xref: /netbsd-src/crypto/external/bsd/openssl/dist/CHANGES.md (revision 06dfa8449cb5e76c0044ec0f3badf7d5180af0f5)
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 3.0.7 and 3.0.8 [7 Feb 2023]
32
33 * Fixed NULL dereference during PKCS7 data verification.
34
35   A NULL pointer can be dereferenced when signatures are being
36   verified on PKCS7 signed or signedAndEnveloped data. In case the hash
37   algorithm used for the signature is known to the OpenSSL library but
38   the implementation of the hash algorithm is not available the digest
39   initialization will fail. There is a missing check for the return
40   value from the initialization function which later leads to invalid
41   usage of the digest API most likely leading to a crash.
42   ([CVE-2023-0401])
43
44   PKCS7 data is processed by the SMIME library calls and also by the
45   time stamp (TS) library calls. The TLS implementation in OpenSSL does
46   not call these functions however third party applications would be
47   affected if they call these functions to verify signatures on untrusted
48   data.
49
50   *Tomáš Mráz*
51
52 * Fixed X.400 address type confusion in X.509 GeneralName.
53
54   There is a type confusion vulnerability relating to X.400 address processing
55   inside an X.509 GeneralName. X.400 addresses were parsed as an ASN1_STRING
56   but the public structure definition for GENERAL_NAME incorrectly specified
57   the type of the x400Address field as ASN1_TYPE. This field is subsequently
58   interpreted by the OpenSSL function GENERAL_NAME_cmp as an ASN1_TYPE rather
59   than an ASN1_STRING.
60
61   When CRL checking is enabled (i.e. the application sets the
62   X509_V_FLAG_CRL_CHECK flag), this vulnerability may allow an attacker to
63   pass arbitrary pointers to a memcmp call, enabling them to read memory
64   contents or enact a denial of service.
65   ([CVE-2023-0286])
66
67   *Hugo Landau*
68
69 * Fixed NULL dereference validating DSA public key.
70
71   An invalid pointer dereference on read can be triggered when an
72   application tries to check a malformed DSA public key by the
73   EVP_PKEY_public_check() function. This will most likely lead
74   to an application crash. This function can be called on public
75   keys supplied from untrusted sources which could allow an attacker
76   to cause a denial of service attack.
77
78   The TLS implementation in OpenSSL does not call this function
79   but applications might call the function if there are additional
80   security requirements imposed by standards such as FIPS 140-3.
81   ([CVE-2023-0217])
82
83   *Shane Lontis, Tomáš Mráz*
84
85 * Fixed Invalid pointer dereference in d2i_PKCS7 functions.
86
87   An invalid pointer dereference on read can be triggered when an
88   application tries to load malformed PKCS7 data with the
89   d2i_PKCS7(), d2i_PKCS7_bio() or d2i_PKCS7_fp() functions.
90
91   The result of the dereference is an application crash which could
92   lead to a denial of service attack. The TLS implementation in OpenSSL
93   does not call this function however third party applications might
94   call these functions on untrusted data.
95   ([CVE-2023-0216])
96
97   *Tomáš Mráz*
98
99 * Fixed Use-after-free following BIO_new_NDEF.
100
101   The public API function BIO_new_NDEF is a helper function used for
102   streaming ASN.1 data via a BIO. It is primarily used internally to OpenSSL
103   to support the SMIME, CMS and PKCS7 streaming capabilities, but may also
104   be called directly by end user applications.
105
106   The function receives a BIO from the caller, prepends a new BIO_f_asn1
107   filter BIO onto the front of it to form a BIO chain, and then returns
108   the new head of the BIO chain to the caller. Under certain conditions,
109   for example if a CMS recipient public key is invalid, the new filter BIO
110   is freed and the function returns a NULL result indicating a failure.
111   However, in this case, the BIO chain is not properly cleaned up and the
112   BIO passed by the caller still retains internal pointers to the previously
113   freed filter BIO. If the caller then goes on to call BIO_pop() on the BIO
114   then a use-after-free will occur. This will most likely result in a crash.
115   ([CVE-2023-0215])
116
117   *Viktor Dukhovni, Matt Caswell*
118
119 * Fixed Double free after calling PEM_read_bio_ex.
120
121   The function PEM_read_bio_ex() reads a PEM file from a BIO and parses and
122   decodes the "name" (e.g. "CERTIFICATE"), any header data and the payload
123   data. If the function succeeds then the "name_out", "header" and "data"
124   arguments are populated with pointers to buffers containing the relevant
125   decoded data. The caller is responsible for freeing those buffers. It is
126   possible to construct a PEM file that results in 0 bytes of payload data.
127   In this case PEM_read_bio_ex() will return a failure code but will populate
128   the header argument with a pointer to a buffer that has already been freed.
129   If the caller also frees this buffer then a double free will occur. This
130   will most likely lead to a crash.
131
132   The functions PEM_read_bio() and PEM_read() are simple wrappers around
133   PEM_read_bio_ex() and therefore these functions are also directly affected.
134
135   These functions are also called indirectly by a number of other OpenSSL
136   functions including PEM_X509_INFO_read_bio_ex() and
137   SSL_CTX_use_serverinfo_file() which are also vulnerable. Some OpenSSL
138   internal uses of these functions are not vulnerable because the caller does
139   not free the header argument if PEM_read_bio_ex() returns a failure code.
140   ([CVE-2022-4450])
141
142   *Kurt Roeckx, Matt Caswell*
143
144 * Fixed Timing Oracle in RSA Decryption.
145
146   A timing based side channel exists in the OpenSSL RSA Decryption
147   implementation which could be sufficient to recover a plaintext across
148   a network in a Bleichenbacher style attack. To achieve a successful
149   decryption an attacker would have to be able to send a very large number
150   of trial messages for decryption. The vulnerability affects all RSA padding
151   modes: PKCS#1 v1.5, RSA-OEAP and RSASVE.
152   ([CVE-2022-4304])
153
154   *Dmitry Belyavsky, Hubert Kario*
155
156 * Fixed X.509 Name Constraints Read Buffer Overflow.
157
158   A read buffer overrun can be triggered in X.509 certificate verification,
159   specifically in name constraint checking. The read buffer overrun might
160   result in a crash which could lead to a denial of service attack.
161   In a TLS client, this can be triggered by connecting to a malicious
162   server. In a TLS server, this can be triggered if the server requests
163   client authentication and a malicious client connects.
164   ([CVE-2022-4203])
165
166   *Viktor Dukhovni*
167
168 * Fixed X.509 Policy Constraints Double Locking security issue.
169
170   If an X.509 certificate contains a malformed policy constraint and
171   policy processing is enabled, then a write lock will be taken twice
172   recursively.  On some operating systems (most widely: Windows) this
173   results in a denial of service when the affected process hangs.  Policy
174   processing being enabled on a publicly facing server is not considered
175   to be a common setup.
176   ([CVE-2022-3996])
177
178   *Paul Dale*
179
180 * Our provider implementations of `OSSL_FUNC_KEYMGMT_EXPORT` and
181   `OSSL_FUNC_KEYMGMT_GET_PARAMS` for EC and SM2 keys now honor
182   `OSSL_PKEY_PARAM_EC_POINT_CONVERSION_FORMAT` as set (and
183   default to `POINT_CONVERSION_UNCOMPRESSED`) when exporting
184   `OSSL_PKEY_PARAM_PUB_KEY`, instead of unconditionally using
185   `POINT_CONVERSION_COMPRESSED` as in previous 3.x releases.
186   For symmetry, our implementation of `EVP_PKEY_ASN1_METHOD->export_to`
187   for legacy EC and SM2 keys is also changed similarly to honor the
188   equivalent conversion format flag as specified in the underlying
189   `EC_KEY` object being exported to a provider, when this function is
190   called through `EVP_PKEY_export()`.
191
192   *Nicola Tuveri*
193
194### Changes between 3.0.6 and 3.0.7 [1 Nov 2022]
195
196 * Fixed two buffer overflows in punycode decoding functions.
197
198   A buffer overrun can be triggered in X.509 certificate verification,
199   specifically in name constraint checking. Note that this occurs after
200   certificate chain signature verification and requires either a CA to
201   have signed the malicious certificate or for the application to continue
202   certificate verification despite failure to construct a path to a trusted
203   issuer.
204
205   In a TLS client, this can be triggered by connecting to a malicious
206   server.  In a TLS server, this can be triggered if the server requests
207   client authentication and a malicious client connects.
208
209   An attacker can craft a malicious email address to overflow
210   an arbitrary number of bytes containing the `.`  character (decimal 46)
211   on the stack.  This buffer overflow could result in a crash (causing a
212   denial of service).
213   ([CVE-2022-3786])
214
215   An attacker can craft a malicious email address to overflow four
216   attacker-controlled bytes on the stack.  This buffer overflow could
217   result in a crash (causing a denial of service) or potentially remote code
218   execution depending on stack layout for any given platform/compiler.
219   ([CVE-2022-3602])
220
221   *Paul Dale*
222
223 * Removed all references to invalid OSSL_PKEY_PARAM_RSA names for CRT
224   parameters in OpenSSL code.
225   Applications should not use the names OSSL_PKEY_PARAM_RSA_FACTOR,
226   OSSL_PKEY_PARAM_RSA_EXPONENT and OSSL_PKEY_PARAM_RSA_COEFFICIENT.
227   Use the numbered names such as OSSL_PKEY_PARAM_RSA_FACTOR1 instead.
228   Using these invalid names may cause algorithms to use slower methods
229   that ignore the CRT parameters.
230
231   *Shane Lontis*
232
233 * Fixed a regression introduced in 3.0.6 version raising errors on some stack
234   operations.
235
236   *Tomáš Mráz*
237
238 * Fixed a regression introduced in 3.0.6 version not refreshing the certificate
239   data to be signed before signing the certificate.
240
241   *Gibeom Gwon*
242
243 * Added RIPEMD160 to the default provider.
244
245   *Paul Dale*
246
247 * Ensured that the key share group sent or accepted for the key exchange
248   is allowed for the protocol version.
249
250   *Matt Caswell*
251
252### Changes between 3.0.5 and 3.0.6 [11 Oct 2022]
253
254 * OpenSSL supports creating a custom cipher via the legacy
255   EVP_CIPHER_meth_new() function and associated function calls. This function
256   was deprecated in OpenSSL 3.0 and application authors are instead encouraged
257   to use the new provider mechanism in order to implement custom ciphers.
258
259   OpenSSL versions 3.0.0 to 3.0.5 incorrectly handle legacy custom ciphers
260   passed to the EVP_EncryptInit_ex2(), EVP_DecryptInit_ex2() and
261   EVP_CipherInit_ex2() functions (as well as other similarly named encryption
262   and decryption initialisation functions). Instead of using the custom cipher
263   directly it incorrectly tries to fetch an equivalent cipher from the
264   available providers. An equivalent cipher is found based on the NID passed to
265   EVP_CIPHER_meth_new(). This NID is supposed to represent the unique NID for a
266   given cipher. However it is possible for an application to incorrectly pass
267   NID_undef as this value in the call to EVP_CIPHER_meth_new(). When NID_undef
268   is used in this way the OpenSSL encryption/decryption initialisation function
269   will match the NULL cipher as being equivalent and will fetch this from the
270   available providers. This will succeed if the default provider has been
271   loaded (or if a third party provider has been loaded that offers this
272   cipher). Using the NULL cipher means that the plaintext is emitted as the
273   ciphertext.
274
275   Applications are only affected by this issue if they call
276   EVP_CIPHER_meth_new() using NID_undef and subsequently use it in a call to an
277   encryption/decryption initialisation function. Applications that only use
278   SSL/TLS are not impacted by this issue.
279   ([CVE-2022-3358])
280
281   *Matt Caswell*
282
283 * Fix LLVM vs Apple LLVM version numbering confusion that caused build failures
284   on MacOS 10.11
285
286   *Richard Levitte*
287
288 * Fixed the linux-mips64 Configure target which was missing the
289   SIXTY_FOUR_BIT bn_ops flag. This was causing heap corruption on that
290   platform.
291
292   *Adam Joseph*
293
294 * Fix handling of a ticket key callback that returns 0 in TLSv1.3 to not send a
295   ticket
296
297   *Matt Caswell*
298
299 * Correctly handle a retransmitted ClientHello in DTLS
300
301   *Matt Caswell*
302
303 * Fixed detection of ktls support in cross-compile environment on Linux
304
305   *Tomas Mraz*
306
307 * Fixed some regressions and test failures when running the 3.0.0 FIPS provider
308   against 3.0.x
309
310   *Paul Dale*
311
312 * Fixed SSL_pending() and SSL_has_pending() with DTLS which were failing to
313   report correct results in some cases
314
315   *Matt Caswell*
316
317 * Fix UWP builds by defining VirtualLock
318
319   *Charles Milette*
320
321 * For known safe primes use the minimum key length according to RFC 7919.
322   Longer private key sizes unnecessarily raise the cycles needed to compute the
323   shared secret without any increase of the real security. This fixes a
324   regression from 1.1.1 where these shorter keys were generated for the known
325   safe primes.
326
327   *Tomas Mraz*
328
329 * Added the loongarch64 target
330
331   *Shi Pujin*
332
333 * Fixed EC ASM flag passing. Flags for ASM implementations of EC curves were
334   only passed to the FIPS provider and not to the default or legacy provider.
335
336   *Juergen Christ*
337
338 * Fixed reported performance degradation on aarch64. Restored the
339   implementation prior to commit 2621751 ("aes/asm/aesv8-armx.pl: avoid
340   32-bit lane assignment in CTR mode") for 64bit targets only, since it is
341   reportedly 2-17% slower and the silicon errata only affects 32bit targets.
342   The new algorithm is still used for 32 bit targets.
343
344   *Bernd Edlinger*
345
346 * Added a missing header for memcmp that caused compilation failure on some
347   platforms
348
349   *Gregor Jasny*
350
351### Changes between 3.0.4 and 3.0.5 [5 Jul 2022]
352
353 * The OpenSSL 3.0.4 release introduced a serious bug in the RSA
354   implementation for X86_64 CPUs supporting the AVX512IFMA instructions.
355   This issue makes the RSA implementation with 2048 bit private keys
356   incorrect on such machines and memory corruption will happen during
357   the computation. As a consequence of the memory corruption an attacker
358   may be able to trigger a remote code execution on the machine performing
359   the computation.
360
361   SSL/TLS servers or other servers using 2048 bit RSA private keys running
362   on machines supporting AVX512IFMA instructions of the X86_64 architecture
363   are affected by this issue.
364   ([CVE-2022-2274])
365
366   *Xi Ruoyao*
367
368 * AES OCB mode for 32-bit x86 platforms using the AES-NI assembly optimised
369   implementation would not encrypt the entirety of the data under some
370   circumstances.  This could reveal sixteen bytes of data that was
371   preexisting in the memory that wasn't written.  In the special case of
372   "in place" encryption, sixteen bytes of the plaintext would be revealed.
373
374   Since OpenSSL does not support OCB based cipher suites for TLS and DTLS,
375   they are both unaffected.
376   ([CVE-2022-2097])
377
378   *Alex Chernyakhovsky, David Benjamin, Alejandro Sedeño*
379
380### Changes between 3.0.3 and 3.0.4 [21 Jun 2022]
381
382 * In addition to the c_rehash shell command injection identified in
383   CVE-2022-1292, further bugs where the c_rehash script does not
384   properly sanitise shell metacharacters to prevent command injection have been
385   fixed.
386
387   When the CVE-2022-1292 was fixed it was not discovered that there
388   are other places in the script where the file names of certificates
389   being hashed were possibly passed to a command executed through the shell.
390
391   This script is distributed by some operating systems in a manner where
392   it is automatically executed.  On such operating systems, an attacker
393   could execute arbitrary commands with the privileges of the script.
394
395   Use of the c_rehash script is considered obsolete and should be replaced
396   by the OpenSSL rehash command line tool.
397   (CVE-2022-2068)
398
399   *Daniel Fiala, Tomáš Mráz*
400
401 * Case insensitive string comparison no longer uses locales.  It has instead
402   been directly implemented.
403
404   *Paul Dale*
405
406### Changes between 3.0.2 and 3.0.3 [3 May 2022]
407
408 * Case insensitive string comparison is reimplemented via new locale-agnostic
409   comparison functions OPENSSL_str[n]casecmp always using the POSIX locale for
410   comparison. The previous implementation had problems when the Turkish locale
411   was used.
412
413   *Dmitry Belyavskiy*
414
415 * Fixed a bug in the c_rehash script which was not properly sanitising shell
416   metacharacters to prevent command injection.  This script is distributed by
417   some operating systems in a manner where it is automatically executed.  On
418   such operating systems, an attacker could execute arbitrary commands with the
419   privileges of the script.
420
421   Use of the c_rehash script is considered obsolete and should be replaced
422   by the OpenSSL rehash command line tool.
423   (CVE-2022-1292)
424
425   *Tomáš Mráz*
426
427 * Fixed a bug in the function `OCSP_basic_verify` that verifies the signer
428   certificate on an OCSP response. The bug caused the function in the case
429   where the (non-default) flag OCSP_NOCHECKS is used to return a postivie
430   response (meaning a successful verification) even in the case where the
431   response signing certificate fails to verify.
432
433   It is anticipated that most users of `OCSP_basic_verify` will not use the
434   OCSP_NOCHECKS flag. In this case the `OCSP_basic_verify` function will return
435   a negative value (indicating a fatal error) in the case of a certificate
436   verification failure. The normal expected return value in this case would be
437   0.
438
439   This issue also impacts the command line OpenSSL "ocsp" application. When
440   verifying an ocsp response with the "-no_cert_checks" option the command line
441   application will report that the verification is successful even though it
442   has in fact failed. In this case the incorrect successful response will also
443   be accompanied by error messages showing the failure and contradicting the
444   apparently successful result.
445   ([CVE-2022-1343])
446
447   *Matt Caswell*
448
449 * Fixed a bug where the RC4-MD5 ciphersuite incorrectly used the
450   AAD data as the MAC key. This made the MAC key trivially predictable.
451
452   An attacker could exploit this issue by performing a man-in-the-middle attack
453   to modify data being sent from one endpoint to an OpenSSL 3.0 recipient such
454   that the modified data would still pass the MAC integrity check.
455
456   Note that data sent from an OpenSSL 3.0 endpoint to a non-OpenSSL 3.0
457   endpoint will always be rejected by the recipient and the connection will
458   fail at that point. Many application protocols require data to be sent from
459   the client to the server first. Therefore, in such a case, only an OpenSSL
460   3.0 server would be impacted when talking to a non-OpenSSL 3.0 client.
461
462   If both endpoints are OpenSSL 3.0 then the attacker could modify data being
463   sent in both directions. In this case both clients and servers could be
464   affected, regardless of the application protocol.
465
466   Note that in the absence of an attacker this bug means that an OpenSSL 3.0
467   endpoint communicating with a non-OpenSSL 3.0 endpoint will fail to complete
468   the handshake when using this ciphersuite.
469
470   The confidentiality of data is not impacted by this issue, i.e. an attacker
471   cannot decrypt data that has been encrypted using this ciphersuite - they can
472   only modify it.
473
474   In order for this attack to work both endpoints must legitimately negotiate
475   the RC4-MD5 ciphersuite. This ciphersuite is not compiled by default in
476   OpenSSL 3.0, and is not available within the default provider or the default
477   ciphersuite list. This ciphersuite will never be used if TLSv1.3 has been
478   negotiated. In order for an OpenSSL 3.0 endpoint to use this ciphersuite the
479   following must have occurred:
480
481   1) OpenSSL must have been compiled with the (non-default) compile time option
482      enable-weak-ssl-ciphers
483
484   2) OpenSSL must have had the legacy provider explicitly loaded (either
485      through application code or via configuration)
486
487   3) The ciphersuite must have been explicitly added to the ciphersuite list
488
489   4) The libssl security level must have been set to 0 (default is 1)
490
491   5) A version of SSL/TLS below TLSv1.3 must have been negotiated
492
493   6) Both endpoints must negotiate the RC4-MD5 ciphersuite in preference to any
494      others that both endpoints have in common
495   (CVE-2022-1434)
496
497   *Matt Caswell*
498
499 * Fix a bug in the OPENSSL_LH_flush() function that breaks reuse of the memory
500   occuppied by the removed hash table entries.
501
502   This function is used when decoding certificates or keys. If a long lived
503   process periodically decodes certificates or keys its memory usage will
504   expand without bounds and the process might be terminated by the operating
505   system causing a denial of service. Also traversing the empty hash table
506   entries will take increasingly more time.
507
508   Typically such long lived processes might be TLS clients or TLS servers
509   configured to accept client certificate authentication.
510   (CVE-2022-1473)
511
512   *Hugo Landau, Aliaksei Levin*
513
514 * The functions `OPENSSL_LH_stats` and `OPENSSL_LH_stats_bio` now only report
515   the `num_items`, `num_nodes` and `num_alloc_nodes` statistics. All other
516   statistics are no longer supported. For compatibility, these statistics are
517   still listed in the output but are now always reported as zero.
518
519   *Hugo Landau*
520
521### Changes between 3.0.1 and 3.0.2 [15 Mar 2022]
522
523 * Fixed a bug in the BN_mod_sqrt() function that can cause it to loop forever
524   for non-prime moduli.
525
526   Internally this function is used when parsing certificates that contain
527   elliptic curve public keys in compressed form or explicit elliptic curve
528   parameters with a base point encoded in compressed form.
529
530   It is possible to trigger the infinite loop by crafting a certificate that
531   has invalid explicit curve parameters.
532
533   Since certificate parsing happens prior to verification of the certificate
534   signature, any process that parses an externally supplied certificate may thus
535   be subject to a denial of service attack. The infinite loop can also be
536   reached when parsing crafted private keys as they can contain explicit
537   elliptic curve parameters.
538
539   Thus vulnerable situations include:
540
541    - TLS clients consuming server certificates
542    - TLS servers consuming client certificates
543    - Hosting providers taking certificates or private keys from customers
544    - Certificate authorities parsing certification requests from subscribers
545    - Anything else which parses ASN.1 elliptic curve parameters
546
547   Also any other applications that use the BN_mod_sqrt() where the attacker
548   can control the parameter values are vulnerable to this DoS issue.
549   ([CVE-2022-0778])
550
551   *Tomáš Mráz*
552
553 * Add ciphersuites based on DHE_PSK (RFC 4279) and ECDHE_PSK (RFC 5489)
554   to the list of ciphersuites providing Perfect Forward Secrecy as
555   required by SECLEVEL >= 3.
556
557   *Dmitry Belyavskiy, Nicola Tuveri*
558
559 * Made the AES constant time code for no-asm configurations
560   optional due to the resulting 95% performance degradation.
561   The AES constant time code can be enabled, for no assembly
562   builds, with: ./config no-asm -DOPENSSL_AES_CONST_TIME
563
564   *Paul Dale*
565
566 * Fixed PEM_write_bio_PKCS8PrivateKey() to make it possible to use empty
567   passphrase strings.
568
569   *Darshan Sen*
570
571 * The negative return value handling of the certificate verification callback
572   was reverted. The replacement is to set the verification retry state with
573   the SSL_set_retry_verify() function.
574
575   *Tomáš Mráz*
576
577### Changes between 3.0.0 and 3.0.1 [14 Dec 2021]
578
579 * Fixed invalid handling of X509_verify_cert() internal errors in libssl
580   Internally libssl in OpenSSL calls X509_verify_cert() on the client side to
581   verify a certificate supplied by a server. That function may return a
582   negative return value to indicate an internal error (for example out of
583   memory). Such a negative return value is mishandled by OpenSSL and will cause
584   an IO function (such as SSL_connect() or SSL_do_handshake()) to not indicate
585   success and a subsequent call to SSL_get_error() to return the value
586   SSL_ERROR_WANT_RETRY_VERIFY. This return value is only supposed to be
587   returned by OpenSSL if the application has previously called
588   SSL_CTX_set_cert_verify_callback(). Since most applications do not do this
589   the SSL_ERROR_WANT_RETRY_VERIFY return value from SSL_get_error() will be
590   totally unexpected and applications may not behave correctly as a result. The
591   exact behaviour will depend on the application but it could result in
592   crashes, infinite loops or other similar incorrect responses.
593
594   This issue is made more serious in combination with a separate bug in OpenSSL
595   3.0 that will cause X509_verify_cert() to indicate an internal error when
596   processing a certificate chain. This will occur where a certificate does not
597   include the Subject Alternative Name extension but where a Certificate
598   Authority has enforced name constraints. This issue can occur even with valid
599   chains.
600   ([CVE-2021-4044])
601
602   *Matt Caswell*
603
604 * Corrected a few file name and file reference bugs in the build,
605   installation and setup scripts, which lead to installation verification
606   failures.  Slightly enhanced the installation verification script.
607
608   *Richard Levitte*
609
610 * Fixed EVP_PKEY_eq() to make it possible to use it with strictly private
611   keys.
612
613   *Richard Levitte*
614
615 * Fixed PVK encoder to properly query for the passphrase.
616
617   *Tomáš Mráz*
618
619 * Multiple fixes in the OSSL_HTTP API functions.
620
621   *David von Oheimb*
622
623 * Allow sign extension in OSSL_PARAM_allocate_from_text() for the
624   OSSL_PARAM_INTEGER data type and return error on negative numbers
625   used with the OSSL_PARAM_UNSIGNED_INTEGER data type. Make
626   OSSL_PARAM_BLD_push_BN{,_pad}() return an error on negative numbers.
627
628   *Richard Levitte*
629
630 * Allow copying uninitialized digest contexts with EVP_MD_CTX_copy_ex.
631
632   *Tomáš Mráz*
633
634 * Fixed detection of ARMv7 and ARM64 CPU features on FreeBSD.
635
636   *Allan Jude*
637
638 * Multiple threading fixes.
639
640   *Matt Caswell*
641
642 * Added NULL digest implementation to keep compatibility with 1.1.1 version.
643
644   *Tomáš Mráz*
645
646 * Allow fetching an operation from the provider that owns an unexportable key
647   as a fallback if that is still allowed by the property query.
648
649   *Richard Levitte*
650
651### Changes between 1.1.1 and 3.0.0 [7 sep 2021]
652
653 * TLS_MAX_VERSION, DTLS_MAX_VERSION and DTLS_MIN_VERSION constants are now
654   deprecated.
655
656   *Matt Caswell*
657
658 * The `OPENSSL_s390xcap` environment variable can be used to set bits in the
659   S390X capability vector to zero. This simplifies testing of different code
660   paths on S390X architecture.
661
662   *Patrick Steuer*
663
664 * Encrypting more than 2^64 TLS records with AES-GCM is disallowed
665   as per FIPS 140-2 IG A.5 "Key/IV Pair Uniqueness Requirements from
666   SP 800-38D". The communication will fail at this point.
667
668   *Paul Dale*
669
670 * The EC_GROUP_clear_free() function is deprecated as there is nothing
671   confidential in EC_GROUP data.
672
673   *Nicola Tuveri*
674
675 * The byte order mark (BOM) character is ignored if encountered at the
676   beginning of a PEM-formatted file.
677
678   *Dmitry Belyavskiy*
679
680 * Added CMS support for the Russian GOST algorithms.
681
682   *Dmitry Belyavskiy*
683
684 * Due to move of the implementation of cryptographic operations
685   to the providers, validation of various operation parameters can
686   be postponed until the actual operation is executed where previously
687   it happened immediately when an operation parameter was set.
688
689   For example when setting an unsupported curve with
690   EVP_PKEY_CTX_set_ec_paramgen_curve_nid() this function call will not
691   fail but later keygen operations with the EVP_PKEY_CTX will fail.
692
693   *OpenSSL team members and many third party contributors*
694
695 * The EVP_get_cipherbyname() function will return NULL for algorithms such as
696   "AES-128-SIV", "AES-128-CBC-CTS" and "CAMELLIA-128-CBC-CTS" which were
697   previously only accessible via low level interfaces. Use EVP_CIPHER_fetch()
698   instead to retrieve these algorithms from a provider.
699
700   *Shane Lontis*
701
702 * On build targets where the multilib postfix is set in the build
703   configuration the libdir directory was changing based on whether
704   the lib directory with the multilib postfix exists on the system
705   or not. This unpredictable behavior was removed and eventual
706   multilib postfix is now always added to the default libdir. Use
707   `--libdir=lib` to override the libdir if adding the postfix is
708   undesirable.
709
710   *Jan Lána*
711
712 * The triple DES key wrap functionality now conforms to RFC 3217 but is
713   no longer interoperable with OpenSSL 1.1.1.
714
715   *Paul Dale*
716
717 * The ERR_GET_FUNC() function was removed.  With the loss of meaningful
718   function codes, this function can only cause problems for calling
719   applications.
720
721   *Paul Dale*
722
723 * Add a configurable flag to output date formats as ISO 8601. Does not
724   change the default date format.
725
726   *William Edmisten*
727
728 * Version of MSVC earlier than 1300 could get link warnings, which could
729   be suppressed if the undocumented -DI_CAN_LIVE_WITH_LNK4049 was set.
730   Support for this flag has been removed.
731
732   *Rich Salz*
733
734 * Rework and make DEBUG macros consistent. Remove unused -DCONF_DEBUG,
735   -DBN_CTX_DEBUG, and REF_PRINT. Add a new tracing category and use it for
736   printing reference counts. Rename -DDEBUG_UNUSED to -DUNUSED_RESULT_DEBUG
737   Fix BN_DEBUG_RAND so it compiles and, when set, force DEBUG_RAND to be set
738   also. Rename engine_debug_ref to be ENGINE_REF_PRINT also for consistency.
739
740   *Rich Salz*
741
742 * The signatures of the functions to get and set options on SSL and
743   SSL_CTX objects changed from "unsigned long" to "uint64_t" type.
744   Some source code changes may be required.
745
746   *Rich Salz*
747
748 * The public definitions of conf_method_st and conf_st have been
749   deprecated. They will be made opaque in a future release.
750
751   *Rich Salz and Tomáš Mráz*
752
753 * Client-initiated renegotiation is disabled by default. To allow it, use
754   the -client_renegotiation option, the SSL_OP_ALLOW_CLIENT_RENEGOTIATION
755   flag, or the "ClientRenegotiation" config parameter as appropriate.
756
757   *Rich Salz*
758
759 * Add "abspath" and "includedir" pragma's to config files, to prevent,
760   or modify relative pathname inclusion.
761
762   *Rich Salz*
763
764 * OpenSSL includes a cryptographic module that is intended to be FIPS 140-2
765   validated. Please consult the README-FIPS and
766   README-PROVIDERS files, as well as the migration guide.
767
768   *OpenSSL team members and many third party contributors*
769
770 * For the key types DH and DHX the allowed settable parameters are now different.
771
772   *Shane Lontis*
773
774 * The openssl commands that read keys, certificates, and CRLs now
775   automatically detect the PEM or DER format of the input files.
776
777   *David von Oheimb, Richard Levitte, and Tomáš Mráz*
778
779 * Added enhanced PKCS#12 APIs which accept a library context.
780
781   *Jon Spillett*
782
783 * The default manual page suffix ($MANSUFFIX) has been changed to "ossl"
784
785   *Matt Caswell*
786
787 * Added support for Kernel TLS (KTLS).
788
789   *Boris Pismenny, John Baldwin and Andrew Gallatin*
790
791 * Support for RFC 5746 secure renegotiation is now required by default for
792   SSL or TLS connections to succeed.
793
794   *Benjamin Kaduk*
795
796 * The signature of the `copy` functional parameter of the
797   EVP_PKEY_meth_set_copy() function has changed so its `src` argument is
798   now `const EVP_PKEY_CTX *` instead of `EVP_PKEY_CTX *`. Similarly
799   the signature of the `pub_decode` functional parameter of the
800   EVP_PKEY_asn1_set_public() function has changed so its `pub` argument is
801   now `const X509_PUBKEY *` instead of `X509_PUBKEY *`.
802
803   *David von Oheimb*
804
805 * The error return values from some control calls (ctrl) have changed.
806
807   *Paul Dale*
808
809 * A public key check is now performed during EVP_PKEY_derive_set_peer().
810
811   *Shane Lontis*
812
813 * Many functions in the EVP_ namespace that are getters of values from
814   implementations or contexts were renamed to include get or get0 in their
815   names. Old names are provided as macro aliases for compatibility and
816   are not deprecated.
817
818   *Tomáš Mráz*
819
820 * The EVP_PKEY_CTRL_PKCS7_ENCRYPT, EVP_PKEY_CTRL_PKCS7_DECRYPT,
821   EVP_PKEY_CTRL_PKCS7_SIGN, EVP_PKEY_CTRL_CMS_ENCRYPT,
822   EVP_PKEY_CTRL_CMS_DECRYPT, and EVP_PKEY_CTRL_CMS_SIGN control operations
823   are deprecated.
824
825   *Tomáš Mráz*
826
827 * The EVP_PKEY_public_check() and EVP_PKEY_param_check() functions now work for
828   more key types.
829
830 * The output from the command line applications may have minor
831   changes.
832
833   *Paul Dale*
834
835 * The output from numerous "printing" may have minor changes.
836
837   *David von Oheimb*
838
839 * Windows thread synchronization uses read/write primitives (SRWLock) when
840   supported by the OS, otherwise CriticalSection continues to be used.
841
842   *Vincent Drake*
843
844 * Add filter BIO BIO_f_readbuffer() that allows BIO_tell() and BIO_seek() to
845   work on read only BIO source/sinks that do not support these functions.
846   This allows piping or redirection of a file BIO using stdin to be buffered
847   into memory. This is used internally in OSSL_DECODER_from_bio().
848
849   *Shane Lontis*
850
851 * OSSL_STORE_INFO_get_type() may now return an additional value. In 1.1.1
852   this function would return one of the values OSSL_STORE_INFO_NAME,
853   OSSL_STORE_INFO_PKEY, OSSL_STORE_INFO_PARAMS, OSSL_STORE_INFO_CERT or
854   OSSL_STORE_INFO_CRL. Decoded public keys would previously have been reported
855   as type OSSL_STORE_INFO_PKEY in 1.1.1. In 3.0 decoded public keys are now
856   reported as having the new type OSSL_STORE_INFO_PUBKEY. Applications
857   using this function should be amended to handle the changed return value.
858
859   *Richard Levitte*
860
861 * Improved adherence to Enhanced Security Services (ESS, RFC 2634 and RFC 5035)
862   for the TSP and CMS Advanced Electronic Signatures (CAdES) implementations.
863   As required by RFC 5035 check both ESSCertID and ESSCertIDv2 if both present.
864   Correct the semantics of checking the validation chain in case ESSCertID{,v2}
865   contains more than one certificate identifier: This means that all
866   certificates referenced there MUST be part of the validation chain.
867
868   *David von Oheimb*
869
870 * The implementation of older EVP ciphers related to CAST, IDEA, SEED, RC2, RC4,
871   RC5, DESX and DES have been moved to the legacy provider.
872
873   *Matt Caswell*
874
875 * The implementation of the EVP digests MD2, MD4, MDC2, WHIRLPOOL and
876   RIPEMD-160 have been moved to the legacy provider.
877
878   *Matt Caswell*
879
880 * The deprecated function EVP_PKEY_get0() now returns NULL being called for a
881   provided key.
882
883   *Dmitry Belyavskiy*
884
885 * The deprecated functions EVP_PKEY_get0_RSA(),
886   EVP_PKEY_get0_DSA(), EVP_PKEY_get0_EC_KEY(), EVP_PKEY_get0_DH(),
887   EVP_PKEY_get0_hmac(), EVP_PKEY_get0_poly1305() and EVP_PKEY_get0_siphash() as
888   well as the similarly named "get1" functions behave differently in
889   OpenSSL 3.0.
890
891   *Matt Caswell*
892
893 * A number of functions handling low-level keys or engines were deprecated
894   including EVP_PKEY_set1_engine(), EVP_PKEY_get0_engine(), EVP_PKEY_assign(),
895   EVP_PKEY_get0(), EVP_PKEY_get0_hmac(), EVP_PKEY_get0_poly1305() and
896   EVP_PKEY_get0_siphash().
897
898   *Matt Caswell*
899
900 * PKCS#5 PBKDF1 key derivation has been moved from PKCS5_PBE_keyivgen() into
901   the legacy crypto provider as an EVP_KDF. Applications requiring this KDF
902   will need to load the legacy crypto provider. This includes these PBE
903   algorithms which use this KDF:
904   - NID_pbeWithMD2AndDES_CBC
905   - NID_pbeWithMD5AndDES_CBC
906   - NID_pbeWithSHA1AndRC2_CBC
907   - NID_pbeWithMD2AndRC2_CBC
908   - NID_pbeWithMD5AndRC2_CBC
909   - NID_pbeWithSHA1AndDES_CBC
910
911   *Jon Spillett*
912
913 * Deprecated obsolete BIO_set_callback(), BIO_get_callback(), and
914   BIO_debug_callback() functions.
915
916   *Tomáš Mráz*
917
918 * Deprecated obsolete EVP_PKEY_CTX_get0_dh_kdf_ukm() and
919   EVP_PKEY_CTX_get0_ecdh_kdf_ukm() functions.
920
921   *Tomáš Mráz*
922
923 * The RAND_METHOD APIs have been deprecated.
924
925   *Paul Dale*
926
927 * The SRP APIs have been deprecated.
928
929   *Matt Caswell*
930
931 * Add a compile time option to prevent the caching of provider fetched
932   algorithms.  This is enabled by including the no-cached-fetch option
933   at configuration time.
934
935   *Paul Dale*
936
937 * pkcs12 now uses defaults of PBKDF2, AES and SHA-256, with a MAC iteration
938   count of PKCS12_DEFAULT_ITER.
939
940   *Tomáš Mráz and Sahana Prasad*
941
942 * The openssl speed command does not use low-level API calls anymore.
943
944   *Tomáš Mráz*
945
946 * Parallel dual-prime 1024-bit modular exponentiation for AVX512_IFMA
947   capable processors.
948
949   *Ilya Albrekht, Sergey Kirillov, Andrey Matyukov (Intel Corp)*
950
951 * Combining the Configure options no-ec and no-dh no longer disables TLSv1.3.
952
953   *Matt Caswell*
954
955 * Implemented support for fully "pluggable" TLSv1.3 groups. This means that
956   providers may supply their own group implementations (using either the "key
957   exchange" or the "key encapsulation" methods) which will automatically be
958   detected and used by libssl.
959
960   *Matt Caswell, Nicola Tuveri*
961
962 * The undocumented function X509_certificate_type() has been deprecated;
963
964   *Rich Salz*
965
966 * Deprecated the obsolete BN_pseudo_rand() and BN_pseudo_rand_range().
967
968   *Tomáš Mráz*
969
970 * Removed RSA padding mode for SSLv23 (which was only used for
971   SSLv2). This includes the functions RSA_padding_check_SSLv23() and
972   RSA_padding_add_SSLv23() and the `-ssl` option in the deprecated
973   `rsautl` command.
974
975   *Rich Salz*
976
977 * Deprecated the obsolete X9.31 RSA key generation related functions.
978
979 * While a callback function set via `SSL_CTX_set_cert_verify_callback()`
980   is not allowed to return a value > 1, this is no more taken as failure.
981
982   *Viktor Dukhovni and David von Oheimb*
983
984 * Deprecated the obsolete X9.31 RSA key generation related functions
985   BN_X931_generate_Xpq(), BN_X931_derive_prime_ex(), and
986   BN_X931_generate_prime_ex().
987
988   *Tomáš Mráz*
989
990 * The default key generation method for the regular 2-prime RSA keys was
991   changed to the FIPS 186-4 B.3.6 method.
992
993   *Shane Lontis*
994
995 * Deprecated the BN_is_prime_ex() and BN_is_prime_fasttest_ex() functions.
996
997   *Kurt Roeckx*
998
999 * Deprecated EVP_MD_CTX_set_update_fn() and EVP_MD_CTX_update_fn().
1000
1001   *Rich Salz*
1002
1003 * Deprecated the type OCSP_REQ_CTX and the functions OCSP_REQ_CTX_*() and
1004   replaced with OSSL_HTTP_REQ_CTX and the functions OSSL_HTTP_REQ_CTX_*().
1005
1006   *Rich Salz, Richard Levitte, and David von Oheimb*
1007
1008 * Deprecated `X509_http_nbio()` and `X509_CRL_http_nbio()`.
1009
1010   *David von Oheimb*
1011
1012 * Deprecated `OCSP_parse_url()`.
1013
1014   *David von Oheimb*
1015
1016 * Validation of SM2 keys has been separated from the validation of regular EC
1017   keys.
1018
1019   *Nicola Tuveri*
1020
1021 * Behavior of the `pkey` app is changed, when using the `-check` or `-pubcheck`
1022   switches: a validation failure triggers an early exit, returning a failure
1023   exit status to the parent process.
1024
1025   *Nicola Tuveri*
1026
1027 * Changed behavior of SSL_CTX_set_ciphersuites() and SSL_set_ciphersuites()
1028   to ignore unknown ciphers.
1029
1030   *Otto Hollmann*
1031
1032 * The `-cipher-commands` and `-digest-commands` options
1033   of the command line utility `list` have been deprecated.
1034   Instead use the `-cipher-algorithms` and `-digest-algorithms` options.
1035
1036   *Dmitry Belyavskiy*
1037
1038 * Added convenience functions for generating asymmetric key pairs:
1039   The 'quick' one-shot (yet somewhat limited) function L<EVP_PKEY_Q_keygen(3)>
1040   and macros for the most common cases: <EVP_RSA_gen(3)> and L<EVP_EC_gen(3)>.
1041
1042   *David von Oheimb*
1043
1044 * All of the low level EC_KEY functions have been deprecated.
1045
1046   *Shane Lontis, Paul Dale, Richard Levitte, and Tomáš Mráz*
1047
1048 * Deprecated all the libcrypto and libssl error string loading
1049   functions.
1050
1051   *Richard Levitte*
1052
1053 * The functions SSL_CTX_set_tmp_dh_callback and SSL_set_tmp_dh_callback, as
1054   well as the macros SSL_CTX_set_tmp_dh() and SSL_set_tmp_dh() have been
1055   deprecated.
1056
1057   *Matt Caswell*
1058
1059 * The `-crypt` option to the `passwd` command line tool has been removed.
1060
1061   *Paul Dale*
1062
1063 * The -C option to the `x509`, `dhparam`, `dsaparam`, and `ecparam` commands
1064   were removed.
1065
1066   *Rich Salz*
1067
1068 * Add support for AES Key Wrap inverse ciphers to the EVP layer.
1069
1070   *Shane Lontis*
1071
1072 * Deprecated EVP_PKEY_set1_tls_encodedpoint() and
1073   EVP_PKEY_get1_tls_encodedpoint().
1074
1075   *Matt Caswell*
1076
1077 * The security callback, which can be customised by application code, supports
1078   the security operation SSL_SECOP_TMP_DH. One location of the "other" parameter
1079   was incorrectly passing a DH object. It now passed an EVP_PKEY in all cases.
1080
1081   *Matt Caswell*
1082
1083 * Add PKCS7_get_octet_string() and PKCS7_type_is_other() to the public
1084   interface. Their functionality remains unchanged.
1085
1086   *Jordan Montgomery*
1087
1088 * Added new option for 'openssl list', '-providers', which will display the
1089   list of loaded providers, their names, version and status.  It optionally
1090   displays their gettable parameters.
1091
1092   *Paul Dale*
1093
1094 * Removed EVP_PKEY_set_alias_type().
1095
1096   *Richard Levitte*
1097
1098 * Deprecated `EVP_PKEY_CTX_set_rsa_keygen_pubexp()` and introduced
1099   `EVP_PKEY_CTX_set1_rsa_keygen_pubexp()`, which is now preferred.
1100
1101   *Jeremy Walch*
1102
1103 * Changed all "STACK" functions to be macros instead of inline functions. Macro
1104   parameters are still checked for type safety at compile time via helper
1105   inline functions.
1106
1107   *Matt Caswell*
1108
1109 * Remove the RAND_DRBG API
1110
1111   *Paul Dale and Matthias St. Pierre*
1112
1113 * Allow `SSL_set1_host()` and `SSL_add1_host()` to take IP literal addresses
1114   as well as actual hostnames.
1115
1116   *David Woodhouse*
1117
1118 * The 'MinProtocol' and 'MaxProtocol' configuration commands now silently
1119   ignore TLS protocol version bounds when configuring DTLS-based contexts, and
1120   conversely, silently ignore DTLS protocol version bounds when configuring
1121   TLS-based contexts.  The commands can be repeated to set bounds of both
1122   types.  The same applies with the corresponding "min_protocol" and
1123   "max_protocol" command-line switches, in case some application uses both TLS
1124   and DTLS.
1125
1126   SSL_CTX instances that are created for a fixed protocol version (e.g.
1127   `TLSv1_server_method()`) also silently ignore version bounds.  Previously
1128   attempts to apply bounds to these protocol versions would result in an
1129   error.  Now only the "version-flexible" SSL_CTX instances are subject to
1130   limits in configuration files in command-line options.
1131
1132   *Viktor Dukhovni*
1133
1134 * Deprecated the `ENGINE` API.  Engines should be replaced with providers
1135   going forward.
1136
1137   *Paul Dale*
1138
1139 * Reworked the recorded ERR codes to make better space for system errors.
1140   To distinguish them, the macro `ERR_SYSTEM_ERROR()` indicates if the
1141   given code is a system error (true) or an OpenSSL error (false).
1142
1143   *Richard Levitte*
1144
1145 * Reworked the test perl framework to better allow parallel testing.
1146
1147   *Nicola Tuveri and David von Oheimb*
1148
1149 * Added ciphertext stealing algorithms AES-128-CBC-CTS, AES-192-CBC-CTS and
1150   AES-256-CBC-CTS to the providers. CS1, CS2 and CS3 variants are supported.
1151
1152   *Shane Lontis*
1153
1154 * 'Configure' has been changed to figure out the configuration target if
1155   none is given on the command line.  Consequently, the 'config' script is
1156   now only a mere wrapper.  All documentation is changed to only mention
1157   'Configure'.
1158
1159   *Rich Salz and Richard Levitte*
1160
1161 * Added a library context `OSSL_LIB_CTX` that applications as well as
1162   other libraries can use to form a separate context within which
1163   libcrypto operations are performed.
1164
1165   *Richard Levitte*
1166
1167 * Added various `_ex` functions to the OpenSSL API that support using
1168   a non-default `OSSL_LIB_CTX`.
1169
1170   *OpenSSL team*
1171
1172 * Handshake now fails if Extended Master Secret extension is dropped
1173   on renegotiation.
1174
1175   *Tomáš Mráz*
1176
1177 * Dropped interactive mode from the `openssl` program.
1178
1179   *Richard Levitte*
1180
1181 * Deprecated `EVP_PKEY_cmp()` and `EVP_PKEY_cmp_parameters()`.
1182
1183   *David von Oheimb and Shane Lontis*
1184
1185 * Deprecated `EC_METHOD_get_field_type()`.
1186
1187   *Billy Bob Brumley*
1188
1189 * Deprecated EC_GFp_simple_method(), EC_GFp_mont_method(),
1190   EC_GF2m_simple_method(), EC_GFp_nist_method(), EC_GFp_nistp224_method()
1191   EC_GFp_nistp256_method(), and EC_GFp_nistp521_method().
1192
1193   *Billy Bob Brumley*
1194
1195 * Deprecated EC_GROUP_new(), EC_GROUP_method_of(), and EC_POINT_method_of().
1196
1197   *Billy Bob Brumley*
1198
1199 * Add CAdES-BES signature verification support, mostly derived
1200   from ESSCertIDv2 TS (RFC 5816) contribution by Marek Klein.
1201
1202   *Filipe Raimundo da Silva*
1203
1204 * Add CAdES-BES signature scheme and attributes support (RFC 5126) to CMS API.
1205
1206   *Antonio Iacono*
1207
1208 * Added the AuthEnvelopedData content type structure (RFC 5083) with AES-GCM
1209   parameter (RFC 5084) for the Cryptographic Message Syntax (CMS).
1210
1211   *Jakub Zelenka*
1212
1213 * Deprecated EC_POINT_make_affine() and EC_POINTs_make_affine().
1214
1215   *Billy Bob Brumley*
1216
1217 * Deprecated EC_GROUP_precompute_mult(), EC_GROUP_have_precompute_mult(), and
1218   EC_KEY_precompute_mult().
1219
1220   *Billy Bob Brumley*
1221
1222 * Deprecated EC_POINTs_mul().
1223
1224   *Billy Bob Brumley*
1225
1226 * Removed FIPS_mode() and FIPS_mode_set().
1227
1228   *Shane Lontis*
1229
1230 * The SSL option SSL_OP_IGNORE_UNEXPECTED_EOF is introduced.
1231
1232   *Dmitry Belyavskiy*
1233
1234 * Deprecated EC_POINT_set_Jprojective_coordinates_GFp() and
1235   EC_POINT_get_Jprojective_coordinates_GFp().
1236
1237   *Billy Bob Brumley*
1238
1239 * Added OSSL_PARAM_BLD to the public interface.  This allows OSSL_PARAM
1240   arrays to be more easily constructed via a series of utility functions.
1241   Create a parameter builder using OSSL_PARAM_BLD_new(), add parameters using
1242   the various push functions and finally convert to a passable OSSL_PARAM
1243   array using OSSL_PARAM_BLD_to_param().
1244
1245   *Paul Dale*
1246
1247 * The security strength of SHA1 and MD5 based signatures in TLS has been
1248   reduced.
1249
1250   *Kurt Roeckx*
1251
1252 * Added EVP_PKEY_set_type_by_keymgmt(), to initialise an EVP_PKEY to
1253   contain a provider side internal key.
1254
1255   *Richard Levitte*
1256
1257 * ASN1_verify(), ASN1_digest() and ASN1_sign() have been deprecated.
1258
1259   *Richard Levitte*
1260
1261 * Project text documents not yet having a proper file name extension
1262   (`HACKING`, `LICENSE`, `NOTES*`, `README*`, `VERSION`) have been renamed to
1263   `*.md` as far as reasonable, else `*.txt`, for better use with file managers.
1264
1265   *David von Oheimb*
1266
1267 * The main project documents (README, NEWS, CHANGES, INSTALL, SUPPORT)
1268   have been converted to Markdown with the goal to produce documents
1269   which not only look pretty when viewed online in the browser, but
1270   remain well readable inside a plain text editor.
1271
1272   To achieve this goal, a 'minimalistic' Markdown style has been applied
1273   which avoids formatting elements that interfere too much with the
1274   reading flow in the text file. For example, it
1275
1276   * avoids [ATX headings][] and uses [setext headings][] instead
1277     (which works for `<h1>` and `<h2>` headings only).
1278   * avoids [inline links][] and uses [reference links][] instead.
1279   * avoids [fenced code blocks][] and uses [indented code blocks][] instead.
1280
1281     [ATX headings]:         https://github.github.com/gfm/#atx-headings
1282     [setext headings]:      https://github.github.com/gfm/#setext-headings
1283     [inline links]:         https://github.github.com/gfm/#inline-link
1284     [reference links]:      https://github.github.com/gfm/#reference-link
1285     [fenced code blocks]:   https://github.github.com/gfm/#fenced-code-blocks
1286     [indented code blocks]: https://github.github.com/gfm/#indented-code-blocks
1287
1288   *Matthias St. Pierre*
1289
1290 * The test suite is changed to preserve results of each test recipe.
1291   A new directory test-runs/ with subdirectories named like the
1292   test recipes are created in the build tree for this purpose.
1293
1294   *Richard Levitte*
1295
1296 * Added an implementation of CMP and CRMF (RFC 4210, RFC 4211 RFC 6712).
1297   This adds `crypto/cmp/`, `crpyto/crmf/`, `apps/cmp.c`, and `test/cmp_*`.
1298   See L<openssl-cmp(1)> and L<OSSL_CMP_exec_IR_ses(3)> as starting points.
1299
1300   *David von Oheimb, Martin Peylo*
1301
1302 * Generalized the HTTP client code from `crypto/ocsp/` into `crpyto/http/`.
1303   It supports arbitrary request and response content types, GET redirection,
1304   TLS, connections via HTTP(S) proxies, connections and exchange via
1305   user-defined BIOs (allowing implicit connections), persistent connections,
1306   and timeout checks.  See L<OSSL_HTTP_transfer(3)> etc. for details.
1307   The legacy OCSP-focused (and only partly documented) API
1308   is retained for backward compatibility, while most of it is deprecated.
1309
1310   *David von Oheimb*
1311
1312 * Added `util/check-format.pl`, a tool for checking adherence to the
1313   OpenSSL coding style <https://www.openssl.org/policies/codingstyle.html>.
1314   The checks performed are incomplete and yield some false positives.
1315   Still the tool should be useful for detecting most typical glitches.
1316
1317   *David von Oheimb*
1318
1319 * `BIO_do_connect()` and `BIO_do_handshake()` have been extended:
1320   If domain name resolution yields multiple IP addresses all of them are tried
1321   after `connect()` failures.
1322
1323   *David von Oheimb*
1324
1325 * All of the low level RSA functions have been deprecated.
1326
1327   *Paul Dale*
1328
1329 * X509 certificates signed using SHA1 are no longer allowed at security
1330   level 1 and above.
1331
1332   *Kurt Roeckx*
1333
1334 * The command line utilities dhparam, dsa, gendsa and dsaparam have been
1335   modified to use PKEY APIs.  These commands are now in maintenance mode
1336   and no new features will be added to them.
1337
1338   *Paul Dale*
1339
1340 * The command line utility rsautl has been deprecated.
1341
1342   *Paul Dale*
1343
1344 * The command line utilities genrsa and rsa have been modified to use PKEY
1345   APIs. They now write PKCS#8 keys by default. These commands are now in
1346   maintenance mode and no new features will be added to them.
1347
1348   *Paul Dale*
1349
1350 * All of the low level DH functions have been deprecated.
1351
1352   *Paul Dale and Matt Caswell*
1353
1354 * All of the low level DSA functions have been deprecated.
1355
1356   *Paul Dale*
1357
1358 * Reworked the treatment of EC EVP_PKEYs with the SM2 curve to
1359   automatically become EVP_PKEY_SM2 rather than EVP_PKEY_EC.
1360
1361   *Richard Levitte*
1362
1363 * Deprecated low level ECDH and ECDSA functions.
1364
1365   *Paul Dale*
1366
1367 * Deprecated EVP_PKEY_decrypt_old() and EVP_PKEY_encrypt_old().
1368
1369   *Richard Levitte*
1370
1371 * Enhanced the documentation of EVP_PKEY_get_size(), EVP_PKEY_get_bits()
1372   and EVP_PKEY_get_security_bits().  Especially EVP_PKEY_get_size() needed
1373   a new formulation to include all the things it can be used for,
1374   as well as words of caution.
1375
1376   *Richard Levitte*
1377
1378 * The SSL_CTX_set_tlsext_ticket_key_cb(3) function has been deprecated.
1379
1380   *Paul Dale*
1381
1382 * All of the low level HMAC functions have been deprecated.
1383
1384   *Paul Dale and David von Oheimb*
1385
1386 * Over two thousand fixes were made to the documentation, including:
1387   - Common options (such as -rand/-writerand, TLS version control, etc)
1388     were refactored and point to newly-enhanced descriptions in openssl.pod.
1389   - Added style conformance for all options (with help from Richard Levitte),
1390     documented all reported missing options, added a CI build to check
1391     that all options are documented and that no unimplemented options
1392     are documented.
1393   - Documented some internals, such as all use of environment variables.
1394   - Addressed all internal broken L<> references.
1395
1396   *Rich Salz*
1397
1398 * All of the low level CMAC functions have been deprecated.
1399
1400   *Paul Dale*
1401
1402 * The low-level MD2, MD4, MD5, MDC2, RIPEMD160 and Whirlpool digest
1403   functions have been deprecated.
1404
1405   *Paul Dale and David von Oheimb*
1406
1407 * Corrected the documentation of the return values from the `EVP_DigestSign*`
1408   set of functions.  The documentation mentioned negative values for some
1409   errors, but this was never the case, so the mention of negative values
1410   was removed.
1411
1412   Code that followed the documentation and thereby check with something
1413   like `EVP_DigestSignInit(...) <= 0` will continue to work undisturbed.
1414
1415   *Richard Levitte*
1416
1417 * All of the low level cipher functions have been deprecated.
1418
1419   *Matt Caswell and Paul Dale*
1420
1421 * Removed include/openssl/opensslconf.h.in and replaced it with
1422   include/openssl/configuration.h.in, which differs in not including
1423   <openssl/macros.h>.  A short header include/openssl/opensslconf.h
1424   was added to include both.
1425
1426   This allows internal hacks where one might need to modify the set
1427   of configured macros, for example this if deprecated symbols are
1428   still supposed to be available internally:
1429
1430       #include <openssl/configuration.h>
1431
1432       #undef OPENSSL_NO_DEPRECATED
1433       #define OPENSSL_SUPPRESS_DEPRECATED
1434
1435       #include <openssl/macros.h>
1436
1437   This should not be used by applications that use the exported
1438   symbols, as that will lead to linking errors.
1439
1440   *Richard Levitte*
1441
1442 * Fixed an overflow bug in the x64_64 Montgomery squaring procedure
1443   used in exponentiation with 512-bit moduli. No EC algorithms are
1444   affected. Analysis suggests that attacks against 2-prime RSA1024,
1445   3-prime RSA1536, and DSA1024 as a result of this defect would be very
1446   difficult to perform and are not believed likely. Attacks against DH512
1447   are considered just feasible. However, for an attack the target would
1448   have to re-use the DH512 private key, which is not recommended anyway.
1449   Also applications directly using the low-level API BN_mod_exp may be
1450   affected if they use BN_FLG_CONSTTIME.
1451   ([CVE-2019-1551])
1452
1453   *Andy Polyakov*
1454
1455 * Most memory-debug features have been deprecated, and the functionality
1456   replaced with no-ops.
1457
1458   *Rich Salz*
1459
1460 * Added documentation for the STACK API.
1461
1462   *Rich Salz*
1463
1464 * Introduced a new method type and API, OSSL_ENCODER, to represent
1465   generic encoders.  These do the same sort of job that PEM writers
1466   and d2i functions do, but with support for methods supplied by
1467   providers, and the possibility for providers to support other
1468   formats as well.
1469
1470   *Richard Levitte*
1471
1472 * Introduced a new method type and API, OSSL_DECODER, to represent
1473   generic decoders.  These do the same sort of job that PEM readers
1474   and i2d functions do, but with support for methods supplied by
1475   providers, and the possibility for providers to support other
1476   formats as well.
1477
1478   *Richard Levitte*
1479
1480 * Added a .pragma directive to the syntax of configuration files, to
1481   allow varying behavior in a supported and predictable manner.
1482   Currently added pragma:
1483
1484           .pragma dollarid:on
1485
1486   This allows dollar signs to be a keyword character unless it's
1487   followed by a opening brace or parenthesis.  This is useful for
1488   platforms where dollar signs are commonly used in names, such as
1489   volume names and system directory names on VMS.
1490
1491   *Richard Levitte*
1492
1493 * Added functionality to create an EVP_PKEY from user data.
1494
1495   *Richard Levitte*
1496
1497 * Change the interpretation of the '--api' configuration option to
1498   mean that this is a desired API compatibility level with no
1499   further meaning.  The previous interpretation, that this would
1500   also mean to remove all deprecated symbols up to and including
1501   the given version, no requires that 'no-deprecated' is also used
1502   in the configuration.
1503
1504   When building applications, the desired API compatibility level
1505   can be set with the OPENSSL_API_COMPAT macro like before.  For
1506   API compatibility version below 3.0, the old style numerical
1507   value is valid as before, such as -DOPENSSL_API_COMPAT=0x10100000L.
1508   For version 3.0 and on, the value is expected to be the decimal
1509   value calculated from the major and minor version like this:
1510
1511           MAJOR * 10000 + MINOR * 100
1512
1513   Examples:
1514
1515           -DOPENSSL_API_COMPAT=30000             For 3.0
1516           -DOPENSSL_API_COMPAT=30200             For 3.2
1517
1518   To hide declarations that are deprecated up to and including the
1519   given API compatibility level, -DOPENSSL_NO_DEPRECATED must be
1520   given when building the application as well.
1521
1522   *Richard Levitte*
1523
1524 * Added the X509_LOOKUP_METHOD called X509_LOOKUP_store, to allow
1525   access to certificate and CRL stores via URIs and OSSL_STORE
1526   loaders.
1527
1528   This adds the following functions:
1529
1530   - X509_LOOKUP_store()
1531   - X509_STORE_load_file()
1532   - X509_STORE_load_path()
1533   - X509_STORE_load_store()
1534   - SSL_add_store_cert_subjects_to_stack()
1535   - SSL_CTX_set_default_verify_store()
1536   - SSL_CTX_load_verify_file()
1537   - SSL_CTX_load_verify_dir()
1538   - SSL_CTX_load_verify_store()
1539
1540   *Richard Levitte*
1541
1542 * Added a new method to gather entropy on VMS, based on SYS$GET_ENTROPY.
1543   The presence of this system service is determined at run-time.
1544
1545   *Richard Levitte*
1546
1547 * Added functionality to create an EVP_PKEY context based on data
1548   for methods from providers.  This takes an algorithm name and a
1549   property query string and simply stores them, with the intent
1550   that any operation that uses this context will use those strings
1551   to fetch the needed methods implicitly, thereby making the port
1552   of application written for pre-3.0 OpenSSL easier.
1553
1554   *Richard Levitte*
1555
1556 * The undocumented function NCONF_WIN32() has been deprecated; for
1557   conversion details see the HISTORY section of doc/man5/config.pod
1558
1559   *Rich Salz*
1560
1561 * Introduced the new functions EVP_DigestSignInit_ex() and
1562   EVP_DigestVerifyInit_ex(). The macros EVP_DigestSignUpdate() and
1563   EVP_DigestVerifyUpdate() have been converted to functions. See the man
1564   pages for further details.
1565
1566   *Matt Caswell*
1567
1568 * Over two thousand fixes were made to the documentation, including:
1569   adding missing command flags, better style conformance, documentation
1570   of internals, etc.
1571
1572   *Rich Salz, Richard Levitte*
1573
1574 * s390x assembly pack: add hardware-support for P-256, P-384, P-521,
1575   X25519, X448, Ed25519 and Ed448.
1576
1577   *Patrick Steuer*
1578
1579 * Print all values for a PKCS#12 attribute with 'openssl pkcs12', not just
1580   the first value.
1581
1582   *Jon Spillett*
1583
1584 * Deprecated the public definition of `ERR_STATE` as well as the function
1585   `ERR_get_state()`.  This is done in preparation of making `ERR_STATE` an
1586   opaque type.
1587
1588   *Richard Levitte*
1589
1590 * Added ERR functionality to give callers access to the stored function
1591   names that have replaced the older function code based functions.
1592
1593   New functions are ERR_peek_error_func(), ERR_peek_last_error_func(),
1594   ERR_peek_error_data(), ERR_peek_last_error_data(), ERR_get_error_all(),
1595   ERR_peek_error_all() and ERR_peek_last_error_all().
1596
1597   Deprecate ERR functions ERR_get_error_line(), ERR_get_error_line_data(),
1598   ERR_peek_error_line_data(), ERR_peek_last_error_line_data() and
1599   ERR_func_error_string().
1600
1601   *Richard Levitte*
1602
1603 * Extended testing to be verbose for failing tests only.  The make variables
1604   VERBOSE_FAILURE or VF can be used to enable this:
1605
1606           $ make VF=1 test                           # Unix
1607           $ mms /macro=(VF=1) test                   ! OpenVMS
1608           $ nmake VF=1 test                          # Windows
1609
1610   *Richard Levitte*
1611
1612 * Added the `-copy_extensions` option to the `x509` command for use with
1613   `-req` and `-x509toreq`. When given with the `copy` or `copyall` argument,
1614   all extensions in the request are copied to the certificate or vice versa.
1615
1616   *David von Oheimb*, *Kirill Stefanenkov <kirill_stefanenkov@rambler.ru>*
1617
1618 * Added the `-copy_extensions` option to the `req` command for use with
1619   `-x509`. When given with the `copy` or `copyall` argument,
1620   all extensions in the certification request are copied to the certificate.
1621
1622   *David von Oheimb*
1623
1624 * The `x509`, `req`, and `ca` commands now make sure that X.509v3 certificates
1625   they generate are by default RFC 5280 compliant in the following sense:
1626   There is a subjectKeyIdentifier extension with a hash value of the public key
1627   and for not self-signed certs there is an authorityKeyIdentifier extension
1628   with a keyIdentifier field or issuer information identifying the signing key.
1629   This is done unless some configuration overrides the new default behavior,
1630   such as `subjectKeyIdentifier = none` and `authorityKeyIdentifier = none`.
1631
1632   *David von Oheimb*
1633
1634 * Added several checks to `X509_verify_cert()` according to requirements in
1635   RFC 5280 in case `X509_V_FLAG_X509_STRICT` is set
1636   (which may be done by using the CLI option `-x509_strict`):
1637   * The basicConstraints of CA certificates must be marked critical.
1638   * CA certificates must explicitly include the keyUsage extension.
1639   * If a pathlenConstraint is given the key usage keyCertSign must be allowed.
1640   * The issuer name of any certificate must not be empty.
1641   * The subject name of CA certs, certs with keyUsage crlSign,
1642     and certs without subjectAlternativeName must not be empty.
1643   * If a subjectAlternativeName extension is given it must not be empty.
1644   * The signatureAlgorithm field and the cert signature must be consistent.
1645   * Any given authorityKeyIdentifier and any given subjectKeyIdentifier
1646     must not be marked critical.
1647   * The authorityKeyIdentifier must be given for X.509v3 certs
1648     unless they are self-signed.
1649   * The subjectKeyIdentifier must be given for all X.509v3 CA certs.
1650
1651   *David von Oheimb*
1652
1653 * Certificate verification using `X509_verify_cert()` meanwhile rejects EC keys
1654   with explicit curve parameters (specifiedCurve) as required by RFC 5480.
1655
1656   *Tomáš Mráz*
1657
1658 * For built-in EC curves, ensure an EC_GROUP built from the curve name is
1659   used even when parsing explicit parameters, when loading a encoded key
1660   or calling `EC_GROUP_new_from_ecpkparameters()`/
1661   `EC_GROUP_new_from_ecparameters()`.
1662   This prevents bypass of security hardening and performance gains,
1663   especially for curves with specialized EC_METHODs.
1664   By default, if a key encoded with explicit parameters is loaded and later
1665   encoded, the output is still encoded with explicit parameters, even if
1666   internally a "named" EC_GROUP is used for computation.
1667
1668   *Nicola Tuveri*
1669
1670 * Compute ECC cofactors if not provided during EC_GROUP construction. Before
1671   this change, EC_GROUP_set_generator would accept order and/or cofactor as
1672   NULL. After this change, only the cofactor parameter can be NULL. It also
1673   does some minimal sanity checks on the passed order.
1674   ([CVE-2019-1547])
1675
1676   *Billy Bob Brumley*
1677
1678 * Fixed a padding oracle in PKCS7_dataDecode and CMS_decrypt_set1_pkey.
1679   An attack is simple, if the first CMS_recipientInfo is valid but the
1680   second CMS_recipientInfo is chosen ciphertext. If the second
1681   recipientInfo decodes to PKCS #1 v1.5 form plaintext, the correct
1682   encryption key will be replaced by garbage, and the message cannot be
1683   decoded, but if the RSA decryption fails, the correct encryption key is
1684   used and the recipient will not notice the attack.
1685   As a work around for this potential attack the length of the decrypted
1686   key must be equal to the cipher default key length, in case the
1687   certifiate is not given and all recipientInfo are tried out.
1688   The old behaviour can be re-enabled in the CMS code by setting the
1689   CMS_DEBUG_DECRYPT flag.
1690
1691   *Bernd Edlinger*
1692
1693 * Early start up entropy quality from the DEVRANDOM seed source has been
1694   improved for older Linux systems.  The RAND subsystem will wait for
1695   /dev/random to be producing output before seeding from /dev/urandom.
1696   The seeded state is stored for future library initialisations using
1697   a system global shared memory segment.  The shared memory identifier
1698   can be configured by defining OPENSSL_RAND_SEED_DEVRANDOM_SHM_ID to
1699   the desired value.  The default identifier is 114.
1700
1701   *Paul Dale*
1702
1703 * Revised BN_generate_prime_ex to not avoid factors 2..17863 in p-1
1704   when primes for RSA keys are computed.
1705   Since we previously always generated primes == 2 (mod 3) for RSA keys,
1706   the 2-prime and 3-prime RSA modules were easy to distinguish, since
1707   `N = p*q = 1 (mod 3)`, but `N = p*q*r = 2 (mod 3)`. Therefore fingerprinting
1708   2-prime vs. 3-prime RSA keys was possible by computing N mod 3.
1709   This avoids possible fingerprinting of newly generated RSA modules.
1710
1711   *Bernd Edlinger*
1712
1713 * Correct the extended master secret constant on EBCDIC systems. Without this
1714   fix TLS connections between an EBCDIC system and a non-EBCDIC system that
1715   negotiate EMS will fail. Unfortunately this also means that TLS connections
1716   between EBCDIC systems with this fix, and EBCDIC systems without this
1717   fix will fail if they negotiate EMS.
1718
1719   *Matt Caswell*
1720
1721 * Changed the library initialisation so that the config file is now loaded
1722   by default. This was already the case for libssl. It now occurs for both
1723   libcrypto and libssl. Use the OPENSSL_INIT_NO_LOAD_CONFIG option to
1724   `OPENSSL_init_crypto()` to suppress automatic loading of a config file.
1725
1726   *Matt Caswell*
1727
1728 * Introduced new error raising macros, `ERR_raise()` and `ERR_raise_data()`,
1729   where the former acts as a replacement for `ERR_put_error()`, and the
1730   latter replaces the combination `ERR_put_error()` + `ERR_add_error_data()`.
1731   `ERR_raise_data()` adds more flexibility by taking a format string and
1732   an arbitrary number of arguments following it, to be processed with
1733   `BIO_snprintf()`.
1734
1735   *Richard Levitte*
1736
1737 * Introduced a new function, `OSSL_PROVIDER_available()`, which can be used
1738   to check if a named provider is loaded and available.  When called, it
1739   will also activate all fallback providers if such are still present.
1740
1741   *Richard Levitte*
1742
1743 * Enforce a minimum DH modulus size of 512 bits.
1744
1745   *Bernd Edlinger*
1746
1747 * Changed DH parameters to generate the order q subgroup instead of 2q.
1748   Previously generated DH parameters are still accepted by DH_check
1749   but DH_generate_key works around that by clearing bit 0 of the
1750   private key for those. This avoids leaking bit 0 of the private key.
1751
1752   *Bernd Edlinger*
1753
1754 * Significantly reduce secure memory usage by the randomness pools.
1755
1756   *Paul Dale*
1757
1758 * `{CRYPTO,OPENSSL}_mem_debug_{push,pop}` are now no-ops and have been
1759   deprecated.
1760
1761   *Rich Salz*
1762
1763 * A new type, EVP_KEYEXCH, has been introduced to represent key exchange
1764   algorithms. An implementation of a key exchange algorithm can be obtained
1765   by using the function EVP_KEYEXCH_fetch(). An EVP_KEYEXCH algorithm can be
1766   used in a call to EVP_PKEY_derive_init_ex() which works in a similar way to
1767   the older EVP_PKEY_derive_init() function. See the man pages for the new
1768   functions for further details.
1769
1770   *Matt Caswell*
1771
1772 * The EVP_PKEY_CTX_set_dh_pad() macro has now been converted to a function.
1773
1774   *Matt Caswell*
1775
1776 * Removed the function names from error messages and deprecated the
1777   xxx_F_xxx define's.
1778
1779   *Richard Levitte*
1780
1781 * Removed NextStep support and the macro OPENSSL_UNISTD
1782
1783   *Rich Salz*
1784
1785 * Removed DES_check_key.  Also removed OPENSSL_IMPLEMENT_GLOBAL,
1786   OPENSSL_GLOBAL_REF, OPENSSL_DECLARE_GLOBAL.
1787   Also removed "export var as function" capability; we do not export
1788   variables, only functions.
1789
1790   *Rich Salz*
1791
1792 * RC5_32_set_key has been changed to return an int type, with 0 indicating
1793   an error and 1 indicating success. In previous versions of OpenSSL this
1794   was a void type. If a key was set longer than the maximum possible this
1795   would crash.
1796
1797   *Matt Caswell*
1798
1799 * Support SM2 signing and verification schemes with X509 certificate.
1800
1801   *Paul Yang*
1802
1803 * Use SHA256 as the default digest for TS query in the `ts` app.
1804
1805   *Tomáš Mráz*
1806
1807 * Change PBKDF2 to conform to SP800-132 instead of the older PKCS5 RFC2898.
1808
1809   *Shane Lontis*
1810
1811 * Default cipher lists/suites are now available via a function, the
1812   #defines are deprecated.
1813
1814   *Todd Short*
1815
1816 * Add target VC-WIN32-UWP, VC-WIN64A-UWP, VC-WIN32-ARM-UWP and
1817   VC-WIN64-ARM-UWP in Windows OneCore target for making building libraries
1818   for Windows Store apps easier. Also, the "no-uplink" option has been added.
1819
1820   *Kenji Mouri*
1821
1822 * Join the directories crypto/x509 and crypto/x509v3
1823
1824   *Richard Levitte*
1825
1826 * Added command 'openssl kdf' that uses the EVP_KDF API.
1827
1828   *Shane Lontis*
1829
1830 * Added command 'openssl mac' that uses the EVP_MAC API.
1831
1832   *Shane Lontis*
1833
1834 * Added OPENSSL_info() to get diverse built-in OpenSSL data, such
1835   as default directories.  Also added the command 'openssl info'
1836   for scripting purposes.
1837
1838   *Richard Levitte*
1839
1840 * The functions AES_ige_encrypt() and AES_bi_ige_encrypt() have been
1841   deprecated.
1842
1843   *Matt Caswell*
1844
1845 * Add prediction resistance to the DRBG reseeding process.
1846
1847   *Paul Dale*
1848
1849 * Limit the number of blocks in a data unit for AES-XTS to 2^20 as
1850   mandated by IEEE Std 1619-2018.
1851
1852   *Paul Dale*
1853
1854 * Added newline escaping functionality to a filename when using openssl dgst.
1855   This output format is to replicate the output format found in the `*sum`
1856   checksum programs. This aims to preserve backward compatibility.
1857
1858   *Matt Eaton, Richard Levitte, and Paul Dale*
1859
1860 * Removed the heartbeat message in DTLS feature, as it has very
1861   little usage and doesn't seem to fulfill a valuable purpose.
1862   The configuration option is now deprecated.
1863
1864   *Richard Levitte*
1865
1866 * Changed the output of 'openssl {digestname} < file' to display the
1867   digest name in its output.
1868
1869   *Richard Levitte*
1870
1871 * Added a new generic trace API which provides support for enabling
1872   instrumentation through trace output.
1873
1874   *Richard Levitte & Matthias St. Pierre*
1875
1876 * Added build tests for C++.  These are generated files that only do one
1877   thing, to include one public OpenSSL head file each.  This tests that
1878   the public header files can be usefully included in a C++ application.
1879
1880   This test isn't enabled by default.  It can be enabled with the option
1881   'enable-buildtest-c++'.
1882
1883   *Richard Levitte*
1884
1885 * Added KB KDF (EVP_KDF_KB) to EVP_KDF.
1886
1887   *Robbie Harwood*
1888
1889 * Added SSH KDF (EVP_KDF_SSHKDF) and KRB5 KDF (EVP_KDF_KRB5KDF) to EVP_KDF.
1890
1891   *Simo Sorce*
1892
1893 * Added Single Step KDF (EVP_KDF_SS), X963 KDF, and X942 KDF to EVP_KDF.
1894
1895   *Shane Lontis*
1896
1897 * Added KMAC to EVP_MAC.
1898
1899   *Shane Lontis*
1900
1901 * Added property based algorithm implementation selection framework to
1902   the core.
1903
1904   *Paul Dale*
1905
1906 * Added SCA hardening for modular field inversion in EC_GROUP through
1907   a new dedicated field_inv() pointer in EC_METHOD.
1908   This also addresses a leakage affecting conversions from projective
1909   to affine coordinates.
1910
1911   *Billy Bob Brumley, Nicola Tuveri*
1912
1913 * Added EVP_KDF, an EVP layer KDF API, to simplify adding KDF and PRF
1914   implementations.  This includes an EVP_PKEY to EVP_KDF bridge for
1915   those algorithms that were already supported through the EVP_PKEY API
1916   (scrypt, TLS1 PRF and HKDF).  The low-level KDF functions for PBKDF2
1917   and scrypt are now wrappers that call EVP_KDF.
1918
1919   *David Makepeace*
1920
1921 * Build devcrypto engine as a dynamic engine.
1922
1923   *Eneas U de Queiroz*
1924
1925 * Add keyed BLAKE2 to EVP_MAC.
1926
1927   *Antoine Salon*
1928
1929 * Fix a bug in the computation of the endpoint-pair shared secret used
1930   by DTLS over SCTP. This breaks interoperability with older versions
1931   of OpenSSL like OpenSSL 1.1.0 and OpenSSL 1.0.2. There is a runtime
1932   switch SSL_MODE_DTLS_SCTP_LABEL_LENGTH_BUG (off by default) enabling
1933   interoperability with such broken implementations. However, enabling
1934   this switch breaks interoperability with correct implementations.
1935
1936 * Fix a use after free bug in d2i_X509_PUBKEY when overwriting a
1937   re-used X509_PUBKEY object if the second PUBKEY is malformed.
1938
1939   *Bernd Edlinger*
1940
1941 * Move strictness check from EVP_PKEY_asn1_new() to EVP_PKEY_asn1_add0().
1942
1943   *Richard Levitte*
1944
1945 * Changed the license to the Apache License v2.0.
1946
1947   *Richard Levitte*
1948
1949 * Switch to a new version scheme using three numbers MAJOR.MINOR.PATCH.
1950
1951   - Major releases (indicated by incrementing the MAJOR release number)
1952     may introduce incompatible API/ABI changes.
1953   - Minor releases (indicated by incrementing the MINOR release number)
1954     may introduce new features but retain API/ABI compatibility.
1955   - Patch releases (indicated by incrementing the PATCH number)
1956     are intended for bug fixes and other improvements of existing
1957     features only (like improving performance or adding documentation)
1958     and retain API/ABI compatibility.
1959
1960   *Richard Levitte*
1961
1962 * Add support for RFC5297 SIV mode (siv128), including AES-SIV.
1963
1964   *Todd Short*
1965
1966 * Remove the 'dist' target and add a tarball building script.  The
1967   'dist' target has fallen out of use, and it shouldn't be
1968   necessary to configure just to create a source distribution.
1969
1970   *Richard Levitte*
1971
1972 * Recreate the OS390-Unix config target.  It no longer relies on a
1973   special script like it did for OpenSSL pre-1.1.0.
1974
1975   *Richard Levitte*
1976
1977 * Instead of having the source directories listed in Configure, add
1978   a 'build.info' keyword SUBDIRS to indicate what sub-directories to
1979   look into.
1980
1981   *Richard Levitte*
1982
1983 * Add GMAC to EVP_MAC.
1984
1985   *Paul Dale*
1986
1987 * Ported the HMAC, CMAC and SipHash EVP_PKEY_METHODs to EVP_MAC.
1988
1989   *Richard Levitte*
1990
1991 * Added EVP_MAC, an EVP layer MAC API, to simplify adding MAC
1992   implementations.  This includes a generic EVP_PKEY to EVP_MAC bridge,
1993   to facilitate the continued use of MACs through raw private keys in
1994   functionality such as `EVP_DigestSign*` and `EVP_DigestVerify*`.
1995
1996   *Richard Levitte*
1997
1998 * Deprecate ECDH_KDF_X9_62().
1999
2000   *Antoine Salon*
2001
2002 * Added EVP_PKEY_ECDH_KDF_X9_63 and ecdh_KDF_X9_63() as replacements for
2003   the EVP_PKEY_ECDH_KDF_X9_62 KDF type and ECDH_KDF_X9_62(). The old names
2004   are retained for backwards compatibility.
2005
2006   *Antoine Salon*
2007
2008 * AES-XTS mode now enforces that its two keys are different to mitigate
2009   the attacked described in "Efficient Instantiations of Tweakable
2010   Blockciphers and Refinements to Modes OCB and PMAC" by Phillip Rogaway.
2011   Details of this attack can be obtained from:
2012   <http://web.cs.ucdavis.edu/%7Erogaway/papers/offsets.pdf>
2013
2014   *Paul Dale*
2015
2016 * Rename the object files, i.e. give them other names than in previous
2017   versions.  Their names now include the name of the final product, as
2018   well as its type mnemonic (bin, lib, shlib).
2019
2020   *Richard Levitte*
2021
2022 * Added new option for 'openssl list', '-objects', which will display the
2023   list of built in objects, i.e. OIDs with names.
2024
2025   *Richard Levitte*
2026
2027 * Added the options `-crl_lastupdate` and `-crl_nextupdate` to `openssl ca`,
2028   allowing the `lastUpdate` and `nextUpdate` fields in the generated CRL to
2029   be set explicitly.
2030
2031   *Chris Novakovic*
2032
2033 * Added support for Linux Kernel TLS data-path. The Linux Kernel data-path
2034   improves application performance by removing data copies and providing
2035   applications with zero-copy system calls such as sendfile and splice.
2036
2037   *Boris Pismenny*
2038
2039 * The SSL option SSL_OP_CLEANSE_PLAINTEXT is introduced.
2040
2041   *Martin Elshuber*
2042
2043 * `PKCS12_parse` now maintains the order of the parsed certificates
2044   when outputting them via `*ca` (rather than reversing it).
2045
2046   *David von Oheimb*
2047
2048 * Deprecated pthread fork support methods.
2049
2050   *Randall S. Becker*
2051
2052 * Added support for FFDHE key exchange in TLS 1.3.
2053
2054   *Raja Ashok*
2055
2056 * Added a new concept for OpenSSL plugability: providers.  This
2057   functionality is designed to replace the ENGINE API and ENGINE
2058   implementations, and to be much more dynamic, allowing provider
2059   authors to introduce new algorithms among other things, as long as
2060   there's an API that supports the algorithm type.
2061
2062   With this concept comes a new core API for interaction between
2063   libcrypto and provider implementations.  Public libcrypto functions
2064   that want to use providers do so through this core API.
2065
2066   The main documentation for this core API is found in
2067   doc/man7/provider.pod, doc/man7/provider-base.pod, and they in turn
2068   refer to other manuals describing the API specific for supported
2069   algorithm types (also called operations).
2070
2071   *The OpenSSL team*
2072
2073OpenSSL 1.1.1
2074-------------
2075
2076### Changes between 1.1.1l and 1.1.1m [xx XXX xxxx]
2077
2078 * Avoid loading of a dynamic engine twice.
2079
2080   *Bernd Edlinger*
2081
2082 * Prioritise DANE TLSA issuer certs over peer certs
2083
2084   *Viktor Dukhovni*
2085
2086 * Fixed random API for MacOS prior to 10.12
2087
2088   These MacOS versions don't support the CommonCrypto APIs
2089
2090   *Lenny Primak*
2091
2092### Changes between 1.1.1k and 1.1.1l [24 Aug 2021]
2093
2094 * Fixed an SM2 Decryption Buffer Overflow.
2095
2096   In order to decrypt SM2 encrypted data an application is expected to
2097   call the API function EVP_PKEY_decrypt(). Typically an application will
2098   call this function twice. The first time, on entry, the "out" parameter
2099   can be NULL and, on exit, the "outlen" parameter is populated with the
2100   buffer size required to hold the decrypted plaintext. The application
2101   can then allocate a sufficiently sized buffer and call EVP_PKEY_decrypt()
2102   again, but this time passing a non-NULL value for the "out" parameter.
2103
2104   A bug in the implementation of the SM2 decryption code means that the
2105   calculation of the buffer size required to hold the plaintext returned
2106   by the first call to EVP_PKEY_decrypt() can be smaller than the actual
2107   size required by the second call. This can lead to a buffer overflow
2108   when EVP_PKEY_decrypt() is called by the application a second time with
2109   a buffer that is too small.
2110
2111   A malicious attacker who is able present SM2 content for decryption to
2112   an application could cause attacker chosen data to overflow the buffer
2113   by up to a maximum of 62 bytes altering the contents of other data held
2114   after the buffer, possibly changing application behaviour or causing
2115   the application to crash. The location of the buffer is application
2116   dependent but is typically heap allocated.
2117   ([CVE-2021-3711])
2118
2119   *Matt Caswell*
2120
2121 * Fixed various read buffer overruns processing ASN.1 strings
2122
2123   ASN.1 strings are represented internally within OpenSSL as an ASN1_STRING
2124   structure which contains a buffer holding the string data and a field
2125   holding the buffer length. This contrasts with normal C strings which
2126   are repesented as a buffer for the string data which is terminated
2127   with a NUL (0) byte.
2128
2129   Although not a strict requirement, ASN.1 strings that are parsed using
2130   OpenSSL's own "d2i" functions (and other similar parsing functions) as
2131   well as any string whose value has been set with the ASN1_STRING_set()
2132   function will additionally NUL terminate the byte array in the
2133   ASN1_STRING structure.
2134
2135   However, it is possible for applications to directly construct valid
2136   ASN1_STRING structures which do not NUL terminate the byte array by
2137   directly setting the "data" and "length" fields in the ASN1_STRING
2138   array. This can also happen by using the ASN1_STRING_set0() function.
2139
2140   Numerous OpenSSL functions that print ASN.1 data have been found to
2141   assume that the ASN1_STRING byte array will be NUL terminated, even
2142   though this is not guaranteed for strings that have been directly
2143   constructed. Where an application requests an ASN.1 structure to be
2144   printed, and where that ASN.1 structure contains ASN1_STRINGs that have
2145   been directly constructed by the application without NUL terminating
2146   the "data" field, then a read buffer overrun can occur.
2147
2148   The same thing can also occur during name constraints processing
2149   of certificates (for example if a certificate has been directly
2150   constructed by the application instead of loading it via the OpenSSL
2151   parsing functions, and the certificate contains non NUL terminated
2152   ASN1_STRING structures). It can also occur in the X509_get1_email(),
2153   X509_REQ_get1_email() and X509_get1_ocsp() functions.
2154
2155   If a malicious actor can cause an application to directly construct an
2156   ASN1_STRING and then process it through one of the affected OpenSSL
2157   functions then this issue could be hit. This might result in a crash
2158   (causing a Denial of Service attack). It could also result in the
2159   disclosure of private memory contents (such as private keys, or
2160   sensitive plaintext).
2161   ([CVE-2021-3712])
2162
2163   *Matt Caswell*
2164
2165### Changes between 1.1.1j and 1.1.1k [25 Mar 2021]
2166
2167 * Fixed a problem with verifying a certificate chain when using the
2168   X509_V_FLAG_X509_STRICT flag. This flag enables additional security checks of
2169   the certificates present in a certificate chain. It is not set by default.
2170
2171   Starting from OpenSSL version 1.1.1h a check to disallow certificates in
2172   the chain that have explicitly encoded elliptic curve parameters was added
2173   as an additional strict check.
2174
2175   An error in the implementation of this check meant that the result of a
2176   previous check to confirm that certificates in the chain are valid CA
2177   certificates was overwritten. This effectively bypasses the check
2178   that non-CA certificates must not be able to issue other certificates.
2179
2180   If a "purpose" has been configured then there is a subsequent opportunity
2181   for checks that the certificate is a valid CA.  All of the named "purpose"
2182   values implemented in libcrypto perform this check.  Therefore, where
2183   a purpose is set the certificate chain will still be rejected even when the
2184   strict flag has been used. A purpose is set by default in libssl client and
2185   server certificate verification routines, but it can be overridden or
2186   removed by an application.
2187
2188   In order to be affected, an application must explicitly set the
2189   X509_V_FLAG_X509_STRICT verification flag and either not set a purpose
2190   for the certificate verification or, in the case of TLS client or server
2191   applications, override the default purpose.
2192   ([CVE-2021-3450])
2193
2194   *Tomáš Mráz*
2195
2196 * Fixed an issue where an OpenSSL TLS server may crash if sent a maliciously
2197   crafted renegotiation ClientHello message from a client. If a TLSv1.2
2198   renegotiation ClientHello omits the signature_algorithms extension (where it
2199   was present in the initial ClientHello), but includes a
2200   signature_algorithms_cert extension then a NULL pointer dereference will
2201   result, leading to a crash and a denial of service attack.
2202
2203   A server is only vulnerable if it has TLSv1.2 and renegotiation enabled
2204   (which is the default configuration). OpenSSL TLS clients are not impacted by
2205   this issue.
2206   ([CVE-2021-3449])
2207
2208   *Peter Kästle and Samuel Sapalski*
2209
2210### Changes between 1.1.1i and 1.1.1j [16 Feb 2021]
2211
2212 * Fixed the X509_issuer_and_serial_hash() function. It attempts to
2213   create a unique hash value based on the issuer and serial number data
2214   contained within an X509 certificate. However it was failing to correctly
2215   handle any errors that may occur while parsing the issuer field (which might
2216   occur if the issuer field is maliciously constructed). This may subsequently
2217   result in a NULL pointer deref and a crash leading to a potential denial of
2218   service attack.
2219   ([CVE-2021-23841])
2220
2221   *Matt Caswell*
2222
2223 * Fixed the RSA_padding_check_SSLv23() function and the RSA_SSLV23_PADDING
2224   padding mode to correctly check for rollback attacks. This is considered a
2225   bug in OpenSSL 1.1.1 because it does not support SSLv2. In 1.0.2 this is
2226   CVE-2021-23839.
2227
2228   *Matt Caswell*
2229
2230   Fixed the EVP_CipherUpdate, EVP_EncryptUpdate and EVP_DecryptUpdate
2231   functions. Previously they could overflow the output length argument in some
2232   cases where the input length is close to the maximum permissable length for
2233   an integer on the platform. In such cases the return value from the function
2234   call would be 1 (indicating success), but the output length value would be
2235   negative. This could cause applications to behave incorrectly or crash.
2236   ([CVE-2021-23840])
2237
2238   *Matt Caswell*
2239
2240 * Fixed SRP_Calc_client_key so that it runs in constant time. The previous
2241   implementation called BN_mod_exp without setting BN_FLG_CONSTTIME. This
2242   could be exploited in a side channel attack to recover the password. Since
2243   the attack is local host only this is outside of the current OpenSSL
2244   threat model and therefore no CVE is assigned.
2245
2246   Thanks to Mohammed Sabt and Daniel De Almeida Braga for reporting this
2247   issue.
2248
2249   *Matt Caswell*
2250
2251### Changes between 1.1.1h and 1.1.1i [8 Dec 2020]
2252
2253 * Fixed NULL pointer deref in the GENERAL_NAME_cmp function
2254   This function could crash if both GENERAL_NAMEs contain an EDIPARTYNAME.
2255    If an attacker can control both items being compared  then this could lead
2256    to a possible denial of service attack. OpenSSL itself uses the
2257    GENERAL_NAME_cmp function for two purposes:
2258    1) Comparing CRL distribution point names between an available CRL and a
2259       CRL distribution point embedded in an X509 certificate
2260    2) When verifying that a timestamp response token signer matches the
2261       timestamp authority name (exposed via the API functions
2262       TS_RESP_verify_response and TS_RESP_verify_token)
2263   ([CVE-2020-1971])
2264
2265   *Matt Caswell*
2266
2267### Changes between 1.1.1g and 1.1.1h [22 Sep 2020]
2268
2269 * Certificates with explicit curve parameters are now disallowed in
2270   verification chains if the X509_V_FLAG_X509_STRICT flag is used.
2271
2272   *Tomáš Mráz*
2273
2274 * The 'MinProtocol' and 'MaxProtocol' configuration commands now silently
2275   ignore TLS protocol version bounds when configuring DTLS-based contexts, and
2276   conversely, silently ignore DTLS protocol version bounds when configuring
2277   TLS-based contexts.  The commands can be repeated to set bounds of both
2278   types.  The same applies with the corresponding "min_protocol" and
2279   "max_protocol" command-line switches, in case some application uses both TLS
2280   and DTLS.
2281
2282   SSL_CTX instances that are created for a fixed protocol version (e.g.
2283   TLSv1_server_method()) also silently ignore version bounds.  Previously
2284   attempts to apply bounds to these protocol versions would result in an
2285   error.  Now only the "version-flexible" SSL_CTX instances are subject to
2286   limits in configuration files in command-line options.
2287
2288   *Viktor Dukhovni*
2289
2290 * Handshake now fails if Extended Master Secret extension is dropped
2291   on renegotiation.
2292
2293   *Tomáš Mráz*
2294
2295 * The Oracle Developer Studio compiler will start reporting deprecated APIs
2296
2297### Changes between 1.1.1f and 1.1.1g [21 Apr 2020]
2298
2299 * Fixed segmentation fault in SSL_check_chain()
2300   Server or client applications that call the SSL_check_chain() function
2301   during or after a TLS 1.3 handshake may crash due to a NULL pointer
2302   dereference as a result of incorrect handling of the
2303   "signature_algorithms_cert" TLS extension. The crash occurs if an invalid
2304   or unrecognised signature algorithm is received from the peer. This could
2305   be exploited by a malicious peer in a Denial of Service attack.
2306   ([CVE-2020-1967])
2307
2308   *Benjamin Kaduk*
2309
2310 * Added AES consttime code for no-asm configurations
2311   an optional constant time support for AES was added
2312   when building openssl for no-asm.
2313   Enable with: ./config no-asm -DOPENSSL_AES_CONST_TIME
2314   Disable with: ./config no-asm -DOPENSSL_NO_AES_CONST_TIME
2315   At this time this feature is by default disabled.
2316   It will be enabled by default in 3.0.
2317
2318   *Bernd Edlinger*
2319
2320### Changes between 1.1.1e and 1.1.1f [31 Mar 2020]
2321
2322 * Revert the change of EOF detection while reading in libssl to avoid
2323   regressions in applications depending on the current way of reporting
2324   the EOF. As the existing method is not fully accurate the change to
2325   reporting the EOF via SSL_ERROR_SSL is kept on the current development
2326   branch and will be present in the 3.0 release.
2327
2328   *Tomáš Mráz*
2329
2330 * Revised BN_generate_prime_ex to not avoid factors 3..17863 in p-1
2331   when primes for RSA keys are computed.
2332   Since we previously always generated primes == 2 (mod 3) for RSA keys,
2333   the 2-prime and 3-prime RSA modules were easy to distinguish, since
2334   N = p*q = 1 (mod 3), but N = p*q*r = 2 (mod 3). Therefore fingerprinting
2335   2-prime vs. 3-prime RSA keys was possible by computing N mod 3.
2336   This avoids possible fingerprinting of newly generated RSA modules.
2337
2338   *Bernd Edlinger*
2339
2340### Changes between 1.1.1d and 1.1.1e [17 Mar 2020]
2341
2342 * Properly detect EOF while reading in libssl. Previously if we hit an EOF
2343   while reading in libssl then we would report an error back to the
2344   application (SSL_ERROR_SYSCALL) but errno would be 0. We now add
2345   an error to the stack (which means we instead return SSL_ERROR_SSL) and
2346   therefore give a hint as to what went wrong.
2347
2348   *Matt Caswell*
2349
2350 * Check that ed25519 and ed448 are allowed by the security level. Previously
2351   signature algorithms not using an MD were not being checked that they were
2352   allowed by the security level.
2353
2354   *Kurt Roeckx*
2355
2356 * Fixed SSL_get_servername() behaviour. The behaviour of SSL_get_servername()
2357   was not quite right. The behaviour was not consistent between resumption
2358   and normal handshakes, and also not quite consistent with historical
2359   behaviour. The behaviour in various scenarios has been clarified and
2360   it has been updated to make it match historical behaviour as closely as
2361   possible.
2362
2363   *Matt Caswell*
2364
2365 * *[VMS only]* The header files that the VMS compilers include automatically,
2366   `__DECC_INCLUDE_PROLOGUE.H` and `__DECC_INCLUDE_EPILOGUE.H`, use pragmas
2367   that the C++ compiler doesn't understand.  This is a shortcoming in the
2368   compiler, but can be worked around with `__cplusplus` guards.
2369
2370   C++ applications that use OpenSSL libraries must be compiled using the
2371   qualifier `/NAMES=(AS_IS,SHORTENED)` to be able to use all the OpenSSL
2372   functions.  Otherwise, only functions with symbols of less than 31
2373   characters can be used, as the linker will not be able to successfully
2374   resolve symbols with longer names.
2375
2376   *Richard Levitte*
2377
2378 * Added a new method to gather entropy on VMS, based on SYS$GET_ENTROPY.
2379   The presence of this system service is determined at run-time.
2380
2381   *Richard Levitte*
2382
2383 * Print all values for a PKCS#12 attribute with 'openssl pkcs12', not just
2384   the first value.
2385
2386   *Jon Spillett*
2387
2388### Changes between 1.1.1c and 1.1.1d [10 Sep 2019]
2389
2390 * Fixed a fork protection issue. OpenSSL 1.1.1 introduced a rewritten random
2391   number generator (RNG). This was intended to include protection in the
2392   event of a fork() system call in order to ensure that the parent and child
2393   processes did not share the same RNG state. However this protection was not
2394   being used in the default case.
2395
2396   A partial mitigation for this issue is that the output from a high
2397   precision timer is mixed into the RNG state so the likelihood of a parent
2398   and child process sharing state is significantly reduced.
2399
2400   If an application already calls OPENSSL_init_crypto() explicitly using
2401   OPENSSL_INIT_ATFORK then this problem does not occur at all.
2402   ([CVE-2019-1549])
2403
2404   *Matthias St. Pierre*
2405
2406 * For built-in EC curves, ensure an EC_GROUP built from the curve name is
2407   used even when parsing explicit parameters, when loading a encoded key
2408   or calling `EC_GROUP_new_from_ecpkparameters()`/
2409   `EC_GROUP_new_from_ecparameters()`.
2410   This prevents bypass of security hardening and performance gains,
2411   especially for curves with specialized EC_METHODs.
2412   By default, if a key encoded with explicit parameters is loaded and later
2413   encoded, the output is still encoded with explicit parameters, even if
2414   internally a "named" EC_GROUP is used for computation.
2415
2416   *Nicola Tuveri*
2417
2418 * Compute ECC cofactors if not provided during EC_GROUP construction. Before
2419   this change, EC_GROUP_set_generator would accept order and/or cofactor as
2420   NULL. After this change, only the cofactor parameter can be NULL. It also
2421   does some minimal sanity checks on the passed order.
2422   ([CVE-2019-1547])
2423
2424   *Billy Bob Brumley*
2425
2426 * Fixed a padding oracle in PKCS7_dataDecode and CMS_decrypt_set1_pkey.
2427   An attack is simple, if the first CMS_recipientInfo is valid but the
2428   second CMS_recipientInfo is chosen ciphertext. If the second
2429   recipientInfo decodes to PKCS #1 v1.5 form plaintext, the correct
2430   encryption key will be replaced by garbage, and the message cannot be
2431   decoded, but if the RSA decryption fails, the correct encryption key is
2432   used and the recipient will not notice the attack.
2433   As a work around for this potential attack the length of the decrypted
2434   key must be equal to the cipher default key length, in case the
2435   certifiate is not given and all recipientInfo are tried out.
2436   The old behaviour can be re-enabled in the CMS code by setting the
2437   CMS_DEBUG_DECRYPT flag.
2438   ([CVE-2019-1563])
2439
2440   *Bernd Edlinger*
2441
2442 * Early start up entropy quality from the DEVRANDOM seed source has been
2443   improved for older Linux systems.  The RAND subsystem will wait for
2444   /dev/random to be producing output before seeding from /dev/urandom.
2445   The seeded state is stored for future library initialisations using
2446   a system global shared memory segment.  The shared memory identifier
2447   can be configured by defining OPENSSL_RAND_SEED_DEVRANDOM_SHM_ID to
2448   the desired value.  The default identifier is 114.
2449
2450   *Paul Dale*
2451
2452 * Correct the extended master secret constant on EBCDIC systems. Without this
2453   fix TLS connections between an EBCDIC system and a non-EBCDIC system that
2454   negotiate EMS will fail. Unfortunately this also means that TLS connections
2455   between EBCDIC systems with this fix, and EBCDIC systems without this
2456   fix will fail if they negotiate EMS.
2457
2458   *Matt Caswell*
2459
2460 * Use Windows installation paths in the mingw builds
2461
2462   Mingw isn't a POSIX environment per se, which means that Windows
2463   paths should be used for installation.
2464   ([CVE-2019-1552])
2465
2466   *Richard Levitte*
2467
2468 * Changed DH_check to accept parameters with order q and 2q subgroups.
2469   With order 2q subgroups the bit 0 of the private key is not secret
2470   but DH_generate_key works around that by clearing bit 0 of the
2471   private key for those. This avoids leaking bit 0 of the private key.
2472
2473   *Bernd Edlinger*
2474
2475 * Significantly reduce secure memory usage by the randomness pools.
2476
2477   *Paul Dale*
2478
2479 * Revert the DEVRANDOM_WAIT feature for Linux systems
2480
2481   The DEVRANDOM_WAIT feature added a select() call to wait for the
2482   /dev/random device to become readable before reading from the
2483   /dev/urandom device.
2484
2485   It turned out that this change had negative side effects on
2486   performance which were not acceptable. After some discussion it
2487   was decided to revert this feature and leave it up to the OS
2488   resp. the platform maintainer to ensure a proper initialization
2489   during early boot time.
2490
2491   *Matthias St. Pierre*
2492
2493### Changes between 1.1.1b and 1.1.1c [28 May 2019]
2494
2495 * Add build tests for C++.  These are generated files that only do one
2496   thing, to include one public OpenSSL head file each.  This tests that
2497   the public header files can be usefully included in a C++ application.
2498
2499   This test isn't enabled by default.  It can be enabled with the option
2500   'enable-buildtest-c++'.
2501
2502   *Richard Levitte*
2503
2504 * Enable SHA3 pre-hashing for ECDSA and DSA.
2505
2506   *Patrick Steuer*
2507
2508 * Change the default RSA, DSA and DH size to 2048 bit instead of 1024.
2509   This changes the size when using the `genpkey` command when no size is given.
2510   It fixes an omission in earlier changes that changed all RSA, DSA and DH
2511   generation commands to use 2048 bits by default.
