xref: /freebsd/crypto/openssl/CHANGES.md (revision ee3960cba1068e12fb032a68c46d74841d9edab3)
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 3.0.15 and 3.0.16 [11 Feb 2025]
32
33 * Fixed timing side-channel in ECDSA signature computation.
34
35   There is a timing signal of around 300 nanoseconds when the top word of
36   the inverted ECDSA nonce value is zero. This can happen with significant
37   probability only for some of the supported elliptic curves. In particular
38   the NIST P-521 curve is affected. To be able to measure this leak, the
39   attacker process must either be located in the same physical computer or
40   must have a very fast network connection with low latency.
41
42   ([CVE-2024-13176])
43
44   *Tomáš Mráz*
45
46 * Fixed possible OOB memory access with invalid low-level GF(2^m) elliptic
47   curve parameters.
48
49   Use of the low-level GF(2^m) elliptic curve APIs with untrusted
50   explicit values for the field polynomial can lead to out-of-bounds memory
51   reads or writes.
52   Applications working with "exotic" explicit binary (GF(2^m)) curve
53   parameters, that make it possible to represent invalid field polynomials
54   with a zero constant term, via the above or similar APIs, may terminate
55   abruptly as a result of reading or writing outside of array bounds. Remote
56   code execution cannot easily be ruled out.
57
58   ([CVE-2024-9143])
59
60   *Viktor Dukhovni*
61
62### Changes between 3.0.14 and 3.0.15 [3 Sep 2024]
63
64 * Fixed possible denial of service in X.509 name checks.
65
66   Applications performing certificate name checks (e.g., TLS clients checking
67   server certificates) may attempt to read an invalid memory address when
68   comparing the expected name with an `otherName` subject alternative name of
69   an X.509 certificate. This may result in an exception that terminates the
70   application program.
71
72   ([CVE-2024-6119])
73
74   *Viktor Dukhovni*
75
76 * Fixed possible buffer overread in SSL_select_next_proto().
77
78   Calling the OpenSSL API function SSL_select_next_proto with an empty
79   supported client protocols buffer may cause a crash or memory contents
80   to be sent to the peer.
81
82   ([CVE-2024-5535])
83
84   *Matt Caswell*
85
86### Changes between 3.0.13 and 3.0.14 [4 Jun 2024]
87
88 * Fixed potential use after free after SSL_free_buffers() is called.
89
90   The SSL_free_buffers function is used to free the internal OpenSSL
91   buffer used when processing an incoming record from the network.
92   The call is only expected to succeed if the buffer is not currently
93   in use. However, two scenarios have been identified where the buffer
94   is freed even when still in use.
95
96   The first scenario occurs where a record header has been received
97   from the network and processed by OpenSSL, but the full record body
98   has not yet arrived. In this case calling SSL_free_buffers will succeed
99   even though a record has only been partially processed and the buffer
100   is still in use.
101
102   The second scenario occurs where a full record containing application
103   data has been received and processed by OpenSSL but the application has
104   only read part of this data. Again a call to SSL_free_buffers will
105   succeed even though the buffer is still in use.
106
107   ([CVE-2024-4741])
108
109   *Matt Caswell*
110
111 * Fixed an issue where checking excessively long DSA keys or parameters may
112   be very slow.
113
114   Applications that use the functions EVP_PKEY_param_check() or
115   EVP_PKEY_public_check() to check a DSA public key or DSA parameters may
116   experience long delays. Where the key or parameters that are being checked
117   have been obtained from an untrusted source this may lead to a Denial of
118   Service.
119
120   To resolve this issue DSA keys larger than OPENSSL_DSA_MAX_MODULUS_BITS
121   will now fail the check immediately with a DSA_R_MODULUS_TOO_LARGE error
122   reason.
123
124   ([CVE-2024-4603])
125
126   *Tomáš Mráz*
127
128 * Improved EC/DSA nonce generation routines to avoid bias and timing
129   side channel leaks.
130
131   Thanks to Florian Sieck from Universität zu Lübeck and George Pantelakis
132   and Hubert Kario from Red Hat for reporting the issues.
133
134   *Tomáš Mráz and Paul Dale*
135
136 * Fixed an issue where some non-default TLS server configurations can cause
137   unbounded memory growth when processing TLSv1.3 sessions. An attacker may
138   exploit certain server configurations to trigger unbounded memory growth that
139   would lead to a Denial of Service
140
141   This problem can occur in TLSv1.3 if the non-default SSL_OP_NO_TICKET option
142   is being used (but not if early_data is also configured and the default
143   anti-replay protection is in use). In this case, under certain conditions,
144   the session cache can get into an incorrect state and it will fail to flush
145   properly as it fills. The session cache will continue to grow in an unbounded
146   manner. A malicious client could deliberately create the scenario for this
147   failure to force a Denial of Service. It may also happen by accident in
148   normal operation.
149
150   ([CVE-2024-2511])
151
152   *Matt Caswell*
153
154 * New atexit configuration switch, which controls whether the OPENSSL_cleanup
155   is registered when libcrypto is unloaded. This can be used on platforms
156   where using atexit() from shared libraries causes crashes on exit.
157
158   *Randall S. Becker*
159
160### Changes between 3.0.12 and 3.0.13 [30 Jan 2024]
161
162 * A file in PKCS12 format can contain certificates and keys and may come from
163   an untrusted source. The PKCS12 specification allows certain fields to be
164   NULL, but OpenSSL did not correctly check for this case. A fix has been
165   applied to prevent a NULL pointer dereference that results in OpenSSL
166   crashing. If an application processes PKCS12 files from an untrusted source
167   using the OpenSSL APIs then that application will be vulnerable to this
168   issue prior to this fix.
169
170   OpenSSL APIs that were vulnerable to this are: PKCS12_parse(),
171   PKCS12_unpack_p7data(), PKCS12_unpack_p7encdata(), PKCS12_unpack_authsafes()
172   and PKCS12_newpass().
173
174   We have also fixed a similar issue in SMIME_write_PKCS7(). However since this
175   function is related to writing data we do not consider it security
176   significant.
177
178   ([CVE-2024-0727])
179
180   *Matt Caswell*
181
182 * When function EVP_PKEY_public_check() is called on RSA public keys,
183   a computation is done to confirm that the RSA modulus, n, is composite.
184   For valid RSA keys, n is a product of two or more large primes and this
185   computation completes quickly. However, if n is an overly large prime,
186   then this computation would take a long time.
187
188   An application that calls EVP_PKEY_public_check() and supplies an RSA key
189   obtained from an untrusted source could be vulnerable to a Denial of Service
190   attack.
191
192   The function EVP_PKEY_public_check() is not called from other OpenSSL
193   functions however it is called from the OpenSSL pkey command line
194   application. For that reason that application is also vulnerable if used
195   with the "-pubin" and "-check" options on untrusted data.
196
197   To resolve this issue RSA keys larger than OPENSSL_RSA_MAX_MODULUS_BITS will
198   now fail the check immediately with an RSA_R_MODULUS_TOO_LARGE error reason.
199
200   ([CVE-2023-6237])
201
202   *Tomáš Mráz*
203
204 * Restore the encoding of SM2 PrivateKeyInfo and SubjectPublicKeyInfo to
205   have the contained AlgorithmIdentifier.algorithm set to id-ecPublicKey
206   rather than SM2.
207
208   *Richard Levitte*
209
210 * The POLY1305 MAC (message authentication code) implementation in OpenSSL
211   for PowerPC CPUs saves the contents of vector registers in different
212   order than they are restored. Thus the contents of some of these vector
213   registers is corrupted when returning to the caller. The vulnerable code is
214   used only on newer PowerPC processors supporting the PowerISA 2.07
215   instructions.
216
217   The consequences of this kind of internal application state corruption can
218   be various - from no consequences, if the calling application does not
219   depend on the contents of non-volatile XMM registers at all, to the worst
220   consequences, where the attacker could get complete control of the
221   application process. However unless the compiler uses the vector registers
222   for storing pointers, the most likely consequence, if any, would be an
223   incorrect result of some application dependent calculations or a crash
224   leading to a denial of service.
225
226   ([CVE-2023-6129])
227
228   *Rohan McLure*
229
230 * Fix excessive time spent in DH check / generation with large Q parameter
231   value.
232
233   Applications that use the functions DH_generate_key() to generate an
234   X9.42 DH key may experience long delays. Likewise, applications that use
235   DH_check_pub_key(), DH_check_pub_key_ex() or EVP_PKEY_public_check()
236   to check an X9.42 DH key or X9.42 DH parameters may experience long delays.
237   Where the key or parameters that are being checked have been obtained from
238   an untrusted source this may lead to a Denial of Service.
239
240   ([CVE-2023-5678])
241
242   *Richard Levitte*
243
244### Changes between 3.0.11 and 3.0.12 [24 Oct 2023]
245
246 * Fix incorrect key and IV resizing issues when calling EVP_EncryptInit_ex2(),
247   EVP_DecryptInit_ex2() or EVP_CipherInit_ex2() with OSSL_PARAM parameters
248   that alter the key or IV length ([CVE-2023-5363]).
249
250   *Paul Dale*
251
252### Changes between 3.0.10 and 3.0.11 [19 Sep 2023]
253
254 * Fix POLY1305 MAC implementation corrupting XMM registers on Windows.
255
256   The POLY1305 MAC (message authentication code) implementation in OpenSSL
257   does not save the contents of non-volatile XMM registers on Windows 64
258   platform when calculating the MAC of data larger than 64 bytes. Before
259   returning to the caller all the XMM registers are set to zero rather than
260   restoring their previous content. The vulnerable code is used only on newer
261   x86_64 processors supporting the AVX512-IFMA instructions.
262
263   The consequences of this kind of internal application state corruption can
264   be various - from no consequences, if the calling application does not
265   depend on the contents of non-volatile XMM registers at all, to the worst
266   consequences, where the attacker could get complete control of the
267   application process. However given the contents of the registers are just
268   zeroized so the attacker cannot put arbitrary values inside, the most likely
269   consequence, if any, would be an incorrect result of some application
270   dependent calculations or a crash leading to a denial of service.
271
272   ([CVE-2023-4807])
273
274   *Bernd Edlinger*
275
276### Changes between 3.0.9 and 3.0.10 [1 Aug 2023]
277
278 * Fix excessive time spent checking DH q parameter value.
279
280   The function DH_check() performs various checks on DH parameters. After
281   fixing CVE-2023-3446 it was discovered that a large q parameter value can
282   also trigger an overly long computation during some of these checks.
283   A correct q value, if present, cannot be larger than the modulus p
284   parameter, thus it is unnecessary to perform these checks if q is larger
285   than p.
286
287   If DH_check() is called with such q parameter value,
288   DH_CHECK_INVALID_Q_VALUE return flag is set and the computationally
289   intensive checks are skipped.
290
291   ([CVE-2023-3817])
292
293   *Tomáš Mráz*
294
295 * Fix DH_check() excessive time with over sized modulus.
296
297   The function DH_check() performs various checks on DH parameters. One of
298   those checks confirms that the modulus ("p" parameter) is not too large.
299   Trying to use a very large modulus is slow and OpenSSL will not normally use
300   a modulus which is over 10,000 bits in length.
301
302   However the DH_check() function checks numerous aspects of the key or
303   parameters that have been supplied. Some of those checks use the supplied
304   modulus value even if it has already been found to be too large.
305
306   A new limit has been added to DH_check of 32,768 bits. Supplying a
307   key/parameters with a modulus over this size will simply cause DH_check() to
308   fail.
309
310   ([CVE-2023-3446])
311
312   *Matt Caswell*
313
314 * Do not ignore empty associated data entries with AES-SIV.
315
316   The AES-SIV algorithm allows for authentication of multiple associated
317   data entries along with the encryption. To authenticate empty data the
318   application has to call `EVP_EncryptUpdate()` (or `EVP_CipherUpdate()`)
319   with NULL pointer as the output buffer and 0 as the input buffer length.
320   The AES-SIV implementation in OpenSSL just returns success for such call
321   instead of performing the associated data authentication operation.
322   The empty data thus will not be authenticated. ([CVE-2023-2975])
323
324   Thanks to Juerg Wullschleger (Google) for discovering the issue.
325
326   The fix changes the authentication tag value and the ciphertext for
327   applications that use empty associated data entries with AES-SIV.
328   To decrypt data encrypted with previous versions of OpenSSL the application
329   has to skip calls to `EVP_DecryptUpdate()` for empty associated data
330   entries.
331
332   *Tomáš Mráz*
333
334### Changes between 3.0.8 and 3.0.9 [30 May 2023]
335
336 * Mitigate for the time it takes for `OBJ_obj2txt` to translate gigantic
337   OBJECT IDENTIFIER sub-identifiers to canonical numeric text form.
338
339   OBJ_obj2txt() would translate any size OBJECT IDENTIFIER to canonical
340   numeric text form.  For gigantic sub-identifiers, this would take a very
341   long time, the time complexity being O(n^2) where n is the size of that
342   sub-identifier.  ([CVE-2023-2650])
343
344   To mitigitate this, `OBJ_obj2txt()` will only translate an OBJECT
345   IDENTIFIER to canonical numeric text form if the size of that OBJECT
346   IDENTIFIER is 586 bytes or less, and fail otherwise.
347
348   The basis for this restriction is [RFC 2578 (STD 58), section 3.5]. OBJECT
349   IDENTIFIER values, which stipulates that OBJECT IDENTIFIERS may have at
350   most 128 sub-identifiers, and that the maximum value that each sub-
351   identifier may have is 2^32-1 (4294967295 decimal).
352
353   For each byte of every sub-identifier, only the 7 lower bits are part of
354   the value, so the maximum amount of bytes that an OBJECT IDENTIFIER with
355   these restrictions may occupy is 32 * 128 / 7, which is approximately 586
356   bytes.
357
358   *Richard Levitte*
359
360 * Fixed buffer overread in AES-XTS decryption on ARM 64 bit platforms which
361   happens if the buffer size is 4 mod 5 in 16 byte AES blocks. This can
362   trigger a crash of an application using AES-XTS decryption if the memory
363   just after the buffer being decrypted is not mapped.
364   Thanks to Anton Romanov (Amazon) for discovering the issue.
365   ([CVE-2023-1255])
366
367   *Nevine Ebeid*
368
369 * Reworked the Fix for the Timing Oracle in RSA Decryption ([CVE-2022-4304]).
370   The previous fix for this timing side channel turned out to cause
371   a severe 2-3x performance regression in the typical use case
372   compared to 3.0.7. The new fix uses existing constant time
373   code paths, and restores the previous performance level while
374   fully eliminating all existing timing side channels.
375   The fix was developed by Bernd Edlinger with testing support
376   by Hubert Kario.
377
378   *Bernd Edlinger*
379
380 * Corrected documentation of X509_VERIFY_PARAM_add0_policy() to mention
381   that it does not enable policy checking. Thanks to David Benjamin for
382   discovering this issue.
383   ([CVE-2023-0466])
384
385   *Tomáš Mráz*
386
387 * Fixed an issue where invalid certificate policies in leaf certificates are
388   silently ignored by OpenSSL and other certificate policy checks are skipped
389   for that certificate. A malicious CA could use this to deliberately assert
390   invalid certificate policies in order to circumvent policy checking on the
391   certificate altogether.
392   ([CVE-2023-0465])
393
394   *Matt Caswell*
395
396 * Limited the number of nodes created in a policy tree to mitigate
397   against CVE-2023-0464.  The default limit is set to 1000 nodes, which
398   should be sufficient for most installations.  If required, the limit
399   can be adjusted by setting the OPENSSL_POLICY_TREE_NODES_MAX build
400   time define to a desired maximum number of nodes or zero to allow
401   unlimited growth.
402   ([CVE-2023-0464])
403
404   *Paul Dale*
405
406### Changes between 3.0.7 and 3.0.8 [7 Feb 2023]
407
408 * Fixed NULL dereference during PKCS7 data verification.
409
410   A NULL pointer can be dereferenced when signatures are being
411   verified on PKCS7 signed or signedAndEnveloped data. In case the hash
412   algorithm used for the signature is known to the OpenSSL library but
413   the implementation of the hash algorithm is not available the digest
414   initialization will fail. There is a missing check for the return
415   value from the initialization function which later leads to invalid
416   usage of the digest API most likely leading to a crash.
417   ([CVE-2023-0401])
418
419   PKCS7 data is processed by the SMIME library calls and also by the
420   time stamp (TS) library calls. The TLS implementation in OpenSSL does
421   not call these functions however third party applications would be
422   affected if they call these functions to verify signatures on untrusted
423   data.
424
425   *Tomáš Mráz*
426
427 * Fixed X.400 address type confusion in X.509 GeneralName.
428
429   There is a type confusion vulnerability relating to X.400 address processing
430   inside an X.509 GeneralName. X.400 addresses were parsed as an ASN1_STRING
431   but the public structure definition for GENERAL_NAME incorrectly specified
432   the type of the x400Address field as ASN1_TYPE. This field is subsequently
433   interpreted by the OpenSSL function GENERAL_NAME_cmp as an ASN1_TYPE rather
434   than an ASN1_STRING.
435
436   When CRL checking is enabled (i.e. the application sets the
437   X509_V_FLAG_CRL_CHECK flag), this vulnerability may allow an attacker to
438   pass arbitrary pointers to a memcmp call, enabling them to read memory
439   contents or enact a denial of service.
440   ([CVE-2023-0286])
441
442   *Hugo Landau*
443
444 * Fixed NULL dereference validating DSA public key.
445
446   An invalid pointer dereference on read can be triggered when an
447   application tries to check a malformed DSA public key by the
448   EVP_PKEY_public_check() function. This will most likely lead
449   to an application crash. This function can be called on public
450   keys supplied from untrusted sources which could allow an attacker
451   to cause a denial of service attack.
452
453   The TLS implementation in OpenSSL does not call this function
454   but applications might call the function if there are additional
455   security requirements imposed by standards such as FIPS 140-3.
456   ([CVE-2023-0217])
457
458   *Shane Lontis, Tomáš Mráz*
459
460 * Fixed Invalid pointer dereference in d2i_PKCS7 functions.
461
462   An invalid pointer dereference on read can be triggered when an
463   application tries to load malformed PKCS7 data with the
464   d2i_PKCS7(), d2i_PKCS7_bio() or d2i_PKCS7_fp() functions.
465
466   The result of the dereference is an application crash which could
467   lead to a denial of service attack. The TLS implementation in OpenSSL
468   does not call this function however third party applications might
469   call these functions on untrusted data.
470   ([CVE-2023-0216])
471
472   *Tomáš Mráz*
473
474 * Fixed Use-after-free following BIO_new_NDEF.
475
476   The public API function BIO_new_NDEF is a helper function used for
477   streaming ASN.1 data via a BIO. It is primarily used internally to OpenSSL
478   to support the SMIME, CMS and PKCS7 streaming capabilities, but may also
479   be called directly by end user applications.
480
481   The function receives a BIO from the caller, prepends a new BIO_f_asn1
482   filter BIO onto the front of it to form a BIO chain, and then returns
483   the new head of the BIO chain to the caller. Under certain conditions,
484   for example if a CMS recipient public key is invalid, the new filter BIO
485   is freed and the function returns a NULL result indicating a failure.
486   However, in this case, the BIO chain is not properly cleaned up and the
487   BIO passed by the caller still retains internal pointers to the previously
488   freed filter BIO. If the caller then goes on to call BIO_pop() on the BIO
489   then a use-after-free will occur. This will most likely result in a crash.
490   ([CVE-2023-0215])
491
492   *Viktor Dukhovni, Matt Caswell*
493
494 * Fixed Double free after calling PEM_read_bio_ex.
495
496   The function PEM_read_bio_ex() reads a PEM file from a BIO and parses and
497   decodes the "name" (e.g. "CERTIFICATE"), any header data and the payload
498   data. If the function succeeds then the "name_out", "header" and "data"
499   arguments are populated with pointers to buffers containing the relevant
500   decoded data. The caller is responsible for freeing those buffers. It is
501   possible to construct a PEM file that results in 0 bytes of payload data.
502   In this case PEM_read_bio_ex() will return a failure code but will populate
503   the header argument with a pointer to a buffer that has already been freed.
504   If the caller also frees this buffer then a double free will occur. This
505   will most likely lead to a crash.
506
507   The functions PEM_read_bio() and PEM_read() are simple wrappers around
508   PEM_read_bio_ex() and therefore these functions are also directly affected.
509
510   These functions are also called indirectly by a number of other OpenSSL
511   functions including PEM_X509_INFO_read_bio_ex() and
512   SSL_CTX_use_serverinfo_file() which are also vulnerable. Some OpenSSL
513   internal uses of these functions are not vulnerable because the caller does
514   not free the header argument if PEM_read_bio_ex() returns a failure code.
515   ([CVE-2022-4450])
516
517   *Kurt Roeckx, Matt Caswell*
518
519 * Fixed Timing Oracle in RSA Decryption.
520
521   A timing based side channel exists in the OpenSSL RSA Decryption
522   implementation which could be sufficient to recover a plaintext across
523   a network in a Bleichenbacher style attack. To achieve a successful
524   decryption an attacker would have to be able to send a very large number
525   of trial messages for decryption. The vulnerability affects all RSA padding
526   modes: PKCS#1 v1.5, RSA-OEAP and RSASVE.
527   ([CVE-2022-4304])
528
529   *Dmitry Belyavsky, Hubert Kario*
530
531 * Fixed X.509 Name Constraints Read Buffer Overflow.
532
533   A read buffer overrun can be triggered in X.509 certificate verification,
534   specifically in name constraint checking. The read buffer overrun might
535   result in a crash which could lead to a denial of service attack.
536   In a TLS client, this can be triggered by connecting to a malicious
537   server. In a TLS server, this can be triggered if the server requests
538   client authentication and a malicious client connects.
539   ([CVE-2022-4203])
540
541   *Viktor Dukhovni*
542
543 * Fixed X.509 Policy Constraints Double Locking security issue.
544
545   If an X.509 certificate contains a malformed policy constraint and
546   policy processing is enabled, then a write lock will be taken twice
547   recursively.  On some operating systems (most widely: Windows) this
548   results in a denial of service when the affected process hangs.  Policy
549   processing being enabled on a publicly facing server is not considered
550   to be a common setup.
551   ([CVE-2022-3996])
552
553   *Paul Dale*
554
555 * Our provider implementations of `OSSL_FUNC_KEYMGMT_EXPORT` and
556   `OSSL_FUNC_KEYMGMT_GET_PARAMS` for EC and SM2 keys now honor
557   `OSSL_PKEY_PARAM_EC_POINT_CONVERSION_FORMAT` as set (and
558   default to `POINT_CONVERSION_UNCOMPRESSED`) when exporting
559   `OSSL_PKEY_PARAM_PUB_KEY`, instead of unconditionally using
560   `POINT_CONVERSION_COMPRESSED` as in previous 3.x releases.
561   For symmetry, our implementation of `EVP_PKEY_ASN1_METHOD->export_to`
562   for legacy EC and SM2 keys is also changed similarly to honor the
563   equivalent conversion format flag as specified in the underlying
564   `EC_KEY` object being exported to a provider, when this function is
565   called through `EVP_PKEY_export()`.
566
567   *Nicola Tuveri*
568
569### Changes between 3.0.6 and 3.0.7 [1 Nov 2022]
570
571 * Fixed two buffer overflows in punycode decoding functions.
572
573   A buffer overrun can be triggered in X.509 certificate verification,
574   specifically in name constraint checking. Note that this occurs after
575   certificate chain signature verification and requires either a CA to
576   have signed the malicious certificate or for the application to continue
577   certificate verification despite failure to construct a path to a trusted
578   issuer.
579
580   In a TLS client, this can be triggered by connecting to a malicious
581   server.  In a TLS server, this can be triggered if the server requests
582   client authentication and a malicious client connects.
583
584   An attacker can craft a malicious email address to overflow
585   an arbitrary number of bytes containing the `.`  character (decimal 46)
586   on the stack.  This buffer overflow could result in a crash (causing a
587   denial of service).
588   ([CVE-2022-3786])
589
590   An attacker can craft a malicious email address to overflow four
591   attacker-controlled bytes on the stack.  This buffer overflow could
592   result in a crash (causing a denial of service) or potentially remote code
593   execution depending on stack layout for any given platform/compiler.
594   ([CVE-2022-3602])
595
596   *Paul Dale*
597
598 * Removed all references to invalid OSSL_PKEY_PARAM_RSA names for CRT
599   parameters in OpenSSL code.
600   Applications should not use the names OSSL_PKEY_PARAM_RSA_FACTOR,
601   OSSL_PKEY_PARAM_RSA_EXPONENT and OSSL_PKEY_PARAM_RSA_COEFFICIENT.
602   Use the numbered names such as OSSL_PKEY_PARAM_RSA_FACTOR1 instead.
603   Using these invalid names may cause algorithms to use slower methods
604   that ignore the CRT parameters.
605
606   *Shane Lontis*
607
608 * Fixed a regression introduced in 3.0.6 version raising errors on some stack
609   operations.
610
611   *Tomáš Mráz*
612
613 * Fixed a regression introduced in 3.0.6 version not refreshing the certificate
614   data to be signed before signing the certificate.
615
616   *Gibeom Gwon*
617
618 * Added RIPEMD160 to the default provider.
619
620   *Paul Dale*
621
622 * Ensured that the key share group sent or accepted for the key exchange
623   is allowed for the protocol version.
624
625   *Matt Caswell*
626
627### Changes between 3.0.5 and 3.0.6 [11 Oct 2022]
628
629 * OpenSSL supports creating a custom cipher via the legacy
630   EVP_CIPHER_meth_new() function and associated function calls. This function
631   was deprecated in OpenSSL 3.0 and application authors are instead encouraged
632   to use the new provider mechanism in order to implement custom ciphers.
633
634   OpenSSL versions 3.0.0 to 3.0.5 incorrectly handle legacy custom ciphers
635   passed to the EVP_EncryptInit_ex2(), EVP_DecryptInit_ex2() and
636   EVP_CipherInit_ex2() functions (as well as other similarly named encryption
637   and decryption initialisation functions). Instead of using the custom cipher
638   directly it incorrectly tries to fetch an equivalent cipher from the
639   available providers. An equivalent cipher is found based on the NID passed to
640   EVP_CIPHER_meth_new(). This NID is supposed to represent the unique NID for a
641   given cipher. However it is possible for an application to incorrectly pass
642   NID_undef as this value in the call to EVP_CIPHER_meth_new(). When NID_undef
643   is used in this way the OpenSSL encryption/decryption initialisation function
644   will match the NULL cipher as being equivalent and will fetch this from the
645   available providers. This will succeed if the default provider has been
646   loaded (or if a third party provider has been loaded that offers this
647   cipher). Using the NULL cipher means that the plaintext is emitted as the
648   ciphertext.
649
650   Applications are only affected by this issue if they call
651   EVP_CIPHER_meth_new() using NID_undef and subsequently use it in a call to an
652   encryption/decryption initialisation function. Applications that only use
653   SSL/TLS are not impacted by this issue.
654   ([CVE-2022-3358])
655
656   *Matt Caswell*
657
658 * Fix LLVM vs Apple LLVM version numbering confusion that caused build failures
659   on MacOS 10.11
660
661   *Richard Levitte*
662
663 * Fixed the linux-mips64 Configure target which was missing the
664   SIXTY_FOUR_BIT bn_ops flag. This was causing heap corruption on that
665   platform.
666
667   *Adam Joseph*
668
669 * Fix handling of a ticket key callback that returns 0 in TLSv1.3 to not send a
670   ticket
671
672   *Matt Caswell*
673
674 * Correctly handle a retransmitted ClientHello in DTLS
675
676   *Matt Caswell*
677
678 * Fixed detection of ktls support in cross-compile environment on Linux
679
680   *Tomas Mraz*
681
682 * Fixed some regressions and test failures when running the 3.0.0 FIPS provider
683   against 3.0.x
684
685   *Paul Dale*
686
687 * Fixed SSL_pending() and SSL_has_pending() with DTLS which were failing to
688   report correct results in some cases
689
690   *Matt Caswell*
691
692 * Fix UWP builds by defining VirtualLock
693
694   *Charles Milette*
695
696 * For known safe primes use the minimum key length according to RFC 7919.
697   Longer private key sizes unnecessarily raise the cycles needed to compute the
698   shared secret without any increase of the real security. This fixes a
699   regression from 1.1.1 where these shorter keys were generated for the known
700   safe primes.
701
702   *Tomas Mraz*
703
704 * Added the loongarch64 target
705
706   *Shi Pujin*
707
708 * Fixed EC ASM flag passing. Flags for ASM implementations of EC curves were
709   only passed to the FIPS provider and not to the default or legacy provider.
710
711   *Juergen Christ*
712
713 * Fixed reported performance degradation on aarch64. Restored the
714   implementation prior to commit 2621751 ("aes/asm/aesv8-armx.pl: avoid
715   32-bit lane assignment in CTR mode") for 64bit targets only, since it is
716   reportedly 2-17% slower and the silicon errata only affects 32bit targets.
717   The new algorithm is still used for 32 bit targets.
718
719   *Bernd Edlinger*
720
721 * Added a missing header for memcmp that caused compilation failure on some
722   platforms
723
724   *Gregor Jasny*
725
726### Changes between 3.0.4 and 3.0.5 [5 Jul 2022]
727
728 * The OpenSSL 3.0.4 release introduced a serious bug in the RSA
729   implementation for X86_64 CPUs supporting the AVX512IFMA instructions.
730   This issue makes the RSA implementation with 2048 bit private keys
731   incorrect on such machines and memory corruption will happen during
732   the computation. As a consequence of the memory corruption an attacker
733   may be able to trigger a remote code execution on the machine performing
734   the computation.
735
736   SSL/TLS servers or other servers using 2048 bit RSA private keys running
737   on machines supporting AVX512IFMA instructions of the X86_64 architecture
738   are affected by this issue.
739   ([CVE-2022-2274])
740
741   *Xi Ruoyao*
742
743 * AES OCB mode for 32-bit x86 platforms using the AES-NI assembly optimised
744   implementation would not encrypt the entirety of the data under some
745   circumstances.  This could reveal sixteen bytes of data that was
746   preexisting in the memory that wasn't written.  In the special case of
747   "in place" encryption, sixteen bytes of the plaintext would be revealed.
748
749   Since OpenSSL does not support OCB based cipher suites for TLS and DTLS,
750   they are both unaffected.
751   ([CVE-2022-2097])
752
753   *Alex Chernyakhovsky, David Benjamin, Alejandro Sedeño*
754
755### Changes between 3.0.3 and 3.0.4 [21 Jun 2022]
756
757 * In addition to the c_rehash shell command injection identified in
758   CVE-2022-1292, further bugs where the c_rehash script does not
759   properly sanitise shell metacharacters to prevent command injection have been
760   fixed.
761
762   When the CVE-2022-1292 was fixed it was not discovered that there
763   are other places in the script where the file names of certificates
764   being hashed were possibly passed to a command executed through the shell.
