xref: /freebsd/crypto/openssl/CHANGES.md (revision 2008043f386721d58158e37e0d7e50df8095942d)
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 3.0.11 and 3.0.12 [24 Oct 2023]
32
33 * Fix incorrect key and IV resizing issues when calling EVP_EncryptInit_ex2(),
34   EVP_DecryptInit_ex2() or EVP_CipherInit_ex2() with OSSL_PARAM parameters
35   that alter the key or IV length ([CVE-2023-5363]).
36
37   *Paul Dale*
38
39### Changes between 3.0.10 and 3.0.11 [19 Sep 2023]
40
41 * Fix POLY1305 MAC implementation corrupting XMM registers on Windows.
42
43   The POLY1305 MAC (message authentication code) implementation in OpenSSL
44   does not save the contents of non-volatile XMM registers on Windows 64
45   platform when calculating the MAC of data larger than 64 bytes. Before
46   returning to the caller all the XMM registers are set to zero rather than
47   restoring their previous content. The vulnerable code is used only on newer
48   x86_64 processors supporting the AVX512-IFMA instructions.
49
50   The consequences of this kind of internal application state corruption can
51   be various - from no consequences, if the calling application does not
52   depend on the contents of non-volatile XMM registers at all, to the worst
53   consequences, where the attacker could get complete control of the
54   application process. However given the contents of the registers are just
55   zeroized so the attacker cannot put arbitrary values inside, the most likely
56   consequence, if any, would be an incorrect result of some application
57   dependent calculations or a crash leading to a denial of service.
58
59   ([CVE-2023-4807])
60
61   *Bernd Edlinger*
62
63### Changes between 3.0.9 and 3.0.10 [1 Aug 2023]
64
65 * Fix excessive time spent checking DH q parameter value.
66
67   The function DH_check() performs various checks on DH parameters. After
68   fixing CVE-2023-3446 it was discovered that a large q parameter value can
69   also trigger an overly long computation during some of these checks.
70   A correct q value, if present, cannot be larger than the modulus p
71   parameter, thus it is unnecessary to perform these checks if q is larger
72   than p.
73
74   If DH_check() is called with such q parameter value,
75   DH_CHECK_INVALID_Q_VALUE return flag is set and the computationally
76   intensive checks are skipped.
77
78   ([CVE-2023-3817])
79
80   *Tomáš Mráz*
81
82 * Fix DH_check() excessive time with over sized modulus.
83
84   The function DH_check() performs various checks on DH parameters. One of
85   those checks confirms that the modulus ("p" parameter) is not too large.
86   Trying to use a very large modulus is slow and OpenSSL will not normally use
87   a modulus which is over 10,000 bits in length.
88
89   However the DH_check() function checks numerous aspects of the key or
90   parameters that have been supplied. Some of those checks use the supplied
91   modulus value even if it has already been found to be too large.
92
93   A new limit has been added to DH_check of 32,768 bits. Supplying a
94   key/parameters with a modulus over this size will simply cause DH_check() to
95   fail.
96
97   ([CVE-2023-3446])
98
99   *Matt Caswell*
100
101 * Do not ignore empty associated data entries with AES-SIV.
102
103   The AES-SIV algorithm allows for authentication of multiple associated
104   data entries along with the encryption. To authenticate empty data the
105   application has to call `EVP_EncryptUpdate()` (or `EVP_CipherUpdate()`)
106   with NULL pointer as the output buffer and 0 as the input buffer length.
107   The AES-SIV implementation in OpenSSL just returns success for such call
108   instead of performing the associated data authentication operation.
109   The empty data thus will not be authenticated. ([CVE-2023-2975])
110
111   Thanks to Juerg Wullschleger (Google) for discovering the issue.
112
113   The fix changes the authentication tag value and the ciphertext for
114   applications that use empty associated data entries with AES-SIV.
115   To decrypt data encrypted with previous versions of OpenSSL the application
116   has to skip calls to `EVP_DecryptUpdate()` for empty associated data
117   entries.
118
119   *Tomáš Mráz*
120
121### Changes between 3.0.8 and 3.0.9 [30 May 2023]
122
123 * Mitigate for the time it takes for `OBJ_obj2txt` to translate gigantic
124   OBJECT IDENTIFIER sub-identifiers to canonical numeric text form.
125
126   OBJ_obj2txt() would translate any size OBJECT IDENTIFIER to canonical
127   numeric text form.  For gigantic sub-identifiers, this would take a very
128   long time, the time complexity being O(n^2) where n is the size of that
129   sub-identifier.  ([CVE-2023-2650])
130
131   To mitigitate this, `OBJ_obj2txt()` will only translate an OBJECT
132   IDENTIFIER to canonical numeric text form if the size of that OBJECT
133   IDENTIFIER is 586 bytes or less, and fail otherwise.
134
135   The basis for this restriction is [RFC 2578 (STD 58), section 3.5]. OBJECT
136   IDENTIFIER values, which stipulates that OBJECT IDENTIFIERS may have at
137   most 128 sub-identifiers, and that the maximum value that each sub-
138   identifier may have is 2^32-1 (4294967295 decimal).
139
140   For each byte of every sub-identifier, only the 7 lower bits are part of
141   the value, so the maximum amount of bytes that an OBJECT IDENTIFIER with
142   these restrictions may occupy is 32 * 128 / 7, which is approximately 586
143   bytes.
144
145   *Richard Levitte*
146
147 * Fixed buffer overread in AES-XTS decryption on ARM 64 bit platforms which
148   happens if the buffer size is 4 mod 5 in 16 byte AES blocks. This can
149   trigger a crash of an application using AES-XTS decryption if the memory
150   just after the buffer being decrypted is not mapped.
151   Thanks to Anton Romanov (Amazon) for discovering the issue.
152   ([CVE-2023-1255])
153
154   *Nevine Ebeid*
155
156 * Reworked the Fix for the Timing Oracle in RSA Decryption ([CVE-2022-4304]).
157   The previous fix for this timing side channel turned out to cause
158   a severe 2-3x performance regression in the typical use case
159   compared to 3.0.7. The new fix uses existing constant time
160   code paths, and restores the previous performance level while
161   fully eliminating all existing timing side channels.
162   The fix was developed by Bernd Edlinger with testing support
163   by Hubert Kario.
164
165   *Bernd Edlinger*
166
167 * Corrected documentation of X509_VERIFY_PARAM_add0_policy() to mention
168   that it does not enable policy checking. Thanks to David Benjamin for
169   discovering this issue.
170   ([CVE-2023-0466])
171
172   *Tomáš Mráz*
173
174 * Fixed an issue where invalid certificate policies in leaf certificates are
175   silently ignored by OpenSSL and other certificate policy checks are skipped
176   for that certificate. A malicious CA could use this to deliberately assert
177   invalid certificate policies in order to circumvent policy checking on the
178   certificate altogether.
179   ([CVE-2023-0465])
180
181   *Matt Caswell*
182
183 * Limited the number of nodes created in a policy tree to mitigate
184   against CVE-2023-0464.  The default limit is set to 1000 nodes, which
185   should be sufficient for most installations.  If required, the limit
186   can be adjusted by setting the OPENSSL_POLICY_TREE_NODES_MAX build
187   time define to a desired maximum number of nodes or zero to allow
188   unlimited growth.
189   ([CVE-2023-0464])
190
191   *Paul Dale*
192
193### Changes between 3.0.7 and 3.0.8 [7 Feb 2023]
194
195 * Fixed NULL dereference during PKCS7 data verification.
196
197   A NULL pointer can be dereferenced when signatures are being
198   verified on PKCS7 signed or signedAndEnveloped data. In case the hash
199   algorithm used for the signature is known to the OpenSSL library but
200   the implementation of the hash algorithm is not available the digest
201   initialization will fail. There is a missing check for the return
202   value from the initialization function which later leads to invalid
203   usage of the digest API most likely leading to a crash.
204   ([CVE-2023-0401])
205
206   PKCS7 data is processed by the SMIME library calls and also by the
207   time stamp (TS) library calls. The TLS implementation in OpenSSL does
208   not call these functions however third party applications would be
209   affected if they call these functions to verify signatures on untrusted
210   data.
211
212   *Tomáš Mráz*
213
214 * Fixed X.400 address type confusion in X.509 GeneralName.
215
216   There is a type confusion vulnerability relating to X.400 address processing
217   inside an X.509 GeneralName. X.400 addresses were parsed as an ASN1_STRING
218   but the public structure definition for GENERAL_NAME incorrectly specified
219   the type of the x400Address field as ASN1_TYPE. This field is subsequently
220   interpreted by the OpenSSL function GENERAL_NAME_cmp as an ASN1_TYPE rather
221   than an ASN1_STRING.
222
223   When CRL checking is enabled (i.e. the application sets the
224   X509_V_FLAG_CRL_CHECK flag), this vulnerability may allow an attacker to
225   pass arbitrary pointers to a memcmp call, enabling them to read memory
226   contents or enact a denial of service.
227   ([CVE-2023-0286])
228
229   *Hugo Landau*
230
231 * Fixed NULL dereference validating DSA public key.
232
233   An invalid pointer dereference on read can be triggered when an
234   application tries to check a malformed DSA public key by the
235   EVP_PKEY_public_check() function. This will most likely lead
236   to an application crash. This function can be called on public
237   keys supplied from untrusted sources which could allow an attacker
238   to cause a denial of service attack.
239
240   The TLS implementation in OpenSSL does not call this function
241   but applications might call the function if there are additional
242   security requirements imposed by standards such as FIPS 140-3.
243   ([CVE-2023-0217])
244
245   *Shane Lontis, Tomáš Mráz*
246
247 * Fixed Invalid pointer dereference in d2i_PKCS7 functions.
248
249   An invalid pointer dereference on read can be triggered when an
250   application tries to load malformed PKCS7 data with the
251   d2i_PKCS7(), d2i_PKCS7_bio() or d2i_PKCS7_fp() functions.
252
253   The result of the dereference is an application crash which could
254   lead to a denial of service attack. The TLS implementation in OpenSSL
255   does not call this function however third party applications might
256   call these functions on untrusted data.
257   ([CVE-2023-0216])
258
259   *Tomáš Mráz*
260
261 * Fixed Use-after-free following BIO_new_NDEF.
262
263   The public API function BIO_new_NDEF is a helper function used for
264   streaming ASN.1 data via a BIO. It is primarily used internally to OpenSSL
265   to support the SMIME, CMS and PKCS7 streaming capabilities, but may also
266   be called directly by end user applications.
267
268   The function receives a BIO from the caller, prepends a new BIO_f_asn1
269   filter BIO onto the front of it to form a BIO chain, and then returns
270   the new head of the BIO chain to the caller. Under certain conditions,
271   for example if a CMS recipient public key is invalid, the new filter BIO
272   is freed and the function returns a NULL result indicating a failure.
273   However, in this case, the BIO chain is not properly cleaned up and the
274   BIO passed by the caller still retains internal pointers to the previously
275   freed filter BIO. If the caller then goes on to call BIO_pop() on the BIO
276   then a use-after-free will occur. This will most likely result in a crash.
277   ([CVE-2023-0215])
278
279   *Viktor Dukhovni, Matt Caswell*
280
281 * Fixed Double free after calling PEM_read_bio_ex.
282
283   The function PEM_read_bio_ex() reads a PEM file from a BIO and parses and
284   decodes the "name" (e.g. "CERTIFICATE"), any header data and the payload
285   data. If the function succeeds then the "name_out", "header" and "data"
286   arguments are populated with pointers to buffers containing the relevant
287   decoded data. The caller is responsible for freeing those buffers. It is
288   possible to construct a PEM file that results in 0 bytes of payload data.
289   In this case PEM_read_bio_ex() will return a failure code but will populate
290   the header argument with a pointer to a buffer that has already been freed.
291   If the caller also frees this buffer then a double free will occur. This
292   will most likely lead to a crash.
293
294   The functions PEM_read_bio() and PEM_read() are simple wrappers around
295   PEM_read_bio_ex() and therefore these functions are also directly affected.
296
297   These functions are also called indirectly by a number of other OpenSSL
298   functions including PEM_X509_INFO_read_bio_ex() and
299   SSL_CTX_use_serverinfo_file() which are also vulnerable. Some OpenSSL
300   internal uses of these functions are not vulnerable because the caller does
301   not free the header argument if PEM_read_bio_ex() returns a failure code.
302   ([CVE-2022-4450])
303
304   *Kurt Roeckx, Matt Caswell*
305
306 * Fixed Timing Oracle in RSA Decryption.
307
308   A timing based side channel exists in the OpenSSL RSA Decryption
309   implementation which could be sufficient to recover a plaintext across
310   a network in a Bleichenbacher style attack. To achieve a successful
311   decryption an attacker would have to be able to send a very large number
312   of trial messages for decryption. The vulnerability affects all RSA padding
313   modes: PKCS#1 v1.5, RSA-OEAP and RSASVE.
314   ([CVE-2022-4304])
315
316   *Dmitry Belyavsky, Hubert Kario*
317
318 * Fixed X.509 Name Constraints Read Buffer Overflow.
319
320   A read buffer overrun can be triggered in X.509 certificate verification,
321   specifically in name constraint checking. The read buffer overrun might
322   result in a crash which could lead to a denial of service attack.
323   In a TLS client, this can be triggered by connecting to a malicious
324   server. In a TLS server, this can be triggered if the server requests
325   client authentication and a malicious client connects.
326   ([CVE-2022-4203])
327
328   *Viktor Dukhovni*
329
330 * Fixed X.509 Policy Constraints Double Locking security issue.
331
332   If an X.509 certificate contains a malformed policy constraint and
333   policy processing is enabled, then a write lock will be taken twice
334   recursively.  On some operating systems (most widely: Windows) this
335   results in a denial of service when the affected process hangs.  Policy
336   processing being enabled on a publicly facing server is not considered
337   to be a common setup.
338   ([CVE-2022-3996])
339
340   *Paul Dale*
341
342 * Our provider implementations of `OSSL_FUNC_KEYMGMT_EXPORT` and
343   `OSSL_FUNC_KEYMGMT_GET_PARAMS` for EC and SM2 keys now honor
344   `OSSL_PKEY_PARAM_EC_POINT_CONVERSION_FORMAT` as set (and
345   default to `POINT_CONVERSION_UNCOMPRESSED`) when exporting
346   `OSSL_PKEY_PARAM_PUB_KEY`, instead of unconditionally using
347   `POINT_CONVERSION_COMPRESSED` as in previous 3.x releases.
348   For symmetry, our implementation of `EVP_PKEY_ASN1_METHOD->export_to`
349   for legacy EC and SM2 keys is also changed similarly to honor the
350   equivalent conversion format flag as specified in the underlying
351   `EC_KEY` object being exported to a provider, when this function is
352   called through `EVP_PKEY_export()`.
353
354   *Nicola Tuveri*
355
356### Changes between 3.0.6 and 3.0.7 [1 Nov 2022]
357
358 * Fixed two buffer overflows in punycode decoding functions.
359
360   A buffer overrun can be triggered in X.509 certificate verification,
361   specifically in name constraint checking. Note that this occurs after
362   certificate chain signature verification and requires either a CA to
363   have signed the malicious certificate or for the application to continue
364   certificate verification despite failure to construct a path to a trusted
365   issuer.
366
367   In a TLS client, this can be triggered by connecting to a malicious
368   server.  In a TLS server, this can be triggered if the server requests
369   client authentication and a malicious client connects.
370
371   An attacker can craft a malicious email address to overflow
372   an arbitrary number of bytes containing the `.`  character (decimal 46)
373   on the stack.  This buffer overflow could result in a crash (causing a
374   denial of service).
375   ([CVE-2022-3786])
376
377   An attacker can craft a malicious email address to overflow four
378   attacker-controlled bytes on the stack.  This buffer overflow could
379   result in a crash (causing a denial of service) or potentially remote code
380   execution depending on stack layout for any given platform/compiler.
381   ([CVE-2022-3602])
382
383   *Paul Dale*
384
385 * Removed all references to invalid OSSL_PKEY_PARAM_RSA names for CRT
386   parameters in OpenSSL code.
387   Applications should not use the names OSSL_PKEY_PARAM_RSA_FACTOR,
388   OSSL_PKEY_PARAM_RSA_EXPONENT and OSSL_PKEY_PARAM_RSA_COEFFICIENT.
389   Use the numbered names such as OSSL_PKEY_PARAM_RSA_FACTOR1 instead.
390   Using these invalid names may cause algorithms to use slower methods
391   that ignore the CRT parameters.
392
393   *Shane Lontis*
394
395 * Fixed a regression introduced in 3.0.6 version raising errors on some stack
396   operations.
397
398   *Tomáš Mráz*
399
400 * Fixed a regression introduced in 3.0.6 version not refreshing the certificate
401   data to be signed before signing the certificate.
402
403   *Gibeom Gwon*
404
405 * Added RIPEMD160 to the default provider.
406
407   *Paul Dale*
408
409 * Ensured that the key share group sent or accepted for the key exchange
410   is allowed for the protocol version.
411
412   *Matt Caswell*
413
414### Changes between 3.0.5 and 3.0.6 [11 Oct 2022]
415
416 * OpenSSL supports creating a custom cipher via the legacy
417   EVP_CIPHER_meth_new() function and associated function calls. This function
418   was deprecated in OpenSSL 3.0 and application authors are instead encouraged
419   to use the new provider mechanism in order to implement custom ciphers.
420
421   OpenSSL versions 3.0.0 to 3.0.5 incorrectly handle legacy custom ciphers
422   passed to the EVP_EncryptInit_ex2(), EVP_DecryptInit_ex2() and
423   EVP_CipherInit_ex2() functions (as well as other similarly named encryption
424   and decryption initialisation functions). Instead of using the custom cipher
425   directly it incorrectly tries to fetch an equivalent cipher from the
426   available providers. An equivalent cipher is found based on the NID passed to
427   EVP_CIPHER_meth_new(). This NID is supposed to represent the unique NID for a
428   given cipher. However it is possible for an application to incorrectly pass
429   NID_undef as this value in the call to EVP_CIPHER_meth_new(). When NID_undef
430   is used in this way the OpenSSL encryption/decryption initialisation function
431   will match the NULL cipher as being equivalent and will fetch this from the
432   available providers. This will succeed if the default provider has been
433   loaded (or if a third party provider has been loaded that offers this
434   cipher). Using the NULL cipher means that the plaintext is emitted as the
435   ciphertext.
436
437   Applications are only affected by this issue if they call
438   EVP_CIPHER_meth_new() using NID_undef and subsequently use it in a call to an
439   encryption/decryption initialisation function. Applications that only use
440   SSL/TLS are not impacted by this issue.
441   ([CVE-2022-3358])
442
443   *Matt Caswell*
444
445 * Fix LLVM vs Apple LLVM version numbering confusion that caused build failures
446   on MacOS 10.11
447
448   *Richard Levitte*
449
450 * Fixed the linux-mips64 Configure target which was missing the
451   SIXTY_FOUR_BIT bn_ops flag. This was causing heap corruption on that
452   platform.
453
454   *Adam Joseph*
455
456 * Fix handling of a ticket key callback that returns 0 in TLSv1.3 to not send a
457   ticket
458
459   *Matt Caswell*
460
461 * Correctly handle a retransmitted ClientHello in DTLS
462
463   *Matt Caswell*
464
465 * Fixed detection of ktls support in cross-compile environment on Linux
466
467   *Tomas Mraz*
468
469 * Fixed some regressions and test failures when running the 3.0.0 FIPS provider
470   against 3.0.x
471
472   *Paul Dale*
473
474 * Fixed SSL_pending() and SSL_has_pending() with DTLS which were failing to
475   report correct results in some cases
476
477   *Matt Caswell*
478
479 * Fix UWP builds by defining VirtualLock
480
481   *Charles Milette*
482
483 * For known safe primes use the minimum key length according to RFC 7919.
484   Longer private key sizes unnecessarily raise the cycles needed to compute the
485   shared secret without any increase of the real security. This fixes a
486   regression from 1.1.1 where these shorter keys were generated for the known
487   safe primes.
488
489   *Tomas Mraz*
490
491 * Added the loongarch64 target
492
493   *Shi Pujin*
494
495 * Fixed EC ASM flag passing. Flags for ASM implementations of EC curves were
496   only passed to the FIPS provider and not to the default or legacy provider.
497
498   *Juergen Christ*
499
500 * Fixed reported performance degradation on aarch64. Restored the
501   implementation prior to commit 2621751 ("aes/asm/aesv8-armx.pl: avoid
502   32-bit lane assignment in CTR mode") for 64bit targets only, since it is
503   reportedly 2-17% slower and the silicon errata only affects 32bit targets.
504   The new algorithm is still used for 32 bit targets.
505
506   *Bernd Edlinger*
507
508 * Added a missing header for memcmp that caused compilation failure on some
509   platforms
510
511   *Gregor Jasny*
512
513### Changes between 3.0.4 and 3.0.5 [5 Jul 2022]
514
515 * The OpenSSL 3.0.4 release introduced a serious bug in the RSA
516   implementation for X86_64 CPUs supporting the AVX512IFMA instructions.
517   This issue makes the RSA implementation with 2048 bit private keys
518   incorrect on such machines and memory corruption will happen during
519   the computation. As a consequence of the memory corruption an attacker
520   may be able to trigger a remote code execution on the machine performing
521   the computation.
522
523   SSL/TLS servers or other servers using 2048 bit RSA private keys running
524   on machines supporting AVX512IFMA instructions of the X86_64 architecture
525   are affected by this issue.
526   ([CVE-2022-2274])
527
528   *Xi Ruoyao*
529
530 * AES OCB mode for 32-bit x86 platforms using the AES-NI assembly optimised
531   implementation would not encrypt the entirety of the data under some
532   circumstances.  This could reveal sixteen bytes of data that was
533   preexisting in the memory that wasn't written.  In the special case of
534   "in place" encryption, sixteen bytes of the plaintext would be revealed.
535
536   Since OpenSSL does not support OCB based cipher suites for TLS and DTLS,
537   they are both unaffected.
538   ([CVE-2022-2097])
539
540   *Alex Chernyakhovsky, David Benjamin, Alejandro Sedeño*
541
542### Changes between 3.0.3 and 3.0.4 [21 Jun 2022]
543
544 * In addition to the c_rehash shell command injection identified in
545   CVE-2022-1292, further bugs where the c_rehash script does not
546   properly sanitise shell metacharacters to prevent command injection have been
547   fixed.
548
549   When the CVE-2022-1292 was fixed it was not discovered that there
550   are other places in the script where the file names of certificates
551   being hashed were possibly passed to a command executed through the shell.
552
553   This script is distributed by some operating systems in a manner where
554   it is automatically executed.  On such operating systems, an attacker
555   could execute arbitrary commands with the privileges of the script.
556
557   Use of the c_rehash script is considered obsolete and should be replaced
558   by the OpenSSL rehash command line tool.
559   (CVE-2022-2068)
560
561   *Daniel Fiala, Tomáš Mráz*
562
563 * Case insensitive string comparison no longer uses locales.  It has instead
564   been directly implemented.
565
566   *Paul Dale*
567
568### Changes between 3.0.2 and 3.0.3 [3 May 2022]
569
570 * Case insensitive string comparison is reimplemented via new locale-agnostic
571   comparison functions OPENSSL_str[n]casecmp always using the POSIX locale for
572   comparison. The previous implementation had problems when the Turkish locale
573   was used.
574
575   *Dmitry Belyavskiy*
576
577 * Fixed a bug in the c_rehash script which was not properly sanitising shell
578   metacharacters to prevent command injection.  This script is distributed by
579   some operating systems in a manner where it is automatically executed.  On
580   such operating systems, an attacker could execute arbitrary commands with the
581   privileges of the script.
582
583   Use of the c_rehash script is considered obsolete and should be replaced
584   by the OpenSSL rehash command line tool.
585   (CVE-2022-1292)
586
587   *Tomáš Mráz*
588
589 * Fixed a bug in the function `OCSP_basic_verify` that verifies the signer
590   certificate on an OCSP response. The bug caused the function in the case
591   where the (non-default) flag OCSP_NOCHECKS is used to return a postivie
592   response (meaning a successful verification) even in the case where the
593   response signing certificate fails to verify.
594
595   It is anticipated that most users of `OCSP_basic_verify` will not use the
596   OCSP_NOCHECKS flag. In this case the `OCSP_basic_verify` function will return
597   a negative value (indicating a fatal error) in the case of a certificate
598   verification failure. The normal expected return value in this case would be
599   0.
600
601   This issue also impacts the command line OpenSSL "ocsp" application. When
602   verifying an ocsp response with the "-no_cert_checks" option the command line
603   application will report that the verification is successful even though it
604   has in fact failed. In this case the incorrect successful response will also
605   be accompanied by error messages showing the failure and contradicting the
606   apparently successful result.
607   ([CVE-2022-1343])
608
609   *Matt Caswell*
610
611 * Fixed a bug where the RC4-MD5 ciphersuite incorrectly used the
612   AAD data as the MAC key. This made the MAC key trivially predictable.
613
614   An attacker could exploit this issue by performing a man-in-the-middle attack
615   to modify data being sent from one endpoint to an OpenSSL 3.0 recipient such
616   that the modified data would still pass the MAC integrity check.
617
618   Note that data sent from an OpenSSL 3.0 endpoint to a non-OpenSSL 3.0
619   endpoint will always be rejected by the recipient and the connection will
620   fail at that point. Many application protocols require data to be sent from
621   the client to the server first. Therefore, in such a case, only an OpenSSL
622   3.0 server would be impacted when talking to a non-OpenSSL 3.0 client.
623
624   If both endpoints are OpenSSL 3.0 then the attacker could modify data being
625   sent in both directions. In this case both clients and servers could be
626   affected, regardless of the application protocol.
627
628   Note that in the absence of an attacker this bug means that an OpenSSL 3.0
629   endpoint communicating with a non-OpenSSL 3.0 endpoint will fail to complete
630   the handshake when using this ciphersuite.
631
632   The confidentiality of data is not impacted by this issue, i.e. an attacker
633   cannot decrypt data that has been encrypted using this ciphersuite - they can
634   only modify it.
635
636   In order for this attack to work both endpoints must legitimately negotiate
637   the RC4-MD5 ciphersuite. This ciphersuite is not compiled by default in
638   OpenSSL 3.0, and is not available within the default provider or the default
639   ciphersuite list. This ciphersuite will never be used if TLSv1.3 has been
640   negotiated. In order for an OpenSSL 3.0 endpoint to use this ciphersuite the
641   following must have occurred:
642
643   1) OpenSSL must have been compiled with the (non-default) compile time option
644      enable-weak-ssl-ciphers
645
646   2) OpenSSL must have had the legacy provider explicitly loaded (either
647      through application code or via configuration)
648
649   3) The ciphersuite must have been explicitly added to the ciphersuite list
650
651   4) The libssl security level must have been set to 0 (default is 1)
652
653   5) A version of SSL/TLS below TLSv1.3 must have been negotiated
654
655   6) Both endpoints must negotiate the RC4-MD5 ciphersuite in preference to any
656      others that both endpoints have in common
657   (CVE-2022-1434)
658
659   *Matt Caswell*
660
661 * Fix a bug in the OPENSSL_LH_flush() function that breaks reuse of the memory
662   occuppied by the removed hash table entries.
663
664   This function is used when decoding certificates or keys. If a long lived
665   process periodically decodes certificates or keys its memory usage will
666   expand without bounds and the process might be terminated by the operating
667   system causing a denial of service. Also traversing the empty hash table
668   entries will take increasingly more time.
669
670   Typically such long lived processes might be TLS clients or TLS servers
671   configured to accept client certificate authentication.
672   (CVE-2022-1473)
673
674   *Hugo Landau, Aliaksei Levin*
675
676 * The functions `OPENSSL_LH_stats` and `OPENSSL_LH_stats_bio` now only report
677   the `num_items`, `num_nodes` and `num_alloc_nodes` statistics. All other
678   statistics are no longer supported. For compatibility, these statistics are
679   still listed in the output but are now always reported as zero.
680
681   *Hugo Landau*
682
683### Changes between 3.0.1 and 3.0.2 [15 Mar 2022]
684
685 * Fixed a bug in the BN_mod_sqrt() function that can cause it to loop forever
686   for non-prime moduli.
687
688   Internally this function is used when parsing certificates that contain
689   elliptic curve public keys in compressed form or explicit elliptic curve
690   parameters with a base point encoded in compressed form.
691
692   It is possible to trigger the infinite loop by crafting a certificate that
693   has invalid explicit curve parameters.
694
695   Since certificate parsing happens prior to verification of the certificate
696   signature, any process that parses an externally supplied certificate may thus
697   be subject to a denial of service attack. The infinite loop can also be
698   reached when parsing crafted private keys as they can contain explicit
699   elliptic curve parameters.
700
701   Thus vulnerable situations include:
702
703    - TLS clients consuming server certificates
704    - TLS servers consuming client certificates
705    - Hosting providers taking certificates or private keys from customers
706    - Certificate authorities parsing certification requests from subscribers
707    - Anything else which parses ASN.1 elliptic curve parameters
708
709   Also any other applications that use the BN_mod_sqrt() where the attacker
710   can control the parameter values are vulnerable to this DoS issue.
711   ([CVE-2022-0778])
712
713   *Tomáš Mráz*
714
715 * Add ciphersuites based on DHE_PSK (RFC 4279) and ECDHE_PSK (RFC 5489)
716   to the list of ciphersuites providing Perfect Forward Secrecy as
717   required by SECLEVEL >= 3.
718
719   *Dmitry Belyavskiy, Nicola Tuveri*
720
721 * Made the AES constant time code for no-asm configurations
722   optional due to the resulting 95% performance degradation.
723   The AES constant time code can be enabled, for no assembly
724   builds, with: ./config no-asm -DOPENSSL_AES_CONST_TIME
725
726   *Paul Dale*
727
728 * Fixed PEM_write_bio_PKCS8PrivateKey() to make it possible to use empty
729   passphrase strings.
730
731   *Darshan Sen*
732
733 * The negative return value handling of the certificate verification callback
734   was reverted. The replacement is to set the verification retry state with
735   the SSL_set_retry_verify() function.
736
737   *Tomáš Mráz*
738
739### Changes between 3.0.0 and 3.0.1 [14 Dec 2021]
740
741 * Fixed invalid handling of X509_verify_cert() internal errors in libssl
742   Internally libssl in OpenSSL calls X509_verify_cert() on the client side to
743   verify a certificate supplied by a server. That function may return a
744   negative return value to indicate an internal error (for example out of
745   memory). Such a negative return value is mishandled by OpenSSL and will cause
746   an IO function (such as SSL_connect() or SSL_do_handshake()) to not indicate
747   success and a subsequent call to SSL_get_error() to return the value
748   SSL_ERROR_WANT_RETRY_VERIFY. This return value is only supposed to be
749   returned by OpenSSL if the application has previously called
750   SSL_CTX_set_cert_verify_callback(). Since most applications do not do this
751   the SSL_ERROR_WANT_RETRY_VERIFY return value from SSL_get_error() will be
752   totally unexpected and applications may not behave correctly as a result. The
753   exact behaviour will depend on the application but it could result in
754   crashes, infinite loops or other similar incorrect responses.
755
756   This issue is made more serious in combination with a separate bug in OpenSSL
757   3.0 that will cause X509_verify_cert() to indicate an internal error when
758   processing a certificate chain. This will occur where a certificate does not
759   include the Subject Alternative Name extension but where a Certificate
760   Authority has enforced name constraints. This issue can occur even with valid
761   chains.
762   ([CVE-2021-4044])
763
764   *Matt Caswell*
765
766 * Corrected a few file name and file reference bugs in the build,
767   installation and setup scripts, which lead to installation verification
768   failures.  Slightly enhanced the installation verification script.
769
770   *Richard Levitte*
771
772 * Fixed EVP_PKEY_eq() to make it possible to use it with strictly private
773   keys.
774
775   *Richard Levitte*
776
777 * Fixed PVK encoder to properly query for the passphrase.
778
779   *Tomáš Mráz*
780
781 * Multiple fixes in the OSSL_HTTP API functions.
782
783   *David von Oheimb*
784
785 * Allow sign extension in OSSL_PARAM_allocate_from_text() for the
786   OSSL_PARAM_INTEGER data type and return error on negative numbers
787   used with the OSSL_PARAM_UNSIGNED_INTEGER data type. Make
788   OSSL_PARAM_BLD_push_BN{,_pad}() return an error on negative numbers.
789
790   *Richard Levitte*
791
792 * Allow copying uninitialized digest contexts with EVP_MD_CTX_copy_ex.
793
794   *Tomáš Mráz*
795
796 * Fixed detection of ARMv7 and ARM64 CPU features on FreeBSD.
797
798   *Allan Jude*
799
800 * Multiple threading fixes.
801
802   *Matt Caswell*
803
804 * Added NULL digest implementation to keep compatibility with 1.1.1 version.
805
806   *Tomáš Mráz*
807
808 * Allow fetching an operation from the provider that owns an unexportable key
809   as a fallback if that is still allowed by the property query.
810
811   *Richard Levitte*
812
813### Changes between 1.1.1 and 3.0.0 [7 sep 2021]
814
815 * TLS_MAX_VERSION, DTLS_MAX_VERSION and DTLS_MIN_VERSION constants are now
816   deprecated.
817
818   *Matt Caswell*
819
820 * The `OPENSSL_s390xcap` environment variable can be used to set bits in the
821   S390X capability vector to zero. This simplifies testing of different code
822   paths on S390X architecture.
823
824   *Patrick Steuer*
825
826 * Encrypting more than 2^64 TLS records with AES-GCM is disallowed
827   as per FIPS 140-2 IG A.5 "Key/IV Pair Uniqueness Requirements from
828   SP 800-38D". The communication will fail at this point.
829
830   *Paul Dale*
831
832 * The EC_GROUP_clear_free() function is deprecated as there is nothing
833   confidential in EC_GROUP data.
834
835   *Nicola Tuveri*
836
837 * The byte order mark (BOM) character is ignored if encountered at the
838   beginning of a PEM-formatted file.
839
840   *Dmitry Belyavskiy*
841
842 * Added CMS support for the Russian GOST algorithms.
843
844   *Dmitry Belyavskiy*
845
846 * Due to move of the implementation of cryptographic operations
847   to the providers, validation of various operation parameters can
848   be postponed until the actual operation is executed where previously
849   it happened immediately when an operation parameter was set.
850
851   For example when setting an unsupported curve with
852   EVP_PKEY_CTX_set_ec_paramgen_curve_nid() this function call will not
853   fail but later keygen operations with the EVP_PKEY_CTX will fail.
854
855   *OpenSSL team members and many third party contributors*
856
857 * The EVP_get_cipherbyname() function will return NULL for algorithms such as
858   "AES-128-SIV", "AES-128-CBC-CTS" and "CAMELLIA-128-CBC-CTS" which were
859   previously only accessible via low level interfaces. Use EVP_CIPHER_fetch()
860   instead to retrieve these algorithms from a provider.
861
862   *Shane Lontis*
863
864 * On build targets where the multilib postfix is set in the build
865   configuration the libdir directory was changing based on whether
866   the lib directory with the multilib postfix exists on the system
867   or not. This unpredictable behavior was removed and eventual
868   multilib postfix is now always added to the default libdir. Use
869   `--libdir=lib` to override the libdir if adding the postfix is
870   undesirable.
871
872   *Jan Lána*
873
874 * The triple DES key wrap functionality now conforms to RFC 3217 but is
875   no longer interoperable with OpenSSL 1.1.1.
876
877   *Paul Dale*
878
879 * The ERR_GET_FUNC() function was removed.  With the loss of meaningful
880   function codes, this function can only cause problems for calling
881   applications.
882
883   *Paul Dale*
884
885 * Add a configurable flag to output date formats as ISO 8601. Does not
886   change the default date format.
887
888   *William Edmisten*
889
890 * Version of MSVC earlier than 1300 could get link warnings, which could
891   be suppressed if the undocumented -DI_CAN_LIVE_WITH_LNK4049 was set.
892   Support for this flag has been removed.
893
894   *Rich Salz*
895
896 * Rework and make DEBUG macros consistent. Remove unused -DCONF_DEBUG,
897   -DBN_CTX_DEBUG, and REF_PRINT. Add a new tracing category and use it for
898   printing reference counts. Rename -DDEBUG_UNUSED to -DUNUSED_RESULT_DEBUG
899   Fix BN_DEBUG_RAND so it compiles and, when set, force DEBUG_RAND to be set
900   also. Rename engine_debug_ref to be ENGINE_REF_PRINT also for consistency.
901
902   *Rich Salz*
903
904 * The signatures of the functions to get and set options on SSL and
905   SSL_CTX objects changed from "unsigned long" to "uint64_t" type.
906   Some source code changes may be required.
907
908   *Rich Salz*
909
910 * The public definitions of conf_method_st and conf_st have been
911   deprecated. They will be made opaque in a future release.
912
913   *Rich Salz and Tomáš Mráz*
914
915 * Client-initiated renegotiation is disabled by default. To allow it, use
916   the -client_renegotiation option, the SSL_OP_ALLOW_CLIENT_RENEGOTIATION
917   flag, or the "ClientRenegotiation" config parameter as appropriate.
918
919   *Rich Salz*
920
921 * Add "abspath" and "includedir" pragma's to config files, to prevent,
922   or modify relative pathname inclusion.
923
924   *Rich Salz*
925
926 * OpenSSL includes a cryptographic module that is intended to be FIPS 140-2
927   validated. Please consult the README-FIPS and
928   README-PROVIDERS files, as well as the migration guide.
929
930   *OpenSSL team members and many third party contributors*
931
932 * For the key types DH and DHX the allowed settable parameters are now different.
933
934   *Shane Lontis*
935
936 * The openssl commands that read keys, certificates, and CRLs now
937   automatically detect the PEM or DER format of the input files.
938
939   *David von Oheimb, Richard Levitte, and Tomáš Mráz*
940
941 * Added enhanced PKCS#12 APIs which accept a library context.
942
943   *Jon Spillett*
944
945 * The default manual page suffix ($MANSUFFIX) has been changed to "ossl"
946
947   *Matt Caswell*
948
949 * Added support for Kernel TLS (KTLS).
950
951   *Boris Pismenny, John Baldwin and Andrew Gallatin*
952
953 * Support for RFC 5746 secure renegotiation is now required by default for
954   SSL or TLS connections to succeed.
955
956   *Benjamin Kaduk*
957
958 * The signature of the `copy` functional parameter of the
959   EVP_PKEY_meth_set_copy() function has changed so its `src` argument is
960   now `const EVP_PKEY_CTX *` instead of `EVP_PKEY_CTX *`. Similarly
961   the signature of the `pub_decode` functional parameter of the
962   EVP_PKEY_asn1_set_public() function has changed so its `pub` argument is
963   now `const X509_PUBKEY *` instead of `X509_PUBKEY *`.
964
965   *David von Oheimb*
966
967 * The error return values from some control calls (ctrl) have changed.
968
969   *Paul Dale*
970
971 * A public key check is now performed during EVP_PKEY_derive_set_peer().
972
973   *Shane Lontis*
974
975 * Many functions in the EVP_ namespace that are getters of values from
976   implementations or contexts were renamed to include get or get0 in their
977   names. Old names are provided as macro aliases for compatibility and
978   are not deprecated.
979
980   *Tomáš Mráz*
981
982 * The EVP_PKEY_CTRL_PKCS7_ENCRYPT, EVP_PKEY_CTRL_PKCS7_DECRYPT,
983   EVP_PKEY_CTRL_PKCS7_SIGN, EVP_PKEY_CTRL_CMS_ENCRYPT,
984   EVP_PKEY_CTRL_CMS_DECRYPT, and EVP_PKEY_CTRL_CMS_SIGN control operations
985   are deprecated.
986
987   *Tomáš Mráz*
988
989 * The EVP_PKEY_public_check() and EVP_PKEY_param_check() functions now work for
990   more key types.
991
992 * The output from the command line applications may have minor
993   changes.
994
995   *Paul Dale*
996
997 * The output from numerous "printing" may have minor changes.
998
999   *David von Oheimb*
1000
1001 * Windows thread synchronization uses read/write primitives (SRWLock) when
1002   supported by the OS, otherwise CriticalSection continues to be used.
1003
1004   *Vincent Drake*
1005
1006 * Add filter BIO BIO_f_readbuffer() that allows BIO_tell() and BIO_seek() to
1007   work on read only BIO source/sinks that do not support these functions.
1008   This allows piping or redirection of a file BIO using stdin to be buffered
1009   into memory. This is used internally in OSSL_DECODER_from_bio().
1010
1011   *Shane Lontis*
1012
1013 * OSSL_STORE_INFO_get_type() may now return an additional value. In 1.1.1
1014   this function would return one of the values OSSL_STORE_INFO_NAME,
1015   OSSL_STORE_INFO_PKEY, OSSL_STORE_INFO_PARAMS, OSSL_STORE_INFO_CERT or
1016   OSSL_STORE_INFO_CRL. Decoded public keys would previously have been reported
1017   as type OSSL_STORE_INFO_PKEY in 1.1.1. In 3.0 decoded public keys are now
1018   reported as having the new type OSSL_STORE_INFO_PUBKEY. Applications
1019   using this function should be amended to handle the changed return value.
1020
1021   *Richard Levitte*
1022
1023 * Improved adherence to Enhanced Security Services (ESS, RFC 2634 and RFC 5035)
1024   for the TSP and CMS Advanced Electronic Signatures (CAdES) implementations.
1025   As required by RFC 5035 check both ESSCertID and ESSCertIDv2 if both present.
1026   Correct the semantics of checking the validation chain in case ESSCertID{,v2}
1027   contains more than one certificate identifier: This means that all
1028   certificates referenced there MUST be part of the validation chain.
1029
1030   *David von Oheimb*
1031
1032 * The implementation of older EVP ciphers related to CAST, IDEA, SEED, RC2, RC4,
1033   RC5, DESX and DES have been moved to the legacy provider.
1034
1035   *Matt Caswell*
1036
1037 * The implementation of the EVP digests MD2, MD4, MDC2, WHIRLPOOL and
1038   RIPEMD-160 have been moved to the legacy provider.
1039
1040   *Matt Caswell*
1041
1042 * The deprecated function EVP_PKEY_get0() now returns NULL being called for a
1043   provided key.
1044
1045   *Dmitry Belyavskiy*
1046
1047 * The deprecated functions EVP_PKEY_get0_RSA(),
1048   EVP_PKEY_get0_DSA(), EVP_PKEY_get0_EC_KEY(), EVP_PKEY_get0_DH(),
1049   EVP_PKEY_get0_hmac(), EVP_PKEY_get0_poly1305() and EVP_PKEY_get0_siphash() as
1050   well as the similarly named "get1" functions behave differently in
1051   OpenSSL 3.0.
1052
1053   *Matt Caswell*
1054
1055 * A number of functions handling low-level keys or engines were deprecated
1056   including EVP_PKEY_set1_engine(), EVP_PKEY_get0_engine(), EVP_PKEY_assign(),
1057   EVP_PKEY_get0(), EVP_PKEY_get0_hmac(), EVP_PKEY_get0_poly1305() and
1058   EVP_PKEY_get0_siphash().
1059
1060   *Matt Caswell*
1061
1062 * PKCS#5 PBKDF1 key derivation has been moved from PKCS5_PBE_keyivgen() into
1063   the legacy crypto provider as an EVP_KDF. Applications requiring this KDF
1064   will need to load the legacy crypto provider. This includes these PBE
1065   algorithms which use this KDF:
1066   - NID_pbeWithMD2AndDES_CBC
1067   - NID_pbeWithMD5AndDES_CBC
1068   - NID_pbeWithSHA1AndRC2_CBC
1069   - NID_pbeWithMD2AndRC2_CBC
1070   - NID_pbeWithMD5AndRC2_CBC
1071   - NID_pbeWithSHA1AndDES_CBC
1072
1073   *Jon Spillett*
1074
1075 * Deprecated obsolete BIO_set_callback(), BIO_get_callback(), and
1076   BIO_debug_callback() functions.
1077
1078   *Tomáš Mráz*
1079
1080 * Deprecated obsolete EVP_PKEY_CTX_get0_dh_kdf_ukm() and
1081   EVP_PKEY_CTX_get0_ecdh_kdf_ukm() functions.
1082
1083   *Tomáš Mráz*
1084
1085 * The RAND_METHOD APIs have been deprecated.
1086
1087   *Paul Dale*
1088
1089 * The SRP APIs have been deprecated.
1090
1091   *Matt Caswell*
1092
1093 * Add a compile time option to prevent the caching of provider fetched
1094   algorithms.  This is enabled by including the no-cached-fetch option
1095   at configuration time.
1096
1097   *Paul Dale*
1098
1099 * pkcs12 now uses defaults of PBKDF2, AES and SHA-256, with a MAC iteration
1100   count of PKCS12_DEFAULT_ITER.
1101
1102   *Tomáš Mráz and Sahana Prasad*
1103
1104 * The openssl speed command does not use low-level API calls anymore.
1105
1106   *Tomáš Mráz*
1107
1108 * Parallel dual-prime 1024-bit modular exponentiation for AVX512_IFMA
1109   capable processors.
1110
1111   *Ilya Albrekht, Sergey Kirillov, Andrey Matyukov (Intel Corp)*
1112
1113 * Combining the Configure options no-ec and no-dh no longer disables TLSv1.3.
1114
1115   *Matt Caswell*
1116
1117 * Implemented support for fully "pluggable" TLSv1.3 groups. This means that
1118   providers may supply their own group implementations (using either the "key
1119   exchange" or the "key encapsulation" methods) which will automatically be
1120   detected and used by libssl.
1121
1122   *Matt Caswell, Nicola Tuveri*
1123
1124 * The undocumented function X509_certificate_type() has been deprecated;
1125
1126   *Rich Salz*
1127
1128 * Deprecated the obsolete BN_pseudo_rand() and BN_pseudo_rand_range().
1129
1130   *Tomáš Mráz*
1131
1132 * Removed RSA padding mode for SSLv23 (which was only used for
1133   SSLv2). This includes the functions RSA_padding_check_SSLv23() and
1134   RSA_padding_add_SSLv23() and the `-ssl` option in the deprecated
1135   `rsautl` command.
1136
1137   *Rich Salz*
1138
1139 * Deprecated the obsolete X9.31 RSA key generation related functions.
1140
1141 * While a callback function set via `SSL_CTX_set_cert_verify_callback()`
1142   is not allowed to return a value > 1, this is no more taken as failure.
1143
1144   *Viktor Dukhovni and David von Oheimb*
1145
1146 * Deprecated the obsolete X9.31 RSA key generation related functions
1147   BN_X931_generate_Xpq(), BN_X931_derive_prime_ex(), and
1148   BN_X931_generate_prime_ex().
1149
1150   *Tomáš Mráz*
1151
1152 * The default key generation method for the regular 2-prime RSA keys was
1153   changed to the FIPS 186-4 B.3.6 method.
1154
1155   *Shane Lontis*
1156
1157 * Deprecated the BN_is_prime_ex() and BN_is_prime_fasttest_ex() functions.
1158
1159   *Kurt Roeckx*
1160
1161 * Deprecated EVP_MD_CTX_set_update_fn() and EVP_MD_CTX_update_fn().
1162
1163   *Rich Salz*
1164
1165 * Deprecated the type OCSP_REQ_CTX and the functions OCSP_REQ_CTX_*() and
1166   replaced with OSSL_HTTP_REQ_CTX and the functions OSSL_HTTP_REQ_CTX_*().
1167
1168   *Rich Salz, Richard Levitte, and David von Oheimb*
1169
1170 * Deprecated `X509_http_nbio()` and `X509_CRL_http_nbio()`.
1171
1172   *David von Oheimb*
1173
1174 * Deprecated `OCSP_parse_url()`.
1175
1176   *David von Oheimb*
1177
1178 * Validation of SM2 keys has been separated from the validation of regular EC
1179   keys.
1180
1181   *Nicola Tuveri*
1182
1183 * Behavior of the `pkey` app is changed, when using the `-check` or `-pubcheck`
1184   switches: a validation failure triggers an early exit, returning a failure
1185   exit status to the parent process.
1186
1187   *Nicola Tuveri*
1188
1189 * Changed behavior of SSL_CTX_set_ciphersuites() and SSL_set_ciphersuites()
1190   to ignore unknown ciphers.
1191
1192   *Otto Hollmann*
1193
1194 * The `-cipher-commands` and `-digest-commands` options
1195   of the command line utility `list` have been deprecated.
1196   Instead use the `-cipher-algorithms` and `-digest-algorithms` options.
1197
1198   *Dmitry Belyavskiy*
1199
1200 * Added convenience functions for generating asymmetric key pairs:
1201   The 'quick' one-shot (yet somewhat limited) function L<EVP_PKEY_Q_keygen(3)>
1202   and macros for the most common cases: <EVP_RSA_gen(3)> and L<EVP_EC_gen(3)>.
1203
1204   *David von Oheimb*
1205
1206 * All of the low level EC_KEY functions have been deprecated.
1207
1208   *Shane Lontis, Paul Dale, Richard Levitte, and Tomáš Mráz*
1209
1210 * Deprecated all the libcrypto and libssl error string loading
1211   functions.
1212
1213   *Richard Levitte*
1214
1215 * The functions SSL_CTX_set_tmp_dh_callback and SSL_set_tmp_dh_callback, as
1216   well as the macros SSL_CTX_set_tmp_dh() and SSL_set_tmp_dh() have been
1217   deprecated.
1218
1219   *Matt Caswell*
1220
1221 * The `-crypt` option to the `passwd` command line tool has been removed.
1222
1223   *Paul Dale*
1224
1225 * The -C option to the `x509`, `dhparam`, `dsaparam`, and `ecparam` commands
1226   were removed.
1227
1228   *Rich Salz*
1229
1230 * Add support for AES Key Wrap inverse ciphers to the EVP layer.
1231
1232   *Shane Lontis*
1233
1234 * Deprecated EVP_PKEY_set1_tls_encodedpoint() and
1235   EVP_PKEY_get1_tls_encodedpoint().
1236
1237   *Matt Caswell*
1238
1239 * The security callback, which can be customised by application code, supports
1240   the security operation SSL_SECOP_TMP_DH. One location of the "other" parameter
1241   was incorrectly passing a DH object. It now passed an EVP_PKEY in all cases.
1242
1243   *Matt Caswell*
1244
1245 * Add PKCS7_get_octet_string() and PKCS7_type_is_other() to the public
1246   interface. Their functionality remains unchanged.
1247
1248   *Jordan Montgomery*
1249
1250 * Added new option for 'openssl list', '-providers', which will display the
1251   list of loaded providers, their names, version and status.  It optionally
1252   displays their gettable parameters.
1253
1254   *Paul Dale*
1255
1256 * Removed EVP_PKEY_set_alias_type().
1257
1258   *Richard Levitte*
1259
1260 * Deprecated `EVP_PKEY_CTX_set_rsa_keygen_pubexp()` and introduced
1261   `EVP_PKEY_CTX_set1_rsa_keygen_pubexp()`, which is now preferred.
1262
1263   *Jeremy Walch*
1264
1265 * Changed all "STACK" functions to be macros instead of inline functions. Macro
1266   parameters are still checked for type safety at compile time via helper
1267   inline functions.
1268
1269   *Matt Caswell*
1270
1271 * Remove the RAND_DRBG API
1272
1273   *Paul Dale and Matthias St. Pierre*
1274
1275 * Allow `SSL_set1_host()` and `SSL_add1_host()` to take IP literal addresses
1276   as well as actual hostnames.
1277
1278   *David Woodhouse*
1279
1280 * The 'MinProtocol' and 'MaxProtocol' configuration commands now silently
1281   ignore TLS protocol version bounds when configuring DTLS-based contexts, and
1282   conversely, silently ignore DTLS protocol version bounds when configuring
1283   TLS-based contexts.  The commands can be repeated to set bounds of both
1284   types.  The same applies with the corresponding "min_protocol" and
1285   "max_protocol" command-line switches, in case some application uses both TLS
1286   and DTLS.
1287
1288   SSL_CTX instances that are created for a fixed protocol version (e.g.
1289   `TLSv1_server_method()`) also silently ignore version bounds.  Previously
1290   attempts to apply bounds to these protocol versions would result in an
1291   error.  Now only the "version-flexible" SSL_CTX instances are subject to
1292   limits in configuration files in command-line options.
1293
1294   *Viktor Dukhovni*
1295
1296 * Deprecated the `ENGINE` API.  Engines should be replaced with providers
1297   going forward.
1298
1299   *Paul Dale*
1300
1301 * Reworked the recorded ERR codes to make better space for system errors.
1302   To distinguish them, the macro `ERR_SYSTEM_ERROR()` indicates if the
1303   given code is a system error (true) or an OpenSSL error (false).
1304
1305   *Richard Levitte*
1306
1307 * Reworked the test perl framework to better allow parallel testing.
1308
1309   *Nicola Tuveri and David von Oheimb*
1310
1311 * Added ciphertext stealing algorithms AES-128-CBC-CTS, AES-192-CBC-CTS and
1312   AES-256-CBC-CTS to the providers. CS1, CS2 and CS3 variants are supported.
1313
1314   *Shane Lontis*
1315
1316 * 'Configure' has been changed to figure out the configuration target if
1317   none is given on the command line.  Consequently, the 'config' script is
1318   now only a mere wrapper.  All documentation is changed to only mention
1319   'Configure'.
1320
1321   *Rich Salz and Richard Levitte*
1322
1323 * Added a library context `OSSL_LIB_CTX` that applications as well as
1324   other libraries can use to form a separate context within which
1325   libcrypto operations are performed.
1326
1327   *Richard Levitte*
1328
1329 * Added various `_ex` functions to the OpenSSL API that support using
1330   a non-default `OSSL_LIB_CTX`.
1331
1332   *OpenSSL team*
1333
1334 * Handshake now fails if Extended Master Secret extension is dropped
1335   on renegotiation.
1336
1337   *Tomáš Mráz*
1338
1339 * Dropped interactive mode from the `openssl` program.
1340
1341   *Richard Levitte*
1342
1343 * Deprecated `EVP_PKEY_cmp()` and `EVP_PKEY_cmp_parameters()`.
1344
1345   *David von Oheimb and Shane Lontis*
1346
1347 * Deprecated `EC_METHOD_get_field_type()`.
1348
1349   *Billy Bob Brumley*
1350
1351 * Deprecated EC_GFp_simple_method(), EC_GFp_mont_method(),
1352   EC_GF2m_simple_method(), EC_GFp_nist_method(), EC_GFp_nistp224_method()
1353   EC_GFp_nistp256_method(), and EC_GFp_nistp521_method().
1354
1355   *Billy Bob Brumley*
1356
1357 * Deprecated EC_GROUP_new(), EC_GROUP_method_of(), and EC_POINT_method_of().
1358
1359   *Billy Bob Brumley*
1360
1361 * Add CAdES-BES signature verification support, mostly derived
1362   from ESSCertIDv2 TS (RFC 5816) contribution by Marek Klein.
1363
1364   *Filipe Raimundo da Silva*
1365
1366 * Add CAdES-BES signature scheme and attributes support (RFC 5126) to CMS API.
1367
1368   *Antonio Iacono*
1369
1370 * Added the AuthEnvelopedData content type structure (RFC 5083) with AES-GCM
1371   parameter (RFC 5084) for the Cryptographic Message Syntax (CMS).
1372
1373   *Jakub Zelenka*
1374
1375 * Deprecated EC_POINT_make_affine() and EC_POINTs_make_affine().
1376
1377   *Billy Bob Brumley*
1378
1379 * Deprecated EC_GROUP_precompute_mult(), EC_GROUP_have_precompute_mult(), and
1380   EC_KEY_precompute_mult().
1381
1382   *Billy Bob Brumley*
1383
1384 * Deprecated EC_POINTs_mul().
1385
1386   *Billy Bob Brumley*
1387
1388 * Removed FIPS_mode() and FIPS_mode_set().
1389
1390   *Shane Lontis*
1391
1392 * The SSL option SSL_OP_IGNORE_UNEXPECTED_EOF is introduced.
1393
1394   *Dmitry Belyavskiy*
1395
1396 * Deprecated EC_POINT_set_Jprojective_coordinates_GFp() and
1397   EC_POINT_get_Jprojective_coordinates_GFp().
1398
1399   *Billy Bob Brumley*
1400
1401 * Added OSSL_PARAM_BLD to the public interface.  This allows OSSL_PARAM
1402   arrays to be more easily constructed via a series of utility functions.
1403   Create a parameter builder using OSSL_PARAM_BLD_new(), add parameters using
1404   the various push functions and finally convert to a passable OSSL_PARAM
1405   array using OSSL_PARAM_BLD_to_param().
1406
1407   *Paul Dale*
1408
1409 * The security strength of SHA1 and MD5 based signatures in TLS has been
1410   reduced.
1411
1412   *Kurt Roeckx*
1413
1414 * Added EVP_PKEY_set_type_by_keymgmt(), to initialise an EVP_PKEY to
1415   contain a provider side internal key.
1416
1417   *Richard Levitte*
1418
1419 * ASN1_verify(), ASN1_digest() and ASN1_sign() have been deprecated.
1420
1421   *Richard Levitte*
1422
1423 * Project text documents not yet having a proper file name extension
1424   (`HACKING`, `LICENSE`, `NOTES*`, `README*`, `VERSION`) have been renamed to
1425   `*.md` as far as reasonable, else `*.txt`, for better use with file managers.
1426
1427   *David von Oheimb*
1428
1429 * The main project documents (README, NEWS, CHANGES, INSTALL, SUPPORT)
1430   have been converted to Markdown with the goal to produce documents
1431   which not only look pretty when viewed online in the browser, but
1432   remain well readable inside a plain text editor.
1433
1434   To achieve this goal, a 'minimalistic' Markdown style has been applied
1435   which avoids formatting elements that interfere too much with the
1436   reading flow in the text file. For example, it
1437
1438   * avoids [ATX headings][] and uses [setext headings][] instead
1439     (which works for `<h1>` and `<h2>` headings only).
1440   * avoids [inline links][] and uses [reference links][] instead.
1441   * avoids [fenced code blocks][] and uses [indented code blocks][] instead.
1442
1443     [ATX headings]:         https://github.github.com/gfm/#atx-headings
1444     [setext headings]:      https://github.github.com/gfm/#setext-headings
1445     [inline links]:         https://github.github.com/gfm/#inline-link
1446     [reference links]:      https://github.github.com/gfm/#reference-link
1447     [fenced code blocks]:   https://github.github.com/gfm/#fenced-code-blocks
1448     [indented code blocks]: https://github.github.com/gfm/#indented-code-blocks
1449
1450   *Matthias St. Pierre*
1451
1452 * The test suite is changed to preserve results of each test recipe.
1453   A new directory test-runs/ with subdirectories named like the
1454   test recipes are created in the build tree for this purpose.
1455
1456   *Richard Levitte*
1457
1458 * Added an implementation of CMP and CRMF (RFC 4210, RFC 4211 RFC 6712).
1459   This adds `crypto/cmp/`, `crpyto/crmf/`, `apps/cmp.c`, and `test/cmp_*`.
1460   See L<openssl-cmp(1)> and L<OSSL_CMP_exec_IR_ses(3)> as starting points.
1461
1462   *David von Oheimb, Martin Peylo*
1463
1464 * Generalized the HTTP client code from `crypto/ocsp/` into `crpyto/http/`.
1465   It supports arbitrary request and response content types, GET redirection,
1466   TLS, connections via HTTP(S) proxies, connections and exchange via
1467   user-defined BIOs (allowing implicit connections), persistent connections,
1468   and timeout checks.  See L<OSSL_HTTP_transfer(3)> etc. for details.
1469   The legacy OCSP-focused (and only partly documented) API
1470   is retained for backward compatibility, while most of it is deprecated.
1471
1472   *David von Oheimb*
1473
1474 * Added `util/check-format.pl`, a tool for checking adherence to the
1475   OpenSSL coding style <https://www.openssl.org/policies/codingstyle.html>.
1476   The checks performed are incomplete and yield some false positives.
1477   Still the tool should be useful for detecting most typical glitches.
1478
1479   *David von Oheimb*
1480
1481 * `BIO_do_connect()` and `BIO_do_handshake()` have been extended:
1482   If domain name resolution yields multiple IP addresses all of them are tried
1483   after `connect()` failures.
1484
1485   *David von Oheimb*
1486
1487 * All of the low level RSA functions have been deprecated.
1488
1489   *Paul Dale*
1490
1491 * X509 certificates signed using SHA1 are no longer allowed at security
1492   level 1 and above.
1493
1494   *Kurt Roeckx*
1495
1496 * The command line utilities dhparam, dsa, gendsa and dsaparam have been
1497   modified to use PKEY APIs.  These commands are now in maintenance mode
1498   and no new features will be added to them.
1499
1500   *Paul Dale*
1501
1502 * The command line utility rsautl has been deprecated.
1503
1504   *Paul Dale*
1505
1506 * The command line utilities genrsa and rsa have been modified to use PKEY
1507   APIs. They now write PKCS#8 keys by default. These commands are now in
1508   maintenance mode and no new features will be added to them.
1509
1510   *Paul Dale*
1511
1512 * All of the low level DH functions have been deprecated.
1513
1514   *Paul Dale and Matt Caswell*
1515
1516 * All of the low level DSA functions have been deprecated.
1517
1518   *Paul Dale*
1519
1520 * Reworked the treatment of EC EVP_PKEYs with the SM2 curve to
1521   automatically become EVP_PKEY_SM2 rather than EVP_PKEY_EC.
1522
1523   *Richard Levitte*
1524
1525 * Deprecated low level ECDH and ECDSA functions.
1526
1527   *Paul Dale*
1528
1529 * Deprecated EVP_PKEY_decrypt_old() and EVP_PKEY_encrypt_old().
1530
1531   *Richard Levitte*
1532
1533 * Enhanced the documentation of EVP_PKEY_get_size(), EVP_PKEY_get_bits()
1534   and EVP_PKEY_get_security_bits().  Especially EVP_PKEY_get_size() needed
1535   a new formulation to include all the things it can be used for,
1536   as well as words of caution.
1537
1538   *Richard Levitte*
1539
1540 * The SSL_CTX_set_tlsext_ticket_key_cb(3) function has been deprecated.
1541
1542   *Paul Dale*
1543
1544 * All of the low level HMAC functions have been deprecated.
1545
1546   *Paul Dale and David von Oheimb*
1547
1548 * Over two thousand fixes were made to the documentation, including:
1549   - Common options (such as -rand/-writerand, TLS version control, etc)
1550     were refactored and point to newly-enhanced descriptions in openssl.pod.
1551   - Added style conformance for all options (with help from Richard Levitte),
1552     documented all reported missing options, added a CI build to check
1553     that all options are documented and that no unimplemented options
1554     are documented.
1555   - Documented some internals, such as all use of environment variables.
1556   - Addressed all internal broken L<> references.
1557
1558   *Rich Salz*
1559
1560 * All of the low level CMAC functions have been deprecated.
1561
1562   *Paul Dale*
1563
1564 * The low-level MD2, MD4, MD5, MDC2, RIPEMD160 and Whirlpool digest
1565   functions have been deprecated.
1566
1567   *Paul Dale and David von Oheimb*
1568
1569 * Corrected the documentation of the return values from the `EVP_DigestSign*`
1570   set of functions.  The documentation mentioned negative values for some
1571   errors, but this was never the case, so the mention of negative values
1572   was removed.
1573
1574   Code that followed the documentation and thereby check with something
1575   like `EVP_DigestSignInit(...) <= 0` will continue to work undisturbed.
1576
1577   *Richard Levitte*
1578
1579 * All of the low level cipher functions have been deprecated.
1580
1581   *Matt Caswell and Paul Dale*
1582
1583 * Removed include/openssl/opensslconf.h.in and replaced it with
1584   include/openssl/configuration.h.in, which differs in not including
1585   <openssl/macros.h>.  A short header include/openssl/opensslconf.h
1586   was added to include both.
1587
1588   This allows internal hacks where one might need to modify the set
1589   of configured macros, for example this if deprecated symbols are
1590   still supposed to be available internally:
1591
1592       #include <openssl/configuration.h>
1593
1594       #undef OPENSSL_NO_DEPRECATED
1595       #define OPENSSL_SUPPRESS_DEPRECATED
1596
1597       #include <openssl/macros.h>
1598
1599   This should not be used by applications that use the exported
1600   symbols, as that will lead to linking errors.
1601
1602   *Richard Levitte*
1603
1604 * Fixed an overflow bug in the x64_64 Montgomery squaring procedure
1605   used in exponentiation with 512-bit moduli. No EC algorithms are
1606   affected. Analysis suggests that attacks against 2-prime RSA1024,
1607   3-prime RSA1536, and DSA1024 as a result of this defect would be very
1608   difficult to perform and are not believed likely. Attacks against DH512
1609   are considered just feasible. However, for an attack the target would
1610   have to re-use the DH512 private key, which is not recommended anyway.
1611   Also applications directly using the low-level API BN_mod_exp may be
1612   affected if they use BN_FLG_CONSTTIME.
1613   ([CVE-2019-1551])
1614
1615   *Andy Polyakov*
1616
1617 * Most memory-debug features have been deprecated, and the functionality
1618   replaced with no-ops.
1619
1620   *Rich Salz*
1621
1622 * Added documentation for the STACK API.
1623
1624   *Rich Salz*
1625
1626 * Introduced a new method type and API, OSSL_ENCODER, to represent
1627   generic encoders.  These do the same sort of job that PEM writers
1628   and d2i 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 * Introduced a new method type and API, OSSL_DECODER, to represent
1635   generic decoders.  These do the same sort of job that PEM readers
1636   and i2d functions do, but with support for methods supplied by
1637   providers, and the possibility for providers to support other
1638   formats as well.
1639
1640   *Richard Levitte*
1641
1642 * Added a .pragma directive to the syntax of configuration files, to
1643   allow varying behavior in a supported and predictable manner.
1644   Currently added pragma:
1645
1646           .pragma dollarid:on
1647
1648   This allows dollar signs to be a keyword character unless it's
1649   followed by a opening brace or parenthesis.  This is useful for
1650   platforms where dollar signs are commonly used in names, such as
1651   volume names and system directory names on VMS.
1652
1653   *Richard Levitte*
1654
1655 * Added functionality to create an EVP_PKEY from user data.
1656
1657   *Richard Levitte*
1658
1659 * Change the interpretation of the '--api' configuration option to
1660   mean that this is a desired API compatibility level with no
1661   further meaning.  The previous interpretation, that this would
1662   also mean to remove all deprecated symbols up to and including
1663   the given version, no requires that 'no-deprecated' is also used
1664   in the configuration.
1665
1666   When building applications, the desired API compatibility level
1667   can be set with the OPENSSL_API_COMPAT macro like before.  For
1668   API compatibility version below 3.0, the old style numerical
1669   value is valid as before, such as -DOPENSSL_API_COMPAT=0x10100000L.
1670   For version 3.0 and on, the value is expected to be the decimal
1671   value calculated from the major and minor version like this:
1672
1673           MAJOR * 10000 + MINOR * 100
1674
1675   Examples:
1676
1677           -DOPENSSL_API_COMPAT=30000             For 3.0
1678           -DOPENSSL_API_COMPAT=30200             For 3.2
1679
1680   To hide declarations that are deprecated up to and including the
1681   given API compatibility level, -DOPENSSL_NO_DEPRECATED must be
1682   given when building the application as well.
1683
1684   *Richard Levitte*
1685
1686 * Added the X509_LOOKUP_METHOD called X509_LOOKUP_store, to allow
1687   access to certificate and CRL stores via URIs and OSSL_STORE
1688   loaders.
1689
1690   This adds the following functions:
1691
1692   - X509_LOOKUP_store()
1693   - X509_STORE_load_file()
1694   - X509_STORE_load_path()
1695   - X509_STORE_load_store()
1696   - SSL_add_store_cert_subjects_to_stack()
1697   - SSL_CTX_set_default_verify_store()
1698   - SSL_CTX_load_verify_file()
1699   - SSL_CTX_load_verify_dir()
1700   - SSL_CTX_load_verify_store()
1701
1702   *Richard Levitte*
1703
1704 * Added a new method to gather entropy on VMS, based on SYS$GET_ENTROPY.
1705   The presence of this system service is determined at run-time.
1706
1707   *Richard Levitte*
1708
1709 * Added functionality to create an EVP_PKEY context based on data
1710   for methods from providers.  This takes an algorithm name and a
1711   property query string and simply stores them, with the intent
1712   that any operation that uses this context will use those strings
1713   to fetch the needed methods implicitly, thereby making the port
1714   of application written for pre-3.0 OpenSSL easier.
1715
1716   *Richard Levitte*
1717
1718 * The undocumented function NCONF_WIN32() has been deprecated; for
1719   conversion details see the HISTORY section of doc/man5/config.pod
1720
1721   *Rich Salz*
1722
1723 * Introduced the new functions EVP_DigestSignInit_ex() and
1724   EVP_DigestVerifyInit_ex(). The macros EVP_DigestSignUpdate() and
1725   EVP_DigestVerifyUpdate() have been converted to functions. See the man
1726   pages for further details.
1727
1728   *Matt Caswell*
1729
1730 * Over two thousand fixes were made to the documentation, including:
1731   adding missing command flags, better style conformance, documentation
1732   of internals, etc.
1733
1734   *Rich Salz, Richard Levitte*
1735
1736 * s390x assembly pack: add hardware-support for P-256, P-384, P-521,
1737   X25519, X448, Ed25519 and Ed448.
1738
1739   *Patrick Steuer*
1740
1741 * Print all values for a PKCS#12 attribute with 'openssl pkcs12', not just
1742   the first value.
1743
1744   *Jon Spillett*
1745
1746 * Deprecated the public definition of `ERR_STATE` as well as the function
1747   `ERR_get_state()`.  This is done in preparation of making `ERR_STATE` an
1748   opaque type.
1749
1750   *Richard Levitte*
1751
1752 * Added ERR functionality to give callers access to the stored function
1753   names that have replaced the older function code based functions.
1754
1755   New functions are ERR_peek_error_func(), ERR_peek_last_error_func(),
1756   ERR_peek_error_data(), ERR_peek_last_error_data(), ERR_get_error_all(),
1757   ERR_peek_error_all() and ERR_peek_last_error_all().
1758
1759   Deprecate ERR functions ERR_get_error_line(), ERR_get_error_line_data(),
1760   ERR_peek_error_line_data(), ERR_peek_last_error_line_data() and
1761   ERR_func_error_string().
1762
1763   *Richard Levitte*
1764
1765 * Extended testing to be verbose for failing tests only.  The make variables
1766   VERBOSE_FAILURE or VF can be used to enable this:
1767
1768           $ make VF=1 test                           # Unix
1769           $ mms /macro=(VF=1) test                   ! OpenVMS
1770           $ nmake VF=1 test                          # Windows
1771
1772   *Richard Levitte*
1773
1774 * Added the `-copy_extensions` option to the `x509` command for use with
1775   `-req` and `-x509toreq`. When given with the `copy` or `copyall` argument,
1776   all extensions in the request are copied to the certificate or vice versa.
1777
1778   *David von Oheimb*, *Kirill Stefanenkov <kirill_stefanenkov@rambler.ru>*
1779
1780 * Added the `-copy_extensions` option to the `req` command for use with
1781   `-x509`. When given with the `copy` or `copyall` argument,
1782   all extensions in the certification request are copied to the certificate.
1783
1784   *David von Oheimb*
1785
1786 * The `x509`, `req`, and `ca` commands now make sure that X.509v3 certificates
1787   they generate are by default RFC 5280 compliant in the following sense:
1788   There is a subjectKeyIdentifier extension with a hash value of the public key
1789   and for not self-signed certs there is an authorityKeyIdentifier extension
1790   with a keyIdentifier field or issuer information identifying the signing key.
1791   This is done unless some configuration overrides the new default behavior,
1792   such as `subjectKeyIdentifier = none` and `authorityKeyIdentifier = none`.
1793
1794   *David von Oheimb*
1795
1796 * Added several checks to `X509_verify_cert()` according to requirements in
1797   RFC 5280 in case `X509_V_FLAG_X509_STRICT` is set
1798   (which may be done by using the CLI option `-x509_strict`):
1799   * The basicConstraints of CA certificates must be marked critical.
1800   * CA certificates must explicitly include the keyUsage extension.
1801   * If a pathlenConstraint is given the key usage keyCertSign must be allowed.
1802   * The issuer name of any certificate must not be empty.
1803   * The subject name of CA certs, certs with keyUsage crlSign,
1804     and certs without subjectAlternativeName must not be empty.
1805   * If a subjectAlternativeName extension is given it must not be empty.
1806   * The signatureAlgorithm field and the cert signature must be consistent.
1807   * Any given authorityKeyIdentifier and any given subjectKeyIdentifier
1808     must not be marked critical.
1809   * The authorityKeyIdentifier must be given for X.509v3 certs
1810     unless they are self-signed.
1811   * The subjectKeyIdentifier must be given for all X.509v3 CA certs.
1812
1813   *David von Oheimb*
1814
1815 * Certificate verification using `X509_verify_cert()` meanwhile rejects EC keys
1816   with explicit curve parameters (specifiedCurve) as required by RFC 5480.
1817
1818   *Tomáš Mráz*
1819
1820 * For built-in EC curves, ensure an EC_GROUP built from the curve name is
1821   used even when parsing explicit parameters, when loading a encoded key
1822   or calling `EC_GROUP_new_from_ecpkparameters()`/
1823   `EC_GROUP_new_from_ecparameters()`.
1824   This prevents bypass of security hardening and performance gains,
1825   especially for curves with specialized EC_METHODs.
1826   By default, if a key encoded with explicit parameters is loaded and later
1827   encoded, the output is still encoded with explicit parameters, even if
1828   internally a "named" EC_GROUP is used for computation.
1829
1830   *Nicola Tuveri*
1831
1832 * Compute ECC cofactors if not provided during EC_GROUP construction. Before
1833   this change, EC_GROUP_set_generator would accept order and/or cofactor as
1834   NULL. After this change, only the cofactor parameter can be NULL. It also
1835   does some minimal sanity checks on the passed order.
1836   ([CVE-2019-1547])
1837
1838   *Billy Bob Brumley*
1839
1840 * Fixed a padding oracle in PKCS7_dataDecode and CMS_decrypt_set1_pkey.
1841   An attack is simple, if the first CMS_recipientInfo is valid but the
1842   second CMS_recipientInfo is chosen ciphertext. If the second
1843   recipientInfo decodes to PKCS #1 v1.5 form plaintext, the correct
1844   encryption key will be replaced by garbage, and the message cannot be
1845   decoded, but if the RSA decryption fails, the correct encryption key is
1846   used and the recipient will not notice the attack.
1847   As a work around for this potential attack the length of the decrypted
1848   key must be equal to the cipher default key length, in case the
1849   certifiate is not given and all recipientInfo are tried out.
1850   The old behaviour can be re-enabled in the CMS code by setting the
1851   CMS_DEBUG_DECRYPT flag.
1852
1853   *Bernd Edlinger*
1854
1855 * Early start up entropy quality from the DEVRANDOM seed source has been
1856   improved for older Linux systems.  The RAND subsystem will wait for
1857   /dev/random to be producing output before seeding from /dev/urandom.
1858   The seeded state is stored for future library initialisations using
1859   a system global shared memory segment.  The shared memory identifier
1860   can be configured by defining OPENSSL_RAND_SEED_DEVRANDOM_SHM_ID to
1861   the desired value.  The default identifier is 114.
1862
1863   *Paul Dale*
1864
1865 * Revised BN_generate_prime_ex to not avoid factors 2..17863 in p-1
1866   when primes for RSA keys are computed.
1867   Since we previously always generated primes == 2 (mod 3) for RSA keys,
1868   the 2-prime and 3-prime RSA modules were easy to distinguish, since
1869   `N = p*q = 1 (mod 3)`, but `N = p*q*r = 2 (mod 3)`. Therefore fingerprinting
1870   2-prime vs. 3-prime RSA keys was possible by computing N mod 3.
1871   This avoids possible fingerprinting of newly generated RSA modules.
1872
1873   *Bernd Edlinger*
1874
1875 * Correct the extended master secret constant on EBCDIC systems. Without this
1876   fix TLS connections between an EBCDIC system and a non-EBCDIC system that
1877   negotiate EMS will fail. Unfortunately this also means that TLS connections
1878   between EBCDIC systems with this fix, and EBCDIC systems without this
1879   fix will fail if they negotiate EMS.
1880
1881   *Matt Caswell*
1882
1883 * Changed the library initialisation so that the config file is now loaded
1884   by default. This was already the case for libssl. It now occurs for both
1885   libcrypto and libssl. Use the OPENSSL_INIT_NO_LOAD_CONFIG option to
1886   `OPENSSL_init_crypto()` to suppress automatic loading of a config file.
1887
1888   *Matt Caswell*
1889
1890 * Introduced new error raising macros, `ERR_raise()` and `ERR_raise_data()`,
1891   where the former acts as a replacement for `ERR_put_error()`, and the
1892   latter replaces the combination `ERR_put_error()` + `ERR_add_error_data()`.
1893   `ERR_raise_data()` adds more flexibility by taking a format string and
1894   an arbitrary number of arguments following it, to be processed with
1895   `BIO_snprintf()`.
1896
1897   *Richard Levitte*
1898
1899 * Introduced a new function, `OSSL_PROVIDER_available()`, which can be used
1900   to check if a named provider is loaded and available.  When called, it
1901   will also activate all fallback providers if such are still present.
1902
1903   *Richard Levitte*
1904
1905 * Enforce a minimum DH modulus size of 512 bits.
1906
1907   *Bernd Edlinger*
1908
1909 * Changed DH parameters to generate the order q subgroup instead of 2q.
1910   Previously generated DH parameters are still accepted by DH_check
1911   but DH_generate_key works around that by clearing bit 0 of the
1912   private key for those. This avoids leaking bit 0 of the private key.
1913
1914   *Bernd Edlinger*
1915
1916 * Significantly reduce secure memory usage by the randomness pools.
1917
1918   *Paul Dale*
1919
1920 * `{CRYPTO,OPENSSL}_mem_debug_{push,pop}` are now no-ops and have been
1921   deprecated.
1922
1923   *Rich Salz*
1924
1925 * A new type, EVP_KEYEXCH, has been introduced to represent key exchange
1926   algorithms. An implementation of a key exchange algorithm can be obtained
1927   by using the function EVP_KEYEXCH_fetch(). An EVP_KEYEXCH algorithm can be
1928   used in a call to EVP_PKEY_derive_init_ex() which works in a similar way to
1929   the older EVP_PKEY_derive_init() function. See the man pages for the new
1930   functions for further details.
1931
1932   *Matt Caswell*
1933
1934 * The EVP_PKEY_CTX_set_dh_pad() macro has now been converted to a function.
1935
1936   *Matt Caswell*
1937
1938 * Removed the function names from error messages and deprecated the
1939   xxx_F_xxx define's.
1940
1941   *Richard Levitte*
1942
1943 * Removed NextStep support and the macro OPENSSL_UNISTD
1944
1945   *Rich Salz*
1946
1947 * Removed DES_check_key.  Also removed OPENSSL_IMPLEMENT_GLOBAL,
1948   OPENSSL_GLOBAL_REF, OPENSSL_DECLARE_GLOBAL.
1949   Also removed "export var as function" capability; we do not export
1950   variables, only functions.
1951
1952   *Rich Salz*
1953
1954 * RC5_32_set_key has been changed to return an int type, with 0 indicating
1955   an error and 1 indicating success. In previous versions of OpenSSL this
1956   was a void type. If a key was set longer than the maximum possible this
1957   would crash.
1958
1959   *Matt Caswell*
1960
1961 * Support SM2 signing and verification schemes with X509 certificate.
1962
1963   *Paul Yang*
1964
1965 * Use SHA256 as the default digest for TS query in the `ts` app.
1966
1967   *Tomáš Mráz*
1968
1969 * Change PBKDF2 to conform to SP800-132 instead of the older PKCS5 RFC2898.
1970
1971   *Shane Lontis*
1972
1973 * Default cipher lists/suites are now available via a function, the
1974   #defines are deprecated.
1975
1976   *Todd Short*
1977
1978 * Add target VC-WIN32-UWP, VC-WIN64A-UWP, VC-WIN32-ARM-UWP and
1979   VC-WIN64-ARM-UWP in Windows OneCore target for making building libraries
1980   for Windows Store apps easier. Also, the "no-uplink" option has been added.
1981
1982   *Kenji Mouri*
1983
1984 * Join the directories crypto/x509 and crypto/x509v3
1985
1986   *Richard Levitte*
1987
1988 * Added command 'openssl kdf' that uses the EVP_KDF API.
1989
1990   *Shane Lontis*
1991
1992 * Added command 'openssl mac' that uses the EVP_MAC API.
1993
1994   *Shane Lontis*
1995
1996 * Added OPENSSL_info() to get diverse built-in OpenSSL data, such
1997   as default directories.  Also added the command 'openssl info'
1998   for scripting purposes.
1999
2000   *Richard Levitte*
2001
2002 * The functions AES_ige_encrypt() and AES_bi_ige_encrypt() have been
2003   deprecated.
2004
2005   *Matt Caswell*
2006
2007 * Add prediction resistance to the DRBG reseeding process.
2008
2009   *Paul Dale*
2010
2011 * Limit the number of blocks in a data unit for AES-XTS to 2^20 as
2012   mandated by IEEE Std 1619-2018.
2013
2014   *Paul Dale*
2015
2016 * Added newline escaping functionality to a filename when using openssl dgst.
2017   This output format is to replicate the output format found in the `*sum`
2018   checksum programs. This aims to preserve backward compatibility.
2019
2020   *Matt Eaton, Richard Levitte, and Paul Dale*
2021
2022 * Removed the heartbeat message in DTLS feature, as it has very
2023   little usage and doesn't seem to fulfill a valuable purpose.
2024   The configuration option is now deprecated.
2025
2026   *Richard Levitte*
2027
2028 * Changed the output of 'openssl {digestname} < file' to display the
2029   digest name in its output.
2030
2031   *Richard Levitte*
2032
2033 * Added a new generic trace API which provides support for enabling
2034   instrumentation through trace output.
2035
2036   *Richard Levitte & Matthias St. Pierre*
2037
2038 * Added build tests for C++.  These are generated files that only do one
2039   thing, to include one public OpenSSL head file each.  This tests that
2040   the public header files can be usefully included in a C++ application.
2041
2042   This test isn't enabled by default.  It can be enabled with the option
2043   'enable-buildtest-c++'.
2044
2045   *Richard Levitte*
2046
2047 * Added KB KDF (EVP_KDF_KB) to EVP_KDF.
2048
2049   *Robbie Harwood*
2050
2051 * Added SSH KDF (EVP_KDF_SSHKDF) and KRB5 KDF (EVP_KDF_KRB5KDF) to EVP_KDF.
2052
2053   *Simo Sorce*
2054
2055 * Added Single Step KDF (EVP_KDF_SS), X963 KDF, and X942 KDF to EVP_KDF.
2056
2057   *Shane Lontis*
2058
2059 * Added KMAC to EVP_MAC.
2060
2061   *Shane Lontis*
2062
2063 * Added property based algorithm implementation selection framework to
2064   the core.
2065
2066   *Paul Dale*
2067
2068 * Added SCA hardening for modular field inversion in EC_GROUP through
2069   a new dedicated field_inv() pointer in EC_METHOD.
2070   This also addresses a leakage affecting conversions from projective
2071   to affine coordinates.
2072
2073   *Billy Bob Brumley, Nicola Tuveri*
2074
2075 * Added EVP_KDF, an EVP layer KDF API, to simplify adding KDF and PRF
2076   implementations.  This includes an EVP_PKEY to EVP_KDF bridge for
2077   those algorithms that were already supported through the EVP_PKEY API
2078   (scrypt, TLS1 PRF and HKDF).  The low-level KDF functions for PBKDF2
2079   and scrypt are now wrappers that call EVP_KDF.
2080
2081   *David Makepeace*
2082
2083 * Build devcrypto engine as a dynamic engine.
2084
2085   *Eneas U de Queiroz*
2086
2087 * Add keyed BLAKE2 to EVP_MAC.
2088
2089   *Antoine Salon*
2090
2091 * Fix a bug in the computation of the endpoint-pair shared secret used
2092   by DTLS over SCTP. This breaks interoperability with older versions
2093   of OpenSSL like OpenSSL 1.1.0 and OpenSSL 1.0.2. There is a runtime
2094   switch SSL_MODE_DTLS_SCTP_LABEL_LENGTH_BUG (off by default) enabling
2095   interoperability with such broken implementations. However, enabling
2096   this switch breaks interoperability with correct implementations.
2097
2098 * Fix a use after free bug in d2i_X509_PUBKEY when overwriting a
2099   re-used X509_PUBKEY object if the second PUBKEY is malformed.
2100
2101   *Bernd Edlinger*
2102
2103 * Move strictness check from EVP_PKEY_asn1_new() to EVP_PKEY_asn1_add0().
2104
2105   *Richard Levitte*
2106
2107 * Changed the license to the Apache License v2.0.
2108
2109   *Richard Levitte*
2110
2111 * Switch to a new version scheme using three numbers MAJOR.MINOR.PATCH.
2112
2113   - Major releases (indicated by incrementing the MAJOR release number)
2114     may introduce incompatible API/ABI changes.
2115   - Minor releases (indicated by incrementing the MINOR release number)
2116     may introduce new features but retain API/ABI compatibility.
2117   - Patch releases (indicated by incrementing the PATCH number)
2118     are intended for bug fixes and other improvements of existing
2119     features only (like improving performance or adding documentation)
2120     and retain API/ABI compatibility.
2121
2122   *Richard Levitte*
2123
2124 * Add support for RFC5297 SIV mode (siv128), including AES-SIV.
2125
2126   *Todd Short*
2127
2128 * Remove the 'dist' target and add a tarball building script.  The
2129   'dist' target has fallen out of use, and it shouldn't be
2130   necessary to configure just to create a source distribution.
2131
2132   *Richard Levitte*
2133
2134 * Recreate the OS390-Unix config target.  It no longer relies on a
2135   special script like it did for OpenSSL pre-1.1.0.
2136
2137   *Richard Levitte*
2138
2139 * Instead of having the source directories listed in Configure, add
2140   a 'build.info' keyword SUBDIRS to indicate what sub-directories to
2141   look into.
2142
2143   *Richard Levitte*
2144
2145 * Add GMAC to EVP_MAC.
2146
2147   *Paul Dale*
2148
2149 * Ported the HMAC, CMAC and SipHash EVP_PKEY_METHODs to EVP_MAC.
2150
2151   *Richard Levitte*
2152
2153 * Added EVP_MAC, an EVP layer MAC API, to simplify adding MAC
2154   implementations.  This includes a generic EVP_PKEY to EVP_MAC bridge,
2155   to facilitate the continued use of MACs through raw private keys in
2156   functionality such as `EVP_DigestSign*` and `EVP_DigestVerify*`.
2157
2158   *Richard Levitte*
2159
2160 * Deprecate ECDH_KDF_X9_62().
2161
2162   *Antoine Salon*
2163
2164 * Added EVP_PKEY_ECDH_KDF_X9_63 and ecdh_KDF_X9_63() as replacements for
2165   the EVP_PKEY_ECDH_KDF_X9_62 KDF type and ECDH_KDF_X9_62(). The old names
2166   are retained for backwards compatibility.
2167
2168   *Antoine Salon*
2169
2170 * AES-XTS mode now enforces that its two keys are different to mitigate
2171   the attacked described in "Efficient Instantiations of Tweakable
2172   Blockciphers and Refinements to Modes OCB and PMAC" by Phillip Rogaway.
2173   Details of this attack can be obtained from:
2174   <http://web.cs.ucdavis.edu/%7Erogaway/papers/offsets.pdf>
2175
2176   *Paul Dale*
2177
2178 * Rename the object files, i.e. give them other names than in previous
2179   versions.  Their names now include the name of the final product, as
2180   well as its type mnemonic (bin, lib, shlib).
2181
2182   *Richard Levitte*
2183
2184 * Added new option for 'openssl list', '-objects', which will display the
2185   list of built in objects, i.e. OIDs with names.
2186
2187   *Richard Levitte*
2188
2189 * Added the options `-crl_lastupdate` and `-crl_nextupdate` to `openssl ca`,
2190   allowing the `lastUpdate` and `nextUpdate` fields in the generated CRL to
2191   be set explicitly.
2192
2193   *Chris Novakovic*
2194
2195 * Added support for Linux Kernel TLS data-path. The Linux Kernel data-path
2196   improves application performance by removing data copies and providing
2197   applications with zero-copy system calls such as sendfile and splice.
2198
2199   *Boris Pismenny*
2200
2201 * The SSL option SSL_OP_CLEANSE_PLAINTEXT is introduced.
2202
2203   *Martin Elshuber*
2204
2205 * `PKCS12_parse` now maintains the order of the parsed certificates
2206   when outputting them via `*ca` (rather than reversing it).
2207
2208   *David von Oheimb*
2209
2210 * Deprecated pthread fork support methods.
2211
2212   *Randall S. Becker*
2213
2214 * Added support for FFDHE key exchange in TLS 1.3.
2215
2216   *Raja Ashok*
2217
2218 * Added a new concept for OpenSSL plugability: providers.  This
2219   functionality is designed to replace the ENGINE API and ENGINE
2220   implementations, and to be much more dynamic, allowing provider
2221   authors to introduce new algorithms among other things, as long as
2222   there's an API that supports the algorithm type.
2223
2224   With this concept comes a new core API for interaction between
2225   libcrypto and provider implementations.  Public libcrypto functions
2226   that want to use providers do so through this core API.
2227
2228   The main documentation for this core API is found in
2229   doc/man7/provider.pod, doc/man7/provider-base.pod, and they in turn
2230   refer to other manuals describing the API specific for supported
2231   algorithm types (also called operations).
2232
2233   *The OpenSSL team*
2234
2235OpenSSL 1.1.1
2236-------------
2237
2238### Changes between 1.1.1l and 1.1.1m [xx XXX xxxx]
2239
2240 * Avoid loading of a dynamic engine twice.
2241
2242   *Bernd Edlinger*
2243
2244 * Prioritise DANE TLSA issuer certs over peer certs
2245
2246   *Viktor Dukhovni*
2247
2248 * Fixed random API for MacOS prior to 10.12
2249
2250   These MacOS versions don't support the CommonCrypto APIs
2251
2252   *Lenny Primak*
2253
2254### Changes between 1.1.1k and 1.1.1l [24 Aug 2021]
2255
2256 * Fixed an SM2 Decryption Buffer Overflow.
2257
2258   In order to decrypt SM2 encrypted data an application is expected to
2259   call the API function EVP_PKEY_decrypt(). Typically an application will
2260   call this function twice. The first time, on entry, the "out" parameter
2261   can be NULL and, on exit, the "outlen" parameter is populated with the
2262   buffer size required to hold the decrypted plaintext. The application
2263   can then allocate a sufficiently sized buffer and call EVP_PKEY_decrypt()
2264   again, but this time passing a non-NULL value for the "out" parameter.
2265
2266   A bug in the implementation of the SM2 decryption code means that the
2267   calculation of the buffer size required to hold the plaintext returned
2268   by the first call to EVP_PKEY_decrypt() can be smaller than the actual
2269   size required by the second call. This can lead to a buffer overflow
2270   when EVP_PKEY_decrypt() is called by the application a second time with
2271   a buffer that is too small.
2272
2273   A malicious attacker who is able present SM2 content for decryption to
2274   an application could cause attacker chosen data to overflow the buffer
2275   by up to a maximum of 62 bytes altering the contents of other data held
2276   after the buffer, possibly changing application behaviour or causing
2277   the application to crash. The location of the buffer is application
2278   dependent but is typically heap allocated.
2279   ([CVE-2021-3711])
2280
2281   *Matt Caswell*
2282
2283 * Fixed various read buffer overruns processing ASN.1 strings
2284
2285   ASN.1 strings are represented internally within OpenSSL as an ASN1_STRING
2286   structure which contains a buffer holding the string data and a field
2287   holding the buffer length. This contrasts with normal C strings which
2288   are repesented as a buffer for the string data which is terminated
2289   with a NUL (0) byte.
2290
2291   Although not a strict requirement, ASN.1 strings that are parsed using
2292   OpenSSL's own "d2i" functions (and other similar parsing functions) as
2293   well as any string whose value has been set with the ASN1_STRING_set()
2294   function will additionally NUL terminate the byte array in the
2295   ASN1_STRING structure.
2296
2297   However, it is possible for applications to directly construct valid
2298   ASN1_STRING structures which do not NUL terminate the byte array by
2299   directly setting the "data" and "length" fields in the ASN1_STRING
2300   array. This can also happen by using the ASN1_STRING_set0() function.
2301
2302   Numerous OpenSSL functions that print ASN.1 data have been found to
2303   assume that the ASN1_STRING byte array will be NUL terminated, even
2304   though this is not guaranteed for strings that have been directly
2305   constructed. Where an application requests an ASN.1 structure to be
2306   printed, and where that ASN.1 structure contains ASN1_STRINGs that have
2307   been directly constructed by the application without NUL terminating
2308   the "data" field, then a read buffer overrun can occur.
2309
2310   The same thing can also occur during name constraints processing
2311   of certificates (for example if a certificate has been directly
2312   constructed by the application instead of loading it via the OpenSSL
2313   parsing functions, and the certificate contains non NUL terminated
2314   ASN1_STRING structures). It can also occur in the X509_get1_email(),
2315   X509_REQ_get1_email() and X509_get1_ocsp() functions.
2316
2317   If a malicious actor can cause an application to directly construct an
2318   ASN1_STRING and then process it through one of the affected OpenSSL
2319   functions then this issue could be hit. This might result in a crash
2320   (causing a Denial of Service attack). It could also result in the
2321   disclosure of private memory contents (such as private keys, or
2322   sensitive plaintext).
2323   ([CVE-2021-3712])
2324
2325   *Matt Caswell*
2326
2327### Changes between 1.1.1j and 1.1.1k [25 Mar 2021]
2328
2329 * Fixed a problem with verifying a certificate chain when using the
2330   X509_V_FLAG_X509_STRICT flag. This flag enables additional security checks of
2331   the certificates present in a certificate chain. It is not set by default.
2332
2333   Starting from OpenSSL version 1.1.1h a check to disallow certificates in
2334   the chain that have explicitly encoded elliptic curve parameters was added
2335   as an additional strict check.
2336
2337   An error in the implementation of this check meant that the result of a
2338   previous check to confirm that certificates in the chain are valid CA
2339   certificates was overwritten. This effectively bypasses the check
2340   that non-CA certificates must not be able to issue other certificates.
2341
2342   If a "purpose" has been configured then there is a subsequent opportunity
2343   for checks that the certificate is a valid CA.  All of the named "purpose"
2344   values implemented in libcrypto perform this check.  Therefore, where
2345   a purpose is set the certificate chain will still be rejected even when the
2346   strict flag has been used. A purpose is set by default in libssl client and
2347   server certificate verification routines, but it can be overridden or
2348   removed by an application.
2349
2350   In order to be affected, an application must explicitly set the
2351   X509_V_FLAG_X509_STRICT verification flag and either not set a purpose
2352   for the certificate verification or, in the case of TLS client or server
2353   applications, override the default purpose.
2354   ([CVE-2021-3450])
2355
2356   *Tomáš Mráz*
2357
2358 * Fixed an issue where an OpenSSL TLS server may crash if sent a maliciously
2359   crafted renegotiation ClientHello message from a client. If a TLSv1.2
2360   renegotiation ClientHello omits the signature_algorithms extension (where it
2361   was present in the initial ClientHello), but includes a
2362   signature_algorithms_cert extension then a NULL pointer dereference will
2363   result, leading to a crash and a denial of service attack.
2364
2365   A server is only vulnerable if it has TLSv1.2 and renegotiation enabled
2366   (which is the default configuration). OpenSSL TLS clients are not impacted by
2367   this issue.
2368   ([CVE-2021-3449])
2369
2370   *Peter Kästle and Samuel Sapalski*
2371
2372### Changes between 1.1.1i and 1.1.1j [16 Feb 2021]
2373
2374 * Fixed the X509_issuer_and_serial_hash() function. It attempts to
2375   create a unique hash value based on the issuer and serial number data
2376   contained within an X509 certificate. However it was failing to correctly
2377   handle any errors that may occur while parsing the issuer field (which might
2378   occur if the issuer field is maliciously constructed). This may subsequently
2379   result in a NULL pointer deref and a crash leading to a potential denial of
2380   service attack.
2381   ([CVE-2021-23841])
2382
2383   *Matt Caswell*
2384
2385 * Fixed the RSA_padding_check_SSLv23() function and the RSA_SSLV23_PADDING
2386   padding mode to correctly check for rollback attacks. This is considered a
2387   bug in OpenSSL 1.1.1 because it does not support SSLv2. In 1.0.2 this is
2388   CVE-2021-23839.
2389
2390   *Matt Caswell*
2391
2392   Fixed the EVP_CipherUpdate, EVP_EncryptUpdate and EVP_DecryptUpdate
2393   functions. Previously they could overflow the output length argument in some
2394   cases where the input length is close to the maximum permissable length for
2395   an integer on the platform. In such cases the return value from the function
2396   call would be 1 (indicating success), but the output length value would be
2397   negative. This could cause applications to behave incorrectly or crash.
2398   ([CVE-2021-23840])
2399
2400   *Matt Caswell*
2401
2402 * Fixed SRP_Calc_client_key so that it runs in constant time. The previous
2403   implementation called BN_mod_exp without setting BN_FLG_CONSTTIME. This
2404   could be exploited in a side channel attack to recover the password. Since
2405   the attack is local host only this is outside of the current OpenSSL
2406   threat model and therefore no CVE is assigned.
2407
2408   Thanks to Mohammed Sabt and Daniel De Almeida Braga for reporting this
2409   issue.
2410
2411   *Matt Caswell*
2412
2413### Changes between 1.1.1h and 1.1.1i [8 Dec 2020]
2414
2415 * Fixed NULL pointer deref in the GENERAL_NAME_cmp function
2416   This function could crash if both GENERAL_NAMEs contain an EDIPARTYNAME.
2417    If an attacker can control both items being compared  then this could lead
2418    to a possible denial of service attack. OpenSSL itself uses the
2419    GENERAL_NAME_cmp function for two purposes:
2420    1) Comparing CRL distribution point names between an available CRL and a
2421       CRL distribution point embedded in an X509 certificate
2422    2) When verifying that a timestamp response token signer matches the
2423       timestamp authority name (exposed via the API functions
2424       TS_RESP_verify_response and TS_RESP_verify_token)
2425   ([CVE-2020-1971])
2426
2427   *Matt Caswell*
2428
2429### Changes between 1.1.1g and 1.1.1h [22 Sep 2020]
2430
2431 * Certificates with explicit curve parameters are now disallowed in
2432   verification chains if the X509_V_FLAG_X509_STRICT flag is used.
2433
2434   *Tomáš Mráz*
2435
2436 * The 'MinProtocol' and 'MaxProtocol' configuration commands now silently
2437   ignore TLS protocol version bounds when configuring DTLS-based contexts, and
2438   conversely, silently ignore DTLS protocol version bounds when configuring
2439   TLS-based contexts.  The commands can be repeated to set bounds of both
2440   types.  The same applies with the corresponding "min_protocol" and
2441   "max_protocol" command-line switches, in case some application uses both TLS
2442   and DTLS.
2443
2444   SSL_CTX instances that are created for a fixed protocol version (e.g.
2445   TLSv1_server_method()) also silently ignore version bounds.  Previously
2446   attempts to apply bounds to these protocol versions would result in an
2447   error.  Now only the "version-flexible" SSL_CTX instances are subject to
2448   limits in configuration files in command-line options.
2449
2450   *Viktor Dukhovni*
2451
2452 * Handshake now fails if Extended Master Secret extension is dropped
2453   on renegotiation.
2454
2455   *Tomáš Mráz*
2456
2457 * The Oracle Developer Studio compiler will start reporting deprecated APIs
2458
2459### Changes between 1.1.1f and 1.1.1g [21 Apr 2020]
2460
2461 * Fixed segmentation fault in SSL_check_chain()
2462   Server or client applications that call the SSL_check_chain() function
2463   during or after a TLS 1.3 handshake may crash due to a NULL pointer
2464   dereference as a result of incorrect handling of the
2465   "signature_algorithms_cert" TLS extension. The crash occurs if an invalid
2466   or unrecognised signature algorithm is received from the peer. This could
2467   be exploited by a malicious peer in a Denial of Service attack.
2468   ([CVE-2020-1967])
2469
2470   *Benjamin Kaduk*
2471
2472 * Added AES consttime code for no-asm configurations
2473   an optional constant time support for AES was added
2474   when building openssl for no-asm.
2475   Enable with: ./config no-asm -DOPENSSL_AES_CONST_TIME
2476   Disable with: ./config no-asm -DOPENSSL_NO_AES_CONST_TIME
2477   At this time this feature is by default disabled.
2478   It will be enabled by default in 3.0.
2479
2480   *Bernd Edlinger*
2481
2482### Changes between 1.1.1e and 1.1.1f [31 Mar 2020]
2483
2484 * Revert the change of EOF detection while reading in libssl to avoid
2485   regressions in applications depending on the current way of reporting
2486   the EOF. As the existing method is not fully accurate the change to
2487   reporting the EOF via SSL_ERROR_SSL is kept on the current development
2488   branch and will be present in the 3.0 release.
2489
2490   *Tomáš Mráz*
2491
2492 * Revised BN_generate_prime_ex to not avoid factors 3..17863 in p-1
2493   when primes for RSA keys are computed.
2494   Since we previously always generated primes == 2 (mod 3) for RSA keys,
2495   the 2-prime and 3-prime RSA modules were easy to distinguish, since
2496   N = p*q = 1 (mod 3), but N = p*q*r = 2 (mod 3). Therefore fingerprinting
2497   2-prime vs. 3-prime RSA keys was possible by computing N mod 3.
2498   This avoids possible fingerprinting of newly generated RSA modules.
2499
2500   *Bernd Edlinger*
2501
2502### Changes between 1.1.1d and 1.1.1e [17 Mar 2020]
2503
2504 * Properly detect EOF while reading in libssl. Previously if we hit an EOF
2505   while reading in libssl then we would report an error back to the
2506   application (SSL_ERROR_SYSCALL) but errno would be 0. We now add
2507   an error to the stack (which means we instead return SSL_ERROR_SSL) and
2508   therefore give a hint as to what went wrong.
2509
2510   *Matt Caswell*
2511
2512 * Check that ed25519 and ed448 are allowed by the security level. Previously
2513   signature algorithms not using an MD were not being checked that they were
2514   allowed by the security level.
2515
2516   *Kurt Roeckx*
2517
2518 * Fixed SSL_get_servername() behaviour. The behaviour of SSL_get_servername()
2519   was not quite right. The behaviour was not consistent between resumption
2520   and normal handshakes, and also not quite consistent with historical
2521   behaviour. The behaviour in various scenarios has been clarified and
2522   it has been updated to make it match historical behaviour as closely as
2523   possible.
2524
2525   *Matt Caswell*
2526
2527 * *[VMS only]* The header files that the VMS compilers include automatically,
2528   `__DECC_INCLUDE_PROLOGUE.H` and `__DECC_INCLUDE_EPILOGUE.H`, use pragmas
2529   that the C++ compiler doesn't understand.  This is a shortcoming in the
2530   compiler, but can be worked around with `__cplusplus` guards.