2512
2513   *Kurt Roeckx*
2514
2515 * Reorganize the manual pages to consistently have RETURN VALUES,
2516   EXAMPLES, SEE ALSO and HISTORY come in that order, and adjust
2517   util/fix-doc-nits accordingly.
2518
2519   *Paul Yang, Joshua Lock*
2520
2521 * Add the missing accessor EVP_PKEY_get0_engine()
2522
2523   *Matt Caswell*
2524
2525 * Have commands like `s_client` and `s_server` output the signature scheme
2526   along with other cipher suite parameters when debugging.
2527
2528   *Lorinczy Zsigmond*
2529
2530 * Make OPENSSL_config() error agnostic again.
2531
2532   *Richard Levitte*
2533
2534 * Do the error handling in RSA decryption constant time.
2535
2536   *Bernd Edlinger*
2537
2538 * Prevent over long nonces in ChaCha20-Poly1305.
2539
2540   ChaCha20-Poly1305 is an AEAD cipher, and requires a unique nonce input
2541   for every encryption operation. RFC 7539 specifies that the nonce value
2542   (IV) should be 96 bits (12 bytes). OpenSSL allows a variable nonce length
2543   and front pads the nonce with 0 bytes if it is less than 12
2544   bytes. However it also incorrectly allows a nonce to be set of up to 16
2545   bytes. In this case only the last 12 bytes are significant and any
2546   additional leading bytes are ignored.
2547
2548   It is a requirement of using this cipher that nonce values are
2549   unique. Messages encrypted using a reused nonce value are susceptible to
2550   serious confidentiality and integrity attacks. If an application changes
2551   the default nonce length to be longer than 12 bytes and then makes a
2552   change to the leading bytes of the nonce expecting the new value to be a
2553   new unique nonce then such an application could inadvertently encrypt
2554   messages with a reused nonce.
2555
2556   Additionally the ignored bytes in a long nonce are not covered by the
2557   integrity guarantee of this cipher. Any application that relies on the
2558   integrity of these ignored leading bytes of a long nonce may be further
2559   affected. Any OpenSSL internal use of this cipher, including in SSL/TLS,
2560   is safe because no such use sets such a long nonce value. However user
2561   applications that use this cipher directly and set a non-default nonce
2562   length to be longer than 12 bytes may be vulnerable.
2563
2564   This issue was reported to OpenSSL on 16th of March 2019 by Joran Dirk
2565   Greef of Ronomon.
2566   ([CVE-2019-1543])
2567
2568   *Matt Caswell*
2569
2570 * Add DEVRANDOM_WAIT feature for Linux systems
2571
2572   On older Linux systems where the getrandom() system call is not available,
2573   OpenSSL normally uses the /dev/urandom device for seeding its CSPRNG.
2574   Contrary to getrandom(), the /dev/urandom device will not block during
2575   early boot when the kernel CSPRNG has not been seeded yet.
2576
2577   To mitigate this known weakness, use select() to wait for /dev/random to
2578   become readable before reading from /dev/urandom.
2579
2580 * Ensure that SM2 only uses SM3 as digest algorithm
2581
2582   *Paul Yang*
2583
2584### Changes between 1.1.1a and 1.1.1b [26 Feb 2019]
2585
2586 * Change the info callback signals for the start and end of a post-handshake
2587   message exchange in TLSv1.3. In 1.1.1/1.1.1a we used SSL_CB_HANDSHAKE_START
2588   and SSL_CB_HANDSHAKE_DONE. Experience has shown that many applications get
2589   confused by this and assume that a TLSv1.2 renegotiation has started. This
2590   can break KeyUpdate handling. Instead we no longer signal the start and end
2591   of a post handshake message exchange (although the messages themselves are
2592   still signalled). This could break some applications that were expecting
2593   the old signals. However without this KeyUpdate is not usable for many
2594   applications.
2595
2596   *Matt Caswell*
2597
2598### Changes between 1.1.1 and 1.1.1a [20 Nov 2018]
2599
2600 * Timing vulnerability in DSA signature generation
2601
2602   The OpenSSL DSA signature algorithm has been shown to be vulnerable to a
2603   timing side channel attack. An attacker could use variations in the signing
2604   algorithm to recover the private key.
2605
2606   This issue was reported to OpenSSL on 16th October 2018 by Samuel Weiser.
2607   ([CVE-2018-0734])
2608
2609   *Paul Dale*
2610
2611 * Timing vulnerability in ECDSA signature generation
2612
2613   The OpenSSL ECDSA signature algorithm has been shown to be vulnerable to a
2614   timing side channel attack. An attacker could use variations in the signing
2615   algorithm to recover the private key.
2616
2617   This issue was reported to OpenSSL on 25th October 2018 by Samuel Weiser.
2618   ([CVE-2018-0735])
2619
2620   *Paul Dale*
2621
2622 * Fixed the issue that RAND_add()/RAND_seed() silently discards random input
2623   if its length exceeds 4096 bytes. The limit has been raised to a buffer size
2624   of two gigabytes and the error handling improved.
2625
2626   This issue was reported to OpenSSL by Dr. Falko Strenzke. It has been
2627   categorized as a normal bug, not a security issue, because the DRBG reseeds
2628   automatically and is fully functional even without additional randomness
2629   provided by the application.
2630
2631### Changes between 1.1.0i and 1.1.1 [11 Sep 2018]
2632
2633 * Add a new ClientHello callback. Provides a callback interface that gives
2634   the application the ability to adjust the nascent SSL object at the
2635   earliest stage of ClientHello processing, immediately after extensions have
2636   been collected but before they have been processed. In particular, this
2637   callback can adjust the supported TLS versions in response to the contents
2638   of the ClientHello
2639
2640   *Benjamin Kaduk*
2641
2642 * Add SM2 base algorithm support.
2643
2644   *Jack Lloyd*
2645
2646 * s390x assembly pack: add (improved) hardware-support for the following
2647   cryptographic primitives: sha3, shake, aes-gcm, aes-ccm, aes-ctr, aes-ofb,
2648   aes-cfb/cfb8, aes-ecb.
2649
2650   *Patrick Steuer*
2651
2652 * Make EVP_PKEY_asn1_new() a bit stricter about its input.  A NULL pem_str
2653   parameter is no longer accepted, as it leads to a corrupt table.  NULL
2654   pem_str is reserved for alias entries only.
2655
2656   *Richard Levitte*
2657
2658 * Use the new ec_scalar_mul_ladder scaffold to implement a specialized ladder
2659   step for prime curves. The new implementation is based on formulae from
2660   differential addition-and-doubling in homogeneous projective coordinates
2661   from Izu-Takagi "A fast parallel elliptic curve multiplication resistant
2662   against side channel attacks" and Brier-Joye "Weierstrass Elliptic Curves
2663   and Side-Channel Attacks" Eq. (8) for y-coordinate recovery, modified
2664   to work in projective coordinates.
2665
2666   *Billy Bob Brumley, Nicola Tuveri*
2667
2668 * Change generating and checking of primes so that the error rate of not
2669   being prime depends on the intended use based on the size of the input.
2670   For larger primes this will result in more rounds of Miller-Rabin.
2671   The maximal error rate for primes with more than 1080 bits is lowered
2672   to 2^-128.
2673
2674   *Kurt Roeckx, Annie Yousar*
2675
2676 * Increase the number of Miller-Rabin rounds for DSA key generating to 64.
2677
2678   *Kurt Roeckx*
2679
2680 * The 'tsget' script is renamed to 'tsget.pl', to avoid confusion when
2681   moving between systems, and to avoid confusion when a Windows build is
2682   done with mingw vs with MSVC.  For POSIX installs, there's still a
2683   symlink or copy named 'tsget' to avoid that confusion as well.
2684
2685   *Richard Levitte*
2686
2687 * Revert blinding in ECDSA sign and instead make problematic addition
2688   length-invariant. Switch even to fixed-length Montgomery multiplication.
2689
2690   *Andy Polyakov*
2691
2692 * Use the new ec_scalar_mul_ladder scaffold to implement a specialized ladder
2693   step for binary curves. The new implementation is based on formulae from
2694   differential addition-and-doubling in mixed Lopez-Dahab projective
2695   coordinates, modified to independently blind the operands.
2696
2697   *Billy Bob Brumley, Sohaib ul Hassan, Nicola Tuveri*
2698
2699 * Add a scaffold to optionally enhance the Montgomery ladder implementation
2700   for `ec_scalar_mul_ladder` (formerly `ec_mul_consttime`) allowing
2701   EC_METHODs to implement their own specialized "ladder step", to take
2702   advantage of more favorable coordinate systems or more efficient
2703   differential addition-and-doubling algorithms.
2704
2705   *Billy Bob Brumley, Sohaib ul Hassan, Nicola Tuveri*
2706
2707 * Modified the random device based seed sources to keep the relevant
2708   file descriptors open rather than reopening them on each access.
2709   This allows such sources to operate in a chroot() jail without
2710   the associated device nodes being available. This behaviour can be
2711   controlled using RAND_keep_random_devices_open().
2712
2713   *Paul Dale*
2714
2715 * Numerous side-channel attack mitigations have been applied. This may have
2716   performance impacts for some algorithms for the benefit of improved
2717   security. Specific changes are noted in this change log by their respective
2718   authors.
2719
2720   *Matt Caswell*
2721
2722 * AIX shared library support overhaul. Switch to AIX "natural" way of
2723   handling shared libraries, which means collecting shared objects of
2724   different versions and bitnesses in one common archive. This allows to
2725   mitigate conflict between 1.0 and 1.1 side-by-side installations. It
2726   doesn't affect the way 3rd party applications are linked, only how
2727   multi-version installation is managed.
2728
2729   *Andy Polyakov*
2730
2731 * Make ec_group_do_inverse_ord() more robust and available to other
2732   EC cryptosystems, so that irrespective of BN_FLG_CONSTTIME, SCA
2733   mitigations are applied to the fallback BN_mod_inverse().
2734   When using this function rather than BN_mod_inverse() directly, new
2735   EC cryptosystem implementations are then safer-by-default.
2736
2737   *Billy Bob Brumley*
2738
2739 * Add coordinate blinding for EC_POINT and implement projective
2740   coordinate blinding for generic prime curves as a countermeasure to
2741   chosen point SCA attacks.
2742
2743   *Sohaib ul Hassan, Nicola Tuveri, Billy Bob Brumley*
2744
2745 * Add blinding to ECDSA and DSA signatures to protect against side channel
2746   attacks discovered by Keegan Ryan (NCC Group).
2747
2748   *Matt Caswell*
2749
2750 * Enforce checking in the `pkeyutl` command to ensure that the input
2751   length does not exceed the maximum supported digest length when performing
2752   a sign, verify or verifyrecover operation.
2753
2754   *Matt Caswell*
2755
2756 * SSL_MODE_AUTO_RETRY is enabled by default. Applications that use blocking
2757   I/O in combination with something like select() or poll() will hang. This
2758   can be turned off again using SSL_CTX_clear_mode().
2759   Many applications do not properly handle non-application data records, and
2760   TLS 1.3 sends more of such records. Setting SSL_MODE_AUTO_RETRY works
2761   around the problems in those applications, but can also break some.
2762   It's recommended to read the manpages about SSL_read(), SSL_write(),
2763   SSL_get_error(), SSL_shutdown(), SSL_CTX_set_mode() and
2764   SSL_CTX_set_read_ahead() again.
2765
2766   *Kurt Roeckx*
2767
2768 * When unlocking a pass phrase protected PEM file or PKCS#8 container, we
2769   now allow empty (zero character) pass phrases.
2770
2771   *Richard Levitte*
2772
2773 * Apply blinding to binary field modular inversion and remove patent
2774   pending (OPENSSL_SUN_GF2M_DIV) BN_GF2m_mod_div implementation.
2775
2776   *Billy Bob Brumley*
2777
2778 * Deprecate ec2_mult.c and unify scalar multiplication code paths for
2779   binary and prime elliptic curves.
2780
2781   *Billy Bob Brumley*
2782
2783 * Remove ECDSA nonce padding: EC_POINT_mul is now responsible for
2784   constant time fixed point multiplication.
2785
2786   *Billy Bob Brumley*
2787
2788 * Revise elliptic curve scalar multiplication with timing attack
2789   defenses: ec_wNAF_mul redirects to a constant time implementation
2790   when computing fixed point and variable point multiplication (which
2791   in OpenSSL are mostly used with secret scalars in keygen, sign,
2792   ECDH derive operations).
2793   *Billy Bob Brumley, Nicola Tuveri, Cesar Pereida García,
2794    Sohaib ul Hassan*
2795
2796 * Updated CONTRIBUTING
2797
2798   *Rich Salz*
2799
2800 * Updated DRBG / RAND to request nonce and additional low entropy
2801   randomness from the system.
2802
2803   *Matthias St. Pierre*
2804
2805 * Updated 'openssl rehash' to use OpenSSL consistent default.
2806
2807   *Richard Levitte*
2808
2809 * Moved the load of the ssl_conf module to libcrypto, which helps
2810   loading engines that libssl uses before libssl is initialised.
2811
2812   *Matt Caswell*
2813
2814 * Added EVP_PKEY_sign() and EVP_PKEY_verify() for EdDSA
2815
2816   *Matt Caswell*
2817
2818 * Fixed X509_NAME_ENTRY_set to get multi-valued RDNs right in all cases.
2819
2820   *Ingo Schwarze, Rich Salz*
2821
2822 * Added output of accepting IP address and port for 'openssl s_server'
2823
2824   *Richard Levitte*
2825
2826 * Added a new API for TLSv1.3 ciphersuites:
2827      SSL_CTX_set_ciphersuites()
2828      SSL_set_ciphersuites()
2829
2830   *Matt Caswell*
2831
2832 * Memory allocation failures consistently add an error to the error
2833   stack.
2834
2835   *Rich Salz*
2836
2837 * Don't use OPENSSL_ENGINES and OPENSSL_CONF environment values
2838   in libcrypto when run as setuid/setgid.
2839
2840   *Bernd Edlinger*
2841
2842 * Load any config file by default when libssl is used.
2843
2844   *Matt Caswell*
2845
2846 * Added new public header file <openssl/rand_drbg.h> and documentation
2847   for the RAND_DRBG API. See manual page RAND_DRBG(7) for an overview.
2848
2849   *Matthias St. Pierre*
2850
2851 * QNX support removed (cannot find contributors to get their approval
2852   for the license change).
2853
2854   *Rich Salz*
2855
2856 * TLSv1.3 replay protection for early data has been implemented. See the
2857   SSL_read_early_data() man page for further details.
2858
2859   *Matt Caswell*
2860
2861 * Separated TLSv1.3 ciphersuite configuration out from TLSv1.2 ciphersuite
2862   configuration. TLSv1.3 ciphersuites are not compatible with TLSv1.2 and
2863   below. Similarly TLSv1.2 ciphersuites are not compatible with TLSv1.3.
2864   In order to avoid issues where legacy TLSv1.2 ciphersuite configuration
2865   would otherwise inadvertently disable all TLSv1.3 ciphersuites the
2866   configuration has been separated out. See the ciphers man page or the
2867   SSL_CTX_set_ciphersuites() man page for more information.
2868
2869   *Matt Caswell*
2870
2871 * On POSIX (BSD, Linux, ...) systems the ocsp(1) command running
2872   in responder mode now supports the new "-multi" option, which
2873   spawns the specified number of child processes to handle OCSP
2874   requests.  The "-timeout" option now also limits the OCSP
2875   responder's patience to wait to receive the full client request
2876   on a newly accepted connection. Child processes are respawned
2877   as needed, and the CA index file is automatically reloaded
2878   when changed.  This makes it possible to run the "ocsp" responder
2879   as a long-running service, making the OpenSSL CA somewhat more
2880   feature-complete.  In this mode, most diagnostic messages logged
2881   after entering the event loop are logged via syslog(3) rather than
2882   written to stderr.
2883
2884   *Viktor Dukhovni*
2885
2886 * Added support for X448 and Ed448. Heavily based on original work by
2887   Mike Hamburg.
2888
2889   *Matt Caswell*
2890
2891 * Extend OSSL_STORE with capabilities to search and to narrow the set of
2892   objects loaded.  This adds the functions OSSL_STORE_expect() and
2893   OSSL_STORE_find() as well as needed tools to construct searches and
2894   get the search data out of them.
2895
2896   *Richard Levitte*
2897
2898 * Support for TLSv1.3 added. Note that users upgrading from an earlier
2899   version of OpenSSL should review their configuration settings to ensure
2900   that they are still appropriate for TLSv1.3. For further information see:
2901   <https://wiki.openssl.org/index.php/TLS1.3>
2902
2903   *Matt Caswell*
2904
2905 * Grand redesign of the OpenSSL random generator
2906
2907   The default RAND method now utilizes an AES-CTR DRBG according to
2908   NIST standard SP 800-90Ar1. The new random generator is essentially
2909   a port of the default random generator from the OpenSSL FIPS 2.0
2910   object module. It is a hybrid deterministic random bit generator
2911   using an AES-CTR bit stream and which seeds and reseeds itself
2912   automatically using trusted system entropy sources.
2913
2914   Some of its new features are:
2915    - Support for multiple DRBG instances with seed chaining.
2916    - The default RAND method makes use of a DRBG.
2917    - There is a public and private DRBG instance.
2918    - The DRBG instances are fork-safe.
2919    - Keep all global DRBG instances on the secure heap if it is enabled.
2920    - The public and private DRBG instance are per thread for lock free
2921      operation
2922
2923   *Paul Dale, Benjamin Kaduk, Kurt Roeckx, Rich Salz, Matthias St. Pierre*
2924
2925 * Changed Configure so it only says what it does and doesn't dump
2926   so much data.  Instead, ./configdata.pm should be used as a script
2927   to display all sorts of configuration data.
2928
2929   *Richard Levitte*
2930
2931 * Added processing of "make variables" to Configure.
2932
2933   *Richard Levitte*
2934
2935 * Added SHA512/224 and SHA512/256 algorithm support.
2936
2937   *Paul Dale*
2938
2939 * The last traces of Netware support, first removed in 1.1.0, have
2940   now been removed.
2941
2942   *Rich Salz*
2943
2944 * Get rid of Makefile.shared, and in the process, make the processing
2945   of certain files (rc.obj, or the .def/.map/.opt files produced from
2946   the ordinal files) more visible and hopefully easier to trace and
2947   debug (or make silent).
2948
2949   *Richard Levitte*
2950
2951 * Make it possible to have environment variable assignments as
2952   arguments to config / Configure.
2953
2954   *Richard Levitte*
2955
2956 * Add multi-prime RSA (RFC 8017) support.
2957
2958   *Paul Yang*
2959
2960 * Add SM3 implemented according to GB/T 32905-2016
2961   *Jack Lloyd <jack.lloyd@ribose.com>,*
2962   *Ronald Tse <ronald.tse@ribose.com>,*
2963   *Erick Borsboom <erick.borsboom@ribose.com>*
2964
2965 * Add 'Maximum Fragment Length' TLS extension negotiation and support
2966   as documented in RFC6066.
2967   Based on a patch from Tomasz Moń
2968
2969   *Filipe Raimundo da Silva*
2970
2971 * Add SM4 implemented according to GB/T 32907-2016.
2972   *Jack Lloyd <jack.lloyd@ribose.com>,*
2973   *Ronald Tse <ronald.tse@ribose.com>,*
2974   *Erick Borsboom <erick.borsboom@ribose.com>*
2975
2976 * Reimplement -newreq-nodes and ERR_error_string_n; the
2977   original author does not agree with the license change.
2978
2979   *Rich Salz*
2980
2981 * Add ARIA AEAD TLS support.
2982
2983   *Jon Spillett*
2984
2985 * Some macro definitions to support VS6 have been removed.  Visual
2986   Studio 6 has not worked since 1.1.0
2987
2988   *Rich Salz*
2989
2990 * Add ERR_clear_last_mark(), to allow callers to clear the last mark
2991   without clearing the errors.
2992
2993   *Richard Levitte*
2994
2995 * Add "atfork" functions.  If building on a system that without
2996   pthreads, see doc/man3/OPENSSL_fork_prepare.pod for application
2997   requirements.  The RAND facility now uses/requires this.
2998
2999   *Rich Salz*
3000
3001 * Add SHA3.
3002
3003   *Andy Polyakov*
3004
3005 * The UI API becomes a permanent and integral part of libcrypto, i.e.
3006   not possible to disable entirely.  However, it's still possible to
3007   disable the console reading UI method, UI_OpenSSL() (use UI_null()
3008   as a fallback).
3009
3010   To disable, configure with 'no-ui-console'.  'no-ui' is still
3011   possible to use as an alias.  Check at compile time with the
3012   macro OPENSSL_NO_UI_CONSOLE.  The macro OPENSSL_NO_UI is still
3013   possible to check and is an alias for OPENSSL_NO_UI_CONSOLE.
3014
3015   *Richard Levitte*
3016
3017 * Add a STORE module, which implements a uniform and URI based reader of
3018   stores that can contain keys, certificates, CRLs and numerous other
3019   objects.  The main API is loosely based on a few stdio functions,
3020   and includes OSSL_STORE_open, OSSL_STORE_load, OSSL_STORE_eof,
3021   OSSL_STORE_error and OSSL_STORE_close.
3022   The implementation uses backends called "loaders" to implement arbitrary
3023   URI schemes.  There is one built in "loader" for the 'file' scheme.
3024
3025   *Richard Levitte*
3026
3027 * Add devcrypto engine.  This has been implemented against cryptodev-linux,
3028   then adjusted to work on FreeBSD 8.4 as well.
3029   Enable by configuring with 'enable-devcryptoeng'.  This is done by default
3030   on BSD implementations, as cryptodev.h is assumed to exist on all of them.
3031
3032   *Richard Levitte*
3033
3034 * Module names can prefixed with OSSL_ or OPENSSL_.  This affects
3035   util/mkerr.pl, which is adapted to allow those prefixes, leading to
3036   error code calls like this:
3037
3038           OSSL_FOOerr(OSSL_FOO_F_SOMETHING, OSSL_FOO_R_WHATEVER);
3039
3040   With this change, we claim the namespaces OSSL and OPENSSL in a manner
3041   that can be encoded in C.  For the foreseeable future, this will only
3042   affect new modules.
3043
3044   *Richard Levitte and Tim Hudson*
3045
3046 * Removed BSD cryptodev engine.
3047
3048   *Rich Salz*
3049
3050 * Add a build target 'build_all_generated', to build all generated files
3051   and only that.  This can be used to prepare everything that requires
3052   things like perl for a system that lacks perl and then move everything
3053   to that system and do the rest of the build there.
3054
3055   *Richard Levitte*
3056
3057 * In the UI interface, make it possible to duplicate the user data.  This
3058   can be used by engines that need to retain the data for a longer time
3059   than just the call where this user data is passed.
3060
3061   *Richard Levitte*
3062
3063 * Ignore the '-named_curve auto' value for compatibility of applications
3064   with OpenSSL 1.0.2.
3065
3066   *Tomáš Mráz <tmraz@fedoraproject.org>*
3067
3068 * Fragmented SSL/TLS alerts are no longer accepted. An alert message is 2
3069   bytes long. In theory it is permissible in SSLv3 - TLSv1.2 to fragment such
3070   alerts across multiple records (some of which could be empty). In practice
3071   it make no sense to send an empty alert record, or to fragment one. TLSv1.3
3072   prohibits this altogether and other libraries (BoringSSL, NSS) do not
3073   support this at all. Supporting it adds significant complexity to the
3074   record layer, and its removal is unlikely to cause interoperability
3075   issues.
3076
3077   *Matt Caswell*
3078
3079 * Add the ASN.1 types INT32, UINT32, INT64, UINT64 and variants prefixed
3080   with Z.  These are meant to replace LONG and ZLONG and to be size safe.
3081   The use of LONG and ZLONG is discouraged and scheduled for deprecation
3082   in OpenSSL 1.2.0.
3083
3084   *Richard Levitte*
3085
3086 * Add the 'z' and 'j' modifiers to BIO_printf() et al formatting string,
3087   'z' is to be used for [s]size_t, and 'j' - with [u]int64_t.
3088
3089   *Richard Levitte, Andy Polyakov*
3090
3091 * Add EC_KEY_get0_engine(), which does for EC_KEY what RSA_get0_engine()
3092   does for RSA, etc.
3093
3094   *Richard Levitte*
3095
3096 * Have 'config' recognise 64-bit mingw and choose 'mingw64' as the target
3097   platform rather than 'mingw'.
3098
3099   *Richard Levitte*
3100
3101 * The functions X509_STORE_add_cert and X509_STORE_add_crl return
3102   success if they are asked to add an object which already exists
3103   in the store. This change cascades to other functions which load
3104   certificates and CRLs.
3105
3106   *Paul Dale*
3107
3108 * x86_64 assembly pack: annotate code with DWARF CFI directives to
3109   facilitate stack unwinding even from assembly subroutines.
3110
3111   *Andy Polyakov*
3112
3113 * Remove VAX C specific definitions of OPENSSL_EXPORT, OPENSSL_EXTERN.
3114   Also remove OPENSSL_GLOBAL entirely, as it became a no-op.
3115
3116   *Richard Levitte*
3117
3118 * Remove the VMS-specific reimplementation of gmtime from crypto/o_times.c.
3119   VMS C's RTL has a fully up to date gmtime() and gmtime_r() since V7.1,
3120   which is the minimum version we support.
3121
3122   *Richard Levitte*
3123
3124 * Certificate time validation (X509_cmp_time) enforces stricter
3125   compliance with RFC 5280. Fractional seconds and timezone offsets
3126   are no longer allowed.
3127
3128   *Emilia Käsper*
3129
3130 * Add support for ARIA
3131
3132   *Paul Dale*
3133
3134 * s_client will now send the Server Name Indication (SNI) extension by
3135   default unless the new "-noservername" option is used. The server name is
3136   based on the host provided to the "-connect" option unless overridden by
3137   using "-servername".
3138
3139   *Matt Caswell*
3140
3141 * Add support for SipHash
3142
3143   *Todd Short*
3144
3145 * OpenSSL now fails if it receives an unrecognised record type in TLS1.0
3146   or TLS1.1. Previously this only happened in SSLv3 and TLS1.2. This is to
3147   prevent issues where no progress is being made and the peer continually
3148   sends unrecognised record types, using up resources processing them.
3149
3150   *Matt Caswell*
3151
3152 * 'openssl passwd' can now produce SHA256 and SHA512 based output,
3153   using the algorithm defined in
3154   <https://www.akkadia.org/drepper/SHA-crypt.txt>
3155
3156   *Richard Levitte*
3157
3158 * Heartbeat support has been removed; the ABI is changed for now.
3159
3160   *Richard Levitte, Rich Salz*
3161
3162 * Support for SSL_OP_NO_ENCRYPT_THEN_MAC in SSL_CONF_cmd.
3163
3164   *Emilia Käsper*
3165
3166 * The RSA "null" method, which was partially supported to avoid patent
3167   issues, has been replaced to always returns NULL.
3168
3169   *Rich Salz*
3170
3171OpenSSL 1.1.0
3172-------------
3173
3174### Changes between 1.1.0k and 1.1.0l [10 Sep 2019]
3175
3176 * For built-in EC curves, ensure an EC_GROUP built from the curve name is
3177   used even when parsing explicit parameters, when loading a encoded key
3178   or calling `EC_GROUP_new_from_ecpkparameters()`/
3179   `EC_GROUP_new_from_ecparameters()`.
3180   This prevents bypass of security hardening and performance gains,
3181   especially for curves with specialized EC_METHODs.
3182   By default, if a key encoded with explicit parameters is loaded and later
3183   encoded, the output is still encoded with explicit parameters, even if
3184   internally a "named" EC_GROUP is used for computation.
3185
3186   *Nicola Tuveri*
3187
3188 * Compute ECC cofactors if not provided during EC_GROUP construction. Before
3189   this change, EC_GROUP_set_generator would accept order and/or cofactor as
3190   NULL. After this change, only the cofactor parameter can be NULL. It also
3191   does some minimal sanity checks on the passed order.
3192   ([CVE-2019-1547])
3193
3194   *Billy Bob Brumley*
3195
3196 * Fixed a padding oracle in PKCS7_dataDecode and CMS_decrypt_set1_pkey.
3197   An attack is simple, if the first CMS_recipientInfo is valid but the
3198   second CMS_recipientInfo is chosen ciphertext. If the second
3199   recipientInfo decodes to PKCS #1 v1.5 form plaintext, the correct
3200   encryption key will be replaced by garbage, and the message cannot be
3201   decoded, but if the RSA decryption fails, the correct encryption key is
3202   used and the recipient will not notice the attack.
3203   As a work around for this potential attack the length of the decrypted
3204   key must be equal to the cipher default key length, in case the
3205   certifiate is not given and all recipientInfo are tried out.
3206   The old behaviour can be re-enabled in the CMS code by setting the
3207   CMS_DEBUG_DECRYPT flag.
3208   ([CVE-2019-1563])
3209
3210   *Bernd Edlinger*
3211
3212 * Use Windows installation paths in the mingw builds
3213
3214   Mingw isn't a POSIX environment per se, which means that Windows
3215   paths should be used for installation.
3216   ([CVE-2019-1552])
3217
3218   *Richard Levitte*
3219
3220### Changes between 1.1.0j and 1.1.0k [28 May 2019]
3221
3222 * Change the default RSA, DSA and DH size to 2048 bit instead of 1024.
3223   This changes the size when using the `genpkey` command when no size is given.
3224   It fixes an omission in earlier changes that changed all RSA, DSA and DH
3225   generation commands to use 2048 bits by default.
3226
3227   *Kurt Roeckx*
3228
3229 * Prevent over long nonces in ChaCha20-Poly1305.
3230
3231   ChaCha20-Poly1305 is an AEAD cipher, and requires a unique nonce input
3232   for every encryption operation. RFC 7539 specifies that the nonce value
3233   (IV) should be 96 bits (12 bytes). OpenSSL allows a variable nonce length
3234   and front pads the nonce with 0 bytes if it is less than 12
3235   bytes. However it also incorrectly allows a nonce to be set of up to 16
3236   bytes. In this case only the last 12 bytes are significant and any
3237   additional leading bytes are ignored.
3238
3239   It is a requirement of using this cipher that nonce values are
3240   unique. Messages encrypted using a reused nonce value are susceptible to
3241   serious confidentiality and integrity attacks. If an application changes
3242   the default nonce length to be longer than 12 bytes and then makes a
3243   change to the leading bytes of the nonce expecting the new value to be a
3244   new unique nonce then such an application could inadvertently encrypt
3245   messages with a reused nonce.
3246
3247   Additionally the ignored bytes in a long nonce are not covered by the
3248   integrity guarantee of this cipher. Any application that relies on the
3249   integrity of these ignored leading bytes of a long nonce may be further
3250   affected. Any OpenSSL internal use of this cipher, including in SSL/TLS,
3251   is safe because no such use sets such a long nonce value. However user
3252   applications that use this cipher directly and set a non-default nonce
3253   length to be longer than 12 bytes may be vulnerable.
3254
3255   This issue was reported to OpenSSL on 16th of March 2019 by Joran Dirk
3256   Greef of Ronomon.
3257   ([CVE-2019-1543])
3258
3259   *Matt Caswell*
3260
3261 * Added SCA hardening for modular field inversion in EC_GROUP through
3262   a new dedicated field_inv() pointer in EC_METHOD.
3263   This also addresses a leakage affecting conversions from projective
3264   to affine coordinates.
3265
3266   *Billy Bob Brumley, Nicola Tuveri*
3267
3268 * Fix a use after free bug in d2i_X509_PUBKEY when overwriting a
3269   re-used X509_PUBKEY object if the second PUBKEY is malformed.
3270
3271   *Bernd Edlinger*
3272
3273 * Move strictness check from EVP_PKEY_asn1_new() to EVP_PKEY_asn1_add0().
3274
3275   *Richard Levitte*
3276
3277 * Remove the 'dist' target and add a tarball building script.  The
3278   'dist' target has fallen out of use, and it shouldn't be
3279   necessary to configure just to create a source distribution.
3280
3281   *Richard Levitte*
3282
3283### Changes between 1.1.0i and 1.1.0j [20 Nov 2018]
3284
3285 * Timing vulnerability in DSA signature generation
3286
3287   The OpenSSL DSA signature algorithm has been shown to be vulnerable to a
3288   timing side channel attack. An attacker could use variations in the signing
3289   algorithm to recover the private key.
3290
3291   This issue was reported to OpenSSL on 16th October 2018 by Samuel Weiser.
3292   ([CVE-2018-0734])
3293
3294   *Paul Dale*
3295
3296 * Timing vulnerability in ECDSA signature generation
3297
3298   The OpenSSL ECDSA signature algorithm has been shown to be vulnerable to a
3299   timing side channel attack. An attacker could use variations in the signing
3300   algorithm to recover the private key.
3301
3302   This issue was reported to OpenSSL on 25th October 2018 by Samuel Weiser.
3303   ([CVE-2018-0735])
3304
3305   *Paul Dale*
3306
3307 * Add coordinate blinding for EC_POINT and implement projective
3308   coordinate blinding for generic prime curves as a countermeasure to
3309   chosen point SCA attacks.
3310
3311   *Sohaib ul Hassan, Nicola Tuveri, Billy Bob Brumley*
3312
3313### Changes between 1.1.0h and 1.1.0i [14 Aug 2018]
3314
3315 * Client DoS due to large DH parameter
3316
3317   During key agreement in a TLS handshake using a DH(E) based ciphersuite a
3318   malicious server can send a very large prime value to the client. This will
3319   cause the client to spend an unreasonably long period of time generating a
3320   key for this prime resulting in a hang until the client has finished. This
3321   could be exploited in a Denial Of Service attack.
3322
3323   This issue was reported to OpenSSL on 5th June 2018 by Guido Vranken
3324   ([CVE-2018-0732])
3325
3326   *Guido Vranken*
3327
3328 * Cache timing vulnerability in RSA Key Generation
3329
3330   The OpenSSL RSA Key generation algorithm has been shown to be vulnerable to
3331   a cache timing side channel attack. An attacker with sufficient access to
3332   mount cache timing attacks during the RSA key generation process could
3333   recover the private key.
3334
3335   This issue was reported to OpenSSL on 4th April 2018 by Alejandro Cabrera
3336   Aldaya, Billy Brumley, Cesar Pereida Garcia and Luis Manuel Alvarez Tapia.
3337   ([CVE-2018-0737])
3338
3339   *Billy Brumley*
3340
3341 * Make EVP_PKEY_asn1_new() a bit stricter about its input.  A NULL pem_str
3342   parameter is no longer accepted, as it leads to a corrupt table.  NULL
3343   pem_str is reserved for alias entries only.
3344
3345   *Richard Levitte*
3346
3347 * Revert blinding in ECDSA sign and instead make problematic addition
3348   length-invariant. Switch even to fixed-length Montgomery multiplication.
3349
3350   *Andy Polyakov*
3351
3352 * Change generating and checking of primes so that the error rate of not
3353   being prime depends on the intended use based on the size of the input.
3354   For larger primes this will result in more rounds of Miller-Rabin.
3355   The maximal error rate for primes with more than 1080 bits is lowered
3356   to 2^-128.
3357
3358   *Kurt Roeckx, Annie Yousar*
3359
3360 * Increase the number of Miller-Rabin rounds for DSA key generating to 64.
3361
3362   *Kurt Roeckx*
3363
3364 * Add blinding to ECDSA and DSA signatures to protect against side channel
3365   attacks discovered by Keegan Ryan (NCC Group).
3366
3367   *Matt Caswell*
3368
3369 * When unlocking a pass phrase protected PEM file or PKCS#8 container, we
3370   now allow empty (zero character) pass phrases.
3371
3372   *Richard Levitte*
3373
3374 * Certificate time validation (X509_cmp_time) enforces stricter
3375   compliance with RFC 5280. Fractional seconds and timezone offsets
3376   are no longer allowed.
3377
3378   *Emilia Käsper*
3379
3380 * Fixed a text canonicalisation bug in CMS
3381
3382   Where a CMS detached signature is used with text content the text goes
3383   through a canonicalisation process first prior to signing or verifying a
3384   signature. This process strips trailing space at the end of lines, converts
3385   line terminators to CRLF and removes additional trailing line terminators
3386   at the end of a file. A bug in the canonicalisation process meant that
3387   some characters, such as form-feed, were incorrectly treated as whitespace
3388   and removed. This is contrary to the specification (RFC5485). This fix
3389   could mean that detached text data signed with an earlier version of
3390   OpenSSL 1.1.0 may fail to verify using the fixed version, or text data
3391   signed with a fixed OpenSSL may fail to verify with an earlier version of
3392   OpenSSL 1.1.0. A workaround is to only verify the canonicalised text data
3393   and use the "-binary" flag (for the "cms" command line application) or set
3394   the SMIME_BINARY/PKCS7_BINARY/CMS_BINARY flags (if using CMS_verify()).
3395
3396   *Matt Caswell*
3397
3398### Changes between 1.1.0g and 1.1.0h [27 Mar 2018]
3399
3400 * Constructed ASN.1 types with a recursive definition could exceed the stack
3401
3402   Constructed ASN.1 types with a recursive definition (such as can be found
3403   in PKCS7) could eventually exceed the stack given malicious input with
3404   excessive recursion. This could result in a Denial Of Service attack. There
3405   are no such structures used within SSL/TLS that come from untrusted sources
3406   so this is considered safe.
3407
3408   This issue was reported to OpenSSL on 4th January 2018 by the OSS-fuzz
3409   project.
3410   ([CVE-2018-0739])
3411
3412   *Matt Caswell*
3413
3414 * Incorrect CRYPTO_memcmp on HP-UX PA-RISC
3415
3416   Because of an implementation bug the PA-RISC CRYPTO_memcmp function is
3417   effectively reduced to only comparing the least significant bit of each
3418   byte. This allows an attacker to forge messages that would be considered as
3419   authenticated in an amount of tries lower than that guaranteed by the
3420   security claims of the scheme. The module can only be compiled by the
3421   HP-UX assembler, so that only HP-UX PA-RISC targets are affected.
3422
3423   This issue was reported to OpenSSL on 2nd March 2018 by Peter Waltenberg
3424   (IBM).
3425   ([CVE-2018-0733])
3426
3427   *Andy Polyakov*
3428
3429 * Add a build target 'build_all_generated', to build all generated files
3430   and only that.  This can be used to prepare everything that requires
3431   things like perl for a system that lacks perl and then move everything
3432   to that system and do the rest of the build there.
3433
3434   *Richard Levitte*
3435
3436 * Backport SSL_OP_NO_RENGOTIATION
3437
3438   OpenSSL 1.0.2 and below had the ability to disable renegotiation using the
3439   (undocumented) SSL3_FLAGS_NO_RENEGOTIATE_CIPHERS flag. Due to the opacity
3440   changes this is no longer possible in 1.1.0. Therefore the new
3441   SSL_OP_NO_RENEGOTIATION option from 1.1.1-dev has been backported to
3442   1.1.0 to provide equivalent functionality.
3443
3444   Note that if an application built against 1.1.0h headers (or above) is run
3445   using an older version of 1.1.0 (prior to 1.1.0h) then the option will be
3446   accepted but nothing will happen, i.e. renegotiation will not be prevented.
3447
3448   *Matt Caswell*
3449
3450 * Removed the OS390-Unix config target.  It relied on a script that doesn't
3451   exist.
3452
3453   *Rich Salz*
3454
3455 * rsaz_1024_mul_avx2 overflow bug on x86_64
3456
3457   There is an overflow bug in the AVX2 Montgomery multiplication procedure
3458   used in exponentiation with 1024-bit moduli. No EC algorithms are affected.
3459   Analysis suggests that attacks against RSA and DSA as a result of this
3460   defect would be very difficult to perform and are not believed likely.
3461   Attacks against DH1024 are considered just feasible, because most of the
3462   work necessary to deduce information about a private key may be performed
3463   offline. The amount of resources required for such an attack would be
3464   significant. However, for an attack on TLS to be meaningful, the server
3465   would have to share the DH1024 private key among multiple clients, which is
3466   no longer an option since CVE-2016-0701.
3467
3468   This only affects processors that support the AVX2 but not ADX extensions
3469   like Intel Haswell (4th generation).
3470
3471   This issue was reported to OpenSSL by David Benjamin (Google). The issue
3472   was originally found via the OSS-Fuzz project.
3473   ([CVE-2017-3738])
3474
3475   *Andy Polyakov*
3476
3477### Changes between 1.1.0f and 1.1.0g [2 Nov 2017]
3478
3479 * bn_sqrx8x_internal carry bug on x86_64
3480
3481   There is a carry propagating bug in the x86_64 Montgomery squaring
3482   procedure. No EC algorithms are affected. Analysis suggests that attacks
3483   against RSA and DSA as a result of this defect would be very difficult to
3484   perform and are not believed likely. Attacks against DH are considered just
3485   feasible (although very difficult) because most of the work necessary to
3486   deduce information about a private key may be performed offline. The amount
3487   of resources required for such an attack would be very significant and
3488   likely only accessible to a limited number of attackers. An attacker would
3489   additionally need online access to an unpatched system using the target
3490   private key in a scenario with persistent DH parameters and a private
3491   key that is shared between multiple clients.
3492
3493   This only affects processors that support the BMI1, BMI2 and ADX extensions
3494   like Intel Broadwell (5th generation) and later or AMD Ryzen.
3495
3496   This issue was reported to OpenSSL by the OSS-Fuzz project.
3497   ([CVE-2017-3736])
3498
3499   *Andy Polyakov*
3500
3501 * Malformed X.509 IPAddressFamily could cause OOB read
3502
3503   If an X.509 certificate has a malformed IPAddressFamily extension,
3504   OpenSSL could do a one-byte buffer overread. The most likely result
3505   would be an erroneous display of the certificate in text format.
3506
3507   This issue was reported to OpenSSL by the OSS-Fuzz project.
3508   ([CVE-2017-3735])
3509
3510   *Rich Salz*
3511
3512### Changes between 1.1.0e and 1.1.0f [25 May 2017]
3513
3514 * Have 'config' recognise 64-bit mingw and choose 'mingw64' as the target
3515   platform rather than 'mingw'.
3516
3517   *Richard Levitte*
3518
3519 * Remove the VMS-specific reimplementation of gmtime from crypto/o_times.c.
3520   VMS C's RTL has a fully up to date gmtime() and gmtime_r() since V7.1,
3521   which is the minimum version we support.
3522
3523   *Richard Levitte*
3524
3525### Changes between 1.1.0d and 1.1.0e [16 Feb 2017]
3526
3527 * Encrypt-Then-Mac renegotiation crash
3528
3529   During a renegotiation handshake if the Encrypt-Then-Mac extension is
3530   negotiated where it was not in the original handshake (or vice-versa) then
3531   this can cause OpenSSL to crash (dependant on ciphersuite). Both clients
3532   and servers are affected.
3533
3534   This issue was reported to OpenSSL by Joe Orton (Red Hat).
3535   ([CVE-2017-3733])
3536
3537   *Matt Caswell*
3538
3539### Changes between 1.1.0c and 1.1.0d [26 Jan 2017]
3540
3541 * Truncated packet could crash via OOB read
3542
3543   If one side of an SSL/TLS path is running on a 32-bit host and a specific
3544   cipher is being used, then a truncated packet can cause that host to
3545   perform an out-of-bounds read, usually resulting in a crash.
3546
3547   This issue was reported to OpenSSL by Robert Święcki of Google.
3548   ([CVE-2017-3731])
3549
3550   *Andy Polyakov*
3551
3552 * Bad (EC)DHE parameters cause a client crash
3553
3554   If a malicious server supplies bad parameters for a DHE or ECDHE key
3555   exchange then this can result in the client attempting to dereference a
3556   NULL pointer leading to a client crash. This could be exploited in a Denial
3557   of Service attack.
3558
3559   This issue was reported to OpenSSL by Guido Vranken.
3560   ([CVE-2017-3730])
3561
3562   *Matt Caswell*
3563
3564 * BN_mod_exp may produce incorrect results on x86_64
3565
3566   There is a carry propagating bug in the x86_64 Montgomery squaring
3567   procedure. No EC algorithms are affected. Analysis suggests that attacks
3568   against RSA and DSA as a result of this defect would be very difficult to
3569   perform and are not believed likely. Attacks against DH are considered just
3570   feasible (although very difficult) because most of the work necessary to
3571   deduce information about a private key may be performed offline. The amount
3572   of resources required for such an attack would be very significant and
3573   likely only accessible to a limited number of attackers. An attacker would
3574   additionally need online access to an unpatched system using the target
3575   private key in a scenario with persistent DH parameters and a private
3576   key that is shared between multiple clients. For example this can occur by
3577   default in OpenSSL DHE based SSL/TLS ciphersuites. Note: This issue is very
3578   similar to CVE-2015-3193 but must be treated as a separate problem.
3579
3580   This issue was reported to OpenSSL by the OSS-Fuzz project.
3581   ([CVE-2017-3732])
3582
3583   *Andy Polyakov*
3584
3585### Changes between 1.1.0b and 1.1.0c [10 Nov 2016]
3586
3587 * ChaCha20/Poly1305 heap-buffer-overflow
3588
3589   TLS connections using `*-CHACHA20-POLY1305` ciphersuites are susceptible to
3590   a DoS attack by corrupting larger payloads. This can result in an OpenSSL
3591   crash. This issue is not considered to be exploitable beyond a DoS.
3592
3593   This issue was reported to OpenSSL by Robert Święcki (Google Security Team)
3594   ([CVE-2016-7054])
3595
3596   *Richard Levitte*
3597
3598 * CMS Null dereference
3599
3600   Applications parsing invalid CMS structures can crash with a NULL pointer
3601   dereference. This is caused by a bug in the handling of the ASN.1 CHOICE
3602   type in OpenSSL 1.1.0 which can result in a NULL value being passed to the
3603   structure callback if an attempt is made to free certain invalid encodings.
3604   Only CHOICE structures using a callback which do not handle NULL value are
3605   affected.
3606
3607   This issue was reported to OpenSSL by Tyler Nighswander of ForAllSecure.
3608   ([CVE-2016-7053])
3609
3610   *Stephen Henson*
3611
3612 * Montgomery multiplication may produce incorrect results
3613
3614   There is a carry propagating bug in the Broadwell-specific Montgomery
3615   multiplication procedure that handles input lengths divisible by, but
3616   longer than 256 bits. Analysis suggests that attacks against RSA, DSA
3617   and DH private keys are impossible. This is because the subroutine in
3618   question is not used in operations with the private key itself and an input
3619   of the attacker's direct choice. Otherwise the bug can manifest itself as
3620   transient authentication and key negotiation failures or reproducible
3621   erroneous outcome of public-key operations with specially crafted input.
3622   Among EC algorithms only Brainpool P-512 curves are affected and one
3623   presumably can attack ECDH key negotiation. Impact was not analyzed in
3624   detail, because pre-requisites for attack are considered unlikely. Namely
3625   multiple clients have to choose the curve in question and the server has to
3626   share the private key among them, neither of which is default behaviour.
3627   Even then only clients that chose the curve will be affected.
3628
3629   This issue was publicly reported as transient failures and was not
3630   initially recognized as a security issue. Thanks to Richard Morgan for
3631   providing reproducible case.
3632   ([CVE-2016-7055])
3633
3634   *Andy Polyakov*
3635
3636 * Removed automatic addition of RPATH in shared libraries and executables,
3637   as this was a remainder from OpenSSL 1.0.x and isn't needed any more.
3638
3639   *Richard Levitte*
3640
3641### Changes between 1.1.0a and 1.1.0b [26 Sep 2016]
3642
3643 * Fix Use After Free for large message sizes
3644
3645   The patch applied to address CVE-2016-6307 resulted in an issue where if a
3646   message larger than approx 16k is received then the underlying buffer to
3647   store the incoming message is reallocated and moved. Unfortunately a
3648   dangling pointer to the old location is left which results in an attempt to
3649   write to the previously freed location. This is likely to result in a
3650   crash, however it could potentially lead to execution of arbitrary code.
3651
3652   This issue only affects OpenSSL 1.1.0a.
3653
3654   This issue was reported to OpenSSL by Robert Święcki.
3655   ([CVE-2016-6309])
3656
3657   *Matt Caswell*
3658
3659### Changes between 1.1.0 and 1.1.0a [22 Sep 2016]
3660
3661 * OCSP Status Request extension unbounded memory growth
3662
3663   A malicious client can send an excessively large OCSP Status Request
3664   extension. If that client continually requests renegotiation, sending a
3665   large OCSP Status Request extension each time, then there will be unbounded
3666   memory growth on the server. This will eventually lead to a Denial Of
3667   Service attack through memory exhaustion. Servers with a default
3668   configuration are vulnerable even if they do not support OCSP. Builds using
3669   the "no-ocsp" build time option are not affected.
3670
3671   This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.)
3672   ([CVE-2016-6304])
3673
3674   *Matt Caswell*
3675
3676 * SSL_peek() hang on empty record
3677
3678   OpenSSL 1.1.0 SSL/TLS will hang during a call to SSL_peek() if the peer
3679   sends an empty record. This could be exploited by a malicious peer in a
3680   Denial Of Service attack.
3681
3682   This issue was reported to OpenSSL by Alex Gaynor.
3683   ([CVE-2016-6305])
3684
3685   *Matt Caswell*
3686
3687 * Excessive allocation of memory in tls_get_message_header() and
3688   dtls1_preprocess_fragment()
3689
3690   A (D)TLS message includes 3 bytes for its length in the header for the
3691   message. This would allow for messages up to 16Mb in length. Messages of
3692   this length are excessive and OpenSSL includes a check to ensure that a
3693   peer is sending reasonably sized messages in order to avoid too much memory
3694   being consumed to service a connection. A flaw in the logic of version
3695   1.1.0 means that memory for the message is allocated too early, prior to
3696   the excessive message length check. Due to way memory is allocated in
3697   OpenSSL this could mean an attacker could force up to 21Mb to be allocated
3698   to service a connection. This could lead to a Denial of Service through
3699   memory exhaustion. However, the excessive message length check still takes
3700   place, and this would cause the connection to immediately fail. Assuming
3701   that the application calls SSL_free() on the failed connection in a timely
3702   manner then the 21Mb of allocated memory will then be immediately freed
3703   again. Therefore the excessive memory allocation will be transitory in
3704   nature. This then means that there is only a security impact if:
3705
3706   1) The application does not call SSL_free() in a timely manner in the event
3707   that the connection fails
3708   or
3709   2) The application is working in a constrained environment where there is
3710   very little free memory
3711   or
3712   3) The attacker initiates multiple connection attempts such that there are
3713   multiple connections in a state where memory has been allocated for the
3714   connection; SSL_free() has not yet been called; and there is insufficient
3715   memory to service the multiple requests.
3716
3717   Except in the instance of (1) above any Denial Of Service is likely to be
3718   transitory because as soon as the connection fails the memory is
3719   subsequently freed again in the SSL_free() call. However there is an
3720   increased risk during this period of application crashes due to the lack of
3721   memory - which would then mean a more serious Denial of Service.
3722
3723   This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.)
3724   (CVE-2016-6307 and CVE-2016-6308)
3725
3726   *Matt Caswell*
3727
3728 * solaris-x86-cc, i.e. 32-bit configuration with vendor compiler,
3729   had to be removed. Primary reason is that vendor assembler can't
3730   assemble our modules with -KPIC flag. As result it, assembly
3731   support, was not even available as option. But its lack means
3732   lack of side-channel resistant code, which is incompatible with
3733   security by todays standards. Fortunately gcc is readily available
3734   prepackaged option, which we firmly point at...
3735
3736   *Andy Polyakov*
3737
3738### Changes between 1.0.2h and 1.1.0  [25 Aug 2016]
3739
3740 * Windows command-line tool supports UTF-8 opt-in option for arguments
3741   and console input. Setting OPENSSL_WIN32_UTF8 environment variable
3742   (to any value) allows Windows user to access PKCS#12 file generated
3743   with Windows CryptoAPI and protected with non-ASCII password, as well
3744   as files generated under UTF-8 locale on Linux also protected with
3745   non-ASCII password.
3746
3747   *Andy Polyakov*
3748
3749 * To mitigate the SWEET32 attack ([CVE-2016-2183]), 3DES cipher suites
3750   have been disabled by default and removed from DEFAULT, just like RC4.
3751   See the RC4 item below to re-enable both.
3752
3753   *Rich Salz*
3754
3755 * The method for finding the storage location for the Windows RAND seed file
3756   has changed. First we check %RANDFILE%. If that is not set then we check
3757   the directories %HOME%, %USERPROFILE% and %SYSTEMROOT% in that order. If
3758   all else fails we fall back to C:\.
3759
3760   *Matt Caswell*
3761
3762 * The EVP_EncryptUpdate() function has had its return type changed from void
3763   to int. A return of 0 indicates and error while a return of 1 indicates
3764   success.
3765
3766   *Matt Caswell*
3767
3768 * The flags RSA_FLAG_NO_CONSTTIME, DSA_FLAG_NO_EXP_CONSTTIME and
3769   DH_FLAG_NO_EXP_CONSTTIME which previously provided the ability to switch
3770   off the constant time implementation for RSA, DSA and DH have been made
3771   no-ops and deprecated.
3772
3773   *Matt Caswell*
3774
3775 * Windows RAND implementation was simplified to only get entropy by
3776   calling CryptGenRandom(). Various other RAND-related tickets
3777   were also closed.
3778
3779   *Joseph Wylie Yandle, Rich Salz*
3780
3781 * The stack and lhash API's were renamed to start with `OPENSSL_SK_`
3782   and `OPENSSL_LH_`, respectively.  The old names are available
3783   with API compatibility.  They new names are now completely documented.
3784
3785   *Rich Salz*
3786
3787 * Unify TYPE_up_ref(obj) methods signature.
3788   SSL_CTX_up_ref(), SSL_up_ref(), X509_up_ref(), EVP_PKEY_up_ref(),
3789   X509_CRL_up_ref(), X509_OBJECT_up_ref_count() methods are now returning an
3790   int (instead of void) like all others TYPE_up_ref() methods.
3791   So now these methods also check the return value of CRYPTO_atomic_add(),
3792   and the validity of object reference counter.
3793
3794   *fdasilvayy@gmail.com*
3795
3796 * With Windows Visual Studio builds, the .pdb files are installed
3797   alongside the installed libraries and executables.  For a static
3798   library installation, ossl_static.pdb is the associate compiler
3799   generated .pdb file to be used when linking programs.
3800
3801   *Richard Levitte*
3802
3803 * Remove openssl.spec.  Packaging files belong with the packagers.
3804
3805   *Richard Levitte*
3806
3807 * Automatic Darwin/OSX configuration has had a refresh, it will now
3808   recognise x86_64 architectures automatically.  You can still decide
3809   to build for a different bitness with the environment variable
3810   KERNEL_BITS (can be 32 or 64), for example:
3811
3812           KERNEL_BITS=32 ./config
3813
3814   *Richard Levitte*
3815
3816 * Change default algorithms in pkcs8 utility to use PKCS#5 v2.0,
3817   256 bit AES and HMAC with SHA256.
3818
3819   *Steve Henson*
3820
3821 * Remove support for MIPS o32 ABI on IRIX (and IRIX only).
3822
3823   *Andy Polyakov*
3824
3825 * Triple-DES ciphers have been moved from HIGH to MEDIUM.
3826
3827   *Rich Salz*
3828
3829 * To enable users to have their own config files and build file templates,
3830   Configure looks in the directory indicated by the environment variable
3831   OPENSSL_LOCAL_CONFIG_DIR as well as the in-source Configurations/
3832   directory.  On VMS, OPENSSL_LOCAL_CONFIG_DIR is expected to be a logical
3833   name and is used as is.
3834
3835   *Richard Levitte*
3836
3837 * The following datatypes were made opaque: X509_OBJECT, X509_STORE_CTX,
3838   X509_STORE, X509_LOOKUP, and X509_LOOKUP_METHOD.  The unused type
3839   X509_CERT_FILE_CTX was removed.
3840
3841   *Rich Salz*
3842
3843 * "shared" builds are now the default. To create only static libraries use
3844   the "no-shared" Configure option.
3845
3846   *Matt Caswell*
3847
3848 * Remove the no-aes, no-hmac, no-rsa, no-sha and no-md5 Configure options.
3849   All of these option have not worked for some while and are fundamental
3850   algorithms.
3851
3852   *Matt Caswell*
3853
3854 * Make various cleanup routines no-ops and mark them as deprecated. Most
3855   global cleanup functions are no longer required because they are handled
3856   via auto-deinit (see OPENSSL_init_crypto and OPENSSL_init_ssl man pages).
3857   Explicitly de-initing can cause problems (e.g. where a library that uses
3858   OpenSSL de-inits, but an application is still using it). The affected
3859   functions are CONF_modules_free(), ENGINE_cleanup(), OBJ_cleanup(),
3860   EVP_cleanup(), BIO_sock_cleanup(), CRYPTO_cleanup_all_ex_data(),
3861   RAND_cleanup(), SSL_COMP_free_compression_methods(), ERR_free_strings() and
3862   COMP_zlib_cleanup().
3863
3864   *Matt Caswell*
3865
3866 * --strict-warnings no longer enables runtime debugging options
3867   such as REF_DEBUG. Instead, debug options are automatically
3868   enabled with '--debug' builds.
3869
3870   *Andy Polyakov, Emilia Käsper*
3871
3872 * Made DH and DH_METHOD opaque. The structures for managing DH objects
3873   have been moved out of the public header files. New functions for managing
3874   these have been added.
3875
3876   *Matt Caswell*
3877
3878 * Made RSA and RSA_METHOD opaque. The structures for managing RSA
3879   objects have been moved out of the public header files. New
3880   functions for managing these have been added.
3881
3882   *Richard Levitte*
3883
3884 * Made DSA and DSA_METHOD opaque. The structures for managing DSA objects
3885   have been moved out of the public header files. New functions for managing
3886   these have been added.
3887
3888   *Matt Caswell*
3889
3890 * Made BIO and BIO_METHOD opaque. The structures for managing BIOs have been
3891   moved out of the public header files. New functions for managing these
3892   have been added.
3893
3894   *Matt Caswell*
3895
3896 * Removed no-rijndael as a config option. Rijndael is an old name for AES.
3897
3898   *Matt Caswell*
3899
3900 * Removed the mk1mf build scripts.
3901
3902   *Richard Levitte*
3903
3904 * Headers are now wrapped, if necessary, with OPENSSL_NO_xxx, so
3905   it is always safe to #include a header now.
3906
3907   *Rich Salz*
3908
3909 * Removed the aged BC-32 config and all its supporting scripts
3910
3911   *Richard Levitte*
3912
3913 * Removed support for Ultrix, Netware, and OS/2.
3914
3915   *Rich Salz*
3916
3917 * Add support for HKDF.
3918
3919   *Alessandro Ghedini*
3920
3921 * Add support for blake2b and blake2s
3922
3923   *Bill Cox*
3924
3925 * Added support for "pipelining". Ciphers that have the
3926   EVP_CIPH_FLAG_PIPELINE flag set have a capability to process multiple
3927   encryptions/decryptions simultaneously. There are currently no built-in
3928   ciphers with this property but the expectation is that engines will be able
3929   to offer it to significantly improve throughput. Support has been extended
3930   into libssl so that multiple records for a single connection can be
3931   processed in one go (for >=TLS 1.1).
3932
3933   *Matt Caswell*
3934
3935 * Added the AFALG engine. This is an async capable engine which is able to
3936   offload work to the Linux kernel. In this initial version it only supports
3937   AES128-CBC. The kernel must be version 4.1.0 or greater.
3938
3939   *Catriona Lucey*
3940
3941 * OpenSSL now uses a new threading API. It is no longer necessary to
3942   set locking callbacks to use OpenSSL in a multi-threaded environment. There
3943   are two supported threading models: pthreads and windows threads. It is
3944   also possible to configure OpenSSL at compile time for "no-threads". The
3945   old threading API should no longer be used. The functions have been
3946   replaced with "no-op" compatibility macros.
3947
3948   *Alessandro Ghedini, Matt Caswell*
3949
3950 * Modify behavior of ALPN to invoke callback after SNI/servername
3951   callback, such that updates to the SSL_CTX affect ALPN.
3952
3953   *Todd Short*
3954
3955 * Add SSL_CIPHER queries for authentication and key-exchange.
3956
3957   *Todd Short*
3958
3959 * Changes to the DEFAULT cipherlist:
3960   - Prefer (EC)DHE handshakes over plain RSA.
3961   - Prefer AEAD ciphers over legacy ciphers.
3962   - Prefer ECDSA over RSA when both certificates are available.
3963   - Prefer TLSv1.2 ciphers/PRF.
3964   - Remove DSS, SEED, IDEA, CAMELLIA, and AES-CCM from the
3965     default cipherlist.
3966
3967   *Emilia Käsper*
3968
3969 * Change the ECC default curve list to be this, in order: x25519,
3970   secp256r1, secp521r1, secp384r1.
3971
3972   *Rich Salz*
3973
3974 * RC4 based libssl ciphersuites are now classed as "weak" ciphers and are
3975   disabled by default. They can be re-enabled using the
3976   enable-weak-ssl-ciphers option to Configure.
3977
3978   *Matt Caswell*
3979
3980 * If the server has ALPN configured, but supports no protocols that the
3981   client advertises, send a fatal "no_application_protocol" alert.
3982   This behaviour is SHALL in RFC 7301, though it isn't universally
3983   implemented by other servers.
3984
3985   *Emilia Käsper*
3986
3987 * Add X25519 support.
3988   Add ASN.1 and EVP_PKEY methods for X25519. This includes support
3989   for public and private key encoding using the format documented in
3990   draft-ietf-curdle-pkix-02. The corresponding EVP_PKEY method supports
3991   key generation and key derivation.
3992
3993   TLS support complies with draft-ietf-tls-rfc4492bis-08 and uses
3994   X25519(29).
3995
3996   *Steve Henson*
3997
3998 * Deprecate SRP_VBASE_get_by_user.
3999   SRP_VBASE_get_by_user had inconsistent memory management behaviour.
4000   In order to fix an unavoidable memory leak ([CVE-2016-0798]),
4001   SRP_VBASE_get_by_user was changed to ignore the "fake user" SRP
4002   seed, even if the seed is configured.
4003
4004   Users should use SRP_VBASE_get1_by_user instead. Note that in
4005   SRP_VBASE_get1_by_user, caller must free the returned value. Note
4006   also that even though configuring the SRP seed attempts to hide
4007   invalid usernames by continuing the handshake with fake
4008   credentials, this behaviour is not constant time and no strong
4009   guarantees are made that the handshake is indistinguishable from
4010   that of a valid user.
4011
4012   *Emilia Käsper*
4013
4014 * Configuration change; it's now possible to build dynamic engines
4015   without having to build shared libraries and vice versa.  This
4016   only applies to the engines in `engines/`, those in `crypto/engine/`
4017   will always be built into libcrypto (i.e. "static").
4018
4019   Building dynamic engines is enabled by default; to disable, use
4020   the configuration option "disable-dynamic-engine".
4021
4022   The only requirements for building dynamic engines are the
4023   presence of the DSO module and building with position independent
4024   code, so they will also automatically be disabled if configuring
4025   with "disable-dso" or "disable-pic".
4026
4027   The macros OPENSSL_NO_STATIC_ENGINE and OPENSSL_NO_DYNAMIC_ENGINE
4028   are also taken away from openssl/opensslconf.h, as they are
4029   irrelevant.
4030
4031   *Richard Levitte*
4032
4033 * Configuration change; if there is a known flag to compile
4034   position independent code, it will always be applied on the
4035   libcrypto and libssl object files, and never on the application
4036   object files.  This means other libraries that use routines from
4037   libcrypto / libssl can be made into shared libraries regardless
4038   of how OpenSSL was configured.
4039
4040   If this isn't desirable, the configuration options "disable-pic"
4041   or "no-pic" can be used to disable the use of PIC.  This will
4042   also disable building shared libraries and dynamic engines.
4043
4044   *Richard Levitte*
4045
4046 * Removed JPAKE code.  It was experimental and has no wide use.
4047
4048   *Rich Salz*
4049
4050 * The INSTALL_PREFIX Makefile variable has been renamed to
4051   DESTDIR.  That makes for less confusion on what this variable
4052   is for.  Also, the configuration option --install_prefix is
4053   removed.
4054
4055   *Richard Levitte*
4056
4057 * Heartbeat for TLS has been removed and is disabled by default
4058   for DTLS; configure with enable-heartbeats.  Code that uses the
4059   old #define's might need to be updated.
4060
4061   *Emilia Käsper, Rich Salz*
4062
4063 * Rename REF_CHECK to REF_DEBUG.
4064
4065   *Rich Salz*
4066
4067 * New "unified" build system
4068
4069   The "unified" build system is aimed to be a common system for all
4070   platforms we support.  With it comes new support for VMS.
4071
4072   This system builds supports building in a different directory tree
4073   than the source tree.  It produces one Makefile (for unix family
4074   or lookalikes), or one descrip.mms (for VMS).
4075
4076   The source of information to make the Makefile / descrip.mms is
4077   small files called 'build.info', holding the necessary
4078   information for each directory with source to compile, and a
4079   template in Configurations, like unix-Makefile.tmpl or
4080   descrip.mms.tmpl.
4081
4082   With this change, the library names were also renamed on Windows
4083   and on VMS.  They now have names that are closer to the standard
4084   on Unix, and include the major version number, and in certain
4085   cases, the architecture they are built for.  See "Notes on shared
4086   libraries" in INSTALL.
4087
4088   We rely heavily on the perl module Text::Template.
4089
4090   *Richard Levitte*
4091
4092 * Added support for auto-initialisation and de-initialisation of the library.
4093   OpenSSL no longer requires explicit init or deinit routines to be called,
4094   except in certain circumstances. See the OPENSSL_init_crypto() and
4095   OPENSSL_init_ssl() man pages for further information.
4096
4097   *Matt Caswell*
4098
4099 * The arguments to the DTLSv1_listen function have changed. Specifically the
4100   "peer" argument is now expected to be a BIO_ADDR object.
4101
4102 * Rewrite of BIO networking library. The BIO library lacked consistent
4103   support of IPv6, and adding it required some more extensive
4104   modifications.  This introduces the BIO_ADDR and BIO_ADDRINFO types,
4105   which hold all types of addresses and chains of address information.
4106   It also introduces a new API, with functions like BIO_socket,
4107   BIO_connect, BIO_listen, BIO_lookup and a rewrite of BIO_accept.
4108   The source/sink BIOs BIO_s_connect, BIO_s_accept and BIO_s_datagram
4109   have been adapted accordingly.
4110
4111   *Richard Levitte*
4112
4113 * RSA_padding_check_PKCS1_type_1 now accepts inputs with and without
4114   the leading 0-byte.
4115
4116   *Emilia Käsper*
4117
4118 * CRIME protection: disable compression by default, even if OpenSSL is
4119   compiled with zlib enabled. Applications can still enable compression
4120   by calling SSL_CTX_clear_options(ctx, SSL_OP_NO_COMPRESSION), or by
4121   using the SSL_CONF library to configure compression.
4122
4123   *Emilia Käsper*
4124
4125 * The signature of the session callback configured with
4126   SSL_CTX_sess_set_get_cb was changed. The read-only input buffer
4127   was explicitly marked as `const unsigned char*` instead of
4128   `unsigned char*`.
4129
4130   *Emilia Käsper*
4131
4132 * Always DPURIFY. Remove the use of uninitialized memory in the
4133   RNG, and other conditional uses of DPURIFY. This makes -DPURIFY a no-op.
4134
4135   *Emilia Käsper*
4136
4137 * Removed many obsolete configuration items, including
4138      DES_PTR, DES_RISC1, DES_RISC2, DES_INT
4139      MD2_CHAR, MD2_INT, MD2_LONG
4140      BF_PTR, BF_PTR2
4141      IDEA_SHORT, IDEA_LONG
4142      RC2_SHORT, RC2_LONG, RC4_LONG, RC4_CHUNK, RC4_INDEX
4143
4144   *Rich Salz, with advice from Andy Polyakov*
4145
4146 * Many BN internals have been moved to an internal header file.
4147
4148   *Rich Salz with help from Andy Polyakov*
4149
4150 * Configuration and writing out the results from it has changed.
4151   Files such as Makefile include/openssl/opensslconf.h and are now
4152   produced through general templates, such as Makefile.in and
4153   crypto/opensslconf.h.in and some help from the perl module
4154   Text::Template.
4155
4156   Also, the center of configuration information is no longer
4157   Makefile.  Instead, Configure produces a perl module in
4158   configdata.pm which holds most of the config data (in the hash
4159   table %config), the target data that comes from the target
4160   configuration in one of the `Configurations/*.conf` files (in
4161   %target).
4162
4163   *Richard Levitte*
4164
4165 * To clarify their intended purposes, the Configure options
4166   --prefix and --openssldir change their semantics, and become more
4167   straightforward and less interdependent.
4168
4169   --prefix shall be used exclusively to give the location INSTALLTOP
4170   where programs, scripts, libraries, include files and manuals are
4171   going to be installed.  The default is now /usr/local.
4172
4173   --openssldir shall be used exclusively to give the default
4174   location OPENSSLDIR where certificates, private keys, CRLs are
4175   managed.  This is also where the default openssl.cnf gets
4176   installed.
4177   If the directory given with this option is a relative path, the
4178   values of both the --prefix value and the --openssldir value will
4179   be combined to become OPENSSLDIR.
4180   The default for --openssldir is INSTALLTOP/ssl.
4181
4182   Anyone who uses --openssldir to specify where OpenSSL is to be
4183   installed MUST change to use --prefix instead.
4184
4185   *Richard Levitte*
4186
4187 * The GOST engine was out of date and therefore it has been removed. An up
4188   to date GOST engine is now being maintained in an external repository.
4189   See: <https://wiki.openssl.org/index.php/Binaries>. Libssl still retains
4190   support for GOST ciphersuites (these are only activated if a GOST engine
4191   is present).
4192
4193   *Matt Caswell*
4194
4195 * EGD is no longer supported by default; use enable-egd when
4196   configuring.
4197
4198   *Ben Kaduk and Rich Salz*
4199
4200 * The distribution now has Makefile.in files, which are used to
4201   create Makefile's when Configure is run.  *Configure must be run
4202   before trying to build now.*
4203
4204   *Rich Salz*
4205
4206 * The return value for SSL_CIPHER_description() for error conditions
4207   has changed.
4208
4209   *Rich Salz*
4210
4211 * Support for RFC6698/RFC7671 DANE TLSA peer authentication.
4212
4213   Obtaining and performing DNSSEC validation of TLSA records is
4214   the application's responsibility.  The application provides
4215   the TLSA records of its choice to OpenSSL, and these are then
4216   used to authenticate the peer.
4217
4218   The TLSA records need not even come from DNS.  They can, for
4219   example, be used to implement local end-entity certificate or
4220   trust-anchor "pinning", where the "pin" data takes the form
4221   of TLSA records, which can augment or replace verification
4222   based on the usual WebPKI public certification authorities.
4223
4224   *Viktor Dukhovni*
4225
4226 * Revert default OPENSSL_NO_DEPRECATED setting.  Instead OpenSSL
4227   continues to support deprecated interfaces in default builds.
4228   However, applications are strongly advised to compile their
4229   source files with -DOPENSSL_API_COMPAT=0x10100000L, which hides
4230   the declarations of all interfaces deprecated in 0.9.8, 1.0.0
4231   or the 1.1.0 releases.
4232
4233   In environments in which all applications have been ported to
4234   not use any deprecated interfaces OpenSSL's Configure script
4235   should be used with the --api=1.1.0 option to entirely remove
4236   support for the deprecated features from the library and
4237   unconditionally disable them in the installed headers.
4238   Essentially the same effect can be achieved with the "no-deprecated"
4239   argument to Configure, except that this will always restrict
4240   the build to just the latest API, rather than a fixed API
4241   version.
4242
4243   As applications are ported to future revisions of the API,
4244   they should update their compile-time OPENSSL_API_COMPAT define
4245   accordingly, but in most cases should be able to continue to
4246   compile with later releases.
4247
4248   The OPENSSL_API_COMPAT versions for 1.0.0, and 0.9.8 are
4249   0x10000000L and 0x00908000L, respectively.  However those
4250   versions did not support the OPENSSL_API_COMPAT feature, and
4251   so applications are not typically tested for explicit support
4252   of just the undeprecated features of either release.
4253
4254   *Viktor Dukhovni*
4255
4256 * Add support for setting the minimum and maximum supported protocol.
4257   It can bet set via the SSL_set_min_proto_version() and
4258   SSL_set_max_proto_version(), or via the SSL_CONF's MinProtocol and
4259   MaxProtocol.  It's recommended to use the new APIs to disable
4260   protocols instead of disabling individual protocols using
4261   SSL_set_options() or SSL_CONF's Protocol.  This change also
4262   removes support for disabling TLS 1.2 in the OpenSSL TLS
4263   client at compile time by defining OPENSSL_NO_TLS1_2_CLIENT.
4264
4265   *Kurt Roeckx*
4266
4267 * Support for ChaCha20 and Poly1305 added to libcrypto and libssl.
4268
4269   *Andy Polyakov*
4270
4271 * New EC_KEY_METHOD, this replaces the older ECDSA_METHOD and ECDH_METHOD
4272   and integrates ECDSA and ECDH functionality into EC. Implementations can
4273   now redirect key generation and no longer need to convert to or from
4274   ECDSA_SIG format.
4275
4276   Note: the ecdsa.h and ecdh.h headers are now no longer needed and just
4277   include the ec.h header file instead.
4278
4279   *Steve Henson*
4280
4281 * Remove support for all 40 and 56 bit ciphers.  This includes all the export
4282   ciphers who are no longer supported and drops support the ephemeral RSA key
4283   exchange. The LOW ciphers currently doesn't have any ciphers in it.
4284
4285   *Kurt Roeckx*
4286
4287 * Made EVP_MD_CTX, EVP_MD, EVP_CIPHER_CTX, EVP_CIPHER and HMAC_CTX
4288   opaque.  For HMAC_CTX, the following constructors and destructors
4289   were added:
4290
4291       HMAC_CTX *HMAC_CTX_new(void);
4292       void HMAC_CTX_free(HMAC_CTX *ctx);
4293
4294   For EVP_MD and EVP_CIPHER, complete APIs to create, fill and
4295   destroy such methods has been added.  See EVP_MD_meth_new(3) and
4296   EVP_CIPHER_meth_new(3) for documentation.
4297
4298   Additional changes:
4299   1) `EVP_MD_CTX_cleanup()`, `EVP_CIPHER_CTX_cleanup()` and
4300      `HMAC_CTX_cleanup()` were removed. `HMAC_CTX_reset()` and
4301      `EVP_MD_CTX_reset()` should be called instead to reinitialise
4302      an already created structure.
4303   2) For consistency with the majority of our object creators and
4304      destructors, `EVP_MD_CTX_(create|destroy)` were renamed to
4305      `EVP_MD_CTX_(new|free)`.  The old names are retained as macros
4306      for deprecated builds.
4307
4308   *Richard Levitte*
4309
4310 * Added ASYNC support. Libcrypto now includes the async sub-library to enable
4311   cryptographic operations to be performed asynchronously as long as an
4312   asynchronous capable engine is used. See the ASYNC_start_job() man page for
4313   further details. Libssl has also had this capability integrated with the
4314   introduction of the new mode SSL_MODE_ASYNC and associated error
4315   SSL_ERROR_WANT_ASYNC. See the SSL_CTX_set_mode() and SSL_get_error() man
4316   pages. This work was developed in partnership with Intel Corp.
4317
4318   *Matt Caswell*
4319
4320 * SSL_{CTX_}set_ecdh_auto() has been removed and ECDH is support is
4321   always enabled now.  If you want to disable the support you should
4322   exclude it using the list of supported ciphers. This also means that the
4323   "-no_ecdhe" option has been removed from s_server.
4324
4325   *Kurt Roeckx*
4326
4327 * SSL_{CTX}_set_tmp_ecdh() which can set 1 EC curve now internally calls
4328   SSL_{CTX_}set1_curves() which can set a list.
4329
4330   *Kurt Roeckx*
4331
4332 * Remove support for SSL_{CTX_}set_tmp_ecdh_callback().  You should set the
4333   curve you want to support using SSL_{CTX_}set1_curves().
4334
4335   *Kurt Roeckx*
4336
4337 * State machine rewrite. The state machine code has been significantly
4338   refactored in order to remove much duplication of code and solve issues
4339   with the old code (see [ssl/statem/README.md](ssl/statem/README.md) for
4340   further details). This change does have some associated API changes.
4341   Notably the SSL_state() function has been removed and replaced by
4342   SSL_get_state which now returns an "OSSL_HANDSHAKE_STATE" instead of an int.
4343   SSL_set_state() has been removed altogether. The previous handshake states
4344   defined in ssl.h and ssl3.h have also been removed.
4345
4346   *Matt Caswell*
4347
4348 * All instances of the string "ssleay" in the public API were replaced
4349   with OpenSSL (case-matching; e.g., OPENSSL_VERSION for #define's)
4350   Some error codes related to internal RSA_eay API's were renamed.
4351
4352   *Rich Salz*
4353
4354 * The demo files in crypto/threads were moved to demo/threads.
4355
4356   *Rich Salz*
4357
4358 * Removed obsolete engines: 4758cca, aep, atalla, cswift, nuron, gmp,
4359   sureware and ubsec.
4360
4361   *Matt Caswell, Rich Salz*
4362
4363 * New ASN.1 embed macro.
4364
4365   New ASN.1 macro ASN1_EMBED. This is the same as ASN1_SIMPLE except the
4366   structure is not allocated: it is part of the parent. That is instead of
4367
4368           FOO *x;
4369
4370   it must be:
4371
4372           FOO x;
4373
4374   This reduces memory fragmentation and make it impossible to accidentally
4375   set a mandatory field to NULL.
4376
4377   This currently only works for some fields specifically a SEQUENCE, CHOICE,
4378   or ASN1_STRING type which is part of a parent SEQUENCE. Since it is
4379   equivalent to ASN1_SIMPLE it cannot be tagged, OPTIONAL, SET OF or
4380   SEQUENCE OF.
4381
4382   *Steve Henson*
4383
4384 * Remove EVP_CHECK_DES_KEY, a compile-time option that never compiled.
4385
4386   *Emilia Käsper*
4387
4388 * Removed DES and RC4 ciphersuites from DEFAULT. Also removed RC2 although
4389   in 1.0.2 EXPORT was already removed and the only RC2 ciphersuite is also
4390   an EXPORT one. COMPLEMENTOFDEFAULT has been updated accordingly to add
4391   DES and RC4 ciphersuites.
4392
4393   *Matt Caswell*
4394
4395 * Rewrite EVP_DecodeUpdate (base64 decoding) to fix several bugs.
4396   This changes the decoding behaviour for some invalid messages,
4397   though the change is mostly in the more lenient direction, and
4398   legacy behaviour is preserved as much as possible.
4399
4400   *Emilia Käsper*
4401
4402 * Fix no-stdio build.
4403   *David Woodhouse <David.Woodhouse@intel.com> and also*
4404   *Ivan Nestlerode <ivan.nestlerode@sonos.com>*
4405
4406 * New testing framework
4407   The testing framework has been largely rewritten and is now using
4408   perl and the perl modules Test::Harness and an extended variant of
4409   Test::More called OpenSSL::Test to do its work.  All test scripts in
4410   test/ have been rewritten into test recipes, and all direct calls to
4411   executables in test/Makefile have become individual recipes using the
4412   simplified testing OpenSSL::Test::Simple.
4413
4414   For documentation on our testing modules, do:
4415
4416           perldoc test/testlib/OpenSSL/Test/Simple.pm
4417           perldoc test/testlib/OpenSSL/Test.pm
4418
4419   *Richard Levitte*
4420
4421 * Revamped memory debug; only -DCRYPTO_MDEBUG and -DCRYPTO_MDEBUG_ABORT
4422   are used; the latter aborts on memory leaks (usually checked on exit).
4423   Some undocumented "set malloc, etc., hooks" functions were removed
4424   and others were changed.  All are now documented.
4425
4426   *Rich Salz*
4427
4428 * In DSA_generate_parameters_ex, if the provided seed is too short,
4429   return an error
4430
4431   *Rich Salz and Ismo Puustinen <ismo.puustinen@intel.com>*
4432
4433 * Rewrite PSK to support ECDHE_PSK, DHE_PSK and RSA_PSK. Add ciphersuites
4434   from RFC4279, RFC4785, RFC5487, RFC5489.
4435
4436   Thanks to Christian J. Dietrich and Giuseppe D'Angelo for the
4437   original RSA_PSK patch.
4438
4439   *Steve Henson*
4440
4441 * Dropped support for the SSL3_FLAGS_DELAY_CLIENT_FINISHED flag. This SSLeay
4442   era flag was never set throughout the codebase (only read). Also removed
4443   SSL3_FLAGS_POP_BUFFER which was only used if
4444   SSL3_FLAGS_DELAY_CLIENT_FINISHED was also set.
4445
4446   *Matt Caswell*
4447
4448 * Changed the default name options in the "ca", "crl", "req" and "x509"
4449   to be "oneline" instead of "compat".
4450
4451   *Richard Levitte*
4452
4453 * Remove SSL_OP_TLS_BLOCK_PADDING_BUG. This is SSLeay legacy, we're
4454   not aware of clients that still exhibit this bug, and the workaround
4455   hasn't been working properly for a while.
4456
4457   *Emilia Käsper*
4458
4459 * The return type of BIO_number_read() and BIO_number_written() as well as
4460   the corresponding num_read and num_write members in the BIO structure has
4461   changed from unsigned long to uint64_t. On platforms where an unsigned
4462   long is 32 bits (e.g. Windows) these counters could overflow if >4Gb is
4463   transferred.
4464
4465   *Matt Caswell*
4466
4467 * Given the pervasive nature of TLS extensions it is inadvisable to run
4468   OpenSSL without support for them. It also means that maintaining
4469   the OPENSSL_NO_TLSEXT option within the code is very invasive (and probably
4470   not well tested). Therefore the OPENSSL_NO_TLSEXT option has been removed.
4471
4472   *Matt Caswell*
4473
4474 * Removed support for the two export grade static DH ciphersuites
4475   EXP-DH-RSA-DES-CBC-SHA and EXP-DH-DSS-DES-CBC-SHA. These two ciphersuites
4476   were newly added (along with a number of other static DH ciphersuites) to
4477   1.0.2. However the two export ones have *never* worked since they were
4478   introduced. It seems strange in any case to be adding new export
4479   ciphersuites, and given "logjam" it also does not seem correct to fix them.
4480
4481   *Matt Caswell*
4482
4483 * Version negotiation has been rewritten. In particular SSLv23_method(),
4484   SSLv23_client_method() and SSLv23_server_method() have been deprecated,
4485   and turned into macros which simply call the new preferred function names
4486   TLS_method(), TLS_client_method() and TLS_server_method(). All new code
4487   should use the new names instead. Also as part of this change the ssl23.h
4488   header file has been removed.
4489
4490   *Matt Caswell*
4491
4492 * Support for Kerberos ciphersuites in TLS (RFC2712) has been removed. This
4493   code and the associated standard is no longer considered fit-for-purpose.
4494
4495   *Matt Caswell*
4496
4497 * RT2547 was closed.  When generating a private key, try to make the
4498   output file readable only by the owner.  This behavior change might
4499   be noticeable when interacting with other software.
4500
4501 * Documented all exdata functions.  Added CRYPTO_free_ex_index.
4502   Added a test.
4503
4504   *Rich Salz*
4505
4506 * Added HTTP GET support to the ocsp command.
4507
4508   *Rich Salz*
4509
4510 * Changed default digest for the dgst and enc commands from MD5 to
4511   sha256
4512
4513   *Rich Salz*
4514
4515 * RAND_pseudo_bytes has been deprecated. Users should use RAND_bytes instead.
4516
4517   *Matt Caswell*
4518
4519 * Added support for TLS extended master secret from
4520   draft-ietf-tls-session-hash-03.txt. Thanks for Alfredo Pironti for an
4521   initial patch which was a great help during development.
4522
4523   *Steve Henson*
4524
4525 * All libssl internal structures have been removed from the public header
4526   files, and the OPENSSL_NO_SSL_INTERN option has been removed (since it is
4527   now redundant). Users should not attempt to access internal structures
4528   directly. Instead they should use the provided API functions.
4529
4530   *Matt Caswell*
4531
4532 * config has been changed so that by default OPENSSL_NO_DEPRECATED is used.
4533   Access to deprecated functions can be re-enabled by running config with
4534   "enable-deprecated". In addition applications wishing to use deprecated
4535   functions must define OPENSSL_USE_DEPRECATED. Note that this new behaviour
4536   will, by default, disable some transitive includes that previously existed
4537   in the header files (e.g. ec.h will no longer, by default, include bn.h)
4538
4539   *Matt Caswell*
4540
4541 * Added support for OCB mode. OpenSSL has been granted a patent license
4542   compatible with the OpenSSL license for use of OCB. Details are available
4543   at <https://www.openssl.org/source/OCB-patent-grant-OpenSSL.pdf>. Support
4544   for OCB can be removed by calling config with no-ocb.
4545
4546   *Matt Caswell*
4547
4548 * SSLv2 support has been removed.  It still supports receiving a SSLv2
4549   compatible client hello.
4550
4551   *Kurt Roeckx*
4552
4553 * Increased the minimal RSA keysize from 256 to 512 bits [Rich Salz],
4554   done while fixing the error code for the key-too-small case.
4555
4556   *Annie Yousar <a.yousar@informatik.hu-berlin.de>*
4557
4558 * CA.sh has been removed; use CA.pl instead.
4559
4560   *Rich Salz*
4561
4562 * Removed old DES API.
4563
4564   *Rich Salz*
4565
4566 * Remove various unsupported platforms:
4567      Sony NEWS4
4568      BEOS and BEOS_R5
4569      NeXT
4570      SUNOS
4571      MPE/iX
4572      Sinix/ReliantUNIX RM400
4573      DGUX
4574      NCR
4575      Tandem
4576      Cray
4577      16-bit platforms such as WIN16
4578
4579   *Rich Salz*
4580
4581 * Clean up OPENSSL_NO_xxx #define's
4582   - Use setbuf() and remove OPENSSL_NO_SETVBUF_IONBF
4583   - Rename OPENSSL_SYSNAME_xxx to OPENSSL_SYS_xxx
4584   - OPENSSL_NO_EC{DH,DSA} merged into OPENSSL_NO_EC
4585   - OPENSSL_NO_RIPEMD160, OPENSSL_NO_RIPEMD merged into OPENSSL_NO_RMD160
4586   - OPENSSL_NO_FP_API merged into OPENSSL_NO_STDIO
4587   - Remove OPENSSL_NO_BIO OPENSSL_NO_BUFFER OPENSSL_NO_CHAIN_VERIFY
4588     OPENSSL_NO_EVP OPENSSL_NO_FIPS_ERR OPENSSL_NO_HASH_COMP
4589     OPENSSL_NO_LHASH OPENSSL_NO_OBJECT OPENSSL_NO_SPEED OPENSSL_NO_STACK
4590     OPENSSL_NO_X509 OPENSSL_NO_X509_VERIFY
4591   - Remove MS_STATIC; it's a relic from platforms <32 bits.
4592
4593   *Rich Salz*
4594
4595 * Cleaned up dead code
4596     Remove all but one '#ifdef undef' which is to be looked at.
4597
4598   *Rich Salz*
4599
4600 * Clean up calling of xxx_free routines.
4601      Just like free(), fix most of the xxx_free routines to accept
4602      NULL.  Remove the non-null checks from callers.  Save much code.
4603
4604   *Rich Salz*
4605
4606 * Add secure heap for storage of private keys (when possible).
4607   Add BIO_s_secmem(), CBIGNUM, etc.
4608   Contributed by Akamai Technologies under our Corporate CLA.
4609
4610   *Rich Salz*
4611
4612 * Experimental support for a new, fast, unbiased prime candidate generator,
4613   bn_probable_prime_dh_coprime(). Not currently used by any prime generator.
4614
4615   *Felix Laurie von Massenbach <felix@erbridge.co.uk>*
4616
4617 * New output format NSS in the sess_id command line tool. This allows
4618   exporting the session id and the master key in NSS keylog format.
4619
4620   *Martin Kaiser <martin@kaiser.cx>*
4621
4622 * Harmonize version and its documentation. -f flag is used to display
4623   compilation flags.
4624
4625   *mancha <mancha1@zoho.com>*
4626
4627 * Fix eckey_priv_encode so it immediately returns an error upon a failure
4628   in i2d_ECPrivateKey.  Thanks to Ted Unangst for feedback on this issue.
4629
4630   *mancha <mancha1@zoho.com>*
4631
4632 * Fix some double frees. These are not thought to be exploitable.
4633
4634   *mancha <mancha1@zoho.com>*
4635
4636 * A missing bounds check in the handling of the TLS heartbeat extension
4637   can be used to reveal up to 64k of memory to a connected client or
4638   server.
4639
4640   Thanks for Neel Mehta of Google Security for discovering this bug and to
4641   Adam Langley <agl@chromium.org> and Bodo Moeller <bmoeller@acm.org> for
4642   preparing the fix ([CVE-2014-0160])
4643
4644   *Adam Langley, Bodo Moeller*
4645
4646 * Fix for the attack described in the paper "Recovering OpenSSL
4647   ECDSA Nonces Using the FLUSH+RELOAD Cache Side-channel Attack"
4648   by Yuval Yarom and Naomi Benger. Details can be obtained from:
4649   <http://eprint.iacr.org/2014/140>
4650
4651   Thanks to Yuval Yarom and Naomi Benger for discovering this
4652   flaw and to Yuval Yarom for supplying a fix ([CVE-2014-0076])
4653
4654   *Yuval Yarom and Naomi Benger*
4655
4656 * Use algorithm specific chains in SSL_CTX_use_certificate_chain_file():
4657   this fixes a limitation in previous versions of OpenSSL.
4658
4659   *Steve Henson*
4660
4661 * Experimental encrypt-then-mac support.
4662
4663   Experimental support for encrypt then mac from
4664   draft-gutmann-tls-encrypt-then-mac-02.txt
4665
4666   To enable it set the appropriate extension number (0x42 for the test
4667   server) using e.g. -DTLSEXT_TYPE_encrypt_then_mac=0x42
4668
4669   For non-compliant peers (i.e. just about everything) this should have no
4670   effect.
4671
4672   WARNING: EXPERIMENTAL, SUBJECT TO CHANGE.
4673
4674   *Steve Henson*
4675
4676 * Add EVP support for key wrapping algorithms, to avoid problems with
4677   existing code the flag EVP_CIPHER_CTX_WRAP_ALLOW has to be set in
4678   the EVP_CIPHER_CTX or an error is returned. Add AES and DES3 wrap
4679   algorithms and include tests cases.
4680
4681   *Steve Henson*
4682
4683 * Extend CMS code to support RSA-PSS signatures and RSA-OAEP for
4684   enveloped data.
4685
4686   *Steve Henson*
4687
4688 * Extended RSA OAEP support via EVP_PKEY API. Options to specify digest,
4689   MGF1 digest and OAEP label.
4690
4691   *Steve Henson*
4692
4693 * Make openssl verify return errors.
4694
4695   *Chris Palmer <palmer@google.com> and Ben Laurie*
4696
4697 * New function ASN1_TIME_diff to calculate the difference between two
4698   ASN1_TIME structures or one structure and the current time.
4699
4700   *Steve Henson*
4701
4702 * Update fips_test_suite to support multiple command line options. New
4703   test to induce all self test errors in sequence and check expected
4704   failures.
4705
4706   *Steve Henson*
4707
4708 * Add FIPS_{rsa,dsa,ecdsa}_{sign,verify} functions which digest and
4709   sign or verify all in one operation.
4710
4711   *Steve Henson*
4712
4713 * Add fips_algvs: a multicall fips utility incorporating all the algorithm
4714   test programs and fips_test_suite. Includes functionality to parse
4715   the minimal script output of fipsalgest.pl directly.
4716
4717   *Steve Henson*
4718
4719 * Add authorisation parameter to FIPS_module_mode_set().
4720
4721   *Steve Henson*
4722
4723 * Add FIPS selftest for ECDH algorithm using P-224 and B-233 curves.
4724
4725   *Steve Henson*
4726
4727 * Use separate DRBG fields for internal and external flags. New function
4728   FIPS_drbg_health_check() to perform on demand health checking. Add
4729   generation tests to fips_test_suite with reduced health check interval to
4730   demonstrate periodic health checking. Add "nodh" option to
4731   fips_test_suite to skip very slow DH test.
4732
4733   *Steve Henson*
4734
4735 * New function FIPS_get_cipherbynid() to lookup FIPS supported ciphers
4736   based on NID.
4737
4738   *Steve Henson*
4739
4740 * More extensive health check for DRBG checking many more failure modes.
4741   New function FIPS_selftest_drbg_all() to handle every possible DRBG
4742   combination: call this in fips_test_suite.
4743
4744   *Steve Henson*
4745
4746 * Add support for canonical generation of DSA parameter 'g'. See
4747   FIPS 186-3 A.2.3.
4748
4749 * Add support for HMAC DRBG from SP800-90. Update DRBG algorithm test and
4750   POST to handle HMAC cases.
4751
4752   *Steve Henson*
4753
4754 * Add functions FIPS_module_version() and FIPS_module_version_text()
4755   to return numerical and string versions of the FIPS module number.
4756
4757   *Steve Henson*
4758
4759 * Rename FIPS_mode_set and FIPS_mode to FIPS_module_mode_set and
4760   FIPS_module_mode. FIPS_mode and FIPS_mode_set will be implemented
4761   outside the validated module in the FIPS capable OpenSSL.
4762
4763   *Steve Henson*
4764
4765 * Minor change to DRBG entropy callback semantics. In some cases
4766   there is no multiple of the block length between min_len and
4767   max_len. Allow the callback to return more than max_len bytes
4768   of entropy but discard any extra: it is the callback's responsibility
4769   to ensure that the extra data discarded does not impact the
4770   requested amount of entropy.
4771
4772   *Steve Henson*
4773
4774 * Add PRNG security strength checks to RSA, DSA and ECDSA using
4775   information in FIPS186-3, SP800-57 and SP800-131A.
4776
4777   *Steve Henson*
4778
4779 * CCM support via EVP. Interface is very similar to GCM case except we
4780   must supply all data in one chunk (i.e. no update, final) and the
4781   message length must be supplied if AAD is used. Add algorithm test
4782   support.
4783
4784   *Steve Henson*
4785
4786 * Initial version of POST overhaul. Add POST callback to allow the status
4787   of POST to be monitored and/or failures induced. Modify fips_test_suite
4788   to use callback. Always run all selftests even if one fails.
4789
4790   *Steve Henson*
4791
4792 * XTS support including algorithm test driver in the fips_gcmtest program.
4793   Note: this does increase the maximum key length from 32 to 64 bytes but
4794   there should be no binary compatibility issues as existing applications
4795   will never use XTS mode.
4796
4797   *Steve Henson*
4798
4799 * Extensive reorganisation of FIPS PRNG behaviour. Remove all dependencies
4800   to OpenSSL RAND code and replace with a tiny FIPS RAND API which also
4801   performs algorithm blocking for unapproved PRNG types. Also do not
4802   set PRNG type in FIPS_mode_set(): leave this to the application.
4803   Add default OpenSSL DRBG handling: sets up FIPS PRNG and seeds with
4804   the standard OpenSSL PRNG: set additional data to a date time vector.
4805
4806   *Steve Henson*
4807
4808 * Rename old X9.31 PRNG functions of the form `FIPS_rand*` to `FIPS_x931*`.
4809   This shouldn't present any incompatibility problems because applications
4810   shouldn't be using these directly and any that are will need to rethink
4811   anyway as the X9.31 PRNG is now deprecated by FIPS 140-2
4812
4813   *Steve Henson*
4814
4815 * Extensive self tests and health checking required by SP800-90 DRBG.
4816   Remove strength parameter from FIPS_drbg_instantiate and always
4817   instantiate at maximum supported strength.
4818
4819   *Steve Henson*
4820
4821 * Add ECDH code to fips module and fips_ecdhvs for primitives only testing.
4822
4823   *Steve Henson*
4824
4825 * New algorithm test program fips_dhvs to handle DH primitives only testing.
4826
4827   *Steve Henson*
4828
4829 * New function DH_compute_key_padded() to compute a DH key and pad with
4830   leading zeroes if needed: this complies with SP800-56A et al.
4831
4832   *Steve Henson*
4833
4834 * Initial implementation of SP800-90 DRBGs for Hash and CTR. Not used by
4835   anything, incomplete, subject to change and largely untested at present.
4836
4837   *Steve Henson*
4838
4839 * Modify fipscanisteronly build option to only build the necessary object
4840   files by filtering FIPS_EX_OBJ through a perl script in crypto/Makefile.
4841
4842   *Steve Henson*
4843
4844 * Add experimental option FIPSSYMS to give all symbols in
4845   fipscanister.o and FIPS or fips prefix. This will avoid
4846   conflicts with future versions of OpenSSL. Add perl script
4847   util/fipsas.pl to preprocess assembly language source files
4848   and rename any affected symbols.
4849
4850   *Steve Henson*
4851
4852 * Add selftest checks and algorithm block of non-fips algorithms in
4853   FIPS mode. Remove DES2 from selftests.
4854
4855   *Steve Henson*
4856
4857 * Add ECDSA code to fips module. Add tiny fips_ecdsa_check to just
4858   return internal method without any ENGINE dependencies. Add new
4859   tiny fips sign and verify functions.
4860
4861   *Steve Henson*
4862
4863 * New build option no-ec2m to disable characteristic 2 code.
4864
4865   *Steve Henson*
4866
4867 * New build option "fipscanisteronly". This only builds fipscanister.o
4868   and (currently) associated fips utilities. Uses the file Makefile.fips
4869   instead of Makefile.org as the prototype.
4870
4871   *Steve Henson*
4872
4873 * Add some FIPS mode restrictions to GCM. Add internal IV generator.
4874   Update fips_gcmtest to use IV generator.
4875
4876   *Steve Henson*
4877
4878 * Initial, experimental EVP support for AES-GCM. AAD can be input by
4879   setting output buffer to NULL. The `*Final` function must be
4880   called although it will not retrieve any additional data. The tag
4881   can be set or retrieved with a ctrl. The IV length is by default 12
4882   bytes (96 bits) but can be set to an alternative value. If the IV
4883   length exceeds the maximum IV length (currently 16 bytes) it cannot be
4884   set before the key.
4885
4886   *Steve Henson*
4887
4888 * New flag in ciphers: EVP_CIPH_FLAG_CUSTOM_CIPHER. This means the
4889   underlying do_cipher function handles all cipher semantics itself
4890   including padding and finalisation. This is useful if (for example)
4891   an ENGINE cipher handles block padding itself. The behaviour of
4892   do_cipher is subtly changed if this flag is set: the return value
4893   is the number of characters written to the output buffer (zero is
4894   no longer an error code) or a negative error code. Also if the
4895   input buffer is NULL and length 0 finalisation should be performed.
4896
4897   *Steve Henson*
4898
4899 * If a candidate issuer certificate is already part of the constructed
4900   path ignore it: new debug notification X509_V_ERR_PATH_LOOP for this case.
4901
4902   *Steve Henson*
4903
4904 * Improve forward-security support: add functions
4905
4906           void SSL_CTX_set_not_resumable_session_callback(
4907                    SSL_CTX *ctx, int (*cb)(SSL *ssl, int is_forward_secure))
4908           void SSL_set_not_resumable_session_callback(
4909                    SSL *ssl, int (*cb)(SSL *ssl, int is_forward_secure))
4910
4911   for use by SSL/TLS servers; the callback function will be called whenever a
4912   new session is created, and gets to decide whether the session may be
4913   cached to make it resumable (return 0) or not (return 1).  (As by the
4914   SSL/TLS protocol specifications, the session_id sent by the server will be
4915   empty to indicate that the session is not resumable; also, the server will
4916   not generate RFC 4507 (RFC 5077) session tickets.)
4917
4918   A simple reasonable callback implementation is to return is_forward_secure.
4919   This parameter will be set to 1 or 0 depending on the ciphersuite selected
4920   by the SSL/TLS server library, indicating whether it can provide forward
4921   security.
4922
4923   *Emilia Käsper <emilia.kasper@esat.kuleuven.be> (Google)*
4924
4925 * New -verify_name option in command line utilities to set verification
4926   parameters by name.
4927
4928   *Steve Henson*
4929
4930 * Initial CMAC implementation. WARNING: EXPERIMENTAL, API MAY CHANGE.
4931   Add CMAC pkey methods.
4932
4933   *Steve Henson*
4934
4935 * Experimental renegotiation in s_server -www mode. If the client
4936   browses /reneg connection is renegotiated. If /renegcert it is
4937   renegotiated requesting a certificate.
4938
4939   *Steve Henson*
4940
4941 * Add an "external" session cache for debugging purposes to s_server. This
4942   should help trace issues which normally are only apparent in deployed
4943   multi-process servers.
4944
4945   *Steve Henson*
4946
4947 * Extensive audit of libcrypto with DEBUG_UNUSED. Fix many cases where
4948   return value is ignored. NB. The functions RAND_add(), RAND_seed(),
4949   BIO_set_cipher() and some obscure PEM functions were changed so they
4950   can now return an error. The RAND changes required a change to the
4951   RAND_METHOD structure.
4952
4953   *Steve Henson*
4954
4955 * New macro `__owur` for "OpenSSL Warn Unused Result". This makes use of
4956   a gcc attribute to warn if the result of a function is ignored. This
4957   is enable if DEBUG_UNUSED is set. Add to several functions in evp.h
4958   whose return value is often ignored.
4959
4960   *Steve Henson*
4961
4962 * New -noct, -requestct, -requirect and -ctlogfile options for s_client.
4963   These allow SCTs (signed certificate timestamps) to be requested and
4964   validated when establishing a connection.
4965
4966   *Rob Percival <robpercival@google.com>*
4967
4968OpenSSL 1.0.2
4969-------------
4970
4971### Changes between 1.0.2s and 1.0.2t [10 Sep 2019]
4972
4973 * For built-in EC curves, ensure an EC_GROUP built from the curve name is
4974   used even when parsing explicit parameters, when loading a encoded key
4975   or calling `EC_GROUP_new_from_ecpkparameters()`/
4976   `EC_GROUP_new_from_ecparameters()`.
4977   This prevents bypass of security hardening and performance gains,
4978   especially for curves with specialized EC_METHODs.
4979   By default, if a key encoded with explicit parameters is loaded and later
4980   encoded, the output is still encoded with explicit parameters, even if
4981   internally a "named" EC_GROUP is used for computation.