765
766   This script is distributed by some operating systems in a manner where
767   it is automatically executed.  On such operating systems, an attacker
768   could execute arbitrary commands with the privileges of the script.
769
770   Use of the c_rehash script is considered obsolete and should be replaced
771   by the OpenSSL rehash command line tool.
772   (CVE-2022-2068)
773
774   *Daniel Fiala, Tomáš Mráz*
775
776 * Case insensitive string comparison no longer uses locales.  It has instead
777   been directly implemented.
778
779   *Paul Dale*
780
781### Changes between 3.0.2 and 3.0.3 [3 May 2022]
782
783 * Case insensitive string comparison is reimplemented via new locale-agnostic
784   comparison functions OPENSSL_str[n]casecmp always using the POSIX locale for
785   comparison. The previous implementation had problems when the Turkish locale
786   was used.
787
788   *Dmitry Belyavskiy*
789
790 * Fixed a bug in the c_rehash script which was not properly sanitising shell
791   metacharacters to prevent command injection.  This script is distributed by
792   some operating systems in a manner where it is automatically executed.  On
793   such operating systems, an attacker could execute arbitrary commands with the
794   privileges of the script.
795
796   Use of the c_rehash script is considered obsolete and should be replaced
797   by the OpenSSL rehash command line tool.
798   (CVE-2022-1292)
799
800   *Tomáš Mráz*
801
802 * Fixed a bug in the function `OCSP_basic_verify` that verifies the signer
803   certificate on an OCSP response. The bug caused the function in the case
804   where the (non-default) flag OCSP_NOCHECKS is used to return a postivie
805   response (meaning a successful verification) even in the case where the
806   response signing certificate fails to verify.
807
808   It is anticipated that most users of `OCSP_basic_verify` will not use the
809   OCSP_NOCHECKS flag. In this case the `OCSP_basic_verify` function will return
810   a negative value (indicating a fatal error) in the case of a certificate
811   verification failure. The normal expected return value in this case would be
812   0.
813
814   This issue also impacts the command line OpenSSL "ocsp" application. When
815   verifying an ocsp response with the "-no_cert_checks" option the command line
816   application will report that the verification is successful even though it
817   has in fact failed. In this case the incorrect successful response will also
818   be accompanied by error messages showing the failure and contradicting the
819   apparently successful result.
820   ([CVE-2022-1343])
821
822   *Matt Caswell*
823
824 * Fixed a bug where the RC4-MD5 ciphersuite incorrectly used the
825   AAD data as the MAC key. This made the MAC key trivially predictable.
826
827   An attacker could exploit this issue by performing a man-in-the-middle attack
828   to modify data being sent from one endpoint to an OpenSSL 3.0 recipient such
829   that the modified data would still pass the MAC integrity check.
830
831   Note that data sent from an OpenSSL 3.0 endpoint to a non-OpenSSL 3.0
832   endpoint will always be rejected by the recipient and the connection will
833   fail at that point. Many application protocols require data to be sent from
834   the client to the server first. Therefore, in such a case, only an OpenSSL
835   3.0 server would be impacted when talking to a non-OpenSSL 3.0 client.
836
837   If both endpoints are OpenSSL 3.0 then the attacker could modify data being
838   sent in both directions. In this case both clients and servers could be
839   affected, regardless of the application protocol.
840
841   Note that in the absence of an attacker this bug means that an OpenSSL 3.0
842   endpoint communicating with a non-OpenSSL 3.0 endpoint will fail to complete
843   the handshake when using this ciphersuite.
844
845   The confidentiality of data is not impacted by this issue, i.e. an attacker
846   cannot decrypt data that has been encrypted using this ciphersuite - they can
847   only modify it.
848
849   In order for this attack to work both endpoints must legitimately negotiate
850   the RC4-MD5 ciphersuite. This ciphersuite is not compiled by default in
851   OpenSSL 3.0, and is not available within the default provider or the default
852   ciphersuite list. This ciphersuite will never be used if TLSv1.3 has been
853   negotiated. In order for an OpenSSL 3.0 endpoint to use this ciphersuite the
854   following must have occurred:
855
856   1) OpenSSL must have been compiled with the (non-default) compile time option
857      enable-weak-ssl-ciphers
858
859   2) OpenSSL must have had the legacy provider explicitly loaded (either
860      through application code or via configuration)
861
862   3) The ciphersuite must have been explicitly added to the ciphersuite list
863
864   4) The libssl security level must have been set to 0 (default is 1)
865
866   5) A version of SSL/TLS below TLSv1.3 must have been negotiated
867
868   6) Both endpoints must negotiate the RC4-MD5 ciphersuite in preference to any
869      others that both endpoints have in common
870   (CVE-2022-1434)
871
872   *Matt Caswell*
873
874 * Fix a bug in the OPENSSL_LH_flush() function that breaks reuse of the memory
875   occuppied by the removed hash table entries.
876
877   This function is used when decoding certificates or keys. If a long lived
878   process periodically decodes certificates or keys its memory usage will
879   expand without bounds and the process might be terminated by the operating
880   system causing a denial of service. Also traversing the empty hash table
881   entries will take increasingly more time.
882
883   Typically such long lived processes might be TLS clients or TLS servers
884   configured to accept client certificate authentication.
885   (CVE-2022-1473)
886
887   *Hugo Landau, Aliaksei Levin*
888
889 * The functions `OPENSSL_LH_stats` and `OPENSSL_LH_stats_bio` now only report
890   the `num_items`, `num_nodes` and `num_alloc_nodes` statistics. All other
891   statistics are no longer supported. For compatibility, these statistics are
892   still listed in the output but are now always reported as zero.
893
894   *Hugo Landau*
895
896### Changes between 3.0.1 and 3.0.2 [15 Mar 2022]
897
898 * Fixed a bug in the BN_mod_sqrt() function that can cause it to loop forever
899   for non-prime moduli.
900
901   Internally this function is used when parsing certificates that contain
902   elliptic curve public keys in compressed form or explicit elliptic curve
903   parameters with a base point encoded in compressed form.
904
905   It is possible to trigger the infinite loop by crafting a certificate that
906   has invalid explicit curve parameters.
907
908   Since certificate parsing happens prior to verification of the certificate
909   signature, any process that parses an externally supplied certificate may thus
910   be subject to a denial of service attack. The infinite loop can also be
911   reached when parsing crafted private keys as they can contain explicit
912   elliptic curve parameters.
913
914   Thus vulnerable situations include:
915
916    - TLS clients consuming server certificates
917    - TLS servers consuming client certificates
918    - Hosting providers taking certificates or private keys from customers
919    - Certificate authorities parsing certification requests from subscribers
920    - Anything else which parses ASN.1 elliptic curve parameters
921
922   Also any other applications that use the BN_mod_sqrt() where the attacker
923   can control the parameter values are vulnerable to this DoS issue.
924   ([CVE-2022-0778])
925
926   *Tomáš Mráz*
927
928 * Add ciphersuites based on DHE_PSK (RFC 4279) and ECDHE_PSK (RFC 5489)
929   to the list of ciphersuites providing Perfect Forward Secrecy as
930   required by SECLEVEL >= 3.
931
932   *Dmitry Belyavskiy, Nicola Tuveri*
933
934 * Made the AES constant time code for no-asm configurations
935   optional due to the resulting 95% performance degradation.
936   The AES constant time code can be enabled, for no assembly
937   builds, with: ./config no-asm -DOPENSSL_AES_CONST_TIME
938
939   *Paul Dale*
940
941 * Fixed PEM_write_bio_PKCS8PrivateKey() to make it possible to use empty
942   passphrase strings.
943
944   *Darshan Sen*
945
946 * The negative return value handling of the certificate verification callback
947   was reverted. The replacement is to set the verification retry state with
948   the SSL_set_retry_verify() function.
949
950   *Tomáš Mráz*
951
952### Changes between 3.0.0 and 3.0.1 [14 Dec 2021]
953
954 * Fixed invalid handling of X509_verify_cert() internal errors in libssl
955   Internally libssl in OpenSSL calls X509_verify_cert() on the client side to
956   verify a certificate supplied by a server. That function may return a
957   negative return value to indicate an internal error (for example out of
958   memory). Such a negative return value is mishandled by OpenSSL and will cause
959   an IO function (such as SSL_connect() or SSL_do_handshake()) to not indicate
960   success and a subsequent call to SSL_get_error() to return the value
961   SSL_ERROR_WANT_RETRY_VERIFY. This return value is only supposed to be
962   returned by OpenSSL if the application has previously called
963   SSL_CTX_set_cert_verify_callback(). Since most applications do not do this
964   the SSL_ERROR_WANT_RETRY_VERIFY return value from SSL_get_error() will be
965   totally unexpected and applications may not behave correctly as a result. The
966   exact behaviour will depend on the application but it could result in
967   crashes, infinite loops or other similar incorrect responses.
968
969   This issue is made more serious in combination with a separate bug in OpenSSL
970   3.0 that will cause X509_verify_cert() to indicate an internal error when
971   processing a certificate chain. This will occur where a certificate does not
972   include the Subject Alternative Name extension but where a Certificate
973   Authority has enforced name constraints. This issue can occur even with valid
974   chains.
975   ([CVE-2021-4044])
976
977   *Matt Caswell*
978
979 * Corrected a few file name and file reference bugs in the build,
980   installation and setup scripts, which lead to installation verification
981   failures.  Slightly enhanced the installation verification script.
982
983   *Richard Levitte*
984
985 * Fixed EVP_PKEY_eq() to make it possible to use it with strictly private
986   keys.
987
988   *Richard Levitte*
989
990 * Fixed PVK encoder to properly query for the passphrase.
991
992   *Tomáš Mráz*
993
994 * Multiple fixes in the OSSL_HTTP API functions.
995
996   *David von Oheimb*
997
998 * Allow sign extension in OSSL_PARAM_allocate_from_text() for the
999   OSSL_PARAM_INTEGER data type and return error on negative numbers
1000   used with the OSSL_PARAM_UNSIGNED_INTEGER data type. Make
1001   OSSL_PARAM_BLD_push_BN{,_pad}() return an error on negative numbers.
1002
1003   *Richard Levitte*
1004
1005 * Allow copying uninitialized digest contexts with EVP_MD_CTX_copy_ex.
1006
1007   *Tomáš Mráz*
1008
1009 * Fixed detection of ARMv7 and ARM64 CPU features on FreeBSD.
1010
1011   *Allan Jude*
1012
1013 * Multiple threading fixes.
1014
1015   *Matt Caswell*
1016
1017 * Added NULL digest implementation to keep compatibility with 1.1.1 version.
1018
1019   *Tomáš Mráz*
1020
1021 * Allow fetching an operation from the provider that owns an unexportable key
1022   as a fallback if that is still allowed by the property query.
1023
1024   *Richard Levitte*
1025
1026### Changes between 1.1.1 and 3.0.0 [7 sep 2021]
1027
1028 * TLS_MAX_VERSION, DTLS_MAX_VERSION and DTLS_MIN_VERSION constants are now
1029   deprecated.
1030
1031   *Matt Caswell*
1032
1033 * The `OPENSSL_s390xcap` environment variable can be used to set bits in the
1034   S390X capability vector to zero. This simplifies testing of different code
1035   paths on S390X architecture.
1036
1037   *Patrick Steuer*
1038
1039 * Encrypting more than 2^64 TLS records with AES-GCM is disallowed
1040   as per FIPS 140-2 IG A.5 "Key/IV Pair Uniqueness Requirements from
1041   SP 800-38D". The communication will fail at this point.
1042
1043   *Paul Dale*
1044
1045 * The EC_GROUP_clear_free() function is deprecated as there is nothing
1046   confidential in EC_GROUP data.
1047
1048   *Nicola Tuveri*
1049
1050 * The byte order mark (BOM) character is ignored if encountered at the
1051   beginning of a PEM-formatted file.
1052
1053   *Dmitry Belyavskiy*
1054
1055 * Added CMS support for the Russian GOST algorithms.
1056
1057   *Dmitry Belyavskiy*
1058
1059 * Due to move of the implementation of cryptographic operations
1060   to the providers, validation of various operation parameters can
1061   be postponed until the actual operation is executed where previously
1062   it happened immediately when an operation parameter was set.
1063
1064   For example when setting an unsupported curve with
1065   EVP_PKEY_CTX_set_ec_paramgen_curve_nid() this function call will not
1066   fail but later keygen operations with the EVP_PKEY_CTX will fail.
1067
1068   *OpenSSL team members and many third party contributors*
1069
1070 * The EVP_get_cipherbyname() function will return NULL for algorithms such as
1071   "AES-128-SIV", "AES-128-CBC-CTS" and "CAMELLIA-128-CBC-CTS" which were
1072   previously only accessible via low level interfaces. Use EVP_CIPHER_fetch()
1073   instead to retrieve these algorithms from a provider.
1074
1075   *Shane Lontis*
1076
1077 * On build targets where the multilib postfix is set in the build
1078   configuration the libdir directory was changing based on whether
1079   the lib directory with the multilib postfix exists on the system
1080   or not. This unpredictable behavior was removed and eventual
1081   multilib postfix is now always added to the default libdir. Use
1082   `--libdir=lib` to override the libdir if adding the postfix is
1083   undesirable.
1084
1085   *Jan Lána*
1086
1087 * The triple DES key wrap functionality now conforms to RFC 3217 but is
1088   no longer interoperable with OpenSSL 1.1.1.
1089
1090   *Paul Dale*
1091
1092 * The ERR_GET_FUNC() function was removed.  With the loss of meaningful
1093   function codes, this function can only cause problems for calling
1094   applications.
1095
1096   *Paul Dale*
1097
1098 * Add a configurable flag to output date formats as ISO 8601. Does not
1099   change the default date format.
1100
1101   *William Edmisten*
1102
1103 * Version of MSVC earlier than 1300 could get link warnings, which could
1104   be suppressed if the undocumented -DI_CAN_LIVE_WITH_LNK4049 was set.
1105   Support for this flag has been removed.
1106
1107   *Rich Salz*
1108
1109 * Rework and make DEBUG macros consistent. Remove unused -DCONF_DEBUG,
1110   -DBN_CTX_DEBUG, and REF_PRINT. Add a new tracing category and use it for
1111   printing reference counts. Rename -DDEBUG_UNUSED to -DUNUSED_RESULT_DEBUG
1112   Fix BN_DEBUG_RAND so it compiles and, when set, force DEBUG_RAND to be set
1113   also. Rename engine_debug_ref to be ENGINE_REF_PRINT also for consistency.
1114
1115   *Rich Salz*
1116
1117 * The signatures of the functions to get and set options on SSL and
1118   SSL_CTX objects changed from "unsigned long" to "uint64_t" type.
1119   Some source code changes may be required.
1120
1121   *Rich Salz*
1122
1123 * The public definitions of conf_method_st and conf_st have been
1124   deprecated. They will be made opaque in a future release.
1125
1126   *Rich Salz and Tomáš Mráz*
1127
1128 * Client-initiated renegotiation is disabled by default. To allow it, use
1129   the -client_renegotiation option, the SSL_OP_ALLOW_CLIENT_RENEGOTIATION
1130   flag, or the "ClientRenegotiation" config parameter as appropriate.
1131
1132   *Rich Salz*
1133
1134 * Add "abspath" and "includedir" pragma's to config files, to prevent,
1135   or modify relative pathname inclusion.
1136
1137   *Rich Salz*
1138
1139 * OpenSSL includes a cryptographic module that is intended to be FIPS 140-2
1140   validated. Please consult the README-FIPS and
1141   README-PROVIDERS files, as well as the migration guide.
1142
1143   *OpenSSL team members and many third party contributors*
1144
1145 * For the key types DH and DHX the allowed settable parameters are now different.
1146
1147   *Shane Lontis*
1148
1149 * The openssl commands that read keys, certificates, and CRLs now
1150   automatically detect the PEM or DER format of the input files.
1151
1152   *David von Oheimb, Richard Levitte, and Tomáš Mráz*
1153
1154 * Added enhanced PKCS#12 APIs which accept a library context.
1155
1156   *Jon Spillett*
1157
1158 * The default manual page suffix ($MANSUFFIX) has been changed to "ossl"
1159
1160   *Matt Caswell*
1161
1162 * Added support for Kernel TLS (KTLS).
1163
1164   *Boris Pismenny, John Baldwin and Andrew Gallatin*
1165
1166 * Support for RFC 5746 secure renegotiation is now required by default for
1167   SSL or TLS connections to succeed.
1168
1169   *Benjamin Kaduk*
1170
1171 * The signature of the `copy` functional parameter of the
1172   EVP_PKEY_meth_set_copy() function has changed so its `src` argument is
1173   now `const EVP_PKEY_CTX *` instead of `EVP_PKEY_CTX *`. Similarly
1174   the signature of the `pub_decode` functional parameter of the
1175   EVP_PKEY_asn1_set_public() function has changed so its `pub` argument is
1176   now `const X509_PUBKEY *` instead of `X509_PUBKEY *`.
1177
1178   *David von Oheimb*
1179
1180 * The error return values from some control calls (ctrl) have changed.
1181
1182   *Paul Dale*
1183
1184 * A public key check is now performed during EVP_PKEY_derive_set_peer().
1185
1186   *Shane Lontis*
1187
1188 * Many functions in the EVP_ namespace that are getters of values from
1189   implementations or contexts were renamed to include get or get0 in their
1190   names. Old names are provided as macro aliases for compatibility and
1191   are not deprecated.
1192
1193   *Tomáš Mráz*
1194
1195 * The EVP_PKEY_CTRL_PKCS7_ENCRYPT, EVP_PKEY_CTRL_PKCS7_DECRYPT,
1196   EVP_PKEY_CTRL_PKCS7_SIGN, EVP_PKEY_CTRL_CMS_ENCRYPT,
1197   EVP_PKEY_CTRL_CMS_DECRYPT, and EVP_PKEY_CTRL_CMS_SIGN control operations
1198   are deprecated.
1199
1200   *Tomáš Mráz*
1201
1202 * The EVP_PKEY_public_check() and EVP_PKEY_param_check() functions now work for
1203   more key types.
1204
1205 * The output from the command line applications may have minor
1206   changes.
1207
1208   *Paul Dale*
1209
1210 * The output from numerous "printing" may have minor changes.
1211
1212   *David von Oheimb*
1213
1214 * Windows thread synchronization uses read/write primitives (SRWLock) when
1215   supported by the OS, otherwise CriticalSection continues to be used.
1216
1217   *Vincent Drake*
1218
1219 * Add filter BIO BIO_f_readbuffer() that allows BIO_tell() and BIO_seek() to
1220   work on read only BIO source/sinks that do not support these functions.
1221   This allows piping or redirection of a file BIO using stdin to be buffered
1222   into memory. This is used internally in OSSL_DECODER_from_bio().
1223
1224   *Shane Lontis*
1225
1226 * OSSL_STORE_INFO_get_type() may now return an additional value. In 1.1.1
1227   this function would return one of the values OSSL_STORE_INFO_NAME,
1228   OSSL_STORE_INFO_PKEY, OSSL_STORE_INFO_PARAMS, OSSL_STORE_INFO_CERT or
1229   OSSL_STORE_INFO_CRL. Decoded public keys would previously have been reported
1230   as type OSSL_STORE_INFO_PKEY in 1.1.1. In 3.0 decoded public keys are now
1231   reported as having the new type OSSL_STORE_INFO_PUBKEY. Applications
1232   using this function should be amended to handle the changed return value.
1233
1234   *Richard Levitte*
1235
1236 * Improved adherence to Enhanced Security Services (ESS, RFC 2634 and RFC 5035)
1237   for the TSP and CMS Advanced Electronic Signatures (CAdES) implementations.
1238   As required by RFC 5035 check both ESSCertID and ESSCertIDv2 if both present.
1239   Correct the semantics of checking the validation chain in case ESSCertID{,v2}
1240   contains more than one certificate identifier: This means that all
1241   certificates referenced there MUST be part of the validation chain.
1242
1243   *David von Oheimb*
1244
1245 * The implementation of older EVP ciphers related to CAST, IDEA, SEED, RC2, RC4,
1246   RC5, DESX and DES have been moved to the legacy provider.
1247
1248   *Matt Caswell*
1249
1250 * The implementation of the EVP digests MD2, MD4, MDC2, WHIRLPOOL and
1251   RIPEMD-160 have been moved to the legacy provider.
1252
1253   *Matt Caswell*
1254
1255 * The deprecated function EVP_PKEY_get0() now returns NULL being called for a
1256   provided key.
1257
1258   *Dmitry Belyavskiy*
1259
1260 * The deprecated functions EVP_PKEY_get0_RSA(),
1261   EVP_PKEY_get0_DSA(), EVP_PKEY_get0_EC_KEY(), EVP_PKEY_get0_DH(),
1262   EVP_PKEY_get0_hmac(), EVP_PKEY_get0_poly1305() and EVP_PKEY_get0_siphash() as
1263   well as the similarly named "get1" functions behave differently in
1264   OpenSSL 3.0.
1265
1266   *Matt Caswell*
1267
1268 * A number of functions handling low-level keys or engines were deprecated
1269   including EVP_PKEY_set1_engine(), EVP_PKEY_get0_engine(), EVP_PKEY_assign(),
1270   EVP_PKEY_get0(), EVP_PKEY_get0_hmac(), EVP_PKEY_get0_poly1305() and
1271   EVP_PKEY_get0_siphash().
1272
1273   *Matt Caswell*
1274
1275 * PKCS#5 PBKDF1 key derivation has been moved from PKCS5_PBE_keyivgen() into
1276   the legacy crypto provider as an EVP_KDF. Applications requiring this KDF
1277   will need to load the legacy crypto provider. This includes these PBE
1278   algorithms which use this KDF:
1279   - NID_pbeWithMD2AndDES_CBC
1280   - NID_pbeWithMD5AndDES_CBC
1281   - NID_pbeWithSHA1AndRC2_CBC
1282   - NID_pbeWithMD2AndRC2_CBC
1283   - NID_pbeWithMD5AndRC2_CBC
1284   - NID_pbeWithSHA1AndDES_CBC
1285
1286   *Jon Spillett*
1287
1288 * Deprecated obsolete BIO_set_callback(), BIO_get_callback(), and
1289   BIO_debug_callback() functions.
1290
1291   *Tomáš Mráz*
1292
1293 * Deprecated obsolete EVP_PKEY_CTX_get0_dh_kdf_ukm() and
1294   EVP_PKEY_CTX_get0_ecdh_kdf_ukm() functions.
1295
1296   *Tomáš Mráz*
1297
1298 * The RAND_METHOD APIs have been deprecated.
1299
1300   *Paul Dale*
1301
1302 * The SRP APIs have been deprecated.
1303
1304   *Matt Caswell*
1305
1306 * Add a compile time option to prevent the caching of provider fetched
1307   algorithms.  This is enabled by including the no-cached-fetch option
1308   at configuration time.
1309
1310   *Paul Dale*
1311
1312 * pkcs12 now uses defaults of PBKDF2, AES and SHA-256, with a MAC iteration
1313   count of PKCS12_DEFAULT_ITER.
1314
1315   *Tomáš Mráz and Sahana Prasad*
1316
1317 * The openssl speed command does not use low-level API calls anymore.
1318
1319   *Tomáš Mráz*
1320
1321 * Parallel dual-prime 1024-bit modular exponentiation for AVX512_IFMA
1322   capable processors.
1323
1324   *Ilya Albrekht, Sergey Kirillov, Andrey Matyukov (Intel Corp)*
1325
1326 * Combining the Configure options no-ec and no-dh no longer disables TLSv1.3.
1327
1328   *Matt Caswell*
1329
1330 * Implemented support for fully "pluggable" TLSv1.3 groups. This means that
1331   providers may supply their own group implementations (using either the "key
1332   exchange" or the "key encapsulation" methods) which will automatically be
1333   detected and used by libssl.
1334
1335   *Matt Caswell, Nicola Tuveri*
1336
1337 * The undocumented function X509_certificate_type() has been deprecated;
1338
1339   *Rich Salz*
1340
1341 * Deprecated the obsolete BN_pseudo_rand() and BN_pseudo_rand_range().
1342
1343   *Tomáš Mráz*
1344
1345 * Removed RSA padding mode for SSLv23 (which was only used for
1346   SSLv2). This includes the functions RSA_padding_check_SSLv23() and
1347   RSA_padding_add_SSLv23() and the `-ssl` option in the deprecated
1348   `rsautl` command.
1349
1350   *Rich Salz*
1351
1352 * Deprecated the obsolete X9.31 RSA key generation related functions.
1353
1354 * While a callback function set via `SSL_CTX_set_cert_verify_callback()`
1355   is not allowed to return a value > 1, this is no more taken as failure.
1356
1357   *Viktor Dukhovni and David von Oheimb*
1358
1359 * Deprecated the obsolete X9.31 RSA key generation related functions
1360   BN_X931_generate_Xpq(), BN_X931_derive_prime_ex(), and
1361   BN_X931_generate_prime_ex().
1362
1363   *Tomáš Mráz*
1364
1365 * The default key generation method for the regular 2-prime RSA keys was
1366   changed to the FIPS 186-4 B.3.6 method.
1367
1368   *Shane Lontis*
1369
1370 * Deprecated the BN_is_prime_ex() and BN_is_prime_fasttest_ex() functions.
1371
1372   *Kurt Roeckx*
1373
1374 * Deprecated EVP_MD_CTX_set_update_fn() and EVP_MD_CTX_update_fn().
1375
1376   *Rich Salz*
1377
1378 * Deprecated the type OCSP_REQ_CTX and the functions OCSP_REQ_CTX_*() and
1379   replaced with OSSL_HTTP_REQ_CTX and the functions OSSL_HTTP_REQ_CTX_*().
1380
1381   *Rich Salz, Richard Levitte, and David von Oheimb*
1382
1383 * Deprecated `X509_http_nbio()` and `X509_CRL_http_nbio()`.
1384
1385   *David von Oheimb*
1386
1387 * Deprecated `OCSP_parse_url()`.
1388
1389   *David von Oheimb*
1390
1391 * Validation of SM2 keys has been separated from the validation of regular EC
1392   keys.
1393
1394   *Nicola Tuveri*
1395
1396 * Behavior of the `pkey` app is changed, when using the `-check` or `-pubcheck`
1397   switches: a validation failure triggers an early exit, returning a failure
1398   exit status to the parent process.
1399
1400   *Nicola Tuveri*
1401
1402 * Changed behavior of SSL_CTX_set_ciphersuites() and SSL_set_ciphersuites()
1403   to ignore unknown ciphers.
1404
1405   *Otto Hollmann*
1406
1407 * The `-cipher-commands` and `-digest-commands` options
1408   of the command line utility `list` have been deprecated.
1409   Instead use the `-cipher-algorithms` and `-digest-algorithms` options.
1410
1411   *Dmitry Belyavskiy*
1412
1413 * Added convenience functions for generating asymmetric key pairs:
1414   The 'quick' one-shot (yet somewhat limited) function L<EVP_PKEY_Q_keygen(3)>
1415   and macros for the most common cases: <EVP_RSA_gen(3)> and L<EVP_EC_gen(3)>.
1416
1417   *David von Oheimb*
1418
1419 * All of the low level EC_KEY functions have been deprecated.
1420
1421   *Shane Lontis, Paul Dale, Richard Levitte, and Tomáš Mráz*
1422
1423 * Deprecated all the libcrypto and libssl error string loading
1424   functions.
1425
1426   *Richard Levitte*
1427
1428 * The functions SSL_CTX_set_tmp_dh_callback and SSL_set_tmp_dh_callback, as
1429   well as the macros SSL_CTX_set_tmp_dh() and SSL_set_tmp_dh() have been
1430   deprecated.
1431
1432   *Matt Caswell*
1433
1434 * The `-crypt` option to the `passwd` command line tool has been removed.
1435
1436   *Paul Dale*
1437
1438 * The -C option to the `x509`, `dhparam`, `dsaparam`, and `ecparam` commands
1439   were removed.
1440
1441   *Rich Salz*
1442
1443 * Add support for AES Key Wrap inverse ciphers to the EVP layer.
1444
1445   *Shane Lontis*
1446
1447 * Deprecated EVP_PKEY_set1_tls_encodedpoint() and
1448   EVP_PKEY_get1_tls_encodedpoint().
1449
1450   *Matt Caswell*
1451
1452 * The security callback, which can be customised by application code, supports
1453   the security operation SSL_SECOP_TMP_DH. One location of the "other" parameter
1454   was incorrectly passing a DH object. It now passed an EVP_PKEY in all cases.
1455
1456   *Matt Caswell*
1457
1458 * Add PKCS7_get_octet_string() and PKCS7_type_is_other() to the public
1459   interface. Their functionality remains unchanged.
1460
1461   *Jordan Montgomery*
1462
1463 * Added new option for 'openssl list', '-providers', which will display the
1464   list of loaded providers, their names, version and status.  It optionally
1465   displays their gettable parameters.
1466
1467   *Paul Dale*
1468
1469 * Removed EVP_PKEY_set_alias_type().
1470
1471   *Richard Levitte*
1472
1473 * Deprecated `EVP_PKEY_CTX_set_rsa_keygen_pubexp()` and introduced
1474   `EVP_PKEY_CTX_set1_rsa_keygen_pubexp()`, which is now preferred.
1475
1476   *Jeremy Walch*
1477
1478 * Changed all "STACK" functions to be macros instead of inline functions. Macro
1479   parameters are still checked for type safety at compile time via helper
1480   inline functions.
1481
1482   *Matt Caswell*
1483
1484 * Remove the RAND_DRBG API
1485
1486   *Paul Dale and Matthias St. Pierre*
1487
1488 * Allow `SSL_set1_host()` and `SSL_add1_host()` to take IP literal addresses
1489   as well as actual hostnames.
1490
1491   *David Woodhouse*
1492
1493 * The 'MinProtocol' and 'MaxProtocol' configuration commands now silently
1494   ignore TLS protocol version bounds when configuring DTLS-based contexts, and
1495   conversely, silently ignore DTLS protocol version bounds when configuring
1496   TLS-based contexts.  The commands can be repeated to set bounds of both
1497   types.  The same applies with the corresponding "min_protocol" and
1498   "max_protocol" command-line switches, in case some application uses both TLS
1499   and DTLS.
1500
1501   SSL_CTX instances that are created for a fixed protocol version (e.g.
1502   `TLSv1_server_method()`) also silently ignore version bounds.  Previously
1503   attempts to apply bounds to these protocol versions would result in an
1504   error.  Now only the "version-flexible" SSL_CTX instances are subject to
1505   limits in configuration files in command-line options.
1506
1507   *Viktor Dukhovni*
1508
1509 * Deprecated the `ENGINE` API.  Engines should be replaced with providers
1510   going forward.
1511
1512   *Paul Dale*
1513
1514 * Reworked the recorded ERR codes to make better space for system errors.
1515   To distinguish them, the macro `ERR_SYSTEM_ERROR()` indicates if the
1516   given code is a system error (true) or an OpenSSL error (false).
1517
1518   *Richard Levitte*
1519
1520 * Reworked the test perl framework to better allow parallel testing.