2531
2532   C++ applications that use OpenSSL libraries must be compiled using the
2533   qualifier `/NAMES=(AS_IS,SHORTENED)` to be able to use all the OpenSSL
2534   functions.  Otherwise, only functions with symbols of less than 31
2535   characters can be used, as the linker will not be able to successfully
2536   resolve symbols with longer names.
2537
2538   *Richard Levitte*
2539
2540 * Added a new method to gather entropy on VMS, based on SYS$GET_ENTROPY.
2541   The presence of this system service is determined at run-time.
2542
2543   *Richard Levitte*
2544
2545 * Print all values for a PKCS#12 attribute with 'openssl pkcs12', not just
2546   the first value.
2547
2548   *Jon Spillett*
2549
2550### Changes between 1.1.1c and 1.1.1d [10 Sep 2019]
2551
2552 * Fixed a fork protection issue. OpenSSL 1.1.1 introduced a rewritten random
2553   number generator (RNG). This was intended to include protection in the
2554   event of a fork() system call in order to ensure that the parent and child
2555   processes did not share the same RNG state. However this protection was not
2556   being used in the default case.
2557
2558   A partial mitigation for this issue is that the output from a high
2559   precision timer is mixed into the RNG state so the likelihood of a parent
2560   and child process sharing state is significantly reduced.
2561
2562   If an application already calls OPENSSL_init_crypto() explicitly using
2563   OPENSSL_INIT_ATFORK then this problem does not occur at all.
2564   ([CVE-2019-1549])
2565
2566   *Matthias St. Pierre*
2567
2568 * For built-in EC curves, ensure an EC_GROUP built from the curve name is
2569   used even when parsing explicit parameters, when loading a encoded key
2570   or calling `EC_GROUP_new_from_ecpkparameters()`/
2571   `EC_GROUP_new_from_ecparameters()`.
2572   This prevents bypass of security hardening and performance gains,
2573   especially for curves with specialized EC_METHODs.
2574   By default, if a key encoded with explicit parameters is loaded and later
2575   encoded, the output is still encoded with explicit parameters, even if
2576   internally a "named" EC_GROUP is used for computation.
2577
2578   *Nicola Tuveri*
2579
2580 * Compute ECC cofactors if not provided during EC_GROUP construction. Before
2581   this change, EC_GROUP_set_generator would accept order and/or cofactor as
2582   NULL. After this change, only the cofactor parameter can be NULL. It also
2583   does some minimal sanity checks on the passed order.
2584   ([CVE-2019-1547])
2585
2586   *Billy Bob Brumley*
2587
2588 * Fixed a padding oracle in PKCS7_dataDecode and CMS_decrypt_set1_pkey.
2589   An attack is simple, if the first CMS_recipientInfo is valid but the
2590   second CMS_recipientInfo is chosen ciphertext. If the second
2591   recipientInfo decodes to PKCS #1 v1.5 form plaintext, the correct
2592   encryption key will be replaced by garbage, and the message cannot be
2593   decoded, but if the RSA decryption fails, the correct encryption key is
2594   used and the recipient will not notice the attack.
2595   As a work around for this potential attack the length of the decrypted
2596   key must be equal to the cipher default key length, in case the
2597   certifiate is not given and all recipientInfo are tried out.
2598   The old behaviour can be re-enabled in the CMS code by setting the
2599   CMS_DEBUG_DECRYPT flag.
2600   ([CVE-2019-1563])
2601
2602   *Bernd Edlinger*
2603
2604 * Early start up entropy quality from the DEVRANDOM seed source has been
2605   improved for older Linux systems.  The RAND subsystem will wait for
2606   /dev/random to be producing output before seeding from /dev/urandom.
2607   The seeded state is stored for future library initialisations using
2608   a system global shared memory segment.  The shared memory identifier
2609   can be configured by defining OPENSSL_RAND_SEED_DEVRANDOM_SHM_ID to
2610   the desired value.  The default identifier is 114.
2611
2612   *Paul Dale*
2613
2614 * Correct the extended master secret constant on EBCDIC systems. Without this
2615   fix TLS connections between an EBCDIC system and a non-EBCDIC system that
2616   negotiate EMS will fail. Unfortunately this also means that TLS connections
2617   between EBCDIC systems with this fix, and EBCDIC systems without this
2618   fix will fail if they negotiate EMS.
2619
2620   *Matt Caswell*
2621
2622 * Use Windows installation paths in the mingw builds
2623
2624   Mingw isn't a POSIX environment per se, which means that Windows
2625   paths should be used for installation.
2626   ([CVE-2019-1552])
2627
2628   *Richard Levitte*
2629
2630 * Changed DH_check to accept parameters with order q and 2q subgroups.
2631   With order 2q subgroups the bit 0 of the private key is not secret
2632   but DH_generate_key works around that by clearing bit 0 of the
2633   private key for those. This avoids leaking bit 0 of the private key.
2634
2635   *Bernd Edlinger*
2636
2637 * Significantly reduce secure memory usage by the randomness pools.
2638
2639   *Paul Dale*
2640
2641 * Revert the DEVRANDOM_WAIT feature for Linux systems
2642
2643   The DEVRANDOM_WAIT feature added a select() call to wait for the
2644   /dev/random device to become readable before reading from the
2645   /dev/urandom device.
2646
2647   It turned out that this change had negative side effects on
2648   performance which were not acceptable. After some discussion it
2649   was decided to revert this feature and leave it up to the OS
2650   resp. the platform maintainer to ensure a proper initialization
2651   during early boot time.
2652
2653   *Matthias St. Pierre*
2654
2655### Changes between 1.1.1b and 1.1.1c [28 May 2019]
2656
2657 * Add build tests for C++.  These are generated files that only do one
2658   thing, to include one public OpenSSL head file each.  This tests that
2659   the public header files can be usefully included in a C++ application.
2660
2661   This test isn't enabled by default.  It can be enabled with the option
2662   'enable-buildtest-c++'.
2663
2664   *Richard Levitte*
2665
2666 * Enable SHA3 pre-hashing for ECDSA and DSA.
2667
2668   *Patrick Steuer*
2669
2670 * Change the default RSA, DSA and DH size to 2048 bit instead of 1024.
2671   This changes the size when using the `genpkey` command when no size is given.
2672   It fixes an omission in earlier changes that changed all RSA, DSA and DH
2673   generation commands to use 2048 bits by default.
2674
2675   *Kurt Roeckx*
2676
2677 * Reorganize the manual pages to consistently have RETURN VALUES,
2678   EXAMPLES, SEE ALSO and HISTORY come in that order, and adjust
2679   util/fix-doc-nits accordingly.
2680
2681   *Paul Yang, Joshua Lock*
2682
2683 * Add the missing accessor EVP_PKEY_get0_engine()
2684
2685   *Matt Caswell*
2686
2687 * Have commands like `s_client` and `s_server` output the signature scheme
2688   along with other cipher suite parameters when debugging.
2689
2690   *Lorinczy Zsigmond*
2691
2692 * Make OPENSSL_config() error agnostic again.
2693
2694   *Richard Levitte*
2695
2696 * Do the error handling in RSA decryption constant time.
2697
2698   *Bernd Edlinger*
2699
2700 * Prevent over long nonces in ChaCha20-Poly1305.
2701
2702   ChaCha20-Poly1305 is an AEAD cipher, and requires a unique nonce input
2703   for every encryption operation. RFC 7539 specifies that the nonce value
2704   (IV) should be 96 bits (12 bytes). OpenSSL allows a variable nonce length
2705   and front pads the nonce with 0 bytes if it is less than 12
2706   bytes. However it also incorrectly allows a nonce to be set of up to 16
2707   bytes. In this case only the last 12 bytes are significant and any
2708   additional leading bytes are ignored.
2709
2710   It is a requirement of using this cipher that nonce values are
2711   unique. Messages encrypted using a reused nonce value are susceptible to
2712   serious confidentiality and integrity attacks. If an application changes
2713   the default nonce length to be longer than 12 bytes and then makes a
2714   change to the leading bytes of the nonce expecting the new value to be a
2715   new unique nonce then such an application could inadvertently encrypt
2716   messages with a reused nonce.
2717
2718   Additionally the ignored bytes in a long nonce are not covered by the
2719   integrity guarantee of this cipher. Any application that relies on the
2720   integrity of these ignored leading bytes of a long nonce may be further
2721   affected. Any OpenSSL internal use of this cipher, including in SSL/TLS,
2722   is safe because no such use sets such a long nonce value. However user
2723   applications that use this cipher directly and set a non-default nonce
2724   length to be longer than 12 bytes may be vulnerable.
2725
2726   This issue was reported to OpenSSL on 16th of March 2019 by Joran Dirk
2727   Greef of Ronomon.
2728   ([CVE-2019-1543])
2729
2730   *Matt Caswell*
2731
2732 * Add DEVRANDOM_WAIT feature for Linux systems
2733
2734   On older Linux systems where the getrandom() system call is not available,
2735   OpenSSL normally uses the /dev/urandom device for seeding its CSPRNG.
2736   Contrary to getrandom(), the /dev/urandom device will not block during
2737   early boot when the kernel CSPRNG has not been seeded yet.
2738
2739   To mitigate this known weakness, use select() to wait for /dev/random to
2740   become readable before reading from /dev/urandom.
2741
2742 * Ensure that SM2 only uses SM3 as digest algorithm
2743
2744   *Paul Yang*
2745
2746### Changes between 1.1.1a and 1.1.1b [26 Feb 2019]
2747
2748 * Change the info callback signals for the start and end of a post-handshake
2749   message exchange in TLSv1.3. In 1.1.1/1.1.1a we used SSL_CB_HANDSHAKE_START
2750   and SSL_CB_HANDSHAKE_DONE. Experience has shown that many applications get
2751   confused by this and assume that a TLSv1.2 renegotiation has started. This
2752   can break KeyUpdate handling. Instead we no longer signal the start and end
2753   of a post handshake message exchange (although the messages themselves are
2754   still signalled). This could break some applications that were expecting
2755   the old signals. However without this KeyUpdate is not usable for many
2756   applications.
2757
2758   *Matt Caswell*
2759
2760### Changes between 1.1.1 and 1.1.1a [20 Nov 2018]
2761
2762 * Timing vulnerability in DSA signature generation
2763
2764   The OpenSSL DSA signature algorithm has been shown to be vulnerable to a
2765   timing side channel attack. An attacker could use variations in the signing
2766   algorithm to recover the private key.
2767
2768   This issue was reported to OpenSSL on 16th October 2018 by Samuel Weiser.
2769   ([CVE-2018-0734])
2770
2771   *Paul Dale*
2772
2773 * Timing vulnerability in ECDSA signature generation
2774
2775   The OpenSSL ECDSA signature algorithm has been shown to be vulnerable to a
2776   timing side channel attack. An attacker could use variations in the signing
2777   algorithm to recover the private key.
2778
2779   This issue was reported to OpenSSL on 25th October 2018 by Samuel Weiser.
2780   ([CVE-2018-0735])
2781
2782   *Paul Dale*
2783
2784 * Fixed the issue that RAND_add()/RAND_seed() silently discards random input
2785   if its length exceeds 4096 bytes. The limit has been raised to a buffer size
2786   of two gigabytes and the error handling improved.
2787
2788   This issue was reported to OpenSSL by Dr. Falko Strenzke. It has been
2789   categorized as a normal bug, not a security issue, because the DRBG reseeds
2790   automatically and is fully functional even without additional randomness
2791   provided by the application.
2792
2793### Changes between 1.1.0i and 1.1.1 [11 Sep 2018]
2794
2795 * Add a new ClientHello callback. Provides a callback interface that gives
2796   the application the ability to adjust the nascent SSL object at the
2797   earliest stage of ClientHello processing, immediately after extensions have
2798   been collected but before they have been processed. In particular, this
2799   callback can adjust the supported TLS versions in response to the contents
2800   of the ClientHello
2801
2802   *Benjamin Kaduk*
2803
2804 * Add SM2 base algorithm support.
2805
2806   *Jack Lloyd*
2807
2808 * s390x assembly pack: add (improved) hardware-support for the following
2809   cryptographic primitives: sha3, shake, aes-gcm, aes-ccm, aes-ctr, aes-ofb,
2810   aes-cfb/cfb8, aes-ecb.
2811
2812   *Patrick Steuer*
2813
2814 * Make EVP_PKEY_asn1_new() a bit stricter about its input.  A NULL pem_str
2815   parameter is no longer accepted, as it leads to a corrupt table.  NULL
2816   pem_str is reserved for alias entries only.
2817
2818   *Richard Levitte*
2819
2820 * Use the new ec_scalar_mul_ladder scaffold to implement a specialized ladder
2821   step for prime curves. The new implementation is based on formulae from
2822   differential addition-and-doubling in homogeneous projective coordinates
2823   from Izu-Takagi "A fast parallel elliptic curve multiplication resistant
2824   against side channel attacks" and Brier-Joye "Weierstrass Elliptic Curves
2825   and Side-Channel Attacks" Eq. (8) for y-coordinate recovery, modified
2826   to work in projective coordinates.
2827
2828   *Billy Bob Brumley, Nicola Tuveri*
2829
2830 * Change generating and checking of primes so that the error rate of not
2831   being prime depends on the intended use based on the size of the input.
2832   For larger primes this will result in more rounds of Miller-Rabin.
2833   The maximal error rate for primes with more than 1080 bits is lowered
2834   to 2^-128.
2835
2836   *Kurt Roeckx, Annie Yousar*
2837
2838 * Increase the number of Miller-Rabin rounds for DSA key generating to 64.
2839
2840   *Kurt Roeckx*
2841
2842 * The 'tsget' script is renamed to 'tsget.pl', to avoid confusion when
2843   moving between systems, and to avoid confusion when a Windows build is
2844   done with mingw vs with MSVC.  For POSIX installs, there's still a
2845   symlink or copy named 'tsget' to avoid that confusion as well.
2846
2847   *Richard Levitte*
2848
2849 * Revert blinding in ECDSA sign and instead make problematic addition
2850   length-invariant. Switch even to fixed-length Montgomery multiplication.
2851
2852   *Andy Polyakov*
2853
2854 * Use the new ec_scalar_mul_ladder scaffold to implement a specialized ladder
2855   step for binary curves. The new implementation is based on formulae from
2856   differential addition-and-doubling in mixed Lopez-Dahab projective
2857   coordinates, modified to independently blind the operands.
2858
2859   *Billy Bob Brumley, Sohaib ul Hassan, Nicola Tuveri*
2860
2861 * Add a scaffold to optionally enhance the Montgomery ladder implementation
2862   for `ec_scalar_mul_ladder` (formerly `ec_mul_consttime`) allowing
2863   EC_METHODs to implement their own specialized "ladder step", to take
2864   advantage of more favorable coordinate systems or more efficient
2865   differential addition-and-doubling algorithms.
2866
2867   *Billy Bob Brumley, Sohaib ul Hassan, Nicola Tuveri*
2868
2869 * Modified the random device based seed sources to keep the relevant
2870   file descriptors open rather than reopening them on each access.
2871   This allows such sources to operate in a chroot() jail without
2872   the associated device nodes being available. This behaviour can be
2873   controlled using RAND_keep_random_devices_open().
2874
2875   *Paul Dale*
2876
2877 * Numerous side-channel attack mitigations have been applied. This may have
2878   performance impacts for some algorithms for the benefit of improved
2879   security. Specific changes are noted in this change log by their respective
2880   authors.
2881
2882   *Matt Caswell*
2883
2884 * AIX shared library support overhaul. Switch to AIX "natural" way of
2885   handling shared libraries, which means collecting shared objects of
2886   different versions and bitnesses in one common archive. This allows to
2887   mitigate conflict between 1.0 and 1.1 side-by-side installations. It
2888   doesn't affect the way 3rd party applications are linked, only how
2889   multi-version installation is managed.
2890
2891   *Andy Polyakov*
2892
2893 * Make ec_group_do_inverse_ord() more robust and available to other
2894   EC cryptosystems, so that irrespective of BN_FLG_CONSTTIME, SCA
2895   mitigations are applied to the fallback BN_mod_inverse().
2896   When using this function rather than BN_mod_inverse() directly, new
2897   EC cryptosystem implementations are then safer-by-default.
2898
2899   *Billy Bob Brumley*
2900
2901 * Add coordinate blinding for EC_POINT and implement projective
2902   coordinate blinding for generic prime curves as a countermeasure to
2903   chosen point SCA attacks.
2904
2905   *Sohaib ul Hassan, Nicola Tuveri, Billy Bob Brumley*
2906
2907 * Add blinding to ECDSA and DSA signatures to protect against side channel
2908   attacks discovered by Keegan Ryan (NCC Group).
2909
2910   *Matt Caswell*
2911
2912 * Enforce checking in the `pkeyutl` command to ensure that the input
2913   length does not exceed the maximum supported digest length when performing
2914   a sign, verify or verifyrecover operation.
2915
2916   *Matt Caswell*
2917
2918 * SSL_MODE_AUTO_RETRY is enabled by default. Applications that use blocking
2919   I/O in combination with something like select() or poll() will hang. This
2920   can be turned off again using SSL_CTX_clear_mode().
2921   Many applications do not properly handle non-application data records, and
2922   TLS 1.3 sends more of such records. Setting SSL_MODE_AUTO_RETRY works
2923   around the problems in those applications, but can also break some.
2924   It's recommended to read the manpages about SSL_read(), SSL_write(),
2925   SSL_get_error(), SSL_shutdown(), SSL_CTX_set_mode() and
2926   SSL_CTX_set_read_ahead() again.
2927
2928   *Kurt Roeckx*
2929
2930 * When unlocking a pass phrase protected PEM file or PKCS#8 container, we
2931   now allow empty (zero character) pass phrases.
2932
2933   *Richard Levitte*
2934
2935 * Apply blinding to binary field modular inversion and remove patent
2936   pending (OPENSSL_SUN_GF2M_DIV) BN_GF2m_mod_div implementation.
2937
2938   *Billy Bob Brumley*
2939
2940 * Deprecate ec2_mult.c and unify scalar multiplication code paths for
2941   binary and prime elliptic curves.
2942
2943   *Billy Bob Brumley*
2944
2945 * Remove ECDSA nonce padding: EC_POINT_mul is now responsible for
2946   constant time fixed point multiplication.
2947
2948   *Billy Bob Brumley*
2949
2950 * Revise elliptic curve scalar multiplication with timing attack
2951   defenses: ec_wNAF_mul redirects to a constant time implementation
2952   when computing fixed point and variable point multiplication (which
2953   in OpenSSL are mostly used with secret scalars in keygen, sign,
2954   ECDH derive operations).
2955   *Billy Bob Brumley, Nicola Tuveri, Cesar Pereida García,
2956    Sohaib ul Hassan*
2957
2958 * Updated CONTRIBUTING
2959
2960   *Rich Salz*
2961
2962 * Updated DRBG / RAND to request nonce and additional low entropy
2963   randomness from the system.
2964
2965   *Matthias St. Pierre*
2966
2967 * Updated 'openssl rehash' to use OpenSSL consistent default.
2968
2969   *Richard Levitte*
2970
2971 * Moved the load of the ssl_conf module to libcrypto, which helps
2972   loading engines that libssl uses before libssl is initialised.
2973
2974   *Matt Caswell*
2975
2976 * Added EVP_PKEY_sign() and EVP_PKEY_verify() for EdDSA
2977
2978   *Matt Caswell*
2979
2980 * Fixed X509_NAME_ENTRY_set to get multi-valued RDNs right in all cases.
2981
2982   *Ingo Schwarze, Rich Salz*
2983
2984 * Added output of accepting IP address and port for 'openssl s_server'
2985
2986   *Richard Levitte*
2987
2988 * Added a new API for TLSv1.3 ciphersuites:
2989      SSL_CTX_set_ciphersuites()
2990      SSL_set_ciphersuites()
2991
2992   *Matt Caswell*
2993
2994 * Memory allocation failures consistently add an error to the error
2995   stack.
2996
2997   *Rich Salz*
2998
2999 * Don't use OPENSSL_ENGINES and OPENSSL_CONF environment values
3000   in libcrypto when run as setuid/setgid.
3001
3002   *Bernd Edlinger*
3003
3004 * Load any config file by default when libssl is used.
3005
3006   *Matt Caswell*
3007
3008 * Added new public header file <openssl/rand_drbg.h> and documentation
3009   for the RAND_DRBG API. See manual page RAND_DRBG(7) for an overview.
3010
3011   *Matthias St. Pierre*
3012
3013 * QNX support removed (cannot find contributors to get their approval
3014   for the license change).
3015
3016   *Rich Salz*
3017
3018 * TLSv1.3 replay protection for early data has been implemented. See the
3019   SSL_read_early_data() man page for further details.
3020
3021   *Matt Caswell*
3022
3023 * Separated TLSv1.3 ciphersuite configuration out from TLSv1.2 ciphersuite
3024   configuration. TLSv1.3 ciphersuites are not compatible with TLSv1.2 and
3025   below. Similarly TLSv1.2 ciphersuites are not compatible with TLSv1.3.
3026   In order to avoid issues where legacy TLSv1.2 ciphersuite configuration
3027   would otherwise inadvertently disable all TLSv1.3 ciphersuites the
3028   configuration has been separated out. See the ciphers man page or the
3029   SSL_CTX_set_ciphersuites() man page for more information.
3030
3031   *Matt Caswell*
3032
3033 * On POSIX (BSD, Linux, ...) systems the ocsp(1) command running
3034   in responder mode now supports the new "-multi" option, which
3035   spawns the specified number of child processes to handle OCSP
3036   requests.  The "-timeout" option now also limits the OCSP
3037   responder's patience to wait to receive the full client request
3038   on a newly accepted connection. Child processes are respawned
3039   as needed, and the CA index file is automatically reloaded
3040   when changed.  This makes it possible to run the "ocsp" responder
3041   as a long-running service, making the OpenSSL CA somewhat more
3042   feature-complete.  In this mode, most diagnostic messages logged
3043   after entering the event loop are logged via syslog(3) rather than
3044   written to stderr.
3045
3046   *Viktor Dukhovni*
3047
3048 * Added support for X448 and Ed448. Heavily based on original work by
3049   Mike Hamburg.
3050
3051   *Matt Caswell*
3052
3053 * Extend OSSL_STORE with capabilities to search and to narrow the set of
3054   objects loaded.  This adds the functions OSSL_STORE_expect() and
3055   OSSL_STORE_find() as well as needed tools to construct searches and
3056   get the search data out of them.
3057
3058   *Richard Levitte*
3059
3060 * Support for TLSv1.3 added. Note that users upgrading from an earlier
3061   version of OpenSSL should review their configuration settings to ensure
3062   that they are still appropriate for TLSv1.3. For further information see:
3063   <https://wiki.openssl.org/index.php/TLS1.3>
3064
3065   *Matt Caswell*
3066
3067 * Grand redesign of the OpenSSL random generator
3068
3069   The default RAND method now utilizes an AES-CTR DRBG according to
3070   NIST standard SP 800-90Ar1. The new random generator is essentially
3071   a port of the default random generator from the OpenSSL FIPS 2.0
3072   object module. It is a hybrid deterministic random bit generator
3073   using an AES-CTR bit stream and which seeds and reseeds itself
3074   automatically using trusted system entropy sources.
3075
3076   Some of its new features are:
3077    - Support for multiple DRBG instances with seed chaining.
3078    - The default RAND method makes use of a DRBG.
3079    - There is a public and private DRBG instance.
3080    - The DRBG instances are fork-safe.
3081    - Keep all global DRBG instances on the secure heap if it is enabled.
3082    - The public and private DRBG instance are per thread for lock free
3083      operation
3084
3085   *Paul Dale, Benjamin Kaduk, Kurt Roeckx, Rich Salz, Matthias St. Pierre*
3086
3087 * Changed Configure so it only says what it does and doesn't dump
3088   so much data.  Instead, ./configdata.pm should be used as a script
3089   to display all sorts of configuration data.
3090
3091   *Richard Levitte*
3092
3093 * Added processing of "make variables" to Configure.
3094
3095   *Richard Levitte*
3096
3097 * Added SHA512/224 and SHA512/256 algorithm support.
3098
3099   *Paul Dale*
3100
3101 * The last traces of Netware support, first removed in 1.1.0, have
3102   now been removed.
3103
3104   *Rich Salz*
3105
3106 * Get rid of Makefile.shared, and in the process, make the processing
3107   of certain files (rc.obj, or the .def/.map/.opt files produced from
3108   the ordinal files) more visible and hopefully easier to trace and
3109   debug (or make silent).
3110
3111   *Richard Levitte*
3112
3113 * Make it possible to have environment variable assignments as
3114   arguments to config / Configure.
3115
3116   *Richard Levitte*
3117
3118 * Add multi-prime RSA (RFC 8017) support.
3119
3120   *Paul Yang*
3121
3122 * Add SM3 implemented according to GB/T 32905-2016
3123   *Jack Lloyd <jack.lloyd@ribose.com>,*
3124   *Ronald Tse <ronald.tse@ribose.com>,*
3125   *Erick Borsboom <erick.borsboom@ribose.com>*
3126
3127 * Add 'Maximum Fragment Length' TLS extension negotiation and support
3128   as documented in RFC6066.
3129   Based on a patch from Tomasz Moń
3130
3131   *Filipe Raimundo da Silva*
3132
3133 * Add SM4 implemented according to GB/T 32907-2016.
3134   *Jack Lloyd <jack.lloyd@ribose.com>,*
3135   *Ronald Tse <ronald.tse@ribose.com>,*
3136   *Erick Borsboom <erick.borsboom@ribose.com>*
3137
3138 * Reimplement -newreq-nodes and ERR_error_string_n; the
3139   original author does not agree with the license change.
3140
3141   *Rich Salz*
3142
3143 * Add ARIA AEAD TLS support.
3144
3145   *Jon Spillett*
3146
3147 * Some macro definitions to support VS6 have been removed.  Visual
3148   Studio 6 has not worked since 1.1.0
3149
3150   *Rich Salz*
3151
3152 * Add ERR_clear_last_mark(), to allow callers to clear the last mark
3153   without clearing the errors.
3154
3155   *Richard Levitte*
3156
3157 * Add "atfork" functions.  If building on a system that without
3158   pthreads, see doc/man3/OPENSSL_fork_prepare.pod for application
3159   requirements.  The RAND facility now uses/requires this.
3160
3161   *Rich Salz*
3162
3163 * Add SHA3.
3164
3165   *Andy Polyakov*
3166
3167 * The UI API becomes a permanent and integral part of libcrypto, i.e.
3168   not possible to disable entirely.  However, it's still possible to
3169   disable the console reading UI method, UI_OpenSSL() (use UI_null()
3170   as a fallback).
3171
3172   To disable, configure with 'no-ui-console'.  'no-ui' is still
3173   possible to use as an alias.  Check at compile time with the
3174   macro OPENSSL_NO_UI_CONSOLE.  The macro OPENSSL_NO_UI is still
3175   possible to check and is an alias for OPENSSL_NO_UI_CONSOLE.
3176
3177   *Richard Levitte*
3178
3179 * Add a STORE module, which implements a uniform and URI based reader of
3180   stores that can contain keys, certificates, CRLs and numerous other
3181   objects.  The main API is loosely based on a few stdio functions,
3182   and includes OSSL_STORE_open, OSSL_STORE_load, OSSL_STORE_eof,
3183   OSSL_STORE_error and OSSL_STORE_close.
3184   The implementation uses backends called "loaders" to implement arbitrary
3185   URI schemes.  There is one built in "loader" for the 'file' scheme.
3186
3187   *Richard Levitte*
3188
3189 * Add devcrypto engine.  This has been implemented against cryptodev-linux,
3190   then adjusted to work on FreeBSD 8.4 as well.
3191   Enable by configuring with 'enable-devcryptoeng'.  This is done by default
3192   on BSD implementations, as cryptodev.h is assumed to exist on all of them.
3193
3194   *Richard Levitte*
3195
3196 * Module names can prefixed with OSSL_ or OPENSSL_.  This affects
3197   util/mkerr.pl, which is adapted to allow those prefixes, leading to
3198   error code calls like this:
3199
3200           OSSL_FOOerr(OSSL_FOO_F_SOMETHING, OSSL_FOO_R_WHATEVER);
3201
3202   With this change, we claim the namespaces OSSL and OPENSSL in a manner
3203   that can be encoded in C.  For the foreseeable future, this will only
3204   affect new modules.
3205
3206   *Richard Levitte and Tim Hudson*
3207
3208 * Removed BSD cryptodev engine.
3209
3210   *Rich Salz*
3211
3212 * Add a build target 'build_all_generated', to build all generated files
3213   and only that.  This can be used to prepare everything that requires
3214   things like perl for a system that lacks perl and then move everything
3215   to that system and do the rest of the build there.
3216
3217   *Richard Levitte*
3218
3219 * In the UI interface, make it possible to duplicate the user data.  This
3220   can be used by engines that need to retain the data for a longer time
3221   than just the call where this user data is passed.
3222
3223   *Richard Levitte*
3224
3225 * Ignore the '-named_curve auto' value for compatibility of applications
3226   with OpenSSL 1.0.2.
3227
3228   *Tomáš Mráz <tmraz@fedoraproject.org>*
3229
3230 * Fragmented SSL/TLS alerts are no longer accepted. An alert message is 2
3231   bytes long. In theory it is permissible in SSLv3 - TLSv1.2 to fragment such
3232   alerts across multiple records (some of which could be empty). In practice
3233   it make no sense to send an empty alert record, or to fragment one. TLSv1.3
3234   prohibits this altogether and other libraries (BoringSSL, NSS) do not
3235   support this at all. Supporting it adds significant complexity to the
3236   record layer, and its removal is unlikely to cause interoperability
3237   issues.
3238
3239   *Matt Caswell*
3240
3241 * Add the ASN.1 types INT32, UINT32, INT64, UINT64 and variants prefixed
3242   with Z.  These are meant to replace LONG and ZLONG and to be size safe.
3243   The use of LONG and ZLONG is discouraged and scheduled for deprecation
3244   in OpenSSL 1.2.0.
3245
3246   *Richard Levitte*
3247
3248 * Add the 'z' and 'j' modifiers to BIO_printf() et al formatting string,
3249   'z' is to be used for [s]size_t, and 'j' - with [u]int64_t.
3250
3251   *Richard Levitte, Andy Polyakov*
3252
3253 * Add EC_KEY_get0_engine(), which does for EC_KEY what RSA_get0_engine()
3254   does for RSA, etc.
3255
3256   *Richard Levitte*
3257
3258 * Have 'config' recognise 64-bit mingw and choose 'mingw64' as the target
3259   platform rather than 'mingw'.
3260
3261   *Richard Levitte*
3262
3263 * The functions X509_STORE_add_cert and X509_STORE_add_crl return
3264   success if they are asked to add an object which already exists
3265   in the store. This change cascades to other functions which load
3266   certificates and CRLs.
3267
3268   *Paul Dale*
3269
3270 * x86_64 assembly pack: annotate code with DWARF CFI directives to
3271   facilitate stack unwinding even from assembly subroutines.
3272
3273   *Andy Polyakov*
3274
3275 * Remove VAX C specific definitions of OPENSSL_EXPORT, OPENSSL_EXTERN.
3276   Also remove OPENSSL_GLOBAL entirely, as it became a no-op.
3277
3278   *Richard Levitte*
3279
3280 * Remove the VMS-specific reimplementation of gmtime from crypto/o_times.c.
3281   VMS C's RTL has a fully up to date gmtime() and gmtime_r() since V7.1,
3282   which is the minimum version we support.
3283
3284   *Richard Levitte*
3285
3286 * Certificate time validation (X509_cmp_time) enforces stricter
3287   compliance with RFC 5280. Fractional seconds and timezone offsets
3288   are no longer allowed.
3289
3290   *Emilia Käsper*
3291
3292 * Add support for ARIA
3293
3294   *Paul Dale*
3295
3296 * s_client will now send the Server Name Indication (SNI) extension by
3297   default unless the new "-noservername" option is used. The server name is
3298   based on the host provided to the "-connect" option unless overridden by
3299   using "-servername".
3300
3301   *Matt Caswell*
3302
3303 * Add support for SipHash
3304
3305   *Todd Short*
3306
3307 * OpenSSL now fails if it receives an unrecognised record type in TLS1.0
3308   or TLS1.1. Previously this only happened in SSLv3 and TLS1.2. This is to
3309   prevent issues where no progress is being made and the peer continually
3310   sends unrecognised record types, using up resources processing them.
3311
3312   *Matt Caswell*
3313
3314 * 'openssl passwd' can now produce SHA256 and SHA512 based output,
3315   using the algorithm defined in
3316   <https://www.akkadia.org/drepper/SHA-crypt.txt>
3317
3318   *Richard Levitte*
3319
3320 * Heartbeat support has been removed; the ABI is changed for now.
3321
3322   *Richard Levitte, Rich Salz*
3323
3324 * Support for SSL_OP_NO_ENCRYPT_THEN_MAC in SSL_CONF_cmd.
3325
3326   *Emilia Käsper*
3327
3328 * The RSA "null" method, which was partially supported to avoid patent
3329   issues, has been replaced to always returns NULL.
3330
3331   *Rich Salz*
3332
3333OpenSSL 1.1.0
3334-------------
3335
3336### Changes between 1.1.0k and 1.1.0l [10 Sep 2019]
3337
3338 * For built-in EC curves, ensure an EC_GROUP built from the curve name is
3339   used even when parsing explicit parameters, when loading a encoded key
3340   or calling `EC_GROUP_new_from_ecpkparameters()`/
3341   `EC_GROUP_new_from_ecparameters()`.
3342   This prevents bypass of security hardening and performance gains,
3343   especially for curves with specialized EC_METHODs.
3344   By default, if a key encoded with explicit parameters is loaded and later
3345   encoded, the output is still encoded with explicit parameters, even if
3346   internally a "named" EC_GROUP is used for computation.
3347
3348   *Nicola Tuveri*
3349
3350 * Compute ECC cofactors if not provided during EC_GROUP construction. Before
3351   this change, EC_GROUP_set_generator would accept order and/or cofactor as
3352   NULL. After this change, only the cofactor parameter can be NULL. It also
3353   does some minimal sanity checks on the passed order.
3354   ([CVE-2019-1547])
3355
3356   *Billy Bob Brumley*
3357
3358 * Fixed a padding oracle in PKCS7_dataDecode and CMS_decrypt_set1_pkey.
3359   An attack is simple, if the first CMS_recipientInfo is valid but the
3360   second CMS_recipientInfo is chosen ciphertext. If the second
3361   recipientInfo decodes to PKCS #1 v1.5 form plaintext, the correct
3362   encryption key will be replaced by garbage, and the message cannot be
3363   decoded, but if the RSA decryption fails, the correct encryption key is
3364   used and the recipient will not notice the attack.
3365   As a work around for this potential attack the length of the decrypted
3366   key must be equal to the cipher default key length, in case the
3367   certifiate is not given and all recipientInfo are tried out.
3368   The old behaviour can be re-enabled in the CMS code by setting the
3369   CMS_DEBUG_DECRYPT flag.
3370   ([CVE-2019-1563])
3371
3372   *Bernd Edlinger*
3373
3374 * Use Windows installation paths in the mingw builds
3375
3376   Mingw isn't a POSIX environment per se, which means that Windows
3377   paths should be used for installation.
3378   ([CVE-2019-1552])
3379
3380   *Richard Levitte*
3381
3382### Changes between 1.1.0j and 1.1.0k [28 May 2019]
3383
3384 * Change the default RSA, DSA and DH size to 2048 bit instead of 1024.
3385   This changes the size when using the `genpkey` command when no size is given.
3386   It fixes an omission in earlier changes that changed all RSA, DSA and DH
3387   generation commands to use 2048 bits by default.
3388
3389   *Kurt Roeckx*
3390
3391 * Prevent over long nonces in ChaCha20-Poly1305.
3392
3393   ChaCha20-Poly1305 is an AEAD cipher, and requires a unique nonce input
3394   for every encryption operation. RFC 7539 specifies that the nonce value
3395   (IV) should be 96 bits (12 bytes). OpenSSL allows a variable nonce length
3396   and front pads the nonce with 0 bytes if it is less than 12
3397   bytes. However it also incorrectly allows a nonce to be set of up to 16
3398   bytes. In this case only the last 12 bytes are significant and any
3399   additional leading bytes are ignored.
3400
3401   It is a requirement of using this cipher that nonce values are
3402   unique. Messages encrypted using a reused nonce value are susceptible to
3403   serious confidentiality and integrity attacks. If an application changes
3404   the default nonce length to be longer than 12 bytes and then makes a
3405   change to the leading bytes of the nonce expecting the new value to be a
3406   new unique nonce then such an application could inadvertently encrypt
3407   messages with a reused nonce.
3408
3409   Additionally the ignored bytes in a long nonce are not covered by the
3410   integrity guarantee of this cipher. Any application that relies on the
3411   integrity of these ignored leading bytes of a long nonce may be further
3412   affected. Any OpenSSL internal use of this cipher, including in SSL/TLS,
3413   is safe because no such use sets such a long nonce value. However user
3414   applications that use this cipher directly and set a non-default nonce
3415   length to be longer than 12 bytes may be vulnerable.
3416
3417   This issue was reported to OpenSSL on 16th of March 2019 by Joran Dirk
3418   Greef of Ronomon.
3419   ([CVE-2019-1543])
3420
3421   *Matt Caswell*
3422
3423 * Added SCA hardening for modular field inversion in EC_GROUP through
3424   a new dedicated field_inv() pointer in EC_METHOD.
3425   This also addresses a leakage affecting conversions from projective
3426   to affine coordinates.
3427
3428   *Billy Bob Brumley, Nicola Tuveri*
3429
3430 * Fix a use after free bug in d2i_X509_PUBKEY when overwriting a
3431   re-used X509_PUBKEY object if the second PUBKEY is malformed.
3432
3433   *Bernd Edlinger*
3434
3435 * Move strictness check from EVP_PKEY_asn1_new() to EVP_PKEY_asn1_add0().
3436
3437   *Richard Levitte*
3438
3439 * Remove the 'dist' target and add a tarball building script.  The
3440   'dist' target has fallen out of use, and it shouldn't be
3441   necessary to configure just to create a source distribution.
3442
3443   *Richard Levitte*
3444
3445### Changes between 1.1.0i and 1.1.0j [20 Nov 2018]
3446
3447 * Timing vulnerability in DSA signature generation
3448
3449   The OpenSSL DSA signature algorithm has been shown to be vulnerable to a
3450   timing side channel attack. An attacker could use variations in the signing
3451   algorithm to recover the private key.
3452
3453   This issue was reported to OpenSSL on 16th October 2018 by Samuel Weiser.
3454   ([CVE-2018-0734])
3455
3456   *Paul Dale*
3457
3458 * Timing vulnerability in ECDSA signature generation
3459
3460   The OpenSSL ECDSA signature algorithm has been shown to be vulnerable to a
3461   timing side channel attack. An attacker could use variations in the signing
3462   algorithm to recover the private key.
3463
3464   This issue was reported to OpenSSL on 25th October 2018 by Samuel Weiser.
3465   ([CVE-2018-0735])
3466
3467   *Paul Dale*
3468
3469 * Add coordinate blinding for EC_POINT and implement projective
3470   coordinate blinding for generic prime curves as a countermeasure to
3471   chosen point SCA attacks.
3472
3473   *Sohaib ul Hassan, Nicola Tuveri, Billy Bob Brumley*
3474
3475### Changes between 1.1.0h and 1.1.0i [14 Aug 2018]
3476
3477 * Client DoS due to large DH parameter
3478
3479   During key agreement in a TLS handshake using a DH(E) based ciphersuite a
3480   malicious server can send a very large prime value to the client. This will
3481   cause the client to spend an unreasonably long period of time generating a
3482   key for this prime resulting in a hang until the client has finished. This
3483   could be exploited in a Denial Of Service attack.
3484
3485   This issue was reported to OpenSSL on 5th June 2018 by Guido Vranken
3486   ([CVE-2018-0732])
3487
3488   *Guido Vranken*
3489
3490 * Cache timing vulnerability in RSA Key Generation
3491
3492   The OpenSSL RSA Key generation algorithm has been shown to be vulnerable to
3493   a cache timing side channel attack. An attacker with sufficient access to
3494   mount cache timing attacks during the RSA key generation process could
3495   recover the private key.
3496
3497   This issue was reported to OpenSSL on 4th April 2018 by Alejandro Cabrera
3498   Aldaya, Billy Brumley, Cesar Pereida Garcia and Luis Manuel Alvarez Tapia.
3499   ([CVE-2018-0737])
3500
3501   *Billy Brumley*
3502
3503 * Make EVP_PKEY_asn1_new() a bit stricter about its input.  A NULL pem_str
3504   parameter is no longer accepted, as it leads to a corrupt table.  NULL
3505   pem_str is reserved for alias entries only.
3506
3507   *Richard Levitte*
3508
3509 * Revert blinding in ECDSA sign and instead make problematic addition
3510   length-invariant. Switch even to fixed-length Montgomery multiplication.
3511
3512   *Andy Polyakov*
3513
3514 * Change generating and checking of primes so that the error rate of not
3515   being prime depends on the intended use based on the size of the input.
3516   For larger primes this will result in more rounds of Miller-Rabin.
3517   The maximal error rate for primes with more than 1080 bits is lowered
3518   to 2^-128.
3519
3520   *Kurt Roeckx, Annie Yousar*
3521
3522 * Increase the number of Miller-Rabin rounds for DSA key generating to 64.
3523
3524   *Kurt Roeckx*
3525
3526 * Add blinding to ECDSA and DSA signatures to protect against side channel
3527   attacks discovered by Keegan Ryan (NCC Group).
3528
3529   *Matt Caswell*
3530
3531 * When unlocking a pass phrase protected PEM file or PKCS#8 container, we
3532   now allow empty (zero character) pass phrases.
3533
3534   *Richard Levitte*
3535
3536 * Certificate time validation (X509_cmp_time) enforces stricter
3537   compliance with RFC 5280. Fractional seconds and timezone offsets
3538   are no longer allowed.
3539
3540   *Emilia Käsper*
3541
3542 * Fixed a text canonicalisation bug in CMS
3543
3544   Where a CMS detached signature is used with text content the text goes
3545   through a canonicalisation process first prior to signing or verifying a
3546   signature. This process strips trailing space at the end of lines, converts
3547   line terminators to CRLF and removes additional trailing line terminators
3548   at the end of a file. A bug in the canonicalisation process meant that
3549   some characters, such as form-feed, were incorrectly treated as whitespace
3550   and removed. This is contrary to the specification (RFC5485). This fix
3551   could mean that detached text data signed with an earlier version of
3552   OpenSSL 1.1.0 may fail to verify using the fixed version, or text data
3553   signed with a fixed OpenSSL may fail to verify with an earlier version of
3554   OpenSSL 1.1.0. A workaround is to only verify the canonicalised text data
3555   and use the "-binary" flag (for the "cms" command line application) or set
3556   the SMIME_BINARY/PKCS7_BINARY/CMS_BINARY flags (if using CMS_verify()).
3557
3558   *Matt Caswell*
3559
3560### Changes between 1.1.0g and 1.1.0h [27 Mar 2018]
3561
3562 * Constructed ASN.1 types with a recursive definition could exceed the stack
3563
3564   Constructed ASN.1 types with a recursive definition (such as can be found
3565   in PKCS7) could eventually exceed the stack given malicious input with
3566   excessive recursion. This could result in a Denial Of Service attack. There
3567   are no such structures used within SSL/TLS that come from untrusted sources
3568   so this is considered safe.
3569
3570   This issue was reported to OpenSSL on 4th January 2018 by the OSS-fuzz
3571   project.
3572   ([CVE-2018-0739])
3573
3574   *Matt Caswell*
3575
3576 * Incorrect CRYPTO_memcmp on HP-UX PA-RISC
3577
3578   Because of an implementation bug the PA-RISC CRYPTO_memcmp function is
3579   effectively reduced to only comparing the least significant bit of each
3580   byte. This allows an attacker to forge messages that would be considered as
3581   authenticated in an amount of tries lower than that guaranteed by the
3582   security claims of the scheme. The module can only be compiled by the
3583   HP-UX assembler, so that only HP-UX PA-RISC targets are affected.
3584
3585   This issue was reported to OpenSSL on 2nd March 2018 by Peter Waltenberg
3586   (IBM).
3587   ([CVE-2018-0733])
3588
3589   *Andy Polyakov*
3590
3591 * Add a build target 'build_all_generated', to build all generated files
3592   and only that.  This can be used to prepare everything that requires
3593   things like perl for a system that lacks perl and then move everything
3594   to that system and do the rest of the build there.
3595
3596   *Richard Levitte*
3597
3598 * Backport SSL_OP_NO_RENGOTIATION
3599
3600   OpenSSL 1.0.2 and below had the ability to disable renegotiation using the
3601   (undocumented) SSL3_FLAGS_NO_RENEGOTIATE_CIPHERS flag. Due to the opacity
3602   changes this is no longer possible in 1.1.0. Therefore the new
3603   SSL_OP_NO_RENEGOTIATION option from 1.1.1-dev has been backported to
3604   1.1.0 to provide equivalent functionality.
3605
3606   Note that if an application built against 1.1.0h headers (or above) is run
3607   using an older version of 1.1.0 (prior to 1.1.0h) then the option will be
3608   accepted but nothing will happen, i.e. renegotiation will not be prevented.
3609
3610   *Matt Caswell*
3611
3612 * Removed the OS390-Unix config target.  It relied on a script that doesn't
3613   exist.
3614
3615   *Rich Salz*
3616
3617 * rsaz_1024_mul_avx2 overflow bug on x86_64
3618
3619   There is an overflow bug in the AVX2 Montgomery multiplication procedure
3620   used in exponentiation with 1024-bit moduli. No EC algorithms are affected.
3621   Analysis suggests that attacks against RSA and DSA as a result of this
3622   defect would be very difficult to perform and are not believed likely.
3623   Attacks against DH1024 are considered just feasible, because most of the
3624   work necessary to deduce information about a private key may be performed
3625   offline. The amount of resources required for such an attack would be
3626   significant. However, for an attack on TLS to be meaningful, the server
3627   would have to share the DH1024 private key among multiple clients, which is
3628   no longer an option since CVE-2016-0701.
3629
3630   This only affects processors that support the AVX2 but not ADX extensions
3631   like Intel Haswell (4th generation).
3632
3633   This issue was reported to OpenSSL by David Benjamin (Google). The issue
3634   was originally found via the OSS-Fuzz project.
3635   ([CVE-2017-3738])
3636
3637   *Andy Polyakov*
3638
3639### Changes between 1.1.0f and 1.1.0g [2 Nov 2017]
3640
3641 * bn_sqrx8x_internal carry bug on x86_64
3642
3643   There is a carry propagating bug in the x86_64 Montgomery squaring
3644   procedure. No EC algorithms are affected. Analysis suggests that attacks
3645   against RSA and DSA as a result of this defect would be very difficult to
3646   perform and are not believed likely. Attacks against DH are considered just
3647   feasible (although very difficult) because most of the work necessary to
3648   deduce information about a private key may be performed offline. The amount
3649   of resources required for such an attack would be very significant and
3650   likely only accessible to a limited number of attackers. An attacker would
3651   additionally need online access to an unpatched system using the target
3652   private key in a scenario with persistent DH parameters and a private
3653   key that is shared between multiple clients.
3654
3655   This only affects processors that support the BMI1, BMI2 and ADX extensions
3656   like Intel Broadwell (5th generation) and later or AMD Ryzen.
3657
3658   This issue was reported to OpenSSL by the OSS-Fuzz project.
3659   ([CVE-2017-3736])
3660
3661   *Andy Polyakov*
3662
3663 * Malformed X.509 IPAddressFamily could cause OOB read
3664
3665   If an X.509 certificate has a malformed IPAddressFamily extension,
3666   OpenSSL could do a one-byte buffer overread. The most likely result
3667   would be an erroneous display of the certificate in text format.
3668
3669   This issue was reported to OpenSSL by the OSS-Fuzz project.
3670   ([CVE-2017-3735])
3671
3672   *Rich Salz*
3673
3674### Changes between 1.1.0e and 1.1.0f [25 May 2017]
3675
3676 * Have 'config' recognise 64-bit mingw and choose 'mingw64' as the target
3677   platform rather than 'mingw'.
3678
3679   *Richard Levitte*
3680
3681 * Remove the VMS-specific reimplementation of gmtime from crypto/o_times.c.
3682   VMS C's RTL has a fully up to date gmtime() and gmtime_r() since V7.1,
3683   which is the minimum version we support.
3684
3685   *Richard Levitte*
3686
3687### Changes between 1.1.0d and 1.1.0e [16 Feb 2017]
3688
3689 * Encrypt-Then-Mac renegotiation crash
3690
3691   During a renegotiation handshake if the Encrypt-Then-Mac extension is
3692   negotiated where it was not in the original handshake (or vice-versa) then
3693   this can cause OpenSSL to crash (dependant on ciphersuite). Both clients
3694   and servers are affected.
3695
3696   This issue was reported to OpenSSL by Joe Orton (Red Hat).
3697   ([CVE-2017-3733])
3698
3699   *Matt Caswell*
3700
3701### Changes between 1.1.0c and 1.1.0d [26 Jan 2017]
3702
3703 * Truncated packet could crash via OOB read
3704
3705   If one side of an SSL/TLS path is running on a 32-bit host and a specific
3706   cipher is being used, then a truncated packet can cause that host to
3707   perform an out-of-bounds read, usually resulting in a crash.
3708
3709   This issue was reported to OpenSSL by Robert Święcki of Google.
3710   ([CVE-2017-3731])
3711
3712   *Andy Polyakov*
3713
3714 * Bad (EC)DHE parameters cause a client crash
3715
3716   If a malicious server supplies bad parameters for a DHE or ECDHE key
3717   exchange then this can result in the client attempting to dereference a
3718   NULL pointer leading to a client crash. This could be exploited in a Denial
3719   of Service attack.
3720
3721   This issue was reported to OpenSSL by Guido Vranken.
3722   ([CVE-2017-3730])
3723
3724   *Matt Caswell*
3725
3726 * BN_mod_exp may produce incorrect results on x86_64
3727
3728   There is a carry propagating bug in the x86_64 Montgomery squaring
3729   procedure. No EC algorithms are affected. Analysis suggests that attacks
3730   against RSA and DSA as a result of this defect would be very difficult to
3731   perform and are not believed likely. Attacks against DH are considered just
3732   feasible (although very difficult) because most of the work necessary to
3733   deduce information about a private key may be performed offline. The amount
3734   of resources required for such an attack would be very significant and
3735   likely only accessible to a limited number of attackers. An attacker would
3736   additionally need online access to an unpatched system using the target
3737   private key in a scenario with persistent DH parameters and a private
3738   key that is shared between multiple clients. For example this can occur by
3739   default in OpenSSL DHE based SSL/TLS ciphersuites. Note: This issue is very
3740   similar to CVE-2015-3193 but must be treated as a separate problem.
3741
3742   This issue was reported to OpenSSL by the OSS-Fuzz project.
3743   ([CVE-2017-3732])
3744
3745   *Andy Polyakov*
3746
3747### Changes between 1.1.0b and 1.1.0c [10 Nov 2016]
3748
3749 * ChaCha20/Poly1305 heap-buffer-overflow
3750
3751   TLS connections using `*-CHACHA20-POLY1305` ciphersuites are susceptible to
3752   a DoS attack by corrupting larger payloads. This can result in an OpenSSL
3753   crash. This issue is not considered to be exploitable beyond a DoS.
3754
3755   This issue was reported to OpenSSL by Robert Święcki (Google Security Team)
3756   ([CVE-2016-7054])
3757
3758   *Richard Levitte*
3759
3760 * CMS Null dereference
3761
3762   Applications parsing invalid CMS structures can crash with a NULL pointer
3763   dereference. This is caused by a bug in the handling of the ASN.1 CHOICE
3764   type in OpenSSL 1.1.0 which can result in a NULL value being passed to the
3765   structure callback if an attempt is made to free certain invalid encodings.
3766   Only CHOICE structures using a callback which do not handle NULL value are
3767   affected.
3768
3769   This issue was reported to OpenSSL by Tyler Nighswander of ForAllSecure.
3770   ([CVE-2016-7053])
3771
3772   *Stephen Henson*
3773
3774 * Montgomery multiplication may produce incorrect results
3775
3776   There is a carry propagating bug in the Broadwell-specific Montgomery
3777   multiplication procedure that handles input lengths divisible by, but
3778   longer than 256 bits. Analysis suggests that attacks against RSA, DSA
3779   and DH private keys are impossible. This is because the subroutine in
3780   question is not used in operations with the private key itself and an input
3781   of the attacker's direct choice. Otherwise the bug can manifest itself as
3782   transient authentication and key negotiation failures or reproducible
3783   erroneous outcome of public-key operations with specially crafted input.
3784   Among EC algorithms only Brainpool P-512 curves are affected and one
3785   presumably can attack ECDH key negotiation. Impact was not analyzed in
3786   detail, because pre-requisites for attack are considered unlikely. Namely
3787   multiple clients have to choose the curve in question and the server has to
3788   share the private key among them, neither of which is default behaviour.
3789   Even then only clients that chose the curve will be affected.
3790
3791   This issue was publicly reported as transient failures and was not
3792   initially recognized as a security issue. Thanks to Richard Morgan for
3793   providing reproducible case.
3794   ([CVE-2016-7055])
3795
3796   *Andy Polyakov*
3797
3798 * Removed automatic addition of RPATH in shared libraries and executables,
3799   as this was a remainder from OpenSSL 1.0.x and isn't needed any more.
3800
3801   *Richard Levitte*
3802
3803### Changes between 1.1.0a and 1.1.0b [26 Sep 2016]
3804
3805 * Fix Use After Free for large message sizes
3806
3807   The patch applied to address CVE-2016-6307 resulted in an issue where if a
3808   message larger than approx 16k is received then the underlying buffer to
3809   store the incoming message is reallocated and moved. Unfortunately a
3810   dangling pointer to the old location is left which results in an attempt to
3811   write to the previously freed location. This is likely to result in a
3812   crash, however it could potentially lead to execution of arbitrary code.
3813
3814   This issue only affects OpenSSL 1.1.0a.
3815
3816   This issue was reported to OpenSSL by Robert Święcki.
3817   ([CVE-2016-6309])
3818
3819   *Matt Caswell*
3820
3821### Changes between 1.1.0 and 1.1.0a [22 Sep 2016]
3822
3823 * OCSP Status Request extension unbounded memory growth
3824
3825   A malicious client can send an excessively large OCSP Status Request
3826   extension. If that client continually requests renegotiation, sending a
3827   large OCSP Status Request extension each time, then there will be unbounded
3828   memory growth on the server. This will eventually lead to a Denial Of
3829   Service attack through memory exhaustion. Servers with a default
3830   configuration are vulnerable even if they do not support OCSP. Builds using
3831   the "no-ocsp" build time option are not affected.
3832
3833   This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.)
3834   ([CVE-2016-6304])
3835
3836   *Matt Caswell*
3837
3838 * SSL_peek() hang on empty record
3839
3840   OpenSSL 1.1.0 SSL/TLS will hang during a call to SSL_peek() if the peer
3841   sends an empty record. This could be exploited by a malicious peer in a
3842   Denial Of Service attack.
3843
3844   This issue was reported to OpenSSL by Alex Gaynor.
3845   ([CVE-2016-6305])
3846
3847   *Matt Caswell*
3848
3849 * Excessive allocation of memory in tls_get_message_header() and
3850   dtls1_preprocess_fragment()
3851
3852   A (D)TLS message includes 3 bytes for its length in the header for the
3853   message. This would allow for messages up to 16Mb in length. Messages of
3854   this length are excessive and OpenSSL includes a check to ensure that a
3855   peer is sending reasonably sized messages in order to avoid too much memory
3856   being consumed to service a connection. A flaw in the logic of version
3857   1.1.0 means that memory for the message is allocated too early, prior to
3858   the excessive message length check. Due to way memory is allocated in
3859   OpenSSL this could mean an attacker could force up to 21Mb to be allocated
3860   to service a connection. This could lead to a Denial of Service through
3861   memory exhaustion. However, the excessive message length check still takes
3862   place, and this would cause the connection to immediately fail. Assuming
3863   that the application calls SSL_free() on the failed connection in a timely
3864   manner then the 21Mb of allocated memory will then be immediately freed
3865   again. Therefore the excessive memory allocation will be transitory in
3866   nature. This then means that there is only a security impact if:
3867
3868   1) The application does not call SSL_free() in a timely manner in the event
3869   that the connection fails
3870   or
3871   2) The application is working in a constrained environment where there is
3872   very little free memory
3873   or
3874   3) The attacker initiates multiple connection attempts such that there are
3875   multiple connections in a state where memory has been allocated for the
3876   connection; SSL_free() has not yet been called; and there is insufficient
3877   memory to service the multiple requests.
3878
3879   Except in the instance of (1) above any Denial Of Service is likely to be
3880   transitory because as soon as the connection fails the memory is
3881   subsequently freed again in the SSL_free() call. However there is an
3882   increased risk during this period of application crashes due to the lack of
3883   memory - which would then mean a more serious Denial of Service.
3884
3885   This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.)
3886   (CVE-2016-6307 and CVE-2016-6308)
3887
3888   *Matt Caswell*
3889
3890 * solaris-x86-cc, i.e. 32-bit configuration with vendor compiler,
3891   had to be removed. Primary reason is that vendor assembler can't
3892   assemble our modules with -KPIC flag. As result it, assembly
3893   support, was not even available as option. But its lack means
3894   lack of side-channel resistant code, which is incompatible with
3895   security by todays standards. Fortunately gcc is readily available
3896   prepackaged option, which we firmly point at...
3897
3898   *Andy Polyakov*
3899
3900### Changes between 1.0.2h and 1.1.0  [25 Aug 2016]
3901
3902 * Windows command-line tool supports UTF-8 opt-in option for arguments
3903   and console input. Setting OPENSSL_WIN32_UTF8 environment variable
3904   (to any value) allows Windows user to access PKCS#12 file generated
3905   with Windows CryptoAPI and protected with non-ASCII password, as well
3906   as files generated under UTF-8 locale on Linux also protected with
3907   non-ASCII password.
3908
3909   *Andy Polyakov*
3910
3911 * To mitigate the SWEET32 attack ([CVE-2016-2183]), 3DES cipher suites
3912   have been disabled by default and removed from DEFAULT, just like RC4.
3913   See the RC4 item below to re-enable both.
3914
3915   *Rich Salz*
3916
3917 * The method for finding the storage location for the Windows RAND seed file
3918   has changed. First we check %RANDFILE%. If that is not set then we check
3919   the directories %HOME%, %USERPROFILE% and %SYSTEMROOT% in that order. If
3920   all else fails we fall back to C:\.
3921
3922   *Matt Caswell*
3923
3924 * The EVP_EncryptUpdate() function has had its return type changed from void
3925   to int. A return of 0 indicates and error while a return of 1 indicates
3926   success.
3927
3928   *Matt Caswell*
3929
3930 * The flags RSA_FLAG_NO_CONSTTIME, DSA_FLAG_NO_EXP_CONSTTIME and
3931   DH_FLAG_NO_EXP_CONSTTIME which previously provided the ability to switch
3932   off the constant time implementation for RSA, DSA and DH have been made
3933   no-ops and deprecated.
3934
3935   *Matt Caswell*
3936
3937 * Windows RAND implementation was simplified to only get entropy by
3938   calling CryptGenRandom(). Various other RAND-related tickets
3939   were also closed.
3940
3941   *Joseph Wylie Yandle, Rich Salz*
3942
3943 * The stack and lhash API's were renamed to start with `OPENSSL_SK_`
3944   and `OPENSSL_LH_`, respectively.  The old names are available
3945   with API compatibility.  They new names are now completely documented.
3946
3947   *Rich Salz*
3948
3949 * Unify TYPE_up_ref(obj) methods signature.
3950   SSL_CTX_up_ref(), SSL_up_ref(), X509_up_ref(), EVP_PKEY_up_ref(),
3951   X509_CRL_up_ref(), X509_OBJECT_up_ref_count() methods are now returning an
3952   int (instead of void) like all others TYPE_up_ref() methods.
3953   So now these methods also check the return value of CRYPTO_atomic_add(),
3954   and the validity of object reference counter.
3955
3956   *fdasilvayy@gmail.com*
3957
3958 * With Windows Visual Studio builds, the .pdb files are installed
3959   alongside the installed libraries and executables.  For a static
3960   library installation, ossl_static.pdb is the associate compiler
3961   generated .pdb file to be used when linking programs.
3962
3963   *Richard Levitte*
3964
3965 * Remove openssl.spec.  Packaging files belong with the packagers.
3966
3967   *Richard Levitte*
3968
3969 * Automatic Darwin/OSX configuration has had a refresh, it will now
3970   recognise x86_64 architectures automatically.  You can still decide
3971   to build for a different bitness with the environment variable
3972   KERNEL_BITS (can be 32 or 64), for example:
3973
3974           KERNEL_BITS=32 ./config
3975
3976   *Richard Levitte*
3977
3978 * Change default algorithms in pkcs8 utility to use PKCS#5 v2.0,
3979   256 bit AES and HMAC with SHA256.
3980
3981   *Steve Henson*
3982
3983 * Remove support for MIPS o32 ABI on IRIX (and IRIX only).
3984
3985   *Andy Polyakov*
3986
3987 * Triple-DES ciphers have been moved from HIGH to MEDIUM.
3988
3989   *Rich Salz*
3990
3991 * To enable users to have their own config files and build file templates,
3992   Configure looks in the directory indicated by the environment variable
3993   OPENSSL_LOCAL_CONFIG_DIR as well as the in-source Configurations/
3994   directory.  On VMS, OPENSSL_LOCAL_CONFIG_DIR is expected to be a logical
3995   name and is used as is.
3996
3997   *Richard Levitte*
3998
3999 * The following datatypes were made opaque: X509_OBJECT, X509_STORE_CTX,
4000   X509_STORE, X509_LOOKUP, and X509_LOOKUP_METHOD.  The unused type
4001   X509_CERT_FILE_CTX was removed.
4002
4003   *Rich Salz*
4004
4005 * "shared" builds are now the default. To create only static libraries use
4006   the "no-shared" Configure option.
4007
4008   *Matt Caswell*
4009
4010 * Remove the no-aes, no-hmac, no-rsa, no-sha and no-md5 Configure options.
4011   All of these option have not worked for some while and are fundamental
4012   algorithms.
4013
4014   *Matt Caswell*
4015
4016 * Make various cleanup routines no-ops and mark them as deprecated. Most
4017   global cleanup functions are no longer required because they are handled
4018   via auto-deinit (see OPENSSL_init_crypto and OPENSSL_init_ssl man pages).
4019   Explicitly de-initing can cause problems (e.g. where a library that uses
4020   OpenSSL de-inits, but an application is still using it). The affected
4021   functions are CONF_modules_free(), ENGINE_cleanup(), OBJ_cleanup(),
4022   EVP_cleanup(), BIO_sock_cleanup(), CRYPTO_cleanup_all_ex_data(),
4023   RAND_cleanup(), SSL_COMP_free_compression_methods(), ERR_free_strings() and
4024   COMP_zlib_cleanup().
4025
4026   *Matt Caswell*
4027
4028 * --strict-warnings no longer enables runtime debugging options
4029   such as REF_DEBUG. Instead, debug options are automatically
4030   enabled with '--debug' builds.
4031
4032   *Andy Polyakov, Emilia Käsper*
4033
4034 * Made DH and DH_METHOD opaque. The structures for managing DH objects
4035   have been moved out of the public header files. New functions for managing
4036   these have been added.
4037
4038   *Matt Caswell*
4039
4040 * Made RSA and RSA_METHOD opaque. The structures for managing RSA
4041   objects have been moved out of the public header files. New
4042   functions for managing these have been added.
4043
4044   *Richard Levitte*
4045
4046 * Made DSA and DSA_METHOD opaque. The structures for managing DSA objects
4047   have been moved out of the public header files. New functions for managing
4048   these have been added.
4049
4050   *Matt Caswell*
4051
4052 * Made BIO and BIO_METHOD opaque. The structures for managing BIOs have been
4053   moved out of the public header files. New functions for managing these
4054   have been added.
4055
4056   *Matt Caswell*
4057
4058 * Removed no-rijndael as a config option. Rijndael is an old name for AES.
4059
4060   *Matt Caswell*
4061
4062 * Removed the mk1mf build scripts.
4063
4064   *Richard Levitte*
4065
4066 * Headers are now wrapped, if necessary, with OPENSSL_NO_xxx, so
4067   it is always safe to #include a header now.
4068
4069   *Rich Salz*
4070
4071 * Removed the aged BC-32 config and all its supporting scripts
4072
4073   *Richard Levitte*
4074
4075 * Removed support for Ultrix, Netware, and OS/2.
4076
4077   *Rich Salz*
4078
4079 * Add support for HKDF.
4080
4081   *Alessandro Ghedini*
4082
4083 * Add support for blake2b and blake2s
4084
4085   *Bill Cox*
4086
4087 * Added support for "pipelining". Ciphers that have the
4088   EVP_CIPH_FLAG_PIPELINE flag set have a capability to process multiple
4089   encryptions/decryptions simultaneously. There are currently no built-in
4090   ciphers with this property but the expectation is that engines will be able
4091   to offer it to significantly improve throughput. Support has been extended
4092   into libssl so that multiple records for a single connection can be
4093   processed in one go (for >=TLS 1.1).
4094
4095   *Matt Caswell*
4096
4097 * Added the AFALG engine. This is an async capable engine which is able to
4098   offload work to the Linux kernel. In this initial version it only supports
4099   AES128-CBC. The kernel must be version 4.1.0 or greater.
4100
4101   *Catriona Lucey*
4102
4103 * OpenSSL now uses a new threading API. It is no longer necessary to
4104   set locking callbacks to use OpenSSL in a multi-threaded environment. There
4105   are two supported threading models: pthreads and windows threads. It is
4106   also possible to configure OpenSSL at compile time for "no-threads". The
4107   old threading API should no longer be used. The functions have been
4108   replaced with "no-op" compatibility macros.
4109
4110   *Alessandro Ghedini, Matt Caswell*
4111
4112 * Modify behavior of ALPN to invoke callback after SNI/servername
4113   callback, such that updates to the SSL_CTX affect ALPN.
4114
4115   *Todd Short*
4116
4117 * Add SSL_CIPHER queries for authentication and key-exchange.
4118
4119   *Todd Short*
4120
4121 * Changes to the DEFAULT cipherlist:
4122   - Prefer (EC)DHE handshakes over plain RSA.
4123   - Prefer AEAD ciphers over legacy ciphers.
4124   - Prefer ECDSA over RSA when both certificates are available.
4125   - Prefer TLSv1.2 ciphers/PRF.
4126   - Remove DSS, SEED, IDEA, CAMELLIA, and AES-CCM from the
4127     default cipherlist.
4128
4129   *Emilia Käsper*
4130
4131 * Change the ECC default curve list to be this, in order: x25519,
4132   secp256r1, secp521r1, secp384r1.
4133
4134   *Rich Salz*
4135
4136 * RC4 based libssl ciphersuites are now classed as "weak" ciphers and are
4137   disabled by default. They can be re-enabled using the
4138   enable-weak-ssl-ciphers option to Configure.
4139
4140   *Matt Caswell*
4141
4142 * If the server has ALPN configured, but supports no protocols that the
4143   client advertises, send a fatal "no_application_protocol" alert.
4144   This behaviour is SHALL in RFC 7301, though it isn't universally
4145   implemented by other servers.
4146
4147   *Emilia Käsper*
4148
4149 * Add X25519 support.
4150   Add ASN.1 and EVP_PKEY methods for X25519. This includes support
4151   for public and private key encoding using the format documented in
4152   draft-ietf-curdle-pkix-02. The corresponding EVP_PKEY method supports
4153   key generation and key derivation.
4154
4155   TLS support complies with draft-ietf-tls-rfc4492bis-08 and uses
4156   X25519(29).
4157
4158   *Steve Henson*
4159
4160 * Deprecate SRP_VBASE_get_by_user.
4161   SRP_VBASE_get_by_user had inconsistent memory management behaviour.
4162   In order to fix an unavoidable memory leak ([CVE-2016-0798]),
4163   SRP_VBASE_get_by_user was changed to ignore the "fake user" SRP
4164   seed, even if the seed is configured.
4165
4166   Users should use SRP_VBASE_get1_by_user instead. Note that in
4167   SRP_VBASE_get1_by_user, caller must free the returned value. Note
4168   also that even though configuring the SRP seed attempts to hide
4169   invalid usernames by continuing the handshake with fake
4170   credentials, this behaviour is not constant time and no strong
4171   guarantees are made that the handshake is indistinguishable from
4172   that of a valid user.
4173
4174   *Emilia Käsper*
4175
4176 * Configuration change; it's now possible to build dynamic engines
4177   without having to build shared libraries and vice versa.  This
4178   only applies to the engines in `engines/`, those in `crypto/engine/`
4179   will always be built into libcrypto (i.e. "static").
4180
4181   Building dynamic engines is enabled by default; to disable, use
4182   the configuration option "disable-dynamic-engine".
4183
4184   The only requirements for building dynamic engines are the
4185   presence of the DSO module and building with position independent
4186   code, so they will also automatically be disabled if configuring
4187   with "disable-dso" or "disable-pic".
4188
4189   The macros OPENSSL_NO_STATIC_ENGINE and OPENSSL_NO_DYNAMIC_ENGINE
4190   are also taken away from openssl/opensslconf.h, as they are
4191   irrelevant.
4192
4193   *Richard Levitte*
4194
4195 * Configuration change; if there is a known flag to compile
4196   position independent code, it will always be applied on the
4197   libcrypto and libssl object files, and never on the application
4198   object files.  This means other libraries that use routines from
4199   libcrypto / libssl can be made into shared libraries regardless
4200   of how OpenSSL was configured.
4201
4202   If this isn't desirable, the configuration options "disable-pic"
4203   or "no-pic" can be used to disable the use of PIC.  This will
4204   also disable building shared libraries and dynamic engines.
4205
4206   *Richard Levitte*
4207
4208 * Removed JPAKE code.  It was experimental and has no wide use.
4209
4210   *Rich Salz*
4211
4212 * The INSTALL_PREFIX Makefile variable has been renamed to
4213   DESTDIR.  That makes for less confusion on what this variable
4214   is for.  Also, the configuration option --install_prefix is
4215   removed.
4216
4217   *Richard Levitte*
4218
4219 * Heartbeat for TLS has been removed and is disabled by default
4220   for DTLS; configure with enable-heartbeats.  Code that uses the
4221   old #define's might need to be updated.
4222
4223   *Emilia Käsper, Rich Salz*
4224
4225 * Rename REF_CHECK to REF_DEBUG.
4226
4227   *Rich Salz*
4228
4229 * New "unified" build system
4230
4231   The "unified" build system is aimed to be a common system for all
4232   platforms we support.  With it comes new support for VMS.
4233
4234   This system builds supports building in a different directory tree
4235   than the source tree.  It produces one Makefile (for unix family
4236   or lookalikes), or one descrip.mms (for VMS).
4237
4238   The source of information to make the Makefile / descrip.mms is
4239   small files called 'build.info', holding the necessary
4240   information for each directory with source to compile, and a
4241   template in Configurations, like unix-Makefile.tmpl or
4242   descrip.mms.tmpl.
4243
4244   With this change, the library names were also renamed on Windows
4245   and on VMS.  They now have names that are closer to the standard
4246   on Unix, and include the major version number, and in certain
4247   cases, the architecture they are built for.  See "Notes on shared
4248   libraries" in INSTALL.
4249
4250   We rely heavily on the perl module Text::Template.
4251
4252   *Richard Levitte*
4253
4254 * Added support for auto-initialisation and de-initialisation of the library.
4255   OpenSSL no longer requires explicit init or deinit routines to be called,
4256   except in certain circumstances. See the OPENSSL_init_crypto() and
4257   OPENSSL_init_ssl() man pages for further information.
4258
4259   *Matt Caswell*
4260
4261 * The arguments to the DTLSv1_listen function have changed. Specifically the
4262   "peer" argument is now expected to be a BIO_ADDR object.
4263
4264 * Rewrite of BIO networking library. The BIO library lacked consistent
4265   support of IPv6, and adding it required some more extensive
4266   modifications.  This introduces the BIO_ADDR and BIO_ADDRINFO types,
4267   which hold all types of addresses and chains of address information.
4268   It also introduces a new API, with functions like BIO_socket,
4269   BIO_connect, BIO_listen, BIO_lookup and a rewrite of BIO_accept.
4270   The source/sink BIOs BIO_s_connect, BIO_s_accept and BIO_s_datagram
4271   have been adapted accordingly.
4272
4273   *Richard Levitte*
4274
4275 * RSA_padding_check_PKCS1_type_1 now accepts inputs with and without
4276   the leading 0-byte.
4277
4278   *Emilia Käsper*
4279
4280 * CRIME protection: disable compression by default, even if OpenSSL is
4281   compiled with zlib enabled. Applications can still enable compression
4282   by calling SSL_CTX_clear_options(ctx, SSL_OP_NO_COMPRESSION), or by
4283   using the SSL_CONF library to configure compression.
4284
4285   *Emilia Käsper*
4286
4287 * The signature of the session callback configured with
4288   SSL_CTX_sess_set_get_cb was changed. The read-only input buffer
4289   was explicitly marked as `const unsigned char*` instead of
4290   `unsigned char*`.
4291
4292   *Emilia Käsper*
4293
4294 * Always DPURIFY. Remove the use of uninitialized memory in the
4295   RNG, and other conditional uses of DPURIFY. This makes -DPURIFY a no-op.
4296
4297   *Emilia Käsper*
4298
4299 * Removed many obsolete configuration items, including
4300      DES_PTR, DES_RISC1, DES_RISC2, DES_INT
4301      MD2_CHAR, MD2_INT, MD2_LONG
4302      BF_PTR, BF_PTR2
4303      IDEA_SHORT, IDEA_LONG
4304      RC2_SHORT, RC2_LONG, RC4_LONG, RC4_CHUNK, RC4_INDEX
4305
4306   *Rich Salz, with advice from Andy Polyakov*
4307
4308 * Many BN internals have been moved to an internal header file.
4309
4310   *Rich Salz with help from Andy Polyakov*
4311
4312 * Configuration and writing out the results from it has changed.
4313   Files such as Makefile include/openssl/opensslconf.h and are now
4314   produced through general templates, such as Makefile.in and
4315   crypto/opensslconf.h.in and some help from the perl module
4316   Text::Template.
4317
4318   Also, the center of configuration information is no longer
4319   Makefile.  Instead, Configure produces a perl module in
4320   configdata.pm which holds most of the config data (in the hash
4321   table %config), the target data that comes from the target
4322   configuration in one of the `Configurations/*.conf` files (in
4323   %target).
4324
4325   *Richard Levitte*
4326
4327 * To clarify their intended purposes, the Configure options
4328   --prefix and --openssldir change their semantics, and become more
4329   straightforward and less interdependent.
4330
4331   --prefix shall be used exclusively to give the location INSTALLTOP
4332   where programs, scripts, libraries, include files and manuals are
4333   going to be installed.  The default is now /usr/local.
4334
4335   --openssldir shall be used exclusively to give the default
4336   location OPENSSLDIR where certificates, private keys, CRLs are
4337   managed.  This is also where the default openssl.cnf gets
4338   installed.
4339   If the directory given with this option is a relative path, the
4340   values of both the --prefix value and the --openssldir value will
4341   be combined to become OPENSSLDIR.
4342   The default for --openssldir is INSTALLTOP/ssl.
4343
4344   Anyone who uses --openssldir to specify where OpenSSL is to be
4345   installed MUST change to use --prefix instead.
4346
4347   *Richard Levitte*
4348
4349 * The GOST engine was out of date and therefore it has been removed. An up
4350   to date GOST engine is now being maintained in an external repository.
4351   See: <https://wiki.openssl.org/index.php/Binaries>. Libssl still retains
4352   support for GOST ciphersuites (these are only activated if a GOST engine
4353   is present).
4354
4355   *Matt Caswell*
4356
4357 * EGD is no longer supported by default; use enable-egd when
4358   configuring.
4359
4360   *Ben Kaduk and Rich Salz*
4361
4362 * The distribution now has Makefile.in files, which are used to
4363   create Makefile's when Configure is run.  *Configure must be run
4364   before trying to build now.*
4365
4366   *Rich Salz*
4367
4368 * The return value for SSL_CIPHER_description() for error conditions
4369   has changed.
4370
4371   *Rich Salz*
4372
4373 * Support for RFC6698/RFC7671 DANE TLSA peer authentication.
4374
4375   Obtaining and performing DNSSEC validation of TLSA records is
4376   the application's responsibility.  The application provides
4377   the TLSA records of its choice to OpenSSL, and these are then
4378   used to authenticate the peer.
4379
4380   The TLSA records need not even come from DNS.  They can, for
4381   example, be used to implement local end-entity certificate or
4382   trust-anchor "pinning", where the "pin" data takes the form
4383   of TLSA records, which can augment or replace verification
4384   based on the usual WebPKI public certification authorities.
4385
4386   *Viktor Dukhovni*
4387
4388 * Revert default OPENSSL_NO_DEPRECATED setting.  Instead OpenSSL
4389   continues to support deprecated interfaces in default builds.
4390   However, applications are strongly advised to compile their
4391   source files with -DOPENSSL_API_COMPAT=0x10100000L, which hides
4392   the declarations of all interfaces deprecated in 0.9.8, 1.0.0
4393   or the 1.1.0 releases.
4394
4395   In environments in which all applications have been ported to
4396   not use any deprecated interfaces OpenSSL's Configure script
4397   should be used with the --api=1.1.0 option to entirely remove
4398   support for the deprecated features from the library and
4399   unconditionally disable them in the installed headers.
4400   Essentially the same effect can be achieved with the "no-deprecated"
4401   argument to Configure, except that this will always restrict
4402   the build to just the latest API, rather than a fixed API
4403   version.
4404
4405   As applications are ported to future revisions of the API,
4406   they should update their compile-time OPENSSL_API_COMPAT define
4407   accordingly, but in most cases should be able to continue to
4408   compile with later releases.
4409
4410   The OPENSSL_API_COMPAT versions for 1.0.0, and 0.9.8 are
4411   0x10000000L and 0x00908000L, respectively.  However those
4412   versions did not support the OPENSSL_API_COMPAT feature, and
4413   so applications are not typically tested for explicit support
4414   of just the undeprecated features of either release.
4415
4416   *Viktor Dukhovni*
4417
4418 * Add support for setting the minimum and maximum supported protocol.
4419   It can bet set via the SSL_set_min_proto_version() and
4420   SSL_set_max_proto_version(), or via the SSL_CONF's MinProtocol and
4421   MaxProtocol.  It's recommended to use the new APIs to disable
4422   protocols instead of disabling individual protocols using
4423   SSL_set_options() or SSL_CONF's Protocol.  This change also
4424   removes support for disabling TLS 1.2 in the OpenSSL TLS
4425   client at compile time by defining OPENSSL_NO_TLS1_2_CLIENT.
4426
4427   *Kurt Roeckx*
4428
4429 * Support for ChaCha20 and Poly1305 added to libcrypto and libssl.
4430
4431   *Andy Polyakov*
4432
4433 * New EC_KEY_METHOD, this replaces the older ECDSA_METHOD and ECDH_METHOD
4434   and integrates ECDSA and ECDH functionality into EC. Implementations can
4435   now redirect key generation and no longer need to convert to or from
4436   ECDSA_SIG format.
4437
4438   Note: the ecdsa.h and ecdh.h headers are now no longer needed and just
4439   include the ec.h header file instead.
4440
4441   *Steve Henson*
4442
4443 * Remove support for all 40 and 56 bit ciphers.  This includes all the export
4444   ciphers who are no longer supported and drops support the ephemeral RSA key
4445   exchange. The LOW ciphers currently doesn't have any ciphers in it.
4446
4447   *Kurt Roeckx*
4448
4449 * Made EVP_MD_CTX, EVP_MD, EVP_CIPHER_CTX, EVP_CIPHER and HMAC_CTX
4450   opaque.  For HMAC_CTX, the following constructors and destructors
4451   were added:
4452
4453       HMAC_CTX *HMAC_CTX_new(void);
4454       void HMAC_CTX_free(HMAC_CTX *ctx);
4455
4456   For EVP_MD and EVP_CIPHER, complete APIs to create, fill and
4457   destroy such methods has been added.  See EVP_MD_meth_new(3) and
4458   EVP_CIPHER_meth_new(3) for documentation.
4459
4460   Additional changes:
4461   1) `EVP_MD_CTX_cleanup()`, `EVP_CIPHER_CTX_cleanup()` and
4462      `HMAC_CTX_cleanup()` were removed. `HMAC_CTX_reset()` and
4463      `EVP_MD_CTX_reset()` should be called instead to reinitialise
4464      an already created structure.
4465   2) For consistency with the majority of our object creators and
4466      destructors, `EVP_MD_CTX_(create|destroy)` were renamed to
4467      `EVP_MD_CTX_(new|free)`.  The old names are retained as macros
4468      for deprecated builds.
4469
4470   *Richard Levitte*
4471
4472 * Added ASYNC support. Libcrypto now includes the async sub-library to enable
4473   cryptographic operations to be performed asynchronously as long as an
4474   asynchronous capable engine is used. See the ASYNC_start_job() man page for
4475   further details. Libssl has also had this capability integrated with the
4476   introduction of the new mode SSL_MODE_ASYNC and associated error
4477   SSL_ERROR_WANT_ASYNC. See the SSL_CTX_set_mode() and SSL_get_error() man
4478   pages. This work was developed in partnership with Intel Corp.
4479
4480   *Matt Caswell*
4481
4482 * SSL_{CTX_}set_ecdh_auto() has been removed and ECDH is support is
4483   always enabled now.  If you want to disable the support you should
4484   exclude it using the list of supported ciphers. This also means that the
4485   "-no_ecdhe" option has been removed from s_server.
4486
4487   *Kurt Roeckx*
4488
4489 * SSL_{CTX}_set_tmp_ecdh() which can set 1 EC curve now internally calls
4490   SSL_{CTX_}set1_curves() which can set a list.
4491
4492   *Kurt Roeckx*
4493
4494 * Remove support for SSL_{CTX_}set_tmp_ecdh_callback().  You should set the
4495   curve you want to support using SSL_{CTX_}set1_curves().
4496
4497   *Kurt Roeckx*
4498
4499 * State machine rewrite. The state machine code has been significantly
4500   refactored in order to remove much duplication of code and solve issues
4501   with the old code (see [ssl/statem/README.md](ssl/statem/README.md) for
4502   further details). This change does have some associated API changes.
4503   Notably the SSL_state() function has been removed and replaced by
4504   SSL_get_state which now returns an "OSSL_HANDSHAKE_STATE" instead of an int.
4505   SSL_set_state() has been removed altogether. The previous handshake states
4506   defined in ssl.h and ssl3.h have also been removed.
4507
4508   *Matt Caswell*
4509
4510 * All instances of the string "ssleay" in the public API were replaced
4511   with OpenSSL (case-matching; e.g., OPENSSL_VERSION for #define's)
4512   Some error codes related to internal RSA_eay API's were renamed.
4513
4514   *Rich Salz*
4515
4516 * The demo files in crypto/threads were moved to demo/threads.
4517
4518   *Rich Salz*
4519
4520 * Removed obsolete engines: 4758cca, aep, atalla, cswift, nuron, gmp,
4521   sureware and ubsec.
4522
4523   *Matt Caswell, Rich Salz*
4524
4525 * New ASN.1 embed macro.
4526
4527   New ASN.1 macro ASN1_EMBED. This is the same as ASN1_SIMPLE except the
4528   structure is not allocated: it is part of the parent. That is instead of
4529
4530           FOO *x;
4531
4532   it must be:
4533
4534           FOO x;
4535
4536   This reduces memory fragmentation and make it impossible to accidentally
4537   set a mandatory field to NULL.
4538
4539   This currently only works for some fields specifically a SEQUENCE, CHOICE,
4540   or ASN1_STRING type which is part of a parent SEQUENCE. Since it is
4541   equivalent to ASN1_SIMPLE it cannot be tagged, OPTIONAL, SET OF or
4542   SEQUENCE OF.
4543
4544   *Steve Henson*
4545
4546 * Remove EVP_CHECK_DES_KEY, a compile-time option that never compiled.
4547
4548   *Emilia Käsper*
4549
4550 * Removed DES and RC4 ciphersuites from DEFAULT. Also removed RC2 although
4551   in 1.0.2 EXPORT was already removed and the only RC2 ciphersuite is also
4552   an EXPORT one. COMPLEMENTOFDEFAULT has been updated accordingly to add
4553   DES and RC4 ciphersuites.
4554
4555   *Matt Caswell*
4556
4557 * Rewrite EVP_DecodeUpdate (base64 decoding) to fix several bugs.
4558   This changes the decoding behaviour for some invalid messages,
4559   though the change is mostly in the more lenient direction, and
4560   legacy behaviour is preserved as much as possible.
4561
4562   *Emilia Käsper*
4563
4564 * Fix no-stdio build.
4565   *David Woodhouse <David.Woodhouse@intel.com> and also*
4566   *Ivan Nestlerode <ivan.nestlerode@sonos.com>*
4567
4568 * New testing framework
4569   The testing framework has been largely rewritten and is now using
4570   perl and the perl modules Test::Harness and an extended variant of
4571   Test::More called OpenSSL::Test to do its work.  All test scripts in
4572   test/ have been rewritten into test recipes, and all direct calls to
4573   executables in test/Makefile have become individual recipes using the
4574   simplified testing OpenSSL::Test::Simple.
4575
4576   For documentation on our testing modules, do:
4577
4578           perldoc test/testlib/OpenSSL/Test/Simple.pm
4579           perldoc test/testlib/OpenSSL/Test.pm
4580
4581   *Richard Levitte*
4582
4583 * Revamped memory debug; only -DCRYPTO_MDEBUG and -DCRYPTO_MDEBUG_ABORT
4584   are used; the latter aborts on memory leaks (usually checked on exit).
4585   Some undocumented "set malloc, etc., hooks" functions were removed
4586   and others were changed.  All are now documented.
4587
4588   *Rich Salz*
4589
4590 * In DSA_generate_parameters_ex, if the provided seed is too short,
4591   return an error
4592
4593   *Rich Salz and Ismo Puustinen <ismo.puustinen@intel.com>*
4594
4595 * Rewrite PSK to support ECDHE_PSK, DHE_PSK and RSA_PSK. Add ciphersuites
4596   from RFC4279, RFC4785, RFC5487, RFC5489.
4597
4598   Thanks to Christian J. Dietrich and Giuseppe D'Angelo for the
4599   original RSA_PSK patch.
4600
4601   *Steve Henson*
4602
4603 * Dropped support for the SSL3_FLAGS_DELAY_CLIENT_FINISHED flag. This SSLeay
4604   era flag was never set throughout the codebase (only read). Also removed
4605   SSL3_FLAGS_POP_BUFFER which was only used if
4606   SSL3_FLAGS_DELAY_CLIENT_FINISHED was also set.
4607
4608   *Matt Caswell*
4609
4610 * Changed the default name options in the "ca", "crl", "req" and "x509"
4611   to be "oneline" instead of "compat".
4612
4613   *Richard Levitte*
4614
4615 * Remove SSL_OP_TLS_BLOCK_PADDING_BUG. This is SSLeay legacy, we're
4616   not aware of clients that still exhibit this bug, and the workaround
4617   hasn't been working properly for a while.
4618
4619   *Emilia Käsper*
4620
4621 * The return type of BIO_number_read() and BIO_number_written() as well as
4622   the corresponding num_read and num_write members in the BIO structure has
4623   changed from unsigned long to uint64_t. On platforms where an unsigned
4624   long is 32 bits (e.g. Windows) these counters could overflow if >4Gb is
4625   transferred.
4626
4627   *Matt Caswell*
4628
4629 * Given the pervasive nature of TLS extensions it is inadvisable to run
4630   OpenSSL without support for them. It also means that maintaining
4631   the OPENSSL_NO_TLSEXT option within the code is very invasive (and probably
4632   not well tested). Therefore the OPENSSL_NO_TLSEXT option has been removed.
4633
4634   *Matt Caswell*
4635
4636 * Removed support for the two export grade static DH ciphersuites
4637   EXP-DH-RSA-DES-CBC-SHA and EXP-DH-DSS-DES-CBC-SHA. These two ciphersuites
4638   were newly added (along with a number of other static DH ciphersuites) to
4639   1.0.2. However the two export ones have *never* worked since they were
4640   introduced. It seems strange in any case to be adding new export
4641   ciphersuites, and given "logjam" it also does not seem correct to fix them.
4642
4643   *Matt Caswell*
4644
4645 * Version negotiation has been rewritten. In particular SSLv23_method(),
4646   SSLv23_client_method() and SSLv23_server_method() have been deprecated,
4647   and turned into macros which simply call the new preferred function names
4648   TLS_method(), TLS_client_method() and TLS_server_method(). All new code
4649   should use the new names instead. Also as part of this change the ssl23.h
4650   header file has been removed.
4651
4652   *Matt Caswell*
4653
4654 * Support for Kerberos ciphersuites in TLS (RFC2712) has been removed. This
4655   code and the associated standard is no longer considered fit-for-purpose.
4656
4657   *Matt Caswell*
4658
4659 * RT2547 was closed.  When generating a private key, try to make the
4660   output file readable only by the owner.  This behavior change might
4661   be noticeable when interacting with other software.
4662
4663 * Documented all exdata functions.  Added CRYPTO_free_ex_index.
4664   Added a test.
4665
4666   *Rich Salz*
4667
4668 * Added HTTP GET support to the ocsp command.
4669
4670   *Rich Salz*
4671
4672 * Changed default digest for the dgst and enc commands from MD5 to
4673   sha256
4674
4675   *Rich Salz*
4676
4677 * RAND_pseudo_bytes has been deprecated. Users should use RAND_bytes instead.
4678
4679   *Matt Caswell*
4680
4681 * Added support for TLS extended master secret from
4682   draft-ietf-tls-session-hash-03.txt. Thanks for Alfredo Pironti for an
4683   initial patch which was a great help during development.
4684
4685   *Steve Henson*
4686
4687 * All libssl internal structures have been removed from the public header
4688   files, and the OPENSSL_NO_SSL_INTERN option has been removed (since it is
4689   now redundant). Users should not attempt to access internal structures
4690   directly. Instead they should use the provided API functions.
4691
4692   *Matt Caswell*
4693
4694 * config has been changed so that by default OPENSSL_NO_DEPRECATED is used.
4695   Access to deprecated functions can be re-enabled by running config with
4696   "enable-deprecated". In addition applications wishing to use deprecated
4697   functions must define OPENSSL_USE_DEPRECATED. Note that this new behaviour
4698   will, by default, disable some transitive includes that previously existed
4699   in the header files (e.g. ec.h will no longer, by default, include bn.h)
4700
4701   *Matt Caswell*
4702
4703 * Added support for OCB mode. OpenSSL has been granted a patent license
4704   compatible with the OpenSSL license for use of OCB. Details are available
4705   at <https://www.openssl.org/source/OCB-patent-grant-OpenSSL.pdf>. Support
4706   for OCB can be removed by calling config with no-ocb.
4707
4708   *Matt Caswell*
4709
4710 * SSLv2 support has been removed.  It still supports receiving a SSLv2
4711   compatible client hello.
4712
4713   *Kurt Roeckx*
4714
4715 * Increased the minimal RSA keysize from 256 to 512 bits [Rich Salz],
4716   done while fixing the error code for the key-too-small case.
4717
4718   *Annie Yousar <a.yousar@informatik.hu-berlin.de>*
4719
4720 * CA.sh has been removed; use CA.pl instead.
4721
4722   *Rich Salz*
4723
4724 * Removed old DES API.
4725
4726   *Rich Salz*
4727
4728 * Remove various unsupported platforms:
4729      Sony NEWS4
4730      BEOS and BEOS_R5
4731      NeXT
4732      SUNOS
4733      MPE/iX
4734      Sinix/ReliantUNIX RM400
4735      DGUX
4736      NCR
4737      Tandem
4738      Cray
4739      16-bit platforms such as WIN16
4740
4741   *Rich Salz*
4742
4743 * Clean up OPENSSL_NO_xxx #define's
4744   - Use setbuf() and remove OPENSSL_NO_SETVBUF_IONBF
4745   - Rename OPENSSL_SYSNAME_xxx to OPENSSL_SYS_xxx
4746   - OPENSSL_NO_EC{DH,DSA} merged into OPENSSL_NO_EC
4747   - OPENSSL_NO_RIPEMD160, OPENSSL_NO_RIPEMD merged into OPENSSL_NO_RMD160
4748   - OPENSSL_NO_FP_API merged into OPENSSL_NO_STDIO
4749   - Remove OPENSSL_NO_BIO OPENSSL_NO_BUFFER OPENSSL_NO_CHAIN_VERIFY
4750     OPENSSL_NO_EVP OPENSSL_NO_FIPS_ERR OPENSSL_NO_HASH_COMP
4751     OPENSSL_NO_LHASH OPENSSL_NO_OBJECT OPENSSL_NO_SPEED OPENSSL_NO_STACK
4752     OPENSSL_NO_X509 OPENSSL_NO_X509_VERIFY
4753   - Remove MS_STATIC; it's a relic from platforms <32 bits.
4754
4755   *Rich Salz*
4756
4757 * Cleaned up dead code
4758     Remove all but one '#ifdef undef' which is to be looked at.
4759
4760   *Rich Salz*
4761
4762 * Clean up calling of xxx_free routines.
4763      Just like free(), fix most of the xxx_free routines to accept
4764      NULL.  Remove the non-null checks from callers.  Save much code.
4765
4766   *Rich Salz*
4767
4768 * Add secure heap for storage of private keys (when possible).
4769   Add BIO_s_secmem(), CBIGNUM, etc.
4770   Contributed by Akamai Technologies under our Corporate CLA.
4771
4772   *Rich Salz*
4773
4774 * Experimental support for a new, fast, unbiased prime candidate generator,
4775   bn_probable_prime_dh_coprime(). Not currently used by any prime generator.
4776
4777   *Felix Laurie von Massenbach <felix@erbridge.co.uk>*
4778
4779 * New output format NSS in the sess_id command line tool. This allows
4780   exporting the session id and the master key in NSS keylog format.
4781
4782   *Martin Kaiser <martin@kaiser.cx>*
4783
4784 * Harmonize version and its documentation. -f flag is used to display
4785   compilation flags.
4786
4787   *mancha <mancha1@zoho.com>*
4788
4789 * Fix eckey_priv_encode so it immediately returns an error upon a failure
4790   in i2d_ECPrivateKey.  Thanks to Ted Unangst for feedback on this issue.
4791
4792   *mancha <mancha1@zoho.com>*
4793
4794 * Fix some double frees. These are not thought to be exploitable.
4795
4796   *mancha <mancha1@zoho.com>*
4797
4798 * A missing bounds check in the handling of the TLS heartbeat extension
4799   can be used to reveal up to 64k of memory to a connected client or
4800   server.
4801
4802   Thanks for Neel Mehta of Google Security for discovering this bug and to
4803   Adam Langley <agl@chromium.org> and Bodo Moeller <bmoeller@acm.org> for
4804   preparing the fix ([CVE-2014-0160])
4805
4806   *Adam Langley, Bodo Moeller*
4807
4808 * Fix for the attack described in the paper "Recovering OpenSSL
4809   ECDSA Nonces Using the FLUSH+RELOAD Cache Side-channel Attack"
4810   by Yuval Yarom and Naomi Benger. Details can be obtained from:
4811   <http://eprint.iacr.org/2014/140>
4812
4813   Thanks to Yuval Yarom and Naomi Benger for discovering this
4814   flaw and to Yuval Yarom for supplying a fix ([CVE-2014-0076])
4815
4816   *Yuval Yarom and Naomi Benger*
4817
4818 * Use algorithm specific chains in SSL_CTX_use_certificate_chain_file():
4819   this fixes a limitation in previous versions of OpenSSL.
4820
4821   *Steve Henson*
4822
4823 * Experimental encrypt-then-mac support.
4824
4825   Experimental support for encrypt then mac from
4826   draft-gutmann-tls-encrypt-then-mac-02.txt
4827
4828   To enable it set the appropriate extension number (0x42 for the test
4829   server) using e.g. -DTLSEXT_TYPE_encrypt_then_mac=0x42
4830
4831   For non-compliant peers (i.e. just about everything) this should have no
4832   effect.
4833
4834   WARNING: EXPERIMENTAL, SUBJECT TO CHANGE.
4835
4836   *Steve Henson*
4837
4838 * Add EVP support for key wrapping algorithms, to avoid problems with
4839   existing code the flag EVP_CIPHER_CTX_WRAP_ALLOW has to be set in
4840   the EVP_CIPHER_CTX or an error is returned. Add AES and DES3 wrap
4841   algorithms and include tests cases.
4842
4843   *Steve Henson*
4844
4845 * Extend CMS code to support RSA-PSS signatures and RSA-OAEP for
4846   enveloped data.
4847
4848   *Steve Henson*
4849
4850 * Extended RSA OAEP support via EVP_PKEY API. Options to specify digest,
4851   MGF1 digest and OAEP label.
4852
4853   *Steve Henson*
4854
4855 * Make openssl verify return errors.
4856
4857   *Chris Palmer <palmer@google.com> and Ben Laurie*
4858
4859 * New function ASN1_TIME_diff to calculate the difference between two
4860   ASN1_TIME structures or one structure and the current time.
4861
4862   *Steve Henson*
4863
4864 * Update fips_test_suite to support multiple command line options. New
4865   test to induce all self test errors in sequence and check expected
4866   failures.
4867
4868   *Steve Henson*
4869
4870 * Add FIPS_{rsa,dsa,ecdsa}_{sign,verify} functions which digest and
4871   sign or verify all in one operation.
4872
4873   *Steve Henson*
4874
4875 * Add fips_algvs: a multicall fips utility incorporating all the algorithm
4876   test programs and fips_test_suite. Includes functionality to parse
4877   the minimal script output of fipsalgest.pl directly.
4878
4879   *Steve Henson*
4880
4881 * Add authorisation parameter to FIPS_module_mode_set().
4882
4883   *Steve Henson*
4884
4885 * Add FIPS selftest for ECDH algorithm using P-224 and B-233 curves.
4886
4887   *Steve Henson*
4888
4889 * Use separate DRBG fields for internal and external flags. New function
4890   FIPS_drbg_health_check() to perform on demand health checking. Add
4891   generation tests to fips_test_suite with reduced health check interval to
4892   demonstrate periodic health checking. Add "nodh" option to
4893   fips_test_suite to skip very slow DH test.
4894
4895   *Steve Henson*
4896
4897 * New function FIPS_get_cipherbynid() to lookup FIPS supported ciphers
4898   based on NID.
4899
4900   *Steve Henson*
4901
4902 * More extensive health check for DRBG checking many more failure modes.
4903   New function FIPS_selftest_drbg_all() to handle every possible DRBG
4904   combination: call this in fips_test_suite.
4905
4906   *Steve Henson*
4907
4908 * Add support for canonical generation of DSA parameter 'g'. See
4909   FIPS 186-3 A.2.3.
4910
4911 * Add support for HMAC DRBG from SP800-90. Update DRBG algorithm test and
4912   POST to handle HMAC cases.
4913
4914   *Steve Henson*
4915
4916 * Add functions FIPS_module_version() and FIPS_module_version_text()
4917   to return numerical and string versions of the FIPS module number.
4918
4919   *Steve Henson*
4920
4921 * Rename FIPS_mode_set and FIPS_mode to FIPS_module_mode_set and
4922   FIPS_module_mode. FIPS_mode and FIPS_mode_set will be implemented
4923   outside the validated module in the FIPS capable OpenSSL.
4924
4925   *Steve Henson*
4926
4927 * Minor change to DRBG entropy callback semantics. In some cases
4928   there is no multiple of the block length between min_len and
4929   max_len. Allow the callback to return more than max_len bytes
4930   of entropy but discard any extra: it is the callback's responsibility
4931   to ensure that the extra data discarded does not impact the
4932   requested amount of entropy.
4933
4934   *Steve Henson*
4935
4936 * Add PRNG security strength checks to RSA, DSA and ECDSA using
4937   information in FIPS186-3, SP800-57 and SP800-131A.
4938
4939   *Steve Henson*
4940
4941 * CCM support via EVP. Interface is very similar to GCM case except we
4942   must supply all data in one chunk (i.e. no update, final) and the
4943   message length must be supplied if AAD is used. Add algorithm test
4944   support.
4945
4946   *Steve Henson*
4947
4948 * Initial version of POST overhaul. Add POST callback to allow the status
4949   of POST to be monitored and/or failures induced. Modify fips_test_suite
4950   to use callback. Always run all selftests even if one fails.
4951
4952   *Steve Henson*
4953
4954 * XTS support including algorithm test driver in the fips_gcmtest program.
4955   Note: this does increase the maximum key length from 32 to 64 bytes but
4956   there should be no binary compatibility issues as existing applications
4957   will never use XTS mode.
4958
4959   *Steve Henson*
4960
4961 * Extensive reorganisation of FIPS PRNG behaviour. Remove all dependencies
4962   to OpenSSL RAND code and replace with a tiny FIPS RAND API which also
4963   performs algorithm blocking for unapproved PRNG types. Also do not
4964   set PRNG type in FIPS_mode_set(): leave this to the application.
4965   Add default OpenSSL DRBG handling: sets up FIPS PRNG and seeds with
4966   the standard OpenSSL PRNG: set additional data to a date time vector.
4967
4968   *Steve Henson*
4969
4970 * Rename old X9.31 PRNG functions of the form `FIPS_rand*` to `FIPS_x931*`.
4971   This shouldn't present any incompatibility problems because applications
4972   shouldn't be using these directly and any that are will need to rethink
4973   anyway as the X9.31 PRNG is now deprecated by FIPS 140-2
4974
4975   *Steve Henson*
4976
4977 * Extensive self tests and health checking required by SP800-90 DRBG.
4978   Remove strength parameter from FIPS_drbg_instantiate and always
4979   instantiate at maximum supported strength.
4980
4981   *Steve Henson*
4982
4983 * Add ECDH code to fips module and fips_ecdhvs for primitives only testing.
4984
4985   *Steve Henson*
4986
4987 * New algorithm test program fips_dhvs to handle DH primitives only testing.
4988
4989   *Steve Henson*
4990
4991 * New function DH_compute_key_padded() to compute a DH key and pad with
4992   leading zeroes if needed: this complies with SP800-56A et al.
4993
4994   *Steve Henson*
4995
4996 * Initial implementation of SP800-90 DRBGs for Hash and CTR. Not used by
4997   anything, incomplete, subject to change and largely untested at present.
4998
4999   *Steve Henson*
5000
5001 * Modify fipscanisteronly build option to only build the necessary object
5002   files by filtering FIPS_EX_OBJ through a perl script in crypto/Makefile.
5003
5004   *Steve Henson*
5005
5006 * Add experimental option FIPSSYMS to give all symbols in
5007   fipscanister.o and FIPS or fips prefix. This will avoid
5008   conflicts with future versions of OpenSSL. Add perl script
5009   util/fipsas.pl to preprocess assembly language source files
5010   and rename any affected symbols.
5011
5012   *Steve Henson*
5013
5014 * Add selftest checks and algorithm block of non-fips algorithms in
5015   FIPS mode. Remove DES2 from selftests.
5016
5017   *Steve Henson*
5018
5019 * Add ECDSA code to fips module. Add tiny fips_ecdsa_check to just
5020   return internal method without any ENGINE dependencies. Add new
5021   tiny fips sign and verify functions.