4982
4983   *Nicola Tuveri*
4984
4985 * Compute ECC cofactors if not provided during EC_GROUP construction. Before
4986   this change, EC_GROUP_set_generator would accept order and/or cofactor as
4987   NULL. After this change, only the cofactor parameter can be NULL. It also
4988   does some minimal sanity checks on the passed order.
4989   ([CVE-2019-1547])
4990
4991   *Billy Bob Brumley*
4992
4993 * Fixed a padding oracle in PKCS7_dataDecode and CMS_decrypt_set1_pkey.
4994   An attack is simple, if the first CMS_recipientInfo is valid but the
4995   second CMS_recipientInfo is chosen ciphertext. If the second
4996   recipientInfo decodes to PKCS #1 v1.5 form plaintext, the correct
4997   encryption key will be replaced by garbage, and the message cannot be
4998   decoded, but if the RSA decryption fails, the correct encryption key is
4999   used and the recipient will not notice the attack.
5000   As a work around for this potential attack the length of the decrypted
5001   key must be equal to the cipher default key length, in case the
5002   certifiate is not given and all recipientInfo are tried out.
5003   The old behaviour can be re-enabled in the CMS code by setting the
5004   CMS_DEBUG_DECRYPT flag.
5005   ([CVE-2019-1563])
5006
5007   *Bernd Edlinger*
5008
5009 * Document issue with installation paths in diverse Windows builds
5010
5011   '/usr/local/ssl' is an unsafe prefix for location to install OpenSSL
5012   binaries and run-time config file.
5013   ([CVE-2019-1552])
5014
5015   *Richard Levitte*
5016
5017### Changes between 1.0.2r and 1.0.2s [28 May 2019]
5018
5019 * Change the default RSA, DSA and DH size to 2048 bit instead of 1024.
5020   This changes the size when using the `genpkey` command when no size is given.
5021   It fixes an omission in earlier changes that changed all RSA, DSA and DH
5022   generation commands to use 2048 bits by default.
5023
5024   *Kurt Roeckx*
5025
5026 * Add FIPS support for Android Arm 64-bit
5027
5028   Support for Android Arm 64-bit was added to the OpenSSL FIPS Object
5029   Module in Version 2.0.10. For some reason, the corresponding target
5030   'android64-aarch64' was missing OpenSSL 1.0.2, whence it could not be
5031   built with FIPS support on Android Arm 64-bit. This omission has been
5032   fixed.
5033
5034   *Matthias St. Pierre*
5035
5036### Changes between 1.0.2q and 1.0.2r [26 Feb 2019]
5037
5038 * 0-byte record padding oracle
5039
5040   If an application encounters a fatal protocol error and then calls
5041   SSL_shutdown() twice (once to send a close_notify, and once to receive one)
5042   then OpenSSL can respond differently to the calling application if a 0 byte
5043   record is received with invalid padding compared to if a 0 byte record is
5044   received with an invalid MAC. If the application then behaves differently
5045   based on that in a way that is detectable to the remote peer, then this
5046   amounts to a padding oracle that could be used to decrypt data.
5047
5048   In order for this to be exploitable "non-stitched" ciphersuites must be in
5049   use. Stitched ciphersuites are optimised implementations of certain
5050   commonly used ciphersuites. Also the application must call SSL_shutdown()
5051   twice even if a protocol error has occurred (applications should not do
5052   this but some do anyway).
5053
5054   This issue was discovered by Juraj Somorovsky, Robert Merget and Nimrod
5055   Aviram, with additional investigation by Steven Collison and Andrew
5056   Hourselt. It was reported to OpenSSL on 10th December 2018.
5057   ([CVE-2019-1559])
5058
5059   *Matt Caswell*
5060
5061 * Move strictness check from EVP_PKEY_asn1_new() to EVP_PKEY_asn1_add0().
5062
5063   *Richard Levitte*
5064
5065### Changes between 1.0.2p and 1.0.2q [20 Nov 2018]
5066
5067 * Microarchitecture timing vulnerability in ECC scalar multiplication
5068
5069   OpenSSL ECC scalar multiplication, used in e.g. ECDSA and ECDH, has been
5070   shown to be vulnerable to a microarchitecture timing side channel attack.
5071   An attacker with sufficient access to mount local timing attacks during
5072   ECDSA signature generation could recover the private key.
5073
5074   This issue was reported to OpenSSL on 26th October 2018 by Alejandro
5075   Cabrera Aldaya, Billy Brumley, Sohaib ul Hassan, Cesar Pereida Garcia and
5076   Nicola Tuveri.
5077   ([CVE-2018-5407])
5078
5079   *Billy Brumley*
5080
5081 * Timing vulnerability in DSA signature generation
5082
5083   The OpenSSL DSA signature algorithm has been shown to be vulnerable to a
5084   timing side channel attack. An attacker could use variations in the signing
5085   algorithm to recover the private key.
5086
5087   This issue was reported to OpenSSL on 16th October 2018 by Samuel Weiser.
5088   ([CVE-2018-0734])
5089
5090   *Paul Dale*
5091
5092 * Resolve a compatibility issue in EC_GROUP handling with the FIPS Object
5093   Module, accidentally introduced while backporting security fixes from the
5094   development branch and hindering the use of ECC in FIPS mode.
5095
5096   *Nicola Tuveri*
5097
5098### Changes between 1.0.2o and 1.0.2p [14 Aug 2018]
5099
5100 * Client DoS due to large DH parameter
5101
5102   During key agreement in a TLS handshake using a DH(E) based ciphersuite a
5103   malicious server can send a very large prime value to the client. This will
5104   cause the client to spend an unreasonably long period of time generating a
5105   key for this prime resulting in a hang until the client has finished. This
5106   could be exploited in a Denial Of Service attack.
5107
5108   This issue was reported to OpenSSL on 5th June 2018 by Guido Vranken
5109   ([CVE-2018-0732])
5110
5111   *Guido Vranken*
5112
5113 * Cache timing vulnerability in RSA Key Generation
5114
5115   The OpenSSL RSA Key generation algorithm has been shown to be vulnerable to
5116   a cache timing side channel attack. An attacker with sufficient access to
5117   mount cache timing attacks during the RSA key generation process could
5118   recover the private key.
5119
5120   This issue was reported to OpenSSL on 4th April 2018 by Alejandro Cabrera
5121   Aldaya, Billy Brumley, Cesar Pereida Garcia and Luis Manuel Alvarez Tapia.
5122   ([CVE-2018-0737])
5123
5124   *Billy Brumley*
5125
5126 * Make EVP_PKEY_asn1_new() a bit stricter about its input.  A NULL pem_str
5127   parameter is no longer accepted, as it leads to a corrupt table.  NULL
5128   pem_str is reserved for alias entries only.
5129
5130   *Richard Levitte*
5131
5132 * Revert blinding in ECDSA sign and instead make problematic addition
5133   length-invariant. Switch even to fixed-length Montgomery multiplication.
5134
5135   *Andy Polyakov*
5136
5137 * Change generating and checking of primes so that the error rate of not
5138   being prime depends on the intended use based on the size of the input.
5139   For larger primes this will result in more rounds of Miller-Rabin.
5140   The maximal error rate for primes with more than 1080 bits is lowered
5141   to 2^-128.
5142
5143   *Kurt Roeckx, Annie Yousar*
5144
5145 * Increase the number of Miller-Rabin rounds for DSA key generating to 64.
5146
5147   *Kurt Roeckx*
5148
5149 * Add blinding to ECDSA and DSA signatures to protect against side channel
5150   attacks discovered by Keegan Ryan (NCC Group).
5151
5152   *Matt Caswell*
5153
5154 * When unlocking a pass phrase protected PEM file or PKCS#8 container, we
5155   now allow empty (zero character) pass phrases.
5156
5157   *Richard Levitte*
5158
5159 * Certificate time validation (X509_cmp_time) enforces stricter
5160   compliance with RFC 5280. Fractional seconds and timezone offsets
5161   are no longer allowed.
5162
5163   *Emilia Käsper*
5164
5165### Changes between 1.0.2n and 1.0.2o [27 Mar 2018]
5166
5167 * Constructed ASN.1 types with a recursive definition could exceed the stack
5168
5169   Constructed ASN.1 types with a recursive definition (such as can be found
5170   in PKCS7) could eventually exceed the stack given malicious input with
5171   excessive recursion. This could result in a Denial Of Service attack. There
5172   are no such structures used within SSL/TLS that come from untrusted sources
5173   so this is considered safe.
5174
5175   This issue was reported to OpenSSL on 4th January 2018 by the OSS-fuzz
5176   project.
5177   ([CVE-2018-0739])
5178
5179   *Matt Caswell*
5180
5181### Changes between 1.0.2m and 1.0.2n [7 Dec 2017]
5182
5183 * Read/write after SSL object in error state
5184
5185   OpenSSL 1.0.2 (starting from version 1.0.2b) introduced an "error state"
5186   mechanism. The intent was that if a fatal error occurred during a handshake
5187   then OpenSSL would move into the error state and would immediately fail if
5188   you attempted to continue the handshake. This works as designed for the
5189   explicit handshake functions (SSL_do_handshake(), SSL_accept() and
5190   SSL_connect()), however due to a bug it does not work correctly if
5191   SSL_read() or SSL_write() is called directly. In that scenario, if the
5192   handshake fails then a fatal error will be returned in the initial function
5193   call. If SSL_read()/SSL_write() is subsequently called by the application
5194   for the same SSL object then it will succeed and the data is passed without
5195   being decrypted/encrypted directly from the SSL/TLS record layer.
5196
5197   In order to exploit this issue an application bug would have to be present
5198   that resulted in a call to SSL_read()/SSL_write() being issued after having
5199   already received a fatal error.
5200
5201   This issue was reported to OpenSSL by David Benjamin (Google).
5202   ([CVE-2017-3737])
5203
5204   *Matt Caswell*
5205
5206 * rsaz_1024_mul_avx2 overflow bug on x86_64
5207
5208   There is an overflow bug in the AVX2 Montgomery multiplication procedure
5209   used in exponentiation with 1024-bit moduli. No EC algorithms are affected.
5210   Analysis suggests that attacks against RSA and DSA as a result of this
5211   defect would be very difficult to perform and are not believed likely.
5212   Attacks against DH1024 are considered just feasible, because most of the
5213   work necessary to deduce information about a private key may be performed
5214   offline. The amount of resources required for such an attack would be
5215   significant. However, for an attack on TLS to be meaningful, the server
5216   would have to share the DH1024 private key among multiple clients, which is
5217   no longer an option since CVE-2016-0701.
5218
5219   This only affects processors that support the AVX2 but not ADX extensions
5220   like Intel Haswell (4th generation).
5221
5222   This issue was reported to OpenSSL by David Benjamin (Google). The issue
5223   was originally found via the OSS-Fuzz project.
5224   ([CVE-2017-3738])
5225
5226   *Andy Polyakov*
5227
5228### Changes between 1.0.2l and 1.0.2m [2 Nov 2017]
5229
5230 * bn_sqrx8x_internal carry bug on x86_64
5231
5232   There is a carry propagating bug in the x86_64 Montgomery squaring
5233   procedure. No EC algorithms are affected. Analysis suggests that attacks
5234   against RSA and DSA as a result of this defect would be very difficult to
5235   perform and are not believed likely. Attacks against DH are considered just
5236   feasible (although very difficult) because most of the work necessary to
5237   deduce information about a private key may be performed offline. The amount
5238   of resources required for such an attack would be very significant and
5239   likely only accessible to a limited number of attackers. An attacker would
5240   additionally need online access to an unpatched system using the target
5241   private key in a scenario with persistent DH parameters and a private
5242   key that is shared between multiple clients.
5243
5244   This only affects processors that support the BMI1, BMI2 and ADX extensions
5245   like Intel Broadwell (5th generation) and later or AMD Ryzen.
5246
5247   This issue was reported to OpenSSL by the OSS-Fuzz project.
5248   ([CVE-2017-3736])
5249
5250   *Andy Polyakov*
5251
5252 * Malformed X.509 IPAddressFamily could cause OOB read
5253
5254   If an X.509 certificate has a malformed IPAddressFamily extension,
5255   OpenSSL could do a one-byte buffer overread. The most likely result
5256   would be an erroneous display of the certificate in text format.
5257
5258   This issue was reported to OpenSSL by the OSS-Fuzz project.
5259
5260   *Rich Salz*
5261
5262### Changes between 1.0.2k and 1.0.2l [25 May 2017]
5263
5264 * Have 'config' recognise 64-bit mingw and choose 'mingw64' as the target
5265   platform rather than 'mingw'.
5266
5267   *Richard Levitte*
5268
5269### Changes between 1.0.2j and 1.0.2k [26 Jan 2017]
5270
5271 * Truncated packet could crash via OOB read
5272
5273   If one side of an SSL/TLS path is running on a 32-bit host and a specific
5274   cipher is being used, then a truncated packet can cause that host to
5275   perform an out-of-bounds read, usually resulting in a crash.
5276
5277   This issue was reported to OpenSSL by Robert Święcki of Google.
5278   ([CVE-2017-3731])
5279
5280   *Andy Polyakov*
5281
5282 * BN_mod_exp may produce incorrect results on x86_64
5283
5284   There is a carry propagating bug in the x86_64 Montgomery squaring
5285   procedure. No EC algorithms are affected. Analysis suggests that attacks
5286   against RSA and DSA as a result of this defect would be very difficult to
5287   perform and are not believed likely. Attacks against DH are considered just
5288   feasible (although very difficult) because most of the work necessary to
5289   deduce information about a private key may be performed offline. The amount
5290   of resources required for such an attack would be very significant and
5291   likely only accessible to a limited number of attackers. An attacker would
5292   additionally need online access to an unpatched system using the target
5293   private key in a scenario with persistent DH parameters and a private
5294   key that is shared between multiple clients. For example this can occur by
5295   default in OpenSSL DHE based SSL/TLS ciphersuites. Note: This issue is very
5296   similar to CVE-2015-3193 but must be treated as a separate problem.
5297
5298   This issue was reported to OpenSSL by the OSS-Fuzz project.
5299   ([CVE-2017-3732])
5300
5301   *Andy Polyakov*
5302
5303 * Montgomery multiplication may produce incorrect results
5304
5305   There is a carry propagating bug in the Broadwell-specific Montgomery
5306   multiplication procedure that handles input lengths divisible by, but
5307   longer than 256 bits. Analysis suggests that attacks against RSA, DSA
5308   and DH private keys are impossible. This is because the subroutine in
5309   question is not used in operations with the private key itself and an input
5310   of the attacker's direct choice. Otherwise the bug can manifest itself as
5311   transient authentication and key negotiation failures or reproducible
5312   erroneous outcome of public-key operations with specially crafted input.
5313   Among EC algorithms only Brainpool P-512 curves are affected and one
5314   presumably can attack ECDH key negotiation. Impact was not analyzed in
5315   detail, because pre-requisites for attack are considered unlikely. Namely
5316   multiple clients have to choose the curve in question and the server has to
5317   share the private key among them, neither of which is default behaviour.
5318   Even then only clients that chose the curve will be affected.
5319
5320   This issue was publicly reported as transient failures and was not
5321   initially recognized as a security issue. Thanks to Richard Morgan for
5322   providing reproducible case.
5323   ([CVE-2016-7055])
5324
5325   *Andy Polyakov*
5326
5327 * OpenSSL now fails if it receives an unrecognised record type in TLS1.0
5328   or TLS1.1. Previously this only happened in SSLv3 and TLS1.2. This is to
5329   prevent issues where no progress is being made and the peer continually
5330   sends unrecognised record types, using up resources processing them.
5331
5332   *Matt Caswell*
5333
5334### Changes between 1.0.2i and 1.0.2j [26 Sep 2016]
5335
5336 * Missing CRL sanity check
5337
5338   A bug fix which included a CRL sanity check was added to OpenSSL 1.1.0
5339   but was omitted from OpenSSL 1.0.2i. As a result any attempt to use
5340   CRLs in OpenSSL 1.0.2i will crash with a null pointer exception.
5341
5342   This issue only affects the OpenSSL 1.0.2i
5343   ([CVE-2016-7052])
5344
5345   *Matt Caswell*
5346
5347### Changes between 1.0.2h and 1.0.2i [22 Sep 2016]
5348
5349 * OCSP Status Request extension unbounded memory growth
5350
5351   A malicious client can send an excessively large OCSP Status Request
5352   extension. If that client continually requests renegotiation, sending a
5353   large OCSP Status Request extension each time, then there will be unbounded
5354   memory growth on the server. This will eventually lead to a Denial Of
5355   Service attack through memory exhaustion. Servers with a default
5356   configuration are vulnerable even if they do not support OCSP. Builds using
5357   the "no-ocsp" build time option are not affected.
5358
5359   This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.)
5360   ([CVE-2016-6304])
5361
5362   *Matt Caswell*
5363
5364 * In order to mitigate the SWEET32 attack, the DES ciphers were moved from
5365   HIGH to MEDIUM.
5366
5367   This issue was reported to OpenSSL Karthikeyan Bhargavan and Gaetan
5368   Leurent (INRIA)
5369   ([CVE-2016-2183])
5370
5371   *Rich Salz*
5372
5373 * OOB write in MDC2_Update()
5374
5375   An overflow can occur in MDC2_Update() either if called directly or
5376   through the EVP_DigestUpdate() function using MDC2. If an attacker
5377   is able to supply very large amounts of input data after a previous
5378   call to EVP_EncryptUpdate() with a partial block then a length check
5379   can overflow resulting in a heap corruption.
5380
5381   The amount of data needed is comparable to SIZE_MAX which is impractical
5382   on most platforms.
5383
5384   This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.)
5385   ([CVE-2016-6303])
5386
5387   *Stephen Henson*
5388
5389 * Malformed SHA512 ticket DoS
5390
5391   If a server uses SHA512 for TLS session ticket HMAC it is vulnerable to a
5392   DoS attack where a malformed ticket will result in an OOB read which will
5393   ultimately crash.
5394
5395   The use of SHA512 in TLS session tickets is comparatively rare as it requires
5396   a custom server callback and ticket lookup mechanism.
5397
5398   This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.)
5399   ([CVE-2016-6302])
5400
5401   *Stephen Henson*
5402
5403 * OOB write in BN_bn2dec()
5404
5405   The function BN_bn2dec() does not check the return value of BN_div_word().
5406   This can cause an OOB write if an application uses this function with an
5407   overly large BIGNUM. This could be a problem if an overly large certificate
5408   or CRL is printed out from an untrusted source. TLS is not affected because
5409   record limits will reject an oversized certificate before it is parsed.
5410
5411   This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.)
5412   ([CVE-2016-2182])
5413
5414   *Stephen Henson*
5415
5416 * OOB read in TS_OBJ_print_bio()
5417
5418   The function TS_OBJ_print_bio() misuses OBJ_obj2txt(): the return value is
5419   the total length the OID text representation would use and not the amount
5420   of data written. This will result in OOB reads when large OIDs are
5421   presented.
5422
5423   This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.)
5424   ([CVE-2016-2180])
5425
5426   *Stephen Henson*
5427
5428 * Pointer arithmetic undefined behaviour
5429
5430   Avoid some undefined pointer arithmetic
5431
5432   A common idiom in the codebase is to check limits in the following manner:
5433   "p + len > limit"
5434
5435   Where "p" points to some malloc'd data of SIZE bytes and
5436   limit == p + SIZE
5437
5438   "len" here could be from some externally supplied data (e.g. from a TLS
5439   message).
5440
5441   The rules of C pointer arithmetic are such that "p + len" is only well
5442   defined where len <= SIZE. Therefore the above idiom is actually
5443   undefined behaviour.
5444
5445   For example this could cause problems if some malloc implementation
5446   provides an address for "p" such that "p + len" actually overflows for
5447   values of len that are too big and therefore p + len < limit.
5448
5449   This issue was reported to OpenSSL by Guido Vranken
5450   ([CVE-2016-2177])
5451
5452   *Matt Caswell*
5453
5454 * Constant time flag not preserved in DSA signing
5455
5456   Operations in the DSA signing algorithm should run in constant time in
5457   order to avoid side channel attacks. A flaw in the OpenSSL DSA
5458   implementation means that a non-constant time codepath is followed for
5459   certain operations. This has been demonstrated through a cache-timing
5460   attack to be sufficient for an attacker to recover the private DSA key.
5461
5462   This issue was reported by César Pereida (Aalto University), Billy Brumley
5463   (Tampere University of Technology), and Yuval Yarom (The University of
5464   Adelaide and NICTA).
5465   ([CVE-2016-2178])
5466
5467   *César Pereida*
5468
5469 * DTLS buffered message DoS
5470
5471   In a DTLS connection where handshake messages are delivered out-of-order
5472   those messages that OpenSSL is not yet ready to process will be buffered
5473   for later use. Under certain circumstances, a flaw in the logic means that
5474   those messages do not get removed from the buffer even though the handshake
5475   has been completed. An attacker could force up to approx. 15 messages to
5476   remain in the buffer when they are no longer required. These messages will
5477   be cleared when the DTLS connection is closed. The default maximum size for
5478   a message is 100k. Therefore the attacker could force an additional 1500k
5479   to be consumed per connection. By opening many simulataneous connections an
5480   attacker could cause a DoS attack through memory exhaustion.
5481
5482   This issue was reported to OpenSSL by Quan Luo.
5483   ([CVE-2016-2179])
5484
5485   *Matt Caswell*
5486
5487 * DTLS replay protection DoS
5488
5489   A flaw in the DTLS replay attack protection mechanism means that records
5490   that arrive for future epochs update the replay protection "window" before
5491   the MAC for the record has been validated. This could be exploited by an
5492   attacker by sending a record for the next epoch (which does not have to
5493   decrypt or have a valid MAC), with a very large sequence number. This means
5494   that all subsequent legitimate packets are dropped causing a denial of
5495   service for a specific DTLS connection.
5496
5497   This issue was reported to OpenSSL by the OCAP audit team.
5498   ([CVE-2016-2181])
5499
5500   *Matt Caswell*
5501
5502 * Certificate message OOB reads
5503
5504   In OpenSSL 1.0.2 and earlier some missing message length checks can result
5505   in OOB reads of up to 2 bytes beyond an allocated buffer. There is a
5506   theoretical DoS risk but this has not been observed in practice on common
5507   platforms.
5508
5509   The messages affected are client certificate, client certificate request
5510   and server certificate. As a result the attack can only be performed
5511   against a client or a server which enables client authentication.
5512
5513   This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.)
5514   ([CVE-2016-6306])
5515
5516   *Stephen Henson*
5517
5518### Changes between 1.0.2g and 1.0.2h [3 May 2016]
5519
5520 * Prevent padding oracle in AES-NI CBC MAC check
5521
5522   A MITM attacker can use a padding oracle attack to decrypt traffic
5523   when the connection uses an AES CBC cipher and the server support
5524   AES-NI.
5525
5526   This issue was introduced as part of the fix for Lucky 13 padding
5527   attack ([CVE-2013-0169]). The padding check was rewritten to be in
5528   constant time by making sure that always the same bytes are read and
5529   compared against either the MAC or padding bytes. But it no longer
5530   checked that there was enough data to have both the MAC and padding
5531   bytes.
5532
5533   This issue was reported by Juraj Somorovsky using TLS-Attacker.
5534
5535   *Kurt Roeckx*
5536
5537 * Fix EVP_EncodeUpdate overflow
5538
5539   An overflow can occur in the EVP_EncodeUpdate() function which is used for
5540   Base64 encoding of binary data. If an attacker is able to supply very large
5541   amounts of input data then a length check can overflow resulting in a heap
5542   corruption.
5543
5544   Internally to OpenSSL the EVP_EncodeUpdate() function is primarily used by
5545   the `PEM_write_bio*` family of functions. These are mainly used within the
5546   OpenSSL command line applications, so any application which processes data
5547   from an untrusted source and outputs it as a PEM file should be considered
5548   vulnerable to this issue. User applications that call these APIs directly
5549   with large amounts of untrusted data may also be vulnerable.
5550
5551   This issue was reported by Guido Vranken.
5552   ([CVE-2016-2105])
5553
5554   *Matt Caswell*
5555
5556 * Fix EVP_EncryptUpdate overflow
5557
5558   An overflow can occur in the EVP_EncryptUpdate() function. If an attacker
5559   is able to supply very large amounts of input data after a previous call to
5560   EVP_EncryptUpdate() with a partial block then a length check can overflow
5561   resulting in a heap corruption. Following an analysis of all OpenSSL
5562   internal usage of the EVP_EncryptUpdate() function all usage is one of two
5563   forms. The first form is where the EVP_EncryptUpdate() call is known to be
5564   the first called function after an EVP_EncryptInit(), and therefore that
5565   specific call must be safe. The second form is where the length passed to
5566   EVP_EncryptUpdate() can be seen from the code to be some small value and
5567   therefore there is no possibility of an overflow. Since all instances are
5568   one of these two forms, it is believed that there can be no overflows in
5569   internal code due to this problem. It should be noted that
5570   EVP_DecryptUpdate() can call EVP_EncryptUpdate() in certain code paths.
5571   Also EVP_CipherUpdate() is a synonym for EVP_EncryptUpdate(). All instances
5572   of these calls have also been analysed too and it is believed there are no
5573   instances in internal usage where an overflow could occur.
5574
5575   This issue was reported by Guido Vranken.
5576   ([CVE-2016-2106])
5577
5578   *Matt Caswell*
5579
5580 * Prevent ASN.1 BIO excessive memory allocation
5581
5582   When ASN.1 data is read from a BIO using functions such as d2i_CMS_bio()
5583   a short invalid encoding can cause allocation of large amounts of memory
5584   potentially consuming excessive resources or exhausting memory.
5585
5586   Any application parsing untrusted data through d2i BIO functions is
5587   affected. The memory based functions such as d2i_X509() are *not* affected.
5588   Since the memory based functions are used by the TLS library, TLS
5589   applications are not affected.
5590
5591   This issue was reported by Brian Carpenter.
5592   ([CVE-2016-2109])
5593
5594   *Stephen Henson*
5595
5596 * EBCDIC overread
5597
5598   ASN1 Strings that are over 1024 bytes can cause an overread in applications
5599   using the X509_NAME_oneline() function on EBCDIC systems. This could result
5600   in arbitrary stack data being returned in the buffer.
5601
5602   This issue was reported by Guido Vranken.
5603   ([CVE-2016-2176])
5604
5605   *Matt Caswell*
5606
5607 * Modify behavior of ALPN to invoke callback after SNI/servername
5608   callback, such that updates to the SSL_CTX affect ALPN.
5609
5610   *Todd Short*
5611
5612 * Remove LOW from the DEFAULT cipher list.  This removes singles DES from the
5613   default.
5614
5615   *Kurt Roeckx*
5616
5617 * Only remove the SSLv2 methods with the no-ssl2-method option. When the
5618   methods are enabled and ssl2 is disabled the methods return NULL.
5619
5620   *Kurt Roeckx*
5621
5622### Changes between 1.0.2f and 1.0.2g [1 Mar 2016]
5623
5624* Disable weak ciphers in SSLv3 and up in default builds of OpenSSL.
5625  Builds that are not configured with "enable-weak-ssl-ciphers" will not
5626  provide any "EXPORT" or "LOW" strength ciphers.
5627
5628  *Viktor Dukhovni*
5629
5630* Disable SSLv2 default build, default negotiation and weak ciphers.  SSLv2
5631  is by default disabled at build-time.  Builds that are not configured with
5632  "enable-ssl2" will not support SSLv2.  Even if "enable-ssl2" is used,
5633  users who want to negotiate SSLv2 via the version-flexible SSLv23_method()
5634  will need to explicitly call either of:
5635
5636      SSL_CTX_clear_options(ctx, SSL_OP_NO_SSLv2);
5637  or
5638      SSL_clear_options(ssl, SSL_OP_NO_SSLv2);
5639
5640  as appropriate.  Even if either of those is used, or the application
5641  explicitly uses the version-specific SSLv2_method() or its client and
5642  server variants, SSLv2 ciphers vulnerable to exhaustive search key
5643  recovery have been removed.  Specifically, the SSLv2 40-bit EXPORT
5644  ciphers, and SSLv2 56-bit DES are no longer available.
5645  ([CVE-2016-0800])
5646
5647   *Viktor Dukhovni*
5648
5649 * Fix a double-free in DSA code
5650
5651   A double free bug was discovered when OpenSSL parses malformed DSA private
5652   keys and could lead to a DoS attack or memory corruption for applications
5653   that receive DSA private keys from untrusted sources.  This scenario is
5654   considered rare.
5655
5656   This issue was reported to OpenSSL by Adam Langley(Google/BoringSSL) using
5657   libFuzzer.
5658   ([CVE-2016-0705])
5659
5660   *Stephen Henson*
5661
5662 * Disable SRP fake user seed to address a server memory leak.
5663
5664   Add a new method SRP_VBASE_get1_by_user that handles the seed properly.
5665
5666   SRP_VBASE_get_by_user had inconsistent memory management behaviour.
5667   In order to fix an unavoidable memory leak, SRP_VBASE_get_by_user
5668   was changed to ignore the "fake user" SRP seed, even if the seed
5669   is configured.
5670
5671   Users should use SRP_VBASE_get1_by_user instead. Note that in
5672   SRP_VBASE_get1_by_user, caller must free the returned value. Note
5673   also that even though configuring the SRP seed attempts to hide
5674   invalid usernames by continuing the handshake with fake
5675   credentials, this behaviour is not constant time and no strong
5676   guarantees are made that the handshake is indistinguishable from
5677   that of a valid user.
5678   ([CVE-2016-0798])
5679
5680   *Emilia Käsper*
5681
5682 * Fix BN_hex2bn/BN_dec2bn NULL pointer deref/heap corruption
5683
5684   In the BN_hex2bn function the number of hex digits is calculated using an
5685   int value `i`. Later `bn_expand` is called with a value of `i * 4`. For
5686   large values of `i` this can result in `bn_expand` not allocating any
5687   memory because `i * 4` is negative. This can leave the internal BIGNUM data
5688   field as NULL leading to a subsequent NULL ptr deref. For very large values
5689   of `i`, the calculation `i * 4` could be a positive value smaller than `i`.
5690   In this case memory is allocated to the internal BIGNUM data field, but it
5691   is insufficiently sized leading to heap corruption. A similar issue exists
5692   in BN_dec2bn. This could have security consequences if BN_hex2bn/BN_dec2bn
5693   is ever called by user applications with very large untrusted hex/dec data.
5694   This is anticipated to be a rare occurrence.
5695
5696   All OpenSSL internal usage of these functions use data that is not expected
5697   to be untrusted, e.g. config file data or application command line
5698   arguments. If user developed applications generate config file data based
5699   on untrusted data then it is possible that this could also lead to security
5700   consequences. This is also anticipated to be rare.
5701
5702   This issue was reported to OpenSSL by Guido Vranken.
5703   ([CVE-2016-0797])
5704
5705   *Matt Caswell*
5706
5707 * Fix memory issues in `BIO_*printf` functions
5708
5709   The internal `fmtstr` function used in processing a "%s" format string in
5710   the `BIO_*printf` functions could overflow while calculating the length of a
5711   string and cause an OOB read when printing very long strings.
5712
5713   Additionally the internal `doapr_outch` function can attempt to write to an
5714   OOB memory location (at an offset from the NULL pointer) in the event of a
5715   memory allocation failure. In 1.0.2 and below this could be caused where
5716   the size of a buffer to be allocated is greater than INT_MAX. E.g. this
5717   could be in processing a very long "%s" format string. Memory leaks can
5718   also occur.
5719
5720   The first issue may mask the second issue dependent on compiler behaviour.
5721   These problems could enable attacks where large amounts of untrusted data
5722   is passed to the `BIO_*printf` functions. If applications use these functions
5723   in this way then they could be vulnerable. OpenSSL itself uses these
5724   functions when printing out human-readable dumps of ASN.1 data. Therefore
5725   applications that print this data could be vulnerable if the data is from
5726   untrusted sources. OpenSSL command line applications could also be
5727   vulnerable where they print out ASN.1 data, or if untrusted data is passed
5728   as command line arguments.
5729
5730   Libssl is not considered directly vulnerable. Additionally certificates etc
5731   received via remote connections via libssl are also unlikely to be able to
5732   trigger these issues because of message size limits enforced within libssl.
5733
5734   This issue was reported to OpenSSL Guido Vranken.
5735   ([CVE-2016-0799])
5736
5737   *Matt Caswell*
5738
5739 * Side channel attack on modular exponentiation
5740
5741   A side-channel attack was found which makes use of cache-bank conflicts on
5742   the Intel Sandy-Bridge microarchitecture which could lead to the recovery
5743   of RSA keys.  The ability to exploit this issue is limited as it relies on
5744   an attacker who has control of code in a thread running on the same
5745   hyper-threaded core as the victim thread which is performing decryptions.
5746
5747   This issue was reported to OpenSSL by Yuval Yarom, The University of
5748   Adelaide and NICTA, Daniel Genkin, Technion and Tel Aviv University, and
5749   Nadia Heninger, University of Pennsylvania with more information at
5750   <http://cachebleed.info>.
5751   ([CVE-2016-0702])
5752
5753   *Andy Polyakov*
5754
5755 * Change the `req` command to generate a 2048-bit RSA/DSA key by default,
5756   if no keysize is specified with default_bits. This fixes an
5757   omission in an earlier change that changed all RSA/DSA key generation
5758   commands to use 2048 bits by default.
5759
5760   *Emilia Käsper*
5761
5762### Changes between 1.0.2e and 1.0.2f [28 Jan 2016]
5763
5764 * DH small subgroups
5765
5766   Historically OpenSSL only ever generated DH parameters based on "safe"
5767   primes. More recently (in version 1.0.2) support was provided for
5768   generating X9.42 style parameter files such as those required for RFC 5114
5769   support. The primes used in such files may not be "safe". Where an
5770   application is using DH configured with parameters based on primes that are
5771   not "safe" then an attacker could use this fact to find a peer's private
5772   DH exponent. This attack requires that the attacker complete multiple
5773   handshakes in which the peer uses the same private DH exponent. For example
5774   this could be used to discover a TLS server's private DH exponent if it's
5775   reusing the private DH exponent or it's using a static DH ciphersuite.
5776
5777   OpenSSL provides the option SSL_OP_SINGLE_DH_USE for ephemeral DH (DHE) in
5778   TLS. It is not on by default. If the option is not set then the server
5779   reuses the same private DH exponent for the life of the server process and
5780   would be vulnerable to this attack. It is believed that many popular
5781   applications do set this option and would therefore not be at risk.
5782
5783   The fix for this issue adds an additional check where a "q" parameter is
5784   available (as is the case in X9.42 based parameters). This detects the
5785   only known attack, and is the only possible defense for static DH
5786   ciphersuites. This could have some performance impact.
5787
5788   Additionally the SSL_OP_SINGLE_DH_USE option has been switched on by
5789   default and cannot be disabled. This could have some performance impact.
5790
5791   This issue was reported to OpenSSL by Antonio Sanso (Adobe).
5792   ([CVE-2016-0701])
5793
5794   *Matt Caswell*
5795
5796 * SSLv2 doesn't block disabled ciphers
5797
5798   A malicious client can negotiate SSLv2 ciphers that have been disabled on
5799   the server and complete SSLv2 handshakes even if all SSLv2 ciphers have
5800   been disabled, provided that the SSLv2 protocol was not also disabled via
5801   SSL_OP_NO_SSLv2.
5802
5803   This issue was reported to OpenSSL on 26th December 2015 by Nimrod Aviram
5804   and Sebastian Schinzel.
5805   ([CVE-2015-3197])
5806
5807   *Viktor Dukhovni*
5808
5809### Changes between 1.0.2d and 1.0.2e [3 Dec 2015]
5810
5811 * BN_mod_exp may produce incorrect results on x86_64
5812
5813   There is a carry propagating bug in the x86_64 Montgomery squaring
5814   procedure. No EC algorithms are affected. Analysis suggests that attacks
5815   against RSA and DSA as a result of this defect would be very difficult to
5816   perform and are not believed likely. Attacks against DH are considered just
5817   feasible (although very difficult) because most of the work necessary to
5818   deduce information about a private key may be performed offline. The amount
5819   of resources required for such an attack would be very significant and
5820   likely only accessible to a limited number of attackers. An attacker would
5821   additionally need online access to an unpatched system using the target
5822   private key in a scenario with persistent DH parameters and a private
5823   key that is shared between multiple clients. For example this can occur by
5824   default in OpenSSL DHE based SSL/TLS ciphersuites.
5825
5826   This issue was reported to OpenSSL by Hanno Böck.
5827   ([CVE-2015-3193])
5828
5829   *Andy Polyakov*
5830
5831 * Certificate verify crash with missing PSS parameter
5832
5833   The signature verification routines will crash with a NULL pointer
5834   dereference if presented with an ASN.1 signature using the RSA PSS
5835   algorithm and absent mask generation function parameter. Since these
5836   routines are used to verify certificate signature algorithms this can be
5837   used to crash any certificate verification operation and exploited in a
5838   DoS attack. Any application which performs certificate verification is
5839   vulnerable including OpenSSL clients and servers which enable client
5840   authentication.
5841
5842   This issue was reported to OpenSSL by Loïc Jonas Etienne (Qnective AG).
5843   ([CVE-2015-3194])
5844
5845   *Stephen Henson*
5846
5847 * X509_ATTRIBUTE memory leak
5848
5849   When presented with a malformed X509_ATTRIBUTE structure OpenSSL will leak
5850   memory. This structure is used by the PKCS#7 and CMS routines so any
5851   application which reads PKCS#7 or CMS data from untrusted sources is
5852   affected. SSL/TLS is not affected.
5853
5854   This issue was reported to OpenSSL by Adam Langley (Google/BoringSSL) using
5855   libFuzzer.
5856   ([CVE-2015-3195])
5857
5858   *Stephen Henson*
5859
5860 * Rewrite EVP_DecodeUpdate (base64 decoding) to fix several bugs.
5861   This changes the decoding behaviour for some invalid messages,
5862   though the change is mostly in the more lenient direction, and
5863   legacy behaviour is preserved as much as possible.
5864
5865   *Emilia Käsper*
5866
5867 * In DSA_generate_parameters_ex, if the provided seed is too short,
5868   return an error
5869
5870   *Rich Salz and Ismo Puustinen <ismo.puustinen@intel.com>*
5871
5872### Changes between 1.0.2c and 1.0.2d [9 Jul 2015]
5873
5874 * Alternate chains certificate forgery
5875
5876   During certificate verification, OpenSSL will attempt to find an
5877   alternative certificate chain if the first attempt to build such a chain
5878   fails. An error in the implementation of this logic can mean that an
5879   attacker could cause certain checks on untrusted certificates to be
5880   bypassed, such as the CA flag, enabling them to use a valid leaf
5881   certificate to act as a CA and "issue" an invalid certificate.
5882
5883   This issue was reported to OpenSSL by Adam Langley/David Benjamin
5884   (Google/BoringSSL).
5885
5886   *Matt Caswell*
5887
5888### Changes between 1.0.2b and 1.0.2c [12 Jun 2015]
5889
5890 * Fix HMAC ABI incompatibility. The previous version introduced an ABI
5891   incompatibility in the handling of HMAC. The previous ABI has now been
5892   restored.
5893
5894   *Matt Caswell*
5895
5896### Changes between 1.0.2a and 1.0.2b [11 Jun 2015]
5897
5898 * Malformed ECParameters causes infinite loop
5899
5900   When processing an ECParameters structure OpenSSL enters an infinite loop
5901   if the curve specified is over a specially malformed binary polynomial
5902   field.
5903
5904   This can be used to perform denial of service against any
5905   system which processes public keys, certificate requests or
5906   certificates.  This includes TLS clients and TLS servers with
5907   client authentication enabled.
5908
5909   This issue was reported to OpenSSL by Joseph Barr-Pixton.
5910   ([CVE-2015-1788])
5911
5912   *Andy Polyakov*
5913
5914 * Exploitable out-of-bounds read in X509_cmp_time
5915
5916   X509_cmp_time does not properly check the length of the ASN1_TIME
5917   string and can read a few bytes out of bounds. In addition,
5918   X509_cmp_time accepts an arbitrary number of fractional seconds in the
5919   time string.
5920
5921   An attacker can use this to craft malformed certificates and CRLs of
5922   various sizes and potentially cause a segmentation fault, resulting in
5923   a DoS on applications that verify certificates or CRLs. TLS clients
5924   that verify CRLs are affected. TLS clients and servers with client
5925   authentication enabled may be affected if they use custom verification
5926   callbacks.
5927
5928   This issue was reported to OpenSSL by Robert Swiecki (Google), and
5929   independently by Hanno Böck.
5930   ([CVE-2015-1789])
5931
5932   *Emilia Käsper*
5933
5934 * PKCS7 crash with missing EnvelopedContent
5935
5936   The PKCS#7 parsing code does not handle missing inner EncryptedContent
5937   correctly. An attacker can craft malformed ASN.1-encoded PKCS#7 blobs
5938   with missing content and trigger a NULL pointer dereference on parsing.
5939
5940   Applications that decrypt PKCS#7 data or otherwise parse PKCS#7
5941   structures from untrusted sources are affected. OpenSSL clients and
5942   servers are not affected.
5943
5944   This issue was reported to OpenSSL by Michal Zalewski (Google).
5945   ([CVE-2015-1790])
5946
5947   *Emilia Käsper*
5948
5949 * CMS verify infinite loop with unknown hash function
5950
5951   When verifying a signedData message the CMS code can enter an infinite loop
5952   if presented with an unknown hash function OID. This can be used to perform
5953   denial of service against any system which verifies signedData messages using
5954   the CMS code.
5955   This issue was reported to OpenSSL by Johannes Bauer.
5956   ([CVE-2015-1792])
5957
5958   *Stephen Henson*
5959
5960 * Race condition handling NewSessionTicket
5961
5962   If a NewSessionTicket is received by a multi-threaded client when attempting to
5963   reuse a previous ticket then a race condition can occur potentially leading to
5964   a double free of the ticket data.
5965   ([CVE-2015-1791])
5966
5967   *Matt Caswell*
5968
5969 * Only support 256-bit or stronger elliptic curves with the
5970   'ecdh_auto' setting (server) or by default (client). Of supported
5971   curves, prefer P-256 (both).
5972
5973   *Emilia Kasper*
5974
5975### Changes between 1.0.2 and 1.0.2a [19 Mar 2015]
5976
5977 * ClientHello sigalgs DoS fix
5978
5979   If a client connects to an OpenSSL 1.0.2 server and renegotiates with an
5980   invalid signature algorithms extension a NULL pointer dereference will
5981   occur. This can be exploited in a DoS attack against the server.
5982
5983   This issue was was reported to OpenSSL by David Ramos of Stanford
5984   University.
5985   ([CVE-2015-0291])
5986
5987   *Stephen Henson and Matt Caswell*
5988
5989 * Multiblock corrupted pointer fix
5990
5991   OpenSSL 1.0.2 introduced the "multiblock" performance improvement. This
5992   feature only applies on 64 bit x86 architecture platforms that support AES
5993   NI instructions. A defect in the implementation of "multiblock" can cause
5994   OpenSSL's internal write buffer to become incorrectly set to NULL when
5995   using non-blocking IO. Typically, when the user application is using a
5996   socket BIO for writing, this will only result in a failed connection.
5997   However if some other BIO is used then it is likely that a segmentation
5998   fault will be triggered, thus enabling a potential DoS attack.
5999
6000   This issue was reported to OpenSSL by Daniel Danner and Rainer Mueller.
6001   ([CVE-2015-0290])
6002
6003   *Matt Caswell*
6004
6005 * Segmentation fault in DTLSv1_listen fix
6006
6007   The DTLSv1_listen function is intended to be stateless and processes the
6008   initial ClientHello from many peers. It is common for user code to loop
6009   over the call to DTLSv1_listen until a valid ClientHello is received with
6010   an associated cookie. A defect in the implementation of DTLSv1_listen means
6011   that state is preserved in the SSL object from one invocation to the next
6012   that can lead to a segmentation fault. Errors processing the initial
6013   ClientHello can trigger this scenario. An example of such an error could be
6014   that a DTLS1.0 only client is attempting to connect to a DTLS1.2 only
6015   server.
6016
6017   This issue was reported to OpenSSL by Per Allansson.
6018   ([CVE-2015-0207])
6019
6020   *Matt Caswell*
6021
6022 * Segmentation fault in ASN1_TYPE_cmp fix
6023
6024   The function ASN1_TYPE_cmp will crash with an invalid read if an attempt is
6025   made to compare ASN.1 boolean types. Since ASN1_TYPE_cmp is used to check
6026   certificate signature algorithm consistency this can be used to crash any
6027   certificate verification operation and exploited in a DoS attack. Any
6028   application which performs certificate verification is vulnerable including
6029   OpenSSL clients and servers which enable client authentication.
6030   ([CVE-2015-0286])
6031
6032   *Stephen Henson*
6033
6034 * Segmentation fault for invalid PSS parameters fix
6035
6036   The signature verification routines will crash with a NULL pointer
6037   dereference if presented with an ASN.1 signature using the RSA PSS
6038   algorithm and invalid parameters. Since these routines are used to verify
6039   certificate signature algorithms this can be used to crash any
6040   certificate verification operation and exploited in a DoS attack. Any
6041   application which performs certificate verification is vulnerable including
6042   OpenSSL clients and servers which enable client authentication.
6043
6044   This issue was was reported to OpenSSL by Brian Carpenter.
6045   ([CVE-2015-0208])
6046
6047   *Stephen Henson*
6048
6049 * ASN.1 structure reuse memory corruption fix
6050
6051   Reusing a structure in ASN.1 parsing may allow an attacker to cause
6052   memory corruption via an invalid write. Such reuse is and has been
6053   strongly discouraged and is believed to be rare.
6054
6055   Applications that parse structures containing CHOICE or ANY DEFINED BY
6056   components may be affected. Certificate parsing (d2i_X509 and related
6057   functions) are however not affected. OpenSSL clients and servers are
6058   not affected.
6059   ([CVE-2015-0287])
6060
6061   *Stephen Henson*
6062
6063 * PKCS7 NULL pointer dereferences fix
6064
6065   The PKCS#7 parsing code does not handle missing outer ContentInfo
6066   correctly. An attacker can craft malformed ASN.1-encoded PKCS#7 blobs with
6067   missing content and trigger a NULL pointer dereference on parsing.
6068
6069   Applications that verify PKCS#7 signatures, decrypt PKCS#7 data or
6070   otherwise parse PKCS#7 structures from untrusted sources are
6071   affected. OpenSSL clients and servers are not affected.
6072
6073   This issue was reported to OpenSSL by Michal Zalewski (Google).
6074   ([CVE-2015-0289])
6075
6076   *Emilia Käsper*
6077
6078 * DoS via reachable assert in SSLv2 servers fix
6079
6080   A malicious client can trigger an OPENSSL_assert (i.e., an abort) in
6081   servers that both support SSLv2 and enable export cipher suites by sending
6082   a specially crafted SSLv2 CLIENT-MASTER-KEY message.
6083
6084   This issue was discovered by Sean Burford (Google) and Emilia Käsper
6085   (OpenSSL development team).
6086   ([CVE-2015-0293])
6087
6088   *Emilia Käsper*
6089
6090 * Empty CKE with client auth and DHE fix
6091
6092   If client auth is used then a server can seg fault in the event of a DHE
6093   ciphersuite being selected and a zero length ClientKeyExchange message
6094   being sent by the client. This could be exploited in a DoS attack.
6095   ([CVE-2015-1787])
6096
6097   *Matt Caswell*
6098
6099 * Handshake with unseeded PRNG fix
6100
6101   Under certain conditions an OpenSSL 1.0.2 client can complete a handshake
6102   with an unseeded PRNG. The conditions are:
6103   - The client is on a platform where the PRNG has not been seeded
6104   automatically, and the user has not seeded manually
6105   - A protocol specific client method version has been used (i.e. not
6106   SSL_client_methodv23)
6107   - A ciphersuite is used that does not require additional random data from
6108   the PRNG beyond the initial ClientHello client random (e.g. PSK-RC4-SHA).
6109
6110   If the handshake succeeds then the client random that has been used will
6111   have been generated from a PRNG with insufficient entropy and therefore the
6112   output may be predictable.
6113
6114   For example using the following command with an unseeded openssl will
6115   succeed on an unpatched platform:
6116
6117   openssl s_client -psk 1a2b3c4d -tls1_2 -cipher PSK-RC4-SHA
6118   ([CVE-2015-0285])
6119
6120   *Matt Caswell*
6121
6122 * Use After Free following d2i_ECPrivatekey error fix
6123
6124   A malformed EC private key file consumed via the d2i_ECPrivateKey function
6125   could cause a use after free condition. This, in turn, could cause a double
6126   free in several private key parsing functions (such as d2i_PrivateKey
6127   or EVP_PKCS82PKEY) and could lead to a DoS attack or memory corruption
6128   for applications that receive EC private keys from untrusted
6129   sources. This scenario is considered rare.
6130
6131   This issue was discovered by the BoringSSL project and fixed in their
6132   commit 517073cd4b.
6133   ([CVE-2015-0209])
6134
6135   *Matt Caswell*
6136
6137 * X509_to_X509_REQ NULL pointer deref fix
6138
6139   The function X509_to_X509_REQ will crash with a NULL pointer dereference if
6140   the certificate key is invalid. This function is rarely used in practice.
6141
6142   This issue was discovered by Brian Carpenter.
6143   ([CVE-2015-0288])
6144
6145   *Stephen Henson*
6146
6147 * Removed the export ciphers from the DEFAULT ciphers
6148
6149   *Kurt Roeckx*
6150
6151### Changes between 1.0.1l and 1.0.2 [22 Jan 2015]
6152
6153 * Facilitate "universal" ARM builds targeting range of ARM ISAs, e.g.
6154   ARMv5 through ARMv8, as opposite to "locking" it to single one.
6155   So far those who have to target multiple platforms would compromise
6156   and argue that binary targeting say ARMv5 would still execute on
6157   ARMv8. "Universal" build resolves this compromise by providing
6158   near-optimal performance even on newer platforms.
6159
6160   *Andy Polyakov*
6161
6162 * Accelerated NIST P-256 elliptic curve implementation for x86_64
6163   (other platforms pending).
6164
6165   *Shay Gueron & Vlad Krasnov (Intel Corp), Andy Polyakov*
6166
6167 * Add support for the SignedCertificateTimestampList certificate and
6168   OCSP response extensions from RFC6962.
6169
6170   *Rob Stradling*
6171
6172 * Fix ec_GFp_simple_points_make_affine (thus, EC_POINTs_mul etc.)
6173   for corner cases. (Certain input points at infinity could lead to
6174   bogus results, with non-infinity inputs mapped to infinity too.)
6175
6176   *Bodo Moeller*
6177
6178 * Initial support for PowerISA 2.0.7, first implemented in POWER8.
6179   This covers AES, SHA256/512 and GHASH. "Initial" means that most
6180   common cases are optimized and there still is room for further
6181   improvements. Vector Permutation AES for Altivec is also added.
6182
6183   *Andy Polyakov*
6184
6185 * Add support for little-endian ppc64 Linux target.
6186
6187   *Marcelo Cerri (IBM)*
6188
6189 * Initial support for AMRv8 ISA crypto extensions. This covers AES,
6190   SHA1, SHA256 and GHASH. "Initial" means that most common cases
6191   are optimized and there still is room for further improvements.
6192   Both 32- and 64-bit modes are supported.
6193
6194   *Andy Polyakov, Ard Biesheuvel (Linaro)*
6195
6196 * Improved ARMv7 NEON support.
6197
6198   *Andy Polyakov*
6199
6200 * Support for SPARC Architecture 2011 crypto extensions, first
6201   implemented in SPARC T4. This covers AES, DES, Camellia, SHA1,
6202   SHA256/512, MD5, GHASH and modular exponentiation.
6203
6204   *Andy Polyakov, David Miller*
6205
6206 * Accelerated modular exponentiation for Intel processors, a.k.a.
6207   RSAZ.
6208
6209   *Shay Gueron & Vlad Krasnov (Intel Corp)*
6210
6211 * Support for new and upcoming Intel processors, including AVX2,
6212   BMI and SHA ISA extensions. This includes additional "stitched"
6213   implementations, AESNI-SHA256 and GCM, and multi-buffer support
6214   for TLS encrypt.
6215
6216   This work was sponsored by Intel Corp.
6217
6218   *Andy Polyakov*
6219
6220 * Support for DTLS 1.2. This adds two sets of DTLS methods: DTLS_*_method()
6221   supports both DTLS 1.2 and 1.0 and should use whatever version the peer
6222   supports and DTLSv1_2_*_method() which supports DTLS 1.2 only.
6223
6224   *Steve Henson*
6225
6226 * Use algorithm specific chains in SSL_CTX_use_certificate_chain_file():
6227   this fixes a limitation in previous versions of OpenSSL.
6228
6229   *Steve Henson*
6230
6231 * Extended RSA OAEP support via EVP_PKEY API. Options to specify digest,
6232   MGF1 digest and OAEP label.
6233
6234   *Steve Henson*
6235
6236 * Add EVP support for key wrapping algorithms, to avoid problems with
6237   existing code the flag EVP_CIPHER_CTX_WRAP_ALLOW has to be set in
6238   the EVP_CIPHER_CTX or an error is returned. Add AES and DES3 wrap
6239   algorithms and include tests cases.
6240
6241   *Steve Henson*
6242
6243 * Add functions to allocate and set the fields of an ECDSA_METHOD
6244   structure.
6245
6246   *Douglas E. Engert, Steve Henson*
6247
6248 * New functions OPENSSL_gmtime_diff and ASN1_TIME_diff to find the
6249   difference in days and seconds between two tm or ASN1_TIME structures.
6250
6251   *Steve Henson*
6252
6253 * Add -rev test option to s_server to just reverse order of characters
6254   received by client and send back to server. Also prints an abbreviated
6255   summary of the connection parameters.
6256
6257   *Steve Henson*
6258
6259 * New option -brief for s_client and s_server to print out a brief summary
6260   of connection parameters.
6261
6262   *Steve Henson*
6263
6264 * Add callbacks for arbitrary TLS extensions.
6265
6266   *Trevor Perrin <trevp@trevp.net> and Ben Laurie*
6267
6268 * New option -crl_download in several openssl utilities to download CRLs
6269   from CRLDP extension in certificates.
6270
6271   *Steve Henson*
6272
6273 * New options -CRL and -CRLform for s_client and s_server for CRLs.
6274
6275   *Steve Henson*
6276
6277 * New function X509_CRL_diff to generate a delta CRL from the difference
6278   of two full CRLs. Add support to "crl" utility.
6279
6280   *Steve Henson*
6281
6282 * New functions to set lookup_crls function and to retrieve
6283   X509_STORE from X509_STORE_CTX.
6284
6285   *Steve Henson*
6286
6287 * Print out deprecated issuer and subject unique ID fields in
6288   certificates.
6289
6290   *Steve Henson*
6291
6292 * Extend OCSP I/O functions so they can be used for simple general purpose
6293   HTTP as well as OCSP. New wrapper function which can be used to download
6294   CRLs using the OCSP API.
6295
6296   *Steve Henson*
6297
6298 * Delegate command line handling in s_client/s_server to SSL_CONF APIs.
6299
6300   *Steve Henson*
6301
6302 * `SSL_CONF*` functions. These provide a common framework for application
6303   configuration using configuration files or command lines.
6304
6305   *Steve Henson*
6306
6307 * SSL/TLS tracing code. This parses out SSL/TLS records using the
6308   message callback and prints the results. Needs compile time option
6309   "enable-ssl-trace". New options to s_client and s_server to enable
6310   tracing.
6311
6312   *Steve Henson*
6313
6314 * New ctrl and macro to retrieve supported points extensions.
6315   Print out extension in s_server and s_client.
6316
6317   *Steve Henson*
6318
6319 * New functions to retrieve certificate signature and signature
6320   OID NID.
6321
6322   *Steve Henson*
6323
6324 * Add functions to retrieve and manipulate the raw cipherlist sent by a
6325   client to OpenSSL.
6326
6327   *Steve Henson*
6328
6329 * New Suite B modes for TLS code. These use and enforce the requirements
6330   of RFC6460: restrict ciphersuites, only permit Suite B algorithms and
6331   only use Suite B curves. The Suite B modes can be set by using the
6332   strings "SUITEB128", "SUITEB192" or "SUITEB128ONLY" for the cipherstring.
6333
6334   *Steve Henson*
6335
6336 * New chain verification flags for Suite B levels of security. Check
6337   algorithms are acceptable when flags are set in X509_verify_cert.
6338
6339   *Steve Henson*
6340
6341 * Make tls1_check_chain return a set of flags indicating checks passed
6342   by a certificate chain. Add additional tests to handle client
6343   certificates: checks for matching certificate type and issuer name
6344   comparison.
6345
6346   *Steve Henson*
6347
6348 * If an attempt is made to use a signature algorithm not in the peer
6349   preference list abort the handshake. If client has no suitable
6350   signature algorithms in response to a certificate request do not
6351   use the certificate.
6352
6353   *Steve Henson*
6354
6355 * If server EC tmp key is not in client preference list abort handshake.
6356
6357   *Steve Henson*
6358
6359 * Add support for certificate stores in CERT structure. This makes it
6360   possible to have different stores per SSL structure or one store in
6361   the parent SSL_CTX. Include distinct stores for certificate chain
6362   verification and chain building. New ctrl SSL_CTRL_BUILD_CERT_CHAIN
6363   to build and store a certificate chain in CERT structure: returning
6364   an error if the chain cannot be built: this will allow applications
6365   to test if a chain is correctly configured.
6366
6367   Note: if the CERT based stores are not set then the parent SSL_CTX
6368   store is used to retain compatibility with existing behaviour.
6369
6370   *Steve Henson*
6371
6372 * New function ssl_set_client_disabled to set a ciphersuite disabled
6373   mask based on the current session, check mask when sending client
6374   hello and checking the requested ciphersuite.
6375
6376   *Steve Henson*
6377
6378 * New ctrls to retrieve and set certificate types in a certificate
6379   request message. Print out received values in s_client. If certificate
6380   types is not set with custom values set sensible values based on
6381   supported signature algorithms.
6382
6383   *Steve Henson*
6384
6385 * Support for distinct client and server supported signature algorithms.
6386
6387   *Steve Henson*
6388
6389 * Add certificate callback. If set this is called whenever a certificate
6390   is required by client or server. An application can decide which
6391   certificate chain to present based on arbitrary criteria: for example
6392   supported signature algorithms. Add very simple example to s_server.
6393   This fixes many of the problems and restrictions of the existing client
6394   certificate callback: for example you can now clear an existing
6395   certificate and specify the whole chain.
6396
6397   *Steve Henson*
6398
6399 * Add new "valid_flags" field to CERT_PKEY structure which determines what
6400   the certificate can be used for (if anything). Set valid_flags field
6401   in new tls1_check_chain function. Simplify ssl_set_cert_masks which used
6402   to have similar checks in it.
6403
6404   Add new "cert_flags" field to CERT structure and include a "strict mode".
6405   This enforces some TLS certificate requirements (such as only permitting
6406   certificate signature algorithms contained in the supported algorithms
6407   extension) which some implementations ignore: this option should be used
6408   with caution as it could cause interoperability issues.
6409
6410   *Steve Henson*
6411
6412 * Update and tidy signature algorithm extension processing. Work out
6413   shared signature algorithms based on preferences and peer algorithms
6414   and print them out in s_client and s_server. Abort handshake if no
6415   shared signature algorithms.
6416
6417   *Steve Henson*
6418
6419 * Add new functions to allow customised supported signature algorithms
6420   for SSL and SSL_CTX structures. Add options to s_client and s_server
6421   to support them.
6422
6423   *Steve Henson*
6424
6425 * New function SSL_certs_clear() to delete all references to certificates
6426   from an SSL structure. Before this once a certificate had been added
6427   it couldn't be removed.
6428
6429   *Steve Henson*
6430
6431 * Integrate hostname, email address and IP address checking with certificate
6432   verification. New verify options supporting checking in openssl utility.
6433
6434   *Steve Henson*
6435
6436 * Fixes and wildcard matching support to hostname and email checking
6437   functions. Add manual page.
6438
6439   *Florian Weimer (Red Hat Product Security Team)*
6440
6441 * New functions to check a hostname email or IP address against a
6442   certificate. Add options x509 utility to print results of checks against
6443   a certificate.
6444
6445   *Steve Henson*
6446
6447 * Fix OCSP checking.
6448
6449   *Rob Stradling <rob.stradling@comodo.com> and Ben Laurie*
6450
6451 * Initial experimental support for explicitly trusted non-root CAs.
6452   OpenSSL still tries to build a complete chain to a root but if an
6453   intermediate CA has a trust setting included that is used. The first
6454   setting is used: whether to trust (e.g., -addtrust option to the x509
6455   utility) or reject.
6456
6457   *Steve Henson*
6458
6459 * Add -trusted_first option which attempts to find certificates in the
6460   trusted store even if an untrusted chain is also supplied.
6461
6462   *Steve Henson*
6463
6464 * MIPS assembly pack updates: support for MIPS32r2 and SmartMIPS ASE,
6465   platform support for Linux and Android.
6466
6467   *Andy Polyakov*
6468
6469 * Support for linux-x32, ILP32 environment in x86_64 framework.
6470
6471   *Andy Polyakov*
6472
6473 * Experimental multi-implementation support for FIPS capable OpenSSL.
6474   When in FIPS mode the approved implementations are used as normal,
6475   when not in FIPS mode the internal unapproved versions are used instead.
6476   This means that the FIPS capable OpenSSL isn't forced to use the
6477   (often lower performance) FIPS implementations outside FIPS mode.
6478
6479   *Steve Henson*
6480
6481 * Transparently support X9.42 DH parameters when calling
6482   PEM_read_bio_DHparameters. This means existing applications can handle
6483   the new parameter format automatically.
6484
6485   *Steve Henson*
6486
6487 * Initial experimental support for X9.42 DH parameter format: mainly
6488   to support use of 'q' parameter for RFC5114 parameters.
6489
6490   *Steve Henson*
6491
6492 * Add DH parameters from RFC5114 including test data to dhtest.
6493
6494   *Steve Henson*
6495
6496 * Support for automatic EC temporary key parameter selection. If enabled
6497   the most preferred EC parameters are automatically used instead of
6498   hardcoded fixed parameters. Now a server just has to call:
6499   SSL_CTX_set_ecdh_auto(ctx, 1) and the server will automatically
6500   support ECDH and use the most appropriate parameters.
6501
6502   *Steve Henson*
6503
6504 * Enhance and tidy EC curve and point format TLS extension code. Use
6505   static structures instead of allocation if default values are used.
6506   New ctrls to set curves we wish to support and to retrieve shared curves.
6507   Print out shared curves in s_server. New options to s_server and s_client
6508   to set list of supported curves.
6509
6510   *Steve Henson*
6511
6512 * New ctrls to retrieve supported signature algorithms and
6513   supported curve values as an array of NIDs. Extend openssl utility
6514   to print out received values.
6515
6516   *Steve Henson*
6517
6518 * Add new APIs EC_curve_nist2nid and EC_curve_nid2nist which convert
6519   between NIDs and the more common NIST names such as "P-256". Enhance
6520   ecparam utility and ECC method to recognise the NIST names for curves.
6521
6522   *Steve Henson*
6523
6524 * Enhance SSL/TLS certificate chain handling to support different
6525   chains for each certificate instead of one chain in the parent SSL_CTX.
6526
6527   *Steve Henson*
6528
6529 * Support for fixed DH ciphersuite client authentication: where both
6530   server and client use DH certificates with common parameters.
6531
6532   *Steve Henson*
6533
6534 * Support for fixed DH ciphersuites: those requiring DH server
6535   certificates.
6536
6537   *Steve Henson*
6538
6539 * New function i2d_re_X509_tbs for re-encoding the TBS portion of
6540   the certificate.
6541   Note: Related 1.0.2-beta specific macros X509_get_cert_info,
6542   X509_CINF_set_modified, X509_CINF_get_issuer, X509_CINF_get_extensions and
6543   X509_CINF_get_signature were reverted post internal team review.
6544
6545OpenSSL 1.0.1
6546-------------
6547
6548### Changes between 1.0.1t and 1.0.1u [22 Sep 2016]
6549
6550 * OCSP Status Request extension unbounded memory growth
6551
6552   A malicious client can send an excessively large OCSP Status Request
6553   extension. If that client continually requests renegotiation, sending a
6554   large OCSP Status Request extension each time, then there will be unbounded
6555   memory growth on the server. This will eventually lead to a Denial Of
6556   Service attack through memory exhaustion. Servers with a default
6557   configuration are vulnerable even if they do not support OCSP. Builds using
6558   the "no-ocsp" build time option are not affected.
6559
6560   This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.)
6561   ([CVE-2016-6304])
6562
6563   *Matt Caswell*
6564
6565 * In order to mitigate the SWEET32 attack, the DES ciphers were moved from
6566   HIGH to MEDIUM.
6567
6568   This issue was reported to OpenSSL Karthikeyan Bhargavan and Gaetan
6569   Leurent (INRIA)
6570   ([CVE-2016-2183])
6571
6572   *Rich Salz*
6573
6574 * OOB write in MDC2_Update()
6575
6576   An overflow can occur in MDC2_Update() either if called directly or
6577   through the EVP_DigestUpdate() function using MDC2. If an attacker
6578   is able to supply very large amounts of input data after a previous
6579   call to EVP_EncryptUpdate() with a partial block then a length check
6580   can overflow resulting in a heap corruption.
6581
6582   The amount of data needed is comparable to SIZE_MAX which is impractical
6583   on most platforms.
6584
6585   This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.)
6586   ([CVE-2016-6303])
6587
6588   *Stephen Henson*
6589
6590 * Malformed SHA512 ticket DoS
6591
6592   If a server uses SHA512 for TLS session ticket HMAC it is vulnerable to a
6593   DoS attack where a malformed ticket will result in an OOB read which will
6594   ultimately crash.
6595
6596   The use of SHA512 in TLS session tickets is comparatively rare as it requires
6597   a custom server callback and ticket lookup mechanism.
6598
6599   This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.)
6600   ([CVE-2016-6302])
6601
6602   *Stephen Henson*
6603
6604 * OOB write in BN_bn2dec()
6605
6606   The function BN_bn2dec() does not check the return value of BN_div_word().
6607   This can cause an OOB write if an application uses this function with an
6608   overly large BIGNUM. This could be a problem if an overly large certificate
6609   or CRL is printed out from an untrusted source. TLS is not affected because
6610   record limits will reject an oversized certificate before it is parsed.
6611
6612   This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.)
6613   ([CVE-2016-2182])
6614
6615   *Stephen Henson*
6616
6617 * OOB read in TS_OBJ_print_bio()
6618
6619   The function TS_OBJ_print_bio() misuses OBJ_obj2txt(): the return value is
6620   the total length the OID text representation would use and not the amount
6621   of data written. This will result in OOB reads when large OIDs are
6622   presented.
6623
6624   This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.)
6625   ([CVE-2016-2180])
6626
6627   *Stephen Henson*
6628
6629 * Pointer arithmetic undefined behaviour
6630
6631   Avoid some undefined pointer arithmetic
6632
6633   A common idiom in the codebase is to check limits in the following manner:
6634   "p + len > limit"
6635
6636   Where "p" points to some malloc'd data of SIZE bytes and
6637   limit == p + SIZE
6638
6639   "len" here could be from some externally supplied data (e.g. from a TLS
6640   message).
6641
6642   The rules of C pointer arithmetic are such that "p + len" is only well
6643   defined where len <= SIZE. Therefore the above idiom is actually
6644   undefined behaviour.
6645
6646   For example this could cause problems if some malloc implementation
6647   provides an address for "p" such that "p + len" actually overflows for
6648   values of len that are too big and therefore p + len < limit.
6649
6650   This issue was reported to OpenSSL by Guido Vranken
6651   ([CVE-2016-2177])
6652
6653   *Matt Caswell*
6654
6655 * Constant time flag not preserved in DSA signing
6656
6657   Operations in the DSA signing algorithm should run in constant time in
6658   order to avoid side channel attacks. A flaw in the OpenSSL DSA
6659   implementation means that a non-constant time codepath is followed for
6660   certain operations. This has been demonstrated through a cache-timing
6661   attack to be sufficient for an attacker to recover the private DSA key.
6662
6663   This issue was reported by César Pereida (Aalto University), Billy Brumley
6664   (Tampere University of Technology), and Yuval Yarom (The University of
6665   Adelaide and NICTA).
6666   ([CVE-2016-2178])
6667
6668   *César Pereida*
6669
6670 * DTLS buffered message DoS
6671
6672   In a DTLS connection where handshake messages are delivered out-of-order
6673   those messages that OpenSSL is not yet ready to process will be buffered
6674   for later use. Under certain circumstances, a flaw in the logic means that
6675   those messages do not get removed from the buffer even though the handshake
6676   has been completed. An attacker could force up to approx. 15 messages to
6677   remain in the buffer when they are no longer required. These messages will
6678   be cleared when the DTLS connection is closed. The default maximum size for
6679   a message is 100k. Therefore the attacker could force an additional 1500k
6680   to be consumed per connection. By opening many simulataneous connections an
6681   attacker could cause a DoS attack through memory exhaustion.
6682
6683   This issue was reported to OpenSSL by Quan Luo.
6684   ([CVE-2016-2179])
6685
6686   *Matt Caswell*
6687
6688 * DTLS replay protection DoS
6689
6690   A flaw in the DTLS replay attack protection mechanism means that records
6691   that arrive for future epochs update the replay protection "window" before
6692   the MAC for the record has been validated. This could be exploited by an
6693   attacker by sending a record for the next epoch (which does not have to
6694   decrypt or have a valid MAC), with a very large sequence number. This means
6695   that all subsequent legitimate packets are dropped causing a denial of
6696   service for a specific DTLS connection.
6697
6698   This issue was reported to OpenSSL by the OCAP audit team.
6699   ([CVE-2016-2181])
6700
6701   *Matt Caswell*
6702
6703 * Certificate message OOB reads
6704
6705   In OpenSSL 1.0.2 and earlier some missing message length checks can result
6706   in OOB reads of up to 2 bytes beyond an allocated buffer. There is a
6707   theoretical DoS risk but this has not been observed in practice on common
6708   platforms.
6709
6710   The messages affected are client certificate, client certificate request
6711   and server certificate. As a result the attack can only be performed
6712   against a client or a server which enables client authentication.
6713
6714   This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.)
6715   ([CVE-2016-6306])
6716
6717   *Stephen Henson*
6718
6719### Changes between 1.0.1s and 1.0.1t [3 May 2016]
6720
6721 * Prevent padding oracle in AES-NI CBC MAC check
6722
6723   A MITM attacker can use a padding oracle attack to decrypt traffic
6724   when the connection uses an AES CBC cipher and the server support
6725   AES-NI.
6726
6727   This issue was introduced as part of the fix for Lucky 13 padding
6728   attack ([CVE-2013-0169]). The padding check was rewritten to be in
6729   constant time by making sure that always the same bytes are read and
6730   compared against either the MAC or padding bytes. But it no longer
6731   checked that there was enough data to have both the MAC and padding
6732   bytes.
6733
6734   This issue was reported by Juraj Somorovsky using TLS-Attacker.
6735   ([CVE-2016-2107])
6736
6737   *Kurt Roeckx*
6738
6739 * Fix EVP_EncodeUpdate overflow
6740
6741   An overflow can occur in the EVP_EncodeUpdate() function which is used for
6742   Base64 encoding of binary data. If an attacker is able to supply very large
6743   amounts of input data then a length check can overflow resulting in a heap
6744   corruption.
6745
6746   Internally to OpenSSL the EVP_EncodeUpdate() function is primarly used by
6747   the `PEM_write_bio*` family of functions. These are mainly used within the
6748   OpenSSL command line applications, so any application which processes data
6749   from an untrusted source and outputs it as a PEM file should be considered
6750   vulnerable to this issue. User applications that call these APIs directly
6751   with large amounts of untrusted data may also be vulnerable.
6752
6753   This issue was reported by Guido Vranken.
6754   ([CVE-2016-2105])
6755
6756   *Matt Caswell*
6757
6758 * Fix EVP_EncryptUpdate overflow
6759
6760   An overflow can occur in the EVP_EncryptUpdate() function. If an attacker
6761   is able to supply very large amounts of input data after a previous call to
6762   EVP_EncryptUpdate() with a partial block then a length check can overflow
6763   resulting in a heap corruption. Following an analysis of all OpenSSL
6764   internal usage of the EVP_EncryptUpdate() function all usage is one of two
6765   forms. The first form is where the EVP_EncryptUpdate() call is known to be
6766   the first called function after an EVP_EncryptInit(), and therefore that
6767   specific call must be safe. The second form is where the length passed to
6768   EVP_EncryptUpdate() can be seen from the code to be some small value and
6769   therefore there is no possibility of an overflow. Since all instances are
6770   one of these two forms, it is believed that there can be no overflows in
6771   internal code due to this problem. It should be noted that
6772   EVP_DecryptUpdate() can call EVP_EncryptUpdate() in certain code paths.
6773   Also EVP_CipherUpdate() is a synonym for EVP_EncryptUpdate(). All instances
6774   of these calls have also been analysed too and it is believed there are no
6775   instances in internal usage where an overflow could occur.
6776
6777   This issue was reported by Guido Vranken.
6778   ([CVE-2016-2106])
6779
6780   *Matt Caswell*
6781
6782 * Prevent ASN.1 BIO excessive memory allocation
6783
6784   When ASN.1 data is read from a BIO using functions such as d2i_CMS_bio()
6785   a short invalid encoding can casuse allocation of large amounts of memory
6786   potentially consuming excessive resources or exhausting memory.
6787
6788   Any application parsing untrusted data through d2i BIO functions is
6789   affected. The memory based functions such as d2i_X509() are *not* affected.
6790   Since the memory based functions are used by the TLS library, TLS
6791   applications are not affected.
6792
6793   This issue was reported by Brian Carpenter.
6794   ([CVE-2016-2109])
6795
6796   *Stephen Henson*
6797
6798 * EBCDIC overread
6799
6800   ASN1 Strings that are over 1024 bytes can cause an overread in applications
6801   using the X509_NAME_oneline() function on EBCDIC systems. This could result
6802   in arbitrary stack data being returned in the buffer.
6803
6804   This issue was reported by Guido Vranken.
6805   ([CVE-2016-2176])
6806
6807   *Matt Caswell*
6808
6809 * Modify behavior of ALPN to invoke callback after SNI/servername
6810   callback, such that updates to the SSL_CTX affect ALPN.
6811
6812   *Todd Short*
6813
6814 * Remove LOW from the DEFAULT cipher list.  This removes singles DES from the
6815   default.
6816
6817   *Kurt Roeckx*
6818
6819 * Only remove the SSLv2 methods with the no-ssl2-method option. When the
6820   methods are enabled and ssl2 is disabled the methods return NULL.
6821
6822   *Kurt Roeckx*
6823
6824### Changes between 1.0.1r and 1.0.1s [1 Mar 2016]
6825
6826* Disable weak ciphers in SSLv3 and up in default builds of OpenSSL.
6827  Builds that are not configured with "enable-weak-ssl-ciphers" will not
6828  provide any "EXPORT" or "LOW" strength ciphers.
6829
6830  *Viktor Dukhovni*
6831
6832* Disable SSLv2 default build, default negotiation and weak ciphers.  SSLv2
6833  is by default disabled at build-time.  Builds that are not configured with
6834  "enable-ssl2" will not support SSLv2.  Even if "enable-ssl2" is used,
6835  users who want to negotiate SSLv2 via the version-flexible SSLv23_method()
6836  will need to explicitly call either of:
6837
6838      SSL_CTX_clear_options(ctx, SSL_OP_NO_SSLv2);
6839  or
6840      SSL_clear_options(ssl, SSL_OP_NO_SSLv2);
6841
6842  as appropriate.  Even if either of those is used, or the application
6843  explicitly uses the version-specific SSLv2_method() or its client and
6844  server variants, SSLv2 ciphers vulnerable to exhaustive search key
6845  recovery have been removed.  Specifically, the SSLv2 40-bit EXPORT
6846  ciphers, and SSLv2 56-bit DES are no longer available.
6847  ([CVE-2016-0800])
6848
6849  *Viktor Dukhovni*
6850
6851 * Fix a double-free in DSA code
6852
6853   A double free bug was discovered when OpenSSL parses malformed DSA private
6854   keys and could lead to a DoS attack or memory corruption for applications
6855   that receive DSA private keys from untrusted sources.  This scenario is
6856   considered rare.
6857
6858   This issue was reported to OpenSSL by Adam Langley(Google/BoringSSL) using
6859   libFuzzer.
6860   ([CVE-2016-0705])
6861
6862   *Stephen Henson*
6863
6864 * Disable SRP fake user seed to address a server memory leak.
6865
6866   Add a new method SRP_VBASE_get1_by_user that handles the seed properly.
6867
6868   SRP_VBASE_get_by_user had inconsistent memory management behaviour.
6869   In order to fix an unavoidable memory leak, SRP_VBASE_get_by_user
6870   was changed to ignore the "fake user" SRP seed, even if the seed
6871   is configured.
6872
6873   Users should use SRP_VBASE_get1_by_user instead. Note that in
6874   SRP_VBASE_get1_by_user, caller must free the returned value. Note
6875   also that even though configuring the SRP seed attempts to hide
6876   invalid usernames by continuing the handshake with fake
6877   credentials, this behaviour is not constant time and no strong
6878   guarantees are made that the handshake is indistinguishable from
6879   that of a valid user.
6880   ([CVE-2016-0798])
6881
6882   *Emilia Käsper*
6883
6884 * Fix BN_hex2bn/BN_dec2bn NULL pointer deref/heap corruption
6885
6886   In the BN_hex2bn function the number of hex digits is calculated using an
6887   int value `i`. Later `bn_expand` is called with a value of `i * 4`. For
6888   large values of `i` this can result in `bn_expand` not allocating any
6889   memory because `i * 4` is negative. This can leave the internal BIGNUM data
6890   field as NULL leading to a subsequent NULL ptr deref. For very large values
6891   of `i`, the calculation `i * 4` could be a positive value smaller than `i`.
6892   In this case memory is allocated to the internal BIGNUM data field, but it
6893   is insufficiently sized leading to heap corruption. A similar issue exists
6894   in BN_dec2bn. This could have security consequences if BN_hex2bn/BN_dec2bn
6895   is ever called by user applications with very large untrusted hex/dec data.
6896   This is anticipated to be a rare occurrence.
6897
6898   All OpenSSL internal usage of these functions use data that is not expected
6899   to be untrusted, e.g. config file data or application command line
6900   arguments. If user developed applications generate config file data based
6901   on untrusted data then it is possible that this could also lead to security
6902   consequences. This is also anticipated to be rare.
6903
6904   This issue was reported to OpenSSL by Guido Vranken.
6905   ([CVE-2016-0797])
6906
6907   *Matt Caswell*
6908
6909 * Fix memory issues in `BIO_*printf` functions
6910
6911   The internal `fmtstr` function used in processing a "%s" format string in
6912   the `BIO_*printf` functions could overflow while calculating the length of a
6913   string and cause an OOB read when printing very long strings.
6914
6915   Additionally the internal `doapr_outch` function can attempt to write to an
6916   OOB memory location (at an offset from the NULL pointer) in the event of a
6917   memory allocation failure. In 1.0.2 and below this could be caused where
6918   the size of a buffer to be allocated is greater than INT_MAX. E.g. this
6919   could be in processing a very long "%s" format string. Memory leaks can
6920   also occur.
6921
6922   The first issue may mask the second issue dependent on compiler behaviour.
6923   These problems could enable attacks where large amounts of untrusted data
6924   is passed to the `BIO_*printf` functions. If applications use these functions
6925   in this way then they could be vulnerable. OpenSSL itself uses these
6926   functions when printing out human-readable dumps of ASN.1 data. Therefore
6927   applications that print this data could be vulnerable if the data is from
6928   untrusted sources. OpenSSL command line applications could also be
6929   vulnerable where they print out ASN.1 data, or if untrusted data is passed
6930   as command line arguments.
6931
6932   Libssl is not considered directly vulnerable. Additionally certificates etc
6933   received via remote connections via libssl are also unlikely to be able to
6934   trigger these issues because of message size limits enforced within libssl.
6935
6936   This issue was reported to OpenSSL Guido Vranken.
6937   ([CVE-2016-0799])
6938
6939   *Matt Caswell*
6940
6941 * Side channel attack on modular exponentiation
6942
6943   A side-channel attack was found which makes use of cache-bank conflicts on
6944   the Intel Sandy-Bridge microarchitecture which could lead to the recovery
6945   of RSA keys.  The ability to exploit this issue is limited as it relies on
6946   an attacker who has control of code in a thread running on the same
6947   hyper-threaded core as the victim thread which is performing decryptions.
6948
6949   This issue was reported to OpenSSL by Yuval Yarom, The University of
6950   Adelaide and NICTA, Daniel Genkin, Technion and Tel Aviv University, and
6951   Nadia Heninger, University of Pennsylvania with more information at
6952   <http://cachebleed.info>.
6953   ([CVE-2016-0702])
6954
6955   *Andy Polyakov*
6956
6957 * Change the req command to generate a 2048-bit RSA/DSA key by default,
6958   if no keysize is specified with default_bits. This fixes an
6959   omission in an earlier change that changed all RSA/DSA key generation
6960   commands to use 2048 bits by default.
6961
6962   *Emilia Käsper*
6963
6964### Changes between 1.0.1q and 1.0.1r [28 Jan 2016]
6965
6966 * Protection for DH small subgroup attacks
6967
6968   As a precautionary measure the SSL_OP_SINGLE_DH_USE option has been
6969   switched on by default and cannot be disabled. This could have some
6970   performance impact.
6971
6972   *Matt Caswell*
6973
6974 * SSLv2 doesn't block disabled ciphers
6975
6976   A malicious client can negotiate SSLv2 ciphers that have been disabled on
6977   the server and complete SSLv2 handshakes even if all SSLv2 ciphers have
6978   been disabled, provided that the SSLv2 protocol was not also disabled via
6979   SSL_OP_NO_SSLv2.
6980
6981   This issue was reported to OpenSSL on 26th December 2015 by Nimrod Aviram
6982   and Sebastian Schinzel.
6983   ([CVE-2015-3197])
6984
6985   *Viktor Dukhovni*
6986
6987 * Reject DH handshakes with parameters shorter than 1024 bits.
6988
6989   *Kurt Roeckx*
6990
6991### Changes between 1.0.1p and 1.0.1q [3 Dec 2015]
6992
6993 * Certificate verify crash with missing PSS parameter
6994
6995   The signature verification routines will crash with a NULL pointer
6996   dereference if presented with an ASN.1 signature using the RSA PSS
6997   algorithm and absent mask generation function parameter. Since these
6998   routines are used to verify certificate signature algorithms this can be
6999   used to crash any certificate verification operation and exploited in a
7000   DoS attack. Any application which performs certificate verification is
7001   vulnerable including OpenSSL clients and servers which enable client
7002   authentication.
7003
7004   This issue was reported to OpenSSL by Loïc Jonas Etienne (Qnective AG).
7005   ([CVE-2015-3194])
7006
7007   *Stephen Henson*
7008
7009 * X509_ATTRIBUTE memory leak
7010
7011   When presented with a malformed X509_ATTRIBUTE structure OpenSSL will leak
7012   memory. This structure is used by the PKCS#7 and CMS routines so any
7013   application which reads PKCS#7 or CMS data from untrusted sources is
7014   affected. SSL/TLS is not affected.
7015
7016   This issue was reported to OpenSSL by Adam Langley (Google/BoringSSL) using
7017   libFuzzer.
7018   ([CVE-2015-3195])
7019
7020   *Stephen Henson*
7021
7022 * Rewrite EVP_DecodeUpdate (base64 decoding) to fix several bugs.
7023   This changes the decoding behaviour for some invalid messages,
7024   though the change is mostly in the more lenient direction, and
7025   legacy behaviour is preserved as much as possible.
7026
7027   *Emilia Käsper*
7028
7029 * In DSA_generate_parameters_ex, if the provided seed is too short,
7030   use a random seed, as already documented.
7031
7032   *Rich Salz and Ismo Puustinen <ismo.puustinen@intel.com>*
7033
7034### Changes between 1.0.1o and 1.0.1p [9 Jul 2015]
7035
7036 * Alternate chains certificate forgery
7037
7038   During certificate verfification, OpenSSL will attempt to find an
7039   alternative certificate chain if the first attempt to build such a chain
7040   fails. An error in the implementation of this logic can mean that an
7041   attacker could cause certain checks on untrusted certificates to be
7042   bypassed, such as the CA flag, enabling them to use a valid leaf
7043   certificate to act as a CA and "issue" an invalid certificate.
7044
7045   This issue was reported to OpenSSL by Adam Langley/David Benjamin
7046   (Google/BoringSSL).
7047   ([CVE-2015-1793])
7048
7049   *Matt Caswell*
7050
7051 * Race condition handling PSK identify hint
7052
7053   If PSK identity hints are received by a multi-threaded client then
7054   the values are wrongly updated in the parent SSL_CTX structure. This can
7055   result in a race condition potentially leading to a double free of the
7056   identify hint data.
7057   ([CVE-2015-3196])
7058
7059   *Stephen Henson*
7060
7061### Changes between 1.0.1n and 1.0.1o [12 Jun 2015]
7062
7063 * Fix HMAC ABI incompatibility. The previous version introduced an ABI
7064   incompatibility in the handling of HMAC. The previous ABI has now been
7065   restored.
7066
7067### Changes between 1.0.1m and 1.0.1n [11 Jun 2015]
7068
7069 * Malformed ECParameters causes infinite loop
7070
7071   When processing an ECParameters structure OpenSSL enters an infinite loop
7072   if the curve specified is over a specially malformed binary polynomial
7073   field.
7074
7075   This can be used to perform denial of service against any
7076   system which processes public keys, certificate requests or
7077   certificates.  This includes TLS clients and TLS servers with
7078   client authentication enabled.
7079
7080   This issue was reported to OpenSSL by Joseph Barr-Pixton.
7081   ([CVE-2015-1788])
7082
7083   *Andy Polyakov*
7084
7085 * Exploitable out-of-bounds read in X509_cmp_time
7086
7087   X509_cmp_time does not properly check the length of the ASN1_TIME
7088   string and can read a few bytes out of bounds. In addition,
7089   X509_cmp_time accepts an arbitrary number of fractional seconds in the
7090   time string.
7091
7092   An attacker can use this to craft malformed certificates and CRLs of
7093   various sizes and potentially cause a segmentation fault, resulting in
7094   a DoS on applications that verify certificates or CRLs. TLS clients
7095   that verify CRLs are affected. TLS clients and servers with client
7096   authentication enabled may be affected if they use custom verification
7097   callbacks.
7098
7099   This issue was reported to OpenSSL by Robert Swiecki (Google), and
7100   independently by Hanno Böck.
7101   ([CVE-2015-1789])
7102
7103   *Emilia Käsper*
7104
7105 * PKCS7 crash with missing EnvelopedContent
7106
7107   The PKCS#7 parsing code does not handle missing inner EncryptedContent
7108   correctly. An attacker can craft malformed ASN.1-encoded PKCS#7 blobs
7109   with missing content and trigger a NULL pointer dereference on parsing.
7110
7111   Applications that decrypt PKCS#7 data or otherwise parse PKCS#7
7112   structures from untrusted sources are affected. OpenSSL clients and
7113   servers are not affected.
7114
7115   This issue was reported to OpenSSL by Michal Zalewski (Google).
7116   ([CVE-2015-1790])
7117
7118   *Emilia Käsper*
7119
7120 * CMS verify infinite loop with unknown hash function
7121
7122   When verifying a signedData message the CMS code can enter an infinite loop
7123   if presented with an unknown hash function OID. This can be used to perform
7124   denial of service against any system which verifies signedData messages using
7125   the CMS code.
7126   This issue was reported to OpenSSL by Johannes Bauer.
7127   ([CVE-2015-1792])
7128
7129   *Stephen Henson*
7130
7131 * Race condition handling NewSessionTicket
7132
7133   If a NewSessionTicket is received by a multi-threaded client when attempting to
7134   reuse a previous ticket then a race condition can occur potentially leading to
7135   a double free of the ticket data.
7136   ([CVE-2015-1791])
7137
7138   *Matt Caswell*
7139
7140 * Reject DH handshakes with parameters shorter than 768 bits.
7141
7142   *Kurt Roeckx and Emilia Kasper*
7143
7144 * dhparam: generate 2048-bit parameters by default.
7145
7146   *Kurt Roeckx and Emilia Kasper*
7147
7148### Changes between 1.0.1l and 1.0.1m [19 Mar 2015]
7149
7150 * Segmentation fault in ASN1_TYPE_cmp fix
7151
7152   The function ASN1_TYPE_cmp will crash with an invalid read if an attempt is
7153   made to compare ASN.1 boolean types. Since ASN1_TYPE_cmp is used to check
7154   certificate signature algorithm consistency this can be used to crash any
7155   certificate verification operation and exploited in a DoS attack. Any
7156   application which performs certificate verification is vulnerable including
7157   OpenSSL clients and servers which enable client authentication.
7158   ([CVE-2015-0286])
7159
7160   *Stephen Henson*
7161
7162 * ASN.1 structure reuse memory corruption fix
7163
7164   Reusing a structure in ASN.1 parsing may allow an attacker to cause
7165   memory corruption via an invalid write. Such reuse is and has been
7166   strongly discouraged and is believed to be rare.
7167
7168   Applications that parse structures containing CHOICE or ANY DEFINED BY
7169   components may be affected. Certificate parsing (d2i_X509 and related
7170   functions) are however not affected. OpenSSL clients and servers are
7171   not affected.
7172   ([CVE-2015-0287])
7173
7174   *Stephen Henson*
7175
7176 * PKCS7 NULL pointer dereferences fix
7177
7178   The PKCS#7 parsing code does not handle missing outer ContentInfo
7179   correctly. An attacker can craft malformed ASN.1-encoded PKCS#7 blobs with
7180   missing content and trigger a NULL pointer dereference on parsing.
7181
7182   Applications that verify PKCS#7 signatures, decrypt PKCS#7 data or
7183   otherwise parse PKCS#7 structures from untrusted sources are
7184   affected. OpenSSL clients and servers are not affected.
7185
7186   This issue was reported to OpenSSL by Michal Zalewski (Google).
7187   ([CVE-2015-0289])
7188
7189   *Emilia Käsper*
7190
7191 * DoS via reachable assert in SSLv2 servers fix
7192
7193   A malicious client can trigger an OPENSSL_assert (i.e., an abort) in
7194   servers that both support SSLv2 and enable export cipher suites by sending
7195   a specially crafted SSLv2 CLIENT-MASTER-KEY message.
7196
7197   This issue was discovered by Sean Burford (Google) and Emilia Käsper
7198   (OpenSSL development team).
7199   ([CVE-2015-0293])
7200
7201   *Emilia Käsper*
7202
7203 * Use After Free following d2i_ECPrivatekey error fix
7204
7205   A malformed EC private key file consumed via the d2i_ECPrivateKey function
7206   could cause a use after free condition. This, in turn, could cause a double
7207   free in several private key parsing functions (such as d2i_PrivateKey
7208   or EVP_PKCS82PKEY) and could lead to a DoS attack or memory corruption
7209   for applications that receive EC private keys from untrusted
7210   sources. This scenario is considered rare.
7211
7212   This issue was discovered by the BoringSSL project and fixed in their
7213   commit 517073cd4b.
7214   ([CVE-2015-0209])
7215
7216   *Matt Caswell*
7217
7218 * X509_to_X509_REQ NULL pointer deref fix
7219
7220   The function X509_to_X509_REQ will crash with a NULL pointer dereference if
7221   the certificate key is invalid. This function is rarely used in practice.
7222
7223   This issue was discovered by Brian Carpenter.
7224   ([CVE-2015-0288])
7225
7226   *Stephen Henson*
7227
7228 * Removed the export ciphers from the DEFAULT ciphers
7229
7230   *Kurt Roeckx*
7231
7232### Changes between 1.0.1k and 1.0.1l [15 Jan 2015]
7233
7234 * Build fixes for the Windows and OpenVMS platforms
7235
7236   *Matt Caswell and Richard Levitte*
7237
7238### Changes between 1.0.1j and 1.0.1k [8 Jan 2015]
7239
7240 * Fix DTLS segmentation fault in dtls1_get_record. A carefully crafted DTLS
7241   message can cause a segmentation fault in OpenSSL due to a NULL pointer
7242   dereference. This could lead to a Denial Of Service attack. Thanks to
7243   Markus Stenberg of Cisco Systems, Inc. for reporting this issue.
7244   ([CVE-2014-3571])
7245
7246   *Steve Henson*
7247
7248 * Fix DTLS memory leak in dtls1_buffer_record. A memory leak can occur in the
7249   dtls1_buffer_record function under certain conditions. In particular this
7250   could occur if an attacker sent repeated DTLS records with the same
7251   sequence number but for the next epoch. The memory leak could be exploited
7252   by an attacker in a Denial of Service attack through memory exhaustion.
7253   Thanks to Chris Mueller for reporting this issue.
7254   ([CVE-2015-0206])
7255
7256   *Matt Caswell*
7257
7258 * Fix issue where no-ssl3 configuration sets method to NULL. When openssl is
7259   built with the no-ssl3 option and a SSL v3 ClientHello is received the ssl
7260   method would be set to NULL which could later result in a NULL pointer
7261   dereference. Thanks to Frank Schmirler for reporting this issue.
7262   ([CVE-2014-3569])
7263
7264   *Kurt Roeckx*
7265
7266 * Abort handshake if server key exchange message is omitted for ephemeral
7267   ECDH ciphersuites.
7268
7269   Thanks to Karthikeyan Bhargavan of the PROSECCO team at INRIA for
7270   reporting this issue.
7271   ([CVE-2014-3572])
7272
7273   *Steve Henson*
7274
7275 * Remove non-export ephemeral RSA code on client and server. This code
7276   violated the TLS standard by allowing the use of temporary RSA keys in
7277   non-export ciphersuites and could be used by a server to effectively
7278   downgrade the RSA key length used to a value smaller than the server
7279   certificate. Thanks for Karthikeyan Bhargavan of the PROSECCO team at
7280   INRIA or reporting this issue.
7281   ([CVE-2015-0204])
7282
7283   *Steve Henson*
7284
7285 * Fixed issue where DH client certificates are accepted without verification.
7286   An OpenSSL server will accept a DH certificate for client authentication
7287   without the certificate verify message. This effectively allows a client to
7288   authenticate without the use of a private key. This only affects servers
7289   which trust a client certificate authority which issues certificates
7290   containing DH keys: these are extremely rare and hardly ever encountered.
7291   Thanks for Karthikeyan Bhargavan of the PROSECCO team at INRIA or reporting
7292   this issue.
7293   ([CVE-2015-0205])
7294
7295   *Steve Henson*
7296
7297 * Ensure that the session ID context of an SSL is updated when its
7298   SSL_CTX is updated via SSL_set_SSL_CTX.
7299
7300   The session ID context is typically set from the parent SSL_CTX,
7301   and can vary with the CTX.
7302
7303   *Adam Langley*
7304
7305 * Fix various certificate fingerprint issues.
7306
7307   By using non-DER or invalid encodings outside the signed portion of a
7308   certificate the fingerprint can be changed without breaking the signature.
7309   Although no details of the signed portion of the certificate can be changed
7310   this can cause problems with some applications: e.g. those using the
7311   certificate fingerprint for blacklists.
7312
7313   1. Reject signatures with non zero unused bits.
7314
7315   If the BIT STRING containing the signature has non zero unused bits reject
7316   the signature. All current signature algorithms require zero unused bits.
7317
7318   2. Check certificate algorithm consistency.
7319
7320   Check the AlgorithmIdentifier inside TBS matches the one in the
7321   certificate signature. NB: this will result in signature failure
7322   errors for some broken certificates.
7323
7324   Thanks to Konrad Kraszewski from Google for reporting this issue.
7325
7326   3. Check DSA/ECDSA signatures use DER.
7327
7328   Re-encode DSA/ECDSA signatures and compare with the original received
7329   signature. Return an error if there is a mismatch.
7330
7331   This will reject various cases including garbage after signature
7332   (thanks to Antti Karjalainen and Tuomo Untinen from the Codenomicon CROSS
7333   program for discovering this case) and use of BER or invalid ASN.1 INTEGERs
7334   (negative or with leading zeroes).
7335
7336   Further analysis was conducted and fixes were developed by Stephen Henson
7337   of the OpenSSL core team.
7338
7339   ([CVE-2014-8275])
7340
7341   *Steve Henson*
7342
7343 * Correct Bignum squaring. Bignum squaring (BN_sqr) may produce incorrect
7344   results on some platforms, including x86_64. This bug occurs at random
7345   with a very low probability, and is not known to be exploitable in any
7346   way, though its exact impact is difficult to determine. Thanks to Pieter
7347   Wuille (Blockstream) who reported this issue and also suggested an initial
7348   fix. Further analysis was conducted by the OpenSSL development team and
7349   Adam Langley of Google. The final fix was developed by Andy Polyakov of
7350   the OpenSSL core team.
7351   ([CVE-2014-3570])
7352
7353   *Andy Polyakov*
7354
7355 * Do not resume sessions on the server if the negotiated protocol
7356   version does not match the session's version. Resuming with a different
7357   version, while not strictly forbidden by the RFC, is of questionable
7358   sanity and breaks all known clients.
7359
7360   *David Benjamin, Emilia Käsper*
7361
7362 * Tighten handling of the ChangeCipherSpec (CCS) message: reject
7363   early CCS messages during renegotiation. (Note that because
7364   renegotiation is encrypted, this early CCS was not exploitable.)
7365
7366   *Emilia Käsper*
7367
7368 * Tighten client-side session ticket handling during renegotiation:
7369   ensure that the client only accepts a session ticket if the server sends
7370   the extension anew in the ServerHello. Previously, a TLS client would
7371   reuse the old extension state and thus accept a session ticket if one was
7372   announced in the initial ServerHello.
7373
7374   Similarly, ensure that the client requires a session ticket if one
7375   was advertised in the ServerHello. Previously, a TLS client would
7376   ignore a missing NewSessionTicket message.
7377
7378   *Emilia Käsper*
7379
7380### Changes between 1.0.1i and 1.0.1j [15 Oct 2014]
7381
7382 * SRTP Memory Leak.
7383
7384   A flaw in the DTLS SRTP extension parsing code allows an attacker, who
7385   sends a carefully crafted handshake message, to cause OpenSSL to fail
7386   to free up to 64k of memory causing a memory leak. This could be
7387   exploited in a Denial Of Service attack. This issue affects OpenSSL
7388   1.0.1 server implementations for both SSL/TLS and DTLS regardless of
7389   whether SRTP is used or configured. Implementations of OpenSSL that
7390   have been compiled with OPENSSL_NO_SRTP defined are not affected.
7391
7392   The fix was developed by the OpenSSL team.
7393   ([CVE-2014-3513])
7394
7395   *OpenSSL team*
7396
7397 * Session Ticket Memory Leak.
7398
7399   When an OpenSSL SSL/TLS/DTLS server receives a session ticket the
7400   integrity of that ticket is first verified. In the event of a session
7401   ticket integrity check failing, OpenSSL will fail to free memory
7402   causing a memory leak. By sending a large number of invalid session
7403   tickets an attacker could exploit this issue in a Denial Of Service
7404   attack.
7405   ([CVE-2014-3567])
7406
7407   *Steve Henson*
7408
7409 * Build option no-ssl3 is incomplete.
7410
7411   When OpenSSL is configured with "no-ssl3" as a build option, servers
7412   could accept and complete a SSL 3.0 handshake, and clients could be
7413   configured to send them.
7414   ([CVE-2014-3568])
7415
7416   *Akamai and the OpenSSL team*
7417
7418 * Add support for TLS_FALLBACK_SCSV.
7419   Client applications doing fallback retries should call
7420   SSL_set_mode(s, SSL_MODE_SEND_FALLBACK_SCSV).
7421   ([CVE-2014-3566])
7422
7423   *Adam Langley, Bodo Moeller*
7424
7425 * Add additional DigestInfo checks.
7426
7427   Re-encode DigestInto in DER and check against the original when
7428   verifying RSA signature: this will reject any improperly encoded
7429   DigestInfo structures.
7430
7431   Note: this is a precautionary measure and no attacks are currently known.
7432
7433   *Steve Henson*
7434
7435### Changes between 1.0.1h and 1.0.1i [6 Aug 2014]
7436
7437 * Fix SRP buffer overrun vulnerability. Invalid parameters passed to the
7438   SRP code can be overrun an internal buffer. Add sanity check that
7439   g, A, B < N to SRP code.
7440
7441   Thanks to Sean Devlin and Watson Ladd of Cryptography Services, NCC
7442   Group for discovering this issue.
7443   ([CVE-2014-3512])
7444
7445   *Steve Henson*
7446
7447 * A flaw in the OpenSSL SSL/TLS server code causes the server to negotiate
7448   TLS 1.0 instead of higher protocol versions when the ClientHello message
7449   is badly fragmented. This allows a man-in-the-middle attacker to force a
7450   downgrade to TLS 1.0 even if both the server and the client support a
7451   higher protocol version, by modifying the client's TLS records.
7452
7453   Thanks to David Benjamin and Adam Langley (Google) for discovering and
7454   researching this issue.
7455   ([CVE-2014-3511])
7456
7457   *David Benjamin*
7458
7459 * OpenSSL DTLS clients enabling anonymous (EC)DH ciphersuites are subject
7460   to a denial of service attack. A malicious server can crash the client
7461   with a null pointer dereference (read) by specifying an anonymous (EC)DH
7462   ciphersuite and sending carefully crafted handshake messages.
7463
7464   Thanks to Felix Gröbert (Google) for discovering and researching this
7465   issue.
7466   ([CVE-2014-3510])
7467
7468   *Emilia Käsper*
7469
7470 * By sending carefully crafted DTLS packets an attacker could cause openssl
7471   to leak memory. This can be exploited through a Denial of Service attack.
7472   Thanks to Adam Langley for discovering and researching this issue.
7473   ([CVE-2014-3507])
7474
7475   *Adam Langley*
7476
7477 * An attacker can force openssl to consume large amounts of memory whilst
7478   processing DTLS handshake messages. This can be exploited through a
7479   Denial of Service attack.
7480   Thanks to Adam Langley for discovering and researching this issue.
7481   ([CVE-2014-3506])
7482
7483   *Adam Langley*
7484
7485 * An attacker can force an error condition which causes openssl to crash
7486   whilst processing DTLS packets due to memory being freed twice. This
7487   can be exploited through a Denial of Service attack.