1521
1522   *Nicola Tuveri and David von Oheimb*
1523
1524 * Added ciphertext stealing algorithms AES-128-CBC-CTS, AES-192-CBC-CTS and
1525   AES-256-CBC-CTS to the providers. CS1, CS2 and CS3 variants are supported.
1526
1527   *Shane Lontis*
1528
1529 * 'Configure' has been changed to figure out the configuration target if
1530   none is given on the command line.  Consequently, the 'config' script is
1531   now only a mere wrapper.  All documentation is changed to only mention
1532   'Configure'.
1533
1534   *Rich Salz and Richard Levitte*
1535
1536 * Added a library context `OSSL_LIB_CTX` that applications as well as
1537   other libraries can use to form a separate context within which
1538   libcrypto operations are performed.
1539
1540   *Richard Levitte*
1541
1542 * Added various `_ex` functions to the OpenSSL API that support using
1543   a non-default `OSSL_LIB_CTX`.
1544
1545   *OpenSSL team*
1546
1547 * Handshake now fails if Extended Master Secret extension is dropped
1548   on renegotiation.
1549
1550   *Tomáš Mráz*
1551
1552 * Dropped interactive mode from the `openssl` program.
1553
1554   *Richard Levitte*
1555
1556 * Deprecated `EVP_PKEY_cmp()` and `EVP_PKEY_cmp_parameters()`.
1557
1558   *David von Oheimb and Shane Lontis*
1559
1560 * Deprecated `EC_METHOD_get_field_type()`.
1561
1562   *Billy Bob Brumley*
1563
1564 * Deprecated EC_GFp_simple_method(), EC_GFp_mont_method(),
1565   EC_GF2m_simple_method(), EC_GFp_nist_method(), EC_GFp_nistp224_method()
1566   EC_GFp_nistp256_method(), and EC_GFp_nistp521_method().
1567
1568   *Billy Bob Brumley*
1569
1570 * Deprecated EC_GROUP_new(), EC_GROUP_method_of(), and EC_POINT_method_of().
1571
1572   *Billy Bob Brumley*
1573
1574 * Add CAdES-BES signature verification support, mostly derived
1575   from ESSCertIDv2 TS (RFC 5816) contribution by Marek Klein.
1576
1577   *Filipe Raimundo da Silva*
1578
1579 * Add CAdES-BES signature scheme and attributes support (RFC 5126) to CMS API.
1580
1581   *Antonio Iacono*
1582
1583 * Added the AuthEnvelopedData content type structure (RFC 5083) with AES-GCM
1584   parameter (RFC 5084) for the Cryptographic Message Syntax (CMS).
1585
1586   *Jakub Zelenka*
1587
1588 * Deprecated EC_POINT_make_affine() and EC_POINTs_make_affine().
1589
1590   *Billy Bob Brumley*
1591
1592 * Deprecated EC_GROUP_precompute_mult(), EC_GROUP_have_precompute_mult(), and
1593   EC_KEY_precompute_mult().
1594
1595   *Billy Bob Brumley*
1596
1597 * Deprecated EC_POINTs_mul().
1598
1599   *Billy Bob Brumley*
1600
1601 * Removed FIPS_mode() and FIPS_mode_set().
1602
1603   *Shane Lontis*
1604
1605 * The SSL option SSL_OP_IGNORE_UNEXPECTED_EOF is introduced.
1606
1607   *Dmitry Belyavskiy*
1608
1609 * Deprecated EC_POINT_set_Jprojective_coordinates_GFp() and
1610   EC_POINT_get_Jprojective_coordinates_GFp().
1611
1612   *Billy Bob Brumley*
1613
1614 * Added OSSL_PARAM_BLD to the public interface.  This allows OSSL_PARAM
1615   arrays to be more easily constructed via a series of utility functions.
1616   Create a parameter builder using OSSL_PARAM_BLD_new(), add parameters using
1617   the various push functions and finally convert to a passable OSSL_PARAM
1618   array using OSSL_PARAM_BLD_to_param().
1619
1620   *Paul Dale*
1621
1622 * The security strength of SHA1 and MD5 based signatures in TLS has been
1623   reduced.
1624
1625   *Kurt Roeckx*
1626
1627 * Added EVP_PKEY_set_type_by_keymgmt(), to initialise an EVP_PKEY to
1628   contain a provider side internal key.
1629
1630   *Richard Levitte*
1631
1632 * ASN1_verify(), ASN1_digest() and ASN1_sign() have been deprecated.
1633
1634   *Richard Levitte*
1635
1636 * Project text documents not yet having a proper file name extension
1637   (`HACKING`, `LICENSE`, `NOTES*`, `README*`, `VERSION`) have been renamed to
1638   `*.md` as far as reasonable, else `*.txt`, for better use with file managers.
1639
1640   *David von Oheimb*
1641
1642 * The main project documents (README, NEWS, CHANGES, INSTALL, SUPPORT)
1643   have been converted to Markdown with the goal to produce documents
1644   which not only look pretty when viewed online in the browser, but
1645   remain well readable inside a plain text editor.
1646
1647   To achieve this goal, a 'minimalistic' Markdown style has been applied
1648   which avoids formatting elements that interfere too much with the
1649   reading flow in the text file. For example, it
1650
1651   * avoids [ATX headings][] and uses [setext headings][] instead
1652     (which works for `<h1>` and `<h2>` headings only).
1653   * avoids [inline links][] and uses [reference links][] instead.
1654   * avoids [fenced code blocks][] and uses [indented code blocks][] instead.
1655
1656     [ATX headings]:         https://github.github.com/gfm/#atx-headings
1657     [setext headings]:      https://github.github.com/gfm/#setext-headings
1658     [inline links]:         https://github.github.com/gfm/#inline-link
1659     [reference links]:      https://github.github.com/gfm/#reference-link
1660     [fenced code blocks]:   https://github.github.com/gfm/#fenced-code-blocks
1661     [indented code blocks]: https://github.github.com/gfm/#indented-code-blocks
1662
1663   *Matthias St. Pierre*
1664
1665 * The test suite is changed to preserve results of each test recipe.
1666   A new directory test-runs/ with subdirectories named like the
1667   test recipes are created in the build tree for this purpose.
1668
1669   *Richard Levitte*
1670
1671 * Added an implementation of CMP and CRMF (RFC 4210, RFC 4211 RFC 6712).
1672   This adds `crypto/cmp/`, `crpyto/crmf/`, `apps/cmp.c`, and `test/cmp_*`.
1673   See L<openssl-cmp(1)> and L<OSSL_CMP_exec_IR_ses(3)> as starting points.
1674
1675   *David von Oheimb, Martin Peylo*
1676
1677 * Generalized the HTTP client code from `crypto/ocsp/` into `crpyto/http/`.
1678   It supports arbitrary request and response content types, GET redirection,
1679   TLS, connections via HTTP(S) proxies, connections and exchange via
1680   user-defined BIOs (allowing implicit connections), persistent connections,
1681   and timeout checks.  See L<OSSL_HTTP_transfer(3)> etc. for details.
1682   The legacy OCSP-focused (and only partly documented) API
1683   is retained for backward compatibility, while most of it is deprecated.
1684
1685   *David von Oheimb*
1686
1687 * Added `util/check-format.pl`, a tool for checking adherence to the
1688   OpenSSL coding style <https://www.openssl.org/policies/codingstyle.html>.
1689   The checks performed are incomplete and yield some false positives.
1690   Still the tool should be useful for detecting most typical glitches.
1691
1692   *David von Oheimb*
1693
1694 * `BIO_do_connect()` and `BIO_do_handshake()` have been extended:
1695   If domain name resolution yields multiple IP addresses all of them are tried
1696   after `connect()` failures.
1697
1698   *David von Oheimb*
1699
1700 * All of the low level RSA functions have been deprecated.
1701
1702   *Paul Dale*
1703
1704 * X509 certificates signed using SHA1 are no longer allowed at security
1705   level 1 and above.
1706
1707   *Kurt Roeckx*
1708
1709 * The command line utilities dhparam, dsa, gendsa and dsaparam have been
1710   modified to use PKEY APIs.  These commands are now in maintenance mode
1711   and no new features will be added to them.
1712
1713   *Paul Dale*
1714
1715 * The command line utility rsautl has been deprecated.
1716
1717   *Paul Dale*
1718
1719 * The command line utilities genrsa and rsa have been modified to use PKEY
1720   APIs. They now write PKCS#8 keys by default. These commands are now in
1721   maintenance mode and no new features will be added to them.
1722
1723   *Paul Dale*
1724
1725 * All of the low level DH functions have been deprecated.
1726
1727   *Paul Dale and Matt Caswell*
1728
1729 * All of the low level DSA functions have been deprecated.
1730
1731   *Paul Dale*
1732
1733 * Reworked the treatment of EC EVP_PKEYs with the SM2 curve to
1734   automatically become EVP_PKEY_SM2 rather than EVP_PKEY_EC.
1735
1736   *Richard Levitte*
1737
1738 * Deprecated low level ECDH and ECDSA functions.
1739
1740   *Paul Dale*
1741
1742 * Deprecated EVP_PKEY_decrypt_old() and EVP_PKEY_encrypt_old().
1743
1744   *Richard Levitte*
1745
1746 * Enhanced the documentation of EVP_PKEY_get_size(), EVP_PKEY_get_bits()
1747   and EVP_PKEY_get_security_bits().  Especially EVP_PKEY_get_size() needed
1748   a new formulation to include all the things it can be used for,
1749   as well as words of caution.
1750
1751   *Richard Levitte*
1752
1753 * The SSL_CTX_set_tlsext_ticket_key_cb(3) function has been deprecated.
1754
1755   *Paul Dale*
1756
1757 * All of the low level HMAC functions have been deprecated.
1758
1759   *Paul Dale and David von Oheimb*
1760
1761 * Over two thousand fixes were made to the documentation, including:
1762   - Common options (such as -rand/-writerand, TLS version control, etc)
1763     were refactored and point to newly-enhanced descriptions in openssl.pod.
1764   - Added style conformance for all options (with help from Richard Levitte),
1765     documented all reported missing options, added a CI build to check
1766     that all options are documented and that no unimplemented options
1767     are documented.
1768   - Documented some internals, such as all use of environment variables.
1769   - Addressed all internal broken L<> references.
1770
1771   *Rich Salz*
1772
1773 * All of the low level CMAC functions have been deprecated.
1774
1775   *Paul Dale*
1776
1777 * The low-level MD2, MD4, MD5, MDC2, RIPEMD160 and Whirlpool digest
1778   functions have been deprecated.
1779
1780   *Paul Dale and David von Oheimb*
1781
1782 * Corrected the documentation of the return values from the `EVP_DigestSign*`
1783   set of functions.  The documentation mentioned negative values for some
1784   errors, but this was never the case, so the mention of negative values
1785   was removed.
1786
1787   Code that followed the documentation and thereby check with something
1788   like `EVP_DigestSignInit(...) <= 0` will continue to work undisturbed.
1789
1790   *Richard Levitte*
1791
1792 * All of the low level cipher functions have been deprecated.
1793
1794   *Matt Caswell and Paul Dale*
1795
1796 * Removed include/openssl/opensslconf.h.in and replaced it with
1797   include/openssl/configuration.h.in, which differs in not including
1798   <openssl/macros.h>.  A short header include/openssl/opensslconf.h
1799   was added to include both.
1800
1801   This allows internal hacks where one might need to modify the set
1802   of configured macros, for example this if deprecated symbols are
1803   still supposed to be available internally:
1804
1805       #include <openssl/configuration.h>
1806
1807       #undef OPENSSL_NO_DEPRECATED
1808       #define OPENSSL_SUPPRESS_DEPRECATED
1809
1810       #include <openssl/macros.h>
1811
1812   This should not be used by applications that use the exported
1813   symbols, as that will lead to linking errors.
1814
1815   *Richard Levitte*
1816
1817 * Fixed an overflow bug in the x64_64 Montgomery squaring procedure
1818   used in exponentiation with 512-bit moduli. No EC algorithms are
1819   affected. Analysis suggests that attacks against 2-prime RSA1024,
1820   3-prime RSA1536, and DSA1024 as a result of this defect would be very
1821   difficult to perform and are not believed likely. Attacks against DH512
1822   are considered just feasible. However, for an attack the target would
1823   have to re-use the DH512 private key, which is not recommended anyway.
1824   Also applications directly using the low-level API BN_mod_exp may be
1825   affected if they use BN_FLG_CONSTTIME.
1826   ([CVE-2019-1551])
1827
1828   *Andy Polyakov*
1829
1830 * Most memory-debug features have been deprecated, and the functionality
1831   replaced with no-ops.
1832
1833   *Rich Salz*
1834
1835 * Added documentation for the STACK API.
1836
1837   *Rich Salz*
1838
1839 * Introduced a new method type and API, OSSL_ENCODER, to represent
1840   generic encoders.  These do the same sort of job that PEM writers
1841   and d2i functions do, but with support for methods supplied by
1842   providers, and the possibility for providers to support other
1843   formats as well.
1844
1845   *Richard Levitte*
1846
1847 * Introduced a new method type and API, OSSL_DECODER, to represent
1848   generic decoders.  These do the same sort of job that PEM readers
1849   and i2d functions do, but with support for methods supplied by
1850   providers, and the possibility for providers to support other
1851   formats as well.
1852
1853   *Richard Levitte*
1854
1855 * Added a .pragma directive to the syntax of configuration files, to
1856   allow varying behavior in a supported and predictable manner.
1857   Currently added pragma:
1858
1859           .pragma dollarid:on
1860
1861   This allows dollar signs to be a keyword character unless it's
1862   followed by a opening brace or parenthesis.  This is useful for
1863   platforms where dollar signs are commonly used in names, such as
1864   volume names and system directory names on VMS.
1865
1866   *Richard Levitte*
1867
1868 * Added functionality to create an EVP_PKEY from user data.
1869
1870   *Richard Levitte*
1871
1872 * Change the interpretation of the '--api' configuration option to
1873   mean that this is a desired API compatibility level with no
1874   further meaning.  The previous interpretation, that this would
1875   also mean to remove all deprecated symbols up to and including
1876   the given version, no requires that 'no-deprecated' is also used
1877   in the configuration.
1878
1879   When building applications, the desired API compatibility level
1880   can be set with the OPENSSL_API_COMPAT macro like before.  For
1881   API compatibility version below 3.0, the old style numerical
1882   value is valid as before, such as -DOPENSSL_API_COMPAT=0x10100000L.
1883   For version 3.0 and on, the value is expected to be the decimal
1884   value calculated from the major and minor version like this:
1885
1886           MAJOR * 10000 + MINOR * 100
1887
1888   Examples:
1889
1890           -DOPENSSL_API_COMPAT=30000             For 3.0
1891           -DOPENSSL_API_COMPAT=30200             For 3.2
1892
1893   To hide declarations that are deprecated up to and including the
1894   given API compatibility level, -DOPENSSL_NO_DEPRECATED must be
1895   given when building the application as well.
1896
1897   *Richard Levitte*
1898
1899 * Added the X509_LOOKUP_METHOD called X509_LOOKUP_store, to allow
1900   access to certificate and CRL stores via URIs and OSSL_STORE
1901   loaders.
1902
1903   This adds the following functions:
1904
1905   - X509_LOOKUP_store()
1906   - X509_STORE_load_file()
1907   - X509_STORE_load_path()
1908   - X509_STORE_load_store()
1909   - SSL_add_store_cert_subjects_to_stack()
1910   - SSL_CTX_set_default_verify_store()
1911   - SSL_CTX_load_verify_file()
1912   - SSL_CTX_load_verify_dir()
1913   - SSL_CTX_load_verify_store()
1914
1915   *Richard Levitte*
1916
1917 * Added a new method to gather entropy on VMS, based on SYS$GET_ENTROPY.
1918   The presence of this system service is determined at run-time.
1919
1920   *Richard Levitte*
1921
1922 * Added functionality to create an EVP_PKEY context based on data
1923   for methods from providers.  This takes an algorithm name and a
1924   property query string and simply stores them, with the intent
1925   that any operation that uses this context will use those strings
1926   to fetch the needed methods implicitly, thereby making the port
1927   of application written for pre-3.0 OpenSSL easier.
1928
1929   *Richard Levitte*
1930
1931 * The undocumented function NCONF_WIN32() has been deprecated; for
1932   conversion details see the HISTORY section of doc/man5/config.pod
1933
1934   *Rich Salz*
1935
1936 * Introduced the new functions EVP_DigestSignInit_ex() and
1937   EVP_DigestVerifyInit_ex(). The macros EVP_DigestSignUpdate() and
1938   EVP_DigestVerifyUpdate() have been converted to functions. See the man
1939   pages for further details.
1940
1941   *Matt Caswell*
1942
1943 * Over two thousand fixes were made to the documentation, including:
1944   adding missing command flags, better style conformance, documentation
1945   of internals, etc.
1946
1947   *Rich Salz, Richard Levitte*
1948
1949 * s390x assembly pack: add hardware-support for P-256, P-384, P-521,
1950   X25519, X448, Ed25519 and Ed448.
1951
1952   *Patrick Steuer*
1953
1954 * Print all values for a PKCS#12 attribute with 'openssl pkcs12', not just
1955   the first value.
1956
1957   *Jon Spillett*
1958
1959 * Deprecated the public definition of `ERR_STATE` as well as the function
1960   `ERR_get_state()`.  This is done in preparation of making `ERR_STATE` an
1961   opaque type.
1962
1963   *Richard Levitte*
1964
1965 * Added ERR functionality to give callers access to the stored function
1966   names that have replaced the older function code based functions.
1967
1968   New functions are ERR_peek_error_func(), ERR_peek_last_error_func(),
1969   ERR_peek_error_data(), ERR_peek_last_error_data(), ERR_get_error_all(),
1970   ERR_peek_error_all() and ERR_peek_last_error_all().
1971
1972   Deprecate ERR functions ERR_get_error_line(), ERR_get_error_line_data(),
1973   ERR_peek_error_line_data(), ERR_peek_last_error_line_data() and
1974   ERR_func_error_string().
1975
1976   *Richard Levitte*
1977
1978 * Extended testing to be verbose for failing tests only.  The make variables
1979   VERBOSE_FAILURE or VF can be used to enable this:
1980
1981           $ make VF=1 test                           # Unix
1982           $ mms /macro=(VF=1) test                   ! OpenVMS
1983           $ nmake VF=1 test                          # Windows
1984
1985   *Richard Levitte*
1986
1987 * Added the `-copy_extensions` option to the `x509` command for use with
1988   `-req` and `-x509toreq`. When given with the `copy` or `copyall` argument,
1989   all extensions in the request are copied to the certificate or vice versa.
1990
1991   *David von Oheimb*, *Kirill Stefanenkov <kirill_stefanenkov@rambler.ru>*
1992
1993 * Added the `-copy_extensions` option to the `req` command for use with
1994   `-x509`. When given with the `copy` or `copyall` argument,
1995   all extensions in the certification request are copied to the certificate.
1996
1997   *David von Oheimb*
1998
1999 * The `x509`, `req`, and `ca` commands now make sure that X.509v3 certificates
2000   they generate are by default RFC 5280 compliant in the following sense:
2001   There is a subjectKeyIdentifier extension with a hash value of the public key
2002   and for not self-signed certs there is an authorityKeyIdentifier extension
2003   with a keyIdentifier field or issuer information identifying the signing key.
2004   This is done unless some configuration overrides the new default behavior,
2005   such as `subjectKeyIdentifier = none` and `authorityKeyIdentifier = none`.
2006
2007   *David von Oheimb*
2008
2009 * Added several checks to `X509_verify_cert()` according to requirements in
2010   RFC 5280 in case `X509_V_FLAG_X509_STRICT` is set
2011   (which may be done by using the CLI option `-x509_strict`):
2012   * The basicConstraints of CA certificates must be marked critical.
2013   * CA certificates must explicitly include the keyUsage extension.
2014   * If a pathlenConstraint is given the key usage keyCertSign must be allowed.
2015   * The issuer name of any certificate must not be empty.
2016   * The subject name of CA certs, certs with keyUsage crlSign,
2017     and certs without subjectAlternativeName must not be empty.
2018   * If a subjectAlternativeName extension is given it must not be empty.
2019   * The signatureAlgorithm field and the cert signature must be consistent.
2020   * Any given authorityKeyIdentifier and any given subjectKeyIdentifier
2021     must not be marked critical.
2022   * The authorityKeyIdentifier must be given for X.509v3 certs
2023     unless they are self-signed.
2024   * The subjectKeyIdentifier must be given for all X.509v3 CA certs.
2025
2026   *David von Oheimb*
2027
2028 * Certificate verification using `X509_verify_cert()` meanwhile rejects EC keys
2029   with explicit curve parameters (specifiedCurve) as required by RFC 5480.
2030
2031   *Tomáš Mráz*
2032
2033 * For built-in EC curves, ensure an EC_GROUP built from the curve name is
2034   used even when parsing explicit parameters, when loading a encoded key
2035   or calling `EC_GROUP_new_from_ecpkparameters()`/
2036   `EC_GROUP_new_from_ecparameters()`.
2037   This prevents bypass of security hardening and performance gains,
2038   especially for curves with specialized EC_METHODs.
2039   By default, if a key encoded with explicit parameters is loaded and later
2040   encoded, the output is still encoded with explicit parameters, even if
2041   internally a "named" EC_GROUP is used for computation.
2042
2043   *Nicola Tuveri*
2044
2045 * Compute ECC cofactors if not provided during EC_GROUP construction. Before
2046   this change, EC_GROUP_set_generator would accept order and/or cofactor as
2047   NULL. After this change, only the cofactor parameter can be NULL. It also
2048   does some minimal sanity checks on the passed order.
2049   ([CVE-2019-1547])
2050
2051   *Billy Bob Brumley*
2052
2053 * Fixed a padding oracle in PKCS7_dataDecode and CMS_decrypt_set1_pkey.
2054   An attack is simple, if the first CMS_recipientInfo is valid but the
2055   second CMS_recipientInfo is chosen ciphertext. If the second
2056   recipientInfo decodes to PKCS #1 v1.5 form plaintext, the correct
2057   encryption key will be replaced by garbage, and the message cannot be
2058   decoded, but if the RSA decryption fails, the correct encryption key is
2059   used and the recipient will not notice the attack.
2060   As a work around for this potential attack the length of the decrypted
2061   key must be equal to the cipher default key length, in case the
2062   certifiate is not given and all recipientInfo are tried out.
2063   The old behaviour can be re-enabled in the CMS code by setting the
2064   CMS_DEBUG_DECRYPT flag.
2065
2066   *Bernd Edlinger*
2067
2068 * Early start up entropy quality from the DEVRANDOM seed source has been
2069   improved for older Linux systems.  The RAND subsystem will wait for
2070   /dev/random to be producing output before seeding from /dev/urandom.
2071   The seeded state is stored for future library initialisations using
2072   a system global shared memory segment.  The shared memory identifier
2073   can be configured by defining OPENSSL_RAND_SEED_DEVRANDOM_SHM_ID to
2074   the desired value.  The default identifier is 114.
2075
2076   *Paul Dale*
2077
2078 * Revised BN_generate_prime_ex to not avoid factors 2..17863 in p-1
2079   when primes for RSA keys are computed.
2080   Since we previously always generated primes == 2 (mod 3) for RSA keys,
2081   the 2-prime and 3-prime RSA modules were easy to distinguish, since
2082   `N = p*q = 1 (mod 3)`, but `N = p*q*r = 2 (mod 3)`. Therefore fingerprinting
2083   2-prime vs. 3-prime RSA keys was possible by computing N mod 3.
2084   This avoids possible fingerprinting of newly generated RSA modules.
2085
2086   *Bernd Edlinger*
2087
2088 * Correct the extended master secret constant on EBCDIC systems. Without this
2089   fix TLS connections between an EBCDIC system and a non-EBCDIC system that
2090   negotiate EMS will fail. Unfortunately this also means that TLS connections
2091   between EBCDIC systems with this fix, and EBCDIC systems without this
2092   fix will fail if they negotiate EMS.
2093
2094   *Matt Caswell*
2095
2096 * Changed the library initialisation so that the config file is now loaded
2097   by default. This was already the case for libssl. It now occurs for both
2098   libcrypto and libssl. Use the OPENSSL_INIT_NO_LOAD_CONFIG option to
2099   `OPENSSL_init_crypto()` to suppress automatic loading of a config file.
2100
2101   *Matt Caswell*
2102
2103 * Introduced new error raising macros, `ERR_raise()` and `ERR_raise_data()`,
2104   where the former acts as a replacement for `ERR_put_error()`, and the
2105   latter replaces the combination `ERR_put_error()` + `ERR_add_error_data()`.
2106   `ERR_raise_data()` adds more flexibility by taking a format string and
2107   an arbitrary number of arguments following it, to be processed with
2108   `BIO_snprintf()`.
2109
2110   *Richard Levitte*
2111
2112 * Introduced a new function, `OSSL_PROVIDER_available()`, which can be used
2113   to check if a named provider is loaded and available.  When called, it
2114   will also activate all fallback providers if such are still present.
2115
2116   *Richard Levitte*
2117
2118 * Enforce a minimum DH modulus size of 512 bits.
2119
2120   *Bernd Edlinger*
2121
2122 * Changed DH parameters to generate the order q subgroup instead of 2q.
2123   Previously generated DH parameters are still accepted by DH_check
2124   but DH_generate_key works around that by clearing bit 0 of the
2125   private key for those. This avoids leaking bit 0 of the private key.
2126
2127   *Bernd Edlinger*
2128
2129 * Significantly reduce secure memory usage by the randomness pools.
2130
2131   *Paul Dale*
2132
2133 * `{CRYPTO,OPENSSL}_mem_debug_{push,pop}` are now no-ops and have been
2134   deprecated.
2135
2136   *Rich Salz*
2137
2138 * A new type, EVP_KEYEXCH, has been introduced to represent key exchange
2139   algorithms. An implementation of a key exchange algorithm can be obtained
2140   by using the function EVP_KEYEXCH_fetch(). An EVP_KEYEXCH algorithm can be
2141   used in a call to EVP_PKEY_derive_init_ex() which works in a similar way to
2142   the older EVP_PKEY_derive_init() function. See the man pages for the new
2143   functions for further details.
2144
2145   *Matt Caswell*
2146
2147 * The EVP_PKEY_CTX_set_dh_pad() macro has now been converted to a function.
2148
2149   *Matt Caswell*
2150
2151 * Removed the function names from error messages and deprecated the
2152   xxx_F_xxx define's.
2153
2154   *Richard Levitte*
2155
2156 * Removed NextStep support and the macro OPENSSL_UNISTD
2157
2158   *Rich Salz*
2159
2160 * Removed DES_check_key.  Also removed OPENSSL_IMPLEMENT_GLOBAL,
2161   OPENSSL_GLOBAL_REF, OPENSSL_DECLARE_GLOBAL.
2162   Also removed "export var as function" capability; we do not export
2163   variables, only functions.
2164
2165   *Rich Salz*
2166
2167 * RC5_32_set_key has been changed to return an int type, with 0 indicating
2168   an error and 1 indicating success. In previous versions of OpenSSL this
2169   was a void type. If a key was set longer than the maximum possible this
2170   would crash.
2171
2172   *Matt Caswell*
2173
2174 * Support SM2 signing and verification schemes with X509 certificate.
2175
2176   *Paul Yang*
2177
2178 * Use SHA256 as the default digest for TS query in the `ts` app.
2179
2180   *Tomáš Mráz*
2181
2182 * Change PBKDF2 to conform to SP800-132 instead of the older PKCS5 RFC2898.
2183
2184   *Shane Lontis*
2185
2186 * Default cipher lists/suites are now available via a function, the
2187   #defines are deprecated.
2188
2189   *Todd Short*
2190
2191 * Add target VC-WIN32-UWP, VC-WIN64A-UWP, VC-WIN32-ARM-UWP and
2192   VC-WIN64-ARM-UWP in Windows OneCore target for making building libraries
2193   for Windows Store apps easier. Also, the "no-uplink" option has been added.
2194
2195   *Kenji Mouri*
2196
2197 * Join the directories crypto/x509 and crypto/x509v3
2198
2199   *Richard Levitte*
2200
2201 * Added command 'openssl kdf' that uses the EVP_KDF API.
2202
2203   *Shane Lontis*
2204
2205 * Added command 'openssl mac' that uses the EVP_MAC API.
2206
2207   *Shane Lontis*
2208
2209 * Added OPENSSL_info() to get diverse built-in OpenSSL data, such
2210   as default directories.  Also added the command 'openssl info'
2211   for scripting purposes.
2212
2213   *Richard Levitte*
2214
2215 * The functions AES_ige_encrypt() and AES_bi_ige_encrypt() have been
2216   deprecated.
2217
2218   *Matt Caswell*
2219
2220 * Add prediction resistance to the DRBG reseeding process.
2221
2222   *Paul Dale*
2223
2224 * Limit the number of blocks in a data unit for AES-XTS to 2^20 as
2225   mandated by IEEE Std 1619-2018.
2226
2227   *Paul Dale*
2228
2229 * Added newline escaping functionality to a filename when using openssl dgst.
2230   This output format is to replicate the output format found in the `*sum`
2231   checksum programs. This aims to preserve backward compatibility.
2232
2233   *Matt Eaton, Richard Levitte, and Paul Dale*
2234
2235 * Removed the heartbeat message in DTLS feature, as it has very
2236   little usage and doesn't seem to fulfill a valuable purpose.
2237   The configuration option is now deprecated.
2238
2239   *Richard Levitte*
2240
2241 * Changed the output of 'openssl {digestname} < file' to display the
2242   digest name in its output.
2243
2244   *Richard Levitte*
2245
2246 * Added a new generic trace API which provides support for enabling
2247   instrumentation through trace output.
2248
2249   *Richard Levitte & Matthias St. Pierre*
2250
2251 * Added build tests for C++.  These are generated files that only do one
2252   thing, to include one public OpenSSL head file each.  This tests that
2253   the public header files can be usefully included in a C++ application.
2254
2255   This test isn't enabled by default.  It can be enabled with the option
2256   'enable-buildtest-c++'.
2257
2258   *Richard Levitte*
2259
2260 * Added KB KDF (EVP_KDF_KB) to EVP_KDF.
2261
2262   *Robbie Harwood*
2263
2264 * Added SSH KDF (EVP_KDF_SSHKDF) and KRB5 KDF (EVP_KDF_KRB5KDF) to EVP_KDF.
2265
2266   *Simo Sorce*
2267
2268 * Added Single Step KDF (EVP_KDF_SS), X963 KDF, and X942 KDF to EVP_KDF.
2269
2270   *Shane Lontis*
2271
2272 * Added KMAC to EVP_MAC.
2273
2274   *Shane Lontis*
2275
2276 * Added property based algorithm implementation selection framework to
2277   the core.
2278
2279   *Paul Dale*
2280
2281 * Added SCA hardening for modular field inversion in EC_GROUP through
2282   a new dedicated field_inv() pointer in EC_METHOD.
2283   This also addresses a leakage affecting conversions from projective
2284   to affine coordinates.