5022
5023   *Steve Henson*
5024
5025 * New build option no-ec2m to disable characteristic 2 code.
5026
5027   *Steve Henson*
5028
5029 * New build option "fipscanisteronly". This only builds fipscanister.o
5030   and (currently) associated fips utilities. Uses the file Makefile.fips
5031   instead of Makefile.org as the prototype.
5032
5033   *Steve Henson*
5034
5035 * Add some FIPS mode restrictions to GCM. Add internal IV generator.
5036   Update fips_gcmtest to use IV generator.
5037
5038   *Steve Henson*
5039
5040 * Initial, experimental EVP support for AES-GCM. AAD can be input by
5041   setting output buffer to NULL. The `*Final` function must be
5042   called although it will not retrieve any additional data. The tag
5043   can be set or retrieved with a ctrl. The IV length is by default 12
5044   bytes (96 bits) but can be set to an alternative value. If the IV
5045   length exceeds the maximum IV length (currently 16 bytes) it cannot be
5046   set before the key.
5047
5048   *Steve Henson*
5049
5050 * New flag in ciphers: EVP_CIPH_FLAG_CUSTOM_CIPHER. This means the
5051   underlying do_cipher function handles all cipher semantics itself
5052   including padding and finalisation. This is useful if (for example)
5053   an ENGINE cipher handles block padding itself. The behaviour of
5054   do_cipher is subtly changed if this flag is set: the return value
5055   is the number of characters written to the output buffer (zero is
5056   no longer an error code) or a negative error code. Also if the
5057   input buffer is NULL and length 0 finalisation should be performed.
5058
5059   *Steve Henson*
5060
5061 * If a candidate issuer certificate is already part of the constructed
5062   path ignore it: new debug notification X509_V_ERR_PATH_LOOP for this case.
5063
5064   *Steve Henson*
5065
5066 * Improve forward-security support: add functions
5067
5068           void SSL_CTX_set_not_resumable_session_callback(
5069                    SSL_CTX *ctx, int (*cb)(SSL *ssl, int is_forward_secure))
5070           void SSL_set_not_resumable_session_callback(
5071                    SSL *ssl, int (*cb)(SSL *ssl, int is_forward_secure))
5072
5073   for use by SSL/TLS servers; the callback function will be called whenever a
5074   new session is created, and gets to decide whether the session may be
5075   cached to make it resumable (return 0) or not (return 1).  (As by the
5076   SSL/TLS protocol specifications, the session_id sent by the server will be
5077   empty to indicate that the session is not resumable; also, the server will
5078   not generate RFC 4507 (RFC 5077) session tickets.)
5079
5080   A simple reasonable callback implementation is to return is_forward_secure.
5081   This parameter will be set to 1 or 0 depending on the ciphersuite selected
5082   by the SSL/TLS server library, indicating whether it can provide forward
5083   security.
5084
5085   *Emilia Käsper <emilia.kasper@esat.kuleuven.be> (Google)*
5086
5087 * New -verify_name option in command line utilities to set verification
5088   parameters by name.
5089
5090   *Steve Henson*
5091
5092 * Initial CMAC implementation. WARNING: EXPERIMENTAL, API MAY CHANGE.
5093   Add CMAC pkey methods.
5094
5095   *Steve Henson*
5096
5097 * Experimental renegotiation in s_server -www mode. If the client
5098   browses /reneg connection is renegotiated. If /renegcert it is
5099   renegotiated requesting a certificate.
5100
5101   *Steve Henson*
5102
5103 * Add an "external" session cache for debugging purposes to s_server. This
5104   should help trace issues which normally are only apparent in deployed
5105   multi-process servers.
5106
5107   *Steve Henson*
5108
5109 * Extensive audit of libcrypto with DEBUG_UNUSED. Fix many cases where
5110   return value is ignored. NB. The functions RAND_add(), RAND_seed(),
5111   BIO_set_cipher() and some obscure PEM functions were changed so they
5112   can now return an error. The RAND changes required a change to the
5113   RAND_METHOD structure.
5114
5115   *Steve Henson*
5116
5117 * New macro `__owur` for "OpenSSL Warn Unused Result". This makes use of
5118   a gcc attribute to warn if the result of a function is ignored. This
5119   is enable if DEBUG_UNUSED is set. Add to several functions in evp.h
5120   whose return value is often ignored.
5121
5122   *Steve Henson*
5123
5124 * New -noct, -requestct, -requirect and -ctlogfile options for s_client.
5125   These allow SCTs (signed certificate timestamps) to be requested and
5126   validated when establishing a connection.
5127
5128   *Rob Percival <robpercival@google.com>*
5129
5130OpenSSL 1.0.2
5131-------------
5132
5133### Changes between 1.0.2s and 1.0.2t [10 Sep 2019]
5134
5135 * For built-in EC curves, ensure an EC_GROUP built from the curve name is
5136   used even when parsing explicit parameters, when loading a encoded key
5137   or calling `EC_GROUP_new_from_ecpkparameters()`/
5138   `EC_GROUP_new_from_ecparameters()`.
5139   This prevents bypass of security hardening and performance gains,
5140   especially for curves with specialized EC_METHODs.
5141   By default, if a key encoded with explicit parameters is loaded and later
5142   encoded, the output is still encoded with explicit parameters, even if
5143   internally a "named" EC_GROUP is used for computation.
5144
5145   *Nicola Tuveri*
5146
5147 * Compute ECC cofactors if not provided during EC_GROUP construction. Before
5148   this change, EC_GROUP_set_generator would accept order and/or cofactor as
5149   NULL. After this change, only the cofactor parameter can be NULL. It also
5150   does some minimal sanity checks on the passed order.
5151   ([CVE-2019-1547])
5152
5153   *Billy Bob Brumley*
5154
5155 * Fixed a padding oracle in PKCS7_dataDecode and CMS_decrypt_set1_pkey.
5156   An attack is simple, if the first CMS_recipientInfo is valid but the
5157   second CMS_recipientInfo is chosen ciphertext. If the second
5158   recipientInfo decodes to PKCS #1 v1.5 form plaintext, the correct
5159   encryption key will be replaced by garbage, and the message cannot be
5160   decoded, but if the RSA decryption fails, the correct encryption key is
5161   used and the recipient will not notice the attack.
5162   As a work around for this potential attack the length of the decrypted
5163   key must be equal to the cipher default key length, in case the
5164   certifiate is not given and all recipientInfo are tried out.
5165   The old behaviour can be re-enabled in the CMS code by setting the
5166   CMS_DEBUG_DECRYPT flag.
5167   ([CVE-2019-1563])
5168
5169   *Bernd Edlinger*
5170
5171 * Document issue with installation paths in diverse Windows builds
5172
5173   '/usr/local/ssl' is an unsafe prefix for location to install OpenSSL
5174   binaries and run-time config file.
5175   ([CVE-2019-1552])
5176
5177   *Richard Levitte*
5178
5179### Changes between 1.0.2r and 1.0.2s [28 May 2019]
5180
5181 * Change the default RSA, DSA and DH size to 2048 bit instead of 1024.
5182   This changes the size when using the `genpkey` command when no size is given.
5183   It fixes an omission in earlier changes that changed all RSA, DSA and DH
5184   generation commands to use 2048 bits by default.
5185
5186   *Kurt Roeckx*
5187
5188 * Add FIPS support for Android Arm 64-bit
5189
5190   Support for Android Arm 64-bit was added to the OpenSSL FIPS Object
5191   Module in Version 2.0.10. For some reason, the corresponding target
5192   'android64-aarch64' was missing OpenSSL 1.0.2, whence it could not be
5193   built with FIPS support on Android Arm 64-bit. This omission has been
5194   fixed.
5195
5196   *Matthias St. Pierre*
5197
5198### Changes between 1.0.2q and 1.0.2r [26 Feb 2019]
5199
5200 * 0-byte record padding oracle
5201
5202   If an application encounters a fatal protocol error and then calls
5203   SSL_shutdown() twice (once to send a close_notify, and once to receive one)
5204   then OpenSSL can respond differently to the calling application if a 0 byte
5205   record is received with invalid padding compared to if a 0 byte record is
5206   received with an invalid MAC. If the application then behaves differently
5207   based on that in a way that is detectable to the remote peer, then this
5208   amounts to a padding oracle that could be used to decrypt data.
5209
5210   In order for this to be exploitable "non-stitched" ciphersuites must be in
5211   use. Stitched ciphersuites are optimised implementations of certain
5212   commonly used ciphersuites. Also the application must call SSL_shutdown()
5213   twice even if a protocol error has occurred (applications should not do
5214   this but some do anyway).
5215
5216   This issue was discovered by Juraj Somorovsky, Robert Merget and Nimrod
5217   Aviram, with additional investigation by Steven Collison and Andrew
5218   Hourselt. It was reported to OpenSSL on 10th December 2018.
5219   ([CVE-2019-1559])
5220
5221   *Matt Caswell*
5222
5223 * Move strictness check from EVP_PKEY_asn1_new() to EVP_PKEY_asn1_add0().
5224
5225   *Richard Levitte*
5226
5227### Changes between 1.0.2p and 1.0.2q [20 Nov 2018]
5228
5229 * Microarchitecture timing vulnerability in ECC scalar multiplication
5230
5231   OpenSSL ECC scalar multiplication, used in e.g. ECDSA and ECDH, has been
5232   shown to be vulnerable to a microarchitecture timing side channel attack.
5233   An attacker with sufficient access to mount local timing attacks during
5234   ECDSA signature generation could recover the private key.
5235
5236   This issue was reported to OpenSSL on 26th October 2018 by Alejandro
5237   Cabrera Aldaya, Billy Brumley, Sohaib ul Hassan, Cesar Pereida Garcia and
5238   Nicola Tuveri.
5239   ([CVE-2018-5407])
5240
5241   *Billy Brumley*
5242
5243 * Timing vulnerability in DSA signature generation
5244
5245   The OpenSSL DSA signature algorithm has been shown to be vulnerable to a
5246   timing side channel attack. An attacker could use variations in the signing
5247   algorithm to recover the private key.
5248
5249   This issue was reported to OpenSSL on 16th October 2018 by Samuel Weiser.
5250   ([CVE-2018-0734])
5251
5252   *Paul Dale*
5253
5254 * Resolve a compatibility issue in EC_GROUP handling with the FIPS Object
5255   Module, accidentally introduced while backporting security fixes from the
5256   development branch and hindering the use of ECC in FIPS mode.
5257
5258   *Nicola Tuveri*
5259
5260### Changes between 1.0.2o and 1.0.2p [14 Aug 2018]
5261
5262 * Client DoS due to large DH parameter
5263
5264   During key agreement in a TLS handshake using a DH(E) based ciphersuite a
5265   malicious server can send a very large prime value to the client. This will
5266   cause the client to spend an unreasonably long period of time generating a
5267   key for this prime resulting in a hang until the client has finished. This
5268   could be exploited in a Denial Of Service attack.
5269
5270   This issue was reported to OpenSSL on 5th June 2018 by Guido Vranken
5271   ([CVE-2018-0732])
5272
5273   *Guido Vranken*
5274
5275 * Cache timing vulnerability in RSA Key Generation
5276
5277   The OpenSSL RSA Key generation algorithm has been shown to be vulnerable to
5278   a cache timing side channel attack. An attacker with sufficient access to
5279   mount cache timing attacks during the RSA key generation process could
5280   recover the private key.
5281
5282   This issue was reported to OpenSSL on 4th April 2018 by Alejandro Cabrera
5283   Aldaya, Billy Brumley, Cesar Pereida Garcia and Luis Manuel Alvarez Tapia.
5284   ([CVE-2018-0737])
5285
5286   *Billy Brumley*
5287
5288 * Make EVP_PKEY_asn1_new() a bit stricter about its input.  A NULL pem_str
5289   parameter is no longer accepted, as it leads to a corrupt table.  NULL
5290   pem_str is reserved for alias entries only.
5291
5292   *Richard Levitte*
5293
5294 * Revert blinding in ECDSA sign and instead make problematic addition
5295   length-invariant. Switch even to fixed-length Montgomery multiplication.
5296
5297   *Andy Polyakov*
5298
5299 * Change generating and checking of primes so that the error rate of not
5300   being prime depends on the intended use based on the size of the input.
5301   For larger primes this will result in more rounds of Miller-Rabin.
5302   The maximal error rate for primes with more than 1080 bits is lowered
5303   to 2^-128.
5304
5305   *Kurt Roeckx, Annie Yousar*
5306
5307 * Increase the number of Miller-Rabin rounds for DSA key generating to 64.
5308
5309   *Kurt Roeckx*
5310
5311 * Add blinding to ECDSA and DSA signatures to protect against side channel
5312   attacks discovered by Keegan Ryan (NCC Group).
5313
5314   *Matt Caswell*
5315
5316 * When unlocking a pass phrase protected PEM file or PKCS#8 container, we
5317   now allow empty (zero character) pass phrases.
5318
5319   *Richard Levitte*
5320
5321 * Certificate time validation (X509_cmp_time) enforces stricter
5322   compliance with RFC 5280. Fractional seconds and timezone offsets
5323   are no longer allowed.
5324
5325   *Emilia Käsper*
5326
5327### Changes between 1.0.2n and 1.0.2o [27 Mar 2018]
5328
5329 * Constructed ASN.1 types with a recursive definition could exceed the stack
5330
5331   Constructed ASN.1 types with a recursive definition (such as can be found
5332   in PKCS7) could eventually exceed the stack given malicious input with
5333   excessive recursion. This could result in a Denial Of Service attack. There
5334   are no such structures used within SSL/TLS that come from untrusted sources
5335   so this is considered safe.
5336
5337   This issue was reported to OpenSSL on 4th January 2018 by the OSS-fuzz
5338   project.
5339   ([CVE-2018-0739])
5340
5341   *Matt Caswell*
5342
5343### Changes between 1.0.2m and 1.0.2n [7 Dec 2017]
5344
5345 * Read/write after SSL object in error state
5346
5347   OpenSSL 1.0.2 (starting from version 1.0.2b) introduced an "error state"
5348   mechanism. The intent was that if a fatal error occurred during a handshake
5349   then OpenSSL would move into the error state and would immediately fail if
5350   you attempted to continue the handshake. This works as designed for the
5351   explicit handshake functions (SSL_do_handshake(), SSL_accept() and
5352   SSL_connect()), however due to a bug it does not work correctly if
5353   SSL_read() or SSL_write() is called directly. In that scenario, if the
5354   handshake fails then a fatal error will be returned in the initial function
5355   call. If SSL_read()/SSL_write() is subsequently called by the application
5356   for the same SSL object then it will succeed and the data is passed without
5357   being decrypted/encrypted directly from the SSL/TLS record layer.
5358
5359   In order to exploit this issue an application bug would have to be present
5360   that resulted in a call to SSL_read()/SSL_write() being issued after having
5361   already received a fatal error.
5362
5363   This issue was reported to OpenSSL by David Benjamin (Google).
5364   ([CVE-2017-3737])
5365
5366   *Matt Caswell*
5367
5368 * rsaz_1024_mul_avx2 overflow bug on x86_64
5369
5370   There is an overflow bug in the AVX2 Montgomery multiplication procedure
5371   used in exponentiation with 1024-bit moduli. No EC algorithms are affected.
5372   Analysis suggests that attacks against RSA and DSA as a result of this
5373   defect would be very difficult to perform and are not believed likely.
5374   Attacks against DH1024 are considered just feasible, because most of the
5375   work necessary to deduce information about a private key may be performed
5376   offline. The amount of resources required for such an attack would be
5377   significant. However, for an attack on TLS to be meaningful, the server
5378   would have to share the DH1024 private key among multiple clients, which is
5379   no longer an option since CVE-2016-0701.
5380
5381   This only affects processors that support the AVX2 but not ADX extensions
5382   like Intel Haswell (4th generation).
5383
5384   This issue was reported to OpenSSL by David Benjamin (Google). The issue
5385   was originally found via the OSS-Fuzz project.
5386   ([CVE-2017-3738])
5387
5388   *Andy Polyakov*
5389
5390### Changes between 1.0.2l and 1.0.2m [2 Nov 2017]
5391
5392 * bn_sqrx8x_internal carry bug on x86_64
5393
5394   There is a carry propagating bug in the x86_64 Montgomery squaring
5395   procedure. No EC algorithms are affected. Analysis suggests that attacks
5396   against RSA and DSA as a result of this defect would be very difficult to
5397   perform and are not believed likely. Attacks against DH are considered just
5398   feasible (although very difficult) because most of the work necessary to
5399   deduce information about a private key may be performed offline. The amount
5400   of resources required for such an attack would be very significant and
5401   likely only accessible to a limited number of attackers. An attacker would
5402   additionally need online access to an unpatched system using the target
5403   private key in a scenario with persistent DH parameters and a private
5404   key that is shared between multiple clients.
5405
5406   This only affects processors that support the BMI1, BMI2 and ADX extensions
5407   like Intel Broadwell (5th generation) and later or AMD Ryzen.
5408
5409   This issue was reported to OpenSSL by the OSS-Fuzz project.
5410   ([CVE-2017-3736])
5411
5412   *Andy Polyakov*
5413
5414 * Malformed X.509 IPAddressFamily could cause OOB read
5415
5416   If an X.509 certificate has a malformed IPAddressFamily extension,
5417   OpenSSL could do a one-byte buffer overread. The most likely result
5418   would be an erroneous display of the certificate in text format.
5419
5420   This issue was reported to OpenSSL by the OSS-Fuzz project.
5421
5422   *Rich Salz*
5423
5424### Changes between 1.0.2k and 1.0.2l [25 May 2017]
5425
5426 * Have 'config' recognise 64-bit mingw and choose 'mingw64' as the target
5427   platform rather than 'mingw'.
5428
5429   *Richard Levitte*
5430
5431### Changes between 1.0.2j and 1.0.2k [26 Jan 2017]
5432
5433 * Truncated packet could crash via OOB read
5434
5435   If one side of an SSL/TLS path is running on a 32-bit host and a specific
5436   cipher is being used, then a truncated packet can cause that host to
5437   perform an out-of-bounds read, usually resulting in a crash.
5438
5439   This issue was reported to OpenSSL by Robert Święcki of Google.
5440   ([CVE-2017-3731])
5441
5442   *Andy Polyakov*
5443
5444 * BN_mod_exp may produce incorrect results on x86_64
5445
5446   There is a carry propagating bug in the x86_64 Montgomery squaring
5447   procedure. No EC algorithms are affected. Analysis suggests that attacks
5448   against RSA and DSA as a result of this defect would be very difficult to
5449   perform and are not believed likely. Attacks against DH are considered just
5450   feasible (although very difficult) because most of the work necessary to
5451   deduce information about a private key may be performed offline. The amount
5452   of resources required for such an attack would be very significant and
5453   likely only accessible to a limited number of attackers. An attacker would
5454   additionally need online access to an unpatched system using the target
5455   private key in a scenario with persistent DH parameters and a private
5456   key that is shared between multiple clients. For example this can occur by
5457   default in OpenSSL DHE based SSL/TLS ciphersuites. Note: This issue is very
5458   similar to CVE-2015-3193 but must be treated as a separate problem.
5459
5460   This issue was reported to OpenSSL by the OSS-Fuzz project.
5461   ([CVE-2017-3732])
5462
5463   *Andy Polyakov*
5464
5465 * Montgomery multiplication may produce incorrect results
5466
5467   There is a carry propagating bug in the Broadwell-specific Montgomery
5468   multiplication procedure that handles input lengths divisible by, but
5469   longer than 256 bits. Analysis suggests that attacks against RSA, DSA
5470   and DH private keys are impossible. This is because the subroutine in
5471   question is not used in operations with the private key itself and an input
5472   of the attacker's direct choice. Otherwise the bug can manifest itself as
5473   transient authentication and key negotiation failures or reproducible
5474   erroneous outcome of public-key operations with specially crafted input.
5475   Among EC algorithms only Brainpool P-512 curves are affected and one
5476   presumably can attack ECDH key negotiation. Impact was not analyzed in
5477   detail, because pre-requisites for attack are considered unlikely. Namely
5478   multiple clients have to choose the curve in question and the server has to
5479   share the private key among them, neither of which is default behaviour.
5480   Even then only clients that chose the curve will be affected.
5481
5482   This issue was publicly reported as transient failures and was not
5483   initially recognized as a security issue. Thanks to Richard Morgan for
5484   providing reproducible case.
5485   ([CVE-2016-7055])
5486
5487   *Andy Polyakov*
5488
5489 * OpenSSL now fails if it receives an unrecognised record type in TLS1.0
5490   or TLS1.1. Previously this only happened in SSLv3 and TLS1.2. This is to
5491   prevent issues where no progress is being made and the peer continually
5492   sends unrecognised record types, using up resources processing them.
5493
5494   *Matt Caswell*
5495
5496### Changes between 1.0.2i and 1.0.2j [26 Sep 2016]
5497
5498 * Missing CRL sanity check
5499
5500   A bug fix which included a CRL sanity check was added to OpenSSL 1.1.0
5501   but was omitted from OpenSSL 1.0.2i. As a result any attempt to use
5502   CRLs in OpenSSL 1.0.2i will crash with a null pointer exception.
5503
5504   This issue only affects the OpenSSL 1.0.2i
5505   ([CVE-2016-7052])
5506
5507   *Matt Caswell*
5508
5509### Changes between 1.0.2h and 1.0.2i [22 Sep 2016]
5510
5511 * OCSP Status Request extension unbounded memory growth
5512
5513   A malicious client can send an excessively large OCSP Status Request
5514   extension. If that client continually requests renegotiation, sending a
5515   large OCSP Status Request extension each time, then there will be unbounded
5516   memory growth on the server. This will eventually lead to a Denial Of
5517   Service attack through memory exhaustion. Servers with a default
5518   configuration are vulnerable even if they do not support OCSP. Builds using
5519   the "no-ocsp" build time option are not affected.
5520
5521   This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.)
5522   ([CVE-2016-6304])
5523
5524   *Matt Caswell*
5525
5526 * In order to mitigate the SWEET32 attack, the DES ciphers were moved from
5527   HIGH to MEDIUM.
5528
5529   This issue was reported to OpenSSL Karthikeyan Bhargavan and Gaetan
5530   Leurent (INRIA)
5531   ([CVE-2016-2183])
5532
5533   *Rich Salz*
5534
5535 * OOB write in MDC2_Update()
5536
5537   An overflow can occur in MDC2_Update() either if called directly or
5538   through the EVP_DigestUpdate() function using MDC2. If an attacker
5539   is able to supply very large amounts of input data after a previous
5540   call to EVP_EncryptUpdate() with a partial block then a length check
5541   can overflow resulting in a heap corruption.
5542
5543   The amount of data needed is comparable to SIZE_MAX which is impractical
5544   on most platforms.
5545
5546   This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.)
5547   ([CVE-2016-6303])
5548
5549   *Stephen Henson*
5550
5551 * Malformed SHA512 ticket DoS
5552
5553   If a server uses SHA512 for TLS session ticket HMAC it is vulnerable to a
5554   DoS attack where a malformed ticket will result in an OOB read which will
5555   ultimately crash.
5556
5557   The use of SHA512 in TLS session tickets is comparatively rare as it requires
5558   a custom server callback and ticket lookup mechanism.
5559
5560   This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.)
5561   ([CVE-2016-6302])
5562
5563   *Stephen Henson*
5564
5565 * OOB write in BN_bn2dec()
5566
5567   The function BN_bn2dec() does not check the return value of BN_div_word().
5568   This can cause an OOB write if an application uses this function with an
5569   overly large BIGNUM. This could be a problem if an overly large certificate
5570   or CRL is printed out from an untrusted source. TLS is not affected because
5571   record limits will reject an oversized certificate before it is parsed.
5572
5573   This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.)
5574   ([CVE-2016-2182])
5575
5576   *Stephen Henson*
5577
5578 * OOB read in TS_OBJ_print_bio()
5579
5580   The function TS_OBJ_print_bio() misuses OBJ_obj2txt(): the return value is
5581   the total length the OID text representation would use and not the amount
5582   of data written. This will result in OOB reads when large OIDs are
5583   presented.
5584
5585   This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.)
5586   ([CVE-2016-2180])
5587
5588   *Stephen Henson*
5589
5590 * Pointer arithmetic undefined behaviour
5591
5592   Avoid some undefined pointer arithmetic
5593
5594   A common idiom in the codebase is to check limits in the following manner:
5595   "p + len > limit"
5596
5597   Where "p" points to some malloc'd data of SIZE bytes and
5598   limit == p + SIZE
5599
5600   "len" here could be from some externally supplied data (e.g. from a TLS
5601   message).
5602
5603   The rules of C pointer arithmetic are such that "p + len" is only well
5604   defined where len <= SIZE. Therefore the above idiom is actually
5605   undefined behaviour.
5606
5607   For example this could cause problems if some malloc implementation
5608   provides an address for "p" such that "p + len" actually overflows for
5609   values of len that are too big and therefore p + len < limit.
5610
5611   This issue was reported to OpenSSL by Guido Vranken
5612   ([CVE-2016-2177])
5613
5614   *Matt Caswell*
5615
5616 * Constant time flag not preserved in DSA signing
5617
5618   Operations in the DSA signing algorithm should run in constant time in
5619   order to avoid side channel attacks. A flaw in the OpenSSL DSA
5620   implementation means that a non-constant time codepath is followed for
5621   certain operations. This has been demonstrated through a cache-timing
5622   attack to be sufficient for an attacker to recover the private DSA key.
5623
5624   This issue was reported by César Pereida (Aalto University), Billy Brumley
5625   (Tampere University of Technology), and Yuval Yarom (The University of
5626   Adelaide and NICTA).
5627   ([CVE-2016-2178])
5628
5629   *César Pereida*
5630
5631 * DTLS buffered message DoS
5632
5633   In a DTLS connection where handshake messages are delivered out-of-order
5634   those messages that OpenSSL is not yet ready to process will be buffered
5635   for later use. Under certain circumstances, a flaw in the logic means that
5636   those messages do not get removed from the buffer even though the handshake
5637   has been completed. An attacker could force up to approx. 15 messages to
5638   remain in the buffer when they are no longer required. These messages will
5639   be cleared when the DTLS connection is closed. The default maximum size for
5640   a message is 100k. Therefore the attacker could force an additional 1500k
5641   to be consumed per connection. By opening many simulataneous connections an
5642   attacker could cause a DoS attack through memory exhaustion.
5643
5644   This issue was reported to OpenSSL by Quan Luo.
5645   ([CVE-2016-2179])
5646
5647   *Matt Caswell*
5648
5649 * DTLS replay protection DoS
5650
5651   A flaw in the DTLS replay attack protection mechanism means that records
5652   that arrive for future epochs update the replay protection "window" before
5653   the MAC for the record has been validated. This could be exploited by an
5654   attacker by sending a record for the next epoch (which does not have to
5655   decrypt or have a valid MAC), with a very large sequence number. This means
5656   that all subsequent legitimate packets are dropped causing a denial of
5657   service for a specific DTLS connection.
5658
5659   This issue was reported to OpenSSL by the OCAP audit team.
5660   ([CVE-2016-2181])
5661
5662   *Matt Caswell*
5663
5664 * Certificate message OOB reads
5665
5666   In OpenSSL 1.0.2 and earlier some missing message length checks can result
5667   in OOB reads of up to 2 bytes beyond an allocated buffer. There is a
5668   theoretical DoS risk but this has not been observed in practice on common
5669   platforms.
5670
5671   The messages affected are client certificate, client certificate request
5672   and server certificate. As a result the attack can only be performed
5673   against a client or a server which enables client authentication.
5674
5675   This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.)
5676   ([CVE-2016-6306])
5677
5678   *Stephen Henson*
5679
5680### Changes between 1.0.2g and 1.0.2h [3 May 2016]
5681
5682 * Prevent padding oracle in AES-NI CBC MAC check
5683
5684   A MITM attacker can use a padding oracle attack to decrypt traffic
5685   when the connection uses an AES CBC cipher and the server support
5686   AES-NI.
5687
5688   This issue was introduced as part of the fix for Lucky 13 padding
5689   attack ([CVE-2013-0169]). The padding check was rewritten to be in
5690   constant time by making sure that always the same bytes are read and
5691   compared against either the MAC or padding bytes. But it no longer
5692   checked that there was enough data to have both the MAC and padding
5693   bytes.
5694
5695   This issue was reported by Juraj Somorovsky using TLS-Attacker.
5696
5697   *Kurt Roeckx*
5698
5699 * Fix EVP_EncodeUpdate overflow
5700
5701   An overflow can occur in the EVP_EncodeUpdate() function which is used for
5702   Base64 encoding of binary data. If an attacker is able to supply very large
5703   amounts of input data then a length check can overflow resulting in a heap
5704   corruption.
5705
5706   Internally to OpenSSL the EVP_EncodeUpdate() function is primarily used by
5707   the `PEM_write_bio*` family of functions. These are mainly used within the
5708   OpenSSL command line applications, so any application which processes data
5709   from an untrusted source and outputs it as a PEM file should be considered
5710   vulnerable to this issue. User applications that call these APIs directly
5711   with large amounts of untrusted data may also be vulnerable.
5712
5713   This issue was reported by Guido Vranken.
5714   ([CVE-2016-2105])
5715
5716   *Matt Caswell*
5717
5718 * Fix EVP_EncryptUpdate overflow
5719
5720   An overflow can occur in the EVP_EncryptUpdate() function. If an attacker
5721   is able to supply very large amounts of input data after a previous call to
5722   EVP_EncryptUpdate() with a partial block then a length check can overflow
5723   resulting in a heap corruption. Following an analysis of all OpenSSL
5724   internal usage of the EVP_EncryptUpdate() function all usage is one of two
5725   forms. The first form is where the EVP_EncryptUpdate() call is known to be
5726   the first called function after an EVP_EncryptInit(), and therefore that
5727   specific call must be safe. The second form is where the length passed to
5728   EVP_EncryptUpdate() can be seen from the code to be some small value and
5729   therefore there is no possibility of an overflow. Since all instances are
5730   one of these two forms, it is believed that there can be no overflows in
5731   internal code due to this problem. It should be noted that
5732   EVP_DecryptUpdate() can call EVP_EncryptUpdate() in certain code paths.
5733   Also EVP_CipherUpdate() is a synonym for EVP_EncryptUpdate(). All instances
5734   of these calls have also been analysed too and it is believed there are no
5735   instances in internal usage where an overflow could occur.
5736
5737   This issue was reported by Guido Vranken.
5738   ([CVE-2016-2106])
5739
5740   *Matt Caswell*
5741
5742 * Prevent ASN.1 BIO excessive memory allocation
5743
5744   When ASN.1 data is read from a BIO using functions such as d2i_CMS_bio()
5745   a short invalid encoding can cause allocation of large amounts of memory
5746   potentially consuming excessive resources or exhausting memory.
5747
5748   Any application parsing untrusted data through d2i BIO functions is
5749   affected. The memory based functions such as d2i_X509() are *not* affected.
5750   Since the memory based functions are used by the TLS library, TLS
5751   applications are not affected.
5752
5753   This issue was reported by Brian Carpenter.
5754   ([CVE-2016-2109])
5755
5756   *Stephen Henson*
5757
5758 * EBCDIC overread
5759
5760   ASN1 Strings that are over 1024 bytes can cause an overread in applications
5761   using the X509_NAME_oneline() function on EBCDIC systems. This could result
5762   in arbitrary stack data being returned in the buffer.
5763
5764   This issue was reported by Guido Vranken.
5765   ([CVE-2016-2176])
5766
5767   *Matt Caswell*
5768
5769 * Modify behavior of ALPN to invoke callback after SNI/servername
5770   callback, such that updates to the SSL_CTX affect ALPN.
5771
5772   *Todd Short*
5773
5774 * Remove LOW from the DEFAULT cipher list.  This removes singles DES from the
5775   default.
5776
5777   *Kurt Roeckx*
5778
5779 * Only remove the SSLv2 methods with the no-ssl2-method option. When the
5780   methods are enabled and ssl2 is disabled the methods return NULL.
5781
5782   *Kurt Roeckx*
5783
5784### Changes between 1.0.2f and 1.0.2g [1 Mar 2016]
5785
5786* Disable weak ciphers in SSLv3 and up in default builds of OpenSSL.
5787  Builds that are not configured with "enable-weak-ssl-ciphers" will not
5788  provide any "EXPORT" or "LOW" strength ciphers.
5789
5790  *Viktor Dukhovni*
5791
5792* Disable SSLv2 default build, default negotiation and weak ciphers.  SSLv2
5793  is by default disabled at build-time.  Builds that are not configured with
5794  "enable-ssl2" will not support SSLv2.  Even if "enable-ssl2" is used,
5795  users who want to negotiate SSLv2 via the version-flexible SSLv23_method()
5796  will need to explicitly call either of:
5797
5798      SSL_CTX_clear_options(ctx, SSL_OP_NO_SSLv2);
5799  or
5800      SSL_clear_options(ssl, SSL_OP_NO_SSLv2);
5801
5802  as appropriate.  Even if either of those is used, or the application
5803  explicitly uses the version-specific SSLv2_method() or its client and
5804  server variants, SSLv2 ciphers vulnerable to exhaustive search key
5805  recovery have been removed.  Specifically, the SSLv2 40-bit EXPORT
5806  ciphers, and SSLv2 56-bit DES are no longer available.
5807  ([CVE-2016-0800])
5808
5809   *Viktor Dukhovni*
5810
5811 * Fix a double-free in DSA code
5812
5813   A double free bug was discovered when OpenSSL parses malformed DSA private
5814   keys and could lead to a DoS attack or memory corruption for applications
5815   that receive DSA private keys from untrusted sources.  This scenario is
5816   considered rare.
5817
5818   This issue was reported to OpenSSL by Adam Langley(Google/BoringSSL) using
5819   libFuzzer.
5820   ([CVE-2016-0705])
5821
5822   *Stephen Henson*
5823
5824 * Disable SRP fake user seed to address a server memory leak.
5825
5826   Add a new method SRP_VBASE_get1_by_user that handles the seed properly.
5827
5828   SRP_VBASE_get_by_user had inconsistent memory management behaviour.
5829   In order to fix an unavoidable memory leak, SRP_VBASE_get_by_user
5830   was changed to ignore the "fake user" SRP seed, even if the seed
5831   is configured.
5832
5833   Users should use SRP_VBASE_get1_by_user instead. Note that in
5834   SRP_VBASE_get1_by_user, caller must free the returned value. Note
5835   also that even though configuring the SRP seed attempts to hide
5836   invalid usernames by continuing the handshake with fake
5837   credentials, this behaviour is not constant time and no strong
5838   guarantees are made that the handshake is indistinguishable from
5839   that of a valid user.
5840   ([CVE-2016-0798])
5841
5842   *Emilia Käsper*
5843
5844 * Fix BN_hex2bn/BN_dec2bn NULL pointer deref/heap corruption
5845
5846   In the BN_hex2bn function the number of hex digits is calculated using an
5847   int value `i`. Later `bn_expand` is called with a value of `i * 4`. For
5848   large values of `i` this can result in `bn_expand` not allocating any
5849   memory because `i * 4` is negative. This can leave the internal BIGNUM data
5850   field as NULL leading to a subsequent NULL ptr deref. For very large values
5851   of `i`, the calculation `i * 4` could be a positive value smaller than `i`.
5852   In this case memory is allocated to the internal BIGNUM data field, but it
5853   is insufficiently sized leading to heap corruption. A similar issue exists
5854   in BN_dec2bn. This could have security consequences if BN_hex2bn/BN_dec2bn
5855   is ever called by user applications with very large untrusted hex/dec data.
5856   This is anticipated to be a rare occurrence.
5857
5858   All OpenSSL internal usage of these functions use data that is not expected
5859   to be untrusted, e.g. config file data or application command line
5860   arguments. If user developed applications generate config file data based
5861   on untrusted data then it is possible that this could also lead to security
5862   consequences. This is also anticipated to be rare.
5863
5864   This issue was reported to OpenSSL by Guido Vranken.
5865   ([CVE-2016-0797])
5866
5867   *Matt Caswell*
5868
5869 * Fix memory issues in `BIO_*printf` functions
5870
5871   The internal `fmtstr` function used in processing a "%s" format string in
5872   the `BIO_*printf` functions could overflow while calculating the length of a
5873   string and cause an OOB read when printing very long strings.
5874
5875   Additionally the internal `doapr_outch` function can attempt to write to an
5876   OOB memory location (at an offset from the NULL pointer) in the event of a
5877   memory allocation failure. In 1.0.2 and below this could be caused where
5878   the size of a buffer to be allocated is greater than INT_MAX. E.g. this
5879   could be in processing a very long "%s" format string. Memory leaks can
5880   also occur.
5881
5882   The first issue may mask the second issue dependent on compiler behaviour.
5883   These problems could enable attacks where large amounts of untrusted data
5884   is passed to the `BIO_*printf` functions. If applications use these functions
5885   in this way then they could be vulnerable. OpenSSL itself uses these
5886   functions when printing out human-readable dumps of ASN.1 data. Therefore
5887   applications that print this data could be vulnerable if the data is from
5888   untrusted sources. OpenSSL command line applications could also be
5889   vulnerable where they print out ASN.1 data, or if untrusted data is passed
5890   as command line arguments.
5891
5892   Libssl is not considered directly vulnerable. Additionally certificates etc
5893   received via remote connections via libssl are also unlikely to be able to
5894   trigger these issues because of message size limits enforced within libssl.
5895
5896   This issue was reported to OpenSSL Guido Vranken.
5897   ([CVE-2016-0799])
5898
5899   *Matt Caswell*
5900
5901 * Side channel attack on modular exponentiation
5902
5903   A side-channel attack was found which makes use of cache-bank conflicts on
5904   the Intel Sandy-Bridge microarchitecture which could lead to the recovery
5905   of RSA keys.  The ability to exploit this issue is limited as it relies on
5906   an attacker who has control of code in a thread running on the same
5907   hyper-threaded core as the victim thread which is performing decryptions.
5908
5909   This issue was reported to OpenSSL by Yuval Yarom, The University of
5910   Adelaide and NICTA, Daniel Genkin, Technion and Tel Aviv University, and
5911   Nadia Heninger, University of Pennsylvania with more information at
5912   <http://cachebleed.info>.
5913   ([CVE-2016-0702])
5914
5915   *Andy Polyakov*
5916
5917 * Change the `req` command to generate a 2048-bit RSA/DSA key by default,
5918   if no keysize is specified with default_bits. This fixes an
5919   omission in an earlier change that changed all RSA/DSA key generation
5920   commands to use 2048 bits by default.
5921
5922   *Emilia Käsper*
5923
5924### Changes between 1.0.2e and 1.0.2f [28 Jan 2016]
5925
5926 * DH small subgroups
5927
5928   Historically OpenSSL only ever generated DH parameters based on "safe"
5929   primes. More recently (in version 1.0.2) support was provided for
5930   generating X9.42 style parameter files such as those required for RFC 5114
5931   support. The primes used in such files may not be "safe". Where an
5932   application is using DH configured with parameters based on primes that are
5933   not "safe" then an attacker could use this fact to find a peer's private
5934   DH exponent. This attack requires that the attacker complete multiple
5935   handshakes in which the peer uses the same private DH exponent. For example
5936   this could be used to discover a TLS server's private DH exponent if it's
5937   reusing the private DH exponent or it's using a static DH ciphersuite.
5938
5939   OpenSSL provides the option SSL_OP_SINGLE_DH_USE for ephemeral DH (DHE) in
5940   TLS. It is not on by default. If the option is not set then the server
5941   reuses the same private DH exponent for the life of the server process and
5942   would be vulnerable to this attack. It is believed that many popular
5943   applications do set this option and would therefore not be at risk.
5944
5945   The fix for this issue adds an additional check where a "q" parameter is
5946   available (as is the case in X9.42 based parameters). This detects the
5947   only known attack, and is the only possible defense for static DH
5948   ciphersuites. This could have some performance impact.
5949
5950   Additionally the SSL_OP_SINGLE_DH_USE option has been switched on by
5951   default and cannot be disabled. This could have some performance impact.
5952
5953   This issue was reported to OpenSSL by Antonio Sanso (Adobe).
5954   ([CVE-2016-0701])
5955
5956   *Matt Caswell*
5957
5958 * SSLv2 doesn't block disabled ciphers
5959
5960   A malicious client can negotiate SSLv2 ciphers that have been disabled on
5961   the server and complete SSLv2 handshakes even if all SSLv2 ciphers have
5962   been disabled, provided that the SSLv2 protocol was not also disabled via
5963   SSL_OP_NO_SSLv2.
5964
5965   This issue was reported to OpenSSL on 26th December 2015 by Nimrod Aviram
5966   and Sebastian Schinzel.
5967   ([CVE-2015-3197])
5968
5969   *Viktor Dukhovni*
5970
5971### Changes between 1.0.2d and 1.0.2e [3 Dec 2015]
5972
5973 * BN_mod_exp may produce incorrect results on x86_64
5974
5975   There is a carry propagating bug in the x86_64 Montgomery squaring
5976   procedure. No EC algorithms are affected. Analysis suggests that attacks
5977   against RSA and DSA as a result of this defect would be very difficult to
5978   perform and are not believed likely. Attacks against DH are considered just
5979   feasible (although very difficult) because most of the work necessary to
5980   deduce information about a private key may be performed offline. The amount
5981   of resources required for such an attack would be very significant and
5982   likely only accessible to a limited number of attackers. An attacker would
5983   additionally need online access to an unpatched system using the target
5984   private key in a scenario with persistent DH parameters and a private
5985   key that is shared between multiple clients. For example this can occur by
5986   default in OpenSSL DHE based SSL/TLS ciphersuites.
5987
5988   This issue was reported to OpenSSL by Hanno Böck.
5989   ([CVE-2015-3193])
5990
5991   *Andy Polyakov*
5992
5993 * Certificate verify crash with missing PSS parameter
5994
5995   The signature verification routines will crash with a NULL pointer
5996   dereference if presented with an ASN.1 signature using the RSA PSS
5997   algorithm and absent mask generation function parameter. Since these
5998   routines are used to verify certificate signature algorithms this can be
5999   used to crash any certificate verification operation and exploited in a
6000   DoS attack. Any application which performs certificate verification is
6001   vulnerable including OpenSSL clients and servers which enable client
6002   authentication.
6003
6004   This issue was reported to OpenSSL by Loïc Jonas Etienne (Qnective AG).
6005   ([CVE-2015-3194])
6006
6007   *Stephen Henson*
6008
6009 * X509_ATTRIBUTE memory leak
6010
6011   When presented with a malformed X509_ATTRIBUTE structure OpenSSL will leak
6012   memory. This structure is used by the PKCS#7 and CMS routines so any
6013   application which reads PKCS#7 or CMS data from untrusted sources is
6014   affected. SSL/TLS is not affected.
6015
6016   This issue was reported to OpenSSL by Adam Langley (Google/BoringSSL) using
6017   libFuzzer.
6018   ([CVE-2015-3195])
6019
6020   *Stephen Henson*
6021
6022 * Rewrite EVP_DecodeUpdate (base64 decoding) to fix several bugs.
6023   This changes the decoding behaviour for some invalid messages,
6024   though the change is mostly in the more lenient direction, and
6025   legacy behaviour is preserved as much as possible.
6026
6027   *Emilia Käsper*
6028
6029 * In DSA_generate_parameters_ex, if the provided seed is too short,
6030   return an error
6031
6032   *Rich Salz and Ismo Puustinen <ismo.puustinen@intel.com>*
6033
6034### Changes between 1.0.2c and 1.0.2d [9 Jul 2015]
6035
6036 * Alternate chains certificate forgery
6037
6038   During certificate verification, OpenSSL will attempt to find an
6039   alternative certificate chain if the first attempt to build such a chain
6040   fails. An error in the implementation of this logic can mean that an
6041   attacker could cause certain checks on untrusted certificates to be
6042   bypassed, such as the CA flag, enabling them to use a valid leaf
6043   certificate to act as a CA and "issue" an invalid certificate.
6044
6045   This issue was reported to OpenSSL by Adam Langley/David Benjamin
6046   (Google/BoringSSL).
6047
6048   *Matt Caswell*
6049
6050### Changes between 1.0.2b and 1.0.2c [12 Jun 2015]
6051
6052 * Fix HMAC ABI incompatibility. The previous version introduced an ABI
6053   incompatibility in the handling of HMAC. The previous ABI has now been
6054   restored.
6055
6056   *Matt Caswell*
6057
6058### Changes between 1.0.2a and 1.0.2b [11 Jun 2015]
6059
6060 * Malformed ECParameters causes infinite loop
6061
6062   When processing an ECParameters structure OpenSSL enters an infinite loop
6063   if the curve specified is over a specially malformed binary polynomial
6064   field.
6065
6066   This can be used to perform denial of service against any
6067   system which processes public keys, certificate requests or
6068   certificates.  This includes TLS clients and TLS servers with
6069   client authentication enabled.
6070
6071   This issue was reported to OpenSSL by Joseph Barr-Pixton.
6072   ([CVE-2015-1788])
6073
6074   *Andy Polyakov*
6075
6076 * Exploitable out-of-bounds read in X509_cmp_time
6077
6078   X509_cmp_time does not properly check the length of the ASN1_TIME
6079   string and can read a few bytes out of bounds. In addition,
6080   X509_cmp_time accepts an arbitrary number of fractional seconds in the
6081   time string.
6082
6083   An attacker can use this to craft malformed certificates and CRLs of
6084   various sizes and potentially cause a segmentation fault, resulting in
6085   a DoS on applications that verify certificates or CRLs. TLS clients
6086   that verify CRLs are affected. TLS clients and servers with client
6087   authentication enabled may be affected if they use custom verification
6088   callbacks.
6089
6090   This issue was reported to OpenSSL by Robert Swiecki (Google), and
6091   independently by Hanno Böck.
6092   ([CVE-2015-1789])
6093
6094   *Emilia Käsper*
6095
6096 * PKCS7 crash with missing EnvelopedContent
6097
6098   The PKCS#7 parsing code does not handle missing inner EncryptedContent
6099   correctly. An attacker can craft malformed ASN.1-encoded PKCS#7 blobs
6100   with missing content and trigger a NULL pointer dereference on parsing.
6101
6102   Applications that decrypt PKCS#7 data or otherwise parse PKCS#7
6103   structures from untrusted sources are affected. OpenSSL clients and
6104   servers are not affected.
6105
6106   This issue was reported to OpenSSL by Michal Zalewski (Google).
6107   ([CVE-2015-1790])
6108
6109   *Emilia Käsper*
6110
6111 * CMS verify infinite loop with unknown hash function
6112
6113   When verifying a signedData message the CMS code can enter an infinite loop
6114   if presented with an unknown hash function OID. This can be used to perform
6115   denial of service against any system which verifies signedData messages using
6116   the CMS code.
6117   This issue was reported to OpenSSL by Johannes Bauer.
6118   ([CVE-2015-1792])
6119
6120   *Stephen Henson*
6121
6122 * Race condition handling NewSessionTicket
6123
6124   If a NewSessionTicket is received by a multi-threaded client when attempting to
6125   reuse a previous ticket then a race condition can occur potentially leading to
6126   a double free of the ticket data.
6127   ([CVE-2015-1791])
6128
6129   *Matt Caswell*
6130
6131 * Only support 256-bit or stronger elliptic curves with the
6132   'ecdh_auto' setting (server) or by default (client). Of supported
6133   curves, prefer P-256 (both).
6134
6135   *Emilia Kasper*
6136
6137### Changes between 1.0.2 and 1.0.2a [19 Mar 2015]
6138
6139 * ClientHello sigalgs DoS fix
6140
6141   If a client connects to an OpenSSL 1.0.2 server and renegotiates with an
6142   invalid signature algorithms extension a NULL pointer dereference will
6143   occur. This can be exploited in a DoS attack against the server.
6144
6145   This issue was was reported to OpenSSL by David Ramos of Stanford
6146   University.
6147   ([CVE-2015-0291])
6148
6149   *Stephen Henson and Matt Caswell*
6150
6151 * Multiblock corrupted pointer fix
6152
6153   OpenSSL 1.0.2 introduced the "multiblock" performance improvement. This
6154   feature only applies on 64 bit x86 architecture platforms that support AES
6155   NI instructions. A defect in the implementation of "multiblock" can cause
6156   OpenSSL's internal write buffer to become incorrectly set to NULL when
6157   using non-blocking IO. Typically, when the user application is using a
6158   socket BIO for writing, this will only result in a failed connection.
6159   However if some other BIO is used then it is likely that a segmentation
6160   fault will be triggered, thus enabling a potential DoS attack.
6161
6162   This issue was reported to OpenSSL by Daniel Danner and Rainer Mueller.
6163   ([CVE-2015-0290])
6164
6165   *Matt Caswell*
6166
6167 * Segmentation fault in DTLSv1_listen fix
6168
6169   The DTLSv1_listen function is intended to be stateless and processes the
6170   initial ClientHello from many peers. It is common for user code to loop
6171   over the call to DTLSv1_listen until a valid ClientHello is received with
6172   an associated cookie. A defect in the implementation of DTLSv1_listen means
6173   that state is preserved in the SSL object from one invocation to the next
6174   that can lead to a segmentation fault. Errors processing the initial
6175   ClientHello can trigger this scenario. An example of such an error could be
6176   that a DTLS1.0 only client is attempting to connect to a DTLS1.2 only
6177   server.
6178
6179   This issue was reported to OpenSSL by Per Allansson.
6180   ([CVE-2015-0207])
6181
6182   *Matt Caswell*
6183
6184 * Segmentation fault in ASN1_TYPE_cmp fix
6185
6186   The function ASN1_TYPE_cmp will crash with an invalid read if an attempt is
6187   made to compare ASN.1 boolean types. Since ASN1_TYPE_cmp is used to check
6188   certificate signature algorithm consistency this can be used to crash any
6189   certificate verification operation and exploited in a DoS attack. Any
6190   application which performs certificate verification is vulnerable including
6191   OpenSSL clients and servers which enable client authentication.
6192   ([CVE-2015-0286])
6193
6194   *Stephen Henson*
6195
6196 * Segmentation fault for invalid PSS parameters fix
6197
6198   The signature verification routines will crash with a NULL pointer
6199   dereference if presented with an ASN.1 signature using the RSA PSS
6200   algorithm and invalid parameters. Since these routines are used to verify
6201   certificate signature algorithms this can be used to crash any
6202   certificate verification operation and exploited in a DoS attack. Any
6203   application which performs certificate verification is vulnerable including
6204   OpenSSL clients and servers which enable client authentication.
6205
6206   This issue was was reported to OpenSSL by Brian Carpenter.
6207   ([CVE-2015-0208])
6208
6209   *Stephen Henson*
6210
6211 * ASN.1 structure reuse memory corruption fix
6212
6213   Reusing a structure in ASN.1 parsing may allow an attacker to cause
6214   memory corruption via an invalid write. Such reuse is and has been
6215   strongly discouraged and is believed to be rare.
6216
6217   Applications that parse structures containing CHOICE or ANY DEFINED BY
6218   components may be affected. Certificate parsing (d2i_X509 and related
6219   functions) are however not affected. OpenSSL clients and servers are
6220   not affected.
6221   ([CVE-2015-0287])
6222
6223   *Stephen Henson*
6224
6225 * PKCS7 NULL pointer dereferences fix
6226
6227   The PKCS#7 parsing code does not handle missing outer ContentInfo
6228   correctly. An attacker can craft malformed ASN.1-encoded PKCS#7 blobs with
6229   missing content and trigger a NULL pointer dereference on parsing.
6230
6231   Applications that verify PKCS#7 signatures, decrypt PKCS#7 data or
6232   otherwise parse PKCS#7 structures from untrusted sources are
6233   affected. OpenSSL clients and servers are not affected.
6234
6235   This issue was reported to OpenSSL by Michal Zalewski (Google).
6236   ([CVE-2015-0289])
6237
6238   *Emilia Käsper*
6239
6240 * DoS via reachable assert in SSLv2 servers fix
6241
6242   A malicious client can trigger an OPENSSL_assert (i.e., an abort) in
6243   servers that both support SSLv2 and enable export cipher suites by sending
6244   a specially crafted SSLv2 CLIENT-MASTER-KEY message.
6245
6246   This issue was discovered by Sean Burford (Google) and Emilia Käsper
6247   (OpenSSL development team).
6248   ([CVE-2015-0293])
6249
6250   *Emilia Käsper*
6251
6252 * Empty CKE with client auth and DHE fix
6253
6254   If client auth is used then a server can seg fault in the event of a DHE
6255   ciphersuite being selected and a zero length ClientKeyExchange message
6256   being sent by the client. This could be exploited in a DoS attack.
6257   ([CVE-2015-1787])
6258
6259   *Matt Caswell*
6260
6261 * Handshake with unseeded PRNG fix
6262
6263   Under certain conditions an OpenSSL 1.0.2 client can complete a handshake
6264   with an unseeded PRNG. The conditions are:
6265   - The client is on a platform where the PRNG has not been seeded
6266   automatically, and the user has not seeded manually
6267   - A protocol specific client method version has been used (i.e. not
6268   SSL_client_methodv23)
6269   - A ciphersuite is used that does not require additional random data from
6270   the PRNG beyond the initial ClientHello client random (e.g. PSK-RC4-SHA).
6271
6272   If the handshake succeeds then the client random that has been used will
6273   have been generated from a PRNG with insufficient entropy and therefore the
6274   output may be predictable.
6275
6276   For example using the following command with an unseeded openssl will
6277   succeed on an unpatched platform:
6278
6279   openssl s_client -psk 1a2b3c4d -tls1_2 -cipher PSK-RC4-SHA
6280   ([CVE-2015-0285])
6281
6282   *Matt Caswell*
6283
6284 * Use After Free following d2i_ECPrivatekey error fix
6285
6286   A malformed EC private key file consumed via the d2i_ECPrivateKey function
6287   could cause a use after free condition. This, in turn, could cause a double
6288   free in several private key parsing functions (such as d2i_PrivateKey
6289   or EVP_PKCS82PKEY) and could lead to a DoS attack or memory corruption
6290   for applications that receive EC private keys from untrusted
6291   sources. This scenario is considered rare.
6292
6293   This issue was discovered by the BoringSSL project and fixed in their
6294   commit 517073cd4b.
6295   ([CVE-2015-0209])
6296
6297   *Matt Caswell*
6298
6299 * X509_to_X509_REQ NULL pointer deref fix
6300
6301   The function X509_to_X509_REQ will crash with a NULL pointer dereference if
6302   the certificate key is invalid. This function is rarely used in practice.
6303
6304   This issue was discovered by Brian Carpenter.
6305   ([CVE-2015-0288])
6306
6307   *Stephen Henson*
6308
6309 * Removed the export ciphers from the DEFAULT ciphers
6310
6311   *Kurt Roeckx*
6312
6313### Changes between 1.0.1l and 1.0.2 [22 Jan 2015]
6314
6315 * Facilitate "universal" ARM builds targeting range of ARM ISAs, e.g.
6316   ARMv5 through ARMv8, as opposite to "locking" it to single one.
6317   So far those who have to target multiple platforms would compromise
6318   and argue that binary targeting say ARMv5 would still execute on
6319   ARMv8. "Universal" build resolves this compromise by providing
6320   near-optimal performance even on newer platforms.
6321
6322   *Andy Polyakov*
6323
6324 * Accelerated NIST P-256 elliptic curve implementation for x86_64
6325   (other platforms pending).
6326
6327   *Shay Gueron & Vlad Krasnov (Intel Corp), Andy Polyakov*
6328
6329 * Add support for the SignedCertificateTimestampList certificate and
6330   OCSP response extensions from RFC6962.
6331
6332   *Rob Stradling*
6333
6334 * Fix ec_GFp_simple_points_make_affine (thus, EC_POINTs_mul etc.)
6335   for corner cases. (Certain input points at infinity could lead to
6336   bogus results, with non-infinity inputs mapped to infinity too.)
6337
6338   *Bodo Moeller*
6339
6340 * Initial support for PowerISA 2.0.7, first implemented in POWER8.
6341   This covers AES, SHA256/512 and GHASH. "Initial" means that most
6342   common cases are optimized and there still is room for further
6343   improvements. Vector Permutation AES for Altivec is also added.
6344
6345   *Andy Polyakov*
6346
6347 * Add support for little-endian ppc64 Linux target.
6348
6349   *Marcelo Cerri (IBM)*
6350
6351 * Initial support for AMRv8 ISA crypto extensions. This covers AES,
6352   SHA1, SHA256 and GHASH. "Initial" means that most common cases
6353   are optimized and there still is room for further improvements.
6354   Both 32- and 64-bit modes are supported.
6355
6356   *Andy Polyakov, Ard Biesheuvel (Linaro)*
6357
6358 * Improved ARMv7 NEON support.
6359
6360   *Andy Polyakov*
6361
6362 * Support for SPARC Architecture 2011 crypto extensions, first
6363   implemented in SPARC T4. This covers AES, DES, Camellia, SHA1,
6364   SHA256/512, MD5, GHASH and modular exponentiation.
6365
6366   *Andy Polyakov, David Miller*
6367
6368 * Accelerated modular exponentiation for Intel processors, a.k.a.
6369   RSAZ.
6370
6371   *Shay Gueron & Vlad Krasnov (Intel Corp)*
6372
6373 * Support for new and upcoming Intel processors, including AVX2,
6374   BMI and SHA ISA extensions. This includes additional "stitched"
6375   implementations, AESNI-SHA256 and GCM, and multi-buffer support
6376   for TLS encrypt.
6377
6378   This work was sponsored by Intel Corp.
6379
6380   *Andy Polyakov*
6381
6382 * Support for DTLS 1.2. This adds two sets of DTLS methods: DTLS_*_method()
6383   supports both DTLS 1.2 and 1.0 and should use whatever version the peer
6384   supports and DTLSv1_2_*_method() which supports DTLS 1.2 only.
6385
6386   *Steve Henson*
6387
6388 * Use algorithm specific chains in SSL_CTX_use_certificate_chain_file():
6389   this fixes a limitation in previous versions of OpenSSL.
6390
6391   *Steve Henson*
6392
6393 * Extended RSA OAEP support via EVP_PKEY API. Options to specify digest,
6394   MGF1 digest and OAEP label.
6395
6396   *Steve Henson*
6397
6398 * Add EVP support for key wrapping algorithms, to avoid problems with
6399   existing code the flag EVP_CIPHER_CTX_WRAP_ALLOW has to be set in
6400   the EVP_CIPHER_CTX or an error is returned. Add AES and DES3 wrap
6401   algorithms and include tests cases.
6402
6403   *Steve Henson*
6404
6405 * Add functions to allocate and set the fields of an ECDSA_METHOD
6406   structure.
6407
6408   *Douglas E. Engert, Steve Henson*
6409
6410 * New functions OPENSSL_gmtime_diff and ASN1_TIME_diff to find the
6411   difference in days and seconds between two tm or ASN1_TIME structures.
6412
6413   *Steve Henson*
6414
6415 * Add -rev test option to s_server to just reverse order of characters
6416   received by client and send back to server. Also prints an abbreviated
6417   summary of the connection parameters.
6418
6419   *Steve Henson*
6420
6421 * New option -brief for s_client and s_server to print out a brief summary
6422   of connection parameters.
6423
6424   *Steve Henson*
6425
6426 * Add callbacks for arbitrary TLS extensions.
6427
6428   *Trevor Perrin <trevp@trevp.net> and Ben Laurie*
6429
6430 * New option -crl_download in several openssl utilities to download CRLs
6431   from CRLDP extension in certificates.
6432
6433   *Steve Henson*
6434
6435 * New options -CRL and -CRLform for s_client and s_server for CRLs.
6436
6437   *Steve Henson*
6438
6439 * New function X509_CRL_diff to generate a delta CRL from the difference
6440   of two full CRLs. Add support to "crl" utility.
6441
6442   *Steve Henson*
6443
6444 * New functions to set lookup_crls function and to retrieve
6445   X509_STORE from X509_STORE_CTX.
6446
6447   *Steve Henson*
6448
6449 * Print out deprecated issuer and subject unique ID fields in
6450   certificates.
6451
6452   *Steve Henson*
6453
6454 * Extend OCSP I/O functions so they can be used for simple general purpose
6455   HTTP as well as OCSP. New wrapper function which can be used to download
6456   CRLs using the OCSP API.
6457
6458   *Steve Henson*
6459
6460 * Delegate command line handling in s_client/s_server to SSL_CONF APIs.
6461
6462   *Steve Henson*
6463
6464 * `SSL_CONF*` functions. These provide a common framework for application
6465   configuration using configuration files or command lines.
6466
6467   *Steve Henson*
6468
6469 * SSL/TLS tracing code. This parses out SSL/TLS records using the
6470   message callback and prints the results. Needs compile time option
6471   "enable-ssl-trace". New options to s_client and s_server to enable
6472   tracing.
6473
6474   *Steve Henson*
6475
6476 * New ctrl and macro to retrieve supported points extensions.
6477   Print out extension in s_server and s_client.
6478
6479   *Steve Henson*
6480
6481 * New functions to retrieve certificate signature and signature
6482   OID NID.
6483
6484   *Steve Henson*
6485
6486 * Add functions to retrieve and manipulate the raw cipherlist sent by a
6487   client to OpenSSL.
6488
6489   *Steve Henson*
6490
6491 * New Suite B modes for TLS code. These use and enforce the requirements
6492   of RFC6460: restrict ciphersuites, only permit Suite B algorithms and
6493   only use Suite B curves. The Suite B modes can be set by using the
6494   strings "SUITEB128", "SUITEB192" or "SUITEB128ONLY" for the cipherstring.
6495
6496   *Steve Henson*
6497
6498 * New chain verification flags for Suite B levels of security. Check
6499   algorithms are acceptable when flags are set in X509_verify_cert.
6500
6501   *Steve Henson*
6502
6503 * Make tls1_check_chain return a set of flags indicating checks passed
6504   by a certificate chain. Add additional tests to handle client
6505   certificates: checks for matching certificate type and issuer name
6506   comparison.
6507
6508   *Steve Henson*
6509
6510 * If an attempt is made to use a signature algorithm not in the peer
6511   preference list abort the handshake. If client has no suitable
6512   signature algorithms in response to a certificate request do not
6513   use the certificate.
6514
6515   *Steve Henson*
6516
6517 * If server EC tmp key is not in client preference list abort handshake.
6518
6519   *Steve Henson*
6520
6521 * Add support for certificate stores in CERT structure. This makes it
6522   possible to have different stores per SSL structure or one store in
6523   the parent SSL_CTX. Include distinct stores for certificate chain
6524   verification and chain building. New ctrl SSL_CTRL_BUILD_CERT_CHAIN
6525   to build and store a certificate chain in CERT structure: returning
6526   an error if the chain cannot be built: this will allow applications
6527   to test if a chain is correctly configured.
6528
6529   Note: if the CERT based stores are not set then the parent SSL_CTX
6530   store is used to retain compatibility with existing behaviour.
6531
6532   *Steve Henson*
6533
6534 * New function ssl_set_client_disabled to set a ciphersuite disabled
6535   mask based on the current session, check mask when sending client
6536   hello and checking the requested ciphersuite.
6537
6538   *Steve Henson*
6539
6540 * New ctrls to retrieve and set certificate types in a certificate
6541   request message. Print out received values in s_client. If certificate
6542   types is not set with custom values set sensible values based on
6543   supported signature algorithms.
6544
6545   *Steve Henson*
6546
6547 * Support for distinct client and server supported signature algorithms.
6548
6549   *Steve Henson*
6550
6551 * Add certificate callback. If set this is called whenever a certificate
6552   is required by client or server. An application can decide which
6553   certificate chain to present based on arbitrary criteria: for example
6554   supported signature algorithms. Add very simple example to s_server.
6555   This fixes many of the problems and restrictions of the existing client
6556   certificate callback: for example you can now clear an existing
6557   certificate and specify the whole chain.
6558
6559   *Steve Henson*
6560
6561 * Add new "valid_flags" field to CERT_PKEY structure which determines what
6562   the certificate can be used for (if anything). Set valid_flags field
6563   in new tls1_check_chain function. Simplify ssl_set_cert_masks which used
6564   to have similar checks in it.
6565
6566   Add new "cert_flags" field to CERT structure and include a "strict mode".
6567   This enforces some TLS certificate requirements (such as only permitting
6568   certificate signature algorithms contained in the supported algorithms
6569   extension) which some implementations ignore: this option should be used
6570   with caution as it could cause interoperability issues.
6571
6572   *Steve Henson*
6573
6574 * Update and tidy signature algorithm extension processing. Work out
6575   shared signature algorithms based on preferences and peer algorithms
6576   and print them out in s_client and s_server. Abort handshake if no
6577   shared signature algorithms.
6578
6579   *Steve Henson*
6580
6581 * Add new functions to allow customised supported signature algorithms
6582   for SSL and SSL_CTX structures. Add options to s_client and s_server
6583   to support them.
6584
6585   *Steve Henson*
6586
6587 * New function SSL_certs_clear() to delete all references to certificates
6588   from an SSL structure. Before this once a certificate had been added
6589   it couldn't be removed.
6590
6591   *Steve Henson*
6592
6593 * Integrate hostname, email address and IP address checking with certificate
6594   verification. New verify options supporting checking in openssl utility.
6595
6596   *Steve Henson*
6597
6598 * Fixes and wildcard matching support to hostname and email checking
6599   functions. Add manual page.
6600
6601   *Florian Weimer (Red Hat Product Security Team)*
6602
6603 * New functions to check a hostname email or IP address against a
6604   certificate. Add options x509 utility to print results of checks against
6605   a certificate.
6606
6607   *Steve Henson*
6608
6609 * Fix OCSP checking.
6610
6611   *Rob Stradling <rob.stradling@comodo.com> and Ben Laurie*
6612
6613 * Initial experimental support for explicitly trusted non-root CAs.
6614   OpenSSL still tries to build a complete chain to a root but if an
6615   intermediate CA has a trust setting included that is used. The first
6616   setting is used: whether to trust (e.g., -addtrust option to the x509
6617   utility) or reject.
6618
6619   *Steve Henson*
6620
6621 * Add -trusted_first option which attempts to find certificates in the
6622   trusted store even if an untrusted chain is also supplied.
6623
6624   *Steve Henson*
6625
6626 * MIPS assembly pack updates: support for MIPS32r2 and SmartMIPS ASE,
6627   platform support for Linux and Android.
6628
6629   *Andy Polyakov*
6630
6631 * Support for linux-x32, ILP32 environment in x86_64 framework.
6632
6633   *Andy Polyakov*
6634
6635 * Experimental multi-implementation support for FIPS capable OpenSSL.
6636   When in FIPS mode the approved implementations are used as normal,
6637   when not in FIPS mode the internal unapproved versions are used instead.
6638   This means that the FIPS capable OpenSSL isn't forced to use the
6639   (often lower performance) FIPS implementations outside FIPS mode.
6640
6641   *Steve Henson*
6642
6643 * Transparently support X9.42 DH parameters when calling
6644   PEM_read_bio_DHparameters. This means existing applications can handle
6645   the new parameter format automatically.
6646
6647   *Steve Henson*
6648
6649 * Initial experimental support for X9.42 DH parameter format: mainly
6650   to support use of 'q' parameter for RFC5114 parameters.
6651
6652   *Steve Henson*
6653
6654 * Add DH parameters from RFC5114 including test data to dhtest.
6655
6656   *Steve Henson*
6657
6658 * Support for automatic EC temporary key parameter selection. If enabled
6659   the most preferred EC parameters are automatically used instead of
6660   hardcoded fixed parameters. Now a server just has to call:
6661   SSL_CTX_set_ecdh_auto(ctx, 1) and the server will automatically
6662   support ECDH and use the most appropriate parameters.
6663
6664   *Steve Henson*
6665
6666 * Enhance and tidy EC curve and point format TLS extension code. Use
6667   static structures instead of allocation if default values are used.
6668   New ctrls to set curves we wish to support and to retrieve shared curves.
6669   Print out shared curves in s_server. New options to s_server and s_client
6670   to set list of supported curves.
6671
6672   *Steve Henson*
6673
6674 * New ctrls to retrieve supported signature algorithms and
6675   supported curve values as an array of NIDs. Extend openssl utility
6676   to print out received values.
6677
6678   *Steve Henson*
6679
6680 * Add new APIs EC_curve_nist2nid and EC_curve_nid2nist which convert
6681   between NIDs and the more common NIST names such as "P-256". Enhance
6682   ecparam utility and ECC method to recognise the NIST names for curves.
6683
6684   *Steve Henson*
6685
6686 * Enhance SSL/TLS certificate chain handling to support different
6687   chains for each certificate instead of one chain in the parent SSL_CTX.
6688
6689   *Steve Henson*
6690
6691 * Support for fixed DH ciphersuite client authentication: where both
6692   server and client use DH certificates with common parameters.
6693
6694   *Steve Henson*
6695
6696 * Support for fixed DH ciphersuites: those requiring DH server
6697   certificates.
6698
6699   *Steve Henson*
6700
6701 * New function i2d_re_X509_tbs for re-encoding the TBS portion of
6702   the certificate.
6703   Note: Related 1.0.2-beta specific macros X509_get_cert_info,
6704   X509_CINF_set_modified, X509_CINF_get_issuer, X509_CINF_get_extensions and
6705   X509_CINF_get_signature were reverted post internal team review.
6706
6707OpenSSL 1.0.1
6708-------------
6709
6710### Changes between 1.0.1t and 1.0.1u [22 Sep 2016]
6711
6712 * OCSP Status Request extension unbounded memory growth
6713
6714   A malicious client can send an excessively large OCSP Status Request
6715   extension. If that client continually requests renegotiation, sending a
6716   large OCSP Status Request extension each time, then there will be unbounded
6717   memory growth on the server. This will eventually lead to a Denial Of
6718   Service attack through memory exhaustion. Servers with a default
6719   configuration are vulnerable even if they do not support OCSP. Builds using
6720   the "no-ocsp" build time option are not affected.
6721
6722   This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.)
6723   ([CVE-2016-6304])
6724
6725   *Matt Caswell*
6726
6727 * In order to mitigate the SWEET32 attack, the DES ciphers were moved from
6728   HIGH to MEDIUM.
6729
6730   This issue was reported to OpenSSL Karthikeyan Bhargavan and Gaetan
6731   Leurent (INRIA)
6732   ([CVE-2016-2183])
6733
6734   *Rich Salz*
6735
6736 * OOB write in MDC2_Update()
6737
6738   An overflow can occur in MDC2_Update() either if called directly or
6739   through the EVP_DigestUpdate() function using MDC2. If an attacker
6740   is able to supply very large amounts of input data after a previous
6741   call to EVP_EncryptUpdate() with a partial block then a length check
6742   can overflow resulting in a heap corruption.
6743
6744   The amount of data needed is comparable to SIZE_MAX which is impractical
6745   on most platforms.
6746
6747   This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.)
6748   ([CVE-2016-6303])
6749
6750   *Stephen Henson*
6751
6752 * Malformed SHA512 ticket DoS
6753
6754   If a server uses SHA512 for TLS session ticket HMAC it is vulnerable to a
6755   DoS attack where a malformed ticket will result in an OOB read which will
6756   ultimately crash.
6757
6758   The use of SHA512 in TLS session tickets is comparatively rare as it requires
6759   a custom server callback and ticket lookup mechanism.
6760
6761   This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.)
6762   ([CVE-2016-6302])
6763
6764   *Stephen Henson*
6765
6766 * OOB write in BN_bn2dec()
6767
6768   The function BN_bn2dec() does not check the return value of BN_div_word().
6769   This can cause an OOB write if an application uses this function with an
6770   overly large BIGNUM. This could be a problem if an overly large certificate
6771   or CRL is printed out from an untrusted source. TLS is not affected because
6772   record limits will reject an oversized certificate before it is parsed.
6773
6774   This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.)
6775   ([CVE-2016-2182])
6776
6777   *Stephen Henson*
6778
6779 * OOB read in TS_OBJ_print_bio()
6780
6781   The function TS_OBJ_print_bio() misuses OBJ_obj2txt(): the return value is
6782   the total length the OID text representation would use and not the amount
6783   of data written. This will result in OOB reads when large OIDs are
6784   presented.
6785
6786   This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.)
6787   ([CVE-2016-2180])
6788
6789   *Stephen Henson*
6790
6791 * Pointer arithmetic undefined behaviour
6792
6793   Avoid some undefined pointer arithmetic
6794
6795   A common idiom in the codebase is to check limits in the following manner:
6796   "p + len > limit"
6797
6798   Where "p" points to some malloc'd data of SIZE bytes and
6799   limit == p + SIZE
6800
6801   "len" here could be from some externally supplied data (e.g. from a TLS
6802   message).
6803
6804   The rules of C pointer arithmetic are such that "p + len" is only well
6805   defined where len <= SIZE. Therefore the above idiom is actually
6806   undefined behaviour.
6807
6808   For example this could cause problems if some malloc implementation
6809   provides an address for "p" such that "p + len" actually overflows for
6810   values of len that are too big and therefore p + len < limit.
6811
6812   This issue was reported to OpenSSL by Guido Vranken
6813   ([CVE-2016-2177])
6814
6815   *Matt Caswell*
6816
6817 * Constant time flag not preserved in DSA signing
6818
6819   Operations in the DSA signing algorithm should run in constant time in
6820   order to avoid side channel attacks. A flaw in the OpenSSL DSA
6821   implementation means that a non-constant time codepath is followed for
6822   certain operations. This has been demonstrated through a cache-timing
6823   attack to be sufficient for an attacker to recover the private DSA key.
6824
6825   This issue was reported by César Pereida (Aalto University), Billy Brumley
6826   (Tampere University of Technology), and Yuval Yarom (The University of
6827   Adelaide and NICTA).
6828   ([CVE-2016-2178])
6829
6830   *César Pereida*
6831
6832 * DTLS buffered message DoS
6833
6834   In a DTLS connection where handshake messages are delivered out-of-order
6835   those messages that OpenSSL is not yet ready to process will be buffered
6836   for later use. Under certain circumstances, a flaw in the logic means that
6837   those messages do not get removed from the buffer even though the handshake
6838   has been completed. An attacker could force up to approx. 15 messages to
6839   remain in the buffer when they are no longer required. These messages will
6840   be cleared when the DTLS connection is closed. The default maximum size for
6841   a message is 100k. Therefore the attacker could force an additional 1500k
6842   to be consumed per connection. By opening many simulataneous connections an
6843   attacker could cause a DoS attack through memory exhaustion.
6844
6845   This issue was reported to OpenSSL by Quan Luo.
6846   ([CVE-2016-2179])
6847
6848   *Matt Caswell*
6849
6850 * DTLS replay protection DoS
6851
6852   A flaw in the DTLS replay attack protection mechanism means that records
6853   that arrive for future epochs update the replay protection "window" before
6854   the MAC for the record has been validated. This could be exploited by an
6855   attacker by sending a record for the next epoch (which does not have to
6856   decrypt or have a valid MAC), with a very large sequence number. This means
6857   that all subsequent legitimate packets are dropped causing a denial of
6858   service for a specific DTLS connection.
6859
6860   This issue was reported to OpenSSL by the OCAP audit team.
6861   ([CVE-2016-2181])
6862
6863   *Matt Caswell*
6864
6865 * Certificate message OOB reads
6866
6867   In OpenSSL 1.0.2 and earlier some missing message length checks can result
6868   in OOB reads of up to 2 bytes beyond an allocated buffer. There is a
6869   theoretical DoS risk but this has not been observed in practice on common
6870   platforms.
6871
6872   The messages affected are client certificate, client certificate request
6873   and server certificate. As a result the attack can only be performed
6874   against a client or a server which enables client authentication.
6875
6876   This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.)
6877   ([CVE-2016-6306])
6878
6879   *Stephen Henson*
6880
6881### Changes between 1.0.1s and 1.0.1t [3 May 2016]
6882
6883 * Prevent padding oracle in AES-NI CBC MAC check
6884
6885   A MITM attacker can use a padding oracle attack to decrypt traffic
6886   when the connection uses an AES CBC cipher and the server support
6887   AES-NI.
6888
6889   This issue was introduced as part of the fix for Lucky 13 padding
6890   attack ([CVE-2013-0169]). The padding check was rewritten to be in
6891   constant time by making sure that always the same bytes are read and
6892   compared against either the MAC or padding bytes. But it no longer
6893   checked that there was enough data to have both the MAC and padding
6894   bytes.
6895
6896   This issue was reported by Juraj Somorovsky using TLS-Attacker.
6897   ([CVE-2016-2107])
6898
6899   *Kurt Roeckx*
6900
6901 * Fix EVP_EncodeUpdate overflow
6902
6903   An overflow can occur in the EVP_EncodeUpdate() function which is used for
6904   Base64 encoding of binary data. If an attacker is able to supply very large
6905   amounts of input data then a length check can overflow resulting in a heap
6906   corruption.
6907
6908   Internally to OpenSSL the EVP_EncodeUpdate() function is primarly used by
6909   the `PEM_write_bio*` family of functions. These are mainly used within the
6910   OpenSSL command line applications, so any application which processes data
6911   from an untrusted source and outputs it as a PEM file should be considered
6912   vulnerable to this issue. User applications that call these APIs directly
6913   with large amounts of untrusted data may also be vulnerable.
6914
6915   This issue was reported by Guido Vranken.
6916   ([CVE-2016-2105])
6917
6918   *Matt Caswell*
6919
6920 * Fix EVP_EncryptUpdate overflow
6921
6922   An overflow can occur in the EVP_EncryptUpdate() function. If an attacker
6923   is able to supply very large amounts of input data after a previous call to
6924   EVP_EncryptUpdate() with a partial block then a length check can overflow
6925   resulting in a heap corruption. Following an analysis of all OpenSSL
6926   internal usage of the EVP_EncryptUpdate() function all usage is one of two
6927   forms. The first form is where the EVP_EncryptUpdate() call is known to be
6928   the first called function after an EVP_EncryptInit(), and therefore that
6929   specific call must be safe. The second form is where the length passed to
6930   EVP_EncryptUpdate() can be seen from the code to be some small value and
6931   therefore there is no possibility of an overflow. Since all instances are
6932   one of these two forms, it is believed that there can be no overflows in
6933   internal code due to this problem. It should be noted that
6934   EVP_DecryptUpdate() can call EVP_EncryptUpdate() in certain code paths.
6935   Also EVP_CipherUpdate() is a synonym for EVP_EncryptUpdate(). All instances
6936   of these calls have also been analysed too and it is believed there are no
6937   instances in internal usage where an overflow could occur.
6938
6939   This issue was reported by Guido Vranken.
6940   ([CVE-2016-2106])
6941
6942   *Matt Caswell*
6943
6944 * Prevent ASN.1 BIO excessive memory allocation
6945
6946   When ASN.1 data is read from a BIO using functions such as d2i_CMS_bio()
6947   a short invalid encoding can casuse allocation of large amounts of memory
6948   potentially consuming excessive resources or exhausting memory.
6949
6950   Any application parsing untrusted data through d2i BIO functions is
6951   affected. The memory based functions such as d2i_X509() are *not* affected.
6952   Since the memory based functions are used by the TLS library, TLS
6953   applications are not affected.
6954
6955   This issue was reported by Brian Carpenter.
6956   ([CVE-2016-2109])
6957
6958   *Stephen Henson*
6959
6960 * EBCDIC overread
6961
6962   ASN1 Strings that are over 1024 bytes can cause an overread in applications
6963   using the X509_NAME_oneline() function on EBCDIC systems. This could result
6964   in arbitrary stack data being returned in the buffer.
6965
6966   This issue was reported by Guido Vranken.
6967   ([CVE-2016-2176])
6968
6969   *Matt Caswell*
6970
6971 * Modify behavior of ALPN to invoke callback after SNI/servername
6972   callback, such that updates to the SSL_CTX affect ALPN.
6973
6974   *Todd Short*
6975
6976 * Remove LOW from the DEFAULT cipher list.  This removes singles DES from the
6977   default.
6978
6979   *Kurt Roeckx*
6980
6981 * Only remove the SSLv2 methods with the no-ssl2-method option. When the
6982   methods are enabled and ssl2 is disabled the methods return NULL.
6983
6984   *Kurt Roeckx*
6985
6986### Changes between 1.0.1r and 1.0.1s [1 Mar 2016]
6987
6988* Disable weak ciphers in SSLv3 and up in default builds of OpenSSL.
6989  Builds that are not configured with "enable-weak-ssl-ciphers" will not
6990  provide any "EXPORT" or "LOW" strength ciphers.
6991
6992  *Viktor Dukhovni*
6993
6994* Disable SSLv2 default build, default negotiation and weak ciphers.  SSLv2
6995  is by default disabled at build-time.  Builds that are not configured with
6996  "enable-ssl2" will not support SSLv2.  Even if "enable-ssl2" is used,
6997  users who want to negotiate SSLv2 via the version-flexible SSLv23_method()
6998  will need to explicitly call either of:
6999
7000      SSL_CTX_clear_options(ctx, SSL_OP_NO_SSLv2);
7001  or
7002      SSL_clear_options(ssl, SSL_OP_NO_SSLv2);
7003
7004  as appropriate.  Even if either of those is used, or the application
7005  explicitly uses the version-specific SSLv2_method() or its client and
7006  server variants, SSLv2 ciphers vulnerable to exhaustive search key
7007  recovery have been removed.  Specifically, the SSLv2 40-bit EXPORT
7008  ciphers, and SSLv2 56-bit DES are no longer available.
7009  ([CVE-2016-0800])
7010
7011  *Viktor Dukhovni*
7012
7013 * Fix a double-free in DSA code
7014
7015   A double free bug was discovered when OpenSSL parses malformed DSA private
7016   keys and could lead to a DoS attack or memory corruption for applications
7017   that receive DSA private keys from untrusted sources.  This scenario is
7018   considered rare.
7019
7020   This issue was reported to OpenSSL by Adam Langley(Google/BoringSSL) using
7021   libFuzzer.
7022   ([CVE-2016-0705])
7023
7024   *Stephen Henson*
7025
7026 * Disable SRP fake user seed to address a server memory leak.
7027
7028   Add a new method SRP_VBASE_get1_by_user that handles the seed properly.
7029
7030   SRP_VBASE_get_by_user had inconsistent memory management behaviour.
7031   In order to fix an unavoidable memory leak, SRP_VBASE_get_by_user
7032   was changed to ignore the "fake user" SRP seed, even if the seed
7033   is configured.
7034
7035   Users should use SRP_VBASE_get1_by_user instead. Note that in
7036   SRP_VBASE_get1_by_user, caller must free the returned value. Note
7037   also that even though configuring the SRP seed attempts to hide
7038   invalid usernames by continuing the handshake with fake
7039   credentials, this behaviour is not constant time and no strong
7040   guarantees are made that the handshake is indistinguishable from
7041   that of a valid user.
7042   ([CVE-2016-0798])
7043
7044   *Emilia Käsper*
7045
7046 * Fix BN_hex2bn/BN_dec2bn NULL pointer deref/heap corruption
7047
7048   In the BN_hex2bn function the number of hex digits is calculated using an
7049   int value `i`. Later `bn_expand` is called with a value of `i * 4`. For
7050   large values of `i` this can result in `bn_expand` not allocating any
7051   memory because `i * 4` is negative. This can leave the internal BIGNUM data
7052   field as NULL leading to a subsequent NULL ptr deref. For very large values
7053   of `i`, the calculation `i * 4` could be a positive value smaller than `i`.
7054   In this case memory is allocated to the internal BIGNUM data field, but it
7055   is insufficiently sized leading to heap corruption. A similar issue exists
7056   in BN_dec2bn. This could have security consequences if BN_hex2bn/BN_dec2bn
7057   is ever called by user applications with very large untrusted hex/dec data.
7058   This is anticipated to be a rare occurrence.
7059
7060   All OpenSSL internal usage of these functions use data that is not expected
7061   to be untrusted, e.g. config file data or application command line
7062   arguments. If user developed applications generate config file data based
7063   on untrusted data then it is possible that this could also lead to security
7064   consequences. This is also anticipated to be rare.
7065
7066   This issue was reported to OpenSSL by Guido Vranken.
7067   ([CVE-2016-0797])
7068
7069   *Matt Caswell*
7070
7071 * Fix memory issues in `BIO_*printf` functions
7072
7073   The internal `fmtstr` function used in processing a "%s" format string in
7074   the `BIO_*printf` functions could overflow while calculating the length of a
7075   string and cause an OOB read when printing very long strings.
7076
7077   Additionally the internal `doapr_outch` function can attempt to write to an
7078   OOB memory location (at an offset from the NULL pointer) in the event of a
7079   memory allocation failure. In 1.0.2 and below this could be caused where
7080   the size of a buffer to be allocated is greater than INT_MAX. E.g. this
7081   could be in processing a very long "%s" format string. Memory leaks can
7082   also occur.
7083
7084   The first issue may mask the second issue dependent on compiler behaviour.
7085   These problems could enable attacks where large amounts of untrusted data
7086   is passed to the `BIO_*printf` functions. If applications use these functions
7087   in this way then they could be vulnerable. OpenSSL itself uses these
7088   functions when printing out human-readable dumps of ASN.1 data. Therefore
7089   applications that print this data could be vulnerable if the data is from
7090   untrusted sources. OpenSSL command line applications could also be
7091   vulnerable where they print out ASN.1 data, or if untrusted data is passed
7092   as command line arguments.
7093
7094   Libssl is not considered directly vulnerable. Additionally certificates etc
7095   received via remote connections via libssl are also unlikely to be able to
7096   trigger these issues because of message size limits enforced within libssl.
7097
7098   This issue was reported to OpenSSL Guido Vranken.
7099   ([CVE-2016-0799])
7100
7101   *Matt Caswell*
7102
7103 * Side channel attack on modular exponentiation
7104
7105   A side-channel attack was found which makes use of cache-bank conflicts on
7106   the Intel Sandy-Bridge microarchitecture which could lead to the recovery
7107   of RSA keys.  The ability to exploit this issue is limited as it relies on
7108   an attacker who has control of code in a thread running on the same
7109   hyper-threaded core as the victim thread which is performing decryptions.
7110
7111   This issue was reported to OpenSSL by Yuval Yarom, The University of
7112   Adelaide and NICTA, Daniel Genkin, Technion and Tel Aviv University, and
7113   Nadia Heninger, University of Pennsylvania with more information at
7114   <http://cachebleed.info>.
7115   ([CVE-2016-0702])
7116
7117   *Andy Polyakov*
7118
7119 * Change the req command to generate a 2048-bit RSA/DSA key by default,
7120   if no keysize is specified with default_bits. This fixes an
7121   omission in an earlier change that changed all RSA/DSA key generation
7122   commands to use 2048 bits by default.
7123
7124   *Emilia Käsper*
7125
7126### Changes between 1.0.1q and 1.0.1r [28 Jan 2016]
7127
7128 * Protection for DH small subgroup attacks
7129
7130   As a precautionary measure the SSL_OP_SINGLE_DH_USE option has been
7131   switched on by default and cannot be disabled. This could have some
7132   performance impact.
7133
7134   *Matt Caswell*
7135
7136 * SSLv2 doesn't block disabled ciphers
7137
7138   A malicious client can negotiate SSLv2 ciphers that have been disabled on
7139   the server and complete SSLv2 handshakes even if all SSLv2 ciphers have
7140   been disabled, provided that the SSLv2 protocol was not also disabled via
7141   SSL_OP_NO_SSLv2.
7142
7143   This issue was reported to OpenSSL on 26th December 2015 by Nimrod Aviram
7144   and Sebastian Schinzel.
7145   ([CVE-2015-3197])
7146
7147   *Viktor Dukhovni*
7148
7149 * Reject DH handshakes with parameters shorter than 1024 bits.
7150
7151   *Kurt Roeckx*
7152
7153### Changes between 1.0.1p and 1.0.1q [3 Dec 2015]
7154
7155 * Certificate verify crash with missing PSS parameter
7156
7157   The signature verification routines will crash with a NULL pointer
7158   dereference if presented with an ASN.1 signature using the RSA PSS
7159   algorithm and absent mask generation function parameter. Since these
7160   routines are used to verify certificate signature algorithms this can be
7161   used to crash any certificate verification operation and exploited in a
7162   DoS attack. Any application which performs certificate verification is
7163   vulnerable including OpenSSL clients and servers which enable client
7164   authentication.
7165
7166   This issue was reported to OpenSSL by Loïc Jonas Etienne (Qnective AG).
7167   ([CVE-2015-3194])
7168
7169   *Stephen Henson*
7170
7171 * X509_ATTRIBUTE memory leak
7172
7173   When presented with a malformed X509_ATTRIBUTE structure OpenSSL will leak
7174   memory. This structure is used by the PKCS#7 and CMS routines so any
7175   application which reads PKCS#7 or CMS data from untrusted sources is
7176   affected. SSL/TLS is not affected.
7177
7178   This issue was reported to OpenSSL by Adam Langley (Google/BoringSSL) using
7179   libFuzzer.
7180   ([CVE-2015-3195])
7181
7182   *Stephen Henson*
7183
7184 * Rewrite EVP_DecodeUpdate (base64 decoding) to fix several bugs.
7185   This changes the decoding behaviour for some invalid messages,
7186   though the change is mostly in the more lenient direction, and
7187   legacy behaviour is preserved as much as possible.
7188
7189   *Emilia Käsper*
7190
7191 * In DSA_generate_parameters_ex, if the provided seed is too short,
7192   use a random seed, as already documented.
7193
7194   *Rich Salz and Ismo Puustinen <ismo.puustinen@intel.com>*
7195
7196### Changes between 1.0.1o and 1.0.1p [9 Jul 2015]
7197
7198 * Alternate chains certificate forgery
7199
7200   During certificate verfification, OpenSSL will attempt to find an
7201   alternative certificate chain if the first attempt to build such a chain
7202   fails. An error in the implementation of this logic can mean that an
7203   attacker could cause certain checks on untrusted certificates to be
7204   bypassed, such as the CA flag, enabling them to use a valid leaf
7205   certificate to act as a CA and "issue" an invalid certificate.
7206
7207   This issue was reported to OpenSSL by Adam Langley/David Benjamin
7208   (Google/BoringSSL).
7209   ([CVE-2015-1793])
7210
7211   *Matt Caswell*
7212
7213 * Race condition handling PSK identify hint
7214
7215   If PSK identity hints are received by a multi-threaded client then
7216   the values are wrongly updated in the parent SSL_CTX structure. This can
7217   result in a race condition potentially leading to a double free of the
7218   identify hint data.
7219   ([CVE-2015-3196])
7220
7221   *Stephen Henson*
7222
7223### Changes between 1.0.1n and 1.0.1o [12 Jun 2015]
7224
7225 * Fix HMAC ABI incompatibility. The previous version introduced an ABI
7226   incompatibility in the handling of HMAC. The previous ABI has now been
7227   restored.
7228
7229### Changes between 1.0.1m and 1.0.1n [11 Jun 2015]
7230
7231 * Malformed ECParameters causes infinite loop
7232
7233   When processing an ECParameters structure OpenSSL enters an infinite loop
7234   if the curve specified is over a specially malformed binary polynomial
7235   field.
7236
7237   This can be used to perform denial of service against any
7238   system which processes public keys, certificate requests or
7239   certificates.  This includes TLS clients and TLS servers with
7240   client authentication enabled.
7241
7242   This issue was reported to OpenSSL by Joseph Barr-Pixton.
7243   ([CVE-2015-1788])
7244
7245   *Andy Polyakov*
7246
7247 * Exploitable out-of-bounds read in X509_cmp_time
7248
7249   X509_cmp_time does not properly check the length of the ASN1_TIME
7250   string and can read a few bytes out of bounds. In addition,
7251   X509_cmp_time accepts an arbitrary number of fractional seconds in the
7252   time string.
7253
7254   An attacker can use this to craft malformed certificates and CRLs of
7255   various sizes and potentially cause a segmentation fault, resulting in
7256   a DoS on applications that verify certificates or CRLs. TLS clients
7257   that verify CRLs are affected. TLS clients and servers with client
7258   authentication enabled may be affected if they use custom verification
7259   callbacks.
7260
7261   This issue was reported to OpenSSL by Robert Swiecki (Google), and
7262   independently by Hanno Böck.
7263   ([CVE-2015-1789])
7264
7265   *Emilia Käsper*
7266
7267 * PKCS7 crash with missing EnvelopedContent
7268
7269   The PKCS#7 parsing code does not handle missing inner EncryptedContent
7270   correctly. An attacker can craft malformed ASN.1-encoded PKCS#7 blobs
7271   with missing content and trigger a NULL pointer dereference on parsing.
7272
7273   Applications that decrypt PKCS#7 data or otherwise parse PKCS#7
7274   structures from untrusted sources are affected. OpenSSL clients and
7275   servers are not affected.
7276
7277   This issue was reported to OpenSSL by Michal Zalewski (Google).
7278   ([CVE-2015-1790])
7279
7280   *Emilia Käsper*
7281
7282 * CMS verify infinite loop with unknown hash function
7283
7284   When verifying a signedData message the CMS code can enter an infinite loop
7285   if presented with an unknown hash function OID. This can be used to perform
7286   denial of service against any system which verifies signedData messages using
7287   the CMS code.
7288   This issue was reported to OpenSSL by Johannes Bauer.
7289   ([CVE-2015-1792])
7290
7291   *Stephen Henson*
7292
7293 * Race condition handling NewSessionTicket
7294
7295   If a NewSessionTicket is received by a multi-threaded client when attempting to
7296   reuse a previous ticket then a race condition can occur potentially leading to
7297   a double free of the ticket data.
7298   ([CVE-2015-1791])
7299
7300   *Matt Caswell*
7301
7302 * Reject DH handshakes with parameters shorter than 768 bits.
7303
7304   *Kurt Roeckx and Emilia Kasper*
7305
7306 * dhparam: generate 2048-bit parameters by default.
7307
7308   *Kurt Roeckx and Emilia Kasper*
7309
7310### Changes between 1.0.1l and 1.0.1m [19 Mar 2015]
7311
7312 * Segmentation fault in ASN1_TYPE_cmp fix
7313
7314   The function ASN1_TYPE_cmp will crash with an invalid read if an attempt is
7315   made to compare ASN.1 boolean types. Since ASN1_TYPE_cmp is used to check
7316   certificate signature algorithm consistency this can be used to crash any
7317   certificate verification operation and exploited in a DoS attack. Any
7318   application which performs certificate verification is vulnerable including
7319   OpenSSL clients and servers which enable client authentication.
7320   ([CVE-2015-0286])
7321
7322   *Stephen Henson*
7323
7324 * ASN.1 structure reuse memory corruption fix
7325
7326   Reusing a structure in ASN.1 parsing may allow an attacker to cause
7327   memory corruption via an invalid write. Such reuse is and has been
7328   strongly discouraged and is believed to be rare.
7329
7330   Applications that parse structures containing CHOICE or ANY DEFINED BY
7331   components may be affected. Certificate parsing (d2i_X509 and related
7332   functions) are however not affected. OpenSSL clients and servers are
7333   not affected.
7334   ([CVE-2015-0287])
7335
7336   *Stephen Henson*
7337
7338 * PKCS7 NULL pointer dereferences fix
7339
7340   The PKCS#7 parsing code does not handle missing outer ContentInfo
7341   correctly. An attacker can craft malformed ASN.1-encoded PKCS#7 blobs with
7342   missing content and trigger a NULL pointer dereference on parsing.
7343
7344   Applications that verify PKCS#7 signatures, decrypt PKCS#7 data or
7345   otherwise parse PKCS#7 structures from untrusted sources are
7346   affected. OpenSSL clients and servers are not affected.
7347
7348   This issue was reported to OpenSSL by Michal Zalewski (Google).
7349   ([CVE-2015-0289])
7350
7351   *Emilia Käsper*
7352
7353 * DoS via reachable assert in SSLv2 servers fix
7354
7355   A malicious client can trigger an OPENSSL_assert (i.e., an abort) in
7356   servers that both support SSLv2 and enable export cipher suites by sending
7357   a specially crafted SSLv2 CLIENT-MASTER-KEY message.
7358
7359   This issue was discovered by Sean Burford (Google) and Emilia Käsper
7360   (OpenSSL development team).
7361   ([CVE-2015-0293])
7362
7363   *Emilia Käsper*
7364
7365 * Use After Free following d2i_ECPrivatekey error fix
7366
7367   A malformed EC private key file consumed via the d2i_ECPrivateKey function
7368   could cause a use after free condition. This, in turn, could cause a double
7369   free in several private key parsing functions (such as d2i_PrivateKey
7370   or EVP_PKCS82PKEY) and could lead to a DoS attack or memory corruption
7371   for applications that receive EC private keys from untrusted
7372   sources. This scenario is considered rare.
7373
7374   This issue was discovered by the BoringSSL project and fixed in their
7375   commit 517073cd4b.
7376   ([CVE-2015-0209])
7377
7378   *Matt Caswell*
7379
7380 * X509_to_X509_REQ NULL pointer deref fix
7381
7382   The function X509_to_X509_REQ will crash with a NULL pointer dereference if
7383   the certificate key is invalid. This function is rarely used in practice.
7384
7385   This issue was discovered by Brian Carpenter.
7386   ([CVE-2015-0288])
7387
7388   *Stephen Henson*
7389
7390 * Removed the export ciphers from the DEFAULT ciphers
7391
7392   *Kurt Roeckx*
7393
7394### Changes between 1.0.1k and 1.0.1l [15 Jan 2015]
7395
7396 * Build fixes for the Windows and OpenVMS platforms
7397
7398   *Matt Caswell and Richard Levitte*
7399
7400### Changes between 1.0.1j and 1.0.1k [8 Jan 2015]
7401
7402 * Fix DTLS segmentation fault in dtls1_get_record. A carefully crafted DTLS
7403   message can cause a segmentation fault in OpenSSL due to a NULL pointer
7404   dereference. This could lead to a Denial Of Service attack. Thanks to
7405   Markus Stenberg of Cisco Systems, Inc. for reporting this issue.
7406   ([CVE-2014-3571])
7407
7408   *Steve Henson*
7409
7410 * Fix DTLS memory leak in dtls1_buffer_record. A memory leak can occur in the
7411   dtls1_buffer_record function under certain conditions. In particular this
7412   could occur if an attacker sent repeated DTLS records with the same
7413   sequence number but for the next epoch. The memory leak could be exploited
7414   by an attacker in a Denial of Service attack through memory exhaustion.
7415   Thanks to Chris Mueller for reporting this issue.
7416   ([CVE-2015-0206])
7417
7418   *Matt Caswell*
7419
7420 * Fix issue where no-ssl3 configuration sets method to NULL. When openssl is
7421   built with the no-ssl3 option and a SSL v3 ClientHello is received the ssl
7422   method would be set to NULL which could later result in a NULL pointer
7423   dereference. Thanks to Frank Schmirler for reporting this issue.
7424   ([CVE-2014-3569])
7425
7426   *Kurt Roeckx*
7427
7428 * Abort handshake if server key exchange message is omitted for ephemeral
7429   ECDH ciphersuites.
7430
7431   Thanks to Karthikeyan Bhargavan of the PROSECCO team at INRIA for
7432   reporting this issue.
7433   ([CVE-2014-3572])
7434
7435   *Steve Henson*
7436
7437 * Remove non-export ephemeral RSA code on client and server. This code
7438   violated the TLS standard by allowing the use of temporary RSA keys in
7439   non-export ciphersuites and could be used by a server to effectively
7440   downgrade the RSA key length used to a value smaller than the server
7441   certificate. Thanks for Karthikeyan Bhargavan of the PROSECCO team at
7442   INRIA or reporting this issue.
7443   ([CVE-2015-0204])
7444
7445   *Steve Henson*
7446
7447 * Fixed issue where DH client certificates are accepted without verification.
7448   An OpenSSL server will accept a DH certificate for client authentication
7449   without the certificate verify message. This effectively allows a client to
7450   authenticate without the use of a private key. This only affects servers
7451   which trust a client certificate authority which issues certificates
7452   containing DH keys: these are extremely rare and hardly ever encountered.
7453   Thanks for Karthikeyan Bhargavan of the PROSECCO team at INRIA or reporting
7454   this issue.
7455   ([CVE-2015-0205])
7456
7457   *Steve Henson*
7458
7459 * Ensure that the session ID context of an SSL is updated when its
7460   SSL_CTX is updated via SSL_set_SSL_CTX.
7461
7462   The session ID context is typically set from the parent SSL_CTX,
7463   and can vary with the CTX.
7464
7465   *Adam Langley*
7466
7467 * Fix various certificate fingerprint issues.
7468
7469   By using non-DER or invalid encodings outside the signed portion of a
7470   certificate the fingerprint can be changed without breaking the signature.
7471   Although no details of the signed portion of the certificate can be changed
7472   this can cause problems with some applications: e.g. those using the
7473   certificate fingerprint for blacklists.
7474
7475   1. Reject signatures with non zero unused bits.
7476
7477   If the BIT STRING containing the signature has non zero unused bits reject
7478   the signature. All current signature algorithms require zero unused bits.
7479
7480   2. Check certificate algorithm consistency.
7481
7482   Check the AlgorithmIdentifier inside TBS matches the one in the
7483   certificate signature. NB: this will result in signature failure
7484   errors for some broken certificates.
7485
7486   Thanks to Konrad Kraszewski from Google for reporting this issue.
7487
7488   3. Check DSA/ECDSA signatures use DER.
7489
7490   Re-encode DSA/ECDSA signatures and compare with the original received
7491   signature. Return an error if there is a mismatch.
7492
7493   This will reject various cases including garbage after signature
7494   (thanks to Antti Karjalainen and Tuomo Untinen from the Codenomicon CROSS
7495   program for discovering this case) and use of BER or invalid ASN.1 INTEGERs
7496   (negative or with leading zeroes).
7497
7498   Further analysis was conducted and fixes were developed by Stephen Henson
7499   of the OpenSSL core team.
7500
7501   ([CVE-2014-8275])
7502
7503   *Steve Henson*
7504
7505 * Correct Bignum squaring. Bignum squaring (BN_sqr) may produce incorrect
7506   results on some platforms, including x86_64. This bug occurs at random
7507   with a very low probability, and is not known to be exploitable in any
7508   way, though its exact impact is difficult to determine. Thanks to Pieter
7509   Wuille (Blockstream) who reported this issue and also suggested an initial
7510   fix. Further analysis was conducted by the OpenSSL development team and
7511   Adam Langley of Google. The final fix was developed by Andy Polyakov of
7512   the OpenSSL core team.
7513   ([CVE-2014-3570])
7514
7515   *Andy Polyakov*
7516
7517 * Do not resume sessions on the server if the negotiated protocol
7518   version does not match the session's version. Resuming with a different
7519   version, while not strictly forbidden by the RFC, is of questionable
7520   sanity and breaks all known clients.
7521
7522   *David Benjamin, Emilia Käsper*
7523
7524 * Tighten handling of the ChangeCipherSpec (CCS) message: reject
7525   early CCS messages during renegotiation. (Note that because
7526   renegotiation is encrypted, this early CCS was not exploitable.)
7527
7528   *Emilia Käsper*
7529
7530 * Tighten client-side session ticket handling during renegotiation:
7531   ensure that the client only accepts a session ticket if the server sends
7532   the extension anew in the ServerHello. Previously, a TLS client would
7533   reuse the old extension state and thus accept a session ticket if one was
7534   announced in the initial ServerHello.
7535
7536   Similarly, ensure that the client requires a session ticket if one
7537   was advertised in the ServerHello. Previously, a TLS client would
7538   ignore a missing NewSessionTicket message.
7539
7540   *Emilia Käsper*
7541
7542### Changes between 1.0.1i and 1.0.1j [15 Oct 2014]
7543
7544 * SRTP Memory Leak.