7488   Thanks to Adam Langley and Wan-Teh Chang for discovering and researching
7489   this issue.
7490   ([CVE-2014-3505])
7491
7492   *Adam Langley*
7493
7494 * If a multithreaded client connects to a malicious server using a resumed
7495   session and the server sends an ec point format extension it could write
7496   up to 255 bytes to freed memory.
7497
7498   Thanks to Gabor Tyukasz (LogMeIn Inc) for discovering and researching this
7499   issue.
7500   ([CVE-2014-3509])
7501
7502   *Gabor Tyukasz*
7503
7504 * A malicious server can crash an OpenSSL client with a null pointer
7505   dereference (read) by specifying an SRP ciphersuite even though it was not
7506   properly negotiated with the client. This can be exploited through a
7507   Denial of Service attack.
7508
7509   Thanks to Joonas Kuorilehto and Riku Hietamäki (Codenomicon) for
7510   discovering and researching this issue.
7511   ([CVE-2014-5139])
7512
7513   *Steve Henson*
7514
7515 * A flaw in OBJ_obj2txt may cause pretty printing functions such as
7516   X509_name_oneline, X509_name_print_ex et al. to leak some information
7517   from the stack. Applications may be affected if they echo pretty printing
7518   output to the attacker.
7519
7520   Thanks to Ivan Fratric (Google) for discovering this issue.
7521   ([CVE-2014-3508])
7522
7523   *Emilia Käsper, and Steve Henson*
7524
7525 * Fix ec_GFp_simple_points_make_affine (thus, EC_POINTs_mul etc.)
7526   for corner cases. (Certain input points at infinity could lead to
7527   bogus results, with non-infinity inputs mapped to infinity too.)
7528
7529   *Bodo Moeller*
7530
7531### Changes between 1.0.1g and 1.0.1h [5 Jun 2014]
7532
7533 * Fix for SSL/TLS MITM flaw. An attacker using a carefully crafted
7534   handshake can force the use of weak keying material in OpenSSL
7535   SSL/TLS clients and servers.
7536
7537   Thanks to KIKUCHI Masashi (Lepidum Co. Ltd.) for discovering and
7538   researching this issue. ([CVE-2014-0224])
7539
7540   *KIKUCHI Masashi, Steve Henson*
7541
7542 * Fix DTLS recursion flaw. By sending an invalid DTLS handshake to an
7543   OpenSSL DTLS client the code can be made to recurse eventually crashing
7544   in a DoS attack.
7545
7546   Thanks to Imre Rad (Search-Lab Ltd.) for discovering this issue.
7547   ([CVE-2014-0221])
7548
7549   *Imre Rad, Steve Henson*
7550
7551 * Fix DTLS invalid fragment vulnerability. A buffer overrun attack can
7552   be triggered by sending invalid DTLS fragments to an OpenSSL DTLS
7553   client or server. This is potentially exploitable to run arbitrary
7554   code on a vulnerable client or server.
7555
7556   Thanks to Jüri Aedla for reporting this issue. ([CVE-2014-0195])
7557
7558   *Jüri Aedla, Steve Henson*
7559
7560 * Fix bug in TLS code where clients enable anonymous ECDH ciphersuites
7561   are subject to a denial of service attack.
7562
7563   Thanks to Felix Gröbert and Ivan Fratric at Google for discovering
7564   this issue. ([CVE-2014-3470])
7565
7566   *Felix Gröbert, Ivan Fratric, Steve Henson*
7567
7568 * Harmonize version and its documentation. -f flag is used to display
7569   compilation flags.
7570
7571   *mancha <mancha1@zoho.com>*
7572
7573 * Fix eckey_priv_encode so it immediately returns an error upon a failure
7574   in i2d_ECPrivateKey.
7575
7576   *mancha <mancha1@zoho.com>*
7577
7578 * Fix some double frees. These are not thought to be exploitable.
7579
7580   *mancha <mancha1@zoho.com>*
7581
7582### Changes between 1.0.1f and 1.0.1g [7 Apr 2014]
7583
7584 * A missing bounds check in the handling of the TLS heartbeat extension
7585   can be used to reveal up to 64k of memory to a connected client or
7586   server.
7587
7588   Thanks for Neel Mehta of Google Security for discovering this bug and to
7589   Adam Langley <agl@chromium.org> and Bodo Moeller <bmoeller@acm.org> for
7590   preparing the fix ([CVE-2014-0160])
7591
7592   *Adam Langley, Bodo Moeller*
7593
7594 * Fix for the attack described in the paper "Recovering OpenSSL
7595   ECDSA Nonces Using the FLUSH+RELOAD Cache Side-channel Attack"
7596   by Yuval Yarom and Naomi Benger. Details can be obtained from:
7597   <http://eprint.iacr.org/2014/140>
7598
7599   Thanks to Yuval Yarom and Naomi Benger for discovering this
7600   flaw and to Yuval Yarom for supplying a fix ([CVE-2014-0076])
7601
7602   *Yuval Yarom and Naomi Benger*
7603
7604 * TLS pad extension: draft-agl-tls-padding-03
7605
7606   Workaround for the "TLS hang bug" (see FAQ and PR#2771): if the
7607   TLS client Hello record length value would otherwise be > 255 and
7608   less that 512 pad with a dummy extension containing zeroes so it
7609   is at least 512 bytes long.
7610
7611   *Adam Langley, Steve Henson*
7612
7613### Changes between 1.0.1e and 1.0.1f [6 Jan 2014]
7614
7615 * Fix for TLS record tampering bug. A carefully crafted invalid
7616   handshake could crash OpenSSL with a NULL pointer exception.
7617   Thanks to Anton Johansson for reporting this issues.
7618   ([CVE-2013-4353])
7619
7620 * Keep original DTLS digest and encryption contexts in retransmission
7621   structures so we can use the previous session parameters if they need
7622   to be resent. ([CVE-2013-6450])
7623
7624   *Steve Henson*
7625
7626 * Add option SSL_OP_SAFARI_ECDHE_ECDSA_BUG (part of SSL_OP_ALL) which
7627   avoids preferring ECDHE-ECDSA ciphers when the client appears to be
7628   Safari on OS X.  Safari on OS X 10.8..10.8.3 advertises support for
7629   several ECDHE-ECDSA ciphers, but fails to negotiate them.  The bug
7630   is fixed in OS X 10.8.4, but Apple have ruled out both hot fixing
7631   10.8..10.8.3 and forcing users to upgrade to 10.8.4 or newer.
7632
7633   *Rob Stradling, Adam Langley*
7634
7635### Changes between 1.0.1d and 1.0.1e [11 Feb 2013]
7636
7637 * Correct fix for CVE-2013-0169. The original didn't work on AES-NI
7638   supporting platforms or when small records were transferred.
7639
7640   *Andy Polyakov, Steve Henson*
7641
7642### Changes between 1.0.1c and 1.0.1d [5 Feb 2013]
7643
7644 * Make the decoding of SSLv3, TLS and DTLS CBC records constant time.
7645
7646   This addresses the flaw in CBC record processing discovered by
7647   Nadhem Alfardan and Kenny Paterson. Details of this attack can be found
7648   at: <http://www.isg.rhul.ac.uk/tls/>
7649
7650   Thanks go to Nadhem Alfardan and Kenny Paterson of the Information
7651   Security Group at Royal Holloway, University of London
7652   (www.isg.rhul.ac.uk) for discovering this flaw and Adam Langley and
7653   Emilia Käsper for the initial patch.
7654   ([CVE-2013-0169])
7655
7656   *Emilia Käsper, Adam Langley, Ben Laurie, Andy Polyakov, Steve Henson*
7657
7658 * Fix flaw in AESNI handling of TLS 1.2 and 1.1 records for CBC mode
7659   ciphersuites which can be exploited in a denial of service attack.
7660   Thanks go to and to Adam Langley <agl@chromium.org> for discovering
7661   and detecting this bug and to Wolfgang Ettlinger
7662   <wolfgang.ettlinger@gmail.com> for independently discovering this issue.
7663   ([CVE-2012-2686])
7664
7665   *Adam Langley*
7666
7667 * Return an error when checking OCSP signatures when key is NULL.
7668   This fixes a DoS attack. ([CVE-2013-0166])
7669
7670   *Steve Henson*
7671
7672 * Make openssl verify return errors.
7673
7674   *Chris Palmer <palmer@google.com> and Ben Laurie*
7675
7676 * Call OCSP Stapling callback after ciphersuite has been chosen, so
7677   the right response is stapled. Also change SSL_get_certificate()
7678   so it returns the certificate actually sent.
7679   See <http://rt.openssl.org/Ticket/Display.html?id=2836>.
7680
7681   *Rob Stradling <rob.stradling@comodo.com>*
7682
7683 * Fix possible deadlock when decoding public keys.
7684
7685   *Steve Henson*
7686
7687 * Don't use TLS 1.0 record version number in initial client hello
7688   if renegotiating.
7689
7690   *Steve Henson*
7691
7692### Changes between 1.0.1b and 1.0.1c [10 May 2012]
7693
7694 * Sanity check record length before skipping explicit IV in TLS
7695   1.2, 1.1 and DTLS to fix DoS attack.
7696
7697   Thanks to Codenomicon for discovering this issue using Fuzz-o-Matic
7698   fuzzing as a service testing platform.
7699   ([CVE-2012-2333])
7700
7701   *Steve Henson*
7702
7703 * Initialise tkeylen properly when encrypting CMS messages.
7704   Thanks to Solar Designer of Openwall for reporting this issue.
7705
7706   *Steve Henson*
7707
7708 * In FIPS mode don't try to use composite ciphers as they are not
7709   approved.
7710
7711   *Steve Henson*
7712
7713### Changes between 1.0.1a and 1.0.1b [26 Apr 2012]
7714
7715 * OpenSSL 1.0.0 sets SSL_OP_ALL to 0x80000FFFL and OpenSSL 1.0.1 and
7716   1.0.1a set SSL_OP_NO_TLSv1_1 to 0x00000400L which would unfortunately
7717   mean any application compiled against OpenSSL 1.0.0 headers setting
7718   SSL_OP_ALL would also set SSL_OP_NO_TLSv1_1, unintentionally disabling
7719   TLS 1.1 also. Fix this by changing the value of SSL_OP_NO_TLSv1_1 to
7720   0x10000000L Any application which was previously compiled against
7721   OpenSSL 1.0.1 or 1.0.1a headers and which cares about SSL_OP_NO_TLSv1_1
7722   will need to be recompiled as a result. Letting be results in
7723   inability to disable specifically TLS 1.1 and in client context,
7724   in unlike event, limit maximum offered version to TLS 1.0 [see below].
7725
7726   *Steve Henson*
7727
7728 * In order to ensure interoperability SSL_OP_NO_protocolX does not
7729   disable just protocol X, but all protocols above X *if* there are
7730   protocols *below* X still enabled. In more practical terms it means
7731   that if application wants to disable TLS1.0 in favor of TLS1.1 and
7732   above, it's not sufficient to pass `SSL_OP_NO_TLSv1`, one has to pass
7733   `SSL_OP_NO_TLSv1|SSL_OP_NO_SSLv3|SSL_OP_NO_SSLv2`. This applies to
7734   client side.
7735
7736   *Andy Polyakov*
7737
7738### Changes between 1.0.1 and 1.0.1a [19 Apr 2012]
7739
7740 * Check for potentially exploitable overflows in asn1_d2i_read_bio
7741   BUF_mem_grow and BUF_mem_grow_clean. Refuse attempts to shrink buffer
7742   in CRYPTO_realloc_clean.
7743
7744   Thanks to Tavis Ormandy, Google Security Team, for discovering this
7745   issue and to Adam Langley <agl@chromium.org> for fixing it.
7746   ([CVE-2012-2110])
7747
7748   *Adam Langley (Google), Tavis Ormandy, Google Security Team*
7749
7750 * Don't allow TLS 1.2 SHA-256 ciphersuites in TLS 1.0, 1.1 connections.
7751
7752   *Adam Langley*
7753
7754 * Workarounds for some broken servers that "hang" if a client hello
7755   record length exceeds 255 bytes.
7756
7757   1. Do not use record version number > TLS 1.0 in initial client
7758      hello: some (but not all) hanging servers will now work.
7759   2. If we set OPENSSL_MAX_TLS1_2_CIPHER_LENGTH this will truncate
7760      the number of ciphers sent in the client hello. This should be
7761      set to an even number, such as 50, for example by passing:
7762      -DOPENSSL_MAX_TLS1_2_CIPHER_LENGTH=50 to config or Configure.
7763      Most broken servers should now work.
7764   3. If all else fails setting OPENSSL_NO_TLS1_2_CLIENT will disable
7765      TLS 1.2 client support entirely.
7766
7767   *Steve Henson*
7768
7769 * Fix SEGV in Vector Permutation AES module observed in OpenSSH.
7770
7771   *Andy Polyakov*
7772
7773### Changes between 1.0.0h and 1.0.1  [14 Mar 2012]
7774
7775 * Add compatibility with old MDC2 signatures which use an ASN1 OCTET
7776   STRING form instead of a DigestInfo.
7777
7778   *Steve Henson*
7779
7780 * The format used for MDC2 RSA signatures is inconsistent between EVP
7781   and the RSA_sign/RSA_verify functions. This was made more apparent when
7782   OpenSSL used RSA_sign/RSA_verify for some RSA signatures in particular
7783   those which went through EVP_PKEY_METHOD in 1.0.0 and later. Detect
7784   the correct format in RSA_verify so both forms transparently work.
7785
7786   *Steve Henson*
7787
7788 * Some servers which support TLS 1.0 can choke if we initially indicate
7789   support for TLS 1.2 and later renegotiate using TLS 1.0 in the RSA
7790   encrypted premaster secret. As a workaround use the maximum permitted
7791   client version in client hello, this should keep such servers happy
7792   and still work with previous versions of OpenSSL.
7793
7794   *Steve Henson*
7795
7796 * Add support for TLS/DTLS heartbeats.
7797
7798   *Robin Seggelmann <seggelmann@fh-muenster.de>*
7799
7800 * Add support for SCTP.
7801
7802   *Robin Seggelmann <seggelmann@fh-muenster.de>*
7803
7804 * Improved PRNG seeding for VOS.
7805
7806   *Paul Green <Paul.Green@stratus.com>*
7807
7808 * Extensive assembler packs updates, most notably:
7809
7810   - x86[_64]:     AES-NI, PCLMULQDQ, RDRAND support;
7811   - x86[_64]:     SSSE3 support (SHA1, vector-permutation AES);
7812   - x86_64:       bit-sliced AES implementation;
7813   - ARM:          NEON support, contemporary platforms optimizations;
7814   - s390x:        z196 support;
7815   - `*`:            GHASH and GF(2^m) multiplication implementations;
7816
7817   *Andy Polyakov*
7818
7819 * Make TLS-SRP code conformant with RFC 5054 API cleanup
7820   (removal of unnecessary code)
7821
7822   *Peter Sylvester <peter.sylvester@edelweb.fr>*
7823
7824 * Add TLS key material exporter from RFC 5705.
7825
7826   *Eric Rescorla*
7827
7828 * Add DTLS-SRTP negotiation from RFC 5764.
7829
7830   *Eric Rescorla*
7831
7832 * Add Next Protocol Negotiation,
7833   <http://tools.ietf.org/html/draft-agl-tls-nextprotoneg-00>. Can be
7834   disabled with a no-npn flag to config or Configure. Code donated
7835   by Google.
7836
7837   *Adam Langley <agl@google.com> and Ben Laurie*
7838
7839 * Add optional 64-bit optimized implementations of elliptic curves NIST-P224,
7840   NIST-P256, NIST-P521, with constant-time single point multiplication on
7841   typical inputs. Compiler support for the nonstandard type `__uint128_t` is
7842   required to use this (present in gcc 4.4 and later, for 64-bit builds).
7843   Code made available under Apache License version 2.0.
7844
7845   Specify "enable-ec_nistp_64_gcc_128" on the Configure (or config) command
7846   line to include this in your build of OpenSSL, and run "make depend" (or
7847   "make update"). This enables the following EC_METHODs:
7848
7849           EC_GFp_nistp224_method()
7850           EC_GFp_nistp256_method()
7851           EC_GFp_nistp521_method()
7852
7853   EC_GROUP_new_by_curve_name() will automatically use these (while
7854   EC_GROUP_new_curve_GFp() currently prefers the more flexible
7855   implementations).
7856
7857   *Emilia Käsper, Adam Langley, Bodo Moeller (Google)*
7858
7859 * Use type ossl_ssize_t instead of ssize_t which isn't available on
7860   all platforms. Move ssize_t definition from e_os.h to the public
7861   header file e_os2.h as it now appears in public header file cms.h
7862
7863   *Steve Henson*
7864
7865 * New -sigopt option to the ca, req and x509 utilities. Additional
7866   signature parameters can be passed using this option and in
7867   particular PSS.
7868
7869   *Steve Henson*
7870
7871 * Add RSA PSS signing function. This will generate and set the
7872   appropriate AlgorithmIdentifiers for PSS based on those in the
7873   corresponding EVP_MD_CTX structure. No application support yet.
7874
7875   *Steve Henson*
7876
7877 * Support for companion algorithm specific ASN1 signing routines.
7878   New function ASN1_item_sign_ctx() signs a pre-initialised
7879   EVP_MD_CTX structure and sets AlgorithmIdentifiers based on
7880   the appropriate parameters.
7881
7882   *Steve Henson*
7883
7884 * Add new algorithm specific ASN1 verification initialisation function
7885   to EVP_PKEY_ASN1_METHOD: this is not in EVP_PKEY_METHOD since the ASN1
7886   handling will be the same no matter what EVP_PKEY_METHOD is used.
7887   Add a PSS handler to support verification of PSS signatures: checked
7888   against a number of sample certificates.
7889
7890   *Steve Henson*
7891
7892 * Add signature printing for PSS. Add PSS OIDs.
7893
7894   *Steve Henson, Martin Kaiser <lists@kaiser.cx>*
7895
7896 * Add algorithm specific signature printing. An individual ASN1 method
7897   can now print out signatures instead of the standard hex dump.
7898
7899   More complex signatures (e.g. PSS) can print out more meaningful
7900   information. Include DSA version that prints out the signature
7901   parameters r, s.
7902
7903   *Steve Henson*
7904
7905 * Password based recipient info support for CMS library: implementing
7906   RFC3211.
7907
7908   *Steve Henson*
7909
7910 * Split password based encryption into PBES2 and PBKDF2 functions. This
7911   neatly separates the code into cipher and PBE sections and is required
7912   for some algorithms that split PBES2 into separate pieces (such as
7913   password based CMS).
7914
7915   *Steve Henson*
7916
7917 * Session-handling fixes:
7918   - Fix handling of connections that are resuming with a session ID,
7919     but also support Session Tickets.
7920   - Fix a bug that suppressed issuing of a new ticket if the client
7921     presented a ticket with an expired session.
7922   - Try to set the ticket lifetime hint to something reasonable.
7923   - Make tickets shorter by excluding irrelevant information.
7924   - On the client side, don't ignore renewed tickets.
7925
7926   *Adam Langley, Bodo Moeller (Google)*
7927
7928 * Fix PSK session representation.
7929
7930   *Bodo Moeller*
7931
7932 * Add RC4-MD5 and AESNI-SHA1 "stitched" implementations.
7933
7934   This work was sponsored by Intel.
7935
7936   *Andy Polyakov*
7937
7938 * Add GCM support to TLS library. Some custom code is needed to split
7939   the IV between the fixed (from PRF) and explicit (from TLS record)
7940   portions. This adds all GCM ciphersuites supported by RFC5288 and
7941   RFC5289. Generalise some `AES*` cipherstrings to include GCM and
7942   add a special AESGCM string for GCM only.
7943
7944   *Steve Henson*
7945
7946 * Expand range of ctrls for AES GCM. Permit setting invocation
7947   field on decrypt and retrieval of invocation field only on encrypt.
7948
7949   *Steve Henson*
7950
7951 * Add HMAC ECC ciphersuites from RFC5289. Include SHA384 PRF support.
7952   As required by RFC5289 these ciphersuites cannot be used if for
7953   versions of TLS earlier than 1.2.
7954
7955   *Steve Henson*
7956
7957 * For FIPS capable OpenSSL interpret a NULL default public key method
7958   as unset and return the appropriate default but do *not* set the default.
7959   This means we can return the appropriate method in applications that
7960   switch between FIPS and non-FIPS modes.
7961
7962   *Steve Henson*
7963
7964 * Redirect HMAC and CMAC operations to FIPS module in FIPS mode. If an
7965   ENGINE is used then we cannot handle that in the FIPS module so we
7966   keep original code iff non-FIPS operations are allowed.
7967
7968   *Steve Henson*
7969
7970 * Add -attime option to openssl utilities.
7971
7972   *Peter Eckersley <pde@eff.org>, Ben Laurie and Steve Henson*
7973
7974 * Redirect DSA and DH operations to FIPS module in FIPS mode.
7975
7976   *Steve Henson*
7977
7978 * Redirect ECDSA and ECDH operations to FIPS module in FIPS mode. Also use
7979   FIPS EC methods unconditionally for now.
7980
7981   *Steve Henson*
7982
7983 * New build option no-ec2m to disable characteristic 2 code.
7984
7985   *Steve Henson*
7986
7987 * Backport libcrypto audit of return value checking from 1.1.0-dev; not
7988   all cases can be covered as some introduce binary incompatibilities.
7989
7990   *Steve Henson*
7991
7992 * Redirect RSA operations to FIPS module including keygen,
7993   encrypt, decrypt, sign and verify. Block use of non FIPS RSA methods.
7994
7995   *Steve Henson*
7996
7997 * Add similar low-level API blocking to ciphers.
7998
7999   *Steve Henson*
8000
8001 * low-level digest APIs are not approved in FIPS mode: any attempt
8002   to use these will cause a fatal error. Applications that *really* want
8003   to use them can use the `private_*` version instead.
8004
8005   *Steve Henson*
8006
8007 * Redirect cipher operations to FIPS module for FIPS builds.
8008
8009   *Steve Henson*
8010
8011 * Redirect digest operations to FIPS module for FIPS builds.
8012
8013   *Steve Henson*
8014
8015 * Update build system to add "fips" flag which will link in fipscanister.o
8016   for static and shared library builds embedding a signature if needed.
8017
8018   *Steve Henson*
8019
8020 * Output TLS supported curves in preference order instead of numerical
8021   order. This is currently hardcoded for the highest order curves first.
8022   This should be configurable so applications can judge speed vs strength.
8023
8024   *Steve Henson*
8025
8026 * Add TLS v1.2 server support for client authentication.
8027
8028   *Steve Henson*
8029
8030 * Add support for FIPS mode in ssl library: disable SSLv3, non-FIPS ciphers
8031   and enable MD5.
8032
8033   *Steve Henson*
8034
8035 * Functions FIPS_mode_set() and FIPS_mode() which call the underlying
8036   FIPS modules versions.
8037
8038   *Steve Henson*
8039
8040 * Add TLS v1.2 client side support for client authentication. Keep cache
8041   of handshake records longer as we don't know the hash algorithm to use
8042   until after the certificate request message is received.
8043
8044   *Steve Henson*
8045
8046 * Initial TLS v1.2 client support. Add a default signature algorithms
8047   extension including all the algorithms we support. Parse new signature
8048   format in client key exchange. Relax some ECC signing restrictions for
8049   TLS v1.2 as indicated in RFC5246.
8050
8051   *Steve Henson*
8052
8053 * Add server support for TLS v1.2 signature algorithms extension. Switch
8054   to new signature format when needed using client digest preference.
8055   All server ciphersuites should now work correctly in TLS v1.2. No client
8056   support yet and no support for client certificates.
8057
8058   *Steve Henson*
8059
8060 * Initial TLS v1.2 support. Add new SHA256 digest to ssl code, switch
8061   to SHA256 for PRF when using TLS v1.2 and later. Add new SHA256 based
8062   ciphersuites. At present only RSA key exchange ciphersuites work with
8063   TLS v1.2. Add new option for TLS v1.2 replacing the old and obsolete
8064   SSL_OP_PKCS1_CHECK flags with SSL_OP_NO_TLSv1_2. New TLSv1.2 methods
8065   and version checking.
8066
8067   *Steve Henson*
8068
8069 * New option OPENSSL_NO_SSL_INTERN. If an application can be compiled
8070   with this defined it will not be affected by any changes to ssl internal
8071   structures. Add several utility functions to allow openssl application
8072   to work with OPENSSL_NO_SSL_INTERN defined.
8073
8074   *Steve Henson*
8075
8076 * A long standing patch to add support for SRP from EdelWeb (Peter
8077   Sylvester and Christophe Renou) was integrated.
8078   *Christophe Renou <christophe.renou@edelweb.fr>, Peter Sylvester
8079   <peter.sylvester@edelweb.fr>, Tom Wu <tjw@cs.stanford.edu>, and
8080   Ben Laurie*
8081
8082 * Add functions to copy EVP_PKEY_METHOD and retrieve flags and id.
8083
8084   *Steve Henson*
8085
8086 * Permit abbreviated handshakes when renegotiating using the function
8087   SSL_renegotiate_abbreviated().
8088
8089   *Robin Seggelmann <seggelmann@fh-muenster.de>*
8090
8091 * Add call to ENGINE_register_all_complete() to
8092   ENGINE_load_builtin_engines(), so some implementations get used
8093   automatically instead of needing explicit application support.
8094
8095   *Steve Henson*
8096
8097 * Add support for TLS key exporter as described in RFC5705.
8098
8099   *Robin Seggelmann <seggelmann@fh-muenster.de>, Steve Henson*
8100
8101 * Initial TLSv1.1 support. Since TLSv1.1 is very similar to TLS v1.0 only
8102   a few changes are required:
8103
8104     Add SSL_OP_NO_TLSv1_1 flag.
8105     Add TLSv1_1 methods.
8106     Update version checking logic to handle version 1.1.
8107     Add explicit IV handling (ported from DTLS code).
8108     Add command line options to s_client/s_server.
8109
8110   *Steve Henson*
8111
8112OpenSSL 1.0.0
8113-------------
8114
8115### Changes between 1.0.0s and 1.0.0t [3 Dec 2015]
8116
8117 * X509_ATTRIBUTE memory leak
8118
8119   When presented with a malformed X509_ATTRIBUTE structure OpenSSL will leak
8120   memory. This structure is used by the PKCS#7 and CMS routines so any
8121   application which reads PKCS#7 or CMS data from untrusted sources is
8122   affected. SSL/TLS is not affected.
8123
8124   This issue was reported to OpenSSL by Adam Langley (Google/BoringSSL) using
8125   libFuzzer.
8126   ([CVE-2015-3195])
8127
8128   *Stephen Henson*
8129
8130 * Race condition handling PSK identify hint
8131
8132   If PSK identity hints are received by a multi-threaded client then
8133   the values are wrongly updated in the parent SSL_CTX structure. This can
8134   result in a race condition potentially leading to a double free of the
8135   identify hint data.
8136   ([CVE-2015-3196])
8137
8138   *Stephen Henson*
8139
8140### Changes between 1.0.0r and 1.0.0s [11 Jun 2015]
8141
8142 * Malformed ECParameters causes infinite loop
8143
8144   When processing an ECParameters structure OpenSSL enters an infinite loop
8145   if the curve specified is over a specially malformed binary polynomial
8146   field.
8147
8148   This can be used to perform denial of service against any
8149   system which processes public keys, certificate requests or
8150   certificates.  This includes TLS clients and TLS servers with
8151   client authentication enabled.
8152
8153   This issue was reported to OpenSSL by Joseph Barr-Pixton.
8154   ([CVE-2015-1788])
8155
8156   *Andy Polyakov*
8157
8158 * Exploitable out-of-bounds read in X509_cmp_time
8159
8160   X509_cmp_time does not properly check the length of the ASN1_TIME
8161   string and can read a few bytes out of bounds. In addition,
8162   X509_cmp_time accepts an arbitrary number of fractional seconds in the
8163   time string.
8164
8165   An attacker can use this to craft malformed certificates and CRLs of
8166   various sizes and potentially cause a segmentation fault, resulting in
8167   a DoS on applications that verify certificates or CRLs. TLS clients
8168   that verify CRLs are affected. TLS clients and servers with client
8169   authentication enabled may be affected if they use custom verification
8170   callbacks.
8171
8172   This issue was reported to OpenSSL by Robert Swiecki (Google), and
8173   independently by Hanno Böck.
8174   ([CVE-2015-1789])
8175
8176   *Emilia Käsper*
8177
8178 * PKCS7 crash with missing EnvelopedContent
8179
8180   The PKCS#7 parsing code does not handle missing inner EncryptedContent
8181   correctly. An attacker can craft malformed ASN.1-encoded PKCS#7 blobs
8182   with missing content and trigger a NULL pointer dereference on parsing.
8183
8184   Applications that decrypt PKCS#7 data or otherwise parse PKCS#7
8185   structures from untrusted sources are affected. OpenSSL clients and
8186   servers are not affected.
8187
8188   This issue was reported to OpenSSL by Michal Zalewski (Google).
8189   ([CVE-2015-1790])
8190
8191   *Emilia Käsper*
8192
8193 * CMS verify infinite loop with unknown hash function
8194
8195   When verifying a signedData message the CMS code can enter an infinite loop
8196   if presented with an unknown hash function OID. This can be used to perform
8197   denial of service against any system which verifies signedData messages using
8198   the CMS code.
8199   This issue was reported to OpenSSL by Johannes Bauer.
8200   ([CVE-2015-1792])
8201
8202   *Stephen Henson*
8203
8204 * Race condition handling NewSessionTicket
8205
8206   If a NewSessionTicket is received by a multi-threaded client when attempting to
8207   reuse a previous ticket then a race condition can occur potentially leading to
8208   a double free of the ticket data.
8209   ([CVE-2015-1791])
8210
8211   *Matt Caswell*
8212
8213### Changes between 1.0.0q and 1.0.0r [19 Mar 2015]
8214
8215 * Segmentation fault in ASN1_TYPE_cmp fix
8216
8217   The function ASN1_TYPE_cmp will crash with an invalid read if an attempt is
8218   made to compare ASN.1 boolean types. Since ASN1_TYPE_cmp is used to check
8219   certificate signature algorithm consistency this can be used to crash any
8220   certificate verification operation and exploited in a DoS attack. Any
8221   application which performs certificate verification is vulnerable including
8222   OpenSSL clients and servers which enable client authentication.
8223   ([CVE-2015-0286])
8224
8225   *Stephen Henson*
8226
8227 * ASN.1 structure reuse memory corruption fix
8228
8229   Reusing a structure in ASN.1 parsing may allow an attacker to cause
8230   memory corruption via an invalid write. Such reuse is and has been
8231   strongly discouraged and is believed to be rare.
8232
8233   Applications that parse structures containing CHOICE or ANY DEFINED BY
8234   components may be affected. Certificate parsing (d2i_X509 and related
8235   functions) are however not affected. OpenSSL clients and servers are
8236   not affected.
8237   ([CVE-2015-0287])
8238
8239   *Stephen Henson*
8240
8241 * PKCS7 NULL pointer dereferences fix
8242
8243   The PKCS#7 parsing code does not handle missing outer ContentInfo
8244   correctly. An attacker can craft malformed ASN.1-encoded PKCS#7 blobs with
8245   missing content and trigger a NULL pointer dereference on parsing.
8246
8247   Applications that verify PKCS#7 signatures, decrypt PKCS#7 data or
8248   otherwise parse PKCS#7 structures from untrusted sources are
8249   affected. OpenSSL clients and servers are not affected.
8250
8251   This issue was reported to OpenSSL by Michal Zalewski (Google).
8252   ([CVE-2015-0289])
8253
8254   *Emilia Käsper*
8255
8256 * DoS via reachable assert in SSLv2 servers fix
8257
8258   A malicious client can trigger an OPENSSL_assert (i.e., an abort) in
8259   servers that both support SSLv2 and enable export cipher suites by sending
8260   a specially crafted SSLv2 CLIENT-MASTER-KEY message.
8261
8262   This issue was discovered by Sean Burford (Google) and Emilia Käsper
8263   (OpenSSL development team).
8264   ([CVE-2015-0293])
8265
8266   *Emilia Käsper*
8267
8268 * Use After Free following d2i_ECPrivatekey error fix
8269
8270   A malformed EC private key file consumed via the d2i_ECPrivateKey function
8271   could cause a use after free condition. This, in turn, could cause a double
8272   free in several private key parsing functions (such as d2i_PrivateKey
8273   or EVP_PKCS82PKEY) and could lead to a DoS attack or memory corruption
8274   for applications that receive EC private keys from untrusted
8275   sources. This scenario is considered rare.
8276
8277   This issue was discovered by the BoringSSL project and fixed in their
8278   commit 517073cd4b.
8279   ([CVE-2015-0209])
8280
8281   *Matt Caswell*
8282
8283 * X509_to_X509_REQ NULL pointer deref fix
8284
8285   The function X509_to_X509_REQ will crash with a NULL pointer dereference if
8286   the certificate key is invalid. This function is rarely used in practice.
8287
8288   This issue was discovered by Brian Carpenter.
8289   ([CVE-2015-0288])
8290
8291   *Stephen Henson*
8292
8293 * Removed the export ciphers from the DEFAULT ciphers
8294
8295   *Kurt Roeckx*
8296
8297### Changes between 1.0.0p and 1.0.0q [15 Jan 2015]
8298
8299 * Build fixes for the Windows and OpenVMS platforms
8300
8301   *Matt Caswell and Richard Levitte*
8302
8303### Changes between 1.0.0o and 1.0.0p [8 Jan 2015]
8304
8305 * Fix DTLS segmentation fault in dtls1_get_record. A carefully crafted DTLS
8306   message can cause a segmentation fault in OpenSSL due to a NULL pointer
8307   dereference. This could lead to a Denial Of Service attack. Thanks to
8308   Markus Stenberg of Cisco Systems, Inc. for reporting this issue.
8309   ([CVE-2014-3571])
8310
8311   *Steve Henson*
8312
8313 * Fix DTLS memory leak in dtls1_buffer_record. A memory leak can occur in the
8314   dtls1_buffer_record function under certain conditions. In particular this
8315   could occur if an attacker sent repeated DTLS records with the same
8316   sequence number but for the next epoch. The memory leak could be exploited
8317   by an attacker in a Denial of Service attack through memory exhaustion.
8318   Thanks to Chris Mueller for reporting this issue.
8319   ([CVE-2015-0206])
8320
8321   *Matt Caswell*
8322
8323 * Fix issue where no-ssl3 configuration sets method to NULL. When openssl is
8324   built with the no-ssl3 option and a SSL v3 ClientHello is received the ssl
8325   method would be set to NULL which could later result in a NULL pointer
8326   dereference. Thanks to Frank Schmirler for reporting this issue.
8327   ([CVE-2014-3569])
8328
8329   *Kurt Roeckx*
8330
8331 * Abort handshake if server key exchange message is omitted for ephemeral
8332   ECDH ciphersuites.
8333
8334   Thanks to Karthikeyan Bhargavan of the PROSECCO team at INRIA for
8335   reporting this issue.
8336   ([CVE-2014-3572])
8337
8338   *Steve Henson*
8339
8340 * Remove non-export ephemeral RSA code on client and server. This code
8341   violated the TLS standard by allowing the use of temporary RSA keys in
8342   non-export ciphersuites and could be used by a server to effectively
8343   downgrade the RSA key length used to a value smaller than the server
8344   certificate. Thanks for Karthikeyan Bhargavan of the PROSECCO team at
8345   INRIA or reporting this issue.
8346   ([CVE-2015-0204])
8347
8348   *Steve Henson*
8349
8350 * Fixed issue where DH client certificates are accepted without verification.
8351   An OpenSSL server will accept a DH certificate for client authentication
8352   without the certificate verify message. This effectively allows a client to
8353   authenticate without the use of a private key. This only affects servers
8354   which trust a client certificate authority which issues certificates
8355   containing DH keys: these are extremely rare and hardly ever encountered.
8356   Thanks for Karthikeyan Bhargavan of the PROSECCO team at INRIA or reporting
8357   this issue.
8358   ([CVE-2015-0205])
8359
8360   *Steve Henson*
8361
8362 * Correct Bignum squaring. Bignum squaring (BN_sqr) may produce incorrect
8363   results on some platforms, including x86_64. This bug occurs at random
8364   with a very low probability, and is not known to be exploitable in any
8365   way, though its exact impact is difficult to determine. Thanks to Pieter
8366   Wuille (Blockstream) who reported this issue and also suggested an initial
8367   fix. Further analysis was conducted by the OpenSSL development team and
8368   Adam Langley of Google. The final fix was developed by Andy Polyakov of
8369   the OpenSSL core team.
8370   ([CVE-2014-3570])
8371
8372   *Andy Polyakov*
8373
8374 * Fix various certificate fingerprint issues.
8375
8376   By using non-DER or invalid encodings outside the signed portion of a
8377   certificate the fingerprint can be changed without breaking the signature.
8378   Although no details of the signed portion of the certificate can be changed
8379   this can cause problems with some applications: e.g. those using the
8380   certificate fingerprint for blacklists.
8381
8382   1. Reject signatures with non zero unused bits.
8383
8384   If the BIT STRING containing the signature has non zero unused bits reject
8385   the signature. All current signature algorithms require zero unused bits.
8386
8387   2. Check certificate algorithm consistency.
8388
8389   Check the AlgorithmIdentifier inside TBS matches the one in the
8390   certificate signature. NB: this will result in signature failure
8391   errors for some broken certificates.
8392
8393   Thanks to Konrad Kraszewski from Google for reporting this issue.
8394
8395   3. Check DSA/ECDSA signatures use DER.
8396
8397   Reencode DSA/ECDSA signatures and compare with the original received
8398   signature. Return an error if there is a mismatch.
8399
8400   This will reject various cases including garbage after signature
8401   (thanks to Antti Karjalainen and Tuomo Untinen from the Codenomicon CROSS
8402   program for discovering this case) and use of BER or invalid ASN.1 INTEGERs
8403   (negative or with leading zeroes).
8404
8405   Further analysis was conducted and fixes were developed by Stephen Henson
8406   of the OpenSSL core team.
8407
8408   ([CVE-2014-8275])
8409
8410   *Steve Henson*
8411
8412### Changes between 1.0.0n and 1.0.0o [15 Oct 2014]
8413
8414 * Session Ticket Memory Leak.
8415
8416   When an OpenSSL SSL/TLS/DTLS server receives a session ticket the
8417   integrity of that ticket is first verified. In the event of a session
8418   ticket integrity check failing, OpenSSL will fail to free memory
8419   causing a memory leak. By sending a large number of invalid session
8420   tickets an attacker could exploit this issue in a Denial Of Service
8421   attack.
8422   ([CVE-2014-3567])
8423
8424   *Steve Henson*
8425
8426 * Build option no-ssl3 is incomplete.
8427
8428   When OpenSSL is configured with "no-ssl3" as a build option, servers
8429   could accept and complete a SSL 3.0 handshake, and clients could be
8430   configured to send them.
8431   ([CVE-2014-3568])
8432
8433   *Akamai and the OpenSSL team*
8434
8435 * Add support for TLS_FALLBACK_SCSV.
8436   Client applications doing fallback retries should call
8437   SSL_set_mode(s, SSL_MODE_SEND_FALLBACK_SCSV).
8438   ([CVE-2014-3566])
8439
8440   *Adam Langley, Bodo Moeller*
8441
8442 * Add additional DigestInfo checks.
8443
8444   Reencode DigestInto in DER and check against the original when
8445   verifying RSA signature: this will reject any improperly encoded
8446   DigestInfo structures.
8447
8448   Note: this is a precautionary measure and no attacks are currently known.
8449
8450   *Steve Henson*
8451
8452### Changes between 1.0.0m and 1.0.0n [6 Aug 2014]
8453
8454 * OpenSSL DTLS clients enabling anonymous (EC)DH ciphersuites are subject
8455   to a denial of service attack. A malicious server can crash the client
8456   with a null pointer dereference (read) by specifying an anonymous (EC)DH
8457   ciphersuite and sending carefully crafted handshake messages.
8458
8459   Thanks to Felix Gröbert (Google) for discovering and researching this
8460   issue.
8461   ([CVE-2014-3510])
8462
8463   *Emilia Käsper*
8464
8465 * By sending carefully crafted DTLS packets an attacker could cause openssl
8466   to leak memory. This can be exploited through a Denial of Service attack.
8467   Thanks to Adam Langley for discovering and researching this issue.
8468   ([CVE-2014-3507])
8469
8470   *Adam Langley*
8471
8472 * An attacker can force openssl to consume large amounts of memory whilst
8473   processing DTLS handshake messages. This can be exploited through a
8474   Denial of Service attack.
8475   Thanks to Adam Langley for discovering and researching this issue.
8476   ([CVE-2014-3506])
8477
8478   *Adam Langley*
8479
8480 * An attacker can force an error condition which causes openssl to crash
8481   whilst processing DTLS packets due to memory being freed twice. This
8482   can be exploited through a Denial of Service attack.
8483   Thanks to Adam Langley and Wan-Teh Chang for discovering and researching
8484   this issue.
8485   ([CVE-2014-3505])
8486
8487   *Adam Langley*
8488
8489 * If a multithreaded client connects to a malicious server using a resumed
8490   session and the server sends an ec point format extension it could write
8491   up to 255 bytes to freed memory.
8492
8493   Thanks to Gabor Tyukasz (LogMeIn Inc) for discovering and researching this
8494   issue.
8495   ([CVE-2014-3509])
8496
8497   *Gabor Tyukasz*
8498
8499 * A flaw in OBJ_obj2txt may cause pretty printing functions such as
8500   X509_name_oneline, X509_name_print_ex et al. to leak some information
8501   from the stack. Applications may be affected if they echo pretty printing
8502   output to the attacker.
8503
8504   Thanks to Ivan Fratric (Google) for discovering this issue.
8505   ([CVE-2014-3508])
8506
8507   *Emilia Käsper, and Steve Henson*
8508
8509 * Fix ec_GFp_simple_points_make_affine (thus, EC_POINTs_mul etc.)
8510   for corner cases. (Certain input points at infinity could lead to
8511   bogus results, with non-infinity inputs mapped to infinity too.)
8512
8513   *Bodo Moeller*
8514
8515### Changes between 1.0.0l and 1.0.0m [5 Jun 2014]
8516
8517 * Fix for SSL/TLS MITM flaw. An attacker using a carefully crafted
8518   handshake can force the use of weak keying material in OpenSSL
8519   SSL/TLS clients and servers.
8520
8521   Thanks to KIKUCHI Masashi (Lepidum Co. Ltd.) for discovering and
8522   researching this issue. ([CVE-2014-0224])
8523
8524   *KIKUCHI Masashi, Steve Henson*
8525
8526 * Fix DTLS recursion flaw. By sending an invalid DTLS handshake to an
8527   OpenSSL DTLS client the code can be made to recurse eventually crashing
8528   in a DoS attack.
8529
8530   Thanks to Imre Rad (Search-Lab Ltd.) for discovering this issue.
8531   ([CVE-2014-0221])
8532
8533   *Imre Rad, Steve Henson*
8534
8535 * Fix DTLS invalid fragment vulnerability. A buffer overrun attack can
8536   be triggered by sending invalid DTLS fragments to an OpenSSL DTLS
8537   client or server. This is potentially exploitable to run arbitrary
8538   code on a vulnerable client or server.
8539
8540   Thanks to Jüri Aedla for reporting this issue. ([CVE-2014-0195])
8541
8542   *Jüri Aedla, Steve Henson*
8543
8544 * Fix bug in TLS code where clients enable anonymous ECDH ciphersuites
8545   are subject to a denial of service attack.
8546
8547   Thanks to Felix Gröbert and Ivan Fratric at Google for discovering
8548   this issue. ([CVE-2014-3470])
8549
8550   *Felix Gröbert, Ivan Fratric, Steve Henson*
8551
8552 * Harmonize version and its documentation. -f flag is used to display
8553   compilation flags.
8554
8555   *mancha <mancha1@zoho.com>*
8556
8557 * Fix eckey_priv_encode so it immediately returns an error upon a failure
8558   in i2d_ECPrivateKey.
8559
8560   *mancha <mancha1@zoho.com>*
8561
8562 * Fix some double frees. These are not thought to be exploitable.
8563
8564   *mancha <mancha1@zoho.com>*
8565
8566 * Fix for the attack described in the paper "Recovering OpenSSL
8567   ECDSA Nonces Using the FLUSH+RELOAD Cache Side-channel Attack"
8568   by Yuval Yarom and Naomi Benger. Details can be obtained from:
8569   <http://eprint.iacr.org/2014/140>
8570
8571   Thanks to Yuval Yarom and Naomi Benger for discovering this
8572   flaw and to Yuval Yarom for supplying a fix ([CVE-2014-0076])
8573
8574   *Yuval Yarom and Naomi Benger*
8575
8576### Changes between 1.0.0k and 1.0.0l [6 Jan 2014]
8577
8578 * Keep original DTLS digest and encryption contexts in retransmission
8579   structures so we can use the previous session parameters if they need
8580   to be resent. ([CVE-2013-6450])
8581
8582   *Steve Henson*
8583
8584 * Add option SSL_OP_SAFARI_ECDHE_ECDSA_BUG (part of SSL_OP_ALL) which
8585   avoids preferring ECDHE-ECDSA ciphers when the client appears to be
8586   Safari on OS X.  Safari on OS X 10.8..10.8.3 advertises support for
8587   several ECDHE-ECDSA ciphers, but fails to negotiate them.  The bug
8588   is fixed in OS X 10.8.4, but Apple have ruled out both hot fixing
8589   10.8..10.8.3 and forcing users to upgrade to 10.8.4 or newer.
8590
8591   *Rob Stradling, Adam Langley*
8592
8593### Changes between 1.0.0j and 1.0.0k [5 Feb 2013]
8594
8595 * Make the decoding of SSLv3, TLS and DTLS CBC records constant time.
8596
8597   This addresses the flaw in CBC record processing discovered by
8598   Nadhem Alfardan and Kenny Paterson. Details of this attack can be found
8599   at: <http://www.isg.rhul.ac.uk/tls/>
8600
8601   Thanks go to Nadhem Alfardan and Kenny Paterson of the Information
8602   Security Group at Royal Holloway, University of London
8603   (www.isg.rhul.ac.uk) for discovering this flaw and Adam Langley and
8604   Emilia Käsper for the initial patch.
8605   ([CVE-2013-0169])
8606
8607   *Emilia Käsper, Adam Langley, Ben Laurie, Andy Polyakov, Steve Henson*
8608
8609 * Return an error when checking OCSP signatures when key is NULL.
8610   This fixes a DoS attack. ([CVE-2013-0166])
8611
8612   *Steve Henson*
8613
8614 * Call OCSP Stapling callback after ciphersuite has been chosen, so
8615   the right response is stapled. Also change SSL_get_certificate()
8616   so it returns the certificate actually sent.
8617   See <http://rt.openssl.org/Ticket/Display.html?id=2836>.
8618   (This is a backport)
8619
8620   *Rob Stradling <rob.stradling@comodo.com>*
8621
8622 * Fix possible deadlock when decoding public keys.
8623
8624   *Steve Henson*
8625
8626### Changes between 1.0.0i and 1.0.0j [10 May 2012]
8627
8628[NB: OpenSSL 1.0.0i and later 1.0.0 patch levels were released after
8629OpenSSL 1.0.1.]
8630
8631 * Sanity check record length before skipping explicit IV in DTLS
8632   to fix DoS attack.
8633
8634   Thanks to Codenomicon for discovering this issue using Fuzz-o-Matic
8635   fuzzing as a service testing platform.
8636   ([CVE-2012-2333])
8637
8638   *Steve Henson*
8639
8640 * Initialise tkeylen properly when encrypting CMS messages.
8641   Thanks to Solar Designer of Openwall for reporting this issue.
8642
8643   *Steve Henson*
8644
8645### Changes between 1.0.0h and 1.0.0i [19 Apr 2012]
8646
8647 * Check for potentially exploitable overflows in asn1_d2i_read_bio
8648   BUF_mem_grow and BUF_mem_grow_clean. Refuse attempts to shrink buffer
8649   in CRYPTO_realloc_clean.
8650
8651   Thanks to Tavis Ormandy, Google Security Team, for discovering this
8652   issue and to Adam Langley <agl@chromium.org> for fixing it.
8653   ([CVE-2012-2110])
8654
8655   *Adam Langley (Google), Tavis Ormandy, Google Security Team*
8656
8657### Changes between 1.0.0g and 1.0.0h [12 Mar 2012]
8658
8659 * Fix MMA (Bleichenbacher's attack on PKCS #1 v1.5 RSA padding) weakness
8660   in CMS and PKCS7 code. When RSA decryption fails use a random key for
8661   content decryption and always return the same error. Note: this attack
8662   needs on average 2^20 messages so it only affects automated senders. The
8663   old behaviour can be re-enabled in the CMS code by setting the
8664   CMS_DEBUG_DECRYPT flag: this is useful for debugging and testing where
8665   an MMA defence is not necessary.
8666   Thanks to Ivan Nestlerode <inestlerode@us.ibm.com> for discovering
8667   this issue. ([CVE-2012-0884])
8668
8669   *Steve Henson*
8670
8671 * Fix CVE-2011-4619: make sure we really are receiving a
8672   client hello before rejecting multiple SGC restarts. Thanks to
8673   Ivan Nestlerode <inestlerode@us.ibm.com> for discovering this bug.
8674
8675   *Steve Henson*
8676
8677### Changes between 1.0.0f and 1.0.0g [18 Jan 2012]
8678
8679 * Fix for DTLS DoS issue introduced by fix for CVE-2011-4109.
8680   Thanks to Antonio Martin, Enterprise Secure Access Research and
8681   Development, Cisco Systems, Inc. for discovering this bug and
8682   preparing a fix. ([CVE-2012-0050])
8683
8684   *Antonio Martin*
8685
8686### Changes between 1.0.0e and 1.0.0f [4 Jan 2012]
8687
8688 * Nadhem Alfardan and Kenny Paterson have discovered an extension
8689   of the Vaudenay padding oracle attack on CBC mode encryption
8690   which enables an efficient plaintext recovery attack against
8691   the OpenSSL implementation of DTLS. Their attack exploits timing
8692   differences arising during decryption processing. A research
8693   paper describing this attack can be found at:
8694   <http://www.isg.rhul.ac.uk/~kp/dtls.pdf>
8695   Thanks go to Nadhem Alfardan and Kenny Paterson of the Information
8696   Security Group at Royal Holloway, University of London
8697   (www.isg.rhul.ac.uk) for discovering this flaw and to Robin Seggelmann
8698   <seggelmann@fh-muenster.de> and Michael Tuexen <tuexen@fh-muenster.de>
8699   for preparing the fix. ([CVE-2011-4108])
8700
8701   *Robin Seggelmann, Michael Tuexen*
8702
8703 * Clear bytes used for block padding of SSL 3.0 records.
8704   ([CVE-2011-4576])
8705
8706   *Adam Langley (Google)*
8707
8708 * Only allow one SGC handshake restart for SSL/TLS. Thanks to George
8709   Kadianakis <desnacked@gmail.com> for discovering this issue and
8710   Adam Langley for preparing the fix. ([CVE-2011-4619])
8711
8712   *Adam Langley (Google)*
8713
8714 * Check parameters are not NULL in GOST ENGINE. ([CVE-2012-0027])
8715
8716   *Andrey Kulikov <amdeich@gmail.com>*
8717
8718 * Prevent malformed RFC3779 data triggering an assertion failure.
8719   Thanks to Andrew Chi, BBN Technologies, for discovering the flaw
8720   and Rob Austein <sra@hactrn.net> for fixing it. ([CVE-2011-4577])
8721
8722   *Rob Austein <sra@hactrn.net>*
8723
8724 * Improved PRNG seeding for VOS.
8725
8726   *Paul Green <Paul.Green@stratus.com>*
8727
8728 * Fix ssl_ciph.c set-up race.
8729
8730   *Adam Langley (Google)*
8731
8732 * Fix spurious failures in ecdsatest.c.
8733
8734   *Emilia Käsper (Google)*
8735
8736 * Fix the BIO_f_buffer() implementation (which was mixing different
8737   interpretations of the `..._len` fields).
8738
8739   *Adam Langley (Google)*
8740
8741 * Fix handling of BN_BLINDING: now BN_BLINDING_invert_ex (rather than
8742   BN_BLINDING_invert_ex) calls BN_BLINDING_update, ensuring that concurrent
8743   threads won't reuse the same blinding coefficients.
8744
8745   This also avoids the need to obtain the CRYPTO_LOCK_RSA_BLINDING
8746   lock to call BN_BLINDING_invert_ex, and avoids one use of
8747   BN_BLINDING_update for each BN_BLINDING structure (previously,
8748   the last update always remained unused).
8749
8750   *Emilia Käsper (Google)*
8751
8752 * In ssl3_clear, preserve s3->init_extra along with s3->rbuf.
8753
8754   *Bob Buckholz (Google)*
8755
8756### Changes between 1.0.0d and 1.0.0e [6 Sep 2011]
8757
8758 * Fix bug where CRLs with nextUpdate in the past are sometimes accepted
8759   by initialising X509_STORE_CTX properly. ([CVE-2011-3207])
8760
8761   *Kaspar Brand <ossl@velox.ch>*
8762
8763 * Fix SSL memory handling for (EC)DH ciphersuites, in particular
8764   for multi-threaded use of ECDH. ([CVE-2011-3210])
8765
8766   *Adam Langley (Google)*
8767
8768 * Fix x509_name_ex_d2i memory leak on bad inputs.
8769
8770   *Bodo Moeller*
8771
8772 * Remove hard coded ecdsaWithSHA1 signature tests in ssl code and check
8773   signature public key algorithm by using OID xref utilities instead.
8774   Before this you could only use some ECC ciphersuites with SHA1 only.
8775
8776   *Steve Henson*
8777
8778 * Add protection against ECDSA timing attacks as mentioned in the paper
8779   by Billy Bob Brumley and Nicola Tuveri, see:
8780   <http://eprint.iacr.org/2011/232.pdf>
8781
8782   *Billy Bob Brumley and Nicola Tuveri*
8783
8784### Changes between 1.0.0c and 1.0.0d [8 Feb 2011]
8785
8786 * Fix parsing of OCSP stapling ClientHello extension. CVE-2011-0014
8787
8788   *Neel Mehta, Adam Langley, Bodo Moeller (Google)*
8789
8790 * Fix bug in string printing code: if *any* escaping is enabled we must
8791   escape the escape character (backslash) or the resulting string is
8792   ambiguous.
8793
8794   *Steve Henson*
8795
8796### Changes between 1.0.0b and 1.0.0c  [2 Dec 2010]
8797
8798 * Disable code workaround for ancient and obsolete Netscape browsers
8799   and servers: an attacker can use it in a ciphersuite downgrade attack.
8800   Thanks to Martin Rex for discovering this bug. CVE-2010-4180
8801
8802   *Steve Henson*
8803
8804 * Fixed J-PAKE implementation error, originally discovered by
8805   Sebastien Martini, further info and confirmation from Stefan
8806   Arentz and Feng Hao. Note that this fix is a security fix. CVE-2010-4252
8807
8808   *Ben Laurie*
8809
8810### Changes between 1.0.0a and 1.0.0b  [16 Nov 2010]
8811
8812 * Fix extension code to avoid race conditions which can result in a buffer
8813   overrun vulnerability: resumed sessions must not be modified as they can
8814   be shared by multiple threads. CVE-2010-3864
8815
8816   *Steve Henson*
8817
8818 * Fix WIN32 build system to correctly link an ENGINE directory into
8819   a DLL.
8820
8821   *Steve Henson*
8822
8823### Changes between 1.0.0 and 1.0.0a  [01 Jun 2010]
8824
8825 * Check return value of int_rsa_verify in pkey_rsa_verifyrecover
8826   ([CVE-2010-1633])
8827
8828   *Steve Henson, Peter-Michael Hager <hager@dortmund.net>*
8829
8830### Changes between 0.9.8n and 1.0.0  [29 Mar 2010]
8831
8832 * Add "missing" function EVP_CIPHER_CTX_copy(). This copies a cipher
8833   context. The operation can be customised via the ctrl mechanism in
8834   case ENGINEs want to include additional functionality.
8835
8836   *Steve Henson*
8837
8838 * Tolerate yet another broken PKCS#8 key format: private key value negative.
8839
8840   *Steve Henson*
8841
8842 * Add new -subject_hash_old and -issuer_hash_old options to x509 utility to
8843   output hashes compatible with older versions of OpenSSL.
8844
8845   *Willy Weisz <weisz@vcpc.univie.ac.at>*
8846
8847 * Fix compression algorithm handling: if resuming a session use the
8848   compression algorithm of the resumed session instead of determining
8849   it from client hello again. Don't allow server to change algorithm.
8850
8851   *Steve Henson*
8852
8853 * Add load_crls() function to commands tidying load_certs() too. Add option
8854   to verify utility to allow additional CRLs to be included.
8855
8856   *Steve Henson*
8857
8858 * Update OCSP request code to permit adding custom headers to the request:
8859   some responders need this.
8860
8861   *Steve Henson*
8862
8863 * The function EVP_PKEY_sign() returns <=0 on error: check return code
8864   correctly.
8865
8866   *Julia Lawall <julia@diku.dk>*
8867
8868 * Update verify callback code in `apps/s_cb.c` and `apps/verify.c`, it
8869   needlessly dereferenced structures, used obsolete functions and
8870   didn't handle all updated verify codes correctly.
8871
8872   *Steve Henson*
8873
8874 * Disable MD2 in the default configuration.
8875
8876   *Steve Henson*
8877
8878 * In BIO_pop() and BIO_push() use the ctrl argument (which was NULL) to
8879   indicate the initial BIO being pushed or popped. This makes it possible
8880   to determine whether the BIO is the one explicitly called or as a result
8881   of the ctrl being passed down the chain. Fix BIO_pop() and SSL BIOs so
8882   it handles reference counts correctly and doesn't zero out the I/O bio
8883   when it is not being explicitly popped. WARNING: applications which
8884   included workarounds for the old buggy behaviour will need to be modified
8885   or they could free up already freed BIOs.
8886
8887   *Steve Henson*
8888
8889 * Extend the uni2asc/asc2uni => OPENSSL_uni2asc/OPENSSL_asc2uni
8890   renaming to all platforms (within the 0.9.8 branch, this was
8891   done conditionally on Netware platforms to avoid a name clash).
8892
8893   *Guenter <lists@gknw.net>*
8894
8895 * Add ECDHE and PSK support to DTLS.
8896
8897   *Michael Tuexen <tuexen@fh-muenster.de>*
8898
8899 * Add CHECKED_STACK_OF macro to safestack.h, otherwise safestack can't
8900   be used on C++.
8901
8902   *Steve Henson*
8903
8904 * Add "missing" function EVP_MD_flags() (without this the only way to
8905   retrieve a digest flags is by accessing the structure directly. Update
8906   `EVP_MD_do_all*()` and `EVP_CIPHER_do_all*()` to include the name a digest
8907   or cipher is registered as in the "from" argument. Print out all
8908   registered digests in the dgst usage message instead of manually
8909   attempting to work them out.
8910
8911   *Steve Henson*
8912
8913 * If no SSLv2 ciphers are used don't use an SSLv2 compatible client hello:
8914   this allows the use of compression and extensions. Change default cipher
8915   string to remove SSLv2 ciphersuites. This effectively avoids ancient SSLv2
8916   by default unless an application cipher string requests it.
8917
8918   *Steve Henson*
8919
8920 * Alter match criteria in PKCS12_parse(). It used to try to use local
8921   key ids to find matching certificates and keys but some PKCS#12 files
8922   don't follow the (somewhat unwritten) rules and this strategy fails.
8923   Now just gather all certificates together and the first private key
8924   then look for the first certificate that matches the key.
8925
8926   *Steve Henson*
8927
8928 * Support use of registered digest and cipher names for dgst and cipher
8929   commands instead of having to add each one as a special case. So now
8930   you can do:
8931
8932           openssl sha256 foo
8933
8934   as well as:
8935
8936           openssl dgst -sha256 foo
8937
8938   and this works for ENGINE based algorithms too.
8939
8940   *Steve Henson*
8941
8942 * Update Gost ENGINE to support parameter files.
8943
8944   *Victor B. Wagner <vitus@cryptocom.ru>*
8945
8946 * Support GeneralizedTime in ca utility.
8947
8948   *Oliver Martin <oliver@volatilevoid.net>, Steve Henson*
8949
8950 * Enhance the hash format used for certificate directory links. The new
8951   form uses the canonical encoding (meaning equivalent names will work
8952   even if they aren't identical) and uses SHA1 instead of MD5. This form
8953   is incompatible with the older format and as a result c_rehash should
8954   be used to rebuild symbolic links.
8955
8956   *Steve Henson*
8957
8958 * Make PKCS#8 the default write format for private keys, replacing the
8959   traditional format. This form is standardised, more secure and doesn't
8960   include an implicit MD5 dependency.
8961
8962   *Steve Henson*
8963
8964 * Add a $gcc_devteam_warn option to Configure. The idea is that any code
8965   committed to OpenSSL should pass this lot as a minimum.
8966
8967   *Steve Henson*
8968
8969 * Add session ticket override functionality for use by EAP-FAST.
8970
8971   *Jouni Malinen <j@w1.fi>*
8972
8973 * Modify HMAC functions to return a value. Since these can be implemented
8974   in an ENGINE errors can occur.
8975
8976   *Steve Henson*
8977
8978 * Type-checked OBJ_bsearch_ex.
8979
8980   *Ben Laurie*
8981
8982 * Type-checked OBJ_bsearch. Also some constification necessitated
8983   by type-checking.  Still to come: TXT_DB, bsearch(?),
8984   OBJ_bsearch_ex, qsort, CRYPTO_EX_DATA, ASN1_VALUE, ASN1_STRING,
8985   CONF_VALUE.
8986
8987   *Ben Laurie*
8988
8989 * New function OPENSSL_gmtime_adj() to add a specific number of days and
8990   seconds to a tm structure directly, instead of going through OS
8991   specific date routines. This avoids any issues with OS routines such
8992   as the year 2038 bug. New `*_adj()` functions for ASN1 time structures
8993   and X509_time_adj_ex() to cover the extended range. The existing
8994   X509_time_adj() is still usable and will no longer have any date issues.
8995
8996   *Steve Henson*
8997
8998 * Delta CRL support. New use deltas option which will attempt to locate
8999   and search any appropriate delta CRLs available.
9000
9001   This work was sponsored by Google.
9002
9003   *Steve Henson*
9004
9005 * Support for CRLs partitioned by reason code. Reorganise CRL processing
9006   code and add additional score elements. Validate alternate CRL paths
9007   as part of the CRL checking and indicate a new error "CRL path validation
9008   error" in this case. Applications wanting additional details can use
9009   the verify callback and check the new "parent" field. If this is not
9010   NULL CRL path validation is taking place. Existing applications won't
9011   see this because it requires extended CRL support which is off by
9012   default.
9013
9014   This work was sponsored by Google.
9015
9016   *Steve Henson*
9017
9018 * Support for freshest CRL extension.
9019
9020   This work was sponsored by Google.
9021
9022   *Steve Henson*
9023
9024 * Initial indirect CRL support. Currently only supported in the CRLs
9025   passed directly and not via lookup. Process certificate issuer
9026   CRL entry extension and lookup CRL entries by bother issuer name
9027   and serial number. Check and process CRL issuer entry in IDP extension.
9028
9029   This work was sponsored by Google.
9030
9031   *Steve Henson*
9032
9033 * Add support for distinct certificate and CRL paths. The CRL issuer
9034   certificate is validated separately in this case. Only enabled if
9035   an extended CRL support flag is set: this flag will enable additional
9036   CRL functionality in future.
9037
9038   This work was sponsored by Google.
9039
9040   *Steve Henson*
9041
9042 * Add support for policy mappings extension.
9043
9044   This work was sponsored by Google.
9045
9046   *Steve Henson*
9047
9048 * Fixes to pathlength constraint, self issued certificate handling,
9049   policy processing to align with RFC3280 and PKITS tests.
9050
9051   This work was sponsored by Google.
9052
9053   *Steve Henson*
9054
9055 * Support for name constraints certificate extension. DN, email, DNS
9056   and URI types are currently supported.
9057
9058   This work was sponsored by Google.
9059
9060   *Steve Henson*
9061
9062 * To cater for systems that provide a pointer-based thread ID rather
9063   than numeric, deprecate the current numeric thread ID mechanism and
9064   replace it with a structure and associated callback type. This
9065   mechanism allows a numeric "hash" to be extracted from a thread ID in
9066   either case, and on platforms where pointers are larger than 'long',
9067   mixing is done to help ensure the numeric 'hash' is usable even if it
9068   can't be guaranteed unique. The default mechanism is to use "&errno"
9069   as a pointer-based thread ID to distinguish between threads.
9070
9071   Applications that want to provide their own thread IDs should now use
9072   CRYPTO_THREADID_set_callback() to register a callback that will call
9073   either CRYPTO_THREADID_set_numeric() or CRYPTO_THREADID_set_pointer().
9074
9075   Note that ERR_remove_state() is now deprecated, because it is tied
9076   to the assumption that thread IDs are numeric.  ERR_remove_state(0)
9077   to free the current thread's error state should be replaced by
9078   ERR_remove_thread_state(NULL).
9079
9080   (This new approach replaces the functions CRYPTO_set_idptr_callback(),
9081   CRYPTO_get_idptr_callback(), and CRYPTO_thread_idptr() that existed in
9082   OpenSSL 0.9.9-dev between June 2006 and August 2008. Also, if an
9083   application was previously providing a numeric thread callback that
9084   was inappropriate for distinguishing threads, then uniqueness might
9085   have been obtained with &errno that happened immediately in the
9086   intermediate development versions of OpenSSL; this is no longer the
9087   case, the numeric thread callback will now override the automatic use
9088   of &errno.)
9089
9090   *Geoff Thorpe, with help from Bodo Moeller*
9091
9092 * Initial support for different CRL issuing certificates. This covers a
9093   simple case where the self issued certificates in the chain exist and
9094   the real CRL issuer is higher in the existing chain.
9095
9096   This work was sponsored by Google.
9097
9098   *Steve Henson*
9099
9100 * Removed effectively defunct crypto/store from the build.
9101
9102   *Ben Laurie*
9103
9104 * Revamp of STACK to provide stronger type-checking. Still to come:
9105   TXT_DB, bsearch(?), OBJ_bsearch, qsort, CRYPTO_EX_DATA, ASN1_VALUE,
9106   ASN1_STRING, CONF_VALUE.
9107
9108   *Ben Laurie*
9109
9110 * Add a new SSL_MODE_RELEASE_BUFFERS mode flag to release unused buffer
9111   RAM on SSL connections.  This option can save about 34k per idle SSL.
9112
9113   *Nick Mathewson*
9114
9115 * Revamp of LHASH to provide stronger type-checking. Still to come:
9116   STACK, TXT_DB, bsearch, qsort.
9117
9118   *Ben Laurie*
9119
9120 * Initial support for Cryptographic Message Syntax (aka CMS) based
9121   on RFC3850, RFC3851 and RFC3852. New cms directory and cms utility,
9122   support for data, signedData, compressedData, digestedData and
9123   encryptedData, envelopedData types included. Scripts to check against
9124   RFC4134 examples draft and interop and consistency checks of many
9125   content types and variants.
9126
9127   *Steve Henson*
9128
9129 * Add options to enc utility to support use of zlib compression BIO.
9130
9131   *Steve Henson*
9132
9133 * Extend mk1mf to support importing of options and assembly language
9134   files from Configure script, currently only included in VC-WIN32.
9135   The assembly language rules can now optionally generate the source
9136   files from the associated perl scripts.
9137
9138   *Steve Henson*
9139
9140 * Implement remaining functionality needed to support GOST ciphersuites.
9141   Interop testing has been performed using CryptoPro implementations.
9142
9143   *Victor B. Wagner <vitus@cryptocom.ru>*
9144
9145 * s390x assembler pack.
9146
9147   *Andy Polyakov*
9148
9149 * ARMv4 assembler pack. ARMv4 refers to v4 and later ISA, not CPU
9150   "family."
9151
9152   *Andy Polyakov*
9153
9154 * Implement Opaque PRF Input TLS extension as specified in
9155   draft-rescorla-tls-opaque-prf-input-00.txt.  Since this is not an
9156   official specification yet and no extension type assignment by
9157   IANA exists, this extension (for now) will have to be explicitly
9158   enabled when building OpenSSL by providing the extension number
9159   to use.  For example, specify an option
9160
9161           -DTLSEXT_TYPE_opaque_prf_input=0x9527
9162
9163   to the "config" or "Configure" script to enable the extension,
9164   assuming extension number 0x9527 (which is a completely arbitrary
9165   and unofficial assignment based on the MD5 hash of the Internet
9166   Draft).  Note that by doing so, you potentially lose
9167   interoperability with other TLS implementations since these might
9168   be using the same extension number for other purposes.
9169
9170   SSL_set_tlsext_opaque_prf_input(ssl, src, len) is used to set the
9171   opaque PRF input value to use in the handshake.  This will create
9172   an internal copy of the length-'len' string at 'src', and will
9173   return non-zero for success.
9174
9175   To get more control and flexibility, provide a callback function
9176   by using
9177
9178           SSL_CTX_set_tlsext_opaque_prf_input_callback(ctx, cb)
9179           SSL_CTX_set_tlsext_opaque_prf_input_callback_arg(ctx, arg)
9180
9181   where
9182
9183           int (*cb)(SSL *, void *peerinput, size_t len, void *arg);
9184           void *arg;
9185
9186   Callback function 'cb' will be called in handshakes, and is
9187   expected to use SSL_set_tlsext_opaque_prf_input() as appropriate.
9188   Argument 'arg' is for application purposes (the value as given to
9189   SSL_CTX_set_tlsext_opaque_prf_input_callback_arg() will directly
9190   be provided to the callback function).  The callback function
9191   has to return non-zero to report success: usually 1 to use opaque
9192   PRF input just if possible, or 2 to enforce use of the opaque PRF
9193   input.  In the latter case, the library will abort the handshake
9194   if opaque PRF input is not successfully negotiated.
9195
9196   Arguments 'peerinput' and 'len' given to the callback function
9197   will always be NULL and 0 in the case of a client.  A server will
9198   see the client's opaque PRF input through these variables if
9199   available (NULL and 0 otherwise).  Note that if the server
9200   provides an opaque PRF input, the length must be the same as the
9201   length of the client's opaque PRF input.
9202
9203   Note that the callback function will only be called when creating
9204   a new session (session resumption can resume whatever was
9205   previously negotiated), and will not be called in SSL 2.0
9206   handshakes; thus, SSL_CTX_set_options(ctx, SSL_OP_NO_SSLv2) or
9207   SSL_set_options(ssl, SSL_OP_NO_SSLv2) is especially recommended
9208   for applications that need to enforce opaque PRF input.
9209
9210   *Bodo Moeller*
9211
9212 * Update ssl code to support digests other than SHA1+MD5 for handshake
9213   MAC.
9214
9215   *Victor B. Wagner <vitus@cryptocom.ru>*
9216
9217 * Add RFC4507 support to OpenSSL. This includes the corrections in
9218   RFC4507bis. The encrypted ticket format is an encrypted encoded
9219   SSL_SESSION structure, that way new session features are automatically
9220   supported.
9221
9222   If a client application caches session in an SSL_SESSION structure
9223   support is transparent because tickets are now stored in the encoded
9224   SSL_SESSION.
9225
9226   The SSL_CTX structure automatically generates keys for ticket
9227   protection in servers so again support should be possible
9228   with no application modification.
9229
9230   If a client or server wishes to disable RFC4507 support then the option
9231   SSL_OP_NO_TICKET can be set.
9232
9233   Add a TLS extension debugging callback to allow the contents of any client
9234   or server extensions to be examined.
9235
9236   This work was sponsored by Google.
9237
9238   *Steve Henson*
9239
9240 * Final changes to avoid use of pointer pointer casts in OpenSSL.
9241   OpenSSL should now compile cleanly on gcc 4.2
9242
9243   *Peter Hartley <pdh@utter.chaos.org.uk>, Steve Henson*
9244
9245 * Update SSL library to use new EVP_PKEY MAC API. Include generic MAC
9246   support including streaming MAC support: this is required for GOST
9247   ciphersuite support.
9248
9249   *Victor B. Wagner <vitus@cryptocom.ru>, Steve Henson*
9250
9251 * Add option -stream to use PKCS#7 streaming in smime utility. New
9252   function i2d_PKCS7_bio_stream() and PEM_write_PKCS7_bio_stream()
9253   to output in BER and PEM format.
9254
9255   *Steve Henson*
9256
9257 * Experimental support for use of HMAC via EVP_PKEY interface. This
9258   allows HMAC to be handled via the `EVP_DigestSign*()` interface. The
9259   EVP_PKEY "key" in this case is the HMAC key, potentially allowing
9260   ENGINE support for HMAC keys which are unextractable. New -mac and
9261   -macopt options to dgst utility.
9262
9263   *Steve Henson*
9264
9265 * New option -sigopt to dgst utility. Update dgst to use
9266   `EVP_Digest{Sign,Verify}*`. These two changes make it possible to use
9267   alternative signing parameters such as X9.31 or PSS in the dgst
9268   utility.
9269
9270   *Steve Henson*
9271
9272 * Change ssl_cipher_apply_rule(), the internal function that does
9273   the work each time a ciphersuite string requests enabling
9274   ("foo+bar"), moving ("+foo+bar"), disabling ("-foo+bar", or
9275   removing ("!foo+bar") a class of ciphersuites: Now it maintains
9276   the order of disabled ciphersuites such that those ciphersuites
9277   that most recently went from enabled to disabled not only stay
9278   in order with respect to each other, but also have higher priority
9279   than other disabled ciphersuites the next time ciphersuites are
9280   enabled again.
9281
9282   This means that you can now say, e.g., "PSK:-PSK:HIGH" to enable
9283   the same ciphersuites as with "HIGH" alone, but in a specific
9284   order where the PSK ciphersuites come first (since they are the
9285   most recently disabled ciphersuites when "HIGH" is parsed).
9286
9287   Also, change ssl_create_cipher_list() (using this new
9288   functionality) such that between otherwise identical
9289   ciphersuites, ephemeral ECDH is preferred over ephemeral DH in
9290   the default order.
9291
9292   *Bodo Moeller*
9293
9294 * Change ssl_create_cipher_list() so that it automatically
9295   arranges the ciphersuites in reasonable order before starting
9296   to process the rule string.  Thus, the definition for "DEFAULT"
9297   (SSL_DEFAULT_CIPHER_LIST) now is just "ALL:!aNULL:!eNULL", but
9298   remains equivalent to `"AES:ALL:!aNULL:!eNULL:+aECDH:+kRSA:+RC4:@STRENGTH"`.
9299   This makes it much easier to arrive at a reasonable default order
9300   in applications for which anonymous ciphers are OK (meaning
9301   that you can't actually use DEFAULT).
9302
9303   *Bodo Moeller; suggested by Victor Duchovni*
9304
9305 * Split the SSL/TLS algorithm mask (as used for ciphersuite string
9306   processing) into multiple integers instead of setting
9307   "SSL_MKEY_MASK" bits, "SSL_AUTH_MASK" bits, "SSL_ENC_MASK",
9308   "SSL_MAC_MASK", and "SSL_SSL_MASK" bits all in a single integer.
9309   (These masks as well as the individual bit definitions are hidden
9310   away into the non-exported interface ssl/ssl_locl.h, so this
9311   change to the definition of the SSL_CIPHER structure shouldn't
9312   affect applications.)  This give us more bits for each of these
9313   categories, so there is no longer a need to coagulate AES128 and
9314   AES256 into a single algorithm bit, and to coagulate Camellia128
9315   and Camellia256 into a single algorithm bit, which has led to all
9316   kinds of kludges.
9317
9318   Thus, among other things, the kludge introduced in 0.9.7m and
9319   0.9.8e for masking out AES256 independently of AES128 or masking
9320   out Camellia256 independently of AES256 is not needed here in 0.9.9.
9321
9322   With the change, we also introduce new ciphersuite aliases that
9323   so far were missing: "AES128", "AES256", "CAMELLIA128", and
9324   "CAMELLIA256".
9325
9326   *Bodo Moeller*
9327
9328 * Add support for dsa-with-SHA224 and dsa-with-SHA256.
9329   Use the leftmost N bytes of the signature input if the input is
9330   larger than the prime q (with N being the size in bytes of q).
9331
9332   *Nils Larsch*
9333
9334 * Very *very* experimental PKCS#7 streaming encoder support. Nothing uses
9335   it yet and it is largely untested.
9336
9337   *Steve Henson*
9338
9339 * Add support for the ecdsa-with-SHA224/256/384/512 signature types.
9340
9341   *Nils Larsch*
9342
9343 * Initial incomplete changes to avoid need for function casts in OpenSSL
9344   some compilers (gcc 4.2 and later) reject their use. Safestack is
9345   reimplemented.  Update ASN1 to avoid use of legacy functions.
9346
9347   *Steve Henson*
9348
9349 * Win32/64 targets are linked with Winsock2.
9350
9351   *Andy Polyakov*
9352
9353 * Add an X509_CRL_METHOD structure to allow CRL processing to be redirected
9354   to external functions. This can be used to increase CRL handling
9355   efficiency especially when CRLs are very large by (for example) storing
9356   the CRL revoked certificates in a database.
9357
9358   *Steve Henson*
9359
9360 * Overhaul of by_dir code. Add support for dynamic loading of CRLs so
9361   new CRLs added to a directory can be used. New command line option
9362   -verify_return_error to s_client and s_server. This causes real errors
9363   to be returned by the verify callback instead of carrying on no matter
9364   what. This reflects the way a "real world" verify callback would behave.
9365
9366   *Steve Henson*
9367
9368 * GOST engine, supporting several GOST algorithms and public key formats.
9369   Kindly donated by Cryptocom.
9370
9371   *Cryptocom*
9372
9373 * Partial support for Issuing Distribution Point CRL extension. CRLs
9374   partitioned by DP are handled but no indirect CRL or reason partitioning
9375   (yet). Complete overhaul of CRL handling: now the most suitable CRL is
9376   selected via a scoring technique which handles IDP and AKID in CRLs.
9377
9378   *Steve Henson*
9379
9380 * New X509_STORE_CTX callbacks lookup_crls() and lookup_certs() which
9381   will ultimately be used for all verify operations: this will remove the
9382   X509_STORE dependency on certificate verification and allow alternative
9383   lookup methods.  X509_STORE based implementations of these two callbacks.
9384
9385   *Steve Henson*
9386
9387 * Allow multiple CRLs to exist in an X509_STORE with matching issuer names.
9388   Modify get_crl() to find a valid (unexpired) CRL if possible.
9389
9390   *Steve Henson*
9391
9392 * New function X509_CRL_match() to check if two CRLs are identical. Normally
9393   this would be called X509_CRL_cmp() but that name is already used by
9394   a function that just compares CRL issuer names. Cache several CRL
9395   extensions in X509_CRL structure and cache CRLDP in X509.
9396
9397   *Steve Henson*
9398
9399 * Store a "canonical" representation of X509_NAME structure (ASN1 Name)
9400   this maps equivalent X509_NAME structures into a consistent structure.
9401   Name comparison can then be performed rapidly using memcmp().
9402
9403   *Steve Henson*
9404
9405 * Non-blocking OCSP request processing. Add -timeout option to ocsp
9406   utility.
9407
9408   *Steve Henson*
9409
9410 * Allow digests to supply their own micalg string for S/MIME type using
9411   the ctrl EVP_MD_CTRL_MICALG.
9412
9413   *Steve Henson*
9414
9415 * During PKCS7 signing pass the PKCS7 SignerInfo structure to the
9416   EVP_PKEY_METHOD before and after signing via the EVP_PKEY_CTRL_PKCS7_SIGN
9417   ctrl. It can then customise the structure before and/or after signing
9418   if necessary.
9419
9420   *Steve Henson*
9421
9422 * New function OBJ_add_sigid() to allow application defined signature OIDs
9423   to be added to OpenSSLs internal tables. New function OBJ_sigid_free()
9424   to free up any added signature OIDs.
9425
9426   *Steve Henson*
9427
9428 * New functions EVP_CIPHER_do_all(), EVP_CIPHER_do_all_sorted(),
9429   EVP_MD_do_all() and EVP_MD_do_all_sorted() to enumerate internal
9430   digest and cipher tables. New options added to openssl utility:
9431   list-message-digest-algorithms and list-cipher-algorithms.
9432
9433   *Steve Henson*
9434
9435 * Change the array representation of binary polynomials: the list
9436   of degrees of non-zero coefficients is now terminated with -1.
9437   Previously it was terminated with 0, which was also part of the
9438   value; thus, the array representation was not applicable to
9439   polynomials where t^0 has coefficient zero.  This change makes
9440   the array representation useful in a more general context.
9441
9442   *Douglas Stebila*
9443
9444 * Various modifications and fixes to SSL/TLS cipher string
9445   handling.  For ECC, the code now distinguishes between fixed ECDH
9446   with RSA certificates on the one hand and with ECDSA certificates
9447   on the other hand, since these are separate ciphersuites.  The
9448   unused code for Fortezza ciphersuites has been removed.
9449
9450   For consistency with EDH, ephemeral ECDH is now called "EECDH"
9451   (not "ECDHE").  For consistency with the code for DH
9452   certificates, use of ECDH certificates is now considered ECDH
9453   authentication, not RSA or ECDSA authentication (the latter is
9454   merely the CA's signing algorithm and not actively used in the
9455   protocol).
9456
9457   The temporary ciphersuite alias "ECCdraft" is no longer
9458   available, and ECC ciphersuites are no longer excluded from "ALL"
9459   and "DEFAULT".  The following aliases now exist for RFC 4492
9460   ciphersuites, most of these by analogy with the DH case:
9461
9462           kECDHr   - ECDH cert, signed with RSA
9463           kECDHe   - ECDH cert, signed with ECDSA
9464           kECDH    - ECDH cert (signed with either RSA or ECDSA)
9465           kEECDH   - ephemeral ECDH
9466           ECDH     - ECDH cert or ephemeral ECDH
9467
9468           aECDH    - ECDH cert
9469           aECDSA   - ECDSA cert
9470           ECDSA    - ECDSA cert
9471
9472           AECDH    - anonymous ECDH
9473           EECDH    - non-anonymous ephemeral ECDH (equivalent to "kEECDH:-AECDH")
9474
9475   *Bodo Moeller*
9476
9477 * Add additional S/MIME capabilities for AES and GOST ciphers if supported.
9478   Use correct micalg parameters depending on digest(s) in signed message.
9479
9480   *Steve Henson*
9481
9482 * Add engine support for EVP_PKEY_ASN1_METHOD. Add functions to process
9483   an ENGINE asn1 method. Support ENGINE lookups in the ASN1 code.
9484
9485   *Steve Henson*
9486
9487 * Initial engine support for EVP_PKEY_METHOD. New functions to permit
9488   an engine to register a method. Add ENGINE lookups for methods and
9489   functional reference processing.
9490
9491   *Steve Henson*
9492
9493 * New functions `EVP_Digest{Sign,Verify)*`. These are enhanced versions of
9494   `EVP_{Sign,Verify}*` which allow an application to customise the signature
9495   process.
9496
9497   *Steve Henson*
9498
9499 * New -resign option to smime utility. This adds one or more signers
9500   to an existing PKCS#7 signedData structure. Also -md option to use an
9501   alternative message digest algorithm for signing.
9502
9503   *Steve Henson*
9504
9505 * Tidy up PKCS#7 routines and add new functions to make it easier to
9506   create PKCS7 structures containing multiple signers. Update smime
9507   application to support multiple signers.
9508
9509   *Steve Henson*
9510
9511 * New -macalg option to pkcs12 utility to allow setting of an alternative
9512   digest MAC.
9513
9514   *Steve Henson*
9515
9516 * Initial support for PKCS#5 v2.0 PRFs other than default SHA1 HMAC.
9517   Reorganize PBE internals to lookup from a static table using NIDs,
9518   add support for HMAC PBE OID translation. Add a EVP_CIPHER ctrl:
9519   EVP_CTRL_PBE_PRF_NID this allows a cipher to specify an alternative
9520   PRF which will be automatically used with PBES2.
9521
9522   *Steve Henson*
9523
9524 * Replace the algorithm specific calls to generate keys in "req" with the
9525   new API.
9526
9527   *Steve Henson*
9528
9529 * Update PKCS#7 enveloped data routines to use new API. This is now
9530   supported by any public key method supporting the encrypt operation. A
9531   ctrl is added to allow the public key algorithm to examine or modify
9532   the PKCS#7 RecipientInfo structure if it needs to: for RSA this is
9533   a no op.
9534
9535   *Steve Henson*
9536
9537 * Add a ctrl to asn1 method to allow a public key algorithm to express
9538   a default digest type to use. In most cases this will be SHA1 but some
9539   algorithms (such as GOST) need to specify an alternative digest. The
9540   return value indicates how strong the preference is 1 means optional and
9541   2 is mandatory (that is it is the only supported type). Modify
9542   ASN1_item_sign() to accept a NULL digest argument to indicate it should
9543   use the default md. Update openssl utilities to use the default digest
9544   type for signing if it is not explicitly indicated.
9545
9546   *Steve Henson*
9547
9548 * Use OID cross reference table in ASN1_sign() and ASN1_verify(). New
9549   EVP_MD flag EVP_MD_FLAG_PKEY_METHOD_SIGNATURE. This uses the relevant
9550   signing method from the key type. This effectively removes the link
9551   between digests and public key types.
9552
9553   *Steve Henson*
9554
9555 * Add an OID cross reference table and utility functions. Its purpose is to
9556   translate between signature OIDs such as SHA1WithrsaEncryption and SHA1,
9557   rsaEncryption. This will allow some of the algorithm specific hackery
9558   needed to use the correct OID to be removed.
9559
9560   *Steve Henson*
9561
9562 * Remove algorithm specific dependencies when setting PKCS7_SIGNER_INFO
9563   structures for PKCS7_sign(). They are now set up by the relevant public
9564   key ASN1 method.
9565
9566   *Steve Henson*
9567
9568 * Add provisional EC pkey method with support for ECDSA and ECDH.
9569
9570   *Steve Henson*
9571
9572 * Add support for key derivation (agreement) in the API, DH method and
9573   pkeyutl.
9574
9575   *Steve Henson*
9576
9577 * Add DSA pkey method and DH pkey methods, extend DH ASN1 method to support
9578   public and private key formats. As a side effect these add additional
9579   command line functionality not previously available: DSA signatures can be
9580   generated and verified using pkeyutl and DH key support and generation in
9581   pkey, genpkey.
9582
9583   *Steve Henson*
9584
9585 * BeOS support.
9586
9587   *Oliver Tappe <zooey@hirschkaefer.de>*
9588
9589 * New make target "install_html_docs" installs HTML renditions of the
9590   manual pages.
9591
9592   *Oliver Tappe <zooey@hirschkaefer.de>*
9593
9594 * New utility "genpkey" this is analogous to "genrsa" etc except it can
9595   generate keys for any algorithm. Extend and update EVP_PKEY_METHOD to
9596   support key and parameter generation and add initial key generation
9597   functionality for RSA.
9598
9599   *Steve Henson*
9600
9601 * Add functions for main EVP_PKEY_method operations. The undocumented
9602   functions `EVP_PKEY_{encrypt,decrypt}` have been renamed to
9603   `EVP_PKEY_{encrypt,decrypt}_old`.
9604
9605   *Steve Henson*
9606
9607 * Initial definitions for EVP_PKEY_METHOD. This will be a high level public
9608   key API, doesn't do much yet.
9609
9610   *Steve Henson*
9611
9612 * New function EVP_PKEY_asn1_get0_info() to retrieve information about
9613   public key algorithms. New option to openssl utility:
9614   "list-public-key-algorithms" to print out info.
9615
9616   *Steve Henson*
9617
9618 * Implement the Supported Elliptic Curves Extension for
9619   ECC ciphersuites from draft-ietf-tls-ecc-12.txt.
9620
9621   *Douglas Stebila*
9622
9623 * Don't free up OIDs in OBJ_cleanup() if they are in use by EVP_MD or
9624   EVP_CIPHER structures to avoid later problems in EVP_cleanup().
9625
9626   *Steve Henson*
9627
9628 * New utilities pkey and pkeyparam. These are similar to algorithm specific
9629   utilities such as rsa, dsa, dsaparam etc except they process any key
9630   type.
9631
9632   *Steve Henson*
9633
9634 * Transfer public key printing routines to EVP_PKEY_ASN1_METHOD. New
9635   functions EVP_PKEY_print_public(), EVP_PKEY_print_private(),
9636   EVP_PKEY_print_param() to print public key data from an EVP_PKEY
9637   structure.
9638
9639   *Steve Henson*
9640
9641 * Initial support for pluggable public key ASN1.
9642   De-spaghettify the public key ASN1 handling. Move public and private
9643   key ASN1 handling to a new EVP_PKEY_ASN1_METHOD structure. Relocate
9644   algorithm specific handling to a single module within the relevant
9645   algorithm directory. Add functions to allow (near) opaque processing
9646   of public and private key structures.
9647
9648   *Steve Henson*
9649
9650 * Implement the Supported Point Formats Extension for
9651   ECC ciphersuites from draft-ietf-tls-ecc-12.txt.
9652
9653   *Douglas Stebila*
9654
9655 * Add initial support for RFC 4279 PSK TLS ciphersuites. Add members
9656   for the psk identity [hint] and the psk callback functions to the
9657   SSL_SESSION, SSL and SSL_CTX structure.
9658
9659   New ciphersuites:
9660           PSK-RC4-SHA, PSK-3DES-EDE-CBC-SHA, PSK-AES128-CBC-SHA,
9661           PSK-AES256-CBC-SHA
9662
9663   New functions:
9664           SSL_CTX_use_psk_identity_hint
9665           SSL_get_psk_identity_hint
9666           SSL_get_psk_identity
9667           SSL_use_psk_identity_hint
9668
9669   *Mika Kousa and Pasi Eronen of Nokia Corporation*
9670
9671 * Add RFC 3161 compliant time stamp request creation, response generation
9672   and response verification functionality.
9673
9674   *Zoltán Glózik <zglozik@opentsa.org>, The OpenTSA Project*
9675
9676 * Add initial support for TLS extensions, specifically for the server_name
9677   extension so far.  The SSL_SESSION, SSL_CTX, and SSL data structures now
9678   have new members for a host name.  The SSL data structure has an
9679   additional member `SSL_CTX *initial_ctx` so that new sessions can be
9680   stored in that context to allow for session resumption, even after the
9681   SSL has been switched to a new SSL_CTX in reaction to a client's
9682   server_name extension.
9683
9684   New functions (subject to change):
9685
9686           SSL_get_servername()
9687           SSL_get_servername_type()
9688           SSL_set_SSL_CTX()
9689
9690   New CTRL codes and macros (subject to change):
9691
9692           SSL_CTRL_SET_TLSEXT_SERVERNAME_CB
9693                               - SSL_CTX_set_tlsext_servername_callback()
9694           SSL_CTRL_SET_TLSEXT_SERVERNAME_ARG
9695                                    - SSL_CTX_set_tlsext_servername_arg()
9696           SSL_CTRL_SET_TLSEXT_HOSTNAME           - SSL_set_tlsext_host_name()
9697
9698   openssl s_client has a new '-servername ...' option.
9699
9700   openssl s_server has new options '-servername_host ...', '-cert2 ...',
9701   '-key2 ...', '-servername_fatal' (subject to change).  This allows
9702   testing the HostName extension for a specific single host name ('-cert'
9703   and '-key' remain fallbacks for handshakes without HostName
9704   negotiation).  If the unrecognized_name alert has to be sent, this by
9705   default is a warning; it becomes fatal with the '-servername_fatal'
9706   option.
9707
9708   *Peter Sylvester,  Remy Allais, Christophe Renou*
9709
9710 * Whirlpool hash implementation is added.
9711
9712   *Andy Polyakov*
9713
9714 * BIGNUM code on 64-bit SPARCv9 targets is switched from bn(64,64) to
9715   bn(64,32). Because of instruction set limitations it doesn't have
9716   any negative impact on performance. This was done mostly in order
9717   to make it possible to share assembler modules, such as bn_mul_mont
9718   implementations, between 32- and 64-bit builds without hassle.
9719
9720   *Andy Polyakov*
9721
9722 * Move code previously exiled into file crypto/ec/ec2_smpt.c
9723   to ec2_smpl.c, and no longer require the OPENSSL_EC_BIN_PT_COMP
9724   macro.
9725
9726   *Bodo Moeller*
9727
9728 * New candidate for BIGNUM assembler implementation, bn_mul_mont,
9729   dedicated Montgomery multiplication procedure, is introduced.
9730   BN_MONT_CTX is modified to allow bn_mul_mont to reach for higher
9731   "64-bit" performance on certain 32-bit targets.
9732
9733   *Andy Polyakov*
9734
9735 * New option SSL_OP_NO_COMP to disable use of compression selectively
9736   in SSL structures. New SSL ctrl to set maximum send fragment size.
9737   Save memory by setting the I/O buffer sizes dynamically instead of
9738   using the maximum available value.
9739
9740   *Steve Henson*
9741
9742 * New option -V for 'openssl ciphers'. This prints the ciphersuite code
9743   in addition to the text details.
9744
9745   *Bodo Moeller*
9746
9747 * Very, very preliminary EXPERIMENTAL support for printing of general
9748   ASN1 structures. This currently produces rather ugly output and doesn't
9749   handle several customised structures at all.
9750
9751   *Steve Henson*
9752
9753 * Integrated support for PVK file format and some related formats such
9754   as MS PUBLICKEYBLOB and PRIVATEKEYBLOB. Command line switches to support
9755   these in the 'rsa' and 'dsa' utilities.
9756
9757   *Steve Henson*
9758
9759 * Support for PKCS#1 RSAPublicKey format on rsa utility command line.
9760
9761   *Steve Henson*
9762
9763 * Remove the ancient ASN1_METHOD code. This was only ever used in one
9764   place for the (very old) "NETSCAPE" format certificates which are now
9765   handled using new ASN1 code equivalents.
9766
9767   *Steve Henson*
9768
9769 * Let the TLSv1_method() etc. functions return a 'const' SSL_METHOD
9770   pointer and make the SSL_METHOD parameter in SSL_CTX_new,
9771   SSL_CTX_set_ssl_version and SSL_set_ssl_method 'const'.
9772
9773   *Nils Larsch*
9774
9775 * Modify CRL distribution points extension code to print out previously
9776   unsupported fields. Enhance extension setting code to allow setting of
9777   all fields.
9778
9779   *Steve Henson*
9780
9781 * Add print and set support for Issuing Distribution Point CRL extension.
9782
9783   *Steve Henson*
9784
9785 * Change 'Configure' script to enable Camellia by default.
9786
9787   *NTT*
9788
9789OpenSSL 0.9.x
9790-------------
9791
9792### Changes between 0.9.8m and 0.9.8n [24 Mar 2010]
9793
9794 * When rejecting SSL/TLS records due to an incorrect version number, never
9795   update s->server with a new major version number.  As of
9796   - OpenSSL 0.9.8m if 'short' is a 16-bit type,
9797   - OpenSSL 0.9.8f if 'short' is longer than 16 bits,
9798   the previous behavior could result in a read attempt at NULL when
9799   receiving specific incorrect SSL/TLS records once record payload
9800   protection is active.  ([CVE-2010-0740])
9801
9802   *Bodo Moeller, Adam Langley <agl@chromium.org>*
9803
9804 * Fix for CVE-2010-0433 where some kerberos enabled versions of OpenSSL
9805   could be crashed if the relevant tables were not present (e.g. chrooted).
9806
9807   *Tomas Hoger <thoger@redhat.com>*
9808
9809### Changes between 0.9.8l and 0.9.8m [25 Feb 2010]
9810
9811 * Always check bn_wexpand() return values for failure.  ([CVE-2009-3245])
9812
9813   *Martin Olsson, Neel Mehta*
9814
9815 * Fix X509_STORE locking: Every 'objs' access requires a lock (to
9816   accommodate for stack sorting, always a write lock!).
9817
9818   *Bodo Moeller*
9819
9820 * On some versions of WIN32 Heap32Next is very slow. This can cause
9821   excessive delays in the RAND_poll(): over a minute. As a workaround
9822   include a time check in the inner Heap32Next loop too.
9823
9824   *Steve Henson*
9825
9826 * The code that handled flushing of data in SSL/TLS originally used the
9827   BIO_CTRL_INFO ctrl to see if any data was pending first. This caused
9828   the problem outlined in PR#1949. The fix suggested there however can
9829   trigger problems with buggy BIO_CTRL_WPENDING (e.g. some versions
9830   of Apache). So instead simplify the code to flush unconditionally.
9831   This should be fine since flushing with no data to flush is a no op.
9832
9833   *Steve Henson*
9834
9835 * Handle TLS versions 2.0 and later properly and correctly use the
9836   highest version of TLS/SSL supported. Although TLS >= 2.0 is some way
9837   off ancient servers have a habit of sticking around for a while...
9838
9839   *Steve Henson*
9840
9841 * Modify compression code so it frees up structures without using the
9842   ex_data callbacks. This works around a problem where some applications
9843   call CRYPTO_cleanup_all_ex_data() before application exit (e.g. when
9844   restarting) then use compression (e.g. SSL with compression) later.
9845   This results in significant per-connection memory leaks and
9846   has caused some security issues including CVE-2008-1678 and
9847   CVE-2009-4355.
9848
9849   *Steve Henson*
9850
9851 * Constify crypto/cast (i.e., <openssl/cast.h>): a CAST_KEY doesn't
9852   change when encrypting or decrypting.
9853
9854   *Bodo Moeller*
9855
9856 * Add option SSL_OP_LEGACY_SERVER_CONNECT which will allow clients to
9857   connect and renegotiate with servers which do not support RI.
9858   Until RI is more widely deployed this option is enabled by default.
9859
9860   *Steve Henson*
9861
9862 * Add "missing" ssl ctrls to clear options and mode.
9863
9864   *Steve Henson*
9865
9866 * If client attempts to renegotiate and doesn't support RI respond with
9867   a no_renegotiation alert as required by RFC5746.  Some renegotiating
9868   TLS clients will continue a connection gracefully when they receive
9869   the alert. Unfortunately OpenSSL mishandled this alert and would hang
9870   waiting for a server hello which it will never receive. Now we treat a
9871   received no_renegotiation alert as a fatal error. This is because
9872   applications requesting a renegotiation might well expect it to succeed
9873   and would have no code in place to handle the server denying it so the
9874   only safe thing to do is to terminate the connection.
9875
9876   *Steve Henson*
9877
9878 * Add ctrl macro SSL_get_secure_renegotiation_support() which returns 1 if
9879   peer supports secure renegotiation and 0 otherwise. Print out peer
9880   renegotiation support in s_client/s_server.
9881
9882   *Steve Henson*
9883
9884 * Replace the highly broken and deprecated SPKAC certification method with
9885   the updated NID creation version. This should correctly handle UTF8.
9886
9887   *Steve Henson*
9888
9889 * Implement RFC5746. Re-enable renegotiation but require the extension
9890   as needed. Unfortunately, SSL3_FLAGS_ALLOW_UNSAFE_LEGACY_RENEGOTIATION
9891   turns out to be a bad idea. It has been replaced by
9892   SSL_OP_ALLOW_UNSAFE_LEGACY_RENEGOTIATION which can be set with
9893   SSL_CTX_set_options(). This is really not recommended unless you
9894   know what you are doing.
9895
9896   *Eric Rescorla <ekr@networkresonance.com>, Ben Laurie, Steve Henson*
9897
9898 * Fixes to stateless session resumption handling. Use initial_ctx when
9899   issuing and attempting to decrypt tickets in case it has changed during
9900   servername handling. Use a non-zero length session ID when attempting
9901   stateless session resumption: this makes it possible to determine if
9902   a resumption has occurred immediately after receiving server hello
9903   (several places in OpenSSL subtly assume this) instead of later in
9904   the handshake.
9905
9906   *Steve Henson*
9907
9908 * The functions ENGINE_ctrl(), OPENSSL_isservice(),
9909   CMS_get1_RecipientRequest() and RAND_bytes() can return <=0 on error
9910   fixes for a few places where the return code is not checked
9911   correctly.
9912
9913   *Julia Lawall <julia@diku.dk>*
9914
9915 * Add --strict-warnings option to Configure script to include devteam
9916   warnings in other configurations.
9917
9918   *Steve Henson*
9919
9920 * Add support for --libdir option and LIBDIR variable in makefiles. This
9921   makes it possible to install openssl libraries in locations which
9922   have names other than "lib", for example "/usr/lib64" which some
9923   systems need.
9924
9925   *Steve Henson, based on patch from Jeremy Utley*
9926
9927 * Don't allow the use of leading 0x80 in OIDs. This is a violation of
9928   X690 8.9.12 and can produce some misleading textual output of OIDs.
9929
9930   *Steve Henson, reported by Dan Kaminsky*
9931
9932 * Delete MD2 from algorithm tables. This follows the recommendation in
9933   several standards that it is not used in new applications due to
9934   several cryptographic weaknesses. For binary compatibility reasons
9935   the MD2 API is still compiled in by default.
9936
9937   *Steve Henson*
9938
9939 * Add compression id to {d2i,i2d}_SSL_SESSION so it is correctly saved
9940   and restored.
9941
9942   *Steve Henson*
9943
9944 * Rename uni2asc and asc2uni functions to OPENSSL_uni2asc and
9945   OPENSSL_asc2uni conditionally on Netware platforms to avoid a name
9946   clash.
9947
9948   *Guenter <lists@gknw.net>*
9949
9950 * Fix the server certificate chain building code to use X509_verify_cert(),
9951   it used to have an ad-hoc builder which was unable to cope with anything
9952   other than a simple chain.
9953
9954   *David Woodhouse <dwmw2@infradead.org>, Steve Henson*
9955
9956 * Don't check self signed certificate signatures in X509_verify_cert()
9957   by default (a flag can override this): it just wastes time without
9958   adding any security. As a useful side effect self signed root CAs
9959   with non-FIPS digests are now usable in FIPS mode.
