756      NULL.  Remove the non-null checks from callers.  Save much code.
4757
4758   *Rich Salz*
4759
4760 * Add secure heap for storage of private keys (when possible).
4761   Add BIO_s_secmem(), CBIGNUM, etc.
4762   Contributed by Akamai Technologies under our Corporate CLA.
4763
4764   *Rich Salz*
4765
4766 * Experimental support for a new, fast, unbiased prime candidate generator,
4767   bn_probable_prime_dh_coprime(). Not currently used by any prime generator.
4768
4769   *Felix Laurie von Massenbach <felix@erbridge.co.uk>*
4770
4771 * New output format NSS in the sess_id command line tool. This allows
4772   exporting the session id and the master key in NSS keylog format.
4773
4774   *Martin Kaiser <martin@kaiser.cx>*
4775
4776 * Harmonize version and its documentation. -f flag is used to display
4777   compilation flags.
4778
4779   *mancha <mancha1@zoho.com>*
4780
4781 * Fix eckey_priv_encode so it immediately returns an error upon a failure
4782   in i2d_ECPrivateKey.  Thanks to Ted Unangst for feedback on this issue.
4783
4784   *mancha <mancha1@zoho.com>*
4785
4786 * Fix some double frees. These are not thought to be exploitable.
4787
4788   *mancha <mancha1@zoho.com>*
4789
4790 * A missing bounds check in the handling of the TLS heartbeat extension
4791   can be used to reveal up to 64k of memory to a connected client or
4792   server.
4793
4794   Thanks for Neel Mehta of Google Security for discovering this bug and to
4795   Adam Langley <agl@chromium.org> and Bodo Moeller <bmoeller@acm.org> for
4796   preparing the fix ([CVE-2014-0160])
4797
4798   *Adam Langley, Bodo Moeller*
4799
4800 * Fix for the attack described in the paper "Recovering OpenSSL
4801   ECDSA Nonces Using the FLUSH+RELOAD Cache Side-channel Attack"
4802   by Yuval Yarom and Naomi Benger. Details can be obtained from:
4803   <http://eprint.iacr.org/2014/140>
4804
4805   Thanks to Yuval Yarom and Naomi Benger for discovering this
4806   flaw and to Yuval Yarom for supplying a fix ([CVE-2014-0076])
4807
4808   *Yuval Yarom and Naomi Benger*
4809
4810 * Use algorithm specific chains in SSL_CTX_use_certificate_chain_file():
4811   this fixes a limitation in previous versions of OpenSSL.
4812
4813   *Steve Henson*
4814
4815 * Experimental encrypt-then-mac support.
4816
4817   Experimental support for encrypt then mac from
4818   draft-gutmann-tls-encrypt-then-mac-02.txt
4819
4820   To enable it set the appropriate extension number (0x42 for the test
4821   server) using e.g. -DTLSEXT_TYPE_encrypt_then_mac=0x42
4822
4823   For non-compliant peers (i.e. just about everything) this should have no
4824   effect.
4825
4826   WARNING: EXPERIMENTAL, SUBJECT TO CHANGE.
4827
4828   *Steve Henson*
4829
4830 * Add EVP support for key wrapping algorithms, to avoid problems with
4831   existing code the flag EVP_CIPHER_CTX_WRAP_ALLOW has to be set in
4832   the EVP_CIPHER_CTX or an error is returned. Add AES and DES3 wrap
4833   algorithms and include tests cases.
4834
4835   *Steve Henson*
4836
4837 * Extend CMS code to support RSA-PSS signatures and RSA-OAEP for
4838   enveloped data.
4839
4840   *Steve Henson*
4841
4842 * Extended RSA OAEP support via EVP_PKEY API. Options to specify digest,
4843   MGF1 digest and OAEP label.
4844
4845   *Steve Henson*
4846
4847 * Make openssl verify return errors.
4848
4849   *Chris Palmer <palmer@google.com> and Ben Laurie*
4850
4851 * New function ASN1_TIME_diff to calculate the difference between two
4852   ASN1_TIME structures or one structure and the current time.
4853
4854   *Steve Henson*
4855
4856 * Update fips_test_suite to support multiple command line options. New
4857   test to induce all self test errors in sequence and check expected
4858   failures.
4859
4860   *Steve Henson*
4861
4862 * Add FIPS_{rsa,dsa,ecdsa}_{sign,verify} functions which digest and
4863   sign or verify all in one operation.
4864
4865   *Steve Henson*
4866
4867 * Add fips_algvs: a multicall fips utility incorporating all the algorithm
4868   test programs and fips_test_suite. Includes functionality to parse
4869   the minimal script output of fipsalgest.pl directly.
4870
4871   *Steve Henson*
4872
4873 * Add authorisation parameter to FIPS_module_mode_set().
4874
4875   *Steve Henson*
4876
4877 * Add FIPS selftest for ECDH algorithm using P-224 and B-233 curves.
4878
4879   *Steve Henson*
4880
4881 * Use separate DRBG fields for internal and external flags. New function
4882   FIPS_drbg_health_check() to perform on demand health checking. Add
4883   generation tests to fips_test_suite with reduced health check interval to
4884   demonstrate periodic health checking. Add "nodh" option to
4885   fips_test_suite to skip very slow DH test.
4886
4887   *Steve Henson*
4888
4889 * New function FIPS_get_cipherbynid() to lookup FIPS supported ciphers
4890   based on NID.
4891
4892   *Steve Henson*
4893
4894 * More extensive health check for DRBG checking many more failure modes.
4895   New function FIPS_selftest_drbg_all() to handle every possible DRBG
4896   combination: call this in fips_test_suite.
4897
4898   *Steve Henson*
4899
4900 * Add support for canonical generation of DSA parameter 'g'. See
4901   FIPS 186-3 A.2.3.
4902
4903 * Add support for HMAC DRBG from SP800-90. Update DRBG algorithm test and
4904   POST to handle HMAC cases.
4905
4906   *Steve Henson*
4907
4908 * Add functions FIPS_module_version() and FIPS_module_version_text()
4909   to return numerical and string versions of the FIPS module number.
4910
4911   *Steve Henson*
4912
4913 * Rename FIPS_mode_set and FIPS_mode to FIPS_module_mode_set and
4914   FIPS_module_mode. FIPS_mode and FIPS_mode_set will be implemented
4915   outside the validated module in the FIPS capable OpenSSL.
4916
4917   *Steve Henson*
4918
4919 * Minor change to DRBG entropy callback semantics. In some cases
4920   there is no multiple of the block length between min_len and
4921   max_len. Allow the callback to return more than max_len bytes
4922   of entropy but discard any extra: it is the callback's responsibility
4923   to ensure that the extra data discarded does not impact the
4924   requested amount of entropy.
4925
4926   *Steve Henson*
4927
4928 * Add PRNG security strength checks to RSA, DSA and ECDSA using
4929   information in FIPS186-3, SP800-57 and SP800-131A.
4930
4931   *Steve Henson*
4932
4933 * CCM support via EVP. Interface is very similar to GCM case except we
4934   must supply all data in one chunk (i.e. no update, final) and the
4935   message length must be supplied if AAD is used. Add algorithm test
4936   support.
4937
4938   *Steve Henson*
4939
4940 * Initial version of POST overhaul. Add POST callback to allow the status
4941   of POST to be monitored and/or failures induced. Modify fips_test_suite
4942   to use callback. Always run all selftests even if one fails.
4943
4944   *Steve Henson*
4945
4946 * XTS support including algorithm test driver in the fips_gcmtest program.
4947   Note: this does increase the maximum key length from 32 to 64 bytes but
4948   there should be no binary compatibility issues as existing applications
4949   will never use XTS mode.
4950
4951   *Steve Henson*
4952
4953 * Extensive reorganisation of FIPS PRNG behaviour. Remove all dependencies
4954   to OpenSSL RAND code and replace with a tiny FIPS RAND API which also
4955   performs algorithm blocking for unapproved PRNG types. Also do not
4956   set PRNG type in FIPS_mode_set(): leave this to the application.
4957   Add default OpenSSL DRBG handling: sets up FIPS PRNG and seeds with
4958   the standard OpenSSL PRNG: set additional data to a date time vector.
4959
4960   *Steve Henson*
4961
4962 * Rename old X9.31 PRNG functions of the form `FIPS_rand*` to `FIPS_x931*`.
4963   This shouldn't present any incompatibility problems because applications
4964   shouldn't be using these directly and any that are will need to rethink
4965   anyway as the X9.31 PRNG is now deprecated by FIPS 140-2
4966
4967   *Steve Henson*
4968
4969 * Extensive self tests and health checking required by SP800-90 DRBG.
4970   Remove strength parameter from FIPS_drbg_instantiate and always
4971   instantiate at maximum supported strength.
4972
4973   *Steve Henson*
4974
4975 * Add ECDH code to fips module and fips_ecdhvs for primitives only testing.
4976
4977   *Steve Henson*
4978
4979 * New algorithm test program fips_dhvs to handle DH primitives only testing.
4980
4981   *Steve Henson*
4982
4983 * New function DH_compute_key_padded() to compute a DH key and pad with
4984   leading zeroes if needed: this complies with SP800-56A et al.
4985
4986   *Steve Henson*
4987
4988 * Initial implementation of SP800-90 DRBGs for Hash and CTR. Not used by
4989   anything, incomplete, subject to change and largely untested at present.
4990
4991   *Steve Henson*
4992
4993 * Modify fipscanisteronly build option to only build the necessary object
4994   files by filtering FIPS_EX_OBJ through a perl script in crypto/Makefile.
4995
4996   *Steve Henson*
4997
4998 * Add experimental option FIPSSYMS to give all symbols in
4999   fipscanister.o and FIPS or fips prefix. This will avoid
5000   conflicts with future versions of OpenSSL. Add perl script
5001   util/fipsas.pl to preprocess assembly language source files
5002   and rename any affected symbols.
5003
5004   *Steve Henson*
5005
5006 * Add selftest checks and algorithm block of non-fips algorithms in
5007   FIPS mode. Remove DES2 from selftests.
5008
5009   *Steve Henson*
5010
5011 * Add ECDSA code to fips module. Add tiny fips_ecdsa_check to just
5012   return internal method without any ENGINE dependencies. Add new
5013   tiny fips sign and verify functions.
5014
5015   *Steve Henson*
5016
5017 * New build option no-ec2m to disable characteristic 2 code.
5018
5019   *Steve Henson*
5020
5021 * New build option "fipscanisteronly". This only builds fipscanister.o
5022   and (currently) associated fips utilities. Uses the file Makefile.fips
5023   instead of Makefile.org as the prototype.
5024
5025   *Steve Henson*
5026
5027 * Add some FIPS mode restrictions to GCM. Add internal IV generator.
5028   Update fips_gcmtest to use IV generator.
5029
5030   *Steve Henson*
5031
5032 * Initial, experimental EVP support for AES-GCM. AAD can be input by
5033   setting output buffer to NULL. The `*Final` function must be
5034   called although it will not retrieve any additional data. The tag
5035   can be set or retrieved with a ctrl. The IV length is by default 12
5036   bytes (96 bits) but can be set to an alternative value. If the IV
5037   length exceeds the maximum IV length (currently 16 bytes) it cannot be
5038   set before the key.
5039
5040   *Steve Henson*
5041
5042 * New flag in ciphers: EVP_CIPH_FLAG_CUSTOM_CIPHER. This means the
5043   underlying do_cipher function handles all cipher semantics itself
5044   including padding and finalisation. This is useful if (for example)
5045   an ENGINE cipher handles block padding itself. The behaviour of
5046   do_cipher is subtly changed if this flag is set: the return value
5047   is the number of characters written to the output buffer (zero is
5048   no longer an error code) or a negative error code. Also if the
5049   input buffer is NULL and length 0 finalisation should be performed.
5050
5051   *Steve Henson*
5052
5053 * If a candidate issuer certificate is already part of the constructed
5054   path ignore it: new debug notification X509_V_ERR_PATH_LOOP for this case.
5055
5056   *Steve Henson*
5057
5058 * Improve forward-security support: add functions
5059
5060           void SSL_CTX_set_not_resumable_session_callback(
5061                    SSL_CTX *ctx, int (*cb)(SSL *ssl, int is_forward_secure))
5062           void SSL_set_not_resumable_session_callback(
5063                    SSL *ssl, int (*cb)(SSL *ssl, int is_forward_secure))
5064
5065   for use by SSL/TLS servers; the callback function will be called whenever a
5066   new session is created, and gets to decide whether the session may be
5067   cached to make it resumable (return 0) or not (return 1).  (As by the
5068   SSL/TLS protocol specifications, the session_id sent by the server will be
5069   empty to indicate that the session is not resumable; also, the server will
5070   not generate RFC 4507 (RFC 5077) session tickets.)
5071
5072   A simple reasonable callback implementation is to return is_forward_secure.
5073   This parameter will be set to 1 or 0 depending on the ciphersuite selected
5074   by the SSL/TLS server library, indicating whether it can provide forward
5075   security.
5076
5077   *Emilia Käsper <emilia.kasper@esat.kuleuven.be> (Google)*
5078
5079 * New -verify_name option in command line utilities to set verification
5080   parameters by name.
5081
5082   *Steve Henson*
5083
5084 * Initial CMAC implementation. WARNING: EXPERIMENTAL, API MAY CHANGE.
5085   Add CMAC pkey methods.
5086
5087   *Steve Henson*
5088
5089 * Experimental renegotiation in s_server -www mode. If the client
5090   browses /reneg connection is renegotiated. If /renegcert it is
5091   renegotiated requesting a certificate.
5092
5093   *Steve Henson*
5094
5095 * Add an "external" session cache for debugging purposes to s_server. This
5096   should help trace issues which normally are only apparent in deployed
5097   multi-process servers.
5098
5099   *Steve Henson*
5100
5101 * Extensive audit of libcrypto with DEBUG_UNUSED. Fix many cases where
5102   return value is ignored. NB. The functions RAND_add(), RAND_seed(),
5103   BIO_set_cipher() and some obscure PEM functions were changed so they
5104   can now return an error. The RAND changes required a change to the
5105   RAND_METHOD structure.
5106
5107   *Steve Henson*
5108
5109 * New macro `__owur` for "OpenSSL Warn Unused Result". This makes use of
5110   a gcc attribute to warn if the result of a function is ignored. This
5111   is enable if DEBUG_UNUSED is set. Add to several functions in evp.h
5112   whose return value is often ignored.
5113
5114   *Steve Henson*
5115
5116 * New -noct, -requestct, -requirect and -ctlogfile options for s_client.
5117   These allow SCTs (signed certificate timestamps) to be requested and
5118   validated when establishing a connection.
5119
5120   *Rob Percival <robpercival@google.com>*
5121
5122OpenSSL 1.0.2
5123-------------
5124
5125### Changes between 1.0.2s and 1.0.2t [10 Sep 2019]
5126
5127 * For built-in EC curves, ensure an EC_GROUP built from the curve name is
5128   used even when parsing explicit parameters, when loading a encoded key
5129   or calling `EC_GROUP_new_from_ecpkparameters()`/
5130   `EC_GROUP_new_from_ecparameters()`.
5131   This prevents bypass of security hardening and performance gains,
5132   especially for curves with specialized EC_METHODs.
5133   By default, if a key encoded with explicit parameters is loaded and later
5134   encoded, the output is still encoded with explicit parameters, even if
5135   internally a "named" EC_GROUP is used for computation.
5136
5137   *Nicola Tuveri*
5138
5139 * Compute ECC cofactors if not provided during EC_GROUP construction. Before
5140   this change, EC_GROUP_set_generator would accept order and/or cofactor as
5141   NULL. After this change, only the cofactor parameter can be NULL. It also
5142   does some minimal sanity checks on the passed order.
5143   ([CVE-2019-1547])
5144
5145   *Billy Bob Brumley*
5146
5147 * Fixed a padding oracle in PKCS7_dataDecode and CMS_decrypt_set1_pkey.
5148   An attack is simple, if the first CMS_recipientInfo is valid but the
5149   second CMS_recipientInfo is chosen ciphertext. If the second
5150   recipientInfo decodes to PKCS #1 v1.5 form plaintext, the correct
5151   encryption key will be replaced by garbage, and the message cannot be
5152   decoded, but if the RSA decryption fails, the correct encryption key is
5153   used and the recipient will not notice the attack.
5154   As a work around for this potential attack the length of the decrypted
5155   key must be equal to the cipher default key length, in case the
5156   certifiate is not given and all recipientInfo are tried out.
5157   The old behaviour can be re-enabled in the CMS code by setting the
5158   CMS_DEBUG_DECRYPT flag.
5159   ([CVE-2019-1563])
5160
5161   *Bernd Edlinger*
5162
5163 * Document issue with installation paths in diverse Windows builds
5164
5165   '/usr/local/ssl' is an unsafe prefix for location to install OpenSSL
5166   binaries and run-time config file.
5167   ([CVE-2019-1552])
5168
5169   *Richard Levitte*
5170
5171### Changes between 1.0.2r and 1.0.2s [28 May 2019]
5172
5173 * Change the default RSA, DSA and DH size to 2048 bit instead of 1024.
5174   This changes the size when using the `genpkey` command when no size is given.
5175   It fixes an omission in earlier changes that changed all RSA, DSA and DH
5176   generation commands to use 2048 bits by default.
5177
5178   *Kurt Roeckx*
5179
5180 * Add FIPS support for Android Arm 64-bit
5181
5182   Support for Android Arm 64-bit was added to the OpenSSL FIPS Object
5183   Module in Version 2.0.10. For some reason, the corresponding target
5184   'android64-aarch64' was missing OpenSSL 1.0.2, whence it could not be
5185   built with FIPS support on Android Arm 64-bit. This omission has been
5186   fixed.
5187
5188   *Matthias St. Pierre*
5189
5190### Changes between 1.0.2q and 1.0.2r [26 Feb 2019]
5191
5192 * 0-byte record padding oracle
5193
5194   If an application encounters a fatal protocol error and then calls
5195   SSL_shutdown() twice (once to send a close_notify, and once to receive one)
5196   then OpenSSL can respond differently to the calling application if a 0 byte
5197   record is received with invalid padding compared to if a 0 byte record is
5198   received with an invalid MAC. If the application then behaves differently
5199   based on that in a way that is detectable to the remote peer, then this
5200   amounts to a padding oracle that could be used to decrypt data.
5201
5202   In order for this to be exploitable "non-stitched" ciphersuites must be in
5203   use. Stitched ciphersuites are optimised implementations of certain
5204   commonly used ciphersuites. Also the application must call SSL_shutdown()
5205   twice even if a protocol error has occurred (applications should not do
5206   this but some do anyway).
5207
5208   This issue was discovered by Juraj Somorovsky, Robert Merget and Nimrod
5209   Aviram, with additional investigation by Steven Collison and Andrew
5210   Hourselt. It was reported to OpenSSL on 10th December 2018.
5211   ([CVE-2019-1559])
5212
5213   *Matt Caswell*
5214
5215 * Move strictness check from EVP_PKEY_asn1_new() to EVP_PKEY_asn1_add0().
5216
5217   *Richard Levitte*
5218
5219### Changes between 1.0.2p and 1.0.2q [20 Nov 2018]
5220
5221 * Microarchitecture timing vulnerability in ECC scalar multiplication
5222
5223   OpenSSL ECC scalar multiplication, used in e.g. ECDSA and ECDH, has been
5224   shown to be vulnerable to a microarchitecture timing side channel attack.
5225   An attacker with sufficient access to mount local timing attacks during
5226   ECDSA signature generation could recover the private key.
5227
5228   This issue was reported to OpenSSL on 26th October 2018 by Alejandro
5229   Cabrera Aldaya, Billy Brumley, Sohaib ul Hassan, Cesar Pereida Garcia and
5230   Nicola Tuveri.
5231   ([CVE-2018-5407])
5232
5233   *Billy Brumley*
5234
5235 * Timing vulnerability in DSA signature generation
5236
5237   The OpenSSL DSA signature algorithm has been shown to be vulnerable to a
5238   timing side channel attack. An attacker could use variations in the signing
5239   algorithm to recover the private key.
5240
5241   This issue was reported to OpenSSL on 16th October 2018 by Samuel Weiser.
5242   ([CVE-2018-0734])
5243
5244   *Paul Dale*
5245
5246 * Resolve a compatibility issue in EC_GROUP handling with the FIPS Object
5247   Module, accidentally introduced while backporting security fixes from the
5248   development branch and hindering the use of ECC in FIPS mode.
5249
5250   *Nicola Tuveri*
5251
5252### Changes between 1.0.2o and 1.0.2p [14 Aug 2018]
5253
5254 * Client DoS due to large DH parameter
5255
5256   During key agreement in a TLS handshake using a DH(E) based ciphersuite a
5257   malicious server can send a very large prime value to the client. This will
5258   cause the client to spend an unreasonably long period of time generating a
5259   key for this prime resulting in a hang until the client has finished. This
5260   could be exploited in a Denial Of Service attack.
5261
5262   This issue was reported to OpenSSL on 5th June 2018 by Guido Vranken
5263   ([CVE-2018-0732])
5264
5265   *Guido Vranken*
5266
5267 * Cache timing vulnerability in RSA Key Generation
5268
5269   The OpenSSL RSA Key generation algorithm has been shown to be vulnerable to
5270   a cache timing side channel attack. An attacker with sufficient access to
5271   mount cache timing attacks during the RSA key generation process could
5272   recover the private key.
5273
5274   This issue was reported to OpenSSL on 4th April 2018 by Alejandro Cabrera
5275   Aldaya, Billy Brumley, Cesar Pereida Garcia and Luis Manuel Alvarez Tapia.
5276   ([CVE-2018-0737])
5277
5278   *Billy Brumley*
5279
5280 * Make EVP_PKEY_asn1_new() a bit stricter about its input.  A NULL pem_str
5281   parameter is no longer accepted, as it leads to a corrupt table.  NULL
5282   pem_str is reserved for alias entries only.
5283
5284   *Richard Levitte*
5285
5286 * Revert blinding in ECDSA sign and instead make problematic addition
5287   length-invariant. Switch even to fixed-length Montgomery multiplication.
5288
5289   *Andy Polyakov*
5290
5291 * Change generating and checking of primes so that the error rate of not
5292   being prime depends on the intended use based on the size of the input.
5293   For larger primes this will result in more rounds of Miller-Rabin.
5294   The maximal error rate for primes with more than 1080 bits is lowered
5295   to 2^-128.
5296
5297   *Kurt Roeckx, Annie Yousar*
5298
5299 * Increase the number of Miller-Rabin rounds for DSA key generating to 64.
5300
5301   *Kurt Roeckx*
5302
5303 * Add blinding to ECDSA and DSA signatures to protect against side channel
5304   attacks discovered by Keegan Ryan (NCC Group).
5305
5306   *Matt Caswell*
5307
5308 * When unlocking a pass phrase protected PEM file or PKCS#8 container, we
5309   now allow empty (zero character) pass phrases.
5310
5311   *Richard Levitte*
5312
5313 * Certificate time validation (X509_cmp_time) enforces stricter
5314   compliance with RFC 5280. Fractional seconds and timezone offsets
5315   are no longer allowed.
5316
5317   *Emilia Käsper*
5318
5319### Changes between 1.0.2n and 1.0.2o [27 Mar 2018]
5320
5321 * Constructed ASN.1 types with a recursive definition could exceed the stack
5322
5323   Constructed ASN.1 types with a recursive definition (such as can be found
5324   in PKCS7) could eventually exceed the stack given malicious input with
5325   excessive recursion. This could result in a Denial Of Service attack. There
5326   are no such structures used within SSL/TLS that come from untrusted sources
5327   so this is considered safe.
5328
5329   This issue was reported to OpenSSL on 4th January 2018 by the OSS-fuzz
5330   project.
5331   ([CVE-2018-0739])
5332
5333   *Matt Caswell*
5334
5335### Changes between 1.0.2m and 1.0.2n [7 Dec 2017]
5336
5337 * Read/write after SSL object in error state
5338
5339   OpenSSL 1.0.2 (starting from version 1.0.2b) introduced an "error state"
5340   mechanism. The intent was that if a fatal error occurred during a handshake
5341   then OpenSSL would move into the error state and would immediately fail if
5342   you attempted to continue the handshake. This works as designed for the
5343   explicit handshake functions (SSL_do_handshake(), SSL_accept() and
5344   SSL_connect()), however due to a bug it does not work correctly if
5345   SSL_read() or SSL_write() is called directly. In that scenario, if the
5346   handshake fails then a fatal error will be returned in the initial function
5347   call. If SSL_read()/SSL_write() is subsequently called by the application
5348   for the same SSL object then it will succeed and the data is passed without
5349   being decrypted/encrypted directly from the SSL/TLS record layer.
5350
5351   In order to exploit this issue an application bug would have to be present
5352   that resulted in a call to SSL_read()/SSL_write() being issued after having
5353   already received a fatal error.
5354
5355   This issue was reported to OpenSSL by David Benjamin (Google).
5356   ([CVE-2017-3737])
5357
5358   *Matt Caswell*
5359
5360 * rsaz_1024_mul_avx2 overflow bug on x86_64
5361
5362   There is an overflow bug in the AVX2 Montgomery multiplication procedure
5363   used in exponentiation with 1024-bit moduli. No EC algorithms are affected.
5364   Analysis suggests that attacks against RSA and DSA as a result of this
5365   defect would be very difficult to perform and are not believed likely.
5366   Attacks against DH1024 are considered just feasible, because most of the
5367   work necessary to deduce information about a private key may be performed
5368   offline. The amount of resources required for such an attack would be
5369   significant. However, for an attack on TLS to be meaningful, the server
5370   would have to share the DH1024 private key among multiple clients, which is
5371   no longer an option since CVE-2016-0701.
5372
5373   This only affects processors that support the AVX2 but not ADX extensions
5374   like Intel Haswell (4th generation).
5375
5376   This issue was reported to OpenSSL by David Benjamin (Google). The issue
5377   was originally found via the OSS-Fuzz project.
5378   ([CVE-2017-3738])
5379
5380   *Andy Polyakov*
5381
5382### Changes between 1.0.2l and 1.0.2m [2 Nov 2017]
5383
5384 * bn_sqrx8x_internal carry bug on x86_64
5385
5386   There is a carry propagating bug in the x86_64 Montgomery squaring
5387   procedure. No EC algorithms are affected. Analysis suggests that attacks
5388   against RSA and DSA as a result of this defect would be very difficult to
5389   perform and are not believed likely. Attacks against DH are considered just
5390   feasible (although very difficult) because most of the work necessary to
5391   deduce information about a private key may be performed offline. The amount
5392   of resources required for such an attack would be very significant and
5393   likely only accessible to a limited number of attackers. An attacker would
5394   additionally need online access to an unpatched system using the target
5395   private key in a scenario with persistent DH parameters and a private
5396   key that is shared between multiple clients.
5397
5398   This only affects processors that support the BMI1, BMI2 and ADX extensions
5399   like Intel Broadwell (5th generation) and later or AMD Ryzen.
5400
5401   This issue was reported to OpenSSL by the OSS-Fuzz project.
5402   ([CVE-2017-3736])
5403
5404   *Andy Polyakov*
5405
5406 * Malformed X.509 IPAddressFamily could cause OOB read
5407
5408   If an X.509 certificate has a malformed IPAddressFamily extension,
5409   OpenSSL could do a one-byte buffer overread. The most likely result
5410   would be an erroneous display of the certificate in text format.
5411
5412   This issue was reported to OpenSSL by the OSS-Fuzz project.
5413
5414   *Rich Salz*
5415
5416### Changes between 1.0.2k and 1.0.2l [25 May 2017]
5417
5418 * Have 'config' recognise 64-bit mingw and choose 'mingw64' as the target
5419   platform rather than 'mingw'.
5420
5421   *Richard Levitte*
5422
5423### Changes between 1.0.2j and 1.0.2k [26 Jan 2017]
5424
5425 * Truncated packet could crash via OOB read
5426
5427   If one side of an SSL/TLS path is running on a 32-bit host and a specific
5428   cipher is being used, then a truncated packet can cause that host to
5429   perform an out-of-bounds read, usually resulting in a crash.
5430
5431   This issue was reported to OpenSSL by Robert Święcki of Google.
5432   ([CVE-2017-3731])
5433
5434   *Andy Polyakov*
5435
5436 * BN_mod_exp may produce incorrect results on x86_64
5437
5438   There is a carry propagating bug in the x86_64 Montgomery squaring
5439   procedure. No EC algorithms are affected. Analysis suggests that attacks
5440   against RSA and DSA as a result of this defect would be very difficult to
5441   perform and are not believed likely. Attacks against DH are considered just
5442   feasible (although very difficult) because most of the work necessary to
5443   deduce information about a private key may be performed offline. The amount
5444   of resources required for such an attack would be very significant and
5445   likely only accessible to a limited number of attackers. An attacker would
5446   additionally need online access to an unpatched system using the target
5447   private key in a scenario with persistent DH parameters and a private
5448   key that is shared between multiple clients. For example this can occur by
5449   default in OpenSSL DHE based SSL/TLS ciphersuites. Note: This issue is very
5450   similar to CVE-2015-3193 but must be treated as a separate problem.
5451
5452   This issue was reported to OpenSSL by the OSS-Fuzz project.
5453   ([CVE-2017-3732])
5454
5455   *Andy Polyakov*
5456
5457 * Montgomery multiplication may produce incorrect results
5458
5459   There is a carry propagating bug in the Broadwell-specific Montgomery
5460   multiplication procedure that handles input lengths divisible by, but
5461   longer than 256 bits. Analysis suggests that attacks against RSA, DSA
5462   and DH private keys are impossible. This is because the subroutine in
5463   question is not used in operations with the private key itself and an input
5464   of the attacker's direct choice. Otherwise the bug can manifest itself as
5465   transient authentication and key negotiation failures or reproducible
5466   erroneous outcome of public-key operations with specially crafted input.
5467   Among EC algorithms only Brainpool P-512 curves are affected and one
5468   presumably can attack ECDH key negotiation. Impact was not analyzed in
5469   detail, because pre-requisites for attack are considered unlikely. Namely
5470   multiple clients have to choose the curve in question and the server has to
5471   share the private key among them, neither of which is default behaviour.
5472   Even then only clients that chose the curve will be affected.
5473
5474   This issue was publicly reported as transient failures and was not
5475   initially recognized as a security issue. Thanks to Richard Morgan for
5476   providing reproducible case.
5477   ([CVE-2016-7055])
5478
5479   *Andy Polyakov*
5480
5481 * OpenSSL now fails if it receives an unrecognised record type in TLS1.0
5482   or TLS1.1. Previously this only happened in SSLv3 and TLS1.2. This is to
5483   prevent issues where no progress is being made and the peer continually
5484   sends unrecognised record types, using up resources processing them.
5485
5486   *Matt Caswell*
5487
5488### Changes between 1.0.2i and 1.0.2j [26 Sep 2016]
5489
5490 * Missing CRL sanity check
5491
5492   A bug fix which included a CRL sanity check was added to OpenSSL 1.1.0
5493   but was omitted from OpenSSL 1.0.2i. As a result any attempt to use
5494   CRLs in OpenSSL 1.0.2i will crash with a null pointer exception.
5495
5496   This issue only affects the OpenSSL 1.0.2i
5497   ([CVE-2016-7052])
5498
5499   *Matt Caswell*
5500
5501### Changes between 1.0.2h and 1.0.2i [22 Sep 2016]
5502
5503 * OCSP Status Request extension unbounded memory growth
5504
5505   A malicious client can send an excessively large OCSP Status Request
5506   extension. If that client continually requests renegotiation, sending a
5507   large OCSP Status Request extension each time, then there will be unbounded
5508   memory growth on the server. This will eventually lead to a Denial Of
5509   Service attack through memory exhaustion. Servers with a default
5510   configuration are vulnerable even if they do not support OCSP. Builds using
5511   the "no-ocsp" build time option are not affected.
5512
5513   This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.)
5514   ([CVE-2016-6304])
5515
5516   *Matt Caswell*
5517
5518 * In order to mitigate the SWEET32 attack, the DES ciphers were moved from
5519   HIGH to MEDIUM.
5520
5521   This issue was reported to OpenSSL Karthikeyan Bhargavan and Gaetan
5522   Leurent (INRIA)
5523   ([CVE-2016-2183])
5524
5525   *Rich Salz*
5526
5527 * OOB write in MDC2_Update()
5528
5529   An overflow can occur in MDC2_Update() either if called directly or
5530   through the EVP_DigestUpdate() function using MDC2. If an attacker
5531   is able to supply very large amounts of input data after a previous
5532   call to EVP_EncryptUpdate() with a partial block then a length check
5533   can overflow resulting in a heap corruption.
5534
5535   The amount of data needed is comparable to SIZE_MAX which is impractical
5536   on most platforms.
5537
5538   This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.)
5539   ([CVE-2016-6303])
5540
5541   *Stephen Henson*
5542
5543 * Malformed SHA512 ticket DoS
5544
5545   If a server uses SHA512 for TLS session ticket HMAC it is vulnerable to a
5546   DoS attack where a malformed ticket will result in an OOB read which will
5547   ultimately crash.
5548
5549   The use of SHA512 in TLS session tickets is comparatively rare as it requires
5550   a custom server callback and ticket lookup mechanism.
5551
5552   This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.)
5553   ([CVE-2016-6302])
5554
5555   *Stephen Henson*
5556
5557 * OOB write in BN_bn2dec()
5558
5559   The function BN_bn2dec() does not check the return value of BN_div_word().
5560   This can cause an OOB write if an application uses this function with an
5561   overly large BIGNUM. This could be a problem if an overly large certificate
5562   or CRL is printed out from an untrusted source. TLS is not affected because
5563   record limits will reject an oversized certificate before it is parsed.
5564
5565   This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.)
5566   ([CVE-2016-2182])
5567
5568   *Stephen Henson*
5569
5570 * OOB read in TS_OBJ_print_bio()
5571
5572   The function TS_OBJ_print_bio() misuses OBJ_obj2txt(): the return value is
5573   the total length the OID text representation would use and not the amount
5574   of data written. This will result in OOB reads when large OIDs are
5575   presented.
5576
5577   This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.)
5578   ([CVE-2016-2180])
5579
5580   *Stephen Henson*
5581
5582 * Pointer arithmetic undefined behaviour
5583
5584   Avoid some undefined pointer arithmetic
5585
5586   A common idiom in the codebase is to check limits in the following manner:
5587   "p + len > limit"
5588
5589   Where "p" points to some malloc'd data of SIZE bytes and
5590   limit == p + SIZE
5591
5592   "len" here could be from some externally supplied data (e.g. from a TLS
5593   message).
5594
5595   The rules of C pointer arithmetic are such that "p + len" is only well
5596   defined where len <= SIZE. Therefore the above idiom is actually
5597   undefined behaviour.
5598
5599   For example this could cause problems if some malloc implementation
5600   provides an address for "p" such that "p + len" actually overflows for
5601   values of len that are too big and therefore p + len < limit.
5602
5603   This issue was reported to OpenSSL by Guido Vranken
5604   ([CVE-2016-2177])
5605
5606   *Matt Caswell*
5607
5608 * Constant time flag not preserved in DSA signing
5609
5610   Operations in the DSA signing algorithm should run in constant time in
5611   order to avoid side channel attacks. A flaw in the OpenSSL DSA
5612   implementation means that a non-constant time codepath is followed for
5613   certain operations. This has been demonstrated through a cache-timing
5614   attack to be sufficient for an attacker to recover the private DSA key.
5615
5616   This issue was reported by César Pereida (Aalto University), Billy Brumley
5617   (Tampere University of Technology), and Yuval Yarom (The University of
5618   Adelaide and NICTA).
5619   ([CVE-2016-2178])
5620
5621   *César Pereida*
5622
5623 * DTLS buffered message DoS
5624
5625   In a DTLS connection where handshake messages are delivered out-of-order
5626   those messages that OpenSSL is not yet ready to process will be buffered
5627   for later use. Under certain circumstances, a flaw in the logic means that
5628   those messages do not get removed from the buffer even though the handshake
5629   has been completed. An attacker could force up to approx. 15 messages to
5630   remain in the buffer when they are no longer required. These messages will
5631   be cleared when the DTLS connection is closed. The default maximum size for
5632   a message is 100k. Therefore the attacker could force an additional 1500k
5633   to be consumed per connection. By opening many simulataneous connections an
5634   attacker could cause a DoS attack through memory exhaustion.
5635
5636   This issue was reported to OpenSSL by Quan Luo.
5637   ([CVE-2016-2179])
5638
5639   *Matt Caswell*
5640
5641 * DTLS replay protection DoS
5642
5643   A flaw in the DTLS replay attack protection mechanism means that records
5644   that arrive for future epochs update the replay protection "window" before
5645   the MAC for the record has been validated. This could be exploited by an
5646   attacker by sending a record for the next epoch (which does not have to
5647   decrypt or have a valid MAC), with a very large sequence number. This means
5648   that all subsequent legitimate packets are dropped causing a denial of
5649   service for a specific DTLS connection.
5650
5651   This issue was reported to OpenSSL by the OCAP audit team.
5652   ([CVE-2016-2181])
5653
5654   *Matt Caswell*
5655
5656 * Certificate message OOB reads
5657
5658   In OpenSSL 1.0.2 and earlier some missing message length checks can result
5659   in OOB reads of up to 2 bytes beyond an allocated buffer. There is a
5660   theoretical DoS risk but this has not been observed in practice on common
5661   platforms.
5662
5663   The messages affected are client certificate, client certificate request
5664   and server certificate. As a result the attack can only be performed
5665   against a client or a server which enables client authentication.
5666
5667   This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.)
5668   ([CVE-2016-6306])
5669
5670   *Stephen Henson*
5671
5672### Changes between 1.0.2g and 1.0.2h [3 May 2016]
5673
5674 * Prevent padding oracle in AES-NI CBC MAC check
5675
5676   A MITM attacker can use a padding oracle attack to decrypt traffic
5677   when the connection uses an AES CBC cipher and the server support
5678   AES-NI.
5679
5680   This issue was introduced as part of the fix for Lucky 13 padding
5681   attack ([CVE-2013-0169]). The padding check was rewritten to be in
5682   constant time by making sure that always the same bytes are read and
5683   compared against either the MAC or padding bytes. But it no longer
5684   checked that there was enough data to have both the MAC and padding
5685   bytes.
5686
5687   This issue was reported by Juraj Somorovsky using TLS-Attacker.
5688
5689   *Kurt Roeckx*
5690
5691 * Fix EVP_EncodeUpdate overflow
5692
5693   An overflow can occur in the EVP_EncodeUpdate() function which is used for
5694   Base64 encoding of binary data. If an attacker is able to supply very large
5695   amounts of input data then a length check can overflow resulting in a heap
5696   corruption.
5697
5698   Internally to OpenSSL the EVP_EncodeUpdate() function is primarily used by
5699   the `PEM_write_bio*` family of functions. These are mainly used within the
5700   OpenSSL command line applications, so any application which processes data
5701   from an untrusted source and outputs it as a PEM file should be considered
5702   vulnerable to this issue. User applications that call these APIs directly
5703   with large amounts of untrusted data may also be vulnerable.
5704
5705   This issue was reported by Guido Vranken.
5706   ([CVE-2016-2105])
5707
5708   *Matt Caswell*
5709
5710 * Fix EVP_EncryptUpdate overflow
5711
5712   An overflow can occur in the EVP_EncryptUpdate() function. If an attacker
5713   is able to supply very large amounts of input data after a previous call to
5714   EVP_EncryptUpdate() with a partial block then a length check can overflow
5715   resulting in a heap corruption. Following an analysis of all OpenSSL
5716   internal usage of the EVP_EncryptUpdate() function all usage is one of two
5717   forms. The first form is where the EVP_EncryptUpdate() call is known to be
5718   the first called function after an EVP_EncryptInit(), and therefore that
5719   specific call must be safe. The second form is where the length passed to
5720   EVP_EncryptUpdate() can be seen from the code to be some small value and
5721   therefore there is no possibility of an overflow. Since all instances are
5722   one of these two forms, it is believed that there can be no overflows in
5723   internal code due to this problem. It should be noted that
5724   EVP_DecryptUpdate() can call EVP_EncryptUpdate() in certain code paths.
5725   Also EVP_CipherUpdate() is a synonym for EVP_EncryptUpdate(). All instances
5726   of these calls have also been analysed too and it is believed there are no
5727   instances in internal usage where an overflow could occur.
5728
5729   This issue was reported by Guido Vranken.
5730   ([CVE-2016-2106])
5731
5732   *Matt Caswell*
5733
5734 * Prevent ASN.1 BIO excessive memory allocation
5735
5736   When ASN.1 data is read from a BIO using functions such as d2i_CMS_bio()
5737   a short invalid encoding can cause allocation of large amounts of memory
5738   potentially consuming excessive resources or exhausting memory.
5739
5740   Any application parsing untrusted data through d2i BIO functions is
5741   affected. The memory based functions such as d2i_X509() are *not* affected.
5742   Since the memory based functions are used by the TLS library, TLS
5743   applications are not affected.
5744
5745   This issue was reported by Brian Carpenter.
5746   ([CVE-2016-2109])
5747
5748   *Stephen Henson*
5749
5750 * EBCDIC overread
5751
5752   ASN1 Strings that are over 1024 bytes can cause an overread in applications
5753   using the X509_NAME_oneline() function on EBCDIC systems. This could result
5754   in arbitrary stack data being returned in the buffer.
5755
5756   This issue was reported by Guido Vranken.
5757   ([CVE-2016-2176])
5758
5759   *Matt Caswell*
5760
5761 * Modify behavior of ALPN to invoke callback after SNI/servername
5762   callback, such that updates to the SSL_CTX affect ALPN.
5763
5764   *Todd Short*
5765
5766 * Remove LOW from the DEFAULT cipher list.  This removes singles DES from the
5767   default.
5768
5769   *Kurt Roeckx*
5770
5771 * Only remove the SSLv2 methods with the no-ssl2-method option. When the
5772   methods are enabled and ssl2 is disabled the methods return NULL.
5773
5774   *Kurt Roeckx*
5775
5776### Changes between 1.0.2f and 1.0.2g [1 Mar 2016]
5777
5778* Disable weak ciphers in SSLv3 and up in default builds of OpenSSL.
5779  Builds that are not configured with "enable-weak-ssl-ciphers" will not
5780  provide any "EXPORT" or "LOW" strength ciphers.
5781
5782  *Viktor Dukhovni*
5783
5784* Disable SSLv2 default build, default negotiation and weak ciphers.  SSLv2
5785  is by default disabled at build-time.  Builds that are not configured with
5786  "enable-ssl2" will not support SSLv2.  Even if "enable-ssl2" is used,
5787  users who want to negotiate SSLv2 via the version-flexible SSLv23_method()
5788  will need to explicitly call either of:
5789
5790      SSL_CTX_clear_options(ctx, SSL_OP_NO_SSLv2);
5791  or
5792      SSL_clear_options(ssl, SSL_OP_NO_SSLv2);
5793
5794  as appropriate.  Even if either of those is used, or the application
5795  explicitly uses the version-specific SSLv2_method() or its client and
5796  server variants, SSLv2 ciphers vulnerable to exhaustive search key
5797  recovery have been removed.  Specifically, the SSLv2 40-bit EXPORT
5798  ciphers, and SSLv2 56-bit DES are no longer available.
5799  ([CVE-2016-0800])
5800
5801   *Viktor Dukhovni*
5802
5803 * Fix a double-free in DSA code
5804
5805   A double free bug was discovered when OpenSSL parses malformed DSA private
5806   keys and could lead to a DoS attack or memory corruption for applications
5807   that receive DSA private keys from untrusted sources.  This scenario is
5808   considered rare.
5809
5810   This issue was reported to OpenSSL by Adam Langley(Google/BoringSSL) using
5811   libFuzzer.
5812   ([CVE-2016-0705])
5813
5814   *Stephen Henson*
5815
5816 * Disable SRP fake user seed to address a server memory leak.
5817
5818   Add a new method SRP_VBASE_get1_by_user that handles the seed properly.
5819
5820   SRP_VBASE_get_by_user had inconsistent memory management behaviour.
5821   In order to fix an unavoidable memory leak, SRP_VBASE_get_by_user
5822   was changed to ignore the "fake user" SRP seed, even if the seed
5823   is configured.
5824
5825   Users should use SRP_VBASE_get1_by_user instead. Note that in
5826   SRP_VBASE_get1_by_user, caller must free the returned value. Note
5827   also that even though configuring the SRP seed attempts to hide
5828   invalid usernames by continuing the handshake with fake
5829   credentials, this behaviour is not constant time and no strong
5830   guarantees are made that the handshake is indistinguishable from
5831   that of a valid user.
5832   ([CVE-2016-0798])
5833
5834   *Emilia Käsper*
5835
5836 * Fix BN_hex2bn/BN_dec2bn NULL pointer deref/heap corruption
5837
5838   In the BN_hex2bn function the number of hex digits is calculated using an
5839   int value `i`. Later `bn_expand` is called with a value of `i * 4`. For
5840   large values of `i` this can result in `bn_expand` not allocating any
5841   memory because `i * 4` is negative. This can leave the internal BIGNUM data
5842   field as NULL leading to a subsequent NULL ptr deref. For very large values
5843   of `i`, the calculation `i * 4` could be a positive value smaller than `i`.
5844   In this case memory is allocated to the internal BIGNUM data field, but it
5845   is insufficiently sized leading to heap corruption. A similar issue exists
5846   in BN_dec2bn. This could have security consequences if BN_hex2bn/BN_dec2bn
5847   is ever called by user applications with very large untrusted hex/dec data.
5848   This is anticipated to be a rare occurrence.
5849
5850   All OpenSSL internal usage of these functions use data that is not expected
5851   to be untrusted, e.g. config file data or application command line
5852   arguments. If user developed applications generate config file data based
5853   on untrusted data then it is possible that this could also lead to security
5854   consequences. This is also anticipated to be rare.
5855
5856   This issue was reported to OpenSSL by Guido Vranken.
5857   ([CVE-2016-0797])
5858
5859   *Matt Caswell*
5860
5861 * Fix memory issues in `BIO_*printf` functions
5862
5863   The internal `fmtstr` function used in processing a "%s" format string in
5864   the `BIO_*printf` functions could overflow while calculating the length of a
5865   string and cause an OOB read when printing very long strings.
5866
5867   Additionally the internal `doapr_outch` function can attempt to write to an
5868   OOB memory location (at an offset from the NULL pointer) in the event of a
5869   memory allocation failure. In 1.0.2 and below this could be caused where
5870   the size of a buffer to be allocated is greater than INT_MAX. E.g. this
5871   could be in processing a very long "%s" format string. Memory leaks can
5872   also occur.
5873
5874   The first issue may mask the second issue dependent on compiler behaviour.
5875   These problems could enable attacks where large amounts of untrusted data
5876   is passed to the `BIO_*printf` functions. If applications use these functions
5877   in this way then they could be vulnerable. OpenSSL itself uses these
5878   functions when printing out human-readable dumps of ASN.1 data. Therefore
5879   applications that print this data could be vulnerable if the data is from
5880   untrusted sources. OpenSSL command line applications could also be
5881   vulnerable where they print out ASN.1 data, or if untrusted data is passed
5882   as command line arguments.
5883
5884   Libssl is not considered directly vulnerable. Additionally certificates etc
5885   received via remote connections via libssl are also unlikely to be able to
5886   trigger these issues because of message size limits enforced within libssl.
5887
5888   This issue was reported to OpenSSL Guido Vranken.
5889   ([CVE-2016-0799])
5890
5891   *Matt Caswell*
5892
5893 * Side channel attack on modular exponentiation
5894
5895   A side-channel attack was found which makes use of cache-bank conflicts on
5896   the Intel Sandy-Bridge microarchitecture which could lead to the recovery
5897   of RSA keys.  The ability to exploit this issue is limited as it relies on
5898   an attacker who has control of code in a thread running on the same
5899   hyper-threaded core as the victim thread which is performing decryptions.
5900
5901   This issue was reported to OpenSSL by Yuval Yarom, The University of
5902   Adelaide and NICTA, Daniel Genkin, Technion and Tel Aviv University, and
5903   Nadia Heninger, University of Pennsylvania with more information at
5904   <http://cachebleed.info>.
5905   ([CVE-2016-0702])
5906
5907   *Andy Polyakov*
5908
5909 * Change the `req` command to generate a 2048-bit RSA/DSA key by default,
5910   if no keysize is specified with default_bits. This fixes an
5911   omission in an earlier change that changed all RSA/DSA key generation
5912   commands to use 2048 bits by default.
5913
5914   *Emilia Käsper*
5915
5916### Changes between 1.0.2e and 1.0.2f [28 Jan 2016]
5917
5918 * DH small subgroups
5919
5920   Historically OpenSSL only ever generated DH parameters based on "safe"
5921   primes. More recently (in version 1.0.2) support was provided for
5922   generating X9.42 style parameter files such as those required for RFC 5114
5923   support. The primes used in such files may not be "safe". Where an
5924   application is using DH configured with parameters based on primes that are
5925   not "safe" then an attacker could use this fact to find a peer's private
5926   DH exponent. This attack requires that the attacker complete multiple
5927   handshakes in which the peer uses the same private DH exponent. For example
5928   this could be used to discover a TLS server's private DH exponent if it's
5929   reusing the private DH exponent or it's using a static DH ciphersuite.
5930
5931   OpenSSL provides the option SSL_OP_SINGLE_DH_USE for ephemeral DH (DHE) in
5932   TLS. It is not on by default. If the option is not set then the server
5933   reuses the same private DH exponent for the life of the server process and
5934   would be vulnerable to this attack. It is believed that many popular
5935   applications do set this option and would therefore not be at risk.
5936
5937   The fix for this issue adds an additional check where a "q" parameter is
5938   available (as is the case in X9.42 based parameters). This detects the
5939   only known attack, and is the only possible defense for static DH
5940   ciphersuites. This could have some performance impact.
5941
5942   Additionally the SSL_OP_SINGLE_DH_USE option has been switched on by
5943   default and cannot be disabled. This could have some performance impact.
5944
5945   This issue was reported to OpenSSL by Antonio Sanso (Adobe).
5946   ([CVE-2016-0701])
5947
5948   *Matt Caswell*
5949
5950 * SSLv2 doesn't block disabled ciphers
5951
5952   A malicious client can negotiate SSLv2 ciphers that have been disabled on
5953   the server and complete SSLv2 handshakes even if all SSLv2 ciphers have
5954   been disabled, provided that the SSLv2 protocol was not also disabled via
5955   SSL_OP_NO_SSLv2.
5956
5957   This issue was reported to OpenSSL on 26th December 2015 by Nimrod Aviram
5958   and Sebastian Schinzel.
5959   ([CVE-2015-3197])
5960
5961   *Viktor Dukhovni*
5962
5963### Changes between 1.0.2d and 1.0.2e [3 Dec 2015]
5964
5965 * BN_mod_exp may produce incorrect results on x86_64
5966
5967   There is a carry propagating bug in the x86_64 Montgomery squaring
5968   procedure. No EC algorithms are affected. Analysis suggests that attacks
5969   against RSA and DSA as a result of this defect would be very difficult to
5970   perform and are not believed likely. Attacks against DH are considered just
5971   feasible (although very difficult) because most of the work necessary to
5972   deduce information about a private key may be performed offline. The amount
5973   of resources required for such an attack would be very significant and
5974   likely only accessible to a limited number of attackers. An attacker would
5975   additionally need online access to an unpatched system using the target
5976   private key in a scenario with persistent DH parameters and a private
5977   key that is shared between multiple clients. For example this can occur by
5978   default in OpenSSL DHE based SSL/TLS ciphersuites.
5979
5980   This issue was reported to OpenSSL by Hanno Böck.
5981   ([CVE-2015-3193])
5982
5983   *Andy Polyakov*
5984
5985 * Certificate verify crash with missing PSS parameter
5986
5987   The signature verification routines will crash with a NULL pointer
5988   dereference if presented with an ASN.1 signature using the RSA PSS
5989   algorithm and absent mask generation function parameter. Since these
5990   routines are used to verify certificate signature algorithms this can be
5991   used to crash any certificate verification operation and exploited in a
5992   DoS attack. Any application which performs certificate verification is
5993   vulnerable including OpenSSL clients and servers which enable client
5994   authentication.
5995
5996   This issue was reported to OpenSSL by Loïc Jonas Etienne (Qnective AG).
5997   ([CVE-2015-3194])
5998
5999   *Stephen Henson*
6000
6001 * X509_ATTRIBUTE memory leak
6002
6003   When presented with a malformed X509_ATTRIBUTE structure OpenSSL will leak
6004   memory. This structure is used by the PKCS#7 and CMS routines so any
6005   application which reads PKCS#7 or CMS data from untrusted sources is
6006   affected. SSL/TLS is not affected.
6007
6008   This issue was reported to OpenSSL by Adam Langley (Google/BoringSSL) using
6009   libFuzzer.
6010   ([CVE-2015-3195])
6011
6012   *Stephen Henson*
6013
6014 * Rewrite EVP_DecodeUpdate (base64 decoding) to fix several bugs.
6015   This changes the decoding behaviour for some invalid messages,
6016   though the change is mostly in the more lenient direction, and
6017   legacy behaviour is preserved as much as possible.
6018
6019   *Emilia Käsper*
6020
6021 * In DSA_generate_parameters_ex, if the provided seed is too short,
6022   return an error
6023
6024   *Rich Salz and Ismo Puustinen <ismo.puustinen@intel.com>*
6025
6026### Changes between 1.0.2c and 1.0.2d [9 Jul 2015]
6027
6028 * Alternate chains certificate forgery
6029
6030   During certificate verification, OpenSSL will attempt to find an
6031   alternative certificate chain if the first attempt to build such a chain
6032   fails. An error in the implementation of this logic can mean that an
6033   attacker could cause certain checks on untrusted certificates to be
6034   bypassed, such as the CA flag, enabling them to use a valid leaf
6035   certificate to act as a CA and "issue" an invalid certificate.
6036
6037   This issue was reported to OpenSSL by Adam Langley/David Benjamin
6038   (Google/BoringSSL).
6039
6040   *Matt Caswell*
6041
6042### Changes between 1.0.2b and 1.0.2c [12 Jun 2015]
6043
6044 * Fix HMAC ABI incompatibility. The previous version introduced an ABI
6045   incompatibility in the handling of HMAC. The previous ABI has now been
6046   restored.
6047
6048   *Matt Caswell*
6049
6050### Changes between 1.0.2a and 1.0.2b [11 Jun 2015]
6051
6052 * Malformed ECParameters causes infinite loop
6053
6054   When processing an ECParameters structure OpenSSL enters an infinite loop
6055   if the curve specified is over a specially malformed binary polynomial
6056   field.
6057
6058   This can be used to perform denial of service against any
6059   system which processes public keys, certificate requests or
6060   certificates.  This includes TLS clients and TLS servers with
6061   client authentication enabled.
6062
6063   This issue was reported to OpenSSL by Joseph Barr-Pixton.
6064   ([CVE-2015-1788])
6065
6066   *Andy Polyakov*
6067
6068 * Exploitable out-of-bounds read in X509_cmp_time
6069
6070   X509_cmp_time does not properly check the length of the ASN1_TIME
6071   string and can read a few bytes out of bounds. In addition,
6072   X509_cmp_time accepts an arbitrary number of fractional seconds in the
6073   time string.
6074
6075   An attacker can use this to craft malformed certificates and CRLs of
6076   various sizes and potentially cause a segmentation fault, resulting in
6077   a DoS on applications that verify certificates or CRLs. TLS clients
6078   that verify CRLs are affected. TLS clients and servers with client
6079   authentication enabled may be affected if they use custom verification
6080   callbacks.
6081
6082   This issue was reported to OpenSSL by Robert Swiecki (Google), and
6083   independently by Hanno Böck.
6084   ([CVE-2015-1789])
6085
6086   *Emilia Käsper*
6087
6088 * PKCS7 crash with missing EnvelopedContent
6089
6090   The PKCS#7 parsing code does not handle missing inner EncryptedContent
6091   correctly. An attacker can craft malformed ASN.1-encoded PKCS#7 blobs
6092   with missing content and trigger a NULL pointer dereference on parsing.
6093
6094   Applications that decrypt PKCS#7 data or otherwise parse PKCS#7
6095   structures from untrusted sources are affected. OpenSSL clients and
6096   servers are not affected.
6097
6098   This issue was reported to OpenSSL by Michal Zalewski (Google).
6099   ([CVE-2015-1790])
6100
6101   *Emilia Käsper*
6102
6103 * CMS verify infinite loop with unknown hash function
6104
6105   When verifying a signedData message the CMS code can enter an infinite loop
6106   if presented with an unknown hash function OID. This can be used to perform
6107   denial of service against any system which verifies signedData messages using
6108   the CMS code.
6109   This issue was reported to OpenSSL by Johannes Bauer.
6110   ([CVE-2015-1792])
6111
6112   *Stephen Henson*
6113
6114 * Race condition handling NewSessionTicket
6115
6116   If a NewSessionTicket is received by a multi-threaded client when attempting to
6117   reuse a previous ticket then a race condition can occur potentially leading to
6118   a double free of the ticket data.
6119   ([CVE-2015-1791])
6120
6121   *Matt Caswell*
6122
6123 * Only support 256-bit or stronger elliptic curves with the
6124   'ecdh_auto' setting (server) or by default (client). Of supported
6125   curves, prefer P-256 (both).
6126
6127   *Emilia Kasper*
6128
6129### Changes between 1.0.2 and 1.0.2a [19 Mar 2015]
6130
6131 * ClientHello sigalgs DoS fix
6132
6133   If a client connects to an OpenSSL 1.0.2 server and renegotiates with an
6134   invalid signature algorithms extension a NULL pointer dereference will
6135   occur. This can be exploited in a DoS attack against the server.
6136
6137   This issue was was reported to OpenSSL by David Ramos of Stanford
6138   University.
6139   ([CVE-2015-0291])
6140
6141   *Stephen Henson and Matt Caswell*
6142
6143 * Multiblock corrupted pointer fix
6144
6145   OpenSSL 1.0.2 introduced the "multiblock" performance improvement. This
6146   feature only applies on 64 bit x86 architecture platforms that support AES
6147   NI instructions. A defect in the implementation of "multiblock" can cause
6148   OpenSSL's internal write buffer to become incorrectly set to NULL when
6149   using non-blocking IO. Typically, when the user application is using a
6150   socket BIO for writing, this will only result in a failed connection.
6151   However if some other BIO is used then it is likely that a segmentation
6152   fault will be triggered, thus enabling a potential DoS attack.
6153
6154   This issue was reported to OpenSSL by Daniel Danner and Rainer Mueller.
6155   ([CVE-2015-0290])
6156
6157   *Matt Caswell*
6158
6159 * Segmentation fault in DTLSv1_listen fix
6160
6161   The DTLSv1_listen function is intended to be stateless and processes the
6162   initial ClientHello from many peers. It is common for user code to loop
6163   over the call to DTLSv1_listen until a valid ClientHello is received with
6164   an associated cookie. A defect in the implementation of DTLSv1_listen means
6165   that state is preserved in the SSL object from one invocation to the next
6166   that can lead to a segmentation fault. Errors processing the initial
6167   ClientHello can trigger this scenario. An example of such an error could be
6168   that a DTLS1.0 only client is attempting to connect to a DTLS1.2 only
6169   server.
6170
6171   This issue was reported to OpenSSL by Per Allansson.
6172   ([CVE-2015-0207])
6173
6174   *Matt Caswell*
6175
6176 * Segmentation fault in ASN1_TYPE_cmp fix
6177
6178   The function ASN1_TYPE_cmp will crash with an invalid read if an attempt is
6179   made to compare ASN.1 boolean types. Since ASN1_TYPE_cmp is used to check
6180   certificate signature algorithm consistency this can be used to crash any
6181   certificate verification operation and exploited in a DoS attack. Any
6182   application which performs certificate verification is vulnerable including
6183   OpenSSL clients and servers which enable client authentication.
6184   ([CVE-2015-0286])
6185
6186   *Stephen Henson*
6187
6188 * Segmentation fault for invalid PSS parameters fix
6189
6190   The signature verification routines will crash with a NULL pointer
6191   dereference if presented with an ASN.1 signature using the RSA PSS
6192   algorithm and invalid parameters. Since these routines are used to verify
6193   certificate signature algorithms this can be used to crash any
6194   certificate verification operation and exploited in a DoS attack. Any
6195   application which performs certificate verification is vulnerable including
6196   OpenSSL clients and servers which enable client authentication.
6197
6198   This issue was was reported to OpenSSL by Brian Carpenter.
6199   ([CVE-2015-0208])
6200
6201   *Stephen Henson*
6202
6203 * ASN.1 structure reuse memory corruption fix
6204
6205   Reusing a structure in ASN.1 parsing may allow an attacker to cause
6206   memory corruption via an invalid write. Such reuse is and has been
6207   strongly discouraged and is believed to be rare.
6208
6209   Applications that parse structures containing CHOICE or ANY DEFINED BY
6210   components may be affected. Certificate parsing (d2i_X509 and related
6211   functions) are however not affected. OpenSSL clients and servers are
6212   not affected.
6213   ([CVE-2015-0287])
6214
6215   *Stephen Henson*
6216
6217 * PKCS7 NULL pointer dereferences fix
6218
6219   The PKCS#7 parsing code does not handle missing outer ContentInfo
6220   correctly. An attacker can craft malformed ASN.1-encoded PKCS#7 blobs with
6221   missing content and trigger a NULL pointer dereference on parsing.
6222
6223   Applications that verify PKCS#7 signatures, decrypt PKCS#7 data or
6224   otherwise parse PKCS#7 structures from untrusted sources are
6225   affected. OpenSSL clients and servers are not affected.
6226
6227   This issue was reported to OpenSSL by Michal Zalewski (Google).
6228   ([CVE-2015-0289])
6229
6230   *Emilia Käsper*
6231
6232 * DoS via reachable assert in SSLv2 servers fix
6233
6234   A malicious client can trigger an OPENSSL_assert (i.e., an abort) in
6235   servers that both support SSLv2 and enable export cipher suites by sending
6236   a specially crafted SSLv2 CLIENT-MASTER-KEY message.
6237
6238   This issue was discovered by Sean Burford (Google) and Emilia Käsper
6239   (OpenSSL development team).
6240   ([CVE-2015-0293])
6241
6242   *Emilia Käsper*
6243
6244 * Empty CKE with client auth and DHE fix
6245
6246   If client auth is used then a server can seg fault in the event of a DHE
6247   ciphersuite being selected and a zero length ClientKeyExchange message
6248   being sent by the client. This could be exploited in a DoS attack.
6249   ([CVE-2015-1787])
6250
6251   *Matt Caswell*
6252
6253 * Handshake with unseeded PRNG fix
6254
6255   Under certain conditions an OpenSSL 1.0.2 client can complete a handshake
6256   with an unseeded PRNG. The conditions are:
6257   - The client is on a platform where the PRNG has not been seeded
6258   automatically, and the user has not seeded manually
6259   - A protocol specific client method version has been used (i.e. not
6260   SSL_client_methodv23)
6261   - A ciphersuite is used that does not require additional random data from
6262   the PRNG beyond the initial ClientHello client random (e.g. PSK-RC4-SHA).
6263
6264   If the handshake succeeds then the client random that has been used will
6265   have been generated from a PRNG with insufficient entropy and therefore the
6266   output may be predictable.
6267
6268   For example using the following command with an unseeded openssl will
6269   succeed on an unpatched platform:
6270
6271   openssl s_client -psk 1a2b3c4d -tls1_2 -cipher PSK-RC4-SHA
6272   ([CVE-2015-0285])
6273
6274   *Matt Caswell*
6275
6276 * Use After Free following d2i_ECPrivatekey error fix
6277
6278   A malformed EC private key file consumed via the d2i_ECPrivateKey function
6279   could cause a use after free condition. This, in turn, could cause a double
6280   free in several private key parsing functions (such as d2i_PrivateKey
6281   or EVP_PKCS82PKEY) and could lead to a DoS attack or memory corruption
6282   for applications that receive EC private keys from untrusted
6283   sources. This scenario is considered rare.
6284
6285   This issue was discovered by the BoringSSL project and fixed in their
6286   commit 517073cd4b.
6287   ([CVE-2015-0209])
6288
6289   *Matt Caswell*
6290
6291 * X509_to_X509_REQ NULL pointer deref fix
6292
6293   The function X509_to_X509_REQ will crash with a NULL pointer dereference if
6294   the certificate key is invalid. This function is rarely used in practice.
6295
6296   This issue was discovered by Brian Carpenter.
6297   ([CVE-2015-0288])
6298
6299   *Stephen Henson*
6300
6301 * Removed the export ciphers from the DEFAULT ciphers
6302
6303   *Kurt Roeckx*
6304
6305### Changes between 1.0.1l and 1.0.2 [22 Jan 2015]
6306
6307 * Facilitate "universal" ARM builds targeting range of ARM ISAs, e.g.
6308   ARMv5 through ARMv8, as opposite to "locking" it to single one.
6309   So far those who have to target multiple platforms would compromise
6310   and argue that binary targeting say ARMv5 would still execute on
6311   ARMv8. "Universal" build resolves this compromise by providing
6312   near-optimal performance even on newer platforms.
6313
6314   *Andy Polyakov*
6315
6316 * Accelerated NIST P-256 elliptic curve implementation for x86_64
6317   (other platforms pending).
6318
6319   *Shay Gueron & Vlad Krasnov (Intel Corp), Andy Polyakov*
6320
6321 * Add support for the SignedCertificateTimestampList certificate and
6322   OCSP response extensions from RFC6962.
6323
6324   *Rob Stradling*
6325
6326 * Fix ec_GFp_simple_points_make_affine (thus, EC_POINTs_mul etc.)
6327   for corner cases. (Certain input points at infinity could lead to
6328   bogus results, with non-infinity inputs mapped to infinity too.)
6329
6330   *Bodo Moeller*
6331
6332 * Initial support for PowerISA 2.0.7, first implemented in POWER8.
6333   This covers AES, SHA256/512 and GHASH. "Initial" means that most
6334   common cases are optimized and there still is room for further
6335   improvements. Vector Permutation AES for Altivec is also added.
6336
6337   *Andy Polyakov*
6338
6339 * Add support for little-endian ppc64 Linux target.
6340
6341   *Marcelo Cerri (IBM)*
6342
6343 * Initial support for AMRv8 ISA crypto extensions. This covers AES,
6344   SHA1, SHA256 and GHASH. "Initial" means that most common cases
6345   are optimized and there still is room for further improvements.
6346   Both 32- and 64-bit modes are supported.
6347
6348   *Andy Polyakov, Ard Biesheuvel (Linaro)*
6349
6350 * Improved ARMv7 NEON support.
6351
6352   *Andy Polyakov*
6353
6354 * Support for SPARC Architecture 2011 crypto extensions, first
6355   implemented in SPARC T4. This covers AES, DES, Camellia, SHA1,
6356   SHA256/512, MD5, GHASH and modular exponentiation.
6357
6358   *Andy Polyakov, David Miller*
6359
6360 * Accelerated modular exponentiation for Intel processors, a.k.a.
6361   RSAZ.
6362
6363   *Shay Gueron & Vlad Krasnov (Intel Corp)*
6364
6365 * Support for new and upcoming Intel processors, including AVX2,
6366   BMI and SHA ISA extensions. This includes additional "stitched"
6367   implementations, AESNI-SHA256 and GCM, and multi-buffer support
6368   for TLS encrypt.
6369
6370   This work was sponsored by Intel Corp.
6371
6372   *Andy Polyakov*
6373
6374 * Support for DTLS 1.2. This adds two sets of DTLS methods: DTLS_*_method()
6375   supports both DTLS 1.2 and 1.0 and should use whatever version the peer
6376   supports and DTLSv1_2_*_method() which supports DTLS 1.2 only.
6377
6378   *Steve Henson*
6379
6380 * Use algorithm specific chains in SSL_CTX_use_certificate_chain_file():
6381   this fixes a limitation in previous versions of OpenSSL.
6382
6383   *Steve Henson*
6384
6385 * Extended RSA OAEP support via EVP_PKEY API. Options to specify digest,
6386   MGF1 digest and OAEP label.
6387
6388   *Steve Henson*
6389
6390 * Add EVP support for key wrapping algorithms, to avoid problems with
6391   existing code the flag EVP_CIPHER_CTX_WRAP_ALLOW has to be set in
6392   the EVP_CIPHER_CTX or an error is returned. Add AES and DES3 wrap
6393   algorithms and include tests cases.
6394
6395   *Steve Henson*
6396
6397 * Add functions to allocate and set the fields of an ECDSA_METHOD
6398   structure.
6399
6400   *Douglas E. Engert, Steve Henson*
6401
6402 * New functions OPENSSL_gmtime_diff and ASN1_TIME_diff to find the
6403   difference in days and seconds between two tm or ASN1_TIME structures.
6404
6405   *Steve Henson*
6406
6407 * Add -rev test option to s_server to just reverse order of characters
6408   received by client and send back to server. Also prints an abbreviated
6409   summary of the connection parameters.
6410
6411   *Steve Henson*
6412
6413 * New option -brief for s_client and s_server to print out a brief summary
6414   of connection parameters.
6415
6416   *Steve Henson*
6417
6418 * Add callbacks for arbitrary TLS extensions.
6419
6420   *Trevor Perrin <trevp@trevp.net> and Ben Laurie*
6421
6422 * New option -crl_download in several openssl utilities to download CRLs
6423   from CRLDP extension in certificates.
6424
6425   *Steve Henson*
6426
6427 * New options -CRL and -CRLform for s_client and s_server for CRLs.
6428
6429   *Steve Henson*
6430
6431 * New function X509_CRL_diff to generate a delta CRL from the difference
6432   of two full CRLs. Add support to "crl" utility.
6433
6434   *Steve Henson*
6435
6436 * New functions to set lookup_crls function and to retrieve
6437   X509_STORE from X509_STORE_CTX.
6438
6439   *Steve Henson*
6440
6441 * Print out deprecated issuer and subject unique ID fields in
6442   certificates.
6443
6444   *Steve Henson*
6445
6446 * Extend OCSP I/O functions so they can be used for simple general purpose
6447   HTTP as well as OCSP. New wrapper function which can be used to download
6448   CRLs using the OCSP API.
6449
6450   *Steve Henson*
6451
6452 * Delegate command line handling in s_client/s_server to SSL_CONF APIs.
6453
6454   *Steve Henson*
6455
6456 * `SSL_CONF*` functions. These provide a common framework for application
6457   configuration using configuration files or command lines.
6458
6459   *Steve Henson*
6460
6461 * SSL/TLS tracing code. This parses out SSL/TLS records using the
6462   message callback and prints the results. Needs compile time option
6463   "enable-ssl-trace". New options to s_client and s_server to enable
6464   tracing.
6465
6466   *Steve Henson*
6467
6468 * New ctrl and macro to retrieve supported points extensions.
6469   Print out extension in s_server and s_client.
6470
6471   *Steve Henson*
6472
6473 * New functions to retrieve certificate signature and signature
6474   OID NID.
6475
6476   *Steve Henson*
6477
6478 * Add functions to retrieve and manipulate the raw cipherlist sent by a
6479   client to OpenSSL.
6480
6481   *Steve Henson*
6482
6483 * New Suite B modes for TLS code. These use and enforce the requirements
6484   of RFC6460: restrict ciphersuites, only permit Suite B algorithms and
6485   only use Suite B curves. The Suite B modes can be set by using the
6486   strings "SUITEB128", "SUITEB192" or "SUITEB128ONLY" for the cipherstring.
6487
6488   *Steve Henson*
6489
6490 * New chain verification flags for Suite B levels of security. Check
6491   algorithms are acceptable when flags are set in X509_verify_cert.
6492
6493   *Steve Henson*
6494
6495 * Make tls1_check_chain return a set of flags indicating checks passed
6496   by a certificate chain. Add additional tests to handle client
6497   certificates: checks for matching certificate type and issuer name
6498   comparison.
6499
6500   *Steve Henson*
6501
6502 * If an attempt is made to use a signature algorithm not in the peer
6503   preference list abort the handshake. If client has no suitable
6504   signature algorithms in response to a certificate request do not
6505   use the certificate.
6506
6507   *Steve Henson*
6508
6509 * If server EC tmp key is not in client preference list abort handshake.
6510
6511   *Steve Henson*
6512
6513 * Add support for certificate stores in CERT structure. This makes it
6514   possible to have different stores per SSL structure or one store in
6515   the parent SSL_CTX. Include distinct stores for certificate chain
6516   verification and chain building. New ctrl SSL_CTRL_BUILD_CERT_CHAIN
6517   to build and store a certificate chain in CERT structure: returning
6518   an error if the chain cannot be built: this will allow applications
6519   to test if a chain is correctly configured.
6520
6521   Note: if the CERT based stores are not set then the parent SSL_CTX
6522   store is used to retain compatibility with existing behaviour.
6523
6524   *Steve Henson*
6525
6526 * New function ssl_set_client_disabled to set a ciphersuite disabled
6527   mask based on the current session, check mask when sending client
6528   hello and checking the requested ciphersuite.
6529
6530   *Steve Henson*
6531
6532 * New ctrls to retrieve and set certificate types in a certificate
6533   request message. Print out received values in s_client. If certificate
6534   types is not set with custom values set sensible values based on
6535   supported signature algorithms.
6536
6537   *Steve Henson*
6538
6539 * Support for distinct client and server supported signature algorithms.
6540
6541   *Steve Henson*
6542
6543 * Add certificate callback. If set this is called whenever a certificate
6544   is required by client or server. An application can decide which
6545   certificate chain to present based on arbitrary criteria: for example
6546   supported signature algorithms. Add very simple example to s_server.
6547   This fixes many of the problems and restrictions of the existing client
6548   certificate callback: for example you can now clear an existing
6549   certificate and specify the whole chain.
6550
6551   *Steve Henson*
6552
6553 * Add new "valid_flags" field to CERT_PKEY structure which determines what
6554   the certificate can be used for (if anything). Set valid_flags field
6555   in new tls1_check_chain function. Simplify ssl_set_cert_masks which used
6556   to have similar checks in it.
6557
6558   Add new "cert_flags" field to CERT structure and include a "strict mode".
6559   This enforces some TLS certificate requirements (such as only permitting
6560   certificate signature algorithms contained in the supported algorithms
6561   extension) which some implementations ignore: this option should be used
6562   with caution as it could cause interoperability issues.
6563
6564   *Steve Henson*
6565
6566 * Update and tidy signature algorithm extension processing. Work out
6567   shared signature algorithms based on preferences and peer algorithms
6568   and print them out in s_client and s_server. Abort handshake if no
6569   shared signature algorithms.
6570
6571   *Steve Henson*
6572
6573 * Add new functions to allow customised supported signature algorithms
6574   for SSL and SSL_CTX structures. Add options to s_client and s_server
6575   to support them.
6576
6577   *Steve Henson*
6578
6579 * New function SSL_certs_clear() to delete all references to certificates
6580   from an SSL structure. Before this once a certificate had been added
6581   it couldn't be removed.
6582
6583   *Steve Henson*
6584
6585 * Integrate hostname, email address and IP address checking with certificate
6586   verification. New verify options supporting checking in openssl utility.
6587
6588   *Steve Henson*
6589
6590 * Fixes and wildcard matching support to hostname and email checking
6591   functions. Add manual page.
6592
6593   *Florian Weimer (Red Hat Product Security Team)*
6594
6595 * New functions to check a hostname email or IP address against a
6596   certificate. Add options x509 utility to print results of checks against
6597   a certificate.
6598
6599   *Steve Henson*
6600
6601 * Fix OCSP checking.
6602
6603   *Rob Stradling <rob.stradling@comodo.com> and Ben Laurie*
6604
6605 * Initial experimental support for explicitly trusted non-root CAs.
6606   OpenSSL still tries to build a complete chain to a root but if an
6607   intermediate CA has a trust setting included that is used. The first
6608   setting is used: whether to trust (e.g., -addtrust option to the x509
6609   utility) or reject.
6610
6611   *Steve Henson*
6612
6613 * Add -trusted_first option which attempts to find certificates in the
6614   trusted store even if an untrusted chain is also supplied.
6615
6616   *Steve Henson*
6617
6618 * MIPS assembly pack updates: support for MIPS32r2 and SmartMIPS ASE,
6619   platform support for Linux and Android.
6620
6621   *Andy Polyakov*
6622
6623 * Support for linux-x32, ILP32 environment in x86_64 framework.
6624
6625   *Andy Polyakov*
6626
6627 * Experimental multi-implementation support for FIPS capable OpenSSL.
6628   When in FIPS mode the approved implementations are used as normal,
6629   when not in FIPS mode the internal unapproved versions are used instead.
6630   This means that the FIPS capable OpenSSL isn't forced to use the
6631   (often lower performance) FIPS implementations outside FIPS mode.
6632
6633   *Steve Henson*
6634
6635 * Transparently support X9.42 DH parameters when calling
6636   PEM_read_bio_DHparameters. This means existing applications can handle
6637   the new parameter format automatically.
6638
6639   *Steve Henson*
6640
6641 * Initial experimental support for X9.42 DH parameter format: mainly
6642   to support use of 'q' parameter for RFC5114 parameters.
6643
6644   *Steve Henson*
6645
6646 * Add DH parameters from RFC5114 including test data to dhtest.
6647
6648   *Steve Henson*
6649
6650 * Support for automatic EC temporary key parameter selection. If enabled
6651   the most preferred EC parameters are automatically used instead of
6652   hardcoded fixed parameters. Now a server just has to call:
6653   SSL_CTX_set_ecdh_auto(ctx, 1) and the server will automatically
6654   support ECDH and use the most appropriate parameters.
6655
6656   *Steve Henson*
6657
6658 * Enhance and tidy EC curve and point format TLS extension code. Use
6659   static structures instead of allocation if default values are used.
6660   New ctrls to set curves we wish to support and to retrieve shared curves.
6661   Print out shared curves in s_server. New options to s_server and s_client
6662   to set list of supported curves.
6663
6664   *Steve Henson*
6665
6666 * New ctrls to retrieve supported signature algorithms and
6667   supported curve values as an array of NIDs. Extend openssl utility
6668   to print out received values.
6669
6670   *Steve Henson*
6671
6672 * Add new APIs EC_curve_nist2nid and EC_curve_nid2nist which convert
6673   between NIDs and the more common NIST names such as "P-256". Enhance
6674   ecparam utility and ECC method to recognise the NIST names for curves.
6675
6676   *Steve Henson*
6677
6678 * Enhance SSL/TLS certificate chain handling to support different
6679   chains for each certificate instead of one chain in the parent SSL_CTX.
6680
6681   *Steve Henson*
6682
6683 * Support for fixed DH ciphersuite client authentication: where both
6684   server and client use DH certificates with common parameters.
6685
6686   *Steve Henson*
6687
6688 * Support for fixed DH ciphersuites: those requiring DH server
6689   certificates.
6690
6691   *Steve Henson*
6692
6693 * New function i2d_re_X509_tbs for re-encoding the TBS portion of
6694   the certificate.