7545
7546   A flaw in the DTLS SRTP extension parsing code allows an attacker, who
7547   sends a carefully crafted handshake message, to cause OpenSSL to fail
7548   to free up to 64k of memory causing a memory leak. This could be
7549   exploited in a Denial Of Service attack. This issue affects OpenSSL
7550   1.0.1 server implementations for both SSL/TLS and DTLS regardless of
7551   whether SRTP is used or configured. Implementations of OpenSSL that
7552   have been compiled with OPENSSL_NO_SRTP defined are not affected.
7553
7554   The fix was developed by the OpenSSL team.
7555   ([CVE-2014-3513])
7556
7557   *OpenSSL team*
7558
7559 * Session Ticket Memory Leak.
7560
7561   When an OpenSSL SSL/TLS/DTLS server receives a session ticket the
7562   integrity of that ticket is first verified. In the event of a session
7563   ticket integrity check failing, OpenSSL will fail to free memory
7564   causing a memory leak. By sending a large number of invalid session
7565   tickets an attacker could exploit this issue in a Denial Of Service
7566   attack.
7567   ([CVE-2014-3567])
7568
7569   *Steve Henson*
7570
7571 * Build option no-ssl3 is incomplete.
7572
7573   When OpenSSL is configured with "no-ssl3" as a build option, servers
7574   could accept and complete a SSL 3.0 handshake, and clients could be
7575   configured to send them.
7576   ([CVE-2014-3568])
7577
7578   *Akamai and the OpenSSL team*
7579
7580 * Add support for TLS_FALLBACK_SCSV.
7581   Client applications doing fallback retries should call
7582   SSL_set_mode(s, SSL_MODE_SEND_FALLBACK_SCSV).
7583   ([CVE-2014-3566])
7584
7585   *Adam Langley, Bodo Moeller*
7586
7587 * Add additional DigestInfo checks.
7588
7589   Re-encode DigestInto in DER and check against the original when
7590   verifying RSA signature: this will reject any improperly encoded
7591   DigestInfo structures.
7592
7593   Note: this is a precautionary measure and no attacks are currently known.
7594
7595   *Steve Henson*
7596
7597### Changes between 1.0.1h and 1.0.1i [6 Aug 2014]
7598
7599 * Fix SRP buffer overrun vulnerability. Invalid parameters passed to the
7600   SRP code can be overrun an internal buffer. Add sanity check that
7601   g, A, B < N to SRP code.
7602
7603   Thanks to Sean Devlin and Watson Ladd of Cryptography Services, NCC
7604   Group for discovering this issue.
7605   ([CVE-2014-3512])
7606
7607   *Steve Henson*
7608
7609 * A flaw in the OpenSSL SSL/TLS server code causes the server to negotiate
7610   TLS 1.0 instead of higher protocol versions when the ClientHello message
7611   is badly fragmented. This allows a man-in-the-middle attacker to force a
7612   downgrade to TLS 1.0 even if both the server and the client support a
7613   higher protocol version, by modifying the client's TLS records.
7614
7615   Thanks to David Benjamin and Adam Langley (Google) for discovering and
7616   researching this issue.
7617   ([CVE-2014-3511])
7618
7619   *David Benjamin*
7620
7621 * OpenSSL DTLS clients enabling anonymous (EC)DH ciphersuites are subject
7622   to a denial of service attack. A malicious server can crash the client
7623   with a null pointer dereference (read) by specifying an anonymous (EC)DH
7624   ciphersuite and sending carefully crafted handshake messages.
7625
7626   Thanks to Felix Gröbert (Google) for discovering and researching this
7627   issue.
7628   ([CVE-2014-3510])
7629
7630   *Emilia Käsper*
7631
7632 * By sending carefully crafted DTLS packets an attacker could cause openssl
7633   to leak memory. This can be exploited through a Denial of Service attack.
7634   Thanks to Adam Langley for discovering and researching this issue.
7635   ([CVE-2014-3507])
7636
7637   *Adam Langley*
7638
7639 * An attacker can force openssl to consume large amounts of memory whilst
7640   processing DTLS handshake messages. This can be exploited through a
7641   Denial of Service attack.
7642   Thanks to Adam Langley for discovering and researching this issue.
7643   ([CVE-2014-3506])
7644
7645   *Adam Langley*
7646
7647 * An attacker can force an error condition which causes openssl to crash
7648   whilst processing DTLS packets due to memory being freed twice. This
7649   can be exploited through a Denial of Service attack.
7650   Thanks to Adam Langley and Wan-Teh Chang for discovering and researching
7651   this issue.
7652   ([CVE-2014-3505])
7653
7654   *Adam Langley*
7655
7656 * If a multithreaded client connects to a malicious server using a resumed
7657   session and the server sends an ec point format extension it could write
7658   up to 255 bytes to freed memory.
7659
7660   Thanks to Gabor Tyukasz (LogMeIn Inc) for discovering and researching this
7661   issue.
7662   ([CVE-2014-3509])
7663
7664   *Gabor Tyukasz*
7665
7666 * A malicious server can crash an OpenSSL client with a null pointer
7667   dereference (read) by specifying an SRP ciphersuite even though it was not
7668   properly negotiated with the client. This can be exploited through a
7669   Denial of Service attack.
7670
7671   Thanks to Joonas Kuorilehto and Riku Hietamäki (Codenomicon) for
7672   discovering and researching this issue.
7673   ([CVE-2014-5139])
7674
7675   *Steve Henson*
7676
7677 * A flaw in OBJ_obj2txt may cause pretty printing functions such as
7678   X509_name_oneline, X509_name_print_ex et al. to leak some information
7679   from the stack. Applications may be affected if they echo pretty printing
7680   output to the attacker.
7681
7682   Thanks to Ivan Fratric (Google) for discovering this issue.
7683   ([CVE-2014-3508])
7684
7685   *Emilia Käsper, and Steve Henson*
7686
7687 * Fix ec_GFp_simple_points_make_affine (thus, EC_POINTs_mul etc.)
7688   for corner cases. (Certain input points at infinity could lead to
7689   bogus results, with non-infinity inputs mapped to infinity too.)
7690
7691   *Bodo Moeller*
7692
7693### Changes between 1.0.1g and 1.0.1h [5 Jun 2014]
7694
7695 * Fix for SSL/TLS MITM flaw. An attacker using a carefully crafted
7696   handshake can force the use of weak keying material in OpenSSL
7697   SSL/TLS clients and servers.
7698
7699   Thanks to KIKUCHI Masashi (Lepidum Co. Ltd.) for discovering and
7700   researching this issue. ([CVE-2014-0224])
7701
7702   *KIKUCHI Masashi, Steve Henson*
7703
7704 * Fix DTLS recursion flaw. By sending an invalid DTLS handshake to an
7705   OpenSSL DTLS client the code can be made to recurse eventually crashing
7706   in a DoS attack.
7707
7708   Thanks to Imre Rad (Search-Lab Ltd.) for discovering this issue.
7709   ([CVE-2014-0221])
7710
7711   *Imre Rad, Steve Henson*
7712
7713 * Fix DTLS invalid fragment vulnerability. A buffer overrun attack can
7714   be triggered by sending invalid DTLS fragments to an OpenSSL DTLS
7715   client or server. This is potentially exploitable to run arbitrary
7716   code on a vulnerable client or server.
7717
7718   Thanks to Jüri Aedla for reporting this issue. ([CVE-2014-0195])
7719
7720   *Jüri Aedla, Steve Henson*
7721
7722 * Fix bug in TLS code where clients enable anonymous ECDH ciphersuites
7723   are subject to a denial of service attack.
7724
7725   Thanks to Felix Gröbert and Ivan Fratric at Google for discovering
7726   this issue. ([CVE-2014-3470])
7727
7728   *Felix Gröbert, Ivan Fratric, Steve Henson*
7729
7730 * Harmonize version and its documentation. -f flag is used to display
7731   compilation flags.
7732
7733   *mancha <mancha1@zoho.com>*
7734
7735 * Fix eckey_priv_encode so it immediately returns an error upon a failure
7736   in i2d_ECPrivateKey.
7737
7738   *mancha <mancha1@zoho.com>*
7739
7740 * Fix some double frees. These are not thought to be exploitable.
7741
7742   *mancha <mancha1@zoho.com>*
7743
7744### Changes between 1.0.1f and 1.0.1g [7 Apr 2014]
7745
7746 * A missing bounds check in the handling of the TLS heartbeat extension
7747   can be used to reveal up to 64k of memory to a connected client or
7748   server.
7749
7750   Thanks for Neel Mehta of Google Security for discovering this bug and to
7751   Adam Langley <agl@chromium.org> and Bodo Moeller <bmoeller@acm.org> for
7752   preparing the fix ([CVE-2014-0160])
7753
7754   *Adam Langley, Bodo Moeller*
7755
7756 * Fix for the attack described in the paper "Recovering OpenSSL
7757   ECDSA Nonces Using the FLUSH+RELOAD Cache Side-channel Attack"
7758   by Yuval Yarom and Naomi Benger. Details can be obtained from:
7759   <http://eprint.iacr.org/2014/140>
7760
7761   Thanks to Yuval Yarom and Naomi Benger for discovering this
7762   flaw and to Yuval Yarom for supplying a fix ([CVE-2014-0076])
7763
7764   *Yuval Yarom and Naomi Benger*
7765
7766 * TLS pad extension: draft-agl-tls-padding-03
7767
7768   Workaround for the "TLS hang bug" (see FAQ and PR#2771): if the
7769   TLS client Hello record length value would otherwise be > 255 and
7770   less that 512 pad with a dummy extension containing zeroes so it
7771   is at least 512 bytes long.
7772
7773   *Adam Langley, Steve Henson*
7774
7775### Changes between 1.0.1e and 1.0.1f [6 Jan 2014]
7776
7777 * Fix for TLS record tampering bug. A carefully crafted invalid
7778   handshake could crash OpenSSL with a NULL pointer exception.
7779   Thanks to Anton Johansson for reporting this issues.
7780   ([CVE-2013-4353])
7781
7782 * Keep original DTLS digest and encryption contexts in retransmission
7783   structures so we can use the previous session parameters if they need
7784   to be resent. ([CVE-2013-6450])
7785
7786   *Steve Henson*
7787
7788 * Add option SSL_OP_SAFARI_ECDHE_ECDSA_BUG (part of SSL_OP_ALL) which
7789   avoids preferring ECDHE-ECDSA ciphers when the client appears to be
7790   Safari on OS X.  Safari on OS X 10.8..10.8.3 advertises support for
7791   several ECDHE-ECDSA ciphers, but fails to negotiate them.  The bug
7792   is fixed in OS X 10.8.4, but Apple have ruled out both hot fixing
7793   10.8..10.8.3 and forcing users to upgrade to 10.8.4 or newer.
7794
7795   *Rob Stradling, Adam Langley*
7796
7797### Changes between 1.0.1d and 1.0.1e [11 Feb 2013]
7798
7799 * Correct fix for CVE-2013-0169. The original didn't work on AES-NI
7800   supporting platforms or when small records were transferred.
7801
7802   *Andy Polyakov, Steve Henson*
7803
7804### Changes between 1.0.1c and 1.0.1d [5 Feb 2013]
7805
7806 * Make the decoding of SSLv3, TLS and DTLS CBC records constant time.
7807
7808   This addresses the flaw in CBC record processing discovered by
7809   Nadhem Alfardan and Kenny Paterson. Details of this attack can be found
7810   at: <http://www.isg.rhul.ac.uk/tls/>
7811
7812   Thanks go to Nadhem Alfardan and Kenny Paterson of the Information
7813   Security Group at Royal Holloway, University of London
7814   (www.isg.rhul.ac.uk) for discovering this flaw and Adam Langley and
7815   Emilia Käsper for the initial patch.
7816   ([CVE-2013-0169])
7817
7818   *Emilia Käsper, Adam Langley, Ben Laurie, Andy Polyakov, Steve Henson*
7819
7820 * Fix flaw in AESNI handling of TLS 1.2 and 1.1 records for CBC mode
7821   ciphersuites which can be exploited in a denial of service attack.
7822   Thanks go to and to Adam Langley <agl@chromium.org> for discovering
7823   and detecting this bug and to Wolfgang Ettlinger
7824   <wolfgang.ettlinger@gmail.com> for independently discovering this issue.
7825   ([CVE-2012-2686])
7826
7827   *Adam Langley*
7828
7829 * Return an error when checking OCSP signatures when key is NULL.
7830   This fixes a DoS attack. ([CVE-2013-0166])
7831
7832   *Steve Henson*
7833
7834 * Make openssl verify return errors.
7835
7836   *Chris Palmer <palmer@google.com> and Ben Laurie*
7837
7838 * Call OCSP Stapling callback after ciphersuite has been chosen, so
7839   the right response is stapled. Also change SSL_get_certificate()
7840   so it returns the certificate actually sent.
7841   See <http://rt.openssl.org/Ticket/Display.html?id=2836>.
7842
7843   *Rob Stradling <rob.stradling@comodo.com>*
7844
7845 * Fix possible deadlock when decoding public keys.
7846
7847   *Steve Henson*
7848
7849 * Don't use TLS 1.0 record version number in initial client hello
7850   if renegotiating.
7851
7852   *Steve Henson*
7853
7854### Changes between 1.0.1b and 1.0.1c [10 May 2012]
7855
7856 * Sanity check record length before skipping explicit IV in TLS
7857   1.2, 1.1 and DTLS to fix DoS attack.
7858
7859   Thanks to Codenomicon for discovering this issue using Fuzz-o-Matic
7860   fuzzing as a service testing platform.
7861   ([CVE-2012-2333])
7862
7863   *Steve Henson*
7864
7865 * Initialise tkeylen properly when encrypting CMS messages.
7866   Thanks to Solar Designer of Openwall for reporting this issue.
7867
7868   *Steve Henson*
7869
7870 * In FIPS mode don't try to use composite ciphers as they are not
7871   approved.
7872
7873   *Steve Henson*
7874
7875### Changes between 1.0.1a and 1.0.1b [26 Apr 2012]
7876
7877 * OpenSSL 1.0.0 sets SSL_OP_ALL to 0x80000FFFL and OpenSSL 1.0.1 and
7878   1.0.1a set SSL_OP_NO_TLSv1_1 to 0x00000400L which would unfortunately
7879   mean any application compiled against OpenSSL 1.0.0 headers setting
7880   SSL_OP_ALL would also set SSL_OP_NO_TLSv1_1, unintentionally disabling
7881   TLS 1.1 also. Fix this by changing the value of SSL_OP_NO_TLSv1_1 to
7882   0x10000000L Any application which was previously compiled against
7883   OpenSSL 1.0.1 or 1.0.1a headers and which cares about SSL_OP_NO_TLSv1_1
7884   will need to be recompiled as a result. Letting be results in
7885   inability to disable specifically TLS 1.1 and in client context,
7886   in unlike event, limit maximum offered version to TLS 1.0 [see below].
7887
7888   *Steve Henson*
7889
7890 * In order to ensure interoperability SSL_OP_NO_protocolX does not
7891   disable just protocol X, but all protocols above X *if* there are
7892   protocols *below* X still enabled. In more practical terms it means
7893   that if application wants to disable TLS1.0 in favor of TLS1.1 and
7894   above, it's not sufficient to pass `SSL_OP_NO_TLSv1`, one has to pass
7895   `SSL_OP_NO_TLSv1|SSL_OP_NO_SSLv3|SSL_OP_NO_SSLv2`. This applies to
7896   client side.
7897
7898   *Andy Polyakov*
7899
7900### Changes between 1.0.1 and 1.0.1a [19 Apr 2012]
7901
7902 * Check for potentially exploitable overflows in asn1_d2i_read_bio
7903   BUF_mem_grow and BUF_mem_grow_clean. Refuse attempts to shrink buffer
7904   in CRYPTO_realloc_clean.
7905
7906   Thanks to Tavis Ormandy, Google Security Team, for discovering this
7907   issue and to Adam Langley <agl@chromium.org> for fixing it.
7908   ([CVE-2012-2110])
7909
7910   *Adam Langley (Google), Tavis Ormandy, Google Security Team*
7911
7912 * Don't allow TLS 1.2 SHA-256 ciphersuites in TLS 1.0, 1.1 connections.
7913
7914   *Adam Langley*
7915
7916 * Workarounds for some broken servers that "hang" if a client hello
7917   record length exceeds 255 bytes.
7918
7919   1. Do not use record version number > TLS 1.0 in initial client
7920      hello: some (but not all) hanging servers will now work.
7921   2. If we set OPENSSL_MAX_TLS1_2_CIPHER_LENGTH this will truncate
7922      the number of ciphers sent in the client hello. This should be
7923      set to an even number, such as 50, for example by passing:
7924      -DOPENSSL_MAX_TLS1_2_CIPHER_LENGTH=50 to config or Configure.
7925      Most broken servers should now work.
7926   3. If all else fails setting OPENSSL_NO_TLS1_2_CLIENT will disable
7927      TLS 1.2 client support entirely.
7928
7929   *Steve Henson*
7930
7931 * Fix SEGV in Vector Permutation AES module observed in OpenSSH.
7932
7933   *Andy Polyakov*
7934
7935### Changes between 1.0.0h and 1.0.1  [14 Mar 2012]
7936
7937 * Add compatibility with old MDC2 signatures which use an ASN1 OCTET
7938   STRING form instead of a DigestInfo.
7939
7940   *Steve Henson*
7941
7942 * The format used for MDC2 RSA signatures is inconsistent between EVP
7943   and the RSA_sign/RSA_verify functions. This was made more apparent when
7944   OpenSSL used RSA_sign/RSA_verify for some RSA signatures in particular
7945   those which went through EVP_PKEY_METHOD in 1.0.0 and later. Detect
7946   the correct format in RSA_verify so both forms transparently work.
7947
7948   *Steve Henson*
7949
7950 * Some servers which support TLS 1.0 can choke if we initially indicate
7951   support for TLS 1.2 and later renegotiate using TLS 1.0 in the RSA
7952   encrypted premaster secret. As a workaround use the maximum permitted
7953   client version in client hello, this should keep such servers happy
7954   and still work with previous versions of OpenSSL.
7955
7956   *Steve Henson*
7957
7958 * Add support for TLS/DTLS heartbeats.
7959
7960   *Robin Seggelmann <seggelmann@fh-muenster.de>*
7961
7962 * Add support for SCTP.
7963
7964   *Robin Seggelmann <seggelmann@fh-muenster.de>*
7965
7966 * Improved PRNG seeding for VOS.
7967
7968   *Paul Green <Paul.Green@stratus.com>*
7969
7970 * Extensive assembler packs updates, most notably:
7971
7972   - x86[_64]:     AES-NI, PCLMULQDQ, RDRAND support;
7973   - x86[_64]:     SSSE3 support (SHA1, vector-permutation AES);
7974   - x86_64:       bit-sliced AES implementation;
7975   - ARM:          NEON support, contemporary platforms optimizations;
7976   - s390x:        z196 support;
7977   - `*`:            GHASH and GF(2^m) multiplication implementations;
7978
7979   *Andy Polyakov*
7980
7981 * Make TLS-SRP code conformant with RFC 5054 API cleanup
7982   (removal of unnecessary code)
7983
7984   *Peter Sylvester <peter.sylvester@edelweb.fr>*
7985
7986 * Add TLS key material exporter from RFC 5705.
7987
7988   *Eric Rescorla*
7989
7990 * Add DTLS-SRTP negotiation from RFC 5764.
7991
7992   *Eric Rescorla*
7993
7994 * Add Next Protocol Negotiation,
7995   <http://tools.ietf.org/html/draft-agl-tls-nextprotoneg-00>. Can be
7996   disabled with a no-npn flag to config or Configure. Code donated
7997   by Google.
7998
7999   *Adam Langley <agl@google.com> and Ben Laurie*
8000
8001 * Add optional 64-bit optimized implementations of elliptic curves NIST-P224,
8002   NIST-P256, NIST-P521, with constant-time single point multiplication on
8003   typical inputs. Compiler support for the nonstandard type `__uint128_t` is
8004   required to use this (present in gcc 4.4 and later, for 64-bit builds).
8005   Code made available under Apache License version 2.0.
8006
8007   Specify "enable-ec_nistp_64_gcc_128" on the Configure (or config) command
8008   line to include this in your build of OpenSSL, and run "make depend" (or
8009   "make update"). This enables the following EC_METHODs:
8010
8011           EC_GFp_nistp224_method()
8012           EC_GFp_nistp256_method()
8013           EC_GFp_nistp521_method()
8014
8015   EC_GROUP_new_by_curve_name() will automatically use these (while
8016   EC_GROUP_new_curve_GFp() currently prefers the more flexible
8017   implementations).
8018
8019   *Emilia Käsper, Adam Langley, Bodo Moeller (Google)*
8020
8021 * Use type ossl_ssize_t instead of ssize_t which isn't available on
8022   all platforms. Move ssize_t definition from e_os.h to the public
8023   header file e_os2.h as it now appears in public header file cms.h
8024
8025   *Steve Henson*
8026
8027 * New -sigopt option to the ca, req and x509 utilities. Additional
8028   signature parameters can be passed using this option and in
8029   particular PSS.
8030
8031   *Steve Henson*
8032
8033 * Add RSA PSS signing function. This will generate and set the
8034   appropriate AlgorithmIdentifiers for PSS based on those in the
8035   corresponding EVP_MD_CTX structure. No application support yet.
8036
8037   *Steve Henson*
8038
8039 * Support for companion algorithm specific ASN1 signing routines.
8040   New function ASN1_item_sign_ctx() signs a pre-initialised
8041   EVP_MD_CTX structure and sets AlgorithmIdentifiers based on
8042   the appropriate parameters.
8043
8044   *Steve Henson*
8045
8046 * Add new algorithm specific ASN1 verification initialisation function
8047   to EVP_PKEY_ASN1_METHOD: this is not in EVP_PKEY_METHOD since the ASN1
8048   handling will be the same no matter what EVP_PKEY_METHOD is used.
8049   Add a PSS handler to support verification of PSS signatures: checked
8050   against a number of sample certificates.
8051
8052   *Steve Henson*
8053
8054 * Add signature printing for PSS. Add PSS OIDs.
8055
8056   *Steve Henson, Martin Kaiser <lists@kaiser.cx>*
8057
8058 * Add algorithm specific signature printing. An individual ASN1 method
8059   can now print out signatures instead of the standard hex dump.
8060
8061   More complex signatures (e.g. PSS) can print out more meaningful
8062   information. Include DSA version that prints out the signature
8063   parameters r, s.
8064
8065   *Steve Henson*
8066
8067 * Password based recipient info support for CMS library: implementing
8068   RFC3211.
8069
8070   *Steve Henson*
8071
8072 * Split password based encryption into PBES2 and PBKDF2 functions. This
8073   neatly separates the code into cipher and PBE sections and is required
8074   for some algorithms that split PBES2 into separate pieces (such as
8075   password based CMS).
8076
8077   *Steve Henson*
8078
8079 * Session-handling fixes:
8080   - Fix handling of connections that are resuming with a session ID,
8081     but also support Session Tickets.
8082   - Fix a bug that suppressed issuing of a new ticket if the client
8083     presented a ticket with an expired session.
8084   - Try to set the ticket lifetime hint to something reasonable.
8085   - Make tickets shorter by excluding irrelevant information.
8086   - On the client side, don't ignore renewed tickets.
8087
8088   *Adam Langley, Bodo Moeller (Google)*
8089
8090 * Fix PSK session representation.
8091
8092   *Bodo Moeller*
8093
8094 * Add RC4-MD5 and AESNI-SHA1 "stitched" implementations.
8095
8096   This work was sponsored by Intel.
8097
8098   *Andy Polyakov*
8099
8100 * Add GCM support to TLS library. Some custom code is needed to split
8101   the IV between the fixed (from PRF) and explicit (from TLS record)
8102   portions. This adds all GCM ciphersuites supported by RFC5288 and
8103   RFC5289. Generalise some `AES*` cipherstrings to include GCM and
8104   add a special AESGCM string for GCM only.
8105
8106   *Steve Henson*
8107
8108 * Expand range of ctrls for AES GCM. Permit setting invocation
8109   field on decrypt and retrieval of invocation field only on encrypt.
8110
8111   *Steve Henson*
8112
8113 * Add HMAC ECC ciphersuites from RFC5289. Include SHA384 PRF support.
8114   As required by RFC5289 these ciphersuites cannot be used if for
8115   versions of TLS earlier than 1.2.
8116
8117   *Steve Henson*
8118
8119 * For FIPS capable OpenSSL interpret a NULL default public key method
8120   as unset and return the appropriate default but do *not* set the default.
8121   This means we can return the appropriate method in applications that
8122   switch between FIPS and non-FIPS modes.
8123
8124   *Steve Henson*
8125
8126 * Redirect HMAC and CMAC operations to FIPS module in FIPS mode. If an
8127   ENGINE is used then we cannot handle that in the FIPS module so we
8128   keep original code iff non-FIPS operations are allowed.
8129
8130   *Steve Henson*
8131
8132 * Add -attime option to openssl utilities.
8133
8134   *Peter Eckersley <pde@eff.org>, Ben Laurie and Steve Henson*
8135
8136 * Redirect DSA and DH operations to FIPS module in FIPS mode.
8137
8138   *Steve Henson*
8139
8140 * Redirect ECDSA and ECDH operations to FIPS module in FIPS mode. Also use
8141   FIPS EC methods unconditionally for now.
8142
8143   *Steve Henson*
8144
8145 * New build option no-ec2m to disable characteristic 2 code.
8146
8147   *Steve Henson*
8148
8149 * Backport libcrypto audit of return value checking from 1.1.0-dev; not
8150   all cases can be covered as some introduce binary incompatibilities.
8151
8152   *Steve Henson*
8153
8154 * Redirect RSA operations to FIPS module including keygen,
8155   encrypt, decrypt, sign and verify. Block use of non FIPS RSA methods.
8156
8157   *Steve Henson*
8158
8159 * Add similar low-level API blocking to ciphers.
8160
8161   *Steve Henson*
8162
8163 * low-level digest APIs are not approved in FIPS mode: any attempt
8164   to use these will cause a fatal error. Applications that *really* want
8165   to use them can use the `private_*` version instead.
8166
8167   *Steve Henson*
8168
8169 * Redirect cipher operations to FIPS module for FIPS builds.
8170
8171   *Steve Henson*
8172
8173 * Redirect digest operations to FIPS module for FIPS builds.
8174
8175   *Steve Henson*
8176
8177 * Update build system to add "fips" flag which will link in fipscanister.o
8178   for static and shared library builds embedding a signature if needed.
8179
8180   *Steve Henson*
8181
8182 * Output TLS supported curves in preference order instead of numerical
8183   order. This is currently hardcoded for the highest order curves first.
8184   This should be configurable so applications can judge speed vs strength.
8185
8186   *Steve Henson*
8187
8188 * Add TLS v1.2 server support for client authentication.
8189
8190   *Steve Henson*
8191
8192 * Add support for FIPS mode in ssl library: disable SSLv3, non-FIPS ciphers
8193   and enable MD5.
8194
8195   *Steve Henson*
8196
8197 * Functions FIPS_mode_set() and FIPS_mode() which call the underlying
8198   FIPS modules versions.
8199
8200   *Steve Henson*
8201
8202 * Add TLS v1.2 client side support for client authentication. Keep cache
8203   of handshake records longer as we don't know the hash algorithm to use
8204   until after the certificate request message is received.
8205
8206   *Steve Henson*
8207
8208 * Initial TLS v1.2 client support. Add a default signature algorithms
8209   extension including all the algorithms we support. Parse new signature
8210   format in client key exchange. Relax some ECC signing restrictions for
8211   TLS v1.2 as indicated in RFC5246.
8212
8213   *Steve Henson*
8214
8215 * Add server support for TLS v1.2 signature algorithms extension. Switch
8216   to new signature format when needed using client digest preference.
8217   All server ciphersuites should now work correctly in TLS v1.2. No client
8218   support yet and no support for client certificates.
8219
8220   *Steve Henson*
8221
8222 * Initial TLS v1.2 support. Add new SHA256 digest to ssl code, switch
8223   to SHA256 for PRF when using TLS v1.2 and later. Add new SHA256 based
8224   ciphersuites. At present only RSA key exchange ciphersuites work with
8225   TLS v1.2. Add new option for TLS v1.2 replacing the old and obsolete
8226   SSL_OP_PKCS1_CHECK flags with SSL_OP_NO_TLSv1_2. New TLSv1.2 methods
8227   and version checking.
8228
8229   *Steve Henson*
8230
8231 * New option OPENSSL_NO_SSL_INTERN. If an application can be compiled
8232   with this defined it will not be affected by any changes to ssl internal
8233   structures. Add several utility functions to allow openssl application
8234   to work with OPENSSL_NO_SSL_INTERN defined.
8235
8236   *Steve Henson*
8237
8238 * A long standing patch to add support for SRP from EdelWeb (Peter
8239   Sylvester and Christophe Renou) was integrated.
8240   *Christophe Renou <christophe.renou@edelweb.fr>, Peter Sylvester
8241   <peter.sylvester@edelweb.fr>, Tom Wu <tjw@cs.stanford.edu>, and
8242   Ben Laurie*
8243
8244 * Add functions to copy EVP_PKEY_METHOD and retrieve flags and id.
8245
8246   *Steve Henson*
8247
8248 * Permit abbreviated handshakes when renegotiating using the function
8249   SSL_renegotiate_abbreviated().
8250
8251   *Robin Seggelmann <seggelmann@fh-muenster.de>*
8252
8253 * Add call to ENGINE_register_all_complete() to
8254   ENGINE_load_builtin_engines(), so some implementations get used
8255   automatically instead of needing explicit application support.
8256
8257   *Steve Henson*
8258
8259 * Add support for TLS key exporter as described in RFC5705.
8260
8261   *Robin Seggelmann <seggelmann@fh-muenster.de>, Steve Henson*
8262
8263 * Initial TLSv1.1 support. Since TLSv1.1 is very similar to TLS v1.0 only
8264   a few changes are required:
8265
8266     Add SSL_OP_NO_TLSv1_1 flag.
8267     Add TLSv1_1 methods.
8268     Update version checking logic to handle version 1.1.
8269     Add explicit IV handling (ported from DTLS code).
8270     Add command line options to s_client/s_server.
8271
8272   *Steve Henson*
8273
8274OpenSSL 1.0.0
8275-------------
8276
8277### Changes between 1.0.0s and 1.0.0t [3 Dec 2015]
8278
8279 * X509_ATTRIBUTE memory leak
8280
8281   When presented with a malformed X509_ATTRIBUTE structure OpenSSL will leak
8282   memory. This structure is used by the PKCS#7 and CMS routines so any
8283   application which reads PKCS#7 or CMS data from untrusted sources is
8284   affected. SSL/TLS is not affected.
8285
8286   This issue was reported to OpenSSL by Adam Langley (Google/BoringSSL) using
8287   libFuzzer.
8288   ([CVE-2015-3195])
8289
8290   *Stephen Henson*
8291
8292 * Race condition handling PSK identify hint
8293
8294   If PSK identity hints are received by a multi-threaded client then
8295   the values are wrongly updated in the parent SSL_CTX structure. This can
8296   result in a race condition potentially leading to a double free of the
8297   identify hint data.
8298   ([CVE-2015-3196])
8299
8300   *Stephen Henson*
8301
8302### Changes between 1.0.0r and 1.0.0s [11 Jun 2015]
8303
8304 * Malformed ECParameters causes infinite loop
8305
8306   When processing an ECParameters structure OpenSSL enters an infinite loop
8307   if the curve specified is over a specially malformed binary polynomial
8308   field.
8309
8310   This can be used to perform denial of service against any
8311   system which processes public keys, certificate requests or
8312   certificates.  This includes TLS clients and TLS servers with
8313   client authentication enabled.
8314
8315   This issue was reported to OpenSSL by Joseph Barr-Pixton.
8316   ([CVE-2015-1788])
8317
8318   *Andy Polyakov*
8319
8320 * Exploitable out-of-bounds read in X509_cmp_time
8321
8322   X509_cmp_time does not properly check the length of the ASN1_TIME
8323   string and can read a few bytes out of bounds. In addition,
8324   X509_cmp_time accepts an arbitrary number of fractional seconds in the
8325   time string.
8326
8327   An attacker can use this to craft malformed certificates and CRLs of
8328   various sizes and potentially cause a segmentation fault, resulting in
8329   a DoS on applications that verify certificates or CRLs. TLS clients
8330   that verify CRLs are affected. TLS clients and servers with client
8331   authentication enabled may be affected if they use custom verification
8332   callbacks.
8333
8334   This issue was reported to OpenSSL by Robert Swiecki (Google), and
8335   independently by Hanno Böck.
8336   ([CVE-2015-1789])
8337
8338   *Emilia Käsper*
8339
8340 * PKCS7 crash with missing EnvelopedContent
8341
8342   The PKCS#7 parsing code does not handle missing inner EncryptedContent
8343   correctly. An attacker can craft malformed ASN.1-encoded PKCS#7 blobs
8344   with missing content and trigger a NULL pointer dereference on parsing.
8345
8346   Applications that decrypt PKCS#7 data or otherwise parse PKCS#7
8347   structures from untrusted sources are affected. OpenSSL clients and
8348   servers are not affected.
8349
8350   This issue was reported to OpenSSL by Michal Zalewski (Google).
8351   ([CVE-2015-1790])
8352
8353   *Emilia Käsper*
8354
8355 * CMS verify infinite loop with unknown hash function
8356
8357   When verifying a signedData message the CMS code can enter an infinite loop
8358   if presented with an unknown hash function OID. This can be used to perform
8359   denial of service against any system which verifies signedData messages using
8360   the CMS code.
8361   This issue was reported to OpenSSL by Johannes Bauer.
8362   ([CVE-2015-1792])
8363
8364   *Stephen Henson*
8365
8366 * Race condition handling NewSessionTicket
8367
8368   If a NewSessionTicket is received by a multi-threaded client when attempting to
8369   reuse a previous ticket then a race condition can occur potentially leading to
8370   a double free of the ticket data.
8371   ([CVE-2015-1791])
8372
8373   *Matt Caswell*
8374
8375### Changes between 1.0.0q and 1.0.0r [19 Mar 2015]
8376
8377 * Segmentation fault in ASN1_TYPE_cmp fix
8378
8379   The function ASN1_TYPE_cmp will crash with an invalid read if an attempt is
8380   made to compare ASN.1 boolean types. Since ASN1_TYPE_cmp is used to check
8381   certificate signature algorithm consistency this can be used to crash any
8382   certificate verification operation and exploited in a DoS attack. Any
8383   application which performs certificate verification is vulnerable including
8384   OpenSSL clients and servers which enable client authentication.
8385   ([CVE-2015-0286])
8386
8387   *Stephen Henson*
8388
8389 * ASN.1 structure reuse memory corruption fix
8390
8391   Reusing a structure in ASN.1 parsing may allow an attacker to cause
8392   memory corruption via an invalid write. Such reuse is and has been
8393   strongly discouraged and is believed to be rare.
8394
8395   Applications that parse structures containing CHOICE or ANY DEFINED BY
8396   components may be affected. Certificate parsing (d2i_X509 and related
8397   functions) are however not affected. OpenSSL clients and servers are
8398   not affected.
8399   ([CVE-2015-0287])
8400
8401   *Stephen Henson*
8402
8403 * PKCS7 NULL pointer dereferences fix
8404
8405   The PKCS#7 parsing code does not handle missing outer ContentInfo
8406   correctly. An attacker can craft malformed ASN.1-encoded PKCS#7 blobs with
8407   missing content and trigger a NULL pointer dereference on parsing.
8408
8409   Applications that verify PKCS#7 signatures, decrypt PKCS#7 data or
8410   otherwise parse PKCS#7 structures from untrusted sources are
8411   affected. OpenSSL clients and servers are not affected.
8412
8413   This issue was reported to OpenSSL by Michal Zalewski (Google).
8414   ([CVE-2015-0289])
8415
8416   *Emilia Käsper*
8417
8418 * DoS via reachable assert in SSLv2 servers fix
8419
8420   A malicious client can trigger an OPENSSL_assert (i.e., an abort) in
8421   servers that both support SSLv2 and enable export cipher suites by sending
8422   a specially crafted SSLv2 CLIENT-MASTER-KEY message.
8423
8424   This issue was discovered by Sean Burford (Google) and Emilia Käsper
8425   (OpenSSL development team).
8426   ([CVE-2015-0293])
8427
8428   *Emilia Käsper*
8429
8430 * Use After Free following d2i_ECPrivatekey error fix
8431
8432   A malformed EC private key file consumed via the d2i_ECPrivateKey function
8433   could cause a use after free condition. This, in turn, could cause a double
8434   free in several private key parsing functions (such as d2i_PrivateKey
8435   or EVP_PKCS82PKEY) and could lead to a DoS attack or memory corruption
8436   for applications that receive EC private keys from untrusted
8437   sources. This scenario is considered rare.
8438
8439   This issue was discovered by the BoringSSL project and fixed in their
8440   commit 517073cd4b.
8441   ([CVE-2015-0209])
8442
8443   *Matt Caswell*
8444
8445 * X509_to_X509_REQ NULL pointer deref fix
8446
8447   The function X509_to_X509_REQ will crash with a NULL pointer dereference if
8448   the certificate key is invalid. This function is rarely used in practice.
8449
8450   This issue was discovered by Brian Carpenter.
8451   ([CVE-2015-0288])
8452
8453   *Stephen Henson*
8454
8455 * Removed the export ciphers from the DEFAULT ciphers
8456
8457   *Kurt Roeckx*
8458
8459### Changes between 1.0.0p and 1.0.0q [15 Jan 2015]
8460
8461 * Build fixes for the Windows and OpenVMS platforms
8462
8463   *Matt Caswell and Richard Levitte*
8464
8465### Changes between 1.0.0o and 1.0.0p [8 Jan 2015]
8466
8467 * Fix DTLS segmentation fault in dtls1_get_record. A carefully crafted DTLS
8468   message can cause a segmentation fault in OpenSSL due to a NULL pointer
8469   dereference. This could lead to a Denial Of Service attack. Thanks to
8470   Markus Stenberg of Cisco Systems, Inc. for reporting this issue.
8471   ([CVE-2014-3571])
8472
8473   *Steve Henson*
8474
8475 * Fix DTLS memory leak in dtls1_buffer_record. A memory leak can occur in the
8476   dtls1_buffer_record function under certain conditions. In particular this
8477   could occur if an attacker sent repeated DTLS records with the same
8478   sequence number but for the next epoch. The memory leak could be exploited
8479   by an attacker in a Denial of Service attack through memory exhaustion.
8480   Thanks to Chris Mueller for reporting this issue.
8481   ([CVE-2015-0206])
8482
8483   *Matt Caswell*
8484
8485 * Fix issue where no-ssl3 configuration sets method to NULL. When openssl is
8486   built with the no-ssl3 option and a SSL v3 ClientHello is received the ssl
8487   method would be set to NULL which could later result in a NULL pointer
8488   dereference. Thanks to Frank Schmirler for reporting this issue.
8489   ([CVE-2014-3569])
8490
8491   *Kurt Roeckx*
8492
8493 * Abort handshake if server key exchange message is omitted for ephemeral
8494   ECDH ciphersuites.
8495
8496   Thanks to Karthikeyan Bhargavan of the PROSECCO team at INRIA for
8497   reporting this issue.
8498   ([CVE-2014-3572])
8499
8500   *Steve Henson*
8501
8502 * Remove non-export ephemeral RSA code on client and server. This code
8503   violated the TLS standard by allowing the use of temporary RSA keys in
8504   non-export ciphersuites and could be used by a server to effectively
8505   downgrade the RSA key length used to a value smaller than the server
8506   certificate. Thanks for Karthikeyan Bhargavan of the PROSECCO team at
8507   INRIA or reporting this issue.
8508   ([CVE-2015-0204])
8509
8510   *Steve Henson*
8511
8512 * Fixed issue where DH client certificates are accepted without verification.
8513   An OpenSSL server will accept a DH certificate for client authentication
8514   without the certificate verify message. This effectively allows a client to
8515   authenticate without the use of a private key. This only affects servers
8516   which trust a client certificate authority which issues certificates
8517   containing DH keys: these are extremely rare and hardly ever encountered.
8518   Thanks for Karthikeyan Bhargavan of the PROSECCO team at INRIA or reporting
8519   this issue.
8520   ([CVE-2015-0205])
8521
8522   *Steve Henson*
8523
8524 * Correct Bignum squaring. Bignum squaring (BN_sqr) may produce incorrect
8525   results on some platforms, including x86_64. This bug occurs at random
8526   with a very low probability, and is not known to be exploitable in any
8527   way, though its exact impact is difficult to determine. Thanks to Pieter
8528   Wuille (Blockstream) who reported this issue and also suggested an initial
8529   fix. Further analysis was conducted by the OpenSSL development team and
8530   Adam Langley of Google. The final fix was developed by Andy Polyakov of
8531   the OpenSSL core team.
8532   ([CVE-2014-3570])
8533
8534   *Andy Polyakov*
8535
8536 * Fix various certificate fingerprint issues.
8537
8538   By using non-DER or invalid encodings outside the signed portion of a
8539   certificate the fingerprint can be changed without breaking the signature.
8540   Although no details of the signed portion of the certificate can be changed
8541   this can cause problems with some applications: e.g. those using the
8542   certificate fingerprint for blacklists.
8543
8544   1. Reject signatures with non zero unused bits.
8545
8546   If the BIT STRING containing the signature has non zero unused bits reject
8547   the signature. All current signature algorithms require zero unused bits.
8548
8549   2. Check certificate algorithm consistency.
8550
8551   Check the AlgorithmIdentifier inside TBS matches the one in the
8552   certificate signature. NB: this will result in signature failure
8553   errors for some broken certificates.
8554
8555   Thanks to Konrad Kraszewski from Google for reporting this issue.
8556
8557   3. Check DSA/ECDSA signatures use DER.
8558
8559   Reencode DSA/ECDSA signatures and compare with the original received
8560   signature. Return an error if there is a mismatch.
8561
8562   This will reject various cases including garbage after signature
8563   (thanks to Antti Karjalainen and Tuomo Untinen from the Codenomicon CROSS
8564   program for discovering this case) and use of BER or invalid ASN.1 INTEGERs
8565   (negative or with leading zeroes).
8566
8567   Further analysis was conducted and fixes were developed by Stephen Henson
8568   of the OpenSSL core team.
8569
8570   ([CVE-2014-8275])
8571
8572   *Steve Henson*
8573
8574### Changes between 1.0.0n and 1.0.0o [15 Oct 2014]
8575
8576 * Session Ticket Memory Leak.
8577
8578   When an OpenSSL SSL/TLS/DTLS server receives a session ticket the
8579   integrity of that ticket is first verified. In the event of a session
8580   ticket integrity check failing, OpenSSL will fail to free memory
8581   causing a memory leak. By sending a large number of invalid session
8582   tickets an attacker could exploit this issue in a Denial Of Service
8583   attack.
8584   ([CVE-2014-3567])
8585
8586   *Steve Henson*
8587
8588 * Build option no-ssl3 is incomplete.
8589
8590   When OpenSSL is configured with "no-ssl3" as a build option, servers
8591   could accept and complete a SSL 3.0 handshake, and clients could be
8592   configured to send them.
8593   ([CVE-2014-3568])
8594
8595   *Akamai and the OpenSSL team*
8596
8597 * Add support for TLS_FALLBACK_SCSV.
8598   Client applications doing fallback retries should call
8599   SSL_set_mode(s, SSL_MODE_SEND_FALLBACK_SCSV).
8600   ([CVE-2014-3566])
8601
8602   *Adam Langley, Bodo Moeller*
8603
8604 * Add additional DigestInfo checks.
8605
8606   Reencode DigestInto in DER and check against the original when
8607   verifying RSA signature: this will reject any improperly encoded
8608   DigestInfo structures.
8609
8610   Note: this is a precautionary measure and no attacks are currently known.
8611
8612   *Steve Henson*
8613
8614### Changes between 1.0.0m and 1.0.0n [6 Aug 2014]
8615
8616 * OpenSSL DTLS clients enabling anonymous (EC)DH ciphersuites are subject
8617   to a denial of service attack. A malicious server can crash the client
8618   with a null pointer dereference (read) by specifying an anonymous (EC)DH
8619   ciphersuite and sending carefully crafted handshake messages.
8620
8621   Thanks to Felix Gröbert (Google) for discovering and researching this
8622   issue.
8623   ([CVE-2014-3510])
8624
8625   *Emilia Käsper*
8626
8627 * By sending carefully crafted DTLS packets an attacker could cause openssl
8628   to leak memory. This can be exploited through a Denial of Service attack.
8629   Thanks to Adam Langley for discovering and researching this issue.
8630   ([CVE-2014-3507])
8631
8632   *Adam Langley*
8633
8634 * An attacker can force openssl to consume large amounts of memory whilst
8635   processing DTLS handshake messages. This can be exploited through a
8636   Denial of Service attack.
8637   Thanks to Adam Langley for discovering and researching this issue.
8638   ([CVE-2014-3506])
8639
8640   *Adam Langley*
8641
8642 * An attacker can force an error condition which causes openssl to crash
8643   whilst processing DTLS packets due to memory being freed twice. This
8644   can be exploited through a Denial of Service attack.
8645   Thanks to Adam Langley and Wan-Teh Chang for discovering and researching
8646   this issue.
8647   ([CVE-2014-3505])
8648
8649   *Adam Langley*
8650
8651 * If a multithreaded client connects to a malicious server using a resumed
8652   session and the server sends an ec point format extension it could write
8653   up to 255 bytes to freed memory.
8654
8655   Thanks to Gabor Tyukasz (LogMeIn Inc) for discovering and researching this
8656   issue.
8657   ([CVE-2014-3509])
8658
8659   *Gabor Tyukasz*
8660
8661 * A flaw in OBJ_obj2txt may cause pretty printing functions such as
8662   X509_name_oneline, X509_name_print_ex et al. to leak some information
8663   from the stack. Applications may be affected if they echo pretty printing
8664   output to the attacker.
8665
8666   Thanks to Ivan Fratric (Google) for discovering this issue.
8667   ([CVE-2014-3508])
8668
8669   *Emilia Käsper, and Steve Henson*
8670
8671 * Fix ec_GFp_simple_points_make_affine (thus, EC_POINTs_mul etc.)
8672   for corner cases. (Certain input points at infinity could lead to
8673   bogus results, with non-infinity inputs mapped to infinity too.)
8674
8675   *Bodo Moeller*
8676
8677### Changes between 1.0.0l and 1.0.0m [5 Jun 2014]
8678
8679 * Fix for SSL/TLS MITM flaw. An attacker using a carefully crafted
8680   handshake can force the use of weak keying material in OpenSSL
8681   SSL/TLS clients and servers.
8682
8683   Thanks to KIKUCHI Masashi (Lepidum Co. Ltd.) for discovering and
8684   researching this issue. ([CVE-2014-0224])
8685
8686   *KIKUCHI Masashi, Steve Henson*
8687
8688 * Fix DTLS recursion flaw. By sending an invalid DTLS handshake to an
8689   OpenSSL DTLS client the code can be made to recurse eventually crashing
8690   in a DoS attack.
8691
8692   Thanks to Imre Rad (Search-Lab Ltd.) for discovering this issue.
8693   ([CVE-2014-0221])
8694
8695   *Imre Rad, Steve Henson*
8696
8697 * Fix DTLS invalid fragment vulnerability. A buffer overrun attack can
8698   be triggered by sending invalid DTLS fragments to an OpenSSL DTLS
8699   client or server. This is potentially exploitable to run arbitrary
8700   code on a vulnerable client or server.
8701
8702   Thanks to Jüri Aedla for reporting this issue. ([CVE-2014-0195])
8703
8704   *Jüri Aedla, Steve Henson*
8705
8706 * Fix bug in TLS code where clients enable anonymous ECDH ciphersuites
8707   are subject to a denial of service attack.
8708
8709   Thanks to Felix Gröbert and Ivan Fratric at Google for discovering
8710   this issue. ([CVE-2014-3470])
8711
8712   *Felix Gröbert, Ivan Fratric, Steve Henson*
8713
8714 * Harmonize version and its documentation. -f flag is used to display
8715   compilation flags.
8716
8717   *mancha <mancha1@zoho.com>*
8718
8719 * Fix eckey_priv_encode so it immediately returns an error upon a failure
8720   in i2d_ECPrivateKey.
8721
8722   *mancha <mancha1@zoho.com>*
8723
8724 * Fix some double frees. These are not thought to be exploitable.
8725
8726   *mancha <mancha1@zoho.com>*
8727
8728 * Fix for the attack described in the paper "Recovering OpenSSL
8729   ECDSA Nonces Using the FLUSH+RELOAD Cache Side-channel Attack"
8730   by Yuval Yarom and Naomi Benger. Details can be obtained from:
8731   <http://eprint.iacr.org/2014/140>
8732
8733   Thanks to Yuval Yarom and Naomi Benger for discovering this
8734   flaw and to Yuval Yarom for supplying a fix ([CVE-2014-0076])
8735
8736   *Yuval Yarom and Naomi Benger*
8737
8738### Changes between 1.0.0k and 1.0.0l [6 Jan 2014]
8739
8740 * Keep original DTLS digest and encryption contexts in retransmission
8741   structures so we can use the previous session parameters if they need
8742   to be resent. ([CVE-2013-6450])
8743
8744   *Steve Henson*
8745
8746 * Add option SSL_OP_SAFARI_ECDHE_ECDSA_BUG (part of SSL_OP_ALL) which
8747   avoids preferring ECDHE-ECDSA ciphers when the client appears to be
8748   Safari on OS X.  Safari on OS X 10.8..10.8.3 advertises support for
8749   several ECDHE-ECDSA ciphers, but fails to negotiate them.  The bug
8750   is fixed in OS X 10.8.4, but Apple have ruled out both hot fixing
8751   10.8..10.8.3 and forcing users to upgrade to 10.8.4 or newer.
8752
8753   *Rob Stradling, Adam Langley*
8754
8755### Changes between 1.0.0j and 1.0.0k [5 Feb 2013]
8756
8757 * Make the decoding of SSLv3, TLS and DTLS CBC records constant time.
8758
8759   This addresses the flaw in CBC record processing discovered by
8760   Nadhem Alfardan and Kenny Paterson. Details of this attack can be found
8761   at: <http://www.isg.rhul.ac.uk/tls/>
8762
8763   Thanks go to Nadhem Alfardan and Kenny Paterson of the Information
8764   Security Group at Royal Holloway, University of London
8765   (www.isg.rhul.ac.uk) for discovering this flaw and Adam Langley and
8766   Emilia Käsper for the initial patch.
8767   ([CVE-2013-0169])
8768
8769   *Emilia Käsper, Adam Langley, Ben Laurie, Andy Polyakov, Steve Henson*
8770
8771 * Return an error when checking OCSP signatures when key is NULL.
8772   This fixes a DoS attack. ([CVE-2013-0166])
8773
8774   *Steve Henson*
8775
8776 * Call OCSP Stapling callback after ciphersuite has been chosen, so
8777   the right response is stapled. Also change SSL_get_certificate()
8778   so it returns the certificate actually sent.
8779   See <http://rt.openssl.org/Ticket/Display.html?id=2836>.
8780   (This is a backport)
8781
8782   *Rob Stradling <rob.stradling@comodo.com>*
8783
8784 * Fix possible deadlock when decoding public keys.
8785
8786   *Steve Henson*
8787
8788### Changes between 1.0.0i and 1.0.0j [10 May 2012]
8789
8790[NB: OpenSSL 1.0.0i and later 1.0.0 patch levels were released after
8791OpenSSL 1.0.1.]
8792
8793 * Sanity check record length before skipping explicit IV in DTLS
8794   to fix DoS attack.
8795
8796   Thanks to Codenomicon for discovering this issue using Fuzz-o-Matic
8797   fuzzing as a service testing platform.
8798   ([CVE-2012-2333])
8799
8800   *Steve Henson*
8801
8802 * Initialise tkeylen properly when encrypting CMS messages.
8803   Thanks to Solar Designer of Openwall for reporting this issue.
8804
8805   *Steve Henson*
8806
8807### Changes between 1.0.0h and 1.0.0i [19 Apr 2012]
8808
8809 * Check for potentially exploitable overflows in asn1_d2i_read_bio
8810   BUF_mem_grow and BUF_mem_grow_clean. Refuse attempts to shrink buffer
8811   in CRYPTO_realloc_clean.
8812
8813   Thanks to Tavis Ormandy, Google Security Team, for discovering this
8814   issue and to Adam Langley <agl@chromium.org> for fixing it.
8815   ([CVE-2012-2110])
8816
8817   *Adam Langley (Google), Tavis Ormandy, Google Security Team*
8818
8819### Changes between 1.0.0g and 1.0.0h [12 Mar 2012]
8820
8821 * Fix MMA (Bleichenbacher's attack on PKCS #1 v1.5 RSA padding) weakness
8822   in CMS and PKCS7 code. When RSA decryption fails use a random key for
8823   content decryption and always return the same error. Note: this attack
8824   needs on average 2^20 messages so it only affects automated senders. The
8825   old behaviour can be re-enabled in the CMS code by setting the
8826   CMS_DEBUG_DECRYPT flag: this is useful for debugging and testing where
8827   an MMA defence is not necessary.
8828   Thanks to Ivan Nestlerode <inestlerode@us.ibm.com> for discovering
8829   this issue. ([CVE-2012-0884])
8830
8831   *Steve Henson*
8832
8833 * Fix CVE-2011-4619: make sure we really are receiving a
8834   client hello before rejecting multiple SGC restarts. Thanks to
8835   Ivan Nestlerode <inestlerode@us.ibm.com> for discovering this bug.
8836
8837   *Steve Henson*
8838
8839### Changes between 1.0.0f and 1.0.0g [18 Jan 2012]
8840
8841 * Fix for DTLS DoS issue introduced by fix for CVE-2011-4109.
8842   Thanks to Antonio Martin, Enterprise Secure Access Research and
8843   Development, Cisco Systems, Inc. for discovering this bug and
8844   preparing a fix. ([CVE-2012-0050])
8845
8846   *Antonio Martin*
8847
8848### Changes between 1.0.0e and 1.0.0f [4 Jan 2012]
8849
8850 * Nadhem Alfardan and Kenny Paterson have discovered an extension
8851   of the Vaudenay padding oracle attack on CBC mode encryption
8852   which enables an efficient plaintext recovery attack against
8853   the OpenSSL implementation of DTLS. Their attack exploits timing
8854   differences arising during decryption processing. A research
8855   paper describing this attack can be found at:
8856   <http://www.isg.rhul.ac.uk/~kp/dtls.pdf>
8857   Thanks go to Nadhem Alfardan and Kenny Paterson of the Information
8858   Security Group at Royal Holloway, University of London
8859   (www.isg.rhul.ac.uk) for discovering this flaw and to Robin Seggelmann
8860   <seggelmann@fh-muenster.de> and Michael Tuexen <tuexen@fh-muenster.de>
8861   for preparing the fix. ([CVE-2011-4108])
8862
8863   *Robin Seggelmann, Michael Tuexen*
8864
8865 * Clear bytes used for block padding of SSL 3.0 records.
8866   ([CVE-2011-4576])
8867
8868   *Adam Langley (Google)*
8869
8870 * Only allow one SGC handshake restart for SSL/TLS. Thanks to George
8871   Kadianakis <desnacked@gmail.com> for discovering this issue and
8872   Adam Langley for preparing the fix. ([CVE-2011-4619])
8873
8874   *Adam Langley (Google)*
8875
8876 * Check parameters are not NULL in GOST ENGINE. ([CVE-2012-0027])
8877
8878   *Andrey Kulikov <amdeich@gmail.com>*
8879
8880 * Prevent malformed RFC3779 data triggering an assertion failure.
8881   Thanks to Andrew Chi, BBN Technologies, for discovering the flaw
8882   and Rob Austein <sra@hactrn.net> for fixing it. ([CVE-2011-4577])
8883
8884   *Rob Austein <sra@hactrn.net>*
8885
8886 * Improved PRNG seeding for VOS.
8887
8888   *Paul Green <Paul.Green@stratus.com>*
8889
8890 * Fix ssl_ciph.c set-up race.
8891
8892   *Adam Langley (Google)*
8893
8894 * Fix spurious failures in ecdsatest.c.
8895
8896   *Emilia Käsper (Google)*
8897
8898 * Fix the BIO_f_buffer() implementation (which was mixing different
8899   interpretations of the `..._len` fields).
8900
8901   *Adam Langley (Google)*
8902
8903 * Fix handling of BN_BLINDING: now BN_BLINDING_invert_ex (rather than
8904   BN_BLINDING_invert_ex) calls BN_BLINDING_update, ensuring that concurrent
8905   threads won't reuse the same blinding coefficients.
8906
8907   This also avoids the need to obtain the CRYPTO_LOCK_RSA_BLINDING
8908   lock to call BN_BLINDING_invert_ex, and avoids one use of
8909   BN_BLINDING_update for each BN_BLINDING structure (previously,
8910   the last update always remained unused).
8911
8912   *Emilia Käsper (Google)*
8913
8914 * In ssl3_clear, preserve s3->init_extra along with s3->rbuf.
8915
8916   *Bob Buckholz (Google)*
8917
8918### Changes between 1.0.0d and 1.0.0e [6 Sep 2011]
8919
8920 * Fix bug where CRLs with nextUpdate in the past are sometimes accepted
8921   by initialising X509_STORE_CTX properly. ([CVE-2011-3207])
8922
8923   *Kaspar Brand <ossl@velox.ch>*
8924
8925 * Fix SSL memory handling for (EC)DH ciphersuites, in particular
8926   for multi-threaded use of ECDH. ([CVE-2011-3210])
8927
8928   *Adam Langley (Google)*
8929
8930 * Fix x509_name_ex_d2i memory leak on bad inputs.
8931
8932   *Bodo Moeller*
8933
8934 * Remove hard coded ecdsaWithSHA1 signature tests in ssl code and check
8935   signature public key algorithm by using OID xref utilities instead.
8936   Before this you could only use some ECC ciphersuites with SHA1 only.
8937
8938   *Steve Henson*
8939
8940 * Add protection against ECDSA timing attacks as mentioned in the paper
8941   by Billy Bob Brumley and Nicola Tuveri, see:
8942   <http://eprint.iacr.org/2011/232.pdf>
8943
8944   *Billy Bob Brumley and Nicola Tuveri*
8945
8946### Changes between 1.0.0c and 1.0.0d [8 Feb 2011]
8947
8948 * Fix parsing of OCSP stapling ClientHello extension. CVE-2011-0014
8949
8950   *Neel Mehta, Adam Langley, Bodo Moeller (Google)*
8951
8952 * Fix bug in string printing code: if *any* escaping is enabled we must
8953   escape the escape character (backslash) or the resulting string is
8954   ambiguous.
8955
8956   *Steve Henson*
8957
8958### Changes between 1.0.0b and 1.0.0c  [2 Dec 2010]
8959
8960 * Disable code workaround for ancient and obsolete Netscape browsers
8961   and servers: an attacker can use it in a ciphersuite downgrade attack.
8962   Thanks to Martin Rex for discovering this bug. CVE-2010-4180
8963
8964   *Steve Henson*
8965
8966 * Fixed J-PAKE implementation error, originally discovered by
8967   Sebastien Martini, further info and confirmation from Stefan
8968   Arentz and Feng Hao. Note that this fix is a security fix. CVE-2010-4252
8969
8970   *Ben Laurie*
8971
8972### Changes between 1.0.0a and 1.0.0b  [16 Nov 2010]
8973
8974 * Fix extension code to avoid race conditions which can result in a buffer
8975   overrun vulnerability: resumed sessions must not be modified as they can
8976   be shared by multiple threads. CVE-2010-3864
8977
8978   *Steve Henson*
8979
8980 * Fix WIN32 build system to correctly link an ENGINE directory into
8981   a DLL.
8982
8983   *Steve Henson*
8984
8985### Changes between 1.0.0 and 1.0.0a  [01 Jun 2010]
8986
8987 * Check return value of int_rsa_verify in pkey_rsa_verifyrecover
8988   ([CVE-2010-1633])
8989
8990   *Steve Henson, Peter-Michael Hager <hager@dortmund.net>*
8991
8992### Changes between 0.9.8n and 1.0.0  [29 Mar 2010]
8993
8994 * Add "missing" function EVP_CIPHER_CTX_copy(). This copies a cipher
8995   context. The operation can be customised via the ctrl mechanism in
8996   case ENGINEs want to include additional functionality.
8997
8998   *Steve Henson*
8999
9000 * Tolerate yet another broken PKCS#8 key format: private key value negative.
9001
9002   *Steve Henson*
9003
9004 * Add new -subject_hash_old and -issuer_hash_old options to x509 utility to
9005   output hashes compatible with older versions of OpenSSL.
9006
9007   *Willy Weisz <weisz@vcpc.univie.ac.at>*
9008
9009 * Fix compression algorithm handling: if resuming a session use the
9010   compression algorithm of the resumed session instead of determining
9011   it from client hello again. Don't allow server to change algorithm.
9012
9013   *Steve Henson*
9014
9015 * Add load_crls() function to commands tidying load_certs() too. Add option
9016   to verify utility to allow additional CRLs to be included.
9017
9018   *Steve Henson*
9019
9020 * Update OCSP request code to permit adding custom headers to the request:
9021   some responders need this.
9022
9023   *Steve Henson*
9024
9025 * The function EVP_PKEY_sign() returns <=0 on error: check return code
9026   correctly.
9027
9028   *Julia Lawall <julia@diku.dk>*
9029
9030 * Update verify callback code in `apps/s_cb.c` and `apps/verify.c`, it
9031   needlessly dereferenced structures, used obsolete functions and
9032   didn't handle all updated verify codes correctly.
9033
9034   *Steve Henson*
9035
9036 * Disable MD2 in the default configuration.
9037
9038   *Steve Henson*
9039
9040 * In BIO_pop() and BIO_push() use the ctrl argument (which was NULL) to
9041   indicate the initial BIO being pushed or popped. This makes it possible
9042   to determine whether the BIO is the one explicitly called or as a result
9043   of the ctrl being passed down the chain. Fix BIO_pop() and SSL BIOs so
9044   it handles reference counts correctly and doesn't zero out the I/O bio
9045   when it is not being explicitly popped. WARNING: applications which
9046   included workarounds for the old buggy behaviour will need to be modified
9047   or they could free up already freed BIOs.
9048
9049   *Steve Henson*
9050
9051 * Extend the uni2asc/asc2uni => OPENSSL_uni2asc/OPENSSL_asc2uni
9052   renaming to all platforms (within the 0.9.8 branch, this was
9053   done conditionally on Netware platforms to avoid a name clash).
9054
9055   *Guenter <lists@gknw.net>*
9056
9057 * Add ECDHE and PSK support to DTLS.
9058
9059   *Michael Tuexen <tuexen@fh-muenster.de>*
9060
9061 * Add CHECKED_STACK_OF macro to safestack.h, otherwise safestack can't
9062   be used on C++.
9063
9064   *Steve Henson*
9065
9066 * Add "missing" function EVP_MD_flags() (without this the only way to
9067   retrieve a digest flags is by accessing the structure directly. Update
9068   `EVP_MD_do_all*()` and `EVP_CIPHER_do_all*()` to include the name a digest
9069   or cipher is registered as in the "from" argument. Print out all
9070   registered digests in the dgst usage message instead of manually
9071   attempting to work them out.
9072
9073   *Steve Henson*
9074
9075 * If no SSLv2 ciphers are used don't use an SSLv2 compatible client hello:
9076   this allows the use of compression and extensions. Change default cipher
9077   string to remove SSLv2 ciphersuites. This effectively avoids ancient SSLv2
9078   by default unless an application cipher string requests it.
9079
9080   *Steve Henson*
9081
9082 * Alter match criteria in PKCS12_parse(). It used to try to use local
9083   key ids to find matching certificates and keys but some PKCS#12 files
9084   don't follow the (somewhat unwritten) rules and this strategy fails.
9085   Now just gather all certificates together and the first private key
9086   then look for the first certificate that matches the key.
9087
9088   *Steve Henson*
9089
9090 * Support use of registered digest and cipher names for dgst and cipher
9091   commands instead of having to add each one as a special case. So now
9092   you can do:
9093
9094           openssl sha256 foo
9095
9096   as well as:
9097
9098           openssl dgst -sha256 foo
9099
9100   and this works for ENGINE based algorithms too.
9101
9102   *Steve Henson*
9103
9104 * Update Gost ENGINE to support parameter files.
9105
9106   *Victor B. Wagner <vitus@cryptocom.ru>*
9107
9108 * Support GeneralizedTime in ca utility.
9109
9110   *Oliver Martin <oliver@volatilevoid.net>, Steve Henson*
9111
9112 * Enhance the hash format used for certificate directory links. The new
9113   form uses the canonical encoding (meaning equivalent names will work
9114   even if they aren't identical) and uses SHA1 instead of MD5. This form
9115   is incompatible with the older format and as a result c_rehash should
9116   be used to rebuild symbolic links.
9117
9118   *Steve Henson*
9119
9120 * Make PKCS#8 the default write format for private keys, replacing the
9121   traditional format. This form is standardised, more secure and doesn't
9122   include an implicit MD5 dependency.
9123
9124   *Steve Henson*
9125
9126 * Add a $gcc_devteam_warn option to Configure. The idea is that any code
9127   committed to OpenSSL should pass this lot as a minimum.
9128
9129   *Steve Henson*
9130
9131 * Add session ticket override functionality for use by EAP-FAST.
9132
9133   *Jouni Malinen <j@w1.fi>*
9134
9135 * Modify HMAC functions to return a value. Since these can be implemented
9136   in an ENGINE errors can occur.
9137
9138   *Steve Henson*
9139
9140 * Type-checked OBJ_bsearch_ex.
9141
9142   *Ben Laurie*
9143
9144 * Type-checked OBJ_bsearch. Also some constification necessitated
9145   by type-checking.  Still to come: TXT_DB, bsearch(?),
9146   OBJ_bsearch_ex, qsort, CRYPTO_EX_DATA, ASN1_VALUE, ASN1_STRING,
9147   CONF_VALUE.
9148
9149   *Ben Laurie*
9150
9151 * New function OPENSSL_gmtime_adj() to add a specific number of days and
9152   seconds to a tm structure directly, instead of going through OS
9153   specific date routines. This avoids any issues with OS routines such
9154   as the year 2038 bug. New `*_adj()` functions for ASN1 time structures
9155   and X509_time_adj_ex() to cover the extended range. The existing
9156   X509_time_adj() is still usable and will no longer have any date issues.
9157
9158   *Steve Henson*
9159
9160 * Delta CRL support. New use deltas option which will attempt to locate
9161   and search any appropriate delta CRLs available.
9162
9163   This work was sponsored by Google.
9164
9165   *Steve Henson*
9166
9167 * Support for CRLs partitioned by reason code. Reorganise CRL processing
9168   code and add additional score elements. Validate alternate CRL paths
9169   as part of the CRL checking and indicate a new error "CRL path validation
9170   error" in this case. Applications wanting additional details can use
9171   the verify callback and check the new "parent" field. If this is not
9172   NULL CRL path validation is taking place. Existing applications won't
9173   see this because it requires extended CRL support which is off by
9174   default.
9175
9176   This work was sponsored by Google.
9177
9178   *Steve Henson*
9179
9180 * Support for freshest CRL extension.
9181
9182   This work was sponsored by Google.
9183
9184   *Steve Henson*
9185
9186 * Initial indirect CRL support. Currently only supported in the CRLs
9187   passed directly and not via lookup. Process certificate issuer
9188   CRL entry extension and lookup CRL entries by bother issuer name
9189   and serial number. Check and process CRL issuer entry in IDP extension.
9190
9191   This work was sponsored by Google.
9192
9193   *Steve Henson*
9194
9195 * Add support for distinct certificate and CRL paths. The CRL issuer
9196   certificate is validated separately in this case. Only enabled if
9197   an extended CRL support flag is set: this flag will enable additional
9198   CRL functionality in future.
9199
9200   This work was sponsored by Google.
9201
9202   *Steve Henson*
9203
9204 * Add support for policy mappings extension.
9205
9206   This work was sponsored by Google.
9207
9208   *Steve Henson*
9209
9210 * Fixes to pathlength constraint, self issued certificate handling,
9211   policy processing to align with RFC3280 and PKITS tests.
9212
9213   This work was sponsored by Google.
9214
9215   *Steve Henson*
9216
9217 * Support for name constraints certificate extension. DN, email, DNS
9218   and URI types are currently supported.
9219
9220   This work was sponsored by Google.
9221
9222   *Steve Henson*
9223
9224 * To cater for systems that provide a pointer-based thread ID rather
9225   than numeric, deprecate the current numeric thread ID mechanism and
9226   replace it with a structure and associated callback type. This
9227   mechanism allows a numeric "hash" to be extracted from a thread ID in
9228   either case, and on platforms where pointers are larger than 'long',
9229   mixing is done to help ensure the numeric 'hash' is usable even if it
9230   can't be guaranteed unique. The default mechanism is to use "&errno"
9231   as a pointer-based thread ID to distinguish between threads.
9232
9233   Applications that want to provide their own thread IDs should now use
9234   CRYPTO_THREADID_set_callback() to register a callback that will call
9235   either CRYPTO_THREADID_set_numeric() or CRYPTO_THREADID_set_pointer().
9236
9237   Note that ERR_remove_state() is now deprecated, because it is tied
9238   to the assumption that thread IDs are numeric.  ERR_remove_state(0)
9239   to free the current thread's error state should be replaced by
9240   ERR_remove_thread_state(NULL).
9241
9242   (This new approach replaces the functions CRYPTO_set_idptr_callback(),
9243   CRYPTO_get_idptr_callback(), and CRYPTO_thread_idptr() that existed in
9244   OpenSSL 0.9.9-dev between June 2006 and August 2008. Also, if an
9245   application was previously providing a numeric thread callback that
9246   was inappropriate for distinguishing threads, then uniqueness might
9247   have been obtained with &errno that happened immediately in the
9248   intermediate development versions of OpenSSL; this is no longer the
9249   case, the numeric thread callback will now override the automatic use
9250   of &errno.)
9251
9252   *Geoff Thorpe, with help from Bodo Moeller*
9253
9254 * Initial support for different CRL issuing certificates. This covers a
9255   simple case where the self issued certificates in the chain exist and
9256   the real CRL issuer is higher in the existing chain.
9257
9258   This work was sponsored by Google.
9259
9260   *Steve Henson*
9261
9262 * Removed effectively defunct crypto/store from the build.
9263
9264   *Ben Laurie*
9265
9266 * Revamp of STACK to provide stronger type-checking. Still to come:
9267   TXT_DB, bsearch(?), OBJ_bsearch, qsort, CRYPTO_EX_DATA, ASN1_VALUE,
9268   ASN1_STRING, CONF_VALUE.
9269
9270   *Ben Laurie*
9271
9272 * Add a new SSL_MODE_RELEASE_BUFFERS mode flag to release unused buffer
9273   RAM on SSL connections.  This option can save about 34k per idle SSL.
9274
9275   *Nick Mathewson*
9276
9277 * Revamp of LHASH to provide stronger type-checking. Still to come:
9278   STACK, TXT_DB, bsearch, qsort.
9279
9280   *Ben Laurie*
9281
9282 * Initial support for Cryptographic Message Syntax (aka CMS) based
9283   on RFC3850, RFC3851 and RFC3852. New cms directory and cms utility,
9284   support for data, signedData, compressedData, digestedData and
9285   encryptedData, envelopedData types included. Scripts to check against
9286   RFC4134 examples draft and interop and consistency checks of many
9287   content types and variants.
9288
9289   *Steve Henson*
9290
9291 * Add options to enc utility to support use of zlib compression BIO.
9292
9293   *Steve Henson*
9294
9295 * Extend mk1mf to support importing of options and assembly language
9296   files from Configure script, currently only included in VC-WIN32.
9297   The assembly language rules can now optionally generate the source
9298   files from the associated perl scripts.
9299
9300   *Steve Henson*
9301
9302 * Implement remaining functionality needed to support GOST ciphersuites.
9303   Interop testing has been performed using CryptoPro implementations.
9304
9305   *Victor B. Wagner <vitus@cryptocom.ru>*
9306
9307 * s390x assembler pack.
9308
9309   *Andy Polyakov*
9310
9311 * ARMv4 assembler pack. ARMv4 refers to v4 and later ISA, not CPU
9312   "family."
9313
9314   *Andy Polyakov*
9315
9316 * Implement Opaque PRF Input TLS extension as specified in
9317   draft-rescorla-tls-opaque-prf-input-00.txt.  Since this is not an
9318   official specification yet and no extension type assignment by
9319   IANA exists, this extension (for now) will have to be explicitly
9320   enabled when building OpenSSL by providing the extension number
9321   to use.  For example, specify an option
9322
9323           -DTLSEXT_TYPE_opaque_prf_input=0x9527
9324
9325   to the "config" or "Configure" script to enable the extension,
9326   assuming extension number 0x9527 (which is a completely arbitrary
9327   and unofficial assignment based on the MD5 hash of the Internet
9328   Draft).  Note that by doing so, you potentially lose
9329   interoperability with other TLS implementations since these might
9330   be using the same extension number for other purposes.
9331
9332   SSL_set_tlsext_opaque_prf_input(ssl, src, len) is used to set the
9333   opaque PRF input value to use in the handshake.  This will create
9334   an internal copy of the length-'len' string at 'src', and will
9335   return non-zero for success.
9336
9337   To get more control and flexibility, provide a callback function
9338   by using
9339
9340           SSL_CTX_set_tlsext_opaque_prf_input_callback(ctx, cb)
9341           SSL_CTX_set_tlsext_opaque_prf_input_callback_arg(ctx, arg)
9342
9343   where
9344
9345           int (*cb)(SSL *, void *peerinput, size_t len, void *arg);
9346           void *arg;
9347
9348   Callback function 'cb' will be called in handshakes, and is
9349   expected to use SSL_set_tlsext_opaque_prf_input() as appropriate.
9350   Argument 'arg' is for application purposes (the value as given to
9351   SSL_CTX_set_tlsext_opaque_prf_input_callback_arg() will directly
9352   be provided to the callback function).  The callback function
9353   has to return non-zero to report success: usually 1 to use opaque
9354   PRF input just if possible, or 2 to enforce use of the opaque PRF
9355   input.  In the latter case, the library will abort the handshake
9356   if opaque PRF input is not successfully negotiated.
9357
9358   Arguments 'peerinput' and 'len' given to the callback function
9359   will always be NULL and 0 in the case of a client.  A server will
9360   see the client's opaque PRF input through these variables if
9361   available (NULL and 0 otherwise).  Note that if the server
9362   provides an opaque PRF input, the length must be the same as the
9363   length of the client's opaque PRF input.
9364
9365   Note that the callback function will only be called when creating
9366   a new session (session resumption can resume whatever was
9367   previously negotiated), and will not be called in SSL 2.0
9368   handshakes; thus, SSL_CTX_set_options(ctx, SSL_OP_NO_SSLv2) or
9369   SSL_set_options(ssl, SSL_OP_NO_SSLv2) is especially recommended
9370   for applications that need to enforce opaque PRF input.
9371
9372   *Bodo Moeller*
9373
9374 * Update ssl code to support digests other than SHA1+MD5 for handshake
9375   MAC.
9376
9377   *Victor B. Wagner <vitus@cryptocom.ru>*
9378
9379 * Add RFC4507 support to OpenSSL. This includes the corrections in
9380   RFC4507bis. The encrypted ticket format is an encrypted encoded
9381   SSL_SESSION structure, that way new session features are automatically
9382   supported.
9383
9384   If a client application caches session in an SSL_SESSION structure
9385   support is transparent because tickets are now stored in the encoded
9386   SSL_SESSION.
9387
9388   The SSL_CTX structure automatically generates keys for ticket
9389   protection in servers so again support should be possible
9390   with no application modification.
9391
9392   If a client or server wishes to disable RFC4507 support then the option
9393   SSL_OP_NO_TICKET can be set.
9394
9395   Add a TLS extension debugging callback to allow the contents of any client
9396   or server extensions to be examined.
9397
9398   This work was sponsored by Google.
9399
9400   *Steve Henson*
9401
9402 * Final changes to avoid use of pointer pointer casts in OpenSSL.
9403   OpenSSL should now compile cleanly on gcc 4.2
9404
9405   *Peter Hartley <pdh@utter.chaos.org.uk>, Steve Henson*
9406
9407 * Update SSL library to use new EVP_PKEY MAC API. Include generic MAC
9408   support including streaming MAC support: this is required for GOST
9409   ciphersuite support.
9410
9411   *Victor B. Wagner <vitus@cryptocom.ru>, Steve Henson*
9412
9413 * Add option -stream to use PKCS#7 streaming in smime utility. New
9414   function i2d_PKCS7_bio_stream() and PEM_write_PKCS7_bio_stream()
9415   to output in BER and PEM format.
9416
9417   *Steve Henson*
9418
9419 * Experimental support for use of HMAC via EVP_PKEY interface. This
9420   allows HMAC to be handled via the `EVP_DigestSign*()` interface. The
9421   EVP_PKEY "key" in this case is the HMAC key, potentially allowing
9422   ENGINE support for HMAC keys which are unextractable. New -mac and
9423   -macopt options to dgst utility.
9424
9425   *Steve Henson*
9426
9427 * New option -sigopt to dgst utility. Update dgst to use
9428   `EVP_Digest{Sign,Verify}*`. These two changes make it possible to use
9429   alternative signing parameters such as X9.31 or PSS in the dgst
9430   utility.
9431
9432   *Steve Henson*
9433
9434 * Change ssl_cipher_apply_rule(), the internal function that does
9435   the work each time a ciphersuite string requests enabling
9436   ("foo+bar"), moving ("+foo+bar"), disabling ("-foo+bar", or
9437   removing ("!foo+bar") a class of ciphersuites: Now it maintains
9438   the order of disabled ciphersuites such that those ciphersuites
9439   that most recently went from enabled to disabled not only stay
9440   in order with respect to each other, but also have higher priority
9441   than other disabled ciphersuites the next time ciphersuites are
9442   enabled again.
9443
9444   This means that you can now say, e.g., "PSK:-PSK:HIGH" to enable
9445   the same ciphersuites as with "HIGH" alone, but in a specific
9446   order where the PSK ciphersuites come first (since they are the
9447   most recently disabled ciphersuites when "HIGH" is parsed).
9448
9449   Also, change ssl_create_cipher_list() (using this new
9450   functionality) such that between otherwise identical
9451   ciphersuites, ephemeral ECDH is preferred over ephemeral DH in
9452   the default order.
9453
9454   *Bodo Moeller*
9455
9456 * Change ssl_create_cipher_list() so that it automatically
9457   arranges the ciphersuites in reasonable order before starting
9458   to process the rule string.  Thus, the definition for "DEFAULT"
9459   (SSL_DEFAULT_CIPHER_LIST) now is just "ALL:!aNULL:!eNULL", but
9460   remains equivalent to `"AES:ALL:!aNULL:!eNULL:+aECDH:+kRSA:+RC4:@STRENGTH"`.
9461   This makes it much easier to arrive at a reasonable default order
9462   in applications for which anonymous ciphers are OK (meaning
9463   that you can't actually use DEFAULT).
9464
9465   *Bodo Moeller; suggested by Victor Duchovni*
9466
9467 * Split the SSL/TLS algorithm mask (as used for ciphersuite string
9468   processing) into multiple integers instead of setting
9469   "SSL_MKEY_MASK" bits, "SSL_AUTH_MASK" bits, "SSL_ENC_MASK",
9470   "SSL_MAC_MASK", and "SSL_SSL_MASK" bits all in a single integer.
9471   (These masks as well as the individual bit definitions are hidden
9472   away into the non-exported interface ssl/ssl_locl.h, so this
9473   change to the definition of the SSL_CIPHER structure shouldn't
9474   affect applications.)  This give us more bits for each of these
9475   categories, so there is no longer a need to coagulate AES128 and
9476   AES256 into a single algorithm bit, and to coagulate Camellia128
9477   and Camellia256 into a single algorithm bit, which has led to all
9478   kinds of kludges.
9479
9480   Thus, among other things, the kludge introduced in 0.9.7m and
9481   0.9.8e for masking out AES256 independently of AES128 or masking
9482   out Camellia256 independently of AES256 is not needed here in 0.9.9.
9483
9484   With the change, we also introduce new ciphersuite aliases that
9485   so far were missing: "AES128", "AES256", "CAMELLIA128", and
9486   "CAMELLIA256".
9487
9488   *Bodo Moeller*
9489
9490 * Add support for dsa-with-SHA224 and dsa-with-SHA256.
9491   Use the leftmost N bytes of the signature input if the input is
9492   larger than the prime q (with N being the size in bytes of q).
9493
9494   *Nils Larsch*
9495
9496 * Very *very* experimental PKCS#7 streaming encoder support. Nothing uses
9497   it yet and it is largely untested.
9498
9499   *Steve Henson*
9500
9501 * Add support for the ecdsa-with-SHA224/256/384/512 signature types.
9502
9503   *Nils Larsch*
9504
9505 * Initial incomplete changes to avoid need for function casts in OpenSSL
9506   some compilers (gcc 4.2 and later) reject their use. Safestack is
9507   reimplemented.  Update ASN1 to avoid use of legacy functions.
9508
9509   *Steve Henson*
9510
9511 * Win32/64 targets are linked with Winsock2.
9512
9513   *Andy Polyakov*
9514
9515 * Add an X509_CRL_METHOD structure to allow CRL processing to be redirected
9516   to external functions. This can be used to increase CRL handling
9517   efficiency especially when CRLs are very large by (for example) storing
9518   the CRL revoked certificates in a database.
9519
9520   *Steve Henson*
9521
9522 * Overhaul of by_dir code. Add support for dynamic loading of CRLs so
9523   new CRLs added to a directory can be used. New command line option
9524   -verify_return_error to s_client and s_server. This causes real errors
9525   to be returned by the verify callback instead of carrying on no matter
9526   what. This reflects the way a "real world" verify callback would behave.
9527
9528   *Steve Henson*
9529
9530 * GOST engine, supporting several GOST algorithms and public key formats.
9531   Kindly donated by Cryptocom.
9532
9533   *Cryptocom*
9534
9535 * Partial support for Issuing Distribution Point CRL extension. CRLs
9536   partitioned by DP are handled but no indirect CRL or reason partitioning
9537   (yet). Complete overhaul of CRL handling: now the most suitable CRL is
9538   selected via a scoring technique which handles IDP and AKID in CRLs.
9539
9540   *Steve Henson*
9541
9542 * New X509_STORE_CTX callbacks lookup_crls() and lookup_certs() which
9543   will ultimately be used for all verify operations: this will remove the
9544   X509_STORE dependency on certificate verification and allow alternative
9545   lookup methods.  X509_STORE based implementations of these two callbacks.
9546
9547   *Steve Henson*
9548
9549 * Allow multiple CRLs to exist in an X509_STORE with matching issuer names.
9550   Modify get_crl() to find a valid (unexpired) CRL if possible.
9551
9552   *Steve Henson*
9553
9554 * New function X509_CRL_match() to check if two CRLs are identical. Normally
9555   this would be called X509_CRL_cmp() but that name is already used by
9556   a function that just compares CRL issuer names. Cache several CRL
9557   extensions in X509_CRL structure and cache CRLDP in X509.
9558
9559   *Steve Henson*
9560
9561 * Store a "canonical" representation of X509_NAME structure (ASN1 Name)
9562   this maps equivalent X509_NAME structures into a consistent structure.
9563   Name comparison can then be performed rapidly using memcmp().
9564
9565   *Steve Henson*
9566
9567 * Non-blocking OCSP request processing. Add -timeout option to ocsp
9568   utility.
9569
9570   *Steve Henson*
9571
9572 * Allow digests to supply their own micalg string for S/MIME type using
9573   the ctrl EVP_MD_CTRL_MICALG.
9574
9575   *Steve Henson*
9576
9577 * During PKCS7 signing pass the PKCS7 SignerInfo structure to the
9578   EVP_PKEY_METHOD before and after signing via the EVP_PKEY_CTRL_PKCS7_SIGN
9579   ctrl. It can then customise the structure before and/or after signing
9580   if necessary.
9581
9582   *Steve Henson*
9583
9584 * New function OBJ_add_sigid() to allow application defined signature OIDs
9585   to be added to OpenSSLs internal tables. New function OBJ_sigid_free()
9586   to free up any added signature OIDs.
9587
9588   *Steve Henson*
9589
9590 * New functions EVP_CIPHER_do_all(), EVP_CIPHER_do_all_sorted(),
9591   EVP_MD_do_all() and EVP_MD_do_all_sorted() to enumerate internal
9592   digest and cipher tables. New options added to openssl utility:
9593   list-message-digest-algorithms and list-cipher-algorithms.
9594
9595   *Steve Henson*
9596
9597 * Change the array representation of binary polynomials: the list
9598   of degrees of non-zero coefficients is now terminated with -1.
9599   Previously it was terminated with 0, which was also part of the
9600   value; thus, the array representation was not applicable to
9601   polynomials where t^0 has coefficient zero.  This change makes
9602   the array representation useful in a more general context.
9603
9604   *Douglas Stebila*
9605
9606 * Various modifications and fixes to SSL/TLS cipher string
9607   handling.  For ECC, the code now distinguishes between fixed ECDH
9608   with RSA certificates on the one hand and with ECDSA certificates
9609   on the other hand, since these are separate ciphersuites.  The
9610   unused code for Fortezza ciphersuites has been removed.
9611
9612   For consistency with EDH, ephemeral ECDH is now called "EECDH"
9613   (not "ECDHE").  For consistency with the code for DH
9614   certificates, use of ECDH certificates is now considered ECDH
9615   authentication, not RSA or ECDSA authentication (the latter is
9616   merely the CA's signing algorithm and not actively used in the
9617   protocol).
9618
9619   The temporary ciphersuite alias "ECCdraft" is no longer
9620   available, and ECC ciphersuites are no longer excluded from "ALL"
9621   and "DEFAULT".  The following aliases now exist for RFC 4492
9622   ciphersuites, most of these by analogy with the DH case:
9623
9624           kECDHr   - ECDH cert, signed with RSA
9625           kECDHe   - ECDH cert, signed with ECDSA
9626           kECDH    - ECDH cert (signed with either RSA or ECDSA)
9627           kEECDH   - ephemeral ECDH
9628           ECDH     - ECDH cert or ephemeral ECDH
9629
9630           aECDH    - ECDH cert
9631           aECDSA   - ECDSA cert
9632           ECDSA    - ECDSA cert
9633
9634           AECDH    - anonymous ECDH
9635           EECDH    - non-anonymous ephemeral ECDH (equivalent to "kEECDH:-AECDH")
9636
9637   *Bodo Moeller*
9638
9639 * Add additional S/MIME capabilities for AES and GOST ciphers if supported.
9640   Use correct micalg parameters depending on digest(s) in signed message.
9641
9642   *Steve Henson*
9643
9644 * Add engine support for EVP_PKEY_ASN1_METHOD. Add functions to process
9645   an ENGINE asn1 method. Support ENGINE lookups in the ASN1 code.
9646
9647   *Steve Henson*
9648
9649 * Initial engine support for EVP_PKEY_METHOD. New functions to permit
9650   an engine to register a method. Add ENGINE lookups for methods and
9651   functional reference processing.
9652
9653   *Steve Henson*
9654
9655 * New functions `EVP_Digest{Sign,Verify)*`. These are enhanced versions of
9656   `EVP_{Sign,Verify}*` which allow an application to customise the signature
9657   process.
9658
9659   *Steve Henson*
9660
9661 * New -resign option to smime utility. This adds one or more signers
9662   to an existing PKCS#7 signedData structure. Also -md option to use an
9663   alternative message digest algorithm for signing.
9664
9665   *Steve Henson*
9666
9667 * Tidy up PKCS#7 routines and add new functions to make it easier to
9668   create PKCS7 structures containing multiple signers. Update smime
9669   application to support multiple signers.
9670
9671   *Steve Henson*
9672
9673 * New -macalg option to pkcs12 utility to allow setting of an alternative
9674   digest MAC.
9675
9676   *Steve Henson*
9677
9678 * Initial support for PKCS#5 v2.0 PRFs other than default SHA1 HMAC.
9679   Reorganize PBE internals to lookup from a static table using NIDs,
9680   add support for HMAC PBE OID translation. Add a EVP_CIPHER ctrl:
9681   EVP_CTRL_PBE_PRF_NID this allows a cipher to specify an alternative
9682   PRF which will be automatically used with PBES2.
9683
9684   *Steve Henson*
9685
9686 * Replace the algorithm specific calls to generate keys in "req" with the
9687   new API.
9688
9689   *Steve Henson*
9690
9691 * Update PKCS#7 enveloped data routines to use new API. This is now
9692   supported by any public key method supporting the encrypt operation. A
9693   ctrl is added to allow the public key algorithm to examine or modify
9694   the PKCS#7 RecipientInfo structure if it needs to: for RSA this is
9695   a no op.
9696
9697   *Steve Henson*
9698
9699 * Add a ctrl to asn1 method to allow a public key algorithm to express
9700   a default digest type to use. In most cases this will be SHA1 but some
9701   algorithms (such as GOST) need to specify an alternative digest. The
9702   return value indicates how strong the preference is 1 means optional and
9703   2 is mandatory (that is it is the only supported type). Modify
9704   ASN1_item_sign() to accept a NULL digest argument to indicate it should
9705   use the default md. Update openssl utilities to use the default digest
9706   type for signing if it is not explicitly indicated.
9707
9708   *Steve Henson*
9709
9710 * Use OID cross reference table in ASN1_sign() and ASN1_verify(). New
9711   EVP_MD flag EVP_MD_FLAG_PKEY_METHOD_SIGNATURE. This uses the relevant
9712   signing method from the key type. This effectively removes the link
9713   between digests and public key types.
9714
9715   *Steve Henson*
9716
9717 * Add an OID cross reference table and utility functions. Its purpose is to
9718   translate between signature OIDs such as SHA1WithrsaEncryption and SHA1,
9719   rsaEncryption. This will allow some of the algorithm specific hackery
9720   needed to use the correct OID to be removed.
9721
9722   *Steve Henson*
9723
9724 * Remove algorithm specific dependencies when setting PKCS7_SIGNER_INFO
9725   structures for PKCS7_sign(). They are now set up by the relevant public
9726   key ASN1 method.
9727
9728   *Steve Henson*
9729
9730 * Add provisional EC pkey method with support for ECDSA and ECDH.
9731
9732   *Steve Henson*
9733
9734 * Add support for key derivation (agreement) in the API, DH method and
9735   pkeyutl.
9736
9737   *Steve Henson*
9738
9739 * Add DSA pkey method and DH pkey methods, extend DH ASN1 method to support
9740   public and private key formats. As a side effect these add additional
9741   command line functionality not previously available: DSA signatures can be
9742   generated and verified using pkeyutl and DH key support and generation in
9743   pkey, genpkey.
9744
9745   *Steve Henson*
9746
9747 * BeOS support.
9748
9749   *Oliver Tappe <zooey@hirschkaefer.de>*
9750
9751 * New make target "install_html_docs" installs HTML renditions of the
9752   manual pages.
9753
9754   *Oliver Tappe <zooey@hirschkaefer.de>*
9755
9756 * New utility "genpkey" this is analogous to "genrsa" etc except it can
9757   generate keys for any algorithm. Extend and update EVP_PKEY_METHOD to
9758   support key and parameter generation and add initial key generation
9759   functionality for RSA.
9760
9761   *Steve Henson*
9762
9763 * Add functions for main EVP_PKEY_method operations. The undocumented
9764   functions `EVP_PKEY_{encrypt,decrypt}` have been renamed to
9765   `EVP_PKEY_{encrypt,decrypt}_old`.
9766
9767   *Steve Henson*
9768
9769 * Initial definitions for EVP_PKEY_METHOD. This will be a high level public
9770   key API, doesn't do much yet.
9771
9772   *Steve Henson*
9773
9774 * New function EVP_PKEY_asn1_get0_info() to retrieve information about
9775   public key algorithms. New option to openssl utility:
9776   "list-public-key-algorithms" to print out info.
9777
9778   *Steve Henson*
9779
9780 * Implement the Supported Elliptic Curves Extension for
9781   ECC ciphersuites from draft-ietf-tls-ecc-12.txt.
9782
9783   *Douglas Stebila*
9784
9785 * Don't free up OIDs in OBJ_cleanup() if they are in use by EVP_MD or
9786   EVP_CIPHER structures to avoid later problems in EVP_cleanup().
9787
9788   *Steve Henson*
9789
9790 * New utilities pkey and pkeyparam. These are similar to algorithm specific
9791   utilities such as rsa, dsa, dsaparam etc except they process any key
9792   type.
9793
9794   *Steve Henson*
9795
9796 * Transfer public key printing routines to EVP_PKEY_ASN1_METHOD. New
9797   functions EVP_PKEY_print_public(), EVP_PKEY_print_private(),
9798   EVP_PKEY_print_param() to print public key data from an EVP_PKEY
9799   structure.
9800
9801   *Steve Henson*
9802
9803 * Initial support for pluggable public key ASN1.
9804   De-spaghettify the public key ASN1 handling. Move public and private
9805   key ASN1 handling to a new EVP_PKEY_ASN1_METHOD structure. Relocate
9806   algorithm specific handling to a single module within the relevant
9807   algorithm directory. Add functions to allow (near) opaque processing
9808   of public and private key structures.
9809
9810   *Steve Henson*
9811
9812 * Implement the Supported Point Formats Extension for
9813   ECC ciphersuites from draft-ietf-tls-ecc-12.txt.
9814
9815   *Douglas Stebila*
9816
9817 * Add initial support for RFC 4279 PSK TLS ciphersuites. Add members
9818   for the psk identity [hint] and the psk callback functions to the
9819   SSL_SESSION, SSL and SSL_CTX structure.
9820
9821   New ciphersuites:
9822           PSK-RC4-SHA, PSK-3DES-EDE-CBC-SHA, PSK-AES128-CBC-SHA,
9823           PSK-AES256-CBC-SHA
9824
9825   New functions:
9826           SSL_CTX_use_psk_identity_hint
9827           SSL_get_psk_identity_hint
9828           SSL_get_psk_identity
9829           SSL_use_psk_identity_hint
9830
9831   *Mika Kousa and Pasi Eronen of Nokia Corporation*
9832
9833 * Add RFC 3161 compliant time stamp request creation, response generation
9834   and response verification functionality.
9835
9836   *Zoltán Glózik <zglozik@opentsa.org>, The OpenTSA Project*
9837
9838 * Add initial support for TLS extensions, specifically for the server_name
9839   extension so far.  The SSL_SESSION, SSL_CTX, and SSL data structures now
9840   have new members for a host name.  The SSL data structure has an
9841   additional member `SSL_CTX *initial_ctx` so that new sessions can be
9842   stored in that context to allow for session resumption, even after the
9843   SSL has been switched to a new SSL_CTX in reaction to a client's
9844   server_name extension.
9845
9846   New functions (subject to change):
9847
9848           SSL_get_servername()
9849           SSL_get_servername_type()
9850           SSL_set_SSL_CTX()
9851
9852   New CTRL codes and macros (subject to change):
9853
9854           SSL_CTRL_SET_TLSEXT_SERVERNAME_CB
9855                               - SSL_CTX_set_tlsext_servername_callback()
9856           SSL_CTRL_SET_TLSEXT_SERVERNAME_ARG
9857                                    - SSL_CTX_set_tlsext_servername_arg()
9858           SSL_CTRL_SET_TLSEXT_HOSTNAME           - SSL_set_tlsext_host_name()
9859
9860   openssl s_client has a new '-servername ...' option.
9861
9862   openssl s_server has new options '-servername_host ...', '-cert2 ...',
9863   '-key2 ...', '-servername_fatal' (subject to change).  This allows
9864   testing the HostName extension for a specific single host name ('-cert'
9865   and '-key' remain fallbacks for handshakes without HostName
9866   negotiation).  If the unrecognized_name alert has to be sent, this by
9867   default is a warning; it becomes fatal with the '-servername_fatal'
9868   option.
9869
9870   *Peter Sylvester,  Remy Allais, Christophe Renou*
9871
9872 * Whirlpool hash implementation is added.
9873
9874   *Andy Polyakov*
9875
9876 * BIGNUM code on 64-bit SPARCv9 targets is switched from bn(64,64) to
9877   bn(64,32). Because of instruction set limitations it doesn't have
9878   any negative impact on performance. This was done mostly in order
9879   to make it possible to share assembler modules, such as bn_mul_mont
9880   implementations, between 32- and 64-bit builds without hassle.
9881
9882   *Andy Polyakov*
9883
9884 * Move code previously exiled into file crypto/ec/ec2_smpt.c
9885   to ec2_smpl.c, and no longer require the OPENSSL_EC_BIN_PT_COMP
9886   macro.
9887
9888   *Bodo Moeller*
9889
9890 * New candidate for BIGNUM assembler implementation, bn_mul_mont,
9891   dedicated Montgomery multiplication procedure, is introduced.
9892   BN_MONT_CTX is modified to allow bn_mul_mont to reach for higher
9893   "64-bit" performance on certain 32-bit targets.
9894
9895   *Andy Polyakov*
9896
9897 * New option SSL_OP_NO_COMP to disable use of compression selectively
9898   in SSL structures. New SSL ctrl to set maximum send fragment size.
9899   Save memory by setting the I/O buffer sizes dynamically instead of
9900   using the maximum available value.
9901
9902   *Steve Henson*
9903
9904 * New option -V for 'openssl ciphers'. This prints the ciphersuite code
9905   in addition to the text details.
9906
9907   *Bodo Moeller*
9908
9909 * Very, very preliminary EXPERIMENTAL support for printing of general
9910   ASN1 structures. This currently produces rather ugly output and doesn't
9911   handle several customised structures at all.
9912
9913   *Steve Henson*
9914
9915 * Integrated support for PVK file format and some related formats such
9916   as MS PUBLICKEYBLOB and PRIVATEKEYBLOB. Command line switches to support
9917   these in the 'rsa' and 'dsa' utilities.
9918
9919   *Steve Henson*
9920
9921 * Support for PKCS#1 RSAPublicKey format on rsa utility command line.
9922
9923   *Steve Henson*
9924
9925 * Remove the ancient ASN1_METHOD code. This was only ever used in one
9926   place for the (very old) "NETSCAPE" format certificates which are now
9927   handled using new ASN1 code equivalents.
9928
9929   *Steve Henson*
9930
9931 * Let the TLSv1_method() etc. functions return a 'const' SSL_METHOD
9932   pointer and make the SSL_METHOD parameter in SSL_CTX_new,
9933   SSL_CTX_set_ssl_version and SSL_set_ssl_method 'const'.
9934
9935   *Nils Larsch*
9936
9937 * Modify CRL distribution points extension code to print out previously
9938   unsupported fields. Enhance extension setting code to allow setting of
9939   all fields.
9940
9941   *Steve Henson*
9942
9943 * Add print and set support for Issuing Distribution Point CRL extension.
9944
9945   *Steve Henson*
9946
9947 * Change 'Configure' script to enable Camellia by default.
9948
9949   *NTT*
9950
9951OpenSSL 0.9.x
9952-------------
9953
9954### Changes between 0.9.8m and 0.9.8n [24 Mar 2010]
9955
9956 * When rejecting SSL/TLS records due to an incorrect version number, never
9957   update s->server with a new major version number.  As of
9958   - OpenSSL 0.9.8m if 'short' is a 16-bit type,
9959   - OpenSSL 0.9.8f if 'short' is longer than 16 bits,
9960   the previous behavior could result in a read attempt at NULL when
9961   receiving specific incorrect SSL/TLS records once record payload
9962   protection is active.  ([CVE-2010-0740])
9963
9964   *Bodo Moeller, Adam Langley <agl@chromium.org>*
9965
9966 * Fix for CVE-2010-0433 where some kerberos enabled versions of OpenSSL
9967   could be crashed if the relevant tables were not present (e.g. chrooted).
9968
9969   *Tomas Hoger <thoger@redhat.com>*
9970
9971### Changes between 0.9.8l and 0.9.8m [25 Feb 2010]
9972
9973 * Always check bn_wexpand() return values for failure.  ([CVE-2009-3245])
9974
9975   *Martin Olsson, Neel Mehta*
9976
9977 * Fix X509_STORE locking: Every 'objs' access requires a lock (to
9978   accommodate for stack sorting, always a write lock!).
9979
9980   *Bodo Moeller*
9981
9982 * On some versions of WIN32 Heap32Next is very slow. This can cause
9983   excessive delays in the RAND_poll(): over a minute. As a workaround
9984   include a time check in the inner Heap32Next loop too.
9985
9986   *Steve Henson*
9987
9988 * The code that handled flushing of data in SSL/TLS originally used the
9989   BIO_CTRL_INFO ctrl to see if any data was pending first. This caused
9990   the problem outlined in PR#1949. The fix suggested there however can
9991   trigger problems with buggy BIO_CTRL_WPENDING (e.g. some versions
9992   of Apache). So instead simplify the code to flush unconditionally.
9993   This should be fine since flushing with no data to flush is a no op.
9994
9995   *Steve Henson*
9996
9997 * Handle TLS versions 2.0 and later properly and correctly use the
9998   highest version of TLS/SSL supported. Although TLS >= 2.0 is some way
9999   off ancient servers have a habit of sticking around for a while...
10000
10001   *Steve Henson*
10002
10003 * Modify compression code so it frees up structures without using the
10004   ex_data callbacks. This works around a problem where some applications
10005   call CRYPTO_cleanup_all_ex_data() before application exit (e.g. when
10006   restarting) then use compression (e.g. SSL with compression) later.
10007   This results in significant per-connection memory leaks and
10008   has caused some security issues including CVE-2008-1678 and
10009   CVE-2009-4355.
10010
10011   *Steve Henson*
10012
10013 * Constify crypto/cast (i.e., <openssl/cast.h>): a CAST_KEY doesn't
10014   change when encrypting or decrypting.
10015
10016   *Bodo Moeller*
10017
10018 * Add option SSL_OP_LEGACY_SERVER_CONNECT which will allow clients to
10019   connect and renegotiate with servers which do not support RI.
10020   Until RI is more widely deployed this option is enabled by default.
10021
10022   *Steve Henson*
10023
10024 * Add "missing" ssl ctrls to clear options and mode.
10025
10026   *Steve Henson*
10027
10028 * If client attempts to renegotiate and doesn't support RI respond with
10029   a no_renegotiation alert as required by RFC5746.  Some renegotiating
10030   TLS clients will continue a connection gracefully when they receive
10031   the alert. Unfortunately OpenSSL mishandled this alert and would hang
10032   waiting for a server hello which it will never receive. Now we treat a
10033   received no_renegotiation alert as a fatal error. This is because
10034   applications requesting a renegotiation might well expect it to succeed
10035   and would have no code in place to handle the server denying it so the
10036   only safe thing to do is to terminate the connection.
10037
10038   *Steve Henson*
10039
10040 * Add ctrl macro SSL_get_secure_renegotiation_support() which returns 1 if
10041   peer supports secure renegotiation and 0 otherwise. Print out peer
10042   renegotiation support in s_client/s_server.
10043
10044   *Steve Henson*
10045
10046 * Replace the highly broken and deprecated SPKAC certification method with
10047   the updated NID creation version. This should correctly handle UTF8.