9960
9961   *Steve Henson*
9962
9963 * In dtls1_process_out_of_seq_message() the check if the current message
9964   is already buffered was missing. For every new message was memory
9965   allocated, allowing an attacker to perform an denial of service attack
9966   with sending out of seq handshake messages until there is no memory
9967   left. Additionally every future message was buffered, even if the
9968   sequence number made no sense and would be part of another handshake.
9969   So only messages with sequence numbers less than 10 in advance will be
9970   buffered.  ([CVE-2009-1378])
9971
9972   *Robin Seggelmann, discovered by Daniel Mentz*
9973
9974 * Records are buffered if they arrive with a future epoch to be
9975   processed after finishing the corresponding handshake. There is
9976   currently no limitation to this buffer allowing an attacker to perform
9977   a DOS attack with sending records with future epochs until there is no
9978   memory left. This patch adds the pqueue_size() function to determine
9979   the size of a buffer and limits the record buffer to 100 entries.
9980   ([CVE-2009-1377])
9981
9982   *Robin Seggelmann, discovered by Daniel Mentz*
9983
9984 * Keep a copy of frag->msg_header.frag_len so it can be used after the
9985   parent structure is freed.  ([CVE-2009-1379])
9986
9987   *Daniel Mentz*
9988
9989 * Handle non-blocking I/O properly in SSL_shutdown() call.
9990
9991   *Darryl Miles <darryl-mailinglists@netbauds.net>*
9992
9993 * Add `2.5.4.*` OIDs
9994
9995   *Ilya O. <vrghost@gmail.com>*
9996
9997### Changes between 0.9.8k and 0.9.8l  [5 Nov 2009]
9998
9999 * Disable renegotiation completely - this fixes a severe security
10000   problem ([CVE-2009-3555]) at the cost of breaking all
10001   renegotiation. Renegotiation can be re-enabled by setting
10002   SSL3_FLAGS_ALLOW_UNSAFE_LEGACY_RENEGOTIATION in s3->flags at
10003   run-time. This is really not recommended unless you know what
10004   you're doing.
10005
10006   *Ben Laurie*
10007
10008### Changes between 0.9.8j and 0.9.8k  [25 Mar 2009]
10009
10010 * Don't set val to NULL when freeing up structures, it is freed up by
10011   underlying code. If `sizeof(void *) > sizeof(long)` this can result in
10012   zeroing past the valid field. ([CVE-2009-0789])
10013
10014   *Paolo Ganci <Paolo.Ganci@AdNovum.CH>*
10015
10016 * Fix bug where return value of CMS_SignerInfo_verify_content() was not
10017   checked correctly. This would allow some invalid signed attributes to
10018   appear to verify correctly. ([CVE-2009-0591])
10019
10020   *Ivan Nestlerode <inestlerode@us.ibm.com>*
10021
10022 * Reject UniversalString and BMPString types with invalid lengths. This
10023   prevents a crash in ASN1_STRING_print_ex() which assumes the strings have
10024   a legal length. ([CVE-2009-0590])
10025
10026   *Steve Henson*
10027
10028 * Set S/MIME signing as the default purpose rather than setting it
10029   unconditionally. This allows applications to override it at the store
10030   level.
10031
10032   *Steve Henson*
10033
10034 * Permit restricted recursion of ASN1 strings. This is needed in practice
10035   to handle some structures.
10036
10037   *Steve Henson*
10038
10039 * Improve efficiency of mem_gets: don't search whole buffer each time
10040   for a '\n'
10041
10042   *Jeremy Shapiro <jnshapir@us.ibm.com>*
10043
10044 * New -hex option for openssl rand.
10045
10046   *Matthieu Herrb*
10047
10048 * Print out UTF8String and NumericString when parsing ASN1.
10049
10050   *Steve Henson*
10051
10052 * Support NumericString type for name components.
10053
10054   *Steve Henson*
10055
10056 * Allow CC in the environment to override the automatically chosen
10057   compiler. Note that nothing is done to ensure flags work with the
10058   chosen compiler.
10059
10060   *Ben Laurie*
10061
10062### Changes between 0.9.8i and 0.9.8j  [07 Jan 2009]
10063
10064 * Properly check EVP_VerifyFinal() and similar return values
10065   ([CVE-2008-5077]).
10066
10067   *Ben Laurie, Bodo Moeller, Google Security Team*
10068
10069 * Enable TLS extensions by default.
10070
10071   *Ben Laurie*
10072
10073 * Allow the CHIL engine to be loaded, whether the application is
10074   multithreaded or not. (This does not release the developer from the
10075   obligation to set up the dynamic locking callbacks.)
10076
10077   *Sander Temme <sander@temme.net>*
10078
10079 * Use correct exit code if there is an error in dgst command.
10080
10081   *Steve Henson; problem pointed out by Roland Dirlewanger*
10082
10083 * Tweak Configure so that you need to say "experimental-jpake" to enable
10084   JPAKE, and need to use -DOPENSSL_EXPERIMENTAL_JPAKE in applications.
10085
10086   *Bodo Moeller*
10087
10088 * Add experimental JPAKE support, including demo authentication in
10089   s_client and s_server.
10090
10091   *Ben Laurie*
10092
10093 * Set the comparison function in v3_addr_canonize().
10094
10095   *Rob Austein <sra@hactrn.net>*
10096
10097 * Add support for XMPP STARTTLS in s_client.
10098
10099   *Philip Paeps <philip@freebsd.org>*
10100
10101 * Change the server-side SSL_OP_NETSCAPE_REUSE_CIPHER_CHANGE_BUG behavior
10102   to ensure that even with this option, only ciphersuites in the
10103   server's preference list will be accepted.  (Note that the option
10104   applies only when resuming a session, so the earlier behavior was
10105   just about the algorithm choice for symmetric cryptography.)
10106
10107   *Bodo Moeller*
10108
10109### Changes between 0.9.8h and 0.9.8i  [15 Sep 2008]
10110
10111 * Fix NULL pointer dereference if a DTLS server received
10112   ChangeCipherSpec as first record ([CVE-2009-1386]).
10113
10114   *PR #1679*
10115
10116 * Fix a state transition in s3_srvr.c and d1_srvr.c
10117   (was using SSL3_ST_CW_CLNT_HELLO_B, should be `..._ST_SW_SRVR_...`).
10118
10119   *Nagendra Modadugu*
10120
10121 * The fix in 0.9.8c that supposedly got rid of unsafe
10122   double-checked locking was incomplete for RSA blinding,
10123   addressing just one layer of what turns out to have been
10124   doubly unsafe triple-checked locking.
10125
10126   So now fix this for real by retiring the MONT_HELPER macro
10127   in crypto/rsa/rsa_eay.c.
10128
10129   *Bodo Moeller; problem pointed out by Marius Schilder*
10130
10131 * Various precautionary measures:
10132
10133   - Avoid size_t integer overflow in HASH_UPDATE (md32_common.h).
10134
10135   - Avoid a buffer overflow in d2i_SSL_SESSION() (ssl_asn1.c).
10136     (NB: This would require knowledge of the secret session ticket key
10137     to exploit, in which case you'd be SOL either way.)
10138
10139   - Change bn_nist.c so that it will properly handle input BIGNUMs
10140     outside the expected range.
10141
10142   - Enforce the 'num' check in BN_div() (bn_div.c) for non-BN_DEBUG
10143     builds.
10144
10145   *Neel Mehta, Bodo Moeller*
10146
10147 * Allow engines to be "soft loaded" - i.e. optionally don't die if
10148   the load fails. Useful for distros.
10149
10150   *Ben Laurie and the FreeBSD team*
10151
10152 * Add support for Local Machine Keyset attribute in PKCS#12 files.
10153
10154   *Steve Henson*
10155
10156 * Fix BN_GF2m_mod_arr() top-bit cleanup code.
10157
10158   *Huang Ying*
10159
10160 * Expand ENGINE to support engine supplied SSL client certificate functions.
10161
10162   This work was sponsored by Logica.
10163
10164   *Steve Henson*
10165
10166 * Add CryptoAPI ENGINE to support use of RSA and DSA keys held in Windows
10167   keystores. Support for SSL/TLS client authentication too.
10168   Not compiled unless enable-capieng specified to Configure.
10169
10170   This work was sponsored by Logica.
10171
10172   *Steve Henson*
10173
10174 * Fix bug in X509_ATTRIBUTE creation: don't set attribute using
10175   ASN1_TYPE_set1 if MBSTRING flag set. This bug would crash certain
10176   attribute creation routines such as certificate requests and PKCS#12
10177   files.
10178
10179   *Steve Henson*
10180
10181### Changes between 0.9.8g and 0.9.8h  [28 May 2008]
10182
10183 * Fix flaw if 'Server Key exchange message' is omitted from a TLS
10184   handshake which could lead to a client crash as found using the
10185   Codenomicon TLS test suite ([CVE-2008-1672])
10186
10187   *Steve Henson, Mark Cox*
10188
10189 * Fix double free in TLS server name extensions which could lead to
10190   a remote crash found by Codenomicon TLS test suite ([CVE-2008-0891])
10191
10192   *Joe Orton*
10193
10194 * Clear error queue in SSL_CTX_use_certificate_chain_file()
10195
10196   Clear the error queue to ensure that error entries left from
10197   older function calls do not interfere with the correct operation.
10198
10199   *Lutz Jaenicke, Erik de Castro Lopo*
10200
10201 * Remove root CA certificates of commercial CAs:
10202
10203   The OpenSSL project does not recommend any specific CA and does not
10204   have any policy with respect to including or excluding any CA.
10205   Therefore it does not make any sense to ship an arbitrary selection
10206   of root CA certificates with the OpenSSL software.
10207
10208   *Lutz Jaenicke*
10209
10210 * RSA OAEP patches to fix two separate invalid memory reads.
10211   The first one involves inputs when 'lzero' is greater than
10212   'SHA_DIGEST_LENGTH' (it would read about SHA_DIGEST_LENGTH bytes
10213   before the beginning of from). The second one involves inputs where
10214   the 'db' section contains nothing but zeroes (there is a one-byte
10215   invalid read after the end of 'db').
10216
10217   *Ivan Nestlerode <inestlerode@us.ibm.com>*
10218
10219 * Partial backport from 0.9.9-dev:
10220
10221   Introduce bn_mul_mont (dedicated Montgomery multiplication
10222   procedure) as a candidate for BIGNUM assembler implementation.
10223   While 0.9.9-dev uses assembler for various architectures, only
10224   x86_64 is available by default here in the 0.9.8 branch, and
10225   32-bit x86 is available through a compile-time setting.
10226
10227   To try the 32-bit x86 assembler implementation, use Configure
10228   option "enable-montasm" (which exists only for this backport).
10229
10230   As "enable-montasm" for 32-bit x86 disclaims code stability
10231   anyway, in this constellation we activate additional code
10232   backported from 0.9.9-dev for further performance improvements,
10233   namely BN_from_montgomery_word.  (To enable this otherwise,
10234   e.g. x86_64, try `-DMONT_FROM_WORD___NON_DEFAULT_0_9_8_BUILD`.)
10235
10236   *Andy Polyakov (backport partially by Bodo Moeller)*
10237
10238 * Add TLS session ticket callback. This allows an application to set
10239   TLS ticket cipher and HMAC keys rather than relying on hardcoded fixed
10240   values. This is useful for key rollover for example where several key
10241   sets may exist with different names.
10242
10243   *Steve Henson*
10244
10245 * Reverse ENGINE-internal logic for caching default ENGINE handles.
10246   This was broken until now in 0.9.8 releases, such that the only way
10247   a registered ENGINE could be used (assuming it initialises
10248   successfully on the host) was to explicitly set it as the default
10249   for the relevant algorithms. This is in contradiction with 0.9.7
10250   behaviour and the documentation. With this fix, when an ENGINE is
10251   registered into a given algorithm's table of implementations, the
10252   'uptodate' flag is reset so that auto-discovery will be used next
10253   time a new context for that algorithm attempts to select an
10254   implementation.
10255
10256   *Ian Lister (tweaked by Geoff Thorpe)*
10257
10258 * Backport of CMS code to OpenSSL 0.9.8. This differs from the 0.9.9
10259   implementation in the following ways:
10260
10261   Lack of EVP_PKEY_ASN1_METHOD means algorithm parameters have to be
10262   hard coded.
10263
10264   Lack of BER streaming support means one pass streaming processing is
10265   only supported if data is detached: setting the streaming flag is
10266   ignored for embedded content.
10267
10268   CMS support is disabled by default and must be explicitly enabled
10269   with the enable-cms configuration option.
10270
10271   *Steve Henson*
10272
10273 * Update the GMP engine glue to do direct copies between BIGNUM and
10274   mpz_t when openssl and GMP use the same limb size. Otherwise the
10275   existing "conversion via a text string export" trick is still used.
10276
10277   *Paul Sheer <paulsheer@gmail.com>*
10278
10279 * Zlib compression BIO. This is a filter BIO which compressed and
10280   uncompresses any data passed through it.
10281
10282   *Steve Henson*
10283
10284 * Add AES_wrap_key() and AES_unwrap_key() functions to implement
10285   RFC3394 compatible AES key wrapping.
10286
10287   *Steve Henson*
10288
10289 * Add utility functions to handle ASN1 structures. ASN1_STRING_set0():
10290   sets string data without copying. X509_ALGOR_set0() and
10291   X509_ALGOR_get0(): set and retrieve X509_ALGOR (AlgorithmIdentifier)
10292   data. Attribute function X509at_get0_data_by_OBJ(): retrieves data
10293   from an X509_ATTRIBUTE structure optionally checking it occurs only
10294   once. ASN1_TYPE_set1(): set and ASN1_TYPE structure copying supplied
10295   data.
10296
10297   *Steve Henson*
10298
10299 * Fix BN flag handling in RSA_eay_mod_exp() and BN_MONT_CTX_set()
10300   to get the expected BN_FLG_CONSTTIME behavior.
10301
10302   *Bodo Moeller (Google)*
10303
10304 * Netware support:
10305
10306   - fixed wrong usage of ioctlsocket() when build for LIBC BSD sockets
10307   - fixed do_tests.pl to run the test suite with CLIB builds too (CLIB_OPT)
10308   - added some more tests to do_tests.pl
10309   - fixed RunningProcess usage so that it works with newer LIBC NDKs too
10310   - removed usage of BN_LLONG for CLIB builds to avoid runtime dependency
10311   - added new Configure targets netware-clib-bsdsock, netware-clib-gcc,
10312     netware-clib-bsdsock-gcc, netware-libc-bsdsock-gcc
10313   - various changes to netware.pl to enable gcc-cross builds on Win32
10314     platform
10315   - changed crypto/bio/b_sock.c to work with macro functions (CLIB BSD)
10316   - various changes to fix missing prototype warnings
10317   - fixed x86nasm.pl to create correct asm files for NASM COFF output
10318   - added AES, WHIRLPOOL and CPUID assembler code to build files
10319   - added missing AES assembler make rules to mk1mf.pl
10320   - fixed order of includes in `apps/ocsp.c` so that `e_os.h` settings apply
10321
10322   *Guenter Knauf <eflash@gmx.net>*
10323
10324 * Implement certificate status request TLS extension defined in RFC3546.
10325   A client can set the appropriate parameters and receive the encoded
10326   OCSP response via a callback. A server can query the supplied parameters
10327   and set the encoded OCSP response in the callback. Add simplified examples
10328   to s_client and s_server.
10329
10330   *Steve Henson*
10331
10332### Changes between 0.9.8f and 0.9.8g  [19 Oct 2007]
10333
10334 * Fix various bugs:
10335   + Binary incompatibility of ssl_ctx_st structure
10336   + DTLS interoperation with non-compliant servers
10337   + Don't call get_session_cb() without proposed session
10338   + Fix ia64 assembler code
10339
10340   *Andy Polyakov, Steve Henson*
10341
10342### Changes between 0.9.8e and 0.9.8f  [11 Oct 2007]
10343
10344 * DTLS Handshake overhaul. There were longstanding issues with
10345   OpenSSL DTLS implementation, which were making it impossible for
10346   RFC 4347 compliant client to communicate with OpenSSL server.
10347   Unfortunately just fixing these incompatibilities would "cut off"
10348   pre-0.9.8f clients. To allow for hassle free upgrade post-0.9.8e
10349   server keeps tolerating non RFC compliant syntax. The opposite is
10350   not true, 0.9.8f client can not communicate with earlier server.
10351   This update even addresses CVE-2007-4995.
10352
10353   *Andy Polyakov*
10354
10355 * Changes to avoid need for function casts in OpenSSL: some compilers
10356   (gcc 4.2 and later) reject their use.
10357   *Kurt Roeckx <kurt@roeckx.be>, Peter Hartley <pdh@utter.chaos.org.uk>,
10358    Steve Henson*
10359
10360 * Add RFC4507 support to OpenSSL. This includes the corrections in
10361   RFC4507bis. The encrypted ticket format is an encrypted encoded
10362   SSL_SESSION structure, that way new session features are automatically
10363   supported.
10364
10365   If a client application caches session in an SSL_SESSION structure
10366   support is transparent because tickets are now stored in the encoded
10367   SSL_SESSION.
10368
10369   The SSL_CTX structure automatically generates keys for ticket
10370   protection in servers so again support should be possible
10371   with no application modification.
10372
10373   If a client or server wishes to disable RFC4507 support then the option
10374   SSL_OP_NO_TICKET can be set.
10375
10376   Add a TLS extension debugging callback to allow the contents of any client
10377   or server extensions to be examined.
10378
10379   This work was sponsored by Google.
10380
10381   *Steve Henson*
10382
10383 * Add initial support for TLS extensions, specifically for the server_name
10384   extension so far.  The SSL_SESSION, SSL_CTX, and SSL data structures now
10385   have new members for a host name.  The SSL data structure has an
10386   additional member `SSL_CTX *initial_ctx` so that new sessions can be
10387   stored in that context to allow for session resumption, even after the
10388   SSL has been switched to a new SSL_CTX in reaction to a client's
10389   server_name extension.
10390
10391   New functions (subject to change):
10392
10393           SSL_get_servername()
10394           SSL_get_servername_type()
10395           SSL_set_SSL_CTX()
10396
10397   New CTRL codes and macros (subject to change):
10398
10399           SSL_CTRL_SET_TLSEXT_SERVERNAME_CB
10400                               - SSL_CTX_set_tlsext_servername_callback()
10401           SSL_CTRL_SET_TLSEXT_SERVERNAME_ARG
10402                                    - SSL_CTX_set_tlsext_servername_arg()
10403           SSL_CTRL_SET_TLSEXT_HOSTNAME           - SSL_set_tlsext_host_name()
10404
10405   openssl s_client has a new '-servername ...' option.
10406
10407   openssl s_server has new options '-servername_host ...', '-cert2 ...',
10408   '-key2 ...', '-servername_fatal' (subject to change).  This allows
10409   testing the HostName extension for a specific single host name ('-cert'
10410   and '-key' remain fallbacks for handshakes without HostName
10411   negotiation).  If the unrecognized_name alert has to be sent, this by
10412   default is a warning; it becomes fatal with the '-servername_fatal'
10413   option.
10414
10415   *Peter Sylvester,  Remy Allais, Christophe Renou, Steve Henson*
10416
10417 * Add AES and SSE2 assembly language support to VC++ build.
10418
10419   *Steve Henson*
10420
10421 * Mitigate attack on final subtraction in Montgomery reduction.
10422
10423   *Andy Polyakov*
10424
10425 * Fix crypto/ec/ec_mult.c to work properly with scalars of value 0
10426   (which previously caused an internal error).
10427
10428   *Bodo Moeller*
10429
10430 * Squeeze another 10% out of IGE mode when in != out.
10431
10432   *Ben Laurie*
10433
10434 * AES IGE mode speedup.
10435
10436   *Dean Gaudet (Google)*
10437
10438 * Add the Korean symmetric 128-bit cipher SEED (see
10439   <http://www.kisa.or.kr/kisa/seed/jsp/seed_eng.jsp>) and
10440   add SEED ciphersuites from RFC 4162:
10441
10442           TLS_RSA_WITH_SEED_CBC_SHA      =  "SEED-SHA"
10443           TLS_DHE_DSS_WITH_SEED_CBC_SHA  =  "DHE-DSS-SEED-SHA"
10444           TLS_DHE_RSA_WITH_SEED_CBC_SHA  =  "DHE-RSA-SEED-SHA"
10445           TLS_DH_anon_WITH_SEED_CBC_SHA  =  "ADH-SEED-SHA"
10446
10447   To minimize changes between patchlevels in the OpenSSL 0.9.8
10448   series, SEED remains excluded from compilation unless OpenSSL
10449   is configured with 'enable-seed'.
10450
10451   *KISA, Bodo Moeller*
10452
10453 * Mitigate branch prediction attacks, which can be practical if a
10454   single processor is shared, allowing a spy process to extract
10455   information.  For detailed background information, see
10456   <http://eprint.iacr.org/2007/039> (O. Aciicmez, S. Gueron,
10457   J.-P. Seifert, "New Branch Prediction Vulnerabilities in OpenSSL
10458   and Necessary Software Countermeasures").  The core of the change
10459   are new versions BN_div_no_branch() and
10460   BN_mod_inverse_no_branch() of BN_div() and BN_mod_inverse(),
10461   respectively, which are slower, but avoid the security-relevant
10462   conditional branches.  These are automatically called by BN_div()
10463   and BN_mod_inverse() if the flag BN_FLG_CONSTTIME is set for one
10464   of the input BIGNUMs.  Also, BN_is_bit_set() has been changed to
10465   remove a conditional branch.
10466
10467   BN_FLG_CONSTTIME is the new name for the previous
10468   BN_FLG_EXP_CONSTTIME flag, since it now affects more than just
10469   modular exponentiation.  (Since OpenSSL 0.9.7h, setting this flag
10470   in the exponent causes BN_mod_exp_mont() to use the alternative
10471   implementation in BN_mod_exp_mont_consttime().)  The old name
10472   remains as a deprecated alias.
10473
10474   Similarly, RSA_FLAG_NO_EXP_CONSTTIME is replaced by a more general
10475   RSA_FLAG_NO_CONSTTIME flag since the RSA implementation now uses
10476   constant-time implementations for more than just exponentiation.
10477   Here too the old name is kept as a deprecated alias.
10478
10479   BN_BLINDING_new() will now use BN_dup() for the modulus so that
10480   the BN_BLINDING structure gets an independent copy of the
10481   modulus.  This means that the previous `BIGNUM *m` argument to
10482   BN_BLINDING_new() and to BN_BLINDING_create_param() now
10483   essentially becomes `const BIGNUM *m`, although we can't actually
10484   change this in the header file before 0.9.9.  It allows
10485   RSA_setup_blinding() to use BN_with_flags() on the modulus to
10486   enable BN_FLG_CONSTTIME.
10487
10488   *Matthew D Wood (Intel Corp)*
10489
10490 * In the SSL/TLS server implementation, be strict about session ID
10491   context matching (which matters if an application uses a single
10492   external cache for different purposes).  Previously,
10493   out-of-context reuse was forbidden only if SSL_VERIFY_PEER was
10494   set.  This did ensure strict client verification, but meant that,
10495   with applications using a single external cache for quite
10496   different requirements, clients could circumvent ciphersuite
10497   restrictions for a given session ID context by starting a session
10498   in a different context.
10499
10500   *Bodo Moeller*
10501
10502 * Include "!eNULL" in SSL_DEFAULT_CIPHER_LIST to make sure that
10503   a ciphersuite string such as "DEFAULT:RSA" cannot enable
10504   authentication-only ciphersuites.
10505
10506   *Bodo Moeller*
10507
10508 * Update the SSL_get_shared_ciphers() fix CVE-2006-3738 which was
10509   not complete and could lead to a possible single byte overflow
10510   ([CVE-2007-5135]) [Ben Laurie]
10511
10512### Changes between 0.9.8d and 0.9.8e  [23 Feb 2007]
10513
10514 * Since AES128 and AES256 (and similarly Camellia128 and
10515   Camellia256) share a single mask bit in the logic of
10516   ssl/ssl_ciph.c, the code for masking out disabled ciphers needs a
10517   kludge to work properly if AES128 is available and AES256 isn't
10518   (or if Camellia128 is available and Camellia256 isn't).
10519
10520   *Victor Duchovni*
10521
10522 * Fix the BIT STRING encoding generated by crypto/ec/ec_asn1.c
10523   (within i2d_ECPrivateKey, i2d_ECPKParameters, i2d_ECParameters):
10524   When a point or a seed is encoded in a BIT STRING, we need to
10525   prevent the removal of trailing zero bits to get the proper DER
10526   encoding.  (By default, crypto/asn1/a_bitstr.c assumes the case
10527   of a NamedBitList, for which trailing 0 bits need to be removed.)
10528
10529   *Bodo Moeller*
10530
10531 * Have SSL/TLS server implementation tolerate "mismatched" record
10532   protocol version while receiving ClientHello even if the
10533   ClientHello is fragmented.  (The server can't insist on the
10534   particular protocol version it has chosen before the ServerHello
10535   message has informed the client about his choice.)
10536
10537   *Bodo Moeller*
10538
10539 * Add RFC 3779 support.
10540
10541   *Rob Austein for ARIN, Ben Laurie*
10542
10543 * Load error codes if they are not already present instead of using a
10544   static variable. This allows them to be cleanly unloaded and reloaded.
10545   Improve header file function name parsing.
10546
10547   *Steve Henson*
10548
10549 * extend SMTP and IMAP protocol emulation in s_client to use EHLO
10550   or CAPABILITY handshake as required by RFCs.
10551
10552   *Goetz Babin-Ebell*
10553
10554### Changes between 0.9.8c and 0.9.8d  [28 Sep 2006]
10555
10556 * Introduce limits to prevent malicious keys being able to
10557   cause a denial of service.  ([CVE-2006-2940])
10558
10559   *Steve Henson, Bodo Moeller*
10560
10561 * Fix ASN.1 parsing of certain invalid structures that can result
10562   in a denial of service.  ([CVE-2006-2937])  [Steve Henson]
10563
10564 * Fix buffer overflow in SSL_get_shared_ciphers() function.
10565   ([CVE-2006-3738]) [Tavis Ormandy and Will Drewry, Google Security Team]
10566
10567 * Fix SSL client code which could crash if connecting to a
10568   malicious SSLv2 server.  ([CVE-2006-4343])
10569
10570   *Tavis Ormandy and Will Drewry, Google Security Team*
10571
10572 * Since 0.9.8b, ciphersuite strings naming explicit ciphersuites
10573   match only those.  Before that, "AES256-SHA" would be interpreted
10574   as a pattern and match "AES128-SHA" too (since AES128-SHA got
10575   the same strength classification in 0.9.7h) as we currently only
10576   have a single AES bit in the ciphersuite description bitmap.
10577   That change, however, also applied to ciphersuite strings such as
10578   "RC4-MD5" that intentionally matched multiple ciphersuites --
10579   namely, SSL 2.0 ciphersuites in addition to the more common ones
10580   from SSL 3.0/TLS 1.0.
10581
10582   So we change the selection algorithm again: Naming an explicit
10583   ciphersuite selects this one ciphersuite, and any other similar
10584   ciphersuite (same bitmap) from *other* protocol versions.
10585   Thus, "RC4-MD5" again will properly select both the SSL 2.0
10586   ciphersuite and the SSL 3.0/TLS 1.0 ciphersuite.
10587
10588   Since SSL 2.0 does not have any ciphersuites for which the
10589   128/256 bit distinction would be relevant, this works for now.
10590   The proper fix will be to use different bits for AES128 and
10591   AES256, which would have avoided the problems from the beginning;
10592   however, bits are scarce, so we can only do this in a new release
10593   (not just a patchlevel) when we can change the SSL_CIPHER
10594   definition to split the single 'unsigned long mask' bitmap into
10595   multiple values to extend the available space.
10596
10597   *Bodo Moeller*
10598
10599### Changes between 0.9.8b and 0.9.8c  [05 Sep 2006]
10600
10601 * Avoid PKCS #1 v1.5 signature attack discovered by Daniel Bleichenbacher
10602   ([CVE-2006-4339])  [Ben Laurie and Google Security Team]
10603
10604 * Add AES IGE and biIGE modes.
10605
10606   *Ben Laurie*
10607
10608 * Change the Unix randomness entropy gathering to use poll() when
10609   possible instead of select(), since the latter has some
10610   undesirable limitations.
10611
10612   *Darryl Miles via Richard Levitte and Bodo Moeller*
10613
10614 * Disable "ECCdraft" ciphersuites more thoroughly.  Now special
10615   treatment in ssl/ssl_ciph.s makes sure that these ciphersuites
10616   cannot be implicitly activated as part of, e.g., the "AES" alias.
10617   However, please upgrade to OpenSSL 0.9.9[-dev] for
10618   non-experimental use of the ECC ciphersuites to get TLS extension
10619   support, which is required for curve and point format negotiation
10620   to avoid potential handshake problems.
10621
10622   *Bodo Moeller*
10623
10624 * Disable rogue ciphersuites:
10625
10626   - SSLv2 0x08 0x00 0x80 ("RC4-64-MD5")
10627   - SSLv3/TLSv1 0x00 0x61 ("EXP1024-RC2-CBC-MD5")
10628   - SSLv3/TLSv1 0x00 0x60 ("EXP1024-RC4-MD5")
10629
10630   The latter two were purportedly from
10631   draft-ietf-tls-56-bit-ciphersuites-0[01].txt, but do not really
10632   appear there.
10633
10634   Also deactivate the remaining ciphersuites from
10635   draft-ietf-tls-56-bit-ciphersuites-01.txt.  These are just as
10636   unofficial, and the ID has long expired.
10637
10638   *Bodo Moeller*
10639
10640 * Fix RSA blinding Heisenbug (problems sometimes occurred on
10641   dual-core machines) and other potential thread-safety issues.
10642
10643   *Bodo Moeller*
10644
10645 * Add the symmetric cipher Camellia (128-bit, 192-bit, 256-bit key
10646   versions), which is now available for royalty-free use
10647   (see <http://info.isl.ntt.co.jp/crypt/eng/info/chiteki.html>).
10648   Also, add Camellia TLS ciphersuites from RFC 4132.
10649
10650   To minimize changes between patchlevels in the OpenSSL 0.9.8
10651   series, Camellia remains excluded from compilation unless OpenSSL
10652   is configured with 'enable-camellia'.
10653
10654   *NTT*
10655
10656 * Disable the padding bug check when compression is in use. The padding
10657   bug check assumes the first packet is of even length, this is not
10658   necessarily true if compression is enabled and can result in false
10659   positives causing handshake failure. The actual bug test is ancient
10660   code so it is hoped that implementations will either have fixed it by
10661   now or any which still have the bug do not support compression.
10662
10663   *Steve Henson*
10664
10665### Changes between 0.9.8a and 0.9.8b  [04 May 2006]
10666
10667 * When applying a cipher rule check to see if string match is an explicit
10668   cipher suite and only match that one cipher suite if it is.
10669
10670   *Steve Henson*
10671
10672 * Link in manifests for VC++ if needed.
10673
10674   *Austin Ziegler <halostatue@gmail.com>*
10675
10676 * Update support for ECC-based TLS ciphersuites according to
10677   draft-ietf-tls-ecc-12.txt with proposed changes (but without
10678   TLS extensions, which are supported starting with the 0.9.9
10679   branch, not in the OpenSSL 0.9.8 branch).
10680
10681   *Douglas Stebila*
10682
10683 * New functions EVP_CIPHER_CTX_new() and EVP_CIPHER_CTX_free() to support
10684   opaque EVP_CIPHER_CTX handling.
10685
10686   *Steve Henson*
10687
10688 * Fixes and enhancements to zlib compression code. We now only use
10689   "zlib1.dll" and use the default `__cdecl` calling convention on Win32
10690   to conform with the standards mentioned here:
10691   <http://www.zlib.net/DLL_FAQ.txt>
10692   Static zlib linking now works on Windows and the new --with-zlib-include
10693   --with-zlib-lib options to Configure can be used to supply the location
10694   of the headers and library. Gracefully handle case where zlib library
10695   can't be loaded.
10696
10697   *Steve Henson*
10698
10699 * Several fixes and enhancements to the OID generation code. The old code
10700   sometimes allowed invalid OIDs (1.X for X >= 40 for example), couldn't
10701   handle numbers larger than ULONG_MAX, truncated printing and had a
10702   non standard OBJ_obj2txt() behaviour.
10703
10704   *Steve Henson*
10705
10706 * Add support for building of engines under engine/ as shared libraries
10707   under VC++ build system.
10708
10709   *Steve Henson*
10710
10711 * Corrected the numerous bugs in the Win32 path splitter in DSO.
10712   Hopefully, we will not see any false combination of paths any more.
10713
10714   *Richard Levitte*
10715
10716### Changes between 0.9.8 and 0.9.8a  [11 Oct 2005]
10717
10718 * Remove the functionality of SSL_OP_MSIE_SSLV2_RSA_PADDING
10719   (part of SSL_OP_ALL).  This option used to disable the
10720   countermeasure against man-in-the-middle protocol-version
10721   rollback in the SSL 2.0 server implementation, which is a bad
10722   idea.  ([CVE-2005-2969])
10723
10724   *Bodo Moeller; problem pointed out by Yutaka Oiwa (Research Center
10725   for Information Security, National Institute of Advanced Industrial
10726   Science and Technology [AIST], Japan)*
10727
10728 * Add two function to clear and return the verify parameter flags.
10729
10730   *Steve Henson*
10731
10732 * Keep cipherlists sorted in the source instead of sorting them at
10733   runtime, thus removing the need for a lock.
10734
10735   *Nils Larsch*
10736
10737 * Avoid some small subgroup attacks in Diffie-Hellman.
10738
10739   *Nick Mathewson and Ben Laurie*
10740
10741 * Add functions for well-known primes.
10742
10743   *Nick Mathewson*
10744
10745 * Extended Windows CE support.
10746
10747   *Satoshi Nakamura and Andy Polyakov*
10748
10749 * Initialize SSL_METHOD structures at compile time instead of during
10750   runtime, thus removing the need for a lock.
10751
10752   *Steve Henson*
10753
10754 * Make PKCS7_decrypt() work even if no certificate is supplied by
10755   attempting to decrypt each encrypted key in turn. Add support to
10756   smime utility.
10757
10758   *Steve Henson*
10759
10760### Changes between 0.9.7h and 0.9.8  [05 Jul 2005]
10761
10762[NB: OpenSSL 0.9.7i and later 0.9.7 patch levels were released after
10763OpenSSL 0.9.8.]
10764
10765 * Add libcrypto.pc and libssl.pc for those who feel they need them.
10766
10767   *Richard Levitte*
10768
10769 * Change CA.sh and CA.pl so they don't bundle the CSR and the private
10770   key into the same file any more.
10771
10772   *Richard Levitte*
10773
10774 * Add initial support for Win64, both IA64 and AMD64/x64 flavors.
10775
10776   *Andy Polyakov*
10777
10778 * Add -utf8 command line and config file option to 'ca'.
10779
10780   *Stefan <stf@udoma.org*
10781
10782 * Removed the macro des_crypt(), as it seems to conflict with some
10783   libraries.  Use DES_crypt().
10784
10785   *Richard Levitte*
10786
10787 * Correct naming of the 'chil' and '4758cca' ENGINEs. This
10788   involves renaming the source and generated shared-libs for
10789   both. The engines will accept the corrected or legacy ids
10790   ('ncipher' and '4758_cca' respectively) when binding. NB,
10791   this only applies when building 'shared'.
10792
10793   *Corinna Vinschen <vinschen@redhat.com> and Geoff Thorpe*
10794
10795 * Add attribute functions to EVP_PKEY structure. Modify
10796   PKCS12_create() to recognize a CSP name attribute and
10797   use it. Make -CSP option work again in pkcs12 utility.
10798
10799   *Steve Henson*
10800
10801 * Add new functionality to the bn blinding code:
10802   - automatic re-creation of the BN_BLINDING parameters after
10803     a fixed number of uses (currently 32)
10804   - add new function for parameter creation
10805   - introduce flags to control the update behaviour of the
10806     BN_BLINDING parameters
10807   - hide BN_BLINDING structure
10808   Add a second BN_BLINDING slot to the RSA structure to improve
10809   performance when a single RSA object is shared among several
10810   threads.
10811
10812   *Nils Larsch*
10813
10814 * Add support for DTLS.
10815
10816   *Nagendra Modadugu <nagendra@cs.stanford.edu> and Ben Laurie*
10817
10818 * Add support for DER encoded private keys (SSL_FILETYPE_ASN1)
10819   to SSL_CTX_use_PrivateKey_file() and SSL_use_PrivateKey_file()
10820
10821   *Walter Goulet*
10822
10823 * Remove buggy and incomplete DH cert support from
10824   ssl/ssl_rsa.c and ssl/s3_both.c
10825
10826   *Nils Larsch*
10827
10828 * Use SHA-1 instead of MD5 as the default digest algorithm for
10829   the `apps/openssl` commands.
10830
10831   *Nils Larsch*
10832
10833 * Compile clean with "-Wall -Wmissing-prototypes
10834   -Wstrict-prototypes -Wmissing-declarations -Werror". Currently
10835   DEBUG_SAFESTACK must also be set.
10836
10837   *Ben Laurie*
10838
10839 * Change ./Configure so that certain algorithms can be disabled by default.
10840   The new counterpiece to "no-xxx" is "enable-xxx".
10841
10842   The patented RC5 and MDC2 algorithms will now be disabled unless
10843   "enable-rc5" and "enable-mdc2", respectively, are specified.
10844
10845   (IDEA remains enabled despite being patented.  This is because IDEA
10846   is frequently required for interoperability, and there is no license
10847   fee for non-commercial use.  As before, "no-idea" can be used to
10848   avoid this algorithm.)
10849
10850   *Bodo Moeller*
10851
10852 * Add processing of proxy certificates (see RFC 3820).  This work was
10853   sponsored by KTH (The Royal Institute of Technology in Stockholm) and
10854   EGEE (Enabling Grids for E-science in Europe).
10855
10856   *Richard Levitte*
10857
10858 * RC4 performance overhaul on modern architectures/implementations, such
10859   as Intel P4, IA-64 and AMD64.
10860
10861   *Andy Polyakov*
10862
10863 * New utility extract-section.pl. This can be used specify an alternative
10864   section number in a pod file instead of having to treat each file as
10865   a separate case in Makefile. This can be done by adding two lines to the
10866   pod file:
10867
10868   =for comment openssl_section:XXX
10869
10870   The blank line is mandatory.
10871
10872   *Steve Henson*
10873
10874 * New arguments -certform, -keyform and -pass for s_client and s_server
10875   to allow alternative format key and certificate files and passphrase
10876   sources.
10877
10878   *Steve Henson*
10879
10880 * New structure X509_VERIFY_PARAM which combines current verify parameters,
10881   update associated structures and add various utility functions.
10882
10883   Add new policy related verify parameters, include policy checking in
10884   standard verify code. Enhance 'smime' application with extra parameters
10885   to support policy checking and print out.
10886
10887   *Steve Henson*
10888
10889 * Add a new engine to support VIA PadLock ACE extensions in the VIA C3
10890   Nehemiah processors. These extensions support AES encryption in hardware
10891   as well as RNG (though RNG support is currently disabled).
10892
10893   *Michal Ludvig <michal@logix.cz>, with help from Andy Polyakov*
10894
10895 * Deprecate `BN_[get|set]_params()` functions (they were ignored internally).
10896
10897   *Geoff Thorpe*
10898
10899 * New FIPS 180-2 algorithms, SHA-224/-256/-384/-512 are implemented.
10900
10901   *Andy Polyakov and a number of other people*
10902
10903 * Improved PowerPC platform support. Most notably BIGNUM assembler
10904   implementation contributed by IBM.
10905
10906   *Suresh Chari, Peter Waltenberg, Andy Polyakov*
10907
10908 * The new 'RSA_generate_key_ex' function now takes a BIGNUM for the public
10909   exponent rather than 'unsigned long'. There is a corresponding change to
10910   the new 'rsa_keygen' element of the RSA_METHOD structure.
10911
10912   *Jelte Jansen, Geoff Thorpe*
10913
10914 * Functionality for creating the initial serial number file is now
10915   moved from CA.pl to the 'ca' utility with a new option -create_serial.
10916
10917   (Before OpenSSL 0.9.7e, CA.pl used to initialize the serial
10918   number file to 1, which is bound to cause problems.  To avoid
10919   the problems while respecting compatibility between different 0.9.7
10920   patchlevels, 0.9.7e  employed 'openssl x509 -next_serial' in
10921   CA.pl for serial number initialization.  With the new release 0.9.8,
10922   we can fix the problem directly in the 'ca' utility.)
10923
10924   *Steve Henson*
10925
10926 * Reduced header interdependencies by declaring more opaque objects in
10927   ossl_typ.h. As a consequence, including some headers (eg. engine.h) will
10928   give fewer recursive includes, which could break lazy source code - so
10929   this change is covered by the OPENSSL_NO_DEPRECATED symbol. As always,
10930   developers should define this symbol when building and using openssl to
10931   ensure they track the recommended behaviour, interfaces, [etc], but
10932   backwards-compatible behaviour prevails when this isn't defined.
10933
10934   *Geoff Thorpe*
10935
10936 * New function X509_POLICY_NODE_print() which prints out policy nodes.
10937
10938   *Steve Henson*
10939
10940 * Add new EVP function EVP_CIPHER_CTX_rand_key and associated functionality.
10941   This will generate a random key of the appropriate length based on the
10942   cipher context. The EVP_CIPHER can provide its own random key generation
10943   routine to support keys of a specific form. This is used in the des and
10944   3des routines to generate a key of the correct parity. Update S/MIME
10945   code to use new functions and hence generate correct parity DES keys.
10946   Add EVP_CHECK_DES_KEY #define to return an error if the key is not
10947   valid (weak or incorrect parity).
10948
10949   *Steve Henson*
10950
10951 * Add a local set of CRLs that can be used by X509_verify_cert() as well
10952   as looking them up. This is useful when the verified structure may contain
10953   CRLs, for example PKCS#7 signedData. Modify PKCS7_verify() to use any CRLs
10954   present unless the new PKCS7_NO_CRL flag is asserted.
10955
10956   *Steve Henson*
10957
10958 * Extend ASN1 oid configuration module. It now additionally accepts the
10959   syntax:
10960
10961   shortName = some long name, 1.2.3.4
10962
10963   *Steve Henson*
10964
10965 * Reimplemented the BN_CTX implementation. There is now no more static
10966   limitation on the number of variables it can handle nor the depth of the
10967   "stack" handling for BN_CTX_start()/BN_CTX_end() pairs. The stack
10968   information can now expand as required, and rather than having a single
10969   static array of bignums, BN_CTX now uses a linked-list of such arrays
10970   allowing it to expand on demand whilst maintaining the usefulness of
10971   BN_CTX's "bundling".
10972
10973   *Geoff Thorpe*
10974
10975 * Add a missing BN_CTX parameter to the 'rsa_mod_exp' callback in RSA_METHOD
10976   to allow all RSA operations to function using a single BN_CTX.
10977
10978   *Geoff Thorpe*
10979
10980 * Preliminary support for certificate policy evaluation and checking. This
10981   is initially intended to pass the tests outlined in "Conformance Testing
10982   of Relying Party Client Certificate Path Processing Logic" v1.07.
10983
10984   *Steve Henson*
10985
10986 * bn_dup_expand() has been deprecated, it was introduced in 0.9.7 and
10987   remained unused and not that useful. A variety of other little bignum
10988   tweaks and fixes have also been made continuing on from the audit (see
10989   below).
10990
10991   *Geoff Thorpe*
10992
10993 * Constify all or almost all d2i, c2i, s2i and r2i functions, along with
10994   associated ASN1, EVP and SSL functions and old ASN1 macros.
10995
10996   *Richard Levitte*
10997
10998 * BN_zero() only needs to set 'top' and 'neg' to zero for correct results,
10999   and this should never fail. So the return value from the use of
11000   BN_set_word() (which can fail due to needless expansion) is now deprecated;
11001   if OPENSSL_NO_DEPRECATED is defined, BN_zero() is a void macro.
11002
11003   *Geoff Thorpe*
11004
11005 * BN_CTX_get() should return zero-valued bignums, providing the same
11006   initialised value as BN_new().
11007
11008   *Geoff Thorpe, suggested by Ulf Möller*
11009
11010 * Support for inhibitAnyPolicy certificate extension.
11011
11012   *Steve Henson*
11013
11014 * An audit of the BIGNUM code is underway, for which debugging code is
11015   enabled when BN_DEBUG is defined. This makes stricter enforcements on what
11016   is considered valid when processing BIGNUMs, and causes execution to
11017   assert() when a problem is discovered. If BN_DEBUG_RAND is defined,
11018   further steps are taken to deliberately pollute unused data in BIGNUM
11019   structures to try and expose faulty code further on. For now, openssl will
11020   (in its default mode of operation) continue to tolerate the inconsistent
11021   forms that it has tolerated in the past, but authors and packagers should
11022   consider trying openssl and their own applications when compiled with
11023   these debugging symbols defined. It will help highlight potential bugs in
11024   their own code, and will improve the test coverage for OpenSSL itself. At
11025   some point, these tighter rules will become openssl's default to improve
11026   maintainability, though the assert()s and other overheads will remain only
11027   in debugging configurations. See bn.h for more details.
11028
11029   *Geoff Thorpe, Nils Larsch, Ulf Möller*
11030
11031 * BN_CTX_init() has been deprecated, as BN_CTX is an opaque structure
11032   that can only be obtained through BN_CTX_new() (which implicitly
11033   initialises it). The presence of this function only made it possible
11034   to overwrite an existing structure (and cause memory leaks).
11035
11036   *Geoff Thorpe*
11037
11038 * Because of the callback-based approach for implementing LHASH as a
11039   template type, lh_insert() adds opaque objects to hash-tables and
11040   lh_doall() or lh_doall_arg() are typically used with a destructor callback
11041   to clean up those corresponding objects before destroying the hash table
11042   (and losing the object pointers). So some over-zealous constifications in
11043   LHASH have been relaxed so that lh_insert() does not take (nor store) the
11044   objects as "const" and the `lh_doall[_arg]` callback wrappers are not
11045   prototyped to have "const" restrictions on the object pointers they are
11046   given (and so aren't required to cast them away any more).
11047
11048   *Geoff Thorpe*
11049
11050 * The tmdiff.h API was so ugly and minimal that our own timing utility
11051   (speed) prefers to use its own implementation. The two implementations
11052   haven't been consolidated as yet (volunteers?) but the tmdiff API has had
11053   its object type properly exposed (MS_TM) instead of casting to/from
11054   `char *`. This may still change yet if someone realises MS_TM and
11055   `ms_time_***`
11056   aren't necessarily the greatest nomenclatures - but this is what was used
11057   internally to the implementation so I've used that for now.
11058
11059   *Geoff Thorpe*
11060
11061 * Ensure that deprecated functions do not get compiled when
11062   OPENSSL_NO_DEPRECATED is defined. Some "openssl" subcommands and a few of
11063   the self-tests were still using deprecated key-generation functions so
11064   these have been updated also.
11065
11066   *Geoff Thorpe*
11067
11068 * Reorganise PKCS#7 code to separate the digest location functionality
11069   into PKCS7_find_digest(), digest addition into PKCS7_bio_add_digest().
11070   New function PKCS7_set_digest() to set the digest type for PKCS#7
11071   digestedData type. Add additional code to correctly generate the
11072   digestedData type and add support for this type in PKCS7 initialization
11073   functions.
11074
11075   *Steve Henson*
11076
11077 * New function PKCS7_set0_type_other() this initializes a PKCS7
11078   structure of type "other".
11079
11080   *Steve Henson*
11081
11082 * Fix prime generation loop in crypto/bn/bn_prime.pl by making
11083   sure the loop does correctly stop and breaking ("division by zero")
11084   modulus operations are not performed. The (pre-generated) prime
11085   table crypto/bn/bn_prime.h was already correct, but it could not be
11086   re-generated on some platforms because of the "division by zero"
11087   situation in the script.
11088
11089   *Ralf S. Engelschall*
11090
11091 * Update support for ECC-based TLS ciphersuites according to
11092   draft-ietf-tls-ecc-03.txt: the KDF1 key derivation function with
11093   SHA-1 now is only used for "small" curves (where the
11094   representation of a field element takes up to 24 bytes); for
11095   larger curves, the field element resulting from ECDH is directly
11096   used as premaster secret.
11097
11098   *Douglas Stebila (Sun Microsystems Laboratories)*
11099
11100 * Add code for kP+lQ timings to crypto/ec/ectest.c, and add SEC2
11101   curve secp160r1 to the tests.
11102
11103   *Douglas Stebila (Sun Microsystems Laboratories)*
11104
11105 * Add the possibility to load symbols globally with DSO.
11106
11107   *Götz Babin-Ebell <babin-ebell@trustcenter.de> via Richard Levitte*
11108
11109 * Add the functions ERR_set_mark() and ERR_pop_to_mark() for better
11110   control of the error stack.
11111
11112   *Richard Levitte*
11113
11114 * Add support for STORE in ENGINE.
11115
11116   *Richard Levitte*
11117
11118 * Add the STORE type.  The intention is to provide a common interface
11119   to certificate and key stores, be they simple file-based stores, or
11120   HSM-type store, or LDAP stores, or...
11121   NOTE: The code is currently UNTESTED and isn't really used anywhere.
11122
11123   *Richard Levitte*
11124
11125 * Add a generic structure called OPENSSL_ITEM.  This can be used to
11126   pass a list of arguments to any function as well as provide a way
11127   for a function to pass data back to the caller.
11128
11129   *Richard Levitte*
11130
11131 * Add the functions BUF_strndup() and BUF_memdup().  BUF_strndup()
11132   works like BUF_strdup() but can be used to duplicate a portion of
11133   a string.  The copy gets NUL-terminated.  BUF_memdup() duplicates
11134   a memory area.
11135
11136   *Richard Levitte*
11137
11138 * Add the function sk_find_ex() which works like sk_find(), but will
11139   return an index to an element even if an exact match couldn't be
11140   found.  The index is guaranteed to point at the element where the
11141   searched-for key would be inserted to preserve sorting order.
11142
11143   *Richard Levitte*
11144
11145 * Add the function OBJ_bsearch_ex() which works like OBJ_bsearch() but
11146   takes an extra flags argument for optional functionality.  Currently,
11147   the following flags are defined:
11148
11149      OBJ_BSEARCH_VALUE_ON_NOMATCH
11150      This one gets OBJ_bsearch_ex() to return a pointer to the first
11151      element where the comparing function returns a negative or zero
11152      number.
11153
11154      OBJ_BSEARCH_FIRST_VALUE_ON_MATCH
11155      This one gets OBJ_bsearch_ex() to return a pointer to the first
11156      element where the comparing function returns zero.  This is useful
11157      if there are more than one element where the comparing function
11158      returns zero.
11159
11160   *Richard Levitte*
11161
11162 * Make it possible to create self-signed certificates with 'openssl ca'
11163   in such a way that the self-signed certificate becomes part of the
11164   CA database and uses the same mechanisms for serial number generation
11165   as all other certificate signing.  The new flag '-selfsign' enables
11166   this functionality.  Adapt CA.sh and CA.pl.in.
11167
11168   *Richard Levitte*
11169
11170 * Add functionality to check the public key of a certificate request
11171   against a given private.  This is useful to check that a certificate
11172   request can be signed by that key (self-signing).
11173
11174   *Richard Levitte*
11175
11176 * Make it possible to have multiple active certificates with the same
11177   subject in the CA index file.  This is done only if the keyword
11178   'unique_subject' is set to 'no' in the main CA section (default
11179   if 'CA_default') of the configuration file.  The value is saved
11180   with the database itself in a separate index attribute file,
11181   named like the index file with '.attr' appended to the name.
11182
11183   *Richard Levitte*
11184
11185 * Generate multi-valued AVAs using '+' notation in config files for
11186   req and dirName.
11187
11188   *Steve Henson*
11189
11190 * Support for nameConstraints certificate extension.
11191
11192   *Steve Henson*
11193
11194 * Support for policyConstraints certificate extension.
11195
11196   *Steve Henson*
11197
11198 * Support for policyMappings certificate extension.
11199
11200   *Steve Henson*
11201
11202 * Make sure the default DSA_METHOD implementation only uses its
11203   dsa_mod_exp() and/or bn_mod_exp() handlers if they are non-NULL,
11204   and change its own handlers to be NULL so as to remove unnecessary
11205   indirection. This lets alternative implementations fallback to the
11206   default implementation more easily.
11207
11208   *Geoff Thorpe*
11209
11210 * Support for directoryName in GeneralName related extensions
11211   in config files.
11212
11213   *Steve Henson*
11214
11215 * Make it possible to link applications using Makefile.shared.
11216   Make that possible even when linking against static libraries!
11217
11218   *Richard Levitte*
11219
11220 * Support for single pass processing for S/MIME signing. This now
11221   means that S/MIME signing can be done from a pipe, in addition
11222   cleartext signing (multipart/signed type) is effectively streaming
11223   and the signed data does not need to be all held in memory.
11224
11225   This is done with a new flag PKCS7_STREAM. When this flag is set
11226   PKCS7_sign() only initializes the PKCS7 structure and the actual signing
11227   is done after the data is output (and digests calculated) in
11228   SMIME_write_PKCS7().
11229
11230   *Steve Henson*
11231
11232 * Add full support for -rpath/-R, both in shared libraries and
11233   applications, at least on the platforms where it's known how
11234   to do it.
11235
11236   *Richard Levitte*
11237
11238 * In crypto/ec/ec_mult.c, implement fast point multiplication with
11239   precomputation, based on wNAF splitting: EC_GROUP_precompute_mult()
11240   will now compute a table of multiples of the generator that
11241   makes subsequent invocations of EC_POINTs_mul() or EC_POINT_mul()
11242   faster (notably in the case of a single point multiplication,
11243   scalar * generator).
11244
11245   *Nils Larsch, Bodo Moeller*
11246
11247 * IPv6 support for certificate extensions. The various extensions
11248   which use the IP:a.b.c.d can now take IPv6 addresses using the
11249   formats of RFC1884 2.2 . IPv6 addresses are now also displayed
11250   correctly.
11251
11252   *Steve Henson*
11253
11254 * Added an ENGINE that implements RSA by performing private key
11255   exponentiations with the GMP library. The conversions to and from
11256   GMP's mpz_t format aren't optimised nor are any montgomery forms
11257   cached, and on x86 it appears OpenSSL's own performance has caught up.
11258   However there are likely to be other architectures where GMP could
11259   provide a boost. This ENGINE is not built in by default, but it can be
11260   specified at Configure time and should be accompanied by the necessary
11261   linker additions, eg;
11262           ./config -DOPENSSL_USE_GMP -lgmp
11263
11264   *Geoff Thorpe*
11265
11266 * "openssl engine" will not display ENGINE/DSO load failure errors when
11267   testing availability of engines with "-t" - the old behaviour is
11268   produced by increasing the feature's verbosity with "-tt".
11269
11270   *Geoff Thorpe*
11271
11272 * ECDSA routines: under certain error conditions uninitialized BN objects
11273   could be freed. Solution: make sure initialization is performed early
11274   enough. (Reported and fix supplied by Nils Larsch <nla@trustcenter.de>
11275   via PR#459)
11276
11277   *Lutz Jaenicke*
11278
11279 * Key-generation can now be implemented in RSA_METHOD, DSA_METHOD
11280   and DH_METHOD (eg. by ENGINE implementations) to override the normal
11281   software implementations. For DSA and DH, parameter generation can
11282   also be overridden by providing the appropriate method callbacks.
11283
11284   *Geoff Thorpe*
11285
11286 * Change the "progress" mechanism used in key-generation and
11287   primality testing to functions that take a new BN_GENCB pointer in
11288   place of callback/argument pairs. The new API functions have `_ex`
11289   postfixes and the older functions are reimplemented as wrappers for
11290   the new ones. The OPENSSL_NO_DEPRECATED symbol can be used to hide
11291   declarations of the old functions to help (graceful) attempts to
11292   migrate to the new functions. Also, the new key-generation API
11293   functions operate on a caller-supplied key-structure and return
11294   success/failure rather than returning a key or NULL - this is to
11295   help make "keygen" another member function of RSA_METHOD etc.
11296
11297   Example for using the new callback interface:
11298
11299           int (*my_callback)(int a, int b, BN_GENCB *cb) = ...;
11300           void *my_arg = ...;
11301           BN_GENCB my_cb;
11302
11303           BN_GENCB_set(&my_cb, my_callback, my_arg);
11304
11305           return BN_is_prime_ex(some_bignum, BN_prime_checks, NULL, &cb);
11306           /* For the meaning of a, b in calls to my_callback(), see the
11307            * documentation of the function that calls the callback.
11308            * cb will point to my_cb; my_arg can be retrieved as cb->arg.
11309            * my_callback should return 1 if it wants BN_is_prime_ex()
11310            * to continue, or 0 to stop.
11311            */
11312
11313   *Geoff Thorpe*
11314
11315 * Change the ZLIB compression method to be stateful, and make it
11316   available to TLS with the number defined in
11317   draft-ietf-tls-compression-04.txt.
11318
11319   *Richard Levitte*
11320
11321 * Add the ASN.1 structures and functions for CertificatePair, which
11322   is defined as follows (according to X.509_4thEditionDraftV6.pdf):
11323
11324           CertificatePair ::= SEQUENCE {
11325              forward         [0]     Certificate OPTIONAL,
11326              reverse         [1]     Certificate OPTIONAL,
11327              -- at least one of the pair shall be present -- }
11328
11329   Also implement the PEM functions to read and write certificate
11330   pairs, and defined the PEM tag as "CERTIFICATE PAIR".
11331
11332   This needed to be defined, mostly for the sake of the LDAP
11333   attribute crossCertificatePair, but may prove useful elsewhere as
11334   well.
11335
11336   *Richard Levitte*
11337
11338 * Make it possible to inhibit symlinking of shared libraries in
11339   Makefile.shared, for Cygwin's sake.
11340
11341   *Richard Levitte*
11342
11343 * Extend the BIGNUM API by creating a function
11344           void BN_set_negative(BIGNUM *a, int neg);
11345   and a macro that behave like
11346           int  BN_is_negative(const BIGNUM *a);
11347
11348   to avoid the need to access 'a->neg' directly in applications.
11349
11350   *Nils Larsch*
11351
11352 * Implement fast modular reduction for pseudo-Mersenne primes
11353   used in NIST curves (crypto/bn/bn_nist.c, crypto/ec/ecp_nist.c).
11354   EC_GROUP_new_curve_GFp() will now automatically use this
11355   if applicable.
11356
11357   *Nils Larsch <nla@trustcenter.de>*
11358
11359 * Add new lock type (CRYPTO_LOCK_BN).
11360
11361   *Bodo Moeller*
11362
11363 * Change the ENGINE framework to automatically load engines
11364   dynamically from specific directories unless they could be
11365   found to already be built in or loaded.  Move all the
11366   current engines except for the cryptodev one to a new
11367   directory engines/.
11368   The engines in engines/ are built as shared libraries if
11369   the "shared" options was given to ./Configure or ./config.
11370   Otherwise, they are inserted in libcrypto.a.
11371   /usr/local/ssl/engines is the default directory for dynamic
11372   engines, but that can be overridden at configure time through
11373   the usual use of --prefix and/or --openssldir, and at run
11374   time with the environment variable OPENSSL_ENGINES.
11375
11376   *Geoff Thorpe and Richard Levitte*
11377
11378 * Add Makefile.shared, a helper makefile to build shared
11379   libraries.  Adapt Makefile.org.
11380
11381   *Richard Levitte*
11382
11383 * Add version info to Win32 DLLs.
11384
11385   *Peter 'Luna' Runestig" <peter@runestig.com>*
11386
11387 * Add new 'medium level' PKCS#12 API. Certificates and keys
11388   can be added using this API to created arbitrary PKCS#12
11389   files while avoiding the low-level API.
11390
11391   New options to PKCS12_create(), key or cert can be NULL and
11392   will then be omitted from the output file. The encryption
11393   algorithm NIDs can be set to -1 for no encryption, the mac
11394   iteration count can be set to 0 to omit the mac.
11395
11396   Enhance pkcs12 utility by making the -nokeys and -nocerts
11397   options work when creating a PKCS#12 file. New option -nomac
11398   to omit the mac, NONE can be set for an encryption algorithm.
11399   New code is modified to use the enhanced PKCS12_create()
11400   instead of the low-level API.
11401
11402   *Steve Henson*
11403
11404 * Extend ASN1 encoder to support indefinite length constructed
11405   encoding. This can output sequences tags and octet strings in
11406   this form. Modify pk7_asn1.c to support indefinite length
11407   encoding. This is experimental and needs additional code to
11408   be useful, such as an ASN1 bio and some enhanced streaming
11409   PKCS#7 code.
11410
11411   Extend template encode functionality so that tagging is passed
11412   down to the template encoder.
11413
11414   *Steve Henson*
11415
11416 * Let 'openssl req' fail if an argument to '-newkey' is not
11417   recognized instead of using RSA as a default.
11418
11419   *Bodo Moeller*
11420
11421 * Add support for ECC-based ciphersuites from draft-ietf-tls-ecc-01.txt.
11422   As these are not official, they are not included in "ALL";
11423   the "ECCdraft" ciphersuite group alias can be used to select them.
11424
11425   *Vipul Gupta and Sumit Gupta (Sun Microsystems Laboratories)*
11426
11427 * Add ECDH engine support.
11428
11429   *Nils Gura and Douglas Stebila (Sun Microsystems Laboratories)*
11430
11431 * Add ECDH in new directory crypto/ecdh/.
11432
11433   *Douglas Stebila (Sun Microsystems Laboratories)*
11434
11435 * Let BN_rand_range() abort with an error after 100 iterations
11436   without success (which indicates a broken PRNG).
11437
11438   *Bodo Moeller*
11439
11440 * Change BN_mod_sqrt() so that it verifies that the input value
11441   is really the square of the return value.  (Previously,
11442   BN_mod_sqrt would show GIGO behaviour.)
11443
11444   *Bodo Moeller*
11445
11446 * Add named elliptic curves over binary fields from X9.62, SECG,
11447   and WAP/WTLS; add OIDs that were still missing.
11448
11449   *Sheueling Chang Shantz and Douglas Stebila (Sun Microsystems Laboratories)*
11450
11451 * Extend the EC library for elliptic curves over binary fields
11452   (new files ec2_smpl.c, ec2_smpt.c, ec2_mult.c in crypto/ec/).
11453   New EC_METHOD:
11454
11455           EC_GF2m_simple_method
11456
11457   New API functions:
11458
11459           EC_GROUP_new_curve_GF2m
11460           EC_GROUP_set_curve_GF2m
11461           EC_GROUP_get_curve_GF2m
11462           EC_POINT_set_affine_coordinates_GF2m
11463           EC_POINT_get_affine_coordinates_GF2m
11464           EC_POINT_set_compressed_coordinates_GF2m
11465
11466   Point compression for binary fields is disabled by default for
11467   patent reasons (compile with OPENSSL_EC_BIN_PT_COMP defined to
11468   enable it).
11469
11470   As binary polynomials are represented as BIGNUMs, various members
11471   of the EC_GROUP and EC_POINT data structures can be shared
11472   between the implementations for prime fields and binary fields;
11473   the above `..._GF2m functions` (except for EX_GROUP_new_curve_GF2m)
11474   are essentially identical to their `..._GFp` counterparts.
11475   (For simplicity, the `..._GFp` prefix has been dropped from
11476   various internal method names.)
11477
11478   An internal 'field_div' method (similar to 'field_mul' and
11479   'field_sqr') has been added; this is used only for binary fields.
11480
11481   *Sheueling Chang Shantz and Douglas Stebila (Sun Microsystems Laboratories)*
11482
11483 * Optionally dispatch EC_POINT_mul(), EC_POINT_precompute_mult()
11484   through methods ('mul', 'precompute_mult').
11485
11486   The generic implementations (now internally called 'ec_wNAF_mul'
11487   and 'ec_wNAF_precomputed_mult') remain the default if these
11488   methods are undefined.
11489
11490   *Sheueling Chang Shantz and Douglas Stebila (Sun Microsystems Laboratories)*
11491
11492 * New function EC_GROUP_get_degree, which is defined through
11493   EC_METHOD.  For curves over prime fields, this returns the bit
11494   length of the modulus.
11495
11496   *Sheueling Chang Shantz and Douglas Stebila (Sun Microsystems Laboratories)*
11497
11498 * New functions EC_GROUP_dup, EC_POINT_dup.
11499   (These simply call ..._new  and ..._copy).
11500
11501   *Sheueling Chang Shantz and Douglas Stebila (Sun Microsystems Laboratories)*
11502
11503 * Add binary polynomial arithmetic software in crypto/bn/bn_gf2m.c.
11504   Polynomials are represented as BIGNUMs (where the sign bit is not
11505   used) in the following functions [macros]:
11506
11507           BN_GF2m_add
11508           BN_GF2m_sub             [= BN_GF2m_add]
11509           BN_GF2m_mod             [wrapper for BN_GF2m_mod_arr]
11510           BN_GF2m_mod_mul         [wrapper for BN_GF2m_mod_mul_arr]
11511           BN_GF2m_mod_sqr         [wrapper for BN_GF2m_mod_sqr_arr]
11512           BN_GF2m_mod_inv
11513           BN_GF2m_mod_exp         [wrapper for BN_GF2m_mod_exp_arr]
11514           BN_GF2m_mod_sqrt        [wrapper for BN_GF2m_mod_sqrt_arr]
11515           BN_GF2m_mod_solve_quad  [wrapper for BN_GF2m_mod_solve_quad_arr]
11516           BN_GF2m_cmp             [= BN_ucmp]
11517
11518   (Note that only the 'mod' functions are actually for fields GF(2^m).
11519   BN_GF2m_add() is misnomer, but this is for the sake of consistency.)
11520
11521   For some functions, an the irreducible polynomial defining a
11522   field can be given as an 'unsigned int[]' with strictly
11523   decreasing elements giving the indices of those bits that are set;
11524   i.e., p[] represents the polynomial
11525           f(t) = t^p[0] + t^p[1] + ... + t^p[k]
11526   where
11527           p[0] > p[1] > ... > p[k] = 0.
11528   This applies to the following functions:
11529
11530           BN_GF2m_mod_arr
11531           BN_GF2m_mod_mul_arr
11532           BN_GF2m_mod_sqr_arr
11533           BN_GF2m_mod_inv_arr        [wrapper for BN_GF2m_mod_inv]
11534           BN_GF2m_mod_div_arr        [wrapper for BN_GF2m_mod_div]
11535           BN_GF2m_mod_exp_arr
11536           BN_GF2m_mod_sqrt_arr
11537           BN_GF2m_mod_solve_quad_arr
11538           BN_GF2m_poly2arr
11539           BN_GF2m_arr2poly
11540
11541   Conversion can be performed by the following functions:
11542
11543           BN_GF2m_poly2arr
11544           BN_GF2m_arr2poly
11545
11546   bntest.c has additional tests for binary polynomial arithmetic.
11547
11548   Two implementations for BN_GF2m_mod_div() are available.
11549   The default algorithm simply uses BN_GF2m_mod_inv() and
11550   BN_GF2m_mod_mul().  The alternative algorithm is compiled in only
11551   if OPENSSL_SUN_GF2M_DIV is defined (patent pending; read the
11552   copyright notice in crypto/bn/bn_gf2m.c before enabling it).
11553
11554   *Sheueling Chang Shantz and Douglas Stebila (Sun Microsystems Laboratories)*
11555
11556 * Add new error code 'ERR_R_DISABLED' that can be used when some
11557   functionality is disabled at compile-time.
11558
11559   *Douglas Stebila <douglas.stebila@sun.com>*
11560
11561 * Change default behaviour of 'openssl asn1parse' so that more
11562   information is visible when viewing, e.g., a certificate:
11563
11564   Modify asn1_parse2 (crypto/asn1/asn1_par.c) so that in non-'dump'
11565   mode the content of non-printable OCTET STRINGs is output in a
11566   style similar to INTEGERs, but with '[HEX DUMP]' prepended to
11567   avoid the appearance of a printable string.
11568
11569   *Nils Larsch <nla@trustcenter.de>*
11570
11571 * Add 'asn1_flag' and 'asn1_form' member to EC_GROUP with access
11572   functions
11573           EC_GROUP_set_asn1_flag()
11574           EC_GROUP_get_asn1_flag()
11575           EC_GROUP_set_point_conversion_form()
11576           EC_GROUP_get_point_conversion_form()
11577   These control ASN1 encoding details:
11578   - Curves (i.e., groups) are encoded explicitly unless asn1_flag
11579     has been set to OPENSSL_EC_NAMED_CURVE.
11580   - Points are encoded in uncompressed form by default; options for
11581     asn1_for are as for point2oct, namely
11582           POINT_CONVERSION_COMPRESSED
11583           POINT_CONVERSION_UNCOMPRESSED
11584           POINT_CONVERSION_HYBRID
11585
11586   Also add 'seed' and 'seed_len' members to EC_GROUP with access
11587   functions
11588           EC_GROUP_set_seed()
11589           EC_GROUP_get0_seed()
11590           EC_GROUP_get_seed_len()
11591   This is used only for ASN1 purposes (so far).
11592
11593   *Nils Larsch <nla@trustcenter.de>*
11594
11595 * Add 'field_type' member to EC_METHOD, which holds the NID
11596   of the appropriate field type OID.  The new function
11597   EC_METHOD_get_field_type() returns this value.
11598
11599   *Nils Larsch <nla@trustcenter.de>*
11600
11601 * Add functions
11602           EC_POINT_point2bn()
11603           EC_POINT_bn2point()
11604           EC_POINT_point2hex()
11605           EC_POINT_hex2point()
11606   providing useful interfaces to EC_POINT_point2oct() and
11607   EC_POINT_oct2point().
11608
11609   *Nils Larsch <nla@trustcenter.de>*
11610
11611 * Change internals of the EC library so that the functions
11612           EC_GROUP_set_generator()
11613           EC_GROUP_get_generator()
11614           EC_GROUP_get_order()
11615           EC_GROUP_get_cofactor()
11616   are implemented directly in crypto/ec/ec_lib.c and not dispatched
11617   to methods, which would lead to unnecessary code duplication when
11618   adding different types of curves.
11619
11620   *Nils Larsch <nla@trustcenter.de> with input by Bodo Moeller*
11621
11622 * Implement compute_wNAF (crypto/ec/ec_mult.c) without BIGNUM
11623   arithmetic, and such that modified wNAFs are generated
11624   (which avoid length expansion in many cases).
11625
11626   *Bodo Moeller*
11627
11628 * Add a function EC_GROUP_check_discriminant() (defined via
11629   EC_METHOD) that verifies that the curve discriminant is non-zero.
11630
11631   Add a function EC_GROUP_check() that makes some sanity tests
11632   on a EC_GROUP, its generator and order.  This includes
11633   EC_GROUP_check_discriminant().
11634
11635   *Nils Larsch <nla@trustcenter.de>*
11636
11637 * Add ECDSA in new directory crypto/ecdsa/.
11638
11639   Add applications 'openssl ecparam' and 'openssl ecdsa'
11640   (these are based on 'openssl dsaparam' and 'openssl dsa').
11641
11642   ECDSA support is also included in various other files across the
11643   library.  Most notably,
11644   - 'openssl req' now has a '-newkey ecdsa:file' option;
11645   - EVP_PKCS82PKEY (crypto/evp/evp_pkey.c) now can handle ECDSA;
11646   - X509_PUBKEY_get (crypto/asn1/x_pubkey.c) and
11647     d2i_PublicKey (crypto/asn1/d2i_pu.c) have been modified to make
11648     them suitable for ECDSA where domain parameters must be
11649     extracted before the specific public key;
11650   - ECDSA engine support has been added.
11651
11652   *Nils Larsch <nla@trustcenter.de>*
11653
11654 * Include some named elliptic curves, and add OIDs from X9.62,
11655   SECG, and WAP/WTLS.  Each curve can be obtained from the new
11656   function
11657           EC_GROUP_new_by_curve_name(),
11658   and the list of available named curves can be obtained with
11659           EC_get_builtin_curves().
11660   Also add a 'curve_name' member to EC_GROUP objects, which can be
11661   accessed via
11662           EC_GROUP_set_curve_name()
11663           EC_GROUP_get_curve_name()
11664
11665   *Nils Larsch <larsch@trustcenter.de, Bodo Moeller*
11666
11667 * Remove a few calls to bn_wexpand() in BN_sqr() (the one in there
11668   was actually never needed) and in BN_mul().  The removal in BN_mul()
11669   required a small change in bn_mul_part_recursive() and the addition
11670   of the functions bn_cmp_part_words(), bn_sub_part_words() and
11671   bn_add_part_words(), which do the same thing as bn_cmp_words(),
11672   bn_sub_words() and bn_add_words() except they take arrays with
11673   differing sizes.
11674
11675   *Richard Levitte*
11676
11677### Changes between 0.9.7l and 0.9.7m  [23 Feb 2007]
11678
11679 * Cleanse PEM buffers before freeing them since they may contain
11680   sensitive data.
11681
11682   *Benjamin Bennett <ben@psc.edu>*
11683
11684 * Include "!eNULL" in SSL_DEFAULT_CIPHER_LIST to make sure that
11685   a ciphersuite string such as "DEFAULT:RSA" cannot enable
11686   authentication-only ciphersuites.
11687
11688   *Bodo Moeller*
11689
11690 * Since AES128 and AES256 share a single mask bit in the logic of
11691   ssl/ssl_ciph.c, the code for masking out disabled ciphers needs a
11692   kludge to work properly if AES128 is available and AES256 isn't.
11693
11694   *Victor Duchovni*
11695
11696 * Expand security boundary to match 1.1.1 module.
11697
11698   *Steve Henson*
11699
11700 * Remove redundant features: hash file source, editing of test vectors
11701   modify fipsld to use external fips_premain.c signature.
11702
11703   *Steve Henson*
11704
11705 * New perl script mkfipsscr.pl to create shell scripts or batch files to
11706   run algorithm test programs.
11707
11708   *Steve Henson*
11709
11710 * Make algorithm test programs more tolerant of whitespace.
11711
11712   *Steve Henson*
11713
11714 * Have SSL/TLS server implementation tolerate "mismatched" record
11715   protocol version while receiving ClientHello even if the
11716   ClientHello is fragmented.  (The server can't insist on the
11717   particular protocol version it has chosen before the ServerHello
11718   message has informed the client about his choice.)
11719
11720   *Bodo Moeller*
11721
11722 * Load error codes if they are not already present instead of using a
11723   static variable. This allows them to be cleanly unloaded and reloaded.
11724
11725   *Steve Henson*
11726
11727### Changes between 0.9.7k and 0.9.7l  [28 Sep 2006]
11728
11729 * Introduce limits to prevent malicious keys being able to
11730   cause a denial of service.  ([CVE-2006-2940])
11731
11732   *Steve Henson, Bodo Moeller*
11733
11734 * Fix ASN.1 parsing of certain invalid structures that can result
11735   in a denial of service.  ([CVE-2006-2937])  [Steve Henson]
11736
11737 * Fix buffer overflow in SSL_get_shared_ciphers() function.
11738   ([CVE-2006-3738]) [Tavis Ormandy and Will Drewry, Google Security Team]
11739
11740 * Fix SSL client code which could crash if connecting to a
11741   malicious SSLv2 server.  ([CVE-2006-4343])
11742
11743   *Tavis Ormandy and Will Drewry, Google Security Team*
11744
11745 * Change ciphersuite string processing so that an explicit
11746   ciphersuite selects this one ciphersuite (so that "AES256-SHA"
11747   will no longer include "AES128-SHA"), and any other similar
11748   ciphersuite (same bitmap) from *other* protocol versions (so that
11749   "RC4-MD5" will still include both the SSL 2.0 ciphersuite and the
11750   SSL 3.0/TLS 1.0 ciphersuite).  This is a backport combining
11751   changes from 0.9.8b and 0.9.8d.
11752
11753   *Bodo Moeller*
11754
11755### Changes between 0.9.7j and 0.9.7k  [05 Sep 2006]
11756
11757 * Avoid PKCS #1 v1.5 signature attack discovered by Daniel Bleichenbacher
11758   ([CVE-2006-4339])  [Ben Laurie and Google Security Team]
11759
11760 * Change the Unix randomness entropy gathering to use poll() when
11761   possible instead of select(), since the latter has some
11762   undesirable limitations.
11763
11764   *Darryl Miles via Richard Levitte and Bodo Moeller*
11765
11766 * Disable rogue ciphersuites:
11767
11768   - SSLv2 0x08 0x00 0x80 ("RC4-64-MD5")
11769   - SSLv3/TLSv1 0x00 0x61 ("EXP1024-RC2-CBC-MD5")
11770   - SSLv3/TLSv1 0x00 0x60 ("EXP1024-RC4-MD5")
11771
11772   The latter two were purportedly from
11773   draft-ietf-tls-56-bit-ciphersuites-0[01].txt, but do not really
11774   appear there.
11775
11776   Also deactivate the remaining ciphersuites from
11777   draft-ietf-tls-56-bit-ciphersuites-01.txt.  These are just as
11778   unofficial, and the ID has long expired.
11779
11780   *Bodo Moeller*
11781
11782 * Fix RSA blinding Heisenbug (problems sometimes occurred on
11783   dual-core machines) and other potential thread-safety issues.
11784
11785   *Bodo Moeller*
11786
11787### Changes between 0.9.7i and 0.9.7j  [04 May 2006]
11788
11789 * Adapt fipsld and the build system to link against the validated FIPS
11790   module in FIPS mode.
11791
11792   *Steve Henson*
11793
11794 * Fixes for VC++ 2005 build under Windows.
11795
11796   *Steve Henson*
11797
11798 * Add new Windows build target VC-32-GMAKE for VC++. This uses GNU make
11799   from a Windows bash shell such as MSYS. It is autodetected from the
11800   "config" script when run from a VC++ environment. Modify standard VC++
11801   build to use fipscanister.o from the GNU make build.
11802
11803   *Steve Henson*
11804
11805### Changes between 0.9.7h and 0.9.7i  [14 Oct 2005]
11806
11807 * Wrapped the definition of EVP_MAX_MD_SIZE in a #ifdef OPENSSL_FIPS.
11808   The value now differs depending on if you build for FIPS or not.
11809   BEWARE!  A program linked with a shared FIPSed libcrypto can't be
11810   safely run with a non-FIPSed libcrypto, as it may crash because of
11811   the difference induced by this change.
11812
11813   *Andy Polyakov*
11814
11815### Changes between 0.9.7g and 0.9.7h  [11 Oct 2005]
11816
11817 * Remove the functionality of SSL_OP_MSIE_SSLV2_RSA_PADDING
11818   (part of SSL_OP_ALL).  This option used to disable the
11819   countermeasure against man-in-the-middle protocol-version
11820   rollback in the SSL 2.0 server implementation, which is a bad
11821   idea.  ([CVE-2005-2969])
11822
11823   *Bodo Moeller; problem pointed out by Yutaka Oiwa (Research Center
11824   for Information Security, National Institute of Advanced Industrial
11825   Science and Technology [AIST, Japan)]*
11826
11827 * Minimal support for X9.31 signatures and PSS padding modes. This is
11828   mainly for FIPS compliance and not fully integrated at this stage.
11829
11830   *Steve Henson*
11831
11832 * For DSA signing, unless DSA_FLAG_NO_EXP_CONSTTIME is set, perform
11833   the exponentiation using a fixed-length exponent.  (Otherwise,
11834   the information leaked through timing could expose the secret key
11835   after many signatures; cf. Bleichenbacher's attack on DSA with
11836   biased k.)
11837
11838   *Bodo Moeller*
11839
11840 * Make a new fixed-window mod_exp implementation the default for
11841   RSA, DSA, and DH private-key operations so that the sequence of
11842   squares and multiplies and the memory access pattern are
11843   independent of the particular secret key.  This will mitigate
11844   cache-timing and potential related attacks.
11845
11846   BN_mod_exp_mont_consttime() is the new exponentiation implementation,
11847   and this is automatically used by BN_mod_exp_mont() if the new flag
11848   BN_FLG_EXP_CONSTTIME is set for the exponent.  RSA, DSA, and DH
11849   will use this BN flag for private exponents unless the flag
11850   RSA_FLAG_NO_EXP_CONSTTIME, DSA_FLAG_NO_EXP_CONSTTIME, or
11851   DH_FLAG_NO_EXP_CONSTTIME, respectively, is set.
11852
11853   *Matthew D Wood (Intel Corp), with some changes by Bodo Moeller*
11854
11855 * Change the client implementation for SSLv23_method() and
11856   SSLv23_client_method() so that is uses the SSL 3.0/TLS 1.0
11857   Client Hello message format if the SSL_OP_NO_SSLv2 option is set.
11858   (Previously, the SSL 2.0 backwards compatible Client Hello
11859   message format would be used even with SSL_OP_NO_SSLv2.)
11860
11861   *Bodo Moeller*
11862
11863 * Add support for smime-type MIME parameter in S/MIME messages which some
11864   clients need.
11865
11866   *Steve Henson*
11867
11868 * New function BN_MONT_CTX_set_locked() to set montgomery parameters in
11869   a threadsafe manner. Modify rsa code to use new function and add calls
11870   to dsa and dh code (which had race conditions before).
11871
11872   *Steve Henson*
11873
11874 * Include the fixed error library code in the C error file definitions
11875   instead of fixing them up at runtime. This keeps the error code
11876   structures constant.
11877
11878   *Steve Henson*
11879
11880### Changes between 0.9.7f and 0.9.7g  [11 Apr 2005]
11881
11882[NB: OpenSSL 0.9.7h and later 0.9.7 patch levels were released after
11883OpenSSL 0.9.8.]
11884
11885 * Fixes for newer kerberos headers. NB: the casts are needed because
11886   the 'length' field is signed on one version and unsigned on another
11887   with no (?) obvious way to tell the difference, without these VC++
11888   complains. Also the "definition" of FAR (blank) is no longer included
11889   nor is the error ENOMEM. KRB5_PRIVATE has to be set to 1 to pick up
11890   some needed definitions.
11891
11892   *Steve Henson*
11893
11894 * Undo Cygwin change.
11895
11896   *Ulf Möller*
11897
11898 * Added support for proxy certificates according to RFC 3820.
11899   Because they may be a security thread to unaware applications,
11900   they must be explicitly allowed in run-time.  See
11901   docs/HOWTO/proxy_certificates.txt for further information.
11902
11903   *Richard Levitte*
11904
11905### Changes between 0.9.7e and 0.9.7f  [22 Mar 2005]
11906
11907 * Use (SSL_RANDOM_VALUE - 4) bytes of pseudo random data when generating
11908   server and client random values. Previously
11909   (SSL_RANDOM_VALUE - sizeof(time_t)) would be used which would result in
11910   less random data when sizeof(time_t) > 4 (some 64 bit platforms).
11911
11912   This change has negligible security impact because:
11913
11914   1. Server and client random values still have 24 bytes of pseudo random
11915      data.
11916
11917   2. Server and client random values are sent in the clear in the initial
11918      handshake.
11919
11920   3. The master secret is derived using the premaster secret (48 bytes in
11921      size for static RSA ciphersuites) as well as client server and random
11922      values.
11923
11924   The OpenSSL team would like to thank the UK NISCC for bringing this issue
11925   to our attention.
11926
11927   *Stephen Henson, reported by UK NISCC*
11928
11929 * Use Windows randomness collection on Cygwin.
11930
11931   *Ulf Möller*
11932
11933 * Fix hang in EGD/PRNGD query when communication socket is closed
11934   prematurely by EGD/PRNGD.
11935
11936   *Darren Tucker <dtucker@zip.com.au> via Lutz Jänicke, resolves #1014*
11937
11938 * Prompt for pass phrases when appropriate for PKCS12 input format.
11939
11940   *Steve Henson*
11941
11942 * Back-port of selected performance improvements from development
11943   branch, as well as improved support for PowerPC platforms.
11944
11945   *Andy Polyakov*
11946
11947 * Add lots of checks for memory allocation failure, error codes to indicate
11948   failure and freeing up memory if a failure occurs.
11949
11950   *Nauticus Networks SSL Team <openssl@nauticusnet.com>, Steve Henson*
11951
11952 * Add new -passin argument to dgst.
11953
11954   *Steve Henson*
11955
11956 * Perform some character comparisons of different types in X509_NAME_cmp:
11957   this is needed for some certificates that re-encode DNs into UTF8Strings
11958   (in violation of RFC3280) and can't or won't issue name rollover
11959   certificates.
11960
11961   *Steve Henson*
11962
11963 * Make an explicit check during certificate validation to see that
11964   the CA setting in each certificate on the chain is correct.  As a
11965   side effect always do the following basic checks on extensions,
11966   not just when there's an associated purpose to the check:
11967
11968   - if there is an unhandled critical extension (unless the user
11969     has chosen to ignore this fault)
11970   - if the path length has been exceeded (if one is set at all)
11971   - that certain extensions fit the associated purpose (if one has
11972     been given)
11973
11974   *Richard Levitte*
11975
11976### Changes between 0.9.7d and 0.9.7e  [25 Oct 2004]
11977
11978 * Avoid a race condition when CRLs are checked in a multi threaded
11979   environment. This would happen due to the reordering of the revoked
11980   entries during signature checking and serial number lookup. Now the
11981   encoding is cached and the serial number sort performed under a lock.
11982   Add new STACK function sk_is_sorted().
11983
11984   *Steve Henson*
11985
11986 * Add Delta CRL to the extension code.
11987
11988   *Steve Henson*
11989
11990 * Various fixes to s3_pkt.c so alerts are sent properly.
11991
11992   *David Holmes <d.holmes@f5.com>*
11993
11994 * Reduce the chances of duplicate issuer name and serial numbers (in
11995   violation of RFC3280) using the OpenSSL certificate creation utilities.
11996   This is done by creating a random 64 bit value for the initial serial
11997   number when a serial number file is created or when a self signed
11998   certificate is created using 'openssl req -x509'. The initial serial
11999   number file is created using 'openssl x509 -next_serial' in CA.pl
12000   rather than being initialized to 1.
12001
12002   *Steve Henson*
12003
12004### Changes between 0.9.7c and 0.9.7d  [17 Mar 2004]
12005
12006 * Fix null-pointer assignment in do_change_cipher_spec() revealed
12007   by using the Codenomicon TLS Test Tool ([CVE-2004-0079])
12008
12009   *Joe Orton, Steve Henson*
12010
12011 * Fix flaw in SSL/TLS handshaking when using Kerberos ciphersuites
12012   ([CVE-2004-0112])
12013
12014   *Joe Orton, Steve Henson*
12015
12016 * Make it possible to have multiple active certificates with the same
12017   subject in the CA index file.  This is done only if the keyword
12018   'unique_subject' is set to 'no' in the main CA section (default
12019   if 'CA_default') of the configuration file.  The value is saved
12020   with the database itself in a separate index attribute file,
12021   named like the index file with '.attr' appended to the name.
12022
12023   *Richard Levitte*
12024
12025 * X509 verify fixes. Disable broken certificate workarounds when
12026   X509_V_FLAGS_X509_STRICT is set. Check CRL issuer has cRLSign set if
12027   keyUsage extension present. Don't accept CRLs with unhandled critical
12028   extensions: since verify currently doesn't process CRL extensions this
12029   rejects a CRL with *any* critical extensions. Add new verify error codes
12030   for these cases.
12031
12032   *Steve Henson*
12033
12034 * When creating an OCSP nonce use an OCTET STRING inside the extnValue.
12035   A clarification of RFC2560 will require the use of OCTET STRINGs and
12036   some implementations cannot handle the current raw format. Since OpenSSL
12037   copies and compares OCSP nonces as opaque blobs without any attempt at
12038   parsing them this should not create any compatibility issues.
12039
12040   *Steve Henson*
12041
12042 * New md flag EVP_MD_CTX_FLAG_REUSE this allows md_data to be reused when
12043   calling EVP_MD_CTX_copy_ex() to avoid calling OPENSSL_malloc(). Without
12044   this HMAC (and other) operations are several times slower than OpenSSL
12045   < 0.9.7.
12046
12047   *Steve Henson*
12048
12049 * Print out GeneralizedTime and UTCTime in ASN1_STRING_print_ex().
12050
12051   *Peter Sylvester <Peter.Sylvester@EdelWeb.fr>*
12052
12053 * Use the correct content when signing type "other".
12054
12055   *Steve Henson*
12056
12057### Changes between 0.9.7b and 0.9.7c  [30 Sep 2003]
12058
12059 * Fix various bugs revealed by running the NISCC test suite:
12060
12061   Stop out of bounds reads in the ASN1 code when presented with
12062   invalid tags (CVE-2003-0543 and CVE-2003-0544).
12063
12064   Free up ASN1_TYPE correctly if ANY type is invalid ([CVE-2003-0545]).
12065
12066   If verify callback ignores invalid public key errors don't try to check
12067   certificate signature with the NULL public key.
12068
12069   *Steve Henson*
12070
12071 * New -ignore_err option in ocsp application to stop the server
12072   exiting on the first error in a request.
12073
12074   *Steve Henson*
12075
12076 * In ssl3_accept() (ssl/s3_srvr.c) only accept a client certificate
12077   if the server requested one: as stated in TLS 1.0 and SSL 3.0
12078   specifications.
12079
12080   *Steve Henson*
12081
12082 * In ssl3_get_client_hello() (ssl/s3_srvr.c), tolerate additional
12083   extra data after the compression methods not only for TLS 1.0
12084   but also for SSL 3.0 (as required by the specification).
12085
12086   *Bodo Moeller; problem pointed out by Matthias Loepfe*
12087
12088 * Change X509_certificate_type() to mark the key as exported/exportable
12089   when it's 512 *bits* long, not 512 bytes.
12090
12091   *Richard Levitte*
12092
12093 * Change AES_cbc_encrypt() so it outputs exact multiple of
12094   blocks during encryption.
12095
12096   *Richard Levitte*
12097
12098 * Various fixes to base64 BIO and non blocking I/O. On write
12099   flushes were not handled properly if the BIO retried. On read
12100   data was not being buffered properly and had various logic bugs.
12101   This also affects blocking I/O when the data being decoded is a
12102   certain size.
12103
12104   *Steve Henson*
12105
12106 * Various S/MIME bugfixes and compatibility changes:
12107   output correct application/pkcs7 MIME type if
12108   PKCS7_NOOLDMIMETYPE is set. Tolerate some broken signatures.
12109   Output CR+LF for EOL if PKCS7_CRLFEOL is set (this makes opening
12110   of files as .eml work). Correctly handle very long lines in MIME
12111   parser.
12112
12113   *Steve Henson*
12114
12115### Changes between 0.9.7a and 0.9.7b  [10 Apr 2003]
12116
12117 * Countermeasure against the Klima-Pokorny-Rosa extension of
12118   Bleichbacher's attack on PKCS #1 v1.5 padding: treat
12119   a protocol version number mismatch like a decryption error
12120   in ssl3_get_client_key_exchange (ssl/s3_srvr.c).
12121
12122   *Bodo Moeller*
12123
12124 * Turn on RSA blinding by default in the default implementation
12125   to avoid a timing attack. Applications that don't want it can call
12126   RSA_blinding_off() or use the new flag RSA_FLAG_NO_BLINDING.
12127   They would be ill-advised to do so in most cases.
12128
12129   *Ben Laurie, Steve Henson, Geoff Thorpe, Bodo Moeller*
12130
12131 * Change RSA blinding code so that it works when the PRNG is not
12132   seeded (in this case, the secret RSA exponent is abused as
12133   an unpredictable seed -- if it is not unpredictable, there
12134   is no point in blinding anyway).  Make RSA blinding thread-safe
12135   by remembering the creator's thread ID in rsa->blinding and
12136   having all other threads use local one-time blinding factors
12137   (this requires more computation than sharing rsa->blinding, but
12138   avoids excessive locking; and if an RSA object is not shared
12139   between threads, blinding will still be very fast).
12140
12141   *Bodo Moeller*
12142
12143 * Fixed a typo bug that would cause ENGINE_set_default() to set an
12144   ENGINE as defaults for all supported algorithms irrespective of
12145   the 'flags' parameter. 'flags' is now honoured, so applications
12146   should make sure they are passing it correctly.
12147
12148   *Geoff Thorpe*
12149
12150 * Target "mingw" now allows native Windows code to be generated in
12151   the Cygwin environment as well as with the MinGW compiler.
12152
12153   *Ulf Moeller*
12154
12155### Changes between 0.9.7 and 0.9.7a  [19 Feb 2003]
12156
12157 * In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked
12158   via timing by performing a MAC computation even if incorrect
12159   block cipher padding has been found.  This is a countermeasure
12160   against active attacks where the attacker has to distinguish
12161   between bad padding and a MAC verification error. ([CVE-2003-0078])
12162
12163   *Bodo Moeller; problem pointed out by Brice Canvel (EPFL),
12164   Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and
12165   Martin Vuagnoux (EPFL, Ilion)*
12166
12167 * Make the no-err option work as intended.  The intention with no-err
12168   is not to have the whole error stack handling routines removed from
12169   libcrypto, it's only intended to remove all the function name and
12170   reason texts, thereby removing some of the footprint that may not
12171   be interesting if those errors aren't displayed anyway.
12172
12173   NOTE: it's still possible for any application or module to have its
12174   own set of error texts inserted.  The routines are there, just not
12175   used by default when no-err is given.
12176
12177   *Richard Levitte*
12178
12179 * Add support for FreeBSD on IA64.
12180
12181   *dirk.meyer@dinoex.sub.org via Richard Levitte, resolves #454*
12182
12183 * Adjust DES_cbc_cksum() so it returns the same value as the MIT
12184   Kerberos function mit_des_cbc_cksum().  Before this change,
12185   the value returned by DES_cbc_cksum() was like the one from
12186   mit_des_cbc_cksum(), except the bytes were swapped.
12187
12188   *Kevin Greaney <Kevin.Greaney@hp.com> and Richard Levitte*
12189
12190 * Allow an application to disable the automatic SSL chain building.
12191   Before this a rather primitive chain build was always performed in
12192   ssl3_output_cert_chain(): an application had no way to send the
12193   correct chain if the automatic operation produced an incorrect result.
12194
12195   Now the chain builder is disabled if either:
12196
12197   1. Extra certificates are added via SSL_CTX_add_extra_chain_cert().
12198
12199   2. The mode flag SSL_MODE_NO_AUTO_CHAIN is set.
12200
12201   The reasoning behind this is that an application would not want the
12202   auto chain building to take place if extra chain certificates are
12203   present and it might also want a means of sending no additional
12204   certificates (for example the chain has two certificates and the
12205   root is omitted).
12206
12207   *Steve Henson*
12208
12209 * Add the possibility to build without the ENGINE framework.
12210
12211   *Steven Reddie <smr@essemer.com.au> via Richard Levitte*
12212
12213 * Under Win32 gmtime() can return NULL: check return value in
12214   OPENSSL_gmtime(). Add error code for case where gmtime() fails.
12215
12216   *Steve Henson*
12217
12218 * DSA routines: under certain error conditions uninitialized BN objects
12219   could be freed. Solution: make sure initialization is performed early
12220   enough. (Reported and fix supplied by Ivan D Nestlerode <nestler@MIT.EDU>,
12221   Nils Larsch <nla@trustcenter.de> via PR#459)
12222
12223   *Lutz Jaenicke*
12224
12225 * Another fix for SSLv2 session ID handling: the session ID was incorrectly
12226   checked on reconnect on the client side, therefore session resumption
12227   could still fail with a "ssl session id is different" error. This
12228   behaviour is masked when SSL_OP_ALL is used due to
12229   SSL_OP_MICROSOFT_SESS_ID_BUG being set.
12230   Behaviour observed by Crispin Flowerday <crispin@flowerday.cx> as
12231   followup to PR #377.
12232
12233   *Lutz Jaenicke*
12234
12235 * IA-32 assembler support enhancements: unified ELF targets, support
12236   for SCO/Caldera platforms, fix for Cygwin shared build.
12237
12238   *Andy Polyakov*
12239
12240 * Add support for FreeBSD on sparc64.  As a consequence, support for
12241   FreeBSD on non-x86 processors is separate from x86 processors on
12242   the config script, much like the NetBSD support.
12243
12244   *Richard Levitte & Kris Kennaway <kris@obsecurity.org>*
12245
12246### Changes between 0.9.6h and 0.9.7  [31 Dec 2002]
12247
12248[NB: OpenSSL 0.9.6i and later 0.9.6 patch levels were released after
12249OpenSSL 0.9.7.]
12250
12251 * Fix session ID handling in SSLv2 client code: the SERVER FINISHED
12252   code (06) was taken as the first octet of the session ID and the last
12253   octet was ignored consequently. As a result SSLv2 client side session
12254   caching could not have worked due to the session ID mismatch between
12255   client and server.
12256   Behaviour observed by Crispin Flowerday <crispin@flowerday.cx> as
12257   PR #377.
12258
12259   *Lutz Jaenicke*
12260
12261 * Change the declaration of needed Kerberos libraries to use EX_LIBS
12262   instead of the special (and badly supported) LIBKRB5.  LIBKRB5 is
12263   removed entirely.
12264
12265   *Richard Levitte*
12266
12267 * The hw_ncipher.c engine requires dynamic locks.  Unfortunately, it
12268   seems that in spite of existing for more than a year, many application
12269   author have done nothing to provide the necessary callbacks, which
12270   means that this particular engine will not work properly anywhere.
12271   This is a very unfortunate situation which forces us, in the name
12272   of usability, to give the hw_ncipher.c a static lock, which is part
12273   of libcrypto.
12274   NOTE: This is for the 0.9.7 series ONLY.  This hack will never
12275   appear in 0.9.8 or later.  We EXPECT application authors to have
12276   dealt properly with this when 0.9.8 is released (unless we actually
12277   make such changes in the libcrypto locking code that changes will
12278   have to be made anyway).
12279
12280   *Richard Levitte*
12281
12282 * In asn1_d2i_read_bio() repeatedly call BIO_read() until all content
12283   octets have been read, EOF or an error occurs. Without this change
12284   some truncated ASN1 structures will not produce an error.
12285
12286   *Steve Henson*
12287
12288 * Disable Heimdal support, since it hasn't been fully implemented.
12289   Still give the possibility to force the use of Heimdal, but with
12290   warnings and a request that patches get sent to openssl-dev.
12291
12292   *Richard Levitte*
12293
12294 * Add the VC-CE target, introduce the WINCE sysname, and add
12295   INSTALL.WCE and appropriate conditionals to make it build.
12296
12297   *Steven Reddie <smr@essemer.com.au> via Richard Levitte*
12298
12299 * Change the DLL names for Cygwin to cygcrypto-x.y.z.dll and
12300   cygssl-x.y.z.dll, where x, y and z are the major, minor and
12301   edit numbers of the version.
12302
12303   *Corinna Vinschen <vinschen@redhat.com> and Richard Levitte*
12304
12305 * Introduce safe string copy and catenation functions
12306   (BUF_strlcpy() and BUF_strlcat()).
12307
12308   *Ben Laurie (CHATS) and Richard Levitte*
12309
12310 * Avoid using fixed-size buffers for one-line DNs.
12311
12312   *Ben Laurie (CHATS)*
12313
12314 * Add BUF_MEM_grow_clean() to avoid information leakage when
12315   resizing buffers containing secrets, and use where appropriate.
12316
12317   *Ben Laurie (CHATS)*
12318
12319 * Avoid using fixed size buffers for configuration file location.
12320
12321   *Ben Laurie (CHATS)*
12322
12323 * Avoid filename truncation for various CA files.
12324
12325   *Ben Laurie (CHATS)*
12326
12327 * Use sizeof in preference to magic numbers.
12328
12329   *Ben Laurie (CHATS)*
12330
12331 * Avoid filename truncation in cert requests.
12332
12333   *Ben Laurie (CHATS)*
12334
12335 * Add assertions to check for (supposedly impossible) buffer
12336   overflows.
12337
12338   *Ben Laurie (CHATS)*
12339
12340 * Don't cache truncated DNS entries in the local cache (this could
12341   potentially lead to a spoofing attack).
12342
12343   *Ben Laurie (CHATS)*
12344
12345 * Fix various buffers to be large enough for hex/decimal
12346   representations in a platform independent manner.
12347
12348   *Ben Laurie (CHATS)*
12349
12350 * Add CRYPTO_realloc_clean() to avoid information leakage when
12351   resizing buffers containing secrets, and use where appropriate.
12352
12353   *Ben Laurie (CHATS)*
12354
12355 * Add BIO_indent() to avoid much slightly worrying code to do
12356   indents.
12357
12358   *Ben Laurie (CHATS)*
12359
12360 * Convert sprintf()/BIO_puts() to BIO_printf().
12361
12362   *Ben Laurie (CHATS)*
12363
12364 * buffer_gets() could terminate with the buffer only half
12365   full. Fixed.
12366
12367   *Ben Laurie (CHATS)*
12368
12369 * Add assertions to prevent user-supplied crypto functions from
12370   overflowing internal buffers by having large block sizes, etc.
12371
12372   *Ben Laurie (CHATS)*
12373
12374 * New OPENSSL_assert() macro (similar to assert(), but enabled
12375   unconditionally).
12376
12377   *Ben Laurie (CHATS)*
12378
12379 * Eliminate unused copy of key in RC4.
12380
12381   *Ben Laurie (CHATS)*
12382
12383 * Eliminate unused and incorrectly sized buffers for IV in pem.h.
12384
12385   *Ben Laurie (CHATS)*
12386
12387 * Fix off-by-one error in EGD path.
12388
12389   *Ben Laurie (CHATS)*
12390
12391 * If RANDFILE path is too long, ignore instead of truncating.
12392
12393   *Ben Laurie (CHATS)*
12394
12395 * Eliminate unused and incorrectly sized X.509 structure
12396   CBCParameter.
12397
12398   *Ben Laurie (CHATS)*
12399
12400 * Eliminate unused and dangerous function knumber().
12401
12402   *Ben Laurie (CHATS)*
12403
12404 * Eliminate unused and dangerous structure, KSSL_ERR.
12405
12406   *Ben Laurie (CHATS)*
12407
12408 * Protect against overlong session ID context length in an encoded
12409   session object. Since these are local, this does not appear to be
12410   exploitable.
12411
12412   *Ben Laurie (CHATS)*
12413
12414 * Change from security patch (see 0.9.6e below) that did not affect
12415   the 0.9.6 release series:
12416
12417   Remote buffer overflow in SSL3 protocol - an attacker could
12418   supply an oversized master key in Kerberos-enabled versions.