2285
2286   *Billy Bob Brumley, Nicola Tuveri*
2287
2288 * Added EVP_KDF, an EVP layer KDF API, to simplify adding KDF and PRF
2289   implementations.  This includes an EVP_PKEY to EVP_KDF bridge for
2290   those algorithms that were already supported through the EVP_PKEY API
2291   (scrypt, TLS1 PRF and HKDF).  The low-level KDF functions for PBKDF2
2292   and scrypt are now wrappers that call EVP_KDF.
2293
2294   *David Makepeace*
2295
2296 * Build devcrypto engine as a dynamic engine.
2297
2298   *Eneas U de Queiroz*
2299
2300 * Add keyed BLAKE2 to EVP_MAC.
2301
2302   *Antoine Salon*
2303
2304 * Fix a bug in the computation of the endpoint-pair shared secret used
2305   by DTLS over SCTP. This breaks interoperability with older versions
2306   of OpenSSL like OpenSSL 1.1.0 and OpenSSL 1.0.2. There is a runtime
2307   switch SSL_MODE_DTLS_SCTP_LABEL_LENGTH_BUG (off by default) enabling
2308   interoperability with such broken implementations. However, enabling
2309   this switch breaks interoperability with correct implementations.
2310
2311 * Fix a use after free bug in d2i_X509_PUBKEY when overwriting a
2312   re-used X509_PUBKEY object if the second PUBKEY is malformed.
2313
2314   *Bernd Edlinger*
2315
2316 * Move strictness check from EVP_PKEY_asn1_new() to EVP_PKEY_asn1_add0().
2317
2318   *Richard Levitte*
2319
2320 * Changed the license to the Apache License v2.0.
2321
2322   *Richard Levitte*
2323
2324 * Switch to a new version scheme using three numbers MAJOR.MINOR.PATCH.
2325
2326   - Major releases (indicated by incrementing the MAJOR release number)
2327     may introduce incompatible API/ABI changes.
2328   - Minor releases (indicated by incrementing the MINOR release number)
2329     may introduce new features but retain API/ABI compatibility.
2330   - Patch releases (indicated by incrementing the PATCH number)
2331     are intended for bug fixes and other improvements of existing
2332     features only (like improving performance or adding documentation)
2333     and retain API/ABI compatibility.
2334
2335   *Richard Levitte*
2336
2337 * Add support for RFC5297 SIV mode (siv128), including AES-SIV.
2338
2339   *Todd Short*
2340
2341 * Remove the 'dist' target and add a tarball building script.  The
2342   'dist' target has fallen out of use, and it shouldn't be
2343   necessary to configure just to create a source distribution.
2344
2345   *Richard Levitte*
2346
2347 * Recreate the OS390-Unix config target.  It no longer relies on a
2348   special script like it did for OpenSSL pre-1.1.0.
2349
2350   *Richard Levitte*
2351
2352 * Instead of having the source directories listed in Configure, add
2353   a 'build.info' keyword SUBDIRS to indicate what sub-directories to
2354   look into.
2355
2356   *Richard Levitte*
2357
2358 * Add GMAC to EVP_MAC.
2359
2360   *Paul Dale*
2361
2362 * Ported the HMAC, CMAC and SipHash EVP_PKEY_METHODs to EVP_MAC.
2363
2364   *Richard Levitte*
2365
2366 * Added EVP_MAC, an EVP layer MAC API, to simplify adding MAC
2367   implementations.  This includes a generic EVP_PKEY to EVP_MAC bridge,
2368   to facilitate the continued use of MACs through raw private keys in
2369   functionality such as `EVP_DigestSign*` and `EVP_DigestVerify*`.
2370
2371   *Richard Levitte*
2372
2373 * Deprecate ECDH_KDF_X9_62().
2374
2375   *Antoine Salon*
2376
2377 * Added EVP_PKEY_ECDH_KDF_X9_63 and ecdh_KDF_X9_63() as replacements for
2378   the EVP_PKEY_ECDH_KDF_X9_62 KDF type and ECDH_KDF_X9_62(). The old names
2379   are retained for backwards compatibility.
2380
2381   *Antoine Salon*
2382
2383 * AES-XTS mode now enforces that its two keys are different to mitigate
2384   the attacked described in "Efficient Instantiations of Tweakable
2385   Blockciphers and Refinements to Modes OCB and PMAC" by Phillip Rogaway.
2386   Details of this attack can be obtained from:
2387   <http://web.cs.ucdavis.edu/%7Erogaway/papers/offsets.pdf>
2388
2389   *Paul Dale*
2390
2391 * Rename the object files, i.e. give them other names than in previous
2392   versions.  Their names now include the name of the final product, as
2393   well as its type mnemonic (bin, lib, shlib).
2394
2395   *Richard Levitte*
2396
2397 * Added new option for 'openssl list', '-objects', which will display the
2398   list of built in objects, i.e. OIDs with names.
2399
2400   *Richard Levitte*
2401
2402 * Added the options `-crl_lastupdate` and `-crl_nextupdate` to `openssl ca`,
2403   allowing the `lastUpdate` and `nextUpdate` fields in the generated CRL to
2404   be set explicitly.
2405
2406   *Chris Novakovic*
2407
2408 * Added support for Linux Kernel TLS data-path. The Linux Kernel data-path
2409   improves application performance by removing data copies and providing
2410   applications with zero-copy system calls such as sendfile and splice.
2411
2412   *Boris Pismenny*
2413
2414 * The SSL option SSL_OP_CLEANSE_PLAINTEXT is introduced.
2415
2416   *Martin Elshuber*
2417
2418 * `PKCS12_parse` now maintains the order of the parsed certificates
2419   when outputting them via `*ca` (rather than reversing it).
2420
2421   *David von Oheimb*
2422
2423 * Deprecated pthread fork support methods.
2424
2425   *Randall S. Becker*
2426
2427 * Added support for FFDHE key exchange in TLS 1.3.
2428
2429   *Raja Ashok*
2430
2431 * Added a new concept for OpenSSL plugability: providers.  This
2432   functionality is designed to replace the ENGINE API and ENGINE
2433   implementations, and to be much more dynamic, allowing provider
2434   authors to introduce new algorithms among other things, as long as
2435   there's an API that supports the algorithm type.
2436
2437   With this concept comes a new core API for interaction between
2438   libcrypto and provider implementations.  Public libcrypto functions
2439   that want to use providers do so through this core API.
2440
2441   The main documentation for this core API is found in
2442   doc/man7/provider.pod, doc/man7/provider-base.pod, and they in turn
2443   refer to other manuals describing the API specific for supported
2444   algorithm types (also called operations).
2445
2446   *The OpenSSL team*
2447
2448OpenSSL 1.1.1
2449-------------
2450
2451### Changes between 1.1.1l and 1.1.1m [xx XXX xxxx]
2452
2453 * Avoid loading of a dynamic engine twice.
2454
2455   *Bernd Edlinger*
2456
2457 * Prioritise DANE TLSA issuer certs over peer certs
2458
2459   *Viktor Dukhovni*
2460
2461 * Fixed random API for MacOS prior to 10.12
2462
2463   These MacOS versions don't support the CommonCrypto APIs
2464
2465   *Lenny Primak*
2466
2467### Changes between 1.1.1k and 1.1.1l [24 Aug 2021]
2468
2469 * Fixed an SM2 Decryption Buffer Overflow.
2470
2471   In order to decrypt SM2 encrypted data an application is expected to
2472   call the API function EVP_PKEY_decrypt(). Typically an application will
2473   call this function twice. The first time, on entry, the "out" parameter
2474   can be NULL and, on exit, the "outlen" parameter is populated with the
2475   buffer size required to hold the decrypted plaintext. The application
2476   can then allocate a sufficiently sized buffer and call EVP_PKEY_decrypt()
2477   again, but this time passing a non-NULL value for the "out" parameter.
2478
2479   A bug in the implementation of the SM2 decryption code means that the
2480   calculation of the buffer size required to hold the plaintext returned
2481   by the first call to EVP_PKEY_decrypt() can be smaller than the actual
2482   size required by the second call. This can lead to a buffer overflow
2483   when EVP_PKEY_decrypt() is called by the application a second time with
2484   a buffer that is too small.
2485
2486   A malicious attacker who is able present SM2 content for decryption to
2487   an application could cause attacker chosen data to overflow the buffer
2488   by up to a maximum of 62 bytes altering the contents of other data held
2489   after the buffer, possibly changing application behaviour or causing
2490   the application to crash. The location of the buffer is application
2491   dependent but is typically heap allocated.
2492   ([CVE-2021-3711])
2493
2494   *Matt Caswell*
2495
2496 * Fixed various read buffer overruns processing ASN.1 strings
2497
2498   ASN.1 strings are represented internally within OpenSSL as an ASN1_STRING
2499   structure which contains a buffer holding the string data and a field
2500   holding the buffer length. This contrasts with normal C strings which
2501   are repesented as a buffer for the string data which is terminated
2502   with a NUL (0) byte.
2503
2504   Although not a strict requirement, ASN.1 strings that are parsed using
2505   OpenSSL's own "d2i" functions (and other similar parsing functions) as
2506   well as any string whose value has been set with the ASN1_STRING_set()
2507   function will additionally NUL terminate the byte array in the
2508   ASN1_STRING structure.
2509
2510   However, it is possible for applications to directly construct valid
2511   ASN1_STRING structures which do not NUL terminate the byte array by
2512   directly setting the "data" and "length" fields in the ASN1_STRING
2513   array. This can also happen by using the ASN1_STRING_set0() function.
2514
2515   Numerous OpenSSL functions that print ASN.1 data have been found to
2516   assume that the ASN1_STRING byte array will be NUL terminated, even
2517   though this is not guaranteed for strings that have been directly
2518   constructed. Where an application requests an ASN.1 structure to be
2519   printed, and where that ASN.1 structure contains ASN1_STRINGs that have
2520   been directly constructed by the application without NUL terminating
2521   the "data" field, then a read buffer overrun can occur.
2522
2523   The same thing can also occur during name constraints processing
2524   of certificates (for example if a certificate has been directly
2525   constructed by the application instead of loading it via the OpenSSL
2526   parsing functions, and the certificate contains non NUL terminated
2527   ASN1_STRING structures). It can also occur in the X509_get1_email(),
2528   X509_REQ_get1_email() and X509_get1_ocsp() functions.
2529
2530   If a malicious actor can cause an application to directly construct an
2531   ASN1_STRING and then process it through one of the affected OpenSSL
2532   functions then this issue could be hit. This might result in a crash
2533   (causing a Denial of Service attack). It could also result in the
2534   disclosure of private memory contents (such as private keys, or
2535   sensitive plaintext).
2536   ([CVE-2021-3712])
2537
2538   *Matt Caswell*
2539
2540### Changes between 1.1.1j and 1.1.1k [25 Mar 2021]
2541
2542 * Fixed a problem with verifying a certificate chain when using the
2543   X509_V_FLAG_X509_STRICT flag. This flag enables additional security checks of
2544   the certificates present in a certificate chain. It is not set by default.
2545
2546   Starting from OpenSSL version 1.1.1h a check to disallow certificates in
2547   the chain that have explicitly encoded elliptic curve parameters was added
2548   as an additional strict check.
2549
2550   An error in the implementation of this check meant that the result of a
2551   previous check to confirm that certificates in the chain are valid CA
2552   certificates was overwritten. This effectively bypasses the check
2553   that non-CA certificates must not be able to issue other certificates.
2554
2555   If a "purpose" has been configured then there is a subsequent opportunity
2556   for checks that the certificate is a valid CA.  All of the named "purpose"
2557   values implemented in libcrypto perform this check.  Therefore, where
2558   a purpose is set the certificate chain will still be rejected even when the
2559   strict flag has been used. A purpose is set by default in libssl client and
2560   server certificate verification routines, but it can be overridden or
2561   removed by an application.
2562
2563   In order to be affected, an application must explicitly set the
2564   X509_V_FLAG_X509_STRICT verification flag and either not set a purpose
2565   for the certificate verification or, in the case of TLS client or server
2566   applications, override the default purpose.
2567   ([CVE-2021-3450])
2568
2569   *Tomáš Mráz*
2570
2571 * Fixed an issue where an OpenSSL TLS server may crash if sent a maliciously
2572   crafted renegotiation ClientHello message from a client. If a TLSv1.2
2573   renegotiation ClientHello omits the signature_algorithms extension (where it
2574   was present in the initial ClientHello), but includes a
2575   signature_algorithms_cert extension then a NULL pointer dereference will
2576   result, leading to a crash and a denial of service attack.
2577
2578   A server is only vulnerable if it has TLSv1.2 and renegotiation enabled
2579   (which is the default configuration). OpenSSL TLS clients are not impacted by
2580   this issue.
2581   ([CVE-2021-3449])
2582
2583   *Peter Kästle and Samuel Sapalski*
2584
2585### Changes between 1.1.1i and 1.1.1j [16 Feb 2021]
2586
2587 * Fixed the X509_issuer_and_serial_hash() function. It attempts to
2588   create a unique hash value based on the issuer and serial number data
2589   contained within an X509 certificate. However it was failing to correctly
2590   handle any errors that may occur while parsing the issuer field (which might
2591   occur if the issuer field is maliciously constructed). This may subsequently
2592   result in a NULL pointer deref and a crash leading to a potential denial of
2593   service attack.
2594   ([CVE-2021-23841])
2595
2596   *Matt Caswell*
2597
2598 * Fixed the RSA_padding_check_SSLv23() function and the RSA_SSLV23_PADDING
2599   padding mode to correctly check for rollback attacks. This is considered a
2600   bug in OpenSSL 1.1.1 because it does not support SSLv2. In 1.0.2 this is
2601   CVE-2021-23839.
2602
2603   *Matt Caswell*
2604
2605   Fixed the EVP_CipherUpdate, EVP_EncryptUpdate and EVP_DecryptUpdate
2606   functions. Previously they could overflow the output length argument in some
2607   cases where the input length is close to the maximum permissable length for
2608   an integer on the platform. In such cases the return value from the function
2609   call would be 1 (indicating success), but the output length value would be
2610   negative. This could cause applications to behave incorrectly or crash.
2611   ([CVE-2021-23840])
2612
2613   *Matt Caswell*
2614
2615 * Fixed SRP_Calc_client_key so that it runs in constant time. The previous
2616   implementation called BN_mod_exp without setting BN_FLG_CONSTTIME. This
2617   could be exploited in a side channel attack to recover the password. Since
2618   the attack is local host only this is outside of the current OpenSSL
2619   threat model and therefore no CVE is assigned.
2620
2621   Thanks to Mohammed Sabt and Daniel De Almeida Braga for reporting this
2622   issue.
2623
2624   *Matt Caswell*
2625
2626### Changes between 1.1.1h and 1.1.1i [8 Dec 2020]
2627
2628 * Fixed NULL pointer deref in the GENERAL_NAME_cmp function
2629   This function could crash if both GENERAL_NAMEs contain an EDIPARTYNAME.
2630    If an attacker can control both items being compared  then this could lead
2631    to a possible denial of service attack. OpenSSL itself uses the
2632    GENERAL_NAME_cmp function for two purposes:
2633    1) Comparing CRL distribution point names between an available CRL and a
2634       CRL distribution point embedded in an X509 certificate
2635    2) When verifying that a timestamp response token signer matches the
2636       timestamp authority name (exposed via the API functions
2637       TS_RESP_verify_response and TS_RESP_verify_token)
2638   ([CVE-2020-1971])
2639
2640   *Matt Caswell*
2641
2642### Changes between 1.1.1g and 1.1.1h [22 Sep 2020]
2643
2644 * Certificates with explicit curve parameters are now disallowed in
2645   verification chains if the X509_V_FLAG_X509_STRICT flag is used.
2646
2647   *Tomáš Mráz*
2648
2649 * The 'MinProtocol' and 'MaxProtocol' configuration commands now silently
2650   ignore TLS protocol version bounds when configuring DTLS-based contexts, and
2651   conversely, silently ignore DTLS protocol version bounds when configuring
2652   TLS-based contexts.  The commands can be repeated to set bounds of both
2653   types.  The same applies with the corresponding "min_protocol" and
2654   "max_protocol" command-line switches, in case some application uses both TLS
2655   and DTLS.
2656
2657   SSL_CTX instances that are created for a fixed protocol version (e.g.
2658   TLSv1_server_method()) also silently ignore version bounds.  Previously
2659   attempts to apply bounds to these protocol versions would result in an
2660   error.  Now only the "version-flexible" SSL_CTX instances are subject to
2661   limits in configuration files in command-line options.
2662
2663   *Viktor Dukhovni*
2664
2665 * Handshake now fails if Extended Master Secret extension is dropped
2666   on renegotiation.
2667
2668   *Tomáš Mráz*
2669
2670 * The Oracle Developer Studio compiler will start reporting deprecated APIs
2671
2672### Changes between 1.1.1f and 1.1.1g [21 Apr 2020]
2673
2674 * Fixed segmentation fault in SSL_check_chain()
2675   Server or client applications that call the SSL_check_chain() function
2676   during or after a TLS 1.3 handshake may crash due to a NULL pointer
2677   dereference as a result of incorrect handling of the
2678   "signature_algorithms_cert" TLS extension. The crash occurs if an invalid
2679   or unrecognised signature algorithm is received from the peer. This could
2680   be exploited by a malicious peer in a Denial of Service attack.
2681   ([CVE-2020-1967])
2682
2683   *Benjamin Kaduk*
2684
2685 * Added AES consttime code for no-asm configurations
2686   an optional constant time support for AES was added
2687   when building openssl for no-asm.
2688   Enable with: ./config no-asm -DOPENSSL_AES_CONST_TIME
2689   Disable with: ./config no-asm -DOPENSSL_NO_AES_CONST_TIME
2690   At this time this feature is by default disabled.
2691   It will be enabled by default in 3.0.
2692
2693   *Bernd Edlinger*
2694
2695### Changes between 1.1.1e and 1.1.1f [31 Mar 2020]
2696
2697 * Revert the change of EOF detection while reading in libssl to avoid
2698   regressions in applications depending on the current way of reporting
2699   the EOF. As the existing method is not fully accurate the change to
2700   reporting the EOF via SSL_ERROR_SSL is kept on the current development
2701   branch and will be present in the 3.0 release.
2702
2703   *Tomáš Mráz*
2704
2705 * Revised BN_generate_prime_ex to not avoid factors 3..17863 in p-1
2706   when primes for RSA keys are computed.
2707   Since we previously always generated primes == 2 (mod 3) for RSA keys,
2708   the 2-prime and 3-prime RSA modules were easy to distinguish, since
2709   N = p*q = 1 (mod 3), but N = p*q*r = 2 (mod 3). Therefore fingerprinting
2710   2-prime vs. 3-prime RSA keys was possible by computing N mod 3.
2711   This avoids possible fingerprinting of newly generated RSA modules.
2712
2713   *Bernd Edlinger*
2714
2715### Changes between 1.1.1d and 1.1.1e [17 Mar 2020]
2716
2717 * Properly detect EOF while reading in libssl. Previously if we hit an EOF
2718   while reading in libssl then we would report an error back to the
2719   application (SSL_ERROR_SYSCALL) but errno would be 0. We now add
2720   an error to the stack (which means we instead return SSL_ERROR_SSL) and
2721   therefore give a hint as to what went wrong.
2722
2723   *Matt Caswell*
2724
2725 * Check that ed25519 and ed448 are allowed by the security level. Previously
2726   signature algorithms not using an MD were not being checked that they were
2727   allowed by the security level.
2728
2729   *Kurt Roeckx*
2730
2731 * Fixed SSL_get_servername() behaviour. The behaviour of SSL_get_servername()
2732   was not quite right. The behaviour was not consistent between resumption
2733   and normal handshakes, and also not quite consistent with historical
2734   behaviour. The behaviour in various scenarios has been clarified and
2735   it has been updated to make it match historical behaviour as closely as
2736   possible.
2737
2738   *Matt Caswell*
2739
2740 * *[VMS only]* The header files that the VMS compilers include automatically,
2741   `__DECC_INCLUDE_PROLOGUE.H` and `__DECC_INCLUDE_EPILOGUE.H`, use pragmas
2742   that the C++ compiler doesn't understand.  This is a shortcoming in the
2743   compiler, but can be worked around with `__cplusplus` guards.
2744
2745   C++ applications that use OpenSSL libraries must be compiled using the
2746   qualifier `/NAMES=(AS_IS,SHORTENED)` to be able to use all the OpenSSL
2747   functions.  Otherwise, only functions with symbols of less than 31
2748   characters can be used, as the linker will not be able to successfully
2749   resolve symbols with longer names.
2750
2751   *Richard Levitte*
2752
2753 * Added a new method to gather entropy on VMS, based on SYS$GET_ENTROPY.
2754   The presence of this system service is determined at run-time.
2755
2756   *Richard Levitte*
2757
2758 * Print all values for a PKCS#12 attribute with 'openssl pkcs12', not just
2759   the first value.
2760
2761   *Jon Spillett*
2762
2763### Changes between 1.1.1c and 1.1.1d [10 Sep 2019]
2764
2765 * Fixed a fork protection issue. OpenSSL 1.1.1 introduced a rewritten random
2766   number generator (RNG). This was intended to include protection in the
2767   event of a fork() system call in order to ensure that the parent and child
2768   processes did not share the same RNG state. However this protection was not
2769   being used in the default case.
2770
2771   A partial mitigation for this issue is that the output from a high
2772   precision timer is mixed into the RNG state so the likelihood of a parent
2773   and child process sharing state is significantly reduced.
2774
2775   If an application already calls OPENSSL_init_crypto() explicitly using
2776   OPENSSL_INIT_ATFORK then this problem does not occur at all.
2777   ([CVE-2019-1549])
2778
2779   *Matthias St. Pierre*
2780
2781 * For built-in EC curves, ensure an EC_GROUP built from the curve name is
2782   used even when parsing explicit parameters, when loading a encoded key
2783   or calling `EC_GROUP_new_from_ecpkparameters()`/
2784   `EC_GROUP_new_from_ecparameters()`.
2785   This prevents bypass of security hardening and performance gains,
2786   especially for curves with specialized EC_METHODs.
2787   By default, if a key encoded with explicit parameters is loaded and later
2788   encoded, the output is still encoded with explicit parameters, even if
2789   internally a "named" EC_GROUP is used for computation.
2790
2791   *Nicola Tuveri*
2792
2793 * Compute ECC cofactors if not provided during EC_GROUP construction. Before
2794   this change, EC_GROUP_set_generator would accept order and/or cofactor as
2795   NULL. After this change, only the cofactor parameter can be NULL. It also
2796   does some minimal sanity checks on the passed order.
2797   ([CVE-2019-1547])
2798
2799   *Billy Bob Brumley*
2800
2801 * Fixed a padding oracle in PKCS7_dataDecode and CMS_decrypt_set1_pkey.
2802   An attack is simple, if the first CMS_recipientInfo is valid but the
2803   second CMS_recipientInfo is chosen ciphertext. If the second
2804   recipientInfo decodes to PKCS #1 v1.5 form plaintext, the correct
2805   encryption key will be replaced by garbage, and the message cannot be
2806   decoded, but if the RSA decryption fails, the correct encryption key is
2807   used and the recipient will not notice the attack.
2808   As a work around for this potential attack the length of the decrypted
2809   key must be equal to the cipher default key length, in case the
2810   certifiate is not given and all recipientInfo are tried out.
2811   The old behaviour can be re-enabled in the CMS code by setting the
2812   CMS_DEBUG_DECRYPT flag.
2813   ([CVE-2019-1563])
2814
2815   *Bernd Edlinger*
2816
2817 * Early start up entropy quality from the DEVRANDOM seed source has been
2818   improved for older Linux systems.  The RAND subsystem will wait for
2819   /dev/random to be producing output before seeding from /dev/urandom.
2820   The seeded state is stored for future library initialisations using
2821   a system global shared memory segment.  The shared memory identifier
2822   can be configured by defining OPENSSL_RAND_SEED_DEVRANDOM_SHM_ID to
2823   the desired value.  The default identifier is 114.
2824
2825   *Paul Dale*
2826
2827 * Correct the extended master secret constant on EBCDIC systems. Without this
2828   fix TLS connections between an EBCDIC system and a non-EBCDIC system that
2829   negotiate EMS will fail. Unfortunately this also means that TLS connections
2830   between EBCDIC systems with this fix, and EBCDIC systems without this
2831   fix will fail if they negotiate EMS.
2832
2833   *Matt Caswell*
2834
2835 * Use Windows installation paths in the mingw builds
2836
2837   Mingw isn't a POSIX environment per se, which means that Windows
2838   paths should be used for installation.
2839   ([CVE-2019-1552])
2840
2841   *Richard Levitte*
2842
2843 * Changed DH_check to accept parameters with order q and 2q subgroups.
2844   With order 2q subgroups the bit 0 of the private key is not secret
2845   but DH_generate_key works around that by clearing bit 0 of the
2846   private key for those. This avoids leaking bit 0 of the private key.
2847
2848   *Bernd Edlinger*
2849
2850 * Significantly reduce secure memory usage by the randomness pools.
2851
2852   *Paul Dale*
2853
2854 * Revert the DEVRANDOM_WAIT feature for Linux systems
2855
2856   The DEVRANDOM_WAIT feature added a select() call to wait for the
2857   /dev/random device to become readable before reading from the
2858   /dev/urandom device.
2859
2860   It turned out that this change had negative side effects on
2861   performance which were not acceptable. After some discussion it
2862   was decided to revert this feature and leave it up to the OS
2863   resp. the platform maintainer to ensure a proper initialization
2864   during early boot time.
2865
2866   *Matthias St. Pierre*
2867
2868### Changes between 1.1.1b and 1.1.1c [28 May 2019]
2869
2870 * Add build tests for C++.  These are generated files that only do one
2871   thing, to include one public OpenSSL head file each.  This tests that
2872   the public header files can be usefully included in a C++ application.
2873
2874   This test isn't enabled by default.  It can be enabled with the option
2875   'enable-buildtest-c++'.
2876
2877   *Richard Levitte*
2878
2879 * Enable SHA3 pre-hashing for ECDSA and DSA.
2880
2881   *Patrick Steuer*
2882
2883 * Change the default RSA, DSA and DH size to 2048 bit instead of 1024.
2884   This changes the size when using the `genpkey` command when no size is given.
2885   It fixes an omission in earlier changes that changed all RSA, DSA and DH
2886   generation commands to use 2048 bits by default.
2887
2888   *Kurt Roeckx*
2889
2890 * Reorganize the manual pages to consistently have RETURN VALUES,
2891   EXAMPLES, SEE ALSO and HISTORY come in that order, and adjust
2892   util/fix-doc-nits accordingly.
2893
2894   *Paul Yang, Joshua Lock*
2895
2896 * Add the missing accessor EVP_PKEY_get0_engine()
2897
2898   *Matt Caswell*
2899
2900 * Have commands like `s_client` and `s_server` output the signature scheme
2901   along with other cipher suite parameters when debugging.
2902
2903   *Lorinczy Zsigmond*
2904
2905 * Make OPENSSL_config() error agnostic again.
2906
2907   *Richard Levitte*
2908
2909 * Do the error handling in RSA decryption constant time.
2910
2911   *Bernd Edlinger*
2912
2913 * Prevent over long nonces in ChaCha20-Poly1305.
2914
2915   ChaCha20-Poly1305 is an AEAD cipher, and requires a unique nonce input
2916   for every encryption operation. RFC 7539 specifies that the nonce value
2917   (IV) should be 96 bits (12 bytes). OpenSSL allows a variable nonce length
2918   and front pads the nonce with 0 bytes if it is less than 12
2919   bytes. However it also incorrectly allows a nonce to be set of up to 16
2920   bytes. In this case only the last 12 bytes are significant and any
2921   additional leading bytes are ignored.
2922
2923   It is a requirement of using this cipher that nonce values are
2924   unique. Messages encrypted using a reused nonce value are susceptible to
2925   serious confidentiality and integrity attacks. If an application changes
2926   the default nonce length to be longer than 12 bytes and then makes a
2927   change to the leading bytes of the nonce expecting the new value to be a
2928   new unique nonce then such an application could inadvertently encrypt
2929   messages with a reused nonce.
2930
2931   Additionally the ignored bytes in a long nonce are not covered by the
2932   integrity guarantee of this cipher. Any application that relies on the
2933   integrity of these ignored leading bytes of a long nonce may be further
2934   affected. Any OpenSSL internal use of this cipher, including in SSL/TLS,
2935   is safe because no such use sets such a long nonce value. However user
2936   applications that use this cipher directly and set a non-default nonce
2937   length to be longer than 12 bytes may be vulnerable.
2938
2939   This issue was reported to OpenSSL on 16th of March 2019 by Joran Dirk
2940   Greef of Ronomon.
2941   ([CVE-2019-1543])
2942
2943   *Matt Caswell*
2944
2945 * Add DEVRANDOM_WAIT feature for Linux systems
2946
2947   On older Linux systems where the getrandom() system call is not available,
2948   OpenSSL normally uses the /dev/urandom device for seeding its CSPRNG.
2949   Contrary to getrandom(), the /dev/urandom device will not block during
2950   early boot when the kernel CSPRNG has not been seeded yet.
2951
2952   To mitigate this known weakness, use select() to wait for /dev/random to
2953   become readable before reading from /dev/urandom.
2954
2955 * Ensure that SM2 only uses SM3 as digest algorithm
2956
2957   *Paul Yang*
2958
2959### Changes between 1.1.1a and 1.1.1b [26 Feb 2019]
2960
2961 * Change the info callback signals for the start and end of a post-handshake
2962   message exchange in TLSv1.3. In 1.1.1/1.1.1a we used SSL_CB_HANDSHAKE_START
2963   and SSL_CB_HANDSHAKE_DONE. Experience has shown that many applications get
2964   confused by this and assume that a TLSv1.2 renegotiation has started. This
2965   can break KeyUpdate handling. Instead we no longer signal the start and end
2966   of a post handshake message exchange (although the messages themselves are
2967   still signalled). This could break some applications that were expecting
2968   the old signals. However without this KeyUpdate is not usable for many
2969   applications.
2970
2971   *Matt Caswell*
2972
2973### Changes between 1.1.1 and 1.1.1a [20 Nov 2018]
2974
2975 * Timing vulnerability in DSA signature generation
2976
2977   The OpenSSL DSA signature algorithm has been shown to be vulnerable to a
2978   timing side channel attack. An attacker could use variations in the signing
2979   algorithm to recover the private key.
2980
2981   This issue was reported to OpenSSL on 16th October 2018 by Samuel Weiser.
2982   ([CVE-2018-0734])
2983
2984   *Paul Dale*
2985
2986 * Timing vulnerability in ECDSA signature generation
2987
2988   The OpenSSL ECDSA signature algorithm has been shown to be vulnerable to a
2989   timing side channel attack. An attacker could use variations in the signing
2990   algorithm to recover the private key.