6695   Note: Related 1.0.2-beta specific macros X509_get_cert_info,
6696   X509_CINF_set_modified, X509_CINF_get_issuer, X509_CINF_get_extensions and
6697   X509_CINF_get_signature were reverted post internal team review.
6698
6699OpenSSL 1.0.1
6700-------------
6701
6702### Changes between 1.0.1t and 1.0.1u [22 Sep 2016]
6703
6704 * OCSP Status Request extension unbounded memory growth
6705
6706   A malicious client can send an excessively large OCSP Status Request
6707   extension. If that client continually requests renegotiation, sending a
6708   large OCSP Status Request extension each time, then there will be unbounded
6709   memory growth on the server. This will eventually lead to a Denial Of
6710   Service attack through memory exhaustion. Servers with a default
6711   configuration are vulnerable even if they do not support OCSP. Builds using
6712   the "no-ocsp" build time option are not affected.
6713
6714   This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.)
6715   ([CVE-2016-6304])
6716
6717   *Matt Caswell*
6718
6719 * In order to mitigate the SWEET32 attack, the DES ciphers were moved from
6720   HIGH to MEDIUM.
6721
6722   This issue was reported to OpenSSL Karthikeyan Bhargavan and Gaetan
6723   Leurent (INRIA)
6724   ([CVE-2016-2183])
6725
6726   *Rich Salz*
6727
6728 * OOB write in MDC2_Update()
6729
6730   An overflow can occur in MDC2_Update() either if called directly or
6731   through the EVP_DigestUpdate() function using MDC2. If an attacker
6732   is able to supply very large amounts of input data after a previous
6733   call to EVP_EncryptUpdate() with a partial block then a length check
6734   can overflow resulting in a heap corruption.
6735
6736   The amount of data needed is comparable to SIZE_MAX which is impractical
6737   on most platforms.
6738
6739   This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.)
6740   ([CVE-2016-6303])
6741
6742   *Stephen Henson*
6743
6744 * Malformed SHA512 ticket DoS
6745
6746   If a server uses SHA512 for TLS session ticket HMAC it is vulnerable to a
6747   DoS attack where a malformed ticket will result in an OOB read which will
6748   ultimately crash.
6749
6750   The use of SHA512 in TLS session tickets is comparatively rare as it requires
6751   a custom server callback and ticket lookup mechanism.
6752
6753   This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.)
6754   ([CVE-2016-6302])
6755
6756   *Stephen Henson*
6757
6758 * OOB write in BN_bn2dec()
6759
6760   The function BN_bn2dec() does not check the return value of BN_div_word().
6761   This can cause an OOB write if an application uses this function with an
6762   overly large BIGNUM. This could be a problem if an overly large certificate
6763   or CRL is printed out from an untrusted source. TLS is not affected because
6764   record limits will reject an oversized certificate before it is parsed.
6765
6766   This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.)
6767   ([CVE-2016-2182])
6768
6769   *Stephen Henson*
6770
6771 * OOB read in TS_OBJ_print_bio()
6772
6773   The function TS_OBJ_print_bio() misuses OBJ_obj2txt(): the return value is
6774   the total length the OID text representation would use and not the amount
6775   of data written. This will result in OOB reads when large OIDs are
6776   presented.
6777
6778   This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.)
6779   ([CVE-2016-2180])
6780
6781   *Stephen Henson*
6782
6783 * Pointer arithmetic undefined behaviour
6784
6785   Avoid some undefined pointer arithmetic
6786
6787   A common idiom in the codebase is to check limits in the following manner:
6788   "p + len > limit"
6789
6790   Where "p" points to some malloc'd data of SIZE bytes and
6791   limit == p + SIZE
6792
6793   "len" here could be from some externally supplied data (e.g. from a TLS
6794   message).
6795
6796   The rules of C pointer arithmetic are such that "p + len" is only well
6797   defined where len <= SIZE. Therefore the above idiom is actually
6798   undefined behaviour.
6799
6800   For example this could cause problems if some malloc implementation
6801   provides an address for "p" such that "p + len" actually overflows for
6802   values of len that are too big and therefore p + len < limit.
6803
6804   This issue was reported to OpenSSL by Guido Vranken
6805   ([CVE-2016-2177])
6806
6807   *Matt Caswell*
6808
6809 * Constant time flag not preserved in DSA signing
6810
6811   Operations in the DSA signing algorithm should run in constant time in
6812   order to avoid side channel attacks. A flaw in the OpenSSL DSA
6813   implementation means that a non-constant time codepath is followed for
6814   certain operations. This has been demonstrated through a cache-timing
6815   attack to be sufficient for an attacker to recover the private DSA key.
6816
6817   This issue was reported by César Pereida (Aalto University), Billy Brumley
6818   (Tampere University of Technology), and Yuval Yarom (The University of
6819   Adelaide and NICTA).
6820   ([CVE-2016-2178])
6821
6822   *César Pereida*
6823
6824 * DTLS buffered message DoS
6825
6826   In a DTLS connection where handshake messages are delivered out-of-order
6827   those messages that OpenSSL is not yet ready to process will be buffered
6828   for later use. Under certain circumstances, a flaw in the logic means that
6829   those messages do not get removed from the buffer even though the handshake
6830   has been completed. An attacker could force up to approx. 15 messages to
6831   remain in the buffer when they are no longer required. These messages will
6832   be cleared when the DTLS connection is closed. The default maximum size for
6833   a message is 100k. Therefore the attacker could force an additional 1500k
6834   to be consumed per connection. By opening many simulataneous connections an
6835   attacker could cause a DoS attack through memory exhaustion.
6836
6837   This issue was reported to OpenSSL by Quan Luo.
6838   ([CVE-2016-2179])
6839
6840   *Matt Caswell*
6841
6842 * DTLS replay protection DoS
6843
6844   A flaw in the DTLS replay attack protection mechanism means that records
6845   that arrive for future epochs update the replay protection "window" before
6846   the MAC for the record has been validated. This could be exploited by an
6847   attacker by sending a record for the next epoch (which does not have to
6848   decrypt or have a valid MAC), with a very large sequence number. This means
6849   that all subsequent legitimate packets are dropped causing a denial of
6850   service for a specific DTLS connection.
6851
6852   This issue was reported to OpenSSL by the OCAP audit team.
6853   ([CVE-2016-2181])
6854
6855   *Matt Caswell*
6856
6857 * Certificate message OOB reads
6858
6859   In OpenSSL 1.0.2 and earlier some missing message length checks can result
6860   in OOB reads of up to 2 bytes beyond an allocated buffer. There is a
6861   theoretical DoS risk but this has not been observed in practice on common
6862   platforms.
6863
6864   The messages affected are client certificate, client certificate request
6865   and server certificate. As a result the attack can only be performed
6866   against a client or a server which enables client authentication.
6867
6868   This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.)
6869   ([CVE-2016-6306])
6870
6871   *Stephen Henson*
6872
6873### Changes between 1.0.1s and 1.0.1t [3 May 2016]
6874
6875 * Prevent padding oracle in AES-NI CBC MAC check
6876
6877   A MITM attacker can use a padding oracle attack to decrypt traffic
6878   when the connection uses an AES CBC cipher and the server support
6879   AES-NI.
6880
6881   This issue was introduced as part of the fix for Lucky 13 padding
6882   attack ([CVE-2013-0169]). The padding check was rewritten to be in
6883   constant time by making sure that always the same bytes are read and
6884   compared against either the MAC or padding bytes. But it no longer
6885   checked that there was enough data to have both the MAC and padding
6886   bytes.
6887
6888   This issue was reported by Juraj Somorovsky using TLS-Attacker.
6889   ([CVE-2016-2107])
6890
6891   *Kurt Roeckx*
6892
6893 * Fix EVP_EncodeUpdate overflow
6894
6895   An overflow can occur in the EVP_EncodeUpdate() function which is used for
6896   Base64 encoding of binary data. If an attacker is able to supply very large
6897   amounts of input data then a length check can overflow resulting in a heap
6898   corruption.
6899
6900   Internally to OpenSSL the EVP_EncodeUpdate() function is primarly used by
6901   the `PEM_write_bio*` family of functions. These are mainly used within the
6902   OpenSSL command line applications, so any application which processes data
6903   from an untrusted source and outputs it as a PEM file should be considered
6904   vulnerable to this issue. User applications that call these APIs directly
6905   with large amounts of untrusted data may also be vulnerable.
6906
6907   This issue was reported by Guido Vranken.
6908   ([CVE-2016-2105])
6909
6910   *Matt Caswell*
6911
6912 * Fix EVP_EncryptUpdate overflow
6913
6914   An overflow can occur in the EVP_EncryptUpdate() function. If an attacker
6915   is able to supply very large amounts of input data after a previous call to
6916   EVP_EncryptUpdate() with a partial block then a length check can overflow
6917   resulting in a heap corruption. Following an analysis of all OpenSSL
6918   internal usage of the EVP_EncryptUpdate() function all usage is one of two
6919   forms. The first form is where the EVP_EncryptUpdate() call is known to be
6920   the first called function after an EVP_EncryptInit(), and therefore that
6921   specific call must be safe. The second form is where the length passed to
6922   EVP_EncryptUpdate() can be seen from the code to be some small value and
6923   therefore there is no possibility of an overflow. Since all instances are
6924   one of these two forms, it is believed that there can be no overflows in
6925   internal code due to this problem. It should be noted that
6926   EVP_DecryptUpdate() can call EVP_EncryptUpdate() in certain code paths.
6927   Also EVP_CipherUpdate() is a synonym for EVP_EncryptUpdate(). All instances
6928   of these calls have also been analysed too and it is believed there are no
6929   instances in internal usage where an overflow could occur.
6930
6931   This issue was reported by Guido Vranken.
6932   ([CVE-2016-2106])
6933
6934   *Matt Caswell*
6935
6936 * Prevent ASN.1 BIO excessive memory allocation
6937
6938   When ASN.1 data is read from a BIO using functions such as d2i_CMS_bio()
6939   a short invalid encoding can casuse allocation of large amounts of memory
6940   potentially consuming excessive resources or exhausting memory.
6941
6942   Any application parsing untrusted data through d2i BIO functions is
6943   affected. The memory based functions such as d2i_X509() are *not* affected.
6944   Since the memory based functions are used by the TLS library, TLS
6945   applications are not affected.
6946
6947   This issue was reported by Brian Carpenter.
6948   ([CVE-2016-2109])
6949
6950   *Stephen Henson*
6951
6952 * EBCDIC overread
6953
6954   ASN1 Strings that are over 1024 bytes can cause an overread in applications
6955   using the X509_NAME_oneline() function on EBCDIC systems. This could result
6956   in arbitrary stack data being returned in the buffer.
6957
6958   This issue was reported by Guido Vranken.
6959   ([CVE-2016-2176])
6960
6961   *Matt Caswell*
6962
6963 * Modify behavior of ALPN to invoke callback after SNI/servername
6964   callback, such that updates to the SSL_CTX affect ALPN.
6965
6966   *Todd Short*
6967
6968 * Remove LOW from the DEFAULT cipher list.  This removes singles DES from the
6969   default.
6970
6971   *Kurt Roeckx*
6972
6973 * Only remove the SSLv2 methods with the no-ssl2-method option. When the
6974   methods are enabled and ssl2 is disabled the methods return NULL.
6975
6976   *Kurt Roeckx*
6977
6978### Changes between 1.0.1r and 1.0.1s [1 Mar 2016]
6979
6980* Disable weak ciphers in SSLv3 and up in default builds of OpenSSL.
6981  Builds that are not configured with "enable-weak-ssl-ciphers" will not
6982  provide any "EXPORT" or "LOW" strength ciphers.
6983
6984  *Viktor Dukhovni*
6985
6986* Disable SSLv2 default build, default negotiation and weak ciphers.  SSLv2
6987  is by default disabled at build-time.  Builds that are not configured with
6988  "enable-ssl2" will not support SSLv2.  Even if "enable-ssl2" is used,
6989  users who want to negotiate SSLv2 via the version-flexible SSLv23_method()
6990  will need to explicitly call either of:
6991
6992      SSL_CTX_clear_options(ctx, SSL_OP_NO_SSLv2);
6993  or
6994      SSL_clear_options(ssl, SSL_OP_NO_SSLv2);
6995
6996  as appropriate.  Even if either of those is used, or the application
6997  explicitly uses the version-specific SSLv2_method() or its client and
6998  server variants, SSLv2 ciphers vulnerable to exhaustive search key
6999  recovery have been removed.  Specifically, the SSLv2 40-bit EXPORT
7000  ciphers, and SSLv2 56-bit DES are no longer available.
7001  ([CVE-2016-0800])
7002
7003  *Viktor Dukhovni*
7004
7005 * Fix a double-free in DSA code
7006
7007   A double free bug was discovered when OpenSSL parses malformed DSA private
7008   keys and could lead to a DoS attack or memory corruption for applications
7009   that receive DSA private keys from untrusted sources.  This scenario is
7010   considered rare.
7011
7012   This issue was reported to OpenSSL by Adam Langley(Google/BoringSSL) using
7013   libFuzzer.
7014   ([CVE-2016-0705])
7015
7016   *Stephen Henson*
7017
7018 * Disable SRP fake user seed to address a server memory leak.
7019
7020   Add a new method SRP_VBASE_get1_by_user that handles the seed properly.
7021
7022   SRP_VBASE_get_by_user had inconsistent memory management behaviour.
7023   In order to fix an unavoidable memory leak, SRP_VBASE_get_by_user
7024   was changed to ignore the "fake user" SRP seed, even if the seed
7025   is configured.
7026
7027   Users should use SRP_VBASE_get1_by_user instead. Note that in
7028   SRP_VBASE_get1_by_user, caller must free the returned value. Note
7029   also that even though configuring the SRP seed attempts to hide
7030   invalid usernames by continuing the handshake with fake
7031   credentials, this behaviour is not constant time and no strong
7032   guarantees are made that the handshake is indistinguishable from
7033   that of a valid user.
7034   ([CVE-2016-0798])
7035
7036   *Emilia Käsper*
7037
7038 * Fix BN_hex2bn/BN_dec2bn NULL pointer deref/heap corruption
7039
7040   In the BN_hex2bn function the number of hex digits is calculated using an
7041   int value `i`. Later `bn_expand` is called with a value of `i * 4`. For
7042   large values of `i` this can result in `bn_expand` not allocating any
7043   memory because `i * 4` is negative. This can leave the internal BIGNUM data
7044   field as NULL leading to a subsequent NULL ptr deref. For very large values
7045   of `i`, the calculation `i * 4` could be a positive value smaller than `i`.
7046   In this case memory is allocated to the internal BIGNUM data field, but it
7047   is insufficiently sized leading to heap corruption. A similar issue exists
7048   in BN_dec2bn. This could have security consequences if BN_hex2bn/BN_dec2bn
7049   is ever called by user applications with very large untrusted hex/dec data.
7050   This is anticipated to be a rare occurrence.
7051
7052   All OpenSSL internal usage of these functions use data that is not expected
7053   to be untrusted, e.g. config file data or application command line
7054   arguments. If user developed applications generate config file data based
7055   on untrusted data then it is possible that this could also lead to security
7056   consequences. This is also anticipated to be rare.
7057
7058   This issue was reported to OpenSSL by Guido Vranken.
7059   ([CVE-2016-0797])
7060
7061   *Matt Caswell*
7062
7063 * Fix memory issues in `BIO_*printf` functions
7064
7065   The internal `fmtstr` function used in processing a "%s" format string in
7066   the `BIO_*printf` functions could overflow while calculating the length of a
7067   string and cause an OOB read when printing very long strings.
7068
7069   Additionally the internal `doapr_outch` function can attempt to write to an
7070   OOB memory location (at an offset from the NULL pointer) in the event of a
7071   memory allocation failure. In 1.0.2 and below this could be caused where
7072   the size of a buffer to be allocated is greater than INT_MAX. E.g. this
7073   could be in processing a very long "%s" format string. Memory leaks can
7074   also occur.
7075
7076   The first issue may mask the second issue dependent on compiler behaviour.
7077   These problems could enable attacks where large amounts of untrusted data
7078   is passed to the `BIO_*printf` functions. If applications use these functions
7079   in this way then they could be vulnerable. OpenSSL itself uses these
7080   functions when printing out human-readable dumps of ASN.1 data. Therefore
7081   applications that print this data could be vulnerable if the data is from
7082   untrusted sources. OpenSSL command line applications could also be
7083   vulnerable where they print out ASN.1 data, or if untrusted data is passed
7084   as command line arguments.
7085
7086   Libssl is not considered directly vulnerable. Additionally certificates etc
7087   received via remote connections via libssl are also unlikely to be able to
7088   trigger these issues because of message size limits enforced within libssl.
7089
7090   This issue was reported to OpenSSL Guido Vranken.
7091   ([CVE-2016-0799])
7092
7093   *Matt Caswell*
7094
7095 * Side channel attack on modular exponentiation
7096
7097   A side-channel attack was found which makes use of cache-bank conflicts on
7098   the Intel Sandy-Bridge microarchitecture which could lead to the recovery
7099   of RSA keys.  The ability to exploit this issue is limited as it relies on
7100   an attacker who has control of code in a thread running on the same
7101   hyper-threaded core as the victim thread which is performing decryptions.
7102
7103   This issue was reported to OpenSSL by Yuval Yarom, The University of
7104   Adelaide and NICTA, Daniel Genkin, Technion and Tel Aviv University, and
7105   Nadia Heninger, University of Pennsylvania with more information at
7106   <http://cachebleed.info>.
7107   ([CVE-2016-0702])
7108
7109   *Andy Polyakov*
7110
7111 * Change the req command to generate a 2048-bit RSA/DSA key by default,
7112   if no keysize is specified with default_bits. This fixes an
7113   omission in an earlier change that changed all RSA/DSA key generation
7114   commands to use 2048 bits by default.
7115
7116   *Emilia Käsper*
7117
7118### Changes between 1.0.1q and 1.0.1r [28 Jan 2016]
7119
7120 * Protection for DH small subgroup attacks
7121
7122   As a precautionary measure the SSL_OP_SINGLE_DH_USE option has been
7123   switched on by default and cannot be disabled. This could have some
7124   performance impact.
7125
7126   *Matt Caswell*
7127
7128 * SSLv2 doesn't block disabled ciphers
7129
7130   A malicious client can negotiate SSLv2 ciphers that have been disabled on
7131   the server and complete SSLv2 handshakes even if all SSLv2 ciphers have
7132   been disabled, provided that the SSLv2 protocol was not also disabled via
7133   SSL_OP_NO_SSLv2.
7134
7135   This issue was reported to OpenSSL on 26th December 2015 by Nimrod Aviram
7136   and Sebastian Schinzel.
7137   ([CVE-2015-3197])
7138
7139   *Viktor Dukhovni*
7140
7141 * Reject DH handshakes with parameters shorter than 1024 bits.
7142
7143   *Kurt Roeckx*
7144
7145### Changes between 1.0.1p and 1.0.1q [3 Dec 2015]
7146
7147 * Certificate verify crash with missing PSS parameter
7148
7149   The signature verification routines will crash with a NULL pointer
7150   dereference if presented with an ASN.1 signature using the RSA PSS
7151   algorithm and absent mask generation function parameter. Since these
7152   routines are used to verify certificate signature algorithms this can be
7153   used to crash any certificate verification operation and exploited in a
7154   DoS attack. Any application which performs certificate verification is
7155   vulnerable including OpenSSL clients and servers which enable client
7156   authentication.
7157
7158   This issue was reported to OpenSSL by Loïc Jonas Etienne (Qnective AG).
7159   ([CVE-2015-3194])
7160
7161   *Stephen Henson*
7162
7163 * X509_ATTRIBUTE memory leak
7164
7165   When presented with a malformed X509_ATTRIBUTE structure OpenSSL will leak
7166   memory. This structure is used by the PKCS#7 and CMS routines so any
7167   application which reads PKCS#7 or CMS data from untrusted sources is
7168   affected. SSL/TLS is not affected.
7169
7170   This issue was reported to OpenSSL by Adam Langley (Google/BoringSSL) using
7171   libFuzzer.
7172   ([CVE-2015-3195])
7173
7174   *Stephen Henson*
7175
7176 * Rewrite EVP_DecodeUpdate (base64 decoding) to fix several bugs.
7177   This changes the decoding behaviour for some invalid messages,
7178   though the change is mostly in the more lenient direction, and
7179   legacy behaviour is preserved as much as possible.
7180
7181   *Emilia Käsper*
7182
7183 * In DSA_generate_parameters_ex, if the provided seed is too short,
7184   use a random seed, as already documented.
7185
7186   *Rich Salz and Ismo Puustinen <ismo.puustinen@intel.com>*
7187
7188### Changes between 1.0.1o and 1.0.1p [9 Jul 2015]
7189
7190 * Alternate chains certificate forgery
7191
7192   During certificate verfification, OpenSSL will attempt to find an
7193   alternative certificate chain if the first attempt to build such a chain
7194   fails. An error in the implementation of this logic can mean that an
7195   attacker could cause certain checks on untrusted certificates to be
7196   bypassed, such as the CA flag, enabling them to use a valid leaf
7197   certificate to act as a CA and "issue" an invalid certificate.
7198
7199   This issue was reported to OpenSSL by Adam Langley/David Benjamin
7200   (Google/BoringSSL).
7201   ([CVE-2015-1793])
7202
7203   *Matt Caswell*
7204
7205 * Race condition handling PSK identify hint
7206
7207   If PSK identity hints are received by a multi-threaded client then
7208   the values are wrongly updated in the parent SSL_CTX structure. This can
7209   result in a race condition potentially leading to a double free of the
7210   identify hint data.
7211   ([CVE-2015-3196])
7212
7213   *Stephen Henson*
7214
7215### Changes between 1.0.1n and 1.0.1o [12 Jun 2015]
7216
7217 * Fix HMAC ABI incompatibility. The previous version introduced an ABI
7218   incompatibility in the handling of HMAC. The previous ABI has now been
7219   restored.
7220
7221### Changes between 1.0.1m and 1.0.1n [11 Jun 2015]
7222
7223 * Malformed ECParameters causes infinite loop
7224
7225   When processing an ECParameters structure OpenSSL enters an infinite loop
7226   if the curve specified is over a specially malformed binary polynomial
7227   field.
7228
7229   This can be used to perform denial of service against any
7230   system which processes public keys, certificate requests or
7231   certificates.  This includes TLS clients and TLS servers with
7232   client authentication enabled.
7233
7234   This issue was reported to OpenSSL by Joseph Barr-Pixton.
7235   ([CVE-2015-1788])
7236
7237   *Andy Polyakov*
7238
7239 * Exploitable out-of-bounds read in X509_cmp_time
7240
7241   X509_cmp_time does not properly check the length of the ASN1_TIME
7242   string and can read a few bytes out of bounds. In addition,
7243   X509_cmp_time accepts an arbitrary number of fractional seconds in the
7244   time string.
7245
7246   An attacker can use this to craft malformed certificates and CRLs of
7247   various sizes and potentially cause a segmentation fault, resulting in
7248   a DoS on applications that verify certificates or CRLs. TLS clients
7249   that verify CRLs are affected. TLS clients and servers with client
7250   authentication enabled may be affected if they use custom verification
7251   callbacks.
7252
7253   This issue was reported to OpenSSL by Robert Swiecki (Google), and
7254   independently by Hanno Böck.
7255   ([CVE-2015-1789])
7256
7257   *Emilia Käsper*
7258
7259 * PKCS7 crash with missing EnvelopedContent
7260
7261   The PKCS#7 parsing code does not handle missing inner EncryptedContent
7262   correctly. An attacker can craft malformed ASN.1-encoded PKCS#7 blobs
7263   with missing content and trigger a NULL pointer dereference on parsing.
7264
7265   Applications that decrypt PKCS#7 data or otherwise parse PKCS#7
7266   structures from untrusted sources are affected. OpenSSL clients and
7267   servers are not affected.
7268
7269   This issue was reported to OpenSSL by Michal Zalewski (Google).
7270   ([CVE-2015-1790])
7271
7272   *Emilia Käsper*
7273
7274 * CMS verify infinite loop with unknown hash function
7275
7276   When verifying a signedData message the CMS code can enter an infinite loop
7277   if presented with an unknown hash function OID. This can be used to perform
7278   denial of service against any system which verifies signedData messages using
7279   the CMS code.
7280   This issue was reported to OpenSSL by Johannes Bauer.
7281   ([CVE-2015-1792])
7282
7283   *Stephen Henson*
7284
7285 * Race condition handling NewSessionTicket
7286
7287   If a NewSessionTicket is received by a multi-threaded client when attempting to
7288   reuse a previous ticket then a race condition can occur potentially leading to
7289   a double free of the ticket data.
7290   ([CVE-2015-1791])
7291
7292   *Matt Caswell*
7293
7294 * Reject DH handshakes with parameters shorter than 768 bits.
7295
7296   *Kurt Roeckx and Emilia Kasper*
7297
7298 * dhparam: generate 2048-bit parameters by default.
7299
7300   *Kurt Roeckx and Emilia Kasper*
7301
7302### Changes between 1.0.1l and 1.0.1m [19 Mar 2015]
7303
7304 * Segmentation fault in ASN1_TYPE_cmp fix
7305
7306   The function ASN1_TYPE_cmp will crash with an invalid read if an attempt is
7307   made to compare ASN.1 boolean types. Since ASN1_TYPE_cmp is used to check
7308   certificate signature algorithm consistency this can be used to crash any
7309   certificate verification operation and exploited in a DoS attack. Any
7310   application which performs certificate verification is vulnerable including
7311   OpenSSL clients and servers which enable client authentication.
7312   ([CVE-2015-0286])
7313
7314   *Stephen Henson*
7315
7316 * ASN.1 structure reuse memory corruption fix
7317
7318   Reusing a structure in ASN.1 parsing may allow an attacker to cause
7319   memory corruption via an invalid write. Such reuse is and has been
7320   strongly discouraged and is believed to be rare.
7321
7322   Applications that parse structures containing CHOICE or ANY DEFINED BY
7323   components may be affected. Certificate parsing (d2i_X509 and related
7324   functions) are however not affected. OpenSSL clients and servers are
7325   not affected.
7326   ([CVE-2015-0287])
7327
7328   *Stephen Henson*
7329
7330 * PKCS7 NULL pointer dereferences fix
7331
7332   The PKCS#7 parsing code does not handle missing outer ContentInfo
7333   correctly. An attacker can craft malformed ASN.1-encoded PKCS#7 blobs with
7334   missing content and trigger a NULL pointer dereference on parsing.
7335
7336   Applications that verify PKCS#7 signatures, decrypt PKCS#7 data or
7337   otherwise parse PKCS#7 structures from untrusted sources are
7338   affected. OpenSSL clients and servers are not affected.
7339
7340   This issue was reported to OpenSSL by Michal Zalewski (Google).
7341   ([CVE-2015-0289])
7342
7343   *Emilia Käsper*
7344
7345 * DoS via reachable assert in SSLv2 servers fix
7346
7347   A malicious client can trigger an OPENSSL_assert (i.e., an abort) in
7348   servers that both support SSLv2 and enable export cipher suites by sending
7349   a specially crafted SSLv2 CLIENT-MASTER-KEY message.
7350
7351   This issue was discovered by Sean Burford (Google) and Emilia Käsper
7352   (OpenSSL development team).
7353   ([CVE-2015-0293])
7354
7355   *Emilia Käsper*
7356
7357 * Use After Free following d2i_ECPrivatekey error fix
7358
7359   A malformed EC private key file consumed via the d2i_ECPrivateKey function
7360   could cause a use after free condition. This, in turn, could cause a double
7361   free in several private key parsing functions (such as d2i_PrivateKey
7362   or EVP_PKCS82PKEY) and could lead to a DoS attack or memory corruption
7363   for applications that receive EC private keys from untrusted
7364   sources. This scenario is considered rare.
7365
7366   This issue was discovered by the BoringSSL project and fixed in their
7367   commit 517073cd4b.
7368   ([CVE-2015-0209])
7369
7370   *Matt Caswell*
7371
7372 * X509_to_X509_REQ NULL pointer deref fix
7373
7374   The function X509_to_X509_REQ will crash with a NULL pointer dereference if
7375   the certificate key is invalid. This function is rarely used in practice.
7376
7377   This issue was discovered by Brian Carpenter.
7378   ([CVE-2015-0288])
7379
7380   *Stephen Henson*
7381
7382 * Removed the export ciphers from the DEFAULT ciphers
7383
7384   *Kurt Roeckx*
7385
7386### Changes between 1.0.1k and 1.0.1l [15 Jan 2015]
7387
7388 * Build fixes for the Windows and OpenVMS platforms
7389
7390   *Matt Caswell and Richard Levitte*
7391
7392### Changes between 1.0.1j and 1.0.1k [8 Jan 2015]
7393
7394 * Fix DTLS segmentation fault in dtls1_get_record. A carefully crafted DTLS
7395   message can cause a segmentation fault in OpenSSL due to a NULL pointer
7396   dereference. This could lead to a Denial Of Service attack. Thanks to
7397   Markus Stenberg of Cisco Systems, Inc. for reporting this issue.
7398   ([CVE-2014-3571])
7399
7400   *Steve Henson*
7401
7402 * Fix DTLS memory leak in dtls1_buffer_record. A memory leak can occur in the
7403   dtls1_buffer_record function under certain conditions. In particular this
7404   could occur if an attacker sent repeated DTLS records with the same
7405   sequence number but for the next epoch. The memory leak could be exploited
7406   by an attacker in a Denial of Service attack through memory exhaustion.
7407   Thanks to Chris Mueller for reporting this issue.
7408   ([CVE-2015-0206])
7409
7410   *Matt Caswell*
7411
7412 * Fix issue where no-ssl3 configuration sets method to NULL. When openssl is
7413   built with the no-ssl3 option and a SSL v3 ClientHello is received the ssl
7414   method would be set to NULL which could later result in a NULL pointer
7415   dereference. Thanks to Frank Schmirler for reporting this issue.
7416   ([CVE-2014-3569])
7417
7418   *Kurt Roeckx*
7419
7420 * Abort handshake if server key exchange message is omitted for ephemeral
7421   ECDH ciphersuites.
7422
7423   Thanks to Karthikeyan Bhargavan of the PROSECCO team at INRIA for
7424   reporting this issue.
7425   ([CVE-2014-3572])
7426
7427   *Steve Henson*
7428
7429 * Remove non-export ephemeral RSA code on client and server. This code
7430   violated the TLS standard by allowing the use of temporary RSA keys in
7431   non-export ciphersuites and could be used by a server to effectively
7432   downgrade the RSA key length used to a value smaller than the server
7433   certificate. Thanks for Karthikeyan Bhargavan of the PROSECCO team at
7434   INRIA or reporting this issue.
7435   ([CVE-2015-0204])
7436
7437   *Steve Henson*
7438
7439 * Fixed issue where DH client certificates are accepted without verification.
7440   An OpenSSL server will accept a DH certificate for client authentication
7441   without the certificate verify message. This effectively allows a client to
7442   authenticate without the use of a private key. This only affects servers
7443   which trust a client certificate authority which issues certificates
7444   containing DH keys: these are extremely rare and hardly ever encountered.
7445   Thanks for Karthikeyan Bhargavan of the PROSECCO team at INRIA or reporting
7446   this issue.
7447   ([CVE-2015-0205])
7448
7449   *Steve Henson*
7450
7451 * Ensure that the session ID context of an SSL is updated when its
7452   SSL_CTX is updated via SSL_set_SSL_CTX.
7453
7454   The session ID context is typically set from the parent SSL_CTX,
7455   and can vary with the CTX.
7456
7457   *Adam Langley*
7458
7459 * Fix various certificate fingerprint issues.
7460
7461   By using non-DER or invalid encodings outside the signed portion of a
7462   certificate the fingerprint can be changed without breaking the signature.
7463   Although no details of the signed portion of the certificate can be changed
7464   this can cause problems with some applications: e.g. those using the
7465   certificate fingerprint for blacklists.
7466
7467   1. Reject signatures with non zero unused bits.
7468
7469   If the BIT STRING containing the signature has non zero unused bits reject
7470   the signature. All current signature algorithms require zero unused bits.
7471
7472   2. Check certificate algorithm consistency.
7473
7474   Check the AlgorithmIdentifier inside TBS matches the one in the
7475   certificate signature. NB: this will result in signature failure
7476   errors for some broken certificates.
7477
7478   Thanks to Konrad Kraszewski from Google for reporting this issue.
7479
7480   3. Check DSA/ECDSA signatures use DER.
7481
7482   Re-encode DSA/ECDSA signatures and compare with the original received
7483   signature. Return an error if there is a mismatch.
7484
7485   This will reject various cases including garbage after signature
7486   (thanks to Antti Karjalainen and Tuomo Untinen from the Codenomicon CROSS
7487   program for discovering this case) and use of BER or invalid ASN.1 INTEGERs
7488   (negative or with leading zeroes).
7489
7490   Further analysis was conducted and fixes were developed by Stephen Henson
7491   of the OpenSSL core team.
7492
7493   ([CVE-2014-8275])
7494
7495   *Steve Henson*
7496
7497 * Correct Bignum squaring. Bignum squaring (BN_sqr) may produce incorrect
7498   results on some platforms, including x86_64. This bug occurs at random
7499   with a very low probability, and is not known to be exploitable in any
7500   way, though its exact impact is difficult to determine. Thanks to Pieter
7501   Wuille (Blockstream) who reported this issue and also suggested an initial
7502   fix. Further analysis was conducted by the OpenSSL development team and
7503   Adam Langley of Google. The final fix was developed by Andy Polyakov of
7504   the OpenSSL core team.
7505   ([CVE-2014-3570])
7506
7507   *Andy Polyakov*
7508
7509 * Do not resume sessions on the server if the negotiated protocol
7510   version does not match the session's version. Resuming with a different
7511   version, while not strictly forbidden by the RFC, is of questionable
7512   sanity and breaks all known clients.
7513
7514   *David Benjamin, Emilia Käsper*
7515
7516 * Tighten handling of the ChangeCipherSpec (CCS) message: reject
7517   early CCS messages during renegotiation. (Note that because
7518   renegotiation is encrypted, this early CCS was not exploitable.)
7519
7520   *Emilia Käsper*
7521
7522 * Tighten client-side session ticket handling during renegotiation:
7523   ensure that the client only accepts a session ticket if the server sends
7524   the extension anew in the ServerHello. Previously, a TLS client would
7525   reuse the old extension state and thus accept a session ticket if one was
7526   announced in the initial ServerHello.
7527
7528   Similarly, ensure that the client requires a session ticket if one
7529   was advertised in the ServerHello. Previously, a TLS client would
7530   ignore a missing NewSessionTicket message.
7531
7532   *Emilia Käsper*
7533
7534### Changes between 1.0.1i and 1.0.1j [15 Oct 2014]
7535
7536 * SRTP Memory Leak.
7537
7538   A flaw in the DTLS SRTP extension parsing code allows an attacker, who
7539   sends a carefully crafted handshake message, to cause OpenSSL to fail
7540   to free up to 64k of memory causing a memory leak. This could be
7541   exploited in a Denial Of Service attack. This issue affects OpenSSL
7542   1.0.1 server implementations for both SSL/TLS and DTLS regardless of
7543   whether SRTP is used or configured. Implementations of OpenSSL that
7544   have been compiled with OPENSSL_NO_SRTP defined are not affected.
7545
7546   The fix was developed by the OpenSSL team.
7547   ([CVE-2014-3513])
7548
7549   *OpenSSL team*
7550
7551 * Session Ticket Memory Leak.
7552
7553   When an OpenSSL SSL/TLS/DTLS server receives a session ticket the
7554   integrity of that ticket is first verified. In the event of a session
7555   ticket integrity check failing, OpenSSL will fail to free memory
7556   causing a memory leak. By sending a large number of invalid session
7557   tickets an attacker could exploit this issue in a Denial Of Service
7558   attack.
7559   ([CVE-2014-3567])
7560
7561   *Steve Henson*
7562
7563 * Build option no-ssl3 is incomplete.
7564
7565   When OpenSSL is configured with "no-ssl3" as a build option, servers
7566   could accept and complete a SSL 3.0 handshake, and clients could be
7567   configured to send them.
7568   ([CVE-2014-3568])
7569
7570   *Akamai and the OpenSSL team*
7571
7572 * Add support for TLS_FALLBACK_SCSV.
7573   Client applications doing fallback retries should call
7574   SSL_set_mode(s, SSL_MODE_SEND_FALLBACK_SCSV).
7575   ([CVE-2014-3566])
7576
7577   *Adam Langley, Bodo Moeller*
7578
7579 * Add additional DigestInfo checks.
7580
7581   Re-encode DigestInto in DER and check against the original when
7582   verifying RSA signature: this will reject any improperly encoded
7583   DigestInfo structures.
7584
7585   Note: this is a precautionary measure and no attacks are currently known.
7586
7587   *Steve Henson*
7588
7589### Changes between 1.0.1h and 1.0.1i [6 Aug 2014]
7590
7591 * Fix SRP buffer overrun vulnerability. Invalid parameters passed to the
7592   SRP code can be overrun an internal buffer. Add sanity check that
7593   g, A, B < N to SRP code.
7594
7595   Thanks to Sean Devlin and Watson Ladd of Cryptography Services, NCC
7596   Group for discovering this issue.
7597   ([CVE-2014-3512])
7598
7599   *Steve Henson*
7600
7601 * A flaw in the OpenSSL SSL/TLS server code causes the server to negotiate
7602   TLS 1.0 instead of higher protocol versions when the ClientHello message
7603   is badly fragmented. This allows a man-in-the-middle attacker to force a
7604   downgrade to TLS 1.0 even if both the server and the client support a
7605   higher protocol version, by modifying the client's TLS records.
7606
7607   Thanks to David Benjamin and Adam Langley (Google) for discovering and
7608   researching this issue.
7609   ([CVE-2014-3511])
7610
7611   *David Benjamin*
7612
7613 * OpenSSL DTLS clients enabling anonymous (EC)DH ciphersuites are subject
7614   to a denial of service attack. A malicious server can crash the client
7615   with a null pointer dereference (read) by specifying an anonymous (EC)DH
7616   ciphersuite and sending carefully crafted handshake messages.
7617
7618   Thanks to Felix Gröbert (Google) for discovering and researching this
7619   issue.
7620   ([CVE-2014-3510])
7621
7622   *Emilia Käsper*
7623
7624 * By sending carefully crafted DTLS packets an attacker could cause openssl
7625   to leak memory. This can be exploited through a Denial of Service attack.
7626   Thanks to Adam Langley for discovering and researching this issue.
7627   ([CVE-2014-3507])
7628
7629   *Adam Langley*
7630
7631 * An attacker can force openssl to consume large amounts of memory whilst
7632   processing DTLS handshake messages. This can be exploited through a
7633   Denial of Service attack.
7634   Thanks to Adam Langley for discovering and researching this issue.
7635   ([CVE-2014-3506])
7636
7637   *Adam Langley*
7638
7639 * An attacker can force an error condition which causes openssl to crash
7640   whilst processing DTLS packets due to memory being freed twice. This
7641   can be exploited through a Denial of Service attack.
7642   Thanks to Adam Langley and Wan-Teh Chang for discovering and researching
7643   this issue.
7644   ([CVE-2014-3505])
7645
7646   *Adam Langley*
7647
7648 * If a multithreaded client connects to a malicious server using a resumed
7649   session and the server sends an ec point format extension it could write
7650   up to 255 bytes to freed memory.
7651
7652   Thanks to Gabor Tyukasz (LogMeIn Inc) for discovering and researching this
7653   issue.
7654   ([CVE-2014-3509])
7655
7656   *Gabor Tyukasz*
7657
7658 * A malicious server can crash an OpenSSL client with a null pointer
7659   dereference (read) by specifying an SRP ciphersuite even though it was not
7660   properly negotiated with the client. This can be exploited through a
7661   Denial of Service attack.
7662
7663   Thanks to Joonas Kuorilehto and Riku Hietamäki (Codenomicon) for
7664   discovering and researching this issue.
7665   ([CVE-2014-5139])
7666
7667   *Steve Henson*
7668
7669 * A flaw in OBJ_obj2txt may cause pretty printing functions such as
7670   X509_name_oneline, X509_name_print_ex et al. to leak some information
7671   from the stack. Applications may be affected if they echo pretty printing
7672   output to the attacker.
7673
7674   Thanks to Ivan Fratric (Google) for discovering this issue.
7675   ([CVE-2014-3508])
7676
7677   *Emilia Käsper, and Steve Henson*
7678
7679 * Fix ec_GFp_simple_points_make_affine (thus, EC_POINTs_mul etc.)
7680   for corner cases. (Certain input points at infinity could lead to
7681   bogus results, with non-infinity inputs mapped to infinity too.)
7682
7683   *Bodo Moeller*
7684
7685### Changes between 1.0.1g and 1.0.1h [5 Jun 2014]
7686
7687 * Fix for SSL/TLS MITM flaw. An attacker using a carefully crafted
7688   handshake can force the use of weak keying material in OpenSSL
7689   SSL/TLS clients and servers.
7690
7691   Thanks to KIKUCHI Masashi (Lepidum Co. Ltd.) for discovering and
7692   researching this issue. ([CVE-2014-0224])
7693
7694   *KIKUCHI Masashi, Steve Henson*
7695
7696 * Fix DTLS recursion flaw. By sending an invalid DTLS handshake to an
7697   OpenSSL DTLS client the code can be made to recurse eventually crashing
7698   in a DoS attack.
7699
7700   Thanks to Imre Rad (Search-Lab Ltd.) for discovering this issue.
7701   ([CVE-2014-0221])
7702
7703   *Imre Rad, Steve Henson*
7704
7705 * Fix DTLS invalid fragment vulnerability. A buffer overrun attack can
7706   be triggered by sending invalid DTLS fragments to an OpenSSL DTLS
7707   client or server. This is potentially exploitable to run arbitrary
7708   code on a vulnerable client or server.
7709
7710   Thanks to Jüri Aedla for reporting this issue. ([CVE-2014-0195])
7711
7712   *Jüri Aedla, Steve Henson*
7713
7714 * Fix bug in TLS code where clients enable anonymous ECDH ciphersuites
7715   are subject to a denial of service attack.
7716
7717   Thanks to Felix Gröbert and Ivan Fratric at Google for discovering
7718   this issue. ([CVE-2014-3470])
7719
7720   *Felix Gröbert, Ivan Fratric, Steve Henson*
7721
7722 * Harmonize version and its documentation. -f flag is used to display
7723   compilation flags.
7724
7725   *mancha <mancha1@zoho.com>*
7726
7727 * Fix eckey_priv_encode so it immediately returns an error upon a failure
7728   in i2d_ECPrivateKey.
7729
7730   *mancha <mancha1@zoho.com>*
7731
7732 * Fix some double frees. These are not thought to be exploitable.
7733
7734   *mancha <mancha1@zoho.com>*
7735
7736### Changes between 1.0.1f and 1.0.1g [7 Apr 2014]
7737
7738 * A missing bounds check in the handling of the TLS heartbeat extension
7739   can be used to reveal up to 64k of memory to a connected client or
7740   server.
7741
7742   Thanks for Neel Mehta of Google Security for discovering this bug and to
7743   Adam Langley <agl@chromium.org> and Bodo Moeller <bmoeller@acm.org> for
7744   preparing the fix ([CVE-2014-0160])
7745
7746   *Adam Langley, Bodo Moeller*
7747
7748 * Fix for the attack described in the paper "Recovering OpenSSL
7749   ECDSA Nonces Using the FLUSH+RELOAD Cache Side-channel Attack"
7750   by Yuval Yarom and Naomi Benger. Details can be obtained from:
7751   <http://eprint.iacr.org/2014/140>
7752
7753   Thanks to Yuval Yarom and Naomi Benger for discovering this
7754   flaw and to Yuval Yarom for supplying a fix ([CVE-2014-0076])
7755
7756   *Yuval Yarom and Naomi Benger*
7757
7758 * TLS pad extension: draft-agl-tls-padding-03
7759
7760   Workaround for the "TLS hang bug" (see FAQ and PR#2771): if the
7761   TLS client Hello record length value would otherwise be > 255 and
7762   less that 512 pad with a dummy extension containing zeroes so it
7763   is at least 512 bytes long.
7764
7765   *Adam Langley, Steve Henson*
7766
7767### Changes between 1.0.1e and 1.0.1f [6 Jan 2014]
7768
7769 * Fix for TLS record tampering bug. A carefully crafted invalid
7770   handshake could crash OpenSSL with a NULL pointer exception.
7771   Thanks to Anton Johansson for reporting this issues.
7772   ([CVE-2013-4353])
7773
7774 * Keep original DTLS digest and encryption contexts in retransmission
7775   structures so we can use the previous session parameters if they need
7776   to be resent. ([CVE-2013-6450])
7777
7778   *Steve Henson*
7779
7780 * Add option SSL_OP_SAFARI_ECDHE_ECDSA_BUG (part of SSL_OP_ALL) which
7781   avoids preferring ECDHE-ECDSA ciphers when the client appears to be
7782   Safari on OS X.  Safari on OS X 10.8..10.8.3 advertises support for
7783   several ECDHE-ECDSA ciphers, but fails to negotiate them.  The bug
7784   is fixed in OS X 10.8.4, but Apple have ruled out both hot fixing
7785   10.8..10.8.3 and forcing users to upgrade to 10.8.4 or newer.
7786
7787   *Rob Stradling, Adam Langley*
7788
7789### Changes between 1.0.1d and 1.0.1e [11 Feb 2013]
7790
7791 * Correct fix for CVE-2013-0169. The original didn't work on AES-NI
7792   supporting platforms or when small records were transferred.
7793
7794   *Andy Polyakov, Steve Henson*
7795
7796### Changes between 1.0.1c and 1.0.1d [5 Feb 2013]
7797
7798 * Make the decoding of SSLv3, TLS and DTLS CBC records constant time.
7799
7800   This addresses the flaw in CBC record processing discovered by
7801   Nadhem Alfardan and Kenny Paterson. Details of this attack can be found
7802   at: <http://www.isg.rhul.ac.uk/tls/>
7803
7804   Thanks go to Nadhem Alfardan and Kenny Paterson of the Information
7805   Security Group at Royal Holloway, University of London
7806   (www.isg.rhul.ac.uk) for discovering this flaw and Adam Langley and
7807   Emilia Käsper for the initial patch.
7808   ([CVE-2013-0169])
7809
7810   *Emilia Käsper, Adam Langley, Ben Laurie, Andy Polyakov, Steve Henson*
7811
7812 * Fix flaw in AESNI handling of TLS 1.2 and 1.1 records for CBC mode
7813   ciphersuites which can be exploited in a denial of service attack.
7814   Thanks go to and to Adam Langley <agl@chromium.org> for discovering
7815   and detecting this bug and to Wolfgang Ettlinger
7816   <wolfgang.ettlinger@gmail.com> for independently discovering this issue.
7817   ([CVE-2012-2686])
7818
7819   *Adam Langley*
7820
7821 * Return an error when checking OCSP signatures when key is NULL.
7822   This fixes a DoS attack. ([CVE-2013-0166])
7823
7824   *Steve Henson*
7825
7826 * Make openssl verify return errors.
7827
7828   *Chris Palmer <palmer@google.com> and Ben Laurie*
7829
7830 * Call OCSP Stapling callback after ciphersuite has been chosen, so
7831   the right response is stapled. Also change SSL_get_certificate()
7832   so it returns the certificate actually sent.
7833   See <http://rt.openssl.org/Ticket/Display.html?id=2836>.
7834
7835   *Rob Stradling <rob.stradling@comodo.com>*
7836
7837 * Fix possible deadlock when decoding public keys.
7838
7839   *Steve Henson*
7840
7841 * Don't use TLS 1.0 record version number in initial client hello
7842   if renegotiating.
7843
7844   *Steve Henson*
7845
7846### Changes between 1.0.1b and 1.0.1c [10 May 2012]
7847
7848 * Sanity check record length before skipping explicit IV in TLS
7849   1.2, 1.1 and DTLS to fix DoS attack.
7850
7851   Thanks to Codenomicon for discovering this issue using Fuzz-o-Matic
7852   fuzzing as a service testing platform.
7853   ([CVE-2012-2333])
7854
7855   *Steve Henson*
7856
7857 * Initialise tkeylen properly when encrypting CMS messages.
7858   Thanks to Solar Designer of Openwall for reporting this issue.
7859
7860   *Steve Henson*
7861
7862 * In FIPS mode don't try to use composite ciphers as they are not
7863   approved.
7864
7865   *Steve Henson*
7866
7867### Changes between 1.0.1a and 1.0.1b [26 Apr 2012]
7868
7869 * OpenSSL 1.0.0 sets SSL_OP_ALL to 0x80000FFFL and OpenSSL 1.0.1 and
7870   1.0.1a set SSL_OP_NO_TLSv1_1 to 0x00000400L which would unfortunately
7871   mean any application compiled against OpenSSL 1.0.0 headers setting
7872   SSL_OP_ALL would also set SSL_OP_NO_TLSv1_1, unintentionally disabling
7873   TLS 1.1 also. Fix this by changing the value of SSL_OP_NO_TLSv1_1 to
7874   0x10000000L Any application which was previously compiled against
7875   OpenSSL 1.0.1 or 1.0.1a headers and which cares about SSL_OP_NO_TLSv1_1
7876   will need to be recompiled as a result. Letting be results in
7877   inability to disable specifically TLS 1.1 and in client context,
7878   in unlike event, limit maximum offered version to TLS 1.0 [see below].
7879
7880   *Steve Henson*
7881
7882 * In order to ensure interoperability SSL_OP_NO_protocolX does not
7883   disable just protocol X, but all protocols above X *if* there are
7884   protocols *below* X still enabled. In more practical terms it means
7885   that if application wants to disable TLS1.0 in favor of TLS1.1 and
7886   above, it's not sufficient to pass `SSL_OP_NO_TLSv1`, one has to pass
7887   `SSL_OP_NO_TLSv1|SSL_OP_NO_SSLv3|SSL_OP_NO_SSLv2`. This applies to
7888   client side.
7889
7890   *Andy Polyakov*
7891
7892### Changes between 1.0.1 and 1.0.1a [19 Apr 2012]
7893
7894 * Check for potentially exploitable overflows in asn1_d2i_read_bio
7895   BUF_mem_grow and BUF_mem_grow_clean. Refuse attempts to shrink buffer
7896   in CRYPTO_realloc_clean.
7897
7898   Thanks to Tavis Ormandy, Google Security Team, for discovering this
7899   issue and to Adam Langley <agl@chromium.org> for fixing it.
7900   ([CVE-2012-2110])
7901
7902   *Adam Langley (Google), Tavis Ormandy, Google Security Team*
7903
7904 * Don't allow TLS 1.2 SHA-256 ciphersuites in TLS 1.0, 1.1 connections.
7905
7906   *Adam Langley*
7907
7908 * Workarounds for some broken servers that "hang" if a client hello
7909   record length exceeds 255 bytes.
7910
7911   1. Do not use record version number > TLS 1.0 in initial client
7912      hello: some (but not all) hanging servers will now work.
7913   2. If we set OPENSSL_MAX_TLS1_2_CIPHER_LENGTH this will truncate
7914      the number of ciphers sent in the client hello. This should be
7915      set to an even number, such as 50, for example by passing:
7916      -DOPENSSL_MAX_TLS1_2_CIPHER_LENGTH=50 to config or Configure.
7917      Most broken servers should now work.
7918   3. If all else fails setting OPENSSL_NO_TLS1_2_CLIENT will disable
7919      TLS 1.2 client support entirely.
7920
7921   *Steve Henson*
7922
7923 * Fix SEGV in Vector Permutation AES module observed in OpenSSH.
7924
7925   *Andy Polyakov*
7926
7927### Changes between 1.0.0h and 1.0.1  [14 Mar 2012]
7928
7929 * Add compatibility with old MDC2 signatures which use an ASN1 OCTET
7930   STRING form instead of a DigestInfo.
7931
7932   *Steve Henson*
7933
7934 * The format used for MDC2 RSA signatures is inconsistent between EVP
7935   and the RSA_sign/RSA_verify functions. This was made more apparent when
7936   OpenSSL used RSA_sign/RSA_verify for some RSA signatures in particular
7937   those which went through EVP_PKEY_METHOD in 1.0.0 and later. Detect
7938   the correct format in RSA_verify so both forms transparently work.
7939
7940   *Steve Henson*
7941
7942 * Some servers which support TLS 1.0 can choke if we initially indicate
7943   support for TLS 1.2 and later renegotiate using TLS 1.0 in the RSA
7944   encrypted premaster secret. As a workaround use the maximum permitted
7945   client version in client hello, this should keep such servers happy
7946   and still work with previous versions of OpenSSL.
7947
7948   *Steve Henson*
7949
7950 * Add support for TLS/DTLS heartbeats.
7951
7952   *Robin Seggelmann <seggelmann@fh-muenster.de>*
7953
7954 * Add support for SCTP.
7955
7956   *Robin Seggelmann <seggelmann@fh-muenster.de>*
7957
7958 * Improved PRNG seeding for VOS.
7959
7960   *Paul Green <Paul.Green@stratus.com>*
7961
7962 * Extensive assembler packs updates, most notably:
7963
7964   - x86[_64]:     AES-NI, PCLMULQDQ, RDRAND support;
7965   - x86[_64]:     SSSE3 support (SHA1, vector-permutation AES);
7966   - x86_64:       bit-sliced AES implementation;
7967   - ARM:          NEON support, contemporary platforms optimizations;
7968   - s390x:        z196 support;
7969   - `*`:            GHASH and GF(2^m) multiplication implementations;
7970
7971   *Andy Polyakov*
7972
7973 * Make TLS-SRP code conformant with RFC 5054 API cleanup
7974   (removal of unnecessary code)
7975
7976   *Peter Sylvester <peter.sylvester@edelweb.fr>*
7977
7978 * Add TLS key material exporter from RFC 5705.
7979
7980   *Eric Rescorla*
7981
7982 * Add DTLS-SRTP negotiation from RFC 5764.
7983
7984   *Eric Rescorla*
7985
7986 * Add Next Protocol Negotiation,
7987   <http://tools.ietf.org/html/draft-agl-tls-nextprotoneg-00>. Can be
7988   disabled with a no-npn flag to config or Configure. Code donated
7989   by Google.
7990
7991   *Adam Langley <agl@google.com> and Ben Laurie*
7992
7993 * Add optional 64-bit optimized implementations of elliptic curves NIST-P224,
7994   NIST-P256, NIST-P521, with constant-time single point multiplication on
7995   typical inputs. Compiler support for the nonstandard type `__uint128_t` is
7996   required to use this (present in gcc 4.4 and later, for 64-bit builds).
7997   Code made available under Apache License version 2.0.
7998
7999   Specify "enable-ec_nistp_64_gcc_128" on the Configure (or config) command
8000   line to include this in your build of OpenSSL, and run "make depend" (or
8001   "make update"). This enables the following EC_METHODs:
8002
8003           EC_GFp_nistp224_method()
8004           EC_GFp_nistp256_method()
8005           EC_GFp_nistp521_method()
8006
8007   EC_GROUP_new_by_curve_name() will automatically use these (while
8008   EC_GROUP_new_curve_GFp() currently prefers the more flexible
8009   implementations).
8010
8011   *Emilia Käsper, Adam Langley, Bodo Moeller (Google)*
8012
8013 * Use type ossl_ssize_t instead of ssize_t which isn't available on
8014   all platforms. Move ssize_t definition from e_os.h to the public
8015   header file e_os2.h as it now appears in public header file cms.h
8016
8017   *Steve Henson*
8018
8019 * New -sigopt option to the ca, req and x509 utilities. Additional
8020   signature parameters can be passed using this option and in
8021   particular PSS.
8022
8023   *Steve Henson*
8024
8025 * Add RSA PSS signing function. This will generate and set the
8026   appropriate AlgorithmIdentifiers for PSS based on those in the
8027   corresponding EVP_MD_CTX structure. No application support yet.
8028
8029   *Steve Henson*
8030
8031 * Support for companion algorithm specific ASN1 signing routines.
8032   New function ASN1_item_sign_ctx() signs a pre-initialised
8033   EVP_MD_CTX structure and sets AlgorithmIdentifiers based on
8034   the appropriate parameters.
8035
8036   *Steve Henson*
8037
8038 * Add new algorithm specific ASN1 verification initialisation function
8039   to EVP_PKEY_ASN1_METHOD: this is not in EVP_PKEY_METHOD since the ASN1
8040   handling will be the same no matter what EVP_PKEY_METHOD is used.
8041   Add a PSS handler to support verification of PSS signatures: checked
8042   against a number of sample certificates.
8043
8044   *Steve Henson*
8045
8046 * Add signature printing for PSS. Add PSS OIDs.
8047
8048   *Steve Henson, Martin Kaiser <lists@kaiser.cx>*
8049
8050 * Add algorithm specific signature printing. An individual ASN1 method
8051   can now print out signatures instead of the standard hex dump.
8052
8053   More complex signatures (e.g. PSS) can print out more meaningful
8054   information. Include DSA version that prints out the signature
8055   parameters r, s.
8056
8057   *Steve Henson*
8058
8059 * Password based recipient info support for CMS library: implementing
8060   RFC3211.
8061
8062   *Steve Henson*
8063
8064 * Split password based encryption into PBES2 and PBKDF2 functions. This
8065   neatly separates the code into cipher and PBE sections and is required
8066   for some algorithms that split PBES2 into separate pieces (such as
8067   password based CMS).
8068
8069   *Steve Henson*
8070
8071 * Session-handling fixes:
8072   - Fix handling of connections that are resuming with a session ID,
8073     but also support Session Tickets.
8074   - Fix a bug that suppressed issuing of a new ticket if the client
8075     presented a ticket with an expired session.
8076   - Try to set the ticket lifetime hint to something reasonable.
8077   - Make tickets shorter by excluding irrelevant information.
8078   - On the client side, don't ignore renewed tickets.
8079
8080   *Adam Langley, Bodo Moeller (Google)*
8081
8082 * Fix PSK session representation.
8083
8084   *Bodo Moeller*
8085
8086 * Add RC4-MD5 and AESNI-SHA1 "stitched" implementations.
8087
8088   This work was sponsored by Intel.
8089
8090   *Andy Polyakov*
8091
8092 * Add GCM support to TLS library. Some custom code is needed to split
8093   the IV between the fixed (from PRF) and explicit (from TLS record)
8094   portions. This adds all GCM ciphersuites supported by RFC5288 and
8095   RFC5289. Generalise some `AES*` cipherstrings to include GCM and
8096   add a special AESGCM string for GCM only.
8097
8098   *Steve Henson*
8099
8100 * Expand range of ctrls for AES GCM. Permit setting invocation
8101   field on decrypt and retrieval of invocation field only on encrypt.
8102
8103   *Steve Henson*
8104
8105 * Add HMAC ECC ciphersuites from RFC5289. Include SHA384 PRF support.
8106   As required by RFC5289 these ciphersuites cannot be used if for
8107   versions of TLS earlier than 1.2.
8108
8109   *Steve Henson*
8110
8111 * For FIPS capable OpenSSL interpret a NULL default public key method
8112   as unset and return the appropriate default but do *not* set the default.
8113   This means we can return the appropriate method in applications that
8114   switch between FIPS and non-FIPS modes.
8115
8116   *Steve Henson*
8117
8118 * Redirect HMAC and CMAC operations to FIPS module in FIPS mode. If an
8119   ENGINE is used then we cannot handle that in the FIPS module so we
8120   keep original code iff non-FIPS operations are allowed.
8121
8122   *Steve Henson*
8123
8124 * Add -attime option to openssl utilities.
8125
8126   *Peter Eckersley <pde@eff.org>, Ben Laurie and Steve Henson*
8127
8128 * Redirect DSA and DH operations to FIPS module in FIPS mode.
8129
8130   *Steve Henson*
8131
8132 * Redirect ECDSA and ECDH operations to FIPS module in FIPS mode. Also use
8133   FIPS EC methods unconditionally for now.
8134
8135   *Steve Henson*
8136
8137 * New build option no-ec2m to disable characteristic 2 code.
8138
8139   *Steve Henson*
8140
8141 * Backport libcrypto audit of return value checking from 1.1.0-dev; not
8142   all cases can be covered as some introduce binary incompatibilities.
8143
8144   *Steve Henson*
8145
8146 * Redirect RSA operations to FIPS module including keygen,
8147   encrypt, decrypt, sign and verify. Block use of non FIPS RSA methods.
8148
8149   *Steve Henson*
8150
8151 * Add similar low-level API blocking to ciphers.
8152
8153   *Steve Henson*
8154
8155 * low-level digest APIs are not approved in FIPS mode: any attempt
8156   to use these will cause a fatal error. Applications that *really* want
8157   to use them can use the `private_*` version instead.
8158
8159   *Steve Henson*
8160
8161 * Redirect cipher operations to FIPS module for FIPS builds.
8162
8163   *Steve Henson*
8164
8165 * Redirect digest operations to FIPS module for FIPS builds.
8166
8167   *Steve Henson*
8168
8169 * Update build system to add "fips" flag which will link in fipscanister.o
8170   for static and shared library builds embedding a signature if needed.
8171
8172   *Steve Henson*
8173
8174 * Output TLS supported curves in preference order instead of numerical
8175   order. This is currently hardcoded for the highest order curves first.
8176   This should be configurable so applications can judge speed vs strength.
8177
8178   *Steve Henson*
8179
8180 * Add TLS v1.2 server support for client authentication.
8181
8182   *Steve Henson*
8183
8184 * Add support for FIPS mode in ssl library: disable SSLv3, non-FIPS ciphers
8185   and enable MD5.
8186
8187   *Steve Henson*
8188
8189 * Functions FIPS_mode_set() and FIPS_mode() which call the underlying
8190   FIPS modules versions.
8191
8192   *Steve Henson*
8193
8194 * Add TLS v1.2 client side support for client authentication. Keep cache
8195   of handshake records longer as we don't know the hash algorithm to use
8196   until after the certificate request message is received.
8197
8198   *Steve Henson*
8199
8200 * Initial TLS v1.2 client support. Add a default signature algorithms
8201   extension including all the algorithms we support. Parse new signature
8202   format in client key exchange. Relax some ECC signing restrictions for
8203   TLS v1.2 as indicated in RFC5246.
8204
8205   *Steve Henson*
8206
8207 * Add server support for TLS v1.2 signature algorithms extension. Switch
8208   to new signature format when needed using client digest preference.
8209   All server ciphersuites should now work correctly in TLS v1.2. No client
8210   support yet and no support for client certificates.
8211
8212   *Steve Henson*
8213
8214 * Initial TLS v1.2 support. Add new SHA256 digest to ssl code, switch
8215   to SHA256 for PRF when using TLS v1.2 and later. Add new SHA256 based
8216   ciphersuites. At present only RSA key exchange ciphersuites work with
8217   TLS v1.2. Add new option for TLS v1.2 replacing the old and obsolete
8218   SSL_OP_PKCS1_CHECK flags with SSL_OP_NO_TLSv1_2. New TLSv1.2 methods
8219   and version checking.
8220
8221   *Steve Henson*
8222
8223 * New option OPENSSL_NO_SSL_INTERN. If an application can be compiled
8224   with this defined it will not be affected by any changes to ssl internal
8225   structures. Add several utility functions to allow openssl application
8226   to work with OPENSSL_NO_SSL_INTERN defined.
8227
8228   *Steve Henson*
8229
8230 * A long standing patch to add support for SRP from EdelWeb (Peter
8231   Sylvester and Christophe Renou) was integrated.
8232   *Christophe Renou <christophe.renou@edelweb.fr>, Peter Sylvester
8233   <peter.sylvester@edelweb.fr>, Tom Wu <tjw@cs.stanford.edu>, and
8234   Ben Laurie*
8235
8236 * Add functions to copy EVP_PKEY_METHOD and retrieve flags and id.
8237
8238   *Steve Henson*
8239
8240 * Permit abbreviated handshakes when renegotiating using the function
8241   SSL_renegotiate_abbreviated().
8242
8243   *Robin Seggelmann <seggelmann@fh-muenster.de>*
8244
8245 * Add call to ENGINE_register_all_complete() to
8246   ENGINE_load_builtin_engines(), so some implementations get used
8247   automatically instead of needing explicit application support.
8248
8249   *Steve Henson*
8250
8251 * Add support for TLS key exporter as described in RFC5705.
8252
8253   *Robin Seggelmann <seggelmann@fh-muenster.de>, Steve Henson*
8254
8255 * Initial TLSv1.1 support. Since TLSv1.1 is very similar to TLS v1.0 only
8256   a few changes are required:
8257
8258     Add SSL_OP_NO_TLSv1_1 flag.
8259     Add TLSv1_1 methods.
8260     Update version checking logic to handle version 1.1.
8261     Add explicit IV handling (ported from DTLS code).
8262     Add command line options to s_client/s_server.
8263
8264   *Steve Henson*
8265
8266OpenSSL 1.0.0
8267-------------
8268
8269### Changes between 1.0.0s and 1.0.0t [3 Dec 2015]
8270
8271 * X509_ATTRIBUTE memory leak
8272
8273   When presented with a malformed X509_ATTRIBUTE structure OpenSSL will leak
8274   memory. This structure is used by the PKCS#7 and CMS routines so any
8275   application which reads PKCS#7 or CMS data from untrusted sources is
8276   affected. SSL/TLS is not affected.
8277
8278   This issue was reported to OpenSSL by Adam Langley (Google/BoringSSL) using
8279   libFuzzer.
8280   ([CVE-2015-3195])
8281
8282   *Stephen Henson*
8283
8284 * Race condition handling PSK identify hint
8285
8286   If PSK identity hints are received by a multi-threaded client then
8287   the values are wrongly updated in the parent SSL_CTX structure. This can
8288   result in a race condition potentially leading to a double free of the
8289   identify hint data.
8290   ([CVE-2015-3196])
8291
8292   *Stephen Henson*
8293
8294### Changes between 1.0.0r and 1.0.0s [11 Jun 2015]
8295
8296 * Malformed ECParameters causes infinite loop
8297
8298   When processing an ECParameters structure OpenSSL enters an infinite loop
8299   if the curve specified is over a specially malformed binary polynomial
8300   field.
8301
8302   This can be used to perform denial of service against any
8303   system which processes public keys, certificate requests or
8304   certificates.  This includes TLS clients and TLS servers with
8305   client authentication enabled.
8306
8307   This issue was reported to OpenSSL by Joseph Barr-Pixton.
8308   ([CVE-2015-1788])
8309
8310   *Andy Polyakov*
8311
8312 * Exploitable out-of-bounds read in X509_cmp_time
8313
8314   X509_cmp_time does not properly check the length of the ASN1_TIME
8315   string and can read a few bytes out of bounds. In addition,
8316   X509_cmp_time accepts an arbitrary number of fractional seconds in the
8317   time string.
8318
8319   An attacker can use this to craft malformed certificates and CRLs of
8320   various sizes and potentially cause a segmentation fault, resulting in
8321   a DoS on applications that verify certificates or CRLs. TLS clients
8322   that verify CRLs are affected. TLS clients and servers with client
8323   authentication enabled may be affected if they use custom verification
8324   callbacks.
8325
8326   This issue was reported to OpenSSL by Robert Swiecki (Google), and
8327   independently by Hanno Böck.
8328   ([CVE-2015-1789])
8329
8330   *Emilia Käsper*
8331
8332 * PKCS7 crash with missing EnvelopedContent
8333
8334   The PKCS#7 parsing code does not handle missing inner EncryptedContent
8335   correctly. An attacker can craft malformed ASN.1-encoded PKCS#7 blobs
8336   with missing content and trigger a NULL pointer dereference on parsing.
8337
8338   Applications that decrypt PKCS#7 data or otherwise parse PKCS#7
8339   structures from untrusted sources are affected. OpenSSL clients and
8340   servers are not affected.
8341
8342   This issue was reported to OpenSSL by Michal Zalewski (Google).
8343   ([CVE-2015-1790])
8344
8345   *Emilia Käsper*
8346
8347 * CMS verify infinite loop with unknown hash function
8348
8349   When verifying a signedData message the CMS code can enter an infinite loop
8350   if presented with an unknown hash function OID. This can be used to perform
8351   denial of service against any system which verifies signedData messages using
8352   the CMS code.
8353   This issue was reported to OpenSSL by Johannes Bauer.
8354   ([CVE-2015-1792])
8355
8356   *Stephen Henson*
8357
8358 * Race condition handling NewSessionTicket
8359
8360   If a NewSessionTicket is received by a multi-threaded client when attempting to
8361   reuse a previous ticket then a race condition can occur potentially leading to
8362   a double free of the ticket data.
8363   ([CVE-2015-1791])
8364
8365   *Matt Caswell*
8366
8367### Changes between 1.0.0q and 1.0.0r [19 Mar 2015]
8368
8369 * Segmentation fault in ASN1_TYPE_cmp fix
8370
8371   The function ASN1_TYPE_cmp will crash with an invalid read if an attempt is
8372   made to compare ASN.1 boolean types. Since ASN1_TYPE_cmp is used to check
8373   certificate signature algorithm consistency this can be used to crash any
8374   certificate verification operation and exploited in a DoS attack. Any
8375   application which performs certificate verification is vulnerable including
8376   OpenSSL clients and servers which enable client authentication.
8377   ([CVE-2015-0286])
8378
8379   *Stephen Henson*
8380
8381 * ASN.1 structure reuse memory corruption fix
8382
8383   Reusing a structure in ASN.1 parsing may allow an attacker to cause
8384   memory corruption via an invalid write. Such reuse is and has been
8385   strongly discouraged and is believed to be rare.
8386
8387   Applications that parse structures containing CHOICE or ANY DEFINED BY
8388   components may be affected. Certificate parsing (d2i_X509 and related
8389   functions) are however not affected. OpenSSL clients and servers are
8390   not affected.
8391   ([CVE-2015-0287])
8392
8393   *Stephen Henson*
8394
8395 * PKCS7 NULL pointer dereferences fix
8396
8397   The PKCS#7 parsing code does not handle missing outer ContentInfo
8398   correctly. An attacker can craft malformed ASN.1-encoded PKCS#7 blobs with
8399   missing content and trigger a NULL pointer dereference on parsing.
8400
8401   Applications that verify PKCS#7 signatures, decrypt PKCS#7 data or
8402   otherwise parse PKCS#7 structures from untrusted sources are
8403   affected. OpenSSL clients and servers are not affected.
8404
8405   This issue was reported to OpenSSL by Michal Zalewski (Google).
8406   ([CVE-2015-0289])
8407
8408   *Emilia Käsper*
8409
8410 * DoS via reachable assert in SSLv2 servers fix
8411
8412   A malicious client can trigger an OPENSSL_assert (i.e., an abort) in
8413   servers that both support SSLv2 and enable export cipher suites by sending
8414   a specially crafted SSLv2 CLIENT-MASTER-KEY message.
8415
8416   This issue was discovered by Sean Burford (Google) and Emilia Käsper
8417   (OpenSSL development team).
8418   ([CVE-2015-0293])
8419
8420   *Emilia Käsper*
8421
8422 * Use After Free following d2i_ECPrivatekey error fix
8423
8424   A malformed EC private key file consumed via the d2i_ECPrivateKey function
8425   could cause a use after free condition. This, in turn, could cause a double
8426   free in several private key parsing functions (such as d2i_PrivateKey
8427   or EVP_PKCS82PKEY) and could lead to a DoS attack or memory corruption
8428   for applications that receive EC private keys from untrusted
8429   sources. This scenario is considered rare.
8430
8431   This issue was discovered by the BoringSSL project and fixed in their
8432   commit 517073cd4b.
8433   ([CVE-2015-0209])
8434
8435   *Matt Caswell*
8436
8437 * X509_to_X509_REQ NULL pointer deref fix
8438
8439   The function X509_to_X509_REQ will crash with a NULL pointer dereference if
8440   the certificate key is invalid. This function is rarely used in practice.
8441
8442   This issue was discovered by Brian Carpenter.
8443   ([CVE-2015-0288])
8444
8445   *Stephen Henson*
8446
8447 * Removed the export ciphers from the DEFAULT ciphers
8448
8449   *Kurt Roeckx*
8450
8451### Changes between 1.0.0p and 1.0.0q [15 Jan 2015]
8452
8453 * Build fixes for the Windows and OpenVMS platforms
8454
8455   *Matt Caswell and Richard Levitte*
8456
8457### Changes between 1.0.0o and 1.0.0p [8 Jan 2015]
8458
8459 * Fix DTLS segmentation fault in dtls1_get_record. A carefully crafted DTLS
8460   message can cause a segmentation fault in OpenSSL due to a NULL pointer
8461   dereference. This could lead to a Denial Of Service attack. Thanks to
8462   Markus Stenberg of Cisco Systems, Inc. for reporting this issue.
8463   ([CVE-2014-3571])
8464
8465   *Steve Henson*
8466
8467 * Fix DTLS memory leak in dtls1_buffer_record. A memory leak can occur in the
8468   dtls1_buffer_record function under certain conditions. In particular this
8469   could occur if an attacker sent repeated DTLS records with the same
8470   sequence number but for the next epoch. The memory leak could be exploited
8471   by an attacker in a Denial of Service attack through memory exhaustion.
8472   Thanks to Chris Mueller for reporting this issue.
8473   ([CVE-2015-0206])
8474
8475   *Matt Caswell*
8476
8477 * Fix issue where no-ssl3 configuration sets method to NULL. When openssl is
8478   built with the no-ssl3 option and a SSL v3 ClientHello is received the ssl
8479   method would be set to NULL which could later result in a NULL pointer
8480   dereference. Thanks to Frank Schmirler for reporting this issue.
8481   ([CVE-2014-3569])
8482
8483   *Kurt Roeckx*
8484
8485 * Abort handshake if server key exchange message is omitted for ephemeral
8486   ECDH ciphersuites.
8487
8488   Thanks to Karthikeyan Bhargavan of the PROSECCO team at INRIA for
8489   reporting this issue.
8490   ([CVE-2014-3572])
8491
8492   *Steve Henson*
8493
8494 * Remove non-export ephemeral RSA code on client and server. This code
8495   violated the TLS standard by allowing the use of temporary RSA keys in
8496   non-export ciphersuites and could be used by a server to effectively
8497   downgrade the RSA key length used to a value smaller than the server
8498   certificate. Thanks for Karthikeyan Bhargavan of the PROSECCO team at
8499   INRIA or reporting this issue.
8500   ([CVE-2015-0204])
8501
8502   *Steve Henson*
8503
8504 * Fixed issue where DH client certificates are accepted without verification.
8505   An OpenSSL server will accept a DH certificate for client authentication
8506   without the certificate verify message. This effectively allows a client to
8507   authenticate without the use of a private key. This only affects servers
8508   which trust a client certificate authority which issues certificates
8509   containing DH keys: these are extremely rare and hardly ever encountered.
8510   Thanks for Karthikeyan Bhargavan of the PROSECCO team at INRIA or reporting
8511   this issue.
8512   ([CVE-2015-0205])
8513
8514   *Steve Henson*
8515
8516 * Correct Bignum squaring. Bignum squaring (BN_sqr) may produce incorrect
8517   results on some platforms, including x86_64. This bug occurs at random
8518   with a very low probability, and is not known to be exploitable in any
8519   way, though its exact impact is difficult to determine. Thanks to Pieter
8520   Wuille (Blockstream) who reported this issue and also suggested an initial
8521   fix. Further analysis was conducted by the OpenSSL development team and
8522   Adam Langley of Google. The final fix was developed by Andy Polyakov of
8523   the OpenSSL core team.
8524   ([CVE-2014-3570])
8525
8526   *Andy Polyakov*
8527
8528 * Fix various certificate fingerprint issues.
8529
8530   By using non-DER or invalid encodings outside the signed portion of a
8531   certificate the fingerprint can be changed without breaking the signature.
8532   Although no details of the signed portion of the certificate can be changed
8533   this can cause problems with some applications: e.g. those using the
8534   certificate fingerprint for blacklists.
8535
8536   1. Reject signatures with non zero unused bits.
8537
8538   If the BIT STRING containing the signature has non zero unused bits reject
8539   the signature. All current signature algorithms require zero unused bits.
8540
8541   2. Check certificate algorithm consistency.
8542
8543   Check the AlgorithmIdentifier inside TBS matches the one in the
8544   certificate signature. NB: this will result in signature failure
8545   errors for some broken certificates.
8546
8547   Thanks to Konrad Kraszewski from Google for reporting this issue.
8548
8549   3. Check DSA/ECDSA signatures use DER.
8550
8551   Reencode DSA/ECDSA signatures and compare with the original received
8552   signature. Return an error if there is a mismatch.
8553
8554   This will reject various cases including garbage after signature
8555   (thanks to Antti Karjalainen and Tuomo Untinen from the Codenomicon CROSS
8556   program for discovering this case) and use of BER or invalid ASN.1 INTEGERs
8557   (negative or with leading zeroes).
8558
8559   Further analysis was conducted and fixes were developed by Stephen Henson
8560   of the OpenSSL core team.
8561
8562   ([CVE-2014-8275])
8563
8564   *Steve Henson*
8565
8566### Changes between 1.0.0n and 1.0.0o [15 Oct 2014]
8567
8568 * Session Ticket Memory Leak.
8569
8570   When an OpenSSL SSL/TLS/DTLS server receives a session ticket the
8571   integrity of that ticket is first verified. In the event of a session
8572   ticket integrity check failing, OpenSSL will fail to free memory
8573   causing a memory leak. By sending a large number of invalid session
8574   tickets an attacker could exploit this issue in a Denial Of Service
8575   attack.
8576   ([CVE-2014-3567])
8577
8578   *Steve Henson*
8579
8580 * Build option no-ssl3 is incomplete.
8581
8582   When OpenSSL is configured with "no-ssl3" as a build option, servers
8583   could accept and complete a SSL 3.0 handshake, and clients could be
8584   configured to send them.
8585   ([CVE-2014-3568])
8586
8587   *Akamai and the OpenSSL team*
8588
8589 * Add support for TLS_FALLBACK_SCSV.
8590   Client applications doing fallback retries should call
8591   SSL_set_mode(s, SSL_MODE_SEND_FALLBACK_SCSV).
8592   ([CVE-2014-3566])
8593
8594   *Adam Langley, Bodo Moeller*
8595
8596 * Add additional DigestInfo checks.
8597
8598   Reencode DigestInto in DER and check against the original when
8599   verifying RSA signature: this will reject any improperly encoded
8600   DigestInfo structures.
8601
8602   Note: this is a precautionary measure and no attacks are currently known.
8603
8604   *Steve Henson*
8605
8606### Changes between 1.0.0m and 1.0.0n [6 Aug 2014]
8607
8608 * OpenSSL DTLS clients enabling anonymous (EC)DH ciphersuites are subject
8609   to a denial of service attack. A malicious server can crash the client
8610   with a null pointer dereference (read) by specifying an anonymous (EC)DH
8611   ciphersuite and sending carefully crafted handshake messages.