10048
10049   *Steve Henson*
10050
10051 * Implement RFC5746. Re-enable renegotiation but require the extension
10052   as needed. Unfortunately, SSL3_FLAGS_ALLOW_UNSAFE_LEGACY_RENEGOTIATION
10053   turns out to be a bad idea. It has been replaced by
10054   SSL_OP_ALLOW_UNSAFE_LEGACY_RENEGOTIATION which can be set with
10055   SSL_CTX_set_options(). This is really not recommended unless you
10056   know what you are doing.
10057
10058   *Eric Rescorla <ekr@networkresonance.com>, Ben Laurie, Steve Henson*
10059
10060 * Fixes to stateless session resumption handling. Use initial_ctx when
10061   issuing and attempting to decrypt tickets in case it has changed during
10062   servername handling. Use a non-zero length session ID when attempting
10063   stateless session resumption: this makes it possible to determine if
10064   a resumption has occurred immediately after receiving server hello
10065   (several places in OpenSSL subtly assume this) instead of later in
10066   the handshake.
10067
10068   *Steve Henson*
10069
10070 * The functions ENGINE_ctrl(), OPENSSL_isservice(),
10071   CMS_get1_RecipientRequest() and RAND_bytes() can return <=0 on error
10072   fixes for a few places where the return code is not checked
10073   correctly.
10074
10075   *Julia Lawall <julia@diku.dk>*
10076
10077 * Add --strict-warnings option to Configure script to include devteam
10078   warnings in other configurations.
10079
10080   *Steve Henson*
10081
10082 * Add support for --libdir option and LIBDIR variable in makefiles. This
10083   makes it possible to install openssl libraries in locations which
10084   have names other than "lib", for example "/usr/lib64" which some
10085   systems need.
10086
10087   *Steve Henson, based on patch from Jeremy Utley*
10088
10089 * Don't allow the use of leading 0x80 in OIDs. This is a violation of
10090   X690 8.9.12 and can produce some misleading textual output of OIDs.
10091
10092   *Steve Henson, reported by Dan Kaminsky*
10093
10094 * Delete MD2 from algorithm tables. This follows the recommendation in
10095   several standards that it is not used in new applications due to
10096   several cryptographic weaknesses. For binary compatibility reasons
10097   the MD2 API is still compiled in by default.
10098
10099   *Steve Henson*
10100
10101 * Add compression id to {d2i,i2d}_SSL_SESSION so it is correctly saved
10102   and restored.
10103
10104   *Steve Henson*
10105
10106 * Rename uni2asc and asc2uni functions to OPENSSL_uni2asc and
10107   OPENSSL_asc2uni conditionally on Netware platforms to avoid a name
10108   clash.
10109
10110   *Guenter <lists@gknw.net>*
10111
10112 * Fix the server certificate chain building code to use X509_verify_cert(),
10113   it used to have an ad-hoc builder which was unable to cope with anything
10114   other than a simple chain.
10115
10116   *David Woodhouse <dwmw2@infradead.org>, Steve Henson*
10117
10118 * Don't check self signed certificate signatures in X509_verify_cert()
10119   by default (a flag can override this): it just wastes time without
10120   adding any security. As a useful side effect self signed root CAs
10121   with non-FIPS digests are now usable in FIPS mode.
10122
10123   *Steve Henson*
10124
10125 * In dtls1_process_out_of_seq_message() the check if the current message
10126   is already buffered was missing. For every new message was memory
10127   allocated, allowing an attacker to perform an denial of service attack
10128   with sending out of seq handshake messages until there is no memory
10129   left. Additionally every future message was buffered, even if the
10130   sequence number made no sense and would be part of another handshake.
10131   So only messages with sequence numbers less than 10 in advance will be
10132   buffered.  ([CVE-2009-1378])
10133
10134   *Robin Seggelmann, discovered by Daniel Mentz*
10135
10136 * Records are buffered if they arrive with a future epoch to be
10137   processed after finishing the corresponding handshake. There is
10138   currently no limitation to this buffer allowing an attacker to perform
10139   a DOS attack with sending records with future epochs until there is no
10140   memory left. This patch adds the pqueue_size() function to determine
10141   the size of a buffer and limits the record buffer to 100 entries.
10142   ([CVE-2009-1377])
10143
10144   *Robin Seggelmann, discovered by Daniel Mentz*
10145
10146 * Keep a copy of frag->msg_header.frag_len so it can be used after the
10147   parent structure is freed.  ([CVE-2009-1379])
10148
10149   *Daniel Mentz*
10150
10151 * Handle non-blocking I/O properly in SSL_shutdown() call.
10152
10153   *Darryl Miles <darryl-mailinglists@netbauds.net>*
10154
10155 * Add `2.5.4.*` OIDs
10156
10157   *Ilya O. <vrghost@gmail.com>*
10158
10159### Changes between 0.9.8k and 0.9.8l  [5 Nov 2009]
10160
10161 * Disable renegotiation completely - this fixes a severe security
10162   problem ([CVE-2009-3555]) at the cost of breaking all
10163   renegotiation. Renegotiation can be re-enabled by setting
10164   SSL3_FLAGS_ALLOW_UNSAFE_LEGACY_RENEGOTIATION in s3->flags at
10165   run-time. This is really not recommended unless you know what
10166   you're doing.
10167
10168   *Ben Laurie*
10169
10170### Changes between 0.9.8j and 0.9.8k  [25 Mar 2009]
10171
10172 * Don't set val to NULL when freeing up structures, it is freed up by
10173   underlying code. If `sizeof(void *) > sizeof(long)` this can result in
10174   zeroing past the valid field. ([CVE-2009-0789])
10175
10176   *Paolo Ganci <Paolo.Ganci@AdNovum.CH>*
10177
10178 * Fix bug where return value of CMS_SignerInfo_verify_content() was not
10179   checked correctly. This would allow some invalid signed attributes to
10180   appear to verify correctly. ([CVE-2009-0591])
10181
10182   *Ivan Nestlerode <inestlerode@us.ibm.com>*
10183
10184 * Reject UniversalString and BMPString types with invalid lengths. This
10185   prevents a crash in ASN1_STRING_print_ex() which assumes the strings have
10186   a legal length. ([CVE-2009-0590])
10187
10188   *Steve Henson*
10189
10190 * Set S/MIME signing as the default purpose rather than setting it
10191   unconditionally. This allows applications to override it at the store
10192   level.
10193
10194   *Steve Henson*
10195
10196 * Permit restricted recursion of ASN1 strings. This is needed in practice
10197   to handle some structures.
10198
10199   *Steve Henson*
10200
10201 * Improve efficiency of mem_gets: don't search whole buffer each time
10202   for a '\n'
10203
10204   *Jeremy Shapiro <jnshapir@us.ibm.com>*
10205
10206 * New -hex option for openssl rand.
10207
10208   *Matthieu Herrb*
10209
10210 * Print out UTF8String and NumericString when parsing ASN1.
10211
10212   *Steve Henson*
10213
10214 * Support NumericString type for name components.
10215
10216   *Steve Henson*
10217
10218 * Allow CC in the environment to override the automatically chosen
10219   compiler. Note that nothing is done to ensure flags work with the
10220   chosen compiler.
10221
10222   *Ben Laurie*
10223
10224### Changes between 0.9.8i and 0.9.8j  [07 Jan 2009]
10225
10226 * Properly check EVP_VerifyFinal() and similar return values
10227   ([CVE-2008-5077]).
10228
10229   *Ben Laurie, Bodo Moeller, Google Security Team*
10230
10231 * Enable TLS extensions by default.
10232
10233   *Ben Laurie*
10234
10235 * Allow the CHIL engine to be loaded, whether the application is
10236   multithreaded or not. (This does not release the developer from the
10237   obligation to set up the dynamic locking callbacks.)
10238
10239   *Sander Temme <sander@temme.net>*
10240
10241 * Use correct exit code if there is an error in dgst command.
10242
10243   *Steve Henson; problem pointed out by Roland Dirlewanger*
10244
10245 * Tweak Configure so that you need to say "experimental-jpake" to enable
10246   JPAKE, and need to use -DOPENSSL_EXPERIMENTAL_JPAKE in applications.
10247
10248   *Bodo Moeller*
10249
10250 * Add experimental JPAKE support, including demo authentication in
10251   s_client and s_server.
10252
10253   *Ben Laurie*
10254
10255 * Set the comparison function in v3_addr_canonize().
10256
10257   *Rob Austein <sra@hactrn.net>*
10258
10259 * Add support for XMPP STARTTLS in s_client.
10260
10261   *Philip Paeps <philip@freebsd.org>*
10262
10263 * Change the server-side SSL_OP_NETSCAPE_REUSE_CIPHER_CHANGE_BUG behavior
10264   to ensure that even with this option, only ciphersuites in the
10265   server's preference list will be accepted.  (Note that the option
10266   applies only when resuming a session, so the earlier behavior was
10267   just about the algorithm choice for symmetric cryptography.)
10268
10269   *Bodo Moeller*
10270
10271### Changes between 0.9.8h and 0.9.8i  [15 Sep 2008]
10272
10273 * Fix NULL pointer dereference if a DTLS server received
10274   ChangeCipherSpec as first record ([CVE-2009-1386]).
10275
10276   *PR #1679*
10277
10278 * Fix a state transition in s3_srvr.c and d1_srvr.c
10279   (was using SSL3_ST_CW_CLNT_HELLO_B, should be `..._ST_SW_SRVR_...`).
10280
10281   *Nagendra Modadugu*
10282
10283 * The fix in 0.9.8c that supposedly got rid of unsafe
10284   double-checked locking was incomplete for RSA blinding,
10285   addressing just one layer of what turns out to have been
10286   doubly unsafe triple-checked locking.
10287
10288   So now fix this for real by retiring the MONT_HELPER macro
10289   in crypto/rsa/rsa_eay.c.
10290
10291   *Bodo Moeller; problem pointed out by Marius Schilder*
10292
10293 * Various precautionary measures:
10294
10295   - Avoid size_t integer overflow in HASH_UPDATE (md32_common.h).
10296
10297   - Avoid a buffer overflow in d2i_SSL_SESSION() (ssl_asn1.c).
10298     (NB: This would require knowledge of the secret session ticket key
10299     to exploit, in which case you'd be SOL either way.)
10300
10301   - Change bn_nist.c so that it will properly handle input BIGNUMs
10302     outside the expected range.
10303
10304   - Enforce the 'num' check in BN_div() (bn_div.c) for non-BN_DEBUG
10305     builds.
10306
10307   *Neel Mehta, Bodo Moeller*
10308
10309 * Allow engines to be "soft loaded" - i.e. optionally don't die if
10310   the load fails. Useful for distros.
10311
10312   *Ben Laurie and the FreeBSD team*
10313
10314 * Add support for Local Machine Keyset attribute in PKCS#12 files.
10315
10316   *Steve Henson*
10317
10318 * Fix BN_GF2m_mod_arr() top-bit cleanup code.
10319
10320   *Huang Ying*
10321
10322 * Expand ENGINE to support engine supplied SSL client certificate functions.
10323
10324   This work was sponsored by Logica.
10325
10326   *Steve Henson*
10327
10328 * Add CryptoAPI ENGINE to support use of RSA and DSA keys held in Windows
10329   keystores. Support for SSL/TLS client authentication too.
10330   Not compiled unless enable-capieng specified to Configure.
10331
10332   This work was sponsored by Logica.
10333
10334   *Steve Henson*
10335
10336 * Fix bug in X509_ATTRIBUTE creation: don't set attribute using
10337   ASN1_TYPE_set1 if MBSTRING flag set. This bug would crash certain
10338   attribute creation routines such as certificate requests and PKCS#12
10339   files.
10340
10341   *Steve Henson*
10342
10343### Changes between 0.9.8g and 0.9.8h  [28 May 2008]
10344
10345 * Fix flaw if 'Server Key exchange message' is omitted from a TLS
10346   handshake which could lead to a client crash as found using the
10347   Codenomicon TLS test suite ([CVE-2008-1672])
10348
10349   *Steve Henson, Mark Cox*
10350
10351 * Fix double free in TLS server name extensions which could lead to
10352   a remote crash found by Codenomicon TLS test suite ([CVE-2008-0891])
10353
10354   *Joe Orton*
10355
10356 * Clear error queue in SSL_CTX_use_certificate_chain_file()
10357
10358   Clear the error queue to ensure that error entries left from
10359   older function calls do not interfere with the correct operation.
10360
10361   *Lutz Jaenicke, Erik de Castro Lopo*
10362
10363 * Remove root CA certificates of commercial CAs:
10364
10365   The OpenSSL project does not recommend any specific CA and does not
10366   have any policy with respect to including or excluding any CA.
10367   Therefore it does not make any sense to ship an arbitrary selection
10368   of root CA certificates with the OpenSSL software.
10369
10370   *Lutz Jaenicke*
10371
10372 * RSA OAEP patches to fix two separate invalid memory reads.
10373   The first one involves inputs when 'lzero' is greater than
10374   'SHA_DIGEST_LENGTH' (it would read about SHA_DIGEST_LENGTH bytes
10375   before the beginning of from). The second one involves inputs where
10376   the 'db' section contains nothing but zeroes (there is a one-byte
10377   invalid read after the end of 'db').
10378
10379   *Ivan Nestlerode <inestlerode@us.ibm.com>*
10380
10381 * Partial backport from 0.9.9-dev:
10382
10383   Introduce bn_mul_mont (dedicated Montgomery multiplication
10384   procedure) as a candidate for BIGNUM assembler implementation.
10385   While 0.9.9-dev uses assembler for various architectures, only
10386   x86_64 is available by default here in the 0.9.8 branch, and
10387   32-bit x86 is available through a compile-time setting.
10388
10389   To try the 32-bit x86 assembler implementation, use Configure
10390   option "enable-montasm" (which exists only for this backport).
10391
10392   As "enable-montasm" for 32-bit x86 disclaims code stability
10393   anyway, in this constellation we activate additional code
10394   backported from 0.9.9-dev for further performance improvements,
10395   namely BN_from_montgomery_word.  (To enable this otherwise,
10396   e.g. x86_64, try `-DMONT_FROM_WORD___NON_DEFAULT_0_9_8_BUILD`.)
10397
10398   *Andy Polyakov (backport partially by Bodo Moeller)*
10399
10400 * Add TLS session ticket callback. This allows an application to set
10401   TLS ticket cipher and HMAC keys rather than relying on hardcoded fixed
10402   values. This is useful for key rollover for example where several key
10403   sets may exist with different names.
10404
10405   *Steve Henson*
10406
10407 * Reverse ENGINE-internal logic for caching default ENGINE handles.
10408   This was broken until now in 0.9.8 releases, such that the only way
10409   a registered ENGINE could be used (assuming it initialises
10410   successfully on the host) was to explicitly set it as the default
10411   for the relevant algorithms. This is in contradiction with 0.9.7
10412   behaviour and the documentation. With this fix, when an ENGINE is
10413   registered into a given algorithm's table of implementations, the
10414   'uptodate' flag is reset so that auto-discovery will be used next
10415   time a new context for that algorithm attempts to select an
10416   implementation.
10417
10418   *Ian Lister (tweaked by Geoff Thorpe)*
10419
10420 * Backport of CMS code to OpenSSL 0.9.8. This differs from the 0.9.9
10421   implementation in the following ways:
10422
10423   Lack of EVP_PKEY_ASN1_METHOD means algorithm parameters have to be
10424   hard coded.
10425
10426   Lack of BER streaming support means one pass streaming processing is
10427   only supported if data is detached: setting the streaming flag is
10428   ignored for embedded content.
10429
10430   CMS support is disabled by default and must be explicitly enabled
10431   with the enable-cms configuration option.
10432
10433   *Steve Henson*
10434
10435 * Update the GMP engine glue to do direct copies between BIGNUM and
10436   mpz_t when openssl and GMP use the same limb size. Otherwise the
10437   existing "conversion via a text string export" trick is still used.
10438
10439   *Paul Sheer <paulsheer@gmail.com>*
10440
10441 * Zlib compression BIO. This is a filter BIO which compressed and
10442   uncompresses any data passed through it.
10443
10444   *Steve Henson*
10445
10446 * Add AES_wrap_key() and AES_unwrap_key() functions to implement
10447   RFC3394 compatible AES key wrapping.
10448
10449   *Steve Henson*
10450
10451 * Add utility functions to handle ASN1 structures. ASN1_STRING_set0():
10452   sets string data without copying. X509_ALGOR_set0() and
10453   X509_ALGOR_get0(): set and retrieve X509_ALGOR (AlgorithmIdentifier)
10454   data. Attribute function X509at_get0_data_by_OBJ(): retrieves data
10455   from an X509_ATTRIBUTE structure optionally checking it occurs only
10456   once. ASN1_TYPE_set1(): set and ASN1_TYPE structure copying supplied
10457   data.
10458
10459   *Steve Henson*
10460
10461 * Fix BN flag handling in RSA_eay_mod_exp() and BN_MONT_CTX_set()
10462   to get the expected BN_FLG_CONSTTIME behavior.
10463
10464   *Bodo Moeller (Google)*
10465
10466 * Netware support:
10467
10468   - fixed wrong usage of ioctlsocket() when build for LIBC BSD sockets
10469   - fixed do_tests.pl to run the test suite with CLIB builds too (CLIB_OPT)
10470   - added some more tests to do_tests.pl
10471   - fixed RunningProcess usage so that it works with newer LIBC NDKs too
10472   - removed usage of BN_LLONG for CLIB builds to avoid runtime dependency
10473   - added new Configure targets netware-clib-bsdsock, netware-clib-gcc,
10474     netware-clib-bsdsock-gcc, netware-libc-bsdsock-gcc
10475   - various changes to netware.pl to enable gcc-cross builds on Win32
10476     platform
10477   - changed crypto/bio/b_sock.c to work with macro functions (CLIB BSD)
10478   - various changes to fix missing prototype warnings
10479   - fixed x86nasm.pl to create correct asm files for NASM COFF output
10480   - added AES, WHIRLPOOL and CPUID assembler code to build files
10481   - added missing AES assembler make rules to mk1mf.pl
10482   - fixed order of includes in `apps/ocsp.c` so that `e_os.h` settings apply
10483
10484   *Guenter Knauf <eflash@gmx.net>*
10485
10486 * Implement certificate status request TLS extension defined in RFC3546.
10487   A client can set the appropriate parameters and receive the encoded
10488   OCSP response via a callback. A server can query the supplied parameters
10489   and set the encoded OCSP response in the callback. Add simplified examples
10490   to s_client and s_server.
10491
10492   *Steve Henson*
10493
10494### Changes between 0.9.8f and 0.9.8g  [19 Oct 2007]
10495
10496 * Fix various bugs:
10497   + Binary incompatibility of ssl_ctx_st structure
10498   + DTLS interoperation with non-compliant servers
10499   + Don't call get_session_cb() without proposed session
10500   + Fix ia64 assembler code
10501
10502   *Andy Polyakov, Steve Henson*
10503
10504### Changes between 0.9.8e and 0.9.8f  [11 Oct 2007]
10505
10506 * DTLS Handshake overhaul. There were longstanding issues with
10507   OpenSSL DTLS implementation, which were making it impossible for
10508   RFC 4347 compliant client to communicate with OpenSSL server.
10509   Unfortunately just fixing these incompatibilities would "cut off"
10510   pre-0.9.8f clients. To allow for hassle free upgrade post-0.9.8e
10511   server keeps tolerating non RFC compliant syntax. The opposite is
10512   not true, 0.9.8f client can not communicate with earlier server.
10513   This update even addresses CVE-2007-4995.
10514
10515   *Andy Polyakov*
10516
10517 * Changes to avoid need for function casts in OpenSSL: some compilers
10518   (gcc 4.2 and later) reject their use.
10519   *Kurt Roeckx <kurt@roeckx.be>, Peter Hartley <pdh@utter.chaos.org.uk>,
10520    Steve Henson*
10521
10522 * Add RFC4507 support to OpenSSL. This includes the corrections in
10523   RFC4507bis. The encrypted ticket format is an encrypted encoded
10524   SSL_SESSION structure, that way new session features are automatically
10525   supported.
10526
10527   If a client application caches session in an SSL_SESSION structure
10528   support is transparent because tickets are now stored in the encoded
10529   SSL_SESSION.
10530
10531   The SSL_CTX structure automatically generates keys for ticket
10532   protection in servers so again support should be possible
10533   with no application modification.
10534
10535   If a client or server wishes to disable RFC4507 support then the option
10536   SSL_OP_NO_TICKET can be set.
10537
10538   Add a TLS extension debugging callback to allow the contents of any client
10539   or server extensions to be examined.
10540
10541   This work was sponsored by Google.
10542
10543   *Steve Henson*
10544
10545 * Add initial support for TLS extensions, specifically for the server_name
10546   extension so far.  The SSL_SESSION, SSL_CTX, and SSL data structures now
10547   have new members for a host name.  The SSL data structure has an
10548   additional member `SSL_CTX *initial_ctx` so that new sessions can be
10549   stored in that context to allow for session resumption, even after the
10550   SSL has been switched to a new SSL_CTX in reaction to a client's
10551   server_name extension.
10552
10553   New functions (subject to change):
10554
10555           SSL_get_servername()
10556           SSL_get_servername_type()
10557           SSL_set_SSL_CTX()
10558
10559   New CTRL codes and macros (subject to change):
10560
10561           SSL_CTRL_SET_TLSEXT_SERVERNAME_CB
10562                               - SSL_CTX_set_tlsext_servername_callback()
10563           SSL_CTRL_SET_TLSEXT_SERVERNAME_ARG
10564                                    - SSL_CTX_set_tlsext_servername_arg()
10565           SSL_CTRL_SET_TLSEXT_HOSTNAME           - SSL_set_tlsext_host_name()
10566
10567   openssl s_client has a new '-servername ...' option.
10568
10569   openssl s_server has new options '-servername_host ...', '-cert2 ...',
10570   '-key2 ...', '-servername_fatal' (subject to change).  This allows
10571   testing the HostName extension for a specific single host name ('-cert'
10572   and '-key' remain fallbacks for handshakes without HostName
10573   negotiation).  If the unrecognized_name alert has to be sent, this by
10574   default is a warning; it becomes fatal with the '-servername_fatal'
10575   option.
10576
10577   *Peter Sylvester,  Remy Allais, Christophe Renou, Steve Henson*
10578
10579 * Add AES and SSE2 assembly language support to VC++ build.
10580
10581   *Steve Henson*
10582
10583 * Mitigate attack on final subtraction in Montgomery reduction.
10584
10585   *Andy Polyakov*
10586
10587 * Fix crypto/ec/ec_mult.c to work properly with scalars of value 0
10588   (which previously caused an internal error).
10589
10590   *Bodo Moeller*
10591
10592 * Squeeze another 10% out of IGE mode when in != out.
10593
10594   *Ben Laurie*
10595
10596 * AES IGE mode speedup.
10597
10598   *Dean Gaudet (Google)*
10599
10600 * Add the Korean symmetric 128-bit cipher SEED (see
10601   <http://www.kisa.or.kr/kisa/seed/jsp/seed_eng.jsp>) and
10602   add SEED ciphersuites from RFC 4162:
10603
10604           TLS_RSA_WITH_SEED_CBC_SHA      =  "SEED-SHA"
10605           TLS_DHE_DSS_WITH_SEED_CBC_SHA  =  "DHE-DSS-SEED-SHA"
10606           TLS_DHE_RSA_WITH_SEED_CBC_SHA  =  "DHE-RSA-SEED-SHA"
10607           TLS_DH_anon_WITH_SEED_CBC_SHA  =  "ADH-SEED-SHA"
10608
10609   To minimize changes between patchlevels in the OpenSSL 0.9.8
10610   series, SEED remains excluded from compilation unless OpenSSL
10611   is configured with 'enable-seed'.
10612
10613   *KISA, Bodo Moeller*
10614
10615 * Mitigate branch prediction attacks, which can be practical if a
10616   single processor is shared, allowing a spy process to extract
10617   information.  For detailed background information, see
10618   <http://eprint.iacr.org/2007/039> (O. Aciicmez, S. Gueron,
10619   J.-P. Seifert, "New Branch Prediction Vulnerabilities in OpenSSL
10620   and Necessary Software Countermeasures").  The core of the change
10621   are new versions BN_div_no_branch() and
10622   BN_mod_inverse_no_branch() of BN_div() and BN_mod_inverse(),
10623   respectively, which are slower, but avoid the security-relevant
10624   conditional branches.  These are automatically called by BN_div()
10625   and BN_mod_inverse() if the flag BN_FLG_CONSTTIME is set for one
10626   of the input BIGNUMs.  Also, BN_is_bit_set() has been changed to
10627   remove a conditional branch.
10628
10629   BN_FLG_CONSTTIME is the new name for the previous
10630   BN_FLG_EXP_CONSTTIME flag, since it now affects more than just
10631   modular exponentiation.  (Since OpenSSL 0.9.7h, setting this flag
10632   in the exponent causes BN_mod_exp_mont() to use the alternative
10633   implementation in BN_mod_exp_mont_consttime().)  The old name
10634   remains as a deprecated alias.
10635
10636   Similarly, RSA_FLAG_NO_EXP_CONSTTIME is replaced by a more general
10637   RSA_FLAG_NO_CONSTTIME flag since the RSA implementation now uses
10638   constant-time implementations for more than just exponentiation.
10639   Here too the old name is kept as a deprecated alias.
10640
10641   BN_BLINDING_new() will now use BN_dup() for the modulus so that
10642   the BN_BLINDING structure gets an independent copy of the
10643   modulus.  This means that the previous `BIGNUM *m` argument to
10644   BN_BLINDING_new() and to BN_BLINDING_create_param() now
10645   essentially becomes `const BIGNUM *m`, although we can't actually
10646   change this in the header file before 0.9.9.  It allows
10647   RSA_setup_blinding() to use BN_with_flags() on the modulus to
10648   enable BN_FLG_CONSTTIME.
10649
10650   *Matthew D Wood (Intel Corp)*
10651
10652 * In the SSL/TLS server implementation, be strict about session ID
10653   context matching (which matters if an application uses a single
10654   external cache for different purposes).  Previously,
10655   out-of-context reuse was forbidden only if SSL_VERIFY_PEER was
10656   set.  This did ensure strict client verification, but meant that,
10657   with applications using a single external cache for quite
10658   different requirements, clients could circumvent ciphersuite
10659   restrictions for a given session ID context by starting a session
10660   in a different context.
10661
10662   *Bodo Moeller*
10663
10664 * Include "!eNULL" in SSL_DEFAULT_CIPHER_LIST to make sure that
10665   a ciphersuite string such as "DEFAULT:RSA" cannot enable
10666   authentication-only ciphersuites.
10667
10668   *Bodo Moeller*
10669
10670 * Update the SSL_get_shared_ciphers() fix CVE-2006-3738 which was
10671   not complete and could lead to a possible single byte overflow
10672   ([CVE-2007-5135]) [Ben Laurie]
10673
10674### Changes between 0.9.8d and 0.9.8e  [23 Feb 2007]
10675
10676 * Since AES128 and AES256 (and similarly Camellia128 and
10677   Camellia256) share a single mask bit in the logic of
10678   ssl/ssl_ciph.c, the code for masking out disabled ciphers needs a
10679   kludge to work properly if AES128 is available and AES256 isn't
10680   (or if Camellia128 is available and Camellia256 isn't).
10681
10682   *Victor Duchovni*
10683
10684 * Fix the BIT STRING encoding generated by crypto/ec/ec_asn1.c
10685   (within i2d_ECPrivateKey, i2d_ECPKParameters, i2d_ECParameters):
10686   When a point or a seed is encoded in a BIT STRING, we need to
10687   prevent the removal of trailing zero bits to get the proper DER
10688   encoding.  (By default, crypto/asn1/a_bitstr.c assumes the case
10689   of a NamedBitList, for which trailing 0 bits need to be removed.)
10690
10691   *Bodo Moeller*
10692
10693 * Have SSL/TLS server implementation tolerate "mismatched" record
10694   protocol version while receiving ClientHello even if the
10695   ClientHello is fragmented.  (The server can't insist on the
10696   particular protocol version it has chosen before the ServerHello
10697   message has informed the client about his choice.)
10698
10699   *Bodo Moeller*
10700
10701 * Add RFC 3779 support.
10702
10703   *Rob Austein for ARIN, Ben Laurie*
10704
10705 * Load error codes if they are not already present instead of using a
10706   static variable. This allows them to be cleanly unloaded and reloaded.
10707   Improve header file function name parsing.
10708
10709   *Steve Henson*
10710
10711 * extend SMTP and IMAP protocol emulation in s_client to use EHLO
10712   or CAPABILITY handshake as required by RFCs.
10713
10714   *Goetz Babin-Ebell*
10715
10716### Changes between 0.9.8c and 0.9.8d  [28 Sep 2006]
10717
10718 * Introduce limits to prevent malicious keys being able to
10719   cause a denial of service.  ([CVE-2006-2940])
10720
10721   *Steve Henson, Bodo Moeller*
10722
10723 * Fix ASN.1 parsing of certain invalid structures that can result
10724   in a denial of service.  ([CVE-2006-2937])  [Steve Henson]
10725
10726 * Fix buffer overflow in SSL_get_shared_ciphers() function.
10727   ([CVE-2006-3738]) [Tavis Ormandy and Will Drewry, Google Security Team]
10728
10729 * Fix SSL client code which could crash if connecting to a
10730   malicious SSLv2 server.  ([CVE-2006-4343])
10731
10732   *Tavis Ormandy and Will Drewry, Google Security Team*
10733
10734 * Since 0.9.8b, ciphersuite strings naming explicit ciphersuites
10735   match only those.  Before that, "AES256-SHA" would be interpreted
10736   as a pattern and match "AES128-SHA" too (since AES128-SHA got
10737   the same strength classification in 0.9.7h) as we currently only
10738   have a single AES bit in the ciphersuite description bitmap.
10739   That change, however, also applied to ciphersuite strings such as
10740   "RC4-MD5" that intentionally matched multiple ciphersuites --
10741   namely, SSL 2.0 ciphersuites in addition to the more common ones
10742   from SSL 3.0/TLS 1.0.
10743
10744   So we change the selection algorithm again: Naming an explicit
10745   ciphersuite selects this one ciphersuite, and any other similar
10746   ciphersuite (same bitmap) from *other* protocol versions.
10747   Thus, "RC4-MD5" again will properly select both the SSL 2.0
10748   ciphersuite and the SSL 3.0/TLS 1.0 ciphersuite.
10749
10750   Since SSL 2.0 does not have any ciphersuites for which the
10751   128/256 bit distinction would be relevant, this works for now.
10752   The proper fix will be to use different bits for AES128 and
10753   AES256, which would have avoided the problems from the beginning;
10754   however, bits are scarce, so we can only do this in a new release
10755   (not just a patchlevel) when we can change the SSL_CIPHER
10756   definition to split the single 'unsigned long mask' bitmap into
10757   multiple values to extend the available space.
10758
10759   *Bodo Moeller*
10760
10761### Changes between 0.9.8b and 0.9.8c  [05 Sep 2006]
10762
10763 * Avoid PKCS #1 v1.5 signature attack discovered by Daniel Bleichenbacher
10764   ([CVE-2006-4339])  [Ben Laurie and Google Security Team]
10765
10766 * Add AES IGE and biIGE modes.
10767
10768   *Ben Laurie*
10769
10770 * Change the Unix randomness entropy gathering to use poll() when
10771   possible instead of select(), since the latter has some
10772   undesirable limitations.
10773
10774   *Darryl Miles via Richard Levitte and Bodo Moeller*
10775
10776 * Disable "ECCdraft" ciphersuites more thoroughly.  Now special
10777   treatment in ssl/ssl_ciph.s makes sure that these ciphersuites
10778   cannot be implicitly activated as part of, e.g., the "AES" alias.
10779   However, please upgrade to OpenSSL 0.9.9[-dev] for
10780   non-experimental use of the ECC ciphersuites to get TLS extension
10781   support, which is required for curve and point format negotiation
10782   to avoid potential handshake problems.
10783
10784   *Bodo Moeller*
10785
10786 * Disable rogue ciphersuites:
10787
10788   - SSLv2 0x08 0x00 0x80 ("RC4-64-MD5")
10789   - SSLv3/TLSv1 0x00 0x61 ("EXP1024-RC2-CBC-MD5")
10790   - SSLv3/TLSv1 0x00 0x60 ("EXP1024-RC4-MD5")
10791
10792   The latter two were purportedly from
10793   draft-ietf-tls-56-bit-ciphersuites-0[01].txt, but do not really
10794   appear there.
10795
10796   Also deactivate the remaining ciphersuites from
10797   draft-ietf-tls-56-bit-ciphersuites-01.txt.  These are just as
10798   unofficial, and the ID has long expired.
10799
10800   *Bodo Moeller*
10801
10802 * Fix RSA blinding Heisenbug (problems sometimes occurred on
10803   dual-core machines) and other potential thread-safety issues.
10804
10805   *Bodo Moeller*
10806
10807 * Add the symmetric cipher Camellia (128-bit, 192-bit, 256-bit key
10808   versions), which is now available for royalty-free use
10809   (see <http://info.isl.ntt.co.jp/crypt/eng/info/chiteki.html>).
10810   Also, add Camellia TLS ciphersuites from RFC 4132.
10811
10812   To minimize changes between patchlevels in the OpenSSL 0.9.8
10813   series, Camellia remains excluded from compilation unless OpenSSL
10814   is configured with 'enable-camellia'.
10815
10816   *NTT*
10817
10818 * Disable the padding bug check when compression is in use. The padding
10819   bug check assumes the first packet is of even length, this is not
10820   necessarily true if compression is enabled and can result in false
10821   positives causing handshake failure. The actual bug test is ancient
10822   code so it is hoped that implementations will either have fixed it by
10823   now or any which still have the bug do not support compression.
10824
10825   *Steve Henson*
10826
10827### Changes between 0.9.8a and 0.9.8b  [04 May 2006]
10828
10829 * When applying a cipher rule check to see if string match is an explicit
10830   cipher suite and only match that one cipher suite if it is.
10831
10832   *Steve Henson*
10833
10834 * Link in manifests for VC++ if needed.
10835
10836   *Austin Ziegler <halostatue@gmail.com>*
10837
10838 * Update support for ECC-based TLS ciphersuites according to
10839   draft-ietf-tls-ecc-12.txt with proposed changes (but without
10840   TLS extensions, which are supported starting with the 0.9.9
10841   branch, not in the OpenSSL 0.9.8 branch).
10842
10843   *Douglas Stebila*
10844
10845 * New functions EVP_CIPHER_CTX_new() and EVP_CIPHER_CTX_free() to support
10846   opaque EVP_CIPHER_CTX handling.
10847
10848   *Steve Henson*
10849
10850 * Fixes and enhancements to zlib compression code. We now only use
10851   "zlib1.dll" and use the default `__cdecl` calling convention on Win32
10852   to conform with the standards mentioned here:
10853   <http://www.zlib.net/DLL_FAQ.txt>
10854   Static zlib linking now works on Windows and the new --with-zlib-include
10855   --with-zlib-lib options to Configure can be used to supply the location
10856   of the headers and library. Gracefully handle case where zlib library
10857   can't be loaded.
10858
10859   *Steve Henson*
10860
10861 * Several fixes and enhancements to the OID generation code. The old code
10862   sometimes allowed invalid OIDs (1.X for X >= 40 for example), couldn't
10863   handle numbers larger than ULONG_MAX, truncated printing and had a
10864   non standard OBJ_obj2txt() behaviour.
10865
10866   *Steve Henson*
10867
10868 * Add support for building of engines under engine/ as shared libraries
10869   under VC++ build system.
10870
10871   *Steve Henson*
10872
10873 * Corrected the numerous bugs in the Win32 path splitter in DSO.
10874   Hopefully, we will not see any false combination of paths any more.
10875
10876   *Richard Levitte*
10877
10878### Changes between 0.9.8 and 0.9.8a  [11 Oct 2005]
10879
10880 * Remove the functionality of SSL_OP_MSIE_SSLV2_RSA_PADDING
10881   (part of SSL_OP_ALL).  This option used to disable the
10882   countermeasure against man-in-the-middle protocol-version
10883   rollback in the SSL 2.0 server implementation, which is a bad
10884   idea.  ([CVE-2005-2969])
10885
10886   *Bodo Moeller; problem pointed out by Yutaka Oiwa (Research Center
10887   for Information Security, National Institute of Advanced Industrial
10888   Science and Technology [AIST], Japan)*
10889
10890 * Add two function to clear and return the verify parameter flags.
10891
10892   *Steve Henson*
10893
10894 * Keep cipherlists sorted in the source instead of sorting them at
10895   runtime, thus removing the need for a lock.
10896
10897   *Nils Larsch*
10898
10899 * Avoid some small subgroup attacks in Diffie-Hellman.
10900
10901   *Nick Mathewson and Ben Laurie*
10902
10903 * Add functions for well-known primes.
10904
10905   *Nick Mathewson*
10906
10907 * Extended Windows CE support.
10908
10909   *Satoshi Nakamura and Andy Polyakov*
10910
10911 * Initialize SSL_METHOD structures at compile time instead of during
10912   runtime, thus removing the need for a lock.
10913
10914   *Steve Henson*
10915
10916 * Make PKCS7_decrypt() work even if no certificate is supplied by
10917   attempting to decrypt each encrypted key in turn. Add support to
10918   smime utility.
10919
10920   *Steve Henson*
10921
10922### Changes between 0.9.7h and 0.9.8  [05 Jul 2005]
10923
10924[NB: OpenSSL 0.9.7i and later 0.9.7 patch levels were released after
10925OpenSSL 0.9.8.]
10926
10927 * Add libcrypto.pc and libssl.pc for those who feel they need them.
10928
10929   *Richard Levitte*
10930
10931 * Change CA.sh and CA.pl so they don't bundle the CSR and the private
10932   key into the same file any more.
10933
10934   *Richard Levitte*
10935
10936 * Add initial support for Win64, both IA64 and AMD64/x64 flavors.
10937
10938   *Andy Polyakov*
10939
10940 * Add -utf8 command line and config file option to 'ca'.
10941
10942   *Stefan <stf@udoma.org*
10943
10944 * Removed the macro des_crypt(), as it seems to conflict with some
10945   libraries.  Use DES_crypt().
10946
10947   *Richard Levitte*
10948
10949 * Correct naming of the 'chil' and '4758cca' ENGINEs. This
10950   involves renaming the source and generated shared-libs for
10951   both. The engines will accept the corrected or legacy ids
10952   ('ncipher' and '4758_cca' respectively) when binding. NB,
10953   this only applies when building 'shared'.
10954
10955   *Corinna Vinschen <vinschen@redhat.com> and Geoff Thorpe*
10956
10957 * Add attribute functions to EVP_PKEY structure. Modify
10958   PKCS12_create() to recognize a CSP name attribute and
10959   use it. Make -CSP option work again in pkcs12 utility.
10960
10961   *Steve Henson*
10962
10963 * Add new functionality to the bn blinding code:
10964   - automatic re-creation of the BN_BLINDING parameters after
10965     a fixed number of uses (currently 32)
10966   - add new function for parameter creation
10967   - introduce flags to control the update behaviour of the
10968     BN_BLINDING parameters
10969   - hide BN_BLINDING structure
10970   Add a second BN_BLINDING slot to the RSA structure to improve
10971   performance when a single RSA object is shared among several
10972   threads.
10973
10974   *Nils Larsch*
10975
10976 * Add support for DTLS.
10977
10978   *Nagendra Modadugu <nagendra@cs.stanford.edu> and Ben Laurie*
10979
10980 * Add support for DER encoded private keys (SSL_FILETYPE_ASN1)
10981   to SSL_CTX_use_PrivateKey_file() and SSL_use_PrivateKey_file()
10982
10983   *Walter Goulet*
10984
10985 * Remove buggy and incomplete DH cert support from
10986   ssl/ssl_rsa.c and ssl/s3_both.c
10987
10988   *Nils Larsch*
10989
10990 * Use SHA-1 instead of MD5 as the default digest algorithm for
10991   the `apps/openssl` commands.
10992
10993   *Nils Larsch*
10994
10995 * Compile clean with "-Wall -Wmissing-prototypes
10996   -Wstrict-prototypes -Wmissing-declarations -Werror". Currently
10997   DEBUG_SAFESTACK must also be set.
10998
10999   *Ben Laurie*
11000
11001 * Change ./Configure so that certain algorithms can be disabled by default.
11002   The new counterpiece to "no-xxx" is "enable-xxx".
11003
11004   The patented RC5 and MDC2 algorithms will now be disabled unless
11005   "enable-rc5" and "enable-mdc2", respectively, are specified.
11006
11007   (IDEA remains enabled despite being patented.  This is because IDEA
11008   is frequently required for interoperability, and there is no license
11009   fee for non-commercial use.  As before, "no-idea" can be used to
11010   avoid this algorithm.)
11011
11012   *Bodo Moeller*
11013
11014 * Add processing of proxy certificates (see RFC 3820).  This work was
11015   sponsored by KTH (The Royal Institute of Technology in Stockholm) and
11016   EGEE (Enabling Grids for E-science in Europe).
11017
11018   *Richard Levitte*
11019
11020 * RC4 performance overhaul on modern architectures/implementations, such
11021   as Intel P4, IA-64 and AMD64.
11022
11023   *Andy Polyakov*
11024
11025 * New utility extract-section.pl. This can be used specify an alternative
11026   section number in a pod file instead of having to treat each file as
11027   a separate case in Makefile. This can be done by adding two lines to the
11028   pod file:
11029
11030   =for comment openssl_section:XXX
11031
11032   The blank line is mandatory.
11033
11034   *Steve Henson*
11035
11036 * New arguments -certform, -keyform and -pass for s_client and s_server
11037   to allow alternative format key and certificate files and passphrase
11038   sources.
11039
11040   *Steve Henson*
11041
11042 * New structure X509_VERIFY_PARAM which combines current verify parameters,
11043   update associated structures and add various utility functions.
11044
11045   Add new policy related verify parameters, include policy checking in
11046   standard verify code. Enhance 'smime' application with extra parameters
11047   to support policy checking and print out.
11048
11049   *Steve Henson*
11050
11051 * Add a new engine to support VIA PadLock ACE extensions in the VIA C3
11052   Nehemiah processors. These extensions support AES encryption in hardware
11053   as well as RNG (though RNG support is currently disabled).
11054
11055   *Michal Ludvig <michal@logix.cz>, with help from Andy Polyakov*
11056
11057 * Deprecate `BN_[get|set]_params()` functions (they were ignored internally).
11058
11059   *Geoff Thorpe*
11060
11061 * New FIPS 180-2 algorithms, SHA-224/-256/-384/-512 are implemented.
11062
11063   *Andy Polyakov and a number of other people*
11064
11065 * Improved PowerPC platform support. Most notably BIGNUM assembler
11066   implementation contributed by IBM.
11067
11068   *Suresh Chari, Peter Waltenberg, Andy Polyakov*
11069
11070 * The new 'RSA_generate_key_ex' function now takes a BIGNUM for the public
11071   exponent rather than 'unsigned long'. There is a corresponding change to
11072   the new 'rsa_keygen' element of the RSA_METHOD structure.
11073
11074   *Jelte Jansen, Geoff Thorpe*
11075
11076 * Functionality for creating the initial serial number file is now
11077   moved from CA.pl to the 'ca' utility with a new option -create_serial.
11078
11079   (Before OpenSSL 0.9.7e, CA.pl used to initialize the serial
11080   number file to 1, which is bound to cause problems.  To avoid
11081   the problems while respecting compatibility between different 0.9.7
11082   patchlevels, 0.9.7e  employed 'openssl x509 -next_serial' in
11083   CA.pl for serial number initialization.  With the new release 0.9.8,
11084   we can fix the problem directly in the 'ca' utility.)
11085
11086   *Steve Henson*
11087
11088 * Reduced header interdependencies by declaring more opaque objects in
11089   ossl_typ.h. As a consequence, including some headers (eg. engine.h) will
11090   give fewer recursive includes, which could break lazy source code - so
11091   this change is covered by the OPENSSL_NO_DEPRECATED symbol. As always,
11092   developers should define this symbol when building and using openssl to
11093   ensure they track the recommended behaviour, interfaces, [etc], but
11094   backwards-compatible behaviour prevails when this isn't defined.
11095
11096   *Geoff Thorpe*
11097
11098 * New function X509_POLICY_NODE_print() which prints out policy nodes.
11099
11100   *Steve Henson*
11101
11102 * Add new EVP function EVP_CIPHER_CTX_rand_key and associated functionality.
11103   This will generate a random key of the appropriate length based on the
11104   cipher context. The EVP_CIPHER can provide its own random key generation
11105   routine to support keys of a specific form. This is used in the des and
11106   3des routines to generate a key of the correct parity. Update S/MIME
11107   code to use new functions and hence generate correct parity DES keys.
11108   Add EVP_CHECK_DES_KEY #define to return an error if the key is not
11109   valid (weak or incorrect parity).
11110
11111   *Steve Henson*
11112
11113 * Add a local set of CRLs that can be used by X509_verify_cert() as well
11114   as looking them up. This is useful when the verified structure may contain
11115   CRLs, for example PKCS#7 signedData. Modify PKCS7_verify() to use any CRLs
11116   present unless the new PKCS7_NO_CRL flag is asserted.
11117
11118   *Steve Henson*
11119
11120 * Extend ASN1 oid configuration module. It now additionally accepts the
11121   syntax:
11122
11123   shortName = some long name, 1.2.3.4
11124
11125   *Steve Henson*
11126
11127 * Reimplemented the BN_CTX implementation. There is now no more static
11128   limitation on the number of variables it can handle nor the depth of the
11129   "stack" handling for BN_CTX_start()/BN_CTX_end() pairs. The stack
11130   information can now expand as required, and rather than having a single
11131   static array of bignums, BN_CTX now uses a linked-list of such arrays
11132   allowing it to expand on demand whilst maintaining the usefulness of
11133   BN_CTX's "bundling".
11134
11135   *Geoff Thorpe*
11136
11137 * Add a missing BN_CTX parameter to the 'rsa_mod_exp' callback in RSA_METHOD
11138   to allow all RSA operations to function using a single BN_CTX.
11139
11140   *Geoff Thorpe*
11141
11142 * Preliminary support for certificate policy evaluation and checking. This
11143   is initially intended to pass the tests outlined in "Conformance Testing
11144   of Relying Party Client Certificate Path Processing Logic" v1.07.
11145
11146   *Steve Henson*
11147
11148 * bn_dup_expand() has been deprecated, it was introduced in 0.9.7 and
11149   remained unused and not that useful. A variety of other little bignum
11150   tweaks and fixes have also been made continuing on from the audit (see
11151   below).
11152
11153   *Geoff Thorpe*
11154
11155 * Constify all or almost all d2i, c2i, s2i and r2i functions, along with
11156   associated ASN1, EVP and SSL functions and old ASN1 macros.
11157
11158   *Richard Levitte*
11159
11160 * BN_zero() only needs to set 'top' and 'neg' to zero for correct results,
11161   and this should never fail. So the return value from the use of
11162   BN_set_word() (which can fail due to needless expansion) is now deprecated;
11163   if OPENSSL_NO_DEPRECATED is defined, BN_zero() is a void macro.
11164
11165   *Geoff Thorpe*
11166
11167 * BN_CTX_get() should return zero-valued bignums, providing the same
11168   initialised value as BN_new().
11169
11170   *Geoff Thorpe, suggested by Ulf Möller*
11171
11172 * Support for inhibitAnyPolicy certificate extension.
11173
11174   *Steve Henson*
11175
11176 * An audit of the BIGNUM code is underway, for which debugging code is
11177   enabled when BN_DEBUG is defined. This makes stricter enforcements on what
11178   is considered valid when processing BIGNUMs, and causes execution to
11179   assert() when a problem is discovered. If BN_DEBUG_RAND is defined,
11180   further steps are taken to deliberately pollute unused data in BIGNUM
11181   structures to try and expose faulty code further on. For now, openssl will
11182   (in its default mode of operation) continue to tolerate the inconsistent
11183   forms that it has tolerated in the past, but authors and packagers should
11184   consider trying openssl and their own applications when compiled with
11185   these debugging symbols defined. It will help highlight potential bugs in
11186   their own code, and will improve the test coverage for OpenSSL itself. At
11187   some point, these tighter rules will become openssl's default to improve
11188   maintainability, though the assert()s and other overheads will remain only
11189   in debugging configurations. See bn.h for more details.
11190
11191   *Geoff Thorpe, Nils Larsch, Ulf Möller*
11192
11193 * BN_CTX_init() has been deprecated, as BN_CTX is an opaque structure
11194   that can only be obtained through BN_CTX_new() (which implicitly
11195   initialises it). The presence of this function only made it possible
11196   to overwrite an existing structure (and cause memory leaks).
11197
11198   *Geoff Thorpe*
11199
11200 * Because of the callback-based approach for implementing LHASH as a
11201   template type, lh_insert() adds opaque objects to hash-tables and
11202   lh_doall() or lh_doall_arg() are typically used with a destructor callback
11203   to clean up those corresponding objects before destroying the hash table
11204   (and losing the object pointers). So some over-zealous constifications in
11205   LHASH have been relaxed so that lh_insert() does not take (nor store) the
11206   objects as "const" and the `lh_doall[_arg]` callback wrappers are not
11207   prototyped to have "const" restrictions on the object pointers they are
11208   given (and so aren't required to cast them away any more).
11209
11210   *Geoff Thorpe*
11211
11212 * The tmdiff.h API was so ugly and minimal that our own timing utility
11213   (speed) prefers to use its own implementation. The two implementations
11214   haven't been consolidated as yet (volunteers?) but the tmdiff API has had
11215   its object type properly exposed (MS_TM) instead of casting to/from
11216   `char *`. This may still change yet if someone realises MS_TM and
11217   `ms_time_***`
11218   aren't necessarily the greatest nomenclatures - but this is what was used
11219   internally to the implementation so I've used that for now.
11220
11221   *Geoff Thorpe*
11222
11223 * Ensure that deprecated functions do not get compiled when
11224   OPENSSL_NO_DEPRECATED is defined. Some "openssl" subcommands and a few of
11225   the self-tests were still using deprecated key-generation functions so
11226   these have been updated also.
11227
11228   *Geoff Thorpe*
11229
11230 * Reorganise PKCS#7 code to separate the digest location functionality
11231   into PKCS7_find_digest(), digest addition into PKCS7_bio_add_digest().
11232   New function PKCS7_set_digest() to set the digest type for PKCS#7
11233   digestedData type. Add additional code to correctly generate the
11234   digestedData type and add support for this type in PKCS7 initialization
11235   functions.
11236
11237   *Steve Henson*
11238
11239 * New function PKCS7_set0_type_other() this initializes a PKCS7
11240   structure of type "other".
11241
11242   *Steve Henson*
11243
11244 * Fix prime generation loop in crypto/bn/bn_prime.pl by making
11245   sure the loop does correctly stop and breaking ("division by zero")
11246   modulus operations are not performed. The (pre-generated) prime
11247   table crypto/bn/bn_prime.h was already correct, but it could not be
11248   re-generated on some platforms because of the "division by zero"
11249   situation in the script.
11250
11251   *Ralf S. Engelschall*
11252
11253 * Update support for ECC-based TLS ciphersuites according to
11254   draft-ietf-tls-ecc-03.txt: the KDF1 key derivation function with
11255   SHA-1 now is only used for "small" curves (where the
11256   representation of a field element takes up to 24 bytes); for
11257   larger curves, the field element resulting from ECDH is directly
11258   used as premaster secret.
11259
11260   *Douglas Stebila (Sun Microsystems Laboratories)*
11261
11262 * Add code for kP+lQ timings to crypto/ec/ectest.c, and add SEC2
11263   curve secp160r1 to the tests.
11264
11265   *Douglas Stebila (Sun Microsystems Laboratories)*
11266
11267 * Add the possibility to load symbols globally with DSO.
11268
11269   *Götz Babin-Ebell <babin-ebell@trustcenter.de> via Richard Levitte*
11270
11271 * Add the functions ERR_set_mark() and ERR_pop_to_mark() for better
11272   control of the error stack.
11273
11274   *Richard Levitte*
11275
11276 * Add support for STORE in ENGINE.
11277
11278   *Richard Levitte*
11279
11280 * Add the STORE type.  The intention is to provide a common interface
11281   to certificate and key stores, be they simple file-based stores, or
11282   HSM-type store, or LDAP stores, or...
11283   NOTE: The code is currently UNTESTED and isn't really used anywhere.
11284
11285   *Richard Levitte*
11286
11287 * Add a generic structure called OPENSSL_ITEM.  This can be used to
11288   pass a list of arguments to any function as well as provide a way
11289   for a function to pass data back to the caller.
11290
11291   *Richard Levitte*
11292
11293 * Add the functions BUF_strndup() and BUF_memdup().  BUF_strndup()
11294   works like BUF_strdup() but can be used to duplicate a portion of
11295   a string.  The copy gets NUL-terminated.  BUF_memdup() duplicates
11296   a memory area.
11297
11298   *Richard Levitte*
11299
11300 * Add the function sk_find_ex() which works like sk_find(), but will
11301   return an index to an element even if an exact match couldn't be
11302   found.  The index is guaranteed to point at the element where the
11303   searched-for key would be inserted to preserve sorting order.
11304
11305   *Richard Levitte*
11306
11307 * Add the function OBJ_bsearch_ex() which works like OBJ_bsearch() but
11308   takes an extra flags argument for optional functionality.  Currently,
11309   the following flags are defined:
11310
11311      OBJ_BSEARCH_VALUE_ON_NOMATCH
11312      This one gets OBJ_bsearch_ex() to return a pointer to the first
11313      element where the comparing function returns a negative or zero
11314      number.
11315
11316      OBJ_BSEARCH_FIRST_VALUE_ON_MATCH
11317      This one gets OBJ_bsearch_ex() to return a pointer to the first
11318      element where the comparing function returns zero.  This is useful
11319      if there are more than one element where the comparing function
11320      returns zero.
11321
11322   *Richard Levitte*
11323
11324 * Make it possible to create self-signed certificates with 'openssl ca'
11325   in such a way that the self-signed certificate becomes part of the
11326   CA database and uses the same mechanisms for serial number generation
11327   as all other certificate signing.  The new flag '-selfsign' enables
11328   this functionality.  Adapt CA.sh and CA.pl.in.
11329
11330   *Richard Levitte*
11331
11332 * Add functionality to check the public key of a certificate request
11333   against a given private.  This is useful to check that a certificate
11334   request can be signed by that key (self-signing).
11335
11336   *Richard Levitte*
11337
11338 * Make it possible to have multiple active certificates with the same
11339   subject in the CA index file.  This is done only if the keyword
11340   'unique_subject' is set to 'no' in the main CA section (default
11341   if 'CA_default') of the configuration file.  The value is saved
11342   with the database itself in a separate index attribute file,
11343   named like the index file with '.attr' appended to the name.
11344
11345   *Richard Levitte*
11346
11347 * Generate multi-valued AVAs using '+' notation in config files for
11348   req and dirName.
11349
11350   *Steve Henson*
11351
11352 * Support for nameConstraints certificate extension.
11353
11354   *Steve Henson*
11355
11356 * Support for policyConstraints certificate extension.
11357
11358   *Steve Henson*
11359
11360 * Support for policyMappings certificate extension.
11361
11362   *Steve Henson*
11363
11364 * Make sure the default DSA_METHOD implementation only uses its
11365   dsa_mod_exp() and/or bn_mod_exp() handlers if they are non-NULL,
11366   and change its own handlers to be NULL so as to remove unnecessary
11367   indirection. This lets alternative implementations fallback to the
11368   default implementation more easily.
11369
11370   *Geoff Thorpe*
11371
11372 * Support for directoryName in GeneralName related extensions
11373   in config files.
11374
11375   *Steve Henson*
11376
11377 * Make it possible to link applications using Makefile.shared.
11378   Make that possible even when linking against static libraries!
11379
11380   *Richard Levitte*
11381
11382 * Support for single pass processing for S/MIME signing. This now
11383   means that S/MIME signing can be done from a pipe, in addition
11384   cleartext signing (multipart/signed type) is effectively streaming
11385   and the signed data does not need to be all held in memory.
11386
11387   This is done with a new flag PKCS7_STREAM. When this flag is set
11388   PKCS7_sign() only initializes the PKCS7 structure and the actual signing
11389   is done after the data is output (and digests calculated) in
11390   SMIME_write_PKCS7().
11391
11392   *Steve Henson*
11393
11394 * Add full support for -rpath/-R, both in shared libraries and
11395   applications, at least on the platforms where it's known how
11396   to do it.
11397
11398   *Richard Levitte*
11399
11400 * In crypto/ec/ec_mult.c, implement fast point multiplication with
11401   precomputation, based on wNAF splitting: EC_GROUP_precompute_mult()
11402   will now compute a table of multiples of the generator that
11403   makes subsequent invocations of EC_POINTs_mul() or EC_POINT_mul()
11404   faster (notably in the case of a single point multiplication,
11405   scalar * generator).
11406
11407   *Nils Larsch, Bodo Moeller*
11408
11409 * IPv6 support for certificate extensions. The various extensions
11410   which use the IP:a.b.c.d can now take IPv6 addresses using the
11411   formats of RFC1884 2.2 . IPv6 addresses are now also displayed
11412   correctly.
11413
11414   *Steve Henson*
11415
11416 * Added an ENGINE that implements RSA by performing private key
11417   exponentiations with the GMP library. The conversions to and from
11418   GMP's mpz_t format aren't optimised nor are any montgomery forms
11419   cached, and on x86 it appears OpenSSL's own performance has caught up.
11420   However there are likely to be other architectures where GMP could
11421   provide a boost. This ENGINE is not built in by default, but it can be
11422   specified at Configure time and should be accompanied by the necessary
11423   linker additions, eg;
11424           ./config -DOPENSSL_USE_GMP -lgmp
11425
11426   *Geoff Thorpe*
11427
11428 * "openssl engine" will not display ENGINE/DSO load failure errors when
11429   testing availability of engines with "-t" - the old behaviour is
11430   produced by increasing the feature's verbosity with "-tt".
11431
11432   *Geoff Thorpe*
11433
11434 * ECDSA routines: under certain error conditions uninitialized BN objects
11435   could be freed. Solution: make sure initialization is performed early
11436   enough. (Reported and fix supplied by Nils Larsch <nla@trustcenter.de>
11437   via PR#459)
11438
11439   *Lutz Jaenicke*
11440
11441 * Key-generation can now be implemented in RSA_METHOD, DSA_METHOD
11442   and DH_METHOD (eg. by ENGINE implementations) to override the normal
11443   software implementations. For DSA and DH, parameter generation can
11444   also be overridden by providing the appropriate method callbacks.
11445
11446   *Geoff Thorpe*
11447
11448 * Change the "progress" mechanism used in key-generation and
11449   primality testing to functions that take a new BN_GENCB pointer in
11450   place of callback/argument pairs. The new API functions have `_ex`
11451   postfixes and the older functions are reimplemented as wrappers for
11452   the new ones. The OPENSSL_NO_DEPRECATED symbol can be used to hide
11453   declarations of the old functions to help (graceful) attempts to
11454   migrate to the new functions. Also, the new key-generation API
11455   functions operate on a caller-supplied key-structure and return
11456   success/failure rather than returning a key or NULL - this is to
11457   help make "keygen" another member function of RSA_METHOD etc.
11458
11459   Example for using the new callback interface:
11460
11461           int (*my_callback)(int a, int b, BN_GENCB *cb) = ...;
11462           void *my_arg = ...;
11463           BN_GENCB my_cb;
11464
11465           BN_GENCB_set(&my_cb, my_callback, my_arg);
11466
11467           return BN_is_prime_ex(some_bignum, BN_prime_checks, NULL, &cb);
11468           /* For the meaning of a, b in calls to my_callback(), see the
11469            * documentation of the function that calls the callback.
11470            * cb will point to my_cb; my_arg can be retrieved as cb->arg.
11471            * my_callback should return 1 if it wants BN_is_prime_ex()
11472            * to continue, or 0 to stop.
11473            */
11474
11475   *Geoff Thorpe*
11476
11477 * Change the ZLIB compression method to be stateful, and make it
11478   available to TLS with the number defined in
11479   draft-ietf-tls-compression-04.txt.
11480
11481   *Richard Levitte*
11482
11483 * Add the ASN.1 structures and functions for CertificatePair, which
11484   is defined as follows (according to X.509_4thEditionDraftV6.pdf):
11485
11486           CertificatePair ::= SEQUENCE {
11487              forward         [0]     Certificate OPTIONAL,
11488              reverse         [1]     Certificate OPTIONAL,
11489              -- at least one of the pair shall be present -- }
11490
11491   Also implement the PEM functions to read and write certificate
11492   pairs, and defined the PEM tag as "CERTIFICATE PAIR".
11493
11494   This needed to be defined, mostly for the sake of the LDAP
11495   attribute crossCertificatePair, but may prove useful elsewhere as
11496   well.
11497
11498   *Richard Levitte*
11499
11500 * Make it possible to inhibit symlinking of shared libraries in
11501   Makefile.shared, for Cygwin's sake.
11502
11503   *Richard Levitte*
11504
11505 * Extend the BIGNUM API by creating a function
11506           void BN_set_negative(BIGNUM *a, int neg);
11507   and a macro that behave like
11508           int  BN_is_negative(const BIGNUM *a);
11509
11510   to avoid the need to access 'a->neg' directly in applications.
11511
11512   *Nils Larsch*
11513
11514 * Implement fast modular reduction for pseudo-Mersenne primes
11515   used in NIST curves (crypto/bn/bn_nist.c, crypto/ec/ecp_nist.c).
11516   EC_GROUP_new_curve_GFp() will now automatically use this
11517   if applicable.
11518
11519   *Nils Larsch <nla@trustcenter.de>*
11520
11521 * Add new lock type (CRYPTO_LOCK_BN).
11522
11523   *Bodo Moeller*
11524
11525 * Change the ENGINE framework to automatically load engines
11526   dynamically from specific directories unless they could be
11527   found to already be built in or loaded.  Move all the
11528   current engines except for the cryptodev one to a new
11529   directory engines/.
11530   The engines in engines/ are built as shared libraries if
11531   the "shared" options was given to ./Configure or ./config.
11532   Otherwise, they are inserted in libcrypto.a.
11533   /usr/local/ssl/engines is the default directory for dynamic
11534   engines, but that can be overridden at configure time through
11535   the usual use of --prefix and/or --openssldir, and at run
11536   time with the environment variable OPENSSL_ENGINES.
11537
11538   *Geoff Thorpe and Richard Levitte*
11539
11540 * Add Makefile.shared, a helper makefile to build shared
11541   libraries.  Adapt Makefile.org.
11542
11543   *Richard Levitte*
11544
11545 * Add version info to Win32 DLLs.
11546
11547   *Peter 'Luna' Runestig" <peter@runestig.com>*
11548
11549 * Add new 'medium level' PKCS#12 API. Certificates and keys
11550   can be added using this API to created arbitrary PKCS#12
11551   files while avoiding the low-level API.
11552
11553   New options to PKCS12_create(), key or cert can be NULL and
11554   will then be omitted from the output file. The encryption
11555   algorithm NIDs can be set to -1 for no encryption, the mac
11556   iteration count can be set to 0 to omit the mac.
11557
11558   Enhance pkcs12 utility by making the -nokeys and -nocerts
11559   options work when creating a PKCS#12 file. New option -nomac
11560   to omit the mac, NONE can be set for an encryption algorithm.
11561   New code is modified to use the enhanced PKCS12_create()
11562   instead of the low-level API.
11563
11564   *Steve Henson*
11565
11566 * Extend ASN1 encoder to support indefinite length constructed
11567   encoding. This can output sequences tags and octet strings in
11568   this form. Modify pk7_asn1.c to support indefinite length
11569   encoding. This is experimental and needs additional code to
11570   be useful, such as an ASN1 bio and some enhanced streaming
11571   PKCS#7 code.
11572
11573   Extend template encode functionality so that tagging is passed
11574   down to the template encoder.
11575
11576   *Steve Henson*
11577
11578 * Let 'openssl req' fail if an argument to '-newkey' is not
11579   recognized instead of using RSA as a default.
11580
11581   *Bodo Moeller*
11582
11583 * Add support for ECC-based ciphersuites from draft-ietf-tls-ecc-01.txt.
11584   As these are not official, they are not included in "ALL";
11585   the "ECCdraft" ciphersuite group alias can be used to select them.
11586
11587   *Vipul Gupta and Sumit Gupta (Sun Microsystems Laboratories)*
11588
11589 * Add ECDH engine support.
11590
11591   *Nils Gura and Douglas Stebila (Sun Microsystems Laboratories)*
11592
11593 * Add ECDH in new directory crypto/ecdh/.
11594
11595   *Douglas Stebila (Sun Microsystems Laboratories)*
11596
11597 * Let BN_rand_range() abort with an error after 100 iterations
11598   without success (which indicates a broken PRNG).
11599
11600   *Bodo Moeller*
11601
11602 * Change BN_mod_sqrt() so that it verifies that the input value
11603   is really the square of the return value.  (Previously,
11604   BN_mod_sqrt would show GIGO behaviour.)
11605
11606   *Bodo Moeller*
11607
11608 * Add named elliptic curves over binary fields from X9.62, SECG,
11609   and WAP/WTLS; add OIDs that were still missing.
11610
11611   *Sheueling Chang Shantz and Douglas Stebila (Sun Microsystems Laboratories)*
11612
11613 * Extend the EC library for elliptic curves over binary fields
11614   (new files ec2_smpl.c, ec2_smpt.c, ec2_mult.c in crypto/ec/).
11615   New EC_METHOD:
11616
11617           EC_GF2m_simple_method
11618
11619   New API functions:
11620
11621           EC_GROUP_new_curve_GF2m
11622           EC_GROUP_set_curve_GF2m
11623           EC_GROUP_get_curve_GF2m
11624           EC_POINT_set_affine_coordinates_GF2m
11625           EC_POINT_get_affine_coordinates_GF2m
11626           EC_POINT_set_compressed_coordinates_GF2m
11627
11628   Point compression for binary fields is disabled by default for
11629   patent reasons (compile with OPENSSL_EC_BIN_PT_COMP defined to
11630   enable it).
11631
11632   As binary polynomials are represented as BIGNUMs, various members
11633   of the EC_GROUP and EC_POINT data structures can be shared
11634   between the implementations for prime fields and binary fields;
11635   the above `..._GF2m functions` (except for EX_GROUP_new_curve_GF2m)
11636   are essentially identical to their `..._GFp` counterparts.
11637   (For simplicity, the `..._GFp` prefix has been dropped from
11638   various internal method names.)
11639
11640   An internal 'field_div' method (similar to 'field_mul' and
11641   'field_sqr') has been added; this is used only for binary fields.
11642
11643   *Sheueling Chang Shantz and Douglas Stebila (Sun Microsystems Laboratories)*
11644
11645 * Optionally dispatch EC_POINT_mul(), EC_POINT_precompute_mult()
11646   through methods ('mul', 'precompute_mult').
11647
11648   The generic implementations (now internally called 'ec_wNAF_mul'
11649   and 'ec_wNAF_precomputed_mult') remain the default if these
11650   methods are undefined.
11651
11652   *Sheueling Chang Shantz and Douglas Stebila (Sun Microsystems Laboratories)*
11653
11654 * New function EC_GROUP_get_degree, which is defined through
11655   EC_METHOD.  For curves over prime fields, this returns the bit
11656   length of the modulus.
11657
11658   *Sheueling Chang Shantz and Douglas Stebila (Sun Microsystems Laboratories)*
11659
11660 * New functions EC_GROUP_dup, EC_POINT_dup.
11661   (These simply call ..._new  and ..._copy).
11662
11663   *Sheueling Chang Shantz and Douglas Stebila (Sun Microsystems Laboratories)*
11664
11665 * Add binary polynomial arithmetic software in crypto/bn/bn_gf2m.c.
11666   Polynomials are represented as BIGNUMs (where the sign bit is not
11667   used) in the following functions [macros]:
11668
11669           BN_GF2m_add
11670           BN_GF2m_sub             [= BN_GF2m_add]
11671           BN_GF2m_mod             [wrapper for BN_GF2m_mod_arr]
11672           BN_GF2m_mod_mul         [wrapper for BN_GF2m_mod_mul_arr]
11673           BN_GF2m_mod_sqr         [wrapper for BN_GF2m_mod_sqr_arr]
11674           BN_GF2m_mod_inv
11675           BN_GF2m_mod_exp         [wrapper for BN_GF2m_mod_exp_arr]
11676           BN_GF2m_mod_sqrt        [wrapper for BN_GF2m_mod_sqrt_arr]
11677           BN_GF2m_mod_solve_quad  [wrapper for BN_GF2m_mod_solve_quad_arr]
11678           BN_GF2m_cmp             [= BN_ucmp]
11679
11680   (Note that only the 'mod' functions are actually for fields GF(2^m).
11681   BN_GF2m_add() is misnomer, but this is for the sake of consistency.)
11682
11683   For some functions, an the irreducible polynomial defining a
11684   field can be given as an 'unsigned int[]' with strictly
11685   decreasing elements giving the indices of those bits that are set;
11686   i.e., p[] represents the polynomial
11687           f(t) = t^p[0] + t^p[1] + ... + t^p[k]
11688   where
11689           p[0] > p[1] > ... > p[k] = 0.
11690   This applies to the following functions:
11691
11692           BN_GF2m_mod_arr
11693           BN_GF2m_mod_mul_arr
11694           BN_GF2m_mod_sqr_arr
11695           BN_GF2m_mod_inv_arr        [wrapper for BN_GF2m_mod_inv]
11696           BN_GF2m_mod_div_arr        [wrapper for BN_GF2m_mod_div]
11697           BN_GF2m_mod_exp_arr
11698           BN_GF2m_mod_sqrt_arr
11699           BN_GF2m_mod_solve_quad_arr
11700           BN_GF2m_poly2arr
11701           BN_GF2m_arr2poly
11702
11703   Conversion can be performed by the following functions:
11704
11705           BN_GF2m_poly2arr
11706           BN_GF2m_arr2poly
11707
11708   bntest.c has additional tests for binary polynomial arithmetic.
11709
11710   Two implementations for BN_GF2m_mod_div() are available.
11711   The default algorithm simply uses BN_GF2m_mod_inv() and
11712   BN_GF2m_mod_mul().  The alternative algorithm is compiled in only
11713   if OPENSSL_SUN_GF2M_DIV is defined (patent pending; read the
11714   copyright notice in crypto/bn/bn_gf2m.c before enabling it).
11715
11716   *Sheueling Chang Shantz and Douglas Stebila (Sun Microsystems Laboratories)*
11717
11718 * Add new error code 'ERR_R_DISABLED' that can be used when some
11719   functionality is disabled at compile-time.
11720
11721   *Douglas Stebila <douglas.stebila@sun.com>*
11722
11723 * Change default behaviour of 'openssl asn1parse' so that more
11724   information is visible when viewing, e.g., a certificate:
11725
11726   Modify asn1_parse2 (crypto/asn1/asn1_par.c) so that in non-'dump'
11727   mode the content of non-printable OCTET STRINGs is output in a
11728   style similar to INTEGERs, but with '[HEX DUMP]' prepended to
11729   avoid the appearance of a printable string.
11730
11731   *Nils Larsch <nla@trustcenter.de>*
11732
11733 * Add 'asn1_flag' and 'asn1_form' member to EC_GROUP with access
11734   functions
11735           EC_GROUP_set_asn1_flag()
11736           EC_GROUP_get_asn1_flag()
11737           EC_GROUP_set_point_conversion_form()
11738           EC_GROUP_get_point_conversion_form()
11739   These control ASN1 encoding details:
11740   - Curves (i.e., groups) are encoded explicitly unless asn1_flag
11741     has been set to OPENSSL_EC_NAMED_CURVE.
11742   - Points are encoded in uncompressed form by default; options for
11743     asn1_for are as for point2oct, namely
11744           POINT_CONVERSION_COMPRESSED
11745           POINT_CONVERSION_UNCOMPRESSED
11746           POINT_CONVERSION_HYBRID
11747
11748   Also add 'seed' and 'seed_len' members to EC_GROUP with access
11749   functions
11750           EC_GROUP_set_seed()
11751           EC_GROUP_get0_seed()
11752           EC_GROUP_get_seed_len()
11753   This is used only for ASN1 purposes (so far).
11754
11755   *Nils Larsch <nla@trustcenter.de>*
11756
11757 * Add 'field_type' member to EC_METHOD, which holds the NID
11758   of the appropriate field type OID.  The new function
11759   EC_METHOD_get_field_type() returns this value.
11760
11761   *Nils Larsch <nla@trustcenter.de>*
11762
11763 * Add functions
11764           EC_POINT_point2bn()
11765           EC_POINT_bn2point()
11766           EC_POINT_point2hex()
11767           EC_POINT_hex2point()
11768   providing useful interfaces to EC_POINT_point2oct() and
11769   EC_POINT_oct2point().
11770
11771   *Nils Larsch <nla@trustcenter.de>*
11772
11773 * Change internals of the EC library so that the functions
11774           EC_GROUP_set_generator()
11775           EC_GROUP_get_generator()
11776           EC_GROUP_get_order()
11777           EC_GROUP_get_cofactor()
11778   are implemented directly in crypto/ec/ec_lib.c and not dispatched
11779   to methods, which would lead to unnecessary code duplication when
11780   adding different types of curves.
11781
11782   *Nils Larsch <nla@trustcenter.de> with input by Bodo Moeller*
11783
11784 * Implement compute_wNAF (crypto/ec/ec_mult.c) without BIGNUM
11785   arithmetic, and such that modified wNAFs are generated
11786   (which avoid length expansion in many cases).
11787
11788   *Bodo Moeller*
11789
11790 * Add a function EC_GROUP_check_discriminant() (defined via
11791   EC_METHOD) that verifies that the curve discriminant is non-zero.
11792
11793   Add a function EC_GROUP_check() that makes some sanity tests
11794   on a EC_GROUP, its generator and order.  This includes
11795   EC_GROUP_check_discriminant().
11796
11797   *Nils Larsch <nla@trustcenter.de>*
11798
11799 * Add ECDSA in new directory crypto/ecdsa/.
11800
11801   Add applications 'openssl ecparam' and 'openssl ecdsa'
11802   (these are based on 'openssl dsaparam' and 'openssl dsa').
11803
11804   ECDSA support is also included in various other files across the
11805   library.  Most notably,
11806   - 'openssl req' now has a '-newkey ecdsa:file' option;
11807   - EVP_PKCS82PKEY (crypto/evp/evp_pkey.c) now can handle ECDSA;
11808   - X509_PUBKEY_get (crypto/asn1/x_pubkey.c) and
11809     d2i_PublicKey (crypto/asn1/d2i_pu.c) have been modified to make
11810     them suitable for ECDSA where domain parameters must be
11811     extracted before the specific public key;
11812   - ECDSA engine support has been added.
11813
11814   *Nils Larsch <nla@trustcenter.de>*
11815
11816 * Include some named elliptic curves, and add OIDs from X9.62,
11817   SECG, and WAP/WTLS.  Each curve can be obtained from the new
11818   function
11819           EC_GROUP_new_by_curve_name(),
11820   and the list of available named curves can be obtained with
11821           EC_get_builtin_curves().
11822   Also add a 'curve_name' member to EC_GROUP objects, which can be
11823   accessed via
11824           EC_GROUP_set_curve_name()
11825           EC_GROUP_get_curve_name()
11826
11827   *Nils Larsch <larsch@trustcenter.de, Bodo Moeller*
11828
11829 * Remove a few calls to bn_wexpand() in BN_sqr() (the one in there
11830   was actually never needed) and in BN_mul().  The removal in BN_mul()
11831   required a small change in bn_mul_part_recursive() and the addition
11832   of the functions bn_cmp_part_words(), bn_sub_part_words() and
11833   bn_add_part_words(), which do the same thing as bn_cmp_words(),
11834   bn_sub_words() and bn_add_words() except they take arrays with
11835   differing sizes.
11836
11837   *Richard Levitte*
11838
11839### Changes between 0.9.7l and 0.9.7m  [23 Feb 2007]
11840
11841 * Cleanse PEM buffers before freeing them since they may contain
11842   sensitive data.
11843
11844   *Benjamin Bennett <ben@psc.edu>*
11845
11846 * Include "!eNULL" in SSL_DEFAULT_CIPHER_LIST to make sure that
11847   a ciphersuite string such as "DEFAULT:RSA" cannot enable
11848   authentication-only ciphersuites.
11849
11850   *Bodo Moeller*
11851
11852 * Since AES128 and AES256 share a single mask bit in the logic of
11853   ssl/ssl_ciph.c, the code for masking out disabled ciphers needs a
11854   kludge to work properly if AES128 is available and AES256 isn't.
11855
11856   *Victor Duchovni*
11857
11858 * Expand security boundary to match 1.1.1 module.
11859
11860   *Steve Henson*
11861
11862 * Remove redundant features: hash file source, editing of test vectors
11863   modify fipsld to use external fips_premain.c signature.
11864
11865   *Steve Henson*
11866
11867 * New perl script mkfipsscr.pl to create shell scripts or batch files to
11868   run algorithm test programs.
11869
11870   *Steve Henson*
11871
11872 * Make algorithm test programs more tolerant of whitespace.
11873
11874   *Steve Henson*
11875
11876 * Have SSL/TLS server implementation tolerate "mismatched" record
11877   protocol version while receiving ClientHello even if the
11878   ClientHello is fragmented.  (The server can't insist on the
11879   particular protocol version it has chosen before the ServerHello
11880   message has informed the client about his choice.)
11881
11882   *Bodo Moeller*
11883
11884 * Load error codes if they are not already present instead of using a
11885   static variable. This allows them to be cleanly unloaded and reloaded.
11886
11887   *Steve Henson*
11888
11889### Changes between 0.9.7k and 0.9.7l  [28 Sep 2006]
11890
11891 * Introduce limits to prevent malicious keys being able to
11892   cause a denial of service.  ([CVE-2006-2940])
11893
11894   *Steve Henson, Bodo Moeller*
11895
11896 * Fix ASN.1 parsing of certain invalid structures that can result
11897   in a denial of service.  ([CVE-2006-2937])  [Steve Henson]
11898
11899 * Fix buffer overflow in SSL_get_shared_ciphers() function.
11900   ([CVE-2006-3738]) [Tavis Ormandy and Will Drewry, Google Security Team]
11901
11902 * Fix SSL client code which could crash if connecting to a
11903   malicious SSLv2 server.  ([CVE-2006-4343])
11904
11905   *Tavis Ormandy and Will Drewry, Google Security Team*
11906
11907 * Change ciphersuite string processing so that an explicit
11908   ciphersuite selects this one ciphersuite (so that "AES256-SHA"
11909   will no longer include "AES128-SHA"), and any other similar
11910   ciphersuite (same bitmap) from *other* protocol versions (so that
11911   "RC4-MD5" will still include both the SSL 2.0 ciphersuite and the
11912   SSL 3.0/TLS 1.0 ciphersuite).  This is a backport combining
11913   changes from 0.9.8b and 0.9.8d.
11914
11915   *Bodo Moeller*
11916
11917### Changes between 0.9.7j and 0.9.7k  [05 Sep 2006]
11918
11919 * Avoid PKCS #1 v1.5 signature attack discovered by Daniel Bleichenbacher
11920   ([CVE-2006-4339])  [Ben Laurie and Google Security Team]
11921
11922 * Change the Unix randomness entropy gathering to use poll() when
11923   possible instead of select(), since the latter has some
11924   undesirable limitations.
11925
11926   *Darryl Miles via Richard Levitte and Bodo Moeller*
11927
11928 * Disable rogue ciphersuites:
11929
11930   - SSLv2 0x08 0x00 0x80 ("RC4-64-MD5")
11931   - SSLv3/TLSv1 0x00 0x61 ("EXP1024-RC2-CBC-MD5")
11932   - SSLv3/TLSv1 0x00 0x60 ("EXP1024-RC4-MD5")
11933
11934   The latter two were purportedly from
11935   draft-ietf-tls-56-bit-ciphersuites-0[01].txt, but do not really
11936   appear there.
11937
11938   Also deactivate the remaining ciphersuites from
11939   draft-ietf-tls-56-bit-ciphersuites-01.txt.  These are just as
11940   unofficial, and the ID has long expired.
11941
11942   *Bodo Moeller*
11943
11944 * Fix RSA blinding Heisenbug (problems sometimes occurred on
11945   dual-core machines) and other potential thread-safety issues.
11946
11947   *Bodo Moeller*
11948
11949### Changes between 0.9.7i and 0.9.7j  [04 May 2006]
11950
11951 * Adapt fipsld and the build system to link against the validated FIPS
11952   module in FIPS mode.
11953
11954   *Steve Henson*
11955
11956 * Fixes for VC++ 2005 build under Windows.
11957
11958   *Steve Henson*
11959
11960 * Add new Windows build target VC-32-GMAKE for VC++. This uses GNU make
11961   from a Windows bash shell such as MSYS. It is autodetected from the
11962   "config" script when run from a VC++ environment. Modify standard VC++
11963   build to use fipscanister.o from the GNU make build.
11964
11965   *Steve Henson*
11966
11967### Changes between 0.9.7h and 0.9.7i  [14 Oct 2005]
11968
11969 * Wrapped the definition of EVP_MAX_MD_SIZE in a #ifdef OPENSSL_FIPS.
11970   The value now differs depending on if you build for FIPS or not.
11971   BEWARE!  A program linked with a shared FIPSed libcrypto can't be
11972   safely run with a non-FIPSed libcrypto, as it may crash because of
11973   the difference induced by this change.
11974
11975   *Andy Polyakov*
11976
11977### Changes between 0.9.7g and 0.9.7h  [11 Oct 2005]
11978
11979 * Remove the functionality of SSL_OP_MSIE_SSLV2_RSA_PADDING
11980   (part of SSL_OP_ALL).  This option used to disable the
11981   countermeasure against man-in-the-middle protocol-version
11982   rollback in the SSL 2.0 server implementation, which is a bad
11983   idea.  ([CVE-2005-2969])
11984
11985   *Bodo Moeller; problem pointed out by Yutaka Oiwa (Research Center
11986   for Information Security, National Institute of Advanced Industrial
11987   Science and Technology [AIST, Japan)]*
11988
11989 * Minimal support for X9.31 signatures and PSS padding modes. This is
11990   mainly for FIPS compliance and not fully integrated at this stage.
11991
11992   *Steve Henson*
11993
11994 * For DSA signing, unless DSA_FLAG_NO_EXP_CONSTTIME is set, perform
11995   the exponentiation using a fixed-length exponent.  (Otherwise,
11996   the information leaked through timing could expose the secret key
11997   after many signatures; cf. Bleichenbacher's attack on DSA with
11998   biased k.)
11999
12000   *Bodo Moeller*
12001
12002 * Make a new fixed-window mod_exp implementation the default for
12003   RSA, DSA, and DH private-key operations so that the sequence of
12004   squares and multiplies and the memory access pattern are
12005   independent of the particular secret key.  This will mitigate
12006   cache-timing and potential related attacks.
12007
12008   BN_mod_exp_mont_consttime() is the new exponentiation implementation,
12009   and this is automatically used by BN_mod_exp_mont() if the new flag
12010   BN_FLG_EXP_CONSTTIME is set for the exponent.  RSA, DSA, and DH
12011   will use this BN flag for private exponents unless the flag
12012   RSA_FLAG_NO_EXP_CONSTTIME, DSA_FLAG_NO_EXP_CONSTTIME, or
12013   DH_FLAG_NO_EXP_CONSTTIME, respectively, is set.
12014
12015   *Matthew D Wood (Intel Corp), with some changes by Bodo Moeller*
12016
12017 * Change the client implementation for SSLv23_method() and
12018   SSLv23_client_method() so that is uses the SSL 3.0/TLS 1.0
12019   Client Hello message format if the SSL_OP_NO_SSLv2 option is set.
12020   (Previously, the SSL 2.0 backwards compatible Client Hello
12021   message format would be used even with SSL_OP_NO_SSLv2.)
12022
12023   *Bodo Moeller*
12024
12025 * Add support for smime-type MIME parameter in S/MIME messages which some
12026   clients need.
12027
12028   *Steve Henson*
12029
12030 * New function BN_MONT_CTX_set_locked() to set montgomery parameters in
12031   a threadsafe manner. Modify rsa code to use new function and add calls
12032   to dsa and dh code (which had race conditions before).
12033
12034   *Steve Henson*
12035
12036 * Include the fixed error library code in the C error file definitions
12037   instead of fixing them up at runtime. This keeps the error code
12038   structures constant.
12039
12040   *Steve Henson*
12041
12042### Changes between 0.9.7f and 0.9.7g  [11 Apr 2005]
12043
12044[NB: OpenSSL 0.9.7h and later 0.9.7 patch levels were released after
12045OpenSSL 0.9.8.]
12046
12047 * Fixes for newer kerberos headers. NB: the casts are needed because
12048   the 'length' field is signed on one version and unsigned on another
12049   with no (?) obvious way to tell the difference, without these VC++
12050   complains. Also the "definition" of FAR (blank) is no longer included
12051   nor is the error ENOMEM. KRB5_PRIVATE has to be set to 1 to pick up
12052   some needed definitions.
12053
12054   *Steve Henson*
12055
12056 * Undo Cygwin change.
12057
12058   *Ulf Möller*
12059
12060 * Added support for proxy certificates according to RFC 3820.
12061   Because they may be a security thread to unaware applications,
12062   they must be explicitly allowed in run-time.  See
12063   docs/HOWTO/proxy_certificates.txt for further information.
12064
12065   *Richard Levitte*
12066
12067### Changes between 0.9.7e and 0.9.7f  [22 Mar 2005]
12068
12069 * Use (SSL_RANDOM_VALUE - 4) bytes of pseudo random data when generating
12070   server and client random values. Previously
12071   (SSL_RANDOM_VALUE - sizeof(time_t)) would be used which would result in
12072   less random data when sizeof(time_t) > 4 (some 64 bit platforms).
12073
12074   This change has negligible security impact because:
12075
12076   1. Server and client random values still have 24 bytes of pseudo random
12077      data.
12078
12079   2. Server and client random values are sent in the clear in the initial
12080      handshake.
12081
12082   3. The master secret is derived using the premaster secret (48 bytes in
12083      size for static RSA ciphersuites) as well as client server and random
12084      values.
12085
12086   The OpenSSL team would like to thank the UK NISCC for bringing this issue
12087   to our attention.
12088
12089   *Stephen Henson, reported by UK NISCC*
12090
12091 * Use Windows randomness collection on Cygwin.
12092
12093   *Ulf Möller*
12094
12095 * Fix hang in EGD/PRNGD query when communication socket is closed
12096   prematurely by EGD/PRNGD.
12097
12098   *Darren Tucker <dtucker@zip.com.au> via Lutz Jänicke, resolves #1014*
12099
12100 * Prompt for pass phrases when appropriate for PKCS12 input format.
12101
12102   *Steve Henson*
12103
12104 * Back-port of selected performance improvements from development
12105   branch, as well as improved support for PowerPC platforms.
12106
12107   *Andy Polyakov*
12108
12109 * Add lots of checks for memory allocation failure, error codes to indicate
12110   failure and freeing up memory if a failure occurs.
12111
12112   *Nauticus Networks SSL Team <openssl@nauticusnet.com>, Steve Henson*
12113
12114 * Add new -passin argument to dgst.
12115
12116   *Steve Henson*
12117
12118 * Perform some character comparisons of different types in X509_NAME_cmp:
12119   this is needed for some certificates that re-encode DNs into UTF8Strings
12120   (in violation of RFC3280) and can't or won't issue name rollover
12121   certificates.
12122
12123   *Steve Henson*
12124
12125 * Make an explicit check during certificate validation to see that
12126   the CA setting in each certificate on the chain is correct.  As a
12127   side effect always do the following basic checks on extensions,
12128   not just when there's an associated purpose to the check:
12129
12130   - if there is an unhandled critical extension (unless the user
12131     has chosen to ignore this fault)
12132   - if the path length has been exceeded (if one is set at all)
12133   - that certain extensions fit the associated purpose (if one has
12134     been given)
12135
12136   *Richard Levitte*
12137
12138### Changes between 0.9.7d and 0.9.7e  [25 Oct 2004]
12139
12140 * Avoid a race condition when CRLs are checked in a multi threaded
12141   environment. This would happen due to the reordering of the revoked
12142   entries during signature checking and serial number lookup. Now the
12143   encoding is cached and the serial number sort performed under a lock.
12144   Add new STACK function sk_is_sorted().
12145
12146   *Steve Henson*
12147
12148 * Add Delta CRL to the extension code.
12149
12150   *Steve Henson*
12151
12152 * Various fixes to s3_pkt.c so alerts are sent properly.
12153
12154   *David Holmes <d.holmes@f5.com>*
12155
12156 * Reduce the chances of duplicate issuer name and serial numbers (in
12157   violation of RFC3280) using the OpenSSL certificate creation utilities.
12158   This is done by creating a random 64 bit value for the initial serial
12159   number when a serial number file is created or when a self signed
12160   certificate is created using 'openssl req -x509'. The initial serial
12161   number file is created using 'openssl x509 -next_serial' in CA.pl
12162   rather than being initialized to 1.
12163
12164   *Steve Henson*
12165
12166### Changes between 0.9.7c and 0.9.7d  [17 Mar 2004]
12167
12168 * Fix null-pointer assignment in do_change_cipher_spec() revealed
12169   by using the Codenomicon TLS Test Tool ([CVE-2004-0079])
12170
12171   *Joe Orton, Steve Henson*
12172
12173 * Fix flaw in SSL/TLS handshaking when using Kerberos ciphersuites
12174   ([CVE-2004-0112])
12175
12176   *Joe Orton, Steve Henson*
12177
12178 * Make it possible to have multiple active certificates with the same
12179   subject in the CA index file.  This is done only if the keyword
12180   'unique_subject' is set to 'no' in the main CA section (default
12181   if 'CA_default') of the configuration file.  The value is saved
12182   with the database itself in a separate index attribute file,
12183   named like the index file with '.attr' appended to the name.
12184
12185   *Richard Levitte*
12186
12187 * X509 verify fixes. Disable broken certificate workarounds when
12188   X509_V_FLAGS_X509_STRICT is set. Check CRL issuer has cRLSign set if
12189   keyUsage extension present. Don't accept CRLs with unhandled critical
12190   extensions: since verify currently doesn't process CRL extensions this
12191   rejects a CRL with *any* critical extensions. Add new verify error codes
12192   for these cases.
12193
12194   *Steve Henson*
12195
12196 * When creating an OCSP nonce use an OCTET STRING inside the extnValue.
12197   A clarification of RFC2560 will require the use of OCTET STRINGs and
12198   some implementations cannot handle the current raw format. Since OpenSSL
12199   copies and compares OCSP nonces as opaque blobs without any attempt at
12200   parsing them this should not create any compatibility issues.
12201
12202   *Steve Henson*
12203
12204 * New md flag EVP_MD_CTX_FLAG_REUSE this allows md_data to be reused when
12205   calling EVP_MD_CTX_copy_ex() to avoid calling OPENSSL_malloc(). Without
12206   this HMAC (and other) operations are several times slower than OpenSSL
12207   < 0.9.7.
12208
12209   *Steve Henson*
12210
12211 * Print out GeneralizedTime and UTCTime in ASN1_STRING_print_ex().
12212
12213   *Peter Sylvester <Peter.Sylvester@EdelWeb.fr>*
12214
12215 * Use the correct content when signing type "other".
12216
12217   *Steve Henson*
12218
12219### Changes between 0.9.7b and 0.9.7c  [30 Sep 2003]
12220
12221 * Fix various bugs revealed by running the NISCC test suite:
12222
12223   Stop out of bounds reads in the ASN1 code when presented with
12224   invalid tags (CVE-2003-0543 and CVE-2003-0544).
12225
12226   Free up ASN1_TYPE correctly if ANY type is invalid ([CVE-2003-0545]).
12227
12228   If verify callback ignores invalid public key errors don't try to check
12229   certificate signature with the NULL public key.
12230
12231   *Steve Henson*
12232
12233 * New -ignore_err option in ocsp application to stop the server
12234   exiting on the first error in a request.
12235
12236   *Steve Henson*
12237
12238 * In ssl3_accept() (ssl/s3_srvr.c) only accept a client certificate
12239   if the server requested one: as stated in TLS 1.0 and SSL 3.0
12240   specifications.
12241
12242   *Steve Henson*
12243
12244 * In ssl3_get_client_hello() (ssl/s3_srvr.c), tolerate additional
12245   extra data after the compression methods not only for TLS 1.0
12246   but also for SSL 3.0 (as required by the specification).
12247
12248   *Bodo Moeller; problem pointed out by Matthias Loepfe*
12249
12250 * Change X509_certificate_type() to mark the key as exported/exportable
12251   when it's 512 *bits* long, not 512 bytes.
12252
12253   *Richard Levitte*
12254
12255 * Change AES_cbc_encrypt() so it outputs exact multiple of
12256   blocks during encryption.
12257
12258   *Richard Levitte*
12259
12260 * Various fixes to base64 BIO and non blocking I/O. On write
12261   flushes were not handled properly if the BIO retried. On read
12262   data was not being buffered properly and had various logic bugs.
12263   This also affects blocking I/O when the data being decoded is a
12264   certain size.
12265
12266   *Steve Henson*
12267
12268 * Various S/MIME bugfixes and compatibility changes:
12269   output correct application/pkcs7 MIME type if
12270   PKCS7_NOOLDMIMETYPE is set. Tolerate some broken signatures.
12271   Output CR+LF for EOL if PKCS7_CRLFEOL is set (this makes opening
12272   of files as .eml work). Correctly handle very long lines in MIME
12273   parser.
12274
12275   *Steve Henson*
12276
12277### Changes between 0.9.7a and 0.9.7b  [10 Apr 2003]
12278
12279 * Countermeasure against the Klima-Pokorny-Rosa extension of
12280   Bleichbacher's attack on PKCS #1 v1.5 padding: treat
12281   a protocol version number mismatch like a decryption error
12282   in ssl3_get_client_key_exchange (ssl/s3_srvr.c).
12283
12284   *Bodo Moeller*
12285
12286 * Turn on RSA blinding by default in the default implementation
12287   to avoid a timing attack. Applications that don't want it can call
12288   RSA_blinding_off() or use the new flag RSA_FLAG_NO_BLINDING.
12289   They would be ill-advised to do so in most cases.
12290
12291   *Ben Laurie, Steve Henson, Geoff Thorpe, Bodo Moeller*
12292
12293 * Change RSA blinding code so that it works when the PRNG is not
12294   seeded (in this case, the secret RSA exponent is abused as
12295   an unpredictable seed -- if it is not unpredictable, there
12296   is no point in blinding anyway).  Make RSA blinding thread-safe
12297   by remembering the creator's thread ID in rsa->blinding and
12298   having all other threads use local one-time blinding factors
12299   (this requires more computation than sharing rsa->blinding, but
12300   avoids excessive locking; and if an RSA object is not shared
12301   between threads, blinding will still be very fast).
12302
12303   *Bodo Moeller*
12304
12305 * Fixed a typo bug that would cause ENGINE_set_default() to set an
12306   ENGINE as defaults for all supported algorithms irrespective of
12307   the 'flags' parameter. 'flags' is now honoured, so applications
12308   should make sure they are passing it correctly.
12309
12310   *Geoff Thorpe*
12311
12312 * Target "mingw" now allows native Windows code to be generated in
12313   the Cygwin environment as well as with the MinGW compiler.
12314
12315   *Ulf Moeller*
12316
12317### Changes between 0.9.7 and 0.9.7a  [19 Feb 2003]
12318
12319 * In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked
12320   via timing by performing a MAC computation even if incorrect
12321   block cipher padding has been found.  This is a countermeasure
12322   against active attacks where the attacker has to distinguish
12323   between bad padding and a MAC verification error. ([CVE-2003-0078])
12324
12325   *Bodo Moeller; problem pointed out by Brice Canvel (EPFL),
12326   Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and
12327   Martin Vuagnoux (EPFL, Ilion)*
12328
12329 * Make the no-err option work as intended.  The intention with no-err
12330   is not to have the whole error stack handling routines removed from
12331   libcrypto, it's only intended to remove all the function name and
12332   reason texts, thereby removing some of the footprint that may not
12333   be interesting if those errors aren't displayed anyway.
12334
12335   NOTE: it's still possible for any application or module to have its
12336   own set of error texts inserted.  The routines are there, just not
12337   used by default when no-err is given.
12338
12339   *Richard Levitte*
12340
12341 * Add support for FreeBSD on IA64.
12342
12343   *dirk.meyer@dinoex.sub.org via Richard Levitte, resolves #454*
12344
12345 * Adjust DES_cbc_cksum() so it returns the same value as the MIT
12346   Kerberos function mit_des_cbc_cksum().  Before this change,
12347   the value returned by DES_cbc_cksum() was like the one from
12348   mit_des_cbc_cksum(), except the bytes were swapped.
12349
12350   *Kevin Greaney <Kevin.Greaney@hp.com> and Richard Levitte*
12351
12352 * Allow an application to disable the automatic SSL chain building.
12353   Before this a rather primitive chain build was always performed in
12354   ssl3_output_cert_chain(): an application had no way to send the
12355   correct chain if the automatic operation produced an incorrect result.
12356
12357   Now the chain builder is disabled if either:
12358
12359   1. Extra certificates are added via SSL_CTX_add_extra_chain_cert().
12360
12361   2. The mode flag SSL_MODE_NO_AUTO_CHAIN is set.
12362
12363   The reasoning behind this is that an application would not want the
12364   auto chain building to take place if extra chain certificates are
12365   present and it might also want a means of sending no additional
12366   certificates (for example the chain has two certificates and the
12367   root is omitted).
12368
12369   *Steve Henson*
12370
12371 * Add the possibility to build without the ENGINE framework.
12372
12373   *Steven Reddie <smr@essemer.com.au> via Richard Levitte*
12374
12375 * Under Win32 gmtime() can return NULL: check return value in
12376   OPENSSL_gmtime(). Add error code for case where gmtime() fails.
12377
12378   *Steve Henson*
12379
12380 * DSA routines: under certain error conditions uninitialized BN objects
12381   could be freed. Solution: make sure initialization is performed early
12382   enough. (Reported and fix supplied by Ivan D Nestlerode <nestler@MIT.EDU>,
12383   Nils Larsch <nla@trustcenter.de> via PR#459)
12384
12385   *Lutz Jaenicke*
12386
12387 * Another fix for SSLv2 session ID handling: the session ID was incorrectly
12388   checked on reconnect on the client side, therefore session resumption
12389   could still fail with a "ssl session id is different" error. This
12390   behaviour is masked when SSL_OP_ALL is used due to
12391   SSL_OP_MICROSOFT_SESS_ID_BUG being set.
12392   Behaviour observed by Crispin Flowerday <crispin@flowerday.cx> as
12393   followup to PR #377.
12394
12395   *Lutz Jaenicke*
12396
12397 * IA-32 assembler support enhancements: unified ELF targets, support
12398   for SCO/Caldera platforms, fix for Cygwin shared build.
12399
12400   *Andy Polyakov*
12401
12402 * Add support for FreeBSD on sparc64.  As a consequence, support for
12403   FreeBSD on non-x86 processors is separate from x86 processors on
12404   the config script, much like the NetBSD support.
12405
12406   *Richard Levitte & Kris Kennaway <kris@obsecurity.org>*
12407
12408### Changes between 0.9.6h and 0.9.7  [31 Dec 2002]
12409
12410[NB: OpenSSL 0.9.6i and later 0.9.6 patch levels were released after
12411OpenSSL 0.9.7.]
12412
12413 * Fix session ID handling in SSLv2 client code: the SERVER FINISHED
12414   code (06) was taken as the first octet of the session ID and the last
12415   octet was ignored consequently. As a result SSLv2 client side session
12416   caching could not have worked due to the session ID mismatch between
12417   client and server.
12418   Behaviour observed by Crispin Flowerday <crispin@flowerday.cx> as
12419   PR #377.
12420
12421   *Lutz Jaenicke*
12422
12423 * Change the declaration of needed Kerberos libraries to use EX_LIBS
12424   instead of the special (and badly supported) LIBKRB5.  LIBKRB5 is
12425   removed entirely.
12426
12427   *Richard Levitte*
12428
12429 * The hw_ncipher.c engine requires dynamic locks.  Unfortunately, it
12430   seems that in spite of existing for more than a year, many application
12431   author have done nothing to provide the necessary callbacks, which
12432   means that this particular engine will not work properly anywhere.
12433   This is a very unfortunate situation which forces us, in the name
12434   of usability, to give the hw_ncipher.c a static lock, which is part
12435   of libcrypto.
12436   NOTE: This is for the 0.9.7 series ONLY.  This hack will never
12437   appear in 0.9.8 or later.  We EXPECT application authors to have
12438   dealt properly with this when 0.9.8 is released (unless we actually
12439   make such changes in the libcrypto locking code that changes will
12440   have to be made anyway).
12441
12442   *Richard Levitte*
12443
12444 * In asn1_d2i_read_bio() repeatedly call BIO_read() until all content
12445   octets have been read, EOF or an error occurs. Without this change
12446   some truncated ASN1 structures will not produce an error.
12447
12448   *Steve Henson*
12449
12450 * Disable Heimdal support, since it hasn't been fully implemented.
12451   Still give the possibility to force the use of Heimdal, but with
12452   warnings and a request that patches get sent to openssl-dev.
12453
12454   *Richard Levitte*
12455
12456 * Add the VC-CE target, introduce the WINCE sysname, and add
12457   INSTALL.WCE and appropriate conditionals to make it build.
12458
12459   *Steven Reddie <smr@essemer.com.au> via Richard Levitte*
12460
12461 * Change the DLL names for Cygwin to cygcrypto-x.y.z.dll and
12462   cygssl-x.y.z.dll, where x, y and z are the major, minor and
12463   edit numbers of the version.
12464
12465   *Corinna Vinschen <vinschen@redhat.com> and Richard Levitte*
12466
12467 * Introduce safe string copy and catenation functions
12468   (BUF_strlcpy() and BUF_strlcat()).
12469
12470   *Ben Laurie (CHATS) and Richard Levitte*
12471
12472 * Avoid using fixed-size buffers for one-line DNs.
12473
12474   *Ben Laurie (CHATS)*
12475
12476 * Add BUF_MEM_grow_clean() to avoid information leakage when
12477   resizing buffers containing secrets, and use where appropriate.
12478
12479   *Ben Laurie (CHATS)*
12480
12481 * Avoid using fixed size buffers for configuration file location.
12482
12483   *Ben Laurie (CHATS)*
12484
12485 * Avoid filename truncation for various CA files.
12486
12487   *Ben Laurie (CHATS)*
12488
12489 * Use sizeof in preference to magic numbers.
12490
12491   *Ben Laurie (CHATS)*
12492
12493 * Avoid filename truncation in cert requests.
12494
12495   *Ben Laurie (CHATS)*
12496
12497 * Add assertions to check for (supposedly impossible) buffer
12498   overflows.
12499
12500   *Ben Laurie (CHATS)*
12501
12502 * Don't cache truncated DNS entries in the local cache (this could
12503   potentially lead to a spoofing attack).
12504
12505   *Ben Laurie (CHATS)*
12506
12507 * Fix various buffers to be large enough for hex/decimal
12508   representations in a platform independent manner.
12509
12510   *Ben Laurie (CHATS)*
12511
12512 * Add CRYPTO_realloc_clean() to avoid information leakage when
12513   resizing buffers containing secrets, and use where appropriate.