2991
2992   This issue was reported to OpenSSL on 25th October 2018 by Samuel Weiser.
2993   ([CVE-2018-0735])
2994
2995   *Paul Dale*
2996
2997 * Fixed the issue that RAND_add()/RAND_seed() silently discards random input
2998   if its length exceeds 4096 bytes. The limit has been raised to a buffer size
2999   of two gigabytes and the error handling improved.
3000
3001   This issue was reported to OpenSSL by Dr. Falko Strenzke. It has been
3002   categorized as a normal bug, not a security issue, because the DRBG reseeds
3003   automatically and is fully functional even without additional randomness
3004   provided by the application.
3005
3006### Changes between 1.1.0i and 1.1.1 [11 Sep 2018]
3007
3008 * Add a new ClientHello callback. Provides a callback interface that gives
3009   the application the ability to adjust the nascent SSL object at the
3010   earliest stage of ClientHello processing, immediately after extensions have
3011   been collected but before they have been processed. In particular, this
3012   callback can adjust the supported TLS versions in response to the contents
3013   of the ClientHello
3014
3015   *Benjamin Kaduk*
3016
3017 * Add SM2 base algorithm support.
3018
3019   *Jack Lloyd*
3020
3021 * s390x assembly pack: add (improved) hardware-support for the following
3022   cryptographic primitives: sha3, shake, aes-gcm, aes-ccm, aes-ctr, aes-ofb,
3023   aes-cfb/cfb8, aes-ecb.
3024
3025   *Patrick Steuer*
3026
3027 * Make EVP_PKEY_asn1_new() a bit stricter about its input.  A NULL pem_str
3028   parameter is no longer accepted, as it leads to a corrupt table.  NULL
3029   pem_str is reserved for alias entries only.
3030
3031   *Richard Levitte*
3032
3033 * Use the new ec_scalar_mul_ladder scaffold to implement a specialized ladder
3034   step for prime curves. The new implementation is based on formulae from
3035   differential addition-and-doubling in homogeneous projective coordinates
3036   from Izu-Takagi "A fast parallel elliptic curve multiplication resistant
3037   against side channel attacks" and Brier-Joye "Weierstrass Elliptic Curves
3038   and Side-Channel Attacks" Eq. (8) for y-coordinate recovery, modified
3039   to work in projective coordinates.
3040
3041   *Billy Bob Brumley, Nicola Tuveri*
3042
3043 * Change generating and checking of primes so that the error rate of not
3044   being prime depends on the intended use based on the size of the input.
3045   For larger primes this will result in more rounds of Miller-Rabin.
3046   The maximal error rate for primes with more than 1080 bits is lowered
3047   to 2^-128.
3048
3049   *Kurt Roeckx, Annie Yousar*
3050
3051 * Increase the number of Miller-Rabin rounds for DSA key generating to 64.
3052
3053   *Kurt Roeckx*
3054
3055 * The 'tsget' script is renamed to 'tsget.pl', to avoid confusion when
3056   moving between systems, and to avoid confusion when a Windows build is
3057   done with mingw vs with MSVC.  For POSIX installs, there's still a
3058   symlink or copy named 'tsget' to avoid that confusion as well.
3059
3060   *Richard Levitte*
3061
3062 * Revert blinding in ECDSA sign and instead make problematic addition
3063   length-invariant. Switch even to fixed-length Montgomery multiplication.
3064
3065   *Andy Polyakov*
3066
3067 * Use the new ec_scalar_mul_ladder scaffold to implement a specialized ladder
3068   step for binary curves. The new implementation is based on formulae from
3069   differential addition-and-doubling in mixed Lopez-Dahab projective
3070   coordinates, modified to independently blind the operands.
3071
3072   *Billy Bob Brumley, Sohaib ul Hassan, Nicola Tuveri*
3073
3074 * Add a scaffold to optionally enhance the Montgomery ladder implementation
3075   for `ec_scalar_mul_ladder` (formerly `ec_mul_consttime`) allowing
3076   EC_METHODs to implement their own specialized "ladder step", to take
3077   advantage of more favorable coordinate systems or more efficient
3078   differential addition-and-doubling algorithms.
3079
3080   *Billy Bob Brumley, Sohaib ul Hassan, Nicola Tuveri*
3081
3082 * Modified the random device based seed sources to keep the relevant
3083   file descriptors open rather than reopening them on each access.
3084   This allows such sources to operate in a chroot() jail without
3085   the associated device nodes being available. This behaviour can be
3086   controlled using RAND_keep_random_devices_open().
3087
3088   *Paul Dale*
3089
3090 * Numerous side-channel attack mitigations have been applied. This may have
3091   performance impacts for some algorithms for the benefit of improved
3092   security. Specific changes are noted in this change log by their respective
3093   authors.
3094
3095   *Matt Caswell*
3096
3097 * AIX shared library support overhaul. Switch to AIX "natural" way of
3098   handling shared libraries, which means collecting shared objects of
3099   different versions and bitnesses in one common archive. This allows to
3100   mitigate conflict between 1.0 and 1.1 side-by-side installations. It
3101   doesn't affect the way 3rd party applications are linked, only how
3102   multi-version installation is managed.
3103
3104   *Andy Polyakov*
3105
3106 * Make ec_group_do_inverse_ord() more robust and available to other
3107   EC cryptosystems, so that irrespective of BN_FLG_CONSTTIME, SCA
3108   mitigations are applied to the fallback BN_mod_inverse().
3109   When using this function rather than BN_mod_inverse() directly, new
3110   EC cryptosystem implementations are then safer-by-default.
3111
3112   *Billy Bob Brumley*
3113
3114 * Add coordinate blinding for EC_POINT and implement projective
3115   coordinate blinding for generic prime curves as a countermeasure to
3116   chosen point SCA attacks.
3117
3118   *Sohaib ul Hassan, Nicola Tuveri, Billy Bob Brumley*
3119
3120 * Add blinding to ECDSA and DSA signatures to protect against side channel
3121   attacks discovered by Keegan Ryan (NCC Group).
3122
3123   *Matt Caswell*
3124
3125 * Enforce checking in the `pkeyutl` command to ensure that the input
3126   length does not exceed the maximum supported digest length when performing
3127   a sign, verify or verifyrecover operation.
3128
3129   *Matt Caswell*
3130
3131 * SSL_MODE_AUTO_RETRY is enabled by default. Applications that use blocking
3132   I/O in combination with something like select() or poll() will hang. This
3133   can be turned off again using SSL_CTX_clear_mode().
3134   Many applications do not properly handle non-application data records, and
3135   TLS 1.3 sends more of such records. Setting SSL_MODE_AUTO_RETRY works
3136   around the problems in those applications, but can also break some.
3137   It's recommended to read the manpages about SSL_read(), SSL_write(),
3138   SSL_get_error(), SSL_shutdown(), SSL_CTX_set_mode() and
3139   SSL_CTX_set_read_ahead() again.
3140
3141   *Kurt Roeckx*
3142
3143 * When unlocking a pass phrase protected PEM file or PKCS#8 container, we
3144   now allow empty (zero character) pass phrases.
3145
3146   *Richard Levitte*
3147
3148 * Apply blinding to binary field modular inversion and remove patent
3149   pending (OPENSSL_SUN_GF2M_DIV) BN_GF2m_mod_div implementation.
3150
3151   *Billy Bob Brumley*
3152
3153 * Deprecate ec2_mult.c and unify scalar multiplication code paths for
3154   binary and prime elliptic curves.
3155
3156   *Billy Bob Brumley*
3157
3158 * Remove ECDSA nonce padding: EC_POINT_mul is now responsible for
3159   constant time fixed point multiplication.
3160
3161   *Billy Bob Brumley*
3162
3163 * Revise elliptic curve scalar multiplication with timing attack
3164   defenses: ec_wNAF_mul redirects to a constant time implementation
3165   when computing fixed point and variable point multiplication (which
3166   in OpenSSL are mostly used with secret scalars in keygen, sign,
3167   ECDH derive operations).
3168   *Billy Bob Brumley, Nicola Tuveri, Cesar Pereida García,
3169    Sohaib ul Hassan*
3170
3171 * Updated CONTRIBUTING
3172
3173   *Rich Salz*
3174
3175 * Updated DRBG / RAND to request nonce and additional low entropy
3176   randomness from the system.
3177
3178   *Matthias St. Pierre*
3179
3180 * Updated 'openssl rehash' to use OpenSSL consistent default.
3181
3182   *Richard Levitte*
3183
3184 * Moved the load of the ssl_conf module to libcrypto, which helps
3185   loading engines that libssl uses before libssl is initialised.
3186
3187   *Matt Caswell*
3188
3189 * Added EVP_PKEY_sign() and EVP_PKEY_verify() for EdDSA
3190
3191   *Matt Caswell*
3192
3193 * Fixed X509_NAME_ENTRY_set to get multi-valued RDNs right in all cases.
3194
3195   *Ingo Schwarze, Rich Salz*
3196
3197 * Added output of accepting IP address and port for 'openssl s_server'
3198
3199   *Richard Levitte*
3200
3201 * Added a new API for TLSv1.3 ciphersuites:
3202      SSL_CTX_set_ciphersuites()
3203      SSL_set_ciphersuites()
3204
3205   *Matt Caswell*
3206
3207 * Memory allocation failures consistently add an error to the error
3208   stack.
3209
3210   *Rich Salz*
3211
3212 * Don't use OPENSSL_ENGINES and OPENSSL_CONF environment values
3213   in libcrypto when run as setuid/setgid.
3214
3215   *Bernd Edlinger*
3216
3217 * Load any config file by default when libssl is used.
3218
3219   *Matt Caswell*
3220
3221 * Added new public header file <openssl/rand_drbg.h> and documentation
3222   for the RAND_DRBG API. See manual page RAND_DRBG(7) for an overview.
3223
3224   *Matthias St. Pierre*
3225
3226 * QNX support removed (cannot find contributors to get their approval
3227   for the license change).
3228
3229   *Rich Salz*
3230
3231 * TLSv1.3 replay protection for early data has been implemented. See the
3232   SSL_read_early_data() man page for further details.
3233
3234   *Matt Caswell*
3235
3236 * Separated TLSv1.3 ciphersuite configuration out from TLSv1.2 ciphersuite
3237   configuration. TLSv1.3 ciphersuites are not compatible with TLSv1.2 and
3238   below. Similarly TLSv1.2 ciphersuites are not compatible with TLSv1.3.
3239   In order to avoid issues where legacy TLSv1.2 ciphersuite configuration
3240   would otherwise inadvertently disable all TLSv1.3 ciphersuites the
3241   configuration has been separated out. See the ciphers man page or the
3242   SSL_CTX_set_ciphersuites() man page for more information.
3243
3244   *Matt Caswell*
3245
3246 * On POSIX (BSD, Linux, ...) systems the ocsp(1) command running
3247   in responder mode now supports the new "-multi" option, which
3248   spawns the specified number of child processes to handle OCSP
3249   requests.  The "-timeout" option now also limits the OCSP
3250   responder's patience to wait to receive the full client request
3251   on a newly accepted connection. Child processes are respawned
3252   as needed, and the CA index file is automatically reloaded
3253   when changed.  This makes it possible to run the "ocsp" responder
3254   as a long-running service, making the OpenSSL CA somewhat more
3255   feature-complete.  In this mode, most diagnostic messages logged
3256   after entering the event loop are logged via syslog(3) rather than
3257   written to stderr.
3258
3259   *Viktor Dukhovni*
3260
3261 * Added support for X448 and Ed448. Heavily based on original work by
3262   Mike Hamburg.
3263
3264   *Matt Caswell*
3265
3266 * Extend OSSL_STORE with capabilities to search and to narrow the set of
3267   objects loaded.  This adds the functions OSSL_STORE_expect() and
3268   OSSL_STORE_find() as well as needed tools to construct searches and
3269   get the search data out of them.
3270
3271   *Richard Levitte*
3272
3273 * Support for TLSv1.3 added. Note that users upgrading from an earlier
3274   version of OpenSSL should review their configuration settings to ensure
3275   that they are still appropriate for TLSv1.3. For further information see:
3276   <https://wiki.openssl.org/index.php/TLS1.3>
3277
3278   *Matt Caswell*
3279
3280 * Grand redesign of the OpenSSL random generator
3281
3282   The default RAND method now utilizes an AES-CTR DRBG according to
3283   NIST standard SP 800-90Ar1. The new random generator is essentially
3284   a port of the default random generator from the OpenSSL FIPS 2.0
3285   object module. It is a hybrid deterministic random bit generator
3286   using an AES-CTR bit stream and which seeds and reseeds itself
3287   automatically using trusted system entropy sources.
3288
3289   Some of its new features are:
3290    - Support for multiple DRBG instances with seed chaining.
3291    - The default RAND method makes use of a DRBG.
3292    - There is a public and private DRBG instance.
3293    - The DRBG instances are fork-safe.
3294    - Keep all global DRBG instances on the secure heap if it is enabled.
3295    - The public and private DRBG instance are per thread for lock free
3296      operation
3297
3298   *Paul Dale, Benjamin Kaduk, Kurt Roeckx, Rich Salz, Matthias St. Pierre*
3299
3300 * Changed Configure so it only says what it does and doesn't dump
3301   so much data.  Instead, ./configdata.pm should be used as a script
3302   to display all sorts of configuration data.
3303
3304   *Richard Levitte*
3305
3306 * Added processing of "make variables" to Configure.
3307
3308   *Richard Levitte*
3309
3310 * Added SHA512/224 and SHA512/256 algorithm support.
3311
3312   *Paul Dale*
3313
3314 * The last traces of Netware support, first removed in 1.1.0, have
3315   now been removed.
3316
3317   *Rich Salz*
3318
3319 * Get rid of Makefile.shared, and in the process, make the processing
3320   of certain files (rc.obj, or the .def/.map/.opt files produced from
3321   the ordinal files) more visible and hopefully easier to trace and
3322   debug (or make silent).
3323
3324   *Richard Levitte*
3325
3326 * Make it possible to have environment variable assignments as
3327   arguments to config / Configure.
3328
3329   *Richard Levitte*
3330
3331 * Add multi-prime RSA (RFC 8017) support.
3332
3333   *Paul Yang*
3334
3335 * Add SM3 implemented according to GB/T 32905-2016
3336   *Jack Lloyd <jack.lloyd@ribose.com>,*
3337   *Ronald Tse <ronald.tse@ribose.com>,*
3338   *Erick Borsboom <erick.borsboom@ribose.com>*
3339
3340 * Add 'Maximum Fragment Length' TLS extension negotiation and support
3341   as documented in RFC6066.
3342   Based on a patch from Tomasz Moń
3343
3344   *Filipe Raimundo da Silva*
3345
3346 * Add SM4 implemented according to GB/T 32907-2016.
3347   *Jack Lloyd <jack.lloyd@ribose.com>,*
3348   *Ronald Tse <ronald.tse@ribose.com>,*
3349   *Erick Borsboom <erick.borsboom@ribose.com>*
3350
3351 * Reimplement -newreq-nodes and ERR_error_string_n; the
3352   original author does not agree with the license change.
3353
3354   *Rich Salz*
3355
3356 * Add ARIA AEAD TLS support.
3357
3358   *Jon Spillett*
3359
3360 * Some macro definitions to support VS6 have been removed.  Visual
3361   Studio 6 has not worked since 1.1.0
3362
3363   *Rich Salz*
3364
3365 * Add ERR_clear_last_mark(), to allow callers to clear the last mark
3366   without clearing the errors.
3367
3368   *Richard Levitte*
3369
3370 * Add "atfork" functions.  If building on a system that without
3371   pthreads, see doc/man3/OPENSSL_fork_prepare.pod for application
3372   requirements.  The RAND facility now uses/requires this.
3373
3374   *Rich Salz*
3375
3376 * Add SHA3.
3377
3378   *Andy Polyakov*
3379
3380 * The UI API becomes a permanent and integral part of libcrypto, i.e.
3381   not possible to disable entirely.  However, it's still possible to
3382   disable the console reading UI method, UI_OpenSSL() (use UI_null()
3383   as a fallback).
3384
3385   To disable, configure with 'no-ui-console'.  'no-ui' is still
3386   possible to use as an alias.  Check at compile time with the
3387   macro OPENSSL_NO_UI_CONSOLE.  The macro OPENSSL_NO_UI is still
3388   possible to check and is an alias for OPENSSL_NO_UI_CONSOLE.
3389
3390   *Richard Levitte*
3391
3392 * Add a STORE module, which implements a uniform and URI based reader of
3393   stores that can contain keys, certificates, CRLs and numerous other
3394   objects.  The main API is loosely based on a few stdio functions,
3395   and includes OSSL_STORE_open, OSSL_STORE_load, OSSL_STORE_eof,
3396   OSSL_STORE_error and OSSL_STORE_close.
3397   The implementation uses backends called "loaders" to implement arbitrary
3398   URI schemes.  There is one built in "loader" for the 'file' scheme.
3399
3400   *Richard Levitte*
3401
3402 * Add devcrypto engine.  This has been implemented against cryptodev-linux,
3403   then adjusted to work on FreeBSD 8.4 as well.
3404   Enable by configuring with 'enable-devcryptoeng'.  This is done by default
3405   on BSD implementations, as cryptodev.h is assumed to exist on all of them.
3406
3407   *Richard Levitte*
3408
3409 * Module names can prefixed with OSSL_ or OPENSSL_.  This affects
3410   util/mkerr.pl, which is adapted to allow those prefixes, leading to
3411   error code calls like this:
3412
3413           OSSL_FOOerr(OSSL_FOO_F_SOMETHING, OSSL_FOO_R_WHATEVER);
3414
3415   With this change, we claim the namespaces OSSL and OPENSSL in a manner
3416   that can be encoded in C.  For the foreseeable future, this will only
3417   affect new modules.
3418
3419   *Richard Levitte and Tim Hudson*
3420
3421 * Removed BSD cryptodev engine.
3422
3423   *Rich Salz*
3424
3425 * Add a build target 'build_all_generated', to build all generated files
3426   and only that.  This can be used to prepare everything that requires
3427   things like perl for a system that lacks perl and then move everything
3428   to that system and do the rest of the build there.
3429
3430   *Richard Levitte*
3431
3432 * In the UI interface, make it possible to duplicate the user data.  This
3433   can be used by engines that need to retain the data for a longer time
3434   than just the call where this user data is passed.
3435
3436   *Richard Levitte*
3437
3438 * Ignore the '-named_curve auto' value for compatibility of applications
3439   with OpenSSL 1.0.2.
3440
3441   *Tomáš Mráz <tmraz@fedoraproject.org>*
3442
3443 * Fragmented SSL/TLS alerts are no longer accepted. An alert message is 2
3444   bytes long. In theory it is permissible in SSLv3 - TLSv1.2 to fragment such
3445   alerts across multiple records (some of which could be empty). In practice
3446   it make no sense to send an empty alert record, or to fragment one. TLSv1.3
3447   prohibits this altogether and other libraries (BoringSSL, NSS) do not
3448   support this at all. Supporting it adds significant complexity to the
3449   record layer, and its removal is unlikely to cause interoperability
3450   issues.
3451
3452   *Matt Caswell*
3453
3454 * Add the ASN.1 types INT32, UINT32, INT64, UINT64 and variants prefixed
3455   with Z.  These are meant to replace LONG and ZLONG and to be size safe.
3456   The use of LONG and ZLONG is discouraged and scheduled for deprecation
3457   in OpenSSL 1.2.0.
3458
3459   *Richard Levitte*
3460
3461 * Add the 'z' and 'j' modifiers to BIO_printf() et al formatting string,
3462   'z' is to be used for [s]size_t, and 'j' - with [u]int64_t.
3463
3464   *Richard Levitte, Andy Polyakov*
3465
3466 * Add EC_KEY_get0_engine(), which does for EC_KEY what RSA_get0_engine()
3467   does for RSA, etc.
3468
3469   *Richard Levitte*
3470
3471 * Have 'config' recognise 64-bit mingw and choose 'mingw64' as the target
3472   platform rather than 'mingw'.
3473
3474   *Richard Levitte*
3475
3476 * The functions X509_STORE_add_cert and X509_STORE_add_crl return
3477   success if they are asked to add an object which already exists
3478   in the store. This change cascades to other functions which load
3479   certificates and CRLs.
3480
3481   *Paul Dale*
3482
3483 * x86_64 assembly pack: annotate code with DWARF CFI directives to
3484   facilitate stack unwinding even from assembly subroutines.
3485
3486   *Andy Polyakov*
3487
3488 * Remove VAX C specific definitions of OPENSSL_EXPORT, OPENSSL_EXTERN.
3489   Also remove OPENSSL_GLOBAL entirely, as it became a no-op.
3490
3491   *Richard Levitte*
3492
3493 * Remove the VMS-specific reimplementation of gmtime from crypto/o_times.c.
3494   VMS C's RTL has a fully up to date gmtime() and gmtime_r() since V7.1,
3495   which is the minimum version we support.
3496
3497   *Richard Levitte*
3498
3499 * Certificate time validation (X509_cmp_time) enforces stricter
3500   compliance with RFC 5280. Fractional seconds and timezone offsets
3501   are no longer allowed.
3502
3503   *Emilia Käsper*
3504
3505 * Add support for ARIA
3506
3507   *Paul Dale*
3508
3509 * s_client will now send the Server Name Indication (SNI) extension by
3510   default unless the new "-noservername" option is used. The server name is
3511   based on the host provided to the "-connect" option unless overridden by
3512   using "-servername".
3513
3514   *Matt Caswell*
3515
3516 * Add support for SipHash
3517
3518   *Todd Short*
3519
3520 * OpenSSL now fails if it receives an unrecognised record type in TLS1.0
3521   or TLS1.1. Previously this only happened in SSLv3 and TLS1.2. This is to
3522   prevent issues where no progress is being made and the peer continually
3523   sends unrecognised record types, using up resources processing them.
3524
3525   *Matt Caswell*
3526
3527 * 'openssl passwd' can now produce SHA256 and SHA512 based output,
3528   using the algorithm defined in
3529   <https://www.akkadia.org/drepper/SHA-crypt.txt>
3530
3531   *Richard Levitte*
3532
3533 * Heartbeat support has been removed; the ABI is changed for now.
3534
3535   *Richard Levitte, Rich Salz*
3536
3537 * Support for SSL_OP_NO_ENCRYPT_THEN_MAC in SSL_CONF_cmd.
3538
3539   *Emilia Käsper*
3540
3541 * The RSA "null" method, which was partially supported to avoid patent
3542   issues, has been replaced to always returns NULL.
3543
3544   *Rich Salz*
3545
3546OpenSSL 1.1.0
3547-------------
3548
3549### Changes between 1.1.0k and 1.1.0l [10 Sep 2019]
3550
3551 * For built-in EC curves, ensure an EC_GROUP built from the curve name is
3552   used even when parsing explicit parameters, when loading a encoded key
3553   or calling `EC_GROUP_new_from_ecpkparameters()`/
3554   `EC_GROUP_new_from_ecparameters()`.
3555   This prevents bypass of security hardening and performance gains,
3556   especially for curves with specialized EC_METHODs.
3557   By default, if a key encoded with explicit parameters is loaded and later
3558   encoded, the output is still encoded with explicit parameters, even if
3559   internally a "named" EC_GROUP is used for computation.
3560
3561   *Nicola Tuveri*
3562
3563 * Compute ECC cofactors if not provided during EC_GROUP construction. Before
3564   this change, EC_GROUP_set_generator would accept order and/or cofactor as
3565   NULL. After this change, only the cofactor parameter can be NULL. It also
3566   does some minimal sanity checks on the passed order.
3567   ([CVE-2019-1547])
3568
3569   *Billy Bob Brumley*
3570
3571 * Fixed a padding oracle in PKCS7_dataDecode and CMS_decrypt_set1_pkey.
3572   An attack is simple, if the first CMS_recipientInfo is valid but the
3573   second CMS_recipientInfo is chosen ciphertext. If the second
3574   recipientInfo decodes to PKCS #1 v1.5 form plaintext, the correct
3575   encryption key will be replaced by garbage, and the message cannot be
3576   decoded, but if the RSA decryption fails, the correct encryption key is
3577   used and the recipient will not notice the attack.
3578   As a work around for this potential attack the length of the decrypted
3579   key must be equal to the cipher default key length, in case the
3580   certifiate is not given and all recipientInfo are tried out.
3581   The old behaviour can be re-enabled in the CMS code by setting the
3582   CMS_DEBUG_DECRYPT flag.
3583   ([CVE-2019-1563])
3584
3585   *Bernd Edlinger*
3586
3587 * Use Windows installation paths in the mingw builds
3588
3589   Mingw isn't a POSIX environment per se, which means that Windows
3590   paths should be used for installation.
3591   ([CVE-2019-1552])
3592
3593   *Richard Levitte*
3594
3595### Changes between 1.1.0j and 1.1.0k [28 May 2019]
3596
3597 * Change the default RSA, DSA and DH size to 2048 bit instead of 1024.
3598   This changes the size when using the `genpkey` command when no size is given.
3599   It fixes an omission in earlier changes that changed all RSA, DSA and DH
3600   generation commands to use 2048 bits by default.
3601
3602   *Kurt Roeckx*
3603
3604 * Prevent over long nonces in ChaCha20-Poly1305.
3605
3606   ChaCha20-Poly1305 is an AEAD cipher, and requires a unique nonce input
3607   for every encryption operation. RFC 7539 specifies that the nonce value
3608   (IV) should be 96 bits (12 bytes). OpenSSL allows a variable nonce length
3609   and front pads the nonce with 0 bytes if it is less than 12
3610   bytes. However it also incorrectly allows a nonce to be set of up to 16
3611   bytes. In this case only the last 12 bytes are significant and any
3612   additional leading bytes are ignored.
3613
3614   It is a requirement of using this cipher that nonce values are
3615   unique. Messages encrypted using a reused nonce value are susceptible to
3616   serious confidentiality and integrity attacks. If an application changes
3617   the default nonce length to be longer than 12 bytes and then makes a
3618   change to the leading bytes of the nonce expecting the new value to be a
3619   new unique nonce then such an application could inadvertently encrypt
3620   messages with a reused nonce.
3621
3622   Additionally the ignored bytes in a long nonce are not covered by the
3623   integrity guarantee of this cipher. Any application that relies on the
3624   integrity of these ignored leading bytes of a long nonce may be further
3625   affected. Any OpenSSL internal use of this cipher, including in SSL/TLS,
3626   is safe because no such use sets such a long nonce value. However user
3627   applications that use this cipher directly and set a non-default nonce
3628   length to be longer than 12 bytes may be vulnerable.
3629
3630   This issue was reported to OpenSSL on 16th of March 2019 by Joran Dirk
3631   Greef of Ronomon.
3632   ([CVE-2019-1543])
3633
3634   *Matt Caswell*
3635
3636 * Added SCA hardening for modular field inversion in EC_GROUP through
3637   a new dedicated field_inv() pointer in EC_METHOD.
3638   This also addresses a leakage affecting conversions from projective
3639   to affine coordinates.
3640
3641   *Billy Bob Brumley, Nicola Tuveri*
3642
3643 * Fix a use after free bug in d2i_X509_PUBKEY when overwriting a
3644   re-used X509_PUBKEY object if the second PUBKEY is malformed.
3645
3646   *Bernd Edlinger*
3647
3648 * Move strictness check from EVP_PKEY_asn1_new() to EVP_PKEY_asn1_add0().
3649
3650   *Richard Levitte*
3651
3652 * Remove the 'dist' target and add a tarball building script.  The
3653   'dist' target has fallen out of use, and it shouldn't be
3654   necessary to configure just to create a source distribution.
3655
3656   *Richard Levitte*
3657
3658### Changes between 1.1.0i and 1.1.0j [20 Nov 2018]
3659
3660 * Timing vulnerability in DSA signature generation
3661
3662   The OpenSSL DSA signature algorithm has been shown to be vulnerable to a
3663   timing side channel attack. An attacker could use variations in the signing
3664   algorithm to recover the private key.
3665
3666   This issue was reported to OpenSSL on 16th October 2018 by Samuel Weiser.
3667   ([CVE-2018-0734])
3668
3669   *Paul Dale*
3670
3671 * Timing vulnerability in ECDSA signature generation
3672
3673   The OpenSSL ECDSA signature algorithm has been shown to be vulnerable to a
3674   timing side channel attack. An attacker could use variations in the signing
3675   algorithm to recover the private key.
3676
3677   This issue was reported to OpenSSL on 25th October 2018 by Samuel Weiser.
3678   ([CVE-2018-0735])
3679
3680   *Paul Dale*
3681
3682 * Add coordinate blinding for EC_POINT and implement projective
3683   coordinate blinding for generic prime curves as a countermeasure to
3684   chosen point SCA attacks.
3685
3686   *Sohaib ul Hassan, Nicola Tuveri, Billy Bob Brumley*
3687
3688### Changes between 1.1.0h and 1.1.0i [14 Aug 2018]
3689
3690 * Client DoS due to large DH parameter
3691
3692   During key agreement in a TLS handshake using a DH(E) based ciphersuite a
3693   malicious server can send a very large prime value to the client. This will
3694   cause the client to spend an unreasonably long period of time generating a
3695   key for this prime resulting in a hang until the client has finished. This
3696   could be exploited in a Denial Of Service attack.
3697
3698   This issue was reported to OpenSSL on 5th June 2018 by Guido Vranken
3699   ([CVE-2018-0732])
3700
3701   *Guido Vranken*
3702
3703 * Cache timing vulnerability in RSA Key Generation
3704
3705   The OpenSSL RSA Key generation algorithm has been shown to be vulnerable to
3706   a cache timing side channel attack. An attacker with sufficient access to
3707   mount cache timing attacks during the RSA key generation process could
3708   recover the private key.
3709
3710   This issue was reported to OpenSSL on 4th April 2018 by Alejandro Cabrera
3711   Aldaya, Billy Brumley, Cesar Pereida Garcia and Luis Manuel Alvarez Tapia.
3712   ([CVE-2018-0737])
3713
3714   *Billy Brumley*
3715
3716 * Make EVP_PKEY_asn1_new() a bit stricter about its input.  A NULL pem_str
3717   parameter is no longer accepted, as it leads to a corrupt table.  NULL
3718   pem_str is reserved for alias entries only.
3719
3720   *Richard Levitte*
3721
3722 * Revert blinding in ECDSA sign and instead make problematic addition
3723   length-invariant. Switch even to fixed-length Montgomery multiplication.
3724
3725   *Andy Polyakov*
3726
3727 * Change generating and checking of primes so that the error rate of not
3728   being prime depends on the intended use based on the size of the input.
3729   For larger primes this will result in more rounds of Miller-Rabin.
3730   The maximal error rate for primes with more than 1080 bits is lowered
3731   to 2^-128.
3732
3733   *Kurt Roeckx, Annie Yousar*
3734
3735 * Increase the number of Miller-Rabin rounds for DSA key generating to 64.
3736
3737   *Kurt Roeckx*
3738
3739 * Add blinding to ECDSA and DSA signatures to protect against side channel
3740   attacks discovered by Keegan Ryan (NCC Group).
3741
3742   *Matt Caswell*
3743
3744 * When unlocking a pass phrase protected PEM file or PKCS#8 container, we
3745   now allow empty (zero character) pass phrases.