12419   ([CVE-2002-0657])
12420
12421   *Ben Laurie (CHATS)*
12422
12423 * Change the SSL kerb5 codes to match RFC 2712.
12424
12425   *Richard Levitte*
12426
12427 * Make -nameopt work fully for req and add -reqopt switch.
12428
12429   *Michael Bell <michael.bell@rz.hu-berlin.de>, Steve Henson*
12430
12431 * The "block size" for block ciphers in CFB and OFB mode should be 1.
12432
12433   *Steve Henson, reported by Yngve Nysaeter Pettersen <yngve@opera.com>*
12434
12435 * Make sure tests can be performed even if the corresponding algorithms
12436   have been removed entirely.  This was also the last step to make
12437   OpenSSL compilable with DJGPP under all reasonable conditions.
12438
12439   *Richard Levitte, Doug Kaufman <dkaufman@rahul.net>*
12440
12441 * Add cipher selection rules COMPLEMENTOFALL and COMPLEMENTOFDEFAULT
12442   to allow version independent disabling of normally unselected ciphers,
12443   which may be activated as a side-effect of selecting a single cipher.
12444
12445   (E.g., cipher list string "RSA" enables ciphersuites that are left
12446   out of "ALL" because they do not provide symmetric encryption.
12447   "RSA:!COMPLEMEMENTOFALL" avoids these unsafe ciphersuites.)
12448
12449   *Lutz Jaenicke, Bodo Moeller*
12450
12451 * Add appropriate support for separate platform-dependent build
12452   directories.  The recommended way to make a platform-dependent
12453   build directory is the following (tested on Linux), maybe with
12454   some local tweaks:
12455
12456           # Place yourself outside of the OpenSSL source tree.  In
12457           # this example, the environment variable OPENSSL_SOURCE
12458           # is assumed to contain the absolute OpenSSL source directory.
12459           mkdir -p objtree/"`uname -s`-`uname -r`-`uname -m`"
12460           cd objtree/"`uname -s`-`uname -r`-`uname -m`"
12461           (cd $OPENSSL_SOURCE; find . -type f) | while read F; do
12462                   mkdir -p `dirname $F`
12463                   ln -s $OPENSSL_SOURCE/$F $F
12464           done
12465
12466   To be absolutely sure not to disturb the source tree, a "make clean"
12467   is a good thing.  If it isn't successful, don't worry about it,
12468   it probably means the source directory is very clean.
12469
12470   *Richard Levitte*
12471
12472 * Make sure any ENGINE control commands make local copies of string
12473   pointers passed to them whenever necessary. Otherwise it is possible
12474   the caller may have overwritten (or deallocated) the original string
12475   data when a later ENGINE operation tries to use the stored values.
12476
12477   *Götz Babin-Ebell <babinebell@trustcenter.de>*
12478
12479 * Improve diagnostics in file reading and command-line digests.
12480
12481   *Ben Laurie aided and abetted by Solar Designer <solar@openwall.com>*
12482
12483 * Add AES modes CFB and OFB to the object database.  Correct an
12484   error in AES-CFB decryption.
12485
12486   *Richard Levitte*
12487
12488 * Remove most calls to EVP_CIPHER_CTX_cleanup() in evp_enc.c, this
12489   allows existing EVP_CIPHER_CTX structures to be reused after
12490   calling `EVP_*Final()`. This behaviour is used by encryption
12491   BIOs and some applications. This has the side effect that
12492   applications must explicitly clean up cipher contexts with
12493   EVP_CIPHER_CTX_cleanup() or they will leak memory.
12494
12495   *Steve Henson*
12496
12497 * Check the values of dna and dnb in bn_mul_recursive before calling
12498   bn_mul_comba (a non zero value means the a or b arrays do not contain
12499   n2 elements) and fallback to bn_mul_normal if either is not zero.
12500
12501   *Steve Henson*
12502
12503 * Fix escaping of non-ASCII characters when using the -subj option
12504   of the "openssl req" command line tool. (Robert Joop <joop@fokus.gmd.de>)
12505
12506   *Lutz Jaenicke*
12507
12508 * Make object definitions compliant to LDAP (RFC2256): SN is the short
12509   form for "surname", serialNumber has no short form.
12510   Use "mail" as the short name for "rfc822Mailbox" according to RFC2798;
12511   therefore remove "mail" short name for "internet 7".
12512   The OID for unique identifiers in X509 certificates is
12513   x500UniqueIdentifier, not uniqueIdentifier.
12514   Some more OID additions. (Michael Bell <michael.bell@rz.hu-berlin.de>)
12515
12516   *Lutz Jaenicke*
12517
12518 * Add an "init" command to the ENGINE config module and auto initialize
12519   ENGINEs. Without any "init" command the ENGINE will be initialized
12520   after all ctrl commands have been executed on it. If init=1 the
12521   ENGINE is initialized at that point (ctrls before that point are run
12522   on the uninitialized ENGINE and after on the initialized one). If
12523   init=0 then the ENGINE will not be initialized at all.
12524
12525   *Steve Henson*
12526
12527 * Fix the 'app_verify_callback' interface so that the user-defined
12528   argument is actually passed to the callback: In the
12529   SSL_CTX_set_cert_verify_callback() prototype, the callback
12530   declaration has been changed from
12531           int (*cb)()
12532   into
12533           int (*cb)(X509_STORE_CTX *,void *);
12534   in ssl_verify_cert_chain (ssl/ssl_cert.c), the call
12535           i=s->ctx->app_verify_callback(&ctx)
12536   has been changed into
12537           i=s->ctx->app_verify_callback(&ctx, s->ctx->app_verify_arg).
12538
12539   To update applications using SSL_CTX_set_cert_verify_callback(),
12540   a dummy argument can be added to their callback functions.
12541
12542   *D. K. Smetters <smetters@parc.xerox.com>*
12543
12544 * Added the '4758cca' ENGINE to support IBM 4758 cards.
12545
12546   *Maurice Gittens <maurice@gittens.nl>, touchups by Geoff Thorpe*
12547
12548 * Add and OPENSSL_LOAD_CONF define which will cause
12549   OpenSSL_add_all_algorithms() to load the openssl.cnf config file.
12550   This allows older applications to transparently support certain
12551   OpenSSL features: such as crypto acceleration and dynamic ENGINE loading.
12552   Two new functions OPENSSL_add_all_algorithms_noconf() which will never
12553   load the config file and OPENSSL_add_all_algorithms_conf() which will
12554   always load it have also been added.
12555
12556   *Steve Henson*
12557
12558 * Add the OFB, CFB and CTR (all with 128 bit feedback) to AES.
12559   Adjust NIDs and EVP layer.
12560
12561   *Stephen Sprunk <stephen@sprunk.org> and Richard Levitte*
12562
12563 * Config modules support in openssl utility.
12564
12565   Most commands now load modules from the config file,
12566   though in a few (such as version) this isn't done
12567   because it couldn't be used for anything.
12568
12569   In the case of ca and req the config file used is
12570   the same as the utility itself: that is the -config
12571   command line option can be used to specify an
12572   alternative file.
12573
12574   *Steve Henson*
12575
12576 * Move default behaviour from OPENSSL_config(). If appname is NULL
12577   use "openssl_conf" if filename is NULL use default openssl config file.
12578
12579   *Steve Henson*
12580
12581 * Add an argument to OPENSSL_config() to allow the use of an alternative
12582   config section name. Add a new flag to tolerate a missing config file
12583   and move code to CONF_modules_load_file().
12584
12585   *Steve Henson*
12586
12587 * Support for crypto accelerator cards from Accelerated Encryption
12588   Processing, www.aep.ie.  (Use engine 'aep')
12589   The support was copied from 0.9.6c [engine] and adapted/corrected
12590   to work with the new engine framework.
12591
12592   *AEP Inc. and Richard Levitte*
12593
12594 * Support for SureWare crypto accelerator cards from Baltimore
12595   Technologies.  (Use engine 'sureware')
12596   The support was copied from 0.9.6c [engine] and adapted
12597   to work with the new engine framework.
12598
12599   *Richard Levitte*
12600
12601 * Have the CHIL engine fork-safe (as defined by nCipher) and actually
12602   make the newer ENGINE framework commands for the CHIL engine work.
12603
12604   *Toomas Kiisk <vix@cyber.ee> and Richard Levitte*
12605
12606 * Make it possible to produce shared libraries on ReliantUNIX.
12607
12608   *Robert Dahlem <Robert.Dahlem@ffm2.siemens.de> via Richard Levitte*
12609
12610 * Add the configuration target debug-linux-ppro.
12611   Make 'openssl rsa' use the general key loading routines
12612   implemented in `apps.c`, and make those routines able to
12613   handle the key format FORMAT_NETSCAPE and the variant
12614   FORMAT_IISSGC.
12615
12616   *Toomas Kiisk <vix@cyber.ee> via Richard Levitte*
12617
12618 * Fix a crashbug and a logic bug in hwcrhk_load_pubkey().
12619
12620   *Toomas Kiisk <vix@cyber.ee> via Richard Levitte*
12621
12622 * Add -keyform to rsautl, and document -engine.
12623
12624   *Richard Levitte, inspired by Toomas Kiisk <vix@cyber.ee>*
12625
12626 * Change BIO_new_file (crypto/bio/bss_file.c) to use new
12627   BIO_R_NO_SUCH_FILE error code rather than the generic
12628   ERR_R_SYS_LIB error code if fopen() fails with ENOENT.
12629
12630   *Ben Laurie*
12631
12632 * Add new functions
12633           ERR_peek_last_error
12634           ERR_peek_last_error_line
12635           ERR_peek_last_error_line_data.
12636   These are similar to
12637           ERR_peek_error
12638           ERR_peek_error_line
12639           ERR_peek_error_line_data,
12640   but report on the latest error recorded rather than the first one
12641   still in the error queue.
12642
12643   *Ben Laurie, Bodo Moeller*
12644
12645 * default_algorithms option in ENGINE config module. This allows things
12646   like:
12647   default_algorithms = ALL
12648   default_algorithms = RSA, DSA, RAND, CIPHERS, DIGESTS
12649
12650   *Steve Henson*
12651
12652 * Preliminary ENGINE config module.
12653
12654   *Steve Henson*
12655
12656 * New experimental application configuration code.
12657
12658   *Steve Henson*
12659
12660 * Change the AES code to follow the same name structure as all other
12661   symmetric ciphers, and behave the same way.  Move everything to
12662   the directory crypto/aes, thereby obsoleting crypto/rijndael.
12663
12664   *Stephen Sprunk <stephen@sprunk.org> and Richard Levitte*
12665
12666 * SECURITY: remove unsafe setjmp/signal interaction from ui_openssl.c.
12667
12668   *Ben Laurie and Theo de Raadt*
12669
12670 * Add option to output public keys in req command.
12671
12672   *Massimiliano Pala madwolf@openca.org*
12673
12674 * Use wNAFs in EC_POINTs_mul() for improved efficiency
12675   (up to about 10% better than before for P-192 and P-224).
12676
12677   *Bodo Moeller*
12678
12679 * New functions/macros
12680
12681           SSL_CTX_set_msg_callback(ctx, cb)
12682           SSL_CTX_set_msg_callback_arg(ctx, arg)
12683           SSL_set_msg_callback(ssl, cb)
12684           SSL_set_msg_callback_arg(ssl, arg)
12685
12686   to request calling a callback function
12687
12688           void cb(int write_p, int version, int content_type,
12689                   const void *buf, size_t len, SSL *ssl, void *arg)
12690
12691   whenever a protocol message has been completely received
12692   (write_p == 0) or sent (write_p == 1).  Here 'version' is the
12693   protocol version  according to which the SSL library interprets
12694   the current protocol message (SSL2_VERSION, SSL3_VERSION, or
12695   TLS1_VERSION).  'content_type' is 0 in the case of SSL 2.0, or
12696   the content type as defined in the SSL 3.0/TLS 1.0 protocol
12697   specification (change_cipher_spec(20), alert(21), handshake(22)).
12698   'buf' and 'len' point to the actual message, 'ssl' to the
12699   SSL object, and 'arg' is the application-defined value set by
12700   SSL[_CTX]_set_msg_callback_arg().
12701
12702   'openssl s_client' and 'openssl s_server' have new '-msg' options
12703   to enable a callback that displays all protocol messages.
12704
12705   *Bodo Moeller*
12706
12707 * Change the shared library support so shared libraries are built as
12708   soon as the corresponding static library is finished, and thereby get
12709   openssl and the test programs linked against the shared library.
12710   This still only happens when the keyword "shard" has been given to
12711   the configuration scripts.
12712
12713   NOTE: shared library support is still an experimental thing, and
12714   backward binary compatibility is still not guaranteed.
12715
12716   *"Maciej W. Rozycki" <macro@ds2.pg.gda.pl> and Richard Levitte*
12717
12718 * Add support for Subject Information Access extension.
12719
12720   *Peter Sylvester <Peter.Sylvester@EdelWeb.fr>*
12721
12722 * Make BUF_MEM_grow() behaviour more consistent: Initialise to zero
12723   additional bytes when new memory had to be allocated, not just
12724   when reusing an existing buffer.
12725
12726   *Bodo Moeller*
12727
12728 * New command line and configuration option 'utf8' for the req command.
12729   This allows field values to be specified as UTF8 strings.
12730
12731   *Steve Henson*
12732
12733 * Add -multi and -mr options to "openssl speed" - giving multiple parallel
12734   runs for the former and machine-readable output for the latter.
12735
12736   *Ben Laurie*
12737
12738 * Add '-noemailDN' option to 'openssl ca'.  This prevents inclusion
12739   of the e-mail address in the DN (i.e., it will go into a certificate
12740   extension only).  The new configuration file option 'email_in_dn = no'
12741   has the same effect.
12742
12743   *Massimiliano Pala madwolf@openca.org*
12744
12745 * Change all functions with names starting with `des_` to be starting
12746   with `DES_` instead.  Add wrappers that are compatible with libdes,
12747   but are named `_ossl_old_des_*`.  Finally, add macros that map the
12748   `des_*` symbols to the corresponding `_ossl_old_des_*` if libdes
12749   compatibility is desired.  If OpenSSL 0.9.6c compatibility is
12750   desired, the `des_*` symbols will be mapped to `DES_*`, with one
12751   exception.
12752
12753   Since we provide two compatibility mappings, the user needs to
12754   define the macro OPENSSL_DES_LIBDES_COMPATIBILITY if libdes
12755   compatibility is desired.  The default (i.e., when that macro
12756   isn't defined) is OpenSSL 0.9.6c compatibility.
12757
12758   There are also macros that enable and disable the support of old
12759   des functions altogether.  Those are OPENSSL_ENABLE_OLD_DES_SUPPORT
12760   and OPENSSL_DISABLE_OLD_DES_SUPPORT.  If none or both of those
12761   are defined, the default will apply: to support the old des routines.
12762
12763   In either case, one must include openssl/des.h to get the correct
12764   definitions.  Do not try to just include openssl/des_old.h, that
12765   won't work.
12766
12767   NOTE: This is a major break of an old API into a new one.  Software
12768   authors are encouraged to switch to the `DES_` style functions.  Some
12769   time in the future, des_old.h and the libdes compatibility functions
12770   will be disable (i.e. OPENSSL_DISABLE_OLD_DES_SUPPORT will be the
12771   default), and then completely removed.
12772
12773   *Richard Levitte*
12774
12775 * Test for certificates which contain unsupported critical extensions.
12776   If such a certificate is found during a verify operation it is
12777   rejected by default: this behaviour can be overridden by either
12778   handling the new error X509_V_ERR_UNHANDLED_CRITICAL_EXTENSION or
12779   by setting the verify flag X509_V_FLAG_IGNORE_CRITICAL. A new function
12780   X509_supported_extension() has also been added which returns 1 if a
12781   particular extension is supported.
12782
12783   *Steve Henson*
12784
12785 * Modify the behaviour of EVP cipher functions in similar way to digests
12786   to retain compatibility with existing code.
12787
12788   *Steve Henson*
12789
12790 * Modify the behaviour of EVP_DigestInit() and EVP_DigestFinal() to retain
12791   compatibility with existing code. In particular the 'ctx' parameter does
12792   not have to be to be initialized before the call to EVP_DigestInit() and
12793   it is tidied up after a call to EVP_DigestFinal(). New function
12794   EVP_DigestFinal_ex() which does not tidy up the ctx. Similarly function
12795   EVP_MD_CTX_copy() changed to not require the destination to be
12796   initialized valid and new function EVP_MD_CTX_copy_ex() added which
12797   requires the destination to be valid.
12798
12799   Modify all the OpenSSL digest calls to use EVP_DigestInit_ex(),
12800   EVP_DigestFinal_ex() and EVP_MD_CTX_copy_ex().
12801
12802   *Steve Henson*
12803
12804 * Change ssl3_get_message (ssl/s3_both.c) and the functions using it
12805   so that complete 'Handshake' protocol structures are kept in memory
12806   instead of overwriting 'msg_type' and 'length' with 'body' data.
12807
12808   *Bodo Moeller*
12809
12810 * Add an implementation of SSL_add_dir_cert_subjects_to_stack for Win32.
12811
12812   *Massimo Santin via Richard Levitte*
12813
12814 * Major restructuring to the underlying ENGINE code. This includes
12815   reduction of linker bloat, separation of pure "ENGINE" manipulation
12816   (initialisation, etc) from functionality dealing with implementations
12817   of specific crypto interfaces. This change also introduces integrated
12818   support for symmetric ciphers and digest implementations - so ENGINEs
12819   can now accelerate these by providing EVP_CIPHER and EVP_MD
12820   implementations of their own. This is detailed in
12821   [crypto/engine/README.md](crypto/engine/README.md)
12822   as it couldn't be adequately described here. However, there are a few
12823   API changes worth noting - some RSA, DSA, DH, and RAND functions that
12824   were changed in the original introduction of ENGINE code have now
12825   reverted back - the hooking from this code to ENGINE is now a good
12826   deal more passive and at run-time, operations deal directly with
12827   RSA_METHODs, DSA_METHODs (etc) as they did before, rather than
12828   dereferencing through an ENGINE pointer any more. Also, the ENGINE
12829   functions dealing with `BN_MOD_EXP[_CRT]` handlers have been removed -
12830   they were not being used by the framework as there is no concept of a
12831   BIGNUM_METHOD and they could not be generalised to the new
12832   'ENGINE_TABLE' mechanism that underlies the new code. Similarly,
12833   ENGINE_cpy() has been removed as it cannot be consistently defined in
12834   the new code.
12835
12836   *Geoff Thorpe*
12837
12838 * Change ASN1_GENERALIZEDTIME_check() to allow fractional seconds.
12839
12840   *Steve Henson*
12841
12842 * Change mkdef.pl to sort symbols that get the same entry number,
12843   and make sure the automatically generated functions `ERR_load_*`
12844   become part of libeay.num as well.
12845
12846   *Richard Levitte*
12847
12848 * New function SSL_renegotiate_pending().  This returns true once
12849   renegotiation has been requested (either SSL_renegotiate() call
12850   or HelloRequest/ClientHello received from the peer) and becomes
12851   false once a handshake has been completed.
12852   (For servers, SSL_renegotiate() followed by SSL_do_handshake()
12853   sends a HelloRequest, but does not ensure that a handshake takes
12854   place.  SSL_renegotiate_pending() is useful for checking if the
12855   client has followed the request.)
12856
12857   *Bodo Moeller*
12858
12859 * New SSL option SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION.
12860   By default, clients may request session resumption even during
12861   renegotiation (if session ID contexts permit); with this option,
12862   session resumption is possible only in the first handshake.
12863
12864   SSL_OP_ALL is now 0x00000FFFL instead of 0x000FFFFFL.  This makes
12865   more bits available for options that should not be part of
12866   SSL_OP_ALL (such as SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION).
12867
12868   *Bodo Moeller*
12869
12870 * Add some demos for certificate and certificate request creation.
12871
12872   *Steve Henson*
12873
12874 * Make maximum certificate chain size accepted from the peer application
12875   settable (`SSL*_get/set_max_cert_list()`), as proposed by
12876   "Douglas E. Engert" <deengert@anl.gov>.
12877
12878   *Lutz Jaenicke*
12879
12880 * Add support for shared libraries for Unixware-7
12881   (Boyd Lynn Gerber <gerberb@zenez.com>).
12882
12883   *Lutz Jaenicke*
12884
12885 * Add a "destroy" handler to ENGINEs that allows structural cleanup to
12886   be done prior to destruction. Use this to unload error strings from
12887   ENGINEs that load their own error strings. NB: This adds two new API
12888   functions to "get" and "set" this destroy handler in an ENGINE.
12889
12890   *Geoff Thorpe*
12891
12892 * Alter all existing ENGINE implementations (except "openssl" and
12893   "openbsd") to dynamically instantiate their own error strings. This
12894   makes them more flexible to be built both as statically-linked ENGINEs
12895   and self-contained shared-libraries loadable via the "dynamic" ENGINE.
12896   Also, add stub code to each that makes building them as self-contained
12897   shared-libraries easier (see [README-Engine.md](README-Engine.md)).
12898
12899   *Geoff Thorpe*
12900
12901 * Add a "dynamic" ENGINE that provides a mechanism for binding ENGINE
12902   implementations into applications that are completely implemented in
12903   self-contained shared-libraries. The "dynamic" ENGINE exposes control
12904   commands that can be used to configure what shared-library to load and
12905   to control aspects of the way it is handled. Also, made an update to
12906   the [README-Engine.md](README-Engine.md) file
12907   that brings its information up-to-date and
12908   provides some information and instructions on the "dynamic" ENGINE
12909   (ie. how to use it, how to build "dynamic"-loadable ENGINEs, etc).
12910
12911   *Geoff Thorpe*
12912
12913 * Make it possible to unload ranges of ERR strings with a new
12914   "ERR_unload_strings" function.
12915
12916   *Geoff Thorpe*
12917
12918 * Add a copy() function to EVP_MD.
12919
12920   *Ben Laurie*
12921
12922 * Make EVP_MD routines take a context pointer instead of just the
12923   md_data void pointer.
12924
12925   *Ben Laurie*
12926
12927 * Add flags to EVP_MD and EVP_MD_CTX. EVP_MD_FLAG_ONESHOT indicates
12928   that the digest can only process a single chunk of data
12929   (typically because it is provided by a piece of
12930   hardware). EVP_MD_CTX_FLAG_ONESHOT indicates that the application
12931   is only going to provide a single chunk of data, and hence the
12932   framework needn't accumulate the data for oneshot drivers.
12933
12934   *Ben Laurie*
12935
12936 * As with "ERR", make it possible to replace the underlying "ex_data"
12937   functions. This change also alters the storage and management of global
12938   ex_data state - it's now all inside ex_data.c and all "class" code (eg.
12939   RSA, BIO, SSL_CTX, etc) no longer stores its own STACKS and per-class
12940   index counters. The API functions that use this state have been changed
12941   to take a "class_index" rather than pointers to the class's local STACK
12942   and counter, and there is now an API function to dynamically create new
12943   classes. This centralisation allows us to (a) plug a lot of the
12944   thread-safety problems that existed, and (b) makes it possible to clean
12945   up all allocated state using "CRYPTO_cleanup_all_ex_data()". W.r.t. (b)
12946   such data would previously have always leaked in application code and
12947   workarounds were in place to make the memory debugging turn a blind eye
12948   to it. Application code that doesn't use this new function will still
12949   leak as before, but their memory debugging output will announce it now
12950   rather than letting it slide.
12951
12952   Besides the addition of CRYPTO_cleanup_all_ex_data(), another API change
12953   induced by the "ex_data" overhaul is that X509_STORE_CTX_init() now
12954   has a return value to indicate success or failure.
12955
12956   *Geoff Thorpe*
12957
12958 * Make it possible to replace the underlying "ERR" functions such that the
12959   global state (2 LHASH tables and 2 locks) is only used by the "default"
12960   implementation. This change also adds two functions to "get" and "set"
12961   the implementation prior to it being automatically set the first time
12962   any other ERR function takes place. Ie. an application can call "get",
12963   pass the return value to a module it has just loaded, and that module
12964   can call its own "set" function using that value. This means the
12965   module's "ERR" operations will use (and modify) the error state in the
12966   application and not in its own statically linked copy of OpenSSL code.
12967
12968   *Geoff Thorpe*
12969
12970 * Give DH, DSA, and RSA types their own `*_up_ref()` function to increment
12971   reference counts. This performs normal REF_PRINT/REF_CHECK macros on
12972   the operation, and provides a more encapsulated way for external code
12973   (crypto/evp/ and ssl/) to do this. Also changed the evp and ssl code
12974   to use these functions rather than manually incrementing the counts.
12975
12976   Also rename "DSO_up()" function to more descriptive "DSO_up_ref()".
12977
12978   *Geoff Thorpe*
12979
12980 * Add EVP test program.
12981
12982   *Ben Laurie*
12983
12984 * Add symmetric cipher support to ENGINE. Expect the API to change!
12985
12986   *Ben Laurie*
12987
12988 * New CRL functions: X509_CRL_set_version(), X509_CRL_set_issuer_name()
12989   X509_CRL_set_lastUpdate(), X509_CRL_set_nextUpdate(), X509_CRL_sort(),
12990   X509_REVOKED_set_serialNumber(), and X509_REVOKED_set_revocationDate().
12991   These allow a CRL to be built without having to access X509_CRL fields
12992   directly. Modify 'ca' application to use new functions.
12993
12994   *Steve Henson*
12995
12996 * Move SSL_OP_TLS_ROLLBACK_BUG out of the SSL_OP_ALL list of recommended
12997   bug workarounds. Rollback attack detection is a security feature.
12998   The problem will only arise on OpenSSL servers when TLSv1 is not
12999   available (sslv3_server_method() or SSL_OP_NO_TLSv1).
13000   Software authors not wanting to support TLSv1 will have special reasons
13001   for their choice and can explicitly enable this option.
13002
13003   *Bodo Moeller, Lutz Jaenicke*
13004
13005 * Rationalise EVP so it can be extended: don't include a union of
13006   cipher/digest structures, add init/cleanup functions for EVP_MD_CTX
13007   (similar to those existing for EVP_CIPHER_CTX).
13008   Usage example:
13009
13010           EVP_MD_CTX md;
13011
13012           EVP_MD_CTX_init(&md);             /* new function call */
13013           EVP_DigestInit(&md, EVP_sha1());
13014           EVP_DigestUpdate(&md, in, len);
13015           EVP_DigestFinal(&md, out, NULL);
13016           EVP_MD_CTX_cleanup(&md);          /* new function call */
13017
13018   *Ben Laurie*
13019
13020 * Make DES key schedule conform to the usual scheme, as well as
13021   correcting its structure. This means that calls to DES functions
13022   now have to pass a pointer to a des_key_schedule instead of a
13023   plain des_key_schedule (which was actually always a pointer
13024   anyway): E.g.,
13025
13026           des_key_schedule ks;
13027
13028           des_set_key_checked(..., &ks);
13029           des_ncbc_encrypt(..., &ks, ...);
13030
13031   (Note that a later change renames 'des_...' into 'DES_...'.)
13032
13033   *Ben Laurie*
13034
13035 * Initial reduction of linker bloat: the use of some functions, such as
13036   PEM causes large amounts of unused functions to be linked in due to
13037   poor organisation. For example pem_all.c contains every PEM function
13038   which has a knock on effect of linking in large amounts of (unused)
13039   ASN1 code. Grouping together similar functions and splitting unrelated
13040   functions prevents this.
13041
13042   *Steve Henson*
13043
13044 * Cleanup of EVP macros.
13045
13046   *Ben Laurie*
13047
13048 * Change historical references to `{NID,SN,LN}_des_ede` and ede3 to add the
13049   correct `_ecb suffix`.
13050
13051   *Ben Laurie*
13052
13053 * Add initial OCSP responder support to ocsp application. The
13054   revocation information is handled using the text based index
13055   use by the ca application. The responder can either handle
13056   requests generated internally, supplied in files (for example
13057   via a CGI script) or using an internal minimal server.
13058
13059   *Steve Henson*
13060
13061 * Add configuration choices to get zlib compression for TLS.
13062
13063   *Richard Levitte*
13064
13065 * Changes to Kerberos SSL for RFC 2712 compliance:
13066   1. Implemented real KerberosWrapper, instead of just using
13067      KRB5 AP_REQ message.  [Thanks to Simon Wilkinson <sxw@sxw.org.uk>]
13068   2. Implemented optional authenticator field of KerberosWrapper.
13069
13070   Added openssl-style ASN.1 macros for Kerberos ticket, ap_req,
13071   and authenticator structs; see crypto/krb5/.
13072
13073   Generalized Kerberos calls to support multiple Kerberos libraries.
13074   *Vern Staats <staatsvr@asc.hpc.mil>, Jeffrey Altman <jaltman@columbia.edu>
13075   via Richard Levitte*
13076
13077 * Cause 'openssl speed' to use fully hard-coded DSA keys as it
13078   already does with RSA. testdsa.h now has 'priv_key/pub_key'
13079   values for each of the key sizes rather than having just
13080   parameters (and 'speed' generating keys each time).
13081
13082   *Geoff Thorpe*
13083
13084 * Speed up EVP routines.
13085   Before:
13086crypt
13087pe              8 bytes     64 bytes    256 bytes   1024 bytes   8192 bytes
13088s-cbc           4408.85k     5560.51k     5778.46k     5862.20k     5825.16k
13089s-cbc           4389.55k     5571.17k     5792.23k     5846.91k     5832.11k
13090s-cbc           4394.32k     5575.92k     5807.44k     5848.37k     5841.30k
13091crypt
13092s-cbc           3482.66k     5069.49k     5496.39k     5614.16k     5639.28k
13093s-cbc           3480.74k     5068.76k     5510.34k     5609.87k     5635.52k
13094s-cbc           3483.72k     5067.62k     5504.60k     5708.01k     5724.80k
13095   After:
13096crypt
13097s-cbc           4660.16k     5650.19k     5807.19k     5827.13k     5783.32k
13098crypt
13099s-cbc           3624.96k     5258.21k     5530.91k     5624.30k     5628.26k
13100
13101   *Ben Laurie*
13102
13103 * Added the OS2-EMX target.
13104
13105   *"Brian Havard" <brianh@kheldar.apana.org.au> and Richard Levitte*
13106
13107 * Rewrite commands to use `NCONF` routines instead of the old `CONF`.
13108   New functions to support `NCONF` routines in extension code.
13109   New function `CONF_set_nconf()`
13110   to allow functions which take an `NCONF` to also handle the old `LHASH`
13111   structure: this means that the old `CONF` compatible routines can be
13112   retained (in particular w.rt. extensions) without having to duplicate the
13113   code. New function `X509V3_add_ext_nconf_sk()` to add extensions to a stack.
13114
13115   *Steve Henson*
13116
13117 * Enhance the general user interface with mechanisms for inner control
13118   and with possibilities to have yes/no kind of prompts.
13119
13120   *Richard Levitte*
13121
13122 * Change all calls to low-level digest routines in the library and
13123   applications to use EVP. Add missing calls to HMAC_cleanup() and
13124   don't assume HMAC_CTX can be copied using memcpy().
13125
13126   *Verdon Walker <VWalker@novell.com>, Steve Henson*
13127
13128 * Add the possibility to control engines through control names but with
13129   arbitrary arguments instead of just a string.
13130   Change the key loaders to take a UI_METHOD instead of a callback
13131   function pointer.  NOTE: this breaks binary compatibility with earlier
13132   versions of OpenSSL [engine].
13133   Adapt the nCipher code for these new conditions and add a card insertion
13134   callback.
13135
13136   *Richard Levitte*
13137
13138 * Enhance the general user interface with mechanisms to better support
13139   dialog box interfaces, application-defined prompts, the possibility
13140   to use defaults (for example default passwords from somewhere else)
13141   and interrupts/cancellations.
13142
13143   *Richard Levitte*
13144
13145 * Tidy up PKCS#12 attribute handling. Add support for the CSP name
13146   attribute in PKCS#12 files, add new -CSP option to pkcs12 utility.
13147
13148   *Steve Henson*
13149
13150 * Fix a memory leak in 'sk_dup()' in the case reallocation fails. (Also
13151   tidy up some unnecessarily weird code in 'sk_new()').
13152
13153   *Geoff, reported by Diego Tartara <dtartara@novamens.com>*
13154
13155 * Change the key loading routines for ENGINEs to use the same kind
13156   callback (pem_password_cb) as all other routines that need this
13157   kind of callback.
13158
13159   *Richard Levitte*
13160
13161 * Increase ENTROPY_NEEDED to 32 bytes, as Rijndael can operate with
13162   256 bit (=32 byte) keys. Of course seeding with more entropy bytes
13163   than this minimum value is recommended.
13164
13165   *Lutz Jaenicke*
13166
13167 * New random seeder for OpenVMS, using the system process statistics
13168   that are easily reachable.
13169
13170   *Richard Levitte*
13171
13172 * Windows apparently can't transparently handle global
13173   variables defined in DLLs. Initialisations such as:
13174
13175           const ASN1_ITEM *it = &ASN1_INTEGER_it;
13176
13177   won't compile. This is used by the any applications that need to
13178   declare their own ASN1 modules. This was fixed by adding the option
13179   EXPORT_VAR_AS_FN to all Win32 platforms, although this isn't strictly
13180   needed for static libraries under Win32.
13181
13182   *Steve Henson*
13183
13184 * New functions X509_PURPOSE_set() and X509_TRUST_set() to handle
13185   setting of purpose and trust fields. New X509_STORE trust and
13186   purpose functions and tidy up setting in other SSL functions.
13187
13188   *Steve Henson*
13189
13190 * Add copies of X509_STORE_CTX fields and callbacks to X509_STORE
13191   structure. These are inherited by X509_STORE_CTX when it is
13192   initialised. This allows various defaults to be set in the
13193   X509_STORE structure (such as flags for CRL checking and custom
13194   purpose or trust settings) for functions which only use X509_STORE_CTX
13195   internally such as S/MIME.
13196
13197   Modify X509_STORE_CTX_purpose_inherit() so it only sets purposes and
13198   trust settings if they are not set in X509_STORE. This allows X509_STORE
13199   purposes and trust (in S/MIME for example) to override any set by default.
13200
13201   Add command line options for CRL checking to smime, s_client and s_server
13202   applications.
13203
13204   *Steve Henson*
13205
13206 * Initial CRL based revocation checking. If the CRL checking flag(s)
13207   are set then the CRL is looked up in the X509_STORE structure and
13208   its validity and signature checked, then if the certificate is found
13209   in the CRL the verify fails with a revoked error.
13210
13211   Various new CRL related callbacks added to X509_STORE_CTX structure.
13212
13213   Command line options added to 'verify' application to support this.
13214
13215   This needs some additional work, such as being able to handle multiple
13216   CRLs with different times, extension based lookup (rather than just
13217   by subject name) and ultimately more complete V2 CRL extension
13218   handling.
13219
13220   *Steve Henson*
13221
13222 * Add a general user interface API (crypto/ui/).  This is designed
13223   to replace things like des_read_password and friends (backward
13224   compatibility functions using this new API are provided).
13225   The purpose is to remove prompting functions from the DES code
13226   section as well as provide for prompting through dialog boxes in
13227   a window system and the like.
13228
13229   *Richard Levitte*
13230
13231 * Add "ex_data" support to ENGINE so implementations can add state at a
13232   per-structure level rather than having to store it globally.
13233
13234   *Geoff*
13235
13236 * Make it possible for ENGINE structures to be copied when retrieved by
13237   ENGINE_by_id() if the ENGINE specifies a new flag: ENGINE_FLAGS_BY_ID_COPY.
13238   This causes the "original" ENGINE structure to act like a template,
13239   analogous to the RSA vs. RSA_METHOD type of separation. Because of this
13240   operational state can be localised to each ENGINE structure, despite the
13241   fact they all share the same "methods". New ENGINE structures returned in
13242   this case have no functional references and the return value is the single
13243   structural reference. This matches the single structural reference returned
13244   by ENGINE_by_id() normally, when it is incremented on the pre-existing
13245   ENGINE structure.
13246
13247   *Geoff*
13248
13249 * Fix ASN1 decoder when decoding type ANY and V_ASN1_OTHER: since this
13250   needs to match any other type at all we need to manually clear the
13251   tag cache.
13252
13253   *Steve Henson*
13254
13255 * Changes to the "openssl engine" utility to include;
13256   - verbosity levels ('-v', '-vv', and '-vvv') that provide information
13257     about an ENGINE's available control commands.
13258   - executing control commands from command line arguments using the
13259     '-pre' and '-post' switches. '-post' is only used if '-t' is
13260     specified and the ENGINE is successfully initialised. The syntax for
13261     the individual commands are colon-separated, for example;
13262           openssl engine chil -pre FORK_CHECK:0 -pre SO_PATH:/lib/test.so
13263
13264   *Geoff*
13265
13266 * New dynamic control command support for ENGINEs. ENGINEs can now
13267   declare their own commands (numbers), names (strings), descriptions,
13268   and input types for run-time discovery by calling applications. A
13269   subset of these commands are implicitly classed as "executable"
13270   depending on their input type, and only these can be invoked through
13271   the new string-based API function ENGINE_ctrl_cmd_string(). (Eg. this
13272   can be based on user input, config files, etc). The distinction is
13273   that "executable" commands cannot return anything other than a boolean
13274   result and can only support numeric or string input, whereas some
13275   discoverable commands may only be for direct use through
13276   ENGINE_ctrl(), eg. supporting the exchange of binary data, function
13277   pointers, or other custom uses. The "executable" commands are to
13278   support parameterisations of ENGINE behaviour that can be
13279   unambiguously defined by ENGINEs and used consistently across any
13280   OpenSSL-based application. Commands have been added to all the
13281   existing hardware-supporting ENGINEs, noticeably "SO_PATH" to allow
13282   control over shared-library paths without source code alterations.
13283
13284   *Geoff*
13285
13286 * Changed all ENGINE implementations to dynamically allocate their
13287   ENGINEs rather than declaring them statically. Apart from this being
13288   necessary with the removal of the ENGINE_FLAGS_MALLOCED distinction,
13289   this also allows the implementations to compile without using the
13290   internal engine_int.h header.
13291
13292   *Geoff*
13293
13294 * Minor adjustment to "rand" code. RAND_get_rand_method() now returns a
13295   'const' value. Any code that should be able to modify a RAND_METHOD
13296   should already have non-const pointers to it (ie. they should only
13297   modify their own ones).
13298
13299   *Geoff*
13300
13301 * Made a variety of little tweaks to the ENGINE code.
13302   - "atalla" and "ubsec" string definitions were moved from header files
13303     to C code. "nuron" string definitions were placed in variables
13304     rather than hard-coded - allowing parameterisation of these values
13305     later on via ctrl() commands.
13306   - Removed unused "#if 0"'d code.
13307   - Fixed engine list iteration code so it uses ENGINE_free() to release
13308     structural references.
13309   - Constified the RAND_METHOD element of ENGINE structures.
13310   - Constified various get/set functions as appropriate and added
13311     missing functions (including a catch-all ENGINE_cpy that duplicates
13312     all ENGINE values onto a new ENGINE except reference counts/state).
13313   - Removed NULL parameter checks in get/set functions. Setting a method
13314     or function to NULL is a way of cancelling out a previously set
13315     value.  Passing a NULL ENGINE parameter is just plain stupid anyway
13316     and doesn't justify the extra error symbols and code.
13317   - Deprecate the ENGINE_FLAGS_MALLOCED define and move the area for
13318     flags from engine_int.h to engine.h.
13319   - Changed prototypes for ENGINE handler functions (init(), finish(),
13320     ctrl(), key-load functions, etc) to take an (ENGINE*) parameter.
13321
13322   *Geoff*
13323
13324 * Implement binary inversion algorithm for BN_mod_inverse in addition
13325   to the algorithm using long division.  The binary algorithm can be
13326   used only if the modulus is odd.  On 32-bit systems, it is faster
13327   only for relatively small moduli (roughly 20-30% for 128-bit moduli,
13328   roughly 5-15% for 256-bit moduli), so we use it only for moduli
13329   up to 450 bits.  In 64-bit environments, the binary algorithm
13330   appears to be advantageous for much longer moduli; here we use it
13331   for moduli up to 2048 bits.
13332
13333   *Bodo Moeller*
13334
13335 * Rewrite CHOICE field setting in ASN1_item_ex_d2i(). The old code
13336   could not support the combine flag in choice fields.
13337
13338   *Steve Henson*
13339
13340 * Add a 'copy_extensions' option to the 'ca' utility. This copies
13341   extensions from a certificate request to the certificate.
13342
13343   *Steve Henson*
13344
13345 * Allow multiple 'certopt' and 'nameopt' options to be separated
13346   by commas. Add 'namopt' and 'certopt' options to the 'ca' config
13347   file: this allows the display of the certificate about to be
13348   signed to be customised, to allow certain fields to be included
13349   or excluded and extension details. The old system didn't display
13350   multicharacter strings properly, omitted fields not in the policy
13351   and couldn't display additional details such as extensions.
13352
13353   *Steve Henson*
13354
13355 * Function EC_POINTs_mul for multiple scalar multiplication
13356   of an arbitrary number of elliptic curve points
13357           \sum scalars[i]*points[i],
13358   optionally including the generator defined for the EC_GROUP:
13359           scalar*generator +  \sum scalars[i]*points[i].
13360
13361   EC_POINT_mul is a simple wrapper function for the typical case
13362   that the point list has just one item (besides the optional
13363   generator).
13364
13365   *Bodo Moeller*
13366
13367 * First EC_METHODs for curves over GF(p):
13368
13369   EC_GFp_simple_method() uses the basic BN_mod_mul and BN_mod_sqr
13370   operations and provides various method functions that can also
13371   operate with faster implementations of modular arithmetic.
13372
13373   EC_GFp_mont_method() reuses most functions that are part of
13374   EC_GFp_simple_method, but uses Montgomery arithmetic.
13375
13376   *Bodo Moeller; point addition and point doubling
13377   implementation directly derived from source code provided by
13378   Lenka Fibikova <fibikova@exp-math.uni-essen.de>*
13379
13380 * Framework for elliptic curves (crypto/ec/ec.h, crypto/ec/ec_lcl.h,
13381   crypto/ec/ec_lib.c):
13382
13383   Curves are EC_GROUP objects (with an optional group generator)
13384   based on EC_METHODs that are built into the library.
13385
13386   Points are EC_POINT objects based on EC_GROUP objects.
13387
13388   Most of the framework would be able to handle curves over arbitrary
13389   finite fields, but as there are no obvious types for fields other
13390   than GF(p), some functions are limited to that for now.
13391
13392   *Bodo Moeller*
13393
13394 * Add the -HTTP option to s_server.  It is similar to -WWW, but requires
13395   that the file contains a complete HTTP response.
13396
13397   *Richard Levitte*
13398
13399 * Add the ec directory to mkdef.pl and mkfiles.pl. In mkdef.pl
13400   change the def and num file printf format specifier from "%-40sXXX"
13401   to "%-39s XXX". The latter will always guarantee a space after the
13402   field while the former will cause them to run together if the field
13403   is 40 of more characters long.
13404
13405   *Steve Henson*
13406
13407 * Constify the cipher and digest 'method' functions and structures
13408   and modify related functions to take constant EVP_MD and EVP_CIPHER
13409   pointers.
13410
13411   *Steve Henson*
13412
13413 * Hide BN_CTX structure details in bn_lcl.h instead of publishing them
13414   in <openssl/bn.h>.  Also further increase BN_CTX_NUM to 32.
13415
13416   *Bodo Moeller*
13417
13418 * Modify `EVP_Digest*()` routines so they now return values. Although the
13419   internal software routines can never fail additional hardware versions
13420   might.
13421
13422   *Steve Henson*
13423
13424 * Clean up crypto/err/err.h and change some error codes to avoid conflicts:
13425
13426   Previously ERR_R_FATAL was too small and coincided with ERR_LIB_PKCS7
13427   (= ERR_R_PKCS7_LIB); it is now 64 instead of 32.
13428
13429   ASN1 error codes
13430           ERR_R_NESTED_ASN1_ERROR
13431           ...
13432           ERR_R_MISSING_ASN1_EOS
13433   were 4 .. 9, conflicting with
13434           ERR_LIB_RSA (= ERR_R_RSA_LIB)
13435           ...
13436           ERR_LIB_PEM (= ERR_R_PEM_LIB).
13437   They are now 58 .. 63 (i.e., just below ERR_R_FATAL).
13438
13439   Add new error code 'ERR_R_INTERNAL_ERROR'.
13440
13441   *Bodo Moeller*
13442
13443 * Don't overuse locks in crypto/err/err.c: For data retrieval, CRYPTO_r_lock
13444   suffices.
13445
13446   *Bodo Moeller*
13447
13448 * New option '-subj arg' for 'openssl req' and 'openssl ca'.  This
13449   sets the subject name for a new request or supersedes the
13450   subject name in a given request. Formats that can be parsed are
13451           'CN=Some Name, OU=myOU, C=IT'
13452   and
13453           'CN=Some Name/OU=myOU/C=IT'.
13454
13455   Add options '-batch' and '-verbose' to 'openssl req'.
13456
13457   *Massimiliano Pala <madwolf@hackmasters.net>*
13458
13459 * Introduce the possibility to access global variables through
13460   functions on platform were that's the best way to handle exporting
13461   global variables in shared libraries.  To enable this functionality,
13462   one must configure with "EXPORT_VAR_AS_FN" or defined the C macro
13463   "OPENSSL_EXPORT_VAR_AS_FUNCTION" in crypto/opensslconf.h (the latter
13464   is normally done by Configure or something similar).
13465
13466   To implement a global variable, use the macro OPENSSL_IMPLEMENT_GLOBAL
13467   in the source file (foo.c) like this:
13468
13469           OPENSSL_IMPLEMENT_GLOBAL(int,foo)=1;
13470           OPENSSL_IMPLEMENT_GLOBAL(double,bar);
13471
13472   To declare a global variable, use the macros OPENSSL_DECLARE_GLOBAL
13473   and OPENSSL_GLOBAL_REF in the header file (foo.h) like this:
13474
13475           OPENSSL_DECLARE_GLOBAL(int,foo);
13476           #define foo OPENSSL_GLOBAL_REF(foo)
13477           OPENSSL_DECLARE_GLOBAL(double,bar);
13478           #define bar OPENSSL_GLOBAL_REF(bar)
13479
13480   The #defines are very important, and therefore so is including the
13481   header file everywhere where the defined globals are used.
13482
13483   The macro OPENSSL_EXPORT_VAR_AS_FUNCTION also affects the definition
13484   of ASN.1 items, but that structure is a bit different.
13485
13486   The largest change is in util/mkdef.pl which has been enhanced with
13487   better and easier to understand logic to choose which symbols should
13488   go into the Windows .def files as well as a number of fixes and code
13489   cleanup (among others, algorithm keywords are now sorted
13490   lexicographically to avoid constant rewrites).
13491
13492   *Richard Levitte*
13493
13494 * In BN_div() keep a copy of the sign of 'num' before writing the
13495   result to 'rm' because if rm==num the value will be overwritten
13496   and produce the wrong result if 'num' is negative: this caused
13497   problems with BN_mod() and BN_nnmod().
13498
13499   *Steve Henson*
13500
13501 * Function OCSP_request_verify(). This checks the signature on an
13502   OCSP request and verifies the signer certificate. The signer
13503   certificate is just checked for a generic purpose and OCSP request
13504   trust settings.
13505
13506   *Steve Henson*
13507
13508 * Add OCSP_check_validity() function to check the validity of OCSP
13509   responses. OCSP responses are prepared in real time and may only
13510   be a few seconds old. Simply checking that the current time lies
13511   between thisUpdate and nextUpdate max reject otherwise valid responses
13512   caused by either OCSP responder or client clock inaccuracy. Instead
13513   we allow thisUpdate and nextUpdate to fall within a certain period of
13514   the current time. The age of the response can also optionally be
13515   checked. Two new options -validity_period and -status_age added to
13516   ocsp utility.
13517
13518   *Steve Henson*
13519
13520 * If signature or public key algorithm is unrecognized print out its
13521   OID rather that just UNKNOWN.
13522
13523   *Steve Henson*
13524
13525 * Change OCSP_cert_to_id() to tolerate a NULL subject certificate and
13526   OCSP_cert_id_new() a NULL serialNumber. This allows a partial certificate
13527   ID to be generated from the issuer certificate alone which can then be
13528   passed to OCSP_id_issuer_cmp().
13529
13530   *Steve Henson*
13531
13532 * New compilation option ASN1_ITEM_FUNCTIONS. This causes the new
13533   ASN1 modules to export functions returning ASN1_ITEM pointers
13534   instead of the ASN1_ITEM structures themselves. This adds several
13535   new macros which allow the underlying ASN1 function/structure to
13536   be accessed transparently. As a result code should not use ASN1_ITEM
13537   references directly (such as &X509_it) but instead use the relevant
13538   macros (such as ASN1_ITEM_rptr(X509)). This option is to allow
13539   use of the new ASN1 code on platforms where exporting structures
13540   is problematical (for example in shared libraries) but exporting
13541   functions returning pointers to structures is not.
13542
13543   *Steve Henson*
13544
13545 * Add support for overriding the generation of SSL/TLS session IDs.
13546   These callbacks can be registered either in an SSL_CTX or per SSL.
13547   The purpose of this is to allow applications to control, if they wish,
13548   the arbitrary values chosen for use as session IDs, particularly as it
13549   can be useful for session caching in multiple-server environments. A
13550   command-line switch for testing this (and any client code that wishes
13551   to use such a feature) has been added to "s_server".
13552
13553   *Geoff Thorpe, Lutz Jaenicke*
13554
13555 * Modify mkdef.pl to recognise and parse preprocessor conditionals
13556   of the form `#if defined(...) || defined(...) || ...` and
13557   `#if !defined(...) && !defined(...) && ...`.  This also avoids
13558   the growing number of special cases it was previously handling.
13559
13560   *Richard Levitte*
13561
13562 * Make all configuration macros available for application by making
13563   sure they are available in opensslconf.h, by giving them names starting
13564   with `OPENSSL_` to avoid conflicts with other packages and by making
13565   sure e_os2.h will cover all platform-specific cases together with
13566   opensslconf.h.
13567   Additionally, it is now possible to define configuration/platform-
13568   specific names (called "system identities").  In the C code, these
13569   are prefixed with `OPENSSL_SYSNAME_`.  e_os2.h will create another
13570   macro with the name beginning with `OPENSSL_SYS_`, which is determined
13571   from `OPENSSL_SYSNAME_*` or compiler-specific macros depending on
13572   what is available.
13573
13574   *Richard Levitte*
13575
13576 * New option -set_serial to 'req' and 'x509' this allows the serial
13577   number to use to be specified on the command line. Previously self
13578   signed certificates were hard coded with serial number 0 and the
13579   CA options of 'x509' had to use a serial number in a file which was
13580   auto incremented.
13581
13582   *Steve Henson*
13583
13584 * New options to 'ca' utility to support V2 CRL entry extensions.
13585   Currently CRL reason, invalidity date and hold instruction are
13586   supported. Add new CRL extensions to V3 code and some new objects.
13587
13588   *Steve Henson*
13589
13590 * New function EVP_CIPHER_CTX_set_padding() this is used to
13591   disable standard block padding (aka PKCS#5 padding) in the EVP
13592   API, which was previously mandatory. This means that the data is
13593   not padded in any way and so the total length much be a multiple
13594   of the block size, otherwise an error occurs.
13595
13596   *Steve Henson*
13597
13598 * Initial (incomplete) OCSP SSL support.
13599
13600   *Steve Henson*
13601
13602 * New function OCSP_parse_url(). This splits up a URL into its host,
13603   port and path components: primarily to parse OCSP URLs. New -url
13604   option to ocsp utility.
13605
13606   *Steve Henson*
13607
13608 * New nonce behavior. The return value of OCSP_check_nonce() now
13609   reflects the various checks performed. Applications can decide
13610   whether to tolerate certain situations such as an absent nonce
13611   in a response when one was present in a request: the ocsp application
13612   just prints out a warning. New function OCSP_add1_basic_nonce()
13613   this is to allow responders to include a nonce in a response even if
13614   the request is nonce-less.
13615
13616   *Steve Henson*
13617
13618 * Disable stdin buffering in `load_cert()` (`apps/apps.c`) so that no certs are
13619   skipped when using openssl x509 multiple times on a single input file,
13620   e.g. `(openssl x509 -out cert1; openssl x509 -out cert2) <certs`.
13621
13622   *Bodo Moeller*
13623
13624 * Make ASN1_UTCTIME_set_string() and ASN1_GENERALIZEDTIME_set_string()
13625   set string type: to handle setting ASN1_TIME structures. Fix ca
13626   utility to correctly initialize revocation date of CRLs.
13627
13628   *Steve Henson*
13629
13630 * New option SSL_OP_CIPHER_SERVER_PREFERENCE allows the server to override
13631   the clients preferred ciphersuites and rather use its own preferences.
13632   Should help to work around M$ SGC (Server Gated Cryptography) bug in
13633   Internet Explorer by ensuring unchanged hash method during stepup.
13634   (Also replaces the broken/deactivated SSL_OP_NON_EXPORT_FIRST option.)
13635
13636   *Lutz Jaenicke*
13637
13638 * Make mkdef.pl recognise all DECLARE_ASN1 macros, change rijndael
13639   to aes and add a new 'exist' option to print out symbols that don't
13640   appear to exist.
13641
13642   *Steve Henson*
13643
13644 * Additional options to ocsp utility to allow flags to be set and
13645   additional certificates supplied.
13646
13647   *Steve Henson*
13648
13649 * Add the option -VAfile to 'openssl ocsp', so the user can give the
13650   OCSP client a number of certificate to only verify the response
13651   signature against.
13652
13653   *Richard Levitte*
13654
13655 * Update Rijndael code to version 3.0 and change EVP AES ciphers to
13656   handle the new API. Currently only ECB, CBC modes supported. Add new
13657   AES OIDs.
13658
13659   Add TLS AES ciphersuites as described in RFC3268, "Advanced
13660   Encryption Standard (AES) Ciphersuites for Transport Layer
13661   Security (TLS)".  (In beta versions of OpenSSL 0.9.7, these were
13662   not enabled by default and were not part of the "ALL" ciphersuite
13663   alias because they were not yet official; they could be
13664   explicitly requested by specifying the "AESdraft" ciphersuite
13665   group alias.  In the final release of OpenSSL 0.9.7, the group
13666   alias is called "AES" and is part of "ALL".)
13667
13668   *Ben Laurie, Steve  Henson, Bodo Moeller*
13669
13670 * New function OCSP_copy_nonce() to copy nonce value (if present) from
13671   request to response.
13672
13673   *Steve Henson*
13674
13675 * Functions for OCSP responders. OCSP_request_onereq_count(),
13676   OCSP_request_onereq_get0(), OCSP_onereq_get0_id() and OCSP_id_get0_info()
13677   extract information from a certificate request. OCSP_response_create()
13678   creates a response and optionally adds a basic response structure.
13679   OCSP_basic_add1_status() adds a complete single response to a basic
13680   response and returns the OCSP_SINGLERESP structure just added (to allow
13681   extensions to be included for example). OCSP_basic_add1_cert() adds a
13682   certificate to a basic response and OCSP_basic_sign() signs a basic
13683   response with various flags. New helper functions ASN1_TIME_check()
13684   (checks validity of ASN1_TIME structure) and ASN1_TIME_to_generalizedtime()
13685   (converts ASN1_TIME to GeneralizedTime).
13686
13687   *Steve Henson*
13688
13689 * Various new functions. EVP_Digest() combines EVP_Digest{Init,Update,Final}()
13690   in a single operation. X509_get0_pubkey_bitstr() extracts the public_key
13691   structure from a certificate. X509_pubkey_digest() digests the public_key
13692   contents: this is used in various key identifiers.
13693
13694   *Steve Henson*
13695
13696 * Make sk_sort() tolerate a NULL argument.
13697
13698   *Steve Henson reported by Massimiliano Pala <madwolf@comune.modena.it>*
13699
13700 * New OCSP verify flag OCSP_TRUSTOTHER. When set the "other" certificates
13701   passed by the function are trusted implicitly. If any of them signed the
13702   response then it is assumed to be valid and is not verified.
13703
13704   *Steve Henson*
13705
13706 * In PKCS7_set_type() initialise content_type in PKCS7_ENC_CONTENT
13707   to data. This was previously part of the PKCS7 ASN1 code. This
13708   was causing problems with OpenSSL created PKCS#12 and PKCS#7 structures.
13709   *Steve Henson, reported by Kenneth R. Robinette
13710                              <support@securenetterm.com>*
13711
13712 * Add CRYPTO_push_info() and CRYPTO_pop_info() calls to new ASN1
13713   routines: without these tracing memory leaks is very painful.
13714   Fix leaks in PKCS12 and PKCS7 routines.
13715
13716   *Steve Henson*
13717
13718 * Make X509_time_adj() cope with the new behaviour of ASN1_TIME_new().
13719   Previously it initialised the 'type' argument to V_ASN1_UTCTIME which
13720   effectively meant GeneralizedTime would never be used. Now it
13721   is initialised to -1 but X509_time_adj() now has to check the value
13722   and use ASN1_TIME_set() if the value is not V_ASN1_UTCTIME or
13723   V_ASN1_GENERALIZEDTIME, without this it always uses GeneralizedTime.
13724   *Steve Henson, reported by Kenneth R. Robinette
13725                              <support@securenetterm.com>*
13726
13727 * Fixes to BN_to_ASN1_INTEGER when bn is zero. This would previously
13728   result in a zero length in the ASN1_INTEGER structure which was
13729   not consistent with the structure when d2i_ASN1_INTEGER() was used
13730   and would cause ASN1_INTEGER_cmp() to fail. Enhance s2i_ASN1_INTEGER()
13731   to cope with hex and negative integers. Fix bug in i2a_ASN1_INTEGER()
13732   where it did not print out a minus for negative ASN1_INTEGER.
13733
13734   *Steve Henson*
13735
13736 * Add summary printout to ocsp utility. The various functions which
13737   convert status values to strings have been renamed to:
13738   OCSP_response_status_str(), OCSP_cert_status_str() and
13739   OCSP_crl_reason_str() and are no longer static. New options
13740   to verify nonce values and to disable verification. OCSP response
13741   printout format cleaned up.
13742
13743   *Steve Henson*
13744
13745 * Add additional OCSP certificate checks. These are those specified
13746   in RFC2560. This consists of two separate checks: the CA of the
13747   certificate being checked must either be the OCSP signer certificate
13748   or the issuer of the OCSP signer certificate. In the latter case the
13749   OCSP signer certificate must contain the OCSP signing extended key
13750   usage. This check is performed by attempting to match the OCSP
13751   signer or the OCSP signer CA to the issuerNameHash and issuerKeyHash
13752   in the OCSP_CERTID structures of the response.
13753
13754   *Steve Henson*
13755
13756 * Initial OCSP certificate verification added to OCSP_basic_verify()
13757   and related routines. This uses the standard OpenSSL certificate
13758   verify routines to perform initial checks (just CA validity) and
13759   to obtain the certificate chain. Then additional checks will be
13760   performed on the chain. Currently the root CA is checked to see
13761   if it is explicitly trusted for OCSP signing. This is used to set
13762   a root CA as a global signing root: that is any certificate that
13763   chains to that CA is an acceptable OCSP signing certificate.
13764
13765   *Steve Henson*
13766
13767 * New '-extfile ...' option to 'openssl ca' for reading X.509v3
13768   extensions from a separate configuration file.
13769   As when reading extensions from the main configuration file,
13770   the '-extensions ...' option may be used for specifying the
13771   section to use.
13772
13773   *Massimiliano Pala <madwolf@comune.modena.it>*
13774
13775 * New OCSP utility. Allows OCSP requests to be generated or
13776   read. The request can be sent to a responder and the output
13777   parsed, outputted or printed in text form. Not complete yet:
13778   still needs to check the OCSP response validity.
13779
13780   *Steve Henson*
13781
13782 * New subcommands for 'openssl ca':
13783   `openssl ca -status <serial>` prints the status of the cert with
13784   the given serial number (according to the index file).
13785   `openssl ca -updatedb` updates the expiry status of certificates
13786   in the index file.
13787
13788   *Massimiliano Pala <madwolf@comune.modena.it>*
13789
13790 * New '-newreq-nodes' command option to CA.pl.  This is like
13791   '-newreq', but calls 'openssl req' with the '-nodes' option
13792   so that the resulting key is not encrypted.
13793
13794   *Damien Miller <djm@mindrot.org>*
13795
13796 * New configuration for the GNU Hurd.
13797
13798   *Jonathan Bartlett <johnnyb@wolfram.com> via Richard Levitte*
13799
13800 * Initial code to implement OCSP basic response verify. This
13801   is currently incomplete. Currently just finds the signer's
13802   certificate and verifies the signature on the response.
13803
13804   *Steve Henson*
13805
13806 * New SSLeay_version code SSLEAY_DIR to determine the compiled-in
13807   value of OPENSSLDIR.  This is available via the new '-d' option
13808   to 'openssl version', and is also included in 'openssl version -a'.
13809
13810   *Bodo Moeller*
13811
13812 * Allowing defining memory allocation callbacks that will be given
13813   file name and line number information in additional arguments
13814   (a `const char*` and an int).  The basic functionality remains, as
13815   well as the original possibility to just replace malloc(),
13816   realloc() and free() by functions that do not know about these
13817   additional arguments.  To register and find out the current
13818   settings for extended allocation functions, the following
13819   functions are provided:
13820
13821           CRYPTO_set_mem_ex_functions
13822           CRYPTO_set_locked_mem_ex_functions
13823           CRYPTO_get_mem_ex_functions
13824           CRYPTO_get_locked_mem_ex_functions
13825
13826   These work the same way as CRYPTO_set_mem_functions and friends.
13827   `CRYPTO_get_[locked_]mem_functions` now writes 0 where such an
13828   extended allocation function is enabled.
13829   Similarly, `CRYPTO_get_[locked_]mem_ex_functions` writes 0 where
13830   a conventional allocation function is enabled.
13831
13832   *Richard Levitte, Bodo Moeller*
13833
13834 * Finish off removing the remaining LHASH function pointer casts.
13835   There should no longer be any prototype-casting required when using
13836   the LHASH abstraction, and any casts that remain are "bugs". See
13837   the callback types and macros at the head of lhash.h for details
13838   (and "OBJ_cleanup" in crypto/objects/obj_dat.c as an example).
13839
13840   *Geoff Thorpe*
13841
13842 * Add automatic query of EGD sockets in RAND_poll() for the unix variant.
13843   If /dev/[u]random devices are not available or do not return enough
13844   entropy, EGD style sockets (served by EGD or PRNGD) will automatically
13845   be queried.
13846   The locations /var/run/egd-pool, /dev/egd-pool, /etc/egd-pool, and
13847   /etc/entropy will be queried once each in this sequence, querying stops
13848   when enough entropy was collected without querying more sockets.
13849
13850   *Lutz Jaenicke*
13851
13852 * Change the Unix RAND_poll() variant to be able to poll several
13853   random devices, as specified by DEVRANDOM, until a sufficient amount
13854   of data has been collected.   We spend at most 10 ms on each file
13855   (select timeout) and read in non-blocking mode.  DEVRANDOM now
13856   defaults to the list "/dev/urandom", "/dev/random", "/dev/srandom"
13857   (previously it was just the string "/dev/urandom"), so on typical
13858   platforms the 10 ms delay will never occur.
13859   Also separate out the Unix variant to its own file, rand_unix.c.
13860   For VMS, there's a currently-empty rand_vms.c.
13861
13862   *Richard Levitte*
13863
13864 * Move OCSP client related routines to ocsp_cl.c. These
13865   provide utility functions which an application needing
13866   to issue a request to an OCSP responder and analyse the
13867   response will typically need: as opposed to those which an
13868   OCSP responder itself would need which will be added later.
13869
13870   OCSP_request_sign() signs an OCSP request with an API similar
13871   to PKCS7_sign(). OCSP_response_status() returns status of OCSP
13872   response. OCSP_response_get1_basic() extracts basic response
13873   from response. OCSP_resp_find_status(): finds and extracts status
13874   information from an OCSP_CERTID structure (which will be created
13875   when the request structure is built). These are built from lower
13876   level functions which work on OCSP_SINGLERESP structures but
13877   won't normally be used unless the application wishes to examine
13878   extensions in the OCSP response for example.
13879
13880   Replace nonce routines with a pair of functions.
13881   OCSP_request_add1_nonce() adds a nonce value and optionally
13882   generates a random value. OCSP_check_nonce() checks the
13883   validity of the nonce in an OCSP response.
13884
13885   *Steve Henson*
13886
13887 * Change function OCSP_request_add() to OCSP_request_add0_id().
13888   This doesn't copy the supplied OCSP_CERTID and avoids the
13889   need to free up the newly created id. Change return type
13890   to OCSP_ONEREQ to return the internal OCSP_ONEREQ structure.
13891   This can then be used to add extensions to the request.
13892   Deleted OCSP_request_new(), since most of its functionality
13893   is now in OCSP_REQUEST_new() (and the case insensitive name
13894   clash) apart from the ability to set the request name which
13895   will be added elsewhere.
13896
13897   *Steve Henson*
13898
13899 * Update OCSP API. Remove obsolete extensions argument from
13900   various functions. Extensions are now handled using the new
13901   OCSP extension code. New simple OCSP HTTP function which
13902   can be used to send requests and parse the response.
13903
13904   *Steve Henson*
13905
13906 * Fix the PKCS#7 (S/MIME) code to work with new ASN1. Two new
13907   ASN1_ITEM structures help with sign and verify. PKCS7_ATTR_SIGN
13908   uses the special reorder version of SET OF to sort the attributes
13909   and reorder them to match the encoded order. This resolves a long
13910   standing problem: a verify on a PKCS7 structure just after signing
13911   it used to fail because the attribute order did not match the
13912   encoded order. PKCS7_ATTR_VERIFY does not reorder the attributes:
13913   it uses the received order. This is necessary to tolerate some broken
13914   software that does not order SET OF. This is handled by encoding
13915   as a SEQUENCE OF but using implicit tagging (with UNIVERSAL class)
13916   to produce the required SET OF.
13917
13918   *Steve Henson*
13919
13920 * Have mk1mf.pl generate the macros OPENSSL_BUILD_SHLIBCRYPTO and
13921   OPENSSL_BUILD_SHLIBSSL and use them appropriately in the header
13922   files to get correct declarations of the ASN.1 item variables.
13923
13924   *Richard Levitte*
13925
13926 * Rewrite of PKCS#12 code to use new ASN1 functionality. Replace many
13927   PKCS#12 macros with real functions. Fix two unrelated ASN1 bugs:
13928   asn1_check_tlen() would sometimes attempt to use 'ctx' when it was
13929   NULL and ASN1_TYPE was not dereferenced properly in asn1_ex_c2i().
13930   New ASN1 macro: DECLARE_ASN1_ITEM() which just declares the relevant
13931   ASN1_ITEM and no wrapper functions.
13932
13933   *Steve Henson*
13934
13935 * New functions or ASN1_item_d2i_fp() and ASN1_item_d2i_bio(). These
13936   replace the old function pointer based I/O routines. Change most of
13937   the `*_d2i_bio()` and `*_d2i_fp()` functions to use these.
13938
13939   *Steve Henson*
13940
13941 * Enhance mkdef.pl to be more accepting about spacing in C preprocessor
13942   lines, recognize more "algorithms" that can be deselected, and make
13943   it complain about algorithm deselection that isn't recognised.
13944
13945   *Richard Levitte*
13946
13947 * New ASN1 functions to handle dup, sign, verify, digest, pack and
13948   unpack operations in terms of ASN1_ITEM. Modify existing wrappers
13949   to use new functions. Add NO_ASN1_OLD which can be set to remove
13950   some old style ASN1 functions: this can be used to determine if old
13951   code will still work when these eventually go away.
13952
13953   *Steve Henson*
13954
13955 * New extension functions for OCSP structures, these follow the
13956   same conventions as certificates and CRLs.
13957
13958   *Steve Henson*
13959
13960 * New function X509V3_add1_i2d(). This automatically encodes and
13961   adds an extension. Its behaviour can be customised with various
13962   flags to append, replace or delete. Various wrappers added for
13963   certificates and CRLs.
13964
13965   *Steve Henson*
13966
13967 * Fix to avoid calling the underlying ASN1 print routine when
13968   an extension cannot be parsed. Correct a typo in the
13969   OCSP_SERVICELOC extension. Tidy up print OCSP format.
13970
13971   *Steve Henson*
13972
13973 * Make mkdef.pl parse some of the ASN1 macros and add appropriate
13974   entries for variables.
13975
13976   *Steve Henson*
13977
13978 * Add functionality to `apps/openssl.c` for detecting locking
13979   problems: As the program is single-threaded, all we have
13980   to do is register a locking callback using an array for
13981   storing which locks are currently held by the program.
13982
13983   *Bodo Moeller*
13984
13985 * Use a lock around the call to CRYPTO_get_ex_new_index() in
13986   SSL_get_ex_data_X509_STORE_idx(), which is used in
13987   ssl_verify_cert_chain() and thus can be called at any time
13988   during TLS/SSL handshakes so that thread-safety is essential.
13989   Unfortunately, the ex_data design is not at all suited
13990   for multi-threaded use, so it probably should be abolished.
13991
13992   *Bodo Moeller*
13993
13994 * Added Broadcom "ubsec" ENGINE to OpenSSL.
13995
13996   *Broadcom, tweaked and integrated by Geoff Thorpe*
13997
13998 * Move common extension printing code to new function
13999   X509V3_print_extensions(). Reorganise OCSP print routines and
14000   implement some needed OCSP ASN1 functions. Add OCSP extensions.
14001
14002   *Steve Henson*
14003
14004 * New function X509_signature_print() to remove duplication in some
14005   print routines.
14006
14007   *Steve Henson*
14008
14009 * Add a special meaning when SET OF and SEQUENCE OF flags are both
14010   set (this was treated exactly the same as SET OF previously). This
14011   is used to reorder the STACK representing the structure to match the
14012   encoding. This will be used to get round a problem where a PKCS7
14013   structure which was signed could not be verified because the STACK
14014   order did not reflect the encoded order.
14015
14016   *Steve Henson*
14017
14018 * Reimplement the OCSP ASN1 module using the new code.
14019
14020   *Steve Henson*
14021
14022 * Update the X509V3 code to permit the use of an ASN1_ITEM structure
14023   for its ASN1 operations. The old style function pointers still exist
14024   for now but they will eventually go away.
14025
14026   *Steve Henson*
14027
14028 * Merge in replacement ASN1 code from the ASN1 branch. This almost
14029   completely replaces the old ASN1 functionality with a table driven
14030   encoder and decoder which interprets an ASN1_ITEM structure describing
14031   the ASN1 module. Compatibility with the existing ASN1 API (i2d,d2i) is
14032   largely maintained. Almost all of the old asn1_mac.h macro based ASN1
14033   has also been converted to the new form.
14034
14035   *Steve Henson*
14036
14037 * Change BN_mod_exp_recp so that negative moduli are tolerated
14038   (the sign is ignored).  Similarly, ignore the sign in BN_MONT_CTX_set
14039   so that BN_mod_exp_mont and BN_mod_exp_mont_word work
14040   for negative moduli.
14041
14042   *Bodo Moeller*
14043
14044 * Fix BN_uadd and BN_usub: Always return non-negative results instead
14045   of not touching the result's sign bit.
14046
14047   *Bodo Moeller*
14048
14049 * BN_div bugfix: If the result is 0, the sign (res->neg) must not be
14050   set.
14051
14052   *Bodo Moeller*
14053
14054 * Changed the LHASH code to use prototypes for callbacks, and created
14055   macros to declare and implement thin (optionally static) functions
14056   that provide type-safety and avoid function pointer casting for the
14057   type-specific callbacks.
14058
14059   *Geoff Thorpe*
14060
14061 * Added Kerberos Cipher Suites to be used with TLS, as written in
14062   RFC 2712.
14063   *Veers Staats <staatsvr@asc.hpc.mil>,
14064   Jeffrey Altman <jaltman@columbia.edu>, via Richard Levitte*
14065
14066 * Reformat the FAQ so the different questions and answers can be divided
14067   in sections depending on the subject.
14068
14069   *Richard Levitte*
14070
14071 * Have the zlib compression code load ZLIB.DLL dynamically under
14072   Windows.
14073
14074   *Richard Levitte*
14075
14076 * New function BN_mod_sqrt for computing square roots modulo a prime
14077   (using the probabilistic Tonelli-Shanks algorithm unless
14078   p == 3 (mod 4)  or  p == 5 (mod 8),  which are cases that can
14079   be handled deterministically).
14080
14081   *Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller*
14082
14083 * Make BN_mod_inverse faster by explicitly handling small quotients
14084   in the Euclid loop. (Speed gain about 20% for small moduli [256 or
14085   512 bits], about 30% for larger ones [1024 or 2048 bits].)
14086
14087   *Bodo Moeller*
14088
14089 * New function BN_kronecker.
14090
14091   *Bodo Moeller*
14092
14093 * Fix BN_gcd so that it works on negative inputs; the result is
14094   positive unless both parameters are zero.
14095   Previously something reasonably close to an infinite loop was
14096   possible because numbers could be growing instead of shrinking
14097   in the implementation of Euclid's algorithm.
14098
14099   *Bodo Moeller*
14100
14101 * Fix BN_is_word() and BN_is_one() macros to take into account the
14102   sign of the number in question.
14103
14104   Fix BN_is_word(a,w) to work correctly for w == 0.
14105
14106   The old BN_is_word(a,w) macro is now called BN_abs_is_word(a,w)
14107   because its test if the absolute value of 'a' equals 'w'.
14108   Note that BN_abs_is_word does *not* handle w == 0 reliably;
14109   it exists mostly for use in the implementations of BN_is_zero(),
14110   BN_is_one(), and BN_is_word().
14111
14112   *Bodo Moeller*
14113
14114 * New function BN_swap.
14115
14116   *Bodo Moeller*
14117
14118 * Use BN_nnmod instead of BN_mod in crypto/bn/bn_exp.c so that
14119   the exponentiation functions are more likely to produce reasonable
14120   results on negative inputs.
14121
14122   *Bodo Moeller*
14123
14124 * Change BN_mod_mul so that the result is always non-negative.
14125   Previously, it could be negative if one of the factors was negative;
14126   I don't think anyone really wanted that behaviour.
14127
14128   *Bodo Moeller*
14129
14130 * Move `BN_mod_...` functions into new file `crypto/bn/bn_mod.c`
14131   (except for exponentiation, which stays in `crypto/bn/bn_exp.c`,
14132   and `BN_mod_mul_reciprocal`, which stays in `crypto/bn/bn_recp.c`)
14133   and add new functions:
14134
14135           BN_nnmod
14136           BN_mod_sqr
14137           BN_mod_add
14138           BN_mod_add_quick
14139           BN_mod_sub
14140           BN_mod_sub_quick
14141           BN_mod_lshift1
14142           BN_mod_lshift1_quick
14143           BN_mod_lshift
14144           BN_mod_lshift_quick
14145
14146   These functions always generate non-negative results.
14147
14148   `BN_nnmod` otherwise is `like BN_mod` (if `BN_mod` computes a remainder `r`
14149   such that `|m| < r < 0`, `BN_nnmod` will output `rem + |m|` instead).
14150
14151   `BN_mod_XXX_quick(r, a, [b,] m)` generates the same result as
14152   `BN_mod_XXX(r, a, [b,] m, ctx)`, but requires that `a` [and  `b`]
14153   be reduced modulo `m`.
14154
14155   *Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller*
14156
14157<!--
14158   The following entry accidentally appeared in the CHANGES file
14159   distributed with OpenSSL 0.9.7.  The modifications described in
14160   it do *not* apply to OpenSSL 0.9.7.
14161
14162 * Remove a few calls to bn_wexpand() in BN_sqr() (the one in there
14163   was actually never needed) and in BN_mul().  The removal in BN_mul()
14164   required a small change in bn_mul_part_recursive() and the addition
14165   of the functions bn_cmp_part_words(), bn_sub_part_words() and
14166   bn_add_part_words(), which do the same thing as bn_cmp_words(),
14167   bn_sub_words() and bn_add_words() except they take arrays with
14168   differing sizes.
14169
14170   *Richard Levitte*
14171-->
14172
14173 * In 'openssl passwd', verify passwords read from the terminal
14174   unless the '-salt' option is used (which usually means that
14175   verification would just waste user's time since the resulting
14176   hash is going to be compared with some given password hash)
14177   or the new '-noverify' option is used.
14178
14179   This is an incompatible change, but it does not affect
14180   non-interactive use of 'openssl passwd' (passwords on the command
14181   line, '-stdin' option, '-in ...' option) and thus should not
14182   cause any problems.
14183
14184   *Bodo Moeller*
14185
14186 * Remove all references to RSAref, since there's no more need for it.
14187
14188   *Richard Levitte*
14189
14190 * Make DSO load along a path given through an environment variable
14191   (SHLIB_PATH) with shl_load().
14192
14193   *Richard Levitte*
14194
14195 * Constify the ENGINE code as a result of BIGNUM constification.
14196   Also constify the RSA code and most things related to it.  In a
14197   few places, most notable in the depth of the ASN.1 code, ugly
14198   casts back to non-const were required (to be solved at a later
14199   time)
14200
14201   *Richard Levitte*
14202
14203 * Make it so the openssl application has all engines loaded by default.
14204
14205   *Richard Levitte*
14206
14207 * Constify the BIGNUM routines a little more.
14208
14209   *Richard Levitte*
14210
14211 * Add the following functions:
14212
14213           ENGINE_load_cswift()
14214           ENGINE_load_chil()
14215           ENGINE_load_atalla()
14216           ENGINE_load_nuron()
14217           ENGINE_load_builtin_engines()
14218
14219   That way, an application can itself choose if external engines that
14220   are built-in in OpenSSL shall ever be used or not.  The benefit is
14221   that applications won't have to be linked with libdl or other dso
14222   libraries unless it's really needed.
14223
14224   Changed 'openssl engine' to load all engines on demand.
14225   Changed the engine header files to avoid the duplication of some
14226   declarations (they differed!).
14227
14228   *Richard Levitte*
14229
14230 * 'openssl engine' can now list capabilities.
14231
14232   *Richard Levitte*
14233
14234 * Better error reporting in 'openssl engine'.
14235
14236   *Richard Levitte*
14237
14238 * Never call load_dh_param(NULL) in s_server.
14239
14240   *Bodo Moeller*
14241
14242 * Add engine application.  It can currently list engines by name and
14243   identity, and test if they are actually available.
14244
14245   *Richard Levitte*
14246
14247 * Improve RPM specification file by forcing symbolic linking and making
14248   sure the installed documentation is also owned by root.root.
14249
14250   *Damien Miller <djm@mindrot.org>*
14251
14252 * Give the OpenSSL applications more possibilities to make use of
14253   keys (public as well as private) handled by engines.
14254
14255   *Richard Levitte*
14256
14257 * Add OCSP code that comes from CertCo.
14258
14259   *Richard Levitte*
14260
14261 * Add VMS support for the Rijndael code.
14262
14263   *Richard Levitte*
14264
14265 * Added untested support for Nuron crypto accelerator.
14266
14267   *Ben Laurie*
14268
14269 * Add support for external cryptographic devices.  This code was
14270   previously distributed separately as the "engine" branch.
14271
14272   *Geoff Thorpe, Richard Levitte*
14273
14274 * Rework the filename-translation in the DSO code. It is now possible to
14275   have far greater control over how a "name" is turned into a filename
14276   depending on the operating environment and any oddities about the
14277   different shared library filenames on each system.
14278
14279   *Geoff Thorpe*
14280
14281 * Support threads on FreeBSD-elf in Configure.
14282
14283   *Richard Levitte*
14284
14285 * Fix for SHA1 assembly problem with MASM: it produces
14286   warnings about corrupt line number information when assembling
14287   with debugging information. This is caused by the overlapping
14288   of two sections.
14289
14290   *Bernd Matthes <mainbug@celocom.de>, Steve Henson*
14291
14292 * NCONF changes.
14293   NCONF_get_number() has no error checking at all.  As a replacement,
14294   NCONF_get_number_e() is defined (`_e` for "error checking") and is
14295   promoted strongly.  The old NCONF_get_number is kept around for
14296   binary backward compatibility.
14297   Make it possible for methods to load from something other than a BIO,
14298   by providing a function pointer that is given a name instead of a BIO.
14299   For example, this could be used to load configuration data from an
14300   LDAP server.
14301
14302   *Richard Levitte*
14303
14304 * Fix for non blocking accept BIOs. Added new I/O special reason
14305   BIO_RR_ACCEPT to cover this case. Previously use of accept BIOs
14306   with non blocking I/O was not possible because no retry code was
14307   implemented. Also added new SSL code SSL_WANT_ACCEPT to cover
14308   this case.
14309
14310   *Steve Henson*
14311
14312 * Added the beginnings of Rijndael support.
14313
14314   *Ben Laurie*
14315
14316 * Fix for bug in DirectoryString mask setting. Add support for
14317   X509_NAME_print_ex() in 'req' and X509_print_ex() function
14318   to allow certificate printing to more controllable, additional
14319   'certopt' option to 'x509' to allow new printing options to be
14320   set.
14321
14322   *Steve Henson*
14323
14324 * Clean old EAY MD5 hack from e_os.h.
14325
14326   *Richard Levitte*
14327
14328### Changes between 0.9.6l and 0.9.6m  [17 Mar 2004]
14329
14330 * Fix null-pointer assignment in do_change_cipher_spec() revealed
14331   by using the Codenomicon TLS Test Tool ([CVE-2004-0079])
14332
14333   *Joe Orton, Steve Henson*
14334
14335### Changes between 0.9.6k and 0.9.6l  [04 Nov 2003]
14336
14337 * Fix additional bug revealed by the NISCC test suite:
14338
14339   Stop bug triggering large recursion when presented with
14340   certain ASN.1 tags ([CVE-2003-0851])
14341
14342   *Steve Henson*
14343
14344### Changes between 0.9.6j and 0.9.6k  [30 Sep 2003]
14345
14346 * Fix various bugs revealed by running the NISCC test suite:
14347
14348   Stop out of bounds reads in the ASN1 code when presented with
14349   invalid tags (CVE-2003-0543 and CVE-2003-0544).
14350
14351   If verify callback ignores invalid public key errors don't try to check
14352   certificate signature with the NULL public key.
14353
14354   *Steve Henson*
14355
14356 * In ssl3_accept() (ssl/s3_srvr.c) only accept a client certificate
14357   if the server requested one: as stated in TLS 1.0 and SSL 3.0
14358   specifications.
14359
14360   *Steve Henson*
14361
14362 * In ssl3_get_client_hello() (ssl/s3_srvr.c), tolerate additional
14363   extra data after the compression methods not only for TLS 1.0
14364   but also for SSL 3.0 (as required by the specification).
14365
14366   *Bodo Moeller; problem pointed out by Matthias Loepfe*
14367
14368 * Change X509_certificate_type() to mark the key as exported/exportable
14369   when it's 512 *bits* long, not 512 bytes.
14370
14371   *Richard Levitte*
14372
14373### Changes between 0.9.6i and 0.9.6j  [10 Apr 2003]
14374
14375 * Countermeasure against the Klima-Pokorny-Rosa extension of
14376   Bleichbacher's attack on PKCS #1 v1.5 padding: treat
14377   a protocol version number mismatch like a decryption error
14378   in ssl3_get_client_key_exchange (ssl/s3_srvr.c).
14379
14380   *Bodo Moeller*
14381
14382 * Turn on RSA blinding by default in the default implementation
14383   to avoid a timing attack. Applications that don't want it can call
14384   RSA_blinding_off() or use the new flag RSA_FLAG_NO_BLINDING.
14385   They would be ill-advised to do so in most cases.
14386
14387   *Ben Laurie, Steve Henson, Geoff Thorpe, Bodo Moeller*
14388
14389 * Change RSA blinding code so that it works when the PRNG is not
14390   seeded (in this case, the secret RSA exponent is abused as
14391   an unpredictable seed -- if it is not unpredictable, there
14392   is no point in blinding anyway).  Make RSA blinding thread-safe
14393   by remembering the creator's thread ID in rsa->blinding and
14394   having all other threads use local one-time blinding factors
14395   (this requires more computation than sharing rsa->blinding, but
14396   avoids excessive locking; and if an RSA object is not shared
14397   between threads, blinding will still be very fast).
14398
14399   *Bodo Moeller*
14400
14401### Changes between 0.9.6h and 0.9.6i  [19 Feb 2003]
14402
14403 * In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked
14404   via timing by performing a MAC computation even if incorrect
14405   block cipher padding has been found.  This is a countermeasure
14406   against active attacks where the attacker has to distinguish
14407   between bad padding and a MAC verification error. ([CVE-2003-0078])
14408
14409   *Bodo Moeller; problem pointed out by Brice Canvel (EPFL),
14410   Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and
14411   Martin Vuagnoux (EPFL, Ilion)*
14412
14413### Changes between 0.9.6g and 0.9.6h  [5 Dec 2002]
14414
14415 * New function OPENSSL_cleanse(), which is used to cleanse a section of
14416   memory from its contents.  This is done with a counter that will
14417   place alternating values in each byte.  This can be used to solve
14418   two issues: 1) the removal of calls to memset() by highly optimizing
14419   compilers, and 2) cleansing with other values than 0, since those can
14420   be read through on certain media, for example a swap space on disk.
14421
14422   *Geoff Thorpe*
14423
14424 * Bugfix: client side session caching did not work with external caching,
14425   because the session->cipher setting was not restored when reloading
14426   from the external cache. This problem was masked, when
14427   SSL_OP_NETSCAPE_REUSE_CIPHER_CHANGE_BUG (part of SSL_OP_ALL) was set.
14428   (Found by Steve Haslam <steve@araqnid.ddts.net>.)
14429
14430   *Lutz Jaenicke*
14431
14432 * Fix client_certificate (ssl/s2_clnt.c): The permissible total
14433   length of the REQUEST-CERTIFICATE message is 18 .. 34, not 17 .. 33.
14434
14435   *Zeev Lieber <zeev-l@yahoo.com>*
14436
14437 * Undo an undocumented change introduced in 0.9.6e which caused
14438   repeated calls to OpenSSL_add_all_ciphers() and
14439   OpenSSL_add_all_digests() to be ignored, even after calling
14440   EVP_cleanup().
14441
14442   *Richard Levitte*
14443
14444 * Change the default configuration reader to deal with last line not
14445   being properly terminated.
14446
14447   *Richard Levitte*
14448
14449 * Change X509_NAME_cmp() so it applies the special rules on handling
14450   DN values that are of type PrintableString, as well as RDNs of type
14451   emailAddress where the value has the type ia5String.
14452
14453   *stefank@valicert.com via Richard Levitte*
14454
14455 * Add a SSL_SESS_CACHE_NO_INTERNAL_STORE flag to take over half
14456   the job SSL_SESS_CACHE_NO_INTERNAL_LOOKUP was inconsistently
14457   doing, define a new flag (SSL_SESS_CACHE_NO_INTERNAL) to be
14458   the bitwise-OR of the two for use by the majority of applications
14459   wanting this behaviour, and update the docs. The documented
14460   behaviour and actual behaviour were inconsistent and had been
14461   changing anyway, so this is more a bug-fix than a behavioural
14462   change.
14463
14464   *Geoff Thorpe, diagnosed by Nadav Har'El*
14465
14466 * Don't impose a 16-byte length minimum on session IDs in ssl/s3_clnt.c
14467   (the SSL 3.0 and TLS 1.0 specifications allow any length up to 32 bytes).
14468
14469   *Bodo Moeller*
14470
14471 * Fix initialization code race conditions in
14472           SSLv23_method(),  SSLv23_client_method(),   SSLv23_server_method(),
14473           SSLv2_method(),   SSLv2_client_method(),    SSLv2_server_method(),
14474           SSLv3_method(),   SSLv3_client_method(),    SSLv3_server_method(),
14475           TLSv1_method(),   TLSv1_client_method(),    TLSv1_server_method(),
14476           ssl2_get_cipher_by_char(),
14477           ssl3_get_cipher_by_char().
14478
14479   *Patrick McCormick <patrick@tellme.com>, Bodo Moeller*
14480
14481 * Reorder cleanup sequence in SSL_CTX_free(): only remove the ex_data after
14482   the cached sessions are flushed, as the remove_cb() might use ex_data
14483   contents. Bug found by Sam Varshavchik <mrsam@courier-mta.com>
14484   (see [openssl.org #212]).
14485
14486   *Geoff Thorpe, Lutz Jaenicke*
14487
14488 * Fix typo in OBJ_txt2obj which incorrectly passed the content
14489   length, instead of the encoding length to d2i_ASN1_OBJECT.
14490
14491   *Steve Henson*
14492
14493### Changes between 0.9.6f and 0.9.6g  [9 Aug 2002]
14494
14495 * [In 0.9.6g-engine release:]
14496   Fix crypto/engine/vendor_defns/cswift.h for WIN32 (use `_stdcall`).
14497
14498   *Lynn Gazis <lgazis@rainbow.com>*
14499
14500### Changes between 0.9.6e and 0.9.6f  [8 Aug 2002]
14501
14502 * Fix ASN1 checks. Check for overflow by comparing with LONG_MAX
14503   and get fix the header length calculation.
14504   *Florian Weimer <Weimer@CERT.Uni-Stuttgart.DE>,
14505   Alon Kantor <alonk@checkpoint.com> (and others), Steve Henson*
14506
14507 * Use proper error handling instead of 'assertions' in buffer
14508   overflow checks added in 0.9.6e.  This prevents DoS (the
14509   assertions could call abort()).
14510
14511   *Arne Ansper <arne@ats.cyber.ee>, Bodo Moeller*
14512
14513### Changes between 0.9.6d and 0.9.6e  [30 Jul 2002]
14514
14515 * Add various sanity checks to asn1_get_length() to reject
14516   the ASN1 length bytes if they exceed sizeof(long), will appear
14517   negative or the content length exceeds the length of the
14518   supplied buffer.
14519
14520   *Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>*
14521
14522 * Fix cipher selection routines: ciphers without encryption had no flags
14523   for the cipher strength set and where therefore not handled correctly
14524   by the selection routines (PR #130).
14525
14526   *Lutz Jaenicke*
14527
14528 * Fix EVP_dsa_sha macro.
14529
14530   *Nils Larsch*
14531
14532 * New option
14533        SSL_OP_DONT_INSERT_EMPTY_FRAGMENTS
14534   for disabling the SSL 3.0/TLS 1.0 CBC vulnerability countermeasure
14535   that was added in OpenSSL 0.9.6d.
14536
14537   As the countermeasure turned out to be incompatible with some
14538   broken SSL implementations, the new option is part of SSL_OP_ALL.
14539   SSL_OP_ALL is usually employed when compatibility with weird SSL
14540   implementations is desired (e.g. '-bugs' option to 's_client' and
14541   's_server'), so the new option is automatically set in many
14542   applications.
14543
14544   *Bodo Moeller*
14545
14546 * Changes in security patch:
14547
14548   Changes marked "(CHATS)" were sponsored by the Defense Advanced
14549   Research Projects Agency (DARPA) and Air Force Research Laboratory,
14550   Air Force Materiel Command, USAF, under agreement number
14551   F30602-01-2-0537.
14552
14553 * Add various sanity checks to asn1_get_length() to reject
14554   the ASN1 length bytes if they exceed sizeof(long), will appear
14555   negative or the content length exceeds the length of the
14556   supplied buffer. ([CVE-2002-0659])
14557
14558   *Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>*
14559
14560 * Assertions for various potential buffer overflows, not known to
14561   happen in practice.
14562
14563   *Ben Laurie (CHATS)*
14564
14565 * Various temporary buffers to hold ASCII versions of integers were
14566   too small for 64 bit platforms. ([CVE-2002-0655])
14567   *Matthew Byng-Maddick <mbm@aldigital.co.uk> and Ben Laurie (CHATS)>*
14568
14569 * Remote buffer overflow in SSL3 protocol - an attacker could
14570   supply an oversized session ID to a client. ([CVE-2002-0656])
14571
14572   *Ben Laurie (CHATS)*
14573
14574 * Remote buffer overflow in SSL2 protocol - an attacker could
14575   supply an oversized client master key. ([CVE-2002-0656])
14576
14577   *Ben Laurie (CHATS)*
14578
14579### Changes between 0.9.6c and 0.9.6d  [9 May 2002]
14580
14581 * Fix crypto/asn1/a_sign.c so that 'parameters' is omitted (not
14582   encoded as NULL) with id-dsa-with-sha1.
14583
14584   *Nils Larsch <nla@trustcenter.de>; problem pointed out by Bodo Moeller*
14585
14586 * Check various `X509_...()` return values in `apps/req.c`.
14587
14588   *Nils Larsch <nla@trustcenter.de>*
14589
14590 * Fix BASE64 decode (EVP_DecodeUpdate) for data with CR/LF ended lines:
14591   an end-of-file condition would erroneously be flagged, when the CRLF
14592   was just at the end of a processed block. The bug was discovered when
14593   processing data through a buffering memory BIO handing the data to a
14594   BASE64-decoding BIO. Bug fund and patch submitted by Pavel Tsekov
14595   <ptsekov@syntrex.com> and Nedelcho Stanev.
14596
14597   *Lutz Jaenicke*
14598
14599 * Implement a countermeasure against a vulnerability recently found
14600   in CBC ciphersuites in SSL 3.0/TLS 1.0: Send an empty fragment
14601   before application data chunks to avoid the use of known IVs
14602   with data potentially chosen by the attacker.
14603
14604   *Bodo Moeller*
14605
14606 * Fix length checks in ssl3_get_client_hello().
14607
14608   *Bodo Moeller*
14609
14610 * TLS/SSL library bugfix: use s->s3->in_read_app_data differently
14611   to prevent ssl3_read_internal() from incorrectly assuming that
14612   ssl3_read_bytes() found application data while handshake
14613   processing was enabled when in fact s->s3->in_read_app_data was
14614   merely automatically cleared during the initial handshake.
14615
14616   *Bodo Moeller; problem pointed out by Arne Ansper <arne@ats.cyber.ee>*
14617
14618 * Fix object definitions for Private and Enterprise: they were not
14619   recognized in their shortname (=lowercase) representation. Extend
14620   obj_dat.pl to issue an error when using undefined keywords instead
14621   of silently ignoring the problem (Svenning Sorensen
14622   <sss@sss.dnsalias.net>).
14623
14624   *Lutz Jaenicke*
14625
14626 * Fix DH_generate_parameters() so that it works for 'non-standard'
14627   generators, i.e. generators other than 2 and 5.  (Previously, the
14628   code did not properly initialise the 'add' and 'rem' values to
14629   BN_generate_prime().)
14630
14631   In the new general case, we do not insist that 'generator' is
14632   actually a primitive root: This requirement is rather pointless;
14633   a generator of the order-q subgroup is just as good, if not
14634   better.
14635
14636   *Bodo Moeller*
14637
14638 * Map new X509 verification errors to alerts. Discovered and submitted by
14639   Tom Wu <tom@arcot.com>.
14640
14641   *Lutz Jaenicke*
14642
14643 * Fix ssl3_pending() (ssl/s3_lib.c) to prevent SSL_pending() from
14644   returning non-zero before the data has been completely received
14645   when using non-blocking I/O.
14646
14647   *Bodo Moeller; problem pointed out by John Hughes*
14648
14649 * Some of the ciphers missed the strength entry (SSL_LOW etc).
14650
14651   *Ben Laurie, Lutz Jaenicke*
14652
14653 * Fix bug in SSL_clear(): bad sessions were not removed (found by
14654   Yoram Zahavi <YoramZ@gilian.com>).
14655
14656   *Lutz Jaenicke*
14657
14658 * Add information about CygWin 1.3 and on, and preserve proper
14659   configuration for the versions before that.
14660
14661   *Corinna Vinschen <vinschen@redhat.com> and Richard Levitte*
14662
14663 * Make removal from session cache (SSL_CTX_remove_session()) more robust:
14664   check whether we deal with a copy of a session and do not delete from
14665   the cache in this case. Problem reported by "Izhar Shoshani Levi"
14666   <izhar@checkpoint.com>.
14667
14668   *Lutz Jaenicke*
14669
14670 * Do not store session data into the internal session cache, if it
14671   is never intended to be looked up (SSL_SESS_CACHE_NO_INTERNAL_LOOKUP
14672   flag is set). Proposed by Aslam <aslam@funk.com>.
14673
14674   *Lutz Jaenicke*
14675
14676 * Have ASN1_BIT_STRING_set_bit() really clear a bit when the requested
14677   value is 0.
14678
14679   *Richard Levitte*
14680
14681 * [In 0.9.6d-engine release:]
14682   Fix a crashbug and a logic bug in hwcrhk_load_pubkey().
14683
14684   *Toomas Kiisk <vix@cyber.ee> via Richard Levitte*
14685
14686 * Add the configuration target linux-s390x.
14687
14688   *Neale Ferguson <Neale.Ferguson@SoftwareAG-USA.com> via Richard Levitte*
14689
14690 * The earlier bugfix for the SSL3_ST_SW_HELLO_REQ_C case of
14691   ssl3_accept (ssl/s3_srvr.c) incorrectly used a local flag
14692   variable as an indication that a ClientHello message has been
14693   received.  As the flag value will be lost between multiple
14694   invocations of ssl3_accept when using non-blocking I/O, the
14695   function may not be aware that a handshake has actually taken
14696   place, thus preventing a new session from being added to the
14697   session cache.
14698
14699   To avoid this problem, we now set s->new_session to 2 instead of
14700   using a local variable.
14701
14702   *Lutz Jaenicke, Bodo Moeller*
14703
14704 * Bugfix: Return -1 from ssl3_get_server_done (ssl3/s3_clnt.c)
14705   if the SSL_R_LENGTH_MISMATCH error is detected.
14706
14707   *Geoff Thorpe, Bodo Moeller*
14708
14709 * New 'shared_ldflag' column in Configure platform table.
14710
14711   *Richard Levitte*
14712
14713 * Fix EVP_CIPHER_mode macro.
14714
14715   *"Dan S. Camper" <dan@bti.net>*
14716
14717 * Fix ssl3_read_bytes (ssl/s3_pkt.c): To ignore messages of unknown
14718   type, we must throw them away by setting rr->length to 0.
14719
14720   *D P Chang <dpc@qualys.com>*
14721
14722### Changes between 0.9.6b and 0.9.6c  [21 dec 2001]
14723
14724 * Fix BN_rand_range bug pointed out by Dominikus Scherkl
14725   <Dominikus.Scherkl@biodata.com>.  (The previous implementation
14726   worked incorrectly for those cases where range = `10..._2`  and
14727   `3*range`  is two bits longer than  range.)
14728
14729   *Bodo Moeller*
14730
14731 * Only add signing time to PKCS7 structures if it is not already
14732   present.
14733
14734   *Steve Henson*
14735
14736 * Fix crypto/objects/objects.h: "ld-ce" should be "id-ce",
14737   OBJ_ld_ce should be OBJ_id_ce.
14738   Also some ip-pda OIDs in crypto/objects/objects.txt were
14739   incorrect (cf. RFC 3039).
14740
14741   *Matt Cooper, Frederic Giudicelli, Bodo Moeller*
14742
14743 * Release CRYPTO_LOCK_DYNLOCK when CRYPTO_destroy_dynlockid()
14744   returns early because it has nothing to do.
14745
14746   *Andy Schneider <andy.schneider@bjss.co.uk>*
14747
14748 * [In 0.9.6c-engine release:]
14749   Fix mutex callback return values in crypto/engine/hw_ncipher.c.
14750
14751   *Andy Schneider <andy.schneider@bjss.co.uk>*
14752
14753 * [In 0.9.6c-engine release:]
14754   Add support for Cryptographic Appliance's keyserver technology.
14755   (Use engine 'keyclient')
14756
14757   *Cryptographic Appliances and Geoff Thorpe*
14758
14759 * Add a configuration entry for OS/390 Unix.  The C compiler 'c89'
14760   is called via tools/c89.sh because arguments have to be
14761   rearranged (all '-L' options must appear before the first object
14762   modules).
14763
14764   *Richard Shapiro <rshapiro@abinitio.com>*
14765
14766 * [In 0.9.6c-engine release:]
14767   Add support for Broadcom crypto accelerator cards, backported
14768   from 0.9.7.
14769
14770   *Broadcom, Nalin Dahyabhai <nalin@redhat.com>, Mark Cox*
14771
14772 * [In 0.9.6c-engine release:]
14773   Add support for SureWare crypto accelerator cards from
14774   Baltimore Technologies.  (Use engine 'sureware')
14775
14776   *Baltimore Technologies and Mark Cox*
14777
14778 * [In 0.9.6c-engine release:]
14779   Add support for crypto accelerator cards from Accelerated
14780   Encryption Processing, www.aep.ie.  (Use engine 'aep')
14781
14782   *AEP Inc. and Mark Cox*
14783
14784 * Add a configuration entry for gcc on UnixWare.
14785
14786   *Gary Benson <gbenson@redhat.com>*
14787
14788 * Change ssl/s2_clnt.c and ssl/s2_srvr.c so that received handshake
14789   messages are stored in a single piece (fixed-length part and
14790   variable-length part combined) and fix various bugs found on the way.
14791
14792   *Bodo Moeller*
14793
14794 * Disable caching in BIO_gethostbyname(), directly use gethostbyname()
14795   instead.  BIO_gethostbyname() does not know what timeouts are
14796   appropriate, so entries would stay in cache even when they have
14797   become invalid.
14798   *Bodo Moeller; problem pointed out by Rich Salz <rsalz@zolera.com>*
14799
14800 * Change ssl23_get_client_hello (ssl/s23_srvr.c) behaviour when
14801   faced with a pathologically small ClientHello fragment that does
14802   not contain client_version: Instead of aborting with an error,
14803   simply choose the highest available protocol version (i.e.,
14804   TLS 1.0 unless it is disabled).  In practice, ClientHello
14805   messages are never sent like this, but this change gives us
14806   strictly correct behaviour at least for TLS.
14807
14808   *Bodo Moeller*
14809
14810 * Fix SSL handshake functions and SSL_clear() such that SSL_clear()
14811   never resets s->method to s->ctx->method when called from within
14812   one of the SSL handshake functions.