8612
8613   Thanks to Felix Gröbert (Google) for discovering and researching this
8614   issue.
8615   ([CVE-2014-3510])
8616
8617   *Emilia Käsper*
8618
8619 * By sending carefully crafted DTLS packets an attacker could cause openssl
8620   to leak memory. This can be exploited through a Denial of Service attack.
8621   Thanks to Adam Langley for discovering and researching this issue.
8622   ([CVE-2014-3507])
8623
8624   *Adam Langley*
8625
8626 * An attacker can force openssl to consume large amounts of memory whilst
8627   processing DTLS handshake messages. This can be exploited through a
8628   Denial of Service attack.
8629   Thanks to Adam Langley for discovering and researching this issue.
8630   ([CVE-2014-3506])
8631
8632   *Adam Langley*
8633
8634 * An attacker can force an error condition which causes openssl to crash
8635   whilst processing DTLS packets due to memory being freed twice. This
8636   can be exploited through a Denial of Service attack.
8637   Thanks to Adam Langley and Wan-Teh Chang for discovering and researching
8638   this issue.
8639   ([CVE-2014-3505])
8640
8641   *Adam Langley*
8642
8643 * If a multithreaded client connects to a malicious server using a resumed
8644   session and the server sends an ec point format extension it could write
8645   up to 255 bytes to freed memory.
8646
8647   Thanks to Gabor Tyukasz (LogMeIn Inc) for discovering and researching this
8648   issue.
8649   ([CVE-2014-3509])
8650
8651   *Gabor Tyukasz*
8652
8653 * A flaw in OBJ_obj2txt may cause pretty printing functions such as
8654   X509_name_oneline, X509_name_print_ex et al. to leak some information
8655   from the stack. Applications may be affected if they echo pretty printing
8656   output to the attacker.
8657
8658   Thanks to Ivan Fratric (Google) for discovering this issue.
8659   ([CVE-2014-3508])
8660
8661   *Emilia Käsper, and Steve Henson*
8662
8663 * Fix ec_GFp_simple_points_make_affine (thus, EC_POINTs_mul etc.)
8664   for corner cases. (Certain input points at infinity could lead to
8665   bogus results, with non-infinity inputs mapped to infinity too.)
8666
8667   *Bodo Moeller*
8668
8669### Changes between 1.0.0l and 1.0.0m [5 Jun 2014]
8670
8671 * Fix for SSL/TLS MITM flaw. An attacker using a carefully crafted
8672   handshake can force the use of weak keying material in OpenSSL
8673   SSL/TLS clients and servers.
8674
8675   Thanks to KIKUCHI Masashi (Lepidum Co. Ltd.) for discovering and
8676   researching this issue. ([CVE-2014-0224])
8677
8678   *KIKUCHI Masashi, Steve Henson*
8679
8680 * Fix DTLS recursion flaw. By sending an invalid DTLS handshake to an
8681   OpenSSL DTLS client the code can be made to recurse eventually crashing
8682   in a DoS attack.
8683
8684   Thanks to Imre Rad (Search-Lab Ltd.) for discovering this issue.
8685   ([CVE-2014-0221])
8686
8687   *Imre Rad, Steve Henson*
8688
8689 * Fix DTLS invalid fragment vulnerability. A buffer overrun attack can
8690   be triggered by sending invalid DTLS fragments to an OpenSSL DTLS
8691   client or server. This is potentially exploitable to run arbitrary
8692   code on a vulnerable client or server.
8693
8694   Thanks to Jüri Aedla for reporting this issue. ([CVE-2014-0195])
8695
8696   *Jüri Aedla, Steve Henson*
8697
8698 * Fix bug in TLS code where clients enable anonymous ECDH ciphersuites
8699   are subject to a denial of service attack.
8700
8701   Thanks to Felix Gröbert and Ivan Fratric at Google for discovering
8702   this issue. ([CVE-2014-3470])
8703
8704   *Felix Gröbert, Ivan Fratric, Steve Henson*
8705
8706 * Harmonize version and its documentation. -f flag is used to display
8707   compilation flags.
8708
8709   *mancha <mancha1@zoho.com>*
8710
8711 * Fix eckey_priv_encode so it immediately returns an error upon a failure
8712   in i2d_ECPrivateKey.
8713
8714   *mancha <mancha1@zoho.com>*
8715
8716 * Fix some double frees. These are not thought to be exploitable.
8717
8718   *mancha <mancha1@zoho.com>*
8719
8720 * Fix for the attack described in the paper "Recovering OpenSSL
8721   ECDSA Nonces Using the FLUSH+RELOAD Cache Side-channel Attack"
8722   by Yuval Yarom and Naomi Benger. Details can be obtained from:
8723   <http://eprint.iacr.org/2014/140>
8724
8725   Thanks to Yuval Yarom and Naomi Benger for discovering this
8726   flaw and to Yuval Yarom for supplying a fix ([CVE-2014-0076])
8727
8728   *Yuval Yarom and Naomi Benger*
8729
8730### Changes between 1.0.0k and 1.0.0l [6 Jan 2014]
8731
8732 * Keep original DTLS digest and encryption contexts in retransmission
8733   structures so we can use the previous session parameters if they need
8734   to be resent. ([CVE-2013-6450])
8735
8736   *Steve Henson*
8737
8738 * Add option SSL_OP_SAFARI_ECDHE_ECDSA_BUG (part of SSL_OP_ALL) which
8739   avoids preferring ECDHE-ECDSA ciphers when the client appears to be
8740   Safari on OS X.  Safari on OS X 10.8..10.8.3 advertises support for
8741   several ECDHE-ECDSA ciphers, but fails to negotiate them.  The bug
8742   is fixed in OS X 10.8.4, but Apple have ruled out both hot fixing
8743   10.8..10.8.3 and forcing users to upgrade to 10.8.4 or newer.
8744
8745   *Rob Stradling, Adam Langley*
8746
8747### Changes between 1.0.0j and 1.0.0k [5 Feb 2013]
8748
8749 * Make the decoding of SSLv3, TLS and DTLS CBC records constant time.
8750
8751   This addresses the flaw in CBC record processing discovered by
8752   Nadhem Alfardan and Kenny Paterson. Details of this attack can be found
8753   at: <http://www.isg.rhul.ac.uk/tls/>
8754
8755   Thanks go to Nadhem Alfardan and Kenny Paterson of the Information
8756   Security Group at Royal Holloway, University of London
8757   (www.isg.rhul.ac.uk) for discovering this flaw and Adam Langley and
8758   Emilia Käsper for the initial patch.
8759   ([CVE-2013-0169])
8760
8761   *Emilia Käsper, Adam Langley, Ben Laurie, Andy Polyakov, Steve Henson*
8762
8763 * Return an error when checking OCSP signatures when key is NULL.
8764   This fixes a DoS attack. ([CVE-2013-0166])
8765
8766   *Steve Henson*
8767
8768 * Call OCSP Stapling callback after ciphersuite has been chosen, so
8769   the right response is stapled. Also change SSL_get_certificate()
8770   so it returns the certificate actually sent.
8771   See <http://rt.openssl.org/Ticket/Display.html?id=2836>.
8772   (This is a backport)
8773
8774   *Rob Stradling <rob.stradling@comodo.com>*
8775
8776 * Fix possible deadlock when decoding public keys.
8777
8778   *Steve Henson*
8779
8780### Changes between 1.0.0i and 1.0.0j [10 May 2012]
8781
8782[NB: OpenSSL 1.0.0i and later 1.0.0 patch levels were released after
8783OpenSSL 1.0.1.]
8784
8785 * Sanity check record length before skipping explicit IV in DTLS
8786   to fix DoS attack.
8787
8788   Thanks to Codenomicon for discovering this issue using Fuzz-o-Matic
8789   fuzzing as a service testing platform.
8790   ([CVE-2012-2333])
8791
8792   *Steve Henson*
8793
8794 * Initialise tkeylen properly when encrypting CMS messages.
8795   Thanks to Solar Designer of Openwall for reporting this issue.
8796
8797   *Steve Henson*
8798
8799### Changes between 1.0.0h and 1.0.0i [19 Apr 2012]
8800
8801 * Check for potentially exploitable overflows in asn1_d2i_read_bio
8802   BUF_mem_grow and BUF_mem_grow_clean. Refuse attempts to shrink buffer
8803   in CRYPTO_realloc_clean.
8804
8805   Thanks to Tavis Ormandy, Google Security Team, for discovering this
8806   issue and to Adam Langley <agl@chromium.org> for fixing it.
8807   ([CVE-2012-2110])
8808
8809   *Adam Langley (Google), Tavis Ormandy, Google Security Team*
8810
8811### Changes between 1.0.0g and 1.0.0h [12 Mar 2012]
8812
8813 * Fix MMA (Bleichenbacher's attack on PKCS #1 v1.5 RSA padding) weakness
8814   in CMS and PKCS7 code. When RSA decryption fails use a random key for
8815   content decryption and always return the same error. Note: this attack
8816   needs on average 2^20 messages so it only affects automated senders. The
8817   old behaviour can be re-enabled in the CMS code by setting the
8818   CMS_DEBUG_DECRYPT flag: this is useful for debugging and testing where
8819   an MMA defence is not necessary.
8820   Thanks to Ivan Nestlerode <inestlerode@us.ibm.com> for discovering
8821   this issue. ([CVE-2012-0884])
8822
8823   *Steve Henson*
8824
8825 * Fix CVE-2011-4619: make sure we really are receiving a
8826   client hello before rejecting multiple SGC restarts. Thanks to
8827   Ivan Nestlerode <inestlerode@us.ibm.com> for discovering this bug.
8828
8829   *Steve Henson*
8830
8831### Changes between 1.0.0f and 1.0.0g [18 Jan 2012]
8832
8833 * Fix for DTLS DoS issue introduced by fix for CVE-2011-4109.
8834   Thanks to Antonio Martin, Enterprise Secure Access Research and
8835   Development, Cisco Systems, Inc. for discovering this bug and
8836   preparing a fix. ([CVE-2012-0050])
8837
8838   *Antonio Martin*
8839
8840### Changes between 1.0.0e and 1.0.0f [4 Jan 2012]
8841
8842 * Nadhem Alfardan and Kenny Paterson have discovered an extension
8843   of the Vaudenay padding oracle attack on CBC mode encryption
8844   which enables an efficient plaintext recovery attack against
8845   the OpenSSL implementation of DTLS. Their attack exploits timing
8846   differences arising during decryption processing. A research
8847   paper describing this attack can be found at:
8848   <http://www.isg.rhul.ac.uk/~kp/dtls.pdf>
8849   Thanks go to Nadhem Alfardan and Kenny Paterson of the Information
8850   Security Group at Royal Holloway, University of London
8851   (www.isg.rhul.ac.uk) for discovering this flaw and to Robin Seggelmann
8852   <seggelmann@fh-muenster.de> and Michael Tuexen <tuexen@fh-muenster.de>
8853   for preparing the fix. ([CVE-2011-4108])
8854
8855   *Robin Seggelmann, Michael Tuexen*
8856
8857 * Clear bytes used for block padding of SSL 3.0 records.
8858   ([CVE-2011-4576])
8859
8860   *Adam Langley (Google)*
8861
8862 * Only allow one SGC handshake restart for SSL/TLS. Thanks to George
8863   Kadianakis <desnacked@gmail.com> for discovering this issue and
8864   Adam Langley for preparing the fix. ([CVE-2011-4619])
8865
8866   *Adam Langley (Google)*
8867
8868 * Check parameters are not NULL in GOST ENGINE. ([CVE-2012-0027])
8869
8870   *Andrey Kulikov <amdeich@gmail.com>*
8871
8872 * Prevent malformed RFC3779 data triggering an assertion failure.
8873   Thanks to Andrew Chi, BBN Technologies, for discovering the flaw
8874   and Rob Austein <sra@hactrn.net> for fixing it. ([CVE-2011-4577])
8875
8876   *Rob Austein <sra@hactrn.net>*
8877
8878 * Improved PRNG seeding for VOS.
8879
8880   *Paul Green <Paul.Green@stratus.com>*
8881
8882 * Fix ssl_ciph.c set-up race.
8883
8884   *Adam Langley (Google)*
8885
8886 * Fix spurious failures in ecdsatest.c.
8887
8888   *Emilia Käsper (Google)*
8889
8890 * Fix the BIO_f_buffer() implementation (which was mixing different
8891   interpretations of the `..._len` fields).
8892
8893   *Adam Langley (Google)*
8894
8895 * Fix handling of BN_BLINDING: now BN_BLINDING_invert_ex (rather than
8896   BN_BLINDING_invert_ex) calls BN_BLINDING_update, ensuring that concurrent
8897   threads won't reuse the same blinding coefficients.
8898
8899   This also avoids the need to obtain the CRYPTO_LOCK_RSA_BLINDING
8900   lock to call BN_BLINDING_invert_ex, and avoids one use of
8901   BN_BLINDING_update for each BN_BLINDING structure (previously,
8902   the last update always remained unused).
8903
8904   *Emilia Käsper (Google)*
8905
8906 * In ssl3_clear, preserve s3->init_extra along with s3->rbuf.
8907
8908   *Bob Buckholz (Google)*
8909
8910### Changes between 1.0.0d and 1.0.0e [6 Sep 2011]
8911
8912 * Fix bug where CRLs with nextUpdate in the past are sometimes accepted
8913   by initialising X509_STORE_CTX properly. ([CVE-2011-3207])
8914
8915   *Kaspar Brand <ossl@velox.ch>*
8916
8917 * Fix SSL memory handling for (EC)DH ciphersuites, in particular
8918   for multi-threaded use of ECDH. ([CVE-2011-3210])
8919
8920   *Adam Langley (Google)*
8921
8922 * Fix x509_name_ex_d2i memory leak on bad inputs.
8923
8924   *Bodo Moeller*
8925
8926 * Remove hard coded ecdsaWithSHA1 signature tests in ssl code and check
8927   signature public key algorithm by using OID xref utilities instead.
8928   Before this you could only use some ECC ciphersuites with SHA1 only.
8929
8930   *Steve Henson*
8931
8932 * Add protection against ECDSA timing attacks as mentioned in the paper
8933   by Billy Bob Brumley and Nicola Tuveri, see:
8934   <http://eprint.iacr.org/2011/232.pdf>
8935
8936   *Billy Bob Brumley and Nicola Tuveri*
8937
8938### Changes between 1.0.0c and 1.0.0d [8 Feb 2011]
8939
8940 * Fix parsing of OCSP stapling ClientHello extension. CVE-2011-0014
8941
8942   *Neel Mehta, Adam Langley, Bodo Moeller (Google)*
8943
8944 * Fix bug in string printing code: if *any* escaping is enabled we must
8945   escape the escape character (backslash) or the resulting string is
8946   ambiguous.
8947
8948   *Steve Henson*
8949
8950### Changes between 1.0.0b and 1.0.0c  [2 Dec 2010]
8951
8952 * Disable code workaround for ancient and obsolete Netscape browsers
8953   and servers: an attacker can use it in a ciphersuite downgrade attack.
8954   Thanks to Martin Rex for discovering this bug. CVE-2010-4180
8955
8956   *Steve Henson*
8957
8958 * Fixed J-PAKE implementation error, originally discovered by
8959   Sebastien Martini, further info and confirmation from Stefan
8960   Arentz and Feng Hao. Note that this fix is a security fix. CVE-2010-4252
8961
8962   *Ben Laurie*
8963
8964### Changes between 1.0.0a and 1.0.0b  [16 Nov 2010]
8965
8966 * Fix extension code to avoid race conditions which can result in a buffer
8967   overrun vulnerability: resumed sessions must not be modified as they can
8968   be shared by multiple threads. CVE-2010-3864
8969
8970   *Steve Henson*
8971
8972 * Fix WIN32 build system to correctly link an ENGINE directory into
8973   a DLL.
8974
8975   *Steve Henson*
8976
8977### Changes between 1.0.0 and 1.0.0a  [01 Jun 2010]
8978
8979 * Check return value of int_rsa_verify in pkey_rsa_verifyrecover
8980   ([CVE-2010-1633])
8981
8982   *Steve Henson, Peter-Michael Hager <hager@dortmund.net>*
8983
8984### Changes between 0.9.8n and 1.0.0  [29 Mar 2010]
8985
8986 * Add "missing" function EVP_CIPHER_CTX_copy(). This copies a cipher
8987   context. The operation can be customised via the ctrl mechanism in
8988   case ENGINEs want to include additional functionality.
8989
8990   *Steve Henson*
8991
8992 * Tolerate yet another broken PKCS#8 key format: private key value negative.
8993
8994   *Steve Henson*
8995
8996 * Add new -subject_hash_old and -issuer_hash_old options to x509 utility to
8997   output hashes compatible with older versions of OpenSSL.
8998
8999   *Willy Weisz <weisz@vcpc.univie.ac.at>*
9000
9001 * Fix compression algorithm handling: if resuming a session use the
9002   compression algorithm of the resumed session instead of determining
9003   it from client hello again. Don't allow server to change algorithm.
9004
9005   *Steve Henson*
9006
9007 * Add load_crls() function to commands tidying load_certs() too. Add option
9008   to verify utility to allow additional CRLs to be included.
9009
9010   *Steve Henson*
9011
9012 * Update OCSP request code to permit adding custom headers to the request:
9013   some responders need this.
9014
9015   *Steve Henson*
9016
9017 * The function EVP_PKEY_sign() returns <=0 on error: check return code
9018   correctly.
9019
9020   *Julia Lawall <julia@diku.dk>*
9021
9022 * Update verify callback code in `apps/s_cb.c` and `apps/verify.c`, it
9023   needlessly dereferenced structures, used obsolete functions and
9024   didn't handle all updated verify codes correctly.
9025
9026   *Steve Henson*
9027
9028 * Disable MD2 in the default configuration.
9029
9030   *Steve Henson*
9031
9032 * In BIO_pop() and BIO_push() use the ctrl argument (which was NULL) to
9033   indicate the initial BIO being pushed or popped. This makes it possible
9034   to determine whether the BIO is the one explicitly called or as a result
9035   of the ctrl being passed down the chain. Fix BIO_pop() and SSL BIOs so
9036   it handles reference counts correctly and doesn't zero out the I/O bio
9037   when it is not being explicitly popped. WARNING: applications which
9038   included workarounds for the old buggy behaviour will need to be modified
9039   or they could free up already freed BIOs.
9040
9041   *Steve Henson*
9042
9043 * Extend the uni2asc/asc2uni => OPENSSL_uni2asc/OPENSSL_asc2uni
9044   renaming to all platforms (within the 0.9.8 branch, this was
9045   done conditionally on Netware platforms to avoid a name clash).
9046
9047   *Guenter <lists@gknw.net>*
9048
9049 * Add ECDHE and PSK support to DTLS.
9050
9051   *Michael Tuexen <tuexen@fh-muenster.de>*
9052
9053 * Add CHECKED_STACK_OF macro to safestack.h, otherwise safestack can't
9054   be used on C++.
9055
9056   *Steve Henson*
9057
9058 * Add "missing" function EVP_MD_flags() (without this the only way to
9059   retrieve a digest flags is by accessing the structure directly. Update
9060   `EVP_MD_do_all*()` and `EVP_CIPHER_do_all*()` to include the name a digest
9061   or cipher is registered as in the "from" argument. Print out all
9062   registered digests in the dgst usage message instead of manually
9063   attempting to work them out.
9064
9065   *Steve Henson*
9066
9067 * If no SSLv2 ciphers are used don't use an SSLv2 compatible client hello:
9068   this allows the use of compression and extensions. Change default cipher
9069   string to remove SSLv2 ciphersuites. This effectively avoids ancient SSLv2
9070   by default unless an application cipher string requests it.
9071
9072   *Steve Henson*
9073
9074 * Alter match criteria in PKCS12_parse(). It used to try to use local
9075   key ids to find matching certificates and keys but some PKCS#12 files
9076   don't follow the (somewhat unwritten) rules and this strategy fails.
9077   Now just gather all certificates together and the first private key
9078   then look for the first certificate that matches the key.
9079
9080   *Steve Henson*
9081
9082 * Support use of registered digest and cipher names for dgst and cipher
9083   commands instead of having to add each one as a special case. So now
9084   you can do:
9085
9086           openssl sha256 foo
9087
9088   as well as:
9089
9090           openssl dgst -sha256 foo
9091
9092   and this works for ENGINE based algorithms too.
9093
9094   *Steve Henson*
9095
9096 * Update Gost ENGINE to support parameter files.
9097
9098   *Victor B. Wagner <vitus@cryptocom.ru>*
9099
9100 * Support GeneralizedTime in ca utility.
9101
9102   *Oliver Martin <oliver@volatilevoid.net>, Steve Henson*
9103
9104 * Enhance the hash format used for certificate directory links. The new
9105   form uses the canonical encoding (meaning equivalent names will work
9106   even if they aren't identical) and uses SHA1 instead of MD5. This form
9107   is incompatible with the older format and as a result c_rehash should
9108   be used to rebuild symbolic links.
9109
9110   *Steve Henson*
9111
9112 * Make PKCS#8 the default write format for private keys, replacing the
9113   traditional format. This form is standardised, more secure and doesn't
9114   include an implicit MD5 dependency.
9115
9116   *Steve Henson*
9117
9118 * Add a $gcc_devteam_warn option to Configure. The idea is that any code
9119   committed to OpenSSL should pass this lot as a minimum.
9120
9121   *Steve Henson*
9122
9123 * Add session ticket override functionality for use by EAP-FAST.
9124
9125   *Jouni Malinen <j@w1.fi>*
9126
9127 * Modify HMAC functions to return a value. Since these can be implemented
9128   in an ENGINE errors can occur.
9129
9130   *Steve Henson*
9131
9132 * Type-checked OBJ_bsearch_ex.
9133
9134   *Ben Laurie*
9135
9136 * Type-checked OBJ_bsearch. Also some constification necessitated
9137   by type-checking.  Still to come: TXT_DB, bsearch(?),
9138   OBJ_bsearch_ex, qsort, CRYPTO_EX_DATA, ASN1_VALUE, ASN1_STRING,
9139   CONF_VALUE.
9140
9141   *Ben Laurie*
9142
9143 * New function OPENSSL_gmtime_adj() to add a specific number of days and
9144   seconds to a tm structure directly, instead of going through OS
9145   specific date routines. This avoids any issues with OS routines such
9146   as the year 2038 bug. New `*_adj()` functions for ASN1 time structures
9147   and X509_time_adj_ex() to cover the extended range. The existing
9148   X509_time_adj() is still usable and will no longer have any date issues.
9149
9150   *Steve Henson*
9151
9152 * Delta CRL support. New use deltas option which will attempt to locate
9153   and search any appropriate delta CRLs available.
9154
9155   This work was sponsored by Google.
9156
9157   *Steve Henson*
9158
9159 * Support for CRLs partitioned by reason code. Reorganise CRL processing
9160   code and add additional score elements. Validate alternate CRL paths
9161   as part of the CRL checking and indicate a new error "CRL path validation
9162   error" in this case. Applications wanting additional details can use
9163   the verify callback and check the new "parent" field. If this is not
9164   NULL CRL path validation is taking place. Existing applications won't
9165   see this because it requires extended CRL support which is off by
9166   default.
9167
9168   This work was sponsored by Google.
9169
9170   *Steve Henson*
9171
9172 * Support for freshest CRL extension.
9173
9174   This work was sponsored by Google.
9175
9176   *Steve Henson*
9177
9178 * Initial indirect CRL support. Currently only supported in the CRLs
9179   passed directly and not via lookup. Process certificate issuer
9180   CRL entry extension and lookup CRL entries by bother issuer name
9181   and serial number. Check and process CRL issuer entry in IDP extension.
9182
9183   This work was sponsored by Google.
9184
9185   *Steve Henson*
9186
9187 * Add support for distinct certificate and CRL paths. The CRL issuer
9188   certificate is validated separately in this case. Only enabled if
9189   an extended CRL support flag is set: this flag will enable additional
9190   CRL functionality in future.
9191
9192   This work was sponsored by Google.
9193
9194   *Steve Henson*
9195
9196 * Add support for policy mappings extension.
9197
9198   This work was sponsored by Google.
9199
9200   *Steve Henson*
9201
9202 * Fixes to pathlength constraint, self issued certificate handling,
9203   policy processing to align with RFC3280 and PKITS tests.
9204
9205   This work was sponsored by Google.
9206
9207   *Steve Henson*
9208
9209 * Support for name constraints certificate extension. DN, email, DNS
9210   and URI types are currently supported.
9211
9212   This work was sponsored by Google.
9213
9214   *Steve Henson*
9215
9216 * To cater for systems that provide a pointer-based thread ID rather
9217   than numeric, deprecate the current numeric thread ID mechanism and
9218   replace it with a structure and associated callback type. This
9219   mechanism allows a numeric "hash" to be extracted from a thread ID in
9220   either case, and on platforms where pointers are larger than 'long',
9221   mixing is done to help ensure the numeric 'hash' is usable even if it
9222   can't be guaranteed unique. The default mechanism is to use "&errno"
9223   as a pointer-based thread ID to distinguish between threads.
9224
9225   Applications that want to provide their own thread IDs should now use
9226   CRYPTO_THREADID_set_callback() to register a callback that will call
9227   either CRYPTO_THREADID_set_numeric() or CRYPTO_THREADID_set_pointer().
9228
9229   Note that ERR_remove_state() is now deprecated, because it is tied
9230   to the assumption that thread IDs are numeric.  ERR_remove_state(0)
9231   to free the current thread's error state should be replaced by
9232   ERR_remove_thread_state(NULL).
9233
9234   (This new approach replaces the functions CRYPTO_set_idptr_callback(),
9235   CRYPTO_get_idptr_callback(), and CRYPTO_thread_idptr() that existed in
9236   OpenSSL 0.9.9-dev between June 2006 and August 2008. Also, if an
9237   application was previously providing a numeric thread callback that
9238   was inappropriate for distinguishing threads, then uniqueness might
9239   have been obtained with &errno that happened immediately in the
9240   intermediate development versions of OpenSSL; this is no longer the
9241   case, the numeric thread callback will now override the automatic use
9242   of &errno.)
9243
9244   *Geoff Thorpe, with help from Bodo Moeller*
9245
9246 * Initial support for different CRL issuing certificates. This covers a
9247   simple case where the self issued certificates in the chain exist and
9248   the real CRL issuer is higher in the existing chain.
9249
9250   This work was sponsored by Google.
9251
9252   *Steve Henson*
9253
9254 * Removed effectively defunct crypto/store from the build.
9255
9256   *Ben Laurie*
9257
9258 * Revamp of STACK to provide stronger type-checking. Still to come:
9259   TXT_DB, bsearch(?), OBJ_bsearch, qsort, CRYPTO_EX_DATA, ASN1_VALUE,
9260   ASN1_STRING, CONF_VALUE.
9261
9262   *Ben Laurie*
9263
9264 * Add a new SSL_MODE_RELEASE_BUFFERS mode flag to release unused buffer
9265   RAM on SSL connections.  This option can save about 34k per idle SSL.
9266
9267   *Nick Mathewson*
9268
9269 * Revamp of LHASH to provide stronger type-checking. Still to come:
9270   STACK, TXT_DB, bsearch, qsort.
9271
9272   *Ben Laurie*
9273
9274 * Initial support for Cryptographic Message Syntax (aka CMS) based
9275   on RFC3850, RFC3851 and RFC3852. New cms directory and cms utility,
9276   support for data, signedData, compressedData, digestedData and
9277   encryptedData, envelopedData types included. Scripts to check against
9278   RFC4134 examples draft and interop and consistency checks of many
9279   content types and variants.
9280
9281   *Steve Henson*
9282
9283 * Add options to enc utility to support use of zlib compression BIO.
9284
9285   *Steve Henson*
9286
9287 * Extend mk1mf to support importing of options and assembly language
9288   files from Configure script, currently only included in VC-WIN32.
9289   The assembly language rules can now optionally generate the source
9290   files from the associated perl scripts.
9291
9292   *Steve Henson*
9293
9294 * Implement remaining functionality needed to support GOST ciphersuites.
9295   Interop testing has been performed using CryptoPro implementations.
9296
9297   *Victor B. Wagner <vitus@cryptocom.ru>*
9298
9299 * s390x assembler pack.
9300
9301   *Andy Polyakov*
9302
9303 * ARMv4 assembler pack. ARMv4 refers to v4 and later ISA, not CPU
9304   "family."
9305
9306   *Andy Polyakov*
9307
9308 * Implement Opaque PRF Input TLS extension as specified in
9309   draft-rescorla-tls-opaque-prf-input-00.txt.  Since this is not an
9310   official specification yet and no extension type assignment by
9311   IANA exists, this extension (for now) will have to be explicitly
9312   enabled when building OpenSSL by providing the extension number
9313   to use.  For example, specify an option
9314
9315           -DTLSEXT_TYPE_opaque_prf_input=0x9527
9316
9317   to the "config" or "Configure" script to enable the extension,
9318   assuming extension number 0x9527 (which is a completely arbitrary
9319   and unofficial assignment based on the MD5 hash of the Internet
9320   Draft).  Note that by doing so, you potentially lose
9321   interoperability with other TLS implementations since these might
9322   be using the same extension number for other purposes.
9323
9324   SSL_set_tlsext_opaque_prf_input(ssl, src, len) is used to set the
9325   opaque PRF input value to use in the handshake.  This will create
9326   an internal copy of the length-'len' string at 'src', and will
9327   return non-zero for success.
9328
9329   To get more control and flexibility, provide a callback function
9330   by using
9331
9332           SSL_CTX_set_tlsext_opaque_prf_input_callback(ctx, cb)
9333           SSL_CTX_set_tlsext_opaque_prf_input_callback_arg(ctx, arg)
9334
9335   where
9336
9337           int (*cb)(SSL *, void *peerinput, size_t len, void *arg);
9338           void *arg;
9339
9340   Callback function 'cb' will be called in handshakes, and is
9341   expected to use SSL_set_tlsext_opaque_prf_input() as appropriate.
9342   Argument 'arg' is for application purposes (the value as given to
9343   SSL_CTX_set_tlsext_opaque_prf_input_callback_arg() will directly
9344   be provided to the callback function).  The callback function
9345   has to return non-zero to report success: usually 1 to use opaque
9346   PRF input just if possible, or 2 to enforce use of the opaque PRF
9347   input.  In the latter case, the library will abort the handshake
9348   if opaque PRF input is not successfully negotiated.
9349
9350   Arguments 'peerinput' and 'len' given to the callback function
9351   will always be NULL and 0 in the case of a client.  A server will
9352   see the client's opaque PRF input through these variables if
9353   available (NULL and 0 otherwise).  Note that if the server
9354   provides an opaque PRF input, the length must be the same as the
9355   length of the client's opaque PRF input.
9356
9357   Note that the callback function will only be called when creating
9358   a new session (session resumption can resume whatever was
9359   previously negotiated), and will not be called in SSL 2.0
9360   handshakes; thus, SSL_CTX_set_options(ctx, SSL_OP_NO_SSLv2) or
9361   SSL_set_options(ssl, SSL_OP_NO_SSLv2) is especially recommended
9362   for applications that need to enforce opaque PRF input.
9363
9364   *Bodo Moeller*
9365
9366 * Update ssl code to support digests other than SHA1+MD5 for handshake
9367   MAC.
9368
9369   *Victor B. Wagner <vitus@cryptocom.ru>*
9370
9371 * Add RFC4507 support to OpenSSL. This includes the corrections in
9372   RFC4507bis. The encrypted ticket format is an encrypted encoded
9373   SSL_SESSION structure, that way new session features are automatically
9374   supported.
9375
9376   If a client application caches session in an SSL_SESSION structure
9377   support is transparent because tickets are now stored in the encoded
9378   SSL_SESSION.
9379
9380   The SSL_CTX structure automatically generates keys for ticket
9381   protection in servers so again support should be possible
9382   with no application modification.
9383
9384   If a client or server wishes to disable RFC4507 support then the option
9385   SSL_OP_NO_TICKET can be set.
9386
9387   Add a TLS extension debugging callback to allow the contents of any client
9388   or server extensions to be examined.
9389
9390   This work was sponsored by Google.
9391
9392   *Steve Henson*
9393
9394 * Final changes to avoid use of pointer pointer casts in OpenSSL.
9395   OpenSSL should now compile cleanly on gcc 4.2
9396
9397   *Peter Hartley <pdh@utter.chaos.org.uk>, Steve Henson*
9398
9399 * Update SSL library to use new EVP_PKEY MAC API. Include generic MAC
9400   support including streaming MAC support: this is required for GOST
9401   ciphersuite support.
9402
9403   *Victor B. Wagner <vitus@cryptocom.ru>, Steve Henson*
9404
9405 * Add option -stream to use PKCS#7 streaming in smime utility. New
9406   function i2d_PKCS7_bio_stream() and PEM_write_PKCS7_bio_stream()
9407   to output in BER and PEM format.
9408
9409   *Steve Henson*
9410
9411 * Experimental support for use of HMAC via EVP_PKEY interface. This
9412   allows HMAC to be handled via the `EVP_DigestSign*()` interface. The
9413   EVP_PKEY "key" in this case is the HMAC key, potentially allowing
9414   ENGINE support for HMAC keys which are unextractable. New -mac and
9415   -macopt options to dgst utility.
9416
9417   *Steve Henson*
9418
9419 * New option -sigopt to dgst utility. Update dgst to use
9420   `EVP_Digest{Sign,Verify}*`. These two changes make it possible to use
9421   alternative signing parameters such as X9.31 or PSS in the dgst
9422   utility.
9423
9424   *Steve Henson*
9425
9426 * Change ssl_cipher_apply_rule(), the internal function that does
9427   the work each time a ciphersuite string requests enabling
9428   ("foo+bar"), moving ("+foo+bar"), disabling ("-foo+bar", or
9429   removing ("!foo+bar") a class of ciphersuites: Now it maintains
9430   the order of disabled ciphersuites such that those ciphersuites
9431   that most recently went from enabled to disabled not only stay
9432   in order with respect to each other, but also have higher priority
9433   than other disabled ciphersuites the next time ciphersuites are
9434   enabled again.
9435
9436   This means that you can now say, e.g., "PSK:-PSK:HIGH" to enable
9437   the same ciphersuites as with "HIGH" alone, but in a specific
9438   order where the PSK ciphersuites come first (since they are the
9439   most recently disabled ciphersuites when "HIGH" is parsed).
9440
9441   Also, change ssl_create_cipher_list() (using this new
9442   functionality) such that between otherwise identical
9443   ciphersuites, ephemeral ECDH is preferred over ephemeral DH in
9444   the default order.
9445
9446   *Bodo Moeller*
9447
9448 * Change ssl_create_cipher_list() so that it automatically
9449   arranges the ciphersuites in reasonable order before starting
9450   to process the rule string.  Thus, the definition for "DEFAULT"
9451   (SSL_DEFAULT_CIPHER_LIST) now is just "ALL:!aNULL:!eNULL", but
9452   remains equivalent to `"AES:ALL:!aNULL:!eNULL:+aECDH:+kRSA:+RC4:@STRENGTH"`.
9453   This makes it much easier to arrive at a reasonable default order
9454   in applications for which anonymous ciphers are OK (meaning
9455   that you can't actually use DEFAULT).
9456
9457   *Bodo Moeller; suggested by Victor Duchovni*
9458
9459 * Split the SSL/TLS algorithm mask (as used for ciphersuite string
9460   processing) into multiple integers instead of setting
9461   "SSL_MKEY_MASK" bits, "SSL_AUTH_MASK" bits, "SSL_ENC_MASK",
9462   "SSL_MAC_MASK", and "SSL_SSL_MASK" bits all in a single integer.
9463   (These masks as well as the individual bit definitions are hidden
9464   away into the non-exported interface ssl/ssl_locl.h, so this
9465   change to the definition of the SSL_CIPHER structure shouldn't
9466   affect applications.)  This give us more bits for each of these
9467   categories, so there is no longer a need to coagulate AES128 and
9468   AES256 into a single algorithm bit, and to coagulate Camellia128
9469   and Camellia256 into a single algorithm bit, which has led to all
9470   kinds of kludges.
9471
9472   Thus, among other things, the kludge introduced in 0.9.7m and
9473   0.9.8e for masking out AES256 independently of AES128 or masking
9474   out Camellia256 independently of AES256 is not needed here in 0.9.9.
9475
9476   With the change, we also introduce new ciphersuite aliases that
9477   so far were missing: "AES128", "AES256", "CAMELLIA128", and
9478   "CAMELLIA256".
9479
9480   *Bodo Moeller*
9481
9482 * Add support for dsa-with-SHA224 and dsa-with-SHA256.
9483   Use the leftmost N bytes of the signature input if the input is
9484   larger than the prime q (with N being the size in bytes of q).
9485
9486   *Nils Larsch*
9487
9488 * Very *very* experimental PKCS#7 streaming encoder support. Nothing uses
9489   it yet and it is largely untested.
9490
9491   *Steve Henson*
9492
9493 * Add support for the ecdsa-with-SHA224/256/384/512 signature types.
9494
9495   *Nils Larsch*
9496
9497 * Initial incomplete changes to avoid need for function casts in OpenSSL
9498   some compilers (gcc 4.2 and later) reject their use. Safestack is
9499   reimplemented.  Update ASN1 to avoid use of legacy functions.
9500
9501   *Steve Henson*
9502
9503 * Win32/64 targets are linked with Winsock2.
9504
9505   *Andy Polyakov*
9506
9507 * Add an X509_CRL_METHOD structure to allow CRL processing to be redirected
9508   to external functions. This can be used to increase CRL handling
9509   efficiency especially when CRLs are very large by (for example) storing
9510   the CRL revoked certificates in a database.
9511
9512   *Steve Henson*
9513
9514 * Overhaul of by_dir code. Add support for dynamic loading of CRLs so
9515   new CRLs added to a directory can be used. New command line option
9516   -verify_return_error to s_client and s_server. This causes real errors
9517   to be returned by the verify callback instead of carrying on no matter
9518   what. This reflects the way a "real world" verify callback would behave.
9519
9520   *Steve Henson*
9521
9522 * GOST engine, supporting several GOST algorithms and public key formats.
9523   Kindly donated by Cryptocom.
9524
9525   *Cryptocom*
9526
9527 * Partial support for Issuing Distribution Point CRL extension. CRLs
9528   partitioned by DP are handled but no indirect CRL or reason partitioning
9529   (yet). Complete overhaul of CRL handling: now the most suitable CRL is
9530   selected via a scoring technique which handles IDP and AKID in CRLs.
9531
9532   *Steve Henson*
9533
9534 * New X509_STORE_CTX callbacks lookup_crls() and lookup_certs() which
9535   will ultimately be used for all verify operations: this will remove the
9536   X509_STORE dependency on certificate verification and allow alternative
9537   lookup methods.  X509_STORE based implementations of these two callbacks.
9538
9539   *Steve Henson*
9540
9541 * Allow multiple CRLs to exist in an X509_STORE with matching issuer names.
9542   Modify get_crl() to find a valid (unexpired) CRL if possible.
9543
9544   *Steve Henson*
9545
9546 * New function X509_CRL_match() to check if two CRLs are identical. Normally
9547   this would be called X509_CRL_cmp() but that name is already used by
9548   a function that just compares CRL issuer names. Cache several CRL
9549   extensions in X509_CRL structure and cache CRLDP in X509.
9550
9551   *Steve Henson*
9552
9553 * Store a "canonical" representation of X509_NAME structure (ASN1 Name)
9554   this maps equivalent X509_NAME structures into a consistent structure.
9555   Name comparison can then be performed rapidly using memcmp().
9556
9557   *Steve Henson*
9558
9559 * Non-blocking OCSP request processing. Add -timeout option to ocsp
9560   utility.
9561
9562   *Steve Henson*
9563
9564 * Allow digests to supply their own micalg string for S/MIME type using
9565   the ctrl EVP_MD_CTRL_MICALG.
9566
9567   *Steve Henson*
9568
9569 * During PKCS7 signing pass the PKCS7 SignerInfo structure to the
9570   EVP_PKEY_METHOD before and after signing via the EVP_PKEY_CTRL_PKCS7_SIGN
9571   ctrl. It can then customise the structure before and/or after signing
9572   if necessary.
9573
9574   *Steve Henson*
9575
9576 * New function OBJ_add_sigid() to allow application defined signature OIDs
9577   to be added to OpenSSLs internal tables. New function OBJ_sigid_free()
9578   to free up any added signature OIDs.
9579
9580   *Steve Henson*
9581
9582 * New functions EVP_CIPHER_do_all(), EVP_CIPHER_do_all_sorted(),
9583   EVP_MD_do_all() and EVP_MD_do_all_sorted() to enumerate internal
9584   digest and cipher tables. New options added to openssl utility:
9585   list-message-digest-algorithms and list-cipher-algorithms.
9586
9587   *Steve Henson*
9588
9589 * Change the array representation of binary polynomials: the list
9590   of degrees of non-zero coefficients is now terminated with -1.
9591   Previously it was terminated with 0, which was also part of the
9592   value; thus, the array representation was not applicable to
9593   polynomials where t^0 has coefficient zero.  This change makes
9594   the array representation useful in a more general context.
9595
9596   *Douglas Stebila*
9597
9598 * Various modifications and fixes to SSL/TLS cipher string
9599   handling.  For ECC, the code now distinguishes between fixed ECDH
9600   with RSA certificates on the one hand and with ECDSA certificates
9601   on the other hand, since these are separate ciphersuites.  The
9602   unused code for Fortezza ciphersuites has been removed.
9603
9604   For consistency with EDH, ephemeral ECDH is now called "EECDH"
9605   (not "ECDHE").  For consistency with the code for DH
9606   certificates, use of ECDH certificates is now considered ECDH
9607   authentication, not RSA or ECDSA authentication (the latter is
9608   merely the CA's signing algorithm and not actively used in the
9609   protocol).
9610
9611   The temporary ciphersuite alias "ECCdraft" is no longer
9612   available, and ECC ciphersuites are no longer excluded from "ALL"
9613   and "DEFAULT".  The following aliases now exist for RFC 4492
9614   ciphersuites, most of these by analogy with the DH case:
9615
9616           kECDHr   - ECDH cert, signed with RSA
9617           kECDHe   - ECDH cert, signed with ECDSA
9618           kECDH    - ECDH cert (signed with either RSA or ECDSA)
9619           kEECDH   - ephemeral ECDH
9620           ECDH     - ECDH cert or ephemeral ECDH
9621
9622           aECDH    - ECDH cert
9623           aECDSA   - ECDSA cert
9624           ECDSA    - ECDSA cert
9625
9626           AECDH    - anonymous ECDH
9627           EECDH    - non-anonymous ephemeral ECDH (equivalent to "kEECDH:-AECDH")
9628
9629   *Bodo Moeller*
9630
9631 * Add additional S/MIME capabilities for AES and GOST ciphers if supported.
9632   Use correct micalg parameters depending on digest(s) in signed message.
9633
9634   *Steve Henson*
9635
9636 * Add engine support for EVP_PKEY_ASN1_METHOD. Add functions to process
9637   an ENGINE asn1 method. Support ENGINE lookups in the ASN1 code.
9638
9639   *Steve Henson*
9640
9641 * Initial engine support for EVP_PKEY_METHOD. New functions to permit
9642   an engine to register a method. Add ENGINE lookups for methods and
9643   functional reference processing.
9644
9645   *Steve Henson*
9646
9647 * New functions `EVP_Digest{Sign,Verify)*`. These are enhanced versions of
9648   `EVP_{Sign,Verify}*` which allow an application to customise the signature
9649   process.
9650
9651   *Steve Henson*
9652
9653 * New -resign option to smime utility. This adds one or more signers
9654   to an existing PKCS#7 signedData structure. Also -md option to use an
9655   alternative message digest algorithm for signing.
9656
9657   *Steve Henson*
9658
9659 * Tidy up PKCS#7 routines and add new functions to make it easier to
9660   create PKCS7 structures containing multiple signers. Update smime
9661   application to support multiple signers.
9662
9663   *Steve Henson*
9664
9665 * New -macalg option to pkcs12 utility to allow setting of an alternative
9666   digest MAC.
9667
9668   *Steve Henson*
9669
9670 * Initial support for PKCS#5 v2.0 PRFs other than default SHA1 HMAC.
9671   Reorganize PBE internals to lookup from a static table using NIDs,
9672   add support for HMAC PBE OID translation. Add a EVP_CIPHER ctrl:
9673   EVP_CTRL_PBE_PRF_NID this allows a cipher to specify an alternative
9674   PRF which will be automatically used with PBES2.
9675
9676   *Steve Henson*
9677
9678 * Replace the algorithm specific calls to generate keys in "req" with the
9679   new API.
9680
9681   *Steve Henson*
9682
9683 * Update PKCS#7 enveloped data routines to use new API. This is now
9684   supported by any public key method supporting the encrypt operation. A
9685   ctrl is added to allow the public key algorithm to examine or modify
9686   the PKCS#7 RecipientInfo structure if it needs to: for RSA this is
9687   a no op.
9688
9689   *Steve Henson*
9690
9691 * Add a ctrl to asn1 method to allow a public key algorithm to express
9692   a default digest type to use. In most cases this will be SHA1 but some
9693   algorithms (such as GOST) need to specify an alternative digest. The
9694   return value indicates how strong the preference is 1 means optional and
9695   2 is mandatory (that is it is the only supported type). Modify
9696   ASN1_item_sign() to accept a NULL digest argument to indicate it should
9697   use the default md. Update openssl utilities to use the default digest
9698   type for signing if it is not explicitly indicated.
9699
9700   *Steve Henson*
9701
9702 * Use OID cross reference table in ASN1_sign() and ASN1_verify(). New
9703   EVP_MD flag EVP_MD_FLAG_PKEY_METHOD_SIGNATURE. This uses the relevant
9704   signing method from the key type. This effectively removes the link
9705   between digests and public key types.
9706
9707   *Steve Henson*
9708
9709 * Add an OID cross reference table and utility functions. Its purpose is to
9710   translate between signature OIDs such as SHA1WithrsaEncryption and SHA1,
9711   rsaEncryption. This will allow some of the algorithm specific hackery
9712   needed to use the correct OID to be removed.
9713
9714   *Steve Henson*
9715
9716 * Remove algorithm specific dependencies when setting PKCS7_SIGNER_INFO
9717   structures for PKCS7_sign(). They are now set up by the relevant public
9718   key ASN1 method.
9719
9720   *Steve Henson*
9721
9722 * Add provisional EC pkey method with support for ECDSA and ECDH.
9723
9724   *Steve Henson*
9725
9726 * Add support for key derivation (agreement) in the API, DH method and
9727   pkeyutl.
9728
9729   *Steve Henson*
9730
9731 * Add DSA pkey method and DH pkey methods, extend DH ASN1 method to support
9732   public and private key formats. As a side effect these add additional
9733   command line functionality not previously available: DSA signatures can be
9734   generated and verified using pkeyutl and DH key support and generation in
9735   pkey, genpkey.
9736
9737   *Steve Henson*
9738
9739 * BeOS support.
9740
9741   *Oliver Tappe <zooey@hirschkaefer.de>*
9742
9743 * New make target "install_html_docs" installs HTML renditions of the
9744   manual pages.
9745
9746   *Oliver Tappe <zooey@hirschkaefer.de>*
9747
9748 * New utility "genpkey" this is analogous to "genrsa" etc except it can
9749   generate keys for any algorithm. Extend and update EVP_PKEY_METHOD to
9750   support key and parameter generation and add initial key generation
9751   functionality for RSA.
9752
9753   *Steve Henson*
9754
9755 * Add functions for main EVP_PKEY_method operations. The undocumented
9756   functions `EVP_PKEY_{encrypt,decrypt}` have been renamed to
9757   `EVP_PKEY_{encrypt,decrypt}_old`.
9758
9759   *Steve Henson*
9760
9761 * Initial definitions for EVP_PKEY_METHOD. This will be a high level public
9762   key API, doesn't do much yet.
9763
9764   *Steve Henson*
9765
9766 * New function EVP_PKEY_asn1_get0_info() to retrieve information about
9767   public key algorithms. New option to openssl utility:
9768   "list-public-key-algorithms" to print out info.
9769
9770   *Steve Henson*
9771
9772 * Implement the Supported Elliptic Curves Extension for
9773   ECC ciphersuites from draft-ietf-tls-ecc-12.txt.
9774
9775   *Douglas Stebila*
9776
9777 * Don't free up OIDs in OBJ_cleanup() if they are in use by EVP_MD or
9778   EVP_CIPHER structures to avoid later problems in EVP_cleanup().
9779
9780   *Steve Henson*
9781
9782 * New utilities pkey and pkeyparam. These are similar to algorithm specific
9783   utilities such as rsa, dsa, dsaparam etc except they process any key
9784   type.
9785
9786   *Steve Henson*
9787
9788 * Transfer public key printing routines to EVP_PKEY_ASN1_METHOD. New
9789   functions EVP_PKEY_print_public(), EVP_PKEY_print_private(),
9790   EVP_PKEY_print_param() to print public key data from an EVP_PKEY
9791   structure.
9792
9793   *Steve Henson*
9794
9795 * Initial support for pluggable public key ASN1.
9796   De-spaghettify the public key ASN1 handling. Move public and private
9797   key ASN1 handling to a new EVP_PKEY_ASN1_METHOD structure. Relocate
9798   algorithm specific handling to a single module within the relevant
9799   algorithm directory. Add functions to allow (near) opaque processing
9800   of public and private key structures.
9801
9802   *Steve Henson*
9803
9804 * Implement the Supported Point Formats Extension for
9805   ECC ciphersuites from draft-ietf-tls-ecc-12.txt.
9806
9807   *Douglas Stebila*
9808
9809 * Add initial support for RFC 4279 PSK TLS ciphersuites. Add members
9810   for the psk identity [hint] and the psk callback functions to the
9811   SSL_SESSION, SSL and SSL_CTX structure.
9812
9813   New ciphersuites:
9814           PSK-RC4-SHA, PSK-3DES-EDE-CBC-SHA, PSK-AES128-CBC-SHA,
9815           PSK-AES256-CBC-SHA
9816
9817   New functions:
9818           SSL_CTX_use_psk_identity_hint
9819           SSL_get_psk_identity_hint
9820           SSL_get_psk_identity
9821           SSL_use_psk_identity_hint
9822
9823   *Mika Kousa and Pasi Eronen of Nokia Corporation*
9824
9825 * Add RFC 3161 compliant time stamp request creation, response generation
9826   and response verification functionality.
9827
9828   *Zoltán Glózik <zglozik@opentsa.org>, The OpenTSA Project*
9829
9830 * Add initial support for TLS extensions, specifically for the server_name
9831   extension so far.  The SSL_SESSION, SSL_CTX, and SSL data structures now
9832   have new members for a host name.  The SSL data structure has an
9833   additional member `SSL_CTX *initial_ctx` so that new sessions can be
9834   stored in that context to allow for session resumption, even after the
9835   SSL has been switched to a new SSL_CTX in reaction to a client's
9836   server_name extension.
9837
9838   New functions (subject to change):
9839
9840           SSL_get_servername()
9841           SSL_get_servername_type()
9842           SSL_set_SSL_CTX()
9843
9844   New CTRL codes and macros (subject to change):
9845
9846           SSL_CTRL_SET_TLSEXT_SERVERNAME_CB
9847                               - SSL_CTX_set_tlsext_servername_callback()
9848           SSL_CTRL_SET_TLSEXT_SERVERNAME_ARG
9849                                    - SSL_CTX_set_tlsext_servername_arg()
9850           SSL_CTRL_SET_TLSEXT_HOSTNAME           - SSL_set_tlsext_host_name()
9851
9852   openssl s_client has a new '-servername ...' option.
9853
9854   openssl s_server has new options '-servername_host ...', '-cert2 ...',
9855   '-key2 ...', '-servername_fatal' (subject to change).  This allows
9856   testing the HostName extension for a specific single host name ('-cert'
9857   and '-key' remain fallbacks for handshakes without HostName
9858   negotiation).  If the unrecognized_name alert has to be sent, this by
9859   default is a warning; it becomes fatal with the '-servername_fatal'
9860   option.
9861
9862   *Peter Sylvester,  Remy Allais, Christophe Renou*
9863
9864 * Whirlpool hash implementation is added.
9865
9866   *Andy Polyakov*
9867
9868 * BIGNUM code on 64-bit SPARCv9 targets is switched from bn(64,64) to
9869   bn(64,32). Because of instruction set limitations it doesn't have
9870   any negative impact on performance. This was done mostly in order
9871   to make it possible to share assembler modules, such as bn_mul_mont
9872   implementations, between 32- and 64-bit builds without hassle.
9873
9874   *Andy Polyakov*
9875
9876 * Move code previously exiled into file crypto/ec/ec2_smpt.c
9877   to ec2_smpl.c, and no longer require the OPENSSL_EC_BIN_PT_COMP
9878   macro.
9879
9880   *Bodo Moeller*
9881
9882 * New candidate for BIGNUM assembler implementation, bn_mul_mont,
9883   dedicated Montgomery multiplication procedure, is introduced.
9884   BN_MONT_CTX is modified to allow bn_mul_mont to reach for higher
9885   "64-bit" performance on certain 32-bit targets.
9886
9887   *Andy Polyakov*
9888
9889 * New option SSL_OP_NO_COMP to disable use of compression selectively
9890   in SSL structures. New SSL ctrl to set maximum send fragment size.
9891   Save memory by setting the I/O buffer sizes dynamically instead of
9892   using the maximum available value.
9893
9894   *Steve Henson*
9895
9896 * New option -V for 'openssl ciphers'. This prints the ciphersuite code
9897   in addition to the text details.
9898
9899   *Bodo Moeller*
9900
9901 * Very, very preliminary EXPERIMENTAL support for printing of general
9902   ASN1 structures. This currently produces rather ugly output and doesn't
9903   handle several customised structures at all.
9904
9905   *Steve Henson*
9906
9907 * Integrated support for PVK file format and some related formats such
9908   as MS PUBLICKEYBLOB and PRIVATEKEYBLOB. Command line switches to support
9909   these in the 'rsa' and 'dsa' utilities.
9910
9911   *Steve Henson*
9912
9913 * Support for PKCS#1 RSAPublicKey format on rsa utility command line.
9914
9915   *Steve Henson*
9916
9917 * Remove the ancient ASN1_METHOD code. This was only ever used in one
9918   place for the (very old) "NETSCAPE" format certificates which are now
9919   handled using new ASN1 code equivalents.
9920
9921   *Steve Henson*
9922
9923 * Let the TLSv1_method() etc. functions return a 'const' SSL_METHOD
9924   pointer and make the SSL_METHOD parameter in SSL_CTX_new,
9925   SSL_CTX_set_ssl_version and SSL_set_ssl_method 'const'.
9926
9927   *Nils Larsch*
9928
9929 * Modify CRL distribution points extension code to print out previously
9930   unsupported fields. Enhance extension setting code to allow setting of
9931   all fields.
9932
9933   *Steve Henson*
9934
9935 * Add print and set support for Issuing Distribution Point CRL extension.
9936
9937   *Steve Henson*
9938
9939 * Change 'Configure' script to enable Camellia by default.
9940
9941   *NTT*
9942
9943OpenSSL 0.9.x
9944-------------
9945
9946### Changes between 0.9.8m and 0.9.8n [24 Mar 2010]
9947
9948 * When rejecting SSL/TLS records due to an incorrect version number, never
9949   update s->server with a new major version number.  As of
9950   - OpenSSL 0.9.8m if 'short' is a 16-bit type,
9951   - OpenSSL 0.9.8f if 'short' is longer than 16 bits,
9952   the previous behavior could result in a read attempt at NULL when
9953   receiving specific incorrect SSL/TLS records once record payload
9954   protection is active.  ([CVE-2010-0740])
9955
9956   *Bodo Moeller, Adam Langley <agl@chromium.org>*
9957
9958 * Fix for CVE-2010-0433 where some kerberos enabled versions of OpenSSL
9959   could be crashed if the relevant tables were not present (e.g. chrooted).
9960
9961   *Tomas Hoger <thoger@redhat.com>*
9962
9963### Changes between 0.9.8l and 0.9.8m [25 Feb 2010]
9964
9965 * Always check bn_wexpand() return values for failure.  ([CVE-2009-3245])
9966
9967   *Martin Olsson, Neel Mehta*
9968
9969 * Fix X509_STORE locking: Every 'objs' access requires a lock (to
9970   accommodate for stack sorting, always a write lock!).
9971
9972   *Bodo Moeller*
9973
9974 * On some versions of WIN32 Heap32Next is very slow. This can cause
9975   excessive delays in the RAND_poll(): over a minute. As a workaround
9976   include a time check in the inner Heap32Next loop too.
9977
9978   *Steve Henson*
9979
9980 * The code that handled flushing of data in SSL/TLS originally used the
9981   BIO_CTRL_INFO ctrl to see if any data was pending first. This caused
9982   the problem outlined in PR#1949. The fix suggested there however can
9983   trigger problems with buggy BIO_CTRL_WPENDING (e.g. some versions
9984   of Apache). So instead simplify the code to flush unconditionally.
9985   This should be fine since flushing with no data to flush is a no op.
9986
9987   *Steve Henson*
9988
9989 * Handle TLS versions 2.0 and later properly and correctly use the
9990   highest version of TLS/SSL supported. Although TLS >= 2.0 is some way
9991   off ancient servers have a habit of sticking around for a while...
9992
9993   *Steve Henson*
9994
9995 * Modify compression code so it frees up structures without using the
9996   ex_data callbacks. This works around a problem where some applications
9997   call CRYPTO_cleanup_all_ex_data() before application exit (e.g. when
9998   restarting) then use compression (e.g. SSL with compression) later.
9999   This results in significant per-connection memory leaks and
10000   has caused some security issues including CVE-2008-1678 and
10001   CVE-2009-4355.
10002
10003   *Steve Henson*
10004
10005 * Constify crypto/cast (i.e., <openssl/cast.h>): a CAST_KEY doesn't
10006   change when encrypting or decrypting.
10007
10008   *Bodo Moeller*
10009
10010 * Add option SSL_OP_LEGACY_SERVER_CONNECT which will allow clients to
10011   connect and renegotiate with servers which do not support RI.
10012   Until RI is more widely deployed this option is enabled by default.
10013
10014   *Steve Henson*
10015
10016 * Add "missing" ssl ctrls to clear options and mode.
10017
10018   *Steve Henson*
10019
10020 * If client attempts to renegotiate and doesn't support RI respond with
10021   a no_renegotiation alert as required by RFC5746.  Some renegotiating
10022   TLS clients will continue a connection gracefully when they receive
10023   the alert. Unfortunately OpenSSL mishandled this alert and would hang
10024   waiting for a server hello which it will never receive. Now we treat a
10025   received no_renegotiation alert as a fatal error. This is because
10026   applications requesting a renegotiation might well expect it to succeed
10027   and would have no code in place to handle the server denying it so the
10028   only safe thing to do is to terminate the connection.
10029
10030   *Steve Henson*
10031
10032 * Add ctrl macro SSL_get_secure_renegotiation_support() which returns 1 if
10033   peer supports secure renegotiation and 0 otherwise. Print out peer
10034   renegotiation support in s_client/s_server.
10035
10036   *Steve Henson*
10037
10038 * Replace the highly broken and deprecated SPKAC certification method with
10039   the updated NID creation version. This should correctly handle UTF8.
10040
10041   *Steve Henson*
10042
10043 * Implement RFC5746. Re-enable renegotiation but require the extension
10044   as needed. Unfortunately, SSL3_FLAGS_ALLOW_UNSAFE_LEGACY_RENEGOTIATION
10045   turns out to be a bad idea. It has been replaced by
10046   SSL_OP_ALLOW_UNSAFE_LEGACY_RENEGOTIATION which can be set with
10047   SSL_CTX_set_options(). This is really not recommended unless you
10048   know what you are doing.
10049
10050   *Eric Rescorla <ekr@networkresonance.com>, Ben Laurie, Steve Henson*
10051
10052 * Fixes to stateless session resumption handling. Use initial_ctx when
10053   issuing and attempting to decrypt tickets in case it has changed during
10054   servername handling. Use a non-zero length session ID when attempting
10055   stateless session resumption: this makes it possible to determine if
10056   a resumption has occurred immediately after receiving server hello
10057   (several places in OpenSSL subtly assume this) instead of later in
10058   the handshake.
10059
10060   *Steve Henson*
10061
10062 * The functions ENGINE_ctrl(), OPENSSL_isservice(),
10063   CMS_get1_RecipientRequest() and RAND_bytes() can return <=0 on error
10064   fixes for a few places where the return code is not checked
10065   correctly.
10066
10067   *Julia Lawall <julia@diku.dk>*
10068
10069 * Add --strict-warnings option to Configure script to include devteam
10070   warnings in other configurations.
10071
10072   *Steve Henson*
10073
10074 * Add support for --libdir option and LIBDIR variable in makefiles. This
10075   makes it possible to install openssl libraries in locations which
10076   have names other than "lib", for example "/usr/lib64" which some
10077   systems need.
10078
10079   *Steve Henson, based on patch from Jeremy Utley*
10080
10081 * Don't allow the use of leading 0x80 in OIDs. This is a violation of
10082   X690 8.9.12 and can produce some misleading textual output of OIDs.
10083
10084   *Steve Henson, reported by Dan Kaminsky*
10085
10086 * Delete MD2 from algorithm tables. This follows the recommendation in
10087   several standards that it is not used in new applications due to
10088   several cryptographic weaknesses. For binary compatibility reasons
10089   the MD2 API is still compiled in by default.
10090
10091   *Steve Henson*
10092
10093 * Add compression id to {d2i,i2d}_SSL_SESSION so it is correctly saved
10094   and restored.
10095
10096   *Steve Henson*
10097
10098 * Rename uni2asc and asc2uni functions to OPENSSL_uni2asc and
10099   OPENSSL_asc2uni conditionally on Netware platforms to avoid a name
10100   clash.
10101
10102   *Guenter <lists@gknw.net>*
10103
10104 * Fix the server certificate chain building code to use X509_verify_cert(),
10105   it used to have an ad-hoc builder which was unable to cope with anything
10106   other than a simple chain.
10107
10108   *David Woodhouse <dwmw2@infradead.org>, Steve Henson*
10109
10110 * Don't check self signed certificate signatures in X509_verify_cert()
10111   by default (a flag can override this): it just wastes time without
10112   adding any security. As a useful side effect self signed root CAs
10113   with non-FIPS digests are now usable in FIPS mode.
10114
10115   *Steve Henson*
10116
10117 * In dtls1_process_out_of_seq_message() the check if the current message
10118   is already buffered was missing. For every new message was memory
10119   allocated, allowing an attacker to perform an denial of service attack
10120   with sending out of seq handshake messages until there is no memory
10121   left. Additionally every future message was buffered, even if the
10122   sequence number made no sense and would be part of another handshake.
10123   So only messages with sequence numbers less than 10 in advance will be
10124   buffered.  ([CVE-2009-1378])
10125
10126   *Robin Seggelmann, discovered by Daniel Mentz*
10127
10128 * Records are buffered if they arrive with a future epoch to be
10129   processed after finishing the corresponding handshake. There is
10130   currently no limitation to this buffer allowing an attacker to perform
10131   a DOS attack with sending records with future epochs until there is no
10132   memory left. This patch adds the pqueue_size() function to determine
10133   the size of a buffer and limits the record buffer to 100 entries.
10134   ([CVE-2009-1377])
10135
10136   *Robin Seggelmann, discovered by Daniel Mentz*
10137
10138 * Keep a copy of frag->msg_header.frag_len so it can be used after the
10139   parent structure is freed.  ([CVE-2009-1379])
10140
10141   *Daniel Mentz*
10142
10143 * Handle non-blocking I/O properly in SSL_shutdown() call.
10144
10145   *Darryl Miles <darryl-mailinglists@netbauds.net>*
10146
10147 * Add `2.5.4.*` OIDs
10148
10149   *Ilya O. <vrghost@gmail.com>*
10150
10151### Changes between 0.9.8k and 0.9.8l  [5 Nov 2009]
10152
10153 * Disable renegotiation completely - this fixes a severe security
10154   problem ([CVE-2009-3555]) at the cost of breaking all
10155   renegotiation. Renegotiation can be re-enabled by setting
10156   SSL3_FLAGS_ALLOW_UNSAFE_LEGACY_RENEGOTIATION in s3->flags at
10157   run-time. This is really not recommended unless you know what
10158   you're doing.
10159
10160   *Ben Laurie*
10161
10162### Changes between 0.9.8j and 0.9.8k  [25 Mar 2009]
10163
10164 * Don't set val to NULL when freeing up structures, it is freed up by
10165   underlying code. If `sizeof(void *) > sizeof(long)` this can result in
10166   zeroing past the valid field. ([CVE-2009-0789])
10167
10168   *Paolo Ganci <Paolo.Ganci@AdNovum.CH>*
10169
10170 * Fix bug where return value of CMS_SignerInfo_verify_content() was not
10171   checked correctly. This would allow some invalid signed attributes to
10172   appear to verify correctly. ([CVE-2009-0591])
10173
10174   *Ivan Nestlerode <inestlerode@us.ibm.com>*
10175
10176 * Reject UniversalString and BMPString types with invalid lengths. This
10177   prevents a crash in ASN1_STRING_print_ex() which assumes the strings have
10178   a legal length. ([CVE-2009-0590])
10179
10180   *Steve Henson*
10181
10182 * Set S/MIME signing as the default purpose rather than setting it
10183   unconditionally. This allows applications to override it at the store
10184   level.
10185
10186   *Steve Henson*
10187
10188 * Permit restricted recursion of ASN1 strings. This is needed in practice
10189   to handle some structures.
10190
10191   *Steve Henson*
10192
10193 * Improve efficiency of mem_gets: don't search whole buffer each time
10194   for a '\n'
10195
10196   *Jeremy Shapiro <jnshapir@us.ibm.com>*
10197
10198 * New -hex option for openssl rand.
10199
10200   *Matthieu Herrb*
10201
10202 * Print out UTF8String and NumericString when parsing ASN1.
10203
10204   *Steve Henson*
10205
10206 * Support NumericString type for name components.
10207
10208   *Steve Henson*
10209
10210 * Allow CC in the environment to override the automatically chosen
10211   compiler. Note that nothing is done to ensure flags work with the
10212   chosen compiler.
10213
10214   *Ben Laurie*
10215
10216### Changes between 0.9.8i and 0.9.8j  [07 Jan 2009]
10217
10218 * Properly check EVP_VerifyFinal() and similar return values
10219   ([CVE-2008-5077]).
10220
10221   *Ben Laurie, Bodo Moeller, Google Security Team*
10222
10223 * Enable TLS extensions by default.
10224
10225   *Ben Laurie*
10226
10227 * Allow the CHIL engine to be loaded, whether the application is
10228   multithreaded or not. (This does not release the developer from the
10229   obligation to set up the dynamic locking callbacks.)
10230
10231   *Sander Temme <sander@temme.net>*
10232
10233 * Use correct exit code if there is an error in dgst command.
10234
10235   *Steve Henson; problem pointed out by Roland Dirlewanger*
10236
10237 * Tweak Configure so that you need to say "experimental-jpake" to enable
10238   JPAKE, and need to use -DOPENSSL_EXPERIMENTAL_JPAKE in applications.
10239
10240   *Bodo Moeller*
10241
10242 * Add experimental JPAKE support, including demo authentication in
10243   s_client and s_server.
10244
10245   *Ben Laurie*
10246
10247 * Set the comparison function in v3_addr_canonize().
10248
10249   *Rob Austein <sra@hactrn.net>*
10250
10251 * Add support for XMPP STARTTLS in s_client.
10252
10253   *Philip Paeps <philip@freebsd.org>*
10254
10255 * Change the server-side SSL_OP_NETSCAPE_REUSE_CIPHER_CHANGE_BUG behavior
10256   to ensure that even with this option, only ciphersuites in the
10257   server's preference list will be accepted.  (Note that the option
10258   applies only when resuming a session, so the earlier behavior was
10259   just about the algorithm choice for symmetric cryptography.)
10260
10261   *Bodo Moeller*
10262
10263### Changes between 0.9.8h and 0.9.8i  [15 Sep 2008]
10264
10265 * Fix NULL pointer dereference if a DTLS server received
10266   ChangeCipherSpec as first record ([CVE-2009-1386]).
10267
10268   *PR #1679*
10269
10270 * Fix a state transition in s3_srvr.c and d1_srvr.c
10271   (was using SSL3_ST_CW_CLNT_HELLO_B, should be `..._ST_SW_SRVR_...`).
10272
10273   *Nagendra Modadugu*
10274
10275 * The fix in 0.9.8c that supposedly got rid of unsafe
10276   double-checked locking was incomplete for RSA blinding,
10277   addressing just one layer of what turns out to have been
10278   doubly unsafe triple-checked locking.
10279
10280   So now fix this for real by retiring the MONT_HELPER macro
10281   in crypto/rsa/rsa_eay.c.
10282
10283   *Bodo Moeller; problem pointed out by Marius Schilder*
10284
10285 * Various precautionary measures:
10286
10287   - Avoid size_t integer overflow in HASH_UPDATE (md32_common.h).
10288
10289   - Avoid a buffer overflow in d2i_SSL_SESSION() (ssl_asn1.c).
10290     (NB: This would require knowledge of the secret session ticket key
10291     to exploit, in which case you'd be SOL either way.)
10292
10293   - Change bn_nist.c so that it will properly handle input BIGNUMs
10294     outside the expected range.
10295
10296   - Enforce the 'num' check in BN_div() (bn_div.c) for non-BN_DEBUG
10297     builds.
10298
10299   *Neel Mehta, Bodo Moeller*
10300
10301 * Allow engines to be "soft loaded" - i.e. optionally don't die if
10302   the load fails. Useful for distros.
10303
10304   *Ben Laurie and the FreeBSD team*
10305
10306 * Add support for Local Machine Keyset attribute in PKCS#12 files.
10307
10308   *Steve Henson*
10309
10310 * Fix BN_GF2m_mod_arr() top-bit cleanup code.
10311
10312   *Huang Ying*
10313
10314 * Expand ENGINE to support engine supplied SSL client certificate functions.
10315
10316   This work was sponsored by Logica.
10317
10318   *Steve Henson*
10319
10320 * Add CryptoAPI ENGINE to support use of RSA and DSA keys held in Windows
10321   keystores. Support for SSL/TLS client authentication too.
10322   Not compiled unless enable-capieng specified to Configure.
10323
10324   This work was sponsored by Logica.
10325
10326   *Steve Henson*
10327
10328 * Fix bug in X509_ATTRIBUTE creation: don't set attribute using
10329   ASN1_TYPE_set1 if MBSTRING flag set. This bug would crash certain
10330   attribute creation routines such as certificate requests and PKCS#12
10331   files.
10332
10333   *Steve Henson*
10334
10335### Changes between 0.9.8g and 0.9.8h  [28 May 2008]
10336
10337 * Fix flaw if 'Server Key exchange message' is omitted from a TLS
10338   handshake which could lead to a client crash as found using the
10339   Codenomicon TLS test suite ([CVE-2008-1672])
10340
10341   *Steve Henson, Mark Cox*
10342
10343 * Fix double free in TLS server name extensions which could lead to
10344   a remote crash found by Codenomicon TLS test suite ([CVE-2008-0891])
10345
10346   *Joe Orton*
10347
10348 * Clear error queue in SSL_CTX_use_certificate_chain_file()
10349
10350   Clear the error queue to ensure that error entries left from
10351   older function calls do not interfere with the correct operation.
10352
10353   *Lutz Jaenicke, Erik de Castro Lopo*
10354
10355 * Remove root CA certificates of commercial CAs:
10356
10357   The OpenSSL project does not recommend any specific CA and does not
10358   have any policy with respect to including or excluding any CA.
10359   Therefore it does not make any sense to ship an arbitrary selection
10360   of root CA certificates with the OpenSSL software.
10361
10362   *Lutz Jaenicke*
10363
10364 * RSA OAEP patches to fix two separate invalid memory reads.
10365   The first one involves inputs when 'lzero' is greater than
10366   'SHA_DIGEST_LENGTH' (it would read about SHA_DIGEST_LENGTH bytes
10367   before the beginning of from). The second one involves inputs where
10368   the 'db' section contains nothing but zeroes (there is a one-byte
10369   invalid read after the end of 'db').
10370
10371   *Ivan Nestlerode <inestlerode@us.ibm.com>*
10372
10373 * Partial backport from 0.9.9-dev:
10374
10375   Introduce bn_mul_mont (dedicated Montgomery multiplication
10376   procedure) as a candidate for BIGNUM assembler implementation.
10377   While 0.9.9-dev uses assembler for various architectures, only
10378   x86_64 is available by default here in the 0.9.8 branch, and
10379   32-bit x86 is available through a compile-time setting.
10380
10381   To try the 32-bit x86 assembler implementation, use Configure
10382   option "enable-montasm" (which exists only for this backport).
10383
10384   As "enable-montasm" for 32-bit x86 disclaims code stability
10385   anyway, in this constellation we activate additional code
10386   backported from 0.9.9-dev for further performance improvements,
10387   namely BN_from_montgomery_word.  (To enable this otherwise,
10388   e.g. x86_64, try `-DMONT_FROM_WORD___NON_DEFAULT_0_9_8_BUILD`.)
10389
10390   *Andy Polyakov (backport partially by Bodo Moeller)*
10391
10392 * Add TLS session ticket callback. This allows an application to set
10393   TLS ticket cipher and HMAC keys rather than relying on hardcoded fixed
10394   values. This is useful for key rollover for example where several key
10395   sets may exist with different names.
10396
10397   *Steve Henson*
10398
10399 * Reverse ENGINE-internal logic for caching default ENGINE handles.
10400   This was broken until now in 0.9.8 releases, such that the only way
10401   a registered ENGINE could be used (assuming it initialises
10402   successfully on the host) was to explicitly set it as the default
10403   for the relevant algorithms. This is in contradiction with 0.9.7
10404   behaviour and the documentation. With this fix, when an ENGINE is
10405   registered into a given algorithm's table of implementations, the
10406   'uptodate' flag is reset so that auto-discovery will be used next
10407   time a new context for that algorithm attempts to select an
10408   implementation.
10409
10410   *Ian Lister (tweaked by Geoff Thorpe)*
10411
10412 * Backport of CMS code to OpenSSL 0.9.8. This differs from the 0.9.9
10413   implementation in the following ways:
10414
10415   Lack of EVP_PKEY_ASN1_METHOD means algorithm parameters have to be
10416   hard coded.
10417
10418   Lack of BER streaming support means one pass streaming processing is
10419   only supported if data is detached: setting the streaming flag is
10420   ignored for embedded content.
10421
10422   CMS support is disabled by default and must be explicitly enabled
10423   with the enable-cms configuration option.
10424
10425   *Steve Henson*
10426
10427 * Update the GMP engine glue to do direct copies between BIGNUM and
10428   mpz_t when openssl and GMP use the same limb size. Otherwise the
10429   existing "conversion via a text string export" trick is still used.
10430
10431   *Paul Sheer <paulsheer@gmail.com>*
10432
10433 * Zlib compression BIO. This is a filter BIO which compressed and
10434   uncompresses any data passed through it.
10435
10436   *Steve Henson*
10437
10438 * Add AES_wrap_key() and AES_unwrap_key() functions to implement
10439   RFC3394 compatible AES key wrapping.
10440
10441   *Steve Henson*
10442
10443 * Add utility functions to handle ASN1 structures. ASN1_STRING_set0():
10444   sets string data without copying. X509_ALGOR_set0() and
10445   X509_ALGOR_get0(): set and retrieve X509_ALGOR (AlgorithmIdentifier)
10446   data. Attribute function X509at_get0_data_by_OBJ(): retrieves data
10447   from an X509_ATTRIBUTE structure optionally checking it occurs only
10448   once. ASN1_TYPE_set1(): set and ASN1_TYPE structure copying supplied
10449   data.
10450
10451   *Steve Henson*
10452
10453 * Fix BN flag handling in RSA_eay_mod_exp() and BN_MONT_CTX_set()
10454   to get the expected BN_FLG_CONSTTIME behavior.
10455
10456   *Bodo Moeller (Google)*
10457
10458 * Netware support:
10459
10460   - fixed wrong usage of ioctlsocket() when build for LIBC BSD sockets
10461   - fixed do_tests.pl to run the test suite with CLIB builds too (CLIB_OPT)
10462   - added some more tests to do_tests.pl
10463   - fixed RunningProcess usage so that it works with newer LIBC NDKs too
10464   - removed usage of BN_LLONG for CLIB builds to avoid runtime dependency
10465   - added new Configure targets netware-clib-bsdsock, netware-clib-gcc,
10466     netware-clib-bsdsock-gcc, netware-libc-bsdsock-gcc
10467   - various changes to netware.pl to enable gcc-cross builds on Win32
10468     platform
10469   - changed crypto/bio/b_sock.c to work with macro functions (CLIB BSD)
10470   - various changes to fix missing prototype warnings
10471   - fixed x86nasm.pl to create correct asm files for NASM COFF output
10472   - added AES, WHIRLPOOL and CPUID assembler code to build files
10473   - added missing AES assembler make rules to mk1mf.pl
10474   - fixed order of includes in `apps/ocsp.c` so that `e_os.h` settings apply
10475
10476   *Guenter Knauf <eflash@gmx.net>*
10477
10478 * Implement certificate status request TLS extension defined in RFC3546.
10479   A client can set the appropriate parameters and receive the encoded
10480   OCSP response via a callback. A server can query the supplied parameters
10481   and set the encoded OCSP response in the callback. Add simplified examples
10482   to s_client and s_server.
10483
10484   *Steve Henson*
10485
10486### Changes between 0.9.8f and 0.9.8g  [19 Oct 2007]
10487
10488 * Fix various bugs:
10489   + Binary incompatibility of ssl_ctx_st structure
10490   + DTLS interoperation with non-compliant servers
10491   + Don't call get_session_cb() without proposed session
10492   + Fix ia64 assembler code
10493
10494   *Andy Polyakov, Steve Henson*
10495
10496### Changes between 0.9.8e and 0.9.8f  [11 Oct 2007]
10497
10498 * DTLS Handshake overhaul. There were longstanding issues with
10499   OpenSSL DTLS implementation, which were making it impossible for
10500   RFC 4347 compliant client to communicate with OpenSSL server.
10501   Unfortunately just fixing these incompatibilities would "cut off"
10502   pre-0.9.8f clients. To allow for hassle free upgrade post-0.9.8e
10503   server keeps tolerating non RFC compliant syntax. The opposite is
10504   not true, 0.9.8f client can not communicate with earlier server.
10505   This update even addresses CVE-2007-4995.
10506
10507   *Andy Polyakov*
10508
10509 * Changes to avoid need for function casts in OpenSSL: some compilers
10510   (gcc 4.2 and later) reject their use.
10511   *Kurt Roeckx <kurt@roeckx.be>, Peter Hartley <pdh@utter.chaos.org.uk>,
10512    Steve Henson*
10513
10514 * Add RFC4507 support to OpenSSL. This includes the corrections in
10515   RFC4507bis. The encrypted ticket format is an encrypted encoded
10516   SSL_SESSION structure, that way new session features are automatically
10517   supported.