12514
12515   *Ben Laurie (CHATS)*
12516
12517 * Add BIO_indent() to avoid much slightly worrying code to do
12518   indents.
12519
12520   *Ben Laurie (CHATS)*
12521
12522 * Convert sprintf()/BIO_puts() to BIO_printf().
12523
12524   *Ben Laurie (CHATS)*
12525
12526 * buffer_gets() could terminate with the buffer only half
12527   full. Fixed.
12528
12529   *Ben Laurie (CHATS)*
12530
12531 * Add assertions to prevent user-supplied crypto functions from
12532   overflowing internal buffers by having large block sizes, etc.
12533
12534   *Ben Laurie (CHATS)*
12535
12536 * New OPENSSL_assert() macro (similar to assert(), but enabled
12537   unconditionally).
12538
12539   *Ben Laurie (CHATS)*
12540
12541 * Eliminate unused copy of key in RC4.
12542
12543   *Ben Laurie (CHATS)*
12544
12545 * Eliminate unused and incorrectly sized buffers for IV in pem.h.
12546
12547   *Ben Laurie (CHATS)*
12548
12549 * Fix off-by-one error in EGD path.
12550
12551   *Ben Laurie (CHATS)*
12552
12553 * If RANDFILE path is too long, ignore instead of truncating.
12554
12555   *Ben Laurie (CHATS)*
12556
12557 * Eliminate unused and incorrectly sized X.509 structure
12558   CBCParameter.
12559
12560   *Ben Laurie (CHATS)*
12561
12562 * Eliminate unused and dangerous function knumber().
12563
12564   *Ben Laurie (CHATS)*
12565
12566 * Eliminate unused and dangerous structure, KSSL_ERR.
12567
12568   *Ben Laurie (CHATS)*
12569
12570 * Protect against overlong session ID context length in an encoded
12571   session object. Since these are local, this does not appear to be
12572   exploitable.
12573
12574   *Ben Laurie (CHATS)*
12575
12576 * Change from security patch (see 0.9.6e below) that did not affect
12577   the 0.9.6 release series:
12578
12579   Remote buffer overflow in SSL3 protocol - an attacker could
12580   supply an oversized master key in Kerberos-enabled versions.
12581   ([CVE-2002-0657])
12582
12583   *Ben Laurie (CHATS)*
12584
12585 * Change the SSL kerb5 codes to match RFC 2712.
12586
12587   *Richard Levitte*
12588
12589 * Make -nameopt work fully for req and add -reqopt switch.
12590
12591   *Michael Bell <michael.bell@rz.hu-berlin.de>, Steve Henson*
12592
12593 * The "block size" for block ciphers in CFB and OFB mode should be 1.
12594
12595   *Steve Henson, reported by Yngve Nysaeter Pettersen <yngve@opera.com>*
12596
12597 * Make sure tests can be performed even if the corresponding algorithms
12598   have been removed entirely.  This was also the last step to make
12599   OpenSSL compilable with DJGPP under all reasonable conditions.
12600
12601   *Richard Levitte, Doug Kaufman <dkaufman@rahul.net>*
12602
12603 * Add cipher selection rules COMPLEMENTOFALL and COMPLEMENTOFDEFAULT
12604   to allow version independent disabling of normally unselected ciphers,
12605   which may be activated as a side-effect of selecting a single cipher.
12606
12607   (E.g., cipher list string "RSA" enables ciphersuites that are left
12608   out of "ALL" because they do not provide symmetric encryption.
12609   "RSA:!COMPLEMEMENTOFALL" avoids these unsafe ciphersuites.)
12610
12611   *Lutz Jaenicke, Bodo Moeller*
12612
12613 * Add appropriate support for separate platform-dependent build
12614   directories.  The recommended way to make a platform-dependent
12615   build directory is the following (tested on Linux), maybe with
12616   some local tweaks:
12617
12618           # Place yourself outside of the OpenSSL source tree.  In
12619           # this example, the environment variable OPENSSL_SOURCE
12620           # is assumed to contain the absolute OpenSSL source directory.
12621           mkdir -p objtree/"`uname -s`-`uname -r`-`uname -m`"
12622           cd objtree/"`uname -s`-`uname -r`-`uname -m`"
12623           (cd $OPENSSL_SOURCE; find . -type f) | while read F; do
12624                   mkdir -p `dirname $F`
12625                   ln -s $OPENSSL_SOURCE/$F $F
12626           done
12627
12628   To be absolutely sure not to disturb the source tree, a "make clean"
12629   is a good thing.  If it isn't successful, don't worry about it,
12630   it probably means the source directory is very clean.
12631
12632   *Richard Levitte*
12633
12634 * Make sure any ENGINE control commands make local copies of string
12635   pointers passed to them whenever necessary. Otherwise it is possible
12636   the caller may have overwritten (or deallocated) the original string
12637   data when a later ENGINE operation tries to use the stored values.
12638
12639   *Götz Babin-Ebell <babinebell@trustcenter.de>*
12640
12641 * Improve diagnostics in file reading and command-line digests.
12642
12643   *Ben Laurie aided and abetted by Solar Designer <solar@openwall.com>*
12644
12645 * Add AES modes CFB and OFB to the object database.  Correct an
12646   error in AES-CFB decryption.
12647
12648   *Richard Levitte*
12649
12650 * Remove most calls to EVP_CIPHER_CTX_cleanup() in evp_enc.c, this
12651   allows existing EVP_CIPHER_CTX structures to be reused after
12652   calling `EVP_*Final()`. This behaviour is used by encryption
12653   BIOs and some applications. This has the side effect that
12654   applications must explicitly clean up cipher contexts with
12655   EVP_CIPHER_CTX_cleanup() or they will leak memory.
12656
12657   *Steve Henson*
12658
12659 * Check the values of dna and dnb in bn_mul_recursive before calling
12660   bn_mul_comba (a non zero value means the a or b arrays do not contain
12661   n2 elements) and fallback to bn_mul_normal if either is not zero.
12662
12663   *Steve Henson*
12664
12665 * Fix escaping of non-ASCII characters when using the -subj option
12666   of the "openssl req" command line tool. (Robert Joop <joop@fokus.gmd.de>)
12667
12668   *Lutz Jaenicke*
12669
12670 * Make object definitions compliant to LDAP (RFC2256): SN is the short
12671   form for "surname", serialNumber has no short form.
12672   Use "mail" as the short name for "rfc822Mailbox" according to RFC2798;
12673   therefore remove "mail" short name for "internet 7".
12674   The OID for unique identifiers in X509 certificates is
12675   x500UniqueIdentifier, not uniqueIdentifier.
12676   Some more OID additions. (Michael Bell <michael.bell@rz.hu-berlin.de>)
12677
12678   *Lutz Jaenicke*
12679
12680 * Add an "init" command to the ENGINE config module and auto initialize
12681   ENGINEs. Without any "init" command the ENGINE will be initialized
12682   after all ctrl commands have been executed on it. If init=1 the
12683   ENGINE is initialized at that point (ctrls before that point are run
12684   on the uninitialized ENGINE and after on the initialized one). If
12685   init=0 then the ENGINE will not be initialized at all.
12686
12687   *Steve Henson*
12688
12689 * Fix the 'app_verify_callback' interface so that the user-defined
12690   argument is actually passed to the callback: In the
12691   SSL_CTX_set_cert_verify_callback() prototype, the callback
12692   declaration has been changed from
12693           int (*cb)()
12694   into
12695           int (*cb)(X509_STORE_CTX *,void *);
12696   in ssl_verify_cert_chain (ssl/ssl_cert.c), the call
12697           i=s->ctx->app_verify_callback(&ctx)
12698   has been changed into
12699           i=s->ctx->app_verify_callback(&ctx, s->ctx->app_verify_arg).
12700
12701   To update applications using SSL_CTX_set_cert_verify_callback(),
12702   a dummy argument can be added to their callback functions.
12703
12704   *D. K. Smetters <smetters@parc.xerox.com>*
12705
12706 * Added the '4758cca' ENGINE to support IBM 4758 cards.
12707
12708   *Maurice Gittens <maurice@gittens.nl>, touchups by Geoff Thorpe*
12709
12710 * Add and OPENSSL_LOAD_CONF define which will cause
12711   OpenSSL_add_all_algorithms() to load the openssl.cnf config file.
12712   This allows older applications to transparently support certain
12713   OpenSSL features: such as crypto acceleration and dynamic ENGINE loading.
12714   Two new functions OPENSSL_add_all_algorithms_noconf() which will never
12715   load the config file and OPENSSL_add_all_algorithms_conf() which will
12716   always load it have also been added.
12717
12718   *Steve Henson*
12719
12720 * Add the OFB, CFB and CTR (all with 128 bit feedback) to AES.
12721   Adjust NIDs and EVP layer.
12722
12723   *Stephen Sprunk <stephen@sprunk.org> and Richard Levitte*
12724
12725 * Config modules support in openssl utility.
12726
12727   Most commands now load modules from the config file,
12728   though in a few (such as version) this isn't done
12729   because it couldn't be used for anything.
12730
12731   In the case of ca and req the config file used is
12732   the same as the utility itself: that is the -config
12733   command line option can be used to specify an
12734   alternative file.
12735
12736   *Steve Henson*
12737
12738 * Move default behaviour from OPENSSL_config(). If appname is NULL
12739   use "openssl_conf" if filename is NULL use default openssl config file.
12740
12741   *Steve Henson*
12742
12743 * Add an argument to OPENSSL_config() to allow the use of an alternative
12744   config section name. Add a new flag to tolerate a missing config file
12745   and move code to CONF_modules_load_file().
12746
12747   *Steve Henson*
12748
12749 * Support for crypto accelerator cards from Accelerated Encryption
12750   Processing, www.aep.ie.  (Use engine 'aep')
12751   The support was copied from 0.9.6c [engine] and adapted/corrected
12752   to work with the new engine framework.
12753
12754   *AEP Inc. and Richard Levitte*
12755
12756 * Support for SureWare crypto accelerator cards from Baltimore
12757   Technologies.  (Use engine 'sureware')
12758   The support was copied from 0.9.6c [engine] and adapted
12759   to work with the new engine framework.
12760
12761   *Richard Levitte*
12762
12763 * Have the CHIL engine fork-safe (as defined by nCipher) and actually
12764   make the newer ENGINE framework commands for the CHIL engine work.
12765
12766   *Toomas Kiisk <vix@cyber.ee> and Richard Levitte*
12767
12768 * Make it possible to produce shared libraries on ReliantUNIX.
12769
12770   *Robert Dahlem <Robert.Dahlem@ffm2.siemens.de> via Richard Levitte*
12771
12772 * Add the configuration target debug-linux-ppro.
12773   Make 'openssl rsa' use the general key loading routines
12774   implemented in `apps.c`, and make those routines able to
12775   handle the key format FORMAT_NETSCAPE and the variant
12776   FORMAT_IISSGC.
12777
12778   *Toomas Kiisk <vix@cyber.ee> via Richard Levitte*
12779
12780 * Fix a crashbug and a logic bug in hwcrhk_load_pubkey().
12781
12782   *Toomas Kiisk <vix@cyber.ee> via Richard Levitte*
12783
12784 * Add -keyform to rsautl, and document -engine.
12785
12786   *Richard Levitte, inspired by Toomas Kiisk <vix@cyber.ee>*
12787
12788 * Change BIO_new_file (crypto/bio/bss_file.c) to use new
12789   BIO_R_NO_SUCH_FILE error code rather than the generic
12790   ERR_R_SYS_LIB error code if fopen() fails with ENOENT.
12791
12792   *Ben Laurie*
12793
12794 * Add new functions
12795           ERR_peek_last_error
12796           ERR_peek_last_error_line
12797           ERR_peek_last_error_line_data.
12798   These are similar to
12799           ERR_peek_error
12800           ERR_peek_error_line
12801           ERR_peek_error_line_data,
12802   but report on the latest error recorded rather than the first one
12803   still in the error queue.
12804
12805   *Ben Laurie, Bodo Moeller*
12806
12807 * default_algorithms option in ENGINE config module. This allows things
12808   like:
12809   default_algorithms = ALL
12810   default_algorithms = RSA, DSA, RAND, CIPHERS, DIGESTS
12811
12812   *Steve Henson*
12813
12814 * Preliminary ENGINE config module.
12815
12816   *Steve Henson*
12817
12818 * New experimental application configuration code.
12819
12820   *Steve Henson*
12821
12822 * Change the AES code to follow the same name structure as all other
12823   symmetric ciphers, and behave the same way.  Move everything to
12824   the directory crypto/aes, thereby obsoleting crypto/rijndael.
12825
12826   *Stephen Sprunk <stephen@sprunk.org> and Richard Levitte*
12827
12828 * SECURITY: remove unsafe setjmp/signal interaction from ui_openssl.c.
12829
12830   *Ben Laurie and Theo de Raadt*
12831
12832 * Add option to output public keys in req command.
12833
12834   *Massimiliano Pala madwolf@openca.org*
12835
12836 * Use wNAFs in EC_POINTs_mul() for improved efficiency
12837   (up to about 10% better than before for P-192 and P-224).
12838
12839   *Bodo Moeller*
12840
12841 * New functions/macros
12842
12843           SSL_CTX_set_msg_callback(ctx, cb)
12844           SSL_CTX_set_msg_callback_arg(ctx, arg)
12845           SSL_set_msg_callback(ssl, cb)
12846           SSL_set_msg_callback_arg(ssl, arg)
12847
12848   to request calling a callback function
12849
12850           void cb(int write_p, int version, int content_type,
12851                   const void *buf, size_t len, SSL *ssl, void *arg)
12852
12853   whenever a protocol message has been completely received
12854   (write_p == 0) or sent (write_p == 1).  Here 'version' is the
12855   protocol version  according to which the SSL library interprets
12856   the current protocol message (SSL2_VERSION, SSL3_VERSION, or
12857   TLS1_VERSION).  'content_type' is 0 in the case of SSL 2.0, or
12858   the content type as defined in the SSL 3.0/TLS 1.0 protocol
12859   specification (change_cipher_spec(20), alert(21), handshake(22)).
12860   'buf' and 'len' point to the actual message, 'ssl' to the
12861   SSL object, and 'arg' is the application-defined value set by
12862   SSL[_CTX]_set_msg_callback_arg().
12863
12864   'openssl s_client' and 'openssl s_server' have new '-msg' options
12865   to enable a callback that displays all protocol messages.
12866
12867   *Bodo Moeller*
12868
12869 * Change the shared library support so shared libraries are built as
12870   soon as the corresponding static library is finished, and thereby get
12871   openssl and the test programs linked against the shared library.
12872   This still only happens when the keyword "shard" has been given to
12873   the configuration scripts.
12874
12875   NOTE: shared library support is still an experimental thing, and
12876   backward binary compatibility is still not guaranteed.
12877
12878   *"Maciej W. Rozycki" <macro@ds2.pg.gda.pl> and Richard Levitte*
12879
12880 * Add support for Subject Information Access extension.
12881
12882   *Peter Sylvester <Peter.Sylvester@EdelWeb.fr>*
12883
12884 * Make BUF_MEM_grow() behaviour more consistent: Initialise to zero
12885   additional bytes when new memory had to be allocated, not just
12886   when reusing an existing buffer.
12887
12888   *Bodo Moeller*
12889
12890 * New command line and configuration option 'utf8' for the req command.
12891   This allows field values to be specified as UTF8 strings.
12892
12893   *Steve Henson*
12894
12895 * Add -multi and -mr options to "openssl speed" - giving multiple parallel
12896   runs for the former and machine-readable output for the latter.
12897
12898   *Ben Laurie*
12899
12900 * Add '-noemailDN' option to 'openssl ca'.  This prevents inclusion
12901   of the e-mail address in the DN (i.e., it will go into a certificate
12902   extension only).  The new configuration file option 'email_in_dn = no'
12903   has the same effect.
12904
12905   *Massimiliano Pala madwolf@openca.org*
12906
12907 * Change all functions with names starting with `des_` to be starting
12908   with `DES_` instead.  Add wrappers that are compatible with libdes,
12909   but are named `_ossl_old_des_*`.  Finally, add macros that map the
12910   `des_*` symbols to the corresponding `_ossl_old_des_*` if libdes
12911   compatibility is desired.  If OpenSSL 0.9.6c compatibility is
12912   desired, the `des_*` symbols will be mapped to `DES_*`, with one
12913   exception.
12914
12915   Since we provide two compatibility mappings, the user needs to
12916   define the macro OPENSSL_DES_LIBDES_COMPATIBILITY if libdes
12917   compatibility is desired.  The default (i.e., when that macro
12918   isn't defined) is OpenSSL 0.9.6c compatibility.
12919
12920   There are also macros that enable and disable the support of old
12921   des functions altogether.  Those are OPENSSL_ENABLE_OLD_DES_SUPPORT
12922   and OPENSSL_DISABLE_OLD_DES_SUPPORT.  If none or both of those
12923   are defined, the default will apply: to support the old des routines.
12924
12925   In either case, one must include openssl/des.h to get the correct
12926   definitions.  Do not try to just include openssl/des_old.h, that
12927   won't work.
12928
12929   NOTE: This is a major break of an old API into a new one.  Software
12930   authors are encouraged to switch to the `DES_` style functions.  Some
12931   time in the future, des_old.h and the libdes compatibility functions
12932   will be disable (i.e. OPENSSL_DISABLE_OLD_DES_SUPPORT will be the
12933   default), and then completely removed.
12934
12935   *Richard Levitte*
12936
12937 * Test for certificates which contain unsupported critical extensions.
12938   If such a certificate is found during a verify operation it is
12939   rejected by default: this behaviour can be overridden by either
12940   handling the new error X509_V_ERR_UNHANDLED_CRITICAL_EXTENSION or
12941   by setting the verify flag X509_V_FLAG_IGNORE_CRITICAL. A new function
12942   X509_supported_extension() has also been added which returns 1 if a
12943   particular extension is supported.
12944
12945   *Steve Henson*
12946
12947 * Modify the behaviour of EVP cipher functions in similar way to digests
12948   to retain compatibility with existing code.
12949
12950   *Steve Henson*
12951
12952 * Modify the behaviour of EVP_DigestInit() and EVP_DigestFinal() to retain
12953   compatibility with existing code. In particular the 'ctx' parameter does
12954   not have to be to be initialized before the call to EVP_DigestInit() and
12955   it is tidied up after a call to EVP_DigestFinal(). New function
12956   EVP_DigestFinal_ex() which does not tidy up the ctx. Similarly function
12957   EVP_MD_CTX_copy() changed to not require the destination to be
12958   initialized valid and new function EVP_MD_CTX_copy_ex() added which
12959   requires the destination to be valid.
12960
12961   Modify all the OpenSSL digest calls to use EVP_DigestInit_ex(),
12962   EVP_DigestFinal_ex() and EVP_MD_CTX_copy_ex().
12963
12964   *Steve Henson*
12965
12966 * Change ssl3_get_message (ssl/s3_both.c) and the functions using it
12967   so that complete 'Handshake' protocol structures are kept in memory
12968   instead of overwriting 'msg_type' and 'length' with 'body' data.
12969
12970   *Bodo Moeller*
12971
12972 * Add an implementation of SSL_add_dir_cert_subjects_to_stack for Win32.
12973
12974   *Massimo Santin via Richard Levitte*
12975
12976 * Major restructuring to the underlying ENGINE code. This includes
12977   reduction of linker bloat, separation of pure "ENGINE" manipulation
12978   (initialisation, etc) from functionality dealing with implementations
12979   of specific crypto interfaces. This change also introduces integrated
12980   support for symmetric ciphers and digest implementations - so ENGINEs
12981   can now accelerate these by providing EVP_CIPHER and EVP_MD
12982   implementations of their own. This is detailed in
12983   [crypto/engine/README.md](crypto/engine/README.md)
12984   as it couldn't be adequately described here. However, there are a few
12985   API changes worth noting - some RSA, DSA, DH, and RAND functions that
12986   were changed in the original introduction of ENGINE code have now
12987   reverted back - the hooking from this code to ENGINE is now a good
12988   deal more passive and at run-time, operations deal directly with
12989   RSA_METHODs, DSA_METHODs (etc) as they did before, rather than
12990   dereferencing through an ENGINE pointer any more. Also, the ENGINE
12991   functions dealing with `BN_MOD_EXP[_CRT]` handlers have been removed -
12992   they were not being used by the framework as there is no concept of a
12993   BIGNUM_METHOD and they could not be generalised to the new
12994   'ENGINE_TABLE' mechanism that underlies the new code. Similarly,
12995   ENGINE_cpy() has been removed as it cannot be consistently defined in
12996   the new code.
12997
12998   *Geoff Thorpe*
12999
13000 * Change ASN1_GENERALIZEDTIME_check() to allow fractional seconds.
13001
13002   *Steve Henson*
13003
13004 * Change mkdef.pl to sort symbols that get the same entry number,
13005   and make sure the automatically generated functions `ERR_load_*`
13006   become part of libeay.num as well.
13007
13008   *Richard Levitte*
13009
13010 * New function SSL_renegotiate_pending().  This returns true once
13011   renegotiation has been requested (either SSL_renegotiate() call
13012   or HelloRequest/ClientHello received from the peer) and becomes
13013   false once a handshake has been completed.
13014   (For servers, SSL_renegotiate() followed by SSL_do_handshake()
13015   sends a HelloRequest, but does not ensure that a handshake takes
13016   place.  SSL_renegotiate_pending() is useful for checking if the
13017   client has followed the request.)
13018
13019   *Bodo Moeller*
13020
13021 * New SSL option SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION.
13022   By default, clients may request session resumption even during
13023   renegotiation (if session ID contexts permit); with this option,
13024   session resumption is possible only in the first handshake.
13025
13026   SSL_OP_ALL is now 0x00000FFFL instead of 0x000FFFFFL.  This makes
13027   more bits available for options that should not be part of
13028   SSL_OP_ALL (such as SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION).
13029
13030   *Bodo Moeller*
13031
13032 * Add some demos for certificate and certificate request creation.
13033
13034   *Steve Henson*
13035
13036 * Make maximum certificate chain size accepted from the peer application
13037   settable (`SSL*_get/set_max_cert_list()`), as proposed by
13038   "Douglas E. Engert" <deengert@anl.gov>.
13039
13040   *Lutz Jaenicke*
13041
13042 * Add support for shared libraries for Unixware-7
13043   (Boyd Lynn Gerber <gerberb@zenez.com>).
13044
13045   *Lutz Jaenicke*
13046
13047 * Add a "destroy" handler to ENGINEs that allows structural cleanup to
13048   be done prior to destruction. Use this to unload error strings from
13049   ENGINEs that load their own error strings. NB: This adds two new API
13050   functions to "get" and "set" this destroy handler in an ENGINE.
13051
13052   *Geoff Thorpe*
13053
13054 * Alter all existing ENGINE implementations (except "openssl" and
13055   "openbsd") to dynamically instantiate their own error strings. This
13056   makes them more flexible to be built both as statically-linked ENGINEs
13057   and self-contained shared-libraries loadable via the "dynamic" ENGINE.
13058   Also, add stub code to each that makes building them as self-contained
13059   shared-libraries easier (see [README-Engine.md](README-Engine.md)).
13060
13061   *Geoff Thorpe*
13062
13063 * Add a "dynamic" ENGINE that provides a mechanism for binding ENGINE
13064   implementations into applications that are completely implemented in
13065   self-contained shared-libraries. The "dynamic" ENGINE exposes control
13066   commands that can be used to configure what shared-library to load and
13067   to control aspects of the way it is handled. Also, made an update to
13068   the [README-Engine.md](README-Engine.md) file
13069   that brings its information up-to-date and
13070   provides some information and instructions on the "dynamic" ENGINE
13071   (ie. how to use it, how to build "dynamic"-loadable ENGINEs, etc).
13072
13073   *Geoff Thorpe*
13074
13075 * Make it possible to unload ranges of ERR strings with a new
13076   "ERR_unload_strings" function.
13077
13078   *Geoff Thorpe*
13079
13080 * Add a copy() function to EVP_MD.
13081
13082   *Ben Laurie*
13083
13084 * Make EVP_MD routines take a context pointer instead of just the
13085   md_data void pointer.
13086
13087   *Ben Laurie*
13088
13089 * Add flags to EVP_MD and EVP_MD_CTX. EVP_MD_FLAG_ONESHOT indicates
13090   that the digest can only process a single chunk of data
13091   (typically because it is provided by a piece of
13092   hardware). EVP_MD_CTX_FLAG_ONESHOT indicates that the application
13093   is only going to provide a single chunk of data, and hence the
13094   framework needn't accumulate the data for oneshot drivers.
13095
13096   *Ben Laurie*
13097
13098 * As with "ERR", make it possible to replace the underlying "ex_data"
13099   functions. This change also alters the storage and management of global
13100   ex_data state - it's now all inside ex_data.c and all "class" code (eg.
13101   RSA, BIO, SSL_CTX, etc) no longer stores its own STACKS and per-class
13102   index counters. The API functions that use this state have been changed
13103   to take a "class_index" rather than pointers to the class's local STACK
13104   and counter, and there is now an API function to dynamically create new
13105   classes. This centralisation allows us to (a) plug a lot of the
13106   thread-safety problems that existed, and (b) makes it possible to clean
13107   up all allocated state using "CRYPTO_cleanup_all_ex_data()". W.r.t. (b)
13108   such data would previously have always leaked in application code and
13109   workarounds were in place to make the memory debugging turn a blind eye
13110   to it. Application code that doesn't use this new function will still
13111   leak as before, but their memory debugging output will announce it now
13112   rather than letting it slide.
13113
13114   Besides the addition of CRYPTO_cleanup_all_ex_data(), another API change
13115   induced by the "ex_data" overhaul is that X509_STORE_CTX_init() now
13116   has a return value to indicate success or failure.
13117
13118   *Geoff Thorpe*
13119
13120 * Make it possible to replace the underlying "ERR" functions such that the
13121   global state (2 LHASH tables and 2 locks) is only used by the "default"
13122   implementation. This change also adds two functions to "get" and "set"
13123   the implementation prior to it being automatically set the first time
13124   any other ERR function takes place. Ie. an application can call "get",
13125   pass the return value to a module it has just loaded, and that module
13126   can call its own "set" function using that value. This means the
13127   module's "ERR" operations will use (and modify) the error state in the
13128   application and not in its own statically linked copy of OpenSSL code.
13129
13130   *Geoff Thorpe*
13131
13132 * Give DH, DSA, and RSA types their own `*_up_ref()` function to increment
13133   reference counts. This performs normal REF_PRINT/REF_CHECK macros on
13134   the operation, and provides a more encapsulated way for external code
13135   (crypto/evp/ and ssl/) to do this. Also changed the evp and ssl code
13136   to use these functions rather than manually incrementing the counts.
13137
13138   Also rename "DSO_up()" function to more descriptive "DSO_up_ref()".
13139
13140   *Geoff Thorpe*
13141
13142 * Add EVP test program.
13143
13144   *Ben Laurie*
13145
13146 * Add symmetric cipher support to ENGINE. Expect the API to change!
13147
13148   *Ben Laurie*
13149
13150 * New CRL functions: X509_CRL_set_version(), X509_CRL_set_issuer_name()
13151   X509_CRL_set_lastUpdate(), X509_CRL_set_nextUpdate(), X509_CRL_sort(),
13152   X509_REVOKED_set_serialNumber(), and X509_REVOKED_set_revocationDate().
13153   These allow a CRL to be built without having to access X509_CRL fields
13154   directly. Modify 'ca' application to use new functions.
13155
13156   *Steve Henson*
13157
13158 * Move SSL_OP_TLS_ROLLBACK_BUG out of the SSL_OP_ALL list of recommended
13159   bug workarounds. Rollback attack detection is a security feature.
13160   The problem will only arise on OpenSSL servers when TLSv1 is not
13161   available (sslv3_server_method() or SSL_OP_NO_TLSv1).
13162   Software authors not wanting to support TLSv1 will have special reasons
13163   for their choice and can explicitly enable this option.
13164
13165   *Bodo Moeller, Lutz Jaenicke*
13166
13167 * Rationalise EVP so it can be extended: don't include a union of
13168   cipher/digest structures, add init/cleanup functions for EVP_MD_CTX
13169   (similar to those existing for EVP_CIPHER_CTX).
13170   Usage example:
13171
13172           EVP_MD_CTX md;
13173
13174           EVP_MD_CTX_init(&md);             /* new function call */
13175           EVP_DigestInit(&md, EVP_sha1());
13176           EVP_DigestUpdate(&md, in, len);
13177           EVP_DigestFinal(&md, out, NULL);
13178           EVP_MD_CTX_cleanup(&md);          /* new function call */
13179
13180   *Ben Laurie*
13181
13182 * Make DES key schedule conform to the usual scheme, as well as
13183   correcting its structure. This means that calls to DES functions
13184   now have to pass a pointer to a des_key_schedule instead of a
13185   plain des_key_schedule (which was actually always a pointer
13186   anyway): E.g.,
13187
13188           des_key_schedule ks;
13189
13190           des_set_key_checked(..., &ks);
13191           des_ncbc_encrypt(..., &ks, ...);
13192
13193   (Note that a later change renames 'des_...' into 'DES_...'.)
13194
13195   *Ben Laurie*
13196
13197 * Initial reduction of linker bloat: the use of some functions, such as
13198   PEM causes large amounts of unused functions to be linked in due to
13199   poor organisation. For example pem_all.c contains every PEM function
13200   which has a knock on effect of linking in large amounts of (unused)
13201   ASN1 code. Grouping together similar functions and splitting unrelated
13202   functions prevents this.
13203
13204   *Steve Henson*
13205
13206 * Cleanup of EVP macros.
13207
13208   *Ben Laurie*
13209
13210 * Change historical references to `{NID,SN,LN}_des_ede` and ede3 to add the
13211   correct `_ecb suffix`.
13212
13213   *Ben Laurie*
13214
13215 * Add initial OCSP responder support to ocsp application. The
13216   revocation information is handled using the text based index
13217   use by the ca application. The responder can either handle
13218   requests generated internally, supplied in files (for example
13219   via a CGI script) or using an internal minimal server.
13220
13221   *Steve Henson*
13222
13223 * Add configuration choices to get zlib compression for TLS.
13224
13225   *Richard Levitte*
13226
13227 * Changes to Kerberos SSL for RFC 2712 compliance:
13228   1. Implemented real KerberosWrapper, instead of just using
13229      KRB5 AP_REQ message.  [Thanks to Simon Wilkinson <sxw@sxw.org.uk>]
13230   2. Implemented optional authenticator field of KerberosWrapper.
13231
13232   Added openssl-style ASN.1 macros for Kerberos ticket, ap_req,
13233   and authenticator structs; see crypto/krb5/.
13234
13235   Generalized Kerberos calls to support multiple Kerberos libraries.
13236   *Vern Staats <staatsvr@asc.hpc.mil>, Jeffrey Altman <jaltman@columbia.edu>
13237   via Richard Levitte*
13238
13239 * Cause 'openssl speed' to use fully hard-coded DSA keys as it
13240   already does with RSA. testdsa.h now has 'priv_key/pub_key'
13241   values for each of the key sizes rather than having just
13242   parameters (and 'speed' generating keys each time).
13243
13244   *Geoff Thorpe*
13245
13246 * Speed up EVP routines.
13247   Before:
13248crypt
13249pe              8 bytes     64 bytes    256 bytes   1024 bytes   8192 bytes
13250s-cbc           4408.85k     5560.51k     5778.46k     5862.20k     5825.16k
13251s-cbc           4389.55k     5571.17k     5792.23k     5846.91k     5832.11k
13252s-cbc           4394.32k     5575.92k     5807.44k     5848.37k     5841.30k
13253crypt
13254s-cbc           3482.66k     5069.49k     5496.39k     5614.16k     5639.28k
13255s-cbc           3480.74k     5068.76k     5510.34k     5609.87k     5635.52k
13256s-cbc           3483.72k     5067.62k     5504.60k     5708.01k     5724.80k
13257   After:
13258crypt
13259s-cbc           4660.16k     5650.19k     5807.19k     5827.13k     5783.32k
13260crypt
13261s-cbc           3624.96k     5258.21k     5530.91k     5624.30k     5628.26k
13262
13263   *Ben Laurie*
13264
13265 * Added the OS2-EMX target.
13266
13267   *"Brian Havard" <brianh@kheldar.apana.org.au> and Richard Levitte*
13268
13269 * Rewrite commands to use `NCONF` routines instead of the old `CONF`.
13270   New functions to support `NCONF` routines in extension code.
13271   New function `CONF_set_nconf()`
13272   to allow functions which take an `NCONF` to also handle the old `LHASH`
13273   structure: this means that the old `CONF` compatible routines can be
13274   retained (in particular w.rt. extensions) without having to duplicate the
13275   code. New function `X509V3_add_ext_nconf_sk()` to add extensions to a stack.
13276
13277   *Steve Henson*
13278
13279 * Enhance the general user interface with mechanisms for inner control
13280   and with possibilities to have yes/no kind of prompts.
13281
13282   *Richard Levitte*
13283
13284 * Change all calls to low-level digest routines in the library and
13285   applications to use EVP. Add missing calls to HMAC_cleanup() and
13286   don't assume HMAC_CTX can be copied using memcpy().
13287
13288   *Verdon Walker <VWalker@novell.com>, Steve Henson*
13289
13290 * Add the possibility to control engines through control names but with
13291   arbitrary arguments instead of just a string.
13292   Change the key loaders to take a UI_METHOD instead of a callback
13293   function pointer.  NOTE: this breaks binary compatibility with earlier
13294   versions of OpenSSL [engine].
13295   Adapt the nCipher code for these new conditions and add a card insertion
13296   callback.
13297
13298   *Richard Levitte*
13299
13300 * Enhance the general user interface with mechanisms to better support
13301   dialog box interfaces, application-defined prompts, the possibility
13302   to use defaults (for example default passwords from somewhere else)
13303   and interrupts/cancellations.
13304
13305   *Richard Levitte*
13306
13307 * Tidy up PKCS#12 attribute handling. Add support for the CSP name
13308   attribute in PKCS#12 files, add new -CSP option to pkcs12 utility.
13309
13310   *Steve Henson*
13311
13312 * Fix a memory leak in 'sk_dup()' in the case reallocation fails. (Also
13313   tidy up some unnecessarily weird code in 'sk_new()').
13314
13315   *Geoff, reported by Diego Tartara <dtartara@novamens.com>*
13316
13317 * Change the key loading routines for ENGINEs to use the same kind
13318   callback (pem_password_cb) as all other routines that need this
13319   kind of callback.
13320
13321   *Richard Levitte*
13322
13323 * Increase ENTROPY_NEEDED to 32 bytes, as Rijndael can operate with
13324   256 bit (=32 byte) keys. Of course seeding with more entropy bytes
13325   than this minimum value is recommended.
13326
13327   *Lutz Jaenicke*
13328
13329 * New random seeder for OpenVMS, using the system process statistics
13330   that are easily reachable.
13331
13332   *Richard Levitte*
13333
13334 * Windows apparently can't transparently handle global
13335   variables defined in DLLs. Initialisations such as:
13336
13337           const ASN1_ITEM *it = &ASN1_INTEGER_it;
13338
13339   won't compile. This is used by the any applications that need to
13340   declare their own ASN1 modules. This was fixed by adding the option
13341   EXPORT_VAR_AS_FN to all Win32 platforms, although this isn't strictly
13342   needed for static libraries under Win32.
13343
13344   *Steve Henson*
13345
13346 * New functions X509_PURPOSE_set() and X509_TRUST_set() to handle
13347   setting of purpose and trust fields. New X509_STORE trust and
13348   purpose functions and tidy up setting in other SSL functions.
13349
13350   *Steve Henson*
13351
13352 * Add copies of X509_STORE_CTX fields and callbacks to X509_STORE
13353   structure. These are inherited by X509_STORE_CTX when it is
13354   initialised. This allows various defaults to be set in the
13355   X509_STORE structure (such as flags for CRL checking and custom
13356   purpose or trust settings) for functions which only use X509_STORE_CTX
13357   internally such as S/MIME.
13358
13359   Modify X509_STORE_CTX_purpose_inherit() so it only sets purposes and
13360   trust settings if they are not set in X509_STORE. This allows X509_STORE
13361   purposes and trust (in S/MIME for example) to override any set by default.
13362
13363   Add command line options for CRL checking to smime, s_client and s_server
13364   applications.
13365
13366   *Steve Henson*
13367
13368 * Initial CRL based revocation checking. If the CRL checking flag(s)
13369   are set then the CRL is looked up in the X509_STORE structure and
13370   its validity and signature checked, then if the certificate is found
13371   in the CRL the verify fails with a revoked error.
13372
13373   Various new CRL related callbacks added to X509_STORE_CTX structure.
13374
13375   Command line options added to 'verify' application to support this.
13376
13377   This needs some additional work, such as being able to handle multiple
13378   CRLs with different times, extension based lookup (rather than just
13379   by subject name) and ultimately more complete V2 CRL extension
13380   handling.
13381
13382   *Steve Henson*
13383
13384 * Add a general user interface API (crypto/ui/).  This is designed
13385   to replace things like des_read_password and friends (backward
13386   compatibility functions using this new API are provided).
13387   The purpose is to remove prompting functions from the DES code
13388   section as well as provide for prompting through dialog boxes in
13389   a window system and the like.
13390
13391   *Richard Levitte*
13392
13393 * Add "ex_data" support to ENGINE so implementations can add state at a
13394   per-structure level rather than having to store it globally.
13395
13396   *Geoff*
13397
13398 * Make it possible for ENGINE structures to be copied when retrieved by
13399   ENGINE_by_id() if the ENGINE specifies a new flag: ENGINE_FLAGS_BY_ID_COPY.
13400   This causes the "original" ENGINE structure to act like a template,
13401   analogous to the RSA vs. RSA_METHOD type of separation. Because of this
13402   operational state can be localised to each ENGINE structure, despite the
13403   fact they all share the same "methods". New ENGINE structures returned in
13404   this case have no functional references and the return value is the single
13405   structural reference. This matches the single structural reference returned
13406   by ENGINE_by_id() normally, when it is incremented on the pre-existing
13407   ENGINE structure.
13408
13409   *Geoff*
13410
13411 * Fix ASN1 decoder when decoding type ANY and V_ASN1_OTHER: since this
13412   needs to match any other type at all we need to manually clear the
13413   tag cache.
13414
13415   *Steve Henson*
13416
13417 * Changes to the "openssl engine" utility to include;
13418   - verbosity levels ('-v', '-vv', and '-vvv') that provide information
13419     about an ENGINE's available control commands.
13420   - executing control commands from command line arguments using the
13421     '-pre' and '-post' switches. '-post' is only used if '-t' is
13422     specified and the ENGINE is successfully initialised. The syntax for
13423     the individual commands are colon-separated, for example;
13424           openssl engine chil -pre FORK_CHECK:0 -pre SO_PATH:/lib/test.so
13425
13426   *Geoff*
13427
13428 * New dynamic control command support for ENGINEs. ENGINEs can now
13429   declare their own commands (numbers), names (strings), descriptions,
13430   and input types for run-time discovery by calling applications. A
13431   subset of these commands are implicitly classed as "executable"
13432   depending on their input type, and only these can be invoked through
13433   the new string-based API function ENGINE_ctrl_cmd_string(). (Eg. this
13434   can be based on user input, config files, etc). The distinction is
13435   that "executable" commands cannot return anything other than a boolean
13436   result and can only support numeric or string input, whereas some
13437   discoverable commands may only be for direct use through
13438   ENGINE_ctrl(), eg. supporting the exchange of binary data, function
13439   pointers, or other custom uses. The "executable" commands are to
13440   support parameterisations of ENGINE behaviour that can be
13441   unambiguously defined by ENGINEs and used consistently across any
13442   OpenSSL-based application. Commands have been added to all the
13443   existing hardware-supporting ENGINEs, noticeably "SO_PATH" to allow
13444   control over shared-library paths without source code alterations.
13445
13446   *Geoff*
13447
13448 * Changed all ENGINE implementations to dynamically allocate their
13449   ENGINEs rather than declaring them statically. Apart from this being
13450   necessary with the removal of the ENGINE_FLAGS_MALLOCED distinction,
13451   this also allows the implementations to compile without using the
13452   internal engine_int.h header.
13453
13454   *Geoff*
13455
13456 * Minor adjustment to "rand" code. RAND_get_rand_method() now returns a
13457   'const' value. Any code that should be able to modify a RAND_METHOD
13458   should already have non-const pointers to it (ie. they should only
13459   modify their own ones).
13460
13461   *Geoff*
13462
13463 * Made a variety of little tweaks to the ENGINE code.
13464   - "atalla" and "ubsec" string definitions were moved from header files
13465     to C code. "nuron" string definitions were placed in variables
13466     rather than hard-coded - allowing parameterisation of these values
13467     later on via ctrl() commands.
13468   - Removed unused "#if 0"'d code.
13469   - Fixed engine list iteration code so it uses ENGINE_free() to release
13470     structural references.
13471   - Constified the RAND_METHOD element of ENGINE structures.
13472   - Constified various get/set functions as appropriate and added
13473     missing functions (including a catch-all ENGINE_cpy that duplicates
13474     all ENGINE values onto a new ENGINE except reference counts/state).
13475   - Removed NULL parameter checks in get/set functions. Setting a method
13476     or function to NULL is a way of cancelling out a previously set
13477     value.  Passing a NULL ENGINE parameter is just plain stupid anyway
13478     and doesn't justify the extra error symbols and code.
13479   - Deprecate the ENGINE_FLAGS_MALLOCED define and move the area for
13480     flags from engine_int.h to engine.h.
13481   - Changed prototypes for ENGINE handler functions (init(), finish(),
13482     ctrl(), key-load functions, etc) to take an (ENGINE*) parameter.
13483
13484   *Geoff*
13485
13486 * Implement binary inversion algorithm for BN_mod_inverse in addition
13487   to the algorithm using long division.  The binary algorithm can be
13488   used only if the modulus is odd.  On 32-bit systems, it is faster
13489   only for relatively small moduli (roughly 20-30% for 128-bit moduli,
13490   roughly 5-15% for 256-bit moduli), so we use it only for moduli
13491   up to 450 bits.  In 64-bit environments, the binary algorithm
13492   appears to be advantageous for much longer moduli; here we use it
13493   for moduli up to 2048 bits.
13494
13495   *Bodo Moeller*
13496
13497 * Rewrite CHOICE field setting in ASN1_item_ex_d2i(). The old code
13498   could not support the combine flag in choice fields.
13499
13500   *Steve Henson*
13501
13502 * Add a 'copy_extensions' option to the 'ca' utility. This copies
13503   extensions from a certificate request to the certificate.
13504
13505   *Steve Henson*
13506
13507 * Allow multiple 'certopt' and 'nameopt' options to be separated
13508   by commas. Add 'namopt' and 'certopt' options to the 'ca' config
13509   file: this allows the display of the certificate about to be
13510   signed to be customised, to allow certain fields to be included
13511   or excluded and extension details. The old system didn't display
13512   multicharacter strings properly, omitted fields not in the policy
13513   and couldn't display additional details such as extensions.
13514
13515   *Steve Henson*
13516
13517 * Function EC_POINTs_mul for multiple scalar multiplication
13518   of an arbitrary number of elliptic curve points
13519           \sum scalars[i]*points[i],
13520   optionally including the generator defined for the EC_GROUP:
13521           scalar*generator +  \sum scalars[i]*points[i].
13522
13523   EC_POINT_mul is a simple wrapper function for the typical case
13524   that the point list has just one item (besides the optional
13525   generator).
13526
13527   *Bodo Moeller*
13528
13529 * First EC_METHODs for curves over GF(p):
13530
13531   EC_GFp_simple_method() uses the basic BN_mod_mul and BN_mod_sqr
13532   operations and provides various method functions that can also
13533   operate with faster implementations of modular arithmetic.
13534
13535   EC_GFp_mont_method() reuses most functions that are part of
13536   EC_GFp_simple_method, but uses Montgomery arithmetic.
13537
13538   *Bodo Moeller; point addition and point doubling
13539   implementation directly derived from source code provided by
13540   Lenka Fibikova <fibikova@exp-math.uni-essen.de>*
13541
13542 * Framework for elliptic curves (crypto/ec/ec.h, crypto/ec/ec_lcl.h,
13543   crypto/ec/ec_lib.c):
13544
13545   Curves are EC_GROUP objects (with an optional group generator)
13546   based on EC_METHODs that are built into the library.
13547
13548   Points are EC_POINT objects based on EC_GROUP objects.
13549
13550   Most of the framework would be able to handle curves over arbitrary
13551   finite fields, but as there are no obvious types for fields other
13552   than GF(p), some functions are limited to that for now.
13553
13554   *Bodo Moeller*
13555
13556 * Add the -HTTP option to s_server.  It is similar to -WWW, but requires
13557   that the file contains a complete HTTP response.
13558
13559   *Richard Levitte*
13560
13561 * Add the ec directory to mkdef.pl and mkfiles.pl. In mkdef.pl
13562   change the def and num file printf format specifier from "%-40sXXX"
13563   to "%-39s XXX". The latter will always guarantee a space after the
13564   field while the former will cause them to run together if the field
13565   is 40 of more characters long.
13566
13567   *Steve Henson*
13568
13569 * Constify the cipher and digest 'method' functions and structures
13570   and modify related functions to take constant EVP_MD and EVP_CIPHER
13571   pointers.
13572
13573   *Steve Henson*
13574
13575 * Hide BN_CTX structure details in bn_lcl.h instead of publishing them
13576   in <openssl/bn.h>.  Also further increase BN_CTX_NUM to 32.
13577
13578   *Bodo Moeller*
13579
13580 * Modify `EVP_Digest*()` routines so they now return values. Although the
13581   internal software routines can never fail additional hardware versions
13582   might.
13583
13584   *Steve Henson*
13585
13586 * Clean up crypto/err/err.h and change some error codes to avoid conflicts:
13587
13588   Previously ERR_R_FATAL was too small and coincided with ERR_LIB_PKCS7
13589   (= ERR_R_PKCS7_LIB); it is now 64 instead of 32.
13590
13591   ASN1 error codes
13592           ERR_R_NESTED_ASN1_ERROR
13593           ...
13594           ERR_R_MISSING_ASN1_EOS
13595   were 4 .. 9, conflicting with
13596           ERR_LIB_RSA (= ERR_R_RSA_LIB)
13597           ...
13598           ERR_LIB_PEM (= ERR_R_PEM_LIB).
13599   They are now 58 .. 63 (i.e., just below ERR_R_FATAL).
13600
13601   Add new error code 'ERR_R_INTERNAL_ERROR'.
13602
13603   *Bodo Moeller*
13604
13605 * Don't overuse locks in crypto/err/err.c: For data retrieval, CRYPTO_r_lock
13606   suffices.
13607
13608   *Bodo Moeller*
13609
13610 * New option '-subj arg' for 'openssl req' and 'openssl ca'.  This
13611   sets the subject name for a new request or supersedes the
13612   subject name in a given request. Formats that can be parsed are
13613           'CN=Some Name, OU=myOU, C=IT'
13614   and
13615           'CN=Some Name/OU=myOU/C=IT'.
13616
13617   Add options '-batch' and '-verbose' to 'openssl req'.
13618
13619   *Massimiliano Pala <madwolf@hackmasters.net>*
13620
13621 * Introduce the possibility to access global variables through
13622   functions on platform were that's the best way to handle exporting
13623   global variables in shared libraries.  To enable this functionality,
13624   one must configure with "EXPORT_VAR_AS_FN" or defined the C macro
13625   "OPENSSL_EXPORT_VAR_AS_FUNCTION" in crypto/opensslconf.h (the latter
13626   is normally done by Configure or something similar).
13627
13628   To implement a global variable, use the macro OPENSSL_IMPLEMENT_GLOBAL
13629   in the source file (foo.c) like this:
13630
13631           OPENSSL_IMPLEMENT_GLOBAL(int,foo)=1;
13632           OPENSSL_IMPLEMENT_GLOBAL(double,bar);
13633
13634   To declare a global variable, use the macros OPENSSL_DECLARE_GLOBAL
13635   and OPENSSL_GLOBAL_REF in the header file (foo.h) like this:
13636
13637           OPENSSL_DECLARE_GLOBAL(int,foo);
13638           #define foo OPENSSL_GLOBAL_REF(foo)
13639           OPENSSL_DECLARE_GLOBAL(double,bar);
13640           #define bar OPENSSL_GLOBAL_REF(bar)
13641
13642   The #defines are very important, and therefore so is including the
13643   header file everywhere where the defined globals are used.
13644
13645   The macro OPENSSL_EXPORT_VAR_AS_FUNCTION also affects the definition
13646   of ASN.1 items, but that structure is a bit different.
13647
13648   The largest change is in util/mkdef.pl which has been enhanced with
13649   better and easier to understand logic to choose which symbols should
13650   go into the Windows .def files as well as a number of fixes and code
13651   cleanup (among others, algorithm keywords are now sorted
13652   lexicographically to avoid constant rewrites).
13653
13654   *Richard Levitte*
13655
13656 * In BN_div() keep a copy of the sign of 'num' before writing the
13657   result to 'rm' because if rm==num the value will be overwritten
13658   and produce the wrong result if 'num' is negative: this caused
13659   problems with BN_mod() and BN_nnmod().
13660
13661   *Steve Henson*
13662
13663 * Function OCSP_request_verify(). This checks the signature on an
13664   OCSP request and verifies the signer certificate. The signer
13665   certificate is just checked for a generic purpose and OCSP request
13666   trust settings.
13667
13668   *Steve Henson*
13669
13670 * Add OCSP_check_validity() function to check the validity of OCSP
13671   responses. OCSP responses are prepared in real time and may only
13672   be a few seconds old. Simply checking that the current time lies
13673   between thisUpdate and nextUpdate max reject otherwise valid responses
13674   caused by either OCSP responder or client clock inaccuracy. Instead
13675   we allow thisUpdate and nextUpdate to fall within a certain period of
13676   the current time. The age of the response can also optionally be
13677   checked. Two new options -validity_period and -status_age added to
13678   ocsp utility.
13679
13680   *Steve Henson*
13681
13682 * If signature or public key algorithm is unrecognized print out its
13683   OID rather that just UNKNOWN.
13684
13685   *Steve Henson*
13686
13687 * Change OCSP_cert_to_id() to tolerate a NULL subject certificate and
13688   OCSP_cert_id_new() a NULL serialNumber. This allows a partial certificate
13689   ID to be generated from the issuer certificate alone which can then be
13690   passed to OCSP_id_issuer_cmp().
13691
13692   *Steve Henson*
13693
13694 * New compilation option ASN1_ITEM_FUNCTIONS. This causes the new
13695   ASN1 modules to export functions returning ASN1_ITEM pointers
13696   instead of the ASN1_ITEM structures themselves. This adds several
13697   new macros which allow the underlying ASN1 function/structure to
13698   be accessed transparently. As a result code should not use ASN1_ITEM
13699   references directly (such as &X509_it) but instead use the relevant
13700   macros (such as ASN1_ITEM_rptr(X509)). This option is to allow
13701   use of the new ASN1 code on platforms where exporting structures
13702   is problematical (for example in shared libraries) but exporting
13703   functions returning pointers to structures is not.
13704
13705   *Steve Henson*
13706
13707 * Add support for overriding the generation of SSL/TLS session IDs.
13708   These callbacks can be registered either in an SSL_CTX or per SSL.
13709   The purpose of this is to allow applications to control, if they wish,
13710   the arbitrary values chosen for use as session IDs, particularly as it
13711   can be useful for session caching in multiple-server environments. A
13712   command-line switch for testing this (and any client code that wishes
13713   to use such a feature) has been added to "s_server".
13714
13715   *Geoff Thorpe, Lutz Jaenicke*
13716
13717 * Modify mkdef.pl to recognise and parse preprocessor conditionals
13718   of the form `#if defined(...) || defined(...) || ...` and
13719   `#if !defined(...) && !defined(...) && ...`.  This also avoids
13720   the growing number of special cases it was previously handling.
13721
13722   *Richard Levitte*
13723
13724 * Make all configuration macros available for application by making
13725   sure they are available in opensslconf.h, by giving them names starting
13726   with `OPENSSL_` to avoid conflicts with other packages and by making
13727   sure e_os2.h will cover all platform-specific cases together with
13728   opensslconf.h.
13729   Additionally, it is now possible to define configuration/platform-
13730   specific names (called "system identities").  In the C code, these
13731   are prefixed with `OPENSSL_SYSNAME_`.  e_os2.h will create another
13732   macro with the name beginning with `OPENSSL_SYS_`, which is determined
13733   from `OPENSSL_SYSNAME_*` or compiler-specific macros depending on
13734   what is available.
13735
13736   *Richard Levitte*
13737
13738 * New option -set_serial to 'req' and 'x509' this allows the serial
13739   number to use to be specified on the command line. Previously self
13740   signed certificates were hard coded with serial number 0 and the
13741   CA options of 'x509' had to use a serial number in a file which was
13742   auto incremented.
13743
13744   *Steve Henson*
13745
13746 * New options to 'ca' utility to support V2 CRL entry extensions.
13747   Currently CRL reason, invalidity date and hold instruction are
13748   supported. Add new CRL extensions to V3 code and some new objects.
13749
13750   *Steve Henson*
13751
13752 * New function EVP_CIPHER_CTX_set_padding() this is used to
13753   disable standard block padding (aka PKCS#5 padding) in the EVP
13754   API, which was previously mandatory. This means that the data is
13755   not padded in any way and so the total length much be a multiple
13756   of the block size, otherwise an error occurs.
13757
13758   *Steve Henson*
13759
13760 * Initial (incomplete) OCSP SSL support.
13761
13762   *Steve Henson*
13763
13764 * New function OCSP_parse_url(). This splits up a URL into its host,
13765   port and path components: primarily to parse OCSP URLs. New -url
13766   option to ocsp utility.
13767
13768   *Steve Henson*
13769
13770 * New nonce behavior. The return value of OCSP_check_nonce() now
13771   reflects the various checks performed. Applications can decide
13772   whether to tolerate certain situations such as an absent nonce
13773   in a response when one was present in a request: the ocsp application
13774   just prints out a warning. New function OCSP_add1_basic_nonce()
13775   this is to allow responders to include a nonce in a response even if
13776   the request is nonce-less.
13777
13778   *Steve Henson*
13779
13780 * Disable stdin buffering in `load_cert()` (`apps/apps.c`) so that no certs are
13781   skipped when using openssl x509 multiple times on a single input file,
13782   e.g. `(openssl x509 -out cert1; openssl x509 -out cert2) <certs`.
13783
13784   *Bodo Moeller*
13785
13786 * Make ASN1_UTCTIME_set_string() and ASN1_GENERALIZEDTIME_set_string()
13787   set string type: to handle setting ASN1_TIME structures. Fix ca
13788   utility to correctly initialize revocation date of CRLs.
13789
13790   *Steve Henson*
13791
13792 * New option SSL_OP_CIPHER_SERVER_PREFERENCE allows the server to override
13793   the clients preferred ciphersuites and rather use its own preferences.
13794   Should help to work around M$ SGC (Server Gated Cryptography) bug in
13795   Internet Explorer by ensuring unchanged hash method during stepup.
13796   (Also replaces the broken/deactivated SSL_OP_NON_EXPORT_FIRST option.)
13797
13798   *Lutz Jaenicke*
13799
13800 * Make mkdef.pl recognise all DECLARE_ASN1 macros, change rijndael
13801   to aes and add a new 'exist' option to print out symbols that don't
13802   appear to exist.
13803
13804   *Steve Henson*
13805
13806 * Additional options to ocsp utility to allow flags to be set and
13807   additional certificates supplied.
13808
13809   *Steve Henson*
13810
13811 * Add the option -VAfile to 'openssl ocsp', so the user can give the
13812   OCSP client a number of certificate to only verify the response
13813   signature against.
13814
13815   *Richard Levitte*
13816
13817 * Update Rijndael code to version 3.0 and change EVP AES ciphers to
13818   handle the new API. Currently only ECB, CBC modes supported. Add new
13819   AES OIDs.
13820
13821   Add TLS AES ciphersuites as described in RFC3268, "Advanced
13822   Encryption Standard (AES) Ciphersuites for Transport Layer
13823   Security (TLS)".  (In beta versions of OpenSSL 0.9.7, these were
13824   not enabled by default and were not part of the "ALL" ciphersuite
13825   alias because they were not yet official; they could be
13826   explicitly requested by specifying the "AESdraft" ciphersuite
13827   group alias.  In the final release of OpenSSL 0.9.7, the group
13828   alias is called "AES" and is part of "ALL".)
13829
13830   *Ben Laurie, Steve  Henson, Bodo Moeller*
13831
13832 * New function OCSP_copy_nonce() to copy nonce value (if present) from
13833   request to response.
13834
13835   *Steve Henson*
13836
13837 * Functions for OCSP responders. OCSP_request_onereq_count(),
13838   OCSP_request_onereq_get0(), OCSP_onereq_get0_id() and OCSP_id_get0_info()
13839   extract information from a certificate request. OCSP_response_create()
13840   creates a response and optionally adds a basic response structure.
13841   OCSP_basic_add1_status() adds a complete single response to a basic
13842   response and returns the OCSP_SINGLERESP structure just added (to allow
13843   extensions to be included for example). OCSP_basic_add1_cert() adds a
13844   certificate to a basic response and OCSP_basic_sign() signs a basic
13845   response with various flags. New helper functions ASN1_TIME_check()
13846   (checks validity of ASN1_TIME structure) and ASN1_TIME_to_generalizedtime()
13847   (converts ASN1_TIME to GeneralizedTime).
13848
13849   *Steve Henson*
13850
13851 * Various new functions. EVP_Digest() combines EVP_Digest{Init,Update,Final}()
13852   in a single operation. X509_get0_pubkey_bitstr() extracts the public_key
13853   structure from a certificate. X509_pubkey_digest() digests the public_key
13854   contents: this is used in various key identifiers.
13855
13856   *Steve Henson*
13857
13858 * Make sk_sort() tolerate a NULL argument.
13859
13860   *Steve Henson reported by Massimiliano Pala <madwolf@comune.modena.it>*
13861
13862 * New OCSP verify flag OCSP_TRUSTOTHER. When set the "other" certificates
13863   passed by the function are trusted implicitly. If any of them signed the
13864   response then it is assumed to be valid and is not verified.
13865
13866   *Steve Henson*
13867
13868 * In PKCS7_set_type() initialise content_type in PKCS7_ENC_CONTENT
13869   to data. This was previously part of the PKCS7 ASN1 code. This
13870   was causing problems with OpenSSL created PKCS#12 and PKCS#7 structures.
13871   *Steve Henson, reported by Kenneth R. Robinette
13872                              <support@securenetterm.com>*
13873
13874 * Add CRYPTO_push_info() and CRYPTO_pop_info() calls to new ASN1
13875   routines: without these tracing memory leaks is very painful.
13876   Fix leaks in PKCS12 and PKCS7 routines.
13877
13878   *Steve Henson*
13879
13880 * Make X509_time_adj() cope with the new behaviour of ASN1_TIME_new().
13881   Previously it initialised the 'type' argument to V_ASN1_UTCTIME which
13882   effectively meant GeneralizedTime would never be used. Now it
13883   is initialised to -1 but X509_time_adj() now has to check the value
13884   and use ASN1_TIME_set() if the value is not V_ASN1_UTCTIME or
13885   V_ASN1_GENERALIZEDTIME, without this it always uses GeneralizedTime.
13886   *Steve Henson, reported by Kenneth R. Robinette
13887                              <support@securenetterm.com>*
13888
13889 * Fixes to BN_to_ASN1_INTEGER when bn is zero. This would previously
13890   result in a zero length in the ASN1_INTEGER structure which was
13891   not consistent with the structure when d2i_ASN1_INTEGER() was used
13892   and would cause ASN1_INTEGER_cmp() to fail. Enhance s2i_ASN1_INTEGER()
13893   to cope with hex and negative integers. Fix bug in i2a_ASN1_INTEGER()
13894   where it did not print out a minus for negative ASN1_INTEGER.
13895
13896   *Steve Henson*
13897
13898 * Add summary printout to ocsp utility. The various functions which
13899   convert status values to strings have been renamed to:
13900   OCSP_response_status_str(), OCSP_cert_status_str() and
13901   OCSP_crl_reason_str() and are no longer static. New options
13902   to verify nonce values and to disable verification. OCSP response
13903   printout format cleaned up.
13904
13905   *Steve Henson*
13906
13907 * Add additional OCSP certificate checks. These are those specified
13908   in RFC2560. This consists of two separate checks: the CA of the
13909   certificate being checked must either be the OCSP signer certificate
13910   or the issuer of the OCSP signer certificate. In the latter case the
13911   OCSP signer certificate must contain the OCSP signing extended key
13912   usage. This check is performed by attempting to match the OCSP
13913   signer or the OCSP signer CA to the issuerNameHash and issuerKeyHash
13914   in the OCSP_CERTID structures of the response.
13915
13916   *Steve Henson*
13917
13918 * Initial OCSP certificate verification added to OCSP_basic_verify()
13919   and related routines. This uses the standard OpenSSL certificate
13920   verify routines to perform initial checks (just CA validity) and
13921   to obtain the certificate chain. Then additional checks will be
13922   performed on the chain. Currently the root CA is checked to see
13923   if it is explicitly trusted for OCSP signing. This is used to set
13924   a root CA as a global signing root: that is any certificate that
13925   chains to that CA is an acceptable OCSP signing certificate.
13926
13927   *Steve Henson*
13928
13929 * New '-extfile ...' option to 'openssl ca' for reading X.509v3
13930   extensions from a separate configuration file.
13931   As when reading extensions from the main configuration file,
13932   the '-extensions ...' option may be used for specifying the
13933   section to use.
13934
13935   *Massimiliano Pala <madwolf@comune.modena.it>*
13936
13937 * New OCSP utility. Allows OCSP requests to be generated or
13938   read. The request can be sent to a responder and the output
13939   parsed, outputted or printed in text form. Not complete yet:
13940   still needs to check the OCSP response validity.
13941
13942   *Steve Henson*
13943
13944 * New subcommands for 'openssl ca':
13945   `openssl ca -status <serial>` prints the status of the cert with
13946   the given serial number (according to the index file).
13947   `openssl ca -updatedb` updates the expiry status of certificates
13948   in the index file.
13949
13950   *Massimiliano Pala <madwolf@comune.modena.it>*
13951
13952 * New '-newreq-nodes' command option to CA.pl.  This is like
13953   '-newreq', but calls 'openssl req' with the '-nodes' option
13954   so that the resulting key is not encrypted.
13955
13956   *Damien Miller <djm@mindrot.org>*
13957
13958 * New configuration for the GNU Hurd.
13959
13960   *Jonathan Bartlett <johnnyb@wolfram.com> via Richard Levitte*
13961
13962 * Initial code to implement OCSP basic response verify. This
13963   is currently incomplete. Currently just finds the signer's
13964   certificate and verifies the signature on the response.
13965
13966   *Steve Henson*
13967
13968 * New SSLeay_version code SSLEAY_DIR to determine the compiled-in
13969   value of OPENSSLDIR.  This is available via the new '-d' option
13970   to 'openssl version', and is also included in 'openssl version -a'.
13971
13972   *Bodo Moeller*
13973
13974 * Allowing defining memory allocation callbacks that will be given
13975   file name and line number information in additional arguments
13976   (a `const char*` and an int).  The basic functionality remains, as
13977   well as the original possibility to just replace malloc(),
13978   realloc() and free() by functions that do not know about these
13979   additional arguments.  To register and find out the current
13980   settings for extended allocation functions, the following
13981   functions are provided:
13982
13983           CRYPTO_set_mem_ex_functions
13984           CRYPTO_set_locked_mem_ex_functions
13985           CRYPTO_get_mem_ex_functions
13986           CRYPTO_get_locked_mem_ex_functions
13987
13988   These work the same way as CRYPTO_set_mem_functions and friends.
13989   `CRYPTO_get_[locked_]mem_functions` now writes 0 where such an
13990   extended allocation function is enabled.
13991   Similarly, `CRYPTO_get_[locked_]mem_ex_functions` writes 0 where
13992   a conventional allocation function is enabled.
13993
13994   *Richard Levitte, Bodo Moeller*
13995
13996 * Finish off removing the remaining LHASH function pointer casts.
13997   There should no longer be any prototype-casting required when using
13998   the LHASH abstraction, and any casts that remain are "bugs". See
13999   the callback types and macros at the head of lhash.h for details
14000   (and "OBJ_cleanup" in crypto/objects/obj_dat.c as an example).
14001
14002   *Geoff Thorpe*
14003
14004 * Add automatic query of EGD sockets in RAND_poll() for the unix variant.
14005   If /dev/[u]random devices are not available or do not return enough
14006   entropy, EGD style sockets (served by EGD or PRNGD) will automatically
14007   be queried.
14008   The locations /var/run/egd-pool, /dev/egd-pool, /etc/egd-pool, and
14009   /etc/entropy will be queried once each in this sequence, querying stops
14010   when enough entropy was collected without querying more sockets.
14011
14012   *Lutz Jaenicke*
14013
14014 * Change the Unix RAND_poll() variant to be able to poll several
14015   random devices, as specified by DEVRANDOM, until a sufficient amount
14016   of data has been collected.   We spend at most 10 ms on each file
14017   (select timeout) and read in non-blocking mode.  DEVRANDOM now
14018   defaults to the list "/dev/urandom", "/dev/random", "/dev/srandom"
14019   (previously it was just the string "/dev/urandom"), so on typical
14020   platforms the 10 ms delay will never occur.
14021   Also separate out the Unix variant to its own file, rand_unix.c.
14022   For VMS, there's a currently-empty rand_vms.c.
14023
14024   *Richard Levitte*
14025
14026 * Move OCSP client related routines to ocsp_cl.c. These
14027   provide utility functions which an application needing
14028   to issue a request to an OCSP responder and analyse the
14029   response will typically need: as opposed to those which an
14030   OCSP responder itself would need which will be added later.
14031
14032   OCSP_request_sign() signs an OCSP request with an API similar
14033   to PKCS7_sign(). OCSP_response_status() returns status of OCSP
14034   response. OCSP_response_get1_basic() extracts basic response
14035   from response. OCSP_resp_find_status(): finds and extracts status
14036   information from an OCSP_CERTID structure (which will be created
14037   when the request structure is built). These are built from lower
14038   level functions which work on OCSP_SINGLERESP structures but
14039   won't normally be used unless the application wishes to examine
14040   extensions in the OCSP response for example.
14041
14042   Replace nonce routines with a pair of functions.
14043   OCSP_request_add1_nonce() adds a nonce value and optionally
14044   generates a random value. OCSP_check_nonce() checks the
14045   validity of the nonce in an OCSP response.
14046
14047   *Steve Henson*
14048
14049 * Change function OCSP_request_add() to OCSP_request_add0_id().
14050   This doesn't copy the supplied OCSP_CERTID and avoids the
14051   need to free up the newly created id. Change return type
14052   to OCSP_ONEREQ to return the internal OCSP_ONEREQ structure.
14053   This can then be used to add extensions to the request.
14054   Deleted OCSP_request_new(), since most of its functionality
14055   is now in OCSP_REQUEST_new() (and the case insensitive name
14056   clash) apart from the ability to set the request name which
14057   will be added elsewhere.
14058
14059   *Steve Henson*
14060
14061 * Update OCSP API. Remove obsolete extensions argument from
14062   various functions. Extensions are now handled using the new
14063   OCSP extension code. New simple OCSP HTTP function which
14064   can be used to send requests and parse the response.
14065
14066   *Steve Henson*
14067
14068 * Fix the PKCS#7 (S/MIME) code to work with new ASN1. Two new
14069   ASN1_ITEM structures help with sign and verify. PKCS7_ATTR_SIGN
14070   uses the special reorder version of SET OF to sort the attributes
14071   and reorder them to match the encoded order. This resolves a long
14072   standing problem: a verify on a PKCS7 structure just after signing
14073   it used to fail because the attribute order did not match the
14074   encoded order. PKCS7_ATTR_VERIFY does not reorder the attributes:
14075   it uses the received order. This is necessary to tolerate some broken
14076   software that does not order SET OF. This is handled by encoding
14077   as a SEQUENCE OF but using implicit tagging (with UNIVERSAL class)
14078   to produce the required SET OF.
14079
14080   *Steve Henson*
14081
14082 * Have mk1mf.pl generate the macros OPENSSL_BUILD_SHLIBCRYPTO and
14083   OPENSSL_BUILD_SHLIBSSL and use them appropriately in the header
14084   files to get correct declarations of the ASN.1 item variables.
14085
14086   *Richard Levitte*
14087
14088 * Rewrite of PKCS#12 code to use new ASN1 functionality. Replace many
14089   PKCS#12 macros with real functions. Fix two unrelated ASN1 bugs:
14090   asn1_check_tlen() would sometimes attempt to use 'ctx' when it was
14091   NULL and ASN1_TYPE was not dereferenced properly in asn1_ex_c2i().
14092   New ASN1 macro: DECLARE_ASN1_ITEM() which just declares the relevant
14093   ASN1_ITEM and no wrapper functions.
14094
14095   *Steve Henson*
14096
14097 * New functions or ASN1_item_d2i_fp() and ASN1_item_d2i_bio(). These
14098   replace the old function pointer based I/O routines. Change most of
14099   the `*_d2i_bio()` and `*_d2i_fp()` functions to use these.
14100
14101   *Steve Henson*
14102
14103 * Enhance mkdef.pl to be more accepting about spacing in C preprocessor
14104   lines, recognize more "algorithms" that can be deselected, and make
14105   it complain about algorithm deselection that isn't recognised.
14106
14107   *Richard Levitte*
14108
14109 * New ASN1 functions to handle dup, sign, verify, digest, pack and
14110   unpack operations in terms of ASN1_ITEM. Modify existing wrappers
14111   to use new functions. Add NO_ASN1_OLD which can be set to remove
14112   some old style ASN1 functions: this can be used to determine if old
14113   code will still work when these eventually go away.
14114
14115   *Steve Henson*
14116
14117 * New extension functions for OCSP structures, these follow the
14118   same conventions as certificates and CRLs.
14119
14120   *Steve Henson*
14121
14122 * New function X509V3_add1_i2d(). This automatically encodes and
14123   adds an extension. Its behaviour can be customised with various
14124   flags to append, replace or delete. Various wrappers added for
14125   certificates and CRLs.
14126
14127   *Steve Henson*
14128
14129 * Fix to avoid calling the underlying ASN1 print routine when
14130   an extension cannot be parsed. Correct a typo in the
14131   OCSP_SERVICELOC extension. Tidy up print OCSP format.
14132
14133   *Steve Henson*
14134
14135 * Make mkdef.pl parse some of the ASN1 macros and add appropriate
14136   entries for variables.
14137
14138   *Steve Henson*
14139
14140 * Add functionality to `apps/openssl.c` for detecting locking
14141   problems: As the program is single-threaded, all we have
14142   to do is register a locking callback using an array for
14143   storing which locks are currently held by the program.
14144
14145   *Bodo Moeller*
14146
14147 * Use a lock around the call to CRYPTO_get_ex_new_index() in
14148   SSL_get_ex_data_X509_STORE_idx(), which is used in
14149   ssl_verify_cert_chain() and thus can be called at any time
14150   during TLS/SSL handshakes so that thread-safety is essential.
14151   Unfortunately, the ex_data design is not at all suited
14152   for multi-threaded use, so it probably should be abolished.
14153
14154   *Bodo Moeller*
14155
14156 * Added Broadcom "ubsec" ENGINE to OpenSSL.
14157
14158   *Broadcom, tweaked and integrated by Geoff Thorpe*
14159
14160 * Move common extension printing code to new function
14161   X509V3_print_extensions(). Reorganise OCSP print routines and
14162   implement some needed OCSP ASN1 functions. Add OCSP extensions.
14163
14164   *Steve Henson*
14165
14166 * New function X509_signature_print() to remove duplication in some
14167   print routines.
14168
14169   *Steve Henson*
14170
14171 * Add a special meaning when SET OF and SEQUENCE OF flags are both
14172   set (this was treated exactly the same as SET OF previously). This
14173   is used to reorder the STACK representing the structure to match the
14174   encoding. This will be used to get round a problem where a PKCS7
14175   structure which was signed could not be verified because the STACK
14176   order did not reflect the encoded order.
14177
14178   *Steve Henson*
14179
14180 * Reimplement the OCSP ASN1 module using the new code.
14181
14182   *Steve Henson*
14183
14184 * Update the X509V3 code to permit the use of an ASN1_ITEM structure
14185   for its ASN1 operations. The old style function pointers still exist
14186   for now but they will eventually go away.
14187
14188   *Steve Henson*
14189
14190 * Merge in replacement ASN1 code from the ASN1 branch. This almost
14191   completely replaces the old ASN1 functionality with a table driven
14192   encoder and decoder which interprets an ASN1_ITEM structure describing
14193   the ASN1 module. Compatibility with the existing ASN1 API (i2d,d2i) is
14194   largely maintained. Almost all of the old asn1_mac.h macro based ASN1
14195   has also been converted to the new form.
14196
14197   *Steve Henson*
14198
14199 * Change BN_mod_exp_recp so that negative moduli are tolerated
14200   (the sign is ignored).  Similarly, ignore the sign in BN_MONT_CTX_set
14201   so that BN_mod_exp_mont and BN_mod_exp_mont_word work
14202   for negative moduli.
14203
14204   *Bodo Moeller*
14205
14206 * Fix BN_uadd and BN_usub: Always return non-negative results instead
14207   of not touching the result's sign bit.
14208
14209   *Bodo Moeller*
14210
14211 * BN_div bugfix: If the result is 0, the sign (res->neg) must not be
14212   set.
14213
14214   *Bodo Moeller*
14215
14216 * Changed the LHASH code to use prototypes for callbacks, and created
14217   macros to declare and implement thin (optionally static) functions
14218   that provide type-safety and avoid function pointer casting for the
14219   type-specific callbacks.
14220
14221   *Geoff Thorpe*
14222
14223 * Added Kerberos Cipher Suites to be used with TLS, as written in
14224   RFC 2712.
14225   *Veers Staats <staatsvr@asc.hpc.mil>,
14226   Jeffrey Altman <jaltman@columbia.edu>, via Richard Levitte*
14227
14228 * Reformat the FAQ so the different questions and answers can be divided
14229   in sections depending on the subject.
14230
14231   *Richard Levitte*
14232
14233 * Have the zlib compression code load ZLIB.DLL dynamically under
14234   Windows.
14235
14236   *Richard Levitte*
14237
14238 * New function BN_mod_sqrt for computing square roots modulo a prime
14239   (using the probabilistic Tonelli-Shanks algorithm unless
14240   p == 3 (mod 4)  or  p == 5 (mod 8),  which are cases that can
14241   be handled deterministically).
14242
14243   *Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller*
14244
14245 * Make BN_mod_inverse faster by explicitly handling small quotients
14246   in the Euclid loop. (Speed gain about 20% for small moduli [256 or
14247   512 bits], about 30% for larger ones [1024 or 2048 bits].)
14248
14249   *Bodo Moeller*
14250
14251 * New function BN_kronecker.
14252
14253   *Bodo Moeller*
14254
14255 * Fix BN_gcd so that it works on negative inputs; the result is
14256   positive unless both parameters are zero.
14257   Previously something reasonably close to an infinite loop was
14258   possible because numbers could be growing instead of shrinking
14259   in the implementation of Euclid's algorithm.
14260
14261   *Bodo Moeller*
14262
14263 * Fix BN_is_word() and BN_is_one() macros to take into account the
14264   sign of the number in question.
14265
14266   Fix BN_is_word(a,w) to work correctly for w == 0.
14267
14268   The old BN_is_word(a,w) macro is now called BN_abs_is_word(a,w)
14269   because its test if the absolute value of 'a' equals 'w'.
14270   Note that BN_abs_is_word does *not* handle w == 0 reliably;
14271   it exists mostly for use in the implementations of BN_is_zero(),
14272   BN_is_one(), and BN_is_word().
14273
14274   *Bodo Moeller*
14275
14276 * New function BN_swap.
14277
14278   *Bodo Moeller*
14279
14280 * Use BN_nnmod instead of BN_mod in crypto/bn/bn_exp.c so that
14281   the exponentiation functions are more likely to produce reasonable
14282   results on negative inputs.
14283
14284   *Bodo Moeller*
14285
14286 * Change BN_mod_mul so that the result is always non-negative.
14287   Previously, it could be negative if one of the factors was negative;
14288   I don't think anyone really wanted that behaviour.
14289
14290   *Bodo Moeller*
14291
14292 * Move `BN_mod_...` functions into new file `crypto/bn/bn_mod.c`
14293   (except for exponentiation, which stays in `crypto/bn/bn_exp.c`,
14294   and `BN_mod_mul_reciprocal`, which stays in `crypto/bn/bn_recp.c`)
14295   and add new functions:
14296
14297           BN_nnmod
14298           BN_mod_sqr
14299           BN_mod_add
14300           BN_mod_add_quick
14301           BN_mod_sub
14302           BN_mod_sub_quick
14303           BN_mod_lshift1
14304           BN_mod_lshift1_quick
14305           BN_mod_lshift
14306           BN_mod_lshift_quick
14307
14308   These functions always generate non-negative results.
14309
14310   `BN_nnmod` otherwise is `like BN_mod` (if `BN_mod` computes a remainder `r`
14311   such that `|m| < r < 0`, `BN_nnmod` will output `rem + |m|` instead).
14312
14313   `BN_mod_XXX_quick(r, a, [b,] m)` generates the same result as
14314   `BN_mod_XXX(r, a, [b,] m, ctx)`, but requires that `a` [and  `b`]
14315   be reduced modulo `m`.
14316
14317   *Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller*
14318
14319<!--
14320   The following entry accidentally appeared in the CHANGES file
14321   distributed with OpenSSL 0.9.7.  The modifications described in
14322   it do *not* apply to OpenSSL 0.9.7.
14323
14324 * Remove a few calls to bn_wexpand() in BN_sqr() (the one in there
14325   was actually never needed) and in BN_mul().  The removal in BN_mul()
14326   required a small change in bn_mul_part_recursive() and the addition
14327   of the functions bn_cmp_part_words(), bn_sub_part_words() and
14328   bn_add_part_words(), which do the same thing as bn_cmp_words(),
14329   bn_sub_words() and bn_add_words() except they take arrays with
14330   differing sizes.
14331
14332   *Richard Levitte*
14333-->
14334
14335 * In 'openssl passwd', verify passwords read from the terminal
14336   unless the '-salt' option is used (which usually means that
14337   verification would just waste user's time since the resulting
14338   hash is going to be compared with some given password hash)
14339   or the new '-noverify' option is used.
14340
14341   This is an incompatible change, but it does not affect
14342   non-interactive use of 'openssl passwd' (passwords on the command
14343   line, '-stdin' option, '-in ...' option) and thus should not
14344   cause any problems.
14345
14346   *Bodo Moeller*
14347
14348 * Remove all references to RSAref, since there's no more need for it.
14349
14350   *Richard Levitte*
14351
14352 * Make DSO load along a path given through an environment variable
14353   (SHLIB_PATH) with shl_load().
14354
14355   *Richard Levitte*
14356
14357 * Constify the ENGINE code as a result of BIGNUM constification.
14358   Also constify the RSA code and most things related to it.  In a
14359   few places, most notable in the depth of the ASN.1 code, ugly
14360   casts back to non-const were required (to be solved at a later
14361   time)
14362
14363   *Richard Levitte*
14364
14365 * Make it so the openssl application has all engines loaded by default.
14366
14367   *Richard Levitte*
14368
14369 * Constify the BIGNUM routines a little more.
14370
14371   *Richard Levitte*
14372
14373 * Add the following functions:
14374
14375           ENGINE_load_cswift()
14376           ENGINE_load_chil()
14377           ENGINE_load_atalla()
14378           ENGINE_load_nuron()
14379           ENGINE_load_builtin_engines()
14380
14381   That way, an application can itself choose if external engines that
14382   are built-in in OpenSSL shall ever be used or not.  The benefit is
14383   that applications won't have to be linked with libdl or other dso
14384   libraries unless it's really needed.
14385
14386   Changed 'openssl engine' to load all engines on demand.
14387   Changed the engine header files to avoid the duplication of some
14388   declarations (they differed!).
14389
14390   *Richard Levitte*
14391
14392 * 'openssl engine' can now list capabilities.
14393
14394   *Richard Levitte*
14395
14396 * Better error reporting in 'openssl engine'.
14397
14398   *Richard Levitte*
14399
14400 * Never call load_dh_param(NULL) in s_server.
14401
14402   *Bodo Moeller*
14403
14404 * Add engine application.  It can currently list engines by name and
14405   identity, and test if they are actually available.
14406
14407   *Richard Levitte*
14408
14409 * Improve RPM specification file by forcing symbolic linking and making
14410   sure the installed documentation is also owned by root.root.
14411
14412   *Damien Miller <djm@mindrot.org>*
14413
14414 * Give the OpenSSL applications more possibilities to make use of
14415   keys (public as well as private) handled by engines.
14416
14417   *Richard Levitte*
14418
14419 * Add OCSP code that comes from CertCo.
14420
14421   *Richard Levitte*
14422
14423 * Add VMS support for the Rijndael code.
14424
14425   *Richard Levitte*
14426
14427 * Added untested support for Nuron crypto accelerator.
14428
14429   *Ben Laurie*
14430
14431 * Add support for external cryptographic devices.  This code was
14432   previously distributed separately as the "engine" branch.
14433
14434   *Geoff Thorpe, Richard Levitte*
14435
14436 * Rework the filename-translation in the DSO code. It is now possible to
14437   have far greater control over how a "name" is turned into a filename
14438   depending on the operating environment and any oddities about the
14439   different shared library filenames on each system.
14440
14441   *Geoff Thorpe*
14442
14443 * Support threads on FreeBSD-elf in Configure.
14444
14445   *Richard Levitte*
14446
14447 * Fix for SHA1 assembly problem with MASM: it produces
14448   warnings about corrupt line number information when assembling
14449   with debugging information. This is caused by the overlapping
14450   of two sections.
14451
14452   *Bernd Matthes <mainbug@celocom.de>, Steve Henson*
14453
14454 * NCONF changes.
14455   NCONF_get_number() has no error checking at all.  As a replacement,
14456   NCONF_get_number_e() is defined (`_e` for "error checking") and is
14457   promoted strongly.  The old NCONF_get_number is kept around for
14458   binary backward compatibility.
14459   Make it possible for methods to load from something other than a BIO,
14460   by providing a function pointer that is given a name instead of a BIO.
14461   For example, this could be used to load configuration data from an
14462   LDAP server.
14463
14464   *Richard Levitte*
14465
14466 * Fix for non blocking accept BIOs. Added new I/O special reason
14467   BIO_RR_ACCEPT to cover this case. Previously use of accept BIOs
14468   with non blocking I/O was not possible because no retry code was
14469   implemented. Also added new SSL code SSL_WANT_ACCEPT to cover
14470   this case.
14471
14472   *Steve Henson*
14473
14474 * Added the beginnings of Rijndael support.
14475
14476   *Ben Laurie*
14477
14478 * Fix for bug in DirectoryString mask setting. Add support for
14479   X509_NAME_print_ex() in 'req' and X509_print_ex() function
14480   to allow certificate printing to more controllable, additional
14481   'certopt' option to 'x509' to allow new printing options to be
14482   set.
14483
14484   *Steve Henson*
14485
14486 * Clean old EAY MD5 hack from e_os.h.
14487
14488   *Richard Levitte*
14489
14490### Changes between 0.9.6l and 0.9.6m  [17 Mar 2004]
14491
14492 * Fix null-pointer assignment in do_change_cipher_spec() revealed
14493   by using the Codenomicon TLS Test Tool ([CVE-2004-0079])
14494
14495   *Joe Orton, Steve Henson*
14496
14497### Changes between 0.9.6k and 0.9.6l  [04 Nov 2003]
14498
14499 * Fix additional bug revealed by the NISCC test suite:
14500
14501   Stop bug triggering large recursion when presented with
14502   certain ASN.1 tags ([CVE-2003-0851])
14503
14504   *Steve Henson*
14505
14506### Changes between 0.9.6j and 0.9.6k  [30 Sep 2003]
14507
14508 * Fix various bugs revealed by running the NISCC test suite:
14509
14510   Stop out of bounds reads in the ASN1 code when presented with
14511   invalid tags (CVE-2003-0543 and CVE-2003-0544).
14512
14513   If verify callback ignores invalid public key errors don't try to check
14514   certificate signature with the NULL public key.
14515
14516   *Steve Henson*
14517
14518 * In ssl3_accept() (ssl/s3_srvr.c) only accept a client certificate
14519   if the server requested one: as stated in TLS 1.0 and SSL 3.0
14520   specifications.
14521
14522   *Steve Henson*
14523
14524 * In ssl3_get_client_hello() (ssl/s3_srvr.c), tolerate additional
14525   extra data after the compression methods not only for TLS 1.0
14526   but also for SSL 3.0 (as required by the specification).
14527
14528   *Bodo Moeller; problem pointed out by Matthias Loepfe*
14529
14530 * Change X509_certificate_type() to mark the key as exported/exportable
14531   when it's 512 *bits* long, not 512 bytes.
14532
14533   *Richard Levitte*
14534
14535### Changes between 0.9.6i and 0.9.6j  [10 Apr 2003]
14536
14537 * Countermeasure against the Klima-Pokorny-Rosa extension of
14538   Bleichbacher's attack on PKCS #1 v1.5 padding: treat
14539   a protocol version number mismatch like a decryption error
14540   in ssl3_get_client_key_exchange (ssl/s3_srvr.c).
14541
14542   *Bodo Moeller*
14543
14544 * Turn on RSA blinding by default in the default implementation
14545   to avoid a timing attack. Applications that don't want it can call
14546   RSA_blinding_off() or use the new flag RSA_FLAG_NO_BLINDING.
14547   They would be ill-advised to do so in most cases.
14548
14549   *Ben Laurie, Steve Henson, Geoff Thorpe, Bodo Moeller*
14550
14551 * Change RSA blinding code so that it works when the PRNG is not
14552   seeded (in this case, the secret RSA exponent is abused as
14553   an unpredictable seed -- if it is not unpredictable, there
14554   is no point in blinding anyway).  Make RSA blinding thread-safe
14555   by remembering the creator's thread ID in rsa->blinding and
14556   having all other threads use local one-time blinding factors
14557   (this requires more computation than sharing rsa->blinding, but
14558   avoids excessive locking; and if an RSA object is not shared
14559   between threads, blinding will still be very fast).
14560
14561   *Bodo Moeller*
14562
14563### Changes between 0.9.6h and 0.9.6i  [19 Feb 2003]
14564
14565 * In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked
14566   via timing by performing a MAC computation even if incorrect
14567   block cipher padding has been found.  This is a countermeasure
14568   against active attacks where the attacker has to distinguish
14569   between bad padding and a MAC verification error. ([CVE-2003-0078])
14570
14571   *Bodo Moeller; problem pointed out by Brice Canvel (EPFL),
14572   Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and
14573   Martin Vuagnoux (EPFL, Ilion)*
14574
14575### Changes between 0.9.6g and 0.9.6h  [5 Dec 2002]
14576
14577 * New function OPENSSL_cleanse(), which is used to cleanse a section of
14578   memory from its contents.  This is done with a counter that will
14579   place alternating values in each byte.  This can be used to solve
14580   two issues: 1) the removal of calls to memset() by highly optimizing
14581   compilers, and 2) cleansing with other values than 0, since those can
14582   be read through on certain media, for example a swap space on disk.
14583
14584   *Geoff Thorpe*
14585
14586 * Bugfix: client side session caching did not work with external caching,
14587   because the session->cipher setting was not restored when reloading
14588   from the external cache. This problem was masked, when
14589   SSL_OP_NETSCAPE_REUSE_CIPHER_CHANGE_BUG (part of SSL_OP_ALL) was set.
14590   (Found by Steve Haslam <steve@araqnid.ddts.net>.)
14591
14592   *Lutz Jaenicke*
14593
14594 * Fix client_certificate (ssl/s2_clnt.c): The permissible total
14595   length of the REQUEST-CERTIFICATE message is 18 .. 34, not 17 .. 33.
14596
14597   *Zeev Lieber <zeev-l@yahoo.com>*
14598
14599 * Undo an undocumented change introduced in 0.9.6e which caused
14600   repeated calls to OpenSSL_add_all_ciphers() and
14601   OpenSSL_add_all_digests() to be ignored, even after calling
14602   EVP_cleanup().
14603
14604   *Richard Levitte*
14605
14606 * Change the default configuration reader to deal with last line not
14607   being properly terminated.
14608
14609   *Richard Levitte*
14610
14611 * Change X509_NAME_cmp() so it applies the special rules on handling
14612   DN values that are of type PrintableString, as well as RDNs of type
14613   emailAddress where the value has the type ia5String.
14614
14615   *stefank@valicert.com via Richard Levitte*
14616
14617 * Add a SSL_SESS_CACHE_NO_INTERNAL_STORE flag to take over half
14618   the job SSL_SESS_CACHE_NO_INTERNAL_LOOKUP was inconsistently
14619   doing, define a new flag (SSL_SESS_CACHE_NO_INTERNAL) to be
14620   the bitwise-OR of the two for use by the majority of applications
14621   wanting this behaviour, and update the docs. The documented
14622   behaviour and actual behaviour were inconsistent and had been
14623   changing anyway, so this is more a bug-fix than a behavioural
14624   change.
14625
14626   *Geoff Thorpe, diagnosed by Nadav Har'El*
14627
14628 * Don't impose a 16-byte length minimum on session IDs in ssl/s3_clnt.c
14629   (the SSL 3.0 and TLS 1.0 specifications allow any length up to 32 bytes).
14630
14631   *Bodo Moeller*
14632
14633 * Fix initialization code race conditions in
14634           SSLv23_method(),  SSLv23_client_method(),   SSLv23_server_method(),
14635           SSLv2_method(),   SSLv2_client_method(),    SSLv2_server_method(),
14636           SSLv3_method(),   SSLv3_client_method(),    SSLv3_server_method(),
14637           TLSv1_method(),   TLSv1_client_method(),    TLSv1_server_method(),
14638           ssl2_get_cipher_by_char(),
14639           ssl3_get_cipher_by_char().
14640
14641   *Patrick McCormick <patrick@tellme.com>, Bodo Moeller*
14642
14643 * Reorder cleanup sequence in SSL_CTX_free(): only remove the ex_data after
14644   the cached sessions are flushed, as the remove_cb() might use ex_data
14645   contents. Bug found by Sam Varshavchik <mrsam@courier-mta.com>
14646   (see [openssl.org #212]).
14647
14648   *Geoff Thorpe, Lutz Jaenicke*
14649
14650 * Fix typo in OBJ_txt2obj which incorrectly passed the content
14651   length, instead of the encoding length to d2i_ASN1_OBJECT.
14652
14653   *Steve Henson*
14654
14655### Changes between 0.9.6f and 0.9.6g  [9 Aug 2002]
14656
14657 * [In 0.9.6g-engine release:]
14658   Fix crypto/engine/vendor_defns/cswift.h for WIN32 (use `_stdcall`).
14659
14660   *Lynn Gazis <lgazis@rainbow.com>*
14661
14662### Changes between 0.9.6e and 0.9.6f  [8 Aug 2002]
14663
14664 * Fix ASN1 checks. Check for overflow by comparing with LONG_MAX
14665   and get fix the header length calculation.
14666   *Florian Weimer <Weimer@CERT.Uni-Stuttgart.DE>,
14667   Alon Kantor <alonk@checkpoint.com> (and others), Steve Henson*
14668
14669 * Use proper error handling instead of 'assertions' in buffer
14670   overflow checks added in 0.9.6e.  This prevents DoS (the
14671   assertions could call abort()).
14672
14673   *Arne Ansper <arne@ats.cyber.ee>, Bodo Moeller*
14674
14675### Changes between 0.9.6d and 0.9.6e  [30 Jul 2002]
14676
14677 * Add various sanity checks to asn1_get_length() to reject
14678   the ASN1 length bytes if they exceed sizeof(long), will appear
14679   negative or the content length exceeds the length of the
14680   supplied buffer.
14681
14682   *Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>*
14683
14684 * Fix cipher selection routines: ciphers without encryption had no flags
14685   for the cipher strength set and where therefore not handled correctly
14686   by the selection routines (PR #130).
14687
14688   *Lutz Jaenicke*
14689
14690 * Fix EVP_dsa_sha macro.
14691
14692   *Nils Larsch*
14693
14694 * New option
14695        SSL_OP_DONT_INSERT_EMPTY_FRAGMENTS
14696   for disabling the SSL 3.0/TLS 1.0 CBC vulnerability countermeasure
14697   that was added in OpenSSL 0.9.6d.
14698
14699   As the countermeasure turned out to be incompatible with some
14700   broken SSL implementations, the new option is part of SSL_OP_ALL.
14701   SSL_OP_ALL is usually employed when compatibility with weird SSL
14702   implementations is desired (e.g. '-bugs' option to 's_client' and
14703   's_server'), so the new option is automatically set in many
14704   applications.
14705
14706   *Bodo Moeller*
14707
14708 * Changes in security patch:
14709
14710   Changes marked "(CHATS)" were sponsored by the Defense Advanced
14711   Research Projects Agency (DARPA) and Air Force Research Laboratory,
14712   Air Force Materiel Command, USAF, under agreement number
14713   F30602-01-2-0537.
14714
14715 * Add various sanity checks to asn1_get_length() to reject
14716   the ASN1 length bytes if they exceed sizeof(long), will appear
14717   negative or the content length exceeds the length of the
14718   supplied buffer. ([CVE-2002-0659])
14719
14720   *Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>*
14721
14722 * Assertions for various potential buffer overflows, not known to
14723   happen in practice.
14724
14725   *Ben Laurie (CHATS)*
14726
14727 * Various temporary buffers to hold ASCII versions of integers were
14728   too small for 64 bit platforms. ([CVE-2002-0655])
14729   *Matthew Byng-Maddick <mbm@aldigital.co.uk> and Ben Laurie (CHATS)>*
14730
14731 * Remote buffer overflow in SSL3 protocol - an attacker could
14732   supply an oversized session ID to a client. ([CVE-2002-0656])
14733
14734   *Ben Laurie (CHATS)*
14735
14736 * Remote buffer overflow in SSL2 protocol - an attacker could
14737   supply an oversized client master key. ([CVE-2002-0656])
14738
14739   *Ben Laurie (CHATS)*
14740
14741### Changes between 0.9.6c and 0.9.6d  [9 May 2002]
14742
14743 * Fix crypto/asn1/a_sign.c so that 'parameters' is omitted (not
14744   encoded as NULL) with id-dsa-with-sha1.
14745
14746   *Nils Larsch <nla@trustcenter.de>; problem pointed out by Bodo Moeller*
14747
14748 * Check various `X509_...()` return values in `apps/req.c`.
14749
14750   *Nils Larsch <nla@trustcenter.de>*
14751
14752 * Fix BASE64 decode (EVP_DecodeUpdate) for data with CR/LF ended lines:
14753   an end-of-file condition would erroneously be flagged, when the CRLF
14754   was just at the end of a processed block. The bug was discovered when
14755   processing data through a buffering memory BIO handing the data to a
14756   BASE64-decoding BIO. Bug fund and patch submitted by Pavel Tsekov
14757   <ptsekov@syntrex.com> and Nedelcho Stanev.
14758
14759   *Lutz Jaenicke*
14760
14761 * Implement a countermeasure against a vulnerability recently found
14762   in CBC ciphersuites in SSL 3.0/TLS 1.0: Send an empty fragment
14763   before application data chunks to avoid the use of known IVs
14764   with data potentially chosen by the attacker.
14765
14766   *Bodo Moeller*
14767
14768 * Fix length checks in ssl3_get_client_hello().
14769
14770   *Bodo Moeller*
14771
14772 * TLS/SSL library bugfix: use s->s3->in_read_app_data differently
14773   to prevent ssl3_read_internal() from incorrectly assuming that
14774   ssl3_read_bytes() found application data while handshake
14775   processing was enabled when in fact s->s3->in_read_app_data was
14776   merely automatically cleared during the initial handshake.
14777
14778   *Bodo Moeller; problem pointed out by Arne Ansper <arne@ats.cyber.ee>*
14779
14780 * Fix object definitions for Private and Enterprise: they were not
14781   recognized in their shortname (=lowercase) representation. Extend
14782   obj_dat.pl to issue an error when using undefined keywords instead
14783   of silently ignoring the problem (Svenning Sorensen
14784   <sss@sss.dnsalias.net>).
14785
14786   *Lutz Jaenicke*
14787
14788 * Fix DH_generate_parameters() so that it works for 'non-standard'
14789   generators, i.e. generators other than 2 and 5.  (Previously, the
14790   code did not properly initialise the 'add' and 'rem' values to
14791   BN_generate_prime().)
14792
14793   In the new general case, we do not insist that 'generator' is
14794   actually a primitive root: This requirement is rather pointless;
14795   a generator of the order-q subgroup is just as good, if not
14796   better.
14797
14798   *Bodo Moeller*
14799
14800 * Map new X509 verification errors to alerts. Discovered and submitted by
14801   Tom Wu <tom@arcot.com>.
14802
14803   *Lutz Jaenicke*
14804
14805 * Fix ssl3_pending() (ssl/s3_lib.c) to prevent SSL_pending() from
14806   returning non-zero before the data has been completely received
14807   when using non-blocking I/O.
14808
14809   *Bodo Moeller; problem pointed out by John Hughes*
14810
14811 * Some of the ciphers missed the strength entry (SSL_LOW etc).
14812
14813   *Ben Laurie, Lutz Jaenicke*
14814
14815 * Fix bug in SSL_clear(): bad sessions were not removed (found by
14816   Yoram Zahavi <YoramZ@gilian.com>).
14817
14818   *Lutz Jaenicke*
14819
14820 * Add information about CygWin 1.3 and on, and preserve proper
14821   configuration for the versions before that.
14822
14823   *Corinna Vinschen <vinschen@redhat.com> and Richard Levitte*
14824
14825 * Make removal from session cache (SSL_CTX_remove_session()) more robust:
14826   check whether we deal with a copy of a session and do not delete from
14827   the cache in this case. Problem reported by "Izhar Shoshani Levi"
14828   <izhar@checkpoint.com>.
14829
14830   *Lutz Jaenicke*
14831
14832 * Do not store session data into the internal session cache, if it
14833   is never intended to be looked up (SSL_SESS_CACHE_NO_INTERNAL_LOOKUP
14834   flag is set). Proposed by Aslam <aslam@funk.com>.
14835
14836   *Lutz Jaenicke*
14837
14838 * Have ASN1_BIT_STRING_set_bit() really clear a bit when the requested
14839   value is 0.
14840
14841   *Richard Levitte*
14842
14843 * [In 0.9.6d-engine release:]
14844   Fix a crashbug and a logic bug in hwcrhk_load_pubkey().
14845
14846   *Toomas Kiisk <vix@cyber.ee> via Richard Levitte*
14847
14848 * Add the configuration target linux-s390x.
14849
14850   *Neale Ferguson <Neale.Ferguson@SoftwareAG-USA.com> via Richard Levitte*
14851
14852 * The earlier bugfix for the SSL3_ST_SW_HELLO_REQ_C case of
14853   ssl3_accept (ssl/s3_srvr.c) incorrectly used a local flag
14854   variable as an indication that a ClientHello message has been
14855   received.  As the flag value will be lost between multiple
14856   invocations of ssl3_accept when using non-blocking I/O, the
14857   function may not be aware that a handshake has actually taken
14858   place, thus preventing a new session from being added to the
14859   session cache.
14860
14861   To avoid this problem, we now set s->new_session to 2 instead of
14862   using a local variable.
14863
14864   *Lutz Jaenicke, Bodo Moeller*
14865
14866 * Bugfix: Return -1 from ssl3_get_server_done (ssl3/s3_clnt.c)
14867   if the SSL_R_LENGTH_MISMATCH error is detected.
14868
14869   *Geoff Thorpe, Bodo Moeller*
14870
14871 * New 'shared_ldflag' column in Configure platform table.
14872
14873   *Richard Levitte*
14874
14875 * Fix EVP_CIPHER_mode macro.
14876
14877   *"Dan S. Camper" <dan@bti.net>*
14878
14879 * Fix ssl3_read_bytes (ssl/s3_pkt.c): To ignore messages of unknown
14880   type, we must throw them away by setting rr->length to 0.
14881
14882   *D P Chang <dpc@qualys.com>*
14883
14884### Changes between 0.9.6b and 0.9.6c  [21 dec 2001]
14885
14886 * Fix BN_rand_range bug pointed out by Dominikus Scherkl
14887   <Dominikus.Scherkl@biodata.com>.  (The previous implementation
14888   worked incorrectly for those cases where range = `10..._2`  and
14889   `3*range`  is two bits longer than  range.)
14890
14891   *Bodo Moeller*
14892
14893 * Only add signing time to PKCS7 structures if it is not already
14894   present.
14895
14896   *Steve Henson*
14897
14898 * Fix crypto/objects/objects.h: "ld-ce" should be "id-ce",
14899   OBJ_ld_ce should be OBJ_id_ce.
14900   Also some ip-pda OIDs in crypto/objects/objects.txt were
14901   incorrect (cf. RFC 3039).
14902
14903   *Matt Cooper, Frederic Giudicelli, Bodo Moeller*
14904
14905 * Release CRYPTO_LOCK_DYNLOCK when CRYPTO_destroy_dynlockid()
14906   returns early because it has nothing to do.
14907
14908   *Andy Schneider <andy.schneider@bjss.co.uk>*
14909
14910 * [In 0.9.6c-engine release:]
14911   Fix mutex callback return values in crypto/engine/hw_ncipher.c.
14912
14913   *Andy Schneider <andy.schneider@bjss.co.uk>*
14914
14915 * [In 0.9.6c-engine release:]
14916   Add support for Cryptographic Appliance's keyserver technology.
14917   (Use engine 'keyclient')
14918
14919   *Cryptographic Appliances and Geoff Thorpe*
14920
14921 * Add a configuration entry for OS/390 Unix.  The C compiler 'c89'
14922   is called via tools/c89.sh because arguments have to be
14923   rearranged (all '-L' options must appear before the first object
14924   modules).
14925
14926   *Richard Shapiro <rshapiro@abinitio.com>*
14927
14928 * [In 0.9.6c-engine release:]
14929   Add support for Broadcom crypto accelerator cards, backported
14930   from 0.9.7.
14931
14932   *Broadcom, Nalin Dahyabhai <nalin@redhat.com>, Mark Cox*
14933
14934 * [In 0.9.6c-engine release:]
14935   Add support for SureWare crypto accelerator cards from
14936   Baltimore Technologies.  (Use engine 'sureware')
14937
14938   *Baltimore Technologies and Mark Cox*
14939
14940 * [In 0.9.6c-engine release:]
14941   Add support for crypto accelerator cards from Accelerated
14942   Encryption Processing, www.aep.ie.  (Use engine 'aep')
14943
14944   *AEP Inc. and Mark Cox*
14945
14946 * Add a configuration entry for gcc on UnixWare.
14947
14948   *Gary Benson <gbenson@redhat.com>*
14949
14950 * Change ssl/s2_clnt.c and ssl/s2_srvr.c so that received handshake
14951   messages are stored in a single piece (fixed-length part and
14952   variable-length part combined) and fix various bugs found on the way.
14953
14954   *Bodo Moeller*
14955
14956 * Disable caching in BIO_gethostbyname(), directly use gethostbyname()
14957   instead.  BIO_gethostbyname() does not know what timeouts are
14958   appropriate, so entries would stay in cache even when they have
14959   become invalid.