14813
14814   *Bodo Moeller; problem pointed out by Niko Baric*
14815
14816 * In ssl3_get_client_hello (ssl/s3_srvr.c), generate a fatal alert
14817   (sent using the client's version number) if client_version is
14818   smaller than the protocol version in use.  Also change
14819   ssl23_get_client_hello (ssl/s23_srvr.c) to select TLS 1.0 if
14820   the client demanded SSL 3.0 but only TLS 1.0 is enabled; then
14821   the client will at least see that alert.
14822
14823   *Bodo Moeller*
14824
14825 * Fix ssl3_get_message (ssl/s3_both.c) to handle message fragmentation
14826   correctly.
14827
14828   *Bodo Moeller*
14829
14830 * Avoid infinite loop in ssl3_get_message (ssl/s3_both.c) if a
14831   client receives HelloRequest while in a handshake.
14832
14833   *Bodo Moeller; bug noticed by Andy Schneider <andy.schneider@bjss.co.uk>*
14834
14835 * Bugfix in ssl3_accept (ssl/s3_srvr.c): Case SSL3_ST_SW_HELLO_REQ_C
14836   should end in 'break', not 'goto end' which circumvents various
14837   cleanups done in state SSL_ST_OK.   But session related stuff
14838   must be disabled for SSL_ST_OK in the case that we just sent a
14839   HelloRequest.
14840
14841   Also avoid some overhead by not calling ssl_init_wbio_buffer()
14842   before just sending a HelloRequest.
14843
14844   *Bodo Moeller, Eric Rescorla <ekr@rtfm.com>*
14845
14846 * Fix ssl/s3_enc.c, ssl/t1_enc.c and ssl/s3_pkt.c so that we don't
14847   reveal whether illegal block cipher padding was found or a MAC
14848   verification error occurred.  (Neither SSLerr() codes nor alerts
14849   are directly visible to potential attackers, but the information
14850   may leak via logfiles.)
14851
14852   Similar changes are not required for the SSL 2.0 implementation
14853   because the number of padding bytes is sent in clear for SSL 2.0,
14854   and the extra bytes are just ignored.  However ssl/s2_pkt.c
14855   failed to verify that the purported number of padding bytes is in
14856   the legal range.
14857
14858   *Bodo Moeller*
14859
14860 * Add OpenUNIX-8 support including shared libraries
14861   (Boyd Lynn Gerber <gerberb@zenez.com>).
14862
14863   *Lutz Jaenicke*
14864
14865 * Improve RSA_padding_check_PKCS1_OAEP() check again to avoid
14866   'wristwatch attack' using huge encoding parameters (cf.
14867   James H. Manger's CRYPTO 2001 paper).  Note that the
14868   RSA_PKCS1_OAEP_PADDING case of RSA_private_decrypt() does not use
14869   encoding parameters and hence was not vulnerable.
14870
14871   *Bodo Moeller*
14872
14873 * BN_sqr() bug fix.
14874
14875   *Ulf Möller, reported by Jim Ellis <jim.ellis@cavium.com>*
14876
14877 * Rabin-Miller test analyses assume uniformly distributed witnesses,
14878   so use BN_pseudo_rand_range() instead of using BN_pseudo_rand()
14879   followed by modular reduction.
14880
14881   *Bodo Moeller; pointed out by Adam Young <AYoung1@NCSUS.JNJ.COM>*
14882
14883 * Add BN_pseudo_rand_range() with obvious functionality: BN_rand_range()
14884   equivalent based on BN_pseudo_rand() instead of BN_rand().
14885
14886   *Bodo Moeller*
14887
14888 * s3_srvr.c: allow sending of large client certificate lists (> 16 kB).
14889   This function was broken, as the check for a new client hello message
14890   to handle SGC did not allow these large messages.
14891   (Tracked down by "Douglas E. Engert" <deengert@anl.gov>.)
14892
14893   *Lutz Jaenicke*
14894
14895 * Add alert descriptions for TLSv1 to `SSL_alert_desc_string[_long]()`.
14896
14897   *Lutz Jaenicke*
14898
14899 * Fix buggy behaviour of BIO_get_num_renegotiates() and BIO_ctrl()
14900   for BIO_C_GET_WRITE_BUF_SIZE ("Stephen Hinton" <shinton@netopia.com>).
14901
14902   *Lutz Jaenicke*
14903
14904 * Rework the configuration and shared library support for Tru64 Unix.
14905   The configuration part makes use of modern compiler features and
14906   still retains old compiler behavior for those that run older versions
14907   of the OS.  The shared library support part includes a variant that
14908   uses the RPATH feature, and is available through the special
14909   configuration target "alpha-cc-rpath", which will never be selected
14910   automatically.
14911
14912   *Tim Mooney <mooney@dogbert.cc.ndsu.NoDak.edu> via Richard Levitte*
14913
14914 * In ssl3_get_key_exchange (ssl/s3_clnt.c), call ssl3_get_message()
14915   with the same message size as in ssl3_get_certificate_request().
14916   Otherwise, if no ServerKeyExchange message occurs, CertificateRequest
14917   messages might inadvertently be reject as too long.
14918
14919   *Petr Lampa <lampa@fee.vutbr.cz>*
14920
14921 * Enhanced support for IA-64 Unix platforms (well, Linux and HP-UX).
14922
14923   *Andy Polyakov*
14924
14925 * Modified SSL library such that the verify_callback that has been set
14926   specifically for an SSL object with SSL_set_verify() is actually being
14927   used. Before the change, a verify_callback set with this function was
14928   ignored and the verify_callback() set in the SSL_CTX at the time of
14929   the call was used. New function X509_STORE_CTX_set_verify_cb() introduced
14930   to allow the necessary settings.
14931
14932   *Lutz Jaenicke*
14933
14934 * Initialize static variable in crypto/dsa/dsa_lib.c and crypto/dh/dh_lib.c
14935   explicitly to NULL, as at least on Solaris 8 this seems not always to be
14936   done automatically (in contradiction to the requirements of the C
14937   standard). This made problems when used from OpenSSH.
14938
14939   *Lutz Jaenicke*
14940
14941 * In OpenSSL 0.9.6a and 0.9.6b, crypto/dh/dh_key.c ignored
14942   dh->length and always used
14943
14944           BN_rand_range(priv_key, dh->p).
14945
14946   BN_rand_range() is not necessary for Diffie-Hellman, and this
14947   specific range makes Diffie-Hellman unnecessarily inefficient if
14948   dh->length (recommended exponent length) is much smaller than the
14949   length of dh->p.  We could use BN_rand_range() if the order of
14950   the subgroup was stored in the DH structure, but we only have
14951   dh->length.
14952
14953   So switch back to
14954
14955           BN_rand(priv_key, l, ...)
14956
14957   where 'l' is dh->length if this is defined, or BN_num_bits(dh->p)-1
14958   otherwise.
14959
14960   *Bodo Moeller*
14961
14962 * In
14963
14964           RSA_eay_public_encrypt
14965           RSA_eay_private_decrypt
14966           RSA_eay_private_encrypt (signing)
14967           RSA_eay_public_decrypt (signature verification)
14968
14969   (default implementations for RSA_public_encrypt,
14970   RSA_private_decrypt, RSA_private_encrypt, RSA_public_decrypt),
14971   always reject numbers >= n.
14972
14973   *Bodo Moeller*
14974
14975 * In crypto/rand/md_rand.c, use a new short-time lock CRYPTO_LOCK_RAND2
14976   to synchronize access to 'locking_thread'.  This is necessary on
14977   systems where access to 'locking_thread' (an 'unsigned long'
14978   variable) is not atomic.
14979
14980   *Bodo Moeller*
14981
14982 * In crypto/rand/md_rand.c, set 'locking_thread' to current thread's ID
14983   *before* setting the 'crypto_lock_rand' flag.  The previous code had
14984   a race condition if 0 is a valid thread ID.
14985
14986   *Travis Vitek <vitek@roguewave.com>*
14987
14988 * Add support for shared libraries under Irix.
14989
14990   *Albert Chin-A-Young <china@thewrittenword.com>*
14991
14992 * Add configuration option to build on Linux on both big-endian and
14993   little-endian MIPS.
14994
14995   *Ralf Baechle <ralf@uni-koblenz.de>*
14996
14997 * Add the possibility to create shared libraries on HP-UX.
14998
14999   *Richard Levitte*
15000
15001### Changes between 0.9.6a and 0.9.6b  [9 Jul 2001]
15002
15003 * Change ssleay_rand_bytes (crypto/rand/md_rand.c)
15004   to avoid a SSLeay/OpenSSL PRNG weakness pointed out by
15005   Markku-Juhani O. Saarinen <markku-juhani.saarinen@nokia.com>:
15006   PRNG state recovery was possible based on the output of
15007   one PRNG request appropriately sized to gain knowledge on
15008   'md' followed by enough consecutive 1-byte PRNG requests
15009   to traverse all of 'state'.
15010
15011   1. When updating 'md_local' (the current thread's copy of 'md')
15012      during PRNG output generation, hash all of the previous
15013      'md_local' value, not just the half used for PRNG output.
15014
15015   2. Make the number of bytes from 'state' included into the hash
15016      independent from the number of PRNG bytes requested.
15017
15018   The first measure alone would be sufficient to avoid
15019   Markku-Juhani's attack.  (Actually it had never occurred
15020   to me that the half of 'md_local' used for chaining was the
15021   half from which PRNG output bytes were taken -- I had always
15022   assumed that the secret half would be used.)  The second
15023   measure makes sure that additional data from 'state' is never
15024   mixed into 'md_local' in small portions; this heuristically
15025   further strengthens the PRNG.
15026
15027   *Bodo Moeller*
15028
15029 * Fix crypto/bn/asm/mips3.s.
15030
15031   *Andy Polyakov*
15032
15033 * When only the key is given to "enc", the IV is undefined. Print out
15034   an error message in this case.
15035
15036   *Lutz Jaenicke*
15037
15038 * Handle special case when X509_NAME is empty in X509 printing routines.
15039
15040   *Steve Henson*
15041
15042 * In dsa_do_verify (crypto/dsa/dsa_ossl.c), verify that r and s are
15043   positive and less than q.
15044
15045   *Bodo Moeller*
15046
15047 * Don't change `*pointer` in CRYPTO_add_lock() is add_lock_callback is
15048   used: it isn't thread safe and the add_lock_callback should handle
15049   that itself.
15050
15051   *Paul Rose <Paul.Rose@bridge.com>*
15052
15053 * Verify that incoming data obeys the block size in
15054   ssl3_enc (ssl/s3_enc.c) and tls1_enc (ssl/t1_enc.c).
15055
15056   *Bodo Moeller*
15057
15058 * Fix OAEP check.
15059
15060   *Ulf Möller, Bodo Möller*
15061
15062 * The countermeasure against Bleichbacher's attack on PKCS #1 v1.5
15063   RSA encryption was accidentally removed in s3_srvr.c in OpenSSL 0.9.5
15064   when fixing the server behaviour for backwards-compatible 'client
15065   hello' messages.  (Note that the attack is impractical against
15066   SSL 3.0 and TLS 1.0 anyway because length and version checking
15067   means that the probability of guessing a valid ciphertext is
15068   around 2^-40; see section 5 in Bleichenbacher's CRYPTO '98
15069   paper.)
15070
15071   Before 0.9.5, the countermeasure (hide the error by generating a
15072   random 'decryption result') did not work properly because
15073   ERR_clear_error() was missing, meaning that SSL_get_error() would
15074   detect the supposedly ignored error.
15075
15076   Both problems are now fixed.
15077
15078   *Bodo Moeller*
15079
15080 * In crypto/bio/bf_buff.c, increase DEFAULT_BUFFER_SIZE to 4096
15081   (previously it was 1024).
15082
15083   *Bodo Moeller*
15084
15085 * Fix for compatibility mode trust settings: ignore trust settings
15086   unless some valid trust or reject settings are present.
15087
15088   *Steve Henson*
15089
15090 * Fix for blowfish EVP: its a variable length cipher.
15091
15092   *Steve Henson*
15093
15094 * Fix various bugs related to DSA S/MIME verification. Handle missing
15095   parameters in DSA public key structures and return an error in the
15096   DSA routines if parameters are absent.
15097
15098   *Steve Henson*
15099
15100 * In versions up to 0.9.6, RAND_file_name() resorted to file ".rnd"
15101   in the current directory if neither $RANDFILE nor $HOME was set.
15102   RAND_file_name() in 0.9.6a returned NULL in this case.  This has
15103   caused some confusion to Windows users who haven't defined $HOME.
15104   Thus RAND_file_name() is changed again: e_os.h can define a
15105   DEFAULT_HOME, which will be used if $HOME is not set.
15106   For Windows, we use "C:"; on other platforms, we still require
15107   environment variables.
15108
15109 * Move 'if (!initialized) RAND_poll()' into regions protected by
15110   CRYPTO_LOCK_RAND.  This is not strictly necessary, but avoids
15111   having multiple threads call RAND_poll() concurrently.
15112
15113   *Bodo Moeller*
15114
15115 * In crypto/rand/md_rand.c, replace 'add_do_not_lock' flag by a
15116   combination of a flag and a thread ID variable.
15117   Otherwise while one thread is in ssleay_rand_bytes (which sets the
15118   flag), *other* threads can enter ssleay_add_bytes without obeying
15119   the CRYPTO_LOCK_RAND lock (and may even illegally release the lock
15120   that they do not hold after the first thread unsets add_do_not_lock).
15121
15122   *Bodo Moeller*
15123
15124 * Change bctest again: '-x' expressions are not available in all
15125   versions of 'test'.
15126
15127   *Bodo Moeller*
15128
15129### Changes between 0.9.6 and 0.9.6a  [5 Apr 2001]
15130
15131 * Fix a couple of memory leaks in PKCS7_dataDecode()
15132
15133   *Steve Henson, reported by Heyun Zheng <hzheng@atdsprint.com>*
15134
15135 * Change Configure and Makefiles to provide EXE_EXT, which will contain
15136   the default extension for executables, if any.  Also, make the perl
15137   scripts that use symlink() to test if it really exists and use "cp"
15138   if it doesn't.  All this made OpenSSL compilable and installable in
15139   CygWin.
15140
15141   *Richard Levitte*
15142
15143 * Fix for asn1_GetSequence() for indefinite length constructed data.
15144   If SEQUENCE is length is indefinite just set c->slen to the total
15145   amount of data available.
15146
15147   *Steve Henson, reported by shige@FreeBSD.org*
15148
15149   *This change does not apply to 0.9.7.*
15150
15151 * Change bctest to avoid here-documents inside command substitution
15152   (workaround for FreeBSD /bin/sh bug).
15153   For compatibility with Ultrix, avoid shell functions (introduced
15154   in the bctest version that searches along $PATH).
15155
15156   *Bodo Moeller*
15157
15158 * Rename 'des_encrypt' to 'des_encrypt1'.  This avoids the clashes
15159   with des_encrypt() defined on some operating systems, like Solaris
15160   and UnixWare.
15161
15162   *Richard Levitte*
15163
15164 * Check the result of RSA-CRT (see D. Boneh, R. DeMillo, R. Lipton:
15165   On the Importance of Eliminating Errors in Cryptographic
15166   Computations, J. Cryptology 14 (2001) 2, 101-119,
15167   <http://theory.stanford.edu/~dabo/papers/faults.ps.gz>).
15168
15169   *Ulf Moeller*
15170
15171 * MIPS assembler BIGNUM division bug fix.
15172
15173   *Andy Polyakov*
15174
15175 * Disabled incorrect Alpha assembler code.
15176
15177   *Richard Levitte*
15178
15179 * Fix PKCS#7 decode routines so they correctly update the length
15180   after reading an EOC for the EXPLICIT tag.
15181
15182   *Steve Henson*
15183
15184   *This change does not apply to 0.9.7.*
15185
15186 * Fix bug in PKCS#12 key generation routines. This was triggered
15187   if a 3DES key was generated with a 0 initial byte. Include
15188   PKCS12_BROKEN_KEYGEN compilation option to retain the old
15189   (but broken) behaviour.
15190
15191   *Steve Henson*
15192
15193 * Enhance bctest to search for a working bc along $PATH and print
15194   it when found.
15195
15196   *Tim Rice <tim@multitalents.net> via Richard Levitte*
15197
15198 * Fix memory leaks in err.c: free err_data string if necessary;
15199   don't write to the wrong index in ERR_set_error_data.
15200
15201   *Bodo Moeller*
15202
15203 * Implement ssl23_peek (analogous to ssl23_read), which previously
15204   did not exist.
15205
15206   *Bodo Moeller*
15207
15208 * Replace rdtsc with `_emit` statements for VC++ version 5.
15209
15210   *Jeremy Cooper <jeremy@baymoo.org>*
15211
15212 * Make it possible to reuse SSLv2 sessions.
15213
15214   *Richard Levitte*
15215
15216 * In copy_email() check for >= 0 as a return value for
15217   X509_NAME_get_index_by_NID() since 0 is a valid index.
15218
15219   *Steve Henson reported by Massimiliano Pala <madwolf@opensca.org>*
15220
15221 * Avoid coredump with unsupported or invalid public keys by checking if
15222   X509_get_pubkey() fails in PKCS7_verify(). Fix memory leak when
15223   PKCS7_verify() fails with non detached data.
15224
15225   *Steve Henson*
15226
15227 * Don't use getenv in library functions when run as setuid/setgid.
15228   New function OPENSSL_issetugid().
15229
15230   *Ulf Moeller*
15231
15232 * Avoid false positives in memory leak detection code (crypto/mem_dbg.c)
15233   due to incorrect handling of multi-threading:
15234
15235   1. Fix timing glitch in the MemCheck_off() portion of CRYPTO_mem_ctrl().
15236
15237   2. Fix logical glitch in is_MemCheck_on() aka CRYPTO_is_mem_check_on().
15238
15239   3. Count how many times MemCheck_off() has been called so that
15240      nested use can be treated correctly.  This also avoids
15241      inband-signalling in the previous code (which relied on the
15242      assumption that thread ID 0 is impossible).
15243
15244   *Bodo Moeller*
15245
15246 * Add "-rand" option also to s_client and s_server.
15247
15248   *Lutz Jaenicke*
15249
15250 * Fix CPU detection on Irix 6.x.
15251   *Kurt Hockenbury <khockenb@stevens-tech.edu> and
15252   "Bruce W. Forsberg" <bruce.forsberg@baesystems.com>*
15253
15254 * Fix X509_NAME bug which produced incorrect encoding if X509_NAME
15255   was empty.
15256
15257   *Steve Henson*
15258
15259   *This change does not apply to 0.9.7.*
15260
15261 * Use the cached encoding of an X509_NAME structure rather than
15262   copying it. This is apparently the reason for the libsafe "errors"
15263   but the code is actually correct.
15264
15265   *Steve Henson*
15266
15267 * Add new function BN_rand_range(), and fix DSA_sign_setup() to prevent
15268   Bleichenbacher's DSA attack.
15269   Extend BN_[pseudo_]rand: As before, top=1 forces the highest two bits
15270   to be set and top=0 forces the highest bit to be set; top=-1 is new
15271   and leaves the highest bit random.
15272
15273   *Ulf Moeller, Bodo Moeller*
15274
15275 * In the `NCONF_...`-based implementations for `CONF_...` queries
15276   (crypto/conf/conf_lib.c), if the input LHASH is NULL, avoid using
15277   a temporary CONF structure with the data component set to NULL
15278   (which gives segmentation faults in lh_retrieve).
15279   Instead, use NULL for the CONF pointer in CONF_get_string and
15280   CONF_get_number (which may use environment variables) and directly
15281   return NULL from CONF_get_section.
15282
15283   *Bodo Moeller*
15284
15285 * Fix potential buffer overrun for EBCDIC.
15286
15287   *Ulf Moeller*
15288
15289 * Tolerate nonRepudiation as being valid for S/MIME signing and certSign
15290   keyUsage if basicConstraints absent for a CA.
15291
15292   *Steve Henson*
15293
15294 * Make SMIME_write_PKCS7() write mail header values with a format that
15295   is more generally accepted (no spaces before the semicolon), since
15296   some programs can't parse those values properly otherwise.  Also make
15297   sure BIO's that break lines after each write do not create invalid
15298   headers.
15299
15300   *Richard Levitte*
15301
15302 * Make the CRL encoding routines work with empty SEQUENCE OF. The
15303   macros previously used would not encode an empty SEQUENCE OF
15304   and break the signature.
15305
15306   *Steve Henson*
15307
15308   *This change does not apply to 0.9.7.*
15309
15310 * Zero the premaster secret after deriving the master secret in
15311   DH ciphersuites.
15312
15313   *Steve Henson*
15314
15315 * Add some EVP_add_digest_alias registrations (as found in
15316   OpenSSL_add_all_digests()) to SSL_library_init()
15317   aka OpenSSL_add_ssl_algorithms().  This provides improved
15318   compatibility with peers using X.509 certificates
15319   with unconventional AlgorithmIdentifier OIDs.
15320
15321   *Bodo Moeller*
15322
15323 * Fix for Irix with NO_ASM.
15324
15325   *"Bruce W. Forsberg" <bruce.forsberg@baesystems.com>*
15326
15327 * ./config script fixes.
15328
15329   *Ulf Moeller, Richard Levitte*
15330
15331 * Fix 'openssl passwd -1'.
15332
15333   *Bodo Moeller*
15334
15335 * Change PKCS12_key_gen_asc() so it can cope with non null
15336   terminated strings whose length is passed in the passlen
15337   parameter, for example from PEM callbacks. This was done
15338   by adding an extra length parameter to asc2uni().
15339
15340   *Steve Henson, reported by <oddissey@samsung.co.kr>*
15341
15342 * Fix C code generated by 'openssl dsaparam -C': If a BN_bin2bn
15343   call failed, free the DSA structure.
15344
15345   *Bodo Moeller*
15346
15347 * Fix to uni2asc() to cope with zero length Unicode strings.
15348   These are present in some PKCS#12 files.
15349
15350   *Steve Henson*
15351
15352 * Increase s2->wbuf allocation by one byte in ssl2_new (ssl/s2_lib.c).
15353   Otherwise do_ssl_write (ssl/s2_pkt.c) will write beyond buffer limits
15354   when writing a 32767 byte record.
15355
15356   *Bodo Moeller; problem reported by Eric Day <eday@concentric.net>*
15357
15358 * In `RSA_eay_public_{en,ed}crypt` and RSA_eay_mod_exp (rsa_eay.c),
15359   obtain lock CRYPTO_LOCK_RSA before setting `rsa->_method_mod_{n,p,q}`.
15360
15361   (RSA objects have a reference count access to which is protected
15362   by CRYPTO_LOCK_RSA [see rsa_lib.c, s3_srvr.c, ssl_cert.c, ssl_rsa.c],
15363   so they are meant to be shared between threads.)
15364   *Bodo Moeller, Geoff Thorpe; original patch submitted by
15365   "Reddie, Steven" <Steven.Reddie@ca.com>*
15366
15367 * Fix a deadlock in CRYPTO_mem_leaks().
15368
15369   *Bodo Moeller*
15370
15371 * Use better test patterns in bntest.
15372
15373   *Ulf Möller*
15374
15375 * rand_win.c fix for Borland C.
15376
15377   *Ulf Möller*
15378
15379 * BN_rshift bugfix for n == 0.
15380
15381   *Bodo Moeller*
15382
15383 * Add a 'bctest' script that checks for some known 'bc' bugs
15384   so that 'make test' does not abort just because 'bc' is broken.
15385
15386   *Bodo Moeller*
15387
15388 * Store verify_result within SSL_SESSION also for client side to
15389   avoid potential security hole. (Re-used sessions on the client side
15390   always resulted in verify_result==X509_V_OK, not using the original
15391   result of the server certificate verification.)
15392
15393   *Lutz Jaenicke*
15394
15395 * Fix ssl3_pending: If the record in s->s3->rrec is not of type
15396   SSL3_RT_APPLICATION_DATA, return 0.
15397   Similarly, change ssl2_pending to return 0 if SSL_in_init(s) is true.
15398
15399   *Bodo Moeller*
15400
15401 * Fix SSL_peek:
15402   Both ssl2_peek and ssl3_peek, which were totally broken in earlier
15403   releases, have been re-implemented by renaming the previous
15404   implementations of ssl2_read and ssl3_read to ssl2_read_internal
15405   and ssl3_read_internal, respectively, and adding 'peek' parameters
15406   to them.  The new ssl[23]_{read,peek} functions are calls to
15407   ssl[23]_read_internal with the 'peek' flag set appropriately.
15408   A 'peek' parameter has also been added to ssl3_read_bytes, which
15409   does the actual work for ssl3_read_internal.
15410
15411   *Bodo Moeller*
15412
15413 * Initialise "ex_data" member of RSA/DSA/DH structures prior to calling
15414   the method-specific "init()" handler. Also clean up ex_data after
15415   calling the method-specific "finish()" handler. Previously, this was
15416   happening the other way round.
15417
15418   *Geoff Thorpe*
15419
15420 * Increase BN_CTX_NUM (the number of BIGNUMs in a BN_CTX) to 16.
15421   The previous value, 12, was not always sufficient for BN_mod_exp().
15422
15423   *Bodo Moeller*
15424
15425 * Make sure that shared libraries get the internal name engine with
15426   the full version number and not just 0.  This should mark the
15427   shared libraries as not backward compatible.  Of course, this should
15428   be changed again when we can guarantee backward binary compatibility.
15429
15430   *Richard Levitte*
15431
15432 * Fix typo in get_cert_by_subject() in by_dir.c
15433
15434   *Jean-Marc Desperrier <jean-marc.desperrier@certplus.com>*
15435
15436 * Rework the system to generate shared libraries:
15437
15438   - Make note of the expected extension for the shared libraries and
15439     if there is a need for symbolic links from for example libcrypto.so.0
15440     to libcrypto.so.0.9.7.  There is extended info in Configure for
15441     that.
15442
15443   - Make as few rebuilds of the shared libraries as possible.
15444
15445   - Still avoid linking the OpenSSL programs with the shared libraries.
15446
15447   - When installing, install the shared libraries separately from the
15448     static ones.
15449
15450   *Richard Levitte*
15451
15452 * Fix SSL_CTX_set_read_ahead macro to actually use its argument.
15453
15454   Copy SSL_CTX's read_ahead flag to SSL object directly in SSL_new
15455   and not in SSL_clear because the latter is also used by the
15456   accept/connect functions; previously, the settings made by
15457   SSL_set_read_ahead would be lost during the handshake.
15458
15459   *Bodo Moeller; problems reported by Anders Gertz <gertz@epact.se>*
15460
15461 * Correct util/mkdef.pl to be selective about disabled algorithms.
15462   Previously, it would create entries for disabled algorithms no
15463   matter what.
15464
15465   *Richard Levitte*
15466
15467 * Added several new manual pages for SSL_* function.
15468
15469   *Lutz Jaenicke*
15470
15471### Changes between 0.9.5a and 0.9.6  [24 Sep 2000]
15472
15473 * In ssl23_get_client_hello, generate an error message when faced
15474   with an initial SSL 3.0/TLS record that is too small to contain the
15475   first two bytes of the ClientHello message, i.e. client_version.
15476   (Note that this is a pathologic case that probably has never happened
15477   in real life.)  The previous approach was to use the version number
15478   from the record header as a substitute; but our protocol choice
15479   should not depend on that one because it is not authenticated
15480   by the Finished messages.
15481
15482   *Bodo Moeller*
15483
15484 * More robust randomness gathering functions for Windows.
15485
15486   *Jeffrey Altman <jaltman@columbia.edu>*
15487
15488 * For compatibility reasons if the flag X509_V_FLAG_ISSUER_CHECK is
15489   not set then we don't setup the error code for issuer check errors
15490   to avoid possibly overwriting other errors which the callback does
15491   handle. If an application does set the flag then we assume it knows
15492   what it is doing and can handle the new informational codes
15493   appropriately.
15494
15495   *Steve Henson*
15496
15497 * Fix for a nasty bug in ASN1_TYPE handling. ASN1_TYPE is used for
15498   a general "ANY" type, as such it should be able to decode anything
15499   including tagged types. However it didn't check the class so it would
15500   wrongly interpret tagged types in the same way as their universal
15501   counterpart and unknown types were just rejected. Changed so that the
15502   tagged and unknown types are handled in the same way as a SEQUENCE:
15503   that is the encoding is stored intact. There is also a new type
15504   "V_ASN1_OTHER" which is used when the class is not universal, in this
15505   case we have no idea what the actual type is so we just lump them all
15506   together.
15507
15508   *Steve Henson*
15509
15510 * On VMS, stdout may very well lead to a file that is written to
15511   in a record-oriented fashion.  That means that every write() will
15512   write a separate record, which will be read separately by the
15513   programs trying to read from it.  This can be very confusing.
15514
15515   The solution is to put a BIO filter in the way that will buffer
15516   text until a linefeed is reached, and then write everything a
15517   line at a time, so every record written will be an actual line,
15518   not chunks of lines and not (usually doesn't happen, but I've
15519   seen it once) several lines in one record.  BIO_f_linebuffer() is
15520   the answer.
15521
15522   Currently, it's a VMS-only method, because that's where it has
15523   been tested well enough.
15524
15525   *Richard Levitte*
15526
15527 * Remove 'optimized' squaring variant in BN_mod_mul_montgomery,
15528   it can return incorrect results.
15529   (Note: The buggy variant was not enabled in OpenSSL 0.9.5a,
15530   but it was in 0.9.6-beta[12].)
15531
15532   *Bodo Moeller*
15533
15534 * Disable the check for content being present when verifying detached
15535   signatures in pk7_smime.c. Some versions of Netscape (wrongly)
15536   include zero length content when signing messages.
15537
15538   *Steve Henson*
15539
15540 * New BIO_shutdown_wr macro, which invokes the BIO_C_SHUTDOWN_WR
15541   BIO_ctrl (for BIO pairs).
15542
15543   *Bodo Möller*
15544
15545 * Add DSO method for VMS.
15546
15547   *Richard Levitte*
15548
15549 * Bug fix: Montgomery multiplication could produce results with the
15550   wrong sign.
15551
15552   *Ulf Möller*
15553
15554 * Add RPM specification openssl.spec and modify it to build three
15555   packages.  The default package contains applications, application
15556   documentation and run-time libraries.  The devel package contains
15557   include files, static libraries and function documentation.  The
15558   doc package contains the contents of the doc directory.  The original
15559   openssl.spec was provided by Damien Miller <djm@mindrot.org>.
15560
15561   *Richard Levitte*
15562
15563 * Add a large number of documentation files for many SSL routines.
15564
15565   *Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>*
15566
15567 * Add a configuration entry for Sony News 4.
15568
15569   *NAKAJI Hiroyuki <nakaji@tutrp.tut.ac.jp>*
15570
15571 * Don't set the two most significant bits to one when generating a
15572   random number < q in the DSA library.
15573
15574   *Ulf Möller*
15575
15576 * New SSL API mode 'SSL_MODE_AUTO_RETRY'.  This disables the default
15577   behaviour that SSL_read may result in SSL_ERROR_WANT_READ (even if
15578   the underlying transport is blocking) if a handshake took place.
15579   (The default behaviour is needed by applications such as s_client
15580   and s_server that use select() to determine when to use SSL_read;
15581   but for applications that know in advance when to expect data, it
15582   just makes things more complicated.)
15583
15584   *Bodo Moeller*
15585
15586 * Add RAND_egd_bytes(), which gives control over the number of bytes read
15587   from EGD.
15588
15589   *Ben Laurie*
15590
15591 * Add a few more EBCDIC conditionals that make `req` and `x509`
15592   work better on such systems.
15593
15594   *Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>*
15595
15596 * Add two demo programs for PKCS12_parse() and PKCS12_create().
15597   Update PKCS12_parse() so it copies the friendlyName and the
15598   keyid to the certificates aux info.
15599
15600   *Steve Henson*
15601
15602 * Fix bug in PKCS7_verify() which caused an infinite loop
15603   if there was more than one signature.
15604
15605   *Sven Uszpelkat <su@celocom.de>*
15606
15607 * Major change in util/mkdef.pl to include extra information
15608   about each symbol, as well as presenting variables as well
15609   as functions.  This change means that there's n more need
15610   to rebuild the .num files when some algorithms are excluded.
15611
15612   *Richard Levitte*
15613
15614 * Allow the verify time to be set by an application,
15615   rather than always using the current time.
15616
15617   *Steve Henson*
15618
15619 * Phase 2 verify code reorganisation. The certificate
15620   verify code now looks up an issuer certificate by a
15621   number of criteria: subject name, authority key id
15622   and key usage. It also verifies self signed certificates
15623   by the same criteria. The main comparison function is
15624   X509_check_issued() which performs these checks.
15625
15626   Lot of changes were necessary in order to support this
15627   without completely rewriting the lookup code.
15628
15629   Authority and subject key identifier are now cached.
15630
15631   The LHASH 'certs' is X509_STORE has now been replaced
15632   by a STACK_OF(X509_OBJECT). This is mainly because an
15633   LHASH can't store or retrieve multiple objects with
15634   the same hash value.
15635
15636   As a result various functions (which were all internal
15637   use only) have changed to handle the new X509_STORE
15638   structure. This will break anything that messed round
15639   with X509_STORE internally.
15640
15641   The functions X509_STORE_add_cert() now checks for an
15642   exact match, rather than just subject name.
15643
15644   The X509_STORE API doesn't directly support the retrieval
15645   of multiple certificates matching a given criteria, however
15646   this can be worked round by performing a lookup first
15647   (which will fill the cache with candidate certificates)
15648   and then examining the cache for matches. This is probably
15649   the best we can do without throwing out X509_LOOKUP
15650   entirely (maybe later...).
15651
15652   The X509_VERIFY_CTX structure has been enhanced considerably.
15653
15654   All certificate lookup operations now go via a get_issuer()
15655   callback. Although this currently uses an X509_STORE it
15656   can be replaced by custom lookups. This is a simple way
15657   to bypass the X509_STORE hackery necessary to make this
15658   work and makes it possible to use more efficient techniques
15659   in future. A very simple version which uses a simple
15660   STACK for its trusted certificate store is also provided
15661   using X509_STORE_CTX_trusted_stack().
15662
15663   The verify_cb() and verify() callbacks now have equivalents
15664   in the X509_STORE_CTX structure.
15665
15666   X509_STORE_CTX also has a 'flags' field which can be used
15667   to customise the verify behaviour.
15668
15669   *Steve Henson*
15670
15671 * Add new PKCS#7 signing option PKCS7_NOSMIMECAP which
15672   excludes S/MIME capabilities.
15673
15674   *Steve Henson*
15675
15676 * When a certificate request is read in keep a copy of the
15677   original encoding of the signed data and use it when outputting
15678   again. Signatures then use the original encoding rather than
15679   a decoded, encoded version which may cause problems if the
15680   request is improperly encoded.
15681
15682   *Steve Henson*
15683
15684 * For consistency with other BIO_puts implementations, call
15685   buffer_write(b, ...) directly in buffer_puts instead of calling
15686   BIO_write(b, ...).
15687
15688   In BIO_puts, increment b->num_write as in BIO_write.
15689
15690   *Peter.Sylvester@EdelWeb.fr*
15691
15692 * Fix BN_mul_word for the case where the word is 0. (We have to use
15693   BN_zero, we may not return a BIGNUM with an array consisting of
15694   words set to zero.)
15695
15696   *Bodo Moeller*
15697
15698 * Avoid calling abort() from within the library when problems are
15699   detected, except if preprocessor symbols have been defined
15700   (such as REF_CHECK, BN_DEBUG etc.).
15701
15702   *Bodo Moeller*
15703
15704 * New openssl application 'rsautl'. This utility can be
15705   used for low-level RSA operations. DER public key
15706   BIO/fp routines also added.
15707
15708   *Steve Henson*
15709
15710 * New Configure entry and patches for compiling on QNX 4.
15711
15712   *Andreas Schneider <andreas@ds3.etech.fh-hamburg.de>*
15713
15714 * A demo state-machine implementation was sponsored by
15715   Nuron (<http://www.nuron.com/>) and is now available in
15716   demos/state_machine.
15717
15718   *Ben Laurie*
15719
15720 * New options added to the 'dgst' utility for signature
15721   generation and verification.
15722
15723   *Steve Henson*
15724
15725 * Unrecognized PKCS#7 content types are now handled via a
15726   catch all ASN1_TYPE structure. This allows unsupported
15727   types to be stored as a "blob" and an application can
15728   encode and decode it manually.
15729
15730   *Steve Henson*
15731
15732 * Fix various signed/unsigned issues to make a_strex.c
15733   compile under VC++.
15734
15735   *Oscar Jacobsson <oscar.jacobsson@celocom.com>*
15736
15737 * ASN1 fixes. i2d_ASN1_OBJECT was not returning the correct
15738   length if passed a buffer. ASN1_INTEGER_to_BN failed
15739   if passed a NULL BN and its argument was negative.
15740
15741   *Steve Henson, pointed out by Sven Heiberg <sven@tartu.cyber.ee>*
15742
15743 * Modification to PKCS#7 encoding routines to output definite
15744   length encoding. Since currently the whole structures are in
15745   memory there's not real point in using indefinite length
15746   constructed encoding. However if OpenSSL is compiled with
15747   the flag PKCS7_INDEFINITE_ENCODING the old form is used.
15748
15749   *Steve Henson*
15750
15751 * Added BIO_vprintf() and BIO_vsnprintf().
15752
15753   *Richard Levitte*
15754
15755 * Added more prefixes to parse for in the strings written
15756   through a logging bio, to cover all the levels that are available
15757   through syslog.  The prefixes are now:
15758
15759           PANIC, EMERG, EMR       =>      LOG_EMERG
15760           ALERT, ALR              =>      LOG_ALERT
15761           CRIT, CRI               =>      LOG_CRIT
15762           ERROR, ERR              =>      LOG_ERR
15763           WARNING, WARN, WAR      =>      LOG_WARNING
15764           NOTICE, NOTE, NOT       =>      LOG_NOTICE
15765           INFO, INF               =>      LOG_INFO
15766           DEBUG, DBG              =>      LOG_DEBUG
15767
15768   and as before, if none of those prefixes are present at the
15769   beginning of the string, LOG_ERR is chosen.
15770
15771   On Win32, the `LOG_*` levels are mapped according to this:
15772
15773           LOG_EMERG, LOG_ALERT, LOG_CRIT, LOG_ERR => EVENTLOG_ERROR_TYPE
15774           LOG_WARNING                             => EVENTLOG_WARNING_TYPE
15775           LOG_NOTICE, LOG_INFO, LOG_DEBUG         => EVENTLOG_INFORMATION_TYPE
15776
15777   *Richard Levitte*
15778
15779 * Made it possible to reconfigure with just the configuration
15780   argument "reconf" or "reconfigure".  The command line arguments
15781   are stored in Makefile.ssl in the variable CONFIGURE_ARGS,
15782   and are retrieved from there when reconfiguring.
15783
15784   *Richard Levitte*
15785
15786 * MD4 implemented.
15787
15788   *Assar Westerlund <assar@sics.se>, Richard Levitte*
15789
15790 * Add the arguments -CAfile and -CApath to the pkcs12 utility.
15791
15792   *Richard Levitte*
15793
15794 * The obj_dat.pl script was messing up the sorting of object
15795   names. The reason was that it compared the quoted version
15796   of strings as a result "OCSP" > "OCSP Signing" because
15797   " > SPACE. Changed script to store unquoted versions of
15798   names and add quotes on output. It was also omitting some
15799   names from the lookup table if they were given a default
15800   value (that is if SN is missing it is given the same
15801   value as LN and vice versa), these are now added on the
15802   grounds that if an object has a name we should be able to
15803   look it up. Finally added warning output when duplicate
15804   short or long names are found.
15805
15806   *Steve Henson*
15807
15808 * Changes needed for Tandem NSK.
15809
15810   *Scott Uroff <scott@xypro.com>*
15811
15812 * Fix SSL 2.0 rollback checking: Due to an off-by-one error in
15813   RSA_padding_check_SSLv23(), special padding was never detected
15814   and thus the SSL 3.0/TLS 1.0 countermeasure against protocol
15815   version rollback attacks was not effective.
15816
15817   In s23_clnt.c, don't use special rollback-attack detection padding
15818   (RSA_SSLV23_PADDING) if SSL 2.0 is the only protocol enabled in the
15819   client; similarly, in s23_srvr.c, don't do the rollback check if
15820   SSL 2.0 is the only protocol enabled in the server.
15821
15822   *Bodo Moeller*
15823
15824 * Make it possible to get hexdumps of unprintable data with 'openssl
15825   asn1parse'.  By implication, the functions ASN1_parse_dump() and
15826   BIO_dump_indent() are added.
15827
15828   *Richard Levitte*
15829
15830 * New functions ASN1_STRING_print_ex() and X509_NAME_print_ex()
15831   these print out strings and name structures based on various
15832   flags including RFC2253 support and proper handling of
15833   multibyte characters. Added options to the 'x509' utility
15834   to allow the various flags to be set.
15835
15836   *Steve Henson*
15837
15838 * Various fixes to use ASN1_TIME instead of ASN1_UTCTIME.
15839   Also change the functions X509_cmp_current_time() and
15840   X509_gmtime_adj() work with an ASN1_TIME structure,
15841   this will enable certificates using GeneralizedTime in validity
15842   dates to be checked.
15843
15844   *Steve Henson*
15845
15846 * Make the NEG_PUBKEY_BUG code (which tolerates invalid
15847   negative public key encodings) on by default,
15848   NO_NEG_PUBKEY_BUG can be set to disable it.
15849
15850   *Steve Henson*
15851
15852 * New function c2i_ASN1_OBJECT() which acts on ASN1_OBJECT
15853   content octets. An i2c_ASN1_OBJECT is unnecessary because
15854   the encoding can be trivially obtained from the structure.
15855
15856   *Steve Henson*
15857
15858 * crypto/err.c locking bugfix: Use write locks (`CRYPTO_w_[un]lock`),
15859   not read locks (`CRYPTO_r_[un]lock`).
15860
15861   *Bodo Moeller*
15862
15863 * A first attempt at creating official support for shared
15864   libraries through configuration.  I've kept it so the
15865   default is static libraries only, and the OpenSSL programs
15866   are always statically linked for now, but there are
15867   preparations for dynamic linking in place.
15868   This has been tested on Linux and Tru64.
15869
15870   *Richard Levitte*
15871
15872 * Randomness polling function for Win9x, as described in:
15873   Peter Gutmann, Software Generation of Practically Strong
15874   Random Numbers.
15875
15876   *Ulf Möller*
15877
15878 * Fix so PRNG is seeded in req if using an already existing
15879   DSA key.
15880
15881   *Steve Henson*
15882
15883 * New options to smime application. -inform and -outform
15884   allow alternative formats for the S/MIME message including
15885   PEM and DER. The -content option allows the content to be
15886   specified separately. This should allow things like Netscape
15887   form signing output easier to verify.
15888
15889   *Steve Henson*
15890
15891 * Fix the ASN1 encoding of tags using the 'long form'.
15892
15893   *Steve Henson*
15894
15895 * New ASN1 functions, `i2c_*` and `c2i_*` for INTEGER and BIT
15896   STRING types. These convert content octets to and from the
15897   underlying type. The actual tag and length octets are
15898   already assumed to have been read in and checked. These
15899   are needed because all other string types have virtually
15900   identical handling apart from the tag. By having versions
15901   of the ASN1 functions that just operate on content octets
15902   IMPLICIT tagging can be handled properly. It also allows
15903   the ASN1_ENUMERATED code to be cut down because ASN1_ENUMERATED
15904   and ASN1_INTEGER are identical apart from the tag.
15905
15906   *Steve Henson*
15907
15908 * Change the handling of OID objects as follows:
15909
15910   - New object identifiers are inserted in objects.txt, following
15911     the syntax given in [crypto/objects/README.md](crypto/objects/README.md).
15912   - objects.pl is used to process obj_mac.num and create a new
15913     obj_mac.h.
15914   - obj_dat.pl is used to create a new obj_dat.h, using the data in
15915     obj_mac.h.
15916
15917   This is currently kind of a hack, and the perl code in objects.pl
15918   isn't very elegant, but it works as I intended.  The simplest way
15919   to check that it worked correctly is to look in obj_dat.h and
15920   check the array nid_objs and make sure the objects haven't moved
15921   around (this is important!).  Additions are OK, as well as
15922   consistent name changes.
15923
15924   *Richard Levitte*
15925
15926 * Add BSD-style MD5-based passwords to 'openssl passwd' (option '-1').
15927
15928   *Bodo Moeller*
15929
15930 * Addition of the command line parameter '-rand file' to 'openssl req'.
15931   The given file adds to whatever has already been seeded into the
15932   random pool through the RANDFILE configuration file option or
15933   environment variable, or the default random state file.
15934
15935   *Richard Levitte*
15936
15937 * mkstack.pl now sorts each macro group into lexical order.
15938   Previously the output order depended on the order the files
15939   appeared in the directory, resulting in needless rewriting
15940   of safestack.h .
15941
15942   *Steve Henson*
15943
15944 * Patches to make OpenSSL compile under Win32 again. Mostly
15945   work arounds for the VC++ problem that it treats func() as
15946   func(void). Also stripped out the parts of mkdef.pl that
15947   added extra typesafe functions: these no longer exist.
15948
15949   *Steve Henson*
15950
15951 * Reorganisation of the stack code. The macros are now all
15952   collected in safestack.h . Each macro is defined in terms of
15953   a "stack macro" of the form `SKM_<name>(type, a, b)`. The
15954   DEBUG_SAFESTACK is now handled in terms of function casts,
15955   this has the advantage of retaining type safety without the
15956   use of additional functions. If DEBUG_SAFESTACK is not defined
15957   then the non typesafe macros are used instead. Also modified the
15958   mkstack.pl script to handle the new form. Needs testing to see
15959   if which (if any) compilers it chokes and maybe make DEBUG_SAFESTACK
15960   the default if no major problems. Similar behaviour for ASN1_SET_OF
15961   and PKCS12_STACK_OF.
15962
15963   *Steve Henson*
15964
15965 * When some versions of IIS use the 'NET' form of private key the
15966   key derivation algorithm is different. Normally MD5(password) is
15967   used as a 128 bit RC4 key. In the modified case
15968   MD5(MD5(password) + "SGCKEYSALT")  is used instead. Added some
15969   new functions i2d_RSA_NET(), d2i_RSA_NET() etc which are the same
15970   as the old Netscape_RSA functions except they have an additional
15971   'sgckey' parameter which uses the modified algorithm. Also added
15972   an -sgckey command line option to the rsa utility. Thanks to
15973   Adrian Peck <bertie@ncipher.com> for posting details of the modified
15974   algorithm to openssl-dev.
15975
15976   *Steve Henson*
15977
15978 * The evp_local.h macros were using 'c.##kname' which resulted in
15979   invalid expansion on some systems (SCO 5.0.5 for example).
15980   Corrected to 'c.kname'.
15981
15982   *Phillip Porch <root@theporch.com>*
15983
15984 * New X509_get1_email() and X509_REQ_get1_email() functions that return
15985   a STACK of email addresses from a certificate or request, these look
15986   in the subject name and the subject alternative name extensions and
15987   omit any duplicate addresses.
15988
15989   *Steve Henson*
15990
15991 * Re-implement BN_mod_exp2_mont using independent (and larger) windows.
15992   This makes DSA verification about 2 % faster.
15993
15994   *Bodo Moeller*
15995
15996 * Increase maximum window size in `BN_mod_exp_...` to 6 bits instead of 5
15997   (meaning that now 2^5 values will be precomputed, which is only 4 KB
15998   plus overhead for 1024 bit moduli).
15999   This makes exponentiations about 0.5 % faster for 1024 bit
16000   exponents (as measured by "openssl speed rsa2048").
16001
16002   *Bodo Moeller*
16003
16004 * Rename memory handling macros to avoid conflicts with other
16005   software:
16006           Malloc         =>  OPENSSL_malloc
16007           Malloc_locked  =>  OPENSSL_malloc_locked
16008           Realloc        =>  OPENSSL_realloc
16009           Free           =>  OPENSSL_free
16010
16011   *Richard Levitte*
16012
16013 * New function BN_mod_exp_mont_word for small bases (roughly 15%
16014   faster than BN_mod_exp_mont, i.e. 7% for a full DH exchange).
16015
16016   *Bodo Moeller*
16017
16018 * CygWin32 support.
16019
16020   *John Jarvie <jjarvie@newsguy.com>*
16021
16022 * The type-safe stack code has been rejigged. It is now only compiled
16023   in when OpenSSL is configured with the DEBUG_SAFESTACK option and
16024   by default all type-specific stack functions are "#define"d back to
16025   standard stack functions. This results in more streamlined output
16026   but retains the type-safety checking possibilities of the original
16027   approach.
16028
16029   *Geoff Thorpe*
16030
16031 * The STACK code has been cleaned up, and certain type declarations
16032   that didn't make a lot of sense have been brought in line. This has
16033   also involved a cleanup of sorts in safestack.h to more correctly
16034   map type-safe stack functions onto their plain stack counterparts.
16035   This work has also resulted in a variety of "const"ifications of
16036   lots of the code, especially `_cmp` operations which should normally
16037   be prototyped with "const" parameters anyway.
16038
16039   *Geoff Thorpe*
16040
16041 * When generating bytes for the first time in md_rand.c, 'stir the pool'
16042   by seeding with STATE_SIZE dummy bytes (with zero entropy count).
16043   (The PRNG state consists of two parts, the large pool 'state' and 'md',
16044   where all of 'md' is used each time the PRNG is used, but 'state'
16045   is used only indexed by a cyclic counter. As entropy may not be
16046   well distributed from the beginning, 'md' is important as a
16047   chaining variable. However, the output function chains only half
16048   of 'md', i.e. 80 bits.  ssleay_rand_add, on the other hand, chains
16049   all of 'md', and seeding with STATE_SIZE dummy bytes will result
16050   in all of 'state' being rewritten, with the new values depending
16051   on virtually all of 'md'.  This overcomes the 80 bit limitation.)
16052
16053   *Bodo Moeller*
16054
16055 * In ssl/s2_clnt.c and ssl/s3_clnt.c, call ERR_clear_error() when
16056   the handshake is continued after ssl_verify_cert_chain();
16057   otherwise, if SSL_VERIFY_NONE is set, remaining error codes
16058   can lead to 'unexplainable' connection aborts later.
16059
16060   *Bodo Moeller; problem tracked down by Lutz Jaenicke*
16061
16062 * Major EVP API cipher revision.
16063   Add hooks for extra EVP features. This allows various cipher
16064   parameters to be set in the EVP interface. Support added for variable
16065   key length ciphers via the EVP_CIPHER_CTX_set_key_length() function and
16066   setting of RC2 and RC5 parameters.
16067
16068   Modify EVP_OpenInit() and EVP_SealInit() to cope with variable key length
16069   ciphers.
16070
16071   Remove lots of duplicated code from the EVP library. For example *every*
16072   cipher init() function handles the 'iv' in the same way according to the
16073   cipher mode. They also all do nothing if the 'key' parameter is NULL and
16074   for CFB and OFB modes they zero ctx->num.
16075
16076   New functionality allows removal of S/MIME code RC2 hack.
16077
16078   Most of the routines have the same form and so can be declared in terms
16079   of macros.
16080
16081   By shifting this to the top level EVP_CipherInit() it can be removed from
16082   all individual ciphers. If the cipher wants to handle IVs or keys
16083   differently it can set the EVP_CIPH_CUSTOM_IV or EVP_CIPH_ALWAYS_CALL_INIT
16084   flags.
16085
16086   Change lots of functions like EVP_EncryptUpdate() to now return a
16087   value: although software versions of the algorithms cannot fail
16088   any installed hardware versions can.
16089
16090   *Steve Henson*
16091
16092 * Implement SSL_OP_TLS_ROLLBACK_BUG: In ssl3_get_client_key_exchange, if
16093   this option is set, tolerate broken clients that send the negotiated
16094   protocol version number instead of the requested protocol version
16095   number.
16096
16097   *Bodo Moeller*
16098
16099 * Call dh_tmp_cb (set by `..._TMP_DH_CB`) with correct 'is_export' flag;
16100   i.e. non-zero for export ciphersuites, zero otherwise.
16101   Previous versions had this flag inverted, inconsistent with
16102   rsa_tmp_cb (..._TMP_RSA_CB).
16103
16104   *Bodo Moeller; problem reported by Amit Chopra*
16105
16106 * Add missing DSA library text string. Work around for some IIS
16107   key files with invalid SEQUENCE encoding.
16108
16109   *Steve Henson*
16110
16111 * Add a document (doc/standards.txt) that list all kinds of standards
16112   and so on that are implemented in OpenSSL.
16113
16114   *Richard Levitte*
16115
16116 * Enhance c_rehash script. Old version would mishandle certificates
16117   with the same subject name hash and wouldn't handle CRLs at all.
16118   Added -fingerprint option to crl utility, to support new c_rehash
16119   features.
16120
16121   *Steve Henson*
16122
16123 * Eliminate non-ANSI declarations in crypto.h and stack.h.
16124
16125   *Ulf Möller*
16126
16127 * Fix for SSL server purpose checking. Server checking was
16128   rejecting certificates which had extended key usage present
16129   but no ssl client purpose.
16130
16131   *Steve Henson, reported by Rene Grosser <grosser@hisolutions.com>*
16132
16133 * Make PKCS#12 code work with no password. The PKCS#12 spec
16134   is a little unclear about how a blank password is handled.
16135   Since the password in encoded as a BMPString with terminating
16136   double NULL a zero length password would end up as just the
16137   double NULL. However no password at all is different and is
16138   handled differently in the PKCS#12 key generation code. NS
16139   treats a blank password as zero length. MSIE treats it as no
16140   password on export: but it will try both on import. We now do
16141   the same: PKCS12_parse() tries zero length and no password if
16142   the password is set to "" or NULL (NULL is now a valid password:
16143   it wasn't before) as does the pkcs12 application.
16144
16145   *Steve Henson*
16146
16147 * Bugfixes in `apps/x509.c`: Avoid a memory leak; and don't use
16148   perror when PEM_read_bio_X509_REQ fails, the error message must
16149   be obtained from the error queue.
16150
16151   *Bodo Moeller*
16152
16153 * Avoid 'thread_hash' memory leak in crypto/err/err.c by freeing
16154   it in ERR_remove_state if appropriate, and change ERR_get_state
16155   accordingly to avoid race conditions (this is necessary because
16156   thread_hash is no longer constant once set).
16157
16158   *Bodo Moeller*
16159
16160 * Bugfix for linux-elf makefile.one.
16161
16162   *Ulf Möller*
16163
16164 * RSA_get_default_method() will now cause a default
16165   RSA_METHOD to be chosen if one doesn't exist already.
16166   Previously this was only set during a call to RSA_new()
16167   or RSA_new_method(NULL) meaning it was possible for
16168   RSA_get_default_method() to return NULL.
16169
16170   *Geoff Thorpe*
16171
16172 * Added native name translation to the existing DSO code
16173   that will convert (if the flag to do so is set) filenames
16174   that are sufficiently small and have no path information
16175   into a canonical native form. Eg. "blah" converted to
16176   "libblah.so" or "blah.dll" etc.
16177
16178   *Geoff Thorpe*
16179
16180 * New function ERR_error_string_n(e, buf, len) which is like
16181   ERR_error_string(e, buf), but writes at most 'len' bytes
16182   including the 0 terminator.  For ERR_error_string_n, 'buf'
16183   may not be NULL.
16184
16185   *Damien Miller <djm@mindrot.org>, Bodo Moeller*
16186
16187 * CONF library reworked to become more general.  A new CONF
16188   configuration file reader "class" is implemented as well as a
16189   new functions (`NCONF_*`, for "New CONF") to handle it.  The now
16190   old `CONF_*` functions are still there, but are reimplemented to
16191   work in terms of the new functions.  Also, a set of functions
16192   to handle the internal storage of the configuration data is
16193   provided to make it easier to write new configuration file
16194   reader "classes" (I can definitely see something reading a
16195   configuration file in XML format, for example), called `_CONF_*`,
16196   or "the configuration storage API"...
16197
16198   The new configuration file reading functions are:
16199
16200           NCONF_new, NCONF_free, NCONF_load, NCONF_load_fp, NCONF_load_bio,
16201           NCONF_get_section, NCONF_get_string, NCONF_get_numbre
16202
16203           NCONF_default, NCONF_WIN32
16204
16205           NCONF_dump_fp, NCONF_dump_bio
16206
16207   NCONF_default and NCONF_WIN32 are method (or "class") choosers,
16208   NCONF_new creates a new CONF object.  This works in the same way
16209   as other interfaces in OpenSSL, like the BIO interface.
16210   `NCONF_dump_*` dump the internal storage of the configuration file,
16211   which is useful for debugging.  All other functions take the same
16212   arguments as the old `CONF_*` functions with the exception of the
16213   first that must be a `CONF *` instead of a `LHASH *`.
16214
16215   To make it easier to use the new classes with the old `CONF_*` functions,
16216   the function CONF_set_default_method is provided.
16217
16218   *Richard Levitte*
16219
16220 * Add '-tls1' option to 'openssl ciphers', which was already
16221   mentioned in the documentation but had not been implemented.
16222   (This option is not yet really useful because even the additional
16223   experimental TLS 1.0 ciphers are currently treated as SSL 3.0 ciphers.)
16224
16225   *Bodo Moeller*
16226
16227 * Initial DSO code added into libcrypto for letting OpenSSL (and
16228   OpenSSL-based applications) load shared libraries and bind to
16229   them in a portable way.
16230
16231   *Geoff Thorpe, with contributions from Richard Levitte*
16232
16233### Changes between 0.9.5 and 0.9.5a  [1 Apr 2000]
16234
16235 * Make sure _lrotl and _lrotr are only used with MSVC.
16236
16237 * Use lock CRYPTO_LOCK_RAND correctly in ssleay_rand_status
16238   (the default implementation of RAND_status).
16239
16240 * Rename openssl x509 option '-crlext', which was added in 0.9.5,
16241   to '-clrext' (= clear extensions), as intended and documented.
16242   *Bodo Moeller; inconsistency pointed out by Michael Attili
16243   <attili@amaxo.com>*
16244
16245 * Fix for HMAC. It wasn't zeroing the rest of the block if the key length
16246   was larger than the MD block size.
16247
16248   *Steve Henson, pointed out by Yost William <YostW@tce.com>*
16249
16250 * Modernise PKCS12_parse() so it uses STACK_OF(X509) for its ca argument
16251   fix a leak when the ca argument was passed as NULL. Stop X509_PUBKEY_set()
16252   using the passed key: if the passed key was a private key the result
16253   of X509_print(), for example, would be to print out all the private key
16254   components.
16255
16256   *Steve Henson*
16257
16258 * des_quad_cksum() byte order bug fix.
16259   *Ulf Möller, using the problem description in krb4-0.9.7, where
16260   the solution is attributed to Derrick J Brashear <shadow@DEMENTIA.ORG>*
16261
16262 * Fix so V_ASN1_APP_CHOOSE works again: however its use is strongly
16263   discouraged.
16264
16265   *Steve Henson, pointed out by Brian Korver <briank@cs.stanford.edu>*
16266
16267 * For easily testing in shell scripts whether some command
16268   'openssl XXX' exists, the new pseudo-command 'openssl no-XXX'
16269   returns with exit code 0 iff no command of the given name is available.
16270   'no-XXX' is printed in this case, 'XXX' otherwise.  In both cases,
16271   the output goes to stdout and nothing is printed to stderr.
16272   Additional arguments are always ignored.
16273
16274   Since for each cipher there is a command of the same name,
16275   the 'no-cipher' compilation switches can be tested this way.
16276
16277   ('openssl no-XXX' is not able to detect pseudo-commands such
16278   as 'quit', 'list-XXX-commands', or 'no-XXX' itself.)
16279
16280   *Bodo Moeller*
16281
16282 * Update test suite so that 'make test' succeeds in 'no-rsa' configuration.
16283
16284   *Bodo Moeller*
16285
16286 * For SSL_[CTX_]set_tmp_dh, don't create a DH key if SSL_OP_SINGLE_DH_USE
16287   is set; it will be thrown away anyway because each handshake creates
16288   its own key.
16289   ssl_cert_dup, which is used by SSL_new, now copies DH keys in addition
16290   to parameters -- in previous versions (since OpenSSL 0.9.3) the
16291   'default key' from SSL_CTX_set_tmp_dh would always be lost, meaning
16292   you effectively got SSL_OP_SINGLE_DH_USE when using this macro.
16293
16294   *Bodo Moeller*
16295
16296 * New s_client option -ign_eof: EOF at stdin is ignored, and
16297   'Q' and 'R' lose their special meanings (quit/renegotiate).
16298   This is part of what -quiet does; unlike -quiet, -ign_eof
16299   does not suppress any output.
16300
16301   *Richard Levitte*
16302
16303 * Add compatibility options to the purpose and trust code. The
16304   purpose X509_PURPOSE_ANY is "any purpose" which automatically
16305   accepts a certificate or CA, this was the previous behaviour,
16306   with all the associated security issues.
16307
16308   X509_TRUST_COMPAT is the old trust behaviour: only and
16309   automatically trust self signed roots in certificate store. A
16310   new trust setting X509_TRUST_DEFAULT is used to specify that
16311   a purpose has no associated trust setting and it should instead
16312   use the value in the default purpose.
16313
16314   *Steve Henson*
16315
16316 * Fix the PKCS#8 DSA private key code so it decodes keys again
16317   and fix a memory leak.
16318
16319   *Steve Henson*
16320
16321 * In util/mkerr.pl (which implements 'make errors'), preserve
16322   reason strings from the previous version of the .c file, as
16323   the default to have only downcase letters (and digits) in
16324   automatically generated reasons codes is not always appropriate.
16325
16326   *Bodo Moeller*
16327
16328 * In ERR_load_ERR_strings(), build an ERR_LIB_SYS error reason table
16329   using strerror.  Previously, ERR_reason_error_string() returned
16330   library names as reason strings for SYSerr; but SYSerr is a special
16331   case where small numbers are errno values, not library numbers.
16332
16333   *Bodo Moeller*
16334
16335 * Add '-dsaparam' option to 'openssl dhparam' application.  This
16336   converts DSA parameters into DH parameters. (When creating parameters,
16337   DSA_generate_parameters is used.)
16338
16339   *Bodo Moeller*
16340
16341 * Include 'length' (recommended exponent length) in C code generated
16342   by 'openssl dhparam -C'.
16343
16344   *Bodo Moeller*
16345
16346 * The second argument to set_label in perlasm was already being used
16347   so couldn't be used as a "file scope" flag. Moved to third argument
16348   which was free.
16349
16350   *Steve Henson*
16351
16352 * In PEM_ASN1_write_bio and some other functions, use RAND_pseudo_bytes
16353   instead of RAND_bytes for encryption IVs and salts.
16354
16355   *Bodo Moeller*
16356
16357 * Include RAND_status() into RAND_METHOD instead of implementing
16358   it only for md_rand.c  Otherwise replacing the PRNG by calling
16359   RAND_set_rand_method would be impossible.
16360
16361   *Bodo Moeller*
16362
16363 * Don't let DSA_generate_key() enter an infinite loop if the random
16364   number generation fails.
16365
16366   *Bodo Moeller*
16367
16368 * New 'rand' application for creating pseudo-random output.
16369
16370   *Bodo Moeller*
16371
16372 * Added configuration support for Linux/IA64
16373
16374   *Rolf Haberrecker <rolf@suse.de>*
16375
16376 * Assembler module support for Mingw32.
16377
16378   *Ulf Möller*
16379
16380 * Shared library support for HPUX (in shlib/).
16381
16382   *Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE> and Anonymous*
16383
16384 * Shared library support for Solaris gcc.
16385
16386   *Lutz Behnke <behnke@trustcenter.de>*
16387
16388### Changes between 0.9.4 and 0.9.5  [28 Feb 2000]
16389
16390 * PKCS7_encrypt() was adding text MIME headers twice because they
16391   were added manually and by SMIME_crlf_copy().
16392
16393   *Steve Henson*
16394
16395 * In bntest.c don't call BN_rand with zero bits argument.
16396
16397   *Steve Henson, pointed out by Andrew W. Gray <agray@iconsinc.com>*
16398
16399 * BN_mul bugfix: In bn_mul_part_recursion() only the a>a[n] && b>b[n]
16400   case was implemented. This caused BN_div_recp() to fail occasionally.
16401
16402   *Ulf Möller*
16403
16404 * Add an optional second argument to the set_label() in the perl
16405   assembly language builder. If this argument exists and is set
16406   to 1 it signals that the assembler should use a symbol whose
16407   scope is the entire file, not just the current function. This
16408   is needed with MASM which uses the format label:: for this scope.
16409
16410   *Steve Henson, pointed out by Peter Runestig <peter@runestig.com>*
16411
16412 * Change the ASN1 types so they are typedefs by default. Before
16413   almost all types were #define'd to ASN1_STRING which was causing
16414   STACK_OF() problems: you couldn't declare STACK_OF(ASN1_UTF8STRING)
16415   for example.
16416
16417   *Steve Henson*
16418
16419 * Change names of new functions to the new get1/get0 naming
16420   convention: After 'get1', the caller owns a reference count
16421   and has to call `..._free`; 'get0' returns a pointer to some
16422   data structure without incrementing reference counters.
16423   (Some of the existing 'get' functions increment a reference
16424   counter, some don't.)
16425   Similarly, 'set1' and 'add1' functions increase reference
16426   counters or duplicate objects.
16427
16428   *Steve Henson*
16429
16430 * Allow for the possibility of temp RSA key generation failure:
16431   the code used to assume it always worked and crashed on failure.
16432
16433   *Steve Henson*
16434
16435 * Fix potential buffer overrun problem in BIO_printf().
16436   *Ulf Möller, using public domain code by Patrick Powell; problem
16437   pointed out by David Sacerdote <das33@cornell.edu>*
16438
16439 * Support EGD <http://www.lothar.com/tech/crypto/>.  New functions
16440   RAND_egd() and RAND_status().  In the command line application,
16441   the EGD socket can be specified like a seed file using RANDFILE
16442   or -rand.
16443
16444   *Ulf Möller*
16445
16446 * Allow the string CERTIFICATE to be tolerated in PKCS#7 structures.
16447   Some CAs (e.g. Verisign) distribute certificates in this form.
16448
16449   *Steve Henson*
16450
16451 * Remove the SSL_ALLOW_ADH compile option and set the default cipher
16452   list to exclude them. This means that no special compilation option
16453   is needed to use anonymous DH: it just needs to be included in the
16454   cipher list.
16455
16456   *Steve Henson*
16457
16458 * Change the EVP_MD_CTX_type macro so its meaning consistent with
16459   EVP_MD_type. The old functionality is available in a new macro called
16460   EVP_MD_md(). Change code that uses it and update docs.
16461
16462   *Steve Henson*
16463
16464 * `..._ctrl` functions now have corresponding `..._callback_ctrl` functions
16465   where the `void *` argument is replaced by a function pointer argument.
16466   Previously `void *` was abused to point to functions, which works on
16467   many platforms, but is not correct.  As these functions are usually
16468   called by macros defined in OpenSSL header files, most source code
16469   should work without changes.
16470
16471   *Richard Levitte*
16472
16473 * `<openssl/opensslconf.h>` (which is created by Configure) now contains
16474   sections with information on -D... compiler switches used for
16475   compiling the library so that applications can see them.  To enable
16476   one of these sections, a pre-processor symbol `OPENSSL_..._DEFINES`
16477   must be defined.  E.g.,
16478           #define OPENSSL_ALGORITHM_DEFINES
16479           #include <openssl/opensslconf.h>
16480   defines all pertinent `NO_<algo>` symbols, such as NO_IDEA, NO_RSA, etc.
16481
16482   *Richard Levitte, Ulf and Bodo Möller*
16483
16484 * Bugfix: Tolerate fragmentation and interleaving in the SSL 3/TLS
16485   record layer.
16486
16487   *Bodo Moeller*
16488
16489 * Change the 'other' type in certificate aux info to a STACK_OF
16490   X509_ALGOR. Although not an AlgorithmIdentifier as such it has
16491   the required ASN1 format: arbitrary types determined by an OID.
16492
16493   *Steve Henson*
16494
16495 * Add some PEM_write_X509_REQ_NEW() functions and a command line
16496   argument to 'req'. This is not because the function is newer or
16497   better than others it just uses the work 'NEW' in the certificate
16498   request header lines. Some software needs this.
16499
16500   *Steve Henson*
16501
16502 * Reorganise password command line arguments: now passwords can be
16503   obtained from various sources. Delete the PEM_cb function and make
16504   it the default behaviour: i.e. if the callback is NULL and the
16505   usrdata argument is not NULL interpret it as a null terminated pass
16506   phrase. If usrdata and the callback are NULL then the pass phrase
16507   is prompted for as usual.
16508
16509   *Steve Henson*
16510
16511 * Add support for the Compaq Atalla crypto accelerator. If it is installed,
16512   the support is automatically enabled. The resulting binaries will
16513   autodetect the card and use it if present.
16514
16515   *Ben Laurie and Compaq Inc.*
16516
16517 * Work around for Netscape hang bug. This sends certificate request
16518   and server done in one record. Since this is perfectly legal in the
16519   SSL/TLS protocol it isn't a "bug" option and is on by default. See
16520   the bugs/SSLv3 entry for more info.
16521
16522   *Steve Henson*
16523
16524 * HP-UX tune-up: new unified configs, HP C compiler bug workaround.
16525
16526   *Andy Polyakov*
16527
16528 * Add -rand argument to smime and pkcs12 applications and read/write
16529   of seed file.
16530
16531   *Steve Henson*
16532
16533 * New 'passwd' tool for crypt(3) and apr1 password hashes.
16534
16535   *Bodo Moeller*
16536
16537 * Add command line password options to the remaining applications.
16538
16539   *Steve Henson*
16540
16541 * Bug fix for BN_div_recp() for numerators with an even number of
16542   bits.
16543
16544   *Ulf Möller*
16545
16546 * More tests in bntest.c, and changed test_bn output.
16547
16548   *Ulf Möller*
16549
16550 * ./config recognizes MacOS X now.
16551
16552   *Andy Polyakov*
16553
16554 * Bug fix for BN_div() when the first words of num and divisor are
16555   equal (it gave wrong results if `(rem=(n1-q*d0)&BN_MASK2) < d0)`.
16556
16557   *Ulf Möller*
16558
16559 * Add support for various broken PKCS#8 formats, and command line
16560   options to produce them.
16561
16562   *Steve Henson*
16563
16564 * New functions BN_CTX_start(), BN_CTX_get() and BT_CTX_end() to
16565   get temporary BIGNUMs from a BN_CTX.
16566
16567   *Ulf Möller*
16568
16569 * Correct return values in BN_mod_exp_mont() and BN_mod_exp2_mont()
16570   for p == 0.
16571
16572   *Ulf Möller*
16573
16574 * Change the `SSLeay_add_all_*()` functions to `OpenSSL_add_all_*()` and
16575   include a #define from the old name to the new. The original intent
16576   was that statically linked binaries could for example just call
16577   SSLeay_add_all_ciphers() to just add ciphers to the table and not
16578   link with digests. This never worked because SSLeay_add_all_digests()
16579   and SSLeay_add_all_ciphers() were in the same source file so calling
16580   one would link with the other. They are now in separate source files.
16581
16582   *Steve Henson*
16583
16584 * Add a new -notext option to 'ca' and a -pubkey option to 'spkac'.
16585
16586   *Steve Henson*
16587
16588 * Use a less unusual form of the Miller-Rabin primality test (it used
16589   a binary algorithm for exponentiation integrated into the Miller-Rabin
16590   loop, our standard modexp algorithms are faster).
16591
16592   *Bodo Moeller*
16593
16594 * Support for the EBCDIC character set completed.
16595
16596   *Martin Kraemer <Martin.Kraemer@Mch.SNI.De>*
16597
16598 * Source code cleanups: use const where appropriate, eliminate casts,
16599   use `void *` instead of `char *` in lhash.
16600
16601   *Ulf Möller*
16602
16603 * Bugfix: ssl3_send_server_key_exchange was not restartable
16604   (the state was not changed to SSL3_ST_SW_KEY_EXCH_B, and because of
16605   this the server could overwrite ephemeral keys that the client
16606   has already seen).
16607
16608   *Bodo Moeller*
16609
16610 * Turn DSA_is_prime into a macro that calls BN_is_prime,
16611   using 50 iterations of the Rabin-Miller test.
16612
16613   DSA_generate_parameters now uses BN_is_prime_fasttest (with 50
16614   iterations of the Rabin-Miller test as required by the appendix
16615   to FIPS PUB 186[-1]) instead of DSA_is_prime.
16616   As BN_is_prime_fasttest includes trial division, DSA parameter
16617   generation becomes much faster.
16618
16619   This implies a change for the callback functions in DSA_is_prime
16620   and DSA_generate_parameters: The callback function is called once
16621   for each positive witness in the Rabin-Miller test, not just
16622   occasionally in the inner loop; and the parameters to the
16623   callback function now provide an iteration count for the outer
16624   loop rather than for the current invocation of the inner loop.
16625   DSA_generate_parameters additionally can call the callback
16626   function with an 'iteration count' of -1, meaning that a
16627   candidate has passed the trial division test (when q is generated
16628   from an application-provided seed, trial division is skipped).
16629
16630   *Bodo Moeller*
16631
16632 * New function BN_is_prime_fasttest that optionally does trial
16633   division before starting the Rabin-Miller test and has
16634   an additional BN_CTX * argument (whereas BN_is_prime always
16635   has to allocate at least one BN_CTX).
16636   'callback(1, -1, cb_arg)' is called when a number has passed the
16637   trial division stage.
16638
16639   *Bodo Moeller*
16640
16641 * Fix for bug in CRL encoding. The validity dates weren't being handled
16642   as ASN1_TIME.
16643
16644   *Steve Henson*
16645
16646 * New -pkcs12 option to CA.pl script to write out a PKCS#12 file.
16647
16648   *Steve Henson*
16649
16650 * New function BN_pseudo_rand().
16651
16652   *Ulf Möller*
16653
16654 * Clean up BN_mod_mul_montgomery(): replace the broken (and unreadable)
16655   bignum version of BN_from_montgomery() with the working code from
16656   SSLeay 0.9.0 (the word based version is faster anyway), and clean up
16657   the comments.
16658
16659   *Ulf Möller*
16660
16661 * Avoid a race condition in s2_clnt.c (function get_server_hello) that
16662   made it impossible to use the same SSL_SESSION data structure in
16663   SSL2 clients in multiple threads.
16664
16665   *Bodo Moeller*
16666
16667 * The return value of RAND_load_file() no longer counts bytes obtained
16668   by stat().  RAND_load_file(..., -1) is new and uses the complete file
16669   to seed the PRNG (previously an explicit byte count was required).
16670
16671   *Ulf Möller, Bodo Möller*
16672
16673 * Clean up CRYPTO_EX_DATA functions, some of these didn't have prototypes
16674   used `char *` instead of `void *` and had casts all over the place.
16675
16676   *Steve Henson*
16677
16678 * Make BN_generate_prime() return NULL on error if ret!=NULL.
16679
16680   *Ulf Möller*
16681
16682 * Retain source code compatibility for BN_prime_checks macro:
16683   BN_is_prime(..., BN_prime_checks, ...) now uses
16684   BN_prime_checks_for_size to determine the appropriate number of
16685   Rabin-Miller iterations.
16686
16687   *Ulf Möller*
16688
16689 * Diffie-Hellman uses "safe" primes: DH_check() return code renamed to
16690   DH_CHECK_P_NOT_SAFE_PRIME.
16691   (Check if this is true? OpenPGP calls them "strong".)
16692
16693   *Ulf Möller*
16694
16695 * Merge the functionality of "dh" and "gendh" programs into a new program
16696   "dhparam". The old programs are retained for now but will handle DH keys
16697   (instead of parameters) in future.
16698
16699   *Steve Henson*
16700
16701 * Make the ciphers, s_server and s_client programs check the return values
16702   when a new cipher list is set.
16703
16704   *Steve Henson*
16705
16706 * Enhance the SSL/TLS cipher mechanism to correctly handle the TLS 56bit
16707   ciphers. Before when the 56bit ciphers were enabled the sorting was
16708   wrong.
16709
16710   The syntax for the cipher sorting has been extended to support sorting by
16711   cipher-strength (using the strength_bits hard coded in the tables).
16712   The new command is `@STRENGTH` (see also `doc/apps/ciphers.pod`).
16713
16714   Fix a bug in the cipher-command parser: when supplying a cipher command
16715   string with an "undefined" symbol (neither command nor alphanumeric
16716   *A-Za-z0-9*, ssl_set_cipher_list used to hang in an endless loop. Now
16717   an error is flagged.
16718
16719   Due to the strength-sorting extension, the code of the
16720   ssl_create_cipher_list() function was completely rearranged. I hope that
16721   the readability was also increased :-)
16722
16723   *Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>*
16724
16725 * Minor change to 'x509' utility. The -CAcreateserial option now uses 1
16726   for the first serial number and places 2 in the serial number file. This
16727   avoids problems when the root CA is created with serial number zero and
16728   the first user certificate has the same issuer name and serial number
16729   as the root CA.
16730
16731   *Steve Henson*
16732
16733 * Fixes to X509_ATTRIBUTE utilities, change the 'req' program so it uses
16734   the new code. Add documentation for this stuff.
16735
16736   *Steve Henson*
16737
16738 * Changes to X509_ATTRIBUTE utilities. These have been renamed from
16739   `X509_*()` to `X509at_*()` on the grounds that they don't handle X509
16740   structures and behave in an analogous way to the X509v3 functions:
16741   they shouldn't be called directly but wrapper functions should be used
16742   instead.
16743
16744   So we also now have some wrapper functions that call the X509at functions
16745   when passed certificate requests. (TO DO: similar things can be done with
16746   PKCS#7 signed and unsigned attributes, PKCS#12 attributes and a few other
16747   things. Some of these need some d2i or i2d and print functionality
16748   because they handle more complex structures.)
16749
16750   *Steve Henson*
16751
16752 * Add missing #ifndefs that caused missing symbols when building libssl
16753   as a shared library without RSA.  Use #ifndef NO_SSL2 instead of
16754   NO_RSA in `ssl/s2*.c`.
16755
16756   *Kris Kennaway <kris@hub.freebsd.org>, modified by Ulf Möller*
16757
16758 * Precautions against using the PRNG uninitialized: RAND_bytes() now
16759   has a return value which indicates the quality of the random data
16760   (1 = ok, 0 = not seeded).  Also an error is recorded on the thread's
16761   error queue. New function RAND_pseudo_bytes() generates output that is
16762   guaranteed to be unique but not unpredictable. RAND_add is like
16763   RAND_seed, but takes an extra argument for an entropy estimate
16764   (RAND_seed always assumes full entropy).
16765
16766   *Ulf Möller*
16767
16768 * Do more iterations of Rabin-Miller probable prime test (specifically,
16769   3 for 1024-bit primes, 6 for 512-bit primes, 12 for 256-bit primes
16770   instead of only 2 for all lengths; see BN_prime_checks_for_size definition
16771   in crypto/bn/bn_prime.c for the complete table).  This guarantees a
16772   false-positive rate of at most 2^-80 for random input.
16773
16774   *Bodo Moeller*
16775
16776 * Rewrite ssl3_read_n (ssl/s3_pkt.c) avoiding a couple of bugs.
16777
16778   *Bodo Moeller*
16779
16780 * New function X509_CTX_rget_chain() (renamed to X509_CTX_get1_chain
16781   in the 0.9.5 release), this returns the chain
16782   from an X509_CTX structure with a dup of the stack and all
16783   the X509 reference counts upped: so the stack will exist
16784   after X509_CTX_cleanup() has been called. Modify pkcs12.c
16785   to use this.
16786
16787   Also make SSL_SESSION_print() print out the verify return
16788   code.
16789
16790   *Steve Henson*
16791
16792 * Add manpage for the pkcs12 command. Also change the default
16793   behaviour so MAC iteration counts are used unless the new
16794   -nomaciter option is used. This improves file security and
16795   only older versions of MSIE (4.0 for example) need it.
16796
16797   *Steve Henson*
16798
16799 * Honor the no-xxx Configure options when creating .DEF files.
16800
16801   *Ulf Möller*
16802
16803 * Add PKCS#10 attributes to field table: challengePassword,
16804   unstructuredName and unstructuredAddress. These are taken from
16805   draft PKCS#9 v2.0 but are compatible with v1.2 provided no
16806   international characters are used.
16807
16808   More changes to X509_ATTRIBUTE code: allow the setting of types
16809   based on strings. Remove the 'loc' parameter when adding
16810   attributes because these will be a SET OF encoding which is sorted
16811   in ASN1 order.
16812
16813   *Steve Henson*
16814
16815 * Initial changes to the 'req' utility to allow request generation
16816   automation. This will allow an application to just generate a template
16817   file containing all the field values and have req construct the
16818   request.
16819
16820   Initial support for X509_ATTRIBUTE handling. Stacks of these are
16821   used all over the place including certificate requests and PKCS#7
16822   structures. They are currently handled manually where necessary with
16823   some primitive wrappers for PKCS#7. The new functions behave in a
16824   manner analogous to the X509 extension functions: they allow
16825   attributes to be looked up by NID and added.
16826
16827   Later something similar to the X509V3 code would be desirable to
16828   automatically handle the encoding, decoding and printing of the
16829   more complex types. The string types like challengePassword can
16830   be handled by the string table functions.
16831
16832   Also modified the multi byte string table handling. Now there is
16833   a 'global mask' which masks out certain types. The table itself
16834   can use the flag STABLE_NO_MASK to ignore the mask setting: this
16835   is useful when for example there is only one permissible type
16836   (as in countryName) and using the mask might result in no valid
16837   types at all.
16838
16839   *Steve Henson*
16840
16841 * Clean up 'Finished' handling, and add functions SSL_get_finished and
16842   SSL_get_peer_finished to allow applications to obtain the latest
16843   Finished messages sent to the peer or expected from the peer,
16844   respectively.  (SSL_get_peer_finished is usually the Finished message
16845   actually received from the peer, otherwise the protocol will be aborted.)
16846
16847   As the Finished message are message digests of the complete handshake
16848   (with a total of 192 bits for TLS 1.0 and more for SSL 3.0), they can
16849   be used for external authentication procedures when the authentication
16850   provided by SSL/TLS is not desired or is not enough.
16851
16852   *Bodo Moeller*
16853
16854 * Enhanced support for Alpha Linux is added. Now ./config checks if
16855   the host supports BWX extension and if Compaq C is present on the
16856   $PATH. Just exploiting of the BWX extension results in 20-30%
16857   performance kick for some algorithms, e.g. DES and RC4 to mention
16858   a couple. Compaq C in turn generates ~20% faster code for MD5 and
16859   SHA1.
16860
16861   *Andy Polyakov*
16862
16863 * Add support for MS "fast SGC". This is arguably a violation of the
16864   SSL3/TLS protocol. Netscape SGC does two handshakes: the first with
16865   weak crypto and after checking the certificate is SGC a second one
16866   with strong crypto. MS SGC stops the first handshake after receiving
16867   the server certificate message and sends a second client hello. Since
16868   a server will typically do all the time consuming operations before
16869   expecting any further messages from the client (server key exchange
16870   is the most expensive) there is little difference between the two.
16871
16872   To get OpenSSL to support MS SGC we have to permit a second client
16873   hello message after we have sent server done. In addition we have to
16874   reset the MAC if we do get this second client hello.
16875
16876   *Steve Henson*
16877
16878 * Add a function 'd2i_AutoPrivateKey()' this will automatically decide
16879   if a DER encoded private key is RSA or DSA traditional format. Changed
16880   d2i_PrivateKey_bio() to use it. This is only needed for the "traditional"
16881   format DER encoded private key. Newer code should use PKCS#8 format which
16882   has the key type encoded in the ASN1 structure. Added DER private key
16883   support to pkcs8 application.
16884
16885   *Steve Henson*
16886
16887 * SSL 3/TLS 1 servers now don't request certificates when an anonymous
16888   ciphersuites has been selected (as required by the SSL 3/TLS 1
16889   specifications).  Exception: When SSL_VERIFY_FAIL_IF_NO_PEER_CERT
16890   is set, we interpret this as a request to violate the specification
16891   (the worst that can happen is a handshake failure, and 'correct'
16892   behaviour would result in a handshake failure anyway).
16893
16894   *Bodo Moeller*
16895
16896 * In SSL_CTX_add_session, take into account that there might be multiple
16897   SSL_SESSION structures with the same session ID (e.g. when two threads
16898   concurrently obtain them from an external cache).
16899   The internal cache can handle only one SSL_SESSION with a given ID,
16900   so if there's a conflict, we now throw out the old one to achieve
16901   consistency.
16902
16903   *Bodo Moeller*
16904
16905 * Add OIDs for idea and blowfish in CBC mode. This will allow both
16906   to be used in PKCS#5 v2.0 and S/MIME.  Also add checking to
16907   some routines that use cipher OIDs: some ciphers do not have OIDs
16908   defined and so they cannot be used for S/MIME and PKCS#5 v2.0 for
16909   example.
16910
16911   *Steve Henson*
16912
16913 * Simplify the trust setting structure and code. Now we just have
16914   two sequences of OIDs for trusted and rejected settings. These will
16915   typically have values the same as the extended key usage extension
16916   and any application specific purposes.
16917
16918   The trust checking code now has a default behaviour: it will just
16919   check for an object with the same NID as the passed id. Functions can
16920   be provided to override either the default behaviour or the behaviour
16921   for a given id. SSL client, server and email already have functions
16922   in place for compatibility: they check the NID and also return "trusted"
16923   if the certificate is self signed.
16924
16925   *Steve Henson*
16926
16927 * Add d2i,i2d bio/fp functions for PrivateKey: these convert the
16928   traditional format into an EVP_PKEY structure.
16929
16930   *Steve Henson*
16931
16932 * Add a password callback function PEM_cb() which either prompts for
16933   a password if usr_data is NULL or otherwise assumes it is a null
16934   terminated password. Allow passwords to be passed on command line
16935   environment or config files in a few more utilities.
16936
16937   *Steve Henson*
16938
16939 * Add a bunch of DER and PEM functions to handle PKCS#8 format private
16940   keys. Add some short names for PKCS#8 PBE algorithms and allow them
16941   to be specified on the command line for the pkcs8 and pkcs12 utilities.
16942   Update documentation.
16943
16944   *Steve Henson*
16945
16946 * Support for ASN1 "NULL" type. This could be handled before by using
16947   ASN1_TYPE but there wasn't any function that would try to read a NULL
16948   and produce an error if it couldn't. For compatibility we also have
16949   ASN1_NULL_new() and ASN1_NULL_free() functions but these are faked and
16950   don't allocate anything because they don't need to.
16951
16952   *Steve Henson*
16953
16954 * Initial support for MacOS is now provided. Examine INSTALL.MacOS
16955   for details.
16956
16957   *Andy Polyakov, Roy Woods <roy@centicsystems.ca>*
16958
16959 * Rebuild of the memory allocation routines used by OpenSSL code and
16960   possibly others as well.  The purpose is to make an interface that
16961   provide hooks so anyone can build a separate set of allocation and
16962   deallocation routines to be used by OpenSSL, for example memory
16963   pool implementations, or something else, which was previously hard
16964   since Malloc(), Realloc() and Free() were defined as macros having
16965   the values malloc, realloc and free, respectively (except for Win32
16966   compilations).  The same is provided for memory debugging code.
16967   OpenSSL already comes with functionality to find memory leaks, but
16968   this gives people a chance to debug other memory problems.
16969
16970   With these changes, a new set of functions and macros have appeared:
16971
16972     CRYPTO_set_mem_debug_functions()         [F]
16973     CRYPTO_get_mem_debug_functions()         [F]
16974     CRYPTO_dbg_set_options()                 [F]
16975     CRYPTO_dbg_get_options()                 [F]
16976     CRYPTO_malloc_debug_init()               [M]
16977
16978   The memory debug functions are NULL by default, unless the library
16979   is compiled with CRYPTO_MDEBUG or friends is defined.  If someone
16980   wants to debug memory anyway, CRYPTO_malloc_debug_init() (which
16981   gives the standard debugging functions that come with OpenSSL) or
16982   CRYPTO_set_mem_debug_functions() (tells OpenSSL to use functions
16983   provided by the library user) must be used.  When the standard
16984   debugging functions are used, CRYPTO_dbg_set_options can be used to
16985   request additional information:
16986   CRYPTO_dbg_set_options(V_CYRPTO_MDEBUG_xxx) corresponds to setting
16987   the CRYPTO_MDEBUG_xxx macro when compiling the library.
16988
16989   Also, things like CRYPTO_set_mem_functions will always give the
16990   expected result (the new set of functions is used for allocation
16991   and deallocation) at all times, regardless of platform and compiler
16992   options.
16993
16994   To finish it up, some functions that were never use in any other
16995   way than through macros have a new API and new semantic:
16996
16997     CRYPTO_dbg_malloc()
16998     CRYPTO_dbg_realloc()
16999     CRYPTO_dbg_free()
17000
17001   All macros of value have retained their old syntax.
17002
17003   *Richard Levitte and Bodo Moeller*
17004
17005 * Some S/MIME fixes. The OID for SMIMECapabilities was wrong, the
17006   ordering of SMIMECapabilities wasn't in "strength order" and there
17007   was a missing NULL in the AlgorithmIdentifier for the SHA1 signature
17008   algorithm.
17009
17010   *Steve Henson*
17011
17012 * Some ASN1 types with illegal zero length encoding (INTEGER,
17013   ENUMERATED and OBJECT IDENTIFIER) choked the ASN1 routines.
17014
17015   *Frans Heymans <fheymans@isaserver.be>, modified by Steve Henson*
17016
17017 * Merge in my S/MIME library for OpenSSL. This provides a simple
17018   S/MIME API on top of the PKCS#7 code, a MIME parser (with enough
17019   functionality to handle multipart/signed properly) and a utility
17020   called 'smime' to call all this stuff. This is based on code I
17021   originally wrote for Celo who have kindly allowed it to be
17022   included in OpenSSL.
17023
17024   *Steve Henson*
17025
17026 * Add variants des_set_key_checked and des_set_key_unchecked of
17027   des_set_key (aka des_key_sched).  Global variable des_check_key
17028   decides which of these is called by des_set_key; this way
17029   des_check_key behaves as it always did, but applications and
17030   the library itself, which was buggy for des_check_key == 1,
17031   have a cleaner way to pick the version they need.
17032
17033   *Bodo Moeller*
17034
17035 * New function PKCS12_newpass() which changes the password of a
17036   PKCS12 structure.
17037
17038   *Steve Henson*
17039
17040 * Modify X509_TRUST and X509_PURPOSE so it also uses a static and
17041   dynamic mix. In both cases the ids can be used as an index into the
17042   table. Also modified the X509_TRUST_add() and X509_PURPOSE_add()
17043   functions so they accept a list of the field values and the
17044   application doesn't need to directly manipulate the X509_TRUST
17045   structure.
17046
17047   *Steve Henson*
17048
17049 * Modify the ASN1_STRING_TABLE stuff so it also uses bsearch and doesn't
17050   need initialising.
17051
17052   *Steve Henson*
17053
17054 * Modify the way the V3 extension code looks up extensions. This now
17055   works in a similar way to the object code: we have some "standard"
17056   extensions in a static table which is searched with OBJ_bsearch()
17057   and the application can add dynamic ones if needed. The file
17058   crypto/x509v3/ext_dat.h now has the info: this file needs to be
17059   updated whenever a new extension is added to the core code and kept
17060   in ext_nid order. There is a simple program 'tabtest.c' which checks
17061   this. New extensions are not added too often so this file can readily
17062   be maintained manually.
17063
17064   There are two big advantages in doing things this way. The extensions
17065   can be looked up immediately and no longer need to be "added" using
17066   X509V3_add_standard_extensions(): this function now does nothing.
17067   Side note: I get *lots* of email saying the extension code doesn't
17068   work because people forget to call this function.
17069   Also no dynamic allocation is done unless new extensions are added:
17070   so if we don't add custom extensions there is no need to call
17071   X509V3_EXT_cleanup().
17072
17073   *Steve Henson*
17074
17075 * Modify enc utility's salting as follows: make salting the default. Add a
17076   magic header, so unsalted files fail gracefully instead of just decrypting
17077   to garbage. This is because not salting is a big security hole, so people
17078   should be discouraged from doing it.
17079
17080   *Ben Laurie*
17081
17082 * Fixes and enhancements to the 'x509' utility. It allowed a message
17083   digest to be passed on the command line but it only used this
17084   parameter when signing a certificate. Modified so all relevant
17085   operations are affected by the digest parameter including the
17086   -fingerprint and -x509toreq options. Also -x509toreq choked if a
17087   DSA key was used because it didn't fix the digest.
17088
17089   *Steve Henson*
17090
17091 * Initial certificate chain verify code. Currently tests the untrusted
17092   certificates for consistency with the verify purpose (which is set
17093   when the X509_STORE_CTX structure is set up) and checks the pathlength.
17094
17095   There is a NO_CHAIN_VERIFY compilation option to keep the old behaviour:
17096   this is because it will reject chains with invalid extensions whereas
17097   every previous version of OpenSSL and SSLeay made no checks at all.
17098
17099   Trust code: checks the root CA for the relevant trust settings. Trust
17100   settings have an initial value consistent with the verify purpose: e.g.
17101   if the verify purpose is for SSL client use it expects the CA to be
17102   trusted for SSL client use. However the default value can be changed to
17103   permit custom trust settings: one example of this would be to only trust
17104   certificates from a specific "secure" set of CAs.
17105
17106   Also added X509_STORE_CTX_new() and X509_STORE_CTX_free() functions
17107   which should be used for version portability: especially since the
17108   verify structure is likely to change more often now.
17109
17110   SSL integration. Add purpose and trust to SSL_CTX and SSL and functions
17111   to set them. If not set then assume SSL clients will verify SSL servers
17112   and vice versa.
17113
17114   Two new options to the verify program: -untrusted allows a set of
17115   untrusted certificates to be passed in and -purpose which sets the
17116   intended purpose of the certificate. If a purpose is set then the
17117   new chain verify code is used to check extension consistency.
17118
17119   *Steve Henson*
17120
17121 * Support for the authority information access extension.
17122
17123   *Steve Henson*
17124
17125 * Modify RSA and DSA PEM read routines to transparently handle
17126   PKCS#8 format private keys. New *_PUBKEY_* functions that handle
17127   public keys in a format compatible with certificate
17128   SubjectPublicKeyInfo structures. Unfortunately there were already
17129   functions called *_PublicKey_* which used various odd formats so
17130   these are retained for compatibility: however the DSA variants were
17131   never in a public release so they have been deleted. Changed dsa/rsa
17132   utilities to handle the new format: note no releases ever handled public
17133   keys so we should be OK.
17134
17135   The primary motivation for this change is to avoid the same fiasco
17136   that dogs private keys: there are several incompatible private key
17137   formats some of which are standard and some OpenSSL specific and
17138   require various evil hacks to allow partial transparent handling and
17139   even then it doesn't work with DER formats. Given the option anything
17140   other than PKCS#8 should be dumped: but the other formats have to
17141   stay in the name of compatibility.
17142
17143   With public keys and the benefit of hindsight one standard format
17144   is used which works with EVP_PKEY, RSA or DSA structures: though
17145   it clearly returns an error if you try to read the wrong kind of key.
17146
17147   Added a -pubkey option to the 'x509' utility to output the public key.
17148   Also rename the `EVP_PKEY_get_*()` to `EVP_PKEY_rget_*()`
17149   (renamed to `EVP_PKEY_get1_*()` in the OpenSSL 0.9.5 release) and add
17150   `EVP_PKEY_rset_*()` functions (renamed to `EVP_PKEY_set1_*()`)
17151   that do the same as the `EVP_PKEY_assign_*()` except they up the
17152   reference count of the added key (they don't "swallow" the
17153   supplied key).
17154
17155   *Steve Henson*
17156
17157 * Fixes to crypto/x509/by_file.c the code to read in certificates and
17158   CRLs would fail if the file contained no certificates or no CRLs:
17159   added a new function to read in both types and return the number
17160   read: this means that if none are read it will be an error. The
17161   DER versions of the certificate and CRL reader would always fail
17162   because it isn't possible to mix certificates and CRLs in DER format
17163   without choking one or the other routine. Changed this to just read
17164   a certificate: this is the best we can do. Also modified the code
17165   in `apps/verify.c` to take notice of return codes: it was previously
17166   attempting to read in certificates from NULL pointers and ignoring
17167   any errors: this is one reason why the cert and CRL reader seemed
17168   to work. It doesn't check return codes from the default certificate
17169   routines: these may well fail if the certificates aren't installed.
17170
17171   *Steve Henson*
17172
17173 * Code to support otherName option in GeneralName.
17174
17175   *Steve Henson*
17176
17177 * First update to verify code. Change the verify utility
17178   so it warns if it is passed a self signed certificate:
17179   for consistency with the normal behaviour. X509_verify
17180   has been modified to it will now verify a self signed
17181   certificate if *exactly* the same certificate appears
17182   in the store: it was previously impossible to trust a
17183   single self signed certificate. This means that:
17184   openssl verify ss.pem
17185   now gives a warning about a self signed certificate but
17186   openssl verify -CAfile ss.pem ss.pem
17187   is OK.
17188
17189   *Steve Henson*
17190
17191 * For servers, store verify_result in SSL_SESSION data structure
17192   (and add it to external session representation).
17193   This is needed when client certificate verifications fails,
17194   but an application-provided verification callback (set by
17195   SSL_CTX_set_cert_verify_callback) allows accepting the session
17196   anyway (i.e. leaves x509_store_ctx->error != X509_V_OK
17197   but returns 1): When the session is reused, we have to set
17198   ssl->verify_result to the appropriate error code to avoid
17199   security holes.
17200
17201   *Bodo Moeller, problem pointed out by Lutz Jaenicke*
17202
17203 * Fix a bug in the new PKCS#7 code: it didn't consider the
17204   case in PKCS7_dataInit() where the signed PKCS7 structure
17205   didn't contain any existing data because it was being created.
17206
17207   *Po-Cheng Chen <pocheng@nst.com.tw>, slightly modified by Steve Henson*
17208
17209 * Add a salt to the key derivation routines in enc.c. This
17210   forms the first 8 bytes of the encrypted file. Also add a
17211   -S option to allow a salt to be input on the command line.
17212
17213   *Steve Henson*
17214
17215 * New function X509_cmp(). Oddly enough there wasn't a function
17216   to compare two certificates. We do this by working out the SHA1
17217   hash and comparing that. X509_cmp() will be needed by the trust
17218   code.
17219
17220   *Steve Henson*
17221
17222 * SSL_get1_session() is like SSL_get_session(), but increments
17223   the reference count in the SSL_SESSION returned.
17224
17225   *Geoff Thorpe <geoff@eu.c2.net>*
17226
17227 * Fix for 'req': it was adding a null to request attributes.
17228   Also change the X509_LOOKUP and X509_INFO code to handle
17229   certificate auxiliary information.
17230
17231   *Steve Henson*
17232
17233 * Add support for 40 and 64 bit RC2 and RC4 algorithms: document
17234   the 'enc' command.
17235
17236   *Steve Henson*
17237
17238 * Add the possibility to add extra information to the memory leak
17239   detecting output, to form tracebacks, showing from where each
17240   allocation was originated: CRYPTO_push_info("constant string") adds
17241   the string plus current file name and line number to a per-thread
17242   stack, CRYPTO_pop_info() does the obvious, CRYPTO_remove_all_info()
17243   is like calling CYRPTO_pop_info() until the stack is empty.
17244   Also updated memory leak detection code to be multi-thread-safe.
17245
17246   *Richard Levitte*
17247
17248 * Add options -text and -noout to pkcs7 utility and delete the
17249   encryption options which never did anything. Update docs.
17250
17251   *Steve Henson*
17252
17253 * Add options to some of the utilities to allow the pass phrase
17254   to be included on either the command line (not recommended on
17255   OSes like Unix) or read from the environment. Update the
17256   manpages and fix a few bugs.
17257
17258   *Steve Henson*
17259
17260 * Add a few manpages for some of the openssl commands.
17261
17262   *Steve Henson*
17263
17264 * Fix the -revoke option in ca. It was freeing up memory twice,
17265   leaking and not finding already revoked certificates.
17266
17267   *Steve Henson*
17268
17269 * Extensive changes to support certificate auxiliary information.
17270   This involves the use of X509_CERT_AUX structure and X509_AUX
17271   functions. An X509_AUX function such as PEM_read_X509_AUX()
17272   can still read in a certificate file in the usual way but it
17273   will also read in any additional "auxiliary information". By
17274   doing things this way a fair degree of compatibility can be
17275   retained: existing certificates can have this information added
17276   using the new 'x509' options.
17277
17278   Current auxiliary information includes an "alias" and some trust
17279   settings. The trust settings will ultimately be used in enhanced
17280   certificate chain verification routines: currently a certificate
17281   can only be trusted if it is self signed and then it is trusted
17282   for all purposes.
17283
17284   *Steve Henson*
17285
17286 * Fix assembler for Alpha (tested only on DEC OSF not Linux or `*BSD`).
17287   The problem was that one of the replacement routines had not been working
17288   since SSLeay releases.  For now the offending routine has been replaced
17289   with non-optimised assembler.  Even so, this now gives around 95%
17290   performance improvement for 1024 bit RSA signs.
17291
17292   *Mark Cox*
17293
17294 * Hack to fix PKCS#7 decryption when used with some unorthodox RC2
17295   handling. Most clients have the effective key size in bits equal to
17296   the key length in bits: so a 40 bit RC2 key uses a 40 bit (5 byte) key.
17297   A few however don't do this and instead use the size of the decrypted key
17298   to determine the RC2 key length and the AlgorithmIdentifier to determine
17299   the effective key length. In this case the effective key length can still
17300   be 40 bits but the key length can be 168 bits for example. This is fixed
17301   by manually forcing an RC2 key into the EVP_PKEY structure because the
17302   EVP code can't currently handle unusual RC2 key sizes: it always assumes
17303   the key length and effective key length are equal.
17304
17305   *Steve Henson*
17306
17307 * Add a bunch of functions that should simplify the creation of
17308   X509_NAME structures. Now you should be able to do:
17309   X509_NAME_add_entry_by_txt(nm, "CN", MBSTRING_ASC, "Steve", -1, -1, 0);
17310   and have it automatically work out the correct field type and fill in
17311   the structures. The more adventurous can try:
17312   X509_NAME_add_entry_by_txt(nm, field, MBSTRING_UTF8, str, -1, -1, 0);
17313   and it will (hopefully) work out the correct multibyte encoding.