10518
10519   If a client application caches session in an SSL_SESSION structure
10520   support is transparent because tickets are now stored in the encoded
10521   SSL_SESSION.
10522
10523   The SSL_CTX structure automatically generates keys for ticket
10524   protection in servers so again support should be possible
10525   with no application modification.
10526
10527   If a client or server wishes to disable RFC4507 support then the option
10528   SSL_OP_NO_TICKET can be set.
10529
10530   Add a TLS extension debugging callback to allow the contents of any client
10531   or server extensions to be examined.
10532
10533   This work was sponsored by Google.
10534
10535   *Steve Henson*
10536
10537 * Add initial support for TLS extensions, specifically for the server_name
10538   extension so far.  The SSL_SESSION, SSL_CTX, and SSL data structures now
10539   have new members for a host name.  The SSL data structure has an
10540   additional member `SSL_CTX *initial_ctx` so that new sessions can be
10541   stored in that context to allow for session resumption, even after the
10542   SSL has been switched to a new SSL_CTX in reaction to a client's
10543   server_name extension.
10544
10545   New functions (subject to change):
10546
10547           SSL_get_servername()
10548           SSL_get_servername_type()
10549           SSL_set_SSL_CTX()
10550
10551   New CTRL codes and macros (subject to change):
10552
10553           SSL_CTRL_SET_TLSEXT_SERVERNAME_CB
10554                               - SSL_CTX_set_tlsext_servername_callback()
10555           SSL_CTRL_SET_TLSEXT_SERVERNAME_ARG
10556                                    - SSL_CTX_set_tlsext_servername_arg()
10557           SSL_CTRL_SET_TLSEXT_HOSTNAME           - SSL_set_tlsext_host_name()
10558
10559   openssl s_client has a new '-servername ...' option.
10560
10561   openssl s_server has new options '-servername_host ...', '-cert2 ...',
10562   '-key2 ...', '-servername_fatal' (subject to change).  This allows
10563   testing the HostName extension for a specific single host name ('-cert'
10564   and '-key' remain fallbacks for handshakes without HostName
10565   negotiation).  If the unrecognized_name alert has to be sent, this by
10566   default is a warning; it becomes fatal with the '-servername_fatal'
10567   option.
10568
10569   *Peter Sylvester,  Remy Allais, Christophe Renou, Steve Henson*
10570
10571 * Add AES and SSE2 assembly language support to VC++ build.
10572
10573   *Steve Henson*
10574
10575 * Mitigate attack on final subtraction in Montgomery reduction.
10576
10577   *Andy Polyakov*
10578
10579 * Fix crypto/ec/ec_mult.c to work properly with scalars of value 0
10580   (which previously caused an internal error).
10581
10582   *Bodo Moeller*
10583
10584 * Squeeze another 10% out of IGE mode when in != out.
10585
10586   *Ben Laurie*
10587
10588 * AES IGE mode speedup.
10589
10590   *Dean Gaudet (Google)*
10591
10592 * Add the Korean symmetric 128-bit cipher SEED (see
10593   <http://www.kisa.or.kr/kisa/seed/jsp/seed_eng.jsp>) and
10594   add SEED ciphersuites from RFC 4162:
10595
10596           TLS_RSA_WITH_SEED_CBC_SHA      =  "SEED-SHA"
10597           TLS_DHE_DSS_WITH_SEED_CBC_SHA  =  "DHE-DSS-SEED-SHA"
10598           TLS_DHE_RSA_WITH_SEED_CBC_SHA  =  "DHE-RSA-SEED-SHA"
10599           TLS_DH_anon_WITH_SEED_CBC_SHA  =  "ADH-SEED-SHA"
10600
10601   To minimize changes between patchlevels in the OpenSSL 0.9.8
10602   series, SEED remains excluded from compilation unless OpenSSL
10603   is configured with 'enable-seed'.
10604
10605   *KISA, Bodo Moeller*
10606
10607 * Mitigate branch prediction attacks, which can be practical if a
10608   single processor is shared, allowing a spy process to extract
10609   information.  For detailed background information, see
10610   <http://eprint.iacr.org/2007/039> (O. Aciicmez, S. Gueron,
10611   J.-P. Seifert, "New Branch Prediction Vulnerabilities in OpenSSL
10612   and Necessary Software Countermeasures").  The core of the change
10613   are new versions BN_div_no_branch() and
10614   BN_mod_inverse_no_branch() of BN_div() and BN_mod_inverse(),
10615   respectively, which are slower, but avoid the security-relevant
10616   conditional branches.  These are automatically called by BN_div()
10617   and BN_mod_inverse() if the flag BN_FLG_CONSTTIME is set for one
10618   of the input BIGNUMs.  Also, BN_is_bit_set() has been changed to
10619   remove a conditional branch.
10620
10621   BN_FLG_CONSTTIME is the new name for the previous
10622   BN_FLG_EXP_CONSTTIME flag, since it now affects more than just
10623   modular exponentiation.  (Since OpenSSL 0.9.7h, setting this flag
10624   in the exponent causes BN_mod_exp_mont() to use the alternative
10625   implementation in BN_mod_exp_mont_consttime().)  The old name
10626   remains as a deprecated alias.
10627
10628   Similarly, RSA_FLAG_NO_EXP_CONSTTIME is replaced by a more general
10629   RSA_FLAG_NO_CONSTTIME flag since the RSA implementation now uses
10630   constant-time implementations for more than just exponentiation.
10631   Here too the old name is kept as a deprecated alias.
10632
10633   BN_BLINDING_new() will now use BN_dup() for the modulus so that
10634   the BN_BLINDING structure gets an independent copy of the
10635   modulus.  This means that the previous `BIGNUM *m` argument to
10636   BN_BLINDING_new() and to BN_BLINDING_create_param() now
10637   essentially becomes `const BIGNUM *m`, although we can't actually
10638   change this in the header file before 0.9.9.  It allows
10639   RSA_setup_blinding() to use BN_with_flags() on the modulus to
10640   enable BN_FLG_CONSTTIME.
10641
10642   *Matthew D Wood (Intel Corp)*
10643
10644 * In the SSL/TLS server implementation, be strict about session ID
10645   context matching (which matters if an application uses a single
10646   external cache for different purposes).  Previously,
10647   out-of-context reuse was forbidden only if SSL_VERIFY_PEER was
10648   set.  This did ensure strict client verification, but meant that,
10649   with applications using a single external cache for quite
10650   different requirements, clients could circumvent ciphersuite
10651   restrictions for a given session ID context by starting a session
10652   in a different context.
10653
10654   *Bodo Moeller*
10655
10656 * Include "!eNULL" in SSL_DEFAULT_CIPHER_LIST to make sure that
10657   a ciphersuite string such as "DEFAULT:RSA" cannot enable
10658   authentication-only ciphersuites.
10659
10660   *Bodo Moeller*
10661
10662 * Update the SSL_get_shared_ciphers() fix CVE-2006-3738 which was
10663   not complete and could lead to a possible single byte overflow
10664   ([CVE-2007-5135]) [Ben Laurie]
10665
10666### Changes between 0.9.8d and 0.9.8e  [23 Feb 2007]
10667
10668 * Since AES128 and AES256 (and similarly Camellia128 and
10669   Camellia256) share a single mask bit in the logic of
10670   ssl/ssl_ciph.c, the code for masking out disabled ciphers needs a
10671   kludge to work properly if AES128 is available and AES256 isn't
10672   (or if Camellia128 is available and Camellia256 isn't).
10673
10674   *Victor Duchovni*
10675
10676 * Fix the BIT STRING encoding generated by crypto/ec/ec_asn1.c
10677   (within i2d_ECPrivateKey, i2d_ECPKParameters, i2d_ECParameters):
10678   When a point or a seed is encoded in a BIT STRING, we need to
10679   prevent the removal of trailing zero bits to get the proper DER
10680   encoding.  (By default, crypto/asn1/a_bitstr.c assumes the case
10681   of a NamedBitList, for which trailing 0 bits need to be removed.)
10682
10683   *Bodo Moeller*
10684
10685 * Have SSL/TLS server implementation tolerate "mismatched" record
10686   protocol version while receiving ClientHello even if the
10687   ClientHello is fragmented.  (The server can't insist on the
10688   particular protocol version it has chosen before the ServerHello
10689   message has informed the client about his choice.)
10690
10691   *Bodo Moeller*
10692
10693 * Add RFC 3779 support.
10694
10695   *Rob Austein for ARIN, Ben Laurie*
10696
10697 * Load error codes if they are not already present instead of using a
10698   static variable. This allows them to be cleanly unloaded and reloaded.
10699   Improve header file function name parsing.
10700
10701   *Steve Henson*
10702
10703 * extend SMTP and IMAP protocol emulation in s_client to use EHLO
10704   or CAPABILITY handshake as required by RFCs.
10705
10706   *Goetz Babin-Ebell*
10707
10708### Changes between 0.9.8c and 0.9.8d  [28 Sep 2006]
10709
10710 * Introduce limits to prevent malicious keys being able to
10711   cause a denial of service.  ([CVE-2006-2940])
10712
10713   *Steve Henson, Bodo Moeller*
10714
10715 * Fix ASN.1 parsing of certain invalid structures that can result
10716   in a denial of service.  ([CVE-2006-2937])  [Steve Henson]
10717
10718 * Fix buffer overflow in SSL_get_shared_ciphers() function.
10719   ([CVE-2006-3738]) [Tavis Ormandy and Will Drewry, Google Security Team]
10720
10721 * Fix SSL client code which could crash if connecting to a
10722   malicious SSLv2 server.  ([CVE-2006-4343])
10723
10724   *Tavis Ormandy and Will Drewry, Google Security Team*
10725
10726 * Since 0.9.8b, ciphersuite strings naming explicit ciphersuites
10727   match only those.  Before that, "AES256-SHA" would be interpreted
10728   as a pattern and match "AES128-SHA" too (since AES128-SHA got
10729   the same strength classification in 0.9.7h) as we currently only
10730   have a single AES bit in the ciphersuite description bitmap.
10731   That change, however, also applied to ciphersuite strings such as
10732   "RC4-MD5" that intentionally matched multiple ciphersuites --
10733   namely, SSL 2.0 ciphersuites in addition to the more common ones
10734   from SSL 3.0/TLS 1.0.
10735
10736   So we change the selection algorithm again: Naming an explicit
10737   ciphersuite selects this one ciphersuite, and any other similar
10738   ciphersuite (same bitmap) from *other* protocol versions.
10739   Thus, "RC4-MD5" again will properly select both the SSL 2.0
10740   ciphersuite and the SSL 3.0/TLS 1.0 ciphersuite.
10741
10742   Since SSL 2.0 does not have any ciphersuites for which the
10743   128/256 bit distinction would be relevant, this works for now.
10744   The proper fix will be to use different bits for AES128 and
10745   AES256, which would have avoided the problems from the beginning;
10746   however, bits are scarce, so we can only do this in a new release
10747   (not just a patchlevel) when we can change the SSL_CIPHER
10748   definition to split the single 'unsigned long mask' bitmap into
10749   multiple values to extend the available space.
10750
10751   *Bodo Moeller*
10752
10753### Changes between 0.9.8b and 0.9.8c  [05 Sep 2006]
10754
10755 * Avoid PKCS #1 v1.5 signature attack discovered by Daniel Bleichenbacher
10756   ([CVE-2006-4339])  [Ben Laurie and Google Security Team]
10757
10758 * Add AES IGE and biIGE modes.
10759
10760   *Ben Laurie*
10761
10762 * Change the Unix randomness entropy gathering to use poll() when
10763   possible instead of select(), since the latter has some
10764   undesirable limitations.
10765
10766   *Darryl Miles via Richard Levitte and Bodo Moeller*
10767
10768 * Disable "ECCdraft" ciphersuites more thoroughly.  Now special
10769   treatment in ssl/ssl_ciph.s makes sure that these ciphersuites
10770   cannot be implicitly activated as part of, e.g., the "AES" alias.
10771   However, please upgrade to OpenSSL 0.9.9[-dev] for
10772   non-experimental use of the ECC ciphersuites to get TLS extension
10773   support, which is required for curve and point format negotiation
10774   to avoid potential handshake problems.
10775
10776   *Bodo Moeller*
10777
10778 * Disable rogue ciphersuites:
10779
10780   - SSLv2 0x08 0x00 0x80 ("RC4-64-MD5")
10781   - SSLv3/TLSv1 0x00 0x61 ("EXP1024-RC2-CBC-MD5")
10782   - SSLv3/TLSv1 0x00 0x60 ("EXP1024-RC4-MD5")
10783
10784   The latter two were purportedly from
10785   draft-ietf-tls-56-bit-ciphersuites-0[01].txt, but do not really
10786   appear there.
10787
10788   Also deactivate the remaining ciphersuites from
10789   draft-ietf-tls-56-bit-ciphersuites-01.txt.  These are just as
10790   unofficial, and the ID has long expired.
10791
10792   *Bodo Moeller*
10793
10794 * Fix RSA blinding Heisenbug (problems sometimes occurred on
10795   dual-core machines) and other potential thread-safety issues.
10796
10797   *Bodo Moeller*
10798
10799 * Add the symmetric cipher Camellia (128-bit, 192-bit, 256-bit key
10800   versions), which is now available for royalty-free use
10801   (see <http://info.isl.ntt.co.jp/crypt/eng/info/chiteki.html>).
10802   Also, add Camellia TLS ciphersuites from RFC 4132.
10803
10804   To minimize changes between patchlevels in the OpenSSL 0.9.8
10805   series, Camellia remains excluded from compilation unless OpenSSL
10806   is configured with 'enable-camellia'.
10807
10808   *NTT*
10809
10810 * Disable the padding bug check when compression is in use. The padding
10811   bug check assumes the first packet is of even length, this is not
10812   necessarily true if compression is enabled and can result in false
10813   positives causing handshake failure. The actual bug test is ancient
10814   code so it is hoped that implementations will either have fixed it by
10815   now or any which still have the bug do not support compression.
10816
10817   *Steve Henson*
10818
10819### Changes between 0.9.8a and 0.9.8b  [04 May 2006]
10820
10821 * When applying a cipher rule check to see if string match is an explicit
10822   cipher suite and only match that one cipher suite if it is.
10823
10824   *Steve Henson*
10825
10826 * Link in manifests for VC++ if needed.
10827
10828   *Austin Ziegler <halostatue@gmail.com>*
10829
10830 * Update support for ECC-based TLS ciphersuites according to
10831   draft-ietf-tls-ecc-12.txt with proposed changes (but without
10832   TLS extensions, which are supported starting with the 0.9.9
10833   branch, not in the OpenSSL 0.9.8 branch).
10834
10835   *Douglas Stebila*
10836
10837 * New functions EVP_CIPHER_CTX_new() and EVP_CIPHER_CTX_free() to support
10838   opaque EVP_CIPHER_CTX handling.
10839
10840   *Steve Henson*
10841
10842 * Fixes and enhancements to zlib compression code. We now only use
10843   "zlib1.dll" and use the default `__cdecl` calling convention on Win32
10844   to conform with the standards mentioned here:
10845   <http://www.zlib.net/DLL_FAQ.txt>
10846   Static zlib linking now works on Windows and the new --with-zlib-include
10847   --with-zlib-lib options to Configure can be used to supply the location
10848   of the headers and library. Gracefully handle case where zlib library
10849   can't be loaded.
10850
10851   *Steve Henson*
10852
10853 * Several fixes and enhancements to the OID generation code. The old code
10854   sometimes allowed invalid OIDs (1.X for X >= 40 for example), couldn't
10855   handle numbers larger than ULONG_MAX, truncated printing and had a
10856   non standard OBJ_obj2txt() behaviour.
10857
10858   *Steve Henson*
10859
10860 * Add support for building of engines under engine/ as shared libraries
10861   under VC++ build system.
10862
10863   *Steve Henson*
10864
10865 * Corrected the numerous bugs in the Win32 path splitter in DSO.
10866   Hopefully, we will not see any false combination of paths any more.
10867
10868   *Richard Levitte*
10869
10870### Changes between 0.9.8 and 0.9.8a  [11 Oct 2005]
10871
10872 * Remove the functionality of SSL_OP_MSIE_SSLV2_RSA_PADDING
10873   (part of SSL_OP_ALL).  This option used to disable the
10874   countermeasure against man-in-the-middle protocol-version
10875   rollback in the SSL 2.0 server implementation, which is a bad
10876   idea.  ([CVE-2005-2969])
10877
10878   *Bodo Moeller; problem pointed out by Yutaka Oiwa (Research Center
10879   for Information Security, National Institute of Advanced Industrial
10880   Science and Technology [AIST], Japan)*
10881
10882 * Add two function to clear and return the verify parameter flags.
10883
10884   *Steve Henson*
10885
10886 * Keep cipherlists sorted in the source instead of sorting them at
10887   runtime, thus removing the need for a lock.
10888
10889   *Nils Larsch*
10890
10891 * Avoid some small subgroup attacks in Diffie-Hellman.
10892
10893   *Nick Mathewson and Ben Laurie*
10894
10895 * Add functions for well-known primes.
10896
10897   *Nick Mathewson*
10898
10899 * Extended Windows CE support.
10900
10901   *Satoshi Nakamura and Andy Polyakov*
10902
10903 * Initialize SSL_METHOD structures at compile time instead of during
10904   runtime, thus removing the need for a lock.
10905
10906   *Steve Henson*
10907
10908 * Make PKCS7_decrypt() work even if no certificate is supplied by
10909   attempting to decrypt each encrypted key in turn. Add support to
10910   smime utility.
10911
10912   *Steve Henson*
10913
10914### Changes between 0.9.7h and 0.9.8  [05 Jul 2005]
10915
10916[NB: OpenSSL 0.9.7i and later 0.9.7 patch levels were released after
10917OpenSSL 0.9.8.]
10918
10919 * Add libcrypto.pc and libssl.pc for those who feel they need them.
10920
10921   *Richard Levitte*
10922
10923 * Change CA.sh and CA.pl so they don't bundle the CSR and the private
10924   key into the same file any more.
10925
10926   *Richard Levitte*
10927
10928 * Add initial support for Win64, both IA64 and AMD64/x64 flavors.
10929
10930   *Andy Polyakov*
10931
10932 * Add -utf8 command line and config file option to 'ca'.
10933
10934   *Stefan <stf@udoma.org*
10935
10936 * Removed the macro des_crypt(), as it seems to conflict with some
10937   libraries.  Use DES_crypt().
10938
10939   *Richard Levitte*
10940
10941 * Correct naming of the 'chil' and '4758cca' ENGINEs. This
10942   involves renaming the source and generated shared-libs for
10943   both. The engines will accept the corrected or legacy ids
10944   ('ncipher' and '4758_cca' respectively) when binding. NB,
10945   this only applies when building 'shared'.
10946
10947   *Corinna Vinschen <vinschen@redhat.com> and Geoff Thorpe*
10948
10949 * Add attribute functions to EVP_PKEY structure. Modify
10950   PKCS12_create() to recognize a CSP name attribute and
10951   use it. Make -CSP option work again in pkcs12 utility.
10952
10953   *Steve Henson*
10954
10955 * Add new functionality to the bn blinding code:
10956   - automatic re-creation of the BN_BLINDING parameters after
10957     a fixed number of uses (currently 32)
10958   - add new function for parameter creation
10959   - introduce flags to control the update behaviour of the
10960     BN_BLINDING parameters
10961   - hide BN_BLINDING structure
10962   Add a second BN_BLINDING slot to the RSA structure to improve
10963   performance when a single RSA object is shared among several
10964   threads.
10965
10966   *Nils Larsch*
10967
10968 * Add support for DTLS.
10969
10970   *Nagendra Modadugu <nagendra@cs.stanford.edu> and Ben Laurie*
10971
10972 * Add support for DER encoded private keys (SSL_FILETYPE_ASN1)
10973   to SSL_CTX_use_PrivateKey_file() and SSL_use_PrivateKey_file()
10974
10975   *Walter Goulet*
10976
10977 * Remove buggy and incomplete DH cert support from
10978   ssl/ssl_rsa.c and ssl/s3_both.c
10979
10980   *Nils Larsch*
10981
10982 * Use SHA-1 instead of MD5 as the default digest algorithm for
10983   the `apps/openssl` commands.
10984
10985   *Nils Larsch*
10986
10987 * Compile clean with "-Wall -Wmissing-prototypes
10988   -Wstrict-prototypes -Wmissing-declarations -Werror". Currently
10989   DEBUG_SAFESTACK must also be set.
10990
10991   *Ben Laurie*
10992
10993 * Change ./Configure so that certain algorithms can be disabled by default.
10994   The new counterpiece to "no-xxx" is "enable-xxx".
10995
10996   The patented RC5 and MDC2 algorithms will now be disabled unless
10997   "enable-rc5" and "enable-mdc2", respectively, are specified.
10998
10999   (IDEA remains enabled despite being patented.  This is because IDEA
11000   is frequently required for interoperability, and there is no license
11001   fee for non-commercial use.  As before, "no-idea" can be used to
11002   avoid this algorithm.)
11003
11004   *Bodo Moeller*
11005
11006 * Add processing of proxy certificates (see RFC 3820).  This work was
11007   sponsored by KTH (The Royal Institute of Technology in Stockholm) and
11008   EGEE (Enabling Grids for E-science in Europe).
11009
11010   *Richard Levitte*
11011
11012 * RC4 performance overhaul on modern architectures/implementations, such
11013   as Intel P4, IA-64 and AMD64.
11014
11015   *Andy Polyakov*
11016
11017 * New utility extract-section.pl. This can be used specify an alternative
11018   section number in a pod file instead of having to treat each file as
11019   a separate case in Makefile. This can be done by adding two lines to the
11020   pod file:
11021
11022   =for comment openssl_section:XXX
11023
11024   The blank line is mandatory.
11025
11026   *Steve Henson*
11027
11028 * New arguments -certform, -keyform and -pass for s_client and s_server
11029   to allow alternative format key and certificate files and passphrase
11030   sources.
11031
11032   *Steve Henson*
11033
11034 * New structure X509_VERIFY_PARAM which combines current verify parameters,
11035   update associated structures and add various utility functions.
11036
11037   Add new policy related verify parameters, include policy checking in
11038   standard verify code. Enhance 'smime' application with extra parameters
11039   to support policy checking and print out.
11040
11041   *Steve Henson*
11042
11043 * Add a new engine to support VIA PadLock ACE extensions in the VIA C3
11044   Nehemiah processors. These extensions support AES encryption in hardware
11045   as well as RNG (though RNG support is currently disabled).
11046
11047   *Michal Ludvig <michal@logix.cz>, with help from Andy Polyakov*
11048
11049 * Deprecate `BN_[get|set]_params()` functions (they were ignored internally).
11050
11051   *Geoff Thorpe*
11052
11053 * New FIPS 180-2 algorithms, SHA-224/-256/-384/-512 are implemented.
11054
11055   *Andy Polyakov and a number of other people*
11056
11057 * Improved PowerPC platform support. Most notably BIGNUM assembler
11058   implementation contributed by IBM.
11059
11060   *Suresh Chari, Peter Waltenberg, Andy Polyakov*
11061
11062 * The new 'RSA_generate_key_ex' function now takes a BIGNUM for the public
11063   exponent rather than 'unsigned long'. There is a corresponding change to
11064   the new 'rsa_keygen' element of the RSA_METHOD structure.
11065
11066   *Jelte Jansen, Geoff Thorpe*
11067
11068 * Functionality for creating the initial serial number file is now
11069   moved from CA.pl to the 'ca' utility with a new option -create_serial.
11070
11071   (Before OpenSSL 0.9.7e, CA.pl used to initialize the serial
11072   number file to 1, which is bound to cause problems.  To avoid
11073   the problems while respecting compatibility between different 0.9.7
11074   patchlevels, 0.9.7e  employed 'openssl x509 -next_serial' in
11075   CA.pl for serial number initialization.  With the new release 0.9.8,
11076   we can fix the problem directly in the 'ca' utility.)
11077
11078   *Steve Henson*
11079
11080 * Reduced header interdependencies by declaring more opaque objects in
11081   ossl_typ.h. As a consequence, including some headers (eg. engine.h) will
11082   give fewer recursive includes, which could break lazy source code - so
11083   this change is covered by the OPENSSL_NO_DEPRECATED symbol. As always,
11084   developers should define this symbol when building and using openssl to
11085   ensure they track the recommended behaviour, interfaces, [etc], but
11086   backwards-compatible behaviour prevails when this isn't defined.
11087
11088   *Geoff Thorpe*
11089
11090 * New function X509_POLICY_NODE_print() which prints out policy nodes.
11091
11092   *Steve Henson*
11093
11094 * Add new EVP function EVP_CIPHER_CTX_rand_key and associated functionality.
11095   This will generate a random key of the appropriate length based on the
11096   cipher context. The EVP_CIPHER can provide its own random key generation
11097   routine to support keys of a specific form. This is used in the des and
11098   3des routines to generate a key of the correct parity. Update S/MIME
11099   code to use new functions and hence generate correct parity DES keys.
11100   Add EVP_CHECK_DES_KEY #define to return an error if the key is not
11101   valid (weak or incorrect parity).
11102
11103   *Steve Henson*
11104
11105 * Add a local set of CRLs that can be used by X509_verify_cert() as well
11106   as looking them up. This is useful when the verified structure may contain
11107   CRLs, for example PKCS#7 signedData. Modify PKCS7_verify() to use any CRLs
11108   present unless the new PKCS7_NO_CRL flag is asserted.
11109
11110   *Steve Henson*
11111
11112 * Extend ASN1 oid configuration module. It now additionally accepts the
11113   syntax:
11114
11115   shortName = some long name, 1.2.3.4
11116
11117   *Steve Henson*
11118
11119 * Reimplemented the BN_CTX implementation. There is now no more static
11120   limitation on the number of variables it can handle nor the depth of the
11121   "stack" handling for BN_CTX_start()/BN_CTX_end() pairs. The stack
11122   information can now expand as required, and rather than having a single
11123   static array of bignums, BN_CTX now uses a linked-list of such arrays
11124   allowing it to expand on demand whilst maintaining the usefulness of
11125   BN_CTX's "bundling".
11126
11127   *Geoff Thorpe*
11128
11129 * Add a missing BN_CTX parameter to the 'rsa_mod_exp' callback in RSA_METHOD
11130   to allow all RSA operations to function using a single BN_CTX.
11131
11132   *Geoff Thorpe*
11133
11134 * Preliminary support for certificate policy evaluation and checking. This
11135   is initially intended to pass the tests outlined in "Conformance Testing
11136   of Relying Party Client Certificate Path Processing Logic" v1.07.
11137
11138   *Steve Henson*
11139
11140 * bn_dup_expand() has been deprecated, it was introduced in 0.9.7 and
11141   remained unused and not that useful. A variety of other little bignum
11142   tweaks and fixes have also been made continuing on from the audit (see
11143   below).
11144
11145   *Geoff Thorpe*
11146
11147 * Constify all or almost all d2i, c2i, s2i and r2i functions, along with
11148   associated ASN1, EVP and SSL functions and old ASN1 macros.
11149
11150   *Richard Levitte*
11151
11152 * BN_zero() only needs to set 'top' and 'neg' to zero for correct results,
11153   and this should never fail. So the return value from the use of
11154   BN_set_word() (which can fail due to needless expansion) is now deprecated;
11155   if OPENSSL_NO_DEPRECATED is defined, BN_zero() is a void macro.
11156
11157   *Geoff Thorpe*
11158
11159 * BN_CTX_get() should return zero-valued bignums, providing the same
11160   initialised value as BN_new().
11161
11162   *Geoff Thorpe, suggested by Ulf Möller*
11163
11164 * Support for inhibitAnyPolicy certificate extension.
11165
11166   *Steve Henson*
11167
11168 * An audit of the BIGNUM code is underway, for which debugging code is
11169   enabled when BN_DEBUG is defined. This makes stricter enforcements on what
11170   is considered valid when processing BIGNUMs, and causes execution to
11171   assert() when a problem is discovered. If BN_DEBUG_RAND is defined,
11172   further steps are taken to deliberately pollute unused data in BIGNUM
11173   structures to try and expose faulty code further on. For now, openssl will
11174   (in its default mode of operation) continue to tolerate the inconsistent
11175   forms that it has tolerated in the past, but authors and packagers should
11176   consider trying openssl and their own applications when compiled with
11177   these debugging symbols defined. It will help highlight potential bugs in
11178   their own code, and will improve the test coverage for OpenSSL itself. At
11179   some point, these tighter rules will become openssl's default to improve
11180   maintainability, though the assert()s and other overheads will remain only
11181   in debugging configurations. See bn.h for more details.
11182
11183   *Geoff Thorpe, Nils Larsch, Ulf Möller*
11184
11185 * BN_CTX_init() has been deprecated, as BN_CTX is an opaque structure
11186   that can only be obtained through BN_CTX_new() (which implicitly
11187   initialises it). The presence of this function only made it possible
11188   to overwrite an existing structure (and cause memory leaks).
11189
11190   *Geoff Thorpe*
11191
11192 * Because of the callback-based approach for implementing LHASH as a
11193   template type, lh_insert() adds opaque objects to hash-tables and
11194   lh_doall() or lh_doall_arg() are typically used with a destructor callback
11195   to clean up those corresponding objects before destroying the hash table
11196   (and losing the object pointers). So some over-zealous constifications in
11197   LHASH have been relaxed so that lh_insert() does not take (nor store) the
11198   objects as "const" and the `lh_doall[_arg]` callback wrappers are not
11199   prototyped to have "const" restrictions on the object pointers they are
11200   given (and so aren't required to cast them away any more).
11201
11202   *Geoff Thorpe*
11203
11204 * The tmdiff.h API was so ugly and minimal that our own timing utility
11205   (speed) prefers to use its own implementation. The two implementations
11206   haven't been consolidated as yet (volunteers?) but the tmdiff API has had
11207   its object type properly exposed (MS_TM) instead of casting to/from
11208   `char *`. This may still change yet if someone realises MS_TM and
11209   `ms_time_***`
11210   aren't necessarily the greatest nomenclatures - but this is what was used
11211   internally to the implementation so I've used that for now.
11212
11213   *Geoff Thorpe*
11214
11215 * Ensure that deprecated functions do not get compiled when
11216   OPENSSL_NO_DEPRECATED is defined. Some "openssl" subcommands and a few of
11217   the self-tests were still using deprecated key-generation functions so
11218   these have been updated also.
11219
11220   *Geoff Thorpe*
11221
11222 * Reorganise PKCS#7 code to separate the digest location functionality
11223   into PKCS7_find_digest(), digest addition into PKCS7_bio_add_digest().
11224   New function PKCS7_set_digest() to set the digest type for PKCS#7
11225   digestedData type. Add additional code to correctly generate the
11226   digestedData type and add support for this type in PKCS7 initialization
11227   functions.
11228
11229   *Steve Henson*
11230
11231 * New function PKCS7_set0_type_other() this initializes a PKCS7
11232   structure of type "other".
11233
11234   *Steve Henson*
11235
11236 * Fix prime generation loop in crypto/bn/bn_prime.pl by making
11237   sure the loop does correctly stop and breaking ("division by zero")
11238   modulus operations are not performed. The (pre-generated) prime
11239   table crypto/bn/bn_prime.h was already correct, but it could not be
11240   re-generated on some platforms because of the "division by zero"
11241   situation in the script.
11242
11243   *Ralf S. Engelschall*
11244
11245 * Update support for ECC-based TLS ciphersuites according to
11246   draft-ietf-tls-ecc-03.txt: the KDF1 key derivation function with
11247   SHA-1 now is only used for "small" curves (where the
11248   representation of a field element takes up to 24 bytes); for
11249   larger curves, the field element resulting from ECDH is directly
11250   used as premaster secret.
11251
11252   *Douglas Stebila (Sun Microsystems Laboratories)*
11253
11254 * Add code for kP+lQ timings to crypto/ec/ectest.c, and add SEC2
11255   curve secp160r1 to the tests.
11256
11257   *Douglas Stebila (Sun Microsystems Laboratories)*
11258
11259 * Add the possibility to load symbols globally with DSO.
11260
11261   *Götz Babin-Ebell <babin-ebell@trustcenter.de> via Richard Levitte*
11262
11263 * Add the functions ERR_set_mark() and ERR_pop_to_mark() for better
11264   control of the error stack.
11265
11266   *Richard Levitte*
11267
11268 * Add support for STORE in ENGINE.
11269
11270   *Richard Levitte*
11271
11272 * Add the STORE type.  The intention is to provide a common interface
11273   to certificate and key stores, be they simple file-based stores, or
11274   HSM-type store, or LDAP stores, or...
11275   NOTE: The code is currently UNTESTED and isn't really used anywhere.
11276
11277   *Richard Levitte*
11278
11279 * Add a generic structure called OPENSSL_ITEM.  This can be used to
11280   pass a list of arguments to any function as well as provide a way
11281   for a function to pass data back to the caller.
11282
11283   *Richard Levitte*
11284
11285 * Add the functions BUF_strndup() and BUF_memdup().  BUF_strndup()
11286   works like BUF_strdup() but can be used to duplicate a portion of
11287   a string.  The copy gets NUL-terminated.  BUF_memdup() duplicates
11288   a memory area.
11289
11290   *Richard Levitte*
11291
11292 * Add the function sk_find_ex() which works like sk_find(), but will
11293   return an index to an element even if an exact match couldn't be
11294   found.  The index is guaranteed to point at the element where the
11295   searched-for key would be inserted to preserve sorting order.
11296
11297   *Richard Levitte*
11298
11299 * Add the function OBJ_bsearch_ex() which works like OBJ_bsearch() but
11300   takes an extra flags argument for optional functionality.  Currently,
11301   the following flags are defined:
11302
11303      OBJ_BSEARCH_VALUE_ON_NOMATCH
11304      This one gets OBJ_bsearch_ex() to return a pointer to the first
11305      element where the comparing function returns a negative or zero
11306      number.
11307
11308      OBJ_BSEARCH_FIRST_VALUE_ON_MATCH
11309      This one gets OBJ_bsearch_ex() to return a pointer to the first
11310      element where the comparing function returns zero.  This is useful
11311      if there are more than one element where the comparing function
11312      returns zero.
11313
11314   *Richard Levitte*
11315
11316 * Make it possible to create self-signed certificates with 'openssl ca'
11317   in such a way that the self-signed certificate becomes part of the
11318   CA database and uses the same mechanisms for serial number generation
11319   as all other certificate signing.  The new flag '-selfsign' enables
11320   this functionality.  Adapt CA.sh and CA.pl.in.
11321
11322   *Richard Levitte*
11323
11324 * Add functionality to check the public key of a certificate request
11325   against a given private.  This is useful to check that a certificate
11326   request can be signed by that key (self-signing).
11327
11328   *Richard Levitte*
11329
11330 * Make it possible to have multiple active certificates with the same
11331   subject in the CA index file.  This is done only if the keyword
11332   'unique_subject' is set to 'no' in the main CA section (default
11333   if 'CA_default') of the configuration file.  The value is saved
11334   with the database itself in a separate index attribute file,
11335   named like the index file with '.attr' appended to the name.
11336
11337   *Richard Levitte*
11338
11339 * Generate multi-valued AVAs using '+' notation in config files for
11340   req and dirName.
11341
11342   *Steve Henson*
11343
11344 * Support for nameConstraints certificate extension.
11345
11346   *Steve Henson*
11347
11348 * Support for policyConstraints certificate extension.
11349
11350   *Steve Henson*
11351
11352 * Support for policyMappings certificate extension.
11353
11354   *Steve Henson*
11355
11356 * Make sure the default DSA_METHOD implementation only uses its
11357   dsa_mod_exp() and/or bn_mod_exp() handlers if they are non-NULL,
11358   and change its own handlers to be NULL so as to remove unnecessary
11359   indirection. This lets alternative implementations fallback to the
11360   default implementation more easily.
11361
11362   *Geoff Thorpe*
11363
11364 * Support for directoryName in GeneralName related extensions
11365   in config files.
11366
11367   *Steve Henson*
11368
11369 * Make it possible to link applications using Makefile.shared.
11370   Make that possible even when linking against static libraries!
11371
11372   *Richard Levitte*
11373
11374 * Support for single pass processing for S/MIME signing. This now
11375   means that S/MIME signing can be done from a pipe, in addition
11376   cleartext signing (multipart/signed type) is effectively streaming
11377   and the signed data does not need to be all held in memory.
11378
11379   This is done with a new flag PKCS7_STREAM. When this flag is set
11380   PKCS7_sign() only initializes the PKCS7 structure and the actual signing
11381   is done after the data is output (and digests calculated) in
11382   SMIME_write_PKCS7().
11383
11384   *Steve Henson*
11385
11386 * Add full support for -rpath/-R, both in shared libraries and
11387   applications, at least on the platforms where it's known how
11388   to do it.
11389
11390   *Richard Levitte*
11391
11392 * In crypto/ec/ec_mult.c, implement fast point multiplication with
11393   precomputation, based on wNAF splitting: EC_GROUP_precompute_mult()
11394   will now compute a table of multiples of the generator that
11395   makes subsequent invocations of EC_POINTs_mul() or EC_POINT_mul()
11396   faster (notably in the case of a single point multiplication,
11397   scalar * generator).
11398
11399   *Nils Larsch, Bodo Moeller*
11400
11401 * IPv6 support for certificate extensions. The various extensions
11402   which use the IP:a.b.c.d can now take IPv6 addresses using the
11403   formats of RFC1884 2.2 . IPv6 addresses are now also displayed
11404   correctly.
11405
11406   *Steve Henson*
11407
11408 * Added an ENGINE that implements RSA by performing private key
11409   exponentiations with the GMP library. The conversions to and from
11410   GMP's mpz_t format aren't optimised nor are any montgomery forms
11411   cached, and on x86 it appears OpenSSL's own performance has caught up.
11412   However there are likely to be other architectures where GMP could
11413   provide a boost. This ENGINE is not built in by default, but it can be
11414   specified at Configure time and should be accompanied by the necessary
11415   linker additions, eg;
11416           ./config -DOPENSSL_USE_GMP -lgmp
11417
11418   *Geoff Thorpe*
11419
11420 * "openssl engine" will not display ENGINE/DSO load failure errors when
11421   testing availability of engines with "-t" - the old behaviour is
11422   produced by increasing the feature's verbosity with "-tt".
11423
11424   *Geoff Thorpe*
11425
11426 * ECDSA routines: under certain error conditions uninitialized BN objects
11427   could be freed. Solution: make sure initialization is performed early
11428   enough. (Reported and fix supplied by Nils Larsch <nla@trustcenter.de>
11429   via PR#459)
11430
11431   *Lutz Jaenicke*
11432
11433 * Key-generation can now be implemented in RSA_METHOD, DSA_METHOD
11434   and DH_METHOD (eg. by ENGINE implementations) to override the normal
11435   software implementations. For DSA and DH, parameter generation can
11436   also be overridden by providing the appropriate method callbacks.
11437
11438   *Geoff Thorpe*
11439
11440 * Change the "progress" mechanism used in key-generation and
11441   primality testing to functions that take a new BN_GENCB pointer in
11442   place of callback/argument pairs. The new API functions have `_ex`
11443   postfixes and the older functions are reimplemented as wrappers for
11444   the new ones. The OPENSSL_NO_DEPRECATED symbol can be used to hide
11445   declarations of the old functions to help (graceful) attempts to
11446   migrate to the new functions. Also, the new key-generation API
11447   functions operate on a caller-supplied key-structure and return
11448   success/failure rather than returning a key or NULL - this is to
11449   help make "keygen" another member function of RSA_METHOD etc.
11450
11451   Example for using the new callback interface:
11452
11453           int (*my_callback)(int a, int b, BN_GENCB *cb) = ...;
11454           void *my_arg = ...;
11455           BN_GENCB my_cb;
11456
11457           BN_GENCB_set(&my_cb, my_callback, my_arg);
11458
11459           return BN_is_prime_ex(some_bignum, BN_prime_checks, NULL, &cb);
11460           /* For the meaning of a, b in calls to my_callback(), see the
11461            * documentation of the function that calls the callback.
11462            * cb will point to my_cb; my_arg can be retrieved as cb->arg.
11463            * my_callback should return 1 if it wants BN_is_prime_ex()
11464            * to continue, or 0 to stop.
11465            */
11466
11467   *Geoff Thorpe*
11468
11469 * Change the ZLIB compression method to be stateful, and make it
11470   available to TLS with the number defined in
11471   draft-ietf-tls-compression-04.txt.
11472
11473   *Richard Levitte*
11474
11475 * Add the ASN.1 structures and functions for CertificatePair, which
11476   is defined as follows (according to X.509_4thEditionDraftV6.pdf):
11477
11478           CertificatePair ::= SEQUENCE {
11479              forward         [0]     Certificate OPTIONAL,
11480              reverse         [1]     Certificate OPTIONAL,
11481              -- at least one of the pair shall be present -- }
11482
11483   Also implement the PEM functions to read and write certificate
11484   pairs, and defined the PEM tag as "CERTIFICATE PAIR".
11485
11486   This needed to be defined, mostly for the sake of the LDAP
11487   attribute crossCertificatePair, but may prove useful elsewhere as
11488   well.
11489
11490   *Richard Levitte*
11491
11492 * Make it possible to inhibit symlinking of shared libraries in
11493   Makefile.shared, for Cygwin's sake.
11494
11495   *Richard Levitte*
11496
11497 * Extend the BIGNUM API by creating a function
11498           void BN_set_negative(BIGNUM *a, int neg);
11499   and a macro that behave like
11500           int  BN_is_negative(const BIGNUM *a);
11501
11502   to avoid the need to access 'a->neg' directly in applications.
11503
11504   *Nils Larsch*
11505
11506 * Implement fast modular reduction for pseudo-Mersenne primes
11507   used in NIST curves (crypto/bn/bn_nist.c, crypto/ec/ecp_nist.c).
11508   EC_GROUP_new_curve_GFp() will now automatically use this
11509   if applicable.
11510
11511   *Nils Larsch <nla@trustcenter.de>*
11512
11513 * Add new lock type (CRYPTO_LOCK_BN).
11514
11515   *Bodo Moeller*
11516
11517 * Change the ENGINE framework to automatically load engines
11518   dynamically from specific directories unless they could be
11519   found to already be built in or loaded.  Move all the
11520   current engines except for the cryptodev one to a new
11521   directory engines/.
11522   The engines in engines/ are built as shared libraries if
11523   the "shared" options was given to ./Configure or ./config.
11524   Otherwise, they are inserted in libcrypto.a.
11525   /usr/local/ssl/engines is the default directory for dynamic
11526   engines, but that can be overridden at configure time through
11527   the usual use of --prefix and/or --openssldir, and at run
11528   time with the environment variable OPENSSL_ENGINES.
11529
11530   *Geoff Thorpe and Richard Levitte*
11531
11532 * Add Makefile.shared, a helper makefile to build shared
11533   libraries.  Adapt Makefile.org.
11534
11535   *Richard Levitte*
11536
11537 * Add version info to Win32 DLLs.
11538
11539   *Peter 'Luna' Runestig" <peter@runestig.com>*
11540
11541 * Add new 'medium level' PKCS#12 API. Certificates and keys
11542   can be added using this API to created arbitrary PKCS#12
11543   files while avoiding the low-level API.
11544
11545   New options to PKCS12_create(), key or cert can be NULL and
11546   will then be omitted from the output file. The encryption
11547   algorithm NIDs can be set to -1 for no encryption, the mac
11548   iteration count can be set to 0 to omit the mac.
11549
11550   Enhance pkcs12 utility by making the -nokeys and -nocerts
11551   options work when creating a PKCS#12 file. New option -nomac
11552   to omit the mac, NONE can be set for an encryption algorithm.
11553   New code is modified to use the enhanced PKCS12_create()
11554   instead of the low-level API.
11555
11556   *Steve Henson*
11557
11558 * Extend ASN1 encoder to support indefinite length constructed
11559   encoding. This can output sequences tags and octet strings in
11560   this form. Modify pk7_asn1.c to support indefinite length
11561   encoding. This is experimental and needs additional code to
11562   be useful, such as an ASN1 bio and some enhanced streaming
11563   PKCS#7 code.
11564
11565   Extend template encode functionality so that tagging is passed
11566   down to the template encoder.
11567
11568   *Steve Henson*
11569
11570 * Let 'openssl req' fail if an argument to '-newkey' is not
11571   recognized instead of using RSA as a default.
11572
11573   *Bodo Moeller*
11574
11575 * Add support for ECC-based ciphersuites from draft-ietf-tls-ecc-01.txt.
11576   As these are not official, they are not included in "ALL";
11577   the "ECCdraft" ciphersuite group alias can be used to select them.
11578
11579   *Vipul Gupta and Sumit Gupta (Sun Microsystems Laboratories)*
11580
11581 * Add ECDH engine support.
11582
11583   *Nils Gura and Douglas Stebila (Sun Microsystems Laboratories)*
11584
11585 * Add ECDH in new directory crypto/ecdh/.
11586
11587   *Douglas Stebila (Sun Microsystems Laboratories)*
11588
11589 * Let BN_rand_range() abort with an error after 100 iterations
11590   without success (which indicates a broken PRNG).
11591
11592   *Bodo Moeller*
11593
11594 * Change BN_mod_sqrt() so that it verifies that the input value
11595   is really the square of the return value.  (Previously,
11596   BN_mod_sqrt would show GIGO behaviour.)
11597
11598   *Bodo Moeller*
11599
11600 * Add named elliptic curves over binary fields from X9.62, SECG,
11601   and WAP/WTLS; add OIDs that were still missing.
11602
11603   *Sheueling Chang Shantz and Douglas Stebila (Sun Microsystems Laboratories)*
11604
11605 * Extend the EC library for elliptic curves over binary fields
11606   (new files ec2_smpl.c, ec2_smpt.c, ec2_mult.c in crypto/ec/).
11607   New EC_METHOD:
11608
11609           EC_GF2m_simple_method
11610
11611   New API functions:
11612
11613           EC_GROUP_new_curve_GF2m
11614           EC_GROUP_set_curve_GF2m
11615           EC_GROUP_get_curve_GF2m
11616           EC_POINT_set_affine_coordinates_GF2m
11617           EC_POINT_get_affine_coordinates_GF2m
11618           EC_POINT_set_compressed_coordinates_GF2m
11619
11620   Point compression for binary fields is disabled by default for
11621   patent reasons (compile with OPENSSL_EC_BIN_PT_COMP defined to
11622   enable it).
11623
11624   As binary polynomials are represented as BIGNUMs, various members
11625   of the EC_GROUP and EC_POINT data structures can be shared
11626   between the implementations for prime fields and binary fields;
11627   the above `..._GF2m functions` (except for EX_GROUP_new_curve_GF2m)
11628   are essentially identical to their `..._GFp` counterparts.
11629   (For simplicity, the `..._GFp` prefix has been dropped from
11630   various internal method names.)
11631
11632   An internal 'field_div' method (similar to 'field_mul' and
11633   'field_sqr') has been added; this is used only for binary fields.
11634
11635   *Sheueling Chang Shantz and Douglas Stebila (Sun Microsystems Laboratories)*
11636
11637 * Optionally dispatch EC_POINT_mul(), EC_POINT_precompute_mult()
11638   through methods ('mul', 'precompute_mult').
11639
11640   The generic implementations (now internally called 'ec_wNAF_mul'
11641   and 'ec_wNAF_precomputed_mult') remain the default if these
11642   methods are undefined.
11643
11644   *Sheueling Chang Shantz and Douglas Stebila (Sun Microsystems Laboratories)*
11645
11646 * New function EC_GROUP_get_degree, which is defined through
11647   EC_METHOD.  For curves over prime fields, this returns the bit
11648   length of the modulus.
11649
11650   *Sheueling Chang Shantz and Douglas Stebila (Sun Microsystems Laboratories)*
11651
11652 * New functions EC_GROUP_dup, EC_POINT_dup.
11653   (These simply call ..._new  and ..._copy).
11654
11655   *Sheueling Chang Shantz and Douglas Stebila (Sun Microsystems Laboratories)*
11656
11657 * Add binary polynomial arithmetic software in crypto/bn/bn_gf2m.c.
11658   Polynomials are represented as BIGNUMs (where the sign bit is not
11659   used) in the following functions [macros]:
11660
11661           BN_GF2m_add
11662           BN_GF2m_sub             [= BN_GF2m_add]
11663           BN_GF2m_mod             [wrapper for BN_GF2m_mod_arr]
11664           BN_GF2m_mod_mul         [wrapper for BN_GF2m_mod_mul_arr]
11665           BN_GF2m_mod_sqr         [wrapper for BN_GF2m_mod_sqr_arr]
11666           BN_GF2m_mod_inv
11667           BN_GF2m_mod_exp         [wrapper for BN_GF2m_mod_exp_arr]
11668           BN_GF2m_mod_sqrt        [wrapper for BN_GF2m_mod_sqrt_arr]
11669           BN_GF2m_mod_solve_quad  [wrapper for BN_GF2m_mod_solve_quad_arr]
11670           BN_GF2m_cmp             [= BN_ucmp]
11671
11672   (Note that only the 'mod' functions are actually for fields GF(2^m).
11673   BN_GF2m_add() is misnomer, but this is for the sake of consistency.)
11674
11675   For some functions, an the irreducible polynomial defining a
11676   field can be given as an 'unsigned int[]' with strictly
11677   decreasing elements giving the indices of those bits that are set;
11678   i.e., p[] represents the polynomial
11679           f(t) = t^p[0] + t^p[1] + ... + t^p[k]
11680   where
11681           p[0] > p[1] > ... > p[k] = 0.
11682   This applies to the following functions:
11683
11684           BN_GF2m_mod_arr
11685           BN_GF2m_mod_mul_arr
11686           BN_GF2m_mod_sqr_arr
11687           BN_GF2m_mod_inv_arr        [wrapper for BN_GF2m_mod_inv]
11688           BN_GF2m_mod_div_arr        [wrapper for BN_GF2m_mod_div]
11689           BN_GF2m_mod_exp_arr
11690           BN_GF2m_mod_sqrt_arr
11691           BN_GF2m_mod_solve_quad_arr
11692           BN_GF2m_poly2arr
11693           BN_GF2m_arr2poly
11694
11695   Conversion can be performed by the following functions:
11696
11697           BN_GF2m_poly2arr
11698           BN_GF2m_arr2poly
11699
11700   bntest.c has additional tests for binary polynomial arithmetic.
11701
11702   Two implementations for BN_GF2m_mod_div() are available.
11703   The default algorithm simply uses BN_GF2m_mod_inv() and
11704   BN_GF2m_mod_mul().  The alternative algorithm is compiled in only
11705   if OPENSSL_SUN_GF2M_DIV is defined (patent pending; read the
11706   copyright notice in crypto/bn/bn_gf2m.c before enabling it).
11707
11708   *Sheueling Chang Shantz and Douglas Stebila (Sun Microsystems Laboratories)*
11709
11710 * Add new error code 'ERR_R_DISABLED' that can be used when some
11711   functionality is disabled at compile-time.
11712
11713   *Douglas Stebila <douglas.stebila@sun.com>*
11714
11715 * Change default behaviour of 'openssl asn1parse' so that more
11716   information is visible when viewing, e.g., a certificate:
11717
11718   Modify asn1_parse2 (crypto/asn1/asn1_par.c) so that in non-'dump'
11719   mode the content of non-printable OCTET STRINGs is output in a
11720   style similar to INTEGERs, but with '[HEX DUMP]' prepended to
11721   avoid the appearance of a printable string.
11722
11723   *Nils Larsch <nla@trustcenter.de>*
11724
11725 * Add 'asn1_flag' and 'asn1_form' member to EC_GROUP with access
11726   functions
11727           EC_GROUP_set_asn1_flag()
11728           EC_GROUP_get_asn1_flag()
11729           EC_GROUP_set_point_conversion_form()
11730           EC_GROUP_get_point_conversion_form()
11731   These control ASN1 encoding details:
11732   - Curves (i.e., groups) are encoded explicitly unless asn1_flag
11733     has been set to OPENSSL_EC_NAMED_CURVE.
11734   - Points are encoded in uncompressed form by default; options for
11735     asn1_for are as for point2oct, namely
11736           POINT_CONVERSION_COMPRESSED
11737           POINT_CONVERSION_UNCOMPRESSED
11738           POINT_CONVERSION_HYBRID
11739
11740   Also add 'seed' and 'seed_len' members to EC_GROUP with access
11741   functions
11742           EC_GROUP_set_seed()
11743           EC_GROUP_get0_seed()
11744           EC_GROUP_get_seed_len()
11745   This is used only for ASN1 purposes (so far).
11746
11747   *Nils Larsch <nla@trustcenter.de>*
11748
11749 * Add 'field_type' member to EC_METHOD, which holds the NID
11750   of the appropriate field type OID.  The new function
11751   EC_METHOD_get_field_type() returns this value.
11752
11753   *Nils Larsch <nla@trustcenter.de>*
11754
11755 * Add functions
11756           EC_POINT_point2bn()
11757           EC_POINT_bn2point()
11758           EC_POINT_point2hex()
11759           EC_POINT_hex2point()
11760   providing useful interfaces to EC_POINT_point2oct() and
11761   EC_POINT_oct2point().
11762
11763   *Nils Larsch <nla@trustcenter.de>*
11764
11765 * Change internals of the EC library so that the functions
11766           EC_GROUP_set_generator()
11767           EC_GROUP_get_generator()
11768           EC_GROUP_get_order()
11769           EC_GROUP_get_cofactor()
11770   are implemented directly in crypto/ec/ec_lib.c and not dispatched
11771   to methods, which would lead to unnecessary code duplication when
11772   adding different types of curves.
11773
11774   *Nils Larsch <nla@trustcenter.de> with input by Bodo Moeller*
11775
11776 * Implement compute_wNAF (crypto/ec/ec_mult.c) without BIGNUM
11777   arithmetic, and such that modified wNAFs are generated
11778   (which avoid length expansion in many cases).
11779
11780   *Bodo Moeller*
11781
11782 * Add a function EC_GROUP_check_discriminant() (defined via
11783   EC_METHOD) that verifies that the curve discriminant is non-zero.
11784
11785   Add a function EC_GROUP_check() that makes some sanity tests
11786   on a EC_GROUP, its generator and order.  This includes
11787   EC_GROUP_check_discriminant().
11788
11789   *Nils Larsch <nla@trustcenter.de>*
11790
11791 * Add ECDSA in new directory crypto/ecdsa/.
11792
11793   Add applications 'openssl ecparam' and 'openssl ecdsa'
11794   (these are based on 'openssl dsaparam' and 'openssl dsa').
11795
11796   ECDSA support is also included in various other files across the
11797   library.  Most notably,
11798   - 'openssl req' now has a '-newkey ecdsa:file' option;
11799   - EVP_PKCS82PKEY (crypto/evp/evp_pkey.c) now can handle ECDSA;
11800   - X509_PUBKEY_get (crypto/asn1/x_pubkey.c) and
11801     d2i_PublicKey (crypto/asn1/d2i_pu.c) have been modified to make
11802     them suitable for ECDSA where domain parameters must be
11803     extracted before the specific public key;
11804   - ECDSA engine support has been added.
11805
11806   *Nils Larsch <nla@trustcenter.de>*
11807
11808 * Include some named elliptic curves, and add OIDs from X9.62,
11809   SECG, and WAP/WTLS.  Each curve can be obtained from the new
11810   function
11811           EC_GROUP_new_by_curve_name(),
11812   and the list of available named curves can be obtained with
11813           EC_get_builtin_curves().
11814   Also add a 'curve_name' member to EC_GROUP objects, which can be
11815   accessed via
11816           EC_GROUP_set_curve_name()
11817           EC_GROUP_get_curve_name()
11818
11819   *Nils Larsch <larsch@trustcenter.de, Bodo Moeller*
11820
11821 * Remove a few calls to bn_wexpand() in BN_sqr() (the one in there
11822   was actually never needed) and in BN_mul().  The removal in BN_mul()
11823   required a small change in bn_mul_part_recursive() and the addition
11824   of the functions bn_cmp_part_words(), bn_sub_part_words() and
11825   bn_add_part_words(), which do the same thing as bn_cmp_words(),
11826   bn_sub_words() and bn_add_words() except they take arrays with
11827   differing sizes.
11828
11829   *Richard Levitte*
11830
11831### Changes between 0.9.7l and 0.9.7m  [23 Feb 2007]
11832
11833 * Cleanse PEM buffers before freeing them since they may contain
11834   sensitive data.
11835
11836   *Benjamin Bennett <ben@psc.edu>*
11837
11838 * Include "!eNULL" in SSL_DEFAULT_CIPHER_LIST to make sure that
11839   a ciphersuite string such as "DEFAULT:RSA" cannot enable
11840   authentication-only ciphersuites.
11841
11842   *Bodo Moeller*
11843
11844 * Since AES128 and AES256 share a single mask bit in the logic of
11845   ssl/ssl_ciph.c, the code for masking out disabled ciphers needs a
11846   kludge to work properly if AES128 is available and AES256 isn't.
11847
11848   *Victor Duchovni*
11849
11850 * Expand security boundary to match 1.1.1 module.
11851
11852   *Steve Henson*
11853
11854 * Remove redundant features: hash file source, editing of test vectors
11855   modify fipsld to use external fips_premain.c signature.
11856
11857   *Steve Henson*
11858
11859 * New perl script mkfipsscr.pl to create shell scripts or batch files to
11860   run algorithm test programs.
11861
11862   *Steve Henson*
11863
11864 * Make algorithm test programs more tolerant of whitespace.
11865
11866   *Steve Henson*
11867
11868 * Have SSL/TLS server implementation tolerate "mismatched" record
11869   protocol version while receiving ClientHello even if the
11870   ClientHello is fragmented.  (The server can't insist on the
11871   particular protocol version it has chosen before the ServerHello
11872   message has informed the client about his choice.)
11873
11874   *Bodo Moeller*
11875
11876 * Load error codes if they are not already present instead of using a
11877   static variable. This allows them to be cleanly unloaded and reloaded.
11878
11879   *Steve Henson*
11880
11881### Changes between 0.9.7k and 0.9.7l  [28 Sep 2006]
11882
11883 * Introduce limits to prevent malicious keys being able to
11884   cause a denial of service.  ([CVE-2006-2940])
11885
11886   *Steve Henson, Bodo Moeller*
11887
11888 * Fix ASN.1 parsing of certain invalid structures that can result
11889   in a denial of service.  ([CVE-2006-2937])  [Steve Henson]
11890
11891 * Fix buffer overflow in SSL_get_shared_ciphers() function.
11892   ([CVE-2006-3738]) [Tavis Ormandy and Will Drewry, Google Security Team]
11893
11894 * Fix SSL client code which could crash if connecting to a
11895   malicious SSLv2 server.  ([CVE-2006-4343])
11896
11897   *Tavis Ormandy and Will Drewry, Google Security Team*
11898
11899 * Change ciphersuite string processing so that an explicit
11900   ciphersuite selects this one ciphersuite (so that "AES256-SHA"
11901   will no longer include "AES128-SHA"), and any other similar
11902   ciphersuite (same bitmap) from *other* protocol versions (so that
11903   "RC4-MD5" will still include both the SSL 2.0 ciphersuite and the
11904   SSL 3.0/TLS 1.0 ciphersuite).  This is a backport combining
11905   changes from 0.9.8b and 0.9.8d.
11906
11907   *Bodo Moeller*
11908
11909### Changes between 0.9.7j and 0.9.7k  [05 Sep 2006]
11910
11911 * Avoid PKCS #1 v1.5 signature attack discovered by Daniel Bleichenbacher
11912   ([CVE-2006-4339])  [Ben Laurie and Google Security Team]
11913
11914 * Change the Unix randomness entropy gathering to use poll() when
11915   possible instead of select(), since the latter has some
11916   undesirable limitations.
11917
11918   *Darryl Miles via Richard Levitte and Bodo Moeller*
11919
11920 * Disable rogue ciphersuites:
11921
11922   - SSLv2 0x08 0x00 0x80 ("RC4-64-MD5")
11923   - SSLv3/TLSv1 0x00 0x61 ("EXP1024-RC2-CBC-MD5")
11924   - SSLv3/TLSv1 0x00 0x60 ("EXP1024-RC4-MD5")
11925
11926   The latter two were purportedly from
11927   draft-ietf-tls-56-bit-ciphersuites-0[01].txt, but do not really
11928   appear there.
11929
11930   Also deactivate the remaining ciphersuites from
11931   draft-ietf-tls-56-bit-ciphersuites-01.txt.  These are just as
11932   unofficial, and the ID has long expired.
11933
11934   *Bodo Moeller*
11935
11936 * Fix RSA blinding Heisenbug (problems sometimes occurred on
11937   dual-core machines) and other potential thread-safety issues.
11938
11939   *Bodo Moeller*
11940
11941### Changes between 0.9.7i and 0.9.7j  [04 May 2006]
11942
11943 * Adapt fipsld and the build system to link against the validated FIPS
11944   module in FIPS mode.
11945
11946   *Steve Henson*
11947
11948 * Fixes for VC++ 2005 build under Windows.
11949
11950   *Steve Henson*
11951
11952 * Add new Windows build target VC-32-GMAKE for VC++. This uses GNU make
11953   from a Windows bash shell such as MSYS. It is autodetected from the
11954   "config" script when run from a VC++ environment. Modify standard VC++
11955   build to use fipscanister.o from the GNU make build.
11956
11957   *Steve Henson*
11958
11959### Changes between 0.9.7h and 0.9.7i  [14 Oct 2005]
11960
11961 * Wrapped the definition of EVP_MAX_MD_SIZE in a #ifdef OPENSSL_FIPS.
11962   The value now differs depending on if you build for FIPS or not.
11963   BEWARE!  A program linked with a shared FIPSed libcrypto can't be
11964   safely run with a non-FIPSed libcrypto, as it may crash because of
11965   the difference induced by this change.
11966
11967   *Andy Polyakov*
11968
11969### Changes between 0.9.7g and 0.9.7h  [11 Oct 2005]
11970
11971 * Remove the functionality of SSL_OP_MSIE_SSLV2_RSA_PADDING
11972   (part of SSL_OP_ALL).  This option used to disable the
11973   countermeasure against man-in-the-middle protocol-version
11974   rollback in the SSL 2.0 server implementation, which is a bad
11975   idea.  ([CVE-2005-2969])
11976
11977   *Bodo Moeller; problem pointed out by Yutaka Oiwa (Research Center
11978   for Information Security, National Institute of Advanced Industrial
11979   Science and Technology [AIST, Japan)]*
11980
11981 * Minimal support for X9.31 signatures and PSS padding modes. This is
11982   mainly for FIPS compliance and not fully integrated at this stage.
11983
11984   *Steve Henson*
11985
11986 * For DSA signing, unless DSA_FLAG_NO_EXP_CONSTTIME is set, perform
11987   the exponentiation using a fixed-length exponent.  (Otherwise,
11988   the information leaked through timing could expose the secret key
11989   after many signatures; cf. Bleichenbacher's attack on DSA with
11990   biased k.)
11991
11992   *Bodo Moeller*
11993
11994 * Make a new fixed-window mod_exp implementation the default for
11995   RSA, DSA, and DH private-key operations so that the sequence of
11996   squares and multiplies and the memory access pattern are
11997   independent of the particular secret key.  This will mitigate
11998   cache-timing and potential related attacks.
11999
12000   BN_mod_exp_mont_consttime() is the new exponentiation implementation,
12001   and this is automatically used by BN_mod_exp_mont() if the new flag
12002   BN_FLG_EXP_CONSTTIME is set for the exponent.  RSA, DSA, and DH
12003   will use this BN flag for private exponents unless the flag
12004   RSA_FLAG_NO_EXP_CONSTTIME, DSA_FLAG_NO_EXP_CONSTTIME, or
12005   DH_FLAG_NO_EXP_CONSTTIME, respectively, is set.
12006
12007   *Matthew D Wood (Intel Corp), with some changes by Bodo Moeller*
12008
12009 * Change the client implementation for SSLv23_method() and
12010   SSLv23_client_method() so that is uses the SSL 3.0/TLS 1.0
12011   Client Hello message format if the SSL_OP_NO_SSLv2 option is set.
12012   (Previously, the SSL 2.0 backwards compatible Client Hello
12013   message format would be used even with SSL_OP_NO_SSLv2.)
12014
12015   *Bodo Moeller*
12016
12017 * Add support for smime-type MIME parameter in S/MIME messages which some
12018   clients need.
12019
12020   *Steve Henson*
12021
12022 * New function BN_MONT_CTX_set_locked() to set montgomery parameters in
12023   a threadsafe manner. Modify rsa code to use new function and add calls
12024   to dsa and dh code (which had race conditions before).
12025
12026   *Steve Henson*
12027
12028 * Include the fixed error library code in the C error file definitions
12029   instead of fixing them up at runtime. This keeps the error code
12030   structures constant.
12031
12032   *Steve Henson*
12033
12034### Changes between 0.9.7f and 0.9.7g  [11 Apr 2005]
12035
12036[NB: OpenSSL 0.9.7h and later 0.9.7 patch levels were released after
12037OpenSSL 0.9.8.]
12038
12039 * Fixes for newer kerberos headers. NB: the casts are needed because
12040   the 'length' field is signed on one version and unsigned on another
12041   with no (?) obvious way to tell the difference, without these VC++
12042   complains. Also the "definition" of FAR (blank) is no longer included
12043   nor is the error ENOMEM. KRB5_PRIVATE has to be set to 1 to pick up
12044   some needed definitions.
12045
12046   *Steve Henson*
12047
12048 * Undo Cygwin change.
12049
12050   *Ulf Möller*
12051
12052 * Added support for proxy certificates according to RFC 3820.
12053   Because they may be a security thread to unaware applications,
12054   they must be explicitly allowed in run-time.  See
12055   docs/HOWTO/proxy_certificates.txt for further information.
12056
12057   *Richard Levitte*
12058
12059### Changes between 0.9.7e and 0.9.7f  [22 Mar 2005]
12060
12061 * Use (SSL_RANDOM_VALUE - 4) bytes of pseudo random data when generating
12062   server and client random values. Previously
12063   (SSL_RANDOM_VALUE - sizeof(time_t)) would be used which would result in
12064   less random data when sizeof(time_t) > 4 (some 64 bit platforms).
12065
12066   This change has negligible security impact because:
12067
12068   1. Server and client random values still have 24 bytes of pseudo random
12069      data.
12070
12071   2. Server and client random values are sent in the clear in the initial
12072      handshake.
12073
12074   3. The master secret is derived using the premaster secret (48 bytes in
12075      size for static RSA ciphersuites) as well as client server and random
12076      values.
12077
12078   The OpenSSL team would like to thank the UK NISCC for bringing this issue
12079   to our attention.
12080
12081   *Stephen Henson, reported by UK NISCC*
12082
12083 * Use Windows randomness collection on Cygwin.
12084
12085   *Ulf Möller*
12086
12087 * Fix hang in EGD/PRNGD query when communication socket is closed
12088   prematurely by EGD/PRNGD.
12089
12090   *Darren Tucker <dtucker@zip.com.au> via Lutz Jänicke, resolves #1014*
12091
12092 * Prompt for pass phrases when appropriate for PKCS12 input format.
12093
12094   *Steve Henson*
12095
12096 * Back-port of selected performance improvements from development
12097   branch, as well as improved support for PowerPC platforms.
12098
12099   *Andy Polyakov*
12100
12101 * Add lots of checks for memory allocation failure, error codes to indicate
12102   failure and freeing up memory if a failure occurs.
12103
12104   *Nauticus Networks SSL Team <openssl@nauticusnet.com>, Steve Henson*
12105
12106 * Add new -passin argument to dgst.
12107
12108   *Steve Henson*
12109
12110 * Perform some character comparisons of different types in X509_NAME_cmp:
12111   this is needed for some certificates that re-encode DNs into UTF8Strings
12112   (in violation of RFC3280) and can't or won't issue name rollover
12113   certificates.
12114
12115   *Steve Henson*
12116
12117 * Make an explicit check during certificate validation to see that
12118   the CA setting in each certificate on the chain is correct.  As a
12119   side effect always do the following basic checks on extensions,
12120   not just when there's an associated purpose to the check:
12121
12122   - if there is an unhandled critical extension (unless the user
12123     has chosen to ignore this fault)
12124   - if the path length has been exceeded (if one is set at all)
12125   - that certain extensions fit the associated purpose (if one has
12126     been given)
12127
12128   *Richard Levitte*
12129
12130### Changes between 0.9.7d and 0.9.7e  [25 Oct 2004]
12131
12132 * Avoid a race condition when CRLs are checked in a multi threaded
12133   environment. This would happen due to the reordering of the revoked
12134   entries during signature checking and serial number lookup. Now the
12135   encoding is cached and the serial number sort performed under a lock.
12136   Add new STACK function sk_is_sorted().
12137
12138   *Steve Henson*
12139
12140 * Add Delta CRL to the extension code.
12141
12142   *Steve Henson*
12143
12144 * Various fixes to s3_pkt.c so alerts are sent properly.
12145
12146   *David Holmes <d.holmes@f5.com>*
12147
12148 * Reduce the chances of duplicate issuer name and serial numbers (in
12149   violation of RFC3280) using the OpenSSL certificate creation utilities.
12150   This is done by creating a random 64 bit value for the initial serial
12151   number when a serial number file is created or when a self signed
12152   certificate is created using 'openssl req -x509'. The initial serial
12153   number file is created using 'openssl x509 -next_serial' in CA.pl
12154   rather than being initialized to 1.
12155
12156   *Steve Henson*
12157
12158### Changes between 0.9.7c and 0.9.7d  [17 Mar 2004]
12159
12160 * Fix null-pointer assignment in do_change_cipher_spec() revealed
12161   by using the Codenomicon TLS Test Tool ([CVE-2004-0079])
12162
12163   *Joe Orton, Steve Henson*
12164
12165 * Fix flaw in SSL/TLS handshaking when using Kerberos ciphersuites
12166   ([CVE-2004-0112])
12167
12168   *Joe Orton, Steve Henson*
12169
12170 * Make it possible to have multiple active certificates with the same
12171   subject in the CA index file.  This is done only if the keyword
12172   'unique_subject' is set to 'no' in the main CA section (default
12173   if 'CA_default') of the configuration file.  The value is saved
12174   with the database itself in a separate index attribute file,
12175   named like the index file with '.attr' appended to the name.
12176
12177   *Richard Levitte*
12178
12179 * X509 verify fixes. Disable broken certificate workarounds when
12180   X509_V_FLAGS_X509_STRICT is set. Check CRL issuer has cRLSign set if
12181   keyUsage extension present. Don't accept CRLs with unhandled critical
12182   extensions: since verify currently doesn't process CRL extensions this
12183   rejects a CRL with *any* critical extensions. Add new verify error codes
12184   for these cases.
12185
12186   *Steve Henson*
12187
12188 * When creating an OCSP nonce use an OCTET STRING inside the extnValue.
12189   A clarification of RFC2560 will require the use of OCTET STRINGs and
12190   some implementations cannot handle the current raw format. Since OpenSSL
12191   copies and compares OCSP nonces as opaque blobs without any attempt at
12192   parsing them this should not create any compatibility issues.
12193
12194   *Steve Henson*
12195
12196 * New md flag EVP_MD_CTX_FLAG_REUSE this allows md_data to be reused when
12197   calling EVP_MD_CTX_copy_ex() to avoid calling OPENSSL_malloc(). Without
12198   this HMAC (and other) operations are several times slower than OpenSSL
12199   < 0.9.7.
12200
12201   *Steve Henson*
12202
12203 * Print out GeneralizedTime and UTCTime in ASN1_STRING_print_ex().
12204
12205   *Peter Sylvester <Peter.Sylvester@EdelWeb.fr>*
12206
12207 * Use the correct content when signing type "other".
12208
12209   *Steve Henson*
12210
12211### Changes between 0.9.7b and 0.9.7c  [30 Sep 2003]
12212
12213 * Fix various bugs revealed by running the NISCC test suite:
12214
12215   Stop out of bounds reads in the ASN1 code when presented with
12216   invalid tags (CVE-2003-0543 and CVE-2003-0544).
12217
12218   Free up ASN1_TYPE correctly if ANY type is invalid ([CVE-2003-0545]).
12219
12220   If verify callback ignores invalid public key errors don't try to check
12221   certificate signature with the NULL public key.
12222
12223   *Steve Henson*
12224
12225 * New -ignore_err option in ocsp application to stop the server
12226   exiting on the first error in a request.
12227
12228   *Steve Henson*
12229
12230 * In ssl3_accept() (ssl/s3_srvr.c) only accept a client certificate
12231   if the server requested one: as stated in TLS 1.0 and SSL 3.0
12232   specifications.
12233
12234   *Steve Henson*
12235
12236 * In ssl3_get_client_hello() (ssl/s3_srvr.c), tolerate additional
12237   extra data after the compression methods not only for TLS 1.0
12238   but also for SSL 3.0 (as required by the specification).
12239
12240   *Bodo Moeller; problem pointed out by Matthias Loepfe*
12241
12242 * Change X509_certificate_type() to mark the key as exported/exportable
12243   when it's 512 *bits* long, not 512 bytes.
12244
12245   *Richard Levitte*
12246
12247 * Change AES_cbc_encrypt() so it outputs exact multiple of
12248   blocks during encryption.
12249
12250   *Richard Levitte*
12251
12252 * Various fixes to base64 BIO and non blocking I/O. On write
12253   flushes were not handled properly if the BIO retried. On read
12254   data was not being buffered properly and had various logic bugs.
12255   This also affects blocking I/O when the data being decoded is a
12256   certain size.
12257
12258   *Steve Henson*
12259
12260 * Various S/MIME bugfixes and compatibility changes:
12261   output correct application/pkcs7 MIME type if
12262   PKCS7_NOOLDMIMETYPE is set. Tolerate some broken signatures.
12263   Output CR+LF for EOL if PKCS7_CRLFEOL is set (this makes opening
12264   of files as .eml work). Correctly handle very long lines in MIME
12265   parser.
12266
12267   *Steve Henson*
12268
12269### Changes between 0.9.7a and 0.9.7b  [10 Apr 2003]
12270
12271 * Countermeasure against the Klima-Pokorny-Rosa extension of
12272   Bleichbacher's attack on PKCS #1 v1.5 padding: treat
12273   a protocol version number mismatch like a decryption error
12274   in ssl3_get_client_key_exchange (ssl/s3_srvr.c).
12275
12276   *Bodo Moeller*
12277
12278 * Turn on RSA blinding by default in the default implementation
12279   to avoid a timing attack. Applications that don't want it can call
12280   RSA_blinding_off() or use the new flag RSA_FLAG_NO_BLINDING.
12281   They would be ill-advised to do so in most cases.
12282
12283   *Ben Laurie, Steve Henson, Geoff Thorpe, Bodo Moeller*
12284
12285 * Change RSA blinding code so that it works when the PRNG is not
12286   seeded (in this case, the secret RSA exponent is abused as
12287   an unpredictable seed -- if it is not unpredictable, there
12288   is no point in blinding anyway).  Make RSA blinding thread-safe
12289   by remembering the creator's thread ID in rsa->blinding and
12290   having all other threads use local one-time blinding factors
12291   (this requires more computation than sharing rsa->blinding, but
12292   avoids excessive locking; and if an RSA object is not shared
12293   between threads, blinding will still be very fast).
12294
12295   *Bodo Moeller*
12296
12297 * Fixed a typo bug that would cause ENGINE_set_default() to set an
12298   ENGINE as defaults for all supported algorithms irrespective of
12299   the 'flags' parameter. 'flags' is now honoured, so applications
12300   should make sure they are passing it correctly.
12301
12302   *Geoff Thorpe*
12303
12304 * Target "mingw" now allows native Windows code to be generated in
12305   the Cygwin environment as well as with the MinGW compiler.
12306
12307   *Ulf Moeller*
12308
12309### Changes between 0.9.7 and 0.9.7a  [19 Feb 2003]
12310
12311 * In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked
12312   via timing by performing a MAC computation even if incorrect
12313   block cipher padding has been found.  This is a countermeasure
12314   against active attacks where the attacker has to distinguish
12315   between bad padding and a MAC verification error. ([CVE-2003-0078])
12316
12317   *Bodo Moeller; problem pointed out by Brice Canvel (EPFL),
12318   Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and
12319   Martin Vuagnoux (EPFL, Ilion)*
12320
12321 * Make the no-err option work as intended.  The intention with no-err
12322   is not to have the whole error stack handling routines removed from
12323   libcrypto, it's only intended to remove all the function name and
12324   reason texts, thereby removing some of the footprint that may not
12325   be interesting if those errors aren't displayed anyway.
12326
12327   NOTE: it's still possible for any application or module to have its
12328   own set of error texts inserted.  The routines are there, just not
12329   used by default when no-err is given.
12330
12331   *Richard Levitte*
12332
12333 * Add support for FreeBSD on IA64.
12334
12335   *dirk.meyer@dinoex.sub.org via Richard Levitte, resolves #454*
12336
12337 * Adjust DES_cbc_cksum() so it returns the same value as the MIT
12338   Kerberos function mit_des_cbc_cksum().  Before this change,
12339   the value returned by DES_cbc_cksum() was like the one from
12340   mit_des_cbc_cksum(), except the bytes were swapped.
12341
12342   *Kevin Greaney <Kevin.Greaney@hp.com> and Richard Levitte*
12343
12344 * Allow an application to disable the automatic SSL chain building.
12345   Before this a rather primitive chain build was always performed in
12346   ssl3_output_cert_chain(): an application had no way to send the
12347   correct chain if the automatic operation produced an incorrect result.
12348
12349   Now the chain builder is disabled if either:
12350
12351   1. Extra certificates are added via SSL_CTX_add_extra_chain_cert().
12352
12353   2. The mode flag SSL_MODE_NO_AUTO_CHAIN is set.
12354
12355   The reasoning behind this is that an application would not want the
12356   auto chain building to take place if extra chain certificates are
12357   present and it might also want a means of sending no additional
12358   certificates (for example the chain has two certificates and the
12359   root is omitted).
12360
12361   *Steve Henson*
12362
12363 * Add the possibility to build without the ENGINE framework.
12364
12365   *Steven Reddie <smr@essemer.com.au> via Richard Levitte*
12366
12367 * Under Win32 gmtime() can return NULL: check return value in
12368   OPENSSL_gmtime(). Add error code for case where gmtime() fails.
12369
12370   *Steve Henson*
12371
12372 * DSA routines: under certain error conditions uninitialized BN objects
12373   could be freed. Solution: make sure initialization is performed early
12374   enough. (Reported and fix supplied by Ivan D Nestlerode <nestler@MIT.EDU>,
12375   Nils Larsch <nla@trustcenter.de> via PR#459)
12376
12377   *Lutz Jaenicke*
12378
12379 * Another fix for SSLv2 session ID handling: the session ID was incorrectly
12380   checked on reconnect on the client side, therefore session resumption
12381   could still fail with a "ssl session id is different" error. This
12382   behaviour is masked when SSL_OP_ALL is used due to
12383   SSL_OP_MICROSOFT_SESS_ID_BUG being set.
12384   Behaviour observed by Crispin Flowerday <crispin@flowerday.cx> as
12385   followup to PR #377.