14960   *Bodo Moeller; problem pointed out by Rich Salz <rsalz@zolera.com>*
14961
14962 * Change ssl23_get_client_hello (ssl/s23_srvr.c) behaviour when
14963   faced with a pathologically small ClientHello fragment that does
14964   not contain client_version: Instead of aborting with an error,
14965   simply choose the highest available protocol version (i.e.,
14966   TLS 1.0 unless it is disabled).  In practice, ClientHello
14967   messages are never sent like this, but this change gives us
14968   strictly correct behaviour at least for TLS.
14969
14970   *Bodo Moeller*
14971
14972 * Fix SSL handshake functions and SSL_clear() such that SSL_clear()
14973   never resets s->method to s->ctx->method when called from within
14974   one of the SSL handshake functions.
14975
14976   *Bodo Moeller; problem pointed out by Niko Baric*
14977
14978 * In ssl3_get_client_hello (ssl/s3_srvr.c), generate a fatal alert
14979   (sent using the client's version number) if client_version is
14980   smaller than the protocol version in use.  Also change
14981   ssl23_get_client_hello (ssl/s23_srvr.c) to select TLS 1.0 if
14982   the client demanded SSL 3.0 but only TLS 1.0 is enabled; then
14983   the client will at least see that alert.
14984
14985   *Bodo Moeller*
14986
14987 * Fix ssl3_get_message (ssl/s3_both.c) to handle message fragmentation
14988   correctly.
14989
14990   *Bodo Moeller*
14991
14992 * Avoid infinite loop in ssl3_get_message (ssl/s3_both.c) if a
14993   client receives HelloRequest while in a handshake.
14994
14995   *Bodo Moeller; bug noticed by Andy Schneider <andy.schneider@bjss.co.uk>*
14996
14997 * Bugfix in ssl3_accept (ssl/s3_srvr.c): Case SSL3_ST_SW_HELLO_REQ_C
14998   should end in 'break', not 'goto end' which circumvents various
14999   cleanups done in state SSL_ST_OK.   But session related stuff
15000   must be disabled for SSL_ST_OK in the case that we just sent a
15001   HelloRequest.
15002
15003   Also avoid some overhead by not calling ssl_init_wbio_buffer()
15004   before just sending a HelloRequest.
15005
15006   *Bodo Moeller, Eric Rescorla <ekr@rtfm.com>*
15007
15008 * Fix ssl/s3_enc.c, ssl/t1_enc.c and ssl/s3_pkt.c so that we don't
15009   reveal whether illegal block cipher padding was found or a MAC
15010   verification error occurred.  (Neither SSLerr() codes nor alerts
15011   are directly visible to potential attackers, but the information
15012   may leak via logfiles.)
15013
15014   Similar changes are not required for the SSL 2.0 implementation
15015   because the number of padding bytes is sent in clear for SSL 2.0,
15016   and the extra bytes are just ignored.  However ssl/s2_pkt.c
15017   failed to verify that the purported number of padding bytes is in
15018   the legal range.
15019
15020   *Bodo Moeller*
15021
15022 * Add OpenUNIX-8 support including shared libraries
15023   (Boyd Lynn Gerber <gerberb@zenez.com>).
15024
15025   *Lutz Jaenicke*
15026
15027 * Improve RSA_padding_check_PKCS1_OAEP() check again to avoid
15028   'wristwatch attack' using huge encoding parameters (cf.
15029   James H. Manger's CRYPTO 2001 paper).  Note that the
15030   RSA_PKCS1_OAEP_PADDING case of RSA_private_decrypt() does not use
15031   encoding parameters and hence was not vulnerable.
15032
15033   *Bodo Moeller*
15034
15035 * BN_sqr() bug fix.
15036
15037   *Ulf Möller, reported by Jim Ellis <jim.ellis@cavium.com>*
15038
15039 * Rabin-Miller test analyses assume uniformly distributed witnesses,
15040   so use BN_pseudo_rand_range() instead of using BN_pseudo_rand()
15041   followed by modular reduction.
15042
15043   *Bodo Moeller; pointed out by Adam Young <AYoung1@NCSUS.JNJ.COM>*
15044
15045 * Add BN_pseudo_rand_range() with obvious functionality: BN_rand_range()
15046   equivalent based on BN_pseudo_rand() instead of BN_rand().
15047
15048   *Bodo Moeller*
15049
15050 * s3_srvr.c: allow sending of large client certificate lists (> 16 kB).
15051   This function was broken, as the check for a new client hello message
15052   to handle SGC did not allow these large messages.
15053   (Tracked down by "Douglas E. Engert" <deengert@anl.gov>.)
15054
15055   *Lutz Jaenicke*
15056
15057 * Add alert descriptions for TLSv1 to `SSL_alert_desc_string[_long]()`.
15058
15059   *Lutz Jaenicke*
15060
15061 * Fix buggy behaviour of BIO_get_num_renegotiates() and BIO_ctrl()
15062   for BIO_C_GET_WRITE_BUF_SIZE ("Stephen Hinton" <shinton@netopia.com>).
15063
15064   *Lutz Jaenicke*
15065
15066 * Rework the configuration and shared library support for Tru64 Unix.
15067   The configuration part makes use of modern compiler features and
15068   still retains old compiler behavior for those that run older versions
15069   of the OS.  The shared library support part includes a variant that
15070   uses the RPATH feature, and is available through the special
15071   configuration target "alpha-cc-rpath", which will never be selected
15072   automatically.
15073
15074   *Tim Mooney <mooney@dogbert.cc.ndsu.NoDak.edu> via Richard Levitte*
15075
15076 * In ssl3_get_key_exchange (ssl/s3_clnt.c), call ssl3_get_message()
15077   with the same message size as in ssl3_get_certificate_request().
15078   Otherwise, if no ServerKeyExchange message occurs, CertificateRequest
15079   messages might inadvertently be reject as too long.
15080
15081   *Petr Lampa <lampa@fee.vutbr.cz>*
15082
15083 * Enhanced support for IA-64 Unix platforms (well, Linux and HP-UX).
15084
15085   *Andy Polyakov*
15086
15087 * Modified SSL library such that the verify_callback that has been set
15088   specifically for an SSL object with SSL_set_verify() is actually being
15089   used. Before the change, a verify_callback set with this function was
15090   ignored and the verify_callback() set in the SSL_CTX at the time of
15091   the call was used. New function X509_STORE_CTX_set_verify_cb() introduced
15092   to allow the necessary settings.
15093
15094   *Lutz Jaenicke*
15095
15096 * Initialize static variable in crypto/dsa/dsa_lib.c and crypto/dh/dh_lib.c
15097   explicitly to NULL, as at least on Solaris 8 this seems not always to be
15098   done automatically (in contradiction to the requirements of the C
15099   standard). This made problems when used from OpenSSH.
15100
15101   *Lutz Jaenicke*
15102
15103 * In OpenSSL 0.9.6a and 0.9.6b, crypto/dh/dh_key.c ignored
15104   dh->length and always used
15105
15106           BN_rand_range(priv_key, dh->p).
15107
15108   BN_rand_range() is not necessary for Diffie-Hellman, and this
15109   specific range makes Diffie-Hellman unnecessarily inefficient if
15110   dh->length (recommended exponent length) is much smaller than the
15111   length of dh->p.  We could use BN_rand_range() if the order of
15112   the subgroup was stored in the DH structure, but we only have
15113   dh->length.
15114
15115   So switch back to
15116
15117           BN_rand(priv_key, l, ...)
15118
15119   where 'l' is dh->length if this is defined, or BN_num_bits(dh->p)-1
15120   otherwise.
15121
15122   *Bodo Moeller*
15123
15124 * In
15125
15126           RSA_eay_public_encrypt
15127           RSA_eay_private_decrypt
15128           RSA_eay_private_encrypt (signing)
15129           RSA_eay_public_decrypt (signature verification)
15130
15131   (default implementations for RSA_public_encrypt,
15132   RSA_private_decrypt, RSA_private_encrypt, RSA_public_decrypt),
15133   always reject numbers >= n.
15134
15135   *Bodo Moeller*
15136
15137 * In crypto/rand/md_rand.c, use a new short-time lock CRYPTO_LOCK_RAND2
15138   to synchronize access to 'locking_thread'.  This is necessary on
15139   systems where access to 'locking_thread' (an 'unsigned long'
15140   variable) is not atomic.
15141
15142   *Bodo Moeller*
15143
15144 * In crypto/rand/md_rand.c, set 'locking_thread' to current thread's ID
15145   *before* setting the 'crypto_lock_rand' flag.  The previous code had
15146   a race condition if 0 is a valid thread ID.
15147
15148   *Travis Vitek <vitek@roguewave.com>*
15149
15150 * Add support for shared libraries under Irix.
15151
15152   *Albert Chin-A-Young <china@thewrittenword.com>*
15153
15154 * Add configuration option to build on Linux on both big-endian and
15155   little-endian MIPS.
15156
15157   *Ralf Baechle <ralf@uni-koblenz.de>*
15158
15159 * Add the possibility to create shared libraries on HP-UX.
15160
15161   *Richard Levitte*
15162
15163### Changes between 0.9.6a and 0.9.6b  [9 Jul 2001]
15164
15165 * Change ssleay_rand_bytes (crypto/rand/md_rand.c)
15166   to avoid a SSLeay/OpenSSL PRNG weakness pointed out by
15167   Markku-Juhani O. Saarinen <markku-juhani.saarinen@nokia.com>:
15168   PRNG state recovery was possible based on the output of
15169   one PRNG request appropriately sized to gain knowledge on
15170   'md' followed by enough consecutive 1-byte PRNG requests
15171   to traverse all of 'state'.
15172
15173   1. When updating 'md_local' (the current thread's copy of 'md')
15174      during PRNG output generation, hash all of the previous
15175      'md_local' value, not just the half used for PRNG output.
15176
15177   2. Make the number of bytes from 'state' included into the hash
15178      independent from the number of PRNG bytes requested.
15179
15180   The first measure alone would be sufficient to avoid
15181   Markku-Juhani's attack.  (Actually it had never occurred
15182   to me that the half of 'md_local' used for chaining was the
15183   half from which PRNG output bytes were taken -- I had always
15184   assumed that the secret half would be used.)  The second
15185   measure makes sure that additional data from 'state' is never
15186   mixed into 'md_local' in small portions; this heuristically
15187   further strengthens the PRNG.
15188
15189   *Bodo Moeller*
15190
15191 * Fix crypto/bn/asm/mips3.s.
15192
15193   *Andy Polyakov*
15194
15195 * When only the key is given to "enc", the IV is undefined. Print out
15196   an error message in this case.
15197
15198   *Lutz Jaenicke*
15199
15200 * Handle special case when X509_NAME is empty in X509 printing routines.
15201
15202   *Steve Henson*
15203
15204 * In dsa_do_verify (crypto/dsa/dsa_ossl.c), verify that r and s are
15205   positive and less than q.
15206
15207   *Bodo Moeller*
15208
15209 * Don't change `*pointer` in CRYPTO_add_lock() is add_lock_callback is
15210   used: it isn't thread safe and the add_lock_callback should handle
15211   that itself.
15212
15213   *Paul Rose <Paul.Rose@bridge.com>*
15214
15215 * Verify that incoming data obeys the block size in
15216   ssl3_enc (ssl/s3_enc.c) and tls1_enc (ssl/t1_enc.c).
15217
15218   *Bodo Moeller*
15219
15220 * Fix OAEP check.
15221
15222   *Ulf Möller, Bodo Möller*
15223
15224 * The countermeasure against Bleichbacher's attack on PKCS #1 v1.5
15225   RSA encryption was accidentally removed in s3_srvr.c in OpenSSL 0.9.5
15226   when fixing the server behaviour for backwards-compatible 'client
15227   hello' messages.  (Note that the attack is impractical against
15228   SSL 3.0 and TLS 1.0 anyway because length and version checking
15229   means that the probability of guessing a valid ciphertext is
15230   around 2^-40; see section 5 in Bleichenbacher's CRYPTO '98
15231   paper.)
15232
15233   Before 0.9.5, the countermeasure (hide the error by generating a
15234   random 'decryption result') did not work properly because
15235   ERR_clear_error() was missing, meaning that SSL_get_error() would
15236   detect the supposedly ignored error.
15237
15238   Both problems are now fixed.
15239
15240   *Bodo Moeller*
15241
15242 * In crypto/bio/bf_buff.c, increase DEFAULT_BUFFER_SIZE to 4096
15243   (previously it was 1024).
15244
15245   *Bodo Moeller*
15246
15247 * Fix for compatibility mode trust settings: ignore trust settings
15248   unless some valid trust or reject settings are present.
15249
15250   *Steve Henson*
15251
15252 * Fix for blowfish EVP: its a variable length cipher.
15253
15254   *Steve Henson*
15255
15256 * Fix various bugs related to DSA S/MIME verification. Handle missing
15257   parameters in DSA public key structures and return an error in the
15258   DSA routines if parameters are absent.
15259
15260   *Steve Henson*
15261
15262 * In versions up to 0.9.6, RAND_file_name() resorted to file ".rnd"
15263   in the current directory if neither $RANDFILE nor $HOME was set.
15264   RAND_file_name() in 0.9.6a returned NULL in this case.  This has
15265   caused some confusion to Windows users who haven't defined $HOME.
15266   Thus RAND_file_name() is changed again: e_os.h can define a
15267   DEFAULT_HOME, which will be used if $HOME is not set.
15268   For Windows, we use "C:"; on other platforms, we still require
15269   environment variables.
15270
15271 * Move 'if (!initialized) RAND_poll()' into regions protected by
15272   CRYPTO_LOCK_RAND.  This is not strictly necessary, but avoids
15273   having multiple threads call RAND_poll() concurrently.
15274
15275   *Bodo Moeller*
15276
15277 * In crypto/rand/md_rand.c, replace 'add_do_not_lock' flag by a
15278   combination of a flag and a thread ID variable.
15279   Otherwise while one thread is in ssleay_rand_bytes (which sets the
15280   flag), *other* threads can enter ssleay_add_bytes without obeying
15281   the CRYPTO_LOCK_RAND lock (and may even illegally release the lock
15282   that they do not hold after the first thread unsets add_do_not_lock).
15283
15284   *Bodo Moeller*
15285
15286 * Change bctest again: '-x' expressions are not available in all
15287   versions of 'test'.
15288
15289   *Bodo Moeller*
15290
15291### Changes between 0.9.6 and 0.9.6a  [5 Apr 2001]
15292
15293 * Fix a couple of memory leaks in PKCS7_dataDecode()
15294
15295   *Steve Henson, reported by Heyun Zheng <hzheng@atdsprint.com>*
15296
15297 * Change Configure and Makefiles to provide EXE_EXT, which will contain
15298   the default extension for executables, if any.  Also, make the perl
15299   scripts that use symlink() to test if it really exists and use "cp"
15300   if it doesn't.  All this made OpenSSL compilable and installable in
15301   CygWin.
15302
15303   *Richard Levitte*
15304
15305 * Fix for asn1_GetSequence() for indefinite length constructed data.
15306   If SEQUENCE is length is indefinite just set c->slen to the total
15307   amount of data available.
15308
15309   *Steve Henson, reported by shige@FreeBSD.org*
15310
15311   *This change does not apply to 0.9.7.*
15312
15313 * Change bctest to avoid here-documents inside command substitution
15314   (workaround for FreeBSD /bin/sh bug).
15315   For compatibility with Ultrix, avoid shell functions (introduced
15316   in the bctest version that searches along $PATH).
15317
15318   *Bodo Moeller*
15319
15320 * Rename 'des_encrypt' to 'des_encrypt1'.  This avoids the clashes
15321   with des_encrypt() defined on some operating systems, like Solaris
15322   and UnixWare.
15323
15324   *Richard Levitte*
15325
15326 * Check the result of RSA-CRT (see D. Boneh, R. DeMillo, R. Lipton:
15327   On the Importance of Eliminating Errors in Cryptographic
15328   Computations, J. Cryptology 14 (2001) 2, 101-119,
15329   <http://theory.stanford.edu/~dabo/papers/faults.ps.gz>).
15330
15331   *Ulf Moeller*
15332
15333 * MIPS assembler BIGNUM division bug fix.
15334
15335   *Andy Polyakov*
15336
15337 * Disabled incorrect Alpha assembler code.
15338
15339   *Richard Levitte*
15340
15341 * Fix PKCS#7 decode routines so they correctly update the length
15342   after reading an EOC for the EXPLICIT tag.
15343
15344   *Steve Henson*
15345
15346   *This change does not apply to 0.9.7.*
15347
15348 * Fix bug in PKCS#12 key generation routines. This was triggered
15349   if a 3DES key was generated with a 0 initial byte. Include
15350   PKCS12_BROKEN_KEYGEN compilation option to retain the old
15351   (but broken) behaviour.
15352
15353   *Steve Henson*
15354
15355 * Enhance bctest to search for a working bc along $PATH and print
15356   it when found.
15357
15358   *Tim Rice <tim@multitalents.net> via Richard Levitte*
15359
15360 * Fix memory leaks in err.c: free err_data string if necessary;
15361   don't write to the wrong index in ERR_set_error_data.
15362
15363   *Bodo Moeller*
15364
15365 * Implement ssl23_peek (analogous to ssl23_read), which previously
15366   did not exist.
15367
15368   *Bodo Moeller*
15369
15370 * Replace rdtsc with `_emit` statements for VC++ version 5.
15371
15372   *Jeremy Cooper <jeremy@baymoo.org>*
15373
15374 * Make it possible to reuse SSLv2 sessions.
15375
15376   *Richard Levitte*
15377
15378 * In copy_email() check for >= 0 as a return value for
15379   X509_NAME_get_index_by_NID() since 0 is a valid index.
15380
15381   *Steve Henson reported by Massimiliano Pala <madwolf@opensca.org>*
15382
15383 * Avoid coredump with unsupported or invalid public keys by checking if
15384   X509_get_pubkey() fails in PKCS7_verify(). Fix memory leak when
15385   PKCS7_verify() fails with non detached data.
15386
15387   *Steve Henson*
15388
15389 * Don't use getenv in library functions when run as setuid/setgid.
15390   New function OPENSSL_issetugid().
15391
15392   *Ulf Moeller*
15393
15394 * Avoid false positives in memory leak detection code (crypto/mem_dbg.c)
15395   due to incorrect handling of multi-threading:
15396
15397   1. Fix timing glitch in the MemCheck_off() portion of CRYPTO_mem_ctrl().
15398
15399   2. Fix logical glitch in is_MemCheck_on() aka CRYPTO_is_mem_check_on().
15400
15401   3. Count how many times MemCheck_off() has been called so that
15402      nested use can be treated correctly.  This also avoids
15403      inband-signalling in the previous code (which relied on the
15404      assumption that thread ID 0 is impossible).
15405
15406   *Bodo Moeller*
15407
15408 * Add "-rand" option also to s_client and s_server.
15409
15410   *Lutz Jaenicke*
15411
15412 * Fix CPU detection on Irix 6.x.
15413   *Kurt Hockenbury <khockenb@stevens-tech.edu> and
15414   "Bruce W. Forsberg" <bruce.forsberg@baesystems.com>*
15415
15416 * Fix X509_NAME bug which produced incorrect encoding if X509_NAME
15417   was empty.
15418
15419   *Steve Henson*
15420
15421   *This change does not apply to 0.9.7.*
15422
15423 * Use the cached encoding of an X509_NAME structure rather than
15424   copying it. This is apparently the reason for the libsafe "errors"
15425   but the code is actually correct.
15426
15427   *Steve Henson*
15428
15429 * Add new function BN_rand_range(), and fix DSA_sign_setup() to prevent
15430   Bleichenbacher's DSA attack.
15431   Extend BN_[pseudo_]rand: As before, top=1 forces the highest two bits
15432   to be set and top=0 forces the highest bit to be set; top=-1 is new
15433   and leaves the highest bit random.
15434
15435   *Ulf Moeller, Bodo Moeller*
15436
15437 * In the `NCONF_...`-based implementations for `CONF_...` queries
15438   (crypto/conf/conf_lib.c), if the input LHASH is NULL, avoid using
15439   a temporary CONF structure with the data component set to NULL
15440   (which gives segmentation faults in lh_retrieve).
15441   Instead, use NULL for the CONF pointer in CONF_get_string and
15442   CONF_get_number (which may use environment variables) and directly
15443   return NULL from CONF_get_section.
15444
15445   *Bodo Moeller*
15446
15447 * Fix potential buffer overrun for EBCDIC.
15448
15449   *Ulf Moeller*
15450
15451 * Tolerate nonRepudiation as being valid for S/MIME signing and certSign
15452   keyUsage if basicConstraints absent for a CA.
15453
15454   *Steve Henson*
15455
15456 * Make SMIME_write_PKCS7() write mail header values with a format that
15457   is more generally accepted (no spaces before the semicolon), since
15458   some programs can't parse those values properly otherwise.  Also make
15459   sure BIO's that break lines after each write do not create invalid
15460   headers.
15461
15462   *Richard Levitte*
15463
15464 * Make the CRL encoding routines work with empty SEQUENCE OF. The
15465   macros previously used would not encode an empty SEQUENCE OF
15466   and break the signature.
15467
15468   *Steve Henson*
15469
15470   *This change does not apply to 0.9.7.*
15471
15472 * Zero the premaster secret after deriving the master secret in
15473   DH ciphersuites.
15474
15475   *Steve Henson*
15476
15477 * Add some EVP_add_digest_alias registrations (as found in
15478   OpenSSL_add_all_digests()) to SSL_library_init()
15479   aka OpenSSL_add_ssl_algorithms().  This provides improved
15480   compatibility with peers using X.509 certificates
15481   with unconventional AlgorithmIdentifier OIDs.
15482
15483   *Bodo Moeller*
15484
15485 * Fix for Irix with NO_ASM.
15486
15487   *"Bruce W. Forsberg" <bruce.forsberg@baesystems.com>*
15488
15489 * ./config script fixes.
15490
15491   *Ulf Moeller, Richard Levitte*
15492
15493 * Fix 'openssl passwd -1'.
15494
15495   *Bodo Moeller*
15496
15497 * Change PKCS12_key_gen_asc() so it can cope with non null
15498   terminated strings whose length is passed in the passlen
15499   parameter, for example from PEM callbacks. This was done
15500   by adding an extra length parameter to asc2uni().
15501
15502   *Steve Henson, reported by <oddissey@samsung.co.kr>*
15503
15504 * Fix C code generated by 'openssl dsaparam -C': If a BN_bin2bn
15505   call failed, free the DSA structure.
15506
15507   *Bodo Moeller*
15508
15509 * Fix to uni2asc() to cope with zero length Unicode strings.
15510   These are present in some PKCS#12 files.
15511
15512   *Steve Henson*
15513
15514 * Increase s2->wbuf allocation by one byte in ssl2_new (ssl/s2_lib.c).
15515   Otherwise do_ssl_write (ssl/s2_pkt.c) will write beyond buffer limits
15516   when writing a 32767 byte record.
15517
15518   *Bodo Moeller; problem reported by Eric Day <eday@concentric.net>*
15519
15520 * In `RSA_eay_public_{en,ed}crypt` and RSA_eay_mod_exp (rsa_eay.c),
15521   obtain lock CRYPTO_LOCK_RSA before setting `rsa->_method_mod_{n,p,q}`.
15522
15523   (RSA objects have a reference count access to which is protected
15524   by CRYPTO_LOCK_RSA [see rsa_lib.c, s3_srvr.c, ssl_cert.c, ssl_rsa.c],
15525   so they are meant to be shared between threads.)
15526   *Bodo Moeller, Geoff Thorpe; original patch submitted by
15527   "Reddie, Steven" <Steven.Reddie@ca.com>*
15528
15529 * Fix a deadlock in CRYPTO_mem_leaks().
15530
15531   *Bodo Moeller*
15532
15533 * Use better test patterns in bntest.
15534
15535   *Ulf Möller*
15536
15537 * rand_win.c fix for Borland C.
15538
15539   *Ulf Möller*
15540
15541 * BN_rshift bugfix for n == 0.
15542
15543   *Bodo Moeller*
15544
15545 * Add a 'bctest' script that checks for some known 'bc' bugs
15546   so that 'make test' does not abort just because 'bc' is broken.
15547
15548   *Bodo Moeller*
15549
15550 * Store verify_result within SSL_SESSION also for client side to
15551   avoid potential security hole. (Re-used sessions on the client side
15552   always resulted in verify_result==X509_V_OK, not using the original
15553   result of the server certificate verification.)
15554
15555   *Lutz Jaenicke*
15556
15557 * Fix ssl3_pending: If the record in s->s3->rrec is not of type
15558   SSL3_RT_APPLICATION_DATA, return 0.
15559   Similarly, change ssl2_pending to return 0 if SSL_in_init(s) is true.
15560
15561   *Bodo Moeller*
15562
15563 * Fix SSL_peek:
15564   Both ssl2_peek and ssl3_peek, which were totally broken in earlier
15565   releases, have been re-implemented by renaming the previous
15566   implementations of ssl2_read and ssl3_read to ssl2_read_internal
15567   and ssl3_read_internal, respectively, and adding 'peek' parameters
15568   to them.  The new ssl[23]_{read,peek} functions are calls to
15569   ssl[23]_read_internal with the 'peek' flag set appropriately.
15570   A 'peek' parameter has also been added to ssl3_read_bytes, which
15571   does the actual work for ssl3_read_internal.
15572
15573   *Bodo Moeller*
15574
15575 * Initialise "ex_data" member of RSA/DSA/DH structures prior to calling
15576   the method-specific "init()" handler. Also clean up ex_data after
15577   calling the method-specific "finish()" handler. Previously, this was
15578   happening the other way round.
15579
15580   *Geoff Thorpe*
15581
15582 * Increase BN_CTX_NUM (the number of BIGNUMs in a BN_CTX) to 16.
15583   The previous value, 12, was not always sufficient for BN_mod_exp().
15584
15585   *Bodo Moeller*
15586
15587 * Make sure that shared libraries get the internal name engine with
15588   the full version number and not just 0.  This should mark the
15589   shared libraries as not backward compatible.  Of course, this should
15590   be changed again when we can guarantee backward binary compatibility.
15591
15592   *Richard Levitte*
15593
15594 * Fix typo in get_cert_by_subject() in by_dir.c
15595
15596   *Jean-Marc Desperrier <jean-marc.desperrier@certplus.com>*
15597
15598 * Rework the system to generate shared libraries:
15599
15600   - Make note of the expected extension for the shared libraries and
15601     if there is a need for symbolic links from for example libcrypto.so.0
15602     to libcrypto.so.0.9.7.  There is extended info in Configure for
15603     that.
15604
15605   - Make as few rebuilds of the shared libraries as possible.
15606
15607   - Still avoid linking the OpenSSL programs with the shared libraries.
15608
15609   - When installing, install the shared libraries separately from the
15610     static ones.
15611
15612   *Richard Levitte*
15613
15614 * Fix SSL_CTX_set_read_ahead macro to actually use its argument.
15615
15616   Copy SSL_CTX's read_ahead flag to SSL object directly in SSL_new
15617   and not in SSL_clear because the latter is also used by the
15618   accept/connect functions; previously, the settings made by
15619   SSL_set_read_ahead would be lost during the handshake.
15620
15621   *Bodo Moeller; problems reported by Anders Gertz <gertz@epact.se>*
15622
15623 * Correct util/mkdef.pl to be selective about disabled algorithms.
15624   Previously, it would create entries for disabled algorithms no
15625   matter what.
15626
15627   *Richard Levitte*
15628
15629 * Added several new manual pages for SSL_* function.
15630
15631   *Lutz Jaenicke*
15632
15633### Changes between 0.9.5a and 0.9.6  [24 Sep 2000]
15634
15635 * In ssl23_get_client_hello, generate an error message when faced
15636   with an initial SSL 3.0/TLS record that is too small to contain the
15637   first two bytes of the ClientHello message, i.e. client_version.
15638   (Note that this is a pathologic case that probably has never happened
15639   in real life.)  The previous approach was to use the version number
15640   from the record header as a substitute; but our protocol choice
15641   should not depend on that one because it is not authenticated
15642   by the Finished messages.
15643
15644   *Bodo Moeller*
15645
15646 * More robust randomness gathering functions for Windows.
15647
15648   *Jeffrey Altman <jaltman@columbia.edu>*
15649
15650 * For compatibility reasons if the flag X509_V_FLAG_ISSUER_CHECK is
15651   not set then we don't setup the error code for issuer check errors
15652   to avoid possibly overwriting other errors which the callback does
15653   handle. If an application does set the flag then we assume it knows
15654   what it is doing and can handle the new informational codes
15655   appropriately.
15656
15657   *Steve Henson*
15658
15659 * Fix for a nasty bug in ASN1_TYPE handling. ASN1_TYPE is used for
15660   a general "ANY" type, as such it should be able to decode anything
15661   including tagged types. However it didn't check the class so it would
15662   wrongly interpret tagged types in the same way as their universal
15663   counterpart and unknown types were just rejected. Changed so that the
15664   tagged and unknown types are handled in the same way as a SEQUENCE:
15665   that is the encoding is stored intact. There is also a new type
15666   "V_ASN1_OTHER" which is used when the class is not universal, in this
15667   case we have no idea what the actual type is so we just lump them all
15668   together.
15669
15670   *Steve Henson*
15671
15672 * On VMS, stdout may very well lead to a file that is written to
15673   in a record-oriented fashion.  That means that every write() will
15674   write a separate record, which will be read separately by the
15675   programs trying to read from it.  This can be very confusing.
15676
15677   The solution is to put a BIO filter in the way that will buffer
15678   text until a linefeed is reached, and then write everything a
15679   line at a time, so every record written will be an actual line,
15680   not chunks of lines and not (usually doesn't happen, but I've
15681   seen it once) several lines in one record.  BIO_f_linebuffer() is
15682   the answer.
15683
15684   Currently, it's a VMS-only method, because that's where it has
15685   been tested well enough.
15686
15687   *Richard Levitte*
15688
15689 * Remove 'optimized' squaring variant in BN_mod_mul_montgomery,
15690   it can return incorrect results.
15691   (Note: The buggy variant was not enabled in OpenSSL 0.9.5a,
15692   but it was in 0.9.6-beta[12].)
15693
15694   *Bodo Moeller*
15695
15696 * Disable the check for content being present when verifying detached
15697   signatures in pk7_smime.c. Some versions of Netscape (wrongly)
15698   include zero length content when signing messages.
15699
15700   *Steve Henson*
15701
15702 * New BIO_shutdown_wr macro, which invokes the BIO_C_SHUTDOWN_WR
15703   BIO_ctrl (for BIO pairs).
15704
15705   *Bodo Möller*
15706
15707 * Add DSO method for VMS.
15708
15709   *Richard Levitte*
15710
15711 * Bug fix: Montgomery multiplication could produce results with the
15712   wrong sign.
15713
15714   *Ulf Möller*
15715
15716 * Add RPM specification openssl.spec and modify it to build three
15717   packages.  The default package contains applications, application
15718   documentation and run-time libraries.  The devel package contains
15719   include files, static libraries and function documentation.  The
15720   doc package contains the contents of the doc directory.  The original
15721   openssl.spec was provided by Damien Miller <djm@mindrot.org>.
15722
15723   *Richard Levitte*
15724
15725 * Add a large number of documentation files for many SSL routines.
15726
15727   *Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>*
15728
15729 * Add a configuration entry for Sony News 4.
15730
15731   *NAKAJI Hiroyuki <nakaji@tutrp.tut.ac.jp>*
15732
15733 * Don't set the two most significant bits to one when generating a
15734   random number < q in the DSA library.
15735
15736   *Ulf Möller*
15737
15738 * New SSL API mode 'SSL_MODE_AUTO_RETRY'.  This disables the default
15739   behaviour that SSL_read may result in SSL_ERROR_WANT_READ (even if
15740   the underlying transport is blocking) if a handshake took place.
15741   (The default behaviour is needed by applications such as s_client
15742   and s_server that use select() to determine when to use SSL_read;
15743   but for applications that know in advance when to expect data, it
15744   just makes things more complicated.)
15745
15746   *Bodo Moeller*
15747
15748 * Add RAND_egd_bytes(), which gives control over the number of bytes read
15749   from EGD.
15750
15751   *Ben Laurie*
15752
15753 * Add a few more EBCDIC conditionals that make `req` and `x509`
15754   work better on such systems.
15755
15756   *Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>*
15757
15758 * Add two demo programs for PKCS12_parse() and PKCS12_create().
15759   Update PKCS12_parse() so it copies the friendlyName and the
15760   keyid to the certificates aux info.
15761
15762   *Steve Henson*
15763
15764 * Fix bug in PKCS7_verify() which caused an infinite loop
15765   if there was more than one signature.
15766
15767   *Sven Uszpelkat <su@celocom.de>*
15768
15769 * Major change in util/mkdef.pl to include extra information
15770   about each symbol, as well as presenting variables as well
15771   as functions.  This change means that there's n more need
15772   to rebuild the .num files when some algorithms are excluded.
15773
15774   *Richard Levitte*
15775
15776 * Allow the verify time to be set by an application,
15777   rather than always using the current time.
15778
15779   *Steve Henson*
15780
15781 * Phase 2 verify code reorganisation. The certificate
15782   verify code now looks up an issuer certificate by a
15783   number of criteria: subject name, authority key id
15784   and key usage. It also verifies self signed certificates
15785   by the same criteria. The main comparison function is
15786   X509_check_issued() which performs these checks.
15787
15788   Lot of changes were necessary in order to support this
15789   without completely rewriting the lookup code.
15790
15791   Authority and subject key identifier are now cached.
15792
15793   The LHASH 'certs' is X509_STORE has now been replaced
15794   by a STACK_OF(X509_OBJECT). This is mainly because an
15795   LHASH can't store or retrieve multiple objects with
15796   the same hash value.
15797
15798   As a result various functions (which were all internal
15799   use only) have changed to handle the new X509_STORE
15800   structure. This will break anything that messed round
15801   with X509_STORE internally.
15802
15803   The functions X509_STORE_add_cert() now checks for an
15804   exact match, rather than just subject name.
15805
15806   The X509_STORE API doesn't directly support the retrieval
15807   of multiple certificates matching a given criteria, however
15808   this can be worked round by performing a lookup first
15809   (which will fill the cache with candidate certificates)
15810   and then examining the cache for matches. This is probably
15811   the best we can do without throwing out X509_LOOKUP
15812   entirely (maybe later...).
15813
15814   The X509_VERIFY_CTX structure has been enhanced considerably.
15815
15816   All certificate lookup operations now go via a get_issuer()
15817   callback. Although this currently uses an X509_STORE it
15818   can be replaced by custom lookups. This is a simple way
15819   to bypass the X509_STORE hackery necessary to make this
15820   work and makes it possible to use more efficient techniques
15821   in future. A very simple version which uses a simple
15822   STACK for its trusted certificate store is also provided
15823   using X509_STORE_CTX_trusted_stack().
15824
15825   The verify_cb() and verify() callbacks now have equivalents
15826   in the X509_STORE_CTX structure.
15827
15828   X509_STORE_CTX also has a 'flags' field which can be used
15829   to customise the verify behaviour.
15830
15831   *Steve Henson*
15832
15833 * Add new PKCS#7 signing option PKCS7_NOSMIMECAP which
15834   excludes S/MIME capabilities.
15835
15836   *Steve Henson*
15837
15838 * When a certificate request is read in keep a copy of the
15839   original encoding of the signed data and use it when outputting
15840   again. Signatures then use the original encoding rather than
15841   a decoded, encoded version which may cause problems if the
15842   request is improperly encoded.
15843
15844   *Steve Henson*
15845
15846 * For consistency with other BIO_puts implementations, call
15847   buffer_write(b, ...) directly in buffer_puts instead of calling
15848   BIO_write(b, ...).
15849
15850   In BIO_puts, increment b->num_write as in BIO_write.
15851
15852   *Peter.Sylvester@EdelWeb.fr*
15853
15854 * Fix BN_mul_word for the case where the word is 0. (We have to use
15855   BN_zero, we may not return a BIGNUM with an array consisting of
15856   words set to zero.)
15857
15858   *Bodo Moeller*
15859
15860 * Avoid calling abort() from within the library when problems are
15861   detected, except if preprocessor symbols have been defined
15862   (such as REF_CHECK, BN_DEBUG etc.).
15863
15864   *Bodo Moeller*
15865
15866 * New openssl application 'rsautl'. This utility can be
15867   used for low-level RSA operations. DER public key
15868   BIO/fp routines also added.
15869
15870   *Steve Henson*
15871
15872 * New Configure entry and patches for compiling on QNX 4.
15873
15874   *Andreas Schneider <andreas@ds3.etech.fh-hamburg.de>*
15875
15876 * A demo state-machine implementation was sponsored by
15877   Nuron (<http://www.nuron.com/>) and is now available in
15878   demos/state_machine.
15879
15880   *Ben Laurie*
15881
15882 * New options added to the 'dgst' utility for signature
15883   generation and verification.
15884
15885   *Steve Henson*
15886
15887 * Unrecognized PKCS#7 content types are now handled via a
15888   catch all ASN1_TYPE structure. This allows unsupported
15889   types to be stored as a "blob" and an application can
15890   encode and decode it manually.
15891
15892   *Steve Henson*
15893
15894 * Fix various signed/unsigned issues to make a_strex.c
15895   compile under VC++.
15896
15897   *Oscar Jacobsson <oscar.jacobsson@celocom.com>*
15898
15899 * ASN1 fixes. i2d_ASN1_OBJECT was not returning the correct
15900   length if passed a buffer. ASN1_INTEGER_to_BN failed
15901   if passed a NULL BN and its argument was negative.
15902
15903   *Steve Henson, pointed out by Sven Heiberg <sven@tartu.cyber.ee>*
15904
15905 * Modification to PKCS#7 encoding routines to output definite
15906   length encoding. Since currently the whole structures are in
15907   memory there's not real point in using indefinite length
15908   constructed encoding. However if OpenSSL is compiled with
15909   the flag PKCS7_INDEFINITE_ENCODING the old form is used.
15910
15911   *Steve Henson*
15912
15913 * Added BIO_vprintf() and BIO_vsnprintf().
15914
15915   *Richard Levitte*
15916
15917 * Added more prefixes to parse for in the strings written
15918   through a logging bio, to cover all the levels that are available
15919   through syslog.  The prefixes are now:
15920
15921           PANIC, EMERG, EMR       =>      LOG_EMERG
15922           ALERT, ALR              =>      LOG_ALERT
15923           CRIT, CRI               =>      LOG_CRIT
15924           ERROR, ERR              =>      LOG_ERR
15925           WARNING, WARN, WAR      =>      LOG_WARNING
15926           NOTICE, NOTE, NOT       =>      LOG_NOTICE
15927           INFO, INF               =>      LOG_INFO
15928           DEBUG, DBG              =>      LOG_DEBUG
15929
15930   and as before, if none of those prefixes are present at the
15931   beginning of the string, LOG_ERR is chosen.
15932
15933   On Win32, the `LOG_*` levels are mapped according to this:
15934
15935           LOG_EMERG, LOG_ALERT, LOG_CRIT, LOG_ERR => EVENTLOG_ERROR_TYPE
15936           LOG_WARNING                             => EVENTLOG_WARNING_TYPE
15937           LOG_NOTICE, LOG_INFO, LOG_DEBUG         => EVENTLOG_INFORMATION_TYPE
15938
15939   *Richard Levitte*
15940
15941 * Made it possible to reconfigure with just the configuration
15942   argument "reconf" or "reconfigure".  The command line arguments
15943   are stored in Makefile.ssl in the variable CONFIGURE_ARGS,
15944   and are retrieved from there when reconfiguring.
15945
15946   *Richard Levitte*
15947
15948 * MD4 implemented.
15949
15950   *Assar Westerlund <assar@sics.se>, Richard Levitte*
15951
15952 * Add the arguments -CAfile and -CApath to the pkcs12 utility.
15953
15954   *Richard Levitte*
15955
15956 * The obj_dat.pl script was messing up the sorting of object
15957   names. The reason was that it compared the quoted version
15958   of strings as a result "OCSP" > "OCSP Signing" because
15959   " > SPACE. Changed script to store unquoted versions of
15960   names and add quotes on output. It was also omitting some
15961   names from the lookup table if they were given a default
15962   value (that is if SN is missing it is given the same
15963   value as LN and vice versa), these are now added on the
15964   grounds that if an object has a name we should be able to
15965   look it up. Finally added warning output when duplicate
15966   short or long names are found.
15967
15968   *Steve Henson*
15969
15970 * Changes needed for Tandem NSK.
15971
15972   *Scott Uroff <scott@xypro.com>*
15973
15974 * Fix SSL 2.0 rollback checking: Due to an off-by-one error in
15975   RSA_padding_check_SSLv23(), special padding was never detected
15976   and thus the SSL 3.0/TLS 1.0 countermeasure against protocol
15977   version rollback attacks was not effective.
15978
15979   In s23_clnt.c, don't use special rollback-attack detection padding
15980   (RSA_SSLV23_PADDING) if SSL 2.0 is the only protocol enabled in the
15981   client; similarly, in s23_srvr.c, don't do the rollback check if
15982   SSL 2.0 is the only protocol enabled in the server.
15983
15984   *Bodo Moeller*
15985
15986 * Make it possible to get hexdumps of unprintable data with 'openssl
15987   asn1parse'.  By implication, the functions ASN1_parse_dump() and
15988   BIO_dump_indent() are added.
15989
15990   *Richard Levitte*
15991
15992 * New functions ASN1_STRING_print_ex() and X509_NAME_print_ex()
15993   these print out strings and name structures based on various
15994   flags including RFC2253 support and proper handling of
15995   multibyte characters. Added options to the 'x509' utility
15996   to allow the various flags to be set.
15997
15998   *Steve Henson*
15999
16000 * Various fixes to use ASN1_TIME instead of ASN1_UTCTIME.
16001   Also change the functions X509_cmp_current_time() and
16002   X509_gmtime_adj() work with an ASN1_TIME structure,
16003   this will enable certificates using GeneralizedTime in validity
16004   dates to be checked.
16005
16006   *Steve Henson*
16007
16008 * Make the NEG_PUBKEY_BUG code (which tolerates invalid
16009   negative public key encodings) on by default,
16010   NO_NEG_PUBKEY_BUG can be set to disable it.
16011
16012   *Steve Henson*
16013
16014 * New function c2i_ASN1_OBJECT() which acts on ASN1_OBJECT
16015   content octets. An i2c_ASN1_OBJECT is unnecessary because
16016   the encoding can be trivially obtained from the structure.
16017
16018   *Steve Henson*
16019
16020 * crypto/err.c locking bugfix: Use write locks (`CRYPTO_w_[un]lock`),
16021   not read locks (`CRYPTO_r_[un]lock`).
16022
16023   *Bodo Moeller*
16024
16025 * A first attempt at creating official support for shared
16026   libraries through configuration.  I've kept it so the
16027   default is static libraries only, and the OpenSSL programs
16028   are always statically linked for now, but there are
16029   preparations for dynamic linking in place.
16030   This has been tested on Linux and Tru64.
16031
16032   *Richard Levitte*
16033
16034 * Randomness polling function for Win9x, as described in:
16035   Peter Gutmann, Software Generation of Practically Strong
16036   Random Numbers.
16037
16038   *Ulf Möller*
16039
16040 * Fix so PRNG is seeded in req if using an already existing
16041   DSA key.
16042
16043   *Steve Henson*
16044
16045 * New options to smime application. -inform and -outform
16046   allow alternative formats for the S/MIME message including
16047   PEM and DER. The -content option allows the content to be
16048   specified separately. This should allow things like Netscape
16049   form signing output easier to verify.
16050
16051   *Steve Henson*
16052
16053 * Fix the ASN1 encoding of tags using the 'long form'.
16054
16055   *Steve Henson*
16056
16057 * New ASN1 functions, `i2c_*` and `c2i_*` for INTEGER and BIT
16058   STRING types. These convert content octets to and from the
16059   underlying type. The actual tag and length octets are
16060   already assumed to have been read in and checked. These
16061   are needed because all other string types have virtually
16062   identical handling apart from the tag. By having versions
16063   of the ASN1 functions that just operate on content octets
16064   IMPLICIT tagging can be handled properly. It also allows
16065   the ASN1_ENUMERATED code to be cut down because ASN1_ENUMERATED
16066   and ASN1_INTEGER are identical apart from the tag.
16067
16068   *Steve Henson*
16069
16070 * Change the handling of OID objects as follows:
16071
16072   - New object identifiers are inserted in objects.txt, following
16073     the syntax given in [crypto/objects/README.md](crypto/objects/README.md).
16074   - objects.pl is used to process obj_mac.num and create a new
16075     obj_mac.h.
16076   - obj_dat.pl is used to create a new obj_dat.h, using the data in
16077     obj_mac.h.
16078
16079   This is currently kind of a hack, and the perl code in objects.pl
16080   isn't very elegant, but it works as I intended.  The simplest way
16081   to check that it worked correctly is to look in obj_dat.h and
16082   check the array nid_objs and make sure the objects haven't moved
16083   around (this is important!).  Additions are OK, as well as
16084   consistent name changes.
16085
16086   *Richard Levitte*
16087
16088 * Add BSD-style MD5-based passwords to 'openssl passwd' (option '-1').
16089
16090   *Bodo Moeller*
16091
16092 * Addition of the command line parameter '-rand file' to 'openssl req'.
16093   The given file adds to whatever has already been seeded into the
16094   random pool through the RANDFILE configuration file option or
16095   environment variable, or the default random state file.
16096
16097   *Richard Levitte*
16098
16099 * mkstack.pl now sorts each macro group into lexical order.
16100   Previously the output order depended on the order the files
16101   appeared in the directory, resulting in needless rewriting
16102   of safestack.h .
16103
16104   *Steve Henson*
16105
16106 * Patches to make OpenSSL compile under Win32 again. Mostly
16107   work arounds for the VC++ problem that it treats func() as
16108   func(void). Also stripped out the parts of mkdef.pl that
16109   added extra typesafe functions: these no longer exist.
16110
16111   *Steve Henson*
16112
16113 * Reorganisation of the stack code. The macros are now all
16114   collected in safestack.h . Each macro is defined in terms of
16115   a "stack macro" of the form `SKM_<name>(type, a, b)`. The
16116   DEBUG_SAFESTACK is now handled in terms of function casts,
16117   this has the advantage of retaining type safety without the
16118   use of additional functions. If DEBUG_SAFESTACK is not defined
16119   then the non typesafe macros are used instead. Also modified the
16120   mkstack.pl script to handle the new form. Needs testing to see
16121   if which (if any) compilers it chokes and maybe make DEBUG_SAFESTACK
16122   the default if no major problems. Similar behaviour for ASN1_SET_OF
16123   and PKCS12_STACK_OF.
16124
16125   *Steve Henson*
16126
16127 * When some versions of IIS use the 'NET' form of private key the
16128   key derivation algorithm is different. Normally MD5(password) is
16129   used as a 128 bit RC4 key. In the modified case
16130   MD5(MD5(password) + "SGCKEYSALT")  is used instead. Added some
16131   new functions i2d_RSA_NET(), d2i_RSA_NET() etc which are the same
16132   as the old Netscape_RSA functions except they have an additional
16133   'sgckey' parameter which uses the modified algorithm. Also added
16134   an -sgckey command line option to the rsa utility. Thanks to
16135   Adrian Peck <bertie@ncipher.com> for posting details of the modified
16136   algorithm to openssl-dev.
16137
16138   *Steve Henson*
16139
16140 * The evp_local.h macros were using 'c.##kname' which resulted in
16141   invalid expansion on some systems (SCO 5.0.5 for example).
16142   Corrected to 'c.kname'.
16143
16144   *Phillip Porch <root@theporch.com>*
16145
16146 * New X509_get1_email() and X509_REQ_get1_email() functions that return
16147   a STACK of email addresses from a certificate or request, these look
16148   in the subject name and the subject alternative name extensions and
16149   omit any duplicate addresses.
16150
16151   *Steve Henson*
16152
16153 * Re-implement BN_mod_exp2_mont using independent (and larger) windows.
16154   This makes DSA verification about 2 % faster.
16155
16156   *Bodo Moeller*
16157
16158 * Increase maximum window size in `BN_mod_exp_...` to 6 bits instead of 5
16159   (meaning that now 2^5 values will be precomputed, which is only 4 KB
16160   plus overhead for 1024 bit moduli).
16161   This makes exponentiations about 0.5 % faster for 1024 bit
16162   exponents (as measured by "openssl speed rsa2048").
16163
16164   *Bodo Moeller*
16165
16166 * Rename memory handling macros to avoid conflicts with other
16167   software:
16168           Malloc         =>  OPENSSL_malloc
16169           Malloc_locked  =>  OPENSSL_malloc_locked
16170           Realloc        =>  OPENSSL_realloc
16171           Free           =>  OPENSSL_free
16172
16173   *Richard Levitte*
16174
16175 * New function BN_mod_exp_mont_word for small bases (roughly 15%
16176   faster than BN_mod_exp_mont, i.e. 7% for a full DH exchange).
16177
16178   *Bodo Moeller*
16179
16180 * CygWin32 support.
16181
16182   *John Jarvie <jjarvie@newsguy.com>*
16183
16184 * The type-safe stack code has been rejigged. It is now only compiled
16185   in when OpenSSL is configured with the DEBUG_SAFESTACK option and
16186   by default all type-specific stack functions are "#define"d back to
16187   standard stack functions. This results in more streamlined output
16188   but retains the type-safety checking possibilities of the original
16189   approach.
16190
16191   *Geoff Thorpe*
16192
16193 * The STACK code has been cleaned up, and certain type declarations
16194   that didn't make a lot of sense have been brought in line. This has
16195   also involved a cleanup of sorts in safestack.h to more correctly
16196   map type-safe stack functions onto their plain stack counterparts.
16197   This work has also resulted in a variety of "const"ifications of
16198   lots of the code, especially `_cmp` operations which should normally
16199   be prototyped with "const" parameters anyway.
16200
16201   *Geoff Thorpe*
16202
16203 * When generating bytes for the first time in md_rand.c, 'stir the pool'
16204   by seeding with STATE_SIZE dummy bytes (with zero entropy count).
16205   (The PRNG state consists of two parts, the large pool 'state' and 'md',
16206   where all of 'md' is used each time the PRNG is used, but 'state'
16207   is used only indexed by a cyclic counter. As entropy may not be
16208   well distributed from the beginning, 'md' is important as a
16209   chaining variable. However, the output function chains only half
16210   of 'md', i.e. 80 bits.  ssleay_rand_add, on the other hand, chains
16211   all of 'md', and seeding with STATE_SIZE dummy bytes will result
16212   in all of 'state' being rewritten, with the new values depending
16213   on virtually all of 'md'.  This overcomes the 80 bit limitation.)
16214
16215   *Bodo Moeller*
16216
16217 * In ssl/s2_clnt.c and ssl/s3_clnt.c, call ERR_clear_error() when
16218   the handshake is continued after ssl_verify_cert_chain();
16219   otherwise, if SSL_VERIFY_NONE is set, remaining error codes
16220   can lead to 'unexplainable' connection aborts later.
16221
16222   *Bodo Moeller; problem tracked down by Lutz Jaenicke*
16223
16224 * Major EVP API cipher revision.
16225   Add hooks for extra EVP features. This allows various cipher
16226   parameters to be set in the EVP interface. Support added for variable
16227   key length ciphers via the EVP_CIPHER_CTX_set_key_length() function and
16228   setting of RC2 and RC5 parameters.
16229
16230   Modify EVP_OpenInit() and EVP_SealInit() to cope with variable key length
16231   ciphers.
16232
16233   Remove lots of duplicated code from the EVP library. For example *every*
16234   cipher init() function handles the 'iv' in the same way according to the
16235   cipher mode. They also all do nothing if the 'key' parameter is NULL and
16236   for CFB and OFB modes they zero ctx->num.
16237
16238   New functionality allows removal of S/MIME code RC2 hack.
16239
16240   Most of the routines have the same form and so can be declared in terms
16241   of macros.
16242
16243   By shifting this to the top level EVP_CipherInit() it can be removed from
16244   all individual ciphers. If the cipher wants to handle IVs or keys
16245   differently it can set the EVP_CIPH_CUSTOM_IV or EVP_CIPH_ALWAYS_CALL_INIT
16246   flags.
16247
16248   Change lots of functions like EVP_EncryptUpdate() to now return a
16249   value: although software versions of the algorithms cannot fail
16250   any installed hardware versions can.
16251
16252   *Steve Henson*
16253
16254 * Implement SSL_OP_TLS_ROLLBACK_BUG: In ssl3_get_client_key_exchange, if
16255   this option is set, tolerate broken clients that send the negotiated
16256   protocol version number instead of the requested protocol version
16257   number.
16258
16259   *Bodo Moeller*
16260
16261 * Call dh_tmp_cb (set by `..._TMP_DH_CB`) with correct 'is_export' flag;
16262   i.e. non-zero for export ciphersuites, zero otherwise.
16263   Previous versions had this flag inverted, inconsistent with
16264   rsa_tmp_cb (..._TMP_RSA_CB).
16265
16266   *Bodo Moeller; problem reported by Amit Chopra*
16267
16268 * Add missing DSA library text string. Work around for some IIS
16269   key files with invalid SEQUENCE encoding.
16270
16271   *Steve Henson*
16272
16273 * Add a document (doc/standards.txt) that list all kinds of standards
16274   and so on that are implemented in OpenSSL.
16275
16276   *Richard Levitte*
16277
16278 * Enhance c_rehash script. Old version would mishandle certificates
16279   with the same subject name hash and wouldn't handle CRLs at all.
16280   Added -fingerprint option to crl utility, to support new c_rehash
16281   features.
16282
16283   *Steve Henson*
16284
16285 * Eliminate non-ANSI declarations in crypto.h and stack.h.
16286
16287   *Ulf Möller*
16288
16289 * Fix for SSL server purpose checking. Server checking was
16290   rejecting certificates which had extended key usage present
16291   but no ssl client purpose.
16292
16293   *Steve Henson, reported by Rene Grosser <grosser@hisolutions.com>*
16294
16295 * Make PKCS#12 code work with no password. The PKCS#12 spec
16296   is a little unclear about how a blank password is handled.
16297   Since the password in encoded as a BMPString with terminating
16298   double NULL a zero length password would end up as just the
16299   double NULL. However no password at all is different and is
16300   handled differently in the PKCS#12 key generation code. NS
16301   treats a blank password as zero length. MSIE treats it as no
16302   password on export: but it will try both on import. We now do
16303   the same: PKCS12_parse() tries zero length and no password if
16304   the password is set to "" or NULL (NULL is now a valid password:
16305   it wasn't before) as does the pkcs12 application.
16306
16307   *Steve Henson*
16308
16309 * Bugfixes in `apps/x509.c`: Avoid a memory leak; and don't use
16310   perror when PEM_read_bio_X509_REQ fails, the error message must
16311   be obtained from the error queue.
16312
16313   *Bodo Moeller*
16314
16315 * Avoid 'thread_hash' memory leak in crypto/err/err.c by freeing
16316   it in ERR_remove_state if appropriate, and change ERR_get_state
16317   accordingly to avoid race conditions (this is necessary because
16318   thread_hash is no longer constant once set).
16319
16320   *Bodo Moeller*
16321
16322 * Bugfix for linux-elf makefile.one.
16323
16324   *Ulf Möller*
16325
16326 * RSA_get_default_method() will now cause a default
16327   RSA_METHOD to be chosen if one doesn't exist already.
16328   Previously this was only set during a call to RSA_new()
16329   or RSA_new_method(NULL) meaning it was possible for
16330   RSA_get_default_method() to return NULL.
16331
16332   *Geoff Thorpe*
16333
16334 * Added native name translation to the existing DSO code
16335   that will convert (if the flag to do so is set) filenames
16336   that are sufficiently small and have no path information
16337   into a canonical native form. Eg. "blah" converted to
16338   "libblah.so" or "blah.dll" etc.
16339
16340   *Geoff Thorpe*
16341
16342 * New function ERR_error_string_n(e, buf, len) which is like
16343   ERR_error_string(e, buf), but writes at most 'len' bytes
16344   including the 0 terminator.  For ERR_error_string_n, 'buf'
16345   may not be NULL.
16346
16347   *Damien Miller <djm@mindrot.org>, Bodo Moeller*
16348
16349 * CONF library reworked to become more general.  A new CONF
16350   configuration file reader "class" is implemented as well as a
16351   new functions (`NCONF_*`, for "New CONF") to handle it.  The now
16352   old `CONF_*` functions are still there, but are reimplemented to
16353   work in terms of the new functions.  Also, a set of functions
16354   to handle the internal storage of the configuration data is
16355   provided to make it easier to write new configuration file
16356   reader "classes" (I can definitely see something reading a
16357   configuration file in XML format, for example), called `_CONF_*`,
16358   or "the configuration storage API"...
16359
16360   The new configuration file reading functions are:
16361
16362           NCONF_new, NCONF_free, NCONF_load, NCONF_load_fp, NCONF_load_bio,
16363           NCONF_get_section, NCONF_get_string, NCONF_get_numbre
16364
16365           NCONF_default, NCONF_WIN32
16366
16367           NCONF_dump_fp, NCONF_dump_bio
16368
16369   NCONF_default and NCONF_WIN32 are method (or "class") choosers,
16370   NCONF_new creates a new CONF object.  This works in the same way
16371   as other interfaces in OpenSSL, like the BIO interface.
16372   `NCONF_dump_*` dump the internal storage of the configuration file,
16373   which is useful for debugging.  All other functions take the same
16374   arguments as the old `CONF_*` functions with the exception of the
16375   first that must be a `CONF *` instead of a `LHASH *`.
16376
16377   To make it easier to use the new classes with the old `CONF_*` functions,
16378   the function CONF_set_default_method is provided.
16379
16380   *Richard Levitte*
16381
16382 * Add '-tls1' option to 'openssl ciphers', which was already
16383   mentioned in the documentation but had not been implemented.
16384   (This option is not yet really useful because even the additional
16385   experimental TLS 1.0 ciphers are currently treated as SSL 3.0 ciphers.)
16386
16387   *Bodo Moeller*
16388
16389 * Initial DSO code added into libcrypto for letting OpenSSL (and
16390   OpenSSL-based applications) load shared libraries and bind to
16391   them in a portable way.
16392
16393   *Geoff Thorpe, with contributions from Richard Levitte*
16394
16395### Changes between 0.9.5 and 0.9.5a  [1 Apr 2000]
16396
16397 * Make sure _lrotl and _lrotr are only used with MSVC.
16398
16399 * Use lock CRYPTO_LOCK_RAND correctly in ssleay_rand_status
16400   (the default implementation of RAND_status).
16401
16402 * Rename openssl x509 option '-crlext', which was added in 0.9.5,
16403   to '-clrext' (= clear extensions), as intended and documented.
16404   *Bodo Moeller; inconsistency pointed out by Michael Attili
16405   <attili@amaxo.com>*
16406
16407 * Fix for HMAC. It wasn't zeroing the rest of the block if the key length
16408   was larger than the MD block size.
16409
16410   *Steve Henson, pointed out by Yost William <YostW@tce.com>*
16411
16412 * Modernise PKCS12_parse() so it uses STACK_OF(X509) for its ca argument
16413   fix a leak when the ca argument was passed as NULL. Stop X509_PUBKEY_set()
16414   using the passed key: if the passed key was a private key the result
16415   of X509_print(), for example, would be to print out all the private key
16416   components.
16417
16418   *Steve Henson*
16419
16420 * des_quad_cksum() byte order bug fix.
16421   *Ulf Möller, using the problem description in krb4-0.9.7, where
16422   the solution is attributed to Derrick J Brashear <shadow@DEMENTIA.ORG>*
16423
16424 * Fix so V_ASN1_APP_CHOOSE works again: however its use is strongly
16425   discouraged.
16426
16427   *Steve Henson, pointed out by Brian Korver <briank@cs.stanford.edu>*
16428
16429 * For easily testing in shell scripts whether some command
16430   'openssl XXX' exists, the new pseudo-command 'openssl no-XXX'
16431   returns with exit code 0 iff no command of the given name is available.
16432   'no-XXX' is printed in this case, 'XXX' otherwise.  In both cases,
16433   the output goes to stdout and nothing is printed to stderr.
16434   Additional arguments are always ignored.
16435
16436   Since for each cipher there is a command of the same name,
16437   the 'no-cipher' compilation switches can be tested this way.
16438
16439   ('openssl no-XXX' is not able to detect pseudo-commands such
16440   as 'quit', 'list-XXX-commands', or 'no-XXX' itself.)
16441
16442   *Bodo Moeller*
16443
16444 * Update test suite so that 'make test' succeeds in 'no-rsa' configuration.
16445
16446   *Bodo Moeller*
16447
16448 * For SSL_[CTX_]set_tmp_dh, don't create a DH key if SSL_OP_SINGLE_DH_USE
16449   is set; it will be thrown away anyway because each handshake creates
16450   its own key.
16451   ssl_cert_dup, which is used by SSL_new, now copies DH keys in addition
16452   to parameters -- in previous versions (since OpenSSL 0.9.3) the
16453   'default key' from SSL_CTX_set_tmp_dh would always be lost, meaning
16454   you effectively got SSL_OP_SINGLE_DH_USE when using this macro.
16455
16456   *Bodo Moeller*
16457
16458 * New s_client option -ign_eof: EOF at stdin is ignored, and
16459   'Q' and 'R' lose their special meanings (quit/renegotiate).
16460   This is part of what -quiet does; unlike -quiet, -ign_eof
16461   does not suppress any output.
16462
16463   *Richard Levitte*
16464
16465 * Add compatibility options to the purpose and trust code. The
16466   purpose X509_PURPOSE_ANY is "any purpose" which automatically
16467   accepts a certificate or CA, this was the previous behaviour,
16468   with all the associated security issues.
16469
16470   X509_TRUST_COMPAT is the old trust behaviour: only and
16471   automatically trust self signed roots in certificate store. A
16472   new trust setting X509_TRUST_DEFAULT is used to specify that
16473   a purpose has no associated trust setting and it should instead
16474   use the value in the default purpose.
16475
16476   *Steve Henson*
16477
16478 * Fix the PKCS#8 DSA private key code so it decodes keys again
16479   and fix a memory leak.
16480
16481   *Steve Henson*
16482
16483 * In util/mkerr.pl (which implements 'make errors'), preserve
16484   reason strings from the previous version of the .c file, as
16485   the default to have only downcase letters (and digits) in
16486   automatically generated reasons codes is not always appropriate.
16487
16488   *Bodo Moeller*
16489
16490 * In ERR_load_ERR_strings(), build an ERR_LIB_SYS error reason table
16491   using strerror.  Previously, ERR_reason_error_string() returned
16492   library names as reason strings for SYSerr; but SYSerr is a special
16493   case where small numbers are errno values, not library numbers.
16494
16495   *Bodo Moeller*
16496
16497 * Add '-dsaparam' option to 'openssl dhparam' application.  This
16498   converts DSA parameters into DH parameters. (When creating parameters,
16499   DSA_generate_parameters is used.)
16500
16501   *Bodo Moeller*
16502
16503 * Include 'length' (recommended exponent length) in C code generated
16504   by 'openssl dhparam -C'.
16505
16506   *Bodo Moeller*
16507
16508 * The second argument to set_label in perlasm was already being used
16509   so couldn't be used as a "file scope" flag. Moved to third argument
16510   which was free.
16511
16512   *Steve Henson*
16513
16514 * In PEM_ASN1_write_bio and some other functions, use RAND_pseudo_bytes
16515   instead of RAND_bytes for encryption IVs and salts.
16516
16517   *Bodo Moeller*
16518
16519 * Include RAND_status() into RAND_METHOD instead of implementing
16520   it only for md_rand.c  Otherwise replacing the PRNG by calling
16521   RAND_set_rand_method would be impossible.
16522
16523   *Bodo Moeller*
16524
16525 * Don't let DSA_generate_key() enter an infinite loop if the random
16526   number generation fails.
16527
16528   *Bodo Moeller*
16529
16530 * New 'rand' application for creating pseudo-random output.
16531
16532   *Bodo Moeller*
16533
16534 * Added configuration support for Linux/IA64
16535
16536   *Rolf Haberrecker <rolf@suse.de>*
16537
16538 * Assembler module support for Mingw32.
16539
16540   *Ulf Möller*
16541
16542 * Shared library support for HPUX (in shlib/).
16543
16544   *Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE> and Anonymous*
16545
16546 * Shared library support for Solaris gcc.
16547
16548   *Lutz Behnke <behnke@trustcenter.de>*
16549
16550### Changes between 0.9.4 and 0.9.5  [28 Feb 2000]
16551
16552 * PKCS7_encrypt() was adding text MIME headers twice because they
16553   were added manually and by SMIME_crlf_copy().
16554
16555   *Steve Henson*
16556
16557 * In bntest.c don't call BN_rand with zero bits argument.
16558
16559   *Steve Henson, pointed out by Andrew W. Gray <agray@iconsinc.com>*
16560
16561 * BN_mul bugfix: In bn_mul_part_recursion() only the a>a[n] && b>b[n]
16562   case was implemented. This caused BN_div_recp() to fail occasionally.
16563
16564   *Ulf Möller*
16565
16566 * Add an optional second argument to the set_label() in the perl
16567   assembly language builder. If this argument exists and is set
16568   to 1 it signals that the assembler should use a symbol whose
16569   scope is the entire file, not just the current function. This
16570   is needed with MASM which uses the format label:: for this scope.
16571
16572   *Steve Henson, pointed out by Peter Runestig <peter@runestig.com>*
16573
16574 * Change the ASN1 types so they are typedefs by default. Before
16575   almost all types were #define'd to ASN1_STRING which was causing
16576   STACK_OF() problems: you couldn't declare STACK_OF(ASN1_UTF8STRING)
16577   for example.
16578
16579   *Steve Henson*
16580
16581 * Change names of new functions to the new get1/get0 naming
16582   convention: After 'get1', the caller owns a reference count
16583   and has to call `..._free`; 'get0' returns a pointer to some
16584   data structure without incrementing reference counters.
16585   (Some of the existing 'get' functions increment a reference
16586   counter, some don't.)
16587   Similarly, 'set1' and 'add1' functions increase reference
16588   counters or duplicate objects.
16589
16590   *Steve Henson*
16591
16592 * Allow for the possibility of temp RSA key generation failure:
16593   the code used to assume it always worked and crashed on failure.
16594
16595   *Steve Henson*
16596
16597 * Fix potential buffer overrun problem in BIO_printf().
16598   *Ulf Möller, using public domain code by Patrick Powell; problem
16599   pointed out by David Sacerdote <das33@cornell.edu>*
16600
16601 * Support EGD <http://www.lothar.com/tech/crypto/>.  New functions
16602   RAND_egd() and RAND_status().  In the command line application,
16603   the EGD socket can be specified like a seed file using RANDFILE
16604   or -rand.
16605
16606   *Ulf Möller*
16607
16608 * Allow the string CERTIFICATE to be tolerated in PKCS#7 structures.
16609   Some CAs (e.g. Verisign) distribute certificates in this form.
16610
16611   *Steve Henson*
16612
16613 * Remove the SSL_ALLOW_ADH compile option and set the default cipher
16614   list to exclude them. This means that no special compilation option
16615   is needed to use anonymous DH: it just needs to be included in the
16616   cipher list.
16617
16618   *Steve Henson*
16619
16620 * Change the EVP_MD_CTX_type macro so its meaning consistent with
16621   EVP_MD_type. The old functionality is available in a new macro called
16622   EVP_MD_md(). Change code that uses it and update docs.
16623
16624   *Steve Henson*
16625
16626 * `..._ctrl` functions now have corresponding `..._callback_ctrl` functions
16627   where the `void *` argument is replaced by a function pointer argument.
16628   Previously `void *` was abused to point to functions, which works on
16629   many platforms, but is not correct.  As these functions are usually
16630   called by macros defined in OpenSSL header files, most source code
16631   should work without changes.
16632
16633   *Richard Levitte*
16634
16635 * `<openssl/opensslconf.h>` (which is created by Configure) now contains
16636   sections with information on -D... compiler switches used for
16637   compiling the library so that applications can see them.  To enable
16638   one of these sections, a pre-processor symbol `OPENSSL_..._DEFINES`
16639   must be defined.  E.g.,
16640           #define OPENSSL_ALGORITHM_DEFINES
16641           #include <openssl/opensslconf.h>
16642   defines all pertinent `NO_<algo>` symbols, such as NO_IDEA, NO_RSA, etc.
16643
16644   *Richard Levitte, Ulf and Bodo Möller*
16645
16646 * Bugfix: Tolerate fragmentation and interleaving in the SSL 3/TLS
16647   record layer.
16648
16649   *Bodo Moeller*
16650
16651 * Change the 'other' type in certificate aux info to a STACK_OF
16652   X509_ALGOR. Although not an AlgorithmIdentifier as such it has
16653   the required ASN1 format: arbitrary types determined by an OID.
16654
16655   *Steve Henson*
16656
16657 * Add some PEM_write_X509_REQ_NEW() functions and a command line
16658   argument to 'req'. This is not because the function is newer or
16659   better than others it just uses the work 'NEW' in the certificate
16660   request header lines. Some software needs this.
16661
16662   *Steve Henson*
16663
16664 * Reorganise password command line arguments: now passwords can be
16665   obtained from various sources. Delete the PEM_cb function and make
16666   it the default behaviour: i.e. if the callback is NULL and the
16667   usrdata argument is not NULL interpret it as a null terminated pass
16668   phrase. If usrdata and the callback are NULL then the pass phrase
16669   is prompted for as usual.
16670
16671   *Steve Henson*
16672
16673 * Add support for the Compaq Atalla crypto accelerator. If it is installed,
16674   the support is automatically enabled. The resulting binaries will
16675   autodetect the card and use it if present.
16676
16677   *Ben Laurie and Compaq Inc.*
16678
16679 * Work around for Netscape hang bug. This sends certificate request
16680   and server done in one record. Since this is perfectly legal in the
16681   SSL/TLS protocol it isn't a "bug" option and is on by default. See
16682   the bugs/SSLv3 entry for more info.
16683
16684   *Steve Henson*
16685
16686 * HP-UX tune-up: new unified configs, HP C compiler bug workaround.
16687
16688   *Andy Polyakov*
16689
16690 * Add -rand argument to smime and pkcs12 applications and read/write
16691   of seed file.
16692
16693   *Steve Henson*
16694
16695 * New 'passwd' tool for crypt(3) and apr1 password hashes.
16696
16697   *Bodo Moeller*
16698
16699 * Add command line password options to the remaining applications.
16700
16701   *Steve Henson*
16702
16703 * Bug fix for BN_div_recp() for numerators with an even number of
16704   bits.
16705
16706   *Ulf Möller*
16707
16708 * More tests in bntest.c, and changed test_bn output.
16709
16710   *Ulf Möller*
16711
16712 * ./config recognizes MacOS X now.
16713
16714   *Andy Polyakov*
16715
16716 * Bug fix for BN_div() when the first words of num and divisor are
16717   equal (it gave wrong results if `(rem=(n1-q*d0)&BN_MASK2) < d0)`.
16718
16719   *Ulf Möller*
16720
16721 * Add support for various broken PKCS#8 formats, and command line
16722   options to produce them.
16723
16724   *Steve Henson*
16725
16726 * New functions BN_CTX_start(), BN_CTX_get() and BT_CTX_end() to
16727   get temporary BIGNUMs from a BN_CTX.
16728
16729   *Ulf Möller*
16730
16731 * Correct return values in BN_mod_exp_mont() and BN_mod_exp2_mont()
16732   for p == 0.
16733
16734   *Ulf Möller*
16735
16736 * Change the `SSLeay_add_all_*()` functions to `OpenSSL_add_all_*()` and
16737   include a #define from the old name to the new. The original intent
16738   was that statically linked binaries could for example just call
16739   SSLeay_add_all_ciphers() to just add ciphers to the table and not
16740   link with digests. This never worked because SSLeay_add_all_digests()
16741   and SSLeay_add_all_ciphers() were in the same source file so calling
16742   one would link with the other. They are now in separate source files.
16743
16744   *Steve Henson*
16745
16746 * Add a new -notext option to 'ca' and a -pubkey option to 'spkac'.
16747
16748   *Steve Henson*
16749
16750 * Use a less unusual form of the Miller-Rabin primality test (it used
16751   a binary algorithm for exponentiation integrated into the Miller-Rabin
16752   loop, our standard modexp algorithms are faster).
16753
16754   *Bodo Moeller*
16755
16756 * Support for the EBCDIC character set completed.
16757
16758   *Martin Kraemer <Martin.Kraemer@Mch.SNI.De>*
16759
16760 * Source code cleanups: use const where appropriate, eliminate casts,
16761   use `void *` instead of `char *` in lhash.
16762
16763   *Ulf Möller*
16764
16765 * Bugfix: ssl3_send_server_key_exchange was not restartable
16766   (the state was not changed to SSL3_ST_SW_KEY_EXCH_B, and because of
16767   this the server could overwrite ephemeral keys that the client
16768   has already seen).
16769
16770   *Bodo Moeller*
16771
16772 * Turn DSA_is_prime into a macro that calls BN_is_prime,
16773   using 50 iterations of the Rabin-Miller test.
16774
16775   DSA_generate_parameters now uses BN_is_prime_fasttest (with 50
16776   iterations of the Rabin-Miller test as required by the appendix
16777   to FIPS PUB 186[-1]) instead of DSA_is_prime.
16778   As BN_is_prime_fasttest includes trial division, DSA parameter
16779   generation becomes much faster.
16780
16781   This implies a change for the callback functions in DSA_is_prime
16782   and DSA_generate_parameters: The callback function is called once
16783   for each positive witness in the Rabin-Miller test, not just
16784   occasionally in the inner loop; and the parameters to the
16785   callback function now provide an iteration count for the outer
16786   loop rather than for the current invocation of the inner loop.
16787   DSA_generate_parameters additionally can call the callback
16788   function with an 'iteration count' of -1, meaning that a
16789   candidate has passed the trial division test (when q is generated
16790   from an application-provided seed, trial division is skipped).
16791
16792   *Bodo Moeller*
16793
16794 * New function BN_is_prime_fasttest that optionally does trial
16795   division before starting the Rabin-Miller test and has
16796   an additional BN_CTX * argument (whereas BN_is_prime always
16797   has to allocate at least one BN_CTX).
16798   'callback(1, -1, cb_arg)' is called when a number has passed the
16799   trial division stage.
16800
16801   *Bodo Moeller*
16802
16803 * Fix for bug in CRL encoding. The validity dates weren't being handled
16804   as ASN1_TIME.
16805
16806   *Steve Henson*
16807
16808 * New -pkcs12 option to CA.pl script to write out a PKCS#12 file.
16809
16810   *Steve Henson*
16811
16812 * New function BN_pseudo_rand().
16813
16814   *Ulf Möller*
16815
16816 * Clean up BN_mod_mul_montgomery(): replace the broken (and unreadable)
16817   bignum version of BN_from_montgomery() with the working code from
16818   SSLeay 0.9.0 (the word based version is faster anyway), and clean up
16819   the comments.
16820
16821   *Ulf Möller*
16822
16823 * Avoid a race condition in s2_clnt.c (function get_server_hello) that
16824   made it impossible to use the same SSL_SESSION data structure in
16825   SSL2 clients in multiple threads.
16826
16827   *Bodo Moeller*
16828
16829 * The return value of RAND_load_file() no longer counts bytes obtained
16830   by stat().  RAND_load_file(..., -1) is new and uses the complete file
16831   to seed the PRNG (previously an explicit byte count was required).
16832
16833   *Ulf Möller, Bodo Möller*
16834
16835 * Clean up CRYPTO_EX_DATA functions, some of these didn't have prototypes
16836   used `char *` instead of `void *` and had casts all over the place.
16837
16838   *Steve Henson*
16839
16840 * Make BN_generate_prime() return NULL on error if ret!=NULL.
16841
16842   *Ulf Möller*
16843
16844 * Retain source code compatibility for BN_prime_checks macro:
16845   BN_is_prime(..., BN_prime_checks, ...) now uses
16846   BN_prime_checks_for_size to determine the appropriate number of
16847   Rabin-Miller iterations.
16848
16849   *Ulf Möller*
16850
16851 * Diffie-Hellman uses "safe" primes: DH_check() return code renamed to
16852   DH_CHECK_P_NOT_SAFE_PRIME.
16853   (Check if this is true? OpenPGP calls them "strong".)
16854
16855   *Ulf Möller*
16856
16857 * Merge the functionality of "dh" and "gendh" programs into a new program
16858   "dhparam". The old programs are retained for now but will handle DH keys
16859   (instead of parameters) in future.
16860
16861   *Steve Henson*
16862
16863 * Make the ciphers, s_server and s_client programs check the return values
16864   when a new cipher list is set.
16865
16866   *Steve Henson*
16867
16868 * Enhance the SSL/TLS cipher mechanism to correctly handle the TLS 56bit
16869   ciphers. Before when the 56bit ciphers were enabled the sorting was
16870   wrong.
16871
16872   The syntax for the cipher sorting has been extended to support sorting by
16873   cipher-strength (using the strength_bits hard coded in the tables).
16874   The new command is `@STRENGTH` (see also `doc/apps/ciphers.pod`).
16875
16876   Fix a bug in the cipher-command parser: when supplying a cipher command
16877   string with an "undefined" symbol (neither command nor alphanumeric
16878   *A-Za-z0-9*, ssl_set_cipher_list used to hang in an endless loop. Now
16879   an error is flagged.
16880
16881   Due to the strength-sorting extension, the code of the
16882   ssl_create_cipher_list() function was completely rearranged. I hope that
16883   the readability was also increased :-)
16884
16885   *Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>*
16886
16887 * Minor change to 'x509' utility. The -CAcreateserial option now uses 1
16888   for the first serial number and places 2 in the serial number file. This
16889   avoids problems when the root CA is created with serial number zero and
16890   the first user certificate has the same issuer name and serial number
16891   as the root CA.
16892
16893   *Steve Henson*
16894
16895 * Fixes to X509_ATTRIBUTE utilities, change the 'req' program so it uses
16896   the new code. Add documentation for this stuff.
16897
16898   *Steve Henson*
16899
16900 * Changes to X509_ATTRIBUTE utilities. These have been renamed from
16901   `X509_*()` to `X509at_*()` on the grounds that they don't handle X509
16902   structures and behave in an analogous way to the X509v3 functions:
16903   they shouldn't be called directly but wrapper functions should be used
16904   instead.
16905
16906   So we also now have some wrapper functions that call the X509at functions
16907   when passed certificate requests. (TO DO: similar things can be done with
16908   PKCS#7 signed and unsigned attributes, PKCS#12 attributes and a few other
16909   things. Some of these need some d2i or i2d and print functionality
16910   because they handle more complex structures.)
16911
16912   *Steve Henson*
16913
16914 * Add missing #ifndefs that caused missing symbols when building libssl
16915   as a shared library without RSA.  Use #ifndef NO_SSL2 instead of
16916   NO_RSA in `ssl/s2*.c`.
16917
16918   *Kris Kennaway <kris@hub.freebsd.org>, modified by Ulf Möller*
16919
16920 * Precautions against using the PRNG uninitialized: RAND_bytes() now
16921   has a return value which indicates the quality of the random data
16922   (1 = ok, 0 = not seeded).  Also an error is recorded on the thread's
16923   error queue. New function RAND_pseudo_bytes() generates output that is
16924   guaranteed to be unique but not unpredictable. RAND_add is like
16925   RAND_seed, but takes an extra argument for an entropy estimate
16926   (RAND_seed always assumes full entropy).
16927
16928   *Ulf Möller*
16929
16930 * Do more iterations of Rabin-Miller probable prime test (specifically,
16931   3 for 1024-bit primes, 6 for 512-bit primes, 12 for 256-bit primes
16932   instead of only 2 for all lengths; see BN_prime_checks_for_size definition
16933   in crypto/bn/bn_prime.c for the complete table).  This guarantees a
16934   false-positive rate of at most 2^-80 for random input.
16935
16936   *Bodo Moeller*
16937
16938 * Rewrite ssl3_read_n (ssl/s3_pkt.c) avoiding a couple of bugs.
16939
16940   *Bodo Moeller*
16941
16942 * New function X509_CTX_rget_chain() (renamed to X509_CTX_get1_chain
16943   in the 0.9.5 release), this returns the chain
16944   from an X509_CTX structure with a dup of the stack and all
16945   the X509 reference counts upped: so the stack will exist
16946   after X509_CTX_cleanup() has been called. Modify pkcs12.c
16947   to use this.
16948
16949   Also make SSL_SESSION_print() print out the verify return
16950   code.
16951
16952   *Steve Henson*
16953
16954 * Add manpage for the pkcs12 command. Also change the default
16955   behaviour so MAC iteration counts are used unless the new
16956   -nomaciter option is used. This improves file security and
16957   only older versions of MSIE (4.0 for example) need it.
16958
16959   *Steve Henson*
16960
16961 * Honor the no-xxx Configure options when creating .DEF files.
16962
16963   *Ulf Möller*
16964
16965 * Add PKCS#10 attributes to field table: challengePassword,
16966   unstructuredName and unstructuredAddress. These are taken from
16967   draft PKCS#9 v2.0 but are compatible with v1.2 provided no
16968   international characters are used.
16969
16970   More changes to X509_ATTRIBUTE code: allow the setting of types
16971   based on strings. Remove the 'loc' parameter when adding
16972   attributes because these will be a SET OF encoding which is sorted
16973   in ASN1 order.
16974
16975   *Steve Henson*
16976
16977 * Initial changes to the 'req' utility to allow request generation
16978   automation. This will allow an application to just generate a template
16979   file containing all the field values and have req construct the
16980   request.
16981
16982   Initial support for X509_ATTRIBUTE handling. Stacks of these are
16983   used all over the place including certificate requests and PKCS#7
16984   structures. They are currently handled manually where necessary with
16985   some primitive wrappers for PKCS#7. The new functions behave in a
16986   manner analogous to the X509 extension functions: they allow
16987   attributes to be looked up by NID and added.
16988
16989   Later something similar to the X509V3 code would be desirable to
16990   automatically handle the encoding, decoding and printing of the
16991   more complex types. The string types like challengePassword can
16992   be handled by the string table functions.
16993
16994   Also modified the multi byte string table handling. Now there is
16995   a 'global mask' which masks out certain types. The table itself
16996   can use the flag STABLE_NO_MASK to ignore the mask setting: this
16997   is useful when for example there is only one permissible type
16998   (as in countryName) and using the mask might result in no valid
16999   types at all.
17000
17001   *Steve Henson*
17002
17003 * Clean up 'Finished' handling, and add functions SSL_get_finished and
17004   SSL_get_peer_finished to allow applications to obtain the latest
17005   Finished messages sent to the peer or expected from the peer,
17006   respectively.  (SSL_get_peer_finished is usually the Finished message
17007   actually received from the peer, otherwise the protocol will be aborted.)
17008
17009   As the Finished message are message digests of the complete handshake
17010   (with a total of 192 bits for TLS 1.0 and more for SSL 3.0), they can
17011   be used for external authentication procedures when the authentication
17012   provided by SSL/TLS is not desired or is not enough.
17013
17014   *Bodo Moeller*
17015
17016 * Enhanced support for Alpha Linux is added. Now ./config checks if
17017   the host supports BWX extension and if Compaq C is present on the
17018   $PATH. Just exploiting of the BWX extension results in 20-30%
17019   performance kick for some algorithms, e.g. DES and RC4 to mention
17020   a couple. Compaq C in turn generates ~20% faster code for MD5 and
17021   SHA1.
17022
17023   *Andy Polyakov*
17024
17025 * Add support for MS "fast SGC". This is arguably a violation of the
17026   SSL3/TLS protocol. Netscape SGC does two handshakes: the first with
17027   weak crypto and after checking the certificate is SGC a second one
17028   with strong crypto. MS SGC stops the first handshake after receiving
17029   the server certificate message and sends a second client hello. Since
17030   a server will typically do all the time consuming operations before
17031   expecting any further messages from the client (server key exchange
17032   is the most expensive) there is little difference between the two.
17033
17034   To get OpenSSL to support MS SGC we have to permit a second client
17035   hello message after we have sent server done. In addition we have to
17036   reset the MAC if we do get this second client hello.
17037
17038   *Steve Henson*
17039
17040 * Add a function 'd2i_AutoPrivateKey()' this will automatically decide
17041   if a DER encoded private key is RSA or DSA traditional format. Changed
17042   d2i_PrivateKey_bio() to use it. This is only needed for the "traditional"
17043   format DER encoded private key. Newer code should use PKCS#8 format which
17044   has the key type encoded in the ASN1 structure. Added DER private key
17045   support to pkcs8 application.
17046
17047   *Steve Henson*
17048
17049 * SSL 3/TLS 1 servers now don't request certificates when an anonymous
17050   ciphersuites has been selected (as required by the SSL 3/TLS 1
17051   specifications).  Exception: When SSL_VERIFY_FAIL_IF_NO_PEER_CERT
17052   is set, we interpret this as a request to violate the specification
17053   (the worst that can happen is a handshake failure, and 'correct'
17054   behaviour would result in a handshake failure anyway).
17055
17056   *Bodo Moeller*
17057
17058 * In SSL_CTX_add_session, take into account that there might be multiple
17059   SSL_SESSION structures with the same session ID (e.g. when two threads
17060   concurrently obtain them from an external cache).
17061   The internal cache can handle only one SSL_SESSION with a given ID,
17062   so if there's a conflict, we now throw out the old one to achieve
17063   consistency.
17064
17065   *Bodo Moeller*
17066
17067 * Add OIDs for idea and blowfish in CBC mode. This will allow both
17068   to be used in PKCS#5 v2.0 and S/MIME.  Also add checking to
17069   some routines that use cipher OIDs: some ciphers do not have OIDs
17070   defined and so they cannot be used for S/MIME and PKCS#5 v2.0 for
17071   example.
17072
17073   *Steve Henson*
17074
17075 * Simplify the trust setting structure and code. Now we just have
17076   two sequences of OIDs for trusted and rejected settings. These will
17077   typically have values the same as the extended key usage extension
17078   and any application specific purposes.
17079
17080   The trust checking code now has a default behaviour: it will just
17081   check for an object with the same NID as the passed id. Functions can
17082   be provided to override either the default behaviour or the behaviour
17083   for a given id. SSL client, server and email already have functions
17084   in place for compatibility: they check the NID and also return "trusted"
17085   if the certificate is self signed.
17086
17087   *Steve Henson*
17088
17089 * Add d2i,i2d bio/fp functions for PrivateKey: these convert the
17090   traditional format into an EVP_PKEY structure.
17091
17092   *Steve Henson*
17093
17094 * Add a password callback function PEM_cb() which either prompts for
17095   a password if usr_data is NULL or otherwise assumes it is a null
17096   terminated password. Allow passwords to be passed on command line
17097   environment or config files in a few more utilities.
17098
17099   *Steve Henson*
17100
17101 * Add a bunch of DER and PEM functions to handle PKCS#8 format private
17102   keys. Add some short names for PKCS#8 PBE algorithms and allow them
17103   to be specified on the command line for the pkcs8 and pkcs12 utilities.
17104   Update documentation.
17105
17106   *Steve Henson*
17107
17108 * Support for ASN1 "NULL" type. This could be handled before by using
17109   ASN1_TYPE but there wasn't any function that would try to read a NULL
17110   and produce an error if it couldn't. For compatibility we also have
17111   ASN1_NULL_new() and ASN1_NULL_free() functions but these are faked and
17112   don't allocate anything because they don't need to.
17113
17114   *Steve Henson*
17115
17116 * Initial support for MacOS is now provided. Examine INSTALL.MacOS
17117   for details.
17118
17119   *Andy Polyakov, Roy Woods <roy@centicsystems.ca>*
17120
17121 * Rebuild of the memory allocation routines used by OpenSSL code and
17122   possibly others as well.  The purpose is to make an interface that
17123   provide hooks so anyone can build a separate set of allocation and
17124   deallocation routines to be used by OpenSSL, for example memory
17125   pool implementations, or something else, which was previously hard
17126   since Malloc(), Realloc() and Free() were defined as macros having
17127   the values malloc, realloc and free, respectively (except for Win32
17128   compilations).  The same is provided for memory debugging code.
17129   OpenSSL already comes with functionality to find memory leaks, but
17130   this gives people a chance to debug other memory problems.
17131
17132   With these changes, a new set of functions and macros have appeared:
17133
17134     CRYPTO_set_mem_debug_functions()         [F]
17135     CRYPTO_get_mem_debug_functions()         [F]
17136     CRYPTO_dbg_set_options()                 [F]
17137     CRYPTO_dbg_get_options()                 [F]
17138     CRYPTO_malloc_debug_init()               [M]
17139
17140   The memory debug functions are NULL by default, unless the library
17141   is compiled with CRYPTO_MDEBUG or friends is defined.  If someone
17142   wants to debug memory anyway, CRYPTO_malloc_debug_init() (which
17143   gives the standard debugging functions that come with OpenSSL) or
17144   CRYPTO_set_mem_debug_functions() (tells OpenSSL to use functions
17145   provided by the library user) must be used.  When the standard
17146   debugging functions are used, CRYPTO_dbg_set_options can be used to
17147   request additional information:
17148   CRYPTO_dbg_set_options(V_CYRPTO_MDEBUG_xxx) corresponds to setting
17149   the CRYPTO_MDEBUG_xxx macro when compiling the library.
17150
17151   Also, things like CRYPTO_set_mem_functions will always give the
17152   expected result (the new set of functions is used for allocation
17153   and deallocation) at all times, regardless of platform and compiler
17154   options.
17155
17156   To finish it up, some functions that were never use in any other
17157   way than through macros have a new API and new semantic:
17158
17159     CRYPTO_dbg_malloc()
17160     CRYPTO_dbg_realloc()
17161     CRYPTO_dbg_free()
17162
17163   All macros of value have retained their old syntax.
17164
17165   *Richard Levitte and Bodo Moeller*
17166
17167 * Some S/MIME fixes. The OID for SMIMECapabilities was wrong, the
17168   ordering of SMIMECapabilities wasn't in "strength order" and there
17169   was a missing NULL in the AlgorithmIdentifier for the SHA1 signature
17170   algorithm.
17171
17172   *Steve Henson*
17173
17174 * Some ASN1 types with illegal zero length encoding (INTEGER,
17175   ENUMERATED and OBJECT IDENTIFIER) choked the ASN1 routines.
17176
17177   *Frans Heymans <fheymans@isaserver.be>, modified by Steve Henson*
17178
17179 * Merge in my S/MIME library for OpenSSL. This provides a simple
17180   S/MIME API on top of the PKCS#7 code, a MIME parser (with enough
17181   functionality to handle multipart/signed properly) and a utility
17182   called 'smime' to call all this stuff. This is based on code I
17183   originally wrote for Celo who have kindly allowed it to be
17184   included in OpenSSL.
17185
17186   *Steve Henson*
17187
17188 * Add variants des_set_key_checked and des_set_key_unchecked of
17189   des_set_key (aka des_key_sched).  Global variable des_check_key
17190   decides which of these is called by des_set_key; this way
17191   des_check_key behaves as it always did, but applications and
17192   the library itself, which was buggy for des_check_key == 1,
17193   have a cleaner way to pick the version they need.
17194
17195   *Bodo Moeller*
17196
17197 * New function PKCS12_newpass() which changes the password of a
17198   PKCS12 structure.
17199
17200   *Steve Henson*
17201
17202 * Modify X509_TRUST and X509_PURPOSE so it also uses a static and
17203   dynamic mix. In both cases the ids can be used as an index into the
17204   table. Also modified the X509_TRUST_add() and X509_PURPOSE_add()
17205   functions so they accept a list of the field values and the
17206   application doesn't need to directly manipulate the X509_TRUST
17207   structure.
17208
17209   *Steve Henson*
17210
17211 * Modify the ASN1_STRING_TABLE stuff so it also uses bsearch and doesn't
17212   need initialising.
17213
17214   *Steve Henson*
17215
17216 * Modify the way the V3 extension code looks up extensions. This now
17217   works in a similar way to the object code: we have some "standard"
17218   extensions in a static table which is searched with OBJ_bsearch()
17219   and the application can add dynamic ones if needed. The file
17220   crypto/x509v3/ext_dat.h now has the info: this file needs to be
17221   updated whenever a new extension is added to the core code and kept
17222   in ext_nid order. There is a simple program 'tabtest.c' which checks
17223   this. New extensions are not added too often so this file can readily
17224   be maintained manually.
17225
17226   There are two big advantages in doing things this way. The extensions
17227   can be looked up immediately and no longer need to be "added" using
17228   X509V3_add_standard_extensions(): this function now does nothing.
17229   Side note: I get *lots* of email saying the extension code doesn't
17230   work because people forget to call this function.
17231   Also no dynamic allocation is done unless new extensions are added:
17232   so if we don't add custom extensions there is no need to call
17233   X509V3_EXT_cleanup().
17234
17235   *Steve Henson*
17236
17237 * Modify enc utility's salting as follows: make salting the default. Add a
17238   magic header, so unsalted files fail gracefully instead of just decrypting
17239   to garbage. This is because not salting is a big security hole, so people
17240   should be discouraged from doing it.
17241
17242   *Ben Laurie*
17243
17244 * Fixes and enhancements to the 'x509' utility. It allowed a message
17245   digest to be passed on the command line but it only used this
17246   parameter when signing a certificate. Modified so all relevant
17247   operations are affected by the digest parameter including the
17248   -fingerprint and -x509toreq options. Also -x509toreq choked if a
17249   DSA key was used because it didn't fix the digest.
17250
17251   *Steve Henson*
17252
17253 * Initial certificate chain verify code. Currently tests the untrusted
17254   certificates for consistency with the verify purpose (which is set
17255   when the X509_STORE_CTX structure is set up) and checks the pathlength.
17256
17257   There is a NO_CHAIN_VERIFY compilation option to keep the old behaviour:
17258   this is because it will reject chains with invalid extensions whereas
17259   every previous version of OpenSSL and SSLeay made no checks at all.
17260
17261   Trust code: checks the root CA for the relevant trust settings. Trust
17262   settings have an initial value consistent with the verify purpose: e.g.
17263   if the verify purpose is for SSL client use it expects the CA to be
17264   trusted for SSL client use. However the default value can be changed to
17265   permit custom trust settings: one example of this would be to only trust
17266   certificates from a specific "secure" set of CAs.
17267
17268   Also added X509_STORE_CTX_new() and X509_STORE_CTX_free() functions
17269   which should be used for version portability: especially since the
17270   verify structure is likely to change more often now.
17271
17272   SSL integration. Add purpose and trust to SSL_CTX and SSL and functions
17273   to set them. If not set then assume SSL clients will verify SSL servers
17274   and vice versa.
17275
17276   Two new options to the verify program: -untrusted allows a set of
17277   untrusted certificates to be passed in and -purpose which sets the
17278   intended purpose of the certificate. If a purpose is set then the
17279   new chain verify code is used to check extension consistency.
17280
17281   *Steve Henson*
17282
17283 * Support for the authority information access extension.
17284
17285   *Steve Henson*
17286
17287 * Modify RSA and DSA PEM read routines to transparently handle
17288   PKCS#8 format private keys. New *_PUBKEY_* functions that handle
17289   public keys in a format compatible with certificate
17290   SubjectPublicKeyInfo structures. Unfortunately there were already
17291   functions called *_PublicKey_* which used various odd formats so
17292   these are retained for compatibility: however the DSA variants were
17293   never in a public release so they have been deleted. Changed dsa/rsa
17294   utilities to handle the new format: note no releases ever handled public
17295   keys so we should be OK.
17296
17297   The primary motivation for this change is to avoid the same fiasco
17298   that dogs private keys: there are several incompatible private key
17299   formats some of which are standard and some OpenSSL specific and
17300   require various evil hacks to allow partial transparent handling and
17301   even then it doesn't work with DER formats. Given the option anything
17302   other than PKCS#8 should be dumped: but the other formats have to
17303   stay in the name of compatibility.
17304
17305   With public keys and the benefit of hindsight one standard format
17306   is used which works with EVP_PKEY, RSA or DSA structures: though
17307   it clearly returns an error if you try to read the wrong kind of key.
17308
17309   Added a -pubkey option to the 'x509' utility to output the public key.
17310   Also rename the `EVP_PKEY_get_*()` to `EVP_PKEY_rget_*()`
17311   (renamed to `EVP_PKEY_get1_*()` in the OpenSSL 0.9.5 release) and add
17312   `EVP_PKEY_rset_*()` functions (renamed to `EVP_PKEY_set1_*()`)
17313   that do the same as the `EVP_PKEY_assign_*()` except they up the
17314   reference count of the added key (they don't "swallow" the
17315   supplied key).
17316
17317   *Steve Henson*
17318
17319 * Fixes to crypto/x509/by_file.c the code to read in certificates and
17320   CRLs would fail if the file contained no certificates or no CRLs:
17321   added a new function to read in both types and return the number
17322   read: this means that if none are read it will be an error. The
17323   DER versions of the certificate and CRL reader would always fail
17324   because it isn't possible to mix certificates and CRLs in DER format
17325   without choking one or the other routine. Changed this to just read
17326   a certificate: this is the best we can do. Also modified the code
17327   in `apps/verify.c` to take notice of return codes: it was previously
17328   attempting to read in certificates from NULL pointers and ignoring
17329   any errors: this is one reason why the cert and CRL reader seemed
17330   to work. It doesn't check return codes from the default certificate
17331   routines: these may well fail if the certificates aren't installed.
17332
17333   *Steve Henson*
17334
17335 * Code to support otherName option in GeneralName.
17336
17337   *Steve Henson*
17338
17339 * First update to verify code. Change the verify utility
17340   so it warns if it is passed a self signed certificate:
17341   for consistency with the normal behaviour. X509_verify
17342   has been modified to it will now verify a self signed
17343   certificate if *exactly* the same certificate appears
17344   in the store: it was previously impossible to trust a
17345   single self signed certificate. This means that:
17346   openssl verify ss.pem
17347   now gives a warning about a self signed certificate but
17348   openssl verify -CAfile ss.pem ss.pem
17349   is OK.
17350
17351   *Steve Henson*
17352
17353 * For servers, store verify_result in SSL_SESSION data structure
17354   (and add it to external session representation).
17355   This is needed when client certificate verifications fails,
17356   but an application-provided verification callback (set by
17357   SSL_CTX_set_cert_verify_callback) allows accepting the session
17358   anyway (i.e. leaves x509_store_ctx->error != X509_V_OK
17359   but returns 1): When the session is reused, we have to set
17360   ssl->verify_result to the appropriate error code to avoid
17361   security holes.
17362
17363   *Bodo Moeller, problem pointed out by Lutz Jaenicke*
17364
17365 * Fix a bug in the new PKCS#7 code: it didn't consider the
17366   case in PKCS7_dataInit() where the signed PKCS7 structure
17367   didn't contain any existing data because it was being created.
17368
17369   *Po-Cheng Chen <pocheng@nst.com.tw>, slightly modified by Steve Henson*
17370
17371 * Add a salt to the key derivation routines in enc.c. This
17372   forms the first 8 bytes of the encrypted file. Also add a
17373   -S option to allow a salt to be input on the command line.
17374
17375   *Steve Henson*
17376
17377 * New function X509_cmp(). Oddly enough there wasn't a function
17378   to compare two certificates. We do this by working out the SHA1
17379   hash and comparing that. X509_cmp() will be needed by the trust
17380   code.
17381
17382   *Steve Henson*
17383
17384 * SSL_get1_session() is like SSL_get_session(), but increments
17385   the reference count in the SSL_SESSION returned.
17386
17387   *Geoff Thorpe <geoff@eu.c2.net>*
17388
17389 * Fix for 'req': it was adding a null to request attributes.
17390   Also change the X509_LOOKUP and X509_INFO code to handle
17391   certificate auxiliary information.
17392
17393   *Steve Henson*
17394
17395 * Add support for 40 and 64 bit RC2 and RC4 algorithms: document
17396   the 'enc' command.
17397
17398   *Steve Henson*
17399
17400 * Add the possibility to add extra information to the memory leak
17401   detecting output, to form tracebacks, showing from where each
17402   allocation was originated: CRYPTO_push_info("constant string") adds
17403   the string plus current file name and line number to a per-thread
17404   stack, CRYPTO_pop_info() does the obvious, CRYPTO_remove_all_info()
17405   is like calling CYRPTO_pop_info() until the stack is empty.
17406   Also updated memory leak detection code to be multi-thread-safe.
17407
17408   *Richard Levitte*
17409
17410 * Add options -text and -noout to pkcs7 utility and delete the
17411   encryption options which never did anything. Update docs.
17412
17413   *Steve Henson*
17414
17415 * Add options to some of the utilities to allow the pass phrase
17416   to be included on either the command line (not recommended on
17417   OSes like Unix) or read from the environment. Update the
17418   manpages and fix a few bugs.
17419
17420   *Steve Henson*
17421
17422 * Add a few manpages for some of the openssl commands.
17423
17424   *Steve Henson*
17425
17426 * Fix the -revoke option in ca. It was freeing up memory twice,
17427   leaking and not finding already revoked certificates.
17428
17429   *Steve Henson*
17430
17431 * Extensive changes to support certificate auxiliary information.
17432   This involves the use of X509_CERT_AUX structure and X509_AUX
17433   functions. An X509_AUX function such as PEM_read_X509_AUX()
17434   can still read in a certificate file in the usual way but it
17435   will also read in any additional "auxiliary information". By
17436   doing things this way a fair degree of compatibility can be
17437   retained: existing certificates can have this information added
17438   using the new 'x509' options.
17439
17440   Current auxiliary information includes an "alias" and some trust
17441   settings. The trust settings will ultimately be used in enhanced
17442   certificate chain verification routines: currently a certificate
17443   can only be trusted if it is self signed and then it is trusted
17444   for all purposes.
17445
17446   *Steve Henson*
17447
17448 * Fix assembler for Alpha (tested only on DEC OSF not Linux or `*BSD`).
17449   The problem was that one of the replacement routines had not been working
17450   since SSLeay releases.  For now the offending routine has been replaced
17451   with non-optimised assembler.  Even so, this now gives around 95%
17452   performance improvement for 1024 bit RSA signs.
17453
17454   *Mark Cox*
17455
17456 * Hack to fix PKCS#7 decryption when used with some unorthodox RC2
17457   handling. Most clients have the effective key size in bits equal to
17458   the key length in bits: so a 40 bit RC2 key uses a 40 bit (5 byte) key.
17459   A few however don't do this and instead use the size of the decrypted key
17460   to determine the RC2 key length and the AlgorithmIdentifier to determine
17461   the effective key length. In this case the effective key length can still
17462   be 40 bits but the key length can be 168 bits for example. This is fixed
17463   by manually forcing an RC2 key into the EVP_PKEY structure because the
17464   EVP code can't currently handle unusual RC2 key sizes: it always assumes
17465   the key length and effective key length are equal.
17466
17467   *Steve Henson*
17468
17469 * Add a bunch of functions that should simplify the creation of
17470   X509_NAME structures. Now you should be able to do:
17471   X509_NAME_add_entry_by_txt(nm, "CN", MBSTRING_ASC, "Steve", -1, -1, 0);
17472   and have it automatically work out the correct field type and fill in
17473   the structures. The more adventurous can try:
17474   X509_NAME_add_entry_by_txt(nm, field, MBSTRING_UTF8, str, -1, -1, 0);
17475   and it will (hopefully) work out the correct multibyte encoding.