3746
3747   *Richard Levitte*
3748
3749 * Certificate time validation (X509_cmp_time) enforces stricter
3750   compliance with RFC 5280. Fractional seconds and timezone offsets
3751   are no longer allowed.
3752
3753   *Emilia Käsper*
3754
3755 * Fixed a text canonicalisation bug in CMS
3756
3757   Where a CMS detached signature is used with text content the text goes
3758   through a canonicalisation process first prior to signing or verifying a
3759   signature. This process strips trailing space at the end of lines, converts
3760   line terminators to CRLF and removes additional trailing line terminators
3761   at the end of a file. A bug in the canonicalisation process meant that
3762   some characters, such as form-feed, were incorrectly treated as whitespace
3763   and removed. This is contrary to the specification (RFC5485). This fix
3764   could mean that detached text data signed with an earlier version of
3765   OpenSSL 1.1.0 may fail to verify using the fixed version, or text data
3766   signed with a fixed OpenSSL may fail to verify with an earlier version of
3767   OpenSSL 1.1.0. A workaround is to only verify the canonicalised text data
3768   and use the "-binary" flag (for the "cms" command line application) or set
3769   the SMIME_BINARY/PKCS7_BINARY/CMS_BINARY flags (if using CMS_verify()).
3770
3771   *Matt Caswell*
3772
3773### Changes between 1.1.0g and 1.1.0h [27 Mar 2018]
3774
3775 * Constructed ASN.1 types with a recursive definition could exceed the stack
3776
3777   Constructed ASN.1 types with a recursive definition (such as can be found
3778   in PKCS7) could eventually exceed the stack given malicious input with
3779   excessive recursion. This could result in a Denial Of Service attack. There
3780   are no such structures used within SSL/TLS that come from untrusted sources
3781   so this is considered safe.
3782
3783   This issue was reported to OpenSSL on 4th January 2018 by the OSS-fuzz
3784   project.
3785   ([CVE-2018-0739])
3786
3787   *Matt Caswell*
3788
3789 * Incorrect CRYPTO_memcmp on HP-UX PA-RISC
3790
3791   Because of an implementation bug the PA-RISC CRYPTO_memcmp function is
3792   effectively reduced to only comparing the least significant bit of each
3793   byte. This allows an attacker to forge messages that would be considered as
3794   authenticated in an amount of tries lower than that guaranteed by the
3795   security claims of the scheme. The module can only be compiled by the
3796   HP-UX assembler, so that only HP-UX PA-RISC targets are affected.
3797
3798   This issue was reported to OpenSSL on 2nd March 2018 by Peter Waltenberg
3799   (IBM).
3800   ([CVE-2018-0733])
3801
3802   *Andy Polyakov*
3803
3804 * Add a build target 'build_all_generated', to build all generated files
3805   and only that.  This can be used to prepare everything that requires
3806   things like perl for a system that lacks perl and then move everything
3807   to that system and do the rest of the build there.
3808
3809   *Richard Levitte*
3810
3811 * Backport SSL_OP_NO_RENGOTIATION
3812
3813   OpenSSL 1.0.2 and below had the ability to disable renegotiation using the
3814   (undocumented) SSL3_FLAGS_NO_RENEGOTIATE_CIPHERS flag. Due to the opacity
3815   changes this is no longer possible in 1.1.0. Therefore the new
3816   SSL_OP_NO_RENEGOTIATION option from 1.1.1-dev has been backported to
3817   1.1.0 to provide equivalent functionality.
3818
3819   Note that if an application built against 1.1.0h headers (or above) is run
3820   using an older version of 1.1.0 (prior to 1.1.0h) then the option will be
3821   accepted but nothing will happen, i.e. renegotiation will not be prevented.
3822
3823   *Matt Caswell*
3824
3825 * Removed the OS390-Unix config target.  It relied on a script that doesn't
3826   exist.
3827
3828   *Rich Salz*
3829
3830 * rsaz_1024_mul_avx2 overflow bug on x86_64
3831
3832   There is an overflow bug in the AVX2 Montgomery multiplication procedure
3833   used in exponentiation with 1024-bit moduli. No EC algorithms are affected.
3834   Analysis suggests that attacks against RSA and DSA as a result of this
3835   defect would be very difficult to perform and are not believed likely.
3836   Attacks against DH1024 are considered just feasible, because most of the
3837   work necessary to deduce information about a private key may be performed
3838   offline. The amount of resources required for such an attack would be
3839   significant. However, for an attack on TLS to be meaningful, the server
3840   would have to share the DH1024 private key among multiple clients, which is
3841   no longer an option since CVE-2016-0701.
3842
3843   This only affects processors that support the AVX2 but not ADX extensions
3844   like Intel Haswell (4th generation).
3845
3846   This issue was reported to OpenSSL by David Benjamin (Google). The issue
3847   was originally found via the OSS-Fuzz project.
3848   ([CVE-2017-3738])
3849
3850   *Andy Polyakov*
3851
3852### Changes between 1.1.0f and 1.1.0g [2 Nov 2017]
3853
3854 * bn_sqrx8x_internal carry bug on x86_64
3855
3856   There is a carry propagating bug in the x86_64 Montgomery squaring
3857   procedure. No EC algorithms are affected. Analysis suggests that attacks
3858   against RSA and DSA as a result of this defect would be very difficult to
3859   perform and are not believed likely. Attacks against DH are considered just
3860   feasible (although very difficult) because most of the work necessary to
3861   deduce information about a private key may be performed offline. The amount
3862   of resources required for such an attack would be very significant and
3863   likely only accessible to a limited number of attackers. An attacker would
3864   additionally need online access to an unpatched system using the target
3865   private key in a scenario with persistent DH parameters and a private
3866   key that is shared between multiple clients.
3867
3868   This only affects processors that support the BMI1, BMI2 and ADX extensions
3869   like Intel Broadwell (5th generation) and later or AMD Ryzen.
3870
3871   This issue was reported to OpenSSL by the OSS-Fuzz project.
3872   ([CVE-2017-3736])
3873
3874   *Andy Polyakov*
3875
3876 * Malformed X.509 IPAddressFamily could cause OOB read
3877
3878   If an X.509 certificate has a malformed IPAddressFamily extension,
3879   OpenSSL could do a one-byte buffer overread. The most likely result
3880   would be an erroneous display of the certificate in text format.
3881
3882   This issue was reported to OpenSSL by the OSS-Fuzz project.
3883   ([CVE-2017-3735])
3884
3885   *Rich Salz*
3886
3887### Changes between 1.1.0e and 1.1.0f [25 May 2017]
3888
3889 * Have 'config' recognise 64-bit mingw and choose 'mingw64' as the target
3890   platform rather than 'mingw'.
3891
3892   *Richard Levitte*
3893
3894 * Remove the VMS-specific reimplementation of gmtime from crypto/o_times.c.
3895   VMS C's RTL has a fully up to date gmtime() and gmtime_r() since V7.1,
3896   which is the minimum version we support.
3897
3898   *Richard Levitte*
3899
3900### Changes between 1.1.0d and 1.1.0e [16 Feb 2017]
3901
3902 * Encrypt-Then-Mac renegotiation crash
3903
3904   During a renegotiation handshake if the Encrypt-Then-Mac extension is
3905   negotiated where it was not in the original handshake (or vice-versa) then
3906   this can cause OpenSSL to crash (dependant on ciphersuite). Both clients
3907   and servers are affected.
3908
3909   This issue was reported to OpenSSL by Joe Orton (Red Hat).
3910   ([CVE-2017-3733])
3911
3912   *Matt Caswell*
3913
3914### Changes between 1.1.0c and 1.1.0d [26 Jan 2017]
3915
3916 * Truncated packet could crash via OOB read
3917
3918   If one side of an SSL/TLS path is running on a 32-bit host and a specific
3919   cipher is being used, then a truncated packet can cause that host to
3920   perform an out-of-bounds read, usually resulting in a crash.
3921
3922   This issue was reported to OpenSSL by Robert Święcki of Google.
3923   ([CVE-2017-3731])
3924
3925   *Andy Polyakov*
3926
3927 * Bad (EC)DHE parameters cause a client crash
3928
3929   If a malicious server supplies bad parameters for a DHE or ECDHE key
3930   exchange then this can result in the client attempting to dereference a
3931   NULL pointer leading to a client crash. This could be exploited in a Denial
3932   of Service attack.
3933
3934   This issue was reported to OpenSSL by Guido Vranken.
3935   ([CVE-2017-3730])
3936
3937   *Matt Caswell*
3938
3939 * BN_mod_exp may produce incorrect results on x86_64
3940
3941   There is a carry propagating bug in the x86_64 Montgomery squaring
3942   procedure. No EC algorithms are affected. Analysis suggests that attacks
3943   against RSA and DSA as a result of this defect would be very difficult to
3944   perform and are not believed likely. Attacks against DH are considered just
3945   feasible (although very difficult) because most of the work necessary to
3946   deduce information about a private key may be performed offline. The amount
3947   of resources required for such an attack would be very significant and
3948   likely only accessible to a limited number of attackers. An attacker would
3949   additionally need online access to an unpatched system using the target
3950   private key in a scenario with persistent DH parameters and a private
3951   key that is shared between multiple clients. For example this can occur by
3952   default in OpenSSL DHE based SSL/TLS ciphersuites. Note: This issue is very
3953   similar to CVE-2015-3193 but must be treated as a separate problem.
3954
3955   This issue was reported to OpenSSL by the OSS-Fuzz project.
3956   ([CVE-2017-3732])
3957
3958   *Andy Polyakov*
3959
3960### Changes between 1.1.0b and 1.1.0c [10 Nov 2016]
3961
3962 * ChaCha20/Poly1305 heap-buffer-overflow
3963
3964   TLS connections using `*-CHACHA20-POLY1305` ciphersuites are susceptible to
3965   a DoS attack by corrupting larger payloads. This can result in an OpenSSL
3966   crash. This issue is not considered to be exploitable beyond a DoS.
3967
3968   This issue was reported to OpenSSL by Robert Święcki (Google Security Team)
3969   ([CVE-2016-7054])
3970
3971   *Richard Levitte*
3972
3973 * CMS Null dereference
3974
3975   Applications parsing invalid CMS structures can crash with a NULL pointer
3976   dereference. This is caused by a bug in the handling of the ASN.1 CHOICE
3977   type in OpenSSL 1.1.0 which can result in a NULL value being passed to the
3978   structure callback if an attempt is made to free certain invalid encodings.
3979   Only CHOICE structures using a callback which do not handle NULL value are
3980   affected.
3981
3982   This issue was reported to OpenSSL by Tyler Nighswander of ForAllSecure.
3983   ([CVE-2016-7053])
3984
3985   *Stephen Henson*
3986
3987 * Montgomery multiplication may produce incorrect results
3988
3989   There is a carry propagating bug in the Broadwell-specific Montgomery
3990   multiplication procedure that handles input lengths divisible by, but
3991   longer than 256 bits. Analysis suggests that attacks against RSA, DSA
3992   and DH private keys are impossible. This is because the subroutine in
3993   question is not used in operations with the private key itself and an input
3994   of the attacker's direct choice. Otherwise the bug can manifest itself as
3995   transient authentication and key negotiation failures or reproducible
3996   erroneous outcome of public-key operations with specially crafted input.
3997   Among EC algorithms only Brainpool P-512 curves are affected and one
3998   presumably can attack ECDH key negotiation. Impact was not analyzed in
3999   detail, because pre-requisites for attack are considered unlikely. Namely
4000   multiple clients have to choose the curve in question and the server has to
4001   share the private key among them, neither of which is default behaviour.
4002   Even then only clients that chose the curve will be affected.
4003
4004   This issue was publicly reported as transient failures and was not
4005   initially recognized as a security issue. Thanks to Richard Morgan for
4006   providing reproducible case.
4007   ([CVE-2016-7055])
4008
4009   *Andy Polyakov*
4010
4011 * Removed automatic addition of RPATH in shared libraries and executables,
4012   as this was a remainder from OpenSSL 1.0.x and isn't needed any more.
4013
4014   *Richard Levitte*
4015
4016### Changes between 1.1.0a and 1.1.0b [26 Sep 2016]
4017
4018 * Fix Use After Free for large message sizes
4019
4020   The patch applied to address CVE-2016-6307 resulted in an issue where if a
4021   message larger than approx 16k is received then the underlying buffer to
4022   store the incoming message is reallocated and moved. Unfortunately a
4023   dangling pointer to the old location is left which results in an attempt to
4024   write to the previously freed location. This is likely to result in a
4025   crash, however it could potentially lead to execution of arbitrary code.
4026
4027   This issue only affects OpenSSL 1.1.0a.
4028
4029   This issue was reported to OpenSSL by Robert Święcki.
4030   ([CVE-2016-6309])
4031
4032   *Matt Caswell*
4033
4034### Changes between 1.1.0 and 1.1.0a [22 Sep 2016]
4035
4036 * OCSP Status Request extension unbounded memory growth
4037
4038   A malicious client can send an excessively large OCSP Status Request
4039   extension. If that client continually requests renegotiation, sending a
4040   large OCSP Status Request extension each time, then there will be unbounded
4041   memory growth on the server. This will eventually lead to a Denial Of
4042   Service attack through memory exhaustion. Servers with a default
4043   configuration are vulnerable even if they do not support OCSP. Builds using
4044   the "no-ocsp" build time option are not affected.
4045
4046   This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.)
4047   ([CVE-2016-6304])
4048
4049   *Matt Caswell*
4050
4051 * SSL_peek() hang on empty record
4052
4053   OpenSSL 1.1.0 SSL/TLS will hang during a call to SSL_peek() if the peer
4054   sends an empty record. This could be exploited by a malicious peer in a
4055   Denial Of Service attack.
4056
4057   This issue was reported to OpenSSL by Alex Gaynor.
4058   ([CVE-2016-6305])
4059
4060   *Matt Caswell*
4061
4062 * Excessive allocation of memory in tls_get_message_header() and
4063   dtls1_preprocess_fragment()
4064
4065   A (D)TLS message includes 3 bytes for its length in the header for the
4066   message. This would allow for messages up to 16Mb in length. Messages of
4067   this length are excessive and OpenSSL includes a check to ensure that a
4068   peer is sending reasonably sized messages in order to avoid too much memory
4069   being consumed to service a connection. A flaw in the logic of version
4070   1.1.0 means that memory for the message is allocated too early, prior to
4071   the excessive message length check. Due to way memory is allocated in
4072   OpenSSL this could mean an attacker could force up to 21Mb to be allocated
4073   to service a connection. This could lead to a Denial of Service through
4074   memory exhaustion. However, the excessive message length check still takes
4075   place, and this would cause the connection to immediately fail. Assuming
4076   that the application calls SSL_free() on the failed connection in a timely
4077   manner then the 21Mb of allocated memory will then be immediately freed
4078   again. Therefore the excessive memory allocation will be transitory in
4079   nature. This then means that there is only a security impact if:
4080
4081   1) The application does not call SSL_free() in a timely manner in the event
4082   that the connection fails
4083   or
4084   2) The application is working in a constrained environment where there is
4085   very little free memory
4086   or
4087   3) The attacker initiates multiple connection attempts such that there are
4088   multiple connections in a state where memory has been allocated for the
4089   connection; SSL_free() has not yet been called; and there is insufficient
4090   memory to service the multiple requests.
4091
4092   Except in the instance of (1) above any Denial Of Service is likely to be
4093   transitory because as soon as the connection fails the memory is
4094   subsequently freed again in the SSL_free() call. However there is an
4095   increased risk during this period of application crashes due to the lack of
4096   memory - which would then mean a more serious Denial of Service.
4097
4098   This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.)
4099   (CVE-2016-6307 and CVE-2016-6308)
4100
4101   *Matt Caswell*
4102
4103 * solaris-x86-cc, i.e. 32-bit configuration with vendor compiler,
4104   had to be removed. Primary reason is that vendor assembler can't
4105   assemble our modules with -KPIC flag. As result it, assembly
4106   support, was not even available as option. But its lack means
4107   lack of side-channel resistant code, which is incompatible with
4108   security by todays standards. Fortunately gcc is readily available
4109   prepackaged option, which we firmly point at...
4110
4111   *Andy Polyakov*
4112
4113### Changes between 1.0.2h and 1.1.0  [25 Aug 2016]
4114
4115 * Windows command-line tool supports UTF-8 opt-in option for arguments
4116   and console input. Setting OPENSSL_WIN32_UTF8 environment variable
4117   (to any value) allows Windows user to access PKCS#12 file generated
4118   with Windows CryptoAPI and protected with non-ASCII password, as well
4119   as files generated under UTF-8 locale on Linux also protected with
4120   non-ASCII password.
4121
4122   *Andy Polyakov*
4123
4124 * To mitigate the SWEET32 attack ([CVE-2016-2183]), 3DES cipher suites
4125   have been disabled by default and removed from DEFAULT, just like RC4.
4126   See the RC4 item below to re-enable both.
4127
4128   *Rich Salz*
4129
4130 * The method for finding the storage location for the Windows RAND seed file
4131   has changed. First we check %RANDFILE%. If that is not set then we check
4132   the directories %HOME%, %USERPROFILE% and %SYSTEMROOT% in that order. If
4133   all else fails we fall back to C:\.
4134
4135   *Matt Caswell*
4136
4137 * The EVP_EncryptUpdate() function has had its return type changed from void
4138   to int. A return of 0 indicates and error while a return of 1 indicates
4139   success.
4140
4141   *Matt Caswell*
4142
4143 * The flags RSA_FLAG_NO_CONSTTIME, DSA_FLAG_NO_EXP_CONSTTIME and
4144   DH_FLAG_NO_EXP_CONSTTIME which previously provided the ability to switch
4145   off the constant time implementation for RSA, DSA and DH have been made
4146   no-ops and deprecated.
4147
4148   *Matt Caswell*
4149
4150 * Windows RAND implementation was simplified to only get entropy by
4151   calling CryptGenRandom(). Various other RAND-related tickets
4152   were also closed.
4153
4154   *Joseph Wylie Yandle, Rich Salz*
4155
4156 * The stack and lhash API's were renamed to start with `OPENSSL_SK_`
4157   and `OPENSSL_LH_`, respectively.  The old names are available
4158   with API compatibility.  They new names are now completely documented.
4159
4160   *Rich Salz*
4161
4162 * Unify TYPE_up_ref(obj) methods signature.
4163   SSL_CTX_up_ref(), SSL_up_ref(), X509_up_ref(), EVP_PKEY_up_ref(),
4164   X509_CRL_up_ref(), X509_OBJECT_up_ref_count() methods are now returning an
4165   int (instead of void) like all others TYPE_up_ref() methods.
4166   So now these methods also check the return value of CRYPTO_atomic_add(),
4167   and the validity of object reference counter.
4168
4169   *fdasilvayy@gmail.com*
4170
4171 * With Windows Visual Studio builds, the .pdb files are installed
4172   alongside the installed libraries and executables.  For a static
4173   library installation, ossl_static.pdb is the associate compiler
4174   generated .pdb file to be used when linking programs.
4175
4176   *Richard Levitte*
4177
4178 * Remove openssl.spec.  Packaging files belong with the packagers.
4179
4180   *Richard Levitte*
4181
4182 * Automatic Darwin/OSX configuration has had a refresh, it will now
4183   recognise x86_64 architectures automatically.  You can still decide
4184   to build for a different bitness with the environment variable
4185   KERNEL_BITS (can be 32 or 64), for example:
4186
4187           KERNEL_BITS=32 ./config
4188
4189   *Richard Levitte*
4190
4191 * Change default algorithms in pkcs8 utility to use PKCS#5 v2.0,
4192   256 bit AES and HMAC with SHA256.
4193
4194   *Steve Henson*
4195
4196 * Remove support for MIPS o32 ABI on IRIX (and IRIX only).
4197
4198   *Andy Polyakov*
4199
4200 * Triple-DES ciphers have been moved from HIGH to MEDIUM.
4201
4202   *Rich Salz*
4203
4204 * To enable users to have their own config files and build file templates,
4205   Configure looks in the directory indicated by the environment variable
4206   OPENSSL_LOCAL_CONFIG_DIR as well as the in-source Configurations/
4207   directory.  On VMS, OPENSSL_LOCAL_CONFIG_DIR is expected to be a logical
4208   name and is used as is.
4209
4210   *Richard Levitte*
4211
4212 * The following datatypes were made opaque: X509_OBJECT, X509_STORE_CTX,
4213   X509_STORE, X509_LOOKUP, and X509_LOOKUP_METHOD.  The unused type
4214   X509_CERT_FILE_CTX was removed.
4215
4216   *Rich Salz*
4217
4218 * "shared" builds are now the default. To create only static libraries use
4219   the "no-shared" Configure option.
4220
4221   *Matt Caswell*
4222
4223 * Remove the no-aes, no-hmac, no-rsa, no-sha and no-md5 Configure options.
4224   All of these option have not worked for some while and are fundamental
4225   algorithms.
4226
4227   *Matt Caswell*
4228
4229 * Make various cleanup routines no-ops and mark them as deprecated. Most
4230   global cleanup functions are no longer required because they are handled
4231   via auto-deinit (see OPENSSL_init_crypto and OPENSSL_init_ssl man pages).
4232   Explicitly de-initing can cause problems (e.g. where a library that uses
4233   OpenSSL de-inits, but an application is still using it). The affected
4234   functions are CONF_modules_free(), ENGINE_cleanup(), OBJ_cleanup(),
4235   EVP_cleanup(), BIO_sock_cleanup(), CRYPTO_cleanup_all_ex_data(),
4236   RAND_cleanup(), SSL_COMP_free_compression_methods(), ERR_free_strings() and
4237   COMP_zlib_cleanup().
4238
4239   *Matt Caswell*
4240
4241 * --strict-warnings no longer enables runtime debugging options
4242   such as REF_DEBUG. Instead, debug options are automatically
4243   enabled with '--debug' builds.
4244
4245   *Andy Polyakov, Emilia Käsper*
4246
4247 * Made DH and DH_METHOD opaque. The structures for managing DH objects
4248   have been moved out of the public header files. New functions for managing
4249   these have been added.
4250
4251   *Matt Caswell*
4252
4253 * Made RSA and RSA_METHOD opaque. The structures for managing RSA
4254   objects have been moved out of the public header files. New
4255   functions for managing these have been added.
4256
4257   *Richard Levitte*
4258
4259 * Made DSA and DSA_METHOD opaque. The structures for managing DSA objects
4260   have been moved out of the public header files. New functions for managing
4261   these have been added.
4262
4263   *Matt Caswell*
4264
4265 * Made BIO and BIO_METHOD opaque. The structures for managing BIOs have been
4266   moved out of the public header files. New functions for managing these
4267   have been added.
4268
4269   *Matt Caswell*
4270
4271 * Removed no-rijndael as a config option. Rijndael is an old name for AES.
4272
4273   *Matt Caswell*
4274
4275 * Removed the mk1mf build scripts.
4276
4277   *Richard Levitte*
4278
4279 * Headers are now wrapped, if necessary, with OPENSSL_NO_xxx, so
4280   it is always safe to #include a header now.
4281
4282   *Rich Salz*
4283
4284 * Removed the aged BC-32 config and all its supporting scripts
4285
4286   *Richard Levitte*
4287
4288 * Removed support for Ultrix, Netware, and OS/2.
4289
4290   *Rich Salz*
4291
4292 * Add support for HKDF.
4293
4294   *Alessandro Ghedini*
4295
4296 * Add support for blake2b and blake2s
4297
4298   *Bill Cox*
4299
4300 * Added support for "pipelining". Ciphers that have the
4301   EVP_CIPH_FLAG_PIPELINE flag set have a capability to process multiple
4302   encryptions/decryptions simultaneously. There are currently no built-in
4303   ciphers with this property but the expectation is that engines will be able
4304   to offer it to significantly improve throughput. Support has been extended
4305   into libssl so that multiple records for a single connection can be
4306   processed in one go (for >=TLS 1.1).
4307
4308   *Matt Caswell*
4309
4310 * Added the AFALG engine. This is an async capable engine which is able to
4311   offload work to the Linux kernel. In this initial version it only supports
4312   AES128-CBC. The kernel must be version 4.1.0 or greater.
4313
4314   *Catriona Lucey*
4315
4316 * OpenSSL now uses a new threading API. It is no longer necessary to
4317   set locking callbacks to use OpenSSL in a multi-threaded environment. There
4318   are two supported threading models: pthreads and windows threads. It is
4319   also possible to configure OpenSSL at compile time for "no-threads". The
4320   old threading API should no longer be used. The functions have been
4321   replaced with "no-op" compatibility macros.
4322
4323   *Alessandro Ghedini, Matt Caswell*
4324
4325 * Modify behavior of ALPN to invoke callback after SNI/servername
4326   callback, such that updates to the SSL_CTX affect ALPN.
4327
4328   *Todd Short*
4329
4330 * Add SSL_CIPHER queries for authentication and key-exchange.
4331
4332   *Todd Short*
4333
4334 * Changes to the DEFAULT cipherlist:
4335   - Prefer (EC)DHE handshakes over plain RSA.
4336   - Prefer AEAD ciphers over legacy ciphers.
4337   - Prefer ECDSA over RSA when both certificates are available.
4338   - Prefer TLSv1.2 ciphers/PRF.
4339   - Remove DSS, SEED, IDEA, CAMELLIA, and AES-CCM from the
4340     default cipherlist.
4341
4342   *Emilia Käsper*
4343
4344 * Change the ECC default curve list to be this, in order: x25519,
4345   secp256r1, secp521r1, secp384r1.
4346
4347   *Rich Salz*
4348
4349 * RC4 based libssl ciphersuites are now classed as "weak" ciphers and are
4350   disabled by default. They can be re-enabled using the
4351   enable-weak-ssl-ciphers option to Configure.
4352
4353   *Matt Caswell*
4354
4355 * If the server has ALPN configured, but supports no protocols that the
4356   client advertises, send a fatal "no_application_protocol" alert.
4357   This behaviour is SHALL in RFC 7301, though it isn't universally
4358   implemented by other servers.
4359
4360   *Emilia Käsper*
4361
4362 * Add X25519 support.
4363   Add ASN.1 and EVP_PKEY methods for X25519. This includes support
4364   for public and private key encoding using the format documented in
4365   draft-ietf-curdle-pkix-02. The corresponding EVP_PKEY method supports
4366   key generation and key derivation.
4367
4368   TLS support complies with draft-ietf-tls-rfc4492bis-08 and uses
4369   X25519(29).
4370
4371   *Steve Henson*
4372
4373 * Deprecate SRP_VBASE_get_by_user.
4374   SRP_VBASE_get_by_user had inconsistent memory management behaviour.
4375   In order to fix an unavoidable memory leak ([CVE-2016-0798]),
4376   SRP_VBASE_get_by_user was changed to ignore the "fake user" SRP
4377   seed, even if the seed is configured.
4378
4379   Users should use SRP_VBASE_get1_by_user instead. Note that in
4380   SRP_VBASE_get1_by_user, caller must free the returned value. Note
4381   also that even though configuring the SRP seed attempts to hide
4382   invalid usernames by continuing the handshake with fake
4383   credentials, this behaviour is not constant time and no strong
4384   guarantees are made that the handshake is indistinguishable from
4385   that of a valid user.
4386
4387   *Emilia Käsper*
4388
4389 * Configuration change; it's now possible to build dynamic engines
4390   without having to build shared libraries and vice versa.  This
4391   only applies to the engines in `engines/`, those in `crypto/engine/`
4392   will always be built into libcrypto (i.e. "static").
4393
4394   Building dynamic engines is enabled by default; to disable, use
4395   the configuration option "disable-dynamic-engine".
4396
4397   The only requirements for building dynamic engines are the
4398   presence of the DSO module and building with position independent
4399   code, so they will also automatically be disabled if configuring
4400   with "disable-dso" or "disable-pic".
4401
4402   The macros OPENSSL_NO_STATIC_ENGINE and OPENSSL_NO_DYNAMIC_ENGINE
4403   are also taken away from openssl/opensslconf.h, as they are
4404   irrelevant.
4405
4406   *Richard Levitte*
4407
4408 * Configuration change; if there is a known flag to compile
4409   position independent code, it will always be applied on the
4410   libcrypto and libssl object files, and never on the application
4411   object files.  This means other libraries that use routines from
4412   libcrypto / libssl can be made into shared libraries regardless
4413   of how OpenSSL was configured.
4414
4415   If this isn't desirable, the configuration options "disable-pic"
4416   or "no-pic" can be used to disable the use of PIC.  This will
4417   also disable building shared libraries and dynamic engines.
4418
4419   *Richard Levitte*
4420
4421 * Removed JPAKE code.  It was experimental and has no wide use.
4422
4423   *Rich Salz*
4424
4425 * The INSTALL_PREFIX Makefile variable has been renamed to
4426   DESTDIR.  That makes for less confusion on what this variable
4427   is for.  Also, the configuration option --install_prefix is
4428   removed.
4429
4430   *Richard Levitte*
4431
4432 * Heartbeat for TLS has been removed and is disabled by default
4433   for DTLS; configure with enable-heartbeats.  Code that uses the
4434   old #define's might need to be updated.
4435
4436   *Emilia Käsper, Rich Salz*
4437
4438 * Rename REF_CHECK to REF_DEBUG.
4439
4440   *Rich Salz*
4441
4442 * New "unified" build system
4443
4444   The "unified" build system is aimed to be a common system for all
4445   platforms we support.  With it comes new support for VMS.
4446
4447   This system builds supports building in a different directory tree
4448   than the source tree.  It produces one Makefile (for unix family
4449   or lookalikes), or one descrip.mms (for VMS).
4450
4451   The source of information to make the Makefile / descrip.mms is
4452   small files called 'build.info', holding the necessary
4453   information for each directory with source to compile, and a
4454   template in Configurations, like unix-Makefile.tmpl or
4455   descrip.mms.tmpl.
4456
4457   With this change, the library names were also renamed on Windows
4458   and on VMS.  They now have names that are closer to the standard
4459   on Unix, and include the major version number, and in certain
4460   cases, the architecture they are built for.  See "Notes on shared
4461   libraries" in INSTALL.
4462
4463   We rely heavily on the perl module Text::Template.
4464
4465   *Richard Levitte*
4466
4467 * Added support for auto-initialisation and de-initialisation of the library.
4468   OpenSSL no longer requires explicit init or deinit routines to be called,
4469   except in certain circumstances. See the OPENSSL_init_crypto() and
4470   OPENSSL_init_ssl() man pages for further information.
4471
4472   *Matt Caswell*
4473
4474 * The arguments to the DTLSv1_listen function have changed. Specifically the
4475   "peer" argument is now expected to be a BIO_ADDR object.
4476
4477 * Rewrite of BIO networking library. The BIO library lacked consistent
4478   support of IPv6, and adding it required some more extensive
4479   modifications.  This introduces the BIO_ADDR and BIO_ADDRINFO types,
4480   which hold all types of addresses and chains of address information.