12386
12387   *Lutz Jaenicke*
12388
12389 * IA-32 assembler support enhancements: unified ELF targets, support
12390   for SCO/Caldera platforms, fix for Cygwin shared build.
12391
12392   *Andy Polyakov*
12393
12394 * Add support for FreeBSD on sparc64.  As a consequence, support for
12395   FreeBSD on non-x86 processors is separate from x86 processors on
12396   the config script, much like the NetBSD support.
12397
12398   *Richard Levitte & Kris Kennaway <kris@obsecurity.org>*
12399
12400### Changes between 0.9.6h and 0.9.7  [31 Dec 2002]
12401
12402[NB: OpenSSL 0.9.6i and later 0.9.6 patch levels were released after
12403OpenSSL 0.9.7.]
12404
12405 * Fix session ID handling in SSLv2 client code: the SERVER FINISHED
12406   code (06) was taken as the first octet of the session ID and the last
12407   octet was ignored consequently. As a result SSLv2 client side session
12408   caching could not have worked due to the session ID mismatch between
12409   client and server.
12410   Behaviour observed by Crispin Flowerday <crispin@flowerday.cx> as
12411   PR #377.
12412
12413   *Lutz Jaenicke*
12414
12415 * Change the declaration of needed Kerberos libraries to use EX_LIBS
12416   instead of the special (and badly supported) LIBKRB5.  LIBKRB5 is
12417   removed entirely.
12418
12419   *Richard Levitte*
12420
12421 * The hw_ncipher.c engine requires dynamic locks.  Unfortunately, it
12422   seems that in spite of existing for more than a year, many application
12423   author have done nothing to provide the necessary callbacks, which
12424   means that this particular engine will not work properly anywhere.
12425   This is a very unfortunate situation which forces us, in the name
12426   of usability, to give the hw_ncipher.c a static lock, which is part
12427   of libcrypto.
12428   NOTE: This is for the 0.9.7 series ONLY.  This hack will never
12429   appear in 0.9.8 or later.  We EXPECT application authors to have
12430   dealt properly with this when 0.9.8 is released (unless we actually
12431   make such changes in the libcrypto locking code that changes will
12432   have to be made anyway).
12433
12434   *Richard Levitte*
12435
12436 * In asn1_d2i_read_bio() repeatedly call BIO_read() until all content
12437   octets have been read, EOF or an error occurs. Without this change
12438   some truncated ASN1 structures will not produce an error.
12439
12440   *Steve Henson*
12441
12442 * Disable Heimdal support, since it hasn't been fully implemented.
12443   Still give the possibility to force the use of Heimdal, but with
12444   warnings and a request that patches get sent to openssl-dev.
12445
12446   *Richard Levitte*
12447
12448 * Add the VC-CE target, introduce the WINCE sysname, and add
12449   INSTALL.WCE and appropriate conditionals to make it build.
12450
12451   *Steven Reddie <smr@essemer.com.au> via Richard Levitte*
12452
12453 * Change the DLL names for Cygwin to cygcrypto-x.y.z.dll and
12454   cygssl-x.y.z.dll, where x, y and z are the major, minor and
12455   edit numbers of the version.
12456
12457   *Corinna Vinschen <vinschen@redhat.com> and Richard Levitte*
12458
12459 * Introduce safe string copy and catenation functions
12460   (BUF_strlcpy() and BUF_strlcat()).
12461
12462   *Ben Laurie (CHATS) and Richard Levitte*
12463
12464 * Avoid using fixed-size buffers for one-line DNs.
12465
12466   *Ben Laurie (CHATS)*
12467
12468 * Add BUF_MEM_grow_clean() to avoid information leakage when
12469   resizing buffers containing secrets, and use where appropriate.
12470
12471   *Ben Laurie (CHATS)*
12472
12473 * Avoid using fixed size buffers for configuration file location.
12474
12475   *Ben Laurie (CHATS)*
12476
12477 * Avoid filename truncation for various CA files.
12478
12479   *Ben Laurie (CHATS)*
12480
12481 * Use sizeof in preference to magic numbers.
12482
12483   *Ben Laurie (CHATS)*
12484
12485 * Avoid filename truncation in cert requests.
12486
12487   *Ben Laurie (CHATS)*
12488
12489 * Add assertions to check for (supposedly impossible) buffer
12490   overflows.
12491
12492   *Ben Laurie (CHATS)*
12493
12494 * Don't cache truncated DNS entries in the local cache (this could
12495   potentially lead to a spoofing attack).
12496
12497   *Ben Laurie (CHATS)*
12498
12499 * Fix various buffers to be large enough for hex/decimal
12500   representations in a platform independent manner.
12501
12502   *Ben Laurie (CHATS)*
12503
12504 * Add CRYPTO_realloc_clean() to avoid information leakage when
12505   resizing buffers containing secrets, and use where appropriate.
12506
12507   *Ben Laurie (CHATS)*
12508
12509 * Add BIO_indent() to avoid much slightly worrying code to do
12510   indents.
12511
12512   *Ben Laurie (CHATS)*
12513
12514 * Convert sprintf()/BIO_puts() to BIO_printf().
12515
12516   *Ben Laurie (CHATS)*
12517
12518 * buffer_gets() could terminate with the buffer only half
12519   full. Fixed.
12520
12521   *Ben Laurie (CHATS)*
12522
12523 * Add assertions to prevent user-supplied crypto functions from
12524   overflowing internal buffers by having large block sizes, etc.
12525
12526   *Ben Laurie (CHATS)*
12527
12528 * New OPENSSL_assert() macro (similar to assert(), but enabled
12529   unconditionally).
12530
12531   *Ben Laurie (CHATS)*
12532
12533 * Eliminate unused copy of key in RC4.
12534
12535   *Ben Laurie (CHATS)*
12536
12537 * Eliminate unused and incorrectly sized buffers for IV in pem.h.
12538
12539   *Ben Laurie (CHATS)*
12540
12541 * Fix off-by-one error in EGD path.
12542
12543   *Ben Laurie (CHATS)*
12544
12545 * If RANDFILE path is too long, ignore instead of truncating.
12546
12547   *Ben Laurie (CHATS)*
12548
12549 * Eliminate unused and incorrectly sized X.509 structure
12550   CBCParameter.
12551
12552   *Ben Laurie (CHATS)*
12553
12554 * Eliminate unused and dangerous function knumber().
12555
12556   *Ben Laurie (CHATS)*
12557
12558 * Eliminate unused and dangerous structure, KSSL_ERR.
12559
12560   *Ben Laurie (CHATS)*
12561
12562 * Protect against overlong session ID context length in an encoded
12563   session object. Since these are local, this does not appear to be
12564   exploitable.
12565
12566   *Ben Laurie (CHATS)*
12567
12568 * Change from security patch (see 0.9.6e below) that did not affect
12569   the 0.9.6 release series:
12570
12571   Remote buffer overflow in SSL3 protocol - an attacker could
12572   supply an oversized master key in Kerberos-enabled versions.
12573   ([CVE-2002-0657])
12574
12575   *Ben Laurie (CHATS)*
12576
12577 * Change the SSL kerb5 codes to match RFC 2712.
12578
12579   *Richard Levitte*
12580
12581 * Make -nameopt work fully for req and add -reqopt switch.
12582
12583   *Michael Bell <michael.bell@rz.hu-berlin.de>, Steve Henson*
12584
12585 * The "block size" for block ciphers in CFB and OFB mode should be 1.
12586
12587   *Steve Henson, reported by Yngve Nysaeter Pettersen <yngve@opera.com>*
12588
12589 * Make sure tests can be performed even if the corresponding algorithms
12590   have been removed entirely.  This was also the last step to make
12591   OpenSSL compilable with DJGPP under all reasonable conditions.
12592
12593   *Richard Levitte, Doug Kaufman <dkaufman@rahul.net>*
12594
12595 * Add cipher selection rules COMPLEMENTOFALL and COMPLEMENTOFDEFAULT
12596   to allow version independent disabling of normally unselected ciphers,
12597   which may be activated as a side-effect of selecting a single cipher.
12598
12599   (E.g., cipher list string "RSA" enables ciphersuites that are left
12600   out of "ALL" because they do not provide symmetric encryption.
12601   "RSA:!COMPLEMEMENTOFALL" avoids these unsafe ciphersuites.)
12602
12603   *Lutz Jaenicke, Bodo Moeller*
12604
12605 * Add appropriate support for separate platform-dependent build
12606   directories.  The recommended way to make a platform-dependent
12607   build directory is the following (tested on Linux), maybe with
12608   some local tweaks:
12609
12610           # Place yourself outside of the OpenSSL source tree.  In
12611           # this example, the environment variable OPENSSL_SOURCE
12612           # is assumed to contain the absolute OpenSSL source directory.
12613           mkdir -p objtree/"`uname -s`-`uname -r`-`uname -m`"
12614           cd objtree/"`uname -s`-`uname -r`-`uname -m`"
12615           (cd $OPENSSL_SOURCE; find . -type f) | while read F; do
12616                   mkdir -p `dirname $F`
12617                   ln -s $OPENSSL_SOURCE/$F $F
12618           done
12619
12620   To be absolutely sure not to disturb the source tree, a "make clean"
12621   is a good thing.  If it isn't successful, don't worry about it,
12622   it probably means the source directory is very clean.
12623
12624   *Richard Levitte*
12625
12626 * Make sure any ENGINE control commands make local copies of string
12627   pointers passed to them whenever necessary. Otherwise it is possible
12628   the caller may have overwritten (or deallocated) the original string
12629   data when a later ENGINE operation tries to use the stored values.
12630
12631   *Götz Babin-Ebell <babinebell@trustcenter.de>*
12632
12633 * Improve diagnostics in file reading and command-line digests.
12634
12635   *Ben Laurie aided and abetted by Solar Designer <solar@openwall.com>*
12636
12637 * Add AES modes CFB and OFB to the object database.  Correct an
12638   error in AES-CFB decryption.
12639
12640   *Richard Levitte*
12641
12642 * Remove most calls to EVP_CIPHER_CTX_cleanup() in evp_enc.c, this
12643   allows existing EVP_CIPHER_CTX structures to be reused after
12644   calling `EVP_*Final()`. This behaviour is used by encryption
12645   BIOs and some applications. This has the side effect that
12646   applications must explicitly clean up cipher contexts with
12647   EVP_CIPHER_CTX_cleanup() or they will leak memory.
12648
12649   *Steve Henson*
12650
12651 * Check the values of dna and dnb in bn_mul_recursive before calling
12652   bn_mul_comba (a non zero value means the a or b arrays do not contain
12653   n2 elements) and fallback to bn_mul_normal if either is not zero.
12654
12655   *Steve Henson*
12656
12657 * Fix escaping of non-ASCII characters when using the -subj option
12658   of the "openssl req" command line tool. (Robert Joop <joop@fokus.gmd.de>)
12659
12660   *Lutz Jaenicke*
12661
12662 * Make object definitions compliant to LDAP (RFC2256): SN is the short
12663   form for "surname", serialNumber has no short form.
12664   Use "mail" as the short name for "rfc822Mailbox" according to RFC2798;
12665   therefore remove "mail" short name for "internet 7".
12666   The OID for unique identifiers in X509 certificates is
12667   x500UniqueIdentifier, not uniqueIdentifier.
12668   Some more OID additions. (Michael Bell <michael.bell@rz.hu-berlin.de>)
12669
12670   *Lutz Jaenicke*
12671
12672 * Add an "init" command to the ENGINE config module and auto initialize
12673   ENGINEs. Without any "init" command the ENGINE will be initialized
12674   after all ctrl commands have been executed on it. If init=1 the
12675   ENGINE is initialized at that point (ctrls before that point are run
12676   on the uninitialized ENGINE and after on the initialized one). If
12677   init=0 then the ENGINE will not be initialized at all.
12678
12679   *Steve Henson*
12680
12681 * Fix the 'app_verify_callback' interface so that the user-defined
12682   argument is actually passed to the callback: In the
12683   SSL_CTX_set_cert_verify_callback() prototype, the callback
12684   declaration has been changed from
12685           int (*cb)()
12686   into
12687           int (*cb)(X509_STORE_CTX *,void *);
12688   in ssl_verify_cert_chain (ssl/ssl_cert.c), the call
12689           i=s->ctx->app_verify_callback(&ctx)
12690   has been changed into
12691           i=s->ctx->app_verify_callback(&ctx, s->ctx->app_verify_arg).
12692
12693   To update applications using SSL_CTX_set_cert_verify_callback(),
12694   a dummy argument can be added to their callback functions.
12695
12696   *D. K. Smetters <smetters@parc.xerox.com>*
12697
12698 * Added the '4758cca' ENGINE to support IBM 4758 cards.
12699
12700   *Maurice Gittens <maurice@gittens.nl>, touchups by Geoff Thorpe*
12701
12702 * Add and OPENSSL_LOAD_CONF define which will cause
12703   OpenSSL_add_all_algorithms() to load the openssl.cnf config file.
12704   This allows older applications to transparently support certain
12705   OpenSSL features: such as crypto acceleration and dynamic ENGINE loading.
12706   Two new functions OPENSSL_add_all_algorithms_noconf() which will never
12707   load the config file and OPENSSL_add_all_algorithms_conf() which will
12708   always load it have also been added.
12709
12710   *Steve Henson*
12711
12712 * Add the OFB, CFB and CTR (all with 128 bit feedback) to AES.
12713   Adjust NIDs and EVP layer.
12714
12715   *Stephen Sprunk <stephen@sprunk.org> and Richard Levitte*
12716
12717 * Config modules support in openssl utility.
12718
12719   Most commands now load modules from the config file,
12720   though in a few (such as version) this isn't done
12721   because it couldn't be used for anything.
12722
12723   In the case of ca and req the config file used is
12724   the same as the utility itself: that is the -config
12725   command line option can be used to specify an
12726   alternative file.
12727
12728   *Steve Henson*
12729
12730 * Move default behaviour from OPENSSL_config(). If appname is NULL
12731   use "openssl_conf" if filename is NULL use default openssl config file.
12732
12733   *Steve Henson*
12734
12735 * Add an argument to OPENSSL_config() to allow the use of an alternative
12736   config section name. Add a new flag to tolerate a missing config file
12737   and move code to CONF_modules_load_file().
12738
12739   *Steve Henson*
12740
12741 * Support for crypto accelerator cards from Accelerated Encryption
12742   Processing, www.aep.ie.  (Use engine 'aep')
12743   The support was copied from 0.9.6c [engine] and adapted/corrected
12744   to work with the new engine framework.
12745
12746   *AEP Inc. and Richard Levitte*
12747
12748 * Support for SureWare crypto accelerator cards from Baltimore
12749   Technologies.  (Use engine 'sureware')
12750   The support was copied from 0.9.6c [engine] and adapted
12751   to work with the new engine framework.
12752
12753   *Richard Levitte*
12754
12755 * Have the CHIL engine fork-safe (as defined by nCipher) and actually
12756   make the newer ENGINE framework commands for the CHIL engine work.
12757
12758   *Toomas Kiisk <vix@cyber.ee> and Richard Levitte*
12759
12760 * Make it possible to produce shared libraries on ReliantUNIX.
12761
12762   *Robert Dahlem <Robert.Dahlem@ffm2.siemens.de> via Richard Levitte*
12763
12764 * Add the configuration target debug-linux-ppro.
12765   Make 'openssl rsa' use the general key loading routines
12766   implemented in `apps.c`, and make those routines able to
12767   handle the key format FORMAT_NETSCAPE and the variant
12768   FORMAT_IISSGC.
12769
12770   *Toomas Kiisk <vix@cyber.ee> via Richard Levitte*
12771
12772 * Fix a crashbug and a logic bug in hwcrhk_load_pubkey().
12773
12774   *Toomas Kiisk <vix@cyber.ee> via Richard Levitte*
12775
12776 * Add -keyform to rsautl, and document -engine.
12777
12778   *Richard Levitte, inspired by Toomas Kiisk <vix@cyber.ee>*
12779
12780 * Change BIO_new_file (crypto/bio/bss_file.c) to use new
12781   BIO_R_NO_SUCH_FILE error code rather than the generic
12782   ERR_R_SYS_LIB error code if fopen() fails with ENOENT.
12783
12784   *Ben Laurie*
12785
12786 * Add new functions
12787           ERR_peek_last_error
12788           ERR_peek_last_error_line
12789           ERR_peek_last_error_line_data.
12790   These are similar to
12791           ERR_peek_error
12792           ERR_peek_error_line
12793           ERR_peek_error_line_data,
12794   but report on the latest error recorded rather than the first one
12795   still in the error queue.
12796
12797   *Ben Laurie, Bodo Moeller*
12798
12799 * default_algorithms option in ENGINE config module. This allows things
12800   like:
12801   default_algorithms = ALL
12802   default_algorithms = RSA, DSA, RAND, CIPHERS, DIGESTS
12803
12804   *Steve Henson*
12805
12806 * Preliminary ENGINE config module.
12807
12808   *Steve Henson*
12809
12810 * New experimental application configuration code.
12811
12812   *Steve Henson*
12813
12814 * Change the AES code to follow the same name structure as all other
12815   symmetric ciphers, and behave the same way.  Move everything to
12816   the directory crypto/aes, thereby obsoleting crypto/rijndael.
12817
12818   *Stephen Sprunk <stephen@sprunk.org> and Richard Levitte*
12819
12820 * SECURITY: remove unsafe setjmp/signal interaction from ui_openssl.c.
12821
12822   *Ben Laurie and Theo de Raadt*
12823
12824 * Add option to output public keys in req command.
12825
12826   *Massimiliano Pala madwolf@openca.org*
12827
12828 * Use wNAFs in EC_POINTs_mul() for improved efficiency
12829   (up to about 10% better than before for P-192 and P-224).
12830
12831   *Bodo Moeller*
12832
12833 * New functions/macros
12834
12835           SSL_CTX_set_msg_callback(ctx, cb)
12836           SSL_CTX_set_msg_callback_arg(ctx, arg)
12837           SSL_set_msg_callback(ssl, cb)
12838           SSL_set_msg_callback_arg(ssl, arg)
12839
12840   to request calling a callback function
12841
12842           void cb(int write_p, int version, int content_type,
12843                   const void *buf, size_t len, SSL *ssl, void *arg)
12844
12845   whenever a protocol message has been completely received
12846   (write_p == 0) or sent (write_p == 1).  Here 'version' is the
12847   protocol version  according to which the SSL library interprets
12848   the current protocol message (SSL2_VERSION, SSL3_VERSION, or
12849   TLS1_VERSION).  'content_type' is 0 in the case of SSL 2.0, or
12850   the content type as defined in the SSL 3.0/TLS 1.0 protocol
12851   specification (change_cipher_spec(20), alert(21), handshake(22)).
12852   'buf' and 'len' point to the actual message, 'ssl' to the
12853   SSL object, and 'arg' is the application-defined value set by
12854   SSL[_CTX]_set_msg_callback_arg().
12855
12856   'openssl s_client' and 'openssl s_server' have new '-msg' options
12857   to enable a callback that displays all protocol messages.
12858
12859   *Bodo Moeller*
12860
12861 * Change the shared library support so shared libraries are built as
12862   soon as the corresponding static library is finished, and thereby get
12863   openssl and the test programs linked against the shared library.
12864   This still only happens when the keyword "shard" has been given to
12865   the configuration scripts.
12866
12867   NOTE: shared library support is still an experimental thing, and
12868   backward binary compatibility is still not guaranteed.
12869
12870   *"Maciej W. Rozycki" <macro@ds2.pg.gda.pl> and Richard Levitte*
12871
12872 * Add support for Subject Information Access extension.
12873
12874   *Peter Sylvester <Peter.Sylvester@EdelWeb.fr>*
12875
12876 * Make BUF_MEM_grow() behaviour more consistent: Initialise to zero
12877   additional bytes when new memory had to be allocated, not just
12878   when reusing an existing buffer.
12879
12880   *Bodo Moeller*
12881
12882 * New command line and configuration option 'utf8' for the req command.
12883   This allows field values to be specified as UTF8 strings.
12884
12885   *Steve Henson*
12886
12887 * Add -multi and -mr options to "openssl speed" - giving multiple parallel
12888   runs for the former and machine-readable output for the latter.
12889
12890   *Ben Laurie*
12891
12892 * Add '-noemailDN' option to 'openssl ca'.  This prevents inclusion
12893   of the e-mail address in the DN (i.e., it will go into a certificate
12894   extension only).  The new configuration file option 'email_in_dn = no'
12895   has the same effect.
12896
12897   *Massimiliano Pala madwolf@openca.org*
12898
12899 * Change all functions with names starting with `des_` to be starting
12900   with `DES_` instead.  Add wrappers that are compatible with libdes,
12901   but are named `_ossl_old_des_*`.  Finally, add macros that map the
12902   `des_*` symbols to the corresponding `_ossl_old_des_*` if libdes
12903   compatibility is desired.  If OpenSSL 0.9.6c compatibility is
12904   desired, the `des_*` symbols will be mapped to `DES_*`, with one
12905   exception.
12906
12907   Since we provide two compatibility mappings, the user needs to
12908   define the macro OPENSSL_DES_LIBDES_COMPATIBILITY if libdes
12909   compatibility is desired.  The default (i.e., when that macro
12910   isn't defined) is OpenSSL 0.9.6c compatibility.
12911
12912   There are also macros that enable and disable the support of old
12913   des functions altogether.  Those are OPENSSL_ENABLE_OLD_DES_SUPPORT
12914   and OPENSSL_DISABLE_OLD_DES_SUPPORT.  If none or both of those
12915   are defined, the default will apply: to support the old des routines.
12916
12917   In either case, one must include openssl/des.h to get the correct
12918   definitions.  Do not try to just include openssl/des_old.h, that
12919   won't work.
12920
12921   NOTE: This is a major break of an old API into a new one.  Software
12922   authors are encouraged to switch to the `DES_` style functions.  Some
12923   time in the future, des_old.h and the libdes compatibility functions
12924   will be disable (i.e. OPENSSL_DISABLE_OLD_DES_SUPPORT will be the
12925   default), and then completely removed.
12926
12927   *Richard Levitte*
12928
12929 * Test for certificates which contain unsupported critical extensions.
12930   If such a certificate is found during a verify operation it is
12931   rejected by default: this behaviour can be overridden by either
12932   handling the new error X509_V_ERR_UNHANDLED_CRITICAL_EXTENSION or
12933   by setting the verify flag X509_V_FLAG_IGNORE_CRITICAL. A new function
12934   X509_supported_extension() has also been added which returns 1 if a
12935   particular extension is supported.
12936
12937   *Steve Henson*
12938
12939 * Modify the behaviour of EVP cipher functions in similar way to digests
12940   to retain compatibility with existing code.
12941
12942   *Steve Henson*
12943
12944 * Modify the behaviour of EVP_DigestInit() and EVP_DigestFinal() to retain
12945   compatibility with existing code. In particular the 'ctx' parameter does
12946   not have to be to be initialized before the call to EVP_DigestInit() and
12947   it is tidied up after a call to EVP_DigestFinal(). New function
12948   EVP_DigestFinal_ex() which does not tidy up the ctx. Similarly function
12949   EVP_MD_CTX_copy() changed to not require the destination to be
12950   initialized valid and new function EVP_MD_CTX_copy_ex() added which
12951   requires the destination to be valid.
12952
12953   Modify all the OpenSSL digest calls to use EVP_DigestInit_ex(),
12954   EVP_DigestFinal_ex() and EVP_MD_CTX_copy_ex().
12955
12956   *Steve Henson*
12957
12958 * Change ssl3_get_message (ssl/s3_both.c) and the functions using it
12959   so that complete 'Handshake' protocol structures are kept in memory
12960   instead of overwriting 'msg_type' and 'length' with 'body' data.
12961
12962   *Bodo Moeller*
12963
12964 * Add an implementation of SSL_add_dir_cert_subjects_to_stack for Win32.
12965
12966   *Massimo Santin via Richard Levitte*
12967
12968 * Major restructuring to the underlying ENGINE code. This includes
12969   reduction of linker bloat, separation of pure "ENGINE" manipulation
12970   (initialisation, etc) from functionality dealing with implementations
12971   of specific crypto interfaces. This change also introduces integrated
12972   support for symmetric ciphers and digest implementations - so ENGINEs
12973   can now accelerate these by providing EVP_CIPHER and EVP_MD
12974   implementations of their own. This is detailed in
12975   [crypto/engine/README.md](crypto/engine/README.md)
12976   as it couldn't be adequately described here. However, there are a few
12977   API changes worth noting - some RSA, DSA, DH, and RAND functions that
12978   were changed in the original introduction of ENGINE code have now
12979   reverted back - the hooking from this code to ENGINE is now a good
12980   deal more passive and at run-time, operations deal directly with
12981   RSA_METHODs, DSA_METHODs (etc) as they did before, rather than
12982   dereferencing through an ENGINE pointer any more. Also, the ENGINE
12983   functions dealing with `BN_MOD_EXP[_CRT]` handlers have been removed -
12984   they were not being used by the framework as there is no concept of a
12985   BIGNUM_METHOD and they could not be generalised to the new
12986   'ENGINE_TABLE' mechanism that underlies the new code. Similarly,
12987   ENGINE_cpy() has been removed as it cannot be consistently defined in
12988   the new code.
12989
12990   *Geoff Thorpe*
12991
12992 * Change ASN1_GENERALIZEDTIME_check() to allow fractional seconds.
12993
12994   *Steve Henson*
12995
12996 * Change mkdef.pl to sort symbols that get the same entry number,
12997   and make sure the automatically generated functions `ERR_load_*`
12998   become part of libeay.num as well.
12999
13000   *Richard Levitte*
13001
13002 * New function SSL_renegotiate_pending().  This returns true once
13003   renegotiation has been requested (either SSL_renegotiate() call
13004   or HelloRequest/ClientHello received from the peer) and becomes
13005   false once a handshake has been completed.
13006   (For servers, SSL_renegotiate() followed by SSL_do_handshake()
13007   sends a HelloRequest, but does not ensure that a handshake takes
13008   place.  SSL_renegotiate_pending() is useful for checking if the
13009   client has followed the request.)
13010
13011   *Bodo Moeller*
13012
13013 * New SSL option SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION.
13014   By default, clients may request session resumption even during
13015   renegotiation (if session ID contexts permit); with this option,
13016   session resumption is possible only in the first handshake.
13017
13018   SSL_OP_ALL is now 0x00000FFFL instead of 0x000FFFFFL.  This makes
13019   more bits available for options that should not be part of
13020   SSL_OP_ALL (such as SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION).
13021
13022   *Bodo Moeller*
13023
13024 * Add some demos for certificate and certificate request creation.
13025
13026   *Steve Henson*
13027
13028 * Make maximum certificate chain size accepted from the peer application
13029   settable (`SSL*_get/set_max_cert_list()`), as proposed by
13030   "Douglas E. Engert" <deengert@anl.gov>.
13031
13032   *Lutz Jaenicke*
13033
13034 * Add support for shared libraries for Unixware-7
13035   (Boyd Lynn Gerber <gerberb@zenez.com>).
13036
13037   *Lutz Jaenicke*
13038
13039 * Add a "destroy" handler to ENGINEs that allows structural cleanup to
13040   be done prior to destruction. Use this to unload error strings from
13041   ENGINEs that load their own error strings. NB: This adds two new API
13042   functions to "get" and "set" this destroy handler in an ENGINE.
13043
13044   *Geoff Thorpe*
13045
13046 * Alter all existing ENGINE implementations (except "openssl" and
13047   "openbsd") to dynamically instantiate their own error strings. This
13048   makes them more flexible to be built both as statically-linked ENGINEs
13049   and self-contained shared-libraries loadable via the "dynamic" ENGINE.
13050   Also, add stub code to each that makes building them as self-contained
13051   shared-libraries easier (see [README-Engine.md](README-Engine.md)).
13052
13053   *Geoff Thorpe*
13054
13055 * Add a "dynamic" ENGINE that provides a mechanism for binding ENGINE
13056   implementations into applications that are completely implemented in
13057   self-contained shared-libraries. The "dynamic" ENGINE exposes control
13058   commands that can be used to configure what shared-library to load and
13059   to control aspects of the way it is handled. Also, made an update to
13060   the [README-Engine.md](README-Engine.md) file
13061   that brings its information up-to-date and
13062   provides some information and instructions on the "dynamic" ENGINE
13063   (ie. how to use it, how to build "dynamic"-loadable ENGINEs, etc).
13064
13065   *Geoff Thorpe*
13066
13067 * Make it possible to unload ranges of ERR strings with a new
13068   "ERR_unload_strings" function.
13069
13070   *Geoff Thorpe*
13071
13072 * Add a copy() function to EVP_MD.
13073
13074   *Ben Laurie*
13075
13076 * Make EVP_MD routines take a context pointer instead of just the
13077   md_data void pointer.
13078
13079   *Ben Laurie*
13080
13081 * Add flags to EVP_MD and EVP_MD_CTX. EVP_MD_FLAG_ONESHOT indicates
13082   that the digest can only process a single chunk of data
13083   (typically because it is provided by a piece of
13084   hardware). EVP_MD_CTX_FLAG_ONESHOT indicates that the application
13085   is only going to provide a single chunk of data, and hence the
13086   framework needn't accumulate the data for oneshot drivers.
13087
13088   *Ben Laurie*
13089
13090 * As with "ERR", make it possible to replace the underlying "ex_data"
13091   functions. This change also alters the storage and management of global
13092   ex_data state - it's now all inside ex_data.c and all "class" code (eg.
13093   RSA, BIO, SSL_CTX, etc) no longer stores its own STACKS and per-class
13094   index counters. The API functions that use this state have been changed
13095   to take a "class_index" rather than pointers to the class's local STACK
13096   and counter, and there is now an API function to dynamically create new
13097   classes. This centralisation allows us to (a) plug a lot of the
13098   thread-safety problems that existed, and (b) makes it possible to clean
13099   up all allocated state using "CRYPTO_cleanup_all_ex_data()". W.r.t. (b)
13100   such data would previously have always leaked in application code and
13101   workarounds were in place to make the memory debugging turn a blind eye
13102   to it. Application code that doesn't use this new function will still
13103   leak as before, but their memory debugging output will announce it now
13104   rather than letting it slide.
13105
13106   Besides the addition of CRYPTO_cleanup_all_ex_data(), another API change
13107   induced by the "ex_data" overhaul is that X509_STORE_CTX_init() now
13108   has a return value to indicate success or failure.
13109
13110   *Geoff Thorpe*
13111
13112 * Make it possible to replace the underlying "ERR" functions such that the
13113   global state (2 LHASH tables and 2 locks) is only used by the "default"
13114   implementation. This change also adds two functions to "get" and "set"
13115   the implementation prior to it being automatically set the first time
13116   any other ERR function takes place. Ie. an application can call "get",
13117   pass the return value to a module it has just loaded, and that module
13118   can call its own "set" function using that value. This means the
13119   module's "ERR" operations will use (and modify) the error state in the
13120   application and not in its own statically linked copy of OpenSSL code.
13121
13122   *Geoff Thorpe*
13123
13124 * Give DH, DSA, and RSA types their own `*_up_ref()` function to increment
13125   reference counts. This performs normal REF_PRINT/REF_CHECK macros on
13126   the operation, and provides a more encapsulated way for external code
13127   (crypto/evp/ and ssl/) to do this. Also changed the evp and ssl code
13128   to use these functions rather than manually incrementing the counts.
13129
13130   Also rename "DSO_up()" function to more descriptive "DSO_up_ref()".
13131
13132   *Geoff Thorpe*
13133
13134 * Add EVP test program.
13135
13136   *Ben Laurie*
13137
13138 * Add symmetric cipher support to ENGINE. Expect the API to change!
13139
13140   *Ben Laurie*
13141
13142 * New CRL functions: X509_CRL_set_version(), X509_CRL_set_issuer_name()
13143   X509_CRL_set_lastUpdate(), X509_CRL_set_nextUpdate(), X509_CRL_sort(),
13144   X509_REVOKED_set_serialNumber(), and X509_REVOKED_set_revocationDate().
13145   These allow a CRL to be built without having to access X509_CRL fields
13146   directly. Modify 'ca' application to use new functions.
13147
13148   *Steve Henson*
13149
13150 * Move SSL_OP_TLS_ROLLBACK_BUG out of the SSL_OP_ALL list of recommended
13151   bug workarounds. Rollback attack detection is a security feature.
13152   The problem will only arise on OpenSSL servers when TLSv1 is not
13153   available (sslv3_server_method() or SSL_OP_NO_TLSv1).
13154   Software authors not wanting to support TLSv1 will have special reasons
13155   for their choice and can explicitly enable this option.
13156
13157   *Bodo Moeller, Lutz Jaenicke*
13158
13159 * Rationalise EVP so it can be extended: don't include a union of
13160   cipher/digest structures, add init/cleanup functions for EVP_MD_CTX
13161   (similar to those existing for EVP_CIPHER_CTX).
13162   Usage example:
13163
13164           EVP_MD_CTX md;
13165
13166           EVP_MD_CTX_init(&md);             /* new function call */
13167           EVP_DigestInit(&md, EVP_sha1());
13168           EVP_DigestUpdate(&md, in, len);
13169           EVP_DigestFinal(&md, out, NULL);
13170           EVP_MD_CTX_cleanup(&md);          /* new function call */
13171
13172   *Ben Laurie*
13173
13174 * Make DES key schedule conform to the usual scheme, as well as
13175   correcting its structure. This means that calls to DES functions
13176   now have to pass a pointer to a des_key_schedule instead of a
13177   plain des_key_schedule (which was actually always a pointer
13178   anyway): E.g.,
13179
13180           des_key_schedule ks;
13181
13182           des_set_key_checked(..., &ks);
13183           des_ncbc_encrypt(..., &ks, ...);
13184
13185   (Note that a later change renames 'des_...' into 'DES_...'.)
13186
13187   *Ben Laurie*
13188
13189 * Initial reduction of linker bloat: the use of some functions, such as
13190   PEM causes large amounts of unused functions to be linked in due to
13191   poor organisation. For example pem_all.c contains every PEM function
13192   which has a knock on effect of linking in large amounts of (unused)
13193   ASN1 code. Grouping together similar functions and splitting unrelated
13194   functions prevents this.
13195
13196   *Steve Henson*
13197
13198 * Cleanup of EVP macros.
13199
13200   *Ben Laurie*
13201
13202 * Change historical references to `{NID,SN,LN}_des_ede` and ede3 to add the
13203   correct `_ecb suffix`.
13204
13205   *Ben Laurie*
13206
13207 * Add initial OCSP responder support to ocsp application. The
13208   revocation information is handled using the text based index
13209   use by the ca application. The responder can either handle
13210   requests generated internally, supplied in files (for example
13211   via a CGI script) or using an internal minimal server.
13212
13213   *Steve Henson*
13214
13215 * Add configuration choices to get zlib compression for TLS.
13216
13217   *Richard Levitte*
13218
13219 * Changes to Kerberos SSL for RFC 2712 compliance:
13220   1. Implemented real KerberosWrapper, instead of just using
13221      KRB5 AP_REQ message.  [Thanks to Simon Wilkinson <sxw@sxw.org.uk>]
13222   2. Implemented optional authenticator field of KerberosWrapper.
13223
13224   Added openssl-style ASN.1 macros for Kerberos ticket, ap_req,
13225   and authenticator structs; see crypto/krb5/.
13226
13227   Generalized Kerberos calls to support multiple Kerberos libraries.
13228   *Vern Staats <staatsvr@asc.hpc.mil>, Jeffrey Altman <jaltman@columbia.edu>
13229   via Richard Levitte*
13230
13231 * Cause 'openssl speed' to use fully hard-coded DSA keys as it
13232   already does with RSA. testdsa.h now has 'priv_key/pub_key'
13233   values for each of the key sizes rather than having just
13234   parameters (and 'speed' generating keys each time).
13235
13236   *Geoff Thorpe*
13237
13238 * Speed up EVP routines.
13239   Before:
13240crypt
13241pe              8 bytes     64 bytes    256 bytes   1024 bytes   8192 bytes
13242s-cbc           4408.85k     5560.51k     5778.46k     5862.20k     5825.16k
13243s-cbc           4389.55k     5571.17k     5792.23k     5846.91k     5832.11k
13244s-cbc           4394.32k     5575.92k     5807.44k     5848.37k     5841.30k
13245crypt
13246s-cbc           3482.66k     5069.49k     5496.39k     5614.16k     5639.28k
13247s-cbc           3480.74k     5068.76k     5510.34k     5609.87k     5635.52k
13248s-cbc           3483.72k     5067.62k     5504.60k     5708.01k     5724.80k
13249   After:
13250crypt
13251s-cbc           4660.16k     5650.19k     5807.19k     5827.13k     5783.32k
13252crypt
13253s-cbc           3624.96k     5258.21k     5530.91k     5624.30k     5628.26k
13254
13255   *Ben Laurie*
13256
13257 * Added the OS2-EMX target.
13258
13259   *"Brian Havard" <brianh@kheldar.apana.org.au> and Richard Levitte*
13260
13261 * Rewrite commands to use `NCONF` routines instead of the old `CONF`.
13262   New functions to support `NCONF` routines in extension code.
13263   New function `CONF_set_nconf()`
13264   to allow functions which take an `NCONF` to also handle the old `LHASH`
13265   structure: this means that the old `CONF` compatible routines can be
13266   retained (in particular w.rt. extensions) without having to duplicate the
13267   code. New function `X509V3_add_ext_nconf_sk()` to add extensions to a stack.
13268
13269   *Steve Henson*
13270
13271 * Enhance the general user interface with mechanisms for inner control
13272   and with possibilities to have yes/no kind of prompts.
13273
13274   *Richard Levitte*
13275
13276 * Change all calls to low-level digest routines in the library and
13277   applications to use EVP. Add missing calls to HMAC_cleanup() and
13278   don't assume HMAC_CTX can be copied using memcpy().
13279
13280   *Verdon Walker <VWalker@novell.com>, Steve Henson*
13281
13282 * Add the possibility to control engines through control names but with
13283   arbitrary arguments instead of just a string.
13284   Change the key loaders to take a UI_METHOD instead of a callback
13285   function pointer.  NOTE: this breaks binary compatibility with earlier
13286   versions of OpenSSL [engine].
13287   Adapt the nCipher code for these new conditions and add a card insertion
13288   callback.
13289
13290   *Richard Levitte*
13291
13292 * Enhance the general user interface with mechanisms to better support
13293   dialog box interfaces, application-defined prompts, the possibility
13294   to use defaults (for example default passwords from somewhere else)
13295   and interrupts/cancellations.
13296
13297   *Richard Levitte*
13298
13299 * Tidy up PKCS#12 attribute handling. Add support for the CSP name
13300   attribute in PKCS#12 files, add new -CSP option to pkcs12 utility.
13301
13302   *Steve Henson*
13303
13304 * Fix a memory leak in 'sk_dup()' in the case reallocation fails. (Also
13305   tidy up some unnecessarily weird code in 'sk_new()').
13306
13307   *Geoff, reported by Diego Tartara <dtartara@novamens.com>*
13308
13309 * Change the key loading routines for ENGINEs to use the same kind
13310   callback (pem_password_cb) as all other routines that need this
13311   kind of callback.
13312
13313   *Richard Levitte*
13314
13315 * Increase ENTROPY_NEEDED to 32 bytes, as Rijndael can operate with
13316   256 bit (=32 byte) keys. Of course seeding with more entropy bytes
13317   than this minimum value is recommended.
13318
13319   *Lutz Jaenicke*
13320
13321 * New random seeder for OpenVMS, using the system process statistics
13322   that are easily reachable.
13323
13324   *Richard Levitte*
13325
13326 * Windows apparently can't transparently handle global
13327   variables defined in DLLs. Initialisations such as:
13328
13329           const ASN1_ITEM *it = &ASN1_INTEGER_it;
13330
13331   won't compile. This is used by the any applications that need to
13332   declare their own ASN1 modules. This was fixed by adding the option
13333   EXPORT_VAR_AS_FN to all Win32 platforms, although this isn't strictly
13334   needed for static libraries under Win32.
13335
13336   *Steve Henson*
13337
13338 * New functions X509_PURPOSE_set() and X509_TRUST_set() to handle
13339   setting of purpose and trust fields. New X509_STORE trust and
13340   purpose functions and tidy up setting in other SSL functions.
13341
13342   *Steve Henson*
13343
13344 * Add copies of X509_STORE_CTX fields and callbacks to X509_STORE
13345   structure. These are inherited by X509_STORE_CTX when it is
13346   initialised. This allows various defaults to be set in the
13347   X509_STORE structure (such as flags for CRL checking and custom
13348   purpose or trust settings) for functions which only use X509_STORE_CTX
13349   internally such as S/MIME.
13350
13351   Modify X509_STORE_CTX_purpose_inherit() so it only sets purposes and
13352   trust settings if they are not set in X509_STORE. This allows X509_STORE
13353   purposes and trust (in S/MIME for example) to override any set by default.
13354
13355   Add command line options for CRL checking to smime, s_client and s_server
13356   applications.
13357
13358   *Steve Henson*
13359
13360 * Initial CRL based revocation checking. If the CRL checking flag(s)
13361   are set then the CRL is looked up in the X509_STORE structure and
13362   its validity and signature checked, then if the certificate is found
13363   in the CRL the verify fails with a revoked error.
13364
13365   Various new CRL related callbacks added to X509_STORE_CTX structure.
13366
13367   Command line options added to 'verify' application to support this.
13368
13369   This needs some additional work, such as being able to handle multiple
13370   CRLs with different times, extension based lookup (rather than just
13371   by subject name) and ultimately more complete V2 CRL extension
13372   handling.
13373
13374   *Steve Henson*
13375
13376 * Add a general user interface API (crypto/ui/).  This is designed
13377   to replace things like des_read_password and friends (backward
13378   compatibility functions using this new API are provided).
13379   The purpose is to remove prompting functions from the DES code
13380   section as well as provide for prompting through dialog boxes in
13381   a window system and the like.
13382
13383   *Richard Levitte*
13384
13385 * Add "ex_data" support to ENGINE so implementations can add state at a
13386   per-structure level rather than having to store it globally.
13387
13388   *Geoff*
13389
13390 * Make it possible for ENGINE structures to be copied when retrieved by
13391   ENGINE_by_id() if the ENGINE specifies a new flag: ENGINE_FLAGS_BY_ID_COPY.
13392   This causes the "original" ENGINE structure to act like a template,
13393   analogous to the RSA vs. RSA_METHOD type of separation. Because of this
13394   operational state can be localised to each ENGINE structure, despite the
13395   fact they all share the same "methods". New ENGINE structures returned in
13396   this case have no functional references and the return value is the single
13397   structural reference. This matches the single structural reference returned
13398   by ENGINE_by_id() normally, when it is incremented on the pre-existing
13399   ENGINE structure.
13400
13401   *Geoff*
13402
13403 * Fix ASN1 decoder when decoding type ANY and V_ASN1_OTHER: since this
13404   needs to match any other type at all we need to manually clear the
13405   tag cache.
13406
13407   *Steve Henson*
13408
13409 * Changes to the "openssl engine" utility to include;
13410   - verbosity levels ('-v', '-vv', and '-vvv') that provide information
13411     about an ENGINE's available control commands.
13412   - executing control commands from command line arguments using the
13413     '-pre' and '-post' switches. '-post' is only used if '-t' is
13414     specified and the ENGINE is successfully initialised. The syntax for
13415     the individual commands are colon-separated, for example;
13416           openssl engine chil -pre FORK_CHECK:0 -pre SO_PATH:/lib/test.so
13417
13418   *Geoff*
13419
13420 * New dynamic control command support for ENGINEs. ENGINEs can now
13421   declare their own commands (numbers), names (strings), descriptions,
13422   and input types for run-time discovery by calling applications. A
13423   subset of these commands are implicitly classed as "executable"
13424   depending on their input type, and only these can be invoked through
13425   the new string-based API function ENGINE_ctrl_cmd_string(). (Eg. this
13426   can be based on user input, config files, etc). The distinction is
13427   that "executable" commands cannot return anything other than a boolean
13428   result and can only support numeric or string input, whereas some
13429   discoverable commands may only be for direct use through
13430   ENGINE_ctrl(), eg. supporting the exchange of binary data, function
13431   pointers, or other custom uses. The "executable" commands are to
13432   support parameterisations of ENGINE behaviour that can be
13433   unambiguously defined by ENGINEs and used consistently across any
13434   OpenSSL-based application. Commands have been added to all the
13435   existing hardware-supporting ENGINEs, noticeably "SO_PATH" to allow
13436   control over shared-library paths without source code alterations.
13437
13438   *Geoff*
13439
13440 * Changed all ENGINE implementations to dynamically allocate their
13441   ENGINEs rather than declaring them statically. Apart from this being
13442   necessary with the removal of the ENGINE_FLAGS_MALLOCED distinction,
13443   this also allows the implementations to compile without using the
13444   internal engine_int.h header.
13445
13446   *Geoff*
13447
13448 * Minor adjustment to "rand" code. RAND_get_rand_method() now returns a
13449   'const' value. Any code that should be able to modify a RAND_METHOD
13450   should already have non-const pointers to it (ie. they should only
13451   modify their own ones).
13452
13453   *Geoff*
13454
13455 * Made a variety of little tweaks to the ENGINE code.
13456   - "atalla" and "ubsec" string definitions were moved from header files
13457     to C code. "nuron" string definitions were placed in variables
13458     rather than hard-coded - allowing parameterisation of these values
13459     later on via ctrl() commands.
13460   - Removed unused "#if 0"'d code.
13461   - Fixed engine list iteration code so it uses ENGINE_free() to release
13462     structural references.
13463   - Constified the RAND_METHOD element of ENGINE structures.
13464   - Constified various get/set functions as appropriate and added
13465     missing functions (including a catch-all ENGINE_cpy that duplicates
13466     all ENGINE values onto a new ENGINE except reference counts/state).
13467   - Removed NULL parameter checks in get/set functions. Setting a method
13468     or function to NULL is a way of cancelling out a previously set
13469     value.  Passing a NULL ENGINE parameter is just plain stupid anyway
13470     and doesn't justify the extra error symbols and code.
13471   - Deprecate the ENGINE_FLAGS_MALLOCED define and move the area for
13472     flags from engine_int.h to engine.h.
13473   - Changed prototypes for ENGINE handler functions (init(), finish(),
13474     ctrl(), key-load functions, etc) to take an (ENGINE*) parameter.
13475
13476   *Geoff*
13477
13478 * Implement binary inversion algorithm for BN_mod_inverse in addition
13479   to the algorithm using long division.  The binary algorithm can be
13480   used only if the modulus is odd.  On 32-bit systems, it is faster
13481   only for relatively small moduli (roughly 20-30% for 128-bit moduli,
13482   roughly 5-15% for 256-bit moduli), so we use it only for moduli
13483   up to 450 bits.  In 64-bit environments, the binary algorithm
13484   appears to be advantageous for much longer moduli; here we use it
13485   for moduli up to 2048 bits.
13486
13487   *Bodo Moeller*
13488
13489 * Rewrite CHOICE field setting in ASN1_item_ex_d2i(). The old code
13490   could not support the combine flag in choice fields.
13491
13492   *Steve Henson*
13493
13494 * Add a 'copy_extensions' option to the 'ca' utility. This copies
13495   extensions from a certificate request to the certificate.
13496
13497   *Steve Henson*
13498
13499 * Allow multiple 'certopt' and 'nameopt' options to be separated
13500   by commas. Add 'namopt' and 'certopt' options to the 'ca' config
13501   file: this allows the display of the certificate about to be
13502   signed to be customised, to allow certain fields to be included
13503   or excluded and extension details. The old system didn't display
13504   multicharacter strings properly, omitted fields not in the policy
13505   and couldn't display additional details such as extensions.
13506
13507   *Steve Henson*
13508
13509 * Function EC_POINTs_mul for multiple scalar multiplication
13510   of an arbitrary number of elliptic curve points
13511           \sum scalars[i]*points[i],
13512   optionally including the generator defined for the EC_GROUP:
13513           scalar*generator +  \sum scalars[i]*points[i].
13514
13515   EC_POINT_mul is a simple wrapper function for the typical case
13516   that the point list has just one item (besides the optional
13517   generator).
13518
13519   *Bodo Moeller*
13520
13521 * First EC_METHODs for curves over GF(p):
13522
13523   EC_GFp_simple_method() uses the basic BN_mod_mul and BN_mod_sqr
13524   operations and provides various method functions that can also
13525   operate with faster implementations of modular arithmetic.
13526
13527   EC_GFp_mont_method() reuses most functions that are part of
13528   EC_GFp_simple_method, but uses Montgomery arithmetic.
13529
13530   *Bodo Moeller; point addition and point doubling
13531   implementation directly derived from source code provided by
13532   Lenka Fibikova <fibikova@exp-math.uni-essen.de>*
13533
13534 * Framework for elliptic curves (crypto/ec/ec.h, crypto/ec/ec_lcl.h,
13535   crypto/ec/ec_lib.c):
13536
13537   Curves are EC_GROUP objects (with an optional group generator)
13538   based on EC_METHODs that are built into the library.
13539
13540   Points are EC_POINT objects based on EC_GROUP objects.
13541
13542   Most of the framework would be able to handle curves over arbitrary
13543   finite fields, but as there are no obvious types for fields other
13544   than GF(p), some functions are limited to that for now.
13545
13546   *Bodo Moeller*
13547
13548 * Add the -HTTP option to s_server.  It is similar to -WWW, but requires
13549   that the file contains a complete HTTP response.
13550
13551   *Richard Levitte*
13552
13553 * Add the ec directory to mkdef.pl and mkfiles.pl. In mkdef.pl
13554   change the def and num file printf format specifier from "%-40sXXX"
13555   to "%-39s XXX". The latter will always guarantee a space after the
13556   field while the former will cause them to run together if the field
13557   is 40 of more characters long.
13558
13559   *Steve Henson*
13560
13561 * Constify the cipher and digest 'method' functions and structures
13562   and modify related functions to take constant EVP_MD and EVP_CIPHER
13563   pointers.
13564
13565   *Steve Henson*
13566
13567 * Hide BN_CTX structure details in bn_lcl.h instead of publishing them
13568   in <openssl/bn.h>.  Also further increase BN_CTX_NUM to 32.
13569
13570   *Bodo Moeller*
13571
13572 * Modify `EVP_Digest*()` routines so they now return values. Although the
13573   internal software routines can never fail additional hardware versions
13574   might.
13575
13576   *Steve Henson*
13577
13578 * Clean up crypto/err/err.h and change some error codes to avoid conflicts:
13579
13580   Previously ERR_R_FATAL was too small and coincided with ERR_LIB_PKCS7
13581   (= ERR_R_PKCS7_LIB); it is now 64 instead of 32.
13582
13583   ASN1 error codes
13584           ERR_R_NESTED_ASN1_ERROR
13585           ...
13586           ERR_R_MISSING_ASN1_EOS
13587   were 4 .. 9, conflicting with
13588           ERR_LIB_RSA (= ERR_R_RSA_LIB)
13589           ...
13590           ERR_LIB_PEM (= ERR_R_PEM_LIB).
13591   They are now 58 .. 63 (i.e., just below ERR_R_FATAL).
13592
13593   Add new error code 'ERR_R_INTERNAL_ERROR'.
13594
13595   *Bodo Moeller*
13596
13597 * Don't overuse locks in crypto/err/err.c: For data retrieval, CRYPTO_r_lock
13598   suffices.
13599
13600   *Bodo Moeller*
13601
13602 * New option '-subj arg' for 'openssl req' and 'openssl ca'.  This
13603   sets the subject name for a new request or supersedes the
13604   subject name in a given request. Formats that can be parsed are
13605           'CN=Some Name, OU=myOU, C=IT'
13606   and
13607           'CN=Some Name/OU=myOU/C=IT'.
13608
13609   Add options '-batch' and '-verbose' to 'openssl req'.
13610
13611   *Massimiliano Pala <madwolf@hackmasters.net>*
13612
13613 * Introduce the possibility to access global variables through
13614   functions on platform were that's the best way to handle exporting
13615   global variables in shared libraries.  To enable this functionality,
13616   one must configure with "EXPORT_VAR_AS_FN" or defined the C macro
13617   "OPENSSL_EXPORT_VAR_AS_FUNCTION" in crypto/opensslconf.h (the latter
13618   is normally done by Configure or something similar).
13619
13620   To implement a global variable, use the macro OPENSSL_IMPLEMENT_GLOBAL
13621   in the source file (foo.c) like this:
13622
13623           OPENSSL_IMPLEMENT_GLOBAL(int,foo)=1;
13624           OPENSSL_IMPLEMENT_GLOBAL(double,bar);
13625
13626   To declare a global variable, use the macros OPENSSL_DECLARE_GLOBAL
13627   and OPENSSL_GLOBAL_REF in the header file (foo.h) like this:
13628
13629           OPENSSL_DECLARE_GLOBAL(int,foo);
13630           #define foo OPENSSL_GLOBAL_REF(foo)
13631           OPENSSL_DECLARE_GLOBAL(double,bar);
13632           #define bar OPENSSL_GLOBAL_REF(bar)
13633
13634   The #defines are very important, and therefore so is including the
13635   header file everywhere where the defined globals are used.
13636
13637   The macro OPENSSL_EXPORT_VAR_AS_FUNCTION also affects the definition
13638   of ASN.1 items, but that structure is a bit different.
13639
13640   The largest change is in util/mkdef.pl which has been enhanced with
13641   better and easier to understand logic to choose which symbols should
13642   go into the Windows .def files as well as a number of fixes and code
13643   cleanup (among others, algorithm keywords are now sorted
13644   lexicographically to avoid constant rewrites).
13645
13646   *Richard Levitte*
13647
13648 * In BN_div() keep a copy of the sign of 'num' before writing the
13649   result to 'rm' because if rm==num the value will be overwritten
13650   and produce the wrong result if 'num' is negative: this caused
13651   problems with BN_mod() and BN_nnmod().
13652
13653   *Steve Henson*
13654
13655 * Function OCSP_request_verify(). This checks the signature on an
13656   OCSP request and verifies the signer certificate. The signer
13657   certificate is just checked for a generic purpose and OCSP request
13658   trust settings.
13659
13660   *Steve Henson*
13661
13662 * Add OCSP_check_validity() function to check the validity of OCSP
13663   responses. OCSP responses are prepared in real time and may only
13664   be a few seconds old. Simply checking that the current time lies
13665   between thisUpdate and nextUpdate max reject otherwise valid responses
13666   caused by either OCSP responder or client clock inaccuracy. Instead
13667   we allow thisUpdate and nextUpdate to fall within a certain period of
13668   the current time. The age of the response can also optionally be
13669   checked. Two new options -validity_period and -status_age added to
13670   ocsp utility.
13671
13672   *Steve Henson*
13673
13674 * If signature or public key algorithm is unrecognized print out its
13675   OID rather that just UNKNOWN.
13676
13677   *Steve Henson*
13678
13679 * Change OCSP_cert_to_id() to tolerate a NULL subject certificate and
13680   OCSP_cert_id_new() a NULL serialNumber. This allows a partial certificate
13681   ID to be generated from the issuer certificate alone which can then be
13682   passed to OCSP_id_issuer_cmp().
13683
13684   *Steve Henson*
13685
13686 * New compilation option ASN1_ITEM_FUNCTIONS. This causes the new
13687   ASN1 modules to export functions returning ASN1_ITEM pointers
13688   instead of the ASN1_ITEM structures themselves. This adds several
13689   new macros which allow the underlying ASN1 function/structure to
13690   be accessed transparently. As a result code should not use ASN1_ITEM
13691   references directly (such as &X509_it) but instead use the relevant
13692   macros (such as ASN1_ITEM_rptr(X509)). This option is to allow
13693   use of the new ASN1 code on platforms where exporting structures
13694   is problematical (for example in shared libraries) but exporting
13695   functions returning pointers to structures is not.
13696
13697   *Steve Henson*
13698
13699 * Add support for overriding the generation of SSL/TLS session IDs.
13700   These callbacks can be registered either in an SSL_CTX or per SSL.
13701   The purpose of this is to allow applications to control, if they wish,
13702   the arbitrary values chosen for use as session IDs, particularly as it
13703   can be useful for session caching in multiple-server environments. A
13704   command-line switch for testing this (and any client code that wishes
13705   to use such a feature) has been added to "s_server".
13706
13707   *Geoff Thorpe, Lutz Jaenicke*
13708
13709 * Modify mkdef.pl to recognise and parse preprocessor conditionals
13710   of the form `#if defined(...) || defined(...) || ...` and
13711   `#if !defined(...) && !defined(...) && ...`.  This also avoids
13712   the growing number of special cases it was previously handling.
13713
13714   *Richard Levitte*
13715
13716 * Make all configuration macros available for application by making
13717   sure they are available in opensslconf.h, by giving them names starting
13718   with `OPENSSL_` to avoid conflicts with other packages and by making
13719   sure e_os2.h will cover all platform-specific cases together with
13720   opensslconf.h.
13721   Additionally, it is now possible to define configuration/platform-
13722   specific names (called "system identities").  In the C code, these
13723   are prefixed with `OPENSSL_SYSNAME_`.  e_os2.h will create another
13724   macro with the name beginning with `OPENSSL_SYS_`, which is determined
13725   from `OPENSSL_SYSNAME_*` or compiler-specific macros depending on
13726   what is available.
13727
13728   *Richard Levitte*
13729
13730 * New option -set_serial to 'req' and 'x509' this allows the serial
13731   number to use to be specified on the command line. Previously self
13732   signed certificates were hard coded with serial number 0 and the
13733   CA options of 'x509' had to use a serial number in a file which was
13734   auto incremented.
13735
13736   *Steve Henson*
13737
13738 * New options to 'ca' utility to support V2 CRL entry extensions.
13739   Currently CRL reason, invalidity date and hold instruction are
13740   supported. Add new CRL extensions to V3 code and some new objects.
13741
13742   *Steve Henson*
13743
13744 * New function EVP_CIPHER_CTX_set_padding() this is used to
13745   disable standard block padding (aka PKCS#5 padding) in the EVP
13746   API, which was previously mandatory. This means that the data is
13747   not padded in any way and so the total length much be a multiple
13748   of the block size, otherwise an error occurs.
13749
13750   *Steve Henson*
13751
13752 * Initial (incomplete) OCSP SSL support.
13753
13754   *Steve Henson*
13755
13756 * New function OCSP_parse_url(). This splits up a URL into its host,
13757   port and path components: primarily to parse OCSP URLs. New -url
13758   option to ocsp utility.
13759
13760   *Steve Henson*
13761
13762 * New nonce behavior. The return value of OCSP_check_nonce() now
13763   reflects the various checks performed. Applications can decide
13764   whether to tolerate certain situations such as an absent nonce
13765   in a response when one was present in a request: the ocsp application
13766   just prints out a warning. New function OCSP_add1_basic_nonce()
13767   this is to allow responders to include a nonce in a response even if
13768   the request is nonce-less.
13769
13770   *Steve Henson*
13771
13772 * Disable stdin buffering in `load_cert()` (`apps/apps.c`) so that no certs are
13773   skipped when using openssl x509 multiple times on a single input file,
13774   e.g. `(openssl x509 -out cert1; openssl x509 -out cert2) <certs`.
13775
13776   *Bodo Moeller*
13777
13778 * Make ASN1_UTCTIME_set_string() and ASN1_GENERALIZEDTIME_set_string()
13779   set string type: to handle setting ASN1_TIME structures. Fix ca
13780   utility to correctly initialize revocation date of CRLs.
13781
13782   *Steve Henson*
13783
13784 * New option SSL_OP_CIPHER_SERVER_PREFERENCE allows the server to override
13785   the clients preferred ciphersuites and rather use its own preferences.
13786   Should help to work around M$ SGC (Server Gated Cryptography) bug in
13787   Internet Explorer by ensuring unchanged hash method during stepup.
13788   (Also replaces the broken/deactivated SSL_OP_NON_EXPORT_FIRST option.)
13789
13790   *Lutz Jaenicke*
13791
13792 * Make mkdef.pl recognise all DECLARE_ASN1 macros, change rijndael
13793   to aes and add a new 'exist' option to print out symbols that don't
13794   appear to exist.
13795
13796   *Steve Henson*
13797
13798 * Additional options to ocsp utility to allow flags to be set and
13799   additional certificates supplied.
13800
13801   *Steve Henson*
13802
13803 * Add the option -VAfile to 'openssl ocsp', so the user can give the
13804   OCSP client a number of certificate to only verify the response
13805   signature against.
13806
13807   *Richard Levitte*
13808
13809 * Update Rijndael code to version 3.0 and change EVP AES ciphers to
13810   handle the new API. Currently only ECB, CBC modes supported. Add new
13811   AES OIDs.
13812
13813   Add TLS AES ciphersuites as described in RFC3268, "Advanced
13814   Encryption Standard (AES) Ciphersuites for Transport Layer
13815   Security (TLS)".  (In beta versions of OpenSSL 0.9.7, these were
13816   not enabled by default and were not part of the "ALL" ciphersuite
13817   alias because they were not yet official; they could be
13818   explicitly requested by specifying the "AESdraft" ciphersuite
13819   group alias.  In the final release of OpenSSL 0.9.7, the group
13820   alias is called "AES" and is part of "ALL".)
13821
13822   *Ben Laurie, Steve  Henson, Bodo Moeller*
13823
13824 * New function OCSP_copy_nonce() to copy nonce value (if present) from
13825   request to response.
13826
13827   *Steve Henson*
13828
13829 * Functions for OCSP responders. OCSP_request_onereq_count(),
13830   OCSP_request_onereq_get0(), OCSP_onereq_get0_id() and OCSP_id_get0_info()
13831   extract information from a certificate request. OCSP_response_create()
13832   creates a response and optionally adds a basic response structure.
13833   OCSP_basic_add1_status() adds a complete single response to a basic
13834   response and returns the OCSP_SINGLERESP structure just added (to allow
13835   extensions to be included for example). OCSP_basic_add1_cert() adds a
13836   certificate to a basic response and OCSP_basic_sign() signs a basic
13837   response with various flags. New helper functions ASN1_TIME_check()
13838   (checks validity of ASN1_TIME structure) and ASN1_TIME_to_generalizedtime()
13839   (converts ASN1_TIME to GeneralizedTime).
13840
13841   *Steve Henson*
13842
13843 * Various new functions. EVP_Digest() combines EVP_Digest{Init,Update,Final}()
13844   in a single operation. X509_get0_pubkey_bitstr() extracts the public_key
13845   structure from a certificate. X509_pubkey_digest() digests the public_key
13846   contents: this is used in various key identifiers.
13847
13848   *Steve Henson*
13849
13850 * Make sk_sort() tolerate a NULL argument.
13851
13852   *Steve Henson reported by Massimiliano Pala <madwolf@comune.modena.it>*
13853
13854 * New OCSP verify flag OCSP_TRUSTOTHER. When set the "other" certificates
13855   passed by the function are trusted implicitly. If any of them signed the
13856   response then it is assumed to be valid and is not verified.
13857
13858   *Steve Henson*
13859
13860 * In PKCS7_set_type() initialise content_type in PKCS7_ENC_CONTENT
13861   to data. This was previously part of the PKCS7 ASN1 code. This
13862   was causing problems with OpenSSL created PKCS#12 and PKCS#7 structures.
13863   *Steve Henson, reported by Kenneth R. Robinette
13864                              <support@securenetterm.com>*
13865
13866 * Add CRYPTO_push_info() and CRYPTO_pop_info() calls to new ASN1
13867   routines: without these tracing memory leaks is very painful.
13868   Fix leaks in PKCS12 and PKCS7 routines.
13869
13870   *Steve Henson*
13871
13872 * Make X509_time_adj() cope with the new behaviour of ASN1_TIME_new().
13873   Previously it initialised the 'type' argument to V_ASN1_UTCTIME which
13874   effectively meant GeneralizedTime would never be used. Now it
13875   is initialised to -1 but X509_time_adj() now has to check the value
13876   and use ASN1_TIME_set() if the value is not V_ASN1_UTCTIME or
13877   V_ASN1_GENERALIZEDTIME, without this it always uses GeneralizedTime.
13878   *Steve Henson, reported by Kenneth R. Robinette
13879                              <support@securenetterm.com>*
13880
13881 * Fixes to BN_to_ASN1_INTEGER when bn is zero. This would previously
13882   result in a zero length in the ASN1_INTEGER structure which was
13883   not consistent with the structure when d2i_ASN1_INTEGER() was used
13884   and would cause ASN1_INTEGER_cmp() to fail. Enhance s2i_ASN1_INTEGER()
13885   to cope with hex and negative integers. Fix bug in i2a_ASN1_INTEGER()
13886   where it did not print out a minus for negative ASN1_INTEGER.
13887
13888   *Steve Henson*
13889
13890 * Add summary printout to ocsp utility. The various functions which
13891   convert status values to strings have been renamed to:
13892   OCSP_response_status_str(), OCSP_cert_status_str() and
13893   OCSP_crl_reason_str() and are no longer static. New options
13894   to verify nonce values and to disable verification. OCSP response
13895   printout format cleaned up.
13896
13897   *Steve Henson*
13898
13899 * Add additional OCSP certificate checks. These are those specified
13900   in RFC2560. This consists of two separate checks: the CA of the
13901   certificate being checked must either be the OCSP signer certificate
13902   or the issuer of the OCSP signer certificate. In the latter case the
13903   OCSP signer certificate must contain the OCSP signing extended key
13904   usage. This check is performed by attempting to match the OCSP
13905   signer or the OCSP signer CA to the issuerNameHash and issuerKeyHash
13906   in the OCSP_CERTID structures of the response.
13907
13908   *Steve Henson*
13909
13910 * Initial OCSP certificate verification added to OCSP_basic_verify()
13911   and related routines. This uses the standard OpenSSL certificate
13912   verify routines to perform initial checks (just CA validity) and
13913   to obtain the certificate chain. Then additional checks will be
13914   performed on the chain. Currently the root CA is checked to see
13915   if it is explicitly trusted for OCSP signing. This is used to set
13916   a root CA as a global signing root: that is any certificate that
13917   chains to that CA is an acceptable OCSP signing certificate.
13918
13919   *Steve Henson*
13920
13921 * New '-extfile ...' option to 'openssl ca' for reading X.509v3
13922   extensions from a separate configuration file.
13923   As when reading extensions from the main configuration file,
13924   the '-extensions ...' option may be used for specifying the
13925   section to use.
13926
13927   *Massimiliano Pala <madwolf@comune.modena.it>*
13928
13929 * New OCSP utility. Allows OCSP requests to be generated or
13930   read. The request can be sent to a responder and the output
13931   parsed, outputted or printed in text form. Not complete yet:
13932   still needs to check the OCSP response validity.
13933
13934   *Steve Henson*
13935
13936 * New subcommands for 'openssl ca':
13937   `openssl ca -status <serial>` prints the status of the cert with
13938   the given serial number (according to the index file).
13939   `openssl ca -updatedb` updates the expiry status of certificates
13940   in the index file.
13941
13942   *Massimiliano Pala <madwolf@comune.modena.it>*
13943
13944 * New '-newreq-nodes' command option to CA.pl.  This is like
13945   '-newreq', but calls 'openssl req' with the '-nodes' option
13946   so that the resulting key is not encrypted.
13947
13948   *Damien Miller <djm@mindrot.org>*
13949
13950 * New configuration for the GNU Hurd.
13951
13952   *Jonathan Bartlett <johnnyb@wolfram.com> via Richard Levitte*
13953
13954 * Initial code to implement OCSP basic response verify. This
13955   is currently incomplete. Currently just finds the signer's
13956   certificate and verifies the signature on the response.
13957
13958   *Steve Henson*
13959
13960 * New SSLeay_version code SSLEAY_DIR to determine the compiled-in
13961   value of OPENSSLDIR.  This is available via the new '-d' option
13962   to 'openssl version', and is also included in 'openssl version -a'.
13963
13964   *Bodo Moeller*
13965
13966 * Allowing defining memory allocation callbacks that will be given
13967   file name and line number information in additional arguments
13968   (a `const char*` and an int).  The basic functionality remains, as
13969   well as the original possibility to just replace malloc(),
13970   realloc() and free() by functions that do not know about these
13971   additional arguments.  To register and find out the current
13972   settings for extended allocation functions, the following
13973   functions are provided:
13974
13975           CRYPTO_set_mem_ex_functions
13976           CRYPTO_set_locked_mem_ex_functions
13977           CRYPTO_get_mem_ex_functions
13978           CRYPTO_get_locked_mem_ex_functions
13979
13980   These work the same way as CRYPTO_set_mem_functions and friends.
13981   `CRYPTO_get_[locked_]mem_functions` now writes 0 where such an
13982   extended allocation function is enabled.
13983   Similarly, `CRYPTO_get_[locked_]mem_ex_functions` writes 0 where
13984   a conventional allocation function is enabled.
13985
13986   *Richard Levitte, Bodo Moeller*
13987
13988 * Finish off removing the remaining LHASH function pointer casts.
13989   There should no longer be any prototype-casting required when using
13990   the LHASH abstraction, and any casts that remain are "bugs". See
13991   the callback types and macros at the head of lhash.h for details
13992   (and "OBJ_cleanup" in crypto/objects/obj_dat.c as an example).
13993
13994   *Geoff Thorpe*
13995
13996 * Add automatic query of EGD sockets in RAND_poll() for the unix variant.
13997   If /dev/[u]random devices are not available or do not return enough
13998   entropy, EGD style sockets (served by EGD or PRNGD) will automatically
13999   be queried.
14000   The locations /var/run/egd-pool, /dev/egd-pool, /etc/egd-pool, and
14001   /etc/entropy will be queried once each in this sequence, querying stops
14002   when enough entropy was collected without querying more sockets.
14003
14004   *Lutz Jaenicke*
14005
14006 * Change the Unix RAND_poll() variant to be able to poll several
14007   random devices, as specified by DEVRANDOM, until a sufficient amount
14008   of data has been collected.   We spend at most 10 ms on each file
14009   (select timeout) and read in non-blocking mode.  DEVRANDOM now
14010   defaults to the list "/dev/urandom", "/dev/random", "/dev/srandom"
14011   (previously it was just the string "/dev/urandom"), so on typical
14012   platforms the 10 ms delay will never occur.
14013   Also separate out the Unix variant to its own file, rand_unix.c.
14014   For VMS, there's a currently-empty rand_vms.c.
14015
14016   *Richard Levitte*
14017
14018 * Move OCSP client related routines to ocsp_cl.c. These
14019   provide utility functions which an application needing
14020   to issue a request to an OCSP responder and analyse the
14021   response will typically need: as opposed to those which an
14022   OCSP responder itself would need which will be added later.
14023
14024   OCSP_request_sign() signs an OCSP request with an API similar
14025   to PKCS7_sign(). OCSP_response_status() returns status of OCSP
14026   response. OCSP_response_get1_basic() extracts basic response
14027   from response. OCSP_resp_find_status(): finds and extracts status
14028   information from an OCSP_CERTID structure (which will be created
14029   when the request structure is built). These are built from lower
14030   level functions which work on OCSP_SINGLERESP structures but
14031   won't normally be used unless the application wishes to examine
14032   extensions in the OCSP response for example.
14033
14034   Replace nonce routines with a pair of functions.
14035   OCSP_request_add1_nonce() adds a nonce value and optionally
14036   generates a random value. OCSP_check_nonce() checks the
14037   validity of the nonce in an OCSP response.
14038
14039   *Steve Henson*
14040
14041 * Change function OCSP_request_add() to OCSP_request_add0_id().
14042   This doesn't copy the supplied OCSP_CERTID and avoids the
14043   need to free up the newly created id. Change return type
14044   to OCSP_ONEREQ to return the internal OCSP_ONEREQ structure.
14045   This can then be used to add extensions to the request.
14046   Deleted OCSP_request_new(), since most of its functionality
14047   is now in OCSP_REQUEST_new() (and the case insensitive name
14048   clash) apart from the ability to set the request name which
14049   will be added elsewhere.
14050
14051   *Steve Henson*
14052
14053 * Update OCSP API. Remove obsolete extensions argument from
14054   various functions. Extensions are now handled using the new
14055   OCSP extension code. New simple OCSP HTTP function which
14056   can be used to send requests and parse the response.
14057
14058   *Steve Henson*
14059
14060 * Fix the PKCS#7 (S/MIME) code to work with new ASN1. Two new
14061   ASN1_ITEM structures help with sign and verify. PKCS7_ATTR_SIGN
14062   uses the special reorder version of SET OF to sort the attributes
14063   and reorder them to match the encoded order. This resolves a long
14064   standing problem: a verify on a PKCS7 structure just after signing
14065   it used to fail because the attribute order did not match the
14066   encoded order. PKCS7_ATTR_VERIFY does not reorder the attributes:
14067   it uses the received order. This is necessary to tolerate some broken
14068   software that does not order SET OF. This is handled by encoding
14069   as a SEQUENCE OF but using implicit tagging (with UNIVERSAL class)
14070   to produce the required SET OF.
14071
14072   *Steve Henson*
14073
14074 * Have mk1mf.pl generate the macros OPENSSL_BUILD_SHLIBCRYPTO and
14075   OPENSSL_BUILD_SHLIBSSL and use them appropriately in the header
14076   files to get correct declarations of the ASN.1 item variables.
14077
14078   *Richard Levitte*
14079
14080 * Rewrite of PKCS#12 code to use new ASN1 functionality. Replace many
14081   PKCS#12 macros with real functions. Fix two unrelated ASN1 bugs:
14082   asn1_check_tlen() would sometimes attempt to use 'ctx' when it was
14083   NULL and ASN1_TYPE was not dereferenced properly in asn1_ex_c2i().
14084   New ASN1 macro: DECLARE_ASN1_ITEM() which just declares the relevant
14085   ASN1_ITEM and no wrapper functions.
14086
14087   *Steve Henson*
14088
14089 * New functions or ASN1_item_d2i_fp() and ASN1_item_d2i_bio(). These
14090   replace the old function pointer based I/O routines. Change most of
14091   the `*_d2i_bio()` and `*_d2i_fp()` functions to use these.
14092
14093   *Steve Henson*
14094
14095 * Enhance mkdef.pl to be more accepting about spacing in C preprocessor
14096   lines, recognize more "algorithms" that can be deselected, and make
14097   it complain about algorithm deselection that isn't recognised.
14098
14099   *Richard Levitte*
14100
14101 * New ASN1 functions to handle dup, sign, verify, digest, pack and
14102   unpack operations in terms of ASN1_ITEM. Modify existing wrappers
14103   to use new functions. Add NO_ASN1_OLD which can be set to remove
14104   some old style ASN1 functions: this can be used to determine if old
14105   code will still work when these eventually go away.
14106
14107   *Steve Henson*
14108
14109 * New extension functions for OCSP structures, these follow the
14110   same conventions as certificates and CRLs.
14111
14112   *Steve Henson*
14113
14114 * New function X509V3_add1_i2d(). This automatically encodes and
14115   adds an extension. Its behaviour can be customised with various
14116   flags to append, replace or delete. Various wrappers added for
14117   certificates and CRLs.
14118
14119   *Steve Henson*
14120
14121 * Fix to avoid calling the underlying ASN1 print routine when
14122   an extension cannot be parsed. Correct a typo in the
14123   OCSP_SERVICELOC extension. Tidy up print OCSP format.
14124
14125   *Steve Henson*
14126
14127 * Make mkdef.pl parse some of the ASN1 macros and add appropriate
14128   entries for variables.
14129
14130   *Steve Henson*
14131
14132 * Add functionality to `apps/openssl.c` for detecting locking
14133   problems: As the program is single-threaded, all we have
14134   to do is register a locking callback using an array for
14135   storing which locks are currently held by the program.
14136
14137   *Bodo Moeller*
14138
14139 * Use a lock around the call to CRYPTO_get_ex_new_index() in
14140   SSL_get_ex_data_X509_STORE_idx(), which is used in
14141   ssl_verify_cert_chain() and thus can be called at any time
14142   during TLS/SSL handshakes so that thread-safety is essential.
14143   Unfortunately, the ex_data design is not at all suited
14144   for multi-threaded use, so it probably should be abolished.
14145
14146   *Bodo Moeller*
14147
14148 * Added Broadcom "ubsec" ENGINE to OpenSSL.
14149
14150   *Broadcom, tweaked and integrated by Geoff Thorpe*
14151
14152 * Move common extension printing code to new function
14153   X509V3_print_extensions(). Reorganise OCSP print routines and
14154   implement some needed OCSP ASN1 functions. Add OCSP extensions.
14155
14156   *Steve Henson*
14157
14158 * New function X509_signature_print() to remove duplication in some
14159   print routines.
14160
14161   *Steve Henson*
14162
14163 * Add a special meaning when SET OF and SEQUENCE OF flags are both
14164   set (this was treated exactly the same as SET OF previously). This
14165   is used to reorder the STACK representing the structure to match the
14166   encoding. This will be used to get round a problem where a PKCS7
14167   structure which was signed could not be verified because the STACK
14168   order did not reflect the encoded order.
14169
14170   *Steve Henson*
14171
14172 * Reimplement the OCSP ASN1 module using the new code.
14173
14174   *Steve Henson*
14175
14176 * Update the X509V3 code to permit the use of an ASN1_ITEM structure
14177   for its ASN1 operations. The old style function pointers still exist
14178   for now but they will eventually go away.
14179
14180   *Steve Henson*
14181
14182 * Merge in replacement ASN1 code from the ASN1 branch. This almost
14183   completely replaces the old ASN1 functionality with a table driven
14184   encoder and decoder which interprets an ASN1_ITEM structure describing
14185   the ASN1 module. Compatibility with the existing ASN1 API (i2d,d2i) is
14186   largely maintained. Almost all of the old asn1_mac.h macro based ASN1
14187   has also been converted to the new form.
14188
14189   *Steve Henson*
14190
14191 * Change BN_mod_exp_recp so that negative moduli are tolerated
14192   (the sign is ignored).  Similarly, ignore the sign in BN_MONT_CTX_set
14193   so that BN_mod_exp_mont and BN_mod_exp_mont_word work
14194   for negative moduli.
14195
14196   *Bodo Moeller*
14197
14198 * Fix BN_uadd and BN_usub: Always return non-negative results instead
14199   of not touching the result's sign bit.
14200
14201   *Bodo Moeller*
14202
14203 * BN_div bugfix: If the result is 0, the sign (res->neg) must not be
14204   set.
14205
14206   *Bodo Moeller*
14207
14208 * Changed the LHASH code to use prototypes for callbacks, and created
14209   macros to declare and implement thin (optionally static) functions
14210   that provide type-safety and avoid function pointer casting for the
14211   type-specific callbacks.