17314
17315   *Steve Henson*
17316
17317 * Change the 'req' utility to use the new field handling and multibyte
17318   copy routines. Before the DN field creation was handled in an ad hoc
17319   way in req, ca, and x509 which was rather broken and didn't support
17320   BMPStrings or UTF8Strings. Since some software doesn't implement
17321   BMPStrings or UTF8Strings yet, they can be enabled using the config file
17322   using the dirstring_type option. See the new comment in the default
17323   openssl.cnf for more info.
17324
17325   *Steve Henson*
17326
17327 * Make crypto/rand/md_rand.c more robust:
17328   - Assure unique random numbers after fork().
17329   - Make sure that concurrent threads access the global counter and
17330     md serializably so that we never lose entropy in them
17331     or use exactly the same state in multiple threads.
17332     Access to the large state is not always serializable because
17333     the additional locking could be a performance killer, and
17334     md should be large enough anyway.
17335
17336   *Bodo Moeller*
17337
17338 * New file `apps/app_rand.c` with commonly needed functionality
17339   for handling the random seed file.
17340
17341   Use the random seed file in some applications that previously did not:
17342           ca,
17343           dsaparam -genkey (which also ignored its '-rand' option),
17344           s_client,
17345           s_server,
17346           x509 (when signing).
17347   Except on systems with /dev/urandom, it is crucial to have a random
17348   seed file at least for key creation, DSA signing, and for DH exchanges;
17349   for RSA signatures we could do without one.
17350
17351   gendh and gendsa (unlike genrsa) used to read only the first byte
17352   of each file listed in the '-rand' option.  The function as previously
17353   found in genrsa is now in app_rand.c and is used by all programs
17354   that support '-rand'.
17355
17356   *Bodo Moeller*
17357
17358 * In RAND_write_file, use mode 0600 for creating files;
17359   don't just chmod when it may be too late.
17360
17361   *Bodo Moeller*
17362
17363 * Report an error from X509_STORE_load_locations
17364   when X509_LOOKUP_load_file or X509_LOOKUP_add_dir failed.
17365
17366   *Bill Perry*
17367
17368 * New function ASN1_mbstring_copy() this copies a string in either
17369   ASCII, Unicode, Universal (4 bytes per character) or UTF8 format
17370   into an ASN1_STRING type. A mask of permissible types is passed
17371   and it chooses the "minimal" type to use or an error if not type
17372   is suitable.
17373
17374   *Steve Henson*
17375
17376 * Add function equivalents to the various macros in asn1.h. The old
17377   macros are retained with an `M_` prefix. Code inside the library can
17378   use the `M_` macros. External code (including the openssl utility)
17379   should *NOT* in order to be "shared library friendly".
17380
17381   *Steve Henson*
17382
17383 * Add various functions that can check a certificate's extensions
17384   to see if it usable for various purposes such as SSL client,
17385   server or S/MIME and CAs of these types. This is currently
17386   VERY EXPERIMENTAL but will ultimately be used for certificate chain
17387   verification. Also added a -purpose flag to x509 utility to
17388   print out all the purposes.
17389
17390   *Steve Henson*
17391
17392 * Add a CRYPTO_EX_DATA to X509 certificate structure and associated
17393   functions.
17394
17395   *Steve Henson*
17396
17397 * New `X509V3_{X509,CRL,REVOKED}_get_d2i()` functions. These will search
17398   for, obtain and decode and extension and obtain its critical flag.
17399   This allows all the necessary extension code to be handled in a
17400   single function call.
17401
17402   *Steve Henson*
17403
17404 * RC4 tune-up featuring 30-40% performance improvement on most RISC
17405   platforms. See crypto/rc4/rc4_enc.c for further details.
17406
17407   *Andy Polyakov*
17408
17409 * New -noout option to asn1parse. This causes no output to be produced
17410   its main use is when combined with -strparse and -out to extract data
17411   from a file (which may not be in ASN.1 format).
17412
17413   *Steve Henson*
17414
17415 * Fix for pkcs12 program. It was hashing an invalid certificate pointer
17416   when producing the local key id.
17417
17418   *Richard Levitte <levitte@stacken.kth.se>*
17419
17420 * New option -dhparam in s_server. This allows a DH parameter file to be
17421   stated explicitly. If it is not stated then it tries the first server
17422   certificate file. The previous behaviour hard coded the filename
17423   "server.pem".
17424
17425   *Steve Henson*
17426
17427 * Add -pubin and -pubout options to the rsa and dsa commands. These allow
17428   a public key to be input or output. For example:
17429   openssl rsa -in key.pem -pubout -out pubkey.pem
17430   Also added necessary DSA public key functions to handle this.
17431
17432   *Steve Henson*
17433
17434 * Fix so PKCS7_dataVerify() doesn't crash if no certificates are contained
17435   in the message. This was handled by allowing
17436   X509_find_by_issuer_and_serial() to tolerate a NULL passed to it.
17437
17438   *Steve Henson, reported by Sampo Kellomaki <sampo@mail.neuronio.pt>*
17439
17440 * Fix for bug in d2i_ASN1_bytes(): other ASN1 functions add an extra null
17441   to the end of the strings whereas this didn't. This would cause problems
17442   if strings read with d2i_ASN1_bytes() were later modified.
17443
17444   *Steve Henson, reported by Arne Ansper <arne@ats.cyber.ee>*
17445
17446 * Fix for base64 decode bug. When a base64 bio reads only one line of
17447   data and it contains EOF it will end up returning an error. This is
17448   caused by input 46 bytes long. The cause is due to the way base64
17449   BIOs find the start of base64 encoded data. They do this by trying a
17450   trial decode on each line until they find one that works. When they
17451   do a flag is set and it starts again knowing it can pass all the
17452   data directly through the decoder. Unfortunately it doesn't reset
17453   the context it uses. This means that if EOF is reached an attempt
17454   is made to pass two EOFs through the context and this causes the
17455   resulting error. This can also cause other problems as well. As is
17456   usual with these problems it takes *ages* to find and the fix is
17457   trivial: move one line.
17458
17459   *Steve Henson, reported by ian@uns.ns.ac.yu (Ivan Nejgebauer)*
17460
17461 * Ugly workaround to get s_client and s_server working under Windows. The
17462   old code wouldn't work because it needed to select() on sockets and the
17463   tty (for keypresses and to see if data could be written). Win32 only
17464   supports select() on sockets so we select() with a 1s timeout on the
17465   sockets and then see if any characters are waiting to be read, if none
17466   are present then we retry, we also assume we can always write data to
17467   the tty. This isn't nice because the code then blocks until we've
17468   received a complete line of data and it is effectively polling the
17469   keyboard at 1s intervals: however it's quite a bit better than not
17470   working at all :-) A dedicated Windows application might handle this
17471   with an event loop for example.
17472
17473   *Steve Henson*
17474
17475 * Enhance RSA_METHOD structure. Now there are two extra methods, rsa_sign
17476   and rsa_verify. When the RSA_FLAGS_SIGN_VER option is set these functions
17477   will be called when RSA_sign() and RSA_verify() are used. This is useful
17478   if rsa_pub_dec() and rsa_priv_enc() equivalents are not available.
17479   For this to work properly RSA_public_decrypt() and RSA_private_encrypt()
17480   should *not* be used: RSA_sign() and RSA_verify() must be used instead.
17481   This necessitated the support of an extra signature type NID_md5_sha1
17482   for SSL signatures and modifications to the SSL library to use it instead
17483   of calling RSA_public_decrypt() and RSA_private_encrypt().
17484
17485   *Steve Henson*
17486
17487 * Add new -verify -CAfile and -CApath options to the crl program, these
17488   will lookup a CRL issuers certificate and verify the signature in a
17489   similar way to the verify program. Tidy up the crl program so it
17490   no longer accesses structures directly. Make the ASN1 CRL parsing a bit
17491   less strict. It will now permit CRL extensions even if it is not
17492   a V2 CRL: this will allow it to tolerate some broken CRLs.
17493
17494   *Steve Henson*
17495
17496 * Initialize all non-automatic variables each time one of the openssl
17497   sub-programs is started (this is necessary as they may be started
17498   multiple times from the "OpenSSL>" prompt).
17499
17500   *Lennart Bang, Bodo Moeller*
17501
17502 * Preliminary compilation option RSA_NULL which disables RSA crypto without
17503   removing all other RSA functionality (this is what NO_RSA does). This
17504   is so (for example) those in the US can disable those operations covered
17505   by the RSA patent while allowing storage and parsing of RSA keys and RSA
17506   key generation.
17507
17508   *Steve Henson*
17509
17510 * Non-copying interface to BIO pairs.
17511   (still largely untested)
17512
17513   *Bodo Moeller*
17514
17515 * New function ASN1_tag2str() to convert an ASN1 tag to a descriptive
17516   ASCII string. This was handled independently in various places before.
17517
17518   *Steve Henson*
17519
17520 * New functions UTF8_getc() and UTF8_putc() that parse and generate
17521   UTF8 strings a character at a time.
17522
17523   *Steve Henson*
17524
17525 * Use client_version from client hello to select the protocol
17526   (s23_srvr.c) and for RSA client key exchange verification
17527   (s3_srvr.c), as required by the SSL 3.0/TLS 1.0 specifications.
17528
17529   *Bodo Moeller*
17530
17531 * Add various utility functions to handle SPKACs, these were previously
17532   handled by poking round in the structure internals. Added new function
17533   NETSCAPE_SPKI_print() to print out SPKAC and a new utility 'spkac' to
17534   print, verify and generate SPKACs. Based on an original idea from
17535   Massimiliano Pala <madwolf@comune.modena.it> but extensively modified.
17536
17537   *Steve Henson*
17538
17539 * RIPEMD160 is operational on all platforms and is back in 'make test'.
17540
17541   *Andy Polyakov*
17542
17543 * Allow the config file extension section to be overwritten on the
17544   command line. Based on an original idea from Massimiliano Pala
17545   <madwolf@comune.modena.it>. The new option is called -extensions
17546   and can be applied to ca, req and x509. Also -reqexts to override
17547   the request extensions in req and -crlexts to override the crl extensions
17548   in ca.
17549
17550   *Steve Henson*
17551
17552 * Add new feature to the SPKAC handling in ca.  Now you can include
17553   the same field multiple times by preceding it by "XXXX." for example:
17554   1.OU="Unit name 1"
17555   2.OU="Unit name 2"
17556   this is the same syntax as used in the req config file.
17557
17558   *Steve Henson*
17559
17560 * Allow certificate extensions to be added to certificate requests. These
17561   are specified in a 'req_extensions' option of the req section of the
17562   config file. They can be printed out with the -text option to req but
17563   are otherwise ignored at present.
17564
17565   *Steve Henson*
17566
17567 * Fix a horrible bug in enc_read() in crypto/evp/bio_enc.c: if the first
17568   data read consists of only the final block it would not decrypted because
17569   EVP_CipherUpdate() would correctly report zero bytes had been decrypted.
17570   A misplaced 'break' also meant the decrypted final block might not be
17571   copied until the next read.
17572
17573   *Steve Henson*
17574
17575 * Initial support for DH_METHOD. Again based on RSA_METHOD. Also added
17576   a few extra parameters to the DH structure: these will be useful if
17577   for example we want the value of 'q' or implement X9.42 DH.
17578
17579   *Steve Henson*
17580
17581 * Initial support for DSA_METHOD. This is based on the RSA_METHOD and
17582   provides hooks that allow the default DSA functions or functions on a
17583   "per key" basis to be replaced. This allows hardware acceleration and
17584   hardware key storage to be handled without major modification to the
17585   library. Also added low-level modexp hooks and CRYPTO_EX structure and
17586   associated functions.
17587
17588   *Steve Henson*
17589
17590 * Add a new flag to memory BIOs, BIO_FLAG_MEM_RDONLY. This marks the BIO
17591   as "read only": it can't be written to and the buffer it points to will
17592   not be freed. Reading from a read only BIO is much more efficient than
17593   a normal memory BIO. This was added because there are several times when
17594   an area of memory needs to be read from a BIO. The previous method was
17595   to create a memory BIO and write the data to it, this results in two
17596   copies of the data and an O(n^2) reading algorithm. There is a new
17597   function BIO_new_mem_buf() which creates a read only memory BIO from
17598   an area of memory. Also modified the PKCS#7 routines to use read only
17599   memory BIOs.
17600
17601   *Steve Henson*
17602
17603 * Bugfix: ssl23_get_client_hello did not work properly when called in
17604   state SSL23_ST_SR_CLNT_HELLO_B, i.e. when the first 7 bytes of
17605   a SSLv2-compatible client hello for SSLv3 or TLSv1 could be read,
17606   but a retry condition occurred while trying to read the rest.
17607
17608   *Bodo Moeller*
17609
17610 * The PKCS7_ENC_CONTENT_new() function was setting the content type as
17611   NID_pkcs7_encrypted by default: this was wrong since this should almost
17612   always be NID_pkcs7_data. Also modified the PKCS7_set_type() to handle
17613   the encrypted data type: this is a more sensible place to put it and it
17614   allows the PKCS#12 code to be tidied up that duplicated this
17615   functionality.
17616
17617   *Steve Henson*
17618
17619 * Changed obj_dat.pl script so it takes its input and output files on
17620   the command line. This should avoid shell escape redirection problems
17621   under Win32.
17622
17623   *Steve Henson*
17624
17625 * Initial support for certificate extension requests, these are included
17626   in things like Xenroll certificate requests. Included functions to allow
17627   extensions to be obtained and added.
17628
17629   *Steve Henson*
17630
17631 * -crlf option to s_client and s_server for sending newlines as
17632   CRLF (as required by many protocols).
17633
17634   *Bodo Moeller*
17635
17636### Changes between 0.9.3a and 0.9.4  [09 Aug 1999]
17637
17638 * Install libRSAglue.a when OpenSSL is built with RSAref.
17639
17640   *Ralf S. Engelschall*
17641
17642 * A few more `#ifndef NO_FP_API / #endif` pairs for consistency.
17643
17644   *Andrija Antonijevic <TheAntony2@bigfoot.com>*
17645
17646 * Fix -startdate and -enddate (which was missing) arguments to 'ca'
17647   program.
17648
17649   *Steve Henson*
17650
17651 * New function DSA_dup_DH, which duplicates DSA parameters/keys as
17652   DH parameters/keys (q is lost during that conversion, but the resulting
17653   DH parameters contain its length).
17654
17655   For 1024-bit p, DSA_generate_parameters followed by DSA_dup_DH is
17656   much faster than DH_generate_parameters (which creates parameters
17657   where `p = 2*q + 1`), and also the smaller q makes DH computations
17658   much more efficient (160-bit exponentiation instead of 1024-bit
17659   exponentiation); so this provides a convenient way to support DHE
17660   ciphersuites in SSL/TLS servers (see ssl/ssltest.c).  It is of
17661   utter importance to use
17662           SSL_CTX_set_options(s_ctx, SSL_OP_SINGLE_DH_USE);
17663   or
17664           SSL_set_options(s_ctx, SSL_OP_SINGLE_DH_USE);
17665   when such DH parameters are used, because otherwise small subgroup
17666   attacks may become possible!
17667
17668   *Bodo Moeller*
17669
17670 * Avoid memory leak in i2d_DHparams.
17671
17672   *Bodo Moeller*
17673
17674 * Allow the -k option to be used more than once in the enc program:
17675   this allows the same encrypted message to be read by multiple recipients.
17676
17677   *Steve Henson*
17678
17679 * New function OBJ_obj2txt(buf, buf_len, a, no_name), this converts
17680   an ASN1_OBJECT to a text string. If the "no_name" parameter is set then
17681   it will always use the numerical form of the OID, even if it has a short
17682   or long name.
17683
17684   *Steve Henson*
17685
17686 * Added an extra RSA flag: RSA_FLAG_EXT_PKEY. Previously the rsa_mod_exp
17687   method only got called if p,q,dmp1,dmq1,iqmp components were present,
17688   otherwise bn_mod_exp was called. In the case of hardware keys for example
17689   no private key components need be present and it might store extra data
17690   in the RSA structure, which cannot be accessed from bn_mod_exp.
17691   By setting RSA_FLAG_EXT_PKEY rsa_mod_exp will always be called for
17692   private key operations.
17693
17694   *Steve Henson*
17695
17696 * Added support for SPARC Linux.
17697
17698   *Andy Polyakov*
17699
17700 * pem_password_cb function type incompatibly changed from
17701           typedef int pem_password_cb(char *buf, int size, int rwflag);
17702   to
17703           ....(char *buf, int size, int rwflag, void *userdata);
17704   so that applications can pass data to their callbacks:
17705   The `PEM[_ASN1]_{read,write}...` functions and macros now take an
17706   additional void * argument, which is just handed through whenever
17707   the password callback is called.
17708
17709   *Damien Miller <dmiller@ilogic.com.au>; tiny changes by Bodo Moeller*
17710
17711   New function SSL_CTX_set_default_passwd_cb_userdata.
17712
17713   Compatibility note: As many C implementations push function arguments
17714   onto the stack in reverse order, the new library version is likely to
17715   interoperate with programs that have been compiled with the old
17716   pem_password_cb definition (PEM_whatever takes some data that
17717   happens to be on the stack as its last argument, and the callback
17718   just ignores this garbage); but there is no guarantee whatsoever that
17719   this will work.
17720
17721 * The -DPLATFORM="\"$(PLATFORM)\"" definition and the similar -DCFLAGS=...
17722   (both in crypto/Makefile.ssl for use by crypto/cversion.c) caused
17723   problems not only on Windows, but also on some Unix platforms.
17724   To avoid problematic command lines, these definitions are now in an
17725   auto-generated file crypto/buildinf.h (created by crypto/Makefile.ssl
17726   for standard "make" builds, by util/mk1mf.pl for "mk1mf" builds).
17727
17728   *Bodo Moeller*
17729
17730 * MIPS III/IV assembler module is reimplemented.
17731
17732   *Andy Polyakov*
17733
17734 * More DES library cleanups: remove references to srand/rand and
17735   delete an unused file.
17736
17737   *Ulf Möller*
17738
17739 * Add support for the free Netwide assembler (NASM) under Win32,
17740   since not many people have MASM (ml) and it can be hard to obtain.
17741   This is currently experimental but it seems to work OK and pass all
17742   the tests. Check out INSTALL.W32 for info.
17743
17744   *Steve Henson*
17745
17746 * Fix memory leaks in s3_clnt.c: All non-anonymous SSL3/TLS1 connections
17747   without temporary keys kept an extra copy of the server key,
17748   and connections with temporary keys did not free everything in case
17749   of an error.
17750
17751   *Bodo Moeller*
17752
17753 * New function RSA_check_key and new openssl rsa option -check
17754   for verifying the consistency of RSA keys.
17755
17756   *Ulf Moeller, Bodo Moeller*
17757
17758 * Various changes to make Win32 compile work:
17759   1. Casts to avoid "loss of data" warnings in p5_crpt2.c
17760   2. Change unsigned int to int in b_dump.c to avoid "signed/unsigned
17761      comparison" warnings.
17762   3. Add `sk_<TYPE>_sort` to DEF file generator and do make update.
17763
17764   *Steve Henson*
17765
17766 * Add a debugging option to PKCS#5 v2 key generation function: when
17767   you #define DEBUG_PKCS5V2 passwords, salts, iteration counts and
17768   derived keys are printed to stderr.
17769
17770   *Steve Henson*
17771
17772 * Copy the flags in ASN1_STRING_dup().
17773
17774   *Roman E. Pavlov <pre@mo.msk.ru>*
17775
17776 * The x509 application mishandled signing requests containing DSA
17777   keys when the signing key was also DSA and the parameters didn't match.
17778
17779   It was supposed to omit the parameters when they matched the signing key:
17780   the verifying software was then supposed to automatically use the CA's
17781   parameters if they were absent from the end user certificate.
17782
17783   Omitting parameters is no longer recommended. The test was also
17784   the wrong way round! This was probably due to unusual behaviour in
17785   EVP_cmp_parameters() which returns 1 if the parameters match.
17786   This meant that parameters were omitted when they *didn't* match and
17787   the certificate was useless. Certificates signed with 'ca' didn't have
17788   this bug.
17789
17790   *Steve Henson, reported by Doug Erickson <Doug.Erickson@Part.NET>*
17791
17792 * Memory leak checking (-DCRYPTO_MDEBUG) had some problems.
17793   The interface is as follows:
17794   Applications can use
17795           CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_ON) aka MemCheck_start(),
17796           CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_OFF) aka MemCheck_stop();
17797   "off" is now the default.
17798   The library internally uses
17799           CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_DISABLE) aka MemCheck_off(),
17800           CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_ENABLE) aka MemCheck_on()
17801   to disable memory-checking temporarily.
17802
17803   Some inconsistent states that previously were possible (and were
17804   even the default) are now avoided.
17805
17806   -DCRYPTO_MDEBUG_TIME is new and additionally stores the current time
17807   with each memory chunk allocated; this is occasionally more helpful
17808   than just having a counter.
17809
17810   -DCRYPTO_MDEBUG_THREAD is also new and adds the thread ID.
17811
17812   -DCRYPTO_MDEBUG_ALL enables all of the above, plus any future
17813   extensions.
17814
17815   *Bodo Moeller*
17816
17817 * Introduce "mode" for SSL structures (with defaults in SSL_CTX),
17818   which largely parallels "options", but is for changing API behaviour,
17819   whereas "options" are about protocol behaviour.
17820   Initial "mode" flags are:
17821
17822   SSL_MODE_ENABLE_PARTIAL_WRITE   Allow SSL_write to report success when
17823                                   a single record has been written.
17824   SSL_MODE_ACCEPT_MOVING_WRITE_BUFFER  Don't insist that SSL_write
17825                                   retries use the same buffer location.
17826                                   (But all of the contents must be
17827                                   copied!)
17828
17829   *Bodo Moeller*
17830
17831 * Bugfix: SSL_set_options ignored its parameter, only SSL_CTX_set_options
17832   worked.
17833
17834 * Fix problems with no-hmac etc.
17835
17836   *Ulf Möller, pointed out by Brian Wellington <bwelling@tislabs.com>*
17837
17838 * New functions RSA_get_default_method(), RSA_set_method() and
17839   RSA_get_method(). These allows replacement of RSA_METHODs without having
17840   to mess around with the internals of an RSA structure.
17841
17842   *Steve Henson*
17843
17844 * Fix memory leaks in DSA_do_sign and DSA_is_prime.
17845   Also really enable memory leak checks in openssl.c and in some
17846   test programs.
17847
17848   *Chad C. Mulligan, Bodo Moeller*
17849
17850 * Fix a bug in d2i_ASN1_INTEGER() and i2d_ASN1_INTEGER() which can mess
17851   up the length of negative integers. This has now been simplified to just
17852   store the length when it is first determined and use it later, rather
17853   than trying to keep track of where data is copied and updating it to
17854   point to the end.
17855   *Steve Henson, reported by Brien Wheeler <bwheeler@authentica-security.com>*
17856
17857 * Add a new function PKCS7_signatureVerify. This allows the verification
17858   of a PKCS#7 signature but with the signing certificate passed to the
17859   function itself. This contrasts with PKCS7_dataVerify which assumes the
17860   certificate is present in the PKCS#7 structure. This isn't always the
17861   case: certificates can be omitted from a PKCS#7 structure and be
17862   distributed by "out of band" means (such as a certificate database).
17863
17864   *Steve Henson*
17865
17866 * Complete the `PEM_*` macros with DECLARE_PEM versions to replace the
17867   function prototypes in pem.h, also change util/mkdef.pl to add the
17868   necessary function names.
17869
17870   *Steve Henson*
17871
17872 * mk1mf.pl (used by Windows builds) did not properly read the
17873   options set by Configure in the top level Makefile, and Configure
17874   was not even able to write more than one option correctly.
17875   Fixed, now "no-idea no-rc5 -DCRYPTO_MDEBUG" etc. works as intended.
17876
17877   *Bodo Moeller*
17878
17879 * New functions CONF_load_bio() and CONF_load_fp() to allow a config
17880   file to be loaded from a BIO or FILE pointer. The BIO version will
17881   for example allow memory BIOs to contain config info.
17882
17883   *Steve Henson*
17884
17885 * New function "CRYPTO_num_locks" that returns CRYPTO_NUM_LOCKS.
17886   Whoever hopes to achieve shared-library compatibility across versions
17887   must use this, not the compile-time macro.
17888   (Exercise 0.9.4: Which is the minimum library version required by
17889   such programs?)
17890   Note: All this applies only to multi-threaded programs, others don't
17891   need locks.
17892
17893   *Bodo Moeller*
17894
17895 * Add missing case to s3_clnt.c state machine -- one of the new SSL tests
17896   through a BIO pair triggered the default case, i.e.
17897   SSLerr(...,SSL_R_UNKNOWN_STATE).
17898
17899   *Bodo Moeller*
17900
17901 * New "BIO pair" concept (crypto/bio/bss_bio.c) so that applications
17902   can use the SSL library even if none of the specific BIOs is
17903   appropriate.
17904
17905   *Bodo Moeller*
17906
17907 * Fix a bug in i2d_DSAPublicKey() which meant it returned the wrong value
17908   for the encoded length.
17909
17910   *Jeon KyoungHo <khjeon@sds.samsung.co.kr>*
17911
17912 * Add initial documentation of the X509V3 functions.
17913
17914   *Steve Henson*
17915
17916 * Add a new pair of functions PEM_write_PKCS8PrivateKey() and
17917   PEM_write_bio_PKCS8PrivateKey() that are equivalent to
17918   PEM_write_PrivateKey() and PEM_write_bio_PrivateKey() but use the more
17919   secure PKCS#8 private key format with a high iteration count.
17920
17921   *Steve Henson*
17922
17923 * Fix determination of Perl interpreter: A perl or perl5
17924   *directory* in $PATH was also accepted as the interpreter.
17925
17926   *Ralf S. Engelschall*
17927
17928 * Fix demos/sign/sign.c: well there wasn't anything strictly speaking
17929   wrong with it but it was very old and did things like calling
17930   PEM_ASN1_read() directly and used MD5 for the hash not to mention some
17931   unusual formatting.
17932
17933   *Steve Henson*
17934
17935 * Fix demos/selfsign.c: it used obsolete and deleted functions, changed
17936   to use the new extension code.
17937
17938   *Steve Henson*
17939
17940 * Implement the PEM_read/PEM_write functions in crypto/pem/pem_all.c
17941   with macros. This should make it easier to change their form, add extra
17942   arguments etc. Fix a few PEM prototypes which didn't have cipher as a
17943   constant.
17944
17945   *Steve Henson*
17946
17947 * Add to configuration table a new entry that can specify an alternative
17948   name for unistd.h (for pre-POSIX systems); we need this for NeXTstep,
17949   according to Mark Crispin <MRC@Panda.COM>.
17950
17951   *Bodo Moeller*
17952
17953 * DES CBC did not update the IV. Weird.
17954
17955   *Ben Laurie*
17956lse
17957   des_cbc_encrypt does not update the IV, but des_ncbc_encrypt does.
17958   Changing the behaviour of the former might break existing programs --
17959   where IV updating is needed, des_ncbc_encrypt can be used.
17960ndif
17961
17962 * When bntest is run from "make test" it drives bc to check its
17963   calculations, as well as internally checking them. If an internal check
17964   fails, it needs to cause bc to give a non-zero result or make test carries
17965   on without noticing the failure. Fixed.
17966
17967   *Ben Laurie*
17968
17969 * DES library cleanups.
17970
17971   *Ulf Möller*
17972
17973 * Add support for PKCS#5 v2.0 PBE algorithms. This will permit PKCS#8 to be
17974   used with any cipher unlike PKCS#5 v1.5 which can at most handle 64 bit
17975   ciphers. NOTE: although the key derivation function has been verified
17976   against some published test vectors it has not been extensively tested
17977   yet. Added a -v2 "cipher" option to pkcs8 application to allow the use
17978   of v2.0.
17979
17980   *Steve Henson*
17981
17982 * Instead of "mkdir -p", which is not fully portable, use new
17983   Perl script "util/mkdir-p.pl".
17984
17985   *Bodo Moeller*
17986
17987 * Rewrite the way password based encryption (PBE) is handled. It used to
17988   assume that the ASN1 AlgorithmIdentifier parameter was a PBEParameter
17989   structure. This was true for the PKCS#5 v1.5 and PKCS#12 PBE algorithms
17990   but doesn't apply to PKCS#5 v2.0 where it can be something else. Now
17991   the 'parameter' field of the AlgorithmIdentifier is passed to the
17992   underlying key generation function so it must do its own ASN1 parsing.
17993   This has also changed the EVP_PBE_CipherInit() function which now has a
17994   'parameter' argument instead of literal salt and iteration count values
17995   and the function EVP_PBE_ALGOR_CipherInit() has been deleted.
17996
17997   *Steve Henson*
17998
17999 * Support for PKCS#5 v1.5 compatible password based encryption algorithms
18000   and PKCS#8 functionality. New 'pkcs8' application linked to openssl.
18001   Needed to change the PEM_STRING_EVP_PKEY value which was just "PRIVATE
18002   KEY" because this clashed with PKCS#8 unencrypted string. Since this
18003   value was just used as a "magic string" and not used directly its
18004   value doesn't matter.
18005
18006   *Steve Henson*
18007
18008 * Introduce some semblance of const correctness to BN. Shame C doesn't
18009   support mutable.
18010
18011   *Ben Laurie*
18012
18013 * "linux-sparc64" configuration (ultrapenguin).
18014
18015   *Ray Miller <ray.miller@oucs.ox.ac.uk>*
18016   "linux-sparc" configuration.
18017
18018   *Christian Forster <fo@hawo.stw.uni-erlangen.de>*
18019
18020 * config now generates no-xxx options for missing ciphers.
18021
18022   *Ulf Möller*
18023
18024 * Support the EBCDIC character set (work in progress).
18025   File ebcdic.c not yet included because it has a different license.
18026
18027   *Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>*
18028
18029 * Support BS2000/OSD-POSIX.
18030
18031   *Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>*
18032
18033 * Make callbacks for key generation use `void *` instead of `char *`.
18034
18035   *Ben Laurie*
18036
18037 * Make S/MIME samples compile (not yet tested).
18038
18039   *Ben Laurie*
18040
18041 * Additional typesafe stacks.
18042
18043   *Ben Laurie*
18044
18045 * New configuration variants "bsdi-elf-gcc" (BSD/OS 4.x).
18046
18047   *Bodo Moeller*
18048
18049### Changes between 0.9.3 and 0.9.3a  [29 May 1999]
18050
18051 * New configuration variant "sco5-gcc".
18052
18053 * Updated some demos.
18054
18055   *Sean O Riordain, Wade Scholine*
18056
18057 * Add missing BIO_free at exit of pkcs12 application.
18058
18059   *Wu Zhigang*
18060
18061 * Fix memory leak in conf.c.
18062
18063   *Steve Henson*
18064
18065 * Updates for Win32 to assembler version of MD5.
18066
18067   *Steve Henson*
18068
18069 * Set #! path to perl in `apps/der_chop` to where we found it
18070   instead of using a fixed path.
18071
18072   *Bodo Moeller*
18073
18074 * SHA library changes for irix64-mips4-cc.
18075
18076   *Andy Polyakov*
18077
18078 * Improvements for VMS support.
18079
18080   *Richard Levitte*
18081
18082### Changes between 0.9.2b and 0.9.3  [24 May 1999]
18083
18084 * Bignum library bug fix. IRIX 6 passes "make test" now!
18085   This also avoids the problems with SC4.2 and unpatched SC5.
18086
18087   *Andy Polyakov <appro@fy.chalmers.se>*
18088
18089 * New functions sk_num, sk_value and sk_set to replace the previous macros.
18090   These are required because of the typesafe stack would otherwise break
18091   existing code. If old code used a structure member which used to be STACK
18092   and is now STACK_OF (for example cert in a PKCS7_SIGNED structure) with
18093   sk_num or sk_value it would produce an error because the num, data members
18094   are not present in STACK_OF. Now it just produces a warning. sk_set
18095   replaces the old method of assigning a value to sk_value
18096   (e.g. sk_value(x, i) = y) which the library used in a few cases. Any code
18097   that does this will no longer work (and should use sk_set instead) but
18098   this could be regarded as a "questionable" behaviour anyway.
18099
18100   *Steve Henson*
18101
18102 * Fix most of the other PKCS#7 bugs. The "experimental" code can now
18103   correctly handle encrypted S/MIME data.
18104
18105   *Steve Henson*
18106
18107 * Change type of various DES function arguments from des_cblock
18108   (which means, in function argument declarations, pointer to char)
18109   to des_cblock * (meaning pointer to array with 8 char elements),
18110   which allows the compiler to do more typechecking; it was like
18111   that back in SSLeay, but with lots of ugly casts.
18112
18113   Introduce new type const_des_cblock.
18114
18115   *Bodo Moeller*
18116
18117 * Reorganise the PKCS#7 library and get rid of some of the more obvious
18118   problems: find RecipientInfo structure that matches recipient certificate
18119   and initialise the ASN1 structures properly based on passed cipher.
18120
18121   *Steve Henson*
18122
18123 * Belatedly make the BN tests actually check the results.
18124
18125   *Ben Laurie*
18126
18127 * Fix the encoding and decoding of negative ASN1 INTEGERS and conversion
18128   to and from BNs: it was completely broken. New compilation option
18129   NEG_PUBKEY_BUG to allow for some broken certificates that encode public
18130   key elements as negative integers.
18131
18132   *Steve Henson*
18133
18134 * Reorganize and speed up MD5.
18135
18136   *Andy Polyakov <appro@fy.chalmers.se>*
18137
18138 * VMS support.
18139
18140   *Richard Levitte <richard@levitte.org>*
18141
18142 * New option -out to asn1parse to allow the parsed structure to be
18143   output to a file. This is most useful when combined with the -strparse
18144   option to examine the output of things like OCTET STRINGS.
18145
18146   *Steve Henson*
18147
18148 * Make SSL library a little more fool-proof by not requiring any longer
18149   that `SSL_set_{accept,connect}_state` be called before
18150   `SSL_{accept,connect}` may be used (`SSL_set_..._state` is omitted
18151   in many applications because usually everything *appeared* to work as
18152   intended anyway -- now it really works as intended).
18153
18154   *Bodo Moeller*
18155
18156 * Move openssl.cnf out of lib/.
18157
18158   *Ulf Möller*
18159
18160 * Fix various things to let OpenSSL even pass "egcc -pipe -O2 -Wall
18161   -Wshadow -Wpointer-arith -Wcast-align -Wmissing-prototypes
18162   -Wmissing-declarations -Wnested-externs -Winline" with EGCS 1.1.2+
18163
18164   *Ralf S. Engelschall*
18165
18166 * Various fixes to the EVP and PKCS#7 code. It may now be able to
18167   handle PKCS#7 enveloped data properly.
18168
18169   *Sebastian Akerman <sak@parallelconsulting.com>, modified by Steve*
18170
18171 * Create a duplicate of the SSL_CTX's CERT in SSL_new instead of
18172   copying pointers.  The cert_st handling is changed by this in
18173   various ways (and thus what used to be known as ctx->default_cert
18174   is now called ctx->cert, since we don't resort to `s->ctx->[default_]cert`
18175   any longer when s->cert does not give us what we need).
18176   ssl_cert_instantiate becomes obsolete by this change.
18177   As soon as we've got the new code right (possibly it already is?),
18178   we have solved a couple of bugs of the earlier code where s->cert
18179   was used as if it could not have been shared with other SSL structures.
18180
18181   Note that using the SSL API in certain dirty ways now will result
18182   in different behaviour than observed with earlier library versions:
18183   Changing settings for an `SSL_CTX *ctx` after having done s = SSL_new(ctx)
18184   does not influence s as it used to.
18185
18186   In order to clean up things more thoroughly, inside SSL_SESSION
18187   we don't use CERT any longer, but a new structure SESS_CERT
18188   that holds per-session data (if available); currently, this is
18189   the peer's certificate chain and, for clients, the server's certificate
18190   and temporary key.  CERT holds only those values that can have
18191   meaningful defaults in an SSL_CTX.
18192
18193   *Bodo Moeller*
18194
18195 * New function X509V3_EXT_i2d() to create an X509_EXTENSION structure
18196   from the internal representation. Various PKCS#7 fixes: remove some
18197   evil casts and set the enc_dig_alg field properly based on the signing
18198   key type.
18199
18200   *Steve Henson*
18201
18202 * Allow PKCS#12 password to be set from the command line or the
18203   environment. Let 'ca' get its config file name from the environment
18204   variables "OPENSSL_CONF" or "SSLEAY_CONF" (for consistency with 'req'
18205   and 'x509').
18206
18207   *Steve Henson*
18208
18209 * Allow certificate policies extension to use an IA5STRING for the
18210   organization field. This is contrary to the PKIX definition but
18211   VeriSign uses it and IE5 only recognises this form. Document 'x509'
18212   extension option.
18213
18214   *Steve Henson*
18215
18216 * Add PEDANTIC compiler flag to allow compilation with gcc -pedantic,
18217   without disallowing inline assembler and the like for non-pedantic builds.
18218
18219   *Ben Laurie*
18220
18221 * Support Borland C++ builder.
18222
18223   *Janez Jere <jj@void.si>, modified by Ulf Möller*
18224
18225 * Support Mingw32.
18226
18227   *Ulf Möller*
18228
18229 * SHA-1 cleanups and performance enhancements.
18230
18231   *Andy Polyakov <appro@fy.chalmers.se>*
18232
18233 * Sparc v8plus assembler for the bignum library.
18234
18235   *Andy Polyakov <appro@fy.chalmers.se>*
18236
18237 * Accept any -xxx and +xxx compiler options in Configure.
18238
18239   *Ulf Möller*
18240
18241 * Update HPUX configuration.
18242
18243   *Anonymous*
18244
18245 * Add missing `sk_<type>_unshift()` function to safestack.h
18246
18247   *Ralf S. Engelschall*
18248
18249 * New function SSL_CTX_use_certificate_chain_file that sets the
18250   "extra_cert"s in addition to the certificate.  (This makes sense
18251   only for "PEM" format files, as chains as a whole are not
18252   DER-encoded.)
18253
18254   *Bodo Moeller*
18255
18256 * Support verify_depth from the SSL API.
18257   x509_vfy.c had what can be considered an off-by-one-error:
18258   Its depth (which was not part of the external interface)
18259   was actually counting the number of certificates in a chain;
18260   now it really counts the depth.
18261
18262   *Bodo Moeller*
18263
18264 * Bugfix in crypto/x509/x509_cmp.c: The SSLerr macro was used
18265   instead of X509err, which often resulted in confusing error
18266   messages since the error codes are not globally unique
18267   (e.g. an alleged error in ssl3_accept when a certificate
18268   didn't match the private key).
18269
18270 * New function SSL_CTX_set_session_id_context that allows to set a default
18271   value (so that you don't need SSL_set_session_id_context for each
18272   connection using the SSL_CTX).
18273
18274   *Bodo Moeller*
18275
18276 * OAEP decoding bug fix.
18277
18278   *Ulf Möller*
18279
18280 * Support INSTALL_PREFIX for package builders, as proposed by
18281   David Harris.
18282
18283   *Bodo Moeller*
18284
18285 * New Configure options "threads" and "no-threads".  For systems
18286   where the proper compiler options are known (currently Solaris
18287   and Linux), "threads" is the default.
18288
18289   *Bodo Moeller*
18290
18291 * New script util/mklink.pl as a faster substitute for util/mklink.sh.
18292
18293   *Bodo Moeller*
18294
18295 * Install various scripts to $(OPENSSLDIR)/misc, not to
18296   $(INSTALLTOP)/bin -- they shouldn't clutter directories
18297   such as /usr/local/bin.
18298
18299   *Bodo Moeller*
18300
18301 * "make linux-shared" to build shared libraries.
18302
18303   *Niels Poppe <niels@netbox.org>*
18304
18305 * New Configure option `no-<cipher>` (rsa, idea, rc5, ...).
18306
18307   *Ulf Möller*
18308
18309 * Add the PKCS#12 API documentation to openssl.txt. Preliminary support for
18310   extension adding in x509 utility.
18311
18312   *Steve Henson*
18313
18314 * Remove NOPROTO sections and error code comments.
18315
18316   *Ulf Möller*
18317
18318 * Partial rewrite of the DEF file generator to now parse the ANSI
18319   prototypes.
18320
18321   *Steve Henson*
18322
18323 * New Configure options --prefix=DIR and --openssldir=DIR.
18324
18325   *Ulf Möller*
18326
18327 * Complete rewrite of the error code script(s). It is all now handled
18328   by one script at the top level which handles error code gathering,
18329   header rewriting and C source file generation. It should be much better
18330   than the old method: it now uses a modified version of Ulf's parser to
18331   read the ANSI prototypes in all header files (thus the old K&R definitions
18332   aren't needed for error creation any more) and do a better job of
18333   translating function codes into names. The old 'ASN1 error code embedded
18334   in a comment' is no longer necessary and it doesn't use .err files which
18335   have now been deleted. Also the error code call doesn't have to appear all
18336   on one line (which resulted in some large lines...).
18337
18338   *Steve Henson*
18339
18340 * Change #include filenames from `<foo.h>` to `<openssl/foo.h>`.
18341
18342   *Bodo Moeller*
18343
18344 * Change behaviour of ssl2_read when facing length-0 packets: Don't return
18345   0 (which usually indicates a closed connection), but continue reading.
18346
18347   *Bodo Moeller*
18348
18349 * Fix some race conditions.
18350
18351   *Bodo Moeller*
18352
18353 * Add support for CRL distribution points extension. Add Certificate
18354   Policies and CRL distribution points documentation.
18355
18356   *Steve Henson*
18357
18358 * Move the autogenerated header file parts to crypto/opensslconf.h.
18359
18360   *Ulf Möller*
18361
18362 * Fix new 56-bit DES export ciphersuites: they were using 7 bytes instead of
18363   8 of keying material. Merlin has also confirmed interop with this fix
18364   between OpenSSL and Baltimore C/SSL 2.0 and J/SSL 2.0.
18365
18366   *Merlin Hughes <merlin@baltimore.ie>*
18367
18368 * Fix lots of warnings.
18369
18370   *Richard Levitte <levitte@stacken.kth.se>*
18371
18372 * In add_cert_dir() in crypto/x509/by_dir.c, break out of the loop if
18373   the directory spec didn't end with a LIST_SEPARATOR_CHAR.
18374
18375   *Richard Levitte <levitte@stacken.kth.se>*
18376
18377 * Fix problems with sizeof(long) == 8.
18378
18379   *Andy Polyakov <appro@fy.chalmers.se>*
18380
18381 * Change functions to ANSI C.
18382
18383   *Ulf Möller*
18384
18385 * Fix typos in error codes.
18386
18387   *Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>, Ulf Möller*
18388
18389 * Remove defunct assembler files from Configure.
18390
18391   *Ulf Möller*
18392
18393 * SPARC v8 assembler BIGNUM implementation.
18394
18395   *Andy Polyakov <appro@fy.chalmers.se>*
18396
18397 * Support for Certificate Policies extension: both print and set.
18398   Various additions to support the r2i method this uses.
18399
18400   *Steve Henson*
18401
18402 * A lot of constification, and fix a bug in X509_NAME_oneline() that could
18403   return a const string when you are expecting an allocated buffer.
18404
18405   *Ben Laurie*
18406
18407 * Add support for ASN1 types UTF8String and VISIBLESTRING, also the CHOICE
18408   types DirectoryString and DisplayText.
18409
18410   *Steve Henson*
18411
18412 * Add code to allow r2i extensions to access the configuration database,
18413   add an LHASH database driver and add several ctx helper functions.
18414
18415   *Steve Henson*
18416
18417 * Fix an evil bug in bn_expand2() which caused various BN functions to
18418   fail when they extended the size of a BIGNUM.
18419
18420   *Steve Henson*
18421
18422 * Various utility functions to handle SXNet extension. Modify mkdef.pl to
18423   support typesafe stack.
18424
18425   *Steve Henson*
18426
18427 * Fix typo in SSL_[gs]et_options().
18428
18429   *Nils Frostberg <nils@medcom.se>*
18430
18431 * Delete various functions and files that belonged to the (now obsolete)
18432   old X509V3 handling code.
18433
18434   *Steve Henson*
18435
18436 * New Configure option "rsaref".
18437
18438   *Ulf Möller*
18439
18440 * Don't auto-generate pem.h.
18441
18442   *Bodo Moeller*
18443
18444 * Introduce type-safe ASN.1 SETs.
18445
18446   *Ben Laurie*
18447
18448 * Convert various additional casted stacks to type-safe STACK_OF() variants.
18449
18450   *Ben Laurie, Ralf S. Engelschall, Steve Henson*
18451
18452 * Introduce type-safe STACKs. This will almost certainly break lots of code
18453   that links with OpenSSL (well at least cause lots of warnings), but fear
18454   not: the conversion is trivial, and it eliminates loads of evil casts. A
18455   few STACKed things have been converted already. Feel free to convert more.
18456   In the fullness of time, I'll do away with the STACK type altogether.
18457
18458   *Ben Laurie*
18459
18460 * Add `openssl ca -revoke <certfile>` facility which revokes a certificate
18461   specified in `<certfile>` by updating the entry in the index.txt file.
18462   This way one no longer has to edit the index.txt file manually for
18463   revoking a certificate. The -revoke option does the gory details now.
18464
18465   *Massimiliano Pala <madwolf@openca.org>, Ralf S. Engelschall*
18466
18467 * Fix `openssl crl -noout -text` combination where `-noout` killed the
18468   `-text` option at all and this way the `-noout -text` combination was
18469   inconsistent in `openssl crl` with the friends in `openssl x509|rsa|dsa`.
18470
18471   *Ralf S. Engelschall*
18472
18473 * Make sure a corresponding plain text error message exists for the
18474   X509_V_ERR_CERT_REVOKED/23 error number which can occur when a
18475   verify callback function determined that a certificate was revoked.
18476
18477   *Ralf S. Engelschall*
18478
18479 * Bugfix: In test/testenc, don't test `openssl <cipher>` for
18480   ciphers that were excluded, e.g. by -DNO_IDEA.  Also, test
18481   all available ciphers including rc5, which was forgotten until now.
18482   In order to let the testing shell script know which algorithms
18483   are available, a new (up to now undocumented) command
18484   `openssl list-cipher-commands` is used.
18485
18486   *Bodo Moeller*
18487
18488 * Bugfix: s_client occasionally would sleep in select() when
18489   it should have checked SSL_pending() first.
18490
18491   *Bodo Moeller*
18492
18493 * New functions DSA_do_sign and DSA_do_verify to provide access to
18494   the raw DSA values prior to ASN.1 encoding.
18495
18496   *Ulf Möller*
18497
18498 * Tweaks to Configure
18499
18500   *Niels Poppe <niels@netbox.org>*
18501
18502 * Add support for PKCS#5 v2.0 ASN1 PBES2 structures. No other support,
18503   yet...
18504
18505   *Steve Henson*
18506
18507 * New variables $(RANLIB) and $(PERL) in the Makefiles.
18508
18509   *Ulf Möller*
18510
18511 * New config option to avoid instructions that are illegal on the 80386.
18512   The default code is faster, but requires at least a 486.
18513
18514   *Ulf Möller*
18515
18516 * Got rid of old SSL2_CLIENT_VERSION (inconsistently used) and
18517   SSL2_SERVER_VERSION (not used at all) macros, which are now the
18518   same as SSL2_VERSION anyway.
18519
18520   *Bodo Moeller*
18521
18522 * New "-showcerts" option for s_client.
18523
18524   *Bodo Moeller*
18525
18526 * Still more PKCS#12 integration. Add pkcs12 application to openssl
18527   application. Various cleanups and fixes.
18528
18529   *Steve Henson*
18530
18531 * More PKCS#12 integration. Add new pkcs12 directory with Makefile.ssl and
18532   modify error routines to work internally. Add error codes and PBE init
18533   to library startup routines.
18534
18535   *Steve Henson*
18536
18537 * Further PKCS#12 integration. Added password based encryption, PKCS#8 and
18538   packing functions to asn1 and evp. Changed function names and error
18539   codes along the way.
18540
18541   *Steve Henson*
18542
18543 * PKCS12 integration: and so it begins... First of several patches to
18544   slowly integrate PKCS#12 functionality into OpenSSL. Add PKCS#12
18545   objects to objects.h
18546
18547   *Steve Henson*
18548
18549 * Add a new 'indent' option to some X509V3 extension code. Initial ASN1
18550   and display support for Thawte strong extranet extension.
18551
18552   *Steve Henson*
18553
18554 * Add LinuxPPC support.
18555
18556   *Jeff Dubrule <igor@pobox.org>*
18557
18558 * Get rid of redundant BN file bn_mulw.c, and rename bn_div64 to
18559   bn_div_words in alpha.s.
18560
18561   *Hannes Reinecke <H.Reinecke@hw.ac.uk> and Ben Laurie*
18562
18563 * Make sure the RSA OAEP test is skipped under -DRSAref because
18564   OAEP isn't supported when OpenSSL is built with RSAref.
18565
18566   *Ulf Moeller <ulf@fitug.de>*
18567
18568 * Move definitions of IS_SET/IS_SEQUENCE inside crypto/asn1/asn1.h
18569   so they no longer are missing under -DNOPROTO.
18570
18571   *Soren S. Jorvang <soren@t.dk>*
18572
18573### Changes between 0.9.1c and 0.9.2b  [22 Mar 1999]
18574
18575 * Make SSL_get_peer_cert_chain() work in servers. Unfortunately, it still
18576   doesn't work when the session is reused. Coming soon!
18577
18578   *Ben Laurie*
18579
18580 * Fix a security hole, that allows sessions to be reused in the wrong
18581   context thus bypassing client cert protection! All software that uses
18582   client certs and session caches in multiple contexts NEEDS PATCHING to
18583   allow session reuse! A fuller solution is in the works.
18584
18585   *Ben Laurie, problem pointed out by Holger Reif, Bodo Moeller (and ???)*
18586
18587 * Some more source tree cleanups (removed obsolete files
18588   crypto/bf/asm/bf586.pl, test/test.txt and crypto/sha/asm/f.s; changed
18589   permission on "config" script to be executable) and a fix for the INSTALL
18590   document.
18591
18592   *Ulf Moeller <ulf@fitug.de>*
18593
18594 * Remove some legacy and erroneous uses of malloc, free instead of
18595   Malloc, Free.
18596
18597   *Lennart Bang <lob@netstream.se>, with minor changes by Steve*
18598
18599 * Make rsa_oaep_test return non-zero on error.
18600
18601   *Ulf Moeller <ulf@fitug.de>*
18602
18603 * Add support for native Solaris shared libraries. Configure
18604   solaris-sparc-sc4-pic, make, then run shlib/solaris-sc4.sh. It'd be nice
18605   if someone would make that last step automatic.
18606
18607   *Matthias Loepfe <Matthias.Loepfe@AdNovum.CH>*
18608
18609 * ctx_size was not built with the right compiler during "make links". Fixed.
18610
18611   *Ben Laurie*
18612
18613 * Change the meaning of 'ALL' in the cipher list. It now means "everything
18614   except NULL ciphers". This means the default cipher list will no longer
18615   enable NULL ciphers. They need to be specifically enabled e.g. with
18616   the string "DEFAULT:eNULL".
18617
18618   *Steve Henson*
18619
18620 * Fix to RSA private encryption routines: if p < q then it would
18621   occasionally produce an invalid result. This will only happen with
18622   externally generated keys because OpenSSL (and SSLeay) ensure p > q.
18623
18624   *Steve Henson*
18625
18626 * Be less restrictive and allow also `perl util/perlpath.pl
18627   /path/to/bin/perl` in addition to `perl util/perlpath.pl /path/to/bin`,
18628   because this way one can also use an interpreter named `perl5` (which is
18629   usually the name of Perl 5.xxx on platforms where an Perl 4.x is still
18630   installed as `perl`).
18631
18632   *Matthias Loepfe <Matthias.Loepfe@adnovum.ch>*
18633
18634 * Let util/clean-depend.pl work also with older Perl 5.00x versions.
18635
18636   *Matthias Loepfe <Matthias.Loepfe@adnovum.ch>*
18637
18638 * Fix Makefile.org so CC,CFLAG etc are passed to 'make links' add
18639   advapi32.lib to Win32 build and change the pem test comparison
18640   to fc.exe (thanks to Ulrich Kroener <kroneru@yahoo.com> for the
18641   suggestion). Fix misplaced ASNI prototypes and declarations in evp.h
18642   and crypto/des/ede_cbcm_enc.c.
18643
18644   *Steve Henson*
18645
18646 * DES quad checksum was broken on big-endian architectures. Fixed.
18647
18648   *Ben Laurie*
18649
18650 * Comment out two functions in bio.h that aren't implemented. Fix up the
18651   Win32 test batch file so it (might) work again. The Win32 test batch file
18652   is horrible: I feel ill....
18653
18654   *Steve Henson*
18655
18656 * Move various #ifdefs around so NO_SYSLOG, NO_DIRENT etc are now selected
18657   in e_os.h. Audit of header files to check ANSI and non ANSI
18658   sections: 10 functions were absent from non ANSI section and not exported
18659   from Windows DLLs. Fixed up libeay.num for new functions.
18660
18661   *Steve Henson*
18662
18663 * Make `openssl version` output lines consistent.
18664
18665   *Ralf S. Engelschall*
18666
18667 * Fix Win32 symbol export lists for BIO functions: Added
18668   BIO_get_ex_new_index, BIO_get_ex_num, BIO_get_ex_data and BIO_set_ex_data
18669   to ms/libeay{16,32}.def.
18670
18671   *Ralf S. Engelschall*
18672
18673 * Second round of fixing the OpenSSL perl/ stuff. It now at least compiled
18674   fine under Unix and passes some trivial tests I've now added. But the
18675   whole stuff is horribly incomplete, so a README.1ST with a disclaimer was
18676   added to make sure no one expects that this stuff really works in the
18677   OpenSSL 0.9.2 release.  Additionally I've started to clean the XS sources
18678   up and fixed a few little bugs and inconsistencies in OpenSSL.{pm,xs} and
18679   openssl_bio.xs.
18680
18681   *Ralf S. Engelschall*
18682
18683 * Fix the generation of two part addresses in perl.
18684
18685   *Kenji Miyake <kenji@miyake.org>, integrated by Ben Laurie*
18686
18687 * Add config entry for Linux on MIPS.
18688
18689   *John Tobey <jtobey@channel1.com>*
18690
18691 * Make links whenever Configure is run, unless we are on Windoze.
18692
18693   *Ben Laurie*
18694
18695 * Permit extensions to be added to CRLs using crl_section in openssl.cnf.
18696   Currently only issuerAltName and AuthorityKeyIdentifier make any sense
18697   in CRLs.
18698
18699   *Steve Henson*
18700
18701 * Add a useful kludge to allow package maintainers to specify compiler and
18702   other platforms details on the command line without having to patch the
18703   Configure script every time: One now can use
18704   `perl Configure <id>:<details>`,
18705   i.e. platform ids are allowed to have details appended
18706   to them (separated by colons). This is treated as there would be a static
18707   pre-configured entry in Configure's %table under key `<id>` with value
18708   `<details>` and `perl Configure <id>` is called.  So, when you want to
18709   perform a quick test-compile under FreeBSD 3.1 with pgcc and without
18710   assembler stuff you can use `perl Configure "FreeBSD-elf:pgcc:-O6:::"`
18711   now, which overrides the FreeBSD-elf entry on-the-fly.
18712
18713   *Ralf S. Engelschall*
18714
18715 * Disable new TLS1 ciphersuites by default: they aren't official yet.
18716
18717   *Ben Laurie*
18718
18719 * Allow DSO flags like -fpic, -fPIC, -KPIC etc. to be specified
18720   on the `perl Configure ...` command line. This way one can compile
18721   OpenSSL libraries with Position Independent Code (PIC) which is needed
18722   for linking it into DSOs.
18723
18724   *Ralf S. Engelschall*
18725
18726 * Remarkably, export ciphers were totally broken and no-one had noticed!
18727   Fixed.
18728
18729   *Ben Laurie*
18730
18731 * Cleaned up the LICENSE document: The official contact for any license
18732   questions now is the OpenSSL core team under openssl-core@openssl.org.
18733   And add a paragraph about the dual-license situation to make sure people
18734   recognize that _BOTH_ the OpenSSL license _AND_ the SSLeay license apply
18735   to the OpenSSL toolkit.
18736
18737   *Ralf S. Engelschall*
18738
18739 * General source tree makefile cleanups: Made `making xxx in yyy...`
18740   display consistent in the source tree and replaced `/bin/rm` by `rm`.
18741   Additionally cleaned up the `make links` target: Remove unnecessary
18742   semicolons, subsequent redundant removes, inline point.sh into mklink.sh
18743   to speed processing and no longer clutter the display with confusing
18744   stuff. Instead only the actually done links are displayed.
18745
18746   *Ralf S. Engelschall*
18747
18748 * Permit null encryption ciphersuites, used for authentication only. It used
18749   to be necessary to set the preprocessor define SSL_ALLOW_ENULL to do this.
18750   It is now necessary to set SSL_FORBID_ENULL to prevent the use of null
18751   encryption.
18752
18753   *Ben Laurie*
18754
18755 * Add a bunch of fixes to the PKCS#7 stuff. It used to sometimes reorder
18756   signed attributes when verifying signatures (this would break them),
18757   the detached data encoding was wrong and public keys obtained using
18758   X509_get_pubkey() weren't freed.
18759
18760   *Steve Henson*
18761
18762 * Add text documentation for the BUFFER functions. Also added a work around
18763   to a Win95 console bug. This was triggered by the password read stuff: the
18764   last character typed gets carried over to the next fread(). If you were
18765   generating a new cert request using 'req' for example then the last
18766   character of the passphrase would be CR which would then enter the first
18767   field as blank.
18768
18769   *Steve Henson*
18770
18771 * Added the new 'Includes OpenSSL Cryptography Software' button as
18772   doc/openssl_button.{gif,html} which is similar in style to the old SSLeay
18773   button and can be used by applications based on OpenSSL to show the
18774   relationship to the OpenSSL project.
18775
18776   *Ralf S. Engelschall*
18777
18778 * Remove confusing variables in function signatures in files
18779   ssl/ssl_lib.c and ssl/ssl.h.
18780
18781   *Lennart Bong <lob@kulthea.stacken.kth.se>*
18782
18783 * Don't install bss_file.c under PREFIX/include/
18784
18785   *Lennart Bong <lob@kulthea.stacken.kth.se>*
18786
18787 * Get the Win32 compile working again. Modify mkdef.pl so it can handle
18788   functions that return function pointers and has support for NT specific
18789   stuff. Fix mk1mf.pl and VC-32.pl to support NT differences also. Various
18790   #ifdef WIN32 and WINNTs sprinkled about the place and some changes from
18791   unsigned to signed types: this was killing the Win32 compile.
18792
18793   *Steve Henson*
18794
18795 * Add new certificate file to stack functions,
18796   SSL_add_dir_cert_subjects_to_stack() and
18797   SSL_add_file_cert_subjects_to_stack().  These largely supplant
18798   SSL_load_client_CA_file(), and can be used to add multiple certs easily
18799   to a stack (usually this is then handed to SSL_CTX_set_client_CA_list()).
18800   This means that Apache-SSL and similar packages don't have to mess around
18801   to add as many CAs as they want to the preferred list.
18802
18803   *Ben Laurie*
18804
18805 * Experiment with doxygen documentation. Currently only partially applied to
18806   ssl/ssl_lib.c.
18807   See <http://www.stack.nl/~dimitri/doxygen/index.html>, and run doxygen with
18808   openssl.doxy as the configuration file.
18809
18810   *Ben Laurie*
18811
18812 * Get rid of remaining C++-style comments which strict C compilers hate.
18813
18814   *Ralf S. Engelschall, pointed out by Carlos Amengual*
18815
18816 * Changed BN_RECURSION in bn_mont.c to BN_RECURSION_MONT so it is not
18817   compiled in by default: it has problems with large keys.
18818
18819   *Steve Henson*
18820
18821 * Add a bunch of SSL_xxx() functions for configuring the temporary RSA and
18822   DH private keys and/or callback functions which directly correspond to
18823   their SSL_CTX_xxx() counterparts but work on a per-connection basis. This
18824   is needed for applications which have to configure certificates on a
18825   per-connection basis (e.g. Apache+mod_ssl) instead of a per-context basis
18826   (e.g. s_server).
18827      For the RSA certificate situation is makes no difference, but
18828   for the DSA certificate situation this fixes the "no shared cipher"
18829   problem where the OpenSSL cipher selection procedure failed because the
18830   temporary keys were not overtaken from the context and the API provided
18831   no way to reconfigure them.
18832      The new functions now let applications reconfigure the stuff and they
18833   are in detail: SSL_need_tmp_RSA, SSL_set_tmp_rsa, SSL_set_tmp_dh,
18834   SSL_set_tmp_rsa_callback and SSL_set_tmp_dh_callback.  Additionally a new
18835   non-public-API function ssl_cert_instantiate() is used as a helper
18836   function and also to reduce code redundancy inside ssl_rsa.c.
18837
18838   *Ralf S. Engelschall*
18839
18840 * Move s_server -dcert and -dkey options out of the undocumented feature
18841   area because they are useful for the DSA situation and should be
18842   recognized by the users.
18843
18844   *Ralf S. Engelschall*
18845
18846 * Fix the cipher decision scheme for export ciphers: the export bits are
18847   *not* within SSL_MKEY_MASK or SSL_AUTH_MASK, they are within
18848   SSL_EXP_MASK.  So, the original variable has to be used instead of the
18849   already masked variable.
18850
18851   *Richard Levitte <levitte@stacken.kth.se>*
18852
18853 * Fix `port` variable from `int` to `unsigned int` in crypto/bio/b_sock.c
18854
18855   *Richard Levitte <levitte@stacken.kth.se>*
18856
18857 * Change type of another md_len variable in pk7_doit.c:PKCS7_dataFinal()
18858   from `int` to `unsigned int` because it is a length and initialized by
18859   EVP_DigestFinal() which expects an `unsigned int *`.
18860
18861   *Richard Levitte <levitte@stacken.kth.se>*
18862
18863 * Don't hard-code path to Perl interpreter on shebang line of Configure
18864   script. Instead use the usual Shell->Perl transition trick.
18865
18866   *Ralf S. Engelschall*
18867
18868 * Make `openssl x509 -noout -modulus`' functional also for DSA certificates
18869   (in addition to RSA certificates) to match the behaviour of `openssl dsa
18870   -noout -modulus` as it's already the case for `openssl rsa -noout
18871   -modulus`.  For RSA the -modulus is the real "modulus" while for DSA
18872   currently the public key is printed (a decision which was already done by
18873   `openssl dsa -modulus` in the past) which serves a similar purpose.
18874   Additionally the NO_RSA no longer completely removes the whole -modulus
18875   option; it now only avoids using the RSA stuff. Same applies to NO_DSA
18876   now, too.
18877
18878   *Ralf S.  Engelschall*
18879
18880 * Add Arne Ansper's reliable BIO - this is an encrypted, block-digested
18881   BIO. See the source (crypto/evp/bio_ok.c) for more info.
18882
18883   *Arne Ansper <arne@ats.cyber.ee>*
18884
18885 * Dump the old yucky req code that tried (and failed) to allow raw OIDs
18886   to be added. Now both 'req' and 'ca' can use new objects defined in the
18887   config file.
18888
18889   *Steve Henson*
18890
18891 * Add cool BIO that does syslog (or event log on NT).
18892
18893   *Arne Ansper <arne@ats.cyber.ee>, integrated by Ben Laurie*
18894
18895 * Add support for new TLS ciphersuites, TLS_RSA_EXPORT56_WITH_RC4_56_MD5,
18896   TLS_RSA_EXPORT56_WITH_RC2_CBC_56_MD5 and
18897   TLS_RSA_EXPORT56_WITH_DES_CBC_SHA, as specified in "56-bit Export Cipher
18898   Suites For TLS", draft-ietf-tls-56-bit-ciphersuites-00.txt.
18899
18900   *Ben Laurie*
18901
18902 * Add preliminary config info for new extension code.
18903
18904   *Steve Henson*
18905
18906 * Make RSA_NO_PADDING really use no padding.
18907
18908   *Ulf Moeller <ulf@fitug.de>*
18909
18910 * Generate errors when private/public key check is done.
18911
18912   *Ben Laurie*
18913
18914 * Overhaul for 'crl' utility. New function X509_CRL_print. Partial support
18915   for some CRL extensions and new objects added.
18916
18917   *Steve Henson*
18918
18919 * Really fix the ASN1 IMPLICIT bug this time... Partial support for private
18920   key usage extension and fuller support for authority key id.
18921
18922   *Steve Henson*
18923
18924 * Add OAEP encryption for the OpenSSL crypto library. OAEP is the improved
18925   padding method for RSA, which is recommended for new applications in PKCS
18926   #1 v2.0 (RFC 2437, October 1998).
18927   OAEP (Optimal Asymmetric Encryption Padding) has better theoretical
18928   foundations than the ad-hoc padding used in PKCS #1 v1.5. It is secure
18929   against Bleichbacher's attack on RSA.
18930   *Ulf Moeller <ulf@fitug.de>, reformatted, corrected and integrated by
18931   Ben Laurie*
18932
18933 * Updates to the new SSL compression code
18934
18935   *Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)*
18936
18937 * Fix so that the version number in the master secret, when passed
18938   via RSA, checks that if TLS was proposed, but we roll back to SSLv3
18939   (because the server will not accept higher), that the version number
18940   is 0x03,0x01, not 0x03,0x00
18941
18942   *Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)*
18943
18944 * Run extensive memory leak checks on SSL commands. Fixed *lots* of memory
18945   leaks in `ssl/` relating to new `X509_get_pubkey()` behaviour. Also fixes
18946   in `apps/` and an unrelated leak in `crypto/dsa/dsa_vrf.c`.
18947
18948   *Steve Henson*
18949
18950 * Support for RAW extensions where an arbitrary extension can be
18951   created by including its DER encoding. See `apps/openssl.cnf` for
18952   an example.
18953
18954   *Steve Henson*
18955
18956 * Make sure latest Perl versions don't interpret some generated C array
18957   code as Perl array code in the crypto/err/err_genc.pl script.
18958
18959   *Lars Weber <3weber@informatik.uni-hamburg.de>*
18960
18961 * Modify ms/do_ms.bat to not generate assembly language makefiles since
18962   not many people have the assembler. Various Win32 compilation fixes and
18963   update to the INSTALL.W32 file with (hopefully) more accurate Win32
18964   build instructions.
18965
18966   *Steve Henson*
18967
18968 * Modify configure script 'Configure' to automatically create crypto/date.h
18969   file under Win32 and also build pem.h from pem.org. New script
18970   util/mkfiles.pl to create the MINFO file on environments that can't do a
18971   'make files': perl util/mkfiles.pl >MINFO should work.
18972
18973   *Steve Henson*
18974
18975 * Major rework of DES function declarations, in the pursuit of correctness
18976   and purity. As a result, many evil casts evaporated, and some weirdness,
18977   too. You may find this causes warnings in your code. Zapping your evil
18978   casts will probably fix them. Mostly.
18979
18980   *Ben Laurie*
18981
18982 * Fix for a typo in asn1.h. Bug fix to object creation script
18983   obj_dat.pl. It considered a zero in an object definition to mean
18984   "end of object": none of the objects in objects.h have any zeros
18985   so it wasn't spotted.
18986
18987   *Steve Henson, reported by Erwann ABALEA <eabalea@certplus.com>*
18988
18989 * Add support for Triple DES Cipher Block Chaining with Output Feedback
18990   Masking (CBCM). In the absence of test vectors, the best I have been able
18991   to do is check that the decrypt undoes the encrypt, so far. Send me test
18992   vectors if you have them.
18993
18994   *Ben Laurie*
18995
18996 * Correct calculation of key length for export ciphers (too much space was
18997   allocated for null ciphers). This has not been tested!
18998
18999   *Ben Laurie*
19000
19001 * Modifications to the mkdef.pl for Win32 DEF file creation. The usage
19002   message is now correct (it understands "crypto" and "ssl" on its
19003   command line). There is also now an "update" option. This will update
19004   the util/ssleay.num and util/libeay.num files with any new functions.
19005   If you do a:
19006   perl util/mkdef.pl crypto ssl update
19007   it will update them.
19008
19009   *Steve Henson*
19010
19011 * Overhauled the Perl interface:
19012   - ported BN stuff to OpenSSL's different BN library
19013   - made the perl/ source tree CVS-aware
19014   - renamed the package from SSLeay to OpenSSL (the files still contain
19015     their history because I've copied them in the repository)
19016   - removed obsolete files (the test scripts will be replaced
19017     by better Test::Harness variants in the future)
19018
19019   *Ralf S. Engelschall*
19020
19021 * First cut for a very conservative source tree cleanup:
19022   1. merge various obsolete readme texts into doc/ssleay.txt
19023   where we collect the old documents and readme texts.
19024   2. remove the first part of files where I'm already sure that we no
19025   longer need them because of three reasons: either they are just temporary
19026   files which were left by Eric or they are preserved original files where
19027   I've verified that the diff is also available in the CVS via "cvs diff
19028   -rSSLeay_0_8_1b" or they were renamed (as it was definitely the case for
19029   the crypto/md/ stuff).
19030
19031   *Ralf S. Engelschall*
19032
19033 * More extension code. Incomplete support for subject and issuer alt
19034   name, issuer and authority key id. Change the i2v function parameters
19035   and add an extra 'crl' parameter in the X509V3_CTX structure: guess
19036   what that's for :-) Fix to ASN1 macro which messed up
19037   IMPLICIT tag and add f_enum.c which adds a2i, i2a for ENUMERATED.
19038
19039   *Steve Henson*
19040
19041 * Preliminary support for ENUMERATED type. This is largely copied from the
19042   INTEGER code.
19043
19044   *Steve Henson*
19045
19046 * Add new function, EVP_MD_CTX_copy() to replace frequent use of memcpy.
19047
19048   *Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)*
19049
19050 * Make sure `make rehash` target really finds the `openssl` program.
19051
19052   *Ralf S. Engelschall, Matthias Loepfe <Matthias.Loepfe@adnovum.ch>*
19053
19054 * Squeeze another 7% of speed out of MD5 assembler, at least on a P2. I'd
19055   like to hear about it if this slows down other processors.
19056
19057   *Ben Laurie*
19058
19059 * Add CygWin32 platform information to Configure script.
19060
19061   *Alan Batie <batie@aahz.jf.intel.com>*
19062
19063 * Fixed ms/32all.bat script: `no_asm` -> `no-asm`
19064
19065   *Rainer W. Gerling <gerling@mpg-gv.mpg.de>*
19066
19067 * New program nseq to manipulate netscape certificate sequences
19068
19069   *Steve Henson*
19070
19071 * Modify crl2pkcs7 so it supports multiple -certfile arguments. Fix a
19072   few typos.
19073
19074   *Steve Henson*
19075
19076 * Fixes to BN code.  Previously the default was to define BN_RECURSION
19077   but the BN code had some problems that would cause failures when
19078   doing certificate verification and some other functions.
19079
19080   *Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)*
19081
19082 * Add ASN1 and PEM code to support netscape certificate sequences.
19083
19084   *Steve Henson*
19085
19086 * Add ASN1 and PEM code to support netscape certificate sequences.
19087
19088   *Steve Henson*
19089
19090 * Add several PKIX and private extended key usage OIDs.
19091
19092   *Steve Henson*
19093
19094 * Modify the 'ca' program to handle the new extension code. Modify
19095   openssl.cnf for new extension format, add comments.
19096
19097   *Steve Henson*
19098
19099 * More X509 V3 changes. Fix typo in v3_bitstr.c. Add support to 'req'
19100   and add a sample to openssl.cnf so req -x509 now adds appropriate
19101   CA extensions.
19102
19103   *Steve Henson*
19104
19105 * Continued X509 V3 changes. Add to other makefiles, integrate with the
19106   error code, add initial support to X509_print() and x509 application.
19107
19108   *Steve Henson*
19109
19110 * Takes a deep breath and start adding X509 V3 extension support code. Add
19111   files in crypto/x509v3. Move original stuff to crypto/x509v3/old. All this
19112   stuff is currently isolated and isn't even compiled yet.
19113
19114   *Steve Henson*
19115
19116 * Continuing patches for GeneralizedTime. Fix up certificate and CRL
19117   ASN1 to use ASN1_TIME and modify print routines to use ASN1_TIME_print.
19118   Removed the versions check from X509 routines when loading extensions:
19119   this allows certain broken certificates that don't set the version
19120   properly to be processed.
19121
19122   *Steve Henson*
19123
19124 * Deal with irritating shit to do with dependencies, in YAAHW (Yet Another
19125   Ad Hoc Way) - Makefile.ssls now all contain local dependencies, which
19126   can still be regenerated with "make depend".
19127
19128   *Ben Laurie*
19129
19130 * Spelling mistake in C version of CAST-128.
19131
19132   *Ben Laurie, reported by Jeremy Hylton <jeremy@cnri.reston.va.us>*
19133
19134 * Changes to the error generation code. The perl script err-code.pl
19135   now reads in the old error codes and retains the old numbers, only
19136   adding new ones if necessary. It also only changes the .err files if new
19137   codes are added. The makefiles have been modified to only insert errors
19138   when needed (to avoid needlessly modifying header files). This is done
19139   by only inserting errors if the .err file is newer than the auto generated
19140   C file. To rebuild all the error codes from scratch (the old behaviour)
19141   either modify crypto/Makefile.ssl to pass the -regen flag to err_code.pl
19142   or delete all the .err files.
19143
19144   *Steve Henson*
19145
19146 * CAST-128 was incorrectly implemented for short keys. The C version has
19147   been fixed, but is untested. The assembler versions are also fixed, but
19148   new assembler HAS NOT BEEN GENERATED FOR WIN32 - the Makefile needs fixing
19149   to regenerate it if needed.
19150   *Ben Laurie, reported (with fix for C version) by Jun-ichiro itojun
19151    Hagino <itojun@kame.net>*
19152
19153 * File was opened incorrectly in randfile.c.
19154
19155   *Ulf Möller <ulf@fitug.de>*
19156
19157 * Beginning of support for GeneralizedTime. d2i, i2d, check and print
19158   functions. Also ASN1_TIME suite which is a CHOICE of UTCTime or
19159   GeneralizedTime. ASN1_TIME is the proper type used in certificates et
19160   al: it's just almost always a UTCTime. Note this patch adds new error
19161   codes so do a "make errors" if there are problems.
19162
19163   *Steve Henson*
19164
19165 * Correct Linux 1 recognition in config.
19166
19167   *Ulf Möller <ulf@fitug.de>*
19168
19169 * Remove pointless MD5 hash when using DSA keys in ca.
19170
19171   *Anonymous <nobody@replay.com>*
19172
19173 * Generate an error if given an empty string as a cert directory. Also
19174   generate an error if handed NULL (previously returned 0 to indicate an
19175   error, but didn't set one).
19176
19177   *Ben Laurie, reported by Anonymous <nobody@replay.com>*
19178
19179 * Add prototypes to SSL methods. Make SSL_write's buffer const, at last.
19180
19181   *Ben Laurie*
19182
19183 * Fix the dummy function BN_ref_mod_exp() in rsaref.c to have the correct
19184   parameters. This was causing a warning which killed off the Win32 compile.
19185
19186   *Steve Henson*
19187
19188 * Remove C++ style comments from crypto/bn/bn_local.h.
19189
19190   *Neil Costigan <neil.costigan@celocom.com>*
19191
19192 * The function OBJ_txt2nid was broken. It was supposed to return a nid
19193   based on a text string, looking up short and long names and finally
19194   "dot" format. The "dot" format stuff didn't work. Added new function
19195   OBJ_txt2obj to do the same but return an ASN1_OBJECT and rewrote
19196   OBJ_txt2nid to use it. OBJ_txt2obj can also return objects even if the
19197   OID is not part of the table.
19198
19199   *Steve Henson*
19200
19201 * Add prototypes to X509 lookup/verify methods, fixing a bug in
19202   X509_LOOKUP_by_alias().
19203
19204   *Ben Laurie*
19205
19206 * Sort openssl functions by name.
19207
19208   *Ben Laurie*
19209
19210 * Get the `gendsa` command working and add it to the `list` command. Remove
19211   encryption from sample DSA keys (in case anyone is interested the password
19212   was "1234").
19213
19214   *Steve Henson*
19215
19216 * Make *all* `*_free` functions accept a NULL pointer.
19217
19218   *Frans Heymans <fheymans@isaserver.be>*
19219
19220 * If a DH key is generated in s3_srvr.c, don't blow it by trying to use
19221   NULL pointers.
19222
19223   *Anonymous <nobody@replay.com>*
19224
19225 * s_server should send the CAfile as acceptable CAs, not its own cert.
19226
19227   *Bodo Moeller <3moeller@informatik.uni-hamburg.de>*
19228
19229 * Don't blow it for numeric `-newkey` arguments to `apps/req`.
19230
19231   *Bodo Moeller <3moeller@informatik.uni-hamburg.de>*
19232
19233 * Temp key "for export" tests were wrong in s3_srvr.c.
19234
19235   *Anonymous <nobody@replay.com>*
19236
19237 * Add prototype for temp key callback functions
19238   SSL_CTX_set_tmp_{rsa,dh}_callback().
19239
19240   *Ben Laurie*
19241
19242 * Make DH_free() tolerate being passed a NULL pointer (like RSA_free() and
19243   DSA_free()). Make X509_PUBKEY_set() check for errors in d2i_PublicKey().
19244
19245   *Steve Henson*
19246
19247 * X509_name_add_entry() freed the wrong thing after an error.
19248
19249   *Arne Ansper <arne@ats.cyber.ee>*
19250
19251 * rsa_eay.c would attempt to free a NULL context.
19252
19253   *Arne Ansper <arne@ats.cyber.ee>*
19254
19255 * BIO_s_socket() had a broken should_retry() on Windoze.
19256
19257   *Arne Ansper <arne@ats.cyber.ee>*
19258
19259 * BIO_f_buffer() didn't pass on BIO_CTRL_FLUSH.
19260
19261   *Arne Ansper <arne@ats.cyber.ee>*
19262
19263 * Make sure the already existing X509_STORE->depth variable is initialized
19264   in X509_STORE_new(), but document the fact that this variable is still
19265   unused in the certificate verification process.
19266
19267   *Ralf S. Engelschall*
19268
19269 * Fix the various library and `apps/` files to free up pkeys obtained from
19270   X509_PUBKEY_get() et al. Also allow x509.c to handle netscape extensions.
19271
19272   *Steve Henson*
19273
19274 * Fix reference counting in X509_PUBKEY_get(). This makes
19275   demos/maurice/example2.c work, amongst others, probably.
19276
19277   *Steve Henson and Ben Laurie*
19278
19279 * First cut of a cleanup for `apps/`. First the `ssleay` program is now named
19280   `openssl` and second, the shortcut symlinks for the `openssl <command>`
19281   are no longer created. This way we have a single and consistent command
19282   line interface `openssl <command>`, similar to `cvs <command>`.
19283
19284   *Ralf S. Engelschall, Paul Sutton and Ben Laurie*
19285
19286 * ca.c: move test for DSA keys inside #ifndef NO_DSA. Make pubkey
19287   BIT STRING wrapper always have zero unused bits.
19288
19289   *Steve Henson*
19290
19291 * Add CA.pl, perl version of CA.sh, add extended key usage OID.
19292
19293   *Steve Henson*
19294
19295 * Make the top-level INSTALL documentation easier to understand.
19296
19297   *Paul Sutton*
19298
19299 * Makefiles updated to exit if an error occurs in a sub-directory
19300   make (including if user presses ^C) [Paul Sutton]
19301
19302 * Make Montgomery context stuff explicit in RSA data structure.
19303
19304   *Ben Laurie*
19305
19306 * Fix build order of pem and err to allow for generated pem.h.
19307
19308   *Ben Laurie*
19309
19310 * Fix renumbering bug in X509_NAME_delete_entry().
19311
19312   *Ben Laurie*
19313
19314 * Enhanced the err-ins.pl script so it makes the error library number
19315   global and can add a library name. This is needed for external ASN1 and
19316   other error libraries.
19317
19318   *Steve Henson*
19319
19320 * Fixed sk_insert which never worked properly.
19321
19322   *Steve Henson*
19323
19324 * Fix ASN1 macros so they can handle indefinite length constructed
19325   EXPLICIT tags. Some non standard certificates use these: they can now
19326   be read in.
19327
19328   *Steve Henson*
19329
19330 * Merged the various old/obsolete SSLeay documentation files (doc/xxx.doc)
19331   into a single doc/ssleay.txt bundle. This way the information is still
19332   preserved but no longer messes up this directory. Now it's new room for
19333   the new set of documentation files.
19334
19335   *Ralf S. Engelschall*
19336
19337 * SETs were incorrectly DER encoded. This was a major pain, because they
19338   shared code with SEQUENCEs, which aren't coded the same. This means that
19339   almost everything to do with SETs or SEQUENCEs has either changed name or
19340   number of arguments.
19341
19342   *Ben Laurie, based on a partial fix by GP Jayan <gp@nsj.co.jp>*
19343
19344 * Fix test data to work with the above.
19345
19346   *Ben Laurie*
19347
19348 * Fix the RSA header declarations that hid a bug I fixed in 0.9.0b but
19349   was already fixed by Eric for 0.9.1 it seems.
19350
19351   *Ben Laurie - pointed out by Ulf Möller <ulf@fitug.de>*
19352
19353 * Autodetect FreeBSD3.
19354
19355   *Ben Laurie*
19356
19357 * Fix various bugs in Configure. This affects the following platforms:
19358   nextstep
19359   ncr-scde
19360   unixware-2.0
19361   unixware-2.0-pentium
19362   sco5-cc.
19363
19364   *Ben Laurie*
19365
19366 * Eliminate generated files from CVS. Reorder tests to regenerate files
19367   before they are needed.
19368
19369   *Ben Laurie*
19370
19371 * Generate Makefile.ssl from Makefile.org (to keep CVS happy).
19372
19373   *Ben Laurie*
19374
19375### Changes between 0.9.1b and 0.9.1c  [23-Dec-1998]
19376
19377 * Added OPENSSL_VERSION_NUMBER to crypto/crypto.h and
19378   changed SSLeay to OpenSSL in version strings.
19379
19380   *Ralf S. Engelschall*
19381
19382 * Some fixups to the top-level documents.
19383
19384   *Paul Sutton*
19385
19386 * Fixed the nasty bug where rsaref.h was not found under compile-time
19387   because the symlink to include/ was missing.
19388
19389   *Ralf S. Engelschall*
19390
19391 * Incorporated the popular no-RSA/DSA-only patches
19392   which allow to compile an RSA-free SSLeay.
19393
19394   *Andrew Cooke / Interrader Ldt., Ralf S. Engelschall*
19395
19396 * Fixed nasty rehash problem under `make -f Makefile.ssl links`
19397   when "ssleay" is still not found.
19398
19399   *Ralf S. Engelschall*
19400
19401 * Added more platforms to Configure: Cray T3E, HPUX 11,
19402
19403   *Ralf S. Engelschall, Beckmann <beckman@acl.lanl.gov>*
19404
19405 * Updated the README file.
19406
19407   *Ralf S. Engelschall*
19408
19409 * Added various .cvsignore files in the CVS repository subdirs
19410   to make a "cvs update" really silent.
19411
19412   *Ralf S. Engelschall*
19413
19414 * Recompiled the error-definition header files and added
19415   missing symbols to the Win32 linker tables.
19416
19417   *Ralf S. Engelschall*
19418
19419 * Cleaned up the top-level documents;
19420   o new files: CHANGES and LICENSE
19421   o merged VERSION, HISTORY* and README* files a CHANGES.SSLeay
19422   o merged COPYRIGHT into LICENSE
19423   o removed obsolete TODO file
19424   o renamed MICROSOFT to INSTALL.W32
19425
19426   *Ralf S. Engelschall*
19427
19428 * Removed dummy files from the 0.9.1b source tree:
19429   crypto/asn1/x crypto/bio/cd crypto/bio/fg crypto/bio/grep crypto/bio/vi
19430   crypto/bn/asm/......add.c crypto/bn/asm/a.out crypto/dsa/f crypto/md5/f
19431   crypto/pem/gmon.out crypto/perlasm/f crypto/pkcs7/build crypto/rsa/f
19432   crypto/sha/asm/f crypto/threads/f ms/zzz ssl/f ssl/f.mak test/f
19433   util/f.mak util/pl/f util/pl/f.mak crypto/bf/bf_locl.old apps/f
19434
19435   *Ralf S. Engelschall*
19436
19437 * Added various platform portability fixes.
19438
19439   *Mark J. Cox*
19440
19441 * The Genesis of the OpenSSL rpject:
19442   We start with the latest (unreleased) SSLeay version 0.9.1b which Eric A.
19443   Young and Tim J. Hudson created while they were working for C2Net until
19444   summer 1998.
19445
19446   *The OpenSSL Project*
19447
19448### Changes between 0.9.0b and 0.9.1b  [not released]
19449
19450 * Updated a few CA certificates under certs/
19451
19452   *Eric A. Young*
19453
19454 * Changed some BIGNUM api stuff.
19455
19456   *Eric A. Young*
19457
19458 * Various platform ports: OpenBSD, Ultrix, IRIX 64bit, NetBSD,
19459   DGUX x86, Linux Alpha, etc.
19460
19461   *Eric A. Young*
19462
19463 * New COMP library [crypto/comp/] for SSL Record Layer Compression:
19464   RLE (dummy implemented) and ZLIB (really implemented when ZLIB is
19465   available).
19466
19467   *Eric A. Young*
19468
19469 * Add -strparse option to asn1pars program which parses nested
19470   binary structures
19471
19472   *Dr Stephen Henson <shenson@bigfoot.com>*
19473
19474 * Added "oid_file" to ssleay.cnf for "ca" and "req" programs.
19475
19476   *Eric A. Young*
19477
19478 * DSA fix for "ca" program.
19479
19480   *Eric A. Young*
19481
19482 * Added "-genkey" option to "dsaparam" program.
19483
19484   *Eric A. Young*
19485
19486 * Added RIPE MD160 (rmd160) message digest.
19487
19488   *Eric A. Young*
19489
19490 * Added -a (all) option to "ssleay version" command.
19491
19492   *Eric A. Young*
19493
19494 * Added PLATFORM define which is the id given to Configure.
19495
19496   *Eric A. Young*
19497
19498 * Added MemCheck_XXXX functions to crypto/mem.c for memory checking.
19499
19500   *Eric A. Young*
19501
19502 * Extended the ASN.1 parser routines.
19503
19504   *Eric A. Young*
19505
19506 * Extended BIO routines to support REUSEADDR, seek, tell, etc.
19507
19508   *Eric A. Young*
19509
19510 * Added a BN_CTX to the BN library.
19511
19512   *Eric A. Young*
19513
19514 * Fixed the weak key values in DES library
19515
19516   *Eric A. Young*
19517
19518 * Changed API in EVP library for cipher aliases.
19519
19520   *Eric A. Young*
19521
19522 * Added support for RC2/64bit cipher.
19523
19524   *Eric A. Young*
19525
19526 * Converted the lhash library to the crypto/mem.c functions.
19527
19528   *Eric A. Young*
19529
19530 * Added more recognized ASN.1 object ids.
19531
19532   *Eric A. Young*
19533
19534 * Added more RSA padding checks for SSL/TLS.
19535
19536   *Eric A. Young*
19537
19538 * Added BIO proxy/filter functionality.
19539
19540   *Eric A. Young*
19541
19542 * Added extra_certs to SSL_CTX which can be used
19543   send extra CA certificates to the client in the CA cert chain sending
19544   process. It can be configured with SSL_CTX_add_extra_chain_cert().
19545
19546   *Eric A. Young*
19547
19548 * Now Fortezza is denied in the authentication phase because
19549   this is key exchange mechanism is not supported by SSLeay at all.
19550
19551   *Eric A. Young*
19552
19553 * Additional PKCS1 checks.
19554
19555   *Eric A. Young*
19556
19557 * Support the string "TLSv1" for all TLS v1 ciphers.
19558
19559   *Eric A. Young*
19560
19561 * Added function SSL_get_ex_data_X509_STORE_CTX_idx() which gives the
19562   ex_data index of the SSL context in the X509_STORE_CTX ex_data.
19563
19564   *Eric A. Young*
19565
19566 * Fixed a few memory leaks.
19567
19568   *Eric A. Young*
19569
19570 * Fixed various code and comment typos.
19571
19572   *Eric A. Young*
19573
19574 * A minor bug in ssl/s3_clnt.c where there would always be 4 0
19575   bytes sent in the client random.
19576
19577   *Edward Bishop <ebishop@spyglass.com>*
19578
19579<!-- Links -->
19580
19581[CVE-2023-0401]: https://www.openssl.org/news/vulnerabilities.html#CVE-2023-0401
19582[CVE-2023-0286]: https://www.openssl.org/news/vulnerabilities.html#CVE-2023-0286
19583[CVE-2023-0217]: https://www.openssl.org/news/vulnerabilities.html#CVE-2023-0217
19584[CVE-2023-0216]: https://www.openssl.org/news/vulnerabilities.html#CVE-2023-0216
19585[CVE-2023-0215]: https://www.openssl.org/news/vulnerabilities.html#CVE-2023-0215
19586[CVE-2022-4450]: https://www.openssl.org/news/vulnerabilities.html#CVE-2022-4450
19587[CVE-2022-4304]: https://www.openssl.org/news/vulnerabilities.html#CVE-2022-4304
19588[CVE-2022-4203]: https://www.openssl.org/news/vulnerabilities.html#CVE-2022-4203
19589[CVE-2022-3996]: https://www.openssl.org/news/vulnerabilities.html#CVE-2022-3996
19590[CVE-2022-2274]: https://www.openssl.org/news/vulnerabilities.html#CVE-2022-2274
19591[CVE-2022-2097]: https://www.openssl.org/news/vulnerabilities.html#CVE-2022-2274
19592[CVE-2020-1971]: https://www.openssl.org/news/vulnerabilities.html#CVE-2020-1971
19593[CVE-2020-1967]: https://www.openssl.org/news/vulnerabilities.html#CVE-2020-1967
19594[CVE-2019-1563]: https://www.openssl.org/news/vulnerabilities.html#CVE-2019-1563
19595[CVE-2019-1559]: https://www.openssl.org/news/vulnerabilities.html#CVE-2019-1559
19596[CVE-2019-1552]: https://www.openssl.org/news/vulnerabilities.html#CVE-2019-1552
19597[CVE-2019-1551]: https://www.openssl.org/news/vulnerabilities.html#CVE-2019-1551
19598[CVE-2019-1549]: https://www.openssl.org/news/vulnerabilities.html#CVE-2019-1549
19599[CVE-2019-1547]: https://www.openssl.org/news/vulnerabilities.html#CVE-2019-1547
19600[CVE-2019-1543]: https://www.openssl.org/news/vulnerabilities.html#CVE-2019-1543
19601[CVE-2018-5407]: https://www.openssl.org/news/vulnerabilities.html#CVE-2018-5407
19602[CVE-2018-0739]: https://www.openssl.org/news/vulnerabilities.html#CVE-2018-0739
19603[CVE-2018-0737]: https://www.openssl.org/news/vulnerabilities.html#CVE-2018-0737
19604[CVE-2018-0735]: https://www.openssl.org/news/vulnerabilities.html#CVE-2018-0735
19605[CVE-2018-0734]: https://www.openssl.org/news/vulnerabilities.html#CVE-2018-0734
19606[CVE-2018-0733]: https://www.openssl.org/news/vulnerabilities.html#CVE-2018-0733
19607[CVE-2018-0732]: https://www.openssl.org/news/vulnerabilities.html#CVE-2018-0732
19608[CVE-2017-3738]: https://www.openssl.org/news/vulnerabilities.html#CVE-2017-3738
19609[CVE-2017-3737]: https://www.openssl.org/news/vulnerabilities.html#CVE-2017-3737
19610[CVE-2017-3736]: https://www.openssl.org/news/vulnerabilities.html#CVE-2017-3736
19611[CVE-2017-3735]: https://www.openssl.org/news/vulnerabilities.html#CVE-2017-3735
19612[CVE-2017-3733]: https://www.openssl.org/news/vulnerabilities.html#CVE-2017-3733
19613[CVE-2017-3732]: https://www.openssl.org/news/vulnerabilities.html#CVE-2017-3732
19614[CVE-2017-3731]: https://www.openssl.org/news/vulnerabilities.html#CVE-2017-3731
19615[CVE-2017-3730]: https://www.openssl.org/news/vulnerabilities.html#CVE-2017-3730
19616[CVE-2016-7055]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-7055
19617[CVE-2016-7054]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-7054
19618[CVE-2016-7053]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-7053
19619[CVE-2016-7052]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-7052
19620[CVE-2016-6309]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-6309
19621[CVE-2016-6308]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-6308
19622[CVE-2016-6307]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-6307
19623[CVE-2016-6306]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-6306
19624[CVE-2016-6305]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-6305
19625[CVE-2016-6304]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-6304
19626[CVE-2016-6303]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-6303
19627[CVE-2016-6302]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-6302
19628[CVE-2016-2183]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-2183
19629[CVE-2016-2182]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-2182
19630[CVE-2016-2181]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-2181
19631[CVE-2016-2180]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-2180
19632[CVE-2016-2179]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-2179
19633[CVE-2016-2178]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-2178
19634[CVE-2016-2177]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-2177
19635[CVE-2016-2176]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-2176
19636[CVE-2016-2109]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-2109
19637[CVE-2016-2107]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-2107
19638[CVE-2016-2106]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-2106
19639[CVE-2016-2105]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-2105
19640[CVE-2016-0800]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-0800
19641[CVE-2016-0799]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-0799
19642[CVE-2016-0798]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-0798
19643[CVE-2016-0797]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-0797
19644[CVE-2016-0705]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-0705
19645[CVE-2016-0702]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-0702
19646[CVE-2016-0701]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-0701
19647[CVE-2015-3197]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-3197
19648[CVE-2015-3196]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-3196
19649[CVE-2015-3195]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-3195
19650[CVE-2015-3194]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-3194
19651[CVE-2015-3193]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-3193
19652[CVE-2015-1793]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-1793
19653[CVE-2015-1792]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-1792
19654[CVE-2015-1791]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-1791
19655[CVE-2015-1790]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-1790
19656[CVE-2015-1789]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-1789
19657[CVE-2015-1788]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-1788
19658[CVE-2015-1787]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-1787
19659[CVE-2015-0293]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0293
19660[CVE-2015-0291]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0291
19661[CVE-2015-0290]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0290
19662[CVE-2015-0289]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0289
19663[CVE-2015-0288]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0288
19664[CVE-2015-0287]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0287
19665[CVE-2015-0286]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0286
19666[CVE-2015-0285]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0285
19667[CVE-2015-0209]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0209
19668[CVE-2015-0208]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0208
19669[CVE-2015-0207]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0207
19670[CVE-2015-0206]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0206
19671[CVE-2015-0205]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0205
19672[CVE-2015-0204]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0204
19673[CVE-2014-8275]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-8275
19674[CVE-2014-5139]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-5139
19675[CVE-2014-3572]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3572
19676[CVE-2014-3571]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3571
19677[CVE-2014-3570]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3570
19678[CVE-2014-3569]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3569
19679[CVE-2014-3568]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3568
19680[CVE-2014-3567]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3567
19681[CVE-2014-3566]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3566
19682[CVE-2014-3513]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3513
19683[CVE-2014-3512]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3512
19684[CVE-2014-3511]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3511
19685[CVE-2014-3510]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3510
19686[CVE-2014-3509]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3509
19687[CVE-2014-3508]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3508
19688[CVE-2014-3507]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3507
19689[CVE-2014-3506]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3506
19690[CVE-2014-3505]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3505
19691[CVE-2014-3470]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3470
19692[CVE-2014-0224]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-0224
19693[CVE-2014-0221]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-0221
19694[CVE-2014-0195]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-0195
19695[CVE-2014-0160]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-0160
19696[CVE-2014-0076]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-0076
19697[CVE-2013-6450]: https://www.openssl.org/news/vulnerabilities.html#CVE-2013-6450
19698[CVE-2013-4353]: https://www.openssl.org/news/vulnerabilities.html#CVE-2013-4353
19699[CVE-2013-0169]: https://www.openssl.org/news/vulnerabilities.html#CVE-2013-0169
19700[CVE-2013-0166]: https://www.openssl.org/news/vulnerabilities.html#CVE-2013-0166
19701[CVE-2012-2686]: https://www.openssl.org/news/vulnerabilities.html#CVE-2012-2686
19702[CVE-2012-2333]: https://www.openssl.org/news/vulnerabilities.html#CVE-2012-2333
19703[CVE-2012-2110]: https://www.openssl.org/news/vulnerabilities.html#CVE-2012-2110
19704[CVE-2012-0884]: https://www.openssl.org/news/vulnerabilities.html#CVE-2012-0884
19705[CVE-2012-0050]: https://www.openssl.org/news/vulnerabilities.html#CVE-2012-0050
19706[CVE-2012-0027]: https://www.openssl.org/news/vulnerabilities.html#CVE-2012-0027
19707[CVE-2011-4619]: https://www.openssl.org/news/vulnerabilities.html#CVE-2011-4619
19708[CVE-2011-4577]: https://www.openssl.org/news/vulnerabilities.html#CVE-2011-4577
19709[CVE-2011-4576]: https://www.openssl.org/news/vulnerabilities.html#CVE-2011-4576
19710[CVE-2011-4109]: https://www.openssl.org/news/vulnerabilities.html#CVE-2011-4109
19711[CVE-2011-4108]: https://www.openssl.org/news/vulnerabilities.html#CVE-2011-4108
19712[CVE-2011-3210]: https://www.openssl.org/news/vulnerabilities.html#CVE-2011-3210
19713[CVE-2011-3207]: https://www.openssl.org/news/vulnerabilities.html#CVE-2011-3207
19714[CVE-2011-0014]: https://www.openssl.org/news/vulnerabilities.html#CVE-2011-0014
19715[CVE-2010-4252]: https://www.openssl.org/news/vulnerabilities.html#CVE-2010-4252
19716[CVE-2010-4180]: https://www.openssl.org/news/vulnerabilities.html#CVE-2010-4180
19717[CVE-2010-3864]: https://www.openssl.org/news/vulnerabilities.html#CVE-2010-3864
19718[CVE-2010-1633]: https://www.openssl.org/news/vulnerabilities.html#CVE-2010-1633
19719[CVE-2010-0740]: https://www.openssl.org/news/vulnerabilities.html#CVE-2010-0740
19720[CVE-2010-0433]: https://www.openssl.org/news/vulnerabilities.html#CVE-2010-0433
19721[CVE-2009-4355]: https://www.openssl.org/news/vulnerabilities.html#CVE-2009-4355
19722[CVE-2009-3555]: https://www.openssl.org/news/vulnerabilities.html#CVE-2009-3555
19723[CVE-2009-3245]: https://www.openssl.org/news/vulnerabilities.html#CVE-2009-3245
19724[CVE-2009-1386]: https://www.openssl.org/news/vulnerabilities.html#CVE-2009-1386
19725[CVE-2009-1379]: https://www.openssl.org/news/vulnerabilities.html#CVE-2009-1379
19726[CVE-2009-1378]: https://www.openssl.org/news/vulnerabilities.html#CVE-2009-1378
19727[CVE-2009-1377]: https://www.openssl.org/news/vulnerabilities.html#CVE-2009-1377
19728[CVE-2009-0789]: https://www.openssl.org/news/vulnerabilities.html#CVE-2009-0789
19729[CVE-2009-0591]: https://www.openssl.org/news/vulnerabilities.html#CVE-2009-0591
19730[CVE-2009-0590]: https://www.openssl.org/news/vulnerabilities.html#CVE-2009-0590
19731[CVE-2008-5077]: https://www.openssl.org/news/vulnerabilities.html#CVE-2008-5077
19732[CVE-2008-1678]: https://www.openssl.org/news/vulnerabilities.html#CVE-2008-1678
19733[CVE-2008-1672]: https://www.openssl.org/news/vulnerabilities.html#CVE-2008-1672
19734[CVE-2008-0891]: https://www.openssl.org/news/vulnerabilities.html#CVE-2008-0891
19735[CVE-2007-5135]: https://www.openssl.org/news/vulnerabilities.html#CVE-2007-5135
19736[CVE-2007-4995]: https://www.openssl.org/news/vulnerabilities.html#CVE-2007-4995
19737[CVE-2006-4343]: https://www.openssl.org/news/vulnerabilities.html#CVE-2006-4343
19738[CVE-2006-4339]: https://www.openssl.org/news/vulnerabilities.html#CVE-2006-4339
19739[CVE-2006-3738]: https://www.openssl.org/news/vulnerabilities.html#CVE-2006-3738
19740[CVE-2006-2940]: https://www.openssl.org/news/vulnerabilities.html#CVE-2006-2940
19741[CVE-2006-2937]: https://www.openssl.org/news/vulnerabilities.html#CVE-2006-2937
19742[CVE-2005-2969]: https://www.openssl.org/news/vulnerabilities.html#CVE-2005-2969
19743[CVE-2004-0112]: https://www.openssl.org/news/vulnerabilities.html#CVE-2004-0112
19744[CVE-2004-0079]: https://www.openssl.org/news/vulnerabilities.html#CVE-2004-0079
19745[CVE-2003-0851]: https://www.openssl.org/news/vulnerabilities.html#CVE-2003-0851
19746[CVE-2003-0545]: https://www.openssl.org/news/vulnerabilities.html#CVE-2003-0545
19747[CVE-2003-0544]: https://www.openssl.org/news/vulnerabilities.html#CVE-2003-0544
19748[CVE-2003-0543]: https://www.openssl.org/news/vulnerabilities.html#CVE-2003-0543
19749[CVE-2003-0078]: https://www.openssl.org/news/vulnerabilities.html#CVE-2003-0078
19750[CVE-2002-0659]: https://www.openssl.org/news/vulnerabilities.html#CVE-2002-0659
19751[CVE-2002-0657]: https://www.openssl.org/news/vulnerabilities.html#CVE-2002-0657
19752[CVE-2002-0656]: https://www.openssl.org/news/vulnerabilities.html#CVE-2002-0656
19753[CVE-2002-0655]: https://www.openssl.org/news/vulnerabilities.html#CVE-2002-0655
19754