4481   It also introduces a new API, with functions like BIO_socket,
4482   BIO_connect, BIO_listen, BIO_lookup and a rewrite of BIO_accept.
4483   The source/sink BIOs BIO_s_connect, BIO_s_accept and BIO_s_datagram
4484   have been adapted accordingly.
4485
4486   *Richard Levitte*
4487
4488 * RSA_padding_check_PKCS1_type_1 now accepts inputs with and without
4489   the leading 0-byte.
4490
4491   *Emilia Käsper*
4492
4493 * CRIME protection: disable compression by default, even if OpenSSL is
4494   compiled with zlib enabled. Applications can still enable compression
4495   by calling SSL_CTX_clear_options(ctx, SSL_OP_NO_COMPRESSION), or by
4496   using the SSL_CONF library to configure compression.
4497
4498   *Emilia Käsper*
4499
4500 * The signature of the session callback configured with
4501   SSL_CTX_sess_set_get_cb was changed. The read-only input buffer
4502   was explicitly marked as `const unsigned char*` instead of
4503   `unsigned char*`.
4504
4505   *Emilia Käsper*
4506
4507 * Always DPURIFY. Remove the use of uninitialized memory in the
4508   RNG, and other conditional uses of DPURIFY. This makes -DPURIFY a no-op.
4509
4510   *Emilia Käsper*
4511
4512 * Removed many obsolete configuration items, including
4513      DES_PTR, DES_RISC1, DES_RISC2, DES_INT
4514      MD2_CHAR, MD2_INT, MD2_LONG
4515      BF_PTR, BF_PTR2
4516      IDEA_SHORT, IDEA_LONG
4517      RC2_SHORT, RC2_LONG, RC4_LONG, RC4_CHUNK, RC4_INDEX
4518
4519   *Rich Salz, with advice from Andy Polyakov*
4520
4521 * Many BN internals have been moved to an internal header file.
4522
4523   *Rich Salz with help from Andy Polyakov*
4524
4525 * Configuration and writing out the results from it has changed.
4526   Files such as Makefile include/openssl/opensslconf.h and are now
4527   produced through general templates, such as Makefile.in and
4528   crypto/opensslconf.h.in and some help from the perl module
4529   Text::Template.
4530
4531   Also, the center of configuration information is no longer
4532   Makefile.  Instead, Configure produces a perl module in
4533   configdata.pm which holds most of the config data (in the hash
4534   table %config), the target data that comes from the target
4535   configuration in one of the `Configurations/*.conf` files (in
4536   %target).
4537
4538   *Richard Levitte*
4539
4540 * To clarify their intended purposes, the Configure options
4541   --prefix and --openssldir change their semantics, and become more
4542   straightforward and less interdependent.
4543
4544   --prefix shall be used exclusively to give the location INSTALLTOP
4545   where programs, scripts, libraries, include files and manuals are
4546   going to be installed.  The default is now /usr/local.
4547
4548   --openssldir shall be used exclusively to give the default
4549   location OPENSSLDIR where certificates, private keys, CRLs are
4550   managed.  This is also where the default openssl.cnf gets
4551   installed.
4552   If the directory given with this option is a relative path, the
4553   values of both the --prefix value and the --openssldir value will
4554   be combined to become OPENSSLDIR.
4555   The default for --openssldir is INSTALLTOP/ssl.
4556
4557   Anyone who uses --openssldir to specify where OpenSSL is to be
4558   installed MUST change to use --prefix instead.
4559
4560   *Richard Levitte*
4561
4562 * The GOST engine was out of date and therefore it has been removed. An up
4563   to date GOST engine is now being maintained in an external repository.
4564   See: <https://wiki.openssl.org/index.php/Binaries>. Libssl still retains
4565   support for GOST ciphersuites (these are only activated if a GOST engine
4566   is present).
4567
4568   *Matt Caswell*
4569
4570 * EGD is no longer supported by default; use enable-egd when
4571   configuring.
4572
4573   *Ben Kaduk and Rich Salz*
4574
4575 * The distribution now has Makefile.in files, which are used to
4576   create Makefile's when Configure is run.  *Configure must be run
4577   before trying to build now.*
4578
4579   *Rich Salz*
4580
4581 * The return value for SSL_CIPHER_description() for error conditions
4582   has changed.
4583
4584   *Rich Salz*
4585
4586 * Support for RFC6698/RFC7671 DANE TLSA peer authentication.
4587
4588   Obtaining and performing DNSSEC validation of TLSA records is
4589   the application's responsibility.  The application provides
4590   the TLSA records of its choice to OpenSSL, and these are then
4591   used to authenticate the peer.
4592
4593   The TLSA records need not even come from DNS.  They can, for
4594   example, be used to implement local end-entity certificate or
4595   trust-anchor "pinning", where the "pin" data takes the form
4596   of TLSA records, which can augment or replace verification
4597   based on the usual WebPKI public certification authorities.
4598
4599   *Viktor Dukhovni*
4600
4601 * Revert default OPENSSL_NO_DEPRECATED setting.  Instead OpenSSL
4602   continues to support deprecated interfaces in default builds.
4603   However, applications are strongly advised to compile their
4604   source files with -DOPENSSL_API_COMPAT=0x10100000L, which hides
4605   the declarations of all interfaces deprecated in 0.9.8, 1.0.0
4606   or the 1.1.0 releases.
4607
4608   In environments in which all applications have been ported to
4609   not use any deprecated interfaces OpenSSL's Configure script
4610   should be used with the --api=1.1.0 option to entirely remove
4611   support for the deprecated features from the library and
4612   unconditionally disable them in the installed headers.
4613   Essentially the same effect can be achieved with the "no-deprecated"
4614   argument to Configure, except that this will always restrict
4615   the build to just the latest API, rather than a fixed API
4616   version.
4617
4618   As applications are ported to future revisions of the API,
4619   they should update their compile-time OPENSSL_API_COMPAT define
4620   accordingly, but in most cases should be able to continue to
4621   compile with later releases.
4622
4623   The OPENSSL_API_COMPAT versions for 1.0.0, and 0.9.8 are
4624   0x10000000L and 0x00908000L, respectively.  However those
4625   versions did not support the OPENSSL_API_COMPAT feature, and
4626   so applications are not typically tested for explicit support
4627   of just the undeprecated features of either release.
4628
4629   *Viktor Dukhovni*
4630
4631 * Add support for setting the minimum and maximum supported protocol.
4632   It can bet set via the SSL_set_min_proto_version() and
4633   SSL_set_max_proto_version(), or via the SSL_CONF's MinProtocol and
4634   MaxProtocol.  It's recommended to use the new APIs to disable
4635   protocols instead of disabling individual protocols using
4636   SSL_set_options() or SSL_CONF's Protocol.  This change also
4637   removes support for disabling TLS 1.2 in the OpenSSL TLS
4638   client at compile time by defining OPENSSL_NO_TLS1_2_CLIENT.
4639
4640   *Kurt Roeckx*
4641
4642 * Support for ChaCha20 and Poly1305 added to libcrypto and libssl.
4643
4644   *Andy Polyakov*
4645
4646 * New EC_KEY_METHOD, this replaces the older ECDSA_METHOD and ECDH_METHOD
4647   and integrates ECDSA and ECDH functionality into EC. Implementations can
4648   now redirect key generation and no longer need to convert to or from
4649   ECDSA_SIG format.
4650
4651   Note: the ecdsa.h and ecdh.h headers are now no longer needed and just
4652   include the ec.h header file instead.
4653
4654   *Steve Henson*
4655
4656 * Remove support for all 40 and 56 bit ciphers.  This includes all the export
4657   ciphers who are no longer supported and drops support the ephemeral RSA key
4658   exchange. The LOW ciphers currently doesn't have any ciphers in it.
4659
4660   *Kurt Roeckx*
4661
4662 * Made EVP_MD_CTX, EVP_MD, EVP_CIPHER_CTX, EVP_CIPHER and HMAC_CTX
4663   opaque.  For HMAC_CTX, the following constructors and destructors
4664   were added:
4665
4666       HMAC_CTX *HMAC_CTX_new(void);
4667       void HMAC_CTX_free(HMAC_CTX *ctx);
4668
4669   For EVP_MD and EVP_CIPHER, complete APIs to create, fill and
4670   destroy such methods has been added.  See EVP_MD_meth_new(3) and
4671   EVP_CIPHER_meth_new(3) for documentation.
4672
4673   Additional changes:
4674   1) `EVP_MD_CTX_cleanup()`, `EVP_CIPHER_CTX_cleanup()` and
4675      `HMAC_CTX_cleanup()` were removed. `HMAC_CTX_reset()` and
4676      `EVP_MD_CTX_reset()` should be called instead to reinitialise
4677      an already created structure.
4678   2) For consistency with the majority of our object creators and
4679      destructors, `EVP_MD_CTX_(create|destroy)` were renamed to
4680      `EVP_MD_CTX_(new|free)`.  The old names are retained as macros
4681      for deprecated builds.
4682
4683   *Richard Levitte*
4684
4685 * Added ASYNC support. Libcrypto now includes the async sub-library to enable
4686   cryptographic operations to be performed asynchronously as long as an
4687   asynchronous capable engine is used. See the ASYNC_start_job() man page for
4688   further details. Libssl has also had this capability integrated with the
4689   introduction of the new mode SSL_MODE_ASYNC and associated error
4690   SSL_ERROR_WANT_ASYNC. See the SSL_CTX_set_mode() and SSL_get_error() man
4691   pages. This work was developed in partnership with Intel Corp.
4692
4693   *Matt Caswell*
4694
4695 * SSL_{CTX_}set_ecdh_auto() has been removed and ECDH is support is
4696   always enabled now.  If you want to disable the support you should
4697   exclude it using the list of supported ciphers. This also means that the
4698   "-no_ecdhe" option has been removed from s_server.
4699
4700   *Kurt Roeckx*
4701
4702 * SSL_{CTX}_set_tmp_ecdh() which can set 1 EC curve now internally calls
4703   SSL_{CTX_}set1_curves() which can set a list.
4704
4705   *Kurt Roeckx*
4706
4707 * Remove support for SSL_{CTX_}set_tmp_ecdh_callback().  You should set the
4708   curve you want to support using SSL_{CTX_}set1_curves().
4709
4710   *Kurt Roeckx*
4711
4712 * State machine rewrite. The state machine code has been significantly
4713   refactored in order to remove much duplication of code and solve issues
4714   with the old code (see [ssl/statem/README.md](ssl/statem/README.md) for
4715   further details). This change does have some associated API changes.
4716   Notably the SSL_state() function has been removed and replaced by
4717   SSL_get_state which now returns an "OSSL_HANDSHAKE_STATE" instead of an int.
4718   SSL_set_state() has been removed altogether. The previous handshake states
4719   defined in ssl.h and ssl3.h have also been removed.
4720
4721   *Matt Caswell*
4722
4723 * All instances of the string "ssleay" in the public API were replaced
4724   with OpenSSL (case-matching; e.g., OPENSSL_VERSION for #define's)
4725   Some error codes related to internal RSA_eay API's were renamed.
4726
4727   *Rich Salz*
4728
4729 * The demo files in crypto/threads were moved to demo/threads.
4730
4731   *Rich Salz*
4732
4733 * Removed obsolete engines: 4758cca, aep, atalla, cswift, nuron, gmp,
4734   sureware and ubsec.
4735
4736   *Matt Caswell, Rich Salz*
4737
4738 * New ASN.1 embed macro.
4739
4740   New ASN.1 macro ASN1_EMBED. This is the same as ASN1_SIMPLE except the
4741   structure is not allocated: it is part of the parent. That is instead of
4742
4743           FOO *x;
4744
4745   it must be:
4746
4747           FOO x;
4748
4749   This reduces memory fragmentation and make it impossible to accidentally
4750   set a mandatory field to NULL.
4751
4752   This currently only works for some fields specifically a SEQUENCE, CHOICE,
4753   or ASN1_STRING type which is part of a parent SEQUENCE. Since it is
4754   equivalent to ASN1_SIMPLE it cannot be tagged, OPTIONAL, SET OF or
4755   SEQUENCE OF.
4756
4757   *Steve Henson*
4758
4759 * Remove EVP_CHECK_DES_KEY, a compile-time option that never compiled.
4760
4761   *Emilia Käsper*
4762
4763 * Removed DES and RC4 ciphersuites from DEFAULT. Also removed RC2 although
4764   in 1.0.2 EXPORT was already removed and the only RC2 ciphersuite is also
4765   an EXPORT one. COMPLEMENTOFDEFAULT has been updated accordingly to add
4766   DES and RC4 ciphersuites.
4767
4768   *Matt Caswell*
4769
4770 * Rewrite EVP_DecodeUpdate (base64 decoding) to fix several bugs.
4771   This changes the decoding behaviour for some invalid messages,
4772   though the change is mostly in the more lenient direction, and
4773   legacy behaviour is preserved as much as possible.
4774
4775   *Emilia Käsper*
4776
4777 * Fix no-stdio build.
4778   *David Woodhouse <David.Woodhouse@intel.com> and also*
4779   *Ivan Nestlerode <ivan.nestlerode@sonos.com>*
4780
4781 * New testing framework
4782   The testing framework has been largely rewritten and is now using
4783   perl and the perl modules Test::Harness and an extended variant of
4784   Test::More called OpenSSL::Test to do its work.  All test scripts in
4785   test/ have been rewritten into test recipes, and all direct calls to
4786   executables in test/Makefile have become individual recipes using the
4787   simplified testing OpenSSL::Test::Simple.
4788
4789   For documentation on our testing modules, do:
4790
4791           perldoc test/testlib/OpenSSL/Test/Simple.pm
4792           perldoc test/testlib/OpenSSL/Test.pm
4793
4794   *Richard Levitte*
4795
4796 * Revamped memory debug; only -DCRYPTO_MDEBUG and -DCRYPTO_MDEBUG_ABORT
4797   are used; the latter aborts on memory leaks (usually checked on exit).
4798   Some undocumented "set malloc, etc., hooks" functions were removed
4799   and others were changed.  All are now documented.
4800
4801   *Rich Salz*
4802
4803 * In DSA_generate_parameters_ex, if the provided seed is too short,
4804   return an error
4805
4806   *Rich Salz and Ismo Puustinen <ismo.puustinen@intel.com>*
4807
4808 * Rewrite PSK to support ECDHE_PSK, DHE_PSK and RSA_PSK. Add ciphersuites
4809   from RFC4279, RFC4785, RFC5487, RFC5489.
4810
4811   Thanks to Christian J. Dietrich and Giuseppe D'Angelo for the
4812   original RSA_PSK patch.
4813
4814   *Steve Henson*
4815
4816 * Dropped support for the SSL3_FLAGS_DELAY_CLIENT_FINISHED flag. This SSLeay
4817   era flag was never set throughout the codebase (only read). Also removed
4818   SSL3_FLAGS_POP_BUFFER which was only used if
4819   SSL3_FLAGS_DELAY_CLIENT_FINISHED was also set.
4820
4821   *Matt Caswell*
4822
4823 * Changed the default name options in the "ca", "crl", "req" and "x509"
4824   to be "oneline" instead of "compat".
4825
4826   *Richard Levitte*
4827
4828 * Remove SSL_OP_TLS_BLOCK_PADDING_BUG. This is SSLeay legacy, we're
4829   not aware of clients that still exhibit this bug, and the workaround
4830   hasn't been working properly for a while.
4831
4832   *Emilia Käsper*
4833
4834 * The return type of BIO_number_read() and BIO_number_written() as well as
4835   the corresponding num_read and num_write members in the BIO structure has
4836   changed from unsigned long to uint64_t. On platforms where an unsigned
4837   long is 32 bits (e.g. Windows) these counters could overflow if >4Gb is
4838   transferred.
4839
4840   *Matt Caswell*
4841
4842 * Given the pervasive nature of TLS extensions it is inadvisable to run
4843   OpenSSL without support for them. It also means that maintaining
4844   the OPENSSL_NO_TLSEXT option within the code is very invasive (and probably
4845   not well tested). Therefore the OPENSSL_NO_TLSEXT option has been removed.
4846
4847   *Matt Caswell*
4848
4849 * Removed support for the two export grade static DH ciphersuites
4850   EXP-DH-RSA-DES-CBC-SHA and EXP-DH-DSS-DES-CBC-SHA. These two ciphersuites
4851   were newly added (along with a number of other static DH ciphersuites) to
4852   1.0.2. However the two export ones have *never* worked since they were
4853   introduced. It seems strange in any case to be adding new export
4854   ciphersuites, and given "logjam" it also does not seem correct to fix them.
4855
4856   *Matt Caswell*
4857
4858 * Version negotiation has been rewritten. In particular SSLv23_method(),
4859   SSLv23_client_method() and SSLv23_server_method() have been deprecated,
4860   and turned into macros which simply call the new preferred function names
4861   TLS_method(), TLS_client_method() and TLS_server_method(). All new code
4862   should use the new names instead. Also as part of this change the ssl23.h
4863   header file has been removed.
4864
4865   *Matt Caswell*
4866
4867 * Support for Kerberos ciphersuites in TLS (RFC2712) has been removed. This
4868   code and the associated standard is no longer considered fit-for-purpose.
4869
4870   *Matt Caswell*
4871
4872 * RT2547 was closed.  When generating a private key, try to make the
4873   output file readable only by the owner.  This behavior change might
4874   be noticeable when interacting with other software.
4875
4876 * Documented all exdata functions.  Added CRYPTO_free_ex_index.
4877   Added a test.
4878
4879   *Rich Salz*
4880
4881 * Added HTTP GET support to the ocsp command.
4882
4883   *Rich Salz*
4884
4885 * Changed default digest for the dgst and enc commands from MD5 to
4886   sha256
4887
4888   *Rich Salz*
4889
4890 * RAND_pseudo_bytes has been deprecated. Users should use RAND_bytes instead.
4891
4892   *Matt Caswell*
4893
4894 * Added support for TLS extended master secret from
4895   draft-ietf-tls-session-hash-03.txt. Thanks for Alfredo Pironti for an
4896   initial patch which was a great help during development.
4897
4898   *Steve Henson*
4899
4900 * All libssl internal structures have been removed from the public header
4901   files, and the OPENSSL_NO_SSL_INTERN option has been removed (since it is
4902   now redundant). Users should not attempt to access internal structures
4903   directly. Instead they should use the provided API functions.
4904
4905   *Matt Caswell*
4906
4907 * config has been changed so that by default OPENSSL_NO_DEPRECATED is used.
4908   Access to deprecated functions can be re-enabled by running config with
4909   "enable-deprecated". In addition applications wishing to use deprecated
4910   functions must define OPENSSL_USE_DEPRECATED. Note that this new behaviour
4911   will, by default, disable some transitive includes that previously existed
4912   in the header files (e.g. ec.h will no longer, by default, include bn.h)
4913
4914   *Matt Caswell*
4915
4916 * Added support for OCB mode. OpenSSL has been granted a patent license
4917   compatible with the OpenSSL license for use of OCB. Details are available
4918   at <https://www.openssl.org/source/OCB-patent-grant-OpenSSL.pdf>. Support
4919   for OCB can be removed by calling config with no-ocb.
4920
4921   *Matt Caswell*
4922
4923 * SSLv2 support has been removed.  It still supports receiving a SSLv2
4924   compatible client hello.
4925
4926   *Kurt Roeckx*
4927
4928 * Increased the minimal RSA keysize from 256 to 512 bits [Rich Salz],
4929   done while fixing the error code for the key-too-small case.
4930
4931   *Annie Yousar <a.yousar@informatik.hu-berlin.de>*
4932
4933 * CA.sh has been removed; use CA.pl instead.
4934
4935   *Rich Salz*
4936
4937 * Removed old DES API.
4938
4939   *Rich Salz*
4940
4941 * Remove various unsupported platforms:
4942      Sony NEWS4
4943      BEOS and BEOS_R5
4944      NeXT
4945      SUNOS
4946      MPE/iX
4947      Sinix/ReliantUNIX RM400
4948      DGUX
4949      NCR
4950      Tandem
4951      Cray
4952      16-bit platforms such as WIN16
4953
4954   *Rich Salz*
4955
4956 * Clean up OPENSSL_NO_xxx #define's
4957   - Use setbuf() and remove OPENSSL_NO_SETVBUF_IONBF
4958   - Rename OPENSSL_SYSNAME_xxx to OPENSSL_SYS_xxx
4959   - OPENSSL_NO_EC{DH,DSA} merged into OPENSSL_NO_EC
4960   - OPENSSL_NO_RIPEMD160, OPENSSL_NO_RIPEMD merged into OPENSSL_NO_RMD160
4961   - OPENSSL_NO_FP_API merged into OPENSSL_NO_STDIO
4962   - Remove OPENSSL_NO_BIO OPENSSL_NO_BUFFER OPENSSL_NO_CHAIN_VERIFY
4963     OPENSSL_NO_EVP OPENSSL_NO_FIPS_ERR OPENSSL_NO_HASH_COMP
4964     OPENSSL_NO_LHASH OPENSSL_NO_OBJECT OPENSSL_NO_SPEED OPENSSL_NO_STACK
4965     OPENSSL_NO_X509 OPENSSL_NO_X509_VERIFY
4966   - Remove MS_STATIC; it's a relic from platforms <32 bits.
4967
4968   *Rich Salz*
4969
4970 * Cleaned up dead code
4971     Remove all but one '#ifdef undef' which is to be looked at.
4972
4973   *Rich Salz*
4974
4975 * Clean up calling of xxx_free routines.
4976      Just like free(), fix most of the xxx_free routines to accept
4977      NULL.  Remove the non-null checks from callers.  Save much code.
4978
4979   *Rich Salz*
4980
4981 * Add secure heap for storage of private keys (when possible).
4982   Add BIO_s_secmem(), CBIGNUM, etc.
4983   Contributed by Akamai Technologies under our Corporate CLA.
4984
4985   *Rich Salz*
4986
4987 * Experimental support for a new, fast, unbiased prime candidate generator,
4988   bn_probable_prime_dh_coprime(). Not currently used by any prime generator.
4989
4990   *Felix Laurie von Massenbach <felix@erbridge.co.uk>*
4991
4992 * New output format NSS in the sess_id command line tool. This allows
4993   exporting the session id and the master key in NSS keylog format.
4994
4995   *Martin Kaiser <martin@kaiser.cx>*
4996
4997 * Harmonize version and its documentation. -f flag is used to display
4998   compilation flags.
4999
5000   *mancha <mancha1@zoho.com>*
5001
5002 * Fix eckey_priv_encode so it immediately returns an error upon a failure
5003   in i2d_ECPrivateKey.  Thanks to Ted Unangst for feedback on this issue.
5004
5005   *mancha <mancha1@zoho.com>*
5006
5007 * Fix some double frees. These are not thought to be exploitable.
5008
5009   *mancha <mancha1@zoho.com>*
5010
5011 * A missing bounds check in the handling of the TLS heartbeat extension
5012   can be used to reveal up to 64k of memory to a connected client or
5013   server.
5014
5015   Thanks for Neel Mehta of Google Security for discovering this bug and to
5016   Adam Langley <agl@chromium.org> and Bodo Moeller <bmoeller@acm.org> for
5017   preparing the fix ([CVE-2014-0160])
5018
5019   *Adam Langley, Bodo Moeller*
5020
5021 * Fix for the attack described in the paper "Recovering OpenSSL
5022   ECDSA Nonces Using the FLUSH+RELOAD Cache Side-channel Attack"
5023   by Yuval Yarom and Naomi Benger. Details can be obtained from:
5024   <http://eprint.iacr.org/2014/140>
5025
5026   Thanks to Yuval Yarom and Naomi Benger for discovering this
5027   flaw and to Yuval Yarom for supplying a fix ([CVE-2014-0076])
5028
5029   *Yuval Yarom and Naomi Benger*
5030
5031 * Use algorithm specific chains in SSL_CTX_use_certificate_chain_file():
5032   this fixes a limitation in previous versions of OpenSSL.
5033
5034   *Steve Henson*
5035
5036 * Experimental encrypt-then-mac support.
5037
5038   Experimental support for encrypt then mac from
5039   draft-gutmann-tls-encrypt-then-mac-02.txt
5040
5041   To enable it set the appropriate extension number (0x42 for the test
5042   server) using e.g. -DTLSEXT_TYPE_encrypt_then_mac=0x42
5043
5044   For non-compliant peers (i.e. just about everything) this should have no
5045   effect.
5046
5047   WARNING: EXPERIMENTAL, SUBJECT TO CHANGE.
5048
5049   *Steve Henson*
5050
5051 * Add EVP support for key wrapping algorithms, to avoid problems with
5052   existing code the flag EVP_CIPHER_CTX_WRAP_ALLOW has to be set in
5053   the EVP_CIPHER_CTX or an error is returned. Add AES and DES3 wrap
5054   algorithms and include tests cases.
5055
5056   *Steve Henson*
5057
5058 * Extend CMS code to support RSA-PSS signatures and RSA-OAEP for
5059   enveloped data.
5060
5061   *Steve Henson*
5062
5063 * Extended RSA OAEP support via EVP_PKEY API. Options to specify digest,
5064   MGF1 digest and OAEP label.
5065
5066   *Steve Henson*
5067
5068 * Make openssl verify return errors.
5069
5070   *Chris Palmer <palmer@google.com> and Ben Laurie*
5071
5072 * New function ASN1_TIME_diff to calculate the difference between two
5073   ASN1_TIME structures or one structure and the current time.
5074
5075   *Steve Henson*
5076
5077 * Update fips_test_suite to support multiple command line options. New
5078   test to induce all self test errors in sequence and check expected
5079   failures.
5080
5081   *Steve Henson*
5082
5083 * Add FIPS_{rsa,dsa,ecdsa}_{sign,verify} functions which digest and
5084   sign or verify all in one operation.
5085
5086   *Steve Henson*
5087
5088 * Add fips_algvs: a multicall fips utility incorporating all the algorithm
5089   test programs and fips_test_suite. Includes functionality to parse
5090   the minimal script output of fipsalgest.pl directly.
5091
5092   *Steve Henson*
5093
5094 * Add authorisation parameter to FIPS_module_mode_set().
5095
5096   *Steve Henson*
5097
5098 * Add FIPS selftest for ECDH algorithm using P-224 and B-233 curves.
5099
5100   *Steve Henson*
5101
5102 * Use separate DRBG fields for internal and external flags. New function
5103   FIPS_drbg_health_check() to perform on demand health checking. Add
5104   generation tests to fips_test_suite with reduced health check interval to
5105   demonstrate periodic health checking. Add "nodh" option to
5106   fips_test_suite to skip very slow DH test.
5107
5108   *Steve Henson*
5109
5110 * New function FIPS_get_cipherbynid() to lookup FIPS supported ciphers
5111   based on NID.
5112
5113   *Steve Henson*
5114
5115 * More extensive health check for DRBG checking many more failure modes.
5116   New function FIPS_selftest_drbg_all() to handle every possible DRBG
5117   combination: call this in fips_test_suite.
5118
5119   *Steve Henson*
5120
5121 * Add support for canonical generation of DSA parameter 'g'. See
5122   FIPS 186-3 A.2.3.
5123
5124 * Add support for HMAC DRBG from SP800-90. Update DRBG algorithm test and
5125   POST to handle HMAC cases.
5126
5127   *Steve Henson*
5128
5129 * Add functions FIPS_module_version() and FIPS_module_version_text()
5130   to return numerical and string versions of the FIPS module number.
5131
5132   *Steve Henson*
5133
5134 * Rename FIPS_mode_set and FIPS_mode to FIPS_module_mode_set and
5135   FIPS_module_mode. FIPS_mode and FIPS_mode_set will be implemented
5136   outside the validated module in the FIPS capable OpenSSL.
5137
5138   *Steve Henson*
5139
5140 * Minor change to DRBG entropy callback semantics. In some cases
5141   there is no multiple of the block length between min_len and
5142   max_len. Allow the callback to return more than max_len bytes
5143   of entropy but discard any extra: it is the callback's responsibility
5144   to ensure that the extra data discarded does not impact the
5145   requested amount of entropy.
5146
5147   *Steve Henson*
5148
5149 * Add PRNG security strength checks to RSA, DSA and ECDSA using
5150   information in FIPS186-3, SP800-57 and SP800-131A.
5151
5152   *Steve Henson*
5153
5154 * CCM support via EVP. Interface is very similar to GCM case except we
5155   must supply all data in one chunk (i.e. no update, final) and the
5156   message length must be supplied if AAD is used. Add algorithm test
5157   support.
5158
5159   *Steve Henson*
5160
5161 * Initial version of POST overhaul. Add POST callback to allow the status
5162   of POST to be monitored and/or failures induced. Modify fips_test_suite
5163   to use callback. Always run all selftests even if one fails.
5164
5165   *Steve Henson*
5166
5167 * XTS support including algorithm test driver in the fips_gcmtest program.
5168   Note: this does increase the maximum key length from 32 to 64 bytes but
5169   there should be no binary compatibility issues as existing applications
5170   will never use XTS mode.
5171
5172   *Steve Henson*
5173
5174 * Extensive reorganisation of FIPS PRNG behaviour. Remove all dependencies
5175   to OpenSSL RAND code and replace with a tiny FIPS RAND API which also
5176   performs algorithm blocking for unapproved PRNG types. Also do not
5177   set PRNG type in FIPS_mode_set(): leave this to the application.
5178   Add default OpenSSL DRBG handling: sets up FIPS PRNG and seeds with
5179   the standard OpenSSL PRNG: set additional data to a date time vector.
5180
5181   *Steve Henson*
5182
5183 * Rename old X9.31 PRNG functions of the form `FIPS_rand*` to `FIPS_x931*`.
5184   This shouldn't present any incompatibility problems because applications
5185   shouldn't be using these directly and any that are will need to rethink
5186   anyway as the X9.31 PRNG is now deprecated by FIPS 140-2
5187
5188   *Steve Henson*
5189
5190 * Extensive self tests and health checking required by SP800-90 DRBG.
5191   Remove strength parameter from FIPS_drbg_instantiate and always
5192   instantiate at maximum supported strength.
5193
5194   *Steve Henson*
5195
5196 * Add ECDH code to fips module and fips_ecdhvs for primitives only testing.
5197
5198   *Steve Henson*
5199
5200 * New algorithm test program fips_dhvs to handle DH primitives only testing.
5201
5202   *Steve Henson*
5203
5204 * New function DH_compute_key_padded() to compute a DH key and pad with
5205   leading zeroes if needed: this complies with SP800-56A et al.
5206
5207   *Steve Henson*
5208
5209 * Initial implementation of SP800-90 DRBGs for Hash and CTR. Not used by
5210   anything, incomplete, subject to change and largely untested at present.
5211
5212   *Steve Henson*
5213
5214 * Modify fipscanisteronly build option to only build the necessary object
5215   files by filtering FIPS_EX_OBJ through a perl script in crypto/Makefile.
5216
5217   *Steve Henson*
5218
5219 * Add experimental option FIPSSYMS to give all symbols in
5220   fipscanister.o and FIPS or fips prefix. This will avoid
5221   conflicts with future versions of OpenSSL. Add perl script
5222   util/fipsas.pl to preprocess assembly language source files
5223   and rename any affected symbols.