14212
14213   *Geoff Thorpe*
14214
14215 * Added Kerberos Cipher Suites to be used with TLS, as written in
14216   RFC 2712.
14217   *Veers Staats <staatsvr@asc.hpc.mil>,
14218   Jeffrey Altman <jaltman@columbia.edu>, via Richard Levitte*
14219
14220 * Reformat the FAQ so the different questions and answers can be divided
14221   in sections depending on the subject.
14222
14223   *Richard Levitte*
14224
14225 * Have the zlib compression code load ZLIB.DLL dynamically under
14226   Windows.
14227
14228   *Richard Levitte*
14229
14230 * New function BN_mod_sqrt for computing square roots modulo a prime
14231   (using the probabilistic Tonelli-Shanks algorithm unless
14232   p == 3 (mod 4)  or  p == 5 (mod 8),  which are cases that can
14233   be handled deterministically).
14234
14235   *Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller*
14236
14237 * Make BN_mod_inverse faster by explicitly handling small quotients
14238   in the Euclid loop. (Speed gain about 20% for small moduli [256 or
14239   512 bits], about 30% for larger ones [1024 or 2048 bits].)
14240
14241   *Bodo Moeller*
14242
14243 * New function BN_kronecker.
14244
14245   *Bodo Moeller*
14246
14247 * Fix BN_gcd so that it works on negative inputs; the result is
14248   positive unless both parameters are zero.
14249   Previously something reasonably close to an infinite loop was
14250   possible because numbers could be growing instead of shrinking
14251   in the implementation of Euclid's algorithm.
14252
14253   *Bodo Moeller*
14254
14255 * Fix BN_is_word() and BN_is_one() macros to take into account the
14256   sign of the number in question.
14257
14258   Fix BN_is_word(a,w) to work correctly for w == 0.
14259
14260   The old BN_is_word(a,w) macro is now called BN_abs_is_word(a,w)
14261   because its test if the absolute value of 'a' equals 'w'.
14262   Note that BN_abs_is_word does *not* handle w == 0 reliably;
14263   it exists mostly for use in the implementations of BN_is_zero(),
14264   BN_is_one(), and BN_is_word().
14265
14266   *Bodo Moeller*
14267
14268 * New function BN_swap.
14269
14270   *Bodo Moeller*
14271
14272 * Use BN_nnmod instead of BN_mod in crypto/bn/bn_exp.c so that
14273   the exponentiation functions are more likely to produce reasonable
14274   results on negative inputs.
14275
14276   *Bodo Moeller*
14277
14278 * Change BN_mod_mul so that the result is always non-negative.
14279   Previously, it could be negative if one of the factors was negative;
14280   I don't think anyone really wanted that behaviour.
14281
14282   *Bodo Moeller*
14283
14284 * Move `BN_mod_...` functions into new file `crypto/bn/bn_mod.c`
14285   (except for exponentiation, which stays in `crypto/bn/bn_exp.c`,
14286   and `BN_mod_mul_reciprocal`, which stays in `crypto/bn/bn_recp.c`)
14287   and add new functions:
14288
14289           BN_nnmod
14290           BN_mod_sqr
14291           BN_mod_add
14292           BN_mod_add_quick
14293           BN_mod_sub
14294           BN_mod_sub_quick
14295           BN_mod_lshift1
14296           BN_mod_lshift1_quick
14297           BN_mod_lshift
14298           BN_mod_lshift_quick
14299
14300   These functions always generate non-negative results.
14301
14302   `BN_nnmod` otherwise is `like BN_mod` (if `BN_mod` computes a remainder `r`
14303   such that `|m| < r < 0`, `BN_nnmod` will output `rem + |m|` instead).
14304
14305   `BN_mod_XXX_quick(r, a, [b,] m)` generates the same result as
14306   `BN_mod_XXX(r, a, [b,] m, ctx)`, but requires that `a` [and  `b`]
14307   be reduced modulo `m`.
14308
14309   *Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller*
14310
14311<!--
14312   The following entry accidentally appeared in the CHANGES file
14313   distributed with OpenSSL 0.9.7.  The modifications described in
14314   it do *not* apply to OpenSSL 0.9.7.
14315
14316 * Remove a few calls to bn_wexpand() in BN_sqr() (the one in there
14317   was actually never needed) and in BN_mul().  The removal in BN_mul()
14318   required a small change in bn_mul_part_recursive() and the addition
14319   of the functions bn_cmp_part_words(), bn_sub_part_words() and
14320   bn_add_part_words(), which do the same thing as bn_cmp_words(),
14321   bn_sub_words() and bn_add_words() except they take arrays with
14322   differing sizes.
14323
14324   *Richard Levitte*
14325-->
14326
14327 * In 'openssl passwd', verify passwords read from the terminal
14328   unless the '-salt' option is used (which usually means that
14329   verification would just waste user's time since the resulting
14330   hash is going to be compared with some given password hash)
14331   or the new '-noverify' option is used.
14332
14333   This is an incompatible change, but it does not affect
14334   non-interactive use of 'openssl passwd' (passwords on the command
14335   line, '-stdin' option, '-in ...' option) and thus should not
14336   cause any problems.
14337
14338   *Bodo Moeller*
14339
14340 * Remove all references to RSAref, since there's no more need for it.
14341
14342   *Richard Levitte*
14343
14344 * Make DSO load along a path given through an environment variable
14345   (SHLIB_PATH) with shl_load().
14346
14347   *Richard Levitte*
14348
14349 * Constify the ENGINE code as a result of BIGNUM constification.
14350   Also constify the RSA code and most things related to it.  In a
14351   few places, most notable in the depth of the ASN.1 code, ugly
14352   casts back to non-const were required (to be solved at a later
14353   time)
14354
14355   *Richard Levitte*
14356
14357 * Make it so the openssl application has all engines loaded by default.
14358
14359   *Richard Levitte*
14360
14361 * Constify the BIGNUM routines a little more.
14362
14363   *Richard Levitte*
14364
14365 * Add the following functions:
14366
14367           ENGINE_load_cswift()
14368           ENGINE_load_chil()
14369           ENGINE_load_atalla()
14370           ENGINE_load_nuron()
14371           ENGINE_load_builtin_engines()
14372
14373   That way, an application can itself choose if external engines that
14374   are built-in in OpenSSL shall ever be used or not.  The benefit is
14375   that applications won't have to be linked with libdl or other dso
14376   libraries unless it's really needed.
14377
14378   Changed 'openssl engine' to load all engines on demand.
14379   Changed the engine header files to avoid the duplication of some
14380   declarations (they differed!).
14381
14382   *Richard Levitte*
14383
14384 * 'openssl engine' can now list capabilities.
14385
14386   *Richard Levitte*
14387
14388 * Better error reporting in 'openssl engine'.
14389
14390   *Richard Levitte*
14391
14392 * Never call load_dh_param(NULL) in s_server.
14393
14394   *Bodo Moeller*
14395
14396 * Add engine application.  It can currently list engines by name and
14397   identity, and test if they are actually available.
14398
14399   *Richard Levitte*
14400
14401 * Improve RPM specification file by forcing symbolic linking and making
14402   sure the installed documentation is also owned by root.root.
14403
14404   *Damien Miller <djm@mindrot.org>*
14405
14406 * Give the OpenSSL applications more possibilities to make use of
14407   keys (public as well as private) handled by engines.
14408
14409   *Richard Levitte*
14410
14411 * Add OCSP code that comes from CertCo.
14412
14413   *Richard Levitte*
14414
14415 * Add VMS support for the Rijndael code.
14416
14417   *Richard Levitte*
14418
14419 * Added untested support for Nuron crypto accelerator.
14420
14421   *Ben Laurie*
14422
14423 * Add support for external cryptographic devices.  This code was
14424   previously distributed separately as the "engine" branch.
14425
14426   *Geoff Thorpe, Richard Levitte*
14427
14428 * Rework the filename-translation in the DSO code. It is now possible to
14429   have far greater control over how a "name" is turned into a filename
14430   depending on the operating environment and any oddities about the
14431   different shared library filenames on each system.
14432
14433   *Geoff Thorpe*
14434
14435 * Support threads on FreeBSD-elf in Configure.
14436
14437   *Richard Levitte*
14438
14439 * Fix for SHA1 assembly problem with MASM: it produces
14440   warnings about corrupt line number information when assembling
14441   with debugging information. This is caused by the overlapping
14442   of two sections.
14443
14444   *Bernd Matthes <mainbug@celocom.de>, Steve Henson*
14445
14446 * NCONF changes.
14447   NCONF_get_number() has no error checking at all.  As a replacement,
14448   NCONF_get_number_e() is defined (`_e` for "error checking") and is
14449   promoted strongly.  The old NCONF_get_number is kept around for
14450   binary backward compatibility.
14451   Make it possible for methods to load from something other than a BIO,
14452   by providing a function pointer that is given a name instead of a BIO.
14453   For example, this could be used to load configuration data from an
14454   LDAP server.
14455
14456   *Richard Levitte*
14457
14458 * Fix for non blocking accept BIOs. Added new I/O special reason
14459   BIO_RR_ACCEPT to cover this case. Previously use of accept BIOs
14460   with non blocking I/O was not possible because no retry code was
14461   implemented. Also added new SSL code SSL_WANT_ACCEPT to cover
14462   this case.
14463
14464   *Steve Henson*
14465
14466 * Added the beginnings of Rijndael support.
14467
14468   *Ben Laurie*
14469
14470 * Fix for bug in DirectoryString mask setting. Add support for
14471   X509_NAME_print_ex() in 'req' and X509_print_ex() function
14472   to allow certificate printing to more controllable, additional
14473   'certopt' option to 'x509' to allow new printing options to be
14474   set.
14475
14476   *Steve Henson*
14477
14478 * Clean old EAY MD5 hack from e_os.h.
14479
14480   *Richard Levitte*
14481
14482### Changes between 0.9.6l and 0.9.6m  [17 Mar 2004]
14483
14484 * Fix null-pointer assignment in do_change_cipher_spec() revealed
14485   by using the Codenomicon TLS Test Tool ([CVE-2004-0079])
14486
14487   *Joe Orton, Steve Henson*
14488
14489### Changes between 0.9.6k and 0.9.6l  [04 Nov 2003]
14490
14491 * Fix additional bug revealed by the NISCC test suite:
14492
14493   Stop bug triggering large recursion when presented with
14494   certain ASN.1 tags ([CVE-2003-0851])
14495
14496   *Steve Henson*
14497
14498### Changes between 0.9.6j and 0.9.6k  [30 Sep 2003]
14499
14500 * Fix various bugs revealed by running the NISCC test suite:
14501
14502   Stop out of bounds reads in the ASN1 code when presented with
14503   invalid tags (CVE-2003-0543 and CVE-2003-0544).
14504
14505   If verify callback ignores invalid public key errors don't try to check
14506   certificate signature with the NULL public key.
14507
14508   *Steve Henson*
14509
14510 * In ssl3_accept() (ssl/s3_srvr.c) only accept a client certificate
14511   if the server requested one: as stated in TLS 1.0 and SSL 3.0
14512   specifications.
14513
14514   *Steve Henson*
14515
14516 * In ssl3_get_client_hello() (ssl/s3_srvr.c), tolerate additional
14517   extra data after the compression methods not only for TLS 1.0
14518   but also for SSL 3.0 (as required by the specification).
14519
14520   *Bodo Moeller; problem pointed out by Matthias Loepfe*
14521
14522 * Change X509_certificate_type() to mark the key as exported/exportable
14523   when it's 512 *bits* long, not 512 bytes.
14524
14525   *Richard Levitte*
14526
14527### Changes between 0.9.6i and 0.9.6j  [10 Apr 2003]
14528
14529 * Countermeasure against the Klima-Pokorny-Rosa extension of
14530   Bleichbacher's attack on PKCS #1 v1.5 padding: treat
14531   a protocol version number mismatch like a decryption error
14532   in ssl3_get_client_key_exchange (ssl/s3_srvr.c).
14533
14534   *Bodo Moeller*
14535
14536 * Turn on RSA blinding by default in the default implementation
14537   to avoid a timing attack. Applications that don't want it can call
14538   RSA_blinding_off() or use the new flag RSA_FLAG_NO_BLINDING.
14539   They would be ill-advised to do so in most cases.
14540
14541   *Ben Laurie, Steve Henson, Geoff Thorpe, Bodo Moeller*
14542
14543 * Change RSA blinding code so that it works when the PRNG is not
14544   seeded (in this case, the secret RSA exponent is abused as
14545   an unpredictable seed -- if it is not unpredictable, there
14546   is no point in blinding anyway).  Make RSA blinding thread-safe
14547   by remembering the creator's thread ID in rsa->blinding and
14548   having all other threads use local one-time blinding factors
14549   (this requires more computation than sharing rsa->blinding, but
14550   avoids excessive locking; and if an RSA object is not shared
14551   between threads, blinding will still be very fast).
14552
14553   *Bodo Moeller*
14554
14555### Changes between 0.9.6h and 0.9.6i  [19 Feb 2003]
14556
14557 * In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked
14558   via timing by performing a MAC computation even if incorrect
14559   block cipher padding has been found.  This is a countermeasure
14560   against active attacks where the attacker has to distinguish
14561   between bad padding and a MAC verification error. ([CVE-2003-0078])
14562
14563   *Bodo Moeller; problem pointed out by Brice Canvel (EPFL),
14564   Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and
14565   Martin Vuagnoux (EPFL, Ilion)*
14566
14567### Changes between 0.9.6g and 0.9.6h  [5 Dec 2002]
14568
14569 * New function OPENSSL_cleanse(), which is used to cleanse a section of
14570   memory from its contents.  This is done with a counter that will
14571   place alternating values in each byte.  This can be used to solve
14572   two issues: 1) the removal of calls to memset() by highly optimizing
14573   compilers, and 2) cleansing with other values than 0, since those can
14574   be read through on certain media, for example a swap space on disk.
14575
14576   *Geoff Thorpe*
14577
14578 * Bugfix: client side session caching did not work with external caching,
14579   because the session->cipher setting was not restored when reloading
14580   from the external cache. This problem was masked, when
14581   SSL_OP_NETSCAPE_REUSE_CIPHER_CHANGE_BUG (part of SSL_OP_ALL) was set.
14582   (Found by Steve Haslam <steve@araqnid.ddts.net>.)
14583
14584   *Lutz Jaenicke*
14585
14586 * Fix client_certificate (ssl/s2_clnt.c): The permissible total
14587   length of the REQUEST-CERTIFICATE message is 18 .. 34, not 17 .. 33.
14588
14589   *Zeev Lieber <zeev-l@yahoo.com>*
14590
14591 * Undo an undocumented change introduced in 0.9.6e which caused
14592   repeated calls to OpenSSL_add_all_ciphers() and
14593   OpenSSL_add_all_digests() to be ignored, even after calling
14594   EVP_cleanup().
14595
14596   *Richard Levitte*
14597
14598 * Change the default configuration reader to deal with last line not
14599   being properly terminated.
14600
14601   *Richard Levitte*
14602
14603 * Change X509_NAME_cmp() so it applies the special rules on handling
14604   DN values that are of type PrintableString, as well as RDNs of type
14605   emailAddress where the value has the type ia5String.
14606
14607   *stefank@valicert.com via Richard Levitte*
14608
14609 * Add a SSL_SESS_CACHE_NO_INTERNAL_STORE flag to take over half
14610   the job SSL_SESS_CACHE_NO_INTERNAL_LOOKUP was inconsistently
14611   doing, define a new flag (SSL_SESS_CACHE_NO_INTERNAL) to be
14612   the bitwise-OR of the two for use by the majority of applications
14613   wanting this behaviour, and update the docs. The documented
14614   behaviour and actual behaviour were inconsistent and had been
14615   changing anyway, so this is more a bug-fix than a behavioural
14616   change.
14617
14618   *Geoff Thorpe, diagnosed by Nadav Har'El*
14619
14620 * Don't impose a 16-byte length minimum on session IDs in ssl/s3_clnt.c
14621   (the SSL 3.0 and TLS 1.0 specifications allow any length up to 32 bytes).
14622
14623   *Bodo Moeller*
14624
14625 * Fix initialization code race conditions in
14626           SSLv23_method(),  SSLv23_client_method(),   SSLv23_server_method(),
14627           SSLv2_method(),   SSLv2_client_method(),    SSLv2_server_method(),
14628           SSLv3_method(),   SSLv3_client_method(),    SSLv3_server_method(),
14629           TLSv1_method(),   TLSv1_client_method(),    TLSv1_server_method(),
14630           ssl2_get_cipher_by_char(),
14631           ssl3_get_cipher_by_char().
14632
14633   *Patrick McCormick <patrick@tellme.com>, Bodo Moeller*
14634
14635 * Reorder cleanup sequence in SSL_CTX_free(): only remove the ex_data after
14636   the cached sessions are flushed, as the remove_cb() might use ex_data
14637   contents. Bug found by Sam Varshavchik <mrsam@courier-mta.com>
14638   (see [openssl.org #212]).
14639
14640   *Geoff Thorpe, Lutz Jaenicke*
14641
14642 * Fix typo in OBJ_txt2obj which incorrectly passed the content
14643   length, instead of the encoding length to d2i_ASN1_OBJECT.
14644
14645   *Steve Henson*
14646
14647### Changes between 0.9.6f and 0.9.6g  [9 Aug 2002]
14648
14649 * [In 0.9.6g-engine release:]
14650   Fix crypto/engine/vendor_defns/cswift.h for WIN32 (use `_stdcall`).
14651
14652   *Lynn Gazis <lgazis@rainbow.com>*
14653
14654### Changes between 0.9.6e and 0.9.6f  [8 Aug 2002]
14655
14656 * Fix ASN1 checks. Check for overflow by comparing with LONG_MAX
14657   and get fix the header length calculation.
14658   *Florian Weimer <Weimer@CERT.Uni-Stuttgart.DE>,
14659   Alon Kantor <alonk@checkpoint.com> (and others), Steve Henson*
14660
14661 * Use proper error handling instead of 'assertions' in buffer
14662   overflow checks added in 0.9.6e.  This prevents DoS (the
14663   assertions could call abort()).
14664
14665   *Arne Ansper <arne@ats.cyber.ee>, Bodo Moeller*
14666
14667### Changes between 0.9.6d and 0.9.6e  [30 Jul 2002]
14668
14669 * Add various sanity checks to asn1_get_length() to reject
14670   the ASN1 length bytes if they exceed sizeof(long), will appear
14671   negative or the content length exceeds the length of the
14672   supplied buffer.
14673
14674   *Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>*
14675
14676 * Fix cipher selection routines: ciphers without encryption had no flags
14677   for the cipher strength set and where therefore not handled correctly
14678   by the selection routines (PR #130).
14679
14680   *Lutz Jaenicke*
14681
14682 * Fix EVP_dsa_sha macro.
14683
14684   *Nils Larsch*
14685
14686 * New option
14687        SSL_OP_DONT_INSERT_EMPTY_FRAGMENTS
14688   for disabling the SSL 3.0/TLS 1.0 CBC vulnerability countermeasure
14689   that was added in OpenSSL 0.9.6d.
14690
14691   As the countermeasure turned out to be incompatible with some
14692   broken SSL implementations, the new option is part of SSL_OP_ALL.
14693   SSL_OP_ALL is usually employed when compatibility with weird SSL
14694   implementations is desired (e.g. '-bugs' option to 's_client' and
14695   's_server'), so the new option is automatically set in many
14696   applications.
14697
14698   *Bodo Moeller*
14699
14700 * Changes in security patch:
14701
14702   Changes marked "(CHATS)" were sponsored by the Defense Advanced
14703   Research Projects Agency (DARPA) and Air Force Research Laboratory,
14704   Air Force Materiel Command, USAF, under agreement number
14705   F30602-01-2-0537.
14706
14707 * Add various sanity checks to asn1_get_length() to reject
14708   the ASN1 length bytes if they exceed sizeof(long), will appear
14709   negative or the content length exceeds the length of the
14710   supplied buffer. ([CVE-2002-0659])
14711
14712   *Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>*
14713
14714 * Assertions for various potential buffer overflows, not known to
14715   happen in practice.
14716
14717   *Ben Laurie (CHATS)*
14718
14719 * Various temporary buffers to hold ASCII versions of integers were
14720   too small for 64 bit platforms. ([CVE-2002-0655])
14721   *Matthew Byng-Maddick <mbm@aldigital.co.uk> and Ben Laurie (CHATS)>*
14722
14723 * Remote buffer overflow in SSL3 protocol - an attacker could
14724   supply an oversized session ID to a client. ([CVE-2002-0656])
14725
14726   *Ben Laurie (CHATS)*
14727
14728 * Remote buffer overflow in SSL2 protocol - an attacker could
14729   supply an oversized client master key. ([CVE-2002-0656])
14730
14731   *Ben Laurie (CHATS)*
14732
14733### Changes between 0.9.6c and 0.9.6d  [9 May 2002]
14734
14735 * Fix crypto/asn1/a_sign.c so that 'parameters' is omitted (not
14736   encoded as NULL) with id-dsa-with-sha1.
14737
14738   *Nils Larsch <nla@trustcenter.de>; problem pointed out by Bodo Moeller*
14739
14740 * Check various `X509_...()` return values in `apps/req.c`.
14741
14742   *Nils Larsch <nla@trustcenter.de>*
14743
14744 * Fix BASE64 decode (EVP_DecodeUpdate) for data with CR/LF ended lines:
14745   an end-of-file condition would erroneously be flagged, when the CRLF
14746   was just at the end of a processed block. The bug was discovered when
14747   processing data through a buffering memory BIO handing the data to a
14748   BASE64-decoding BIO. Bug fund and patch submitted by Pavel Tsekov
14749   <ptsekov@syntrex.com> and Nedelcho Stanev.
14750
14751   *Lutz Jaenicke*
14752
14753 * Implement a countermeasure against a vulnerability recently found
14754   in CBC ciphersuites in SSL 3.0/TLS 1.0: Send an empty fragment
14755   before application data chunks to avoid the use of known IVs
14756   with data potentially chosen by the attacker.
14757
14758   *Bodo Moeller*
14759
14760 * Fix length checks in ssl3_get_client_hello().
14761
14762   *Bodo Moeller*
14763
14764 * TLS/SSL library bugfix: use s->s3->in_read_app_data differently
14765   to prevent ssl3_read_internal() from incorrectly assuming that
14766   ssl3_read_bytes() found application data while handshake
14767   processing was enabled when in fact s->s3->in_read_app_data was
14768   merely automatically cleared during the initial handshake.
14769
14770   *Bodo Moeller; problem pointed out by Arne Ansper <arne@ats.cyber.ee>*
14771
14772 * Fix object definitions for Private and Enterprise: they were not
14773   recognized in their shortname (=lowercase) representation. Extend
14774   obj_dat.pl to issue an error when using undefined keywords instead
14775   of silently ignoring the problem (Svenning Sorensen
14776   <sss@sss.dnsalias.net>).
14777
14778   *Lutz Jaenicke*
14779
14780 * Fix DH_generate_parameters() so that it works for 'non-standard'
14781   generators, i.e. generators other than 2 and 5.  (Previously, the
14782   code did not properly initialise the 'add' and 'rem' values to
14783   BN_generate_prime().)
14784
14785   In the new general case, we do not insist that 'generator' is
14786   actually a primitive root: This requirement is rather pointless;
14787   a generator of the order-q subgroup is just as good, if not
14788   better.
14789
14790   *Bodo Moeller*
14791
14792 * Map new X509 verification errors to alerts. Discovered and submitted by
14793   Tom Wu <tom@arcot.com>.
14794
14795   *Lutz Jaenicke*
14796
14797 * Fix ssl3_pending() (ssl/s3_lib.c) to prevent SSL_pending() from
14798   returning non-zero before the data has been completely received
14799   when using non-blocking I/O.
14800
14801   *Bodo Moeller; problem pointed out by John Hughes*
14802
14803 * Some of the ciphers missed the strength entry (SSL_LOW etc).
14804
14805   *Ben Laurie, Lutz Jaenicke*
14806
14807 * Fix bug in SSL_clear(): bad sessions were not removed (found by
14808   Yoram Zahavi <YoramZ@gilian.com>).
14809
14810   *Lutz Jaenicke*
14811
14812 * Add information about CygWin 1.3 and on, and preserve proper
14813   configuration for the versions before that.
14814
14815   *Corinna Vinschen <vinschen@redhat.com> and Richard Levitte*
14816
14817 * Make removal from session cache (SSL_CTX_remove_session()) more robust:
14818   check whether we deal with a copy of a session and do not delete from
14819   the cache in this case. Problem reported by "Izhar Shoshani Levi"
14820   <izhar@checkpoint.com>.
14821
14822   *Lutz Jaenicke*
14823
14824 * Do not store session data into the internal session cache, if it
14825   is never intended to be looked up (SSL_SESS_CACHE_NO_INTERNAL_LOOKUP
14826   flag is set). Proposed by Aslam <aslam@funk.com>.
14827
14828   *Lutz Jaenicke*
14829
14830 * Have ASN1_BIT_STRING_set_bit() really clear a bit when the requested
14831   value is 0.
14832
14833   *Richard Levitte*
14834
14835 * [In 0.9.6d-engine release:]
14836   Fix a crashbug and a logic bug in hwcrhk_load_pubkey().
14837
14838   *Toomas Kiisk <vix@cyber.ee> via Richard Levitte*
14839
14840 * Add the configuration target linux-s390x.
14841
14842   *Neale Ferguson <Neale.Ferguson@SoftwareAG-USA.com> via Richard Levitte*
14843
14844 * The earlier bugfix for the SSL3_ST_SW_HELLO_REQ_C case of
14845   ssl3_accept (ssl/s3_srvr.c) incorrectly used a local flag
14846   variable as an indication that a ClientHello message has been
14847   received.  As the flag value will be lost between multiple
14848   invocations of ssl3_accept when using non-blocking I/O, the
14849   function may not be aware that a handshake has actually taken
14850   place, thus preventing a new session from being added to the
14851   session cache.
14852
14853   To avoid this problem, we now set s->new_session to 2 instead of
14854   using a local variable.
14855
14856   *Lutz Jaenicke, Bodo Moeller*
14857
14858 * Bugfix: Return -1 from ssl3_get_server_done (ssl3/s3_clnt.c)
14859   if the SSL_R_LENGTH_MISMATCH error is detected.
14860
14861   *Geoff Thorpe, Bodo Moeller*
14862
14863 * New 'shared_ldflag' column in Configure platform table.
14864
14865   *Richard Levitte*
14866
14867 * Fix EVP_CIPHER_mode macro.
14868
14869   *"Dan S. Camper" <dan@bti.net>*
14870
14871 * Fix ssl3_read_bytes (ssl/s3_pkt.c): To ignore messages of unknown
14872   type, we must throw them away by setting rr->length to 0.
14873
14874   *D P Chang <dpc@qualys.com>*
14875
14876### Changes between 0.9.6b and 0.9.6c  [21 dec 2001]
14877
14878 * Fix BN_rand_range bug pointed out by Dominikus Scherkl
14879   <Dominikus.Scherkl@biodata.com>.  (The previous implementation
14880   worked incorrectly for those cases where range = `10..._2`  and
14881   `3*range`  is two bits longer than  range.)
14882
14883   *Bodo Moeller*
14884
14885 * Only add signing time to PKCS7 structures if it is not already
14886   present.
14887
14888   *Steve Henson*
14889
14890 * Fix crypto/objects/objects.h: "ld-ce" should be "id-ce",
14891   OBJ_ld_ce should be OBJ_id_ce.
14892   Also some ip-pda OIDs in crypto/objects/objects.txt were
14893   incorrect (cf. RFC 3039).
14894
14895   *Matt Cooper, Frederic Giudicelli, Bodo Moeller*
14896
14897 * Release CRYPTO_LOCK_DYNLOCK when CRYPTO_destroy_dynlockid()
14898   returns early because it has nothing to do.
14899
14900   *Andy Schneider <andy.schneider@bjss.co.uk>*
14901
14902 * [In 0.9.6c-engine release:]
14903   Fix mutex callback return values in crypto/engine/hw_ncipher.c.
14904
14905   *Andy Schneider <andy.schneider@bjss.co.uk>*
14906
14907 * [In 0.9.6c-engine release:]
14908   Add support for Cryptographic Appliance's keyserver technology.
14909   (Use engine 'keyclient')
14910
14911   *Cryptographic Appliances and Geoff Thorpe*
14912
14913 * Add a configuration entry for OS/390 Unix.  The C compiler 'c89'
14914   is called via tools/c89.sh because arguments have to be
14915   rearranged (all '-L' options must appear before the first object
14916   modules).
14917
14918   *Richard Shapiro <rshapiro@abinitio.com>*
14919
14920 * [In 0.9.6c-engine release:]
14921   Add support for Broadcom crypto accelerator cards, backported
14922   from 0.9.7.
14923
14924   *Broadcom, Nalin Dahyabhai <nalin@redhat.com>, Mark Cox*
14925
14926 * [In 0.9.6c-engine release:]
14927   Add support for SureWare crypto accelerator cards from
14928   Baltimore Technologies.  (Use engine 'sureware')
14929
14930   *Baltimore Technologies and Mark Cox*
14931
14932 * [In 0.9.6c-engine release:]
14933   Add support for crypto accelerator cards from Accelerated
14934   Encryption Processing, www.aep.ie.  (Use engine 'aep')
14935
14936   *AEP Inc. and Mark Cox*
14937
14938 * Add a configuration entry for gcc on UnixWare.
14939
14940   *Gary Benson <gbenson@redhat.com>*
14941
14942 * Change ssl/s2_clnt.c and ssl/s2_srvr.c so that received handshake
14943   messages are stored in a single piece (fixed-length part and
14944   variable-length part combined) and fix various bugs found on the way.
14945
14946   *Bodo Moeller*
14947
14948 * Disable caching in BIO_gethostbyname(), directly use gethostbyname()
14949   instead.  BIO_gethostbyname() does not know what timeouts are
14950   appropriate, so entries would stay in cache even when they have
14951   become invalid.
14952   *Bodo Moeller; problem pointed out by Rich Salz <rsalz@zolera.com>*
14953
14954 * Change ssl23_get_client_hello (ssl/s23_srvr.c) behaviour when
14955   faced with a pathologically small ClientHello fragment that does
14956   not contain client_version: Instead of aborting with an error,
14957   simply choose the highest available protocol version (i.e.,
14958   TLS 1.0 unless it is disabled).  In practice, ClientHello
14959   messages are never sent like this, but this change gives us
14960   strictly correct behaviour at least for TLS.
14961
14962   *Bodo Moeller*
14963
14964 * Fix SSL handshake functions and SSL_clear() such that SSL_clear()
14965   never resets s->method to s->ctx->method when called from within
14966   one of the SSL handshake functions.
14967
14968   *Bodo Moeller; problem pointed out by Niko Baric*
14969
14970 * In ssl3_get_client_hello (ssl/s3_srvr.c), generate a fatal alert
14971   (sent using the client's version number) if client_version is
14972   smaller than the protocol version in use.  Also change
14973   ssl23_get_client_hello (ssl/s23_srvr.c) to select TLS 1.0 if
14974   the client demanded SSL 3.0 but only TLS 1.0 is enabled; then
14975   the client will at least see that alert.
14976
14977   *Bodo Moeller*
14978
14979 * Fix ssl3_get_message (ssl/s3_both.c) to handle message fragmentation
14980   correctly.
14981
14982   *Bodo Moeller*
14983
14984 * Avoid infinite loop in ssl3_get_message (ssl/s3_both.c) if a
14985   client receives HelloRequest while in a handshake.
14986
14987   *Bodo Moeller; bug noticed by Andy Schneider <andy.schneider@bjss.co.uk>*
14988
14989 * Bugfix in ssl3_accept (ssl/s3_srvr.c): Case SSL3_ST_SW_HELLO_REQ_C
14990   should end in 'break', not 'goto end' which circumvents various
14991   cleanups done in state SSL_ST_OK.   But session related stuff
14992   must be disabled for SSL_ST_OK in the case that we just sent a
14993   HelloRequest.
14994
14995   Also avoid some overhead by not calling ssl_init_wbio_buffer()
14996   before just sending a HelloRequest.
14997
14998   *Bodo Moeller, Eric Rescorla <ekr@rtfm.com>*
14999
15000 * Fix ssl/s3_enc.c, ssl/t1_enc.c and ssl/s3_pkt.c so that we don't
15001   reveal whether illegal block cipher padding was found or a MAC
15002   verification error occurred.  (Neither SSLerr() codes nor alerts
15003   are directly visible to potential attackers, but the information
15004   may leak via logfiles.)
15005
15006   Similar changes are not required for the SSL 2.0 implementation
15007   because the number of padding bytes is sent in clear for SSL 2.0,
15008   and the extra bytes are just ignored.  However ssl/s2_pkt.c
15009   failed to verify that the purported number of padding bytes is in
15010   the legal range.
15011
15012   *Bodo Moeller*
15013
15014 * Add OpenUNIX-8 support including shared libraries
15015   (Boyd Lynn Gerber <gerberb@zenez.com>).
15016
15017   *Lutz Jaenicke*
15018
15019 * Improve RSA_padding_check_PKCS1_OAEP() check again to avoid
15020   'wristwatch attack' using huge encoding parameters (cf.
15021   James H. Manger's CRYPTO 2001 paper).  Note that the
15022   RSA_PKCS1_OAEP_PADDING case of RSA_private_decrypt() does not use
15023   encoding parameters and hence was not vulnerable.
15024
15025   *Bodo Moeller*
15026
15027 * BN_sqr() bug fix.
15028
15029   *Ulf Möller, reported by Jim Ellis <jim.ellis@cavium.com>*
15030
15031 * Rabin-Miller test analyses assume uniformly distributed witnesses,
15032   so use BN_pseudo_rand_range() instead of using BN_pseudo_rand()
15033   followed by modular reduction.
15034
15035   *Bodo Moeller; pointed out by Adam Young <AYoung1@NCSUS.JNJ.COM>*
15036
15037 * Add BN_pseudo_rand_range() with obvious functionality: BN_rand_range()
15038   equivalent based on BN_pseudo_rand() instead of BN_rand().
15039
15040   *Bodo Moeller*
15041
15042 * s3_srvr.c: allow sending of large client certificate lists (> 16 kB).
15043   This function was broken, as the check for a new client hello message
15044   to handle SGC did not allow these large messages.
15045   (Tracked down by "Douglas E. Engert" <deengert@anl.gov>.)
15046
15047   *Lutz Jaenicke*
15048
15049 * Add alert descriptions for TLSv1 to `SSL_alert_desc_string[_long]()`.
15050
15051   *Lutz Jaenicke*
15052
15053 * Fix buggy behaviour of BIO_get_num_renegotiates() and BIO_ctrl()
15054   for BIO_C_GET_WRITE_BUF_SIZE ("Stephen Hinton" <shinton@netopia.com>).
15055
15056   *Lutz Jaenicke*
15057
15058 * Rework the configuration and shared library support for Tru64 Unix.
15059   The configuration part makes use of modern compiler features and
15060   still retains old compiler behavior for those that run older versions
15061   of the OS.  The shared library support part includes a variant that
15062   uses the RPATH feature, and is available through the special
15063   configuration target "alpha-cc-rpath", which will never be selected
15064   automatically.
15065
15066   *Tim Mooney <mooney@dogbert.cc.ndsu.NoDak.edu> via Richard Levitte*
15067
15068 * In ssl3_get_key_exchange (ssl/s3_clnt.c), call ssl3_get_message()
15069   with the same message size as in ssl3_get_certificate_request().
15070   Otherwise, if no ServerKeyExchange message occurs, CertificateRequest
15071   messages might inadvertently be reject as too long.
15072
15073   *Petr Lampa <lampa@fee.vutbr.cz>*
15074
15075 * Enhanced support for IA-64 Unix platforms (well, Linux and HP-UX).
15076
15077   *Andy Polyakov*
15078
15079 * Modified SSL library such that the verify_callback that has been set
15080   specifically for an SSL object with SSL_set_verify() is actually being
15081   used. Before the change, a verify_callback set with this function was
15082   ignored and the verify_callback() set in the SSL_CTX at the time of
15083   the call was used. New function X509_STORE_CTX_set_verify_cb() introduced
15084   to allow the necessary settings.
15085
15086   *Lutz Jaenicke*
15087
15088 * Initialize static variable in crypto/dsa/dsa_lib.c and crypto/dh/dh_lib.c
15089   explicitly to NULL, as at least on Solaris 8 this seems not always to be
15090   done automatically (in contradiction to the requirements of the C
15091   standard). This made problems when used from OpenSSH.
15092
15093   *Lutz Jaenicke*
15094
15095 * In OpenSSL 0.9.6a and 0.9.6b, crypto/dh/dh_key.c ignored
15096   dh->length and always used
15097
15098           BN_rand_range(priv_key, dh->p).
15099
15100   BN_rand_range() is not necessary for Diffie-Hellman, and this
15101   specific range makes Diffie-Hellman unnecessarily inefficient if
15102   dh->length (recommended exponent length) is much smaller than the
15103   length of dh->p.  We could use BN_rand_range() if the order of
15104   the subgroup was stored in the DH structure, but we only have
15105   dh->length.
15106
15107   So switch back to
15108
15109           BN_rand(priv_key, l, ...)
15110
15111   where 'l' is dh->length if this is defined, or BN_num_bits(dh->p)-1
15112   otherwise.
15113
15114   *Bodo Moeller*
15115
15116 * In
15117
15118           RSA_eay_public_encrypt
15119           RSA_eay_private_decrypt
15120           RSA_eay_private_encrypt (signing)
15121           RSA_eay_public_decrypt (signature verification)
15122
15123   (default implementations for RSA_public_encrypt,
15124   RSA_private_decrypt, RSA_private_encrypt, RSA_public_decrypt),
15125   always reject numbers >= n.
15126
15127   *Bodo Moeller*
15128
15129 * In crypto/rand/md_rand.c, use a new short-time lock CRYPTO_LOCK_RAND2
15130   to synchronize access to 'locking_thread'.  This is necessary on
15131   systems where access to 'locking_thread' (an 'unsigned long'
15132   variable) is not atomic.
15133
15134   *Bodo Moeller*
15135
15136 * In crypto/rand/md_rand.c, set 'locking_thread' to current thread's ID
15137   *before* setting the 'crypto_lock_rand' flag.  The previous code had
15138   a race condition if 0 is a valid thread ID.
15139
15140   *Travis Vitek <vitek@roguewave.com>*
15141
15142 * Add support for shared libraries under Irix.
15143
15144   *Albert Chin-A-Young <china@thewrittenword.com>*
15145
15146 * Add configuration option to build on Linux on both big-endian and
15147   little-endian MIPS.
15148
15149   *Ralf Baechle <ralf@uni-koblenz.de>*
15150
15151 * Add the possibility to create shared libraries on HP-UX.
15152
15153   *Richard Levitte*
15154
15155### Changes between 0.9.6a and 0.9.6b  [9 Jul 2001]
15156
15157 * Change ssleay_rand_bytes (crypto/rand/md_rand.c)
15158   to avoid a SSLeay/OpenSSL PRNG weakness pointed out by
15159   Markku-Juhani O. Saarinen <markku-juhani.saarinen@nokia.com>:
15160   PRNG state recovery was possible based on the output of
15161   one PRNG request appropriately sized to gain knowledge on
15162   'md' followed by enough consecutive 1-byte PRNG requests
15163   to traverse all of 'state'.
15164
15165   1. When updating 'md_local' (the current thread's copy of 'md')
15166      during PRNG output generation, hash all of the previous
15167      'md_local' value, not just the half used for PRNG output.
15168
15169   2. Make the number of bytes from 'state' included into the hash
15170      independent from the number of PRNG bytes requested.
15171
15172   The first measure alone would be sufficient to avoid
15173   Markku-Juhani's attack.  (Actually it had never occurred
15174   to me that the half of 'md_local' used for chaining was the
15175   half from which PRNG output bytes were taken -- I had always
15176   assumed that the secret half would be used.)  The second
15177   measure makes sure that additional data from 'state' is never
15178   mixed into 'md_local' in small portions; this heuristically
15179   further strengthens the PRNG.
15180
15181   *Bodo Moeller*
15182
15183 * Fix crypto/bn/asm/mips3.s.
15184
15185   *Andy Polyakov*
15186
15187 * When only the key is given to "enc", the IV is undefined. Print out
15188   an error message in this case.
15189
15190   *Lutz Jaenicke*
15191
15192 * Handle special case when X509_NAME is empty in X509 printing routines.
15193
15194   *Steve Henson*
15195
15196 * In dsa_do_verify (crypto/dsa/dsa_ossl.c), verify that r and s are
15197   positive and less than q.
15198
15199   *Bodo Moeller*
15200
15201 * Don't change `*pointer` in CRYPTO_add_lock() is add_lock_callback is
15202   used: it isn't thread safe and the add_lock_callback should handle
15203   that itself.
15204
15205   *Paul Rose <Paul.Rose@bridge.com>*
15206
15207 * Verify that incoming data obeys the block size in
15208   ssl3_enc (ssl/s3_enc.c) and tls1_enc (ssl/t1_enc.c).
15209
15210   *Bodo Moeller*
15211
15212 * Fix OAEP check.
15213
15214   *Ulf Möller, Bodo Möller*
15215
15216 * The countermeasure against Bleichbacher's attack on PKCS #1 v1.5
15217   RSA encryption was accidentally removed in s3_srvr.c in OpenSSL 0.9.5
15218   when fixing the server behaviour for backwards-compatible 'client
15219   hello' messages.  (Note that the attack is impractical against
15220   SSL 3.0 and TLS 1.0 anyway because length and version checking
15221   means that the probability of guessing a valid ciphertext is
15222   around 2^-40; see section 5 in Bleichenbacher's CRYPTO '98
15223   paper.)
15224
15225   Before 0.9.5, the countermeasure (hide the error by generating a
15226   random 'decryption result') did not work properly because
15227   ERR_clear_error() was missing, meaning that SSL_get_error() would
15228   detect the supposedly ignored error.
15229
15230   Both problems are now fixed.
15231
15232   *Bodo Moeller*
15233
15234 * In crypto/bio/bf_buff.c, increase DEFAULT_BUFFER_SIZE to 4096
15235   (previously it was 1024).
15236
15237   *Bodo Moeller*
15238
15239 * Fix for compatibility mode trust settings: ignore trust settings
15240   unless some valid trust or reject settings are present.
15241
15242   *Steve Henson*
15243
15244 * Fix for blowfish EVP: its a variable length cipher.
15245
15246   *Steve Henson*
15247
15248 * Fix various bugs related to DSA S/MIME verification. Handle missing
15249   parameters in DSA public key structures and return an error in the
15250   DSA routines if parameters are absent.
15251
15252   *Steve Henson*
15253
15254 * In versions up to 0.9.6, RAND_file_name() resorted to file ".rnd"
15255   in the current directory if neither $RANDFILE nor $HOME was set.
15256   RAND_file_name() in 0.9.6a returned NULL in this case.  This has
15257   caused some confusion to Windows users who haven't defined $HOME.
15258   Thus RAND_file_name() is changed again: e_os.h can define a
15259   DEFAULT_HOME, which will be used if $HOME is not set.
15260   For Windows, we use "C:"; on other platforms, we still require
15261   environment variables.
15262
15263 * Move 'if (!initialized) RAND_poll()' into regions protected by
15264   CRYPTO_LOCK_RAND.  This is not strictly necessary, but avoids
15265   having multiple threads call RAND_poll() concurrently.
15266
15267   *Bodo Moeller*
15268
15269 * In crypto/rand/md_rand.c, replace 'add_do_not_lock' flag by a
15270   combination of a flag and a thread ID variable.
15271   Otherwise while one thread is in ssleay_rand_bytes (which sets the
15272   flag), *other* threads can enter ssleay_add_bytes without obeying
15273   the CRYPTO_LOCK_RAND lock (and may even illegally release the lock
15274   that they do not hold after the first thread unsets add_do_not_lock).
15275
15276   *Bodo Moeller*
15277
15278 * Change bctest again: '-x' expressions are not available in all
15279   versions of 'test'.
15280
15281   *Bodo Moeller*
15282
15283### Changes between 0.9.6 and 0.9.6a  [5 Apr 2001]
15284
15285 * Fix a couple of memory leaks in PKCS7_dataDecode()
15286
15287   *Steve Henson, reported by Heyun Zheng <hzheng@atdsprint.com>*
15288
15289 * Change Configure and Makefiles to provide EXE_EXT, which will contain
15290   the default extension for executables, if any.  Also, make the perl
15291   scripts that use symlink() to test if it really exists and use "cp"
15292   if it doesn't.  All this made OpenSSL compilable and installable in
15293   CygWin.
15294
15295   *Richard Levitte*
15296
15297 * Fix for asn1_GetSequence() for indefinite length constructed data.
15298   If SEQUENCE is length is indefinite just set c->slen to the total
15299   amount of data available.
15300
15301   *Steve Henson, reported by shige@FreeBSD.org*
15302
15303   *This change does not apply to 0.9.7.*
15304
15305 * Change bctest to avoid here-documents inside command substitution
15306   (workaround for FreeBSD /bin/sh bug).
15307   For compatibility with Ultrix, avoid shell functions (introduced
15308   in the bctest version that searches along $PATH).
15309
15310   *Bodo Moeller*
15311
15312 * Rename 'des_encrypt' to 'des_encrypt1'.  This avoids the clashes
15313   with des_encrypt() defined on some operating systems, like Solaris
15314   and UnixWare.
15315
15316   *Richard Levitte*
15317
15318 * Check the result of RSA-CRT (see D. Boneh, R. DeMillo, R. Lipton:
15319   On the Importance of Eliminating Errors in Cryptographic
15320   Computations, J. Cryptology 14 (2001) 2, 101-119,
15321   <http://theory.stanford.edu/~dabo/papers/faults.ps.gz>).
15322
15323   *Ulf Moeller*
15324
15325 * MIPS assembler BIGNUM division bug fix.
15326
15327   *Andy Polyakov*
15328
15329 * Disabled incorrect Alpha assembler code.
15330
15331   *Richard Levitte*
15332
15333 * Fix PKCS#7 decode routines so they correctly update the length
15334   after reading an EOC for the EXPLICIT tag.
15335
15336   *Steve Henson*
15337
15338   *This change does not apply to 0.9.7.*
15339
15340 * Fix bug in PKCS#12 key generation routines. This was triggered
15341   if a 3DES key was generated with a 0 initial byte. Include
15342   PKCS12_BROKEN_KEYGEN compilation option to retain the old
15343   (but broken) behaviour.
15344
15345   *Steve Henson*
15346
15347 * Enhance bctest to search for a working bc along $PATH and print
15348   it when found.
15349
15350   *Tim Rice <tim@multitalents.net> via Richard Levitte*
15351
15352 * Fix memory leaks in err.c: free err_data string if necessary;
15353   don't write to the wrong index in ERR_set_error_data.
15354
15355   *Bodo Moeller*
15356
15357 * Implement ssl23_peek (analogous to ssl23_read), which previously
15358   did not exist.
15359
15360   *Bodo Moeller*
15361
15362 * Replace rdtsc with `_emit` statements for VC++ version 5.
15363
15364   *Jeremy Cooper <jeremy@baymoo.org>*
15365
15366 * Make it possible to reuse SSLv2 sessions.
15367
15368   *Richard Levitte*
15369
15370 * In copy_email() check for >= 0 as a return value for
15371   X509_NAME_get_index_by_NID() since 0 is a valid index.
15372
15373   *Steve Henson reported by Massimiliano Pala <madwolf@opensca.org>*
15374
15375 * Avoid coredump with unsupported or invalid public keys by checking if
15376   X509_get_pubkey() fails in PKCS7_verify(). Fix memory leak when
15377   PKCS7_verify() fails with non detached data.
15378
15379   *Steve Henson*
15380
15381 * Don't use getenv in library functions when run as setuid/setgid.
15382   New function OPENSSL_issetugid().
15383
15384   *Ulf Moeller*
15385
15386 * Avoid false positives in memory leak detection code (crypto/mem_dbg.c)
15387   due to incorrect handling of multi-threading:
15388
15389   1. Fix timing glitch in the MemCheck_off() portion of CRYPTO_mem_ctrl().
15390
15391   2. Fix logical glitch in is_MemCheck_on() aka CRYPTO_is_mem_check_on().
15392
15393   3. Count how many times MemCheck_off() has been called so that
15394      nested use can be treated correctly.  This also avoids
15395      inband-signalling in the previous code (which relied on the
15396      assumption that thread ID 0 is impossible).
15397
15398   *Bodo Moeller*
15399
15400 * Add "-rand" option also to s_client and s_server.
15401
15402   *Lutz Jaenicke*
15403
15404 * Fix CPU detection on Irix 6.x.
15405   *Kurt Hockenbury <khockenb@stevens-tech.edu> and
15406   "Bruce W. Forsberg" <bruce.forsberg@baesystems.com>*
15407
15408 * Fix X509_NAME bug which produced incorrect encoding if X509_NAME
15409   was empty.
15410
15411   *Steve Henson*
15412
15413   *This change does not apply to 0.9.7.*
15414
15415 * Use the cached encoding of an X509_NAME structure rather than
15416   copying it. This is apparently the reason for the libsafe "errors"
15417   but the code is actually correct.
15418
15419   *Steve Henson*
15420
15421 * Add new function BN_rand_range(), and fix DSA_sign_setup() to prevent
15422   Bleichenbacher's DSA attack.
15423   Extend BN_[pseudo_]rand: As before, top=1 forces the highest two bits
15424   to be set and top=0 forces the highest bit to be set; top=-1 is new
15425   and leaves the highest bit random.
15426
15427   *Ulf Moeller, Bodo Moeller*
15428
15429 * In the `NCONF_...`-based implementations for `CONF_...` queries
15430   (crypto/conf/conf_lib.c), if the input LHASH is NULL, avoid using
15431   a temporary CONF structure with the data component set to NULL
15432   (which gives segmentation faults in lh_retrieve).
15433   Instead, use NULL for the CONF pointer in CONF_get_string and
15434   CONF_get_number (which may use environment variables) and directly
15435   return NULL from CONF_get_section.
15436
15437   *Bodo Moeller*
15438
15439 * Fix potential buffer overrun for EBCDIC.
15440
15441   *Ulf Moeller*
15442
15443 * Tolerate nonRepudiation as being valid for S/MIME signing and certSign
15444   keyUsage if basicConstraints absent for a CA.
15445
15446   *Steve Henson*
15447
15448 * Make SMIME_write_PKCS7() write mail header values with a format that
15449   is more generally accepted (no spaces before the semicolon), since
15450   some programs can't parse those values properly otherwise.  Also make
15451   sure BIO's that break lines after each write do not create invalid
15452   headers.
15453
15454   *Richard Levitte*
15455
15456 * Make the CRL encoding routines work with empty SEQUENCE OF. The
15457   macros previously used would not encode an empty SEQUENCE OF
15458   and break the signature.
15459
15460   *Steve Henson*
15461
15462   *This change does not apply to 0.9.7.*
15463
15464 * Zero the premaster secret after deriving the master secret in
15465   DH ciphersuites.
15466
15467   *Steve Henson*
15468
15469 * Add some EVP_add_digest_alias registrations (as found in
15470   OpenSSL_add_all_digests()) to SSL_library_init()
15471   aka OpenSSL_add_ssl_algorithms().  This provides improved
15472   compatibility with peers using X.509 certificates
15473   with unconventional AlgorithmIdentifier OIDs.
15474
15475   *Bodo Moeller*
15476
15477 * Fix for Irix with NO_ASM.
15478
15479   *"Bruce W. Forsberg" <bruce.forsberg@baesystems.com>*
15480
15481 * ./config script fixes.
15482
15483   *Ulf Moeller, Richard Levitte*
15484
15485 * Fix 'openssl passwd -1'.
15486
15487   *Bodo Moeller*
15488
15489 * Change PKCS12_key_gen_asc() so it can cope with non null
15490   terminated strings whose length is passed in the passlen
15491   parameter, for example from PEM callbacks. This was done
15492   by adding an extra length parameter to asc2uni().
15493
15494   *Steve Henson, reported by <oddissey@samsung.co.kr>*
15495
15496 * Fix C code generated by 'openssl dsaparam -C': If a BN_bin2bn
15497   call failed, free the DSA structure.
15498
15499   *Bodo Moeller*
15500
15501 * Fix to uni2asc() to cope with zero length Unicode strings.
15502   These are present in some PKCS#12 files.
15503
15504   *Steve Henson*
15505
15506 * Increase s2->wbuf allocation by one byte in ssl2_new (ssl/s2_lib.c).
15507   Otherwise do_ssl_write (ssl/s2_pkt.c) will write beyond buffer limits
15508   when writing a 32767 byte record.
15509
15510   *Bodo Moeller; problem reported by Eric Day <eday@concentric.net>*
15511
15512 * In `RSA_eay_public_{en,ed}crypt` and RSA_eay_mod_exp (rsa_eay.c),
15513   obtain lock CRYPTO_LOCK_RSA before setting `rsa->_method_mod_{n,p,q}`.
15514
15515   (RSA objects have a reference count access to which is protected
15516   by CRYPTO_LOCK_RSA [see rsa_lib.c, s3_srvr.c, ssl_cert.c, ssl_rsa.c],
15517   so they are meant to be shared between threads.)
15518   *Bodo Moeller, Geoff Thorpe; original patch submitted by
15519   "Reddie, Steven" <Steven.Reddie@ca.com>*
15520
15521 * Fix a deadlock in CRYPTO_mem_leaks().
15522
15523   *Bodo Moeller*
15524
15525 * Use better test patterns in bntest.
15526
15527   *Ulf Möller*
15528
15529 * rand_win.c fix for Borland C.
15530
15531   *Ulf Möller*
15532
15533 * BN_rshift bugfix for n == 0.
15534
15535   *Bodo Moeller*
15536
15537 * Add a 'bctest' script that checks for some known 'bc' bugs
15538   so that 'make test' does not abort just because 'bc' is broken.
15539
15540   *Bodo Moeller*
15541
15542 * Store verify_result within SSL_SESSION also for client side to
15543   avoid potential security hole. (Re-used sessions on the client side
15544   always resulted in verify_result==X509_V_OK, not using the original
15545   result of the server certificate verification.)
15546
15547   *Lutz Jaenicke*
15548
15549 * Fix ssl3_pending: If the record in s->s3->rrec is not of type
15550   SSL3_RT_APPLICATION_DATA, return 0.
15551   Similarly, change ssl2_pending to return 0 if SSL_in_init(s) is true.
15552
15553   *Bodo Moeller*
15554
15555 * Fix SSL_peek:
15556   Both ssl2_peek and ssl3_peek, which were totally broken in earlier
15557   releases, have been re-implemented by renaming the previous
15558   implementations of ssl2_read and ssl3_read to ssl2_read_internal
15559   and ssl3_read_internal, respectively, and adding 'peek' parameters
15560   to them.  The new ssl[23]_{read,peek} functions are calls to
15561   ssl[23]_read_internal with the 'peek' flag set appropriately.
15562   A 'peek' parameter has also been added to ssl3_read_bytes, which
15563   does the actual work for ssl3_read_internal.
15564
15565   *Bodo Moeller*
15566
15567 * Initialise "ex_data" member of RSA/DSA/DH structures prior to calling
15568   the method-specific "init()" handler. Also clean up ex_data after
15569   calling the method-specific "finish()" handler. Previously, this was
15570   happening the other way round.
15571
15572   *Geoff Thorpe*
15573
15574 * Increase BN_CTX_NUM (the number of BIGNUMs in a BN_CTX) to 16.
15575   The previous value, 12, was not always sufficient for BN_mod_exp().
15576
15577   *Bodo Moeller*
15578
15579 * Make sure that shared libraries get the internal name engine with
15580   the full version number and not just 0.  This should mark the
15581   shared libraries as not backward compatible.  Of course, this should
15582   be changed again when we can guarantee backward binary compatibility.
15583
15584   *Richard Levitte*
15585
15586 * Fix typo in get_cert_by_subject() in by_dir.c
15587
15588   *Jean-Marc Desperrier <jean-marc.desperrier@certplus.com>*
15589
15590 * Rework the system to generate shared libraries:
15591
15592   - Make note of the expected extension for the shared libraries and
15593     if there is a need for symbolic links from for example libcrypto.so.0
15594     to libcrypto.so.0.9.7.  There is extended info in Configure for
15595     that.
15596
15597   - Make as few rebuilds of the shared libraries as possible.
15598
15599   - Still avoid linking the OpenSSL programs with the shared libraries.
15600
15601   - When installing, install the shared libraries separately from the
15602     static ones.
15603
15604   *Richard Levitte*
15605
15606 * Fix SSL_CTX_set_read_ahead macro to actually use its argument.
15607
15608   Copy SSL_CTX's read_ahead flag to SSL object directly in SSL_new
15609   and not in SSL_clear because the latter is also used by the
15610   accept/connect functions; previously, the settings made by
15611   SSL_set_read_ahead would be lost during the handshake.
15612
15613   *Bodo Moeller; problems reported by Anders Gertz <gertz@epact.se>*
15614
15615 * Correct util/mkdef.pl to be selective about disabled algorithms.
15616   Previously, it would create entries for disabled algorithms no
15617   matter what.
15618
15619   *Richard Levitte*
15620
15621 * Added several new manual pages for SSL_* function.
15622
15623   *Lutz Jaenicke*
15624
15625### Changes between 0.9.5a and 0.9.6  [24 Sep 2000]
15626
15627 * In ssl23_get_client_hello, generate an error message when faced
15628   with an initial SSL 3.0/TLS record that is too small to contain the
15629   first two bytes of the ClientHello message, i.e. client_version.
15630   (Note that this is a pathologic case that probably has never happened
15631   in real life.)  The previous approach was to use the version number
15632   from the record header as a substitute; but our protocol choice
15633   should not depend on that one because it is not authenticated
15634   by the Finished messages.
15635
15636   *Bodo Moeller*
15637
15638 * More robust randomness gathering functions for Windows.
15639
15640   *Jeffrey Altman <jaltman@columbia.edu>*
15641
15642 * For compatibility reasons if the flag X509_V_FLAG_ISSUER_CHECK is
15643   not set then we don't setup the error code for issuer check errors
15644   to avoid possibly overwriting other errors which the callback does
15645   handle. If an application does set the flag then we assume it knows
15646   what it is doing and can handle the new informational codes
15647   appropriately.
15648
15649   *Steve Henson*
15650
15651 * Fix for a nasty bug in ASN1_TYPE handling. ASN1_TYPE is used for
15652   a general "ANY" type, as such it should be able to decode anything
15653   including tagged types. However it didn't check the class so it would
15654   wrongly interpret tagged types in the same way as their universal
15655   counterpart and unknown types were just rejected. Changed so that the
15656   tagged and unknown types are handled in the same way as a SEQUENCE:
15657   that is the encoding is stored intact. There is also a new type
15658   "V_ASN1_OTHER" which is used when the class is not universal, in this
15659   case we have no idea what the actual type is so we just lump them all
15660   together.
15661
15662   *Steve Henson*
15663
15664 * On VMS, stdout may very well lead to a file that is written to
15665   in a record-oriented fashion.  That means that every write() will
15666   write a separate record, which will be read separately by the
15667   programs trying to read from it.  This can be very confusing.
15668
15669   The solution is to put a BIO filter in the way that will buffer
15670   text until a linefeed is reached, and then write everything a
15671   line at a time, so every record written will be an actual line,
15672   not chunks of lines and not (usually doesn't happen, but I've
15673   seen it once) several lines in one record.  BIO_f_linebuffer() is
15674   the answer.
15675
15676   Currently, it's a VMS-only method, because that's where it has
15677   been tested well enough.
15678
15679   *Richard Levitte*
15680
15681 * Remove 'optimized' squaring variant in BN_mod_mul_montgomery,
15682   it can return incorrect results.
15683   (Note: The buggy variant was not enabled in OpenSSL 0.9.5a,
15684   but it was in 0.9.6-beta[12].)
15685
15686   *Bodo Moeller*
15687
15688 * Disable the check for content being present when verifying detached
15689   signatures in pk7_smime.c. Some versions of Netscape (wrongly)
15690   include zero length content when signing messages.
15691
15692   *Steve Henson*
15693
15694 * New BIO_shutdown_wr macro, which invokes the BIO_C_SHUTDOWN_WR
15695   BIO_ctrl (for BIO pairs).
15696
15697   *Bodo Möller*
15698
15699 * Add DSO method for VMS.
15700
15701   *Richard Levitte*
15702
15703 * Bug fix: Montgomery multiplication could produce results with the
15704   wrong sign.
15705
15706   *Ulf Möller*
15707
15708 * Add RPM specification openssl.spec and modify it to build three
15709   packages.  The default package contains applications, application
15710   documentation and run-time libraries.  The devel package contains
15711   include files, static libraries and function documentation.  The
15712   doc package contains the contents of the doc directory.  The original
15713   openssl.spec was provided by Damien Miller <djm@mindrot.org>.
15714
15715   *Richard Levitte*
15716
15717 * Add a large number of documentation files for many SSL routines.
15718
15719   *Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>*
15720
15721 * Add a configuration entry for Sony News 4.
15722
15723   *NAKAJI Hiroyuki <nakaji@tutrp.tut.ac.jp>*
15724
15725 * Don't set the two most significant bits to one when generating a
15726   random number < q in the DSA library.
15727
15728   *Ulf Möller*
15729
15730 * New SSL API mode 'SSL_MODE_AUTO_RETRY'.  This disables the default
15731   behaviour that SSL_read may result in SSL_ERROR_WANT_READ (even if
15732   the underlying transport is blocking) if a handshake took place.
15733   (The default behaviour is needed by applications such as s_client
15734   and s_server that use select() to determine when to use SSL_read;
15735   but for applications that know in advance when to expect data, it
15736   just makes things more complicated.)
15737
15738   *Bodo Moeller*
15739
15740 * Add RAND_egd_bytes(), which gives control over the number of bytes read
15741   from EGD.
15742
15743   *Ben Laurie*
15744
15745 * Add a few more EBCDIC conditionals that make `req` and `x509`
15746   work better on such systems.
15747
15748   *Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>*
15749
15750 * Add two demo programs for PKCS12_parse() and PKCS12_create().
15751   Update PKCS12_parse() so it copies the friendlyName and the
15752   keyid to the certificates aux info.
15753
15754   *Steve Henson*
15755
15756 * Fix bug in PKCS7_verify() which caused an infinite loop
15757   if there was more than one signature.
15758
15759   *Sven Uszpelkat <su@celocom.de>*
15760
15761 * Major change in util/mkdef.pl to include extra information
15762   about each symbol, as well as presenting variables as well
15763   as functions.  This change means that there's n more need
15764   to rebuild the .num files when some algorithms are excluded.
15765
15766   *Richard Levitte*
15767
15768 * Allow the verify time to be set by an application,
15769   rather than always using the current time.
15770
15771   *Steve Henson*
15772
15773 * Phase 2 verify code reorganisation. The certificate
15774   verify code now looks up an issuer certificate by a
15775   number of criteria: subject name, authority key id
15776   and key usage. It also verifies self signed certificates
15777   by the same criteria. The main comparison function is
15778   X509_check_issued() which performs these checks.
15779
15780   Lot of changes were necessary in order to support this
15781   without completely rewriting the lookup code.
15782
15783   Authority and subject key identifier are now cached.
15784
15785   The LHASH 'certs' is X509_STORE has now been replaced
15786   by a STACK_OF(X509_OBJECT). This is mainly because an
15787   LHASH can't store or retrieve multiple objects with
15788   the same hash value.
15789
15790   As a result various functions (which were all internal
15791   use only) have changed to handle the new X509_STORE
15792   structure. This will break anything that messed round
15793   with X509_STORE internally.
15794
15795   The functions X509_STORE_add_cert() now checks for an
15796   exact match, rather than just subject name.
15797
15798   The X509_STORE API doesn't directly support the retrieval
15799   of multiple certificates matching a given criteria, however
15800   this can be worked round by performing a lookup first
15801   (which will fill the cache with candidate certificates)
15802   and then examining the cache for matches. This is probably
15803   the best we can do without throwing out X509_LOOKUP
15804   entirely (maybe later...).
15805
15806   The X509_VERIFY_CTX structure has been enhanced considerably.
15807
15808   All certificate lookup operations now go via a get_issuer()
15809   callback. Although this currently uses an X509_STORE it
15810   can be replaced by custom lookups. This is a simple way
15811   to bypass the X509_STORE hackery necessary to make this
15812   work and makes it possible to use more efficient techniques
15813   in future. A very simple version which uses a simple
15814   STACK for its trusted certificate store is also provided
15815   using X509_STORE_CTX_trusted_stack().
15816
15817   The verify_cb() and verify() callbacks now have equivalents
15818   in the X509_STORE_CTX structure.
15819
15820   X509_STORE_CTX also has a 'flags' field which can be used
15821   to customise the verify behaviour.
15822
15823   *Steve Henson*
15824
15825 * Add new PKCS#7 signing option PKCS7_NOSMIMECAP which
15826   excludes S/MIME capabilities.
15827
15828   *Steve Henson*
15829
15830 * When a certificate request is read in keep a copy of the
15831   original encoding of the signed data and use it when outputting
15832   again. Signatures then use the original encoding rather than
15833   a decoded, encoded version which may cause problems if the
15834   request is improperly encoded.
15835
15836   *Steve Henson*
15837
15838 * For consistency with other BIO_puts implementations, call
15839   buffer_write(b, ...) directly in buffer_puts instead of calling
15840   BIO_write(b, ...).
15841
15842   In BIO_puts, increment b->num_write as in BIO_write.
15843
15844   *Peter.Sylvester@EdelWeb.fr*
15845
15846 * Fix BN_mul_word for the case where the word is 0. (We have to use
15847   BN_zero, we may not return a BIGNUM with an array consisting of
15848   words set to zero.)
15849
15850   *Bodo Moeller*
15851
15852 * Avoid calling abort() from within the library when problems are
15853   detected, except if preprocessor symbols have been defined
15854   (such as REF_CHECK, BN_DEBUG etc.).
15855
15856   *Bodo Moeller*
15857
15858 * New openssl application 'rsautl'. This utility can be
15859   used for low-level RSA operations. DER public key
15860   BIO/fp routines also added.
15861
15862   *Steve Henson*
15863
15864 * New Configure entry and patches for compiling on QNX 4.
15865
15866   *Andreas Schneider <andreas@ds3.etech.fh-hamburg.de>*
15867
15868 * A demo state-machine implementation was sponsored by
15869   Nuron (<http://www.nuron.com/>) and is now available in
15870   demos/state_machine.
15871
15872   *Ben Laurie*
15873
15874 * New options added to the 'dgst' utility for signature
15875   generation and verification.
15876
15877   *Steve Henson*
15878
15879 * Unrecognized PKCS#7 content types are now handled via a
15880   catch all ASN1_TYPE structure. This allows unsupported
15881   types to be stored as a "blob" and an application can
15882   encode and decode it manually.
15883
15884   *Steve Henson*
15885
15886 * Fix various signed/unsigned issues to make a_strex.c
15887   compile under VC++.
15888
15889   *Oscar Jacobsson <oscar.jacobsson@celocom.com>*
15890
15891 * ASN1 fixes. i2d_ASN1_OBJECT was not returning the correct
15892   length if passed a buffer. ASN1_INTEGER_to_BN failed
15893   if passed a NULL BN and its argument was negative.
15894
15895   *Steve Henson, pointed out by Sven Heiberg <sven@tartu.cyber.ee>*
15896
15897 * Modification to PKCS#7 encoding routines to output definite
15898   length encoding. Since currently the whole structures are in
15899   memory there's not real point in using indefinite length
15900   constructed encoding. However if OpenSSL is compiled with
15901   the flag PKCS7_INDEFINITE_ENCODING the old form is used.
15902
15903   *Steve Henson*
15904
15905 * Added BIO_vprintf() and BIO_vsnprintf().
15906
15907   *Richard Levitte*
15908
15909 * Added more prefixes to parse for in the strings written
15910   through a logging bio, to cover all the levels that are available
15911   through syslog.  The prefixes are now:
15912
15913           PANIC, EMERG, EMR       =>      LOG_EMERG
15914           ALERT, ALR              =>      LOG_ALERT
15915           CRIT, CRI               =>      LOG_CRIT
15916           ERROR, ERR              =>      LOG_ERR
15917           WARNING, WARN, WAR      =>      LOG_WARNING
15918           NOTICE, NOTE, NOT       =>      LOG_NOTICE
15919           INFO, INF               =>      LOG_INFO
15920           DEBUG, DBG              =>      LOG_DEBUG
15921
15922   and as before, if none of those prefixes are present at the
15923   beginning of the string, LOG_ERR is chosen.
15924
15925   On Win32, the `LOG_*` levels are mapped according to this:
15926
15927           LOG_EMERG, LOG_ALERT, LOG_CRIT, LOG_ERR => EVENTLOG_ERROR_TYPE
15928           LOG_WARNING                             => EVENTLOG_WARNING_TYPE
15929           LOG_NOTICE, LOG_INFO, LOG_DEBUG         => EVENTLOG_INFORMATION_TYPE
15930
15931   *Richard Levitte*
15932
15933 * Made it possible to reconfigure with just the configuration
15934   argument "reconf" or "reconfigure".  The command line arguments
15935   are stored in Makefile.ssl in the variable CONFIGURE_ARGS,
15936   and are retrieved from there when reconfiguring.
15937
15938   *Richard Levitte*
15939
15940 * MD4 implemented.
15941
15942   *Assar Westerlund <assar@sics.se>, Richard Levitte*
15943
15944 * Add the arguments -CAfile and -CApath to the pkcs12 utility.
15945
15946   *Richard Levitte*
15947
15948 * The obj_dat.pl script was messing up the sorting of object
15949   names. The reason was that it compared the quoted version
15950   of strings as a result "OCSP" > "OCSP Signing" because
15951   " > SPACE. Changed script to store unquoted versions of
15952   names and add quotes on output. It was also omitting some
15953   names from the lookup table if they were given a default
15954   value (that is if SN is missing it is given the same
15955   value as LN and vice versa), these are now added on the
15956   grounds that if an object has a name we should be able to
15957   look it up. Finally added warning output when duplicate
15958   short or long names are found.
15959
15960   *Steve Henson*
15961
15962 * Changes needed for Tandem NSK.
15963
15964   *Scott Uroff <scott@xypro.com>*
15965
15966 * Fix SSL 2.0 rollback checking: Due to an off-by-one error in
15967   RSA_padding_check_SSLv23(), special padding was never detected
15968   and thus the SSL 3.0/TLS 1.0 countermeasure against protocol
15969   version rollback attacks was not effective.
15970
15971   In s23_clnt.c, don't use special rollback-attack detection padding
15972   (RSA_SSLV23_PADDING) if SSL 2.0 is the only protocol enabled in the
15973   client; similarly, in s23_srvr.c, don't do the rollback check if
15974   SSL 2.0 is the only protocol enabled in the server.
15975
15976   *Bodo Moeller*
15977
15978 * Make it possible to get hexdumps of unprintable data with 'openssl
15979   asn1parse'.  By implication, the functions ASN1_parse_dump() and
15980   BIO_dump_indent() are added.
15981
15982   *Richard Levitte*
15983
15984 * New functions ASN1_STRING_print_ex() and X509_NAME_print_ex()
15985   these print out strings and name structures based on various
15986   flags including RFC2253 support and proper handling of
15987   multibyte characters. Added options to the 'x509' utility
15988   to allow the various flags to be set.
15989
15990   *Steve Henson*
15991
15992 * Various fixes to use ASN1_TIME instead of ASN1_UTCTIME.
15993   Also change the functions X509_cmp_current_time() and
15994   X509_gmtime_adj() work with an ASN1_TIME structure,
15995   this will enable certificates using GeneralizedTime in validity
15996   dates to be checked.
15997
15998   *Steve Henson*
15999
16000 * Make the NEG_PUBKEY_BUG code (which tolerates invalid
16001   negative public key encodings) on by default,
16002   NO_NEG_PUBKEY_BUG can be set to disable it.
16003
16004   *Steve Henson*
16005
16006 * New function c2i_ASN1_OBJECT() which acts on ASN1_OBJECT
16007   content octets. An i2c_ASN1_OBJECT is unnecessary because
16008   the encoding can be trivially obtained from the structure.
16009
16010   *Steve Henson*
16011
16012 * crypto/err.c locking bugfix: Use write locks (`CRYPTO_w_[un]lock`),
16013   not read locks (`CRYPTO_r_[un]lock`).
16014
16015   *Bodo Moeller*
16016
16017 * A first attempt at creating official support for shared
16018   libraries through configuration.  I've kept it so the
16019   default is static libraries only, and the OpenSSL programs
16020   are always statically linked for now, but there are
16021   preparations for dynamic linking in place.
16022   This has been tested on Linux and Tru64.
16023
16024   *Richard Levitte*
16025
16026 * Randomness polling function for Win9x, as described in:
16027   Peter Gutmann, Software Generation of Practically Strong
16028   Random Numbers.
16029
16030   *Ulf Möller*
16031
16032 * Fix so PRNG is seeded in req if using an already existing
16033   DSA key.
16034
16035   *Steve Henson*
16036
16037 * New options to smime application. -inform and -outform
16038   allow alternative formats for the S/MIME message including
16039   PEM and DER. The -content option allows the content to be
16040   specified separately. This should allow things like Netscape
16041   form signing output easier to verify.
16042
16043   *Steve Henson*
16044
16045 * Fix the ASN1 encoding of tags using the 'long form'.
16046
16047   *Steve Henson*
16048
16049 * New ASN1 functions, `i2c_*` and `c2i_*` for INTEGER and BIT
16050   STRING types. These convert content octets to and from the
16051   underlying type. The actual tag and length octets are
16052   already assumed to have been read in and checked. These
16053   are needed because all other string types have virtually
16054   identical handling apart from the tag. By having versions
16055   of the ASN1 functions that just operate on content octets
16056   IMPLICIT tagging can be handled properly. It also allows
16057   the ASN1_ENUMERATED code to be cut down because ASN1_ENUMERATED
16058   and ASN1_INTEGER are identical apart from the tag.
16059
16060   *Steve Henson*
16061
16062 * Change the handling of OID objects as follows:
16063
16064   - New object identifiers are inserted in objects.txt, following
16065     the syntax given in [crypto/objects/README.md](crypto/objects/README.md).
16066   - objects.pl is used to process obj_mac.num and create a new
16067     obj_mac.h.
16068   - obj_dat.pl is used to create a new obj_dat.h, using the data in
16069     obj_mac.h.
16070
16071   This is currently kind of a hack, and the perl code in objects.pl
16072   isn't very elegant, but it works as I intended.  The simplest way
16073   to check that it worked correctly is to look in obj_dat.h and
16074   check the array nid_objs and make sure the objects haven't moved
16075   around (this is important!).  Additions are OK, as well as
16076   consistent name changes.
16077
16078   *Richard Levitte*
16079
16080 * Add BSD-style MD5-based passwords to 'openssl passwd' (option '-1').
16081
16082   *Bodo Moeller*
16083
16084 * Addition of the command line parameter '-rand file' to 'openssl req'.
16085   The given file adds to whatever has already been seeded into the
16086   random pool through the RANDFILE configuration file option or
16087   environment variable, or the default random state file.
16088
16089   *Richard Levitte*
16090
16091 * mkstack.pl now sorts each macro group into lexical order.
16092   Previously the output order depended on the order the files
16093   appeared in the directory, resulting in needless rewriting
16094   of safestack.h .
16095
16096   *Steve Henson*
16097
16098 * Patches to make OpenSSL compile under Win32 again. Mostly
16099   work arounds for the VC++ problem that it treats func() as
16100   func(void). Also stripped out the parts of mkdef.pl that
16101   added extra typesafe functions: these no longer exist.
16102
16103   *Steve Henson*
16104
16105 * Reorganisation of the stack code. The macros are now all
16106   collected in safestack.h . Each macro is defined in terms of
16107   a "stack macro" of the form `SKM_<name>(type, a, b)`. The
16108   DEBUG_SAFESTACK is now handled in terms of function casts,
16109   this has the advantage of retaining type safety without the
16110   use of additional functions. If DEBUG_SAFESTACK is not defined
16111   then the non typesafe macros are used instead. Also modified the
16112   mkstack.pl script to handle the new form. Needs testing to see
16113   if which (if any) compilers it chokes and maybe make DEBUG_SAFESTACK
16114   the default if no major problems. Similar behaviour for ASN1_SET_OF
16115   and PKCS12_STACK_OF.
16116
16117   *Steve Henson*
16118
16119 * When some versions of IIS use the 'NET' form of private key the
16120   key derivation algorithm is different. Normally MD5(password) is
16121   used as a 128 bit RC4 key. In the modified case
16122   MD5(MD5(password) + "SGCKEYSALT")  is used instead. Added some
16123   new functions i2d_RSA_NET(), d2i_RSA_NET() etc which are the same
16124   as the old Netscape_RSA functions except they have an additional
16125   'sgckey' parameter which uses the modified algorithm. Also added
16126   an -sgckey command line option to the rsa utility. Thanks to
16127   Adrian Peck <bertie@ncipher.com> for posting details of the modified
16128   algorithm to openssl-dev.
16129
16130   *Steve Henson*
16131
16132 * The evp_local.h macros were using 'c.##kname' which resulted in
16133   invalid expansion on some systems (SCO 5.0.5 for example).
16134   Corrected to 'c.kname'.
16135
16136   *Phillip Porch <root@theporch.com>*
16137
16138 * New X509_get1_email() and X509_REQ_get1_email() functions that return
16139   a STACK of email addresses from a certificate or request, these look
16140   in the subject name and the subject alternative name extensions and
16141   omit any duplicate addresses.
16142
16143   *Steve Henson*
16144
16145 * Re-implement BN_mod_exp2_mont using independent (and larger) windows.
16146   This makes DSA verification about 2 % faster.
16147
16148   *Bodo Moeller*
16149
16150 * Increase maximum window size in `BN_mod_exp_...` to 6 bits instead of 5
16151   (meaning that now 2^5 values will be precomputed, which is only 4 KB
16152   plus overhead for 1024 bit moduli).
16153   This makes exponentiations about 0.5 % faster for 1024 bit
16154   exponents (as measured by "openssl speed rsa2048").
16155
16156   *Bodo Moeller*
16157
16158 * Rename memory handling macros to avoid conflicts with other
16159   software:
16160           Malloc         =>  OPENSSL_malloc
16161           Malloc_locked  =>  OPENSSL_malloc_locked
16162           Realloc        =>  OPENSSL_realloc
16163           Free           =>  OPENSSL_free
16164
16165   *Richard Levitte*
16166
16167 * New function BN_mod_exp_mont_word for small bases (roughly 15%
16168   faster than BN_mod_exp_mont, i.e. 7% for a full DH exchange).
16169
16170   *Bodo Moeller*
16171
16172 * CygWin32 support.
16173
16174   *John Jarvie <jjarvie@newsguy.com>*
16175
16176 * The type-safe stack code has been rejigged. It is now only compiled
16177   in when OpenSSL is configured with the DEBUG_SAFESTACK option and
16178   by default all type-specific stack functions are "#define"d back to
16179   standard stack functions. This results in more streamlined output
16180   but retains the type-safety checking possibilities of the original
16181   approach.