17476
17477   *Steve Henson*
17478
17479 * Change the 'req' utility to use the new field handling and multibyte
17480   copy routines. Before the DN field creation was handled in an ad hoc
17481   way in req, ca, and x509 which was rather broken and didn't support
17482   BMPStrings or UTF8Strings. Since some software doesn't implement
17483   BMPStrings or UTF8Strings yet, they can be enabled using the config file
17484   using the dirstring_type option. See the new comment in the default
17485   openssl.cnf for more info.
17486
17487   *Steve Henson*
17488
17489 * Make crypto/rand/md_rand.c more robust:
17490   - Assure unique random numbers after fork().
17491   - Make sure that concurrent threads access the global counter and
17492     md serializably so that we never lose entropy in them
17493     or use exactly the same state in multiple threads.
17494     Access to the large state is not always serializable because
17495     the additional locking could be a performance killer, and
17496     md should be large enough anyway.
17497
17498   *Bodo Moeller*
17499
17500 * New file `apps/app_rand.c` with commonly needed functionality
17501   for handling the random seed file.
17502
17503   Use the random seed file in some applications that previously did not:
17504           ca,
17505           dsaparam -genkey (which also ignored its '-rand' option),
17506           s_client,
17507           s_server,
17508           x509 (when signing).
17509   Except on systems with /dev/urandom, it is crucial to have a random
17510   seed file at least for key creation, DSA signing, and for DH exchanges;
17511   for RSA signatures we could do without one.
17512
17513   gendh and gendsa (unlike genrsa) used to read only the first byte
17514   of each file listed in the '-rand' option.  The function as previously
17515   found in genrsa is now in app_rand.c and is used by all programs
17516   that support '-rand'.
17517
17518   *Bodo Moeller*
17519
17520 * In RAND_write_file, use mode 0600 for creating files;
17521   don't just chmod when it may be too late.
17522
17523   *Bodo Moeller*
17524
17525 * Report an error from X509_STORE_load_locations
17526   when X509_LOOKUP_load_file or X509_LOOKUP_add_dir failed.
17527
17528   *Bill Perry*
17529
17530 * New function ASN1_mbstring_copy() this copies a string in either
17531   ASCII, Unicode, Universal (4 bytes per character) or UTF8 format
17532   into an ASN1_STRING type. A mask of permissible types is passed
17533   and it chooses the "minimal" type to use or an error if not type
17534   is suitable.
17535
17536   *Steve Henson*
17537
17538 * Add function equivalents to the various macros in asn1.h. The old
17539   macros are retained with an `M_` prefix. Code inside the library can
17540   use the `M_` macros. External code (including the openssl utility)
17541   should *NOT* in order to be "shared library friendly".
17542
17543   *Steve Henson*
17544
17545 * Add various functions that can check a certificate's extensions
17546   to see if it usable for various purposes such as SSL client,
17547   server or S/MIME and CAs of these types. This is currently
17548   VERY EXPERIMENTAL but will ultimately be used for certificate chain
17549   verification. Also added a -purpose flag to x509 utility to
17550   print out all the purposes.
17551
17552   *Steve Henson*
17553
17554 * Add a CRYPTO_EX_DATA to X509 certificate structure and associated
17555   functions.
17556
17557   *Steve Henson*
17558
17559 * New `X509V3_{X509,CRL,REVOKED}_get_d2i()` functions. These will search
17560   for, obtain and decode and extension and obtain its critical flag.
17561   This allows all the necessary extension code to be handled in a
17562   single function call.
17563
17564   *Steve Henson*
17565
17566 * RC4 tune-up featuring 30-40% performance improvement on most RISC
17567   platforms. See crypto/rc4/rc4_enc.c for further details.
17568
17569   *Andy Polyakov*
17570
17571 * New -noout option to asn1parse. This causes no output to be produced
17572   its main use is when combined with -strparse and -out to extract data
17573   from a file (which may not be in ASN.1 format).
17574
17575   *Steve Henson*
17576
17577 * Fix for pkcs12 program. It was hashing an invalid certificate pointer
17578   when producing the local key id.
17579
17580   *Richard Levitte <levitte@stacken.kth.se>*
17581
17582 * New option -dhparam in s_server. This allows a DH parameter file to be
17583   stated explicitly. If it is not stated then it tries the first server
17584   certificate file. The previous behaviour hard coded the filename
17585   "server.pem".
17586
17587   *Steve Henson*
17588
17589 * Add -pubin and -pubout options to the rsa and dsa commands. These allow
17590   a public key to be input or output. For example:
17591   openssl rsa -in key.pem -pubout -out pubkey.pem
17592   Also added necessary DSA public key functions to handle this.
17593
17594   *Steve Henson*
17595
17596 * Fix so PKCS7_dataVerify() doesn't crash if no certificates are contained
17597   in the message. This was handled by allowing
17598   X509_find_by_issuer_and_serial() to tolerate a NULL passed to it.
17599
17600   *Steve Henson, reported by Sampo Kellomaki <sampo@mail.neuronio.pt>*
17601
17602 * Fix for bug in d2i_ASN1_bytes(): other ASN1 functions add an extra null
17603   to the end of the strings whereas this didn't. This would cause problems
17604   if strings read with d2i_ASN1_bytes() were later modified.
17605
17606   *Steve Henson, reported by Arne Ansper <arne@ats.cyber.ee>*
17607
17608 * Fix for base64 decode bug. When a base64 bio reads only one line of
17609   data and it contains EOF it will end up returning an error. This is
17610   caused by input 46 bytes long. The cause is due to the way base64
17611   BIOs find the start of base64 encoded data. They do this by trying a
17612   trial decode on each line until they find one that works. When they
17613   do a flag is set and it starts again knowing it can pass all the
17614   data directly through the decoder. Unfortunately it doesn't reset
17615   the context it uses. This means that if EOF is reached an attempt
17616   is made to pass two EOFs through the context and this causes the
17617   resulting error. This can also cause other problems as well. As is
17618   usual with these problems it takes *ages* to find and the fix is
17619   trivial: move one line.
17620
17621   *Steve Henson, reported by ian@uns.ns.ac.yu (Ivan Nejgebauer)*
17622
17623 * Ugly workaround to get s_client and s_server working under Windows. The
17624   old code wouldn't work because it needed to select() on sockets and the
17625   tty (for keypresses and to see if data could be written). Win32 only
17626   supports select() on sockets so we select() with a 1s timeout on the
17627   sockets and then see if any characters are waiting to be read, if none
17628   are present then we retry, we also assume we can always write data to
17629   the tty. This isn't nice because the code then blocks until we've
17630   received a complete line of data and it is effectively polling the
17631   keyboard at 1s intervals: however it's quite a bit better than not
17632   working at all :-) A dedicated Windows application might handle this
17633   with an event loop for example.
17634
17635   *Steve Henson*
17636
17637 * Enhance RSA_METHOD structure. Now there are two extra methods, rsa_sign
17638   and rsa_verify. When the RSA_FLAGS_SIGN_VER option is set these functions
17639   will be called when RSA_sign() and RSA_verify() are used. This is useful
17640   if rsa_pub_dec() and rsa_priv_enc() equivalents are not available.
17641   For this to work properly RSA_public_decrypt() and RSA_private_encrypt()
17642   should *not* be used: RSA_sign() and RSA_verify() must be used instead.
17643   This necessitated the support of an extra signature type NID_md5_sha1
17644   for SSL signatures and modifications to the SSL library to use it instead
17645   of calling RSA_public_decrypt() and RSA_private_encrypt().
17646
17647   *Steve Henson*
17648
17649 * Add new -verify -CAfile and -CApath options to the crl program, these
17650   will lookup a CRL issuers certificate and verify the signature in a
17651   similar way to the verify program. Tidy up the crl program so it
17652   no longer accesses structures directly. Make the ASN1 CRL parsing a bit
17653   less strict. It will now permit CRL extensions even if it is not
17654   a V2 CRL: this will allow it to tolerate some broken CRLs.
17655
17656   *Steve Henson*
17657
17658 * Initialize all non-automatic variables each time one of the openssl
17659   sub-programs is started (this is necessary as they may be started
17660   multiple times from the "OpenSSL>" prompt).
17661
17662   *Lennart Bang, Bodo Moeller*
17663
17664 * Preliminary compilation option RSA_NULL which disables RSA crypto without
17665   removing all other RSA functionality (this is what NO_RSA does). This
17666   is so (for example) those in the US can disable those operations covered
17667   by the RSA patent while allowing storage and parsing of RSA keys and RSA
17668   key generation.
17669
17670   *Steve Henson*
17671
17672 * Non-copying interface to BIO pairs.
17673   (still largely untested)
17674
17675   *Bodo Moeller*
17676
17677 * New function ASN1_tag2str() to convert an ASN1 tag to a descriptive
17678   ASCII string. This was handled independently in various places before.
17679
17680   *Steve Henson*
17681
17682 * New functions UTF8_getc() and UTF8_putc() that parse and generate
17683   UTF8 strings a character at a time.
17684
17685   *Steve Henson*
17686
17687 * Use client_version from client hello to select the protocol
17688   (s23_srvr.c) and for RSA client key exchange verification
17689   (s3_srvr.c), as required by the SSL 3.0/TLS 1.0 specifications.
17690
17691   *Bodo Moeller*
17692
17693 * Add various utility functions to handle SPKACs, these were previously
17694   handled by poking round in the structure internals. Added new function
17695   NETSCAPE_SPKI_print() to print out SPKAC and a new utility 'spkac' to
17696   print, verify and generate SPKACs. Based on an original idea from
17697   Massimiliano Pala <madwolf@comune.modena.it> but extensively modified.
17698
17699   *Steve Henson*
17700
17701 * RIPEMD160 is operational on all platforms and is back in 'make test'.
17702
17703   *Andy Polyakov*
17704
17705 * Allow the config file extension section to be overwritten on the
17706   command line. Based on an original idea from Massimiliano Pala
17707   <madwolf@comune.modena.it>. The new option is called -extensions
17708   and can be applied to ca, req and x509. Also -reqexts to override
17709   the request extensions in req and -crlexts to override the crl extensions
17710   in ca.
17711
17712   *Steve Henson*
17713
17714 * Add new feature to the SPKAC handling in ca.  Now you can include
17715   the same field multiple times by preceding it by "XXXX." for example:
17716   1.OU="Unit name 1"
17717   2.OU="Unit name 2"
17718   this is the same syntax as used in the req config file.
17719
17720   *Steve Henson*
17721
17722 * Allow certificate extensions to be added to certificate requests. These
17723   are specified in a 'req_extensions' option of the req section of the
17724   config file. They can be printed out with the -text option to req but
17725   are otherwise ignored at present.
17726
17727   *Steve Henson*
17728
17729 * Fix a horrible bug in enc_read() in crypto/evp/bio_enc.c: if the first
17730   data read consists of only the final block it would not decrypted because
17731   EVP_CipherUpdate() would correctly report zero bytes had been decrypted.
17732   A misplaced 'break' also meant the decrypted final block might not be
17733   copied until the next read.
17734
17735   *Steve Henson*
17736
17737 * Initial support for DH_METHOD. Again based on RSA_METHOD. Also added
17738   a few extra parameters to the DH structure: these will be useful if
17739   for example we want the value of 'q' or implement X9.42 DH.
17740
17741   *Steve Henson*
17742
17743 * Initial support for DSA_METHOD. This is based on the RSA_METHOD and
17744   provides hooks that allow the default DSA functions or functions on a
17745   "per key" basis to be replaced. This allows hardware acceleration and
17746   hardware key storage to be handled without major modification to the
17747   library. Also added low-level modexp hooks and CRYPTO_EX structure and
17748   associated functions.
17749
17750   *Steve Henson*
17751
17752 * Add a new flag to memory BIOs, BIO_FLAG_MEM_RDONLY. This marks the BIO
17753   as "read only": it can't be written to and the buffer it points to will
17754   not be freed. Reading from a read only BIO is much more efficient than
17755   a normal memory BIO. This was added because there are several times when
17756   an area of memory needs to be read from a BIO. The previous method was
17757   to create a memory BIO and write the data to it, this results in two
17758   copies of the data and an O(n^2) reading algorithm. There is a new
17759   function BIO_new_mem_buf() which creates a read only memory BIO from
17760   an area of memory. Also modified the PKCS#7 routines to use read only
17761   memory BIOs.
17762
17763   *Steve Henson*
17764
17765 * Bugfix: ssl23_get_client_hello did not work properly when called in
17766   state SSL23_ST_SR_CLNT_HELLO_B, i.e. when the first 7 bytes of
17767   a SSLv2-compatible client hello for SSLv3 or TLSv1 could be read,
17768   but a retry condition occurred while trying to read the rest.
17769
17770   *Bodo Moeller*
17771
17772 * The PKCS7_ENC_CONTENT_new() function was setting the content type as
17773   NID_pkcs7_encrypted by default: this was wrong since this should almost
17774   always be NID_pkcs7_data. Also modified the PKCS7_set_type() to handle
17775   the encrypted data type: this is a more sensible place to put it and it
17776   allows the PKCS#12 code to be tidied up that duplicated this
17777   functionality.
17778
17779   *Steve Henson*
17780
17781 * Changed obj_dat.pl script so it takes its input and output files on
17782   the command line. This should avoid shell escape redirection problems
17783   under Win32.
17784
17785   *Steve Henson*
17786
17787 * Initial support for certificate extension requests, these are included
17788   in things like Xenroll certificate requests. Included functions to allow
17789   extensions to be obtained and added.
17790
17791   *Steve Henson*
17792
17793 * -crlf option to s_client and s_server for sending newlines as
17794   CRLF (as required by many protocols).
17795
17796   *Bodo Moeller*
17797
17798### Changes between 0.9.3a and 0.9.4  [09 Aug 1999]
17799
17800 * Install libRSAglue.a when OpenSSL is built with RSAref.
17801
17802   *Ralf S. Engelschall*
17803
17804 * A few more `#ifndef NO_FP_API / #endif` pairs for consistency.
17805
17806   *Andrija Antonijevic <TheAntony2@bigfoot.com>*
17807
17808 * Fix -startdate and -enddate (which was missing) arguments to 'ca'
17809   program.
17810
17811   *Steve Henson*
17812
17813 * New function DSA_dup_DH, which duplicates DSA parameters/keys as
17814   DH parameters/keys (q is lost during that conversion, but the resulting
17815   DH parameters contain its length).
17816
17817   For 1024-bit p, DSA_generate_parameters followed by DSA_dup_DH is
17818   much faster than DH_generate_parameters (which creates parameters
17819   where `p = 2*q + 1`), and also the smaller q makes DH computations
17820   much more efficient (160-bit exponentiation instead of 1024-bit
17821   exponentiation); so this provides a convenient way to support DHE
17822   ciphersuites in SSL/TLS servers (see ssl/ssltest.c).  It is of
17823   utter importance to use
17824           SSL_CTX_set_options(s_ctx, SSL_OP_SINGLE_DH_USE);
17825   or
17826           SSL_set_options(s_ctx, SSL_OP_SINGLE_DH_USE);
17827   when such DH parameters are used, because otherwise small subgroup
17828   attacks may become possible!
17829
17830   *Bodo Moeller*
17831
17832 * Avoid memory leak in i2d_DHparams.
17833
17834   *Bodo Moeller*
17835
17836 * Allow the -k option to be used more than once in the enc program:
17837   this allows the same encrypted message to be read by multiple recipients.
17838
17839   *Steve Henson*
17840
17841 * New function OBJ_obj2txt(buf, buf_len, a, no_name), this converts
17842   an ASN1_OBJECT to a text string. If the "no_name" parameter is set then
17843   it will always use the numerical form of the OID, even if it has a short
17844   or long name.
17845
17846   *Steve Henson*
17847
17848 * Added an extra RSA flag: RSA_FLAG_EXT_PKEY. Previously the rsa_mod_exp
17849   method only got called if p,q,dmp1,dmq1,iqmp components were present,
17850   otherwise bn_mod_exp was called. In the case of hardware keys for example
17851   no private key components need be present and it might store extra data
17852   in the RSA structure, which cannot be accessed from bn_mod_exp.
17853   By setting RSA_FLAG_EXT_PKEY rsa_mod_exp will always be called for
17854   private key operations.
17855
17856   *Steve Henson*
17857
17858 * Added support for SPARC Linux.
17859
17860   *Andy Polyakov*
17861
17862 * pem_password_cb function type incompatibly changed from
17863           typedef int pem_password_cb(char *buf, int size, int rwflag);
17864   to
17865           ....(char *buf, int size, int rwflag, void *userdata);
17866   so that applications can pass data to their callbacks:
17867   The `PEM[_ASN1]_{read,write}...` functions and macros now take an
17868   additional void * argument, which is just handed through whenever
17869   the password callback is called.
17870
17871   *Damien Miller <dmiller@ilogic.com.au>; tiny changes by Bodo Moeller*
17872
17873   New function SSL_CTX_set_default_passwd_cb_userdata.
17874
17875   Compatibility note: As many C implementations push function arguments
17876   onto the stack in reverse order, the new library version is likely to
17877   interoperate with programs that have been compiled with the old
17878   pem_password_cb definition (PEM_whatever takes some data that
17879   happens to be on the stack as its last argument, and the callback
17880   just ignores this garbage); but there is no guarantee whatsoever that
17881   this will work.
17882
17883 * The -DPLATFORM="\"$(PLATFORM)\"" definition and the similar -DCFLAGS=...
17884   (both in crypto/Makefile.ssl for use by crypto/cversion.c) caused
17885   problems not only on Windows, but also on some Unix platforms.
17886   To avoid problematic command lines, these definitions are now in an
17887   auto-generated file crypto/buildinf.h (created by crypto/Makefile.ssl
17888   for standard "make" builds, by util/mk1mf.pl for "mk1mf" builds).
17889
17890   *Bodo Moeller*
17891
17892 * MIPS III/IV assembler module is reimplemented.
17893
17894   *Andy Polyakov*
17895
17896 * More DES library cleanups: remove references to srand/rand and
17897   delete an unused file.
17898
17899   *Ulf Möller*
17900
17901 * Add support for the free Netwide assembler (NASM) under Win32,
17902   since not many people have MASM (ml) and it can be hard to obtain.
17903   This is currently experimental but it seems to work OK and pass all
17904   the tests. Check out INSTALL.W32 for info.
17905
17906   *Steve Henson*
17907
17908 * Fix memory leaks in s3_clnt.c: All non-anonymous SSL3/TLS1 connections
17909   without temporary keys kept an extra copy of the server key,
17910   and connections with temporary keys did not free everything in case
17911   of an error.
17912
17913   *Bodo Moeller*
17914
17915 * New function RSA_check_key and new openssl rsa option -check
17916   for verifying the consistency of RSA keys.
17917
17918   *Ulf Moeller, Bodo Moeller*
17919
17920 * Various changes to make Win32 compile work:
17921   1. Casts to avoid "loss of data" warnings in p5_crpt2.c
17922   2. Change unsigned int to int in b_dump.c to avoid "signed/unsigned
17923      comparison" warnings.
17924   3. Add `sk_<TYPE>_sort` to DEF file generator and do make update.
17925
17926   *Steve Henson*
17927
17928 * Add a debugging option to PKCS#5 v2 key generation function: when
17929   you #define DEBUG_PKCS5V2 passwords, salts, iteration counts and
17930   derived keys are printed to stderr.
17931
17932   *Steve Henson*
17933
17934 * Copy the flags in ASN1_STRING_dup().
17935
17936   *Roman E. Pavlov <pre@mo.msk.ru>*
17937
17938 * The x509 application mishandled signing requests containing DSA
17939   keys when the signing key was also DSA and the parameters didn't match.
17940
17941   It was supposed to omit the parameters when they matched the signing key:
17942   the verifying software was then supposed to automatically use the CA's
17943   parameters if they were absent from the end user certificate.
17944
17945   Omitting parameters is no longer recommended. The test was also
17946   the wrong way round! This was probably due to unusual behaviour in
17947   EVP_cmp_parameters() which returns 1 if the parameters match.
17948   This meant that parameters were omitted when they *didn't* match and
17949   the certificate was useless. Certificates signed with 'ca' didn't have
17950   this bug.
17951
17952   *Steve Henson, reported by Doug Erickson <Doug.Erickson@Part.NET>*
17953
17954 * Memory leak checking (-DCRYPTO_MDEBUG) had some problems.
17955   The interface is as follows:
17956   Applications can use
17957           CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_ON) aka MemCheck_start(),
17958           CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_OFF) aka MemCheck_stop();
17959   "off" is now the default.
17960   The library internally uses
17961           CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_DISABLE) aka MemCheck_off(),
17962           CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_ENABLE) aka MemCheck_on()
17963   to disable memory-checking temporarily.
17964
17965   Some inconsistent states that previously were possible (and were
17966   even the default) are now avoided.
17967
17968   -DCRYPTO_MDEBUG_TIME is new and additionally stores the current time
17969   with each memory chunk allocated; this is occasionally more helpful
17970   than just having a counter.
17971
17972   -DCRYPTO_MDEBUG_THREAD is also new and adds the thread ID.
17973
17974   -DCRYPTO_MDEBUG_ALL enables all of the above, plus any future
17975   extensions.
17976
17977   *Bodo Moeller*
17978
17979 * Introduce "mode" for SSL structures (with defaults in SSL_CTX),
17980   which largely parallels "options", but is for changing API behaviour,
17981   whereas "options" are about protocol behaviour.
17982   Initial "mode" flags are:
17983
17984   SSL_MODE_ENABLE_PARTIAL_WRITE   Allow SSL_write to report success when
17985                                   a single record has been written.
17986   SSL_MODE_ACCEPT_MOVING_WRITE_BUFFER  Don't insist that SSL_write
17987                                   retries use the same buffer location.
17988                                   (But all of the contents must be
17989                                   copied!)
17990
17991   *Bodo Moeller*
17992
17993 * Bugfix: SSL_set_options ignored its parameter, only SSL_CTX_set_options
17994   worked.
17995
17996 * Fix problems with no-hmac etc.
17997
17998   *Ulf Möller, pointed out by Brian Wellington <bwelling@tislabs.com>*
17999
18000 * New functions RSA_get_default_method(), RSA_set_method() and
18001   RSA_get_method(). These allows replacement of RSA_METHODs without having
18002   to mess around with the internals of an RSA structure.
18003
18004   *Steve Henson*
18005
18006 * Fix memory leaks in DSA_do_sign and DSA_is_prime.
18007   Also really enable memory leak checks in openssl.c and in some
18008   test programs.
18009
18010   *Chad C. Mulligan, Bodo Moeller*
18011
18012 * Fix a bug in d2i_ASN1_INTEGER() and i2d_ASN1_INTEGER() which can mess
18013   up the length of negative integers. This has now been simplified to just
18014   store the length when it is first determined and use it later, rather
18015   than trying to keep track of where data is copied and updating it to
18016   point to the end.
18017   *Steve Henson, reported by Brien Wheeler <bwheeler@authentica-security.com>*
18018
18019 * Add a new function PKCS7_signatureVerify. This allows the verification
18020   of a PKCS#7 signature but with the signing certificate passed to the
18021   function itself. This contrasts with PKCS7_dataVerify which assumes the
18022   certificate is present in the PKCS#7 structure. This isn't always the
18023   case: certificates can be omitted from a PKCS#7 structure and be
18024   distributed by "out of band" means (such as a certificate database).
18025
18026   *Steve Henson*
18027
18028 * Complete the `PEM_*` macros with DECLARE_PEM versions to replace the
18029   function prototypes in pem.h, also change util/mkdef.pl to add the
18030   necessary function names.
18031
18032   *Steve Henson*
18033
18034 * mk1mf.pl (used by Windows builds) did not properly read the
18035   options set by Configure in the top level Makefile, and Configure
18036   was not even able to write more than one option correctly.
18037   Fixed, now "no-idea no-rc5 -DCRYPTO_MDEBUG" etc. works as intended.
18038
18039   *Bodo Moeller*
18040
18041 * New functions CONF_load_bio() and CONF_load_fp() to allow a config
18042   file to be loaded from a BIO or FILE pointer. The BIO version will
18043   for example allow memory BIOs to contain config info.
18044
18045   *Steve Henson*
18046
18047 * New function "CRYPTO_num_locks" that returns CRYPTO_NUM_LOCKS.
18048   Whoever hopes to achieve shared-library compatibility across versions
18049   must use this, not the compile-time macro.
18050   (Exercise 0.9.4: Which is the minimum library version required by
18051   such programs?)
18052   Note: All this applies only to multi-threaded programs, others don't
18053   need locks.
18054
18055   *Bodo Moeller*
18056
18057 * Add missing case to s3_clnt.c state machine -- one of the new SSL tests
18058   through a BIO pair triggered the default case, i.e.
18059   SSLerr(...,SSL_R_UNKNOWN_STATE).
18060
18061   *Bodo Moeller*
18062
18063 * New "BIO pair" concept (crypto/bio/bss_bio.c) so that applications
18064   can use the SSL library even if none of the specific BIOs is
18065   appropriate.
18066
18067   *Bodo Moeller*
18068
18069 * Fix a bug in i2d_DSAPublicKey() which meant it returned the wrong value
18070   for the encoded length.
18071
18072   *Jeon KyoungHo <khjeon@sds.samsung.co.kr>*
18073
18074 * Add initial documentation of the X509V3 functions.
18075
18076   *Steve Henson*
18077
18078 * Add a new pair of functions PEM_write_PKCS8PrivateKey() and
18079   PEM_write_bio_PKCS8PrivateKey() that are equivalent to
18080   PEM_write_PrivateKey() and PEM_write_bio_PrivateKey() but use the more
18081   secure PKCS#8 private key format with a high iteration count.
18082
18083   *Steve Henson*
18084
18085 * Fix determination of Perl interpreter: A perl or perl5
18086   *directory* in $PATH was also accepted as the interpreter.
18087
18088   *Ralf S. Engelschall*
18089
18090 * Fix demos/sign/sign.c: well there wasn't anything strictly speaking
18091   wrong with it but it was very old and did things like calling
18092   PEM_ASN1_read() directly and used MD5 for the hash not to mention some
18093   unusual formatting.
18094
18095   *Steve Henson*
18096
18097 * Fix demos/selfsign.c: it used obsolete and deleted functions, changed
18098   to use the new extension code.
18099
18100   *Steve Henson*
18101
18102 * Implement the PEM_read/PEM_write functions in crypto/pem/pem_all.c
18103   with macros. This should make it easier to change their form, add extra
18104   arguments etc. Fix a few PEM prototypes which didn't have cipher as a
18105   constant.
18106
18107   *Steve Henson*
18108
18109 * Add to configuration table a new entry that can specify an alternative
18110   name for unistd.h (for pre-POSIX systems); we need this for NeXTstep,
18111   according to Mark Crispin <MRC@Panda.COM>.
18112
18113   *Bodo Moeller*
18114
18115 * DES CBC did not update the IV. Weird.
18116
18117   *Ben Laurie*
18118lse
18119   des_cbc_encrypt does not update the IV, but des_ncbc_encrypt does.
18120   Changing the behaviour of the former might break existing programs --
18121   where IV updating is needed, des_ncbc_encrypt can be used.
18122ndif
18123
18124 * When bntest is run from "make test" it drives bc to check its
18125   calculations, as well as internally checking them. If an internal check
18126   fails, it needs to cause bc to give a non-zero result or make test carries
18127   on without noticing the failure. Fixed.
18128
18129   *Ben Laurie*
18130
18131 * DES library cleanups.
18132
18133   *Ulf Möller*
18134
18135 * Add support for PKCS#5 v2.0 PBE algorithms. This will permit PKCS#8 to be
18136   used with any cipher unlike PKCS#5 v1.5 which can at most handle 64 bit
18137   ciphers. NOTE: although the key derivation function has been verified
18138   against some published test vectors it has not been extensively tested
18139   yet. Added a -v2 "cipher" option to pkcs8 application to allow the use
18140   of v2.0.
18141
18142   *Steve Henson*
18143
18144 * Instead of "mkdir -p", which is not fully portable, use new
18145   Perl script "util/mkdir-p.pl".
18146
18147   *Bodo Moeller*
18148
18149 * Rewrite the way password based encryption (PBE) is handled. It used to
18150   assume that the ASN1 AlgorithmIdentifier parameter was a PBEParameter
18151   structure. This was true for the PKCS#5 v1.5 and PKCS#12 PBE algorithms
18152   but doesn't apply to PKCS#5 v2.0 where it can be something else. Now
18153   the 'parameter' field of the AlgorithmIdentifier is passed to the
18154   underlying key generation function so it must do its own ASN1 parsing.
18155   This has also changed the EVP_PBE_CipherInit() function which now has a
18156   'parameter' argument instead of literal salt and iteration count values
18157   and the function EVP_PBE_ALGOR_CipherInit() has been deleted.
18158
18159   *Steve Henson*
18160
18161 * Support for PKCS#5 v1.5 compatible password based encryption algorithms
18162   and PKCS#8 functionality. New 'pkcs8' application linked to openssl.
18163   Needed to change the PEM_STRING_EVP_PKEY value which was just "PRIVATE
18164   KEY" because this clashed with PKCS#8 unencrypted string. Since this
18165   value was just used as a "magic string" and not used directly its
18166   value doesn't matter.
18167
18168   *Steve Henson*
18169
18170 * Introduce some semblance of const correctness to BN. Shame C doesn't
18171   support mutable.
18172
18173   *Ben Laurie*
18174
18175 * "linux-sparc64" configuration (ultrapenguin).
18176
18177   *Ray Miller <ray.miller@oucs.ox.ac.uk>*
18178   "linux-sparc" configuration.
18179
18180   *Christian Forster <fo@hawo.stw.uni-erlangen.de>*
18181
18182 * config now generates no-xxx options for missing ciphers.
18183
18184   *Ulf Möller*
18185
18186 * Support the EBCDIC character set (work in progress).
18187   File ebcdic.c not yet included because it has a different license.
18188
18189   *Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>*
18190
18191 * Support BS2000/OSD-POSIX.
18192
18193   *Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>*
18194
18195 * Make callbacks for key generation use `void *` instead of `char *`.
18196
18197   *Ben Laurie*
18198
18199 * Make S/MIME samples compile (not yet tested).
18200
18201   *Ben Laurie*
18202
18203 * Additional typesafe stacks.
18204
18205   *Ben Laurie*
18206
18207 * New configuration variants "bsdi-elf-gcc" (BSD/OS 4.x).
18208
18209   *Bodo Moeller*
18210
18211### Changes between 0.9.3 and 0.9.3a  [29 May 1999]
18212
18213 * New configuration variant "sco5-gcc".
18214
18215 * Updated some demos.
18216
18217   *Sean O Riordain, Wade Scholine*
18218
18219 * Add missing BIO_free at exit of pkcs12 application.
18220
18221   *Wu Zhigang*
18222
18223 * Fix memory leak in conf.c.
18224
18225   *Steve Henson*
18226
18227 * Updates for Win32 to assembler version of MD5.
18228
18229   *Steve Henson*
18230
18231 * Set #! path to perl in `apps/der_chop` to where we found it
18232   instead of using a fixed path.
18233
18234   *Bodo Moeller*
18235
18236 * SHA library changes for irix64-mips4-cc.
18237
18238   *Andy Polyakov*
18239
18240 * Improvements for VMS support.
18241
18242   *Richard Levitte*
18243
18244### Changes between 0.9.2b and 0.9.3  [24 May 1999]
18245
18246 * Bignum library bug fix. IRIX 6 passes "make test" now!
18247   This also avoids the problems with SC4.2 and unpatched SC5.
18248
18249   *Andy Polyakov <appro@fy.chalmers.se>*
18250
18251 * New functions sk_num, sk_value and sk_set to replace the previous macros.
18252   These are required because of the typesafe stack would otherwise break
18253   existing code. If old code used a structure member which used to be STACK
18254   and is now STACK_OF (for example cert in a PKCS7_SIGNED structure) with
18255   sk_num or sk_value it would produce an error because the num, data members
18256   are not present in STACK_OF. Now it just produces a warning. sk_set
18257   replaces the old method of assigning a value to sk_value
18258   (e.g. sk_value(x, i) = y) which the library used in a few cases. Any code
18259   that does this will no longer work (and should use sk_set instead) but
18260   this could be regarded as a "questionable" behaviour anyway.
18261
18262   *Steve Henson*
18263
18264 * Fix most of the other PKCS#7 bugs. The "experimental" code can now
18265   correctly handle encrypted S/MIME data.
18266
18267   *Steve Henson*
18268
18269 * Change type of various DES function arguments from des_cblock
18270   (which means, in function argument declarations, pointer to char)
18271   to des_cblock * (meaning pointer to array with 8 char elements),
18272   which allows the compiler to do more typechecking; it was like
18273   that back in SSLeay, but with lots of ugly casts.
18274
18275   Introduce new type const_des_cblock.
18276
18277   *Bodo Moeller*
18278
18279 * Reorganise the PKCS#7 library and get rid of some of the more obvious
18280   problems: find RecipientInfo structure that matches recipient certificate
18281   and initialise the ASN1 structures properly based on passed cipher.
18282
18283   *Steve Henson*
18284
18285 * Belatedly make the BN tests actually check the results.
18286
18287   *Ben Laurie*
18288
18289 * Fix the encoding and decoding of negative ASN1 INTEGERS and conversion
18290   to and from BNs: it was completely broken. New compilation option
18291   NEG_PUBKEY_BUG to allow for some broken certificates that encode public
18292   key elements as negative integers.
18293
18294   *Steve Henson*
18295
18296 * Reorganize and speed up MD5.
18297
18298   *Andy Polyakov <appro@fy.chalmers.se>*
18299
18300 * VMS support.
18301
18302   *Richard Levitte <richard@levitte.org>*
18303
18304 * New option -out to asn1parse to allow the parsed structure to be
18305   output to a file. This is most useful when combined with the -strparse
18306   option to examine the output of things like OCTET STRINGS.
18307
18308   *Steve Henson*
18309
18310 * Make SSL library a little more fool-proof by not requiring any longer
18311   that `SSL_set_{accept,connect}_state` be called before
18312   `SSL_{accept,connect}` may be used (`SSL_set_..._state` is omitted
18313   in many applications because usually everything *appeared* to work as
18314   intended anyway -- now it really works as intended).
18315
18316   *Bodo Moeller*
18317
18318 * Move openssl.cnf out of lib/.
18319
18320   *Ulf Möller*
18321
18322 * Fix various things to let OpenSSL even pass "egcc -pipe -O2 -Wall
18323   -Wshadow -Wpointer-arith -Wcast-align -Wmissing-prototypes
18324   -Wmissing-declarations -Wnested-externs -Winline" with EGCS 1.1.2+
18325
18326   *Ralf S. Engelschall*
18327
18328 * Various fixes to the EVP and PKCS#7 code. It may now be able to
18329   handle PKCS#7 enveloped data properly.
18330
18331   *Sebastian Akerman <sak@parallelconsulting.com>, modified by Steve*
18332
18333 * Create a duplicate of the SSL_CTX's CERT in SSL_new instead of
18334   copying pointers.  The cert_st handling is changed by this in
18335   various ways (and thus what used to be known as ctx->default_cert
18336   is now called ctx->cert, since we don't resort to `s->ctx->[default_]cert`
18337   any longer when s->cert does not give us what we need).
18338   ssl_cert_instantiate becomes obsolete by this change.
18339   As soon as we've got the new code right (possibly it already is?),
18340   we have solved a couple of bugs of the earlier code where s->cert
18341   was used as if it could not have been shared with other SSL structures.
18342
18343   Note that using the SSL API in certain dirty ways now will result
18344   in different behaviour than observed with earlier library versions:
18345   Changing settings for an `SSL_CTX *ctx` after having done s = SSL_new(ctx)
18346   does not influence s as it used to.
18347
18348   In order to clean up things more thoroughly, inside SSL_SESSION
18349   we don't use CERT any longer, but a new structure SESS_CERT
18350   that holds per-session data (if available); currently, this is
18351   the peer's certificate chain and, for clients, the server's certificate
18352   and temporary key.  CERT holds only those values that can have
18353   meaningful defaults in an SSL_CTX.
18354
18355   *Bodo Moeller*
18356
18357 * New function X509V3_EXT_i2d() to create an X509_EXTENSION structure
18358   from the internal representation. Various PKCS#7 fixes: remove some
18359   evil casts and set the enc_dig_alg field properly based on the signing
18360   key type.
18361
18362   *Steve Henson*
18363
18364 * Allow PKCS#12 password to be set from the command line or the
18365   environment. Let 'ca' get its config file name from the environment
18366   variables "OPENSSL_CONF" or "SSLEAY_CONF" (for consistency with 'req'
18367   and 'x509').
18368
18369   *Steve Henson*
18370
18371 * Allow certificate policies extension to use an IA5STRING for the
18372   organization field. This is contrary to the PKIX definition but
18373   VeriSign uses it and IE5 only recognises this form. Document 'x509'
18374   extension option.
18375
18376   *Steve Henson*
18377
18378 * Add PEDANTIC compiler flag to allow compilation with gcc -pedantic,
18379   without disallowing inline assembler and the like for non-pedantic builds.
18380
18381   *Ben Laurie*
18382
18383 * Support Borland C++ builder.
18384
18385   *Janez Jere <jj@void.si>, modified by Ulf Möller*
18386
18387 * Support Mingw32.
18388
18389   *Ulf Möller*
18390
18391 * SHA-1 cleanups and performance enhancements.
18392
18393   *Andy Polyakov <appro@fy.chalmers.se>*
18394
18395 * Sparc v8plus assembler for the bignum library.
18396
18397   *Andy Polyakov <appro@fy.chalmers.se>*
18398
18399 * Accept any -xxx and +xxx compiler options in Configure.
18400
18401   *Ulf Möller*
18402
18403 * Update HPUX configuration.
18404
18405   *Anonymous*
18406
18407 * Add missing `sk_<type>_unshift()` function to safestack.h
18408
18409   *Ralf S. Engelschall*
18410
18411 * New function SSL_CTX_use_certificate_chain_file that sets the
18412   "extra_cert"s in addition to the certificate.  (This makes sense
18413   only for "PEM" format files, as chains as a whole are not
18414   DER-encoded.)
18415
18416   *Bodo Moeller*
18417
18418 * Support verify_depth from the SSL API.
18419   x509_vfy.c had what can be considered an off-by-one-error:
18420   Its depth (which was not part of the external interface)
18421   was actually counting the number of certificates in a chain;
18422   now it really counts the depth.
18423
18424   *Bodo Moeller*
18425
18426 * Bugfix in crypto/x509/x509_cmp.c: The SSLerr macro was used
18427   instead of X509err, which often resulted in confusing error
18428   messages since the error codes are not globally unique
18429   (e.g. an alleged error in ssl3_accept when a certificate
18430   didn't match the private key).
18431
18432 * New function SSL_CTX_set_session_id_context that allows to set a default
18433   value (so that you don't need SSL_set_session_id_context for each
18434   connection using the SSL_CTX).
18435
18436   *Bodo Moeller*
18437
18438 * OAEP decoding bug fix.
18439
18440   *Ulf Möller*
18441
18442 * Support INSTALL_PREFIX for package builders, as proposed by
18443   David Harris.
18444
18445   *Bodo Moeller*
18446
18447 * New Configure options "threads" and "no-threads".  For systems
18448   where the proper compiler options are known (currently Solaris
18449   and Linux), "threads" is the default.
18450
18451   *Bodo Moeller*
18452
18453 * New script util/mklink.pl as a faster substitute for util/mklink.sh.
18454
18455   *Bodo Moeller*
18456
18457 * Install various scripts to $(OPENSSLDIR)/misc, not to
18458   $(INSTALLTOP)/bin -- they shouldn't clutter directories
18459   such as /usr/local/bin.
18460
18461   *Bodo Moeller*
18462
18463 * "make linux-shared" to build shared libraries.
18464
18465   *Niels Poppe <niels@netbox.org>*
18466
18467 * New Configure option `no-<cipher>` (rsa, idea, rc5, ...).
18468
18469   *Ulf Möller*
18470
18471 * Add the PKCS#12 API documentation to openssl.txt. Preliminary support for
18472   extension adding in x509 utility.
18473
18474   *Steve Henson*
18475
18476 * Remove NOPROTO sections and error code comments.
18477
18478   *Ulf Möller*
18479
18480 * Partial rewrite of the DEF file generator to now parse the ANSI
18481   prototypes.
18482
18483   *Steve Henson*
18484
18485 * New Configure options --prefix=DIR and --openssldir=DIR.
18486
18487   *Ulf Möller*
18488
18489 * Complete rewrite of the error code script(s). It is all now handled
18490   by one script at the top level which handles error code gathering,
18491   header rewriting and C source file generation. It should be much better
18492   than the old method: it now uses a modified version of Ulf's parser to
18493   read the ANSI prototypes in all header files (thus the old K&R definitions
18494   aren't needed for error creation any more) and do a better job of
18495   translating function codes into names. The old 'ASN1 error code embedded
18496   in a comment' is no longer necessary and it doesn't use .err files which
18497   have now been deleted. Also the error code call doesn't have to appear all
18498   on one line (which resulted in some large lines...).
18499
18500   *Steve Henson*
18501
18502 * Change #include filenames from `<foo.h>` to `<openssl/foo.h>`.
18503
18504   *Bodo Moeller*
18505
18506 * Change behaviour of ssl2_read when facing length-0 packets: Don't return
18507   0 (which usually indicates a closed connection), but continue reading.
18508
18509   *Bodo Moeller*
18510
18511 * Fix some race conditions.
18512
18513   *Bodo Moeller*
18514
18515 * Add support for CRL distribution points extension. Add Certificate
18516   Policies and CRL distribution points documentation.
18517
18518   *Steve Henson*
18519
18520 * Move the autogenerated header file parts to crypto/opensslconf.h.
18521
18522   *Ulf Möller*
18523
18524 * Fix new 56-bit DES export ciphersuites: they were using 7 bytes instead of
18525   8 of keying material. Merlin has also confirmed interop with this fix
18526   between OpenSSL and Baltimore C/SSL 2.0 and J/SSL 2.0.
18527
18528   *Merlin Hughes <merlin@baltimore.ie>*
18529
18530 * Fix lots of warnings.
18531
18532   *Richard Levitte <levitte@stacken.kth.se>*
18533
18534 * In add_cert_dir() in crypto/x509/by_dir.c, break out of the loop if
18535   the directory spec didn't end with a LIST_SEPARATOR_CHAR.
18536
18537   *Richard Levitte <levitte@stacken.kth.se>*
18538
18539 * Fix problems with sizeof(long) == 8.
18540
18541   *Andy Polyakov <appro@fy.chalmers.se>*
18542
18543 * Change functions to ANSI C.
18544
18545   *Ulf Möller*
18546
18547 * Fix typos in error codes.
18548
18549   *Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>, Ulf Möller*
18550
18551 * Remove defunct assembler files from Configure.
18552
18553   *Ulf Möller*
18554
18555 * SPARC v8 assembler BIGNUM implementation.
18556
18557   *Andy Polyakov <appro@fy.chalmers.se>*
18558
18559 * Support for Certificate Policies extension: both print and set.
18560   Various additions to support the r2i method this uses.
18561
18562   *Steve Henson*
18563
18564 * A lot of constification, and fix a bug in X509_NAME_oneline() that could
18565   return a const string when you are expecting an allocated buffer.
18566
18567   *Ben Laurie*
18568
18569 * Add support for ASN1 types UTF8String and VISIBLESTRING, also the CHOICE
18570   types DirectoryString and DisplayText.
18571
18572   *Steve Henson*
18573
18574 * Add code to allow r2i extensions to access the configuration database,
18575   add an LHASH database driver and add several ctx helper functions.
18576
18577   *Steve Henson*
18578
18579 * Fix an evil bug in bn_expand2() which caused various BN functions to
18580   fail when they extended the size of a BIGNUM.
18581
18582   *Steve Henson*
18583
18584 * Various utility functions to handle SXNet extension. Modify mkdef.pl to
18585   support typesafe stack.
18586
18587   *Steve Henson*
18588
18589 * Fix typo in SSL_[gs]et_options().
18590
18591   *Nils Frostberg <nils@medcom.se>*
18592
18593 * Delete various functions and files that belonged to the (now obsolete)
18594   old X509V3 handling code.
18595
18596   *Steve Henson*
18597
18598 * New Configure option "rsaref".
18599
18600   *Ulf Möller*
18601
18602 * Don't auto-generate pem.h.
18603
18604   *Bodo Moeller*
18605
18606 * Introduce type-safe ASN.1 SETs.
18607
18608   *Ben Laurie*
18609
18610 * Convert various additional casted stacks to type-safe STACK_OF() variants.
18611
18612   *Ben Laurie, Ralf S. Engelschall, Steve Henson*
18613
18614 * Introduce type-safe STACKs. This will almost certainly break lots of code
18615   that links with OpenSSL (well at least cause lots of warnings), but fear
18616   not: the conversion is trivial, and it eliminates loads of evil casts. A
18617   few STACKed things have been converted already. Feel free to convert more.
18618   In the fullness of time, I'll do away with the STACK type altogether.
18619
18620   *Ben Laurie*
18621
18622 * Add `openssl ca -revoke <certfile>` facility which revokes a certificate
18623   specified in `<certfile>` by updating the entry in the index.txt file.
18624   This way one no longer has to edit the index.txt file manually for
18625   revoking a certificate. The -revoke option does the gory details now.
18626
18627   *Massimiliano Pala <madwolf@openca.org>, Ralf S. Engelschall*
18628
18629 * Fix `openssl crl -noout -text` combination where `-noout` killed the
18630   `-text` option at all and this way the `-noout -text` combination was
18631   inconsistent in `openssl crl` with the friends in `openssl x509|rsa|dsa`.
18632
18633   *Ralf S. Engelschall*
18634
18635 * Make sure a corresponding plain text error message exists for the
18636   X509_V_ERR_CERT_REVOKED/23 error number which can occur when a
18637   verify callback function determined that a certificate was revoked.
18638
18639   *Ralf S. Engelschall*
18640
18641 * Bugfix: In test/testenc, don't test `openssl <cipher>` for
18642   ciphers that were excluded, e.g. by -DNO_IDEA.  Also, test
18643   all available ciphers including rc5, which was forgotten until now.
18644   In order to let the testing shell script know which algorithms
18645   are available, a new (up to now undocumented) command
18646   `openssl list-cipher-commands` is used.
18647
18648   *Bodo Moeller*
18649
18650 * Bugfix: s_client occasionally would sleep in select() when
18651   it should have checked SSL_pending() first.
18652
18653   *Bodo Moeller*
18654
18655 * New functions DSA_do_sign and DSA_do_verify to provide access to
18656   the raw DSA values prior to ASN.1 encoding.
18657
18658   *Ulf Möller*
18659
18660 * Tweaks to Configure
18661
18662   *Niels Poppe <niels@netbox.org>*
18663
18664 * Add support for PKCS#5 v2.0 ASN1 PBES2 structures. No other support,
18665   yet...
18666
18667   *Steve Henson*
18668
18669 * New variables $(RANLIB) and $(PERL) in the Makefiles.
18670
18671   *Ulf Möller*
18672
18673 * New config option to avoid instructions that are illegal on the 80386.
18674   The default code is faster, but requires at least a 486.
18675
18676   *Ulf Möller*
18677
18678 * Got rid of old SSL2_CLIENT_VERSION (inconsistently used) and
18679   SSL2_SERVER_VERSION (not used at all) macros, which are now the
18680   same as SSL2_VERSION anyway.
18681
18682   *Bodo Moeller*
18683
18684 * New "-showcerts" option for s_client.
18685
18686   *Bodo Moeller*
18687
18688 * Still more PKCS#12 integration. Add pkcs12 application to openssl
18689   application. Various cleanups and fixes.
18690
18691   *Steve Henson*
18692
18693 * More PKCS#12 integration. Add new pkcs12 directory with Makefile.ssl and
18694   modify error routines to work internally. Add error codes and PBE init
18695   to library startup routines.
18696
18697   *Steve Henson*
18698
18699 * Further PKCS#12 integration. Added password based encryption, PKCS#8 and
18700   packing functions to asn1 and evp. Changed function names and error
18701   codes along the way.
18702
18703   *Steve Henson*
18704
18705 * PKCS12 integration: and so it begins... First of several patches to
18706   slowly integrate PKCS#12 functionality into OpenSSL. Add PKCS#12
18707   objects to objects.h
18708
18709   *Steve Henson*
18710
18711 * Add a new 'indent' option to some X509V3 extension code. Initial ASN1
18712   and display support for Thawte strong extranet extension.
18713
18714   *Steve Henson*
18715
18716 * Add LinuxPPC support.
18717
18718   *Jeff Dubrule <igor@pobox.org>*
18719
18720 * Get rid of redundant BN file bn_mulw.c, and rename bn_div64 to
18721   bn_div_words in alpha.s.
18722
18723   *Hannes Reinecke <H.Reinecke@hw.ac.uk> and Ben Laurie*
18724
18725 * Make sure the RSA OAEP test is skipped under -DRSAref because
18726   OAEP isn't supported when OpenSSL is built with RSAref.
18727
18728   *Ulf Moeller <ulf@fitug.de>*
18729
18730 * Move definitions of IS_SET/IS_SEQUENCE inside crypto/asn1/asn1.h
18731   so they no longer are missing under -DNOPROTO.
18732
18733   *Soren S. Jorvang <soren@t.dk>*
18734
18735### Changes between 0.9.1c and 0.9.2b  [22 Mar 1999]
18736
18737 * Make SSL_get_peer_cert_chain() work in servers. Unfortunately, it still
18738   doesn't work when the session is reused. Coming soon!
18739
18740   *Ben Laurie*
18741
18742 * Fix a security hole, that allows sessions to be reused in the wrong
18743   context thus bypassing client cert protection! All software that uses
18744   client certs and session caches in multiple contexts NEEDS PATCHING to
18745   allow session reuse! A fuller solution is in the works.
18746
18747   *Ben Laurie, problem pointed out by Holger Reif, Bodo Moeller (and ???)*
18748
18749 * Some more source tree cleanups (removed obsolete files
18750   crypto/bf/asm/bf586.pl, test/test.txt and crypto/sha/asm/f.s; changed
18751   permission on "config" script to be executable) and a fix for the INSTALL
18752   document.
18753
18754   *Ulf Moeller <ulf@fitug.de>*
18755
18756 * Remove some legacy and erroneous uses of malloc, free instead of
18757   Malloc, Free.
18758
18759   *Lennart Bang <lob@netstream.se>, with minor changes by Steve*
18760
18761 * Make rsa_oaep_test return non-zero on error.
18762
18763   *Ulf Moeller <ulf@fitug.de>*
18764
18765 * Add support for native Solaris shared libraries. Configure
18766   solaris-sparc-sc4-pic, make, then run shlib/solaris-sc4.sh. It'd be nice
18767   if someone would make that last step automatic.
18768
18769   *Matthias Loepfe <Matthias.Loepfe@AdNovum.CH>*
18770
18771 * ctx_size was not built with the right compiler during "make links". Fixed.
18772
18773   *Ben Laurie*
18774
18775 * Change the meaning of 'ALL' in the cipher list. It now means "everything
18776   except NULL ciphers". This means the default cipher list will no longer
18777   enable NULL ciphers. They need to be specifically enabled e.g. with
18778   the string "DEFAULT:eNULL".
18779
18780   *Steve Henson*
18781
18782 * Fix to RSA private encryption routines: if p < q then it would
18783   occasionally produce an invalid result. This will only happen with
18784   externally generated keys because OpenSSL (and SSLeay) ensure p > q.
18785
18786   *Steve Henson*
18787
18788 * Be less restrictive and allow also `perl util/perlpath.pl
18789   /path/to/bin/perl` in addition to `perl util/perlpath.pl /path/to/bin`,
18790   because this way one can also use an interpreter named `perl5` (which is
18791   usually the name of Perl 5.xxx on platforms where an Perl 4.x is still
18792   installed as `perl`).
18793
18794   *Matthias Loepfe <Matthias.Loepfe@adnovum.ch>*
18795
18796 * Let util/clean-depend.pl work also with older Perl 5.00x versions.
18797
18798   *Matthias Loepfe <Matthias.Loepfe@adnovum.ch>*
18799
18800 * Fix Makefile.org so CC,CFLAG etc are passed to 'make links' add
18801   advapi32.lib to Win32 build and change the pem test comparison
18802   to fc.exe (thanks to Ulrich Kroener <kroneru@yahoo.com> for the
18803   suggestion). Fix misplaced ASNI prototypes and declarations in evp.h
18804   and crypto/des/ede_cbcm_enc.c.
18805
18806   *Steve Henson*
18807
18808 * DES quad checksum was broken on big-endian architectures. Fixed.
18809
18810   *Ben Laurie*
18811
18812 * Comment out two functions in bio.h that aren't implemented. Fix up the
18813   Win32 test batch file so it (might) work again. The Win32 test batch file
18814   is horrible: I feel ill....
18815
18816   *Steve Henson*
18817
18818 * Move various #ifdefs around so NO_SYSLOG, NO_DIRENT etc are now selected
18819   in e_os.h. Audit of header files to check ANSI and non ANSI
18820   sections: 10 functions were absent from non ANSI section and not exported
18821   from Windows DLLs. Fixed up libeay.num for new functions.
18822
18823   *Steve Henson*
18824
18825 * Make `openssl version` output lines consistent.
18826
18827   *Ralf S. Engelschall*
18828
18829 * Fix Win32 symbol export lists for BIO functions: Added
18830   BIO_get_ex_new_index, BIO_get_ex_num, BIO_get_ex_data and BIO_set_ex_data
18831   to ms/libeay{16,32}.def.
18832
18833   *Ralf S. Engelschall*
18834
18835 * Second round of fixing the OpenSSL perl/ stuff. It now at least compiled
18836   fine under Unix and passes some trivial tests I've now added. But the
18837   whole stuff is horribly incomplete, so a README.1ST with a disclaimer was
18838   added to make sure no one expects that this stuff really works in the
18839   OpenSSL 0.9.2 release.  Additionally I've started to clean the XS sources
18840   up and fixed a few little bugs and inconsistencies in OpenSSL.{pm,xs} and
18841   openssl_bio.xs.
18842
18843   *Ralf S. Engelschall*
18844
18845 * Fix the generation of two part addresses in perl.
18846
18847   *Kenji Miyake <kenji@miyake.org>, integrated by Ben Laurie*
18848
18849 * Add config entry for Linux on MIPS.
18850
18851   *John Tobey <jtobey@channel1.com>*
18852
18853 * Make links whenever Configure is run, unless we are on Windoze.
18854
18855   *Ben Laurie*
18856
18857 * Permit extensions to be added to CRLs using crl_section in openssl.cnf.
18858   Currently only issuerAltName and AuthorityKeyIdentifier make any sense
18859   in CRLs.
18860
18861   *Steve Henson*
18862
18863 * Add a useful kludge to allow package maintainers to specify compiler and
18864   other platforms details on the command line without having to patch the
18865   Configure script every time: One now can use
18866   `perl Configure <id>:<details>`,
18867   i.e. platform ids are allowed to have details appended
18868   to them (separated by colons). This is treated as there would be a static
18869   pre-configured entry in Configure's %table under key `<id>` with value
18870   `<details>` and `perl Configure <id>` is called.  So, when you want to
18871   perform a quick test-compile under FreeBSD 3.1 with pgcc and without
18872   assembler stuff you can use `perl Configure "FreeBSD-elf:pgcc:-O6:::"`
18873   now, which overrides the FreeBSD-elf entry on-the-fly.
18874
18875   *Ralf S. Engelschall*
18876
18877 * Disable new TLS1 ciphersuites by default: they aren't official yet.
18878
18879   *Ben Laurie*
18880
18881 * Allow DSO flags like -fpic, -fPIC, -KPIC etc. to be specified
18882   on the `perl Configure ...` command line. This way one can compile
18883   OpenSSL libraries with Position Independent Code (PIC) which is needed
18884   for linking it into DSOs.
18885
18886   *Ralf S. Engelschall*
18887
18888 * Remarkably, export ciphers were totally broken and no-one had noticed!
18889   Fixed.
18890
18891   *Ben Laurie*
18892
18893 * Cleaned up the LICENSE document: The official contact for any license
18894   questions now is the OpenSSL core team under openssl-core@openssl.org.
18895   And add a paragraph about the dual-license situation to make sure people
18896   recognize that _BOTH_ the OpenSSL license _AND_ the SSLeay license apply
18897   to the OpenSSL toolkit.
18898
18899   *Ralf S. Engelschall*
18900
18901 * General source tree makefile cleanups: Made `making xxx in yyy...`
18902   display consistent in the source tree and replaced `/bin/rm` by `rm`.
18903   Additionally cleaned up the `make links` target: Remove unnecessary
18904   semicolons, subsequent redundant removes, inline point.sh into mklink.sh
18905   to speed processing and no longer clutter the display with confusing
18906   stuff. Instead only the actually done links are displayed.
18907
18908   *Ralf S. Engelschall*
18909
18910 * Permit null encryption ciphersuites, used for authentication only. It used
18911   to be necessary to set the preprocessor define SSL_ALLOW_ENULL to do this.
18912   It is now necessary to set SSL_FORBID_ENULL to prevent the use of null
18913   encryption.
18914
18915   *Ben Laurie*
18916
18917 * Add a bunch of fixes to the PKCS#7 stuff. It used to sometimes reorder
18918   signed attributes when verifying signatures (this would break them),
18919   the detached data encoding was wrong and public keys obtained using
18920   X509_get_pubkey() weren't freed.
18921
18922   *Steve Henson*
18923
18924 * Add text documentation for the BUFFER functions. Also added a work around
18925   to a Win95 console bug. This was triggered by the password read stuff: the
18926   last character typed gets carried over to the next fread(). If you were
18927   generating a new cert request using 'req' for example then the last
18928   character of the passphrase would be CR which would then enter the first
18929   field as blank.
18930
18931   *Steve Henson*
18932
18933 * Added the new 'Includes OpenSSL Cryptography Software' button as
18934   doc/openssl_button.{gif,html} which is similar in style to the old SSLeay
18935   button and can be used by applications based on OpenSSL to show the
18936   relationship to the OpenSSL project.
18937
18938   *Ralf S. Engelschall*
18939
18940 * Remove confusing variables in function signatures in files
18941   ssl/ssl_lib.c and ssl/ssl.h.
18942
18943   *Lennart Bong <lob@kulthea.stacken.kth.se>*
18944
18945 * Don't install bss_file.c under PREFIX/include/
18946
18947   *Lennart Bong <lob@kulthea.stacken.kth.se>*
18948
18949 * Get the Win32 compile working again. Modify mkdef.pl so it can handle
18950   functions that return function pointers and has support for NT specific
18951   stuff. Fix mk1mf.pl and VC-32.pl to support NT differences also. Various
18952   #ifdef WIN32 and WINNTs sprinkled about the place and some changes from
18953   unsigned to signed types: this was killing the Win32 compile.
18954
18955   *Steve Henson*
18956
18957 * Add new certificate file to stack functions,
18958   SSL_add_dir_cert_subjects_to_stack() and
18959   SSL_add_file_cert_subjects_to_stack().  These largely supplant
18960   SSL_load_client_CA_file(), and can be used to add multiple certs easily
18961   to a stack (usually this is then handed to SSL_CTX_set_client_CA_list()).
18962   This means that Apache-SSL and similar packages don't have to mess around
18963   to add as many CAs as they want to the preferred list.
18964
18965   *Ben Laurie*
18966
18967 * Experiment with doxygen documentation. Currently only partially applied to
18968   ssl/ssl_lib.c.
18969   See <http://www.stack.nl/~dimitri/doxygen/index.html>, and run doxygen with
18970   openssl.doxy as the configuration file.
18971
18972   *Ben Laurie*
18973
18974 * Get rid of remaining C++-style comments which strict C compilers hate.
18975
18976   *Ralf S. Engelschall, pointed out by Carlos Amengual*
18977
18978 * Changed BN_RECURSION in bn_mont.c to BN_RECURSION_MONT so it is not
18979   compiled in by default: it has problems with large keys.
18980
18981   *Steve Henson*
18982
18983 * Add a bunch of SSL_xxx() functions for configuring the temporary RSA and
18984   DH private keys and/or callback functions which directly correspond to
18985   their SSL_CTX_xxx() counterparts but work on a per-connection basis. This
18986   is needed for applications which have to configure certificates on a
18987   per-connection basis (e.g. Apache+mod_ssl) instead of a per-context basis
18988   (e.g. s_server).
18989      For the RSA certificate situation is makes no difference, but
18990   for the DSA certificate situation this fixes the "no shared cipher"
18991   problem where the OpenSSL cipher selection procedure failed because the
18992   temporary keys were not overtaken from the context and the API provided
18993   no way to reconfigure them.
18994      The new functions now let applications reconfigure the stuff and they
18995   are in detail: SSL_need_tmp_RSA, SSL_set_tmp_rsa, SSL_set_tmp_dh,
18996   SSL_set_tmp_rsa_callback and SSL_set_tmp_dh_callback.  Additionally a new
18997   non-public-API function ssl_cert_instantiate() is used as a helper
18998   function and also to reduce code redundancy inside ssl_rsa.c.
18999
19000   *Ralf S. Engelschall*
19001
19002 * Move s_server -dcert and -dkey options out of the undocumented feature
19003   area because they are useful for the DSA situation and should be
19004   recognized by the users.
19005
19006   *Ralf S. Engelschall*
19007
19008 * Fix the cipher decision scheme for export ciphers: the export bits are
19009   *not* within SSL_MKEY_MASK or SSL_AUTH_MASK, they are within
19010   SSL_EXP_MASK.  So, the original variable has to be used instead of the
19011   already masked variable.
19012
19013   *Richard Levitte <levitte@stacken.kth.se>*
19014
19015 * Fix `port` variable from `int` to `unsigned int` in crypto/bio/b_sock.c
19016
19017   *Richard Levitte <levitte@stacken.kth.se>*
19018
19019 * Change type of another md_len variable in pk7_doit.c:PKCS7_dataFinal()
19020   from `int` to `unsigned int` because it is a length and initialized by
19021   EVP_DigestFinal() which expects an `unsigned int *`.
19022
19023   *Richard Levitte <levitte@stacken.kth.se>*
19024
19025 * Don't hard-code path to Perl interpreter on shebang line of Configure
19026   script. Instead use the usual Shell->Perl transition trick.
19027
19028   *Ralf S. Engelschall*
19029
19030 * Make `openssl x509 -noout -modulus`' functional also for DSA certificates
19031   (in addition to RSA certificates) to match the behaviour of `openssl dsa
19032   -noout -modulus` as it's already the case for `openssl rsa -noout
19033   -modulus`.  For RSA the -modulus is the real "modulus" while for DSA
19034   currently the public key is printed (a decision which was already done by
19035   `openssl dsa -modulus` in the past) which serves a similar purpose.
19036   Additionally the NO_RSA no longer completely removes the whole -modulus
19037   option; it now only avoids using the RSA stuff. Same applies to NO_DSA
19038   now, too.
19039
19040   *Ralf S.  Engelschall*
19041
19042 * Add Arne Ansper's reliable BIO - this is an encrypted, block-digested
19043   BIO. See the source (crypto/evp/bio_ok.c) for more info.
19044
19045   *Arne Ansper <arne@ats.cyber.ee>*
19046
19047 * Dump the old yucky req code that tried (and failed) to allow raw OIDs
19048   to be added. Now both 'req' and 'ca' can use new objects defined in the
19049   config file.
19050
19051   *Steve Henson*
19052
19053 * Add cool BIO that does syslog (or event log on NT).
19054
19055   *Arne Ansper <arne@ats.cyber.ee>, integrated by Ben Laurie*
19056
19057 * Add support for new TLS ciphersuites, TLS_RSA_EXPORT56_WITH_RC4_56_MD5,
19058   TLS_RSA_EXPORT56_WITH_RC2_CBC_56_MD5 and
19059   TLS_RSA_EXPORT56_WITH_DES_CBC_SHA, as specified in "56-bit Export Cipher
19060   Suites For TLS", draft-ietf-tls-56-bit-ciphersuites-00.txt.
19061
19062   *Ben Laurie*
19063
19064 * Add preliminary config info for new extension code.
19065
19066   *Steve Henson*
19067
19068 * Make RSA_NO_PADDING really use no padding.
19069
19070   *Ulf Moeller <ulf@fitug.de>*
19071
19072 * Generate errors when private/public key check is done.
19073
19074   *Ben Laurie*
19075
19076 * Overhaul for 'crl' utility. New function X509_CRL_print. Partial support
19077   for some CRL extensions and new objects added.
19078
19079   *Steve Henson*
19080
19081 * Really fix the ASN1 IMPLICIT bug this time... Partial support for private
19082   key usage extension and fuller support for authority key id.
19083
19084   *Steve Henson*
19085
19086 * Add OAEP encryption for the OpenSSL crypto library. OAEP is the improved
19087   padding method for RSA, which is recommended for new applications in PKCS
19088   #1 v2.0 (RFC 2437, October 1998).
19089   OAEP (Optimal Asymmetric Encryption Padding) has better theoretical
19090   foundations than the ad-hoc padding used in PKCS #1 v1.5. It is secure
19091   against Bleichbacher's attack on RSA.
19092   *Ulf Moeller <ulf@fitug.de>, reformatted, corrected and integrated by
19093   Ben Laurie*
19094
19095 * Updates to the new SSL compression code
19096
19097   *Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)*
19098
19099 * Fix so that the version number in the master secret, when passed
19100   via RSA, checks that if TLS was proposed, but we roll back to SSLv3
19101   (because the server will not accept higher), that the version number
19102   is 0x03,0x01, not 0x03,0x00
19103
19104   *Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)*
19105
19106 * Run extensive memory leak checks on SSL commands. Fixed *lots* of memory
19107   leaks in `ssl/` relating to new `X509_get_pubkey()` behaviour. Also fixes
19108   in `apps/` and an unrelated leak in `crypto/dsa/dsa_vrf.c`.
19109
19110   *Steve Henson*
19111
19112 * Support for RAW extensions where an arbitrary extension can be
19113   created by including its DER encoding. See `apps/openssl.cnf` for
19114   an example.
19115
19116   *Steve Henson*
19117
19118 * Make sure latest Perl versions don't interpret some generated C array
19119   code as Perl array code in the crypto/err/err_genc.pl script.
19120
19121   *Lars Weber <3weber@informatik.uni-hamburg.de>*
19122
19123 * Modify ms/do_ms.bat to not generate assembly language makefiles since
19124   not many people have the assembler. Various Win32 compilation fixes and
19125   update to the INSTALL.W32 file with (hopefully) more accurate Win32
19126   build instructions.
19127
19128   *Steve Henson*
19129
19130 * Modify configure script 'Configure' to automatically create crypto/date.h
19131   file under Win32 and also build pem.h from pem.org. New script
19132   util/mkfiles.pl to create the MINFO file on environments that can't do a
19133   'make files': perl util/mkfiles.pl >MINFO should work.
19134
19135   *Steve Henson*
19136
19137 * Major rework of DES function declarations, in the pursuit of correctness
19138   and purity. As a result, many evil casts evaporated, and some weirdness,
19139   too. You may find this causes warnings in your code. Zapping your evil
19140   casts will probably fix them. Mostly.
19141
19142   *Ben Laurie*
19143
19144 * Fix for a typo in asn1.h. Bug fix to object creation script
19145   obj_dat.pl. It considered a zero in an object definition to mean
19146   "end of object": none of the objects in objects.h have any zeros
19147   so it wasn't spotted.
19148
19149   *Steve Henson, reported by Erwann ABALEA <eabalea@certplus.com>*
19150
19151 * Add support for Triple DES Cipher Block Chaining with Output Feedback
19152   Masking (CBCM). In the absence of test vectors, the best I have been able
19153   to do is check that the decrypt undoes the encrypt, so far. Send me test
19154   vectors if you have them.
19155
19156   *Ben Laurie*
19157
19158 * Correct calculation of key length for export ciphers (too much space was
19159   allocated for null ciphers). This has not been tested!
19160
19161   *Ben Laurie*
19162
19163 * Modifications to the mkdef.pl for Win32 DEF file creation. The usage
19164   message is now correct (it understands "crypto" and "ssl" on its
19165   command line). There is also now an "update" option. This will update
19166   the util/ssleay.num and util/libeay.num files with any new functions.
19167   If you do a:
19168   perl util/mkdef.pl crypto ssl update
19169   it will update them.
19170
19171   *Steve Henson*
19172
19173 * Overhauled the Perl interface:
19174   - ported BN stuff to OpenSSL's different BN library
19175   - made the perl/ source tree CVS-aware
19176   - renamed the package from SSLeay to OpenSSL (the files still contain
19177     their history because I've copied them in the repository)
19178   - removed obsolete files (the test scripts will be replaced
19179     by better Test::Harness variants in the future)
19180
19181   *Ralf S. Engelschall*
19182
19183 * First cut for a very conservative source tree cleanup:
19184   1. merge various obsolete readme texts into doc/ssleay.txt
19185   where we collect the old documents and readme texts.
19186   2. remove the first part of files where I'm already sure that we no
19187   longer need them because of three reasons: either they are just temporary
19188   files which were left by Eric or they are preserved original files where
19189   I've verified that the diff is also available in the CVS via "cvs diff
19190   -rSSLeay_0_8_1b" or they were renamed (as it was definitely the case for
19191   the crypto/md/ stuff).
19192
19193   *Ralf S. Engelschall*
19194
19195 * More extension code. Incomplete support for subject and issuer alt
19196   name, issuer and authority key id. Change the i2v function parameters
19197   and add an extra 'crl' parameter in the X509V3_CTX structure: guess
19198   what that's for :-) Fix to ASN1 macro which messed up
19199   IMPLICIT tag and add f_enum.c which adds a2i, i2a for ENUMERATED.
19200
19201   *Steve Henson*
19202
19203 * Preliminary support for ENUMERATED type. This is largely copied from the
19204   INTEGER code.
19205
19206   *Steve Henson*
19207
19208 * Add new function, EVP_MD_CTX_copy() to replace frequent use of memcpy.
19209
19210   *Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)*
19211
19212 * Make sure `make rehash` target really finds the `openssl` program.
19213
19214   *Ralf S. Engelschall, Matthias Loepfe <Matthias.Loepfe@adnovum.ch>*
19215
19216 * Squeeze another 7% of speed out of MD5 assembler, at least on a P2. I'd
19217   like to hear about it if this slows down other processors.
19218
19219   *Ben Laurie*
19220
19221 * Add CygWin32 platform information to Configure script.
19222
19223   *Alan Batie <batie@aahz.jf.intel.com>*
19224
19225 * Fixed ms/32all.bat script: `no_asm` -> `no-asm`
19226
19227   *Rainer W. Gerling <gerling@mpg-gv.mpg.de>*
19228
19229 * New program nseq to manipulate netscape certificate sequences
19230
19231   *Steve Henson*
19232
19233 * Modify crl2pkcs7 so it supports multiple -certfile arguments. Fix a
19234   few typos.
19235
19236   *Steve Henson*
19237
19238 * Fixes to BN code.  Previously the default was to define BN_RECURSION
19239   but the BN code had some problems that would cause failures when
19240   doing certificate verification and some other functions.
19241
19242   *Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)*
19243
19244 * Add ASN1 and PEM code to support netscape certificate sequences.
19245
19246   *Steve Henson*
19247
19248 * Add ASN1 and PEM code to support netscape certificate sequences.
19249
19250   *Steve Henson*
19251
19252 * Add several PKIX and private extended key usage OIDs.
19253
19254   *Steve Henson*
19255
19256 * Modify the 'ca' program to handle the new extension code. Modify
19257   openssl.cnf for new extension format, add comments.
19258
19259   *Steve Henson*
19260
19261 * More X509 V3 changes. Fix typo in v3_bitstr.c. Add support to 'req'
19262   and add a sample to openssl.cnf so req -x509 now adds appropriate
19263   CA extensions.
19264
19265   *Steve Henson*
19266
19267 * Continued X509 V3 changes. Add to other makefiles, integrate with the
19268   error code, add initial support to X509_print() and x509 application.
19269
19270   *Steve Henson*
19271
19272 * Takes a deep breath and start adding X509 V3 extension support code. Add
19273   files in crypto/x509v3. Move original stuff to crypto/x509v3/old. All this
19274   stuff is currently isolated and isn't even compiled yet.
19275
19276   *Steve Henson*
19277
19278 * Continuing patches for GeneralizedTime. Fix up certificate and CRL
19279   ASN1 to use ASN1_TIME and modify print routines to use ASN1_TIME_print.
19280   Removed the versions check from X509 routines when loading extensions:
19281   this allows certain broken certificates that don't set the version
19282   properly to be processed.
19283
19284   *Steve Henson*
19285
19286 * Deal with irritating shit to do with dependencies, in YAAHW (Yet Another
19287   Ad Hoc Way) - Makefile.ssls now all contain local dependencies, which
19288   can still be regenerated with "make depend".
19289
19290   *Ben Laurie*
19291
19292 * Spelling mistake in C version of CAST-128.
19293
19294   *Ben Laurie, reported by Jeremy Hylton <jeremy@cnri.reston.va.us>*
19295
19296 * Changes to the error generation code. The perl script err-code.pl
19297   now reads in the old error codes and retains the old numbers, only
19298   adding new ones if necessary. It also only changes the .err files if new
19299   codes are added. The makefiles have been modified to only insert errors
19300   when needed (to avoid needlessly modifying header files). This is done
19301   by only inserting errors if the .err file is newer than the auto generated
19302   C file. To rebuild all the error codes from scratch (the old behaviour)
19303   either modify crypto/Makefile.ssl to pass the -regen flag to err_code.pl
19304   or delete all the .err files.
19305
19306   *Steve Henson*
19307
19308 * CAST-128 was incorrectly implemented for short keys. The C version has
19309   been fixed, but is untested. The assembler versions are also fixed, but
19310   new assembler HAS NOT BEEN GENERATED FOR WIN32 - the Makefile needs fixing
19311   to regenerate it if needed.
19312   *Ben Laurie, reported (with fix for C version) by Jun-ichiro itojun
19313    Hagino <itojun@kame.net>*
19314
19315 * File was opened incorrectly in randfile.c.
19316
19317   *Ulf Möller <ulf@fitug.de>*
19318
19319 * Beginning of support for GeneralizedTime. d2i, i2d, check and print
19320   functions. Also ASN1_TIME suite which is a CHOICE of UTCTime or
19321   GeneralizedTime. ASN1_TIME is the proper type used in certificates et
19322   al: it's just almost always a UTCTime. Note this patch adds new error
19323   codes so do a "make errors" if there are problems.
19324
19325   *Steve Henson*
19326
19327 * Correct Linux 1 recognition in config.
19328
19329   *Ulf Möller <ulf@fitug.de>*
19330
19331 * Remove pointless MD5 hash when using DSA keys in ca.
19332
19333   *Anonymous <nobody@replay.com>*
19334
19335 * Generate an error if given an empty string as a cert directory. Also
19336   generate an error if handed NULL (previously returned 0 to indicate an
19337   error, but didn't set one).
19338
19339   *Ben Laurie, reported by Anonymous <nobody@replay.com>*
19340
19341 * Add prototypes to SSL methods. Make SSL_write's buffer const, at last.
19342
19343   *Ben Laurie*
19344
19345 * Fix the dummy function BN_ref_mod_exp() in rsaref.c to have the correct
19346   parameters. This was causing a warning which killed off the Win32 compile.
19347
19348   *Steve Henson*
19349
19350 * Remove C++ style comments from crypto/bn/bn_local.h.
19351
19352   *Neil Costigan <neil.costigan@celocom.com>*
19353
19354 * The function OBJ_txt2nid was broken. It was supposed to return a nid
19355   based on a text string, looking up short and long names and finally
19356   "dot" format. The "dot" format stuff didn't work. Added new function
19357   OBJ_txt2obj to do the same but return an ASN1_OBJECT and rewrote
19358   OBJ_txt2nid to use it. OBJ_txt2obj can also return objects even if the
19359   OID is not part of the table.
19360
19361   *Steve Henson*
19362
19363 * Add prototypes to X509 lookup/verify methods, fixing a bug in
19364   X509_LOOKUP_by_alias().
19365
19366   *Ben Laurie*
19367
19368 * Sort openssl functions by name.
19369
19370   *Ben Laurie*
19371
19372 * Get the `gendsa` command working and add it to the `list` command. Remove
19373   encryption from sample DSA keys (in case anyone is interested the password
19374   was "1234").
19375
19376   *Steve Henson*
19377
19378 * Make *all* `*_free` functions accept a NULL pointer.
19379
19380   *Frans Heymans <fheymans@isaserver.be>*
19381
19382 * If a DH key is generated in s3_srvr.c, don't blow it by trying to use
19383   NULL pointers.
19384
19385   *Anonymous <nobody@replay.com>*
19386
19387 * s_server should send the CAfile as acceptable CAs, not its own cert.
19388
19389   *Bodo Moeller <3moeller@informatik.uni-hamburg.de>*
19390
19391 * Don't blow it for numeric `-newkey` arguments to `apps/req`.
19392
19393   *Bodo Moeller <3moeller@informatik.uni-hamburg.de>*
19394
19395 * Temp key "for export" tests were wrong in s3_srvr.c.
19396
19397   *Anonymous <nobody@replay.com>*
19398
19399 * Add prototype for temp key callback functions
19400   SSL_CTX_set_tmp_{rsa,dh}_callback().
19401
19402   *Ben Laurie*
19403
19404 * Make DH_free() tolerate being passed a NULL pointer (like RSA_free() and
19405   DSA_free()). Make X509_PUBKEY_set() check for errors in d2i_PublicKey().
19406
19407   *Steve Henson*
19408
19409 * X509_name_add_entry() freed the wrong thing after an error.
19410
19411   *Arne Ansper <arne@ats.cyber.ee>*
19412
19413 * rsa_eay.c would attempt to free a NULL context.
19414
19415   *Arne Ansper <arne@ats.cyber.ee>*
19416
19417 * BIO_s_socket() had a broken should_retry() on Windoze.
19418
19419   *Arne Ansper <arne@ats.cyber.ee>*
19420
19421 * BIO_f_buffer() didn't pass on BIO_CTRL_FLUSH.
19422
19423   *Arne Ansper <arne@ats.cyber.ee>*
19424
19425 * Make sure the already existing X509_STORE->depth variable is initialized
19426   in X509_STORE_new(), but document the fact that this variable is still
19427   unused in the certificate verification process.
19428
19429   *Ralf S. Engelschall*
19430
19431 * Fix the various library and `apps/` files to free up pkeys obtained from
19432   X509_PUBKEY_get() et al. Also allow x509.c to handle netscape extensions.
19433
19434   *Steve Henson*
19435
19436 * Fix reference counting in X509_PUBKEY_get(). This makes
19437   demos/maurice/example2.c work, amongst others, probably.
19438
19439   *Steve Henson and Ben Laurie*
19440
19441 * First cut of a cleanup for `apps/`. First the `ssleay` program is now named
19442   `openssl` and second, the shortcut symlinks for the `openssl <command>`
19443   are no longer created. This way we have a single and consistent command
19444   line interface `openssl <command>`, similar to `cvs <command>`.
19445
19446   *Ralf S. Engelschall, Paul Sutton and Ben Laurie*
19447
19448 * ca.c: move test for DSA keys inside #ifndef NO_DSA. Make pubkey
19449   BIT STRING wrapper always have zero unused bits.
19450
19451   *Steve Henson*
19452
19453 * Add CA.pl, perl version of CA.sh, add extended key usage OID.
19454
19455   *Steve Henson*
19456
19457 * Make the top-level INSTALL documentation easier to understand.
19458
19459   *Paul Sutton*
19460
19461 * Makefiles updated to exit if an error occurs in a sub-directory
19462   make (including if user presses ^C) [Paul Sutton]
19463
19464 * Make Montgomery context stuff explicit in RSA data structure.
19465
19466   *Ben Laurie*
19467
19468 * Fix build order of pem and err to allow for generated pem.h.
19469
19470   *Ben Laurie*
19471
19472 * Fix renumbering bug in X509_NAME_delete_entry().
19473
19474   *Ben Laurie*
19475
19476 * Enhanced the err-ins.pl script so it makes the error library number
19477   global and can add a library name. This is needed for external ASN1 and
19478   other error libraries.
19479
19480   *Steve Henson*
19481
19482 * Fixed sk_insert which never worked properly.
19483
19484   *Steve Henson*
19485
19486 * Fix ASN1 macros so they can handle indefinite length constructed
19487   EXPLICIT tags. Some non standard certificates use these: they can now
19488   be read in.
19489
19490   *Steve Henson*
19491
19492 * Merged the various old/obsolete SSLeay documentation files (doc/xxx.doc)
19493   into a single doc/ssleay.txt bundle. This way the information is still
19494   preserved but no longer messes up this directory. Now it's new room for
19495   the new set of documentation files.
19496
19497   *Ralf S. Engelschall*
19498
19499 * SETs were incorrectly DER encoded. This was a major pain, because they
19500   shared code with SEQUENCEs, which aren't coded the same. This means that
19501   almost everything to do with SETs or SEQUENCEs has either changed name or
19502   number of arguments.
19503
19504   *Ben Laurie, based on a partial fix by GP Jayan <gp@nsj.co.jp>*
19505
19506 * Fix test data to work with the above.
19507
19508   *Ben Laurie*
19509
19510 * Fix the RSA header declarations that hid a bug I fixed in 0.9.0b but
19511   was already fixed by Eric for 0.9.1 it seems.
19512
19513   *Ben Laurie - pointed out by Ulf Möller <ulf@fitug.de>*
19514
19515 * Autodetect FreeBSD3.
19516
19517   *Ben Laurie*
19518
19519 * Fix various bugs in Configure. This affects the following platforms:
19520   nextstep
19521   ncr-scde
19522   unixware-2.0
19523   unixware-2.0-pentium
19524   sco5-cc.
19525
19526   *Ben Laurie*
19527
19528 * Eliminate generated files from CVS. Reorder tests to regenerate files
19529   before they are needed.
19530
19531   *Ben Laurie*
19532
19533 * Generate Makefile.ssl from Makefile.org (to keep CVS happy).
19534
19535   *Ben Laurie*
19536
19537### Changes between 0.9.1b and 0.9.1c  [23-Dec-1998]
19538
19539 * Added OPENSSL_VERSION_NUMBER to crypto/crypto.h and
19540   changed SSLeay to OpenSSL in version strings.
19541
19542   *Ralf S. Engelschall*
19543
19544 * Some fixups to the top-level documents.
19545
19546   *Paul Sutton*
19547
19548 * Fixed the nasty bug where rsaref.h was not found under compile-time
19549   because the symlink to include/ was missing.
19550
19551   *Ralf S. Engelschall*
19552
19553 * Incorporated the popular no-RSA/DSA-only patches
19554   which allow to compile an RSA-free SSLeay.
19555
19556   *Andrew Cooke / Interrader Ldt., Ralf S. Engelschall*
19557
19558 * Fixed nasty rehash problem under `make -f Makefile.ssl links`
19559   when "ssleay" is still not found.
19560
19561   *Ralf S. Engelschall*
19562
19563 * Added more platforms to Configure: Cray T3E, HPUX 11,
19564
19565   *Ralf S. Engelschall, Beckmann <beckman@acl.lanl.gov>*
19566
19567 * Updated the README file.
19568
19569   *Ralf S. Engelschall*
19570
19571 * Added various .cvsignore files in the CVS repository subdirs
19572   to make a "cvs update" really silent.
19573
19574   *Ralf S. Engelschall*
19575
19576 * Recompiled the error-definition header files and added
19577   missing symbols to the Win32 linker tables.
19578
19579   *Ralf S. Engelschall*
19580
19581 * Cleaned up the top-level documents;
19582   o new files: CHANGES and LICENSE
19583   o merged VERSION, HISTORY* and README* files a CHANGES.SSLeay
19584   o merged COPYRIGHT into LICENSE
19585   o removed obsolete TODO file
19586   o renamed MICROSOFT to INSTALL.W32
19587
19588   *Ralf S. Engelschall*
19589
19590 * Removed dummy files from the 0.9.1b source tree:
19591   crypto/asn1/x crypto/bio/cd crypto/bio/fg crypto/bio/grep crypto/bio/vi
19592   crypto/bn/asm/......add.c crypto/bn/asm/a.out crypto/dsa/f crypto/md5/f
19593   crypto/pem/gmon.out crypto/perlasm/f crypto/pkcs7/build crypto/rsa/f
19594   crypto/sha/asm/f crypto/threads/f ms/zzz ssl/f ssl/f.mak test/f
19595   util/f.mak util/pl/f util/pl/f.mak crypto/bf/bf_locl.old apps/f
19596
19597   *Ralf S. Engelschall*
19598
19599 * Added various platform portability fixes.
19600
19601   *Mark J. Cox*
19602
19603 * The Genesis of the OpenSSL rpject:
19604   We start with the latest (unreleased) SSLeay version 0.9.1b which Eric A.
19605   Young and Tim J. Hudson created while they were working for C2Net until
19606   summer 1998.
19607
19608   *The OpenSSL Project*
19609
19610### Changes between 0.9.0b and 0.9.1b  [not released]
19611
19612 * Updated a few CA certificates under certs/
19613
19614   *Eric A. Young*
19615
19616 * Changed some BIGNUM api stuff.
19617
19618   *Eric A. Young*
19619
19620 * Various platform ports: OpenBSD, Ultrix, IRIX 64bit, NetBSD,
19621   DGUX x86, Linux Alpha, etc.
19622
19623   *Eric A. Young*
19624
19625 * New COMP library [crypto/comp/] for SSL Record Layer Compression:
19626   RLE (dummy implemented) and ZLIB (really implemented when ZLIB is
19627   available).
19628
19629   *Eric A. Young*
19630
19631 * Add -strparse option to asn1pars program which parses nested
19632   binary structures
19633
19634   *Dr Stephen Henson <shenson@bigfoot.com>*
19635
19636 * Added "oid_file" to ssleay.cnf for "ca" and "req" programs.
19637
19638   *Eric A. Young*
19639
19640 * DSA fix for "ca" program.
19641
19642   *Eric A. Young*
19643
19644 * Added "-genkey" option to "dsaparam" program.
19645
19646   *Eric A. Young*
19647
19648 * Added RIPE MD160 (rmd160) message digest.
19649
19650   *Eric A. Young*
19651
19652 * Added -a (all) option to "ssleay version" command.
19653
19654   *Eric A. Young*
19655
19656 * Added PLATFORM define which is the id given to Configure.
19657
19658   *Eric A. Young*
19659
19660 * Added MemCheck_XXXX functions to crypto/mem.c for memory checking.
19661
19662   *Eric A. Young*
19663
19664 * Extended the ASN.1 parser routines.
19665
19666   *Eric A. Young*
19667
19668 * Extended BIO routines to support REUSEADDR, seek, tell, etc.
19669
19670   *Eric A. Young*
19671
19672 * Added a BN_CTX to the BN library.
19673
19674   *Eric A. Young*
19675
19676 * Fixed the weak key values in DES library
19677
19678   *Eric A. Young*
19679
19680 * Changed API in EVP library for cipher aliases.
19681
19682   *Eric A. Young*
19683
19684 * Added support for RC2/64bit cipher.
19685
19686   *Eric A. Young*
19687
19688 * Converted the lhash library to the crypto/mem.c functions.
19689
19690   *Eric A. Young*
19691
19692 * Added more recognized ASN.1 object ids.
19693
19694   *Eric A. Young*
19695
19696 * Added more RSA padding checks for SSL/TLS.
19697
19698   *Eric A. Young*
19699
19700 * Added BIO proxy/filter functionality.
19701
19702   *Eric A. Young*
19703
19704 * Added extra_certs to SSL_CTX which can be used
19705   send extra CA certificates to the client in the CA cert chain sending
19706   process. It can be configured with SSL_CTX_add_extra_chain_cert().
19707
19708   *Eric A. Young*
19709
19710 * Now Fortezza is denied in the authentication phase because
19711   this is key exchange mechanism is not supported by SSLeay at all.
19712
19713   *Eric A. Young*
19714
19715 * Additional PKCS1 checks.
19716
19717   *Eric A. Young*
19718
19719 * Support the string "TLSv1" for all TLS v1 ciphers.
19720
19721   *Eric A. Young*
19722
19723 * Added function SSL_get_ex_data_X509_STORE_CTX_idx() which gives the
19724   ex_data index of the SSL context in the X509_STORE_CTX ex_data.
19725
19726   *Eric A. Young*
19727
19728 * Fixed a few memory leaks.
19729
19730   *Eric A. Young*
19731
19732 * Fixed various code and comment typos.
19733
19734   *Eric A. Young*
19735
19736 * A minor bug in ssl/s3_clnt.c where there would always be 4 0
19737   bytes sent in the client random.
19738
19739   *Edward Bishop <ebishop@spyglass.com>*
19740
19741<!-- Links -->
19742
19743[CVE-2023-5363]: https://www.openssl.org/news/vulnerabilities.html#CVE-2023-5363
19744[CVE-2023-4807]: https://www.openssl.org/news/vulnerabilities.html#CVE-2023-4807
19745[CVE-2023-3817]: https://www.openssl.org/news/vulnerabilities.html#CVE-2023-3817
19746[CVE-2023-3446]: https://www.openssl.org/news/vulnerabilities.html#CVE-2023-3446
19747[CVE-2023-2975]: https://www.openssl.org/news/vulnerabilities.html#CVE-2023-2975
19748[RFC 2578 (STD 58), section 3.5]: https://datatracker.ietf.org/doc/html/rfc2578#section-3.5
19749[CVE-2023-2650]: https://www.openssl.org/news/vulnerabilities.html#CVE-2023-2650
19750[CVE-2023-1255]: https://www.openssl.org/news/vulnerabilities.html#CVE-2023-1255
19751[CVE-2023-0466]: https://www.openssl.org/news/vulnerabilities.html#CVE-2023-0466
19752[CVE-2023-0465]: https://www.openssl.org/news/vulnerabilities.html#CVE-2023-0465
19753[CVE-2023-0464]: https://www.openssl.org/news/vulnerabilities.html#CVE-2023-0464
19754[CVE-2023-0401]: https://www.openssl.org/news/vulnerabilities.html#CVE-2023-0401
19755[CVE-2023-0286]: https://www.openssl.org/news/vulnerabilities.html#CVE-2023-0286
19756[CVE-2023-0217]: https://www.openssl.org/news/vulnerabilities.html#CVE-2023-0217
19757[CVE-2023-0216]: https://www.openssl.org/news/vulnerabilities.html#CVE-2023-0216
19758[CVE-2023-0215]: https://www.openssl.org/news/vulnerabilities.html#CVE-2023-0215
19759[CVE-2022-4450]: https://www.openssl.org/news/vulnerabilities.html#CVE-2022-4450
19760[CVE-2022-4304]: https://www.openssl.org/news/vulnerabilities.html#CVE-2022-4304
19761[CVE-2022-4203]: https://www.openssl.org/news/vulnerabilities.html#CVE-2022-4203
19762[CVE-2022-3996]: https://www.openssl.org/news/vulnerabilities.html#CVE-2022-3996
19763[CVE-2022-2274]: https://www.openssl.org/news/vulnerabilities.html#CVE-2022-2274
19764[CVE-2022-2097]: https://www.openssl.org/news/vulnerabilities.html#CVE-2022-2097
19765[CVE-2020-1971]: https://www.openssl.org/news/vulnerabilities.html#CVE-2020-1971
19766[CVE-2020-1967]: https://www.openssl.org/news/vulnerabilities.html#CVE-2020-1967
19767[CVE-2019-1563]: https://www.openssl.org/news/vulnerabilities.html#CVE-2019-1563
19768[CVE-2019-1559]: https://www.openssl.org/news/vulnerabilities.html#CVE-2019-1559
19769[CVE-2019-1552]: https://www.openssl.org/news/vulnerabilities.html#CVE-2019-1552
19770[CVE-2019-1551]: https://www.openssl.org/news/vulnerabilities.html#CVE-2019-1551
19771[CVE-2019-1549]: https://www.openssl.org/news/vulnerabilities.html#CVE-2019-1549
19772[CVE-2019-1547]: https://www.openssl.org/news/vulnerabilities.html#CVE-2019-1547
19773[CVE-2019-1543]: https://www.openssl.org/news/vulnerabilities.html#CVE-2019-1543
19774[CVE-2018-5407]: https://www.openssl.org/news/vulnerabilities.html#CVE-2018-5407
19775[CVE-2018-0739]: https://www.openssl.org/news/vulnerabilities.html#CVE-2018-0739
19776[CVE-2018-0737]: https://www.openssl.org/news/vulnerabilities.html#CVE-2018-0737
19777[CVE-2018-0735]: https://www.openssl.org/news/vulnerabilities.html#CVE-2018-0735
19778[CVE-2018-0734]: https://www.openssl.org/news/vulnerabilities.html#CVE-2018-0734
19779[CVE-2018-0733]: https://www.openssl.org/news/vulnerabilities.html#CVE-2018-0733
19780[CVE-2018-0732]: https://www.openssl.org/news/vulnerabilities.html#CVE-2018-0732
19781[CVE-2017-3738]: https://www.openssl.org/news/vulnerabilities.html#CVE-2017-3738
19782[CVE-2017-3737]: https://www.openssl.org/news/vulnerabilities.html#CVE-2017-3737
19783[CVE-2017-3736]: https://www.openssl.org/news/vulnerabilities.html#CVE-2017-3736
19784[CVE-2017-3735]: https://www.openssl.org/news/vulnerabilities.html#CVE-2017-3735
19785[CVE-2017-3733]: https://www.openssl.org/news/vulnerabilities.html#CVE-2017-3733
19786[CVE-2017-3732]: https://www.openssl.org/news/vulnerabilities.html#CVE-2017-3732
19787[CVE-2017-3731]: https://www.openssl.org/news/vulnerabilities.html#CVE-2017-3731
19788[CVE-2017-3730]: https://www.openssl.org/news/vulnerabilities.html#CVE-2017-3730
19789[CVE-2016-7055]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-7055
19790[CVE-2016-7054]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-7054
19791[CVE-2016-7053]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-7053
19792[CVE-2016-7052]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-7052
19793[CVE-2016-6309]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-6309
19794[CVE-2016-6308]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-6308
19795[CVE-2016-6307]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-6307
19796[CVE-2016-6306]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-6306
19797[CVE-2016-6305]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-6305
19798[CVE-2016-6304]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-6304
19799[CVE-2016-6303]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-6303
19800[CVE-2016-6302]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-6302
19801[CVE-2016-2183]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-2183
19802[CVE-2016-2182]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-2182
19803[CVE-2016-2181]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-2181
19804[CVE-2016-2180]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-2180
19805[CVE-2016-2179]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-2179
19806[CVE-2016-2178]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-2178
19807[CVE-2016-2177]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-2177
19808[CVE-2016-2176]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-2176
19809[CVE-2016-2109]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-2109
19810[CVE-2016-2107]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-2107
19811[CVE-2016-2106]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-2106
19812[CVE-2016-2105]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-2105
19813[CVE-2016-0800]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-0800
19814[CVE-2016-0799]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-0799
19815[CVE-2016-0798]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-0798
19816[CVE-2016-0797]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-0797
19817[CVE-2016-0705]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-0705
19818[CVE-2016-0702]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-0702
19819[CVE-2016-0701]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-0701
19820[CVE-2015-3197]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-3197
19821[CVE-2015-3196]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-3196
19822[CVE-2015-3195]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-3195
19823[CVE-2015-3194]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-3194
19824[CVE-2015-3193]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-3193
19825[CVE-2015-1793]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-1793
19826[CVE-2015-1792]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-1792
19827[CVE-2015-1791]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-1791
19828[CVE-2015-1790]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-1790
19829[CVE-2015-1789]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-1789
19830[CVE-2015-1788]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-1788
19831[CVE-2015-1787]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-1787
19832[CVE-2015-0293]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0293
19833[CVE-2015-0291]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0291
19834[CVE-2015-0290]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0290
19835[CVE-2015-0289]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0289
19836[CVE-2015-0288]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0288
19837[CVE-2015-0287]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0287
19838[CVE-2015-0286]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0286
19839[CVE-2015-0285]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0285
19840[CVE-2015-0209]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0209
19841[CVE-2015-0208]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0208
19842[CVE-2015-0207]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0207
19843[CVE-2015-0206]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0206
19844[CVE-2015-0205]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0205
19845[CVE-2015-0204]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0204
19846[CVE-2014-8275]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-8275
19847[CVE-2014-5139]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-5139
19848[CVE-2014-3572]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3572
19849[CVE-2014-3571]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3571
19850[CVE-2014-3570]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3570
19851[CVE-2014-3569]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3569
19852[CVE-2014-3568]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3568
19853[CVE-2014-3567]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3567
19854[CVE-2014-3566]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3566
19855[CVE-2014-3513]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3513
19856[CVE-2014-3512]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3512
19857[CVE-2014-3511]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3511
19858[CVE-2014-3510]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3510
19859[CVE-2014-3509]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3509
19860[CVE-2014-3508]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3508
19861[CVE-2014-3507]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3507
19862[CVE-2014-3506]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3506
19863[CVE-2014-3505]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3505
19864[CVE-2014-3470]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3470
19865[CVE-2014-0224]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-0224
19866[CVE-2014-0221]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-0221
19867[CVE-2014-0195]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-0195
19868[CVE-2014-0160]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-0160
19869[CVE-2014-0076]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-0076
19870[CVE-2013-6450]: https://www.openssl.org/news/vulnerabilities.html#CVE-2013-6450
19871[CVE-2013-4353]: https://www.openssl.org/news/vulnerabilities.html#CVE-2013-4353
19872[CVE-2013-0169]: https://www.openssl.org/news/vulnerabilities.html#CVE-2013-0169
19873[CVE-2013-0166]: https://www.openssl.org/news/vulnerabilities.html#CVE-2013-0166
19874[CVE-2012-2686]: https://www.openssl.org/news/vulnerabilities.html#CVE-2012-2686
19875[CVE-2012-2333]: https://www.openssl.org/news/vulnerabilities.html#CVE-2012-2333
19876[CVE-2012-2110]: https://www.openssl.org/news/vulnerabilities.html#CVE-2012-2110
19877[CVE-2012-0884]: https://www.openssl.org/news/vulnerabilities.html#CVE-2012-0884
19878[CVE-2012-0050]: https://www.openssl.org/news/vulnerabilities.html#CVE-2012-0050
19879[CVE-2012-0027]: https://www.openssl.org/news/vulnerabilities.html#CVE-2012-0027
19880[CVE-2011-4619]: https://www.openssl.org/news/vulnerabilities.html#CVE-2011-4619
19881[CVE-2011-4577]: https://www.openssl.org/news/vulnerabilities.html#CVE-2011-4577
19882[CVE-2011-4576]: https://www.openssl.org/news/vulnerabilities.html#CVE-2011-4576
19883[CVE-2011-4109]: https://www.openssl.org/news/vulnerabilities.html#CVE-2011-4109
19884[CVE-2011-4108]: https://www.openssl.org/news/vulnerabilities.html#CVE-2011-4108
19885[CVE-2011-3210]: https://www.openssl.org/news/vulnerabilities.html#CVE-2011-3210
19886[CVE-2011-3207]: https://www.openssl.org/news/vulnerabilities.html#CVE-2011-3207
19887[CVE-2011-0014]: https://www.openssl.org/news/vulnerabilities.html#CVE-2011-0014
19888[CVE-2010-4252]: https://www.openssl.org/news/vulnerabilities.html#CVE-2010-4252
19889[CVE-2010-4180]: https://www.openssl.org/news/vulnerabilities.html#CVE-2010-4180
19890[CVE-2010-3864]: https://www.openssl.org/news/vulnerabilities.html#CVE-2010-3864
19891[CVE-2010-1633]: https://www.openssl.org/news/vulnerabilities.html#CVE-2010-1633
19892[CVE-2010-0740]: https://www.openssl.org/news/vulnerabilities.html#CVE-2010-0740
19893[CVE-2010-0433]: https://www.openssl.org/news/vulnerabilities.html#CVE-2010-0433
19894[CVE-2009-4355]: https://www.openssl.org/news/vulnerabilities.html#CVE-2009-4355
19895[CVE-2009-3555]: https://www.openssl.org/news/vulnerabilities.html#CVE-2009-3555
19896[CVE-2009-3245]: https://www.openssl.org/news/vulnerabilities.html#CVE-2009-3245
19897[CVE-2009-1386]: https://www.openssl.org/news/vulnerabilities.html#CVE-2009-1386
19898[CVE-2009-1379]: https://www.openssl.org/news/vulnerabilities.html#CVE-2009-1379
19899[CVE-2009-1378]: https://www.openssl.org/news/vulnerabilities.html#CVE-2009-1378
19900[CVE-2009-1377]: https://www.openssl.org/news/vulnerabilities.html#CVE-2009-1377
19901[CVE-2009-0789]: https://www.openssl.org/news/vulnerabilities.html#CVE-2009-0789
19902[CVE-2009-0591]: https://www.openssl.org/news/vulnerabilities.html#CVE-2009-0591
19903[CVE-2009-0590]: https://www.openssl.org/news/vulnerabilities.html#CVE-2009-0590
19904[CVE-2008-5077]: https://www.openssl.org/news/vulnerabilities.html#CVE-2008-5077
19905[CVE-2008-1678]: https://www.openssl.org/news/vulnerabilities.html#CVE-2008-1678
19906[CVE-2008-1672]: https://www.openssl.org/news/vulnerabilities.html#CVE-2008-1672
19907[CVE-2008-0891]: https://www.openssl.org/news/vulnerabilities.html#CVE-2008-0891
19908[CVE-2007-5135]: https://www.openssl.org/news/vulnerabilities.html#CVE-2007-5135
19909[CVE-2007-4995]: https://www.openssl.org/news/vulnerabilities.html#CVE-2007-4995
19910[CVE-2006-4343]: https://www.openssl.org/news/vulnerabilities.html#CVE-2006-4343
19911[CVE-2006-4339]: https://www.openssl.org/news/vulnerabilities.html#CVE-2006-4339
19912[CVE-2006-3738]: https://www.openssl.org/news/vulnerabilities.html#CVE-2006-3738
19913[CVE-2006-2940]: https://www.openssl.org/news/vulnerabilities.html#CVE-2006-2940
19914[CVE-2006-2937]: https://www.openssl.org/news/vulnerabilities.html#CVE-2006-2937
19915[CVE-2005-2969]: https://www.openssl.org/news/vulnerabilities.html#CVE-2005-2969
19916[CVE-2004-0112]: https://www.openssl.org/news/vulnerabilities.html#CVE-2004-0112
19917[CVE-2004-0079]: https://www.openssl.org/news/vulnerabilities.html#CVE-2004-0079
19918[CVE-2003-0851]: https://www.openssl.org/news/vulnerabilities.html#CVE-2003-0851
19919[CVE-2003-0545]: https://www.openssl.org/news/vulnerabilities.html#CVE-2003-0545
19920[CVE-2003-0544]: https://www.openssl.org/news/vulnerabilities.html#CVE-2003-0544
19921[CVE-2003-0543]: https://www.openssl.org/news/vulnerabilities.html#CVE-2003-0543
19922[CVE-2003-0078]: https://www.openssl.org/news/vulnerabilities.html#CVE-2003-0078
19923[CVE-2002-0659]: https://www.openssl.org/news/vulnerabilities.html#CVE-2002-0659
19924[CVE-2002-0657]: https://www.openssl.org/news/vulnerabilities.html#CVE-2002-0657
19925[CVE-2002-0656]: https://www.openssl.org/news/vulnerabilities.html#CVE-2002-0656
19926[CVE-2002-0655]: https://www.openssl.org/news/vulnerabilities.html#CVE-2002-0655
19927