5224
5225   *Steve Henson*
5226
5227 * Add selftest checks and algorithm block of non-fips algorithms in
5228   FIPS mode. Remove DES2 from selftests.
5229
5230   *Steve Henson*
5231
5232 * Add ECDSA code to fips module. Add tiny fips_ecdsa_check to just
5233   return internal method without any ENGINE dependencies. Add new
5234   tiny fips sign and verify functions.
5235
5236   *Steve Henson*
5237
5238 * New build option no-ec2m to disable characteristic 2 code.
5239
5240   *Steve Henson*
5241
5242 * New build option "fipscanisteronly". This only builds fipscanister.o
5243   and (currently) associated fips utilities. Uses the file Makefile.fips
5244   instead of Makefile.org as the prototype.
5245
5246   *Steve Henson*
5247
5248 * Add some FIPS mode restrictions to GCM. Add internal IV generator.
5249   Update fips_gcmtest to use IV generator.
5250
5251   *Steve Henson*
5252
5253 * Initial, experimental EVP support for AES-GCM. AAD can be input by
5254   setting output buffer to NULL. The `*Final` function must be
5255   called although it will not retrieve any additional data. The tag
5256   can be set or retrieved with a ctrl. The IV length is by default 12
5257   bytes (96 bits) but can be set to an alternative value. If the IV
5258   length exceeds the maximum IV length (currently 16 bytes) it cannot be
5259   set before the key.
5260
5261   *Steve Henson*
5262
5263 * New flag in ciphers: EVP_CIPH_FLAG_CUSTOM_CIPHER. This means the
5264   underlying do_cipher function handles all cipher semantics itself
5265   including padding and finalisation. This is useful if (for example)
5266   an ENGINE cipher handles block padding itself. The behaviour of
5267   do_cipher is subtly changed if this flag is set: the return value
5268   is the number of characters written to the output buffer (zero is
5269   no longer an error code) or a negative error code. Also if the
5270   input buffer is NULL and length 0 finalisation should be performed.
5271
5272   *Steve Henson*
5273
5274 * If a candidate issuer certificate is already part of the constructed
5275   path ignore it: new debug notification X509_V_ERR_PATH_LOOP for this case.
5276
5277   *Steve Henson*
5278
5279 * Improve forward-security support: add functions
5280
5281           void SSL_CTX_set_not_resumable_session_callback(
5282                    SSL_CTX *ctx, int (*cb)(SSL *ssl, int is_forward_secure))
5283           void SSL_set_not_resumable_session_callback(
5284                    SSL *ssl, int (*cb)(SSL *ssl, int is_forward_secure))
5285
5286   for use by SSL/TLS servers; the callback function will be called whenever a
5287   new session is created, and gets to decide whether the session may be
5288   cached to make it resumable (return 0) or not (return 1).  (As by the
5289   SSL/TLS protocol specifications, the session_id sent by the server will be
5290   empty to indicate that the session is not resumable; also, the server will
5291   not generate RFC 4507 (RFC 5077) session tickets.)
5292
5293   A simple reasonable callback implementation is to return is_forward_secure.
5294   This parameter will be set to 1 or 0 depending on the ciphersuite selected
5295   by the SSL/TLS server library, indicating whether it can provide forward
5296   security.
5297
5298   *Emilia Käsper <emilia.kasper@esat.kuleuven.be> (Google)*
5299
5300 * New -verify_name option in command line utilities to set verification
5301   parameters by name.
5302
5303   *Steve Henson*
5304
5305 * Initial CMAC implementation. WARNING: EXPERIMENTAL, API MAY CHANGE.
5306   Add CMAC pkey methods.
5307
5308   *Steve Henson*
5309
5310 * Experimental renegotiation in s_server -www mode. If the client
5311   browses /reneg connection is renegotiated. If /renegcert it is
5312   renegotiated requesting a certificate.
5313
5314   *Steve Henson*
5315
5316 * Add an "external" session cache for debugging purposes to s_server. This
5317   should help trace issues which normally are only apparent in deployed
5318   multi-process servers.
5319
5320   *Steve Henson*
5321
5322 * Extensive audit of libcrypto with DEBUG_UNUSED. Fix many cases where
5323   return value is ignored. NB. The functions RAND_add(), RAND_seed(),
5324   BIO_set_cipher() and some obscure PEM functions were changed so they
5325   can now return an error. The RAND changes required a change to the
5326   RAND_METHOD structure.
5327
5328   *Steve Henson*
5329
5330 * New macro `__owur` for "OpenSSL Warn Unused Result". This makes use of
5331   a gcc attribute to warn if the result of a function is ignored. This
5332   is enable if DEBUG_UNUSED is set. Add to several functions in evp.h
5333   whose return value is often ignored.
5334
5335   *Steve Henson*
5336
5337 * New -noct, -requestct, -requirect and -ctlogfile options for s_client.
5338   These allow SCTs (signed certificate timestamps) to be requested and
5339   validated when establishing a connection.
5340
5341   *Rob Percival <robpercival@google.com>*
5342
5343OpenSSL 1.0.2
5344-------------
5345
5346### Changes between 1.0.2s and 1.0.2t [10 Sep 2019]
5347
5348 * For built-in EC curves, ensure an EC_GROUP built from the curve name is
5349   used even when parsing explicit parameters, when loading a encoded key
5350   or calling `EC_GROUP_new_from_ecpkparameters()`/
5351   `EC_GROUP_new_from_ecparameters()`.
5352   This prevents bypass of security hardening and performance gains,
5353   especially for curves with specialized EC_METHODs.
5354   By default, if a key encoded with explicit parameters is loaded and later
5355   encoded, the output is still encoded with explicit parameters, even if
5356   internally a "named" EC_GROUP is used for computation.
5357
5358   *Nicola Tuveri*
5359
5360 * Compute ECC cofactors if not provided during EC_GROUP construction. Before
5361   this change, EC_GROUP_set_generator would accept order and/or cofactor as
5362   NULL. After this change, only the cofactor parameter can be NULL. It also
5363   does some minimal sanity checks on the passed order.
5364   ([CVE-2019-1547])
5365
5366   *Billy Bob Brumley*
5367
5368 * Fixed a padding oracle in PKCS7_dataDecode and CMS_decrypt_set1_pkey.
5369   An attack is simple, if the first CMS_recipientInfo is valid but the
5370   second CMS_recipientInfo is chosen ciphertext. If the second
5371   recipientInfo decodes to PKCS #1 v1.5 form plaintext, the correct
5372   encryption key will be replaced by garbage, and the message cannot be
5373   decoded, but if the RSA decryption fails, the correct encryption key is
5374   used and the recipient will not notice the attack.
5375   As a work around for this potential attack the length of the decrypted
5376   key must be equal to the cipher default key length, in case the
5377   certifiate is not given and all recipientInfo are tried out.
5378   The old behaviour can be re-enabled in the CMS code by setting the
5379   CMS_DEBUG_DECRYPT flag.
5380   ([CVE-2019-1563])
5381
5382   *Bernd Edlinger*
5383
5384 * Document issue with installation paths in diverse Windows builds
5385
5386   '/usr/local/ssl' is an unsafe prefix for location to install OpenSSL
5387   binaries and run-time config file.
5388   ([CVE-2019-1552])
5389
5390   *Richard Levitte*
5391
5392### Changes between 1.0.2r and 1.0.2s [28 May 2019]
5393
5394 * Change the default RSA, DSA and DH size to 2048 bit instead of 1024.
5395   This changes the size when using the `genpkey` command when no size is given.
5396   It fixes an omission in earlier changes that changed all RSA, DSA and DH
5397   generation commands to use 2048 bits by default.
5398
5399   *Kurt Roeckx*
5400
5401 * Add FIPS support for Android Arm 64-bit
5402
5403   Support for Android Arm 64-bit was added to the OpenSSL FIPS Object
5404   Module in Version 2.0.10. For some reason, the corresponding target
5405   'android64-aarch64' was missing OpenSSL 1.0.2, whence it could not be
5406   built with FIPS support on Android Arm 64-bit. This omission has been
5407   fixed.
5408
5409   *Matthias St. Pierre*
5410
5411### Changes between 1.0.2q and 1.0.2r [26 Feb 2019]
5412
5413 * 0-byte record padding oracle
5414
5415   If an application encounters a fatal protocol error and then calls
5416   SSL_shutdown() twice (once to send a close_notify, and once to receive one)
5417   then OpenSSL can respond differently to the calling application if a 0 byte
5418   record is received with invalid padding compared to if a 0 byte record is
5419   received with an invalid MAC. If the application then behaves differently
5420   based on that in a way that is detectable to the remote peer, then this
5421   amounts to a padding oracle that could be used to decrypt data.
5422
5423   In order for this to be exploitable "non-stitched" ciphersuites must be in
5424   use. Stitched ciphersuites are optimised implementations of certain
5425   commonly used ciphersuites. Also the application must call SSL_shutdown()
5426   twice even if a protocol error has occurred (applications should not do
5427   this but some do anyway).
5428
5429   This issue was discovered by Juraj Somorovsky, Robert Merget and Nimrod
5430   Aviram, with additional investigation by Steven Collison and Andrew
5431   Hourselt. It was reported to OpenSSL on 10th December 2018.
5432   ([CVE-2019-1559])
5433
5434   *Matt Caswell*
5435
5436 * Move strictness check from EVP_PKEY_asn1_new() to EVP_PKEY_asn1_add0().
5437
5438   *Richard Levitte*
5439
5440### Changes between 1.0.2p and 1.0.2q [20 Nov 2018]
5441
5442 * Microarchitecture timing vulnerability in ECC scalar multiplication
5443
5444   OpenSSL ECC scalar multiplication, used in e.g. ECDSA and ECDH, has been
5445   shown to be vulnerable to a microarchitecture timing side channel attack.
5446   An attacker with sufficient access to mount local timing attacks during
5447   ECDSA signature generation could recover the private key.
5448
5449   This issue was reported to OpenSSL on 26th October 2018 by Alejandro
5450   Cabrera Aldaya, Billy Brumley, Sohaib ul Hassan, Cesar Pereida Garcia and
5451   Nicola Tuveri.
5452   ([CVE-2018-5407])
5453
5454   *Billy Brumley*
5455
5456 * Timing vulnerability in DSA signature generation
5457
5458   The OpenSSL DSA signature algorithm has been shown to be vulnerable to a
5459   timing side channel attack. An attacker could use variations in the signing
5460   algorithm to recover the private key.
5461
5462   This issue was reported to OpenSSL on 16th October 2018 by Samuel Weiser.
5463   ([CVE-2018-0734])
5464
5465   *Paul Dale*
5466
5467 * Resolve a compatibility issue in EC_GROUP handling with the FIPS Object
5468   Module, accidentally introduced while backporting security fixes from the
5469   development branch and hindering the use of ECC in FIPS mode.
5470
5471   *Nicola Tuveri*
5472
5473### Changes between 1.0.2o and 1.0.2p [14 Aug 2018]
5474
5475 * Client DoS due to large DH parameter
5476
5477   During key agreement in a TLS handshake using a DH(E) based ciphersuite a
5478   malicious server can send a very large prime value to the client. This will
5479   cause the client to spend an unreasonably long period of time generating a
5480   key for this prime resulting in a hang until the client has finished. This
5481   could be exploited in a Denial Of Service attack.
5482
5483   This issue was reported to OpenSSL on 5th June 2018 by Guido Vranken
5484   ([CVE-2018-0732])
5485
5486   *Guido Vranken*
5487
5488 * Cache timing vulnerability in RSA Key Generation
5489
5490   The OpenSSL RSA Key generation algorithm has been shown to be vulnerable to
5491   a cache timing side channel attack. An attacker with sufficient access to
5492   mount cache timing attacks during the RSA key generation process could
5493   recover the private key.
5494
5495   This issue was reported to OpenSSL on 4th April 2018 by Alejandro Cabrera
5496   Aldaya, Billy Brumley, Cesar Pereida Garcia and Luis Manuel Alvarez Tapia.
5497   ([CVE-2018-0737])
5498
5499   *Billy Brumley*
5500
5501 * Make EVP_PKEY_asn1_new() a bit stricter about its input.  A NULL pem_str
5502   parameter is no longer accepted, as it leads to a corrupt table.  NULL
5503   pem_str is reserved for alias entries only.
5504
5505   *Richard Levitte*
5506
5507 * Revert blinding in ECDSA sign and instead make problematic addition
5508   length-invariant. Switch even to fixed-length Montgomery multiplication.
5509
5510   *Andy Polyakov*
5511
5512 * Change generating and checking of primes so that the error rate of not
5513   being prime depends on the intended use based on the size of the input.
5514   For larger primes this will result in more rounds of Miller-Rabin.
5515   The maximal error rate for primes with more than 1080 bits is lowered
5516   to 2^-128.
5517
5518   *Kurt Roeckx, Annie Yousar*
5519
5520 * Increase the number of Miller-Rabin rounds for DSA key generating to 64.
5521
5522   *Kurt Roeckx*
5523
5524 * Add blinding to ECDSA and DSA signatures to protect against side channel
5525   attacks discovered by Keegan Ryan (NCC Group).
5526
5527   *Matt Caswell*
5528
5529 * When unlocking a pass phrase protected PEM file or PKCS#8 container, we
5530   now allow empty (zero character) pass phrases.
5531
5532   *Richard Levitte*
5533
5534 * Certificate time validation (X509_cmp_time) enforces stricter
5535   compliance with RFC 5280. Fractional seconds and timezone offsets
5536   are no longer allowed.
5537
5538   *Emilia Käsper*
5539
5540### Changes between 1.0.2n and 1.0.2o [27 Mar 2018]
5541
5542 * Constructed ASN.1 types with a recursive definition could exceed the stack
5543
5544   Constructed ASN.1 types with a recursive definition (such as can be found
5545   in PKCS7) could eventually exceed the stack given malicious input with
5546   excessive recursion. This could result in a Denial Of Service attack. There
5547   are no such structures used within SSL/TLS that come from untrusted sources
5548   so this is considered safe.
5549
5550   This issue was reported to OpenSSL on 4th January 2018 by the OSS-fuzz
5551   project.
5552   ([CVE-2018-0739])
5553
5554   *Matt Caswell*
5555
5556### Changes between 1.0.2m and 1.0.2n [7 Dec 2017]
5557
5558 * Read/write after SSL object in error state
5559
5560   OpenSSL 1.0.2 (starting from version 1.0.2b) introduced an "error state"
5561   mechanism. The intent was that if a fatal error occurred during a handshake
5562   then OpenSSL would move into the error state and would immediately fail if
5563   you attempted to continue the handshake. This works as designed for the
5564   explicit handshake functions (SSL_do_handshake(), SSL_accept() and
5565   SSL_connect()), however due to a bug it does not work correctly if
5566   SSL_read() or SSL_write() is called directly. In that scenario, if the
5567   handshake fails then a fatal error will be returned in the initial function
5568   call. If SSL_read()/SSL_write() is subsequently called by the application
5569   for the same SSL object then it will succeed and the data is passed without
5570   being decrypted/encrypted directly from the SSL/TLS record layer.
5571
5572   In order to exploit this issue an application bug would have to be present
5573   that resulted in a call to SSL_read()/SSL_write() being issued after having
5574   already received a fatal error.
5575
5576   This issue was reported to OpenSSL by David Benjamin (Google).
5577   ([CVE-2017-3737])
5578
5579   *Matt Caswell*
5580
5581 * rsaz_1024_mul_avx2 overflow bug on x86_64
5582
5583   There is an overflow bug in the AVX2 Montgomery multiplication procedure
5584   used in exponentiation with 1024-bit moduli. No EC algorithms are affected.
5585   Analysis suggests that attacks against RSA and DSA as a result of this
5586   defect would be very difficult to perform and are not believed likely.
5587   Attacks against DH1024 are considered just feasible, because most of the
5588   work necessary to deduce information about a private key may be performed
5589   offline. The amount of resources required for such an attack would be
5590   significant. However, for an attack on TLS to be meaningful, the server
5591   would have to share the DH1024 private key among multiple clients, which is
5592   no longer an option since CVE-2016-0701.
5593
5594   This only affects processors that support the AVX2 but not ADX extensions
5595   like Intel Haswell (4th generation).
5596
5597   This issue was reported to OpenSSL by David Benjamin (Google). The issue
5598   was originally found via the OSS-Fuzz project.
5599   ([CVE-2017-3738])
5600
5601   *Andy Polyakov*
5602
5603### Changes between 1.0.2l and 1.0.2m [2 Nov 2017]
5604
5605 * bn_sqrx8x_internal carry bug on x86_64
5606
5607   There is a carry propagating bug in the x86_64 Montgomery squaring
5608   procedure. No EC algorithms are affected. Analysis suggests that attacks
5609   against RSA and DSA as a result of this defect would be very difficult to
5610   perform and are not believed likely. Attacks against DH are considered just
5611   feasible (although very difficult) because most of the work necessary to
5612   deduce information about a private key may be performed offline. The amount
5613   of resources required for such an attack would be very significant and
5614   likely only accessible to a limited number of attackers. An attacker would
5615   additionally need online access to an unpatched system using the target
5616   private key in a scenario with persistent DH parameters and a private
5617   key that is shared between multiple clients.
5618
5619   This only affects processors that support the BMI1, BMI2 and ADX extensions
5620   like Intel Broadwell (5th generation) and later or AMD Ryzen.
5621
5622   This issue was reported to OpenSSL by the OSS-Fuzz project.
5623   ([CVE-2017-3736])
5624
5625   *Andy Polyakov*
5626
5627 * Malformed X.509 IPAddressFamily could cause OOB read
5628
5629   If an X.509 certificate has a malformed IPAddressFamily extension,
5630   OpenSSL could do a one-byte buffer overread. The most likely result
5631   would be an erroneous display of the certificate in text format.
5632
5633   This issue was reported to OpenSSL by the OSS-Fuzz project.
5634
5635   *Rich Salz*
5636
5637### Changes between 1.0.2k and 1.0.2l [25 May 2017]
5638
5639 * Have 'config' recognise 64-bit mingw and choose 'mingw64' as the target
5640   platform rather than 'mingw'.
5641
5642   *Richard Levitte*
5643
5644### Changes between 1.0.2j and 1.0.2k [26 Jan 2017]
5645
5646 * Truncated packet could crash via OOB read
5647
5648   If one side of an SSL/TLS path is running on a 32-bit host and a specific
5649   cipher is being used, then a truncated packet can cause that host to
5650   perform an out-of-bounds read, usually resulting in a crash.
5651
5652   This issue was reported to OpenSSL by Robert Święcki of Google.
5653   ([CVE-2017-3731])
5654
5655   *Andy Polyakov*
5656
5657 * BN_mod_exp may produce incorrect results on x86_64
5658
5659   There is a carry propagating bug in the x86_64 Montgomery squaring
5660   procedure. No EC algorithms are affected. Analysis suggests that attacks
5661   against RSA and DSA as a result of this defect would be very difficult to
5662   perform and are not believed likely. Attacks against DH are considered just
5663   feasible (although very difficult) because most of the work necessary to
5664   deduce information about a private key may be performed offline. The amount
5665   of resources required for such an attack would be very significant and
5666   likely only accessible to a limited number of attackers. An attacker would
5667   additionally need online access to an unpatched system using the target
5668   private key in a scenario with persistent DH parameters and a private
5669   key that is shared between multiple clients. For example this can occur by
5670   default in OpenSSL DHE based SSL/TLS ciphersuites. Note: This issue is very
5671   similar to CVE-2015-3193 but must be treated as a separate problem.
5672
5673   This issue was reported to OpenSSL by the OSS-Fuzz project.
5674   ([CVE-2017-3732])
5675
5676   *Andy Polyakov*
5677
5678 * Montgomery multiplication may produce incorrect results
5679
5680   There is a carry propagating bug in the Broadwell-specific Montgomery
5681   multiplication procedure that handles input lengths divisible by, but
5682   longer than 256 bits. Analysis suggests that attacks against RSA, DSA
5683   and DH private keys are impossible. This is because the subroutine in
5684   question is not used in operations with the private key itself and an input
5685   of the attacker's direct choice. Otherwise the bug can manifest itself as
5686   transient authentication and key negotiation failures or reproducible
5687   erroneous outcome of public-key operations with specially crafted input.
5688   Among EC algorithms only Brainpool P-512 curves are affected and one
5689   presumably can attack ECDH key negotiation. Impact was not analyzed in
5690   detail, because pre-requisites for attack are considered unlikely. Namely
5691   multiple clients have to choose the curve in question and the server has to
5692   share the private key among them, neither of which is default behaviour.
5693   Even then only clients that chose the curve will be affected.
5694
5695   This issue was publicly reported as transient failures and was not
5696   initially recognized as a security issue. Thanks to Richard Morgan for
5697   providing reproducible case.
5698   ([CVE-2016-7055])
5699
5700   *Andy Polyakov*
5701
5702 * OpenSSL now fails if it receives an unrecognised record type in TLS1.0
5703   or TLS1.1. Previously this only happened in SSLv3 and TLS1.2. This is to
5704   prevent issues where no progress is being made and the peer continually
5705   sends unrecognised record types, using up resources processing them.
5706
5707   *Matt Caswell*
5708
5709### Changes between 1.0.2i and 1.0.2j [26 Sep 2016]
5710
5711 * Missing CRL sanity check
5712
5713   A bug fix which included a CRL sanity check was added to OpenSSL 1.1.0
5714   but was omitted from OpenSSL 1.0.2i. As a result any attempt to use
5715   CRLs in OpenSSL 1.0.2i will crash with a null pointer exception.
5716
5717   This issue only affects the OpenSSL 1.0.2i
5718   ([CVE-2016-7052])
5719
5720   *Matt Caswell*
5721
5722### Changes between 1.0.2h and 1.0.2i [22 Sep 2016]
5723
5724 * OCSP Status Request extension unbounded memory growth
5725
5726   A malicious client can send an excessively large OCSP Status Request
5727   extension. If that client continually requests renegotiation, sending a
5728   large OCSP Status Request extension each time, then there will be unbounded
5729   memory growth on the server. This will eventually lead to a Denial Of
5730   Service attack through memory exhaustion. Servers with a default
5731   configuration are vulnerable even if they do not support OCSP. Builds using
5732   the "no-ocsp" build time option are not affected.
5733
5734   This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.)
5735   ([CVE-2016-6304])
5736
5737   *Matt Caswell*
5738
5739 * In order to mitigate the SWEET32 attack, the DES ciphers were moved from
5740   HIGH to MEDIUM.
5741
5742   This issue was reported to OpenSSL Karthikeyan Bhargavan and Gaetan
5743   Leurent (INRIA)
5744   ([CVE-2016-2183])
5745
5746   *Rich Salz*
5747
5748 * OOB write in MDC2_Update()
5749
5750   An overflow can occur in MDC2_Update() either if called directly or
5751   through the EVP_DigestUpdate() function using MDC2. If an attacker
5752   is able to supply very large amounts of input data after a previous
5753   call to EVP_EncryptUpdate() with a partial block then a length check
5754   can overflow resulting in a heap corruption.
5755
5756   The amount of data needed is comparable to SIZE_MAX which is impractical
5757   on most platforms.
5758
5759   This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.)
5760   ([CVE-2016-6303])
5761
5762   *Stephen Henson*
5763
5764 * Malformed SHA512 ticket DoS
5765
5766   If a server uses SHA512 for TLS session ticket HMAC it is vulnerable to a
5767   DoS attack where a malformed ticket will result in an OOB read which will
5768   ultimately crash.
5769
5770   The use of SHA512 in TLS session tickets is comparatively rare as it requires
5771   a custom server callback and ticket lookup mechanism.
5772
5773   This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.)
5774   ([CVE-2016-6302])
5775
5776   *Stephen Henson*
5777
5778 * OOB write in BN_bn2dec()
5779
5780   The function BN_bn2dec() does not check the return value of BN_div_word().
5781   This can cause an OOB write if an application uses this function with an
5782   overly large BIGNUM. This could be a problem if an overly large certificate
5783   or CRL is printed out from an untrusted source. TLS is not affected because
5784   record limits will reject an oversized certificate before it is parsed.
5785
5786   This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.)
5787   ([CVE-2016-2182])
5788
5789   *Stephen Henson*
5790
5791 * OOB read in TS_OBJ_print_bio()
5792
5793   The function TS_OBJ_print_bio() misuses OBJ_obj2txt(): the return value is
5794   the total length the OID text representation would use and not the amount
5795   of data written. This will result in OOB reads when large OIDs are
5796   presented.
5797
5798   This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.)
5799   ([CVE-2016-2180])
5800
5801   *Stephen Henson*
5802
5803 * Pointer arithmetic undefined behaviour
5804
5805   Avoid some undefined pointer arithmetic
5806
5807   A common idiom in the codebase is to check limits in the following manner:
5808   "p + len > limit"
5809
5810   Where "p" points to some malloc'd data of SIZE bytes and
5811   limit == p + SIZE
5812
5813   "len" here could be from some externally supplied data (e.g. from a TLS
5814   message).
5815
5816   The rules of C pointer arithmetic are such that "p + len" is only well
5817   defined where len <= SIZE. Therefore the above idiom is actually
5818   undefined behaviour.
5819
5820   For example this could cause problems if some malloc implementation
5821   provides an address for "p" such that "p + len" actually overflows for
5822   values of len that are too big and therefore p + len < limit.
5823
5824   This issue was reported to OpenSSL by Guido Vranken
5825   ([CVE-2016-2177])
5826
5827   *Matt Caswell*
5828
5829 * Constant time flag not preserved in DSA signing
5830
5831   Operations in the DSA signing algorithm should run in constant time in
5832   order to avoid side channel attacks. A flaw in the OpenSSL DSA
5833   implementation means that a non-constant time codepath is followed for
5834   certain operations. This has been demonstrated through a cache-timing
5835   attack to be sufficient for an attacker to recover the private DSA key.
5836
5837   This issue was reported by César Pereida (Aalto University), Billy Brumley
5838   (Tampere University of Technology), and Yuval Yarom (The University of
5839   Adelaide and NICTA).
5840   ([CVE-2016-2178])
5841
5842   *César Pereida*
5843
5844 * DTLS buffered message DoS
5845
5846   In a DTLS connection where handshake messages are delivered out-of-order
5847   those messages that OpenSSL is not yet ready to process will be buffered
5848   for later use. Under certain circumstances, a flaw in the logic means that
5849   those messages do not get removed from the buffer even though the handshake
5850   has been completed. An attacker could force up to approx. 15 messages to
5851   remain in the buffer when they are no longer required. These messages will
5852   be cleared when the DTLS connection is closed. The default maximum size for
5853   a message is 100k. Therefore the attacker could force an additional 1500k
5854   to be consumed per connection. By opening many simulataneous connections an
5855   attacker could cause a DoS attack through memory exhaustion.
5856
5857   This issue was reported to OpenSSL by Quan Luo.
5858   ([CVE-2016-2179])
5859
5860   *Matt Caswell*
5861
5862 * DTLS replay protection DoS
5863
5864   A flaw in the DTLS replay attack protection mechanism means that records
5865   that arrive for future epochs update the replay protection "window" before
5866   the MAC for the record has been validated. This could be exploited by an
5867   attacker by sending a record for the next epoch (which does not have to
5868   decrypt or have a valid MAC), with a very large sequence number. This means
5869   that all subsequent legitimate packets are dropped causing a denial of
5870   service for a specific DTLS connection.
5871
5872   This issue was reported to OpenSSL by the OCAP audit team.
5873   ([CVE-2016-2181])
5874
5875   *Matt Caswell*
5876
5877 * Certificate message OOB reads
5878
5879   In OpenSSL 1.0.2 and earlier some missing message length checks can result
5880   in OOB reads of up to 2 bytes beyond an allocated buffer. There is a
5881   theoretical DoS risk but this has not been observed in practice on common
5882   platforms.
5883
5884   The messages affected are client certificate, client certificate request
5885   and server certificate. As a result the attack can only be performed
5886   against a client or a server which enables client authentication.
5887
5888   This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.)
5889   ([CVE-2016-6306])
5890
5891   *Stephen Henson*
5892
5893### Changes between 1.0.2g and 1.0.2h [3 May 2016]
5894
5895 * Prevent padding oracle in AES-NI CBC MAC check
5896
5897   A MITM attacker can use a padding oracle attack to decrypt traffic
5898   when the connection uses an AES CBC cipher and the server support
5899   AES-NI.
5900
5901   This issue was introduced as part of the fix for Lucky 13 padding
5902   attack ([CVE-2013-0169]). The padding check was rewritten to be in
5903   constant time by making sure that always the same bytes are read and
5904   compared against either the MAC or padding bytes. But it no longer
5905   checked that there was enough data to have both the MAC and padding
5906   bytes.
5907
5908   This issue was reported by Juraj Somorovsky using TLS-Attacker.
5909
5910   *Kurt Roeckx*
5911
5912 * Fix EVP_EncodeUpdate overflow
5913
5914   An overflow can occur in the EVP_EncodeUpdate() function which is used for
5915   Base64 encoding of binary data. If an attacker is able to supply very large
5916   amounts of input data then a length check can overflow resulting in a heap
5917   corruption.
5918
5919   Internally to OpenSSL the EVP_EncodeUpdate() function is primarily used by
5920   the `PEM_write_bio*` family of functions. These are mainly used within the
5921   OpenSSL command line applications, so any application which processes data
5922   from an untrusted source and outputs it as a PEM file should be considered
5923   vulnerable to this issue. User applications that call these APIs directly
5924   with large amounts of untrusted data may also be vulnerable.
5925
5926   This issue was reported by Guido Vranken.
5927   ([CVE-2016-2105])
5928
5929   *Matt Caswell*
5930
5931 * Fix EVP_EncryptUpdate overflow
5932
5933   An overflow can occur in the EVP_EncryptUpdate() function. If an attacker
5934   is able to supply very large amounts of input data after a previous call to
5935   EVP_EncryptUpdate() with a partial block then a length check can overflow
5936   resulting in a heap corruption. Following an analysis of all OpenSSL
5937   internal usage of the EVP_EncryptUpdate() function all usage is one of two
5938   forms. The first form is where the EVP_EncryptUpdate() call is known to be
5939   the first called function after an EVP_EncryptInit(), and therefore that
5940   specific call must be safe. The second form is where the length passed to
5941   EVP_EncryptUpdate() can be seen from the code to be some small value and
5942   therefore there is no possibility of an overflow. Since all instances are
5943   one of these two forms, it is believed that there can be no overflows in
5944   internal code due to this problem. It should be noted that
5945   EVP_DecryptUpdate() can call EVP_EncryptUpdate() in certain code paths.
5946   Also EVP_CipherUpdate() is a synonym for EVP_EncryptUpdate(). All instances
5947   of these calls have also been analysed too and it is believed there are no
5948   instances in internal usage where an overflow could occur.
5949
5950   This issue was reported by Guido Vranken.
5951   ([CVE-2016-2106])
5952
5953   *Matt Caswell*
5954
5955 * Prevent ASN.1