16182
16183   *Geoff Thorpe*
16184
16185 * The STACK code has been cleaned up, and certain type declarations
16186   that didn't make a lot of sense have been brought in line. This has
16187   also involved a cleanup of sorts in safestack.h to more correctly
16188   map type-safe stack functions onto their plain stack counterparts.
16189   This work has also resulted in a variety of "const"ifications of
16190   lots of the code, especially `_cmp` operations which should normally
16191   be prototyped with "const" parameters anyway.
16192
16193   *Geoff Thorpe*
16194
16195 * When generating bytes for the first time in md_rand.c, 'stir the pool'
16196   by seeding with STATE_SIZE dummy bytes (with zero entropy count).
16197   (The PRNG state consists of two parts, the large pool 'state' and 'md',
16198   where all of 'md' is used each time the PRNG is used, but 'state'
16199   is used only indexed by a cyclic counter. As entropy may not be
16200   well distributed from the beginning, 'md' is important as a
16201   chaining variable. However, the output function chains only half
16202   of 'md', i.e. 80 bits.  ssleay_rand_add, on the other hand, chains
16203   all of 'md', and seeding with STATE_SIZE dummy bytes will result
16204   in all of 'state' being rewritten, with the new values depending
16205   on virtually all of 'md'.  This overcomes the 80 bit limitation.)
16206
16207   *Bodo Moeller*
16208
16209 * In ssl/s2_clnt.c and ssl/s3_clnt.c, call ERR_clear_error() when
16210   the handshake is continued after ssl_verify_cert_chain();
16211   otherwise, if SSL_VERIFY_NONE is set, remaining error codes
16212   can lead to 'unexplainable' connection aborts later.
16213
16214   *Bodo Moeller; problem tracked down by Lutz Jaenicke*
16215
16216 * Major EVP API cipher revision.
16217   Add hooks for extra EVP features. This allows various cipher
16218   parameters to be set in the EVP interface. Support added for variable
16219   key length ciphers via the EVP_CIPHER_CTX_set_key_length() function and
16220   setting of RC2 and RC5 parameters.
16221
16222   Modify EVP_OpenInit() and EVP_SealInit() to cope with variable key length
16223   ciphers.
16224
16225   Remove lots of duplicated code from the EVP library. For example *every*
16226   cipher init() function handles the 'iv' in the same way according to the
16227   cipher mode. They also all do nothing if the 'key' parameter is NULL and
16228   for CFB and OFB modes they zero ctx->num.
16229
16230   New functionality allows removal of S/MIME code RC2 hack.
16231
16232   Most of the routines have the same form and so can be declared in terms
16233   of macros.
16234
16235   By shifting this to the top level EVP_CipherInit() it can be removed from
16236   all individual ciphers. If the cipher wants to handle IVs or keys
16237   differently it can set the EVP_CIPH_CUSTOM_IV or EVP_CIPH_ALWAYS_CALL_INIT
16238   flags.
16239
16240   Change lots of functions like EVP_EncryptUpdate() to now return a
16241   value: although software versions of the algorithms cannot fail
16242   any installed hardware versions can.
16243
16244   *Steve Henson*
16245
16246 * Implement SSL_OP_TLS_ROLLBACK_BUG: In ssl3_get_client_key_exchange, if
16247   this option is set, tolerate broken clients that send the negotiated
16248   protocol version number instead of the requested protocol version
16249   number.
16250
16251   *Bodo Moeller*
16252
16253 * Call dh_tmp_cb (set by `..._TMP_DH_CB`) with correct 'is_export' flag;
16254   i.e. non-zero for export ciphersuites, zero otherwise.
16255   Previous versions had this flag inverted, inconsistent with
16256   rsa_tmp_cb (..._TMP_RSA_CB).
16257
16258   *Bodo Moeller; problem reported by Amit Chopra*
16259
16260 * Add missing DSA library text string. Work around for some IIS
16261   key files with invalid SEQUENCE encoding.
16262
16263   *Steve Henson*
16264
16265 * Add a document (doc/standards.txt) that list all kinds of standards
16266   and so on that are implemented in OpenSSL.
16267
16268   *Richard Levitte*
16269
16270 * Enhance c_rehash script. Old version would mishandle certificates
16271   with the same subject name hash and wouldn't handle CRLs at all.
16272   Added -fingerprint option to crl utility, to support new c_rehash
16273   features.
16274
16275   *Steve Henson*
16276
16277 * Eliminate non-ANSI declarations in crypto.h and stack.h.
16278
16279   *Ulf Möller*
16280
16281 * Fix for SSL server purpose checking. Server checking was
16282   rejecting certificates which had extended key usage present
16283   but no ssl client purpose.
16284
16285   *Steve Henson, reported by Rene Grosser <grosser@hisolutions.com>*
16286
16287 * Make PKCS#12 code work with no password. The PKCS#12 spec
16288   is a little unclear about how a blank password is handled.
16289   Since the password in encoded as a BMPString with terminating
16290   double NULL a zero length password would end up as just the
16291   double NULL. However no password at all is different and is
16292   handled differently in the PKCS#12 key generation code. NS
16293   treats a blank password as zero length. MSIE treats it as no
16294   password on export: but it will try both on import. We now do
16295   the same: PKCS12_parse() tries zero length and no password if
16296   the password is set to "" or NULL (NULL is now a valid password:
16297   it wasn't before) as does the pkcs12 application.
16298
16299   *Steve Henson*
16300
16301 * Bugfixes in `apps/x509.c`: Avoid a memory leak; and don't use
16302   perror when PEM_read_bio_X509_REQ fails, the error message must
16303   be obtained from the error queue.
16304
16305   *Bodo Moeller*
16306
16307 * Avoid 'thread_hash' memory leak in crypto/err/err.c by freeing
16308   it in ERR_remove_state if appropriate, and change ERR_get_state
16309   accordingly to avoid race conditions (this is necessary because
16310   thread_hash is no longer constant once set).
16311
16312   *Bodo Moeller*
16313
16314 * Bugfix for linux-elf makefile.one.
16315
16316   *Ulf Möller*
16317
16318 * RSA_get_default_method() will now cause a default
16319   RSA_METHOD to be chosen if one doesn't exist already.
16320   Previously this was only set during a call to RSA_new()
16321   or RSA_new_method(NULL) meaning it was possible for
16322   RSA_get_default_method() to return NULL.
16323
16324   *Geoff Thorpe*
16325
16326 * Added native name translation to the existing DSO code
16327   that will convert (if the flag to do so is set) filenames
16328   that are sufficiently small and have no path information
16329   into a canonical native form. Eg. "blah" converted to
16330   "libblah.so" or "blah.dll" etc.
16331
16332   *Geoff Thorpe*
16333
16334 * New function ERR_error_string_n(e, buf, len) which is like
16335   ERR_error_string(e, buf), but writes at most 'len' bytes
16336   including the 0 terminator.  For ERR_error_string_n, 'buf'
16337   may not be NULL.
16338
16339   *Damien Miller <djm@mindrot.org>, Bodo Moeller*
16340
16341 * CONF library reworked to become more general.  A new CONF
16342   configuration file reader "class" is implemented as well as a
16343   new functions (`NCONF_*`, for "New CONF") to handle it.  The now
16344   old `CONF_*` functions are still there, but are reimplemented to
16345   work in terms of the new functions.  Also, a set of functions
16346   to handle the internal storage of the configuration data is
16347   provided to make it easier to write new configuration file
16348   reader "classes" (I can definitely see something reading a
16349   configuration file in XML format, for example), called `_CONF_*`,
16350   or "the configuration storage API"...
16351
16352   The new configuration file reading functions are:
16353
16354           NCONF_new, NCONF_free, NCONF_load, NCONF_load_fp, NCONF_load_bio,
16355           NCONF_get_section, NCONF_get_string, NCONF_get_numbre
16356
16357           NCONF_default, NCONF_WIN32
16358
16359           NCONF_dump_fp, NCONF_dump_bio
16360
16361   NCONF_default and NCONF_WIN32 are method (or "class") choosers,
16362   NCONF_new creates a new CONF object.  This works in the same way
16363   as other interfaces in OpenSSL, like the BIO interface.
16364   `NCONF_dump_*` dump the internal storage of the configuration file,
16365   which is useful for debugging.  All other functions take the same
16366   arguments as the old `CONF_*` functions with the exception of the
16367   first that must be a `CONF *` instead of a `LHASH *`.
16368
16369   To make it easier to use the new classes with the old `CONF_*` functions,
16370   the function CONF_set_default_method is provided.
16371
16372   *Richard Levitte*
16373
16374 * Add '-tls1' option to 'openssl ciphers', which was already
16375   mentioned in the documentation but had not been implemented.
16376   (This option is not yet really useful because even the additional
16377   experimental TLS 1.0 ciphers are currently treated as SSL 3.0 ciphers.)
16378
16379   *Bodo Moeller*
16380
16381 * Initial DSO code added into libcrypto for letting OpenSSL (and
16382   OpenSSL-based applications) load shared libraries and bind to
16383   them in a portable way.
16384
16385   *Geoff Thorpe, with contributions from Richard Levitte*
16386
16387### Changes between 0.9.5 and 0.9.5a  [1 Apr 2000]
16388
16389 * Make sure _lrotl and _lrotr are only used with MSVC.
16390
16391 * Use lock CRYPTO_LOCK_RAND correctly in ssleay_rand_status
16392   (the default implementation of RAND_status).
16393
16394 * Rename openssl x509 option '-crlext', which was added in 0.9.5,
16395   to '-clrext' (= clear extensions), as intended and documented.
16396   *Bodo Moeller; inconsistency pointed out by Michael Attili
16397   <attili@amaxo.com>*
16398
16399 * Fix for HMAC. It wasn't zeroing the rest of the block if the key length
16400   was larger than the MD block size.
16401
16402   *Steve Henson, pointed out by Yost William <YostW@tce.com>*
16403
16404 * Modernise PKCS12_parse() so it uses STACK_OF(X509) for its ca argument
16405   fix a leak when the ca argument was passed as NULL. Stop X509_PUBKEY_set()
16406   using the passed key: if the passed key was a private key the result
16407   of X509_print(), for example, would be to print out all the private key
16408   components.
16409
16410   *Steve Henson*
16411
16412 * des_quad_cksum() byte order bug fix.
16413   *Ulf Möller, using the problem description in krb4-0.9.7, where
16414   the solution is attributed to Derrick J Brashear <shadow@DEMENTIA.ORG>*
16415
16416 * Fix so V_ASN1_APP_CHOOSE works again: however its use is strongly
16417   discouraged.
16418
16419   *Steve Henson, pointed out by Brian Korver <briank@cs.stanford.edu>*
16420
16421 * For easily testing in shell scripts whether some command
16422   'openssl XXX' exists, the new pseudo-command 'openssl no-XXX'
16423   returns with exit code 0 iff no command of the given name is available.
16424   'no-XXX' is printed in this case, 'XXX' otherwise.  In both cases,
16425   the output goes to stdout and nothing is printed to stderr.
16426   Additional arguments are always ignored.
16427
16428   Since for each cipher there is a command of the same name,
16429   the 'no-cipher' compilation switches can be tested this way.
16430
16431   ('openssl no-XXX' is not able to detect pseudo-commands such
16432   as 'quit', 'list-XXX-commands', or 'no-XXX' itself.)
16433
16434   *Bodo Moeller*
16435
16436 * Update test suite so that 'make test' succeeds in 'no-rsa' configuration.
16437
16438   *Bodo Moeller*
16439
16440 * For SSL_[CTX_]set_tmp_dh, don't create a DH key if SSL_OP_SINGLE_DH_USE
16441   is set; it will be thrown away anyway because each handshake creates
16442   its own key.
16443   ssl_cert_dup, which is used by SSL_new, now copies DH keys in addition
16444   to parameters -- in previous versions (since OpenSSL 0.9.3) the
16445   'default key' from SSL_CTX_set_tmp_dh would always be lost, meaning
16446   you effectively got SSL_OP_SINGLE_DH_USE when using this macro.
16447
16448   *Bodo Moeller*
16449
16450 * New s_client option -ign_eof: EOF at stdin is ignored, and
16451   'Q' and 'R' lose their special meanings (quit/renegotiate).
16452   This is part of what -quiet does; unlike -quiet, -ign_eof
16453   does not suppress any output.
16454
16455   *Richard Levitte*
16456
16457 * Add compatibility options to the purpose and trust code. The
16458   purpose X509_PURPOSE_ANY is "any purpose" which automatically
16459   accepts a certificate or CA, this was the previous behaviour,
16460   with all the associated security issues.
16461
16462   X509_TRUST_COMPAT is the old trust behaviour: only and
16463   automatically trust self signed roots in certificate store. A
16464   new trust setting X509_TRUST_DEFAULT is used to specify that
16465   a purpose has no associated trust setting and it should instead
16466   use the value in the default purpose.
16467
16468   *Steve Henson*
16469
16470 * Fix the PKCS#8 DSA private key code so it decodes keys again
16471   and fix a memory leak.
16472
16473   *Steve Henson*
16474
16475 * In util/mkerr.pl (which implements 'make errors'), preserve
16476   reason strings from the previous version of the .c file, as
16477   the default to have only downcase letters (and digits) in
16478   automatically generated reasons codes is not always appropriate.
16479
16480   *Bodo Moeller*
16481
16482 * In ERR_load_ERR_strings(), build an ERR_LIB_SYS error reason table
16483   using strerror.  Previously, ERR_reason_error_string() returned
16484   library names as reason strings for SYSerr; but SYSerr is a special
16485   case where small numbers are errno values, not library numbers.
16486
16487   *Bodo Moeller*
16488
16489 * Add '-dsaparam' option to 'openssl dhparam' application.  This
16490   converts DSA parameters into DH parameters. (When creating parameters,
16491   DSA_generate_parameters is used.)
16492
16493   *Bodo Moeller*
16494
16495 * Include 'length' (recommended exponent length) in C code generated
16496   by 'openssl dhparam -C'.
16497
16498   *Bodo Moeller*
16499
16500 * The second argument to set_label in perlasm was already being used
16501   so couldn't be used as a "file scope" flag. Moved to third argument
16502   which was free.
16503
16504   *Steve Henson*
16505
16506 * In PEM_ASN1_write_bio and some other functions, use RAND_pseudo_bytes
16507   instead of RAND_bytes for encryption IVs and salts.
16508
16509   *Bodo Moeller*
16510
16511 * Include RAND_status() into RAND_METHOD instead of implementing
16512   it only for md_rand.c  Otherwise replacing the PRNG by calling
16513   RAND_set_rand_method would be impossible.
16514
16515   *Bodo Moeller*
16516
16517 * Don't let DSA_generate_key() enter an infinite loop if the random
16518   number generation fails.
16519
16520   *Bodo Moeller*
16521
16522 * New 'rand' application for creating pseudo-random output.
16523
16524   *Bodo Moeller*
16525
16526 * Added configuration support for Linux/IA64
16527
16528   *Rolf Haberrecker <rolf@suse.de>*
16529
16530 * Assembler module support for Mingw32.
16531
16532   *Ulf Möller*
16533
16534 * Shared library support for HPUX (in shlib/).
16535
16536   *Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE> and Anonymous*
16537
16538 * Shared library support for Solaris gcc.
16539
16540   *Lutz Behnke <behnke@trustcenter.de>*
16541
16542### Changes between 0.9.4 and 0.9.5  [28 Feb 2000]
16543
16544 * PKCS7_encrypt() was adding text MIME headers twice because they
16545   were added manually and by SMIME_crlf_copy().
16546
16547   *Steve Henson*
16548
16549 * In bntest.c don't call BN_rand with zero bits argument.
16550
16551   *Steve Henson, pointed out by Andrew W. Gray <agray@iconsinc.com>*
16552
16553 * BN_mul bugfix: In bn_mul_part_recursion() only the a>a[n] && b>b[n]
16554   case was implemented. This caused BN_div_recp() to fail occasionally.
16555
16556   *Ulf Möller*
16557
16558 * Add an optional second argument to the set_label() in the perl
16559   assembly language builder. If this argument exists and is set
16560   to 1 it signals that the assembler should use a symbol whose
16561   scope is the entire file, not just the current function. This
16562   is needed with MASM which uses the format label:: for this scope.
16563
16564   *Steve Henson, pointed out by Peter Runestig <peter@runestig.com>*
16565
16566 * Change the ASN1 types so they are typedefs by default. Before
16567   almost all types were #define'd to ASN1_STRING which was causing
16568   STACK_OF() problems: you couldn't declare STACK_OF(ASN1_UTF8STRING)
16569   for example.
16570
16571   *Steve Henson*
16572
16573 * Change names of new functions to the new get1/get0 naming
16574   convention: After 'get1', the caller owns a reference count
16575   and has to call `..._free`; 'get0' returns a pointer to some
16576   data structure without incrementing reference counters.
16577   (Some of the existing 'get' functions increment a reference
16578   counter, some don't.)
16579   Similarly, 'set1' and 'add1' functions increase reference
16580   counters or duplicate objects.
16581
16582   *Steve Henson*
16583
16584 * Allow for the possibility of temp RSA key generation failure:
16585   the code used to assume it always worked and crashed on failure.
16586
16587   *Steve Henson*
16588
16589 * Fix potential buffer overrun problem in BIO_printf().
16590   *Ulf Möller, using public domain code by Patrick Powell; problem
16591   pointed out by David Sacerdote <das33@cornell.edu>*
16592
16593 * Support EGD <http://www.lothar.com/tech/crypto/>.  New functions
16594   RAND_egd() and RAND_status().  In the command line application,
16595   the EGD socket can be specified like a seed file using RANDFILE
16596   or -rand.
16597
16598   *Ulf Möller*
16599
16600 * Allow the string CERTIFICATE to be tolerated in PKCS#7 structures.
16601   Some CAs (e.g. Verisign) distribute certificates in this form.
16602
16603   *Steve Henson*
16604
16605 * Remove the SSL_ALLOW_ADH compile option and set the default cipher
16606   list to exclude them. This means that no special compilation option
16607   is needed to use anonymous DH: it just needs to be included in the
16608   cipher list.
16609
16610   *Steve Henson*
16611
16612 * Change the EVP_MD_CTX_type macro so its meaning consistent with
16613   EVP_MD_type. The old functionality is available in a new macro called
16614   EVP_MD_md(). Change code that uses it and update docs.
16615
16616   *Steve Henson*
16617
16618 * `..._ctrl` functions now have corresponding `..._callback_ctrl` functions
16619   where the `void *` argument is replaced by a function pointer argument.
16620   Previously `void *` was abused to point to functions, which works on
16621   many platforms, but is not correct.  As these functions are usually
16622   called by macros defined in OpenSSL header files, most source code
16623   should work without changes.
16624
16625   *Richard Levitte*
16626
16627 * `<openssl/opensslconf.h>` (which is created by Configure) now contains
16628   sections with information on -D... compiler switches used for
16629   compiling the library so that applications can see them.  To enable
16630   one of these sections, a pre-processor symbol `OPENSSL_..._DEFINES`
16631   must be defined.  E.g.,
16632           #define OPENSSL_ALGORITHM_DEFINES
16633           #include <openssl/opensslconf.h>
16634   defines all pertinent `NO_<algo>` symbols, such as NO_IDEA, NO_RSA, etc.
16635
16636   *Richard Levitte, Ulf and Bodo Möller*
16637
16638 * Bugfix: Tolerate fragmentation and interleaving in the SSL 3/TLS
16639   record layer.
16640
16641   *Bodo Moeller*
16642
16643 * Change the 'other' type in certificate aux info to a STACK_OF
16644   X509_ALGOR. Although not an AlgorithmIdentifier as such it has
16645   the required ASN1 format: arbitrary types determined by an OID.
16646
16647   *Steve Henson*
16648
16649 * Add some PEM_write_X509_REQ_NEW() functions and a command line
16650   argument to 'req'. This is not because the function is newer or
16651   better than others it just uses the work 'NEW' in the certificate
16652   request header lines. Some software needs this.
16653
16654   *Steve Henson*
16655
16656 * Reorganise password command line arguments: now passwords can be
16657   obtained from various sources. Delete the PEM_cb function and make
16658   it the default behaviour: i.e. if the callback is NULL and the
16659   usrdata argument is not NULL interpret it as a null terminated pass
16660   phrase. If usrdata and the callback are NULL then the pass phrase
16661   is prompted for as usual.
16662
16663   *Steve Henson*
16664
16665 * Add support for the Compaq Atalla crypto accelerator. If it is installed,
16666   the support is automatically enabled. The resulting binaries will
16667   autodetect the card and use it if present.
16668
16669   *Ben Laurie and Compaq Inc.*
16670
16671 * Work around for Netscape hang bug. This sends certificate request
16672   and server done in one record. Since this is perfectly legal in the
16673   SSL/TLS protocol it isn't a "bug" option and is on by default. See
16674   the bugs/SSLv3 entry for more info.
16675
16676   *Steve Henson*
16677
16678 * HP-UX tune-up: new unified configs, HP C compiler bug workaround.
16679
16680   *Andy Polyakov*
16681
16682 * Add -rand argument to smime and pkcs12 applications and read/write
16683   of seed file.
16684
16685   *Steve Henson*
16686
16687 * New 'passwd' tool for crypt(3) and apr1 password hashes.
16688
16689   *Bodo Moeller*
16690
16691 * Add command line password options to the remaining applications.
16692
16693   *Steve Henson*
16694
16695 * Bug fix for BN_div_recp() for numerators with an even number of
16696   bits.
16697
16698   *Ulf Möller*
16699
16700 * More tests in bntest.c, and changed test_bn output.
16701
16702   *Ulf Möller*
16703
16704 * ./config recognizes MacOS X now.
16705
16706   *Andy Polyakov*
16707
16708 * Bug fix for BN_div() when the first words of num and divisor are
16709   equal (it gave wrong results if `(rem=(n1-q*d0)&BN_MASK2) < d0)`.
16710
16711   *Ulf Möller*
16712
16713 * Add support for various broken PKCS#8 formats, and command line
16714   options to produce them.
16715
16716   *Steve Henson*
16717
16718 * New functions BN_CTX_start(), BN_CTX_get() and BT_CTX_end() to
16719   get temporary BIGNUMs from a BN_CTX.
16720
16721   *Ulf Möller*
16722
16723 * Correct return values in BN_mod_exp_mont() and BN_mod_exp2_mont()
16724   for p == 0.
16725
16726   *Ulf Möller*
16727
16728 * Change the `SSLeay_add_all_*()` functions to `OpenSSL_add_all_*()` and
16729   include a #define from the old name to the new. The original intent
16730   was that statically linked binaries could for example just call
16731   SSLeay_add_all_ciphers() to just add ciphers to the table and not
16732   link with digests. This never worked because SSLeay_add_all_digests()
16733   and SSLeay_add_all_ciphers() were in the same source file so calling
16734   one would link with the other. They are now in separate source files.
16735
16736   *Steve Henson*
16737
16738 * Add a new -notext option to 'ca' and a -pubkey option to 'spkac'.
16739
16740   *Steve Henson*
16741
16742 * Use a less unusual form of the Miller-Rabin primality test (it used
16743   a binary algorithm for exponentiation integrated into the Miller-Rabin
16744   loop, our standard modexp algorithms are faster).
16745
16746   *Bodo Moeller*
16747
16748 * Support for the EBCDIC character set completed.
16749
16750   *Martin Kraemer <Martin.Kraemer@Mch.SNI.De>*
16751
16752 * Source code cleanups: use const where appropriate, eliminate casts,
16753   use `void *` instead of `char *` in lhash.
16754
16755   *Ulf Möller*
16756
16757 * Bugfix: ssl3_send_server_key_exchange was not restartable
16758   (the state was not changed to SSL3_ST_SW_KEY_EXCH_B, and because of
16759   this the server could overwrite ephemeral keys that the client
16760   has already seen).
16761
16762   *Bodo Moeller*
16763
16764 * Turn DSA_is_prime into a macro that calls BN_is_prime,
16765   using 50 iterations of the Rabin-Miller test.
16766
16767   DSA_generate_parameters now uses BN_is_prime_fasttest (with 50
16768   iterations of the Rabin-Miller test as required by the appendix
16769   to FIPS PUB 186[-1]) instead of DSA_is_prime.
16770   As BN_is_prime_fasttest includes trial division, DSA parameter
16771   generation becomes much faster.
16772
16773   This implies a change for the callback functions in DSA_is_prime
16774   and DSA_generate_parameters: The callback function is called once
16775   for each positive witness in the Rabin-Miller test, not just
16776   occasionally in the inner loop; and the parameters to the
16777   callback function now provide an iteration count for the outer
16778   loop rather than for the current invocation of the inner loop.
16779   DSA_generate_parameters additionally can call the callback
16780   function with an 'iteration count' of -1, meaning that a
16781   candidate has passed the trial division test (when q is generated
16782   from an application-provided seed, trial division is skipped).
16783
16784   *Bodo Moeller*
16785
16786 * New function BN_is_prime_fasttest that optionally does trial
16787   division before starting the Rabin-Miller test and has
16788   an additional BN_CTX * argument (whereas BN_is_prime always
16789   has to allocate at least one BN_CTX).
16790   'callback(1, -1, cb_arg)' is called when a number has passed the
16791   trial division stage.
16792
16793   *Bodo Moeller*
16794
16795 * Fix for bug in CRL encoding. The validity dates weren't being handled
16796   as ASN1_TIME.
16797
16798   *Steve Henson*
16799
16800 * New -pkcs12 option to CA.pl script to write out a PKCS#12 file.
16801
16802   *Steve Henson*
16803
16804 * New function BN_pseudo_rand().
16805
16806   *Ulf Möller*
16807
16808 * Clean up BN_mod_mul_montgomery(): replace the broken (and unreadable)
16809   bignum version of BN_from_montgomery() with the working code from
16810   SSLeay 0.9.0 (the word based version is faster anyway), and clean up
16811   the comments.
16812
16813   *Ulf Möller*
16814
16815 * Avoid a race condition in s2_clnt.c (function get_server_hello) that
16816   made it impossible to use the same SSL_SESSION data structure in
16817   SSL2 clients in multiple threads.
16818
16819   *Bodo Moeller*
16820
16821 * The return value of RAND_load_file() no longer counts bytes obtained
16822   by stat().  RAND_load_file(..., -1) is new and uses the complete file
16823   to seed the PRNG (previously an explicit byte count was required).
16824
16825   *Ulf Möller, Bodo Möller*
16826
16827 * Clean up CRYPTO_EX_DATA functions, some of these didn't have prototypes
16828   used `char *` instead of `void *` and had casts all over the place.
16829
16830   *Steve Henson*
16831
16832 * Make BN_generate_prime() return NULL on error if ret!=NULL.
16833
16834   *Ulf Möller*
16835
16836 * Retain source code compatibility for BN_prime_checks macro:
16837   BN_is_prime(..., BN_prime_checks, ...) now uses
16838   BN_prime_checks_for_size to determine the appropriate number of
16839   Rabin-Miller iterations.
16840
16841   *Ulf Möller*
16842
16843 * Diffie-Hellman uses "safe" primes: DH_check() return code renamed to
16844   DH_CHECK_P_NOT_SAFE_PRIME.
16845   (Check if this is true? OpenPGP calls them "strong".)
16846
16847   *Ulf Möller*
16848
16849 * Merge the functionality of "dh" and "gendh" programs into a new program
16850   "dhparam". The old programs are retained for now but will handle DH keys
16851   (instead of parameters) in future.
16852
16853   *Steve Henson*
16854
16855 * Make the ciphers, s_server and s_client programs check the return values
16856   when a new cipher list is set.
16857
16858   *Steve Henson*
16859
16860 * Enhance the SSL/TLS cipher mechanism to correctly handle the TLS 56bit
16861   ciphers. Before when the 56bit ciphers were enabled the sorting was
16862   wrong.
16863
16864   The syntax for the cipher sorting has been extended to support sorting by
16865   cipher-strength (using the strength_bits hard coded in the tables).
16866   The new command is `@STRENGTH` (see also `doc/apps/ciphers.pod`).
16867
16868   Fix a bug in the cipher-command parser: when supplying a cipher command
16869   string with an "undefined" symbol (neither command nor alphanumeric
16870   *A-Za-z0-9*, ssl_set_cipher_list used to hang in an endless loop. Now
16871   an error is flagged.
16872
16873   Due to the strength-sorting extension, the code of the
16874   ssl_create_cipher_list() function was completely rearranged. I hope that
16875   the readability was also increased :-)
16876
16877   *Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>*
16878
16879 * Minor change to 'x509' utility. The -CAcreateserial option now uses 1
16880   for the first serial number and places 2 in the serial number file. This
16881   avoids problems when the root CA is created with serial number zero and
16882   the first user certificate has the same issuer name and serial number
16883   as the root CA.
16884
16885   *Steve Henson*
16886
16887 * Fixes to X509_ATTRIBUTE utilities, change the 'req' program so it uses
16888   the new code. Add documentation for this stuff.
16889
16890   *Steve Henson*
16891
16892 * Changes to X509_ATTRIBUTE utilities. These have been renamed from
16893   `X509_*()` to `X509at_*()` on the grounds that they don't handle X509
16894   structures and behave in an analogous way to the X509v3 functions:
16895   they shouldn't be called directly but wrapper functions should be used
16896   instead.
16897
16898   So we also now have some wrapper functions that call the X509at functions
16899   when passed certificate requests. (TO DO: similar things can be done with
16900   PKCS#7 signed and unsigned attributes, PKCS#12 attributes and a few other
16901   things. Some of these need some d2i or i2d and print functionality
16902   because they handle more complex structures.)
16903
16904   *Steve Henson*
16905
16906 * Add missing #ifndefs that caused missing symbols when building libssl
16907   as a shared library without RSA.  Use #ifndef NO_SSL2 instead of
16908   NO_RSA in `ssl/s2*.c`.
16909
16910   *Kris Kennaway <kris@hub.freebsd.org>, modified by Ulf Möller*
16911
16912 * Precautions against using the PRNG uninitialized: RAND_bytes() now
16913   has a return value which indicates the quality of the random data
16914   (1 = ok, 0 = not seeded).  Also an error is recorded on the thread's
16915   error queue. New function RAND_pseudo_bytes() generates output that is
16916   guaranteed to be unique but not unpredictable. RAND_add is like
16917   RAND_seed, but takes an extra argument for an entropy estimate
16918   (RAND_seed always assumes full entropy).
16919
16920   *Ulf Möller*
16921
16922 * Do more iterations of Rabin-Miller probable prime test (specifically,
16923   3 for 1024-bit primes, 6 for 512-bit primes, 12 for 256-bit primes
16924   instead of only 2 for all lengths; see BN_prime_checks_for_size definition
16925   in crypto/bn/bn_prime.c for the complete table).  This guarantees a
16926   false-positive rate of at most 2^-80 for random input.
16927
16928   *Bodo Moeller*
16929
16930 * Rewrite ssl3_read_n (ssl/s3_pkt.c) avoiding a couple of bugs.
16931
16932   *Bodo Moeller*
16933
16934 * New function X509_CTX_rget_chain() (renamed to X509_CTX_get1_chain
16935   in the 0.9.5 release), this returns the chain
16936   from an X509_CTX structure with a dup of the stack and all
16937   the X509 reference counts upped: so the stack will exist
16938   after X509_CTX_cleanup() has been called. Modify pkcs12.c
16939   to use this.
16940
16941   Also make SSL_SESSION_print() print out the verify return
16942   code.
16943
16944   *Steve Henson*
16945
16946 * Add manpage for the pkcs12 command. Also change the default
16947   behaviour so MAC iteration counts are used unless the new
16948   -nomaciter option is used. This improves file security and
16949   only older versions of MSIE (4.0 for example) need it.
16950
16951   *Steve Henson*
16952
16953 * Honor the no-xxx Configure options when creating .DEF files.
16954
16955   *Ulf Möller*
16956
16957 * Add PKCS#10 attributes to field table: challengePassword,
16958   unstructuredName and unstructuredAddress. These are taken from
16959   draft PKCS#9 v2.0 but are compatible with v1.2 provided no
16960   international characters are used.
16961
16962   More changes to X509_ATTRIBUTE code: allow the setting of types
16963   based on strings. Remove the 'loc' parameter when adding
16964   attributes because these will be a SET OF encoding which is sorted
16965   in ASN1 order.
16966
16967   *Steve Henson*
16968
16969 * Initial changes to the 'req' utility to allow request generation
16970   automation. This will allow an application to just generate a template
16971   file containing all the field values and have req construct the
16972   request.
16973
16974   Initial support for X509_ATTRIBUTE handling. Stacks of these are
16975   used all over the place including certificate requests and PKCS#7
16976   structures. They are currently handled manually where necessary with
16977   some primitive wrappers for PKCS#7. The new functions behave in a
16978   manner analogous to the X509 extension functions: they allow
16979   attributes to be looked up by NID and added.
16980
16981   Later something similar to the X509V3 code would be desirable to
16982   automatically handle the encoding, decoding and printing of the
16983   more complex types. The string types like challengePassword can
16984   be handled by the string table functions.
16985
16986   Also modified the multi byte string table handling. Now there is
16987   a 'global mask' which masks out certain types. The table itself
16988   can use the flag STABLE_NO_MASK to ignore the mask setting: this
16989   is useful when for example there is only one permissible type
16990   (as in countryName) and using the mask might result in no valid
16991   types at all.
16992
16993   *Steve Henson*
16994
16995 * Clean up 'Finished' handling, and add functions SSL_get_finished and
16996   SSL_get_peer_finished to allow applications to obtain the latest
16997   Finished messages sent to the peer or expected from the peer,
16998   respectively.  (SSL_get_peer_finished is usually the Finished message
16999   actually received from the peer, otherwise the protocol will be aborted.)
17000
17001   As the Finished message are message digests of the complete handshake
17002   (with a total of 192 bits for TLS 1.0 and more for SSL 3.0), they can
17003   be used for external authentication procedures when the authentication
17004   provided by SSL/TLS is not desired or is not enough.
17005
17006   *Bodo Moeller*
17007
17008 * Enhanced support for Alpha Linux is added. Now ./config checks if
17009   the host supports BWX extension and if Compaq C is present on the
17010   $PATH. Just exploiting of the BWX extension results in 20-30%
17011   performance kick for some algorithms, e.g. DES and RC4 to mention
17012   a couple. Compaq C in turn generates ~20% faster code for MD5 and
17013   SHA1.
17014
17015   *Andy Polyakov*
17016
17017 * Add support for MS "fast SGC". This is arguably a violation of the
17018   SSL3/TLS protocol. Netscape SGC does two handshakes: the first with
17019   weak crypto and after checking the certificate is SGC a second one
17020   with strong crypto. MS SGC stops the first handshake after receiving
17021   the server certificate message and sends a second client hello. Since
17022   a server will typically do all the time consuming operations before
17023   expecting any further messages from the client (server key exchange
17024   is the most expensive) there is little difference between the two.
17025
17026   To get OpenSSL to support MS SGC we have to permit a second client
17027   hello message after we have sent server done. In addition we have to
17028   reset the MAC if we do get this second client hello.
17029
17030   *Steve Henson*
17031
17032 * Add a function 'd2i_AutoPrivateKey()' this will automatically decide
17033   if a DER encoded private key is RSA or DSA traditional format. Changed
17034   d2i_PrivateKey_bio() to use it. This is only needed for the "traditional"
17035   format DER encoded private key. Newer code should use PKCS#8 format which
17036   has the key type encoded in the ASN1 structure. Added DER private key
17037   support to pkcs8 application.
17038
17039   *Steve Henson*
17040
17041 * SSL 3/TLS 1 servers now don't request certificates when an anonymous
17042   ciphersuites has been selected (as required by the SSL 3/TLS 1
17043   specifications).  Exception: When SSL_VERIFY_FAIL_IF_NO_PEER_CERT
17044   is set, we interpret this as a request to violate the specification
17045   (the worst that can happen is a handshake failure, and 'correct'
17046   behaviour would result in a handshake failure anyway).
17047
17048   *Bodo Moeller*
17049
17050 * In SSL_CTX_add_session, take into account that there might be multiple
17051   SSL_SESSION structures with the same session ID (e.g. when two threads
17052   concurrently obtain them from an external cache).
17053   The internal cache can handle only one SSL_SESSION with a given ID,
17054   so if there's a conflict, we now throw out the old one to achieve
17055   consistency.
17056
17057   *Bodo Moeller*
17058
17059 * Add OIDs for idea and blowfish in CBC mode. This will allow both
17060   to be used in PKCS#5 v2.0 and S/MIME.  Also add checking to
17061   some routines that use cipher OIDs: some ciphers do not have OIDs
17062   defined and so they cannot be used for S/MIME and PKCS#5 v2.0 for
17063   example.
17064
17065   *Steve Henson*
17066
17067 * Simplify the trust setting structure and code. Now we just have
17068   two sequences of OIDs for trusted and rejected settings. These will
17069   typically have values the same as the extended key usage extension
17070   and any application specific purposes.
17071
17072   The trust checking code now has a default behaviour: it will just
17073   check for an object with the same NID as the passed id. Functions can
17074   be provided to override either the default behaviour or the behaviour
17075   for a given id. SSL client, server and email already have functions
17076   in place for compatibility: they check the NID and also return "trusted"
17077   if the certificate is self signed.
17078
17079   *Steve Henson*
17080
17081 * Add d2i,i2d bio/fp functions for PrivateKey: these convert the
17082   traditional format into an EVP_PKEY structure.
17083
17084   *Steve Henson*
17085
17086 * Add a password callback function PEM_cb() which either prompts for
17087   a password if usr_data is NULL or otherwise assumes it is a null
17088   terminated password. Allow passwords to be passed on command line
17089   environment or config files in a few more utilities.
17090
17091   *Steve Henson*
17092
17093 * Add a bunch of DER and PEM functions to handle PKCS#8 format private
17094   keys. Add some short names for PKCS#8 PBE algorithms and allow them
17095   to be specified on the command line for the pkcs8 and pkcs12 utilities.
17096   Update documentation.
17097
17098   *Steve Henson*
17099
17100 * Support for ASN1 "NULL" type. This could be handled before by using
17101   ASN1_TYPE but there wasn't any function that would try to read a NULL
17102   and produce an error if it couldn't. For compatibility we also have
17103   ASN1_NULL_new() and ASN1_NULL_free() functions but these are faked and
17104   don't allocate anything because they don't need to.
17105
17106   *Steve Henson*
17107
17108 * Initial support for MacOS is now provided. Examine INSTALL.MacOS
17109   for details.
17110
17111   *Andy Polyakov, Roy Woods <roy@centicsystems.ca>*
17112
17113 * Rebuild of the memory allocation routines used by OpenSSL code and
17114   possibly others as well.  The purpose is to make an interface that
17115   provide hooks so anyone can build a separate set of allocation and
17116   deallocation routines to be used by OpenSSL, for example memory
17117   pool implementations, or something else, which was previously hard
17118   since Malloc(), Realloc() and Free() were defined as macros having
17119   the values malloc, realloc and free, respectively (except for Win32
17120   compilations).  The same is provided for memory debugging code.
17121   OpenSSL already comes with functionality to find memory leaks, but
17122   this gives people a chance to debug other memory problems.
17123
17124   With these changes, a new set of functions and macros have appeared:
17125
17126     CRYPTO_set_mem_debug_functions()         [F]
17127     CRYPTO_get_mem_debug_functions()         [F]
17128     CRYPTO_dbg_set_options()                 [F]
17129     CRYPTO_dbg_get_options()                 [F]
17130     CRYPTO_malloc_debug_init()               [M]
17131
17132   The memory debug functions are NULL by default, unless the library
17133   is compiled with CRYPTO_MDEBUG or friends is defined.  If someone
17134   wants to debug memory anyway, CRYPTO_malloc_debug_init() (which
17135   gives the standard debugging functions that come with OpenSSL) or
17136   CRYPTO_set_mem_debug_functions() (tells OpenSSL to use functions
17137   provided by the library user) must be used.  When the standard
17138   debugging functions are used, CRYPTO_dbg_set_options can be used to
17139   request additional information:
17140   CRYPTO_dbg_set_options(V_CYRPTO_MDEBUG_xxx) corresponds to setting
17141   the CRYPTO_MDEBUG_xxx macro when compiling the library.
17142
17143   Also, things like CRYPTO_set_mem_functions will always give the
17144   expected result (the new set of functions is used for allocation
17145   and deallocation) at all times, regardless of platform and compiler
17146   options.
17147
17148   To finish it up, some functions that were never use in any other
17149   way than through macros have a new API and new semantic:
17150
17151     CRYPTO_dbg_malloc()
17152     CRYPTO_dbg_realloc()
17153     CRYPTO_dbg_free()
17154
17155   All macros of value have retained their old syntax.
17156
17157   *Richard Levitte and Bodo Moeller*
17158
17159 * Some S/MIME fixes. The OID for SMIMECapabilities was wrong, the
17160   ordering of SMIMECapabilities wasn't in "strength order" and there
17161   was a missing NULL in the AlgorithmIdentifier for the SHA1 signature
17162   algorithm.
17163
17164   *Steve Henson*
17165
17166 * Some ASN1 types with illegal zero length encoding (INTEGER,
17167   ENUMERATED and OBJECT IDENTIFIER) choked the ASN1 routines.
17168
17169   *Frans Heymans <fheymans@isaserver.be>, modified by Steve Henson*
17170
17171 * Merge in my S/MIME library for OpenSSL. This provides a simple
17172   S/MIME API on top of the PKCS#7 code, a MIME parser (with enough
17173   functionality to handle multipart/signed properly) and a utility
17174   called 'smime' to call all this stuff. This is based on code I
17175   originally wrote for Celo who have kindly allowed it to be
17176   included in OpenSSL.
17177
17178   *Steve Henson*
17179
17180 * Add variants des_set_key_checked and des_set_key_unchecked of
17181   des_set_key (aka des_key_sched).  Global variable des_check_key
17182   decides which of these is called by des_set_key; this way
17183   des_check_key behaves as it always did, but applications and
17184   the library itself, which was buggy for des_check_key == 1,
17185   have a cleaner way to pick the version they need.
17186
17187   *Bodo Moeller*
17188
17189 * New function PKCS12_newpass() which changes the password of a
17190   PKCS12 structure.
17191
17192   *Steve Henson*
17193
17194 * Modify X509_TRUST and X509_PURPOSE so it also uses a static and
17195   dynamic mix. In both cases the ids can be used as an index into the
17196   table. Also modified the X509_TRUST_add() and X509_PURPOSE_add()
17197   functions so they accept a list of the field values and the
17198   application doesn't need to directly manipulate the X509_TRUST
17199   structure.
17200
17201   *Steve Henson*
17202
17203 * Modify the ASN1_STRING_TABLE stuff so it also uses bsearch and doesn't
17204   need initialising.
17205
17206   *Steve Henson*
17207
17208 * Modify the way the V3 extension code looks up extensions. This now
17209   works in a similar way to the object code: we have some "standard"
17210   extensions in a static table which is searched with OBJ_bsearch()
17211   and the application can add dynamic ones if needed. The file
17212   crypto/x509v3/ext_dat.h now has the info: this file needs to be
17213   updated whenever a new extension is added to the core code and kept
17214   in ext_nid order. There is a simple program 'tabtest.c' which checks
17215   this. New extensions are not added too often so this file can readily
17216   be maintained manually.
17217
17218   There are two big advantages in doing things this way. The extensions
17219   can be looked up immediately and no longer need to be "added" using
17220   X509V3_add_standard_extensions(): this function now does nothing.
17221   Side note: I get *lots* of email saying the extension code doesn't
17222   work because people forget to call this function.
17223   Also no dynamic allocation is done unless new extensions are added:
17224   so if we don't add custom extensions there is no need to call
17225   X509V3_EXT_cleanup().
17226
17227   *Steve Henson*
17228
17229 * Modify enc utility's salting as follows: make salting the default. Add a
17230   magic header, so unsalted files fail gracefully instead of just decrypting
17231   to garbage. This is because not salting is a big security hole, so people
17232   should be discouraged from doing it.
17233
17234   *Ben Laurie*
17235
17236 * Fixes and enhancements to the 'x509' utility. It allowed a message
17237   digest to be passed on the command line but it only used this
17238   parameter when signing a certificate. Modified so all relevant
17239   operations are affected by the digest parameter including the
17240   -fingerprint and -x509toreq options. Also -x509toreq choked if a
17241   DSA key was used because it didn't fix the digest.
17242
17243   *Steve Henson*
17244
17245 * Initial certificate chain verify code. Currently tests the untrusted
17246   certificates for consistency with the verify purpose (which is set
17247   when the X509_STORE_CTX structure is set up) and checks the pathlength.
17248
17249   There is a NO_CHAIN_VERIFY compilation option to keep the old behaviour:
17250   this is because it will reject chains with invalid extensions whereas
17251   every previous version of OpenSSL and SSLeay made no checks at all.
17252
17253   Trust code: checks the root CA for the relevant trust settings. Trust
17254   settings have an initial value consistent with the verify purpose: e.g.
17255   if the verify purpose is for SSL client use it expects the CA to be
17256   trusted for SSL client use. However the default value can be changed to
17257   permit custom trust settings: one example of this would be to only trust
17258   certificates from a specific "secure" set of CAs.
17259
17260   Also added X509_STORE_CTX_new() and X509_STORE_CTX_free() functions
17261   which should be used for version portability: especially since the
17262   verify structure is likely to change more often now.
17263
17264   SSL integration. Add purpose and trust to SSL_CTX and SSL and functions
17265   to set them. If not set then assume SSL clients will verify SSL servers
17266   and vice versa.
17267
17268   Two new options to the verify program: -untrusted allows a set of
17269   untrusted certificates to be passed in and -purpose which sets the
17270   intended purpose of the certificate. If a purpose is set then the
17271   new chain verify code is used to check extension consistency.
17272
17273   *Steve Henson*
17274
17275 * Support for the authority information access extension.
17276
17277   *Steve Henson*
17278
17279 * Modify RSA and DSA PEM read routines to transparently handle
17280   PKCS#8 format private keys. New *_PUBKEY_* functions that handle
17281   public keys in a format compatible with certificate
17282   SubjectPublicKeyInfo structures. Unfortunately there were already
17283   functions called *_PublicKey_* which used various odd formats so
17284   these are retained for compatibility: however the DSA variants were
17285   never in a public release so they have been deleted. Changed dsa/rsa
17286   utilities to handle the new format: note no releases ever handled public
17287   keys so we should be OK.
17288
17289   The primary motivation for this change is to avoid the same fiasco
17290   that dogs private keys: there are several incompatible private key
17291   formats some of which are standard and some OpenSSL specific and
17292   require various evil hacks to allow partial transparent handling and
17293   even then it doesn't work with DER formats. Given the option anything
17294   other than PKCS#8 should be dumped: but the other formats have to
17295   stay in the name of compatibility.
17296
17297   With public keys and the benefit of hindsight one standard format
17298   is used which works with EVP_PKEY, RSA or DSA structures: though
17299   it clearly returns an error if you try to read the wrong kind of key.
17300
17301   Added a -pubkey option to the 'x509' utility to output the public key.
17302   Also rename the `EVP_PKEY_get_*()` to `EVP_PKEY_rget_*()`
17303   (renamed to `EVP_PKEY_get1_*()` in the OpenSSL 0.9.5 release) and add
17304   `EVP_PKEY_rset_*()` functions (renamed to `EVP_PKEY_set1_*()`)
17305   that do the same as the `EVP_PKEY_assign_*()` except they up the
17306   reference count of the added key (they don't "swallow" the
17307   supplied key).
17308
17309   *Steve Henson*
17310
17311 * Fixes to crypto/x509/by_file.c the code to read in certificates and
17312   CRLs would fail if the file contained no certificates or no CRLs:
17313   added a new function to read in both types and return the number
17314   read: this means that if none are read it will be an error. The
17315   DER versions of the certificate and CRL reader would always fail
17316   because it isn't possible to mix certificates and CRLs in DER format
17317   without choking one or the other routine. Changed this to just read
17318   a certificate: this is the best we can do. Also modified the code
17319   in `apps/verify.c` to take notice of return codes: it was previously
17320   attempting to read in certificates from NULL pointers and ignoring
17321   any errors: this is one reason why the cert and CRL reader seemed
17322   to work. It doesn't check return codes from the default certificate
17323   routines: these may well fail if the certificates aren't installed.
17324
17325   *Steve Henson*
17326
17327 * Code to support otherName option in GeneralName.
17328
17329   *Steve Henson*
17330
17331 * First update to verify code. Change the verify utility
17332   so it warns if it is passed a self signed certificate:
17333   for consistency with the normal behaviour. X509_verify
17334   has been modified to it will now verify a self signed
17335   certificate if *exactly* the same certificate appears
17336   in the store: it was previously impossible to trust a
17337   single self signed certificate. This means that:
17338   openssl verify ss.pem
17339   now gives a warning about a self signed certificate but
17340   openssl verify -CAfile ss.pem ss.pem
17341   is OK.
17342
17343   *Steve Henson*
17344
17345 * For servers, store verify_result in SSL_SESSION data structure
17346   (and add it to external session representation).
17347   This is needed when client certificate verifications fails,
17348   but an application-provided verification callback (set by
17349   SSL_CTX_set_cert_verify_callback) allows accepting the session
17350   anyway (i.e. leaves x509_store_ctx->error != X509_V_OK
17351   but returns 1): When the session is reused, we have to set
17352   ssl->verify_result to the appropriate error code to avoid
17353   security holes.
17354
17355   *Bodo Moeller, problem pointed out by Lutz Jaenicke*
17356
17357 * Fix a bug in the new PKCS#7 code: it didn't consider the
17358   case in PKCS7_dataInit() where the signed PKCS7 structure
17359   didn't contain any existing data because it was being created.
17360
17361   *Po-Cheng Chen <pocheng@nst.com.tw>, slightly modified by Steve Henson*
17362
17363 * Add a salt to the key derivation routines in enc.c. This
17364   forms the first 8 bytes of the encrypted file. Also add a
17365   -S option to allow a salt to be input on the command line.
17366
17367   *Steve Henson*
17368
17369 * New function X509_cmp(). Oddly enough there wasn't a function
17370   to compare two certificates. We do this by working out the SHA1
17371   hash and comparing that. X509_cmp() will be needed by the trust
17372   code.
17373
17374   *Steve Henson*
17375
17376 * SSL_get1_session() is like SSL_get_session(), but increments
17377   the reference count in the SSL_SESSION returned.
17378
17379   *Geoff Thorpe <geoff@eu.c2.net>*
17380
17381 * Fix for 'req': it was adding a null to request attributes.
17382   Also change the X509_LOOKUP and X509_INFO code to handle
17383   certificate auxiliary information.
17384
17385   *Steve Henson*
17386
17387 * Add support for 40 and 64 bit RC2 and RC4 algorithms: document
17388   the 'enc' command.
17389
17390   *Steve Henson*
17391
17392 * Add the possibility to add extra information to the memory leak
17393   detecting output, to form tracebacks, showing from where each
17394   allocation was originated: CRYPTO_push_info("constant string") adds
17395   the string plus current file name and line number to a per-thread
17396   stack, CRYPTO_pop_info() does the obvious, CRYPTO_remove_all_info()
17397   is like calling CYRPTO_pop_info() until the stack is empty.
17398   Also updated memory leak detection code to be multi-thread-safe.
17399
17400   *Richard Levitte*
17401
17402 * Add options -text and -noout to pkcs7 utility and delete the
17403   encryption options which never did anything. Update docs.
17404
17405   *Steve Henson*
17406
17407 * Add options to some of the utilities to allow the pass phrase
17408   to be included on either the command line (not recommended on
17409   OSes like Unix) or read from the environment. Update the
17410   manpages and fix a few bugs.
17411
17412   *Steve Henson*
17413
17414 * Add a few manpages for some of the openssl commands.
17415
17416   *Steve Henson*
17417
17418 * Fix the -revoke option in ca. It was freeing up memory twice,
17419   leaking and not finding already revoked certificates.
17420
17421   *Steve Henson*
17422
17423 * Extensive changes to support certificate auxiliary information.
17424   This involves the use of X509_CERT_AUX structure and X509_AUX
17425   functions. An X509_AUX function such as PEM_read_X509_AUX()
17426   can still read in a certificate file in the usual way but it
17427   will also read in any additional "auxiliary information". By
17428   doing things this way a fair degree of compatibility can be
17429   retained: existing certificates can have this information added
17430   using the new 'x509' options.
17431
17432   Current auxiliary information includes an "alias" and some trust
17433   settings. The trust settings will ultimately be used in enhanced
17434   certificate chain verification routines: currently a certificate
17435   can only be trusted if it is self signed and then it is trusted
17436   for all purposes.
17437
17438   *Steve Henson*
17439
17440 * Fix assembler for Alpha (tested only on DEC OSF not Linux or `*BSD`).
17441   The problem was that one of the replacement routines had not been working
17442   since SSLeay releases.  For now the offending routine has been replaced
17443   with non-optimised assembler.  Even so, this now gives around 95%
17444   performance improvement for 1024 bit RSA signs.
17445
17446   *Mark Cox*
17447
17448 * Hack to fix PKCS#7 decryption when used with some unorthodox RC2
17449   handling. Most clients have the effective key size in bits equal to
17450   the key length in bits: so a 40 bit RC2 key uses a 40 bit (5 byte) key.
17451   A few however don't do this and instead use the size of the decrypted key
17452   to determine the RC2 key length and the AlgorithmIdentifier to determine
17453   the effective key length. In this case the effective key length can still
17454   be 40 bits but the key length can be 168 bits for example. This is fixed
17455   by manually forcing an RC2 key into the EVP_PKEY structure because the
17456   EVP code can't currently handle unusual RC2 key sizes: it always assumes
17457   the key length and effective key length are equal.
17458
17459   *Steve Henson*
17460
17461 * Add a bunch of functions that should simplify the creation of
17462   X509_NAME structures. Now you should be able to do:
17463   X509_NAME_add_entry_by_txt(nm, "CN", MBSTRING_ASC, "Steve", -1, -1, 0);
17464   and have it automatically work out the correct field type and fill in
17465   the structures. The more adventurous can try:
17466   X509_NAME_add_entry_by_txt(nm, field, MBSTRING_UTF8, str, -1, -1, 0);
17467   and it will (hopefully) work out the correct multibyte encoding.
17468
17469   *Steve Henson*
17470
17471 * Change the 'req' utility to use the new field handling and multibyte
17472   copy routines. Before the DN field creation was handled in an ad hoc
17473   way in req, ca, and x509 which was rather broken and didn't support
17474   BMPStrings or UTF8Strings. Since some software doesn't implement
17475   BMPStrings or UTF8Strings yet, they can be enabled using the config file
17476   using the dirstring_type option. See the new comment in the default
17477   openssl.cnf for more info.
17478
17479   *Steve Henson*
17480
17481 * Make crypto/rand/md_rand.c more robust:
17482   - Assure unique random numbers after fork().
17483   - Make sure that concurrent threads access the global counter and
17484     md serializably so that we never lose entropy in them
17485     or use exactly the same state in multiple threads.
17486     Access to the large state is not always serializable because
17487     the additional locking could be a performance killer, and
17488     md should be large enough anyway.
17489
17490   *Bodo Moeller*
17491
17492 * New file `apps/app_rand.c` with commonly needed functionality
17493   for handling the random seed file.
17494
17495   Use the random seed file in some applications that previously did not:
17496           ca,
17497           dsaparam -genkey (which also ignored its '-rand' option),
17498           s_client,
17499           s_server,
17500           x509 (when signing).
17501   Except on systems with /dev/urandom, it is crucial to have a random
17502   seed file at least for key creation, DSA signing, and for DH exchanges;
17503   for RSA signatures we could do without one.
17504
17505   gendh and gendsa (unlike genrsa) used to read only the first byte
17506   of each file listed in the '-rand' option.  The function as previously
17507   found in genrsa is now in app_rand.c and is used by all programs
17508   that support '-rand'.
17509
17510   *Bodo Moeller*
17511
17512 * In RAND_write_file, use mode 0600 for creating files;
17513   don't just chmod when it may be too late.
17514
17515   *Bodo Moeller*
17516
17517 * Report an error from X509_STORE_load_locations
17518   when X509_LOOKUP_load_file or X509_LOOKUP_add_dir failed.
17519
17520   *Bill Perry*
17521
17522 * New function ASN1_mbstring_copy() this copies a string in either
17523   ASCII, Unicode, Universal (4 bytes per character) or UTF8 format
17524   into an ASN1_STRING type. A mask of permissible types is passed
17525   and it chooses the "minimal" type to use or an error if not type
17526   is suitable.
17527
17528   *Steve Henson*
17529
17530 * Add function equivalents to the various macros in asn1.h. The old
17531   macros are retained with an `M_` prefix. Code inside the library can
17532   use the `M_` macros. External code (including the openssl utility)
17533   should *NOT* in order to be "shared library friendly".
17534
17535   *Steve Henson*
17536
17537 * Add various functions that can check a certificate's extensions
17538   to see if it usable for various purposes such as SSL client,
17539   server or S/MIME and CAs of these types. This is currently
17540   VERY EXPERIMENTAL but will ultimately be used for certificate chain
17541   verification. Also added a -purpose flag to x509 utility to
17542   print out all the purposes.
17543
17544   *Steve Henson*
17545
17546 * Add a CRYPTO_EX_DATA to X509 certificate structure and associated
17547   functions.
17548
17549   *Steve Henson*
17550
17551 * New `X509V3_{X509,CRL,REVOKED}_get_d2i()` functions. These will search
17552   for, obtain and decode and extension and obtain its critical flag.
17553   This allows all the necessary extension code to be handled in a
17554   single function call.
17555
17556   *Steve Henson*
17557
17558 * RC4 tune-up featuring 30-40% performance improvement on most RISC
17559   platforms. See crypto/rc4/rc4_enc.c for further details.
17560
17561   *Andy Polyakov*
17562
17563 * New -noout option to asn1parse. This causes no output to be produced
17564   its main use is when combined with -strparse and -out to extract data
17565   from a file (which may not be in ASN.1 format).
17566
17567   *Steve Henson*
17568
17569 * Fix for pkcs12 program. It was hashing an invalid certificate pointer
17570   when producing the local key id.
17571
17572   *Richard Levitte <levitte@stacken.kth.se>*
17573
17574 * New option -dhparam in s_server. This allows a DH parameter file to be
17575   stated explicitly. If it is not stated then it tries the first server
17576   certificate file. The previous behaviour hard coded the filename
17577   "server.pem".
17578
17579   *Steve Henson*
17580
17581 * Add -pubin and -pubout options to the rsa and dsa commands. These allow
17582   a public key to be input or output. For example:
17583   openssl rsa -in key.pem -pubout -out pubkey.pem
17584   Also added necessary DSA public key functions to handle this.
17585
17586   *Steve Henson*
17587
17588 * Fix so PKCS7_dataVerify() doesn't crash if no certificates are contained
17589   in the message. This was handled by allowing
17590   X509_find_by_issuer_and_serial() to tolerate a NULL passed to it.
17591
17592   *Steve Henson, reported by Sampo Kellomaki <sampo@mail.neuronio.pt>*
17593
17594 * Fix for bug in d2i_ASN1_bytes(): other ASN1 functions add an extra null
17595   to the end of the strings whereas this didn't. This would cause problems
17596   if strings read with d2i_ASN1_bytes() were later modified.
17597
17598   *Steve Henson, reported by Arne Ansper <arne@ats.cyber.ee>*
17599
17600 * Fix for base64 decode bug. When a base64 bio reads only one line of
17601   data and it contains EOF it will end up returning an error. This is
17602   caused by input 46 bytes long. The cause is due to the way base64
17603   BIOs find the start of base64 encoded data. They do this by trying a
17604   trial decode on each line until they find one that works. When they
17605   do a flag is set and it starts again knowing it can pass all the
17606   data directly through the decoder. Unfortunately it doesn't reset
17607   the context it uses. This means that if EOF is reached an attempt
17608   is made to pass two EOFs through the context and this causes the
17609   resulting error. This can also cause other problems as well. As is
17610   usual with these problems it takes *ages* to find and the fix is
17611   trivial: move one line.
17612
17613   *Steve Henson, reported by ian@uns.ns.ac.yu (Ivan Nejgebauer)*
17614
17615 * Ugly workaround to get s_client and s_server working under Windows. The
17616   old code wouldn't work because it needed to select() on sockets and the
17617   tty (for keypresses and to see if data could be written). Win32 only
17618   supports select() on sockets so we select() with a 1s timeout on the
17619   sockets and then see if any characters are waiting to be read, if none
17620   are present then we retry, we also assume we can always write data to
17621   the tty. This isn't nice because the code then blocks until we've
17622   received a complete line of data and it is effectively polling the
17623   keyboard at 1s intervals: however it's quite a bit better than not
17624   working at all :-) A dedicated Windows application might handle this
17625   with an event loop for example.
17626
17627   *Steve Henson*
17628
17629 * Enhance RSA_METHOD structure. Now there are two extra methods, rsa_sign
17630   and rsa_verify. When the RSA_FLAGS_SIGN_VER option is set these functions
17631   will be called when RSA_sign() and RSA_verify() are used. This is useful
17632   if rsa_pub_dec() and rsa_priv_enc() equivalents are not available.
17633   For this to work properly RSA_public_decrypt() and RSA_private_encrypt()
17634   should *not* be used: RSA_sign() and RSA_verify() must be used instead.
17635   This necessitated the support of an extra signature type NID_md5_sha1
17636   for SSL signatures and modifications to the SSL library to use it instead
17637   of calling RSA_public_decrypt() and RSA_private_encrypt().
17638
17639   *Steve Henson*
17640
17641 * Add new -verify -CAfile and -CApath options to the crl program, these
17642   will lookup a CRL issuers certificate and verify the signature in a
17643   similar way to the verify program. Tidy up the crl program so it
17644   no longer accesses structures directly. Make the ASN1 CRL parsing a bit
17645   less strict. It will now permit CRL extensions even if it is not
17646   a V2 CRL: this will allow it to tolerate some broken CRLs.
17647
17648   *Steve Henson*
17649
17650 * Initialize all non-automatic variables each time one of the openssl
17651   sub-programs is started (this is necessary as they may be started
17652   multiple times from the "OpenSSL>" prompt).
17653
17654   *Lennart Bang, Bodo Moeller*
17655
17656 * Preliminary compilation option RSA_NULL which disables RSA crypto without
17657   removing all other RSA functionality (this is what NO_RSA does). This
17658   is so (for example) those in the US can disable those operations covered
17659   by the RSA patent while allowing storage and parsing of RSA keys and RSA
17660   key generation.
17661
17662   *Steve Henson*
17663
17664 * Non-copying interface to BIO pairs.
17665   (still largely untested)
17666
17667   *Bodo Moeller*
17668
17669 * New function ASN1_tag2str() to convert an ASN1 tag to a descriptive
17670   ASCII string. This was handled independently in various places before.
17671
17672   *Steve Henson*
17673
17674 * New functions UTF8_getc() and UTF8_putc() that parse and generate
17675   UTF8 strings a character at a time.
17676
17677   *Steve Henson*
17678
17679 * Use client_version from client hello to select the protocol
17680   (s23_srvr.c) and for RSA client key exchange verification
17681   (s3_srvr.c), as required by the SSL 3.0/TLS 1.0 specifications.
17682
17683   *Bodo Moeller*
17684
17685 * Add various utility functions to handle SPKACs, these were previously
17686   handled by poking round in the structure internals. Added new function
17687   NETSCAPE_SPKI_print() to print out SPKAC and a new utility 'spkac' to
17688   print, verify and generate SPKACs. Based on an original idea from
17689   Massimiliano Pala <madwolf@comune.modena.it> but extensively modified.
17690
17691   *Steve Henson*
17692
17693 * RIPEMD160 is operational on all platforms and is back in 'make test'.
17694
17695   *Andy Polyakov*
17696
17697 * Allow the config file extension section to be overwritten on the
17698   command line. Based on an original idea from Massimiliano Pala
17699   <madwolf@comune.modena.it>. The new option is called -extensions
17700   and can be applied to ca, req and x509. Also -reqexts to override
17701   the request extensions in req and -crlexts to override the crl extensions
17702   in ca.
17703
17704   *Steve Henson*
17705
17706 * Add new feature to the SPKAC handling in ca.  Now you can include
17707   the same field multiple times by preceding it by "XXXX." for example:
17708   1.OU="Unit name 1"
17709   2.OU="Unit name 2"
17710   this is the same syntax as used in the req config file.
17711
17712   *Steve Henson*
17713
17714 * Allow certificate extensions to be added to certificate requests. These
17715   are specified in a 'req_extensions' option of the req section of the
17716   config file. They can be printed out with the -text option to req but
17717   are otherwise ignored at present.
17718
17719   *Steve Henson*
17720
17721 * Fix a horrible bug in enc_read() in crypto/evp/bio_enc.c: if the first
17722   data read consists of only the final block it would not decrypted because
17723   EVP_CipherUpdate() would correctly report zero bytes had been decrypted.
17724   A misplaced 'break' also meant the decrypted final block might not be
17725   copied until the next read.
17726
17727   *Steve Henson*
17728
17729 * Initial support for DH_METHOD. Again based on RSA_METHOD. Also added
17730   a few extra parameters to the DH structure: these will be useful if
17731   for example we want the value of 'q' or implement X9.42 DH.
17732
17733   *Steve Henson*
17734
17735 * Initial support for DSA_METHOD. This is based on the RSA_METHOD and
17736   provides hooks that allow the default DSA functions or functions on a
17737   "per key" basis to be replaced. This allows hardware acceleration and
17738   hardware key storage to be handled without major modification to the
17739   library. Also added low-level modexp hooks and CRYPTO_EX structure and
17740   associated functions.
17741
17742   *Steve Henson*
17743
17744 * Add a new flag to memory BIOs, BIO_FLAG_MEM_RDONLY. This marks the BIO
17745   as "read only": it can't be written to and the buffer it points to will
17746   not be freed. Reading from a read only BIO is much more efficient than
17747   a normal memory BIO. This was added because there are several times when
17748   an area of memory needs to be read from a BIO. The previous method was
17749   to create a memory BIO and write the data to it, this results in two
17750   copies of the data and an O(n^2) reading algorithm. There is a new
17751   function BIO_new_mem_buf() which creates a read only memory BIO from
17752   an area of memory. Also modified the PKCS#7 routines to use read only
17753   memory BIOs.
17754
17755   *Steve Henson*
17756
17757 * Bugfix: ssl23_get_client_hello did not work properly when called in
17758   state SSL23_ST_SR_CLNT_HELLO_B, i.e. when the first 7 bytes of
17759   a SSLv2-compatible client hello for SSLv3 or TLSv1 could be read,
17760   but a retry condition occurred while trying to read the rest.
17761
17762   *Bodo Moeller*
17763
17764 * The PKCS7_ENC_CONTENT_new() function was setting the content type as
17765   NID_pkcs7_encrypted by default: this was wrong since this should almost
17766   always be NID_pkcs7_data. Also modified the PKCS7_set_type() to handle
17767   the encrypted data type: this is a more sensible place to put it and it
17768   allows the PKCS#12 code to be tidied up that duplicated this
17769   functionality.
17770
17771   *Steve Henson*
17772
17773 * Changed obj_dat.pl script so it takes its input and output files on
17774   the command line. This should avoid shell escape redirection problems
17775   under Win32.
17776
17777   *Steve Henson*
17778
17779 * Initial support for certificate extension requests, these are included
17780   in things like Xenroll certificate requests. Included functions to allow
17781   extensions to be obtained and added.
17782
17783   *Steve Henson*
17784
17785 * -crlf option to s_client and s_server for sending newlines as
17786   CRLF (as required by many protocols).
17787
17788   *Bodo Moeller*
17789
17790### Changes between 0.9.3a and 0.9.4  [09 Aug 1999]
17791
17792 * Install libRSAglue.a when OpenSSL is built with RSAref.
17793
17794   *Ralf S. Engelschall*
17795
17796 * A few more `#ifndef NO_FP_API / #endif` pairs for consistency.
17797
17798   *Andrija Antonijevic <TheAntony2@bigfoot.com>*
17799
17800 * Fix -startdate and -enddate (which was missing) arguments to 'ca'
17801   program.
17802
17803   *Steve Henson*
17804
17805 * New function DSA_dup_DH, which duplicates DSA parameters/keys as
17806   DH parameters/keys (q is lost during that conversion, but the resulting
17807   DH parameters contain its length).
17808
17809   For 1024-bit p, DSA_generate_parameters followed by DSA_dup_DH is
17810   much faster than DH_generate_parameters (which creates parameters
17811   where `p = 2*q + 1`), and also the smaller q makes DH computations
17812   much more efficient (160-bit exponentiation instead of 1024-bit
17813   exponentiation); so this provides a convenient way to support DHE
17814   ciphersuites in SSL/TLS servers (see ssl/ssltest.c).  It is of
17815   utter importance to use
17816           SSL_CTX_set_options(s_ctx, SSL_OP_SINGLE_DH_USE);
17817   or
17818           SSL_set_options(s_ctx, SSL_OP_SINGLE_DH_USE);
17819   when such DH parameters are used, because otherwise small subgroup
17820   attacks may become possible!
17821
17822   *Bodo Moeller*
17823
17824 * Avoid memory leak in i2d_DHparams.
17825
17826   *Bodo Moeller*
17827
17828 * Allow the -k option to be used more than once in the enc program:
17829   this allows the same encrypted message to be read by multiple recipients.
17830
17831   *Steve Henson*
17832
17833 * New function OBJ_obj2txt(buf, buf_len, a, no_name), this converts
17834   an ASN1_OBJECT to a text string. If the "no_name" parameter is set then
17835   it will always use the numerical form of the OID, even if it has a short
17836   or long name.
17837
17838   *Steve Henson*
17839
17840 * Added an extra RSA flag: RSA_FLAG_EXT_PKEY. Previously the rsa_mod_exp
17841   method only got called if p,q,dmp1,dmq1,iqmp components were present,
17842   otherwise bn_mod_exp was called. In the case of hardware keys for example
17843   no private key components need be present and it might store extra data
17844   in the RSA structure, which cannot be accessed from bn_mod_exp.
17845   By setting RSA_FLAG_EXT_PKEY rsa_mod_exp will always be called for
17846   private key operations.
17847
17848   *Steve Henson*
17849
17850 * Added support for SPARC Linux.
17851
17852   *Andy Polyakov*
17853
17854 * pem_password_cb function type incompatibly changed from
17855           typedef int pem_password_cb(char *buf, int size, int rwflag);
17856   to
17857           ....(char *buf, int size, int rwflag, void *userdata);
17858   so that applications can pass data to their callbacks:
17859   The `PEM[_ASN1]_{read,write}...` functions and macros now take an
17860   additional void * argument, which is just handed through whenever
17861   the password callback is called.
17862
17863   *Damien Miller <dmiller@ilogic.com.au>; tiny changes by Bodo Moeller*
17864
17865   New function SSL_CTX_set_default_passwd_cb_userdata.
17866
17867   Compatibility note: As many C implementations push function arguments
17868   onto the stack in reverse order, the new library version is likely to
17869   interoperate with programs that have been compiled with the old
17870   pem_password_cb definition (PEM_whatever takes some data that
17871   happens to be on the stack as its last argument, and the callback
17872   just ignores this garbage); but there is no guarantee whatsoever that
17873   this will work.
17874
17875 * The -DPLATFORM="\"$(PLATFORM)\"" definition and the similar -DCFLAGS=...
17876   (both in crypto/Makefile.ssl for use by crypto/cversion.c) caused
17877   problems not only on Windows, but also on some Unix platforms.
17878   To avoid problematic command lines, these definitions are now in an
17879   auto-generated file crypto/buildinf.h (created by crypto/Makefile.ssl
17880   for standard "make" builds, by util/mk1mf.pl for "mk1mf" builds).
17881
17882   *Bodo Moeller*
17883
17884 * MIPS III/IV assembler module is reimplemented.
17885
17886   *Andy Polyakov*
17887
17888 * More DES library cleanups: remove references to srand/rand and
17889   delete an unused file.
17890
17891   *Ulf Möller*
17892
17893 * Add support for the free Netwide assembler (NASM) under Win32,
17894   since not many people have MASM (ml) and it can be hard to obtain.
17895   This is currently experimental but it seems to work OK and pass all
17896   the tests. Check out INSTALL.W32 for info.
17897
17898   *Steve Henson*
17899
17900 * Fix memory leaks in s3_clnt.c: All non-anonymous SSL3/TLS1 connections
17901   without temporary keys kept an extra copy of the server key,
17902   and connections with temporary keys did not free everything in case
17903   of an error.
17904
17905   *Bodo Moeller*
17906
17907 * New function RSA_check_key and new openssl rsa option -check
17908   for verifying the consistency of RSA keys.
17909
17910   *Ulf Moeller, Bodo Moeller*
17911
17912 * Various changes to make Win32 compile work:
17913   1. Casts to avoid "loss of data" warnings in p5_crpt2.c
17914   2. Change unsigned int to int in b_dump.c to avoid "signed/unsigned
17915      comparison" warnings.
17916   3. Add `sk_<TYPE>_sort` to DEF file generator and do make update.
17917
17918   *Steve Henson*
17919
17920 * Add a debugging option to PKCS#5 v2 key generation function: when
17921   you #define DEBUG_PKCS5V2 passwords, salts, iteration counts and
17922   derived keys are printed to stderr.
17923
17924   *Steve Henson*
17925
17926 * Copy the flags in ASN1_STRING_dup().
17927
17928   *Roman E. Pavlov <pre@mo.msk.ru>*
17929
17930 * The x509 application mishandled signing requests containing DSA
17931   keys when the signing key was also DSA and the parameters didn't match.
17932
17933   It was supposed to omit the parameters when they matched the signing key:
17934   the verifying software was then supposed to automatically use the CA's
17935   parameters if they were absent from the end user certificate.
17936
17937   Omitting parameters is no longer recommended. The test was also
17938   the wrong way round! This was probably due to unusual behaviour in
17939   EVP_cmp_parameters() which returns 1 if the parameters match.
17940   This meant that parameters were omitted when they *didn't* match and
17941   the certificate was useless. Certificates signed with 'ca' didn't have
17942   this bug.
17943
17944   *Steve Henson, reported by Doug Erickson <Doug.Erickson@Part.NET>*
17945
17946 * Memory leak checking (-DCRYPTO_MDEBUG) had some problems.
17947   The interface is as follows:
17948   Applications can use
17949           CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_ON) aka MemCheck_start(),
17950           CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_OFF) aka MemCheck_stop();
17951   "off" is now the default.
17952   The library internally uses
17953           CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_DISABLE) aka MemCheck_off(),
17954           CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_ENABLE) aka MemCheck_on()
17955   to disable memory-checking temporarily.
17956
17957   Some inconsistent states that previously were possible (and were
17958   even the default) are now avoided.
17959
17960   -DCRYPTO_MDEBUG_TIME is new and additionally stores the current time
17961   with each memory chunk allocated; this is occasionally more helpful
17962   than just having a counter.
17963
17964   -DCRYPTO_MDEBUG_THREAD is also new and adds the thread ID.
17965
17966   -DCRYPTO_MDEBUG_ALL enables all of the above, plus any future
17967   extensions.
17968
17969   *Bodo Moeller*
17970
17971 * Introduce "mode" for SSL structures (with defaults in SSL_CTX),
17972   which largely parallels "options", but is for changing API behaviour,
17973   whereas "options" are about protocol behaviour.
17974   Initial "mode" flags are:
17975
17976   SSL_MODE_ENABLE_PARTIAL_WRITE   Allow SSL_write to report success when
17977                                   a single record has been written.
17978   SSL_MODE_ACCEPT_MOVING_WRITE_BUFFER  Don't insist that SSL_write
17979                                   retries use the same buffer location.
17980                                   (But all of the contents must be
17981                                   copied!)
17982
17983   *Bodo Moeller*
17984
17985 * Bugfix: SSL_set_options ignored its parameter, only SSL_CTX_set_options
17986   worked.
17987
17988 * Fix problems with no-hmac etc.
17989
17990   *Ulf Möller, pointed out by Brian Wellington <bwelling@tislabs.com>*
17991
17992 * New functions RSA_get_default_method(), RSA_set_method() and
17993   RSA_get_method(). These allows replacement of RSA_METHODs without having
17994   to mess around with the internals of an RSA structure.
17995
17996   *Steve Henson*
17997
17998 * Fix memory leaks in DSA_do_sign and DSA_is_prime.
17999   Also really enable memory leak checks in openssl.c and in some
18000   test programs.
18001
18002   *Chad C. Mulligan, Bodo Moeller*
18003
18004 * Fix a bug in d2i_ASN1_INTEGER() and i2d_ASN1_INTEGER() which can mess
18005   up the length of negative integers. This has now been simplified to just
18006   store the length when it is first determined and use it later, rather
18007   than trying to keep track of where data is copied and updating it to
18008   point to the end.
18009   *Steve Henson, reported by Brien Wheeler <bwheeler@authentica-security.com>*
18010
18011 * Add a new function PKCS7_signatureVerify. This allows the verification
18012   of a PKCS#7 signature but with the signing certificate passed to the
18013   function itself. This contrasts with PKCS7_dataVerify which assumes the
18014   certificate is present in the PKCS#7 structure. This isn't always the
18015   case: certificates can be omitted from a PKCS#7 structure and be
18016   distributed by "out of band" means (such as a certificate database).
18017
18018   *Steve Henson*
18019
18020 * Complete the `PEM_*` macros with DECLARE_PEM versions to replace the
18021   function prototypes in pem.h, also change util/mkdef.pl to add the
18022   necessary function names.
18023
18024   *Steve Henson*
18025
18026 * mk1mf.pl (used by Windows builds) did not properly read the
18027   options set by Configure in the top level Makefile, and Configure
18028   was not even able to write more than one option correctly.
18029   Fixed, now "no-idea no-rc5 -DCRYPTO_MDEBUG" etc. works as intended.
18030
18031   *Bodo Moeller*
18032
18033 * New functions CONF_load_bio() and CONF_load_fp() to allow a config
18034   file to be loaded from a BIO or FILE pointer. The BIO version will
18035   for example allow memory BIOs to contain config info.
18036
18037   *Steve Henson*
18038
18039 * New function "CRYPTO_num_locks" that returns CRYPTO_NUM_LOCKS.
18040   Whoever hopes to achieve shared-library compatibility across versions
18041   must use this, not the compile-time macro.
18042   (Exercise 0.9.4: Which is the minimum library version required by
18043   such programs?)
18044   Note: All this applies only to multi-threaded programs, others don't
18045   need locks.
18046
18047   *Bodo Moeller*
18048
18049 * Add missing case to s3_clnt.c state machine -- one of the new SSL tests
18050   through a BIO pair triggered the default case, i.e.
18051   SSLerr(...,SSL_R_UNKNOWN_STATE).
18052
18053   *Bodo Moeller*
18054
18055 * New "BIO pair" concept (crypto/bio/bss_bio.c) so that applications
18056   can use the SSL library even if none of the specific BIOs is
18057   appropriate.
18058
18059   *Bodo Moeller*
18060
18061 * Fix a bug in i2d_DSAPublicKey() which meant it returned the wrong value
18062   for the encoded length.
18063
18064   *Jeon KyoungHo <khjeon@sds.samsung.co.kr>*
18065
18066 * Add initial documentation of the X509V3 functions.
18067
18068   *Steve Henson*
18069
18070 * Add a new pair of functions PEM_write_PKCS8PrivateKey() and
18071   PEM_write_bio_PKCS8PrivateKey() that are equivalent to
18072   PEM_write_PrivateKey() and PEM_write_bio_PrivateKey() but use the more
18073   secure PKCS#8 private key format with a high iteration count.
18074
18075   *Steve Henson*
18076
18077 * Fix determination of Perl interpreter: A perl or perl5
18078   *directory* in $PATH was also accepted as the interpreter.
18079
18080   *Ralf S. Engelschall*
18081
18082 * Fix demos/sign/sign.c: well there wasn't anything strictly speaking
18083   wrong with it but it was very old and did things like calling
18084   PEM_ASN1_read() directly and used MD5 for the hash not to mention some
18085   unusual formatting.
18086
18087   *Steve Henson*
18088
18089 * Fix demos/selfsign.c: it used obsolete and deleted functions, changed
18090   to use the new extension code.
18091
18092   *Steve Henson*
18093
18094 * Implement the PEM_read/PEM_write functions in crypto/pem/pem_all.c
18095   with macros. This should make it easier to change their form, add extra
18096   arguments etc. Fix a few PEM prototypes which didn't have cipher as a
18097   constant.
18098
18099   *Steve Henson*
18100
18101 * Add to configuration table a new entry that can specify an alternative
18102   name for unistd.h (for pre-POSIX systems); we need this for NeXTstep,
18103   according to Mark Crispin <MRC@Panda.COM>.
18104
18105   *Bodo Moeller*
18106
18107 * DES CBC did not update the IV. Weird.
18108
18109   *Ben Laurie*
18110lse
18111   des_cbc_encrypt does not update the IV, but des_ncbc_encrypt does.
18112   Changing the behaviour of the former might break existing programs --
18113   where IV updating is needed, des_ncbc_encrypt can be used.
18114ndif
18115
18116 * When bntest is run from "make test" it drives bc to check its
18117   calculations, as well as internally checking them. If an internal check
18118   fails, it needs to cause bc to give a non-zero result or make test carries
18119   on without noticing the failure. Fixed.
18120
18121   *Ben Laurie*
18122
18123 * DES library cleanups.
18124
18125   *Ulf Möller*
18126
18127 * Add support for PKCS#5 v2.0 PBE algorithms. This will permit PKCS#8 to be
18128   used with any cipher unlike PKCS#5 v1.5 which can at most handle 64 bit
18129   ciphers. NOTE: although the key derivation function has been verified
18130   against some published test vectors it has not been extensively tested
18131   yet. Added a -v2 "cipher" option to pkcs8 application to allow the use
18132   of v2.0.
18133
18134   *Steve Henson*
18135
18136 * Instead of "mkdir -p", which is not fully portable, use new
18137   Perl script "util/mkdir-p.pl".
18138
18139   *Bodo Moeller*
18140
18141 * Rewrite the way password based encryption (PBE) is handled. It used to
18142   assume that the ASN1 AlgorithmIdentifier parameter was a PBEParameter
18143   structure. This was true for the PKCS#5 v1.5 and PKCS#12 PBE algorithms
18144   but doesn't apply to PKCS#5 v2.0 where it can be something else. Now
18145   the 'parameter' field of the AlgorithmIdentifier is passed to the
18146   underlying key generation function so it must do its own ASN1 parsing.
18147   This has also changed the EVP_PBE_CipherInit() function which now has a
18148   'parameter' argument instead of literal salt and iteration count values
18149   and the function EVP_PBE_ALGOR_CipherInit() has been deleted.
18150
18151   *Steve Henson*
18152
18153 * Support for PKCS#5 v1.5 compatible password based encryption algorithms
18154   and PKCS#8 functionality. New 'pkcs8' application linked to openssl.
18155   Needed to change the PEM_STRING_EVP_PKEY value which was just "PRIVATE
18156   KEY" because this clashed with PKCS#8 unencrypted string. Since this
18157   value was just used as a "magic string" and not used directly its
18158   value doesn't matter.
18159
18160   *Steve Henson*
18161
18162 * Introduce some semblance of const correctness to BN. Shame C doesn't
18163   support mutable.
18164
18165   *Ben Laurie*
18166
18167 * "linux-sparc64" configuration (ultrapenguin).
18168
18169   *Ray Miller <ray.miller@oucs.ox.ac.uk>*
18170   "linux-sparc" configuration.
18171
18172   *Christian Forster <fo@hawo.stw.uni-erlangen.de>*
18173
18174 * config now generates no-xxx options for missing ciphers.
18175
18176   *Ulf Möller*
18177
18178 * Support the EBCDIC character set (work in progress).
18179   File ebcdic.c not yet included because it has a different license.
18180
18181   *Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>*
18182
18183 * Support BS2000/OSD-POSIX.
18184
18185   *Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>*
18186
18187 * Make callbacks for key generation use `void *` instead of `char *`.
18188
18189   *Ben Laurie*
18190
18191 * Make S/MIME samples compile (not yet tested).
18192
18193   *Ben Laurie*
18194
18195 * Additional typesafe stacks.
18196
18197   *Ben Laurie*
18198
18199 * New configuration variants "bsdi-elf-gcc" (BSD/OS 4.x).
18200
18201   *Bodo Moeller*
18202
18203### Changes between 0.9.3 and 0.9.3a  [29 May 1999]
18204
18205 * New configuration variant "sco5-gcc".
18206
18207 * Updated some demos.
18208
18209   *Sean O Riordain, Wade Scholine*
18210
18211 * Add missing BIO_free at exit of pkcs12 application.
18212
18213   *Wu Zhigang*
18214
18215 * Fix memory leak in conf.c.
18216
18217   *Steve Henson*
18218
18219 * Updates for Win32 to assembler version of MD5.
18220
18221   *Steve Henson*
18222
18223 * Set #! path to perl in `apps/der_chop` to where we found it
18224   instead of using a fixed path.
18225
18226   *Bodo Moeller*
18227
18228 * SHA library changes for irix64-mips4-cc.
18229
18230   *Andy Polyakov*
18231
18232 * Improvements for VMS support.
18233
18234   *Richard Levitte*
18235
18236### Changes between 0.9.2b and 0.9.3  [24 May 1999]
18237
18238 * Bignum library bug fix. IRIX 6 passes "make test" now!
18239   This also avoids the problems with SC4.2 and unpatched SC5.
18240
18241   *Andy Polyakov <appro@fy.chalmers.se>*
18242
18243 * New functions sk_num, sk_value and sk_set to replace the previous macros.
18244   These are required because of the typesafe stack would otherwise break
18245   existing code. If old code used a structure member which used to be STACK
18246   and is now STACK_OF (for example cert in a PKCS7_SIGNED structure) with
18247   sk_num or sk_value it would produce an error because the num, data members
18248   are not present in STACK_OF. Now it just produces a warning. sk_set
18249   replaces the old method of assigning a value to sk_value
18250   (e.g. sk_value(x, i) = y) which the library used in a few cases. Any code
18251   that does this will no longer work (and should use sk_set instead) but
18252   this could be regarded as a "questionable" behaviour anyway.
18253
18254   *Steve Henson*
18255
18256 * Fix most of the other PKCS#7 bugs. The "experimental" code can now
18257   correctly handle encrypted S/MIME data.
18258
18259   *Steve Henson*
18260
18261 * Change type of various DES function arguments from des_cblock
18262   (which means, in function argument declarations, pointer to char)
18263   to des_cblock * (meaning pointer to array with 8 char elements),
18264   which allows the compiler to do more typechecking; it was like
18265   that back in SSLeay, but with lots of ugly casts.
18266
18267   Introduce new type const_des_cblock.
18268
18269   *Bodo Moeller*
18270
18271 * Reorganise the PKCS#7 library and get rid of some of the more obvious
18272   problems: find RecipientInfo structure that matches recipient certificate
18273   and initialise the ASN1 structures properly based on passed cipher.
18274
18275   *Steve Henson*
18276
18277 * Belatedly make the BN tests actually check the results.
18278
18279   *Ben Laurie*
18280
18281 * Fix the encoding and decoding of negative ASN1 INTEGERS and conversion
18282   to and from BNs: it was completely broken. New compilation option
18283   NEG_PUBKEY_BUG to allow for some broken certificates that encode public
18284   key elements as negative integers.
18285
18286   *Steve Henson*
18287
18288 * Reorganize and speed up MD5.
18289
18290   *Andy Polyakov <appro@fy.chalmers.se>*
18291
18292 * VMS support.
18293
18294   *Richard Levitte <richard@levitte.org>*
18295
18296 * New option -out to asn1parse to allow the parsed structure to be
18297   output to a file. This is most useful when combined with the -strparse
18298   option to examine the output of things like OCTET STRINGS.
18299
18300   *Steve Henson*
18301
18302 * Make SSL library a little more fool-proof by not requiring any longer
18303   that `SSL_set_{accept,connect}_state` be called before
18304   `SSL_{accept,connect}` may be used (`SSL_set_..._state` is omitted
18305   in many applications because usually everything *appeared* to work as
18306   intended anyway -- now it really works as intended).
18307
18308   *Bodo Moeller*
18309
18310 * Move openssl.cnf out of lib/.
18311
18312   *Ulf Möller*
18313
18314 * Fix various things to let OpenSSL even pass "egcc -pipe -O2 -Wall
18315   -Wshadow -Wpointer-arith -Wcast-align -Wmissing-prototypes
18316   -Wmissing-declarations -Wnested-externs -Winline" with EGCS 1.1.2+
18317
18318   *Ralf S. Engelschall*
18319
18320 * Various fixes to the EVP and PKCS#7 code. It may now be able to
18321   handle PKCS#7 enveloped data properly.
18322
18323   *Sebastian Akerman <sak@parallelconsulting.com>, modified by Steve*
18324
18325 * Create a duplicate of the SSL_CTX's CERT in SSL_new instead of
18326   copying pointers.  The cert_st handling is changed by this in
18327   various ways (and thus what used to be known as ctx->default_cert
18328   is now called ctx->cert, since we don't resort to `s->ctx->[default_]cert`
18329   any longer when s->cert does not give us what we need).
18330   ssl_cert_instantiate becomes obsolete by this change.
18331   As soon as we've got the new code right (possibly it already is?),
18332   we have solved a couple of bugs of the earlier code where s->cert
18333   was used as if it could not have been shared with other SSL structures.
18334
18335   Note that using the SSL API in certain dirty ways now will result
18336   in different behaviour than observed with earlier library versions:
18337   Changing settings for an `SSL_CTX *ctx` after having done s = SSL_new(ctx)
18338   does not influence s as it used to.
18339
18340   In order to clean up things more thoroughly, inside SSL_SESSION
18341   we don't use CERT any longer, but a new structure SESS_CERT
18342   that holds per-session data (if available); currently, this is
18343   the peer's certificate chain and, for clients, the server's certificate
18344   and temporary key.  CERT holds only those values that can have
18345   meaningful defaults in an SSL_CTX.
18346
18347   *Bodo Moeller*
18348
18349 * New function X509V3_EXT_i2d() to create an X509_EXTENSION structure
18350   from the internal representation. Various PKCS#7 fixes: remove some
18351   evil casts and set the enc_dig_alg field properly based on the signing
18352   key type.
18353
18354   *Steve Henson*
18355
18356 * Allow PKCS#12 password to be set from the command line or the
18357   environment. Let 'ca' get its config file name from the environment
18358   variables "OPENSSL_CONF" or "SSLEAY_CONF" (for consistency with 'req'
18359   and 'x509').
18360
18361   *Steve Henson*
18362
18363 * Allow certificate policies extension to use an IA5STRING for the
18364   organization field. This is contrary to the PKIX definition but
18365   VeriSign uses it and IE5 only recognises this form. Document 'x509'
18366   extension option.
18367
18368   *Steve Henson*
18369
18370 * Add PEDANTIC compiler flag to allow compilation with gcc -pedantic,
18371   without disallowing inline assembler and the like for non-pedantic builds.
18372
18373   *Ben Laurie*
18374
18375 * Support Borland C++ builder.
18376
18377   *Janez Jere <jj@void.si>, modified by Ulf Möller*
18378
18379 * Support Mingw32.
18380
18381   *Ulf Möller*
18382
18383 * SHA-1 cleanups and performance enhancements.
18384
18385   *Andy Polyakov <appro@fy.chalmers.se>*
18386
18387 * Sparc v8plus assembler for the bignum library.
18388
18389   *Andy Polyakov <appro@fy.chalmers.se>*
18390
18391 * Accept any -xxx and +xxx compiler options in Configure.
18392
18393   *Ulf Möller*
18394
18395 * Update HPUX configuration.
18396
18397   *Anonymous*
18398
18399 * Add missing `sk_<type>_unshift()` function to safestack.h
18400
18401   *Ralf S. Engelschall*
18402
18403 * New function SSL_CTX_use_certificate_chain_file that sets the
18404   "extra_cert"s in addition to the certificate.  (This makes sense
18405   only for "PEM" format files, as chains as a whole are not
18406   DER-encoded.)
18407
18408   *Bodo Moeller*
18409
18410 * Support verify_depth from the SSL API.
18411   x509_vfy.c had what can be considered an off-by-one-error:
18412   Its depth (which was not part of the external interface)
18413   was actually counting the number of certificates in a chain;
18414   now it really counts the depth.
18415
18416   *Bodo Moeller*
18417
18418 * Bugfix in crypto/x509/x509_cmp.c: The SSLerr macro was used
18419   instead of X509err, which often resulted in confusing error
18420   messages since the error codes are not globally unique
18421   (e.g. an alleged error in ssl3_accept when a certificate
18422   didn't match the private key).
18423
18424 * New function SSL_CTX_set_session_id_context that allows to set a default
18425   value (so that you don't need SSL_set_session_id_context for each
18426   connection using the SSL_CTX).
18427
18428   *Bodo Moeller*
18429
18430 * OAEP decoding bug fix.
18431
18432   *Ulf Möller*
18433
18434 * Support INSTALL_PREFIX for package builders, as proposed by
18435   David Harris.
18436
18437   *Bodo Moeller*
18438
18439 * New Configure options "threads" and "no-threads".  For systems
18440   where the proper compiler options are known (currently Solaris
18441   and Linux), "threads" is the default.
18442
18443   *Bodo Moeller*
18444
18445 * New script util/mklink.pl as a faster substitute for util/mklink.sh.
18446
18447   *Bodo Moeller*
18448
18449 * Install various scripts to $(OPENSSLDIR)/misc, not to
18450   $(INSTALLTOP)/bin -- they shouldn't clutter directories
18451   such as /usr/local/bin.
18452
18453   *Bodo Moeller*
18454
18455 * "make linux-shared" to build shared libraries.
18456
18457   *Niels Poppe <niels@netbox.org>*
18458
18459 * New Configure option `no-<cipher>` (rsa, idea, rc5, ...).
18460
18461   *Ulf Möller*
18462
18463 * Add the PKCS#12 API documentation to openssl.txt. Preliminary support for
18464   extension adding in x509 utility.
18465
18466   *Steve Henson*
18467
18468 * Remove NOPROTO sections and error code comments.
18469
18470   *Ulf Möller*
18471
18472 * Partial rewrite of the DEF file generator to now parse the ANSI
18473   prototypes.
18474
18475   *Steve Henson*
18476
18477 * New Configure options --prefix=DIR and --openssldir=DIR.
18478
18479   *Ulf Möller*
18480
18481 * Complete rewrite of the error code script(s). It is all now handled
18482   by one script at the top level which handles error code gathering,
18483   header rewriting and C source file generation. It should be much better
18484   than the old method: it now uses a modified version of Ulf's parser to
18485   read the ANSI prototypes in all header files (thus the old K&R definitions
18486   aren't needed for error creation any more) and do a better job of
18487   translating function codes into names. The old 'ASN1 error code embedded
18488   in a comment' is no longer necessary and it doesn't use .err files which
18489   have now been deleted. Also the error code call doesn't have to appear all
18490   on one line (which resulted in some large lines...).
18491
18492   *Steve Henson*
18493
18494 * Change #include filenames from `<foo.h>` to `<openssl/foo.h>`.
18495
18496   *Bodo Moeller*
18497
18498 * Change behaviour of ssl2_read when facing length-0 packets: Don't return
18499   0 (which usually indicates a closed connection), but continue reading.
18500
18501   *Bodo Moeller*
18502
18503 * Fix some race conditions.
18504
18505   *Bodo Moeller*
18506
18507 * Add support for CRL distribution points extension. Add Certificate
18508   Policies and CRL distribution points documentation.
18509
18510   *Steve Henson*
18511
18512 * Move the autogenerated header file parts to crypto/opensslconf.h.
18513
18514   *Ulf Möller*
18515
18516 * Fix new 56-bit DES export ciphersuites: they were using 7 bytes instead of
18517   8 of keying material. Merlin has also confirmed interop with this fix
18518   between OpenSSL and Baltimore C/SSL 2.0 and J/SSL 2.0.
18519
18520   *Merlin Hughes <merlin@baltimore.ie>*
18521
18522 * Fix lots of warnings.
18523
18524   *Richard Levitte <levitte@stacken.kth.se>*
18525
18526 * In add_cert_dir() in crypto/x509/by_dir.c, break out of the loop if
18527   the directory spec didn't end with a LIST_SEPARATOR_CHAR.
18528
18529   *Richard Levitte <levitte@stacken.kth.se>*
18530
18531 * Fix problems with sizeof(long) == 8.
18532
18533   *Andy Polyakov <appro@fy.chalmers.se>*
18534
18535 * Change functions to ANSI C.
18536
18537   *Ulf Möller*
18538
18539 * Fix typos in error codes.
18540
18541   *Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>, Ulf Möller*
18542
18543 * Remove defunct assembler files from Configure.
18544
18545   *Ulf Möller*
18546
18547 * SPARC v8 assembler BIGNUM implementation.
18548
18549   *Andy Polyakov <appro@fy.chalmers.se>*
18550
18551 * Support for Certificate Policies extension: both print and set.
18552   Various additions to support the r2i method this uses.
18553
18554   *Steve Henson*
18555
18556 * A lot of constification, and fix a bug in X509_NAME_oneline() that could
18557   return a const string when you are expecting an allocated buffer.
18558
18559   *Ben Laurie*
18560
18561 * Add support for ASN1 types UTF8String and VISIBLESTRING, also the CHOICE
18562   types DirectoryString and DisplayText.
18563
18564   *Steve Henson*
18565
18566 * Add code to allow r2i extensions to access the configuration database,
18567   add an LHASH database driver and add several ctx helper functions.
18568
18569   *Steve Henson*
18570
18571 * Fix an evil bug in bn_expand2() which caused various BN functions to
18572   fail when they extended the size of a BIGNUM.
18573
18574   *Steve Henson*
18575
18576 * Various utility functions to handle SXNet extension. Modify mkdef.pl to
18577   support typesafe stack.
18578
18579   *Steve Henson*
18580
18581 * Fix typo in SSL_[gs]et_options().
18582
18583   *Nils Frostberg <nils@medcom.se>*
18584
18585 * Delete various functions and files that belonged to the (now obsolete)
18586   old X509V3 handling code.
18587
18588   *Steve Henson*
18589
18590 * New Configure option "rsaref".
18591
18592   *Ulf Möller*
18593
18594 * Don't auto-generate pem.h.
18595
18596   *Bodo Moeller*
18597
18598 * Introduce type-safe ASN.1 SETs.
18599
18600   *Ben Laurie*
18601
18602 * Convert various additional casted stacks to type-safe STACK_OF() variants.
18603
18604   *Ben Laurie, Ralf S. Engelschall, Steve Henson*
18605
18606 * Introduce type-safe STACKs. This will almost certainly break lots of code
18607   that links with OpenSSL (well at least cause lots of warnings), but fear
18608   not: the conversion is trivial, and it eliminates loads of evil casts. A
18609   few STACKed things have been converted already. Feel free to convert more.
18610   In the fullness of time, I'll do away with the STACK type altogether.
18611
18612   *Ben Laurie*
18613
18614 * Add `openssl ca -revoke <certfile>` facility which revokes a certificate
18615   specified in `<certfile>` by updating the entry in the index.txt file.
18616   This way one no longer has to edit the index.txt file manually for
18617   revoking a certificate. The -revoke option does the gory details now.
18618
18619   *Massimiliano Pala <madwolf@openca.org>, Ralf S. Engelschall*
18620
18621 * Fix `openssl crl -noout -text` combination where `-noout` killed the
18622   `-text` option at all and this way the `-noout -text` combination was
18623   inconsistent in `openssl crl` with the friends in `openssl x509|rsa|dsa`.
18624
18625   *Ralf S. Engelschall*
18626
18627 * Make sure a corresponding plain text error message exists for the
18628   X509_V_ERR_CERT_REVOKED/23 error number which can occur when a
18629   verify callback function determined that a certificate was revoked.
18630
18631   *Ralf S. Engelschall*
18632
18633 * Bugfix: In test/testenc, don't test `openssl <cipher>` for
18634   ciphers that were excluded, e.g. by -DNO_IDEA.  Also, test
18635   all available ciphers including rc5, which was forgotten until now.
18636   In order to let the testing shell script know which algorithms
18637   are available, a new (up to now undocumented) command
18638   `openssl list-cipher-commands` is used.
18639
18640   *Bodo Moeller*
18641
18642 * Bugfix: s_client occasionally would sleep in select() when
18643   it should have checked SSL_pending() first.
18644
18645   *Bodo Moeller*
18646
18647 * New functions DSA_do_sign and DSA_do_verify to provide access to
18648   the raw DSA values prior to ASN.1 encoding.
18649
18650   *Ulf Möller*
18651
18652 * Tweaks to Configure
18653
18654   *Niels Poppe <niels@netbox.org>*
18655
18656 * Add support for PKCS#5 v2.0 ASN1 PBES2 structures. No other support,
18657   yet...
18658
18659   *Steve Henson*
18660
18661 * New variables $(RANLIB) and $(PERL) in the Makefiles.
18662
18663   *Ulf Möller*
18664
18665 * New config option to avoid instructions that are illegal on the 80386.
18666   The default code is faster, but requires at least a 486.
18667
18668   *Ulf Möller*
18669
18670 * Got rid of old SSL2_CLIENT_VERSION (inconsistently used) and
18671   SSL2_SERVER_VERSION (not used at all) macros, which are now the
18672   same as SSL2_VERSION anyway.
18673
18674   *Bodo Moeller*
18675
18676 * New "-showcerts" option for s_client.
18677
18678   *Bodo Moeller*
18679
18680 * Still more PKCS#12 integration. Add pkcs12 application to openssl
18681   application. Various cleanups and fixes.
18682
18683   *Steve Henson*
18684
18685 * More PKCS#12 integration. Add new pkcs12 directory with Makefile.ssl and
18686   modify error routines to work internally. Add error codes and PBE init
18687   to library startup routines.
18688
18689   *Steve Henson*
18690
18691 * Further PKCS#12 integration. Added password based encryption, PKCS#8 and
18692   packing functions to asn1 and evp. Changed function names and error
18693   codes along the way.
18694
18695   *Steve Henson*
18696
18697 * PKCS12 integration: and so it begins... First of several patches to
18698   slowly integrate PKCS#12 functionality into OpenSSL. Add PKCS#12
18699   objects to objects.h
18700
18701   *Steve Henson*
18702
18703 * Add a new 'indent' option to some X509V3 extension code. Initial ASN1
18704   and display support for Thawte strong extranet extension.
18705
18706   *Steve Henson*
18707
18708 * Add LinuxPPC support.
18709
18710   *Jeff Dubrule <igor@pobox.org>*
18711
18712 * Get rid of redundant BN file bn_mulw.c, and rename bn_div64 to
18713   bn_div_words in alpha.s.
18714
18715   *Hannes Reinecke <H.Reinecke@hw.ac.uk> and Ben Laurie*
18716
18717 * Make sure the RSA OAEP test is skipped under -DRSAref because
18718   OAEP isn't supported when OpenSSL is built with RSAref.
18719
18720   *Ulf Moeller <ulf@fitug.de>*
18721
18722 * Move definitions of IS_SET/IS_SEQUENCE inside crypto/asn1/asn1.h
18723   so they no longer are missing under -DNOPROTO.
18724
18725   *Soren S. Jorvang <soren@t.dk>*
18726
18727### Changes between 0.9.1c and 0.9.2b  [22 Mar 1999]
18728
18729 * Make SSL_get_peer_cert_chain() work in servers. Unfortunately, it still
18730   doesn't work when the session is reused. Coming soon!
18731
18732   *Ben Laurie*
18733
18734 * Fix a security hole, that allows sessions to be reused in the wrong
18735   context thus bypassing client cert protection! All software that uses
18736   client certs and session caches in multiple contexts NEEDS PATCHING to
18737   allow session reuse! A fuller solution is in the works.
18738
18739   *Ben Laurie, problem pointed out by Holger Reif, Bodo Moeller (and ???)*
18740
18741 * Some more source tree cleanups (removed obsolete files
18742   crypto/bf/asm/bf586.pl, test/test.txt and crypto/sha/asm/f.s; changed
18743   permission on "config" script to be executable) and a fix for the INSTALL
18744   document.
18745
18746   *Ulf Moeller <ulf@fitug.de>*
18747
18748 * Remove some legacy and erroneous uses of malloc, free instead of
18749   Malloc, Free.
18750
18751   *Lennart Bang <lob@netstream.se>, with minor changes by Steve*
18752
18753 * Make rsa_oaep_test return non-zero on error.
18754
18755   *Ulf Moeller <ulf@fitug.de>*
18756
18757 * Add support for native Solaris shared libraries. Configure
18758   solaris-sparc-sc4-pic, make, then run shlib/solaris-sc4.sh. It'd be nice
18759   if someone would make that last step automatic.
18760
18761   *Matthias Loepfe <Matthias.Loepfe@AdNovum.CH>*
18762
18763 * ctx_size was not built with the right compiler during "make links". Fixed.
18764
18765   *Ben Laurie*
18766
18767 * Change the meaning of 'ALL' in the cipher list. It now means "everything
18768   except NULL ciphers". This means the default cipher list will no longer
18769   enable NULL ciphers. They need to be specifically enabled e.g. with
18770   the string "DEFAULT:eNULL".
18771
18772   *Steve Henson*
18773
18774 * Fix to RSA private encryption routines: if p < q then it would
18775   occasionally produce an invalid result. This will only happen with
18776   externally generated keys because OpenSSL (and SSLeay) ensure p > q.
18777
18778   *Steve Henson*
18779
18780 * Be less restrictive and allow also `perl util/perlpath.pl
18781   /path/to/bin/perl` in addition to `perl util/perlpath.pl /path/to/bin`,
18782   because this way one can also use an interpreter named `perl5` (which is
18783   usually the name of Perl 5.xxx on platforms where an Perl 4.x is still
18784   installed as `perl`).
18785
18786   *Matthias Loepfe <Matthias.Loepfe@adnovum.ch>*
18787
18788 * Let util/clean-depend.pl work also with older Perl 5.00x versions.
18789
18790   *Matthias Loepfe <Matthias.Loepfe@adnovum.ch>*
18791
18792 * Fix Makefile.org so CC,CFLAG etc are passed to 'make links' add
18793   advapi32.lib to Win32 build and change the pem test comparison
18794   to fc.exe (thanks to Ulrich Kroener <kroneru@yahoo.com> for the
18795   suggestion). Fix misplaced ASNI prototypes and declarations in evp.h
18796   and crypto/des/ede_cbcm_enc.c.
18797
18798   *Steve Henson*
18799
18800 * DES quad checksum was broken on big-endian architectures. Fixed.
18801
18802   *Ben Laurie*
18803
18804 * Comment out two functions in bio.h that aren't implemented. Fix up the
18805   Win32 test batch file so it (might) work again. The Win32 test batch file
18806   is horrible: I feel ill....
18807
18808   *Steve Henson*
18809
18810 * Move various #ifdefs around so NO_SYSLOG, NO_DIRENT etc are now selected
18811   in e_os.h. Audit of header files to check ANSI and non ANSI
18812   sections: 10 functions were absent from non ANSI section and not exported
18813   from Windows DLLs. Fixed up libeay.num for new functions.
18814
18815   *Steve Henson*
18816
18817 * Make `openssl version` output lines consistent.
18818
18819   *Ralf S. Engelschall*
18820
18821 * Fix Win32 symbol export lists for BIO functions: Added
18822   BIO_get_ex_new_index, BIO_get_ex_num, BIO_get_ex_data and BIO_set_ex_data
18823   to ms/libeay{16,32}.def.
18824
18825   *Ralf S. Engelschall*
18826
18827 * Second round of fixing the OpenSSL perl/ stuff. It now at least compiled
18828   fine under Unix and passes some trivial tests I've now added. But the
18829   whole stuff is horribly incomplete, so a README.1ST with a disclaimer was
18830   added to make sure no one expects that this stuff really works in the
18831   OpenSSL 0.9.2 release.  Additionally I've started to clean the XS sources
18832   up and fixed a few little bugs and inconsistencies in OpenSSL.{pm,xs} and
18833   openssl_bio.xs.
18834
18835   *Ralf S. Engelschall*
18836
18837 * Fix the generation of two part addresses in perl.
18838
18839   *Kenji Miyake <kenji@miyake.org>, integrated by Ben Laurie*
18840
18841 * Add config entry for Linux on MIPS.
18842
18843   *John Tobey <jtobey@channel1.com>*
18844
18845 * Make links whenever Configure is run, unless we are on Windoze.
18846
18847   *Ben Laurie*
18848
18849 * Permit extensions to be added to CRLs using crl_section in openssl.cnf.
18850   Currently only issuerAltName and AuthorityKeyIdentifier make any sense
18851   in CRLs.
18852
18853   *Steve Henson*
18854
18855 * Add a useful kludge to allow package maintainers to specify compiler and
18856   other platforms details on the command line without having to patch the
18857   Configure script every time: One now can use
18858   `perl Configure <id>:<details>`,
18859   i.e. platform ids are allowed to have details appended
18860   to them (separated by colons). This is treated as there would be a static
18861   pre-configured entry in Configure's %table under key `<id>` with value
18862   `<details>` and `perl Configure <id>` is called.  So, when you want to
18863   perform a quick test-compile under FreeBSD 3.1 with pgcc and without
18864   assembler stuff you can use `perl Configure "FreeBSD-elf:pgcc:-O6:::"`
18865   now, which overrides the FreeBSD-elf entry on-the-fly.
18866
18867   *Ralf S. Engelschall*
18868
18869 * Disable new TLS1 ciphersuites by default: they aren't official yet.
18870
18871   *Ben Laurie*
18872
18873 * Allow DSO flags like -fpic, -fPIC, -KPIC etc. to be specified
18874   on the `perl Configure ...` command line. This way one can compile
18875   OpenSSL libraries with Position Independent Code (PIC) which is needed
18876   for linking it into DSOs.
18877
18878   *Ralf S. Engelschall*
18879
18880 * Remarkably, export ciphers were totally broken and no-one had noticed!
18881   Fixed.
18882
18883   *Ben Laurie*
18884
18885 * Cleaned up the LICENSE document: The official contact for any license
18886   questions now is the OpenSSL core team under openssl-core@openssl.org.
18887   And add a paragraph about the dual-license situation to make sure people
18888   recognize that _BOTH_ the OpenSSL license _AND_ the SSLeay license apply
18889   to the OpenSSL toolkit.
18890
18891   *Ralf S. Engelschall*
18892
18893 * General source tree makefile cleanups: Made `making xxx in yyy...`
18894   display consistent in the source tree and replaced `/bin/rm` by `rm`.
18895   Additionally cleaned up the `make links` target: Remove unnecessary
18896   semicolons, subsequent redundant removes, inline point.sh into mklink.sh
18897   to speed processing and no longer clutter the display with confusing
18898   stuff. Instead only the actually done links are displayed.
18899
18900   *Ralf S. Engelschall*
18901
18902 * Permit null encryption ciphersuites, used for authentication only. It used
18903   to be necessary to set the preprocessor define SSL_ALLOW_ENULL to do this.
18904   It is now necessary to set SSL_FORBID_ENULL to prevent the use of null
18905   encryption.
18906
18907   *Ben Laurie*
18908
18909 * Add a bunch of fixes to the PKCS#7 stuff. It used to sometimes reorder
18910   signed attributes when verifying signatures (this would break them),
18911   the detached data encoding was wrong and public keys obtained using
18912   X509_get_pubkey() weren't freed.
18913
18914   *Steve Henson*
18915
18916 * Add text documentation for the BUFFER functions. Also added a work around
18917   to a Win95 console bug. This was triggered by the password read stuff: the
18918   last character typed gets carried over to the next fread(). If you were
18919   generating a new cert request using 'req' for example then the last
18920   character of the passphrase would be CR which would then enter the first
18921   field as blank.
18922
18923   *Steve Henson*
18924
18925 * Added the new 'Includes OpenSSL Cryptography Software' button as
18926   doc/openssl_button.{gif,html} which is similar in style to the old SSLeay
18927   button and can be used by applications based on OpenSSL to show the
18928   relationship to the OpenSSL project.
18929
18930   *Ralf S. Engelschall*
18931
18932 * Remove confusing variables in function signatures in files
18933   ssl/ssl_lib.c and ssl/ssl.h.
18934
18935   *Lennart Bong <lob@kulthea.stacken.kth.se>*
18936
18937 * Don't install bss_file.c under PREFIX/include/
18938
18939   *Lennart Bong <lob@kulthea.stacken.kth.se>*
18940
18941 * Get the Win32 compile working again. Modify mkdef.pl so it can handle
18942   functions that return function pointers and has support for NT specific
18943   stuff. Fix mk1mf.pl and VC-32.pl to support NT differences also. Various
18944   #ifdef WIN32 and WINNTs sprinkled about the place and some changes from
18945   unsigned to signed types: this was killing the Win32 compile.
18946
18947   *Steve Henson*
18948
18949 * Add new certificate file to stack functions,
18950   SSL_add_dir_cert_subjects_to_stack() and
18951   SSL_add_file_cert_subjects_to_stack().  These largely supplant
18952   SSL_load_client_CA_file(), and can be used to add multiple certs easily
18953   to a stack (usually this is then handed to SSL_CTX_set_client_CA_list()).
18954   This means that Apache-SSL and similar packages don't have to mess around
18955   to add as many CAs as they want to the preferred list.
18956
18957   *Ben Laurie*
18958
18959 * Experiment with doxygen documentation. Currently only partially applied to
18960   ssl/ssl_lib.c.
18961   See <http://www.stack.nl/~dimitri/doxygen/index.html>, and run doxygen with
18962   openssl.doxy as the configuration file.
18963
18964   *Ben Laurie*
18965
18966 * Get rid of remaining C++-style comments which strict C compilers hate.
18967
18968   *Ralf S. Engelschall, pointed out by Carlos Amengual*
18969
18970 * Changed BN_RECURSION in bn_mont.c to BN_RECURSION_MONT so it is not
18971   compiled in by default: it has problems with large keys.
18972
18973   *Steve Henson*
18974
18975 * Add a bunch of SSL_xxx() functions for configuring the temporary RSA and
18976   DH private keys and/or callback functions which directly correspond to
18977   their SSL_CTX_xxx() counterparts but work on a per-connection basis. This
18978   is needed for applications which have to configure certificates on a
18979   per-connection basis (e.g. Apache+mod_ssl) instead of a per-context basis
18980   (e.g. s_server).
18981      For the RSA certificate situation is makes no difference, but
18982   for the DSA certificate situation this fixes the "no shared cipher"
18983   problem where the OpenSSL cipher selection procedure failed because the
18984   temporary keys were not overtaken from the context and the API provided
18985   no way to reconfigure them.
18986      The new functions now let applications reconfigure the stuff and they
18987   are in detail: SSL_need_tmp_RSA, SSL_set_tmp_rsa, SSL_set_tmp_dh,
18988   SSL_set_tmp_rsa_callback and SSL_set_tmp_dh_callback.  Additionally a new
18989   non-public-API function ssl_cert_instantiate() is used as a helper
18990   function and also to reduce code redundancy inside ssl_rsa.c.
18991
18992   *Ralf S. Engelschall*
18993
18994 * Move s_server -dcert and -dkey options out of the undocumented feature
18995   area because they are useful for the DSA situation and should be
18996   recognized by the users.
18997
18998   *Ralf S. Engelschall*
18999
19000 * Fix the cipher decision scheme for export ciphers: the export bits are
19001   *not* within SSL_MKEY_MASK or SSL_AUTH_MASK, they are within
19002   SSL_EXP_MASK.  So, the original variable has to be used instead of the
19003   already masked variable.
19004
19005   *Richard Levitte <levitte@stacken.kth.se>*
19006
19007 * Fix `port` variable from `int` to `unsigned int` in crypto/bio/b_sock.c
19008
19009   *Richard Levitte <levitte@stacken.kth.se>*
19010
19011 * Change type of another md_len variable in pk7_doit.c:PKCS7_dataFinal()
19012   from `int` to `unsigned int` because it is a length and initialized by
19013   EVP_DigestFinal() which expects an `unsigned int *`.
19014
19015   *Richard Levitte <levitte@stacken.kth.se>*
19016
19017 * Don't hard-code path to Perl interpreter on shebang line of Configure
19018   script. Instead use the usual Shell->Perl transition trick.
19019
19020   *Ralf S. Engelschall*
19021
19022 * Make `openssl x509 -noout -modulus`' functional also for DSA certificates
19023   (in addition to RSA certificates) to match the behaviour of `openssl dsa
19024   -noout -modulus` as it's already the case for `openssl rsa -noout
19025   -modulus`.  For RSA the -modulus is the real "modulus" while for DSA
19026   currently the public key is printed (a decision which was already done by
19027   `openssl dsa -modulus` in the past) which serves a similar purpose.
19028   Additionally the NO_RSA no longer completely removes the whole -modulus
19029   option; it now only avoids using the RSA stuff. Same applies to NO_DSA
19030   now, too.
19031
19032   *Ralf S.  Engelschall*
19033
19034 * Add Arne Ansper's reliable BIO - this is an encrypted, block-digested
19035   BIO. See the source (crypto/evp/bio_ok.c) for more info.
19036
19037   *Arne Ansper <arne@ats.cyber.ee>*
19038
19039 * Dump the old yucky req code that tried (and failed) to allow raw OIDs
19040   to be added. Now both 'req' and 'ca' can use new objects defined in the
19041   config file.
19042
19043   *Steve Henson*
19044
19045 * Add cool BIO that does syslog (or event log on NT).
19046
19047   *Arne Ansper <arne@ats.cyber.ee>, integrated by Ben Laurie*
19048
19049 * Add support for new TLS ciphersuites, TLS_RSA_EXPORT56_WITH_RC4_56_MD5,
19050   TLS_RSA_EXPORT56_WITH_RC2_CBC_56_MD5 and
19051   TLS_RSA_EXPORT56_WITH_DES_CBC_SHA, as specified in "56-bit Export Cipher
19052   Suites For TLS", draft-ietf-tls-56-bit-ciphersuites-00.txt.
19053
19054   *Ben Laurie*
19055
19056 * Add preliminary config info for new extension code.
19057
19058   *Steve Henson*
19059
19060 * Make RSA_NO_PADDING really use no padding.
19061
19062   *Ulf Moeller <ulf@fitug.de>*
19063
19064 * Generate errors when private/public key check is done.
19065
19066   *Ben Laurie*
19067
19068 * Overhaul for 'crl' utility. New function X509_CRL_print. Partial support
19069   for some CRL extensions and new objects added.
19070
19071   *Steve Henson*
19072
19073 * Really fix the ASN1 IMPLICIT bug this time... Partial support for private
19074   key usage extension and fuller support for authority key id.
19075
19076   *Steve Henson*
19077
19078 * Add OAEP encryption for the OpenSSL crypto library. OAEP is the improved
19079   padding method for RSA, which is recommended for new applications in PKCS
19080   #1 v2.0 (RFC 2437, October 1998).
19081   OAEP (Optimal Asymmetric Encryption Padding) has better theoretical
19082   foundations than the ad-hoc padding used in PKCS #1 v1.5. It is secure
19083   against Bleichbacher's attack on RSA.
19084   *Ulf Moeller <ulf@fitug.de>, reformatted, corrected and integrated by
19085   Ben Laurie*
19086
19087 * Updates to the new SSL compression code
19088
19089   *Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)*
19090
19091 * Fix so that the version number in the master secret, when passed
19092   via RSA, checks that if TLS was proposed, but we roll back to SSLv3
19093   (because the server will not accept higher), that the version number
19094   is 0x03,0x01, not 0x03,0x00
19095
19096   *Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)*
19097
19098 * Run extensive memory leak checks on SSL commands. Fixed *lots* of memory
19099   leaks in `ssl/` relating to new `X509_get_pubkey()` behaviour. Also fixes
19100   in `apps/` and an unrelated leak in `crypto/dsa/dsa_vrf.c`.
19101
19102   *Steve Henson*
19103
19104 * Support for RAW extensions where an arbitrary extension can be
19105   created by including its DER encoding. See `apps/openssl.cnf` for
19106   an example.
19107
19108   *Steve Henson*
19109
19110 * Make sure latest Perl versions don't interpret some generated C array
19111   code as Perl array code in the crypto/err/err_genc.pl script.
19112
19113   *Lars Weber <3weber@informatik.uni-hamburg.de>*
19114
19115 * Modify ms/do_ms.bat to not generate assembly language makefiles since
19116   not many people have the assembler. Various Win32 compilation fixes and
19117   update to the INSTALL.W32 file with (hopefully) more accurate Win32
19118   build instructions.
19119
19120   *Steve Henson*
19121
19122 * Modify configure script 'Configure' to automatically create crypto/date.h
19123   file under Win32 and also build pem.h from pem.org. New script
19124   util/mkfiles.pl to create the MINFO file on environments that can't do a
19125   'make files': perl util/mkfiles.pl >MINFO should work.
19126
19127   *Steve Henson*
19128
19129 * Major rework of DES function declarations, in the pursuit of correctness
19130   and purity. As a result, many evil casts evaporated, and some weirdness,
19131   too. You may find this causes warnings in your code. Zapping your evil
19132   casts will probably fix them. Mostly.
19133
19134   *Ben Laurie*
19135
19136 * Fix for a typo in asn1.h. Bug fix to object creation script
19137   obj_dat.pl. It considered a zero in an object definition to mean
19138   "end of object": none of the objects in objects.h have any zeros
19139   so it wasn't spotted.
19140
19141   *Steve Henson, reported by Erwann ABALEA <eabalea@certplus.com>*
19142
19143 * Add support for Triple DES Cipher Block Chaining with Output Feedback
19144   Masking (CBCM). In the absence of test vectors, the best I have been able
19145   to do is check that the decrypt undoes the encrypt, so far. Send me test
19146   vectors if you have them.
19147
19148   *Ben Laurie*
19149
19150 * Correct calculation of key length for export ciphers (too much space was
19151   allocated for null ciphers). This has not been tested!
19152
19153   *Ben Laurie*
19154
19155 * Modifications to the mkdef.pl for Win32 DEF file creation. The usage
19156   message is now correct (it understands "crypto" and "ssl" on its
19157   command line). There is also now an "update" option. This will update
19158   the util/ssleay.num and util/libeay.num files with any new functions.
19159   If you do a:
19160   perl util/mkdef.pl crypto ssl update
19161   it will update them.
19162
19163   *Steve Henson*
19164
19165 * Overhauled the Perl interface:
19166   - ported BN stuff to OpenSSL's different BN library
19167   - made the perl/ source tree CVS-aware
19168   - renamed the package from SSLeay to OpenSSL (the files still contain
19169     their history because I've copied them in the repository)
19170   - removed obsolete files (the test scripts will be replaced
19171     by better Test::Harness variants in the future)
19172
19173   *Ralf S. Engelschall*
19174
19175 * First cut for a very conservative source tree cleanup:
19176   1. merge various obsolete readme texts into doc/ssleay.txt
19177   where we collect the old documents and readme texts.
19178   2. remove the first part of files where I'm already sure that we no
19179   longer need them because of three reasons: either they are just temporary
19180   files which were left by Eric or they are preserved original files where
19181   I've verified that the diff is also available in the CVS via "cvs diff
19182   -rSSLeay_0_8_1b" or they were renamed (as it was definitely the case for
19183   the crypto/md/ stuff).
19184
19185   *Ralf S. Engelschall*
19186
19187 * More extension code. Incomplete support for subject and issuer alt
19188   name, issuer and authority key id. Change the i2v function parameters
19189   and add an extra 'crl' parameter in the X509V3_CTX structure: guess
19190   what that's for :-) Fix to ASN1 macro which messed up
19191   IMPLICIT tag and add f_enum.c which adds a2i, i2a for ENUMERATED.
19192
19193   *Steve Henson*
19194
19195 * Preliminary support for ENUMERATED type. This is largely copied from the
19196   INTEGER code.
19197
19198   *Steve Henson*
19199
19200 * Add new function, EVP_MD_CTX_copy() to replace frequent use of memcpy.
19201
19202   *Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)*
19203
19204 * Make sure `make rehash` target really finds the `openssl` program.
19205
19206   *Ralf S. Engelschall, Matthias Loepfe <Matthias.Loepfe@adnovum.ch>*
19207
19208 * Squeeze another 7% of speed out of MD5 assembler, at least on a P2. I'd
19209   like to hear about it if this slows down other processors.
19210
19211   *Ben Laurie*
19212
19213 * Add CygWin32 platform information to Configure script.
19214
19215   *Alan Batie <batie@aahz.jf.intel.com>*
19216
19217 * Fixed ms/32all.bat script: `no_asm` -> `no-asm`
19218
19219   *Rainer W. Gerling <gerling@mpg-gv.mpg.de>*
19220
19221 * New program nseq to manipulate netscape certificate sequences
19222
19223   *Steve Henson*
19224
19225 * Modify crl2pkcs7 so it supports multiple -certfile arguments. Fix a
19226   few typos.
19227
19228   *Steve Henson*
19229
19230 * Fixes to BN code.  Previously the default was to define BN_RECURSION
19231   but the BN code had some problems that would cause failures when
19232   doing certificate verification and some other functions.
19233
19234   *Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)*
19235
19236 * Add ASN1 and PEM code to support netscape certificate sequences.
19237
19238   *Steve Henson*
19239
19240 * Add ASN1 and PEM code to support netscape certificate sequences.
19241
19242   *Steve Henson*
19243
19244 * Add several PKIX and private extended key usage OIDs.
19245
19246   *Steve Henson*
19247
19248 * Modify the 'ca' program to handle the new extension code. Modify
19249   openssl.cnf for new extension format, add comments.
19250
19251   *Steve Henson*
19252
19253 * More X509 V3 changes. Fix typo in v3_bitstr.c. Add support to 'req'
19254   and add a sample to openssl.cnf so req -x509 now adds appropriate
19255   CA extensions.
19256
19257   *Steve Henson*
19258
19259 * Continued X509 V3 changes. Add to other makefiles, integrate with the
19260   error code, add initial support to X509_print() and x509 application.
19261
19262   *Steve Henson*
19263
19264 * Takes a deep breath and start adding X509 V3 extension support code. Add
19265   files in crypto/x509v3. Move original stuff to crypto/x509v3/old. All this
19266   stuff is currently isolated and isn't even compiled yet.
19267
19268   *Steve Henson*
19269
19270 * Continuing patches for GeneralizedTime. Fix up certificate and CRL
19271   ASN1 to use ASN1_TIME and modify print routines to use ASN1_TIME_print.
19272   Removed the versions check from X509 routines when loading extensions:
19273   this allows certain broken certificates that don't set the version
19274   properly to be processed.
19275
19276   *Steve Henson*
19277
19278 * Deal with irritating shit to do with dependencies, in YAAHW (Yet Another
19279   Ad Hoc Way) - Makefile.ssls now all contain local dependencies, which
19280   can still be regenerated with "make depend".
19281
19282   *Ben Laurie*
19283
19284 * Spelling mistake in C version of CAST-128.
19285
19286   *Ben Laurie, reported by Jeremy Hylton <jeremy@cnri.reston.va.us>*
19287
19288 * Changes to the error generation code. The perl script err-code.pl
19289   now reads in the old error codes and retains the old numbers, only
19290   adding new ones if necessary. It also only changes the .err files if new
19291   codes are added. The makefiles have been modified to only insert errors
19292   when needed (to avoid needlessly modifying header files). This is done
19293   by only inserting errors if the .err file is newer than the auto generated
19294   C file. To rebuild all the error codes from scratch (the old behaviour)
19295   either modify crypto/Makefile.ssl to pass the -regen flag to err_code.pl
19296   or delete all the .err files.
19297
19298   *Steve Henson*
19299
19300 * CAST-128 was incorrectly implemented for short keys. The C version has
19301   been fixed, but is untested. The assembler versions are also fixed, but
19302   new assembler HAS NOT BEEN GENERATED FOR WIN32 - the Makefile needs fixing
19303   to regenerate it if needed.
19304   *Ben Laurie, reported (with fix for C version) by Jun-ichiro itojun
19305    Hagino <itojun@kame.net>*
19306
19307 * File was opened incorrectly in randfile.c.
19308
19309   *Ulf Möller <ulf@fitug.de>*
19310
19311 * Beginning of support for GeneralizedTime. d2i, i2d, check and print
19312   functions. Also ASN1_TIME suite which is a CHOICE of UTCTime or
19313   GeneralizedTime. ASN1_TIME is the proper type used in certificates et
19314   al: it's just almost always a UTCTime. Note this patch adds new error
19315   codes so do a "make errors" if there are problems.
19316
19317   *Steve Henson*
19318
19319 * Correct Linux 1 recognition in config.
19320
19321   *Ulf Möller <ulf@fitug.de>*
19322
19323 * Remove pointless MD5 hash when using DSA keys in ca.
19324
19325   *Anonymous <nobody@replay.com>*
19326
19327 * Generate an error if given an empty string as a cert directory. Also
19328   generate an error if handed NULL (previously returned 0 to indicate an
19329   error, but didn't set one).
19330
19331   *Ben Laurie, reported by Anonymous <nobody@replay.com>*
19332
19333 * Add prototypes to SSL methods. Make SSL_write's buffer const, at last.
19334
19335   *Ben Laurie*
19336
19337 * Fix the dummy function BN_ref_mod_exp() in rsaref.c to have the correct
19338   parameters. This was causing a warning which killed off the Win32 compile.
19339
19340   *Steve Henson*
19341
19342 * Remove C++ style comments from crypto/bn/bn_local.h.
19343
19344   *Neil Costigan <neil.costigan@celocom.com>*
19345
19346 * The function OBJ_txt2nid was broken. It was supposed to return a nid
19347   based on a text string, looking up short and long names and finally
19348   "dot" format. The "dot" format stuff didn't work. Added new function
19349   OBJ_txt2obj to do the same but return an ASN1_OBJECT and rewrote
19350   OBJ_txt2nid to use it. OBJ_txt2obj can also return objects even if the
19351   OID is not part of the table.
19352
19353   *Steve Henson*
19354
19355 * Add prototypes to X509 lookup/verify methods, fixing a bug in
19356   X509_LOOKUP_by_alias().
19357
19358   *Ben Laurie*
19359
19360 * Sort openssl functions by name.
19361
19362   *Ben Laurie*
19363
19364 * Get the `gendsa` command working and add it to the `list` command. Remove
19365   encryption from sample DSA keys (in case anyone is interested the password
19366   was "1234").
19367
19368   *Steve Henson*
19369
19370 * Make *all* `*_free` functions accept a NULL pointer.
19371
19372   *Frans Heymans <fheymans@isaserver.be>*
19373
19374 * If a DH key is generated in s3_srvr.c, don't blow it by trying to use
19375   NULL pointers.
19376
19377   *Anonymous <nobody@replay.com>*
19378
19379 * s_server should send the CAfile as acceptable CAs, not its own cert.
19380
19381   *Bodo Moeller <3moeller@informatik.uni-hamburg.de>*
19382
19383 * Don't blow it for numeric `-newkey` arguments to `apps/req`.
19384
19385   *Bodo Moeller <3moeller@informatik.uni-hamburg.de>*
19386
19387 * Temp key "for export" tests were wrong in s3_srvr.c.
19388
19389   *Anonymous <nobody@replay.com>*
19390
19391 * Add prototype for temp key callback functions
19392   SSL_CTX_set_tmp_{rsa,dh}_callback().
19393
19394   *Ben Laurie*
19395
19396 * Make DH_free() tolerate being passed a NULL pointer (like RSA_free() and
19397   DSA_free()). Make X509_PUBKEY_set() check for errors in d2i_PublicKey().
19398
19399   *Steve Henson*
19400
19401 * X509_name_add_entry() freed the wrong thing after an error.
19402
19403   *Arne Ansper <arne@ats.cyber.ee>*
19404
19405 * rsa_eay.c would attempt to free a NULL context.
19406
19407   *Arne Ansper <arne@ats.cyber.ee>*
19408
19409 * BIO_s_socket() had a broken should_retry() on Windoze.
19410
19411   *Arne Ansper <arne@ats.cyber.ee>*
19412
19413 * BIO_f_buffer() didn't pass on BIO_CTRL_FLUSH.
19414
19415   *Arne Ansper <arne@ats.cyber.ee>*
19416
19417 * Make sure the already existing X509_STORE->depth variable is initialized
19418   in X509_STORE_new(), but document the fact that this variable is still
19419   unused in the certificate verification process.
19420
19421   *Ralf S. Engelschall*
19422
19423 * Fix the various library and `apps/` files to free up pkeys obtained from
19424   X509_PUBKEY_get() et al. Also allow x509.c to handle netscape extensions.
19425
19426   *Steve Henson*
19427
19428 * Fix reference counting in X509_PUBKEY_get(). This makes
19429   demos/maurice/example2.c work, amongst others, probably.
19430
19431   *Steve Henson and Ben Laurie*
19432
19433 * First cut of a cleanup for `apps/`. First the `ssleay` program is now named
19434   `openssl` and second, the shortcut symlinks for the `openssl <command>`
19435   are no longer created. This way we have a single and consistent command
19436   line interface `openssl <command>`, similar to `cvs <command>`.
19437
19438   *Ralf S. Engelschall, Paul Sutton and Ben Laurie*
19439
19440 * ca.c: move test for DSA keys inside #ifndef NO_DSA. Make pubkey
19441   BIT STRING wrapper always have zero unused bits.
19442
19443   *Steve Henson*
19444
19445 * Add CA.pl, perl version of CA.sh, add extended key usage OID.
19446
19447   *Steve Henson*
19448
19449 * Make the top-level INSTALL documentation easier to understand.
19450
19451   *Paul Sutton*
19452
19453 * Makefiles updated to exit if an error occurs in a sub-directory
19454   make (including if user presses ^C) [Paul Sutton]
19455
19456 * Make Montgomery context stuff explicit in RSA data structure.
19457
19458   *Ben Laurie*
19459
19460 * Fix build order of pem and err to allow for generated pem.h.
19461
19462   *Ben Laurie*
19463
19464 * Fix renumbering bug in X509_NAME_delete_entry().
19465
19466   *Ben Laurie*
19467
19468 * Enhanced the err-ins.pl script so it makes the error library number
19469   global and can add a library name. This is needed for external ASN1 and
19470   other error libraries.
19471
19472   *Steve Henson*
19473
19474 * Fixed sk_insert which never worked properly.
19475
19476   *Steve Henson*
19477
19478 * Fix ASN1 macros so they can handle indefinite length constructed
19479   EXPLICIT tags. Some non standard certificates use these: they can now
19480   be read in.
19481
19482   *Steve Henson*
19483
19484 * Merged the various old/obsolete SSLeay documentation files (doc/xxx.doc)
19485   into a single doc/ssleay.txt bundle. This way the information is still
19486   preserved but no longer messes up this directory. Now it's new room for
19487   the new set of documentation files.
19488
19489   *Ralf S. Engelschall*
19490
19491 * SETs were incorrectly DER encoded. This was a major pain, because they
19492   shared code with SEQUENCEs, which aren't coded the same. This means that
19493   almost everything to do with SETs or SEQUENCEs has either changed name or
19494   number of arguments.
19495
19496   *Ben Laurie, based on a partial fix by GP Jayan <gp@nsj.co.jp>*
19497
19498 * Fix test data to work with the above.
19499
19500   *Ben Laurie*
19501
19502 * Fix the RSA header declarations that hid a bug I fixed in 0.9.0b but
19503   was already fixed by Eric for 0.9.1 it seems.
19504
19505   *Ben Laurie - pointed out by Ulf Möller <ulf@fitug.de>*
19506
19507 * Autodetect FreeBSD3.
19508
19509   *Ben Laurie*
19510
19511 * Fix various bugs in Configure. This affects the following platforms:
19512   nextstep
19513   ncr-scde
19514   unixware-2.0
19515   unixware-2.0-pentium
19516   sco5-cc.
19517
19518   *Ben Laurie*
19519
19520 * Eliminate generated files from CVS. Reorder tests to regenerate files
19521   before they are needed.
19522
19523   *Ben Laurie*
19524
19525 * Generate Makefile.ssl from Makefile.org (to keep CVS happy).
19526
19527   *Ben Laurie*
19528
19529### Changes between 0.9.1b and 0.9.1c  [23-Dec-1998]
19530
19531 * Added OPENSSL_VERSION_NUMBER to crypto/crypto.h and
19532   changed SSLeay to OpenSSL in version strings.
19533
19534   *Ralf S. Engelschall*
19535
19536 * Some fixups to the top-level documents.
19537
19538   *Paul Sutton*
19539
19540 * Fixed the nasty bug where rsaref.h was not found under compile-time
19541   because the symlink to include/ was missing.
19542
19543   *Ralf S. Engelschall*
19544
19545 * Incorporated the popular no-RSA/DSA-only patches
19546   which allow to compile an RSA-free SSLeay.
19547
19548   *Andrew Cooke / Interrader Ldt., Ralf S. Engelschall*
19549
19550 * Fixed nasty rehash problem under `make -f Makefile.ssl links`
19551   when "ssleay" is still not found.
19552
19553   *Ralf S. Engelschall*
19554
19555 * Added more platforms to Configure: Cray T3E, HPUX 11,
19556
19557   *Ralf S. Engelschall, Beckmann <beckman@acl.lanl.gov>*
19558
19559 * Updated the README file.
19560
19561   *Ralf S. Engelschall*
19562
19563 * Added various .cvsignore files in the CVS repository subdirs
19564   to make a "cvs update" really silent.
19565
19566   *Ralf S. Engelschall*
19567
19568 * Recompiled the error-definition header files and added
19569   missing symbols to the Win32 linker tables.
19570
19571   *Ralf S. Engelschall*
19572
19573 * Cleaned up the top-level documents;
19574   o new files: CHANGES and LICENSE
19575   o merged VERSION, HISTORY* and README* files a CHANGES.SSLeay
19576   o merged COPYRIGHT into LICENSE
19577   o removed obsolete TODO file
19578   o renamed MICROSOFT to INSTALL.W32
19579
19580   *Ralf S. Engelschall*
19581
19582 * Removed dummy files from the 0.9.1b source tree:
19583   crypto/asn1/x crypto/bio/cd crypto/bio/fg crypto/bio/grep crypto/bio/vi
19584   crypto/bn/asm/......add.c crypto/bn/asm/a.out crypto/dsa/f crypto/md5/f
19585   crypto/pem/gmon.out crypto/perlasm/f crypto/pkcs7/build crypto/rsa/f
19586   crypto/sha/asm/f crypto/threads/f ms/zzz ssl/f ssl/f.mak test/f
19587   util/f.mak util/pl/f util/pl/f.mak crypto/bf/bf_locl.old apps/f
19588
19589   *Ralf S. Engelschall*
19590
19591 * Added various platform portability fixes.
19592
19593   *Mark J. Cox*
19594
19595 * The Genesis of the OpenSSL rpject:
19596   We start with the latest (unreleased) SSLeay version 0.9.1b which Eric A.
19597   Young and Tim J. Hudson created while they were working for C2Net until
19598   summer 1998.
19599
19600   *The OpenSSL Project*
19601
19602### Changes between 0.9.0b and 0.9.1b  [not released]
19603
19604 * Updated a few CA certificates under certs/
19605
19606   *Eric A. Young*
19607
19608 * Changed some BIGNUM api stuff.
19609
19610   *Eric A. Young*
19611
19612 * Various platform ports: OpenBSD, Ultrix, IRIX 64bit, NetBSD,
19613   DGUX x86, Linux Alpha, etc.
19614
19615   *Eric A. Young*
19616
19617 * New COMP library [crypto/comp/] for SSL Record Layer Compression:
19618   RLE (dummy implemented) and ZLIB (really implemented when ZLIB is
19619   available).
19620
19621   *Eric A. Young*
19622
19623 * Add -strparse option to asn1pars program which parses nested
19624   binary structures
19625
19626   *Dr Stephen Henson <shenson@bigfoot.com>*
19627
19628 * Added "oid_file" to ssleay.cnf for "ca" and "req" programs.
19629
19630   *Eric A. Young*
19631
19632 * DSA fix for "ca" program.
19633
19634   *Eric A. Young*
19635
19636 * Added "-genkey" option to "dsaparam" program.
19637
19638   *Eric A. Young*
19639
19640 * Added RIPE MD160 (rmd160) message digest.
19641
19642   *Eric A. Young*
19643
19644 * Added -a (all) option to "ssleay version" command.
19645
19646   *Eric A. Young*
19647
19648 * Added PLATFORM define which is the id given to Configure.
19649
19650   *Eric A. Young*
19651
19652 * Added MemCheck_XXXX functions to crypto/mem.c for memory checking.
19653
19654   *Eric A. Young*
19655
19656 * Extended the ASN.1 parser routines.
19657
19658   *Eric A. Young*
19659
19660 * Extended BIO routines to support REUSEADDR, seek, tell, etc.
19661
19662   *Eric A. Young*
19663
19664 * Added a BN_CTX to the BN library.
19665
19666   *Eric A. Young*
19667
19668 * Fixed the weak key values in DES library
19669
19670   *Eric A. Young*
19671
19672 * Changed API in EVP library for cipher aliases.
19673
19674   *Eric A. Young*
19675
19676 * Added support for RC2/64bit cipher.
19677
19678   *Eric A. Young*
19679
19680 * Converted the lhash library to the crypto/mem.c functions.
19681
19682   *Eric A. Young*
19683
19684 * Added more recognized ASN.1 object ids.
19685
19686   *Eric A. Young*
19687
19688 * Added more RSA padding checks for SSL/TLS.
19689
19690   *Eric A. Young*
19691
19692 * Added BIO proxy/filter functionality.
19693
19694   *Eric A. Young*
19695
19696 * Added extra_certs to SSL_CTX which can be used
19697   send extra CA certificates to the client in the CA cert chain sending
19698   process. It can be configured with SSL_CTX_add_extra_chain_cert().
19699
19700   *Eric A. Young*
19701
19702 * Now Fortezza is denied in the authentication phase because
19703   this is key exchange mechanism is not supported by SSLeay at all.
19704
19705   *Eric A. Young*
19706
19707 * Additional PKCS1 checks.
19708
19709   *Eric A. Young*
19710
19711 * Support the string "TLSv1" for all TLS v1 ciphers.
19712
19713   *Eric A. Young*
19714
19715 * Added function SSL_get_ex_data_X509_STORE_CTX_idx() which gives the
19716   ex_data index of the SSL context in the X509_STORE_CTX ex_data.
19717
19718   *Eric A. Young*
19719
19720 * Fixed a few memory leaks.
19721
19722   *Eric A. Young*
19723
19724 * Fixed various code and comment typos.
19725
19726   *Eric A. Young*
19727
19728 * A minor bug in ssl/s3_clnt.c where there would always be 4 0
19729   bytes sent in the client random.
19730
19731   *Edward Bishop <ebishop@spyglass.com>*
19732
19733<!-- Links -->
19734
19735[CVE-2023-4807]: https://www.openssl.org/news/vulnerabilities.html#CVE-2023-4807
19736[CVE-2023-3817]: https://www.openssl.org/news/vulnerabilities.html#CVE-2023-3817
19737[CVE-2023-3446]: https://www.openssl.org/news/vulnerabilities.html#CVE-2023-3446
19738[CVE-2023-2975]: https://www.openssl.org/news/vulnerabilities.html#CVE-2023-2975
19739[RFC 2578 (STD 58), section 3.5]: https://datatracker.ietf.org/doc/html/rfc2578#section-3.5
19740[CVE-2023-2650]: https://www.openssl.org/news/vulnerabilities.html#CVE-2023-2650
19741[CVE-2023-1255]: https://www.openssl.org/news/vulnerabilities.html#CVE-2023-1255
19742[CVE-2023-0466]: https://www.openssl.org/news/vulnerabilities.html#CVE-2023-0466
19743[CVE-2023-0465]: https://www.openssl.org/news/vulnerabilities.html#CVE-2023-0465
19744[CVE-2023-0464]: https://www.openssl.org/news/vulnerabilities.html#CVE-2023-0464
19745[CVE-2023-0401]: https://www.openssl.org/news/vulnerabilities.html#CVE-2023-0401
19746[CVE-2023-0286]: https://www.openssl.org/news/vulnerabilities.html#CVE-2023-0286
19747[CVE-2023-0217]: https://www.openssl.org/news/vulnerabilities.html#CVE-2023-0217
19748[CVE-2023-0216]: https://www.openssl.org/news/vulnerabilities.html#CVE-2023-0216
19749[CVE-2023-0215]: https://www.openssl.org/news/vulnerabilities.html#CVE-2023-0215
19750[CVE-2022-4450]: https://www.openssl.org/news/vulnerabilities.html#CVE-2022-4450
19751[CVE-2022-4304]: https://www.openssl.org/news/vulnerabilities.html#CVE-2022-4304
19752[CVE-2022-4203]: https://www.openssl.org/news/vulnerabilities.html#CVE-2022-4203
19753[CVE-2022-3996]: https://www.openssl.org/news/vulnerabilities.html#CVE-2022-3996
19754[CVE-2022-2274]: https://www.openssl.org/news/vulnerabilities.html#CVE-2022-2274
19755[CVE-2022-2097]: https://www.openssl.org/news/vulnerabilities.html#CVE-2022-2097
19756[CVE-2020-1971]: https://www.openssl.org/news/vulnerabilities.html#CVE-2020-1971
19757[CVE-2020-1967]: https://www.openssl.org/news/vulnerabilities.html#CVE-2020-1967
19758[CVE-2019-1563]: https://www.openssl.org/news/vulnerabilities.html#CVE-2019-1563
19759[CVE-2019-1559]: https://www.openssl.org/news/vulnerabilities.html#CVE-2019-1559
19760[CVE-2019-1552]: https://www.openssl.org/news/vulnerabilities.html#CVE-2019-1552
19761[CVE-2019-1551]: https://www.openssl.org/news/vulnerabilities.html#CVE-2019-1551
19762[CVE-2019-1549]: https://www.openssl.org/news/vulnerabilities.html#CVE-2019-1549
19763[CVE-2019-1547]: https://www.openssl.org/news/vulnerabilities.html#CVE-2019-1547
19764[CVE-2019-1543]: https://www.openssl.org/news/vulnerabilities.html#CVE-2019-1543
19765[CVE-2018-5407]: https://www.openssl.org/news/vulnerabilities.html#CVE-2018-5407
19766[CVE-2018-0739]: https://www.openssl.org/news/vulnerabilities.html#CVE-2018-0739
19767[CVE-2018-0737]: https://www.openssl.org/news/vulnerabilities.html#CVE-2018-0737
19768[CVE-2018-0735]: https://www.openssl.org/news/vulnerabilities.html#CVE-2018-0735
19769[CVE-2018-0734]: https://www.openssl.org/news/vulnerabilities.html#CVE-2018-0734
19770[CVE-2018-0733]: https://www.openssl.org/news/vulnerabilities.html#CVE-2018-0733
19771[CVE-2018-0732]: https://www.openssl.org/news/vulnerabilities.html#CVE-2018-0732
19772[CVE-2017-3738]: https://www.openssl.org/news/vulnerabilities.html#CVE-2017-3738
19773[CVE-2017-3737]: https://www.openssl.org/news/vulnerabilities.html#CVE-2017-3737
19774[CVE-2017-3736]: https://www.openssl.org/news/vulnerabilities.html#CVE-2017-3736
19775[CVE-2017-3735]: https://www.openssl.org/news/vulnerabilities.html#CVE-2017-3735
19776[CVE-2017-3733]: https://www.openssl.org/news/vulnerabilities.html#CVE-2017-3733
19777[CVE-2017-3732]: https://www.openssl.org/news/vulnerabilities.html#CVE-2017-3732
19778[CVE-2017-3731]: https://www.openssl.org/news/vulnerabilities.html#CVE-2017-3731
19779[CVE-2017-3730]: https://www.openssl.org/news/vulnerabilities.html#CVE-2017-3730
19780[CVE-2016-7055]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-7055
19781[CVE-2016-7054]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-7054
19782[CVE-2016-7053]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-7053
19783[CVE-2016-7052]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-7052
19784[CVE-2016-6309]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-6309
19785[CVE-2016-6308]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-6308
19786[CVE-2016-6307]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-6307
19787[CVE-2016-6306]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-6306
19788[CVE-2016-6305]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-6305
19789[CVE-2016-6304]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-6304
19790[CVE-2016-6303]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-6303
19791[CVE-2016-6302]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-6302
19792[CVE-2016-2183]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-2183
19793[CVE-2016-2182]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-2182
19794[CVE-2016-2181]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-2181
19795[CVE-2016-2180]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-2180
19796[CVE-2016-2179]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-2179
19797[CVE-2016-2178]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-2178
19798[CVE-2016-2177]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-2177
19799[CVE-2016-2176]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-2176
19800[CVE-2016-2109]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-2109
19801[CVE-2016-2107]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-2107
19802[CVE-2016-2106]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-2106
19803[CVE-2016-2105]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-2105
19804[CVE-2016-0800]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-0800
19805[CVE-2016-0799]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-0799
19806[CVE-2016-0798]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-0798
19807[CVE-2016-0797]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-0797
19808[CVE-2016-0705]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-0705
19809[CVE-2016-0702]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-0702
19810[CVE-2016-0701]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-0701
19811[CVE-2015-3197]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-3197
19812[CVE-2015-3196]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-3196
19813[CVE-2015-3195]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-3195
19814[CVE-2015-3194]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-3194
19815[CVE-2015-3193]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-3193
19816[CVE-2015-1793]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-1793
19817[CVE-2015-1792]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-1792
19818[CVE-2015-1791]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-1791
19819[CVE-2015-1790]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-1790
19820[CVE-2015-1789]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-1789
19821[CVE-2015-1788]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-1788
19822[CVE-2015-1787]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-1787
19823[CVE-2015-0293]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0293
19824[CVE-2015-0291]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0291
19825[CVE-2015-0290]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0290
19826[CVE-2015-0289]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0289
19827[CVE-2015-0288]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0288
19828[CVE-2015-0287]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0287
19829[CVE-2015-0286]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0286
19830[CVE-2015-0285]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0285
19831[CVE-2015-0209]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0209
19832[CVE-2015-0208]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0208
19833[CVE-2015-0207]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0207
19834[CVE-2015-0206]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0206
19835[CVE-2015-0205]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0205
19836[CVE-2015-0204]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0204
19837[CVE-2014-8275]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-8275
19838[CVE-2014-5139]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-5139
19839[CVE-2014-3572]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3572
19840[CVE-2014-3571]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3571
19841[CVE-2014-3570]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3570
19842[CVE-2014-3569]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3569
19843[CVE-2014-3568]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3568
19844[CVE-2014-3567]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3567
19845[CVE-2014-3566]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3566
19846[CVE-2014-3513]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3513
19847[CVE-2014-3512]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3512
19848[CVE-2014-3511]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3511
19849[CVE-2014-3510]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3510
19850[CVE-2014-3509]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3509
19851[CVE-2014-3508]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3508
19852[CVE-2014-3507]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3507
19853[CVE-2014-3506]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3506
19854[CVE-2014-3505]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3505
19855[CVE-2014-3470]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3470
19856[CVE-2014-0224]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-0224
19857[CVE-2014-0221]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-0221
19858[CVE-2014-0195]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-0195
19859[CVE-2014-0160]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-0160
19860[CVE-2014-0076]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-0076
19861[CVE-2013-6450]: https://www.openssl.org/news/vulnerabilities.html#CVE-2013-6450
19862[CVE-2013-4353]: https://www.openssl.org/news/vulnerabilities.html#CVE-2013-4353
19863[CVE-2013-0169]: https://www.openssl.org/news/vulnerabilities.html#CVE-2013-0169
19864[CVE-2013-0166]: https://www.openssl.org/news/vulnerabilities.html#CVE-2013-0166
19865[CVE-2012-2686]: https://www.openssl.org/news/vulnerabilities.html#CVE-2012-2686
19866[CVE-2012-2333]: https://www.openssl.org/news/vulnerabilities.html#CVE-2012-2333
19867[CVE-2012-2110]: https://www.openssl.org/news/vulnerabilities.html#CVE-2012-2110
19868[CVE-2012-0884]: https://www.openssl.org/news/vulnerabilities.html#CVE-2012-0884
19869[CVE-2012-0050]: https://www.openssl.org/news/vulnerabilities.html#CVE-2012-0050
19870[CVE-2012-0027]: https://www.openssl.org/news/vulnerabilities.html#CVE-2012-0027
19871[CVE-2011-4619]: https://www.openssl.org/news/vulnerabilities.html#CVE-2011-4619
19872[CVE-2011-4577]: https://www.openssl.org/news/vulnerabilities.html#CVE-2011-4577
19873[CVE-2011-4576]: https://www.openssl.org/news/vulnerabilities.html#CVE-2011-4576
19874[CVE-2011-4109]: https://www.openssl.org/news/vulnerabilities.html#CVE-2011-4109
19875[CVE-2011-4108]: https://www.openssl.org/news/vulnerabilities.html#CVE-2011-4108
19876[CVE-2011-3210]: https://www.openssl.org/news/vulnerabilities.html#CVE-2011-3210
19877[CVE-2011-3207]: https://www.openssl.org/news/vulnerabilities.html#CVE-2011-3207
19878[CVE-2011-0014]: https://www.openssl.org/news/vulnerabilities.html#CVE-2011-0014
19879[CVE-2010-4252]: https://www.openssl.org/news/vulnerabilities.html#CVE-2010-4252
19880[CVE-2010-4180]: https://www.openssl.org/news/vulnerabilities.html#CVE-2010-4180
19881[CVE-2010-3864]: https://www.openssl.org/news/vulnerabilities.html#CVE-2010-3864
19882[CVE-2010-1633]: https://www.openssl.org/news/vulnerabilities.html#CVE-2010-1633
19883[CVE-2010-0740]: https://www.openssl.org/news/vulnerabilities.html#CVE-2010-0740
19884[CVE-2010-0433]: https://www.openssl.org/news/vulnerabilities.html#CVE-2010-0433
19885[CVE-2009-4355]: https://www.openssl.org/news/vulnerabilities.html#CVE-2009-4355
19886[CVE-2009-3555]: https://www.openssl.org/news/vulnerabilities.html#CVE-2009-3555
19887[CVE-2009-3245]: https://www.openssl.org/news/vulnerabilities.html#CVE-2009-3245
19888[CVE-2009-1386]: https://www.openssl.org/news/vulnerabilities.html#CVE-2009-1386
19889[CVE-2009-1379]: https://www.openssl.org/news/vulnerabilities.html#CVE-2009-1379
19890[CVE-2009-1378]: https://www.openssl.org/news/vulnerabilities.html#CVE-2009-1378
19891[CVE-2009-1377]: https://www.openssl.org/news/vulnerabilities.html#CVE-2009-1377
19892[CVE-2009-0789]: https://www.openssl.org/news/vulnerabilities.html#CVE-2009-0789
19893[CVE-2009-0591]: https://www.openssl.org/news/vulnerabilities.html#CVE-2009-0591
19894[CVE-2009-0590]: https://www.openssl.org/news/vulnerabilities.html#CVE-2009-0590
19895[CVE-2008-5077]: https://www.openssl.org/news/vulnerabilities.html#CVE-2008-5077
19896[CVE-2008-1678]: https://www.openssl.org/news/vulnerabilities.html#CVE-2008-1678
19897[CVE-2008-1672]: https://www.openssl.org/news/vulnerabilities.html#CVE-2008-1672
19898[CVE-2008-0891]: https://www.openssl.org/news/vulnerabilities.html#CVE-2008-0891
19899[CVE-2007-5135]: https://www.openssl.org/news/vulnerabilities.html#CVE-2007-5135
19900[CVE-2007-4995]: https://www.openssl.org/news/vulnerabilities.html#CVE-2007-4995
19901[CVE-2006-4343]: https://www.openssl.org/news/vulnerabilities.html#CVE-2006-4343
19902[CVE-2006-4339]: https://www.openssl.org/news/vulnerabilities.html#CVE-2006-4339
19903[CVE-2006-3738]: https://www.openssl.org/news/vulnerabilities.html#CVE-2006-3738
19904[CVE-2006-2940]: https://www.openssl.org/news/vulnerabilities.html#CVE-2006-2940
19905[CVE-2006-2937]: https://www.openssl.org/news/vulnerabilities.html#CVE-2006-2937
19906[CVE-2005-2969]: https://www.openssl.org/news/vulnerabilities.html#CVE-2005-2969
19907[CVE-2004-0112]: https://www.openssl.org/news/vulnerabilities.html#CVE-2004-0112
19908[CVE-2004-0079]: https://www.openssl.org/news/vulnerabilities.html#CVE-2004-0079
19909[CVE-2003-0851]: https://www.openssl.org/news/vulnerabilities.html#CVE-2003-0851
19910[CVE-2003-0545]: https://www.openssl.org/news/vulnerabilities.html#CVE-2003-0545
19911[CVE-2003-0544]: https://www.openssl.org/news/vulnerabilities.html#CVE-2003-0544
19912[CVE-2003-0543]: https://www.openssl.org/news/vulnerabilities.html#CVE-2003-0543
19913[CVE-2003-0078]: https://www.openssl.org/news/vulnerabilities.html#CVE-2003-0078
19914[CVE-2002-0659]: https://www.openssl.org/news/vulnerabilities.html#CVE-2002-0659
19915[CVE-2002-0657]: https://www.openssl.org/news/vulnerabilities.html#CVE-2002-0657
19916[CVE-2002-0656]: https://www.openssl.org/news/vulnerabilities.html#CVE-2002-0656
19917[CVE-2002-0655]: https://www.openssl.org/news/vulnerabilities.html#CVE-2002-0655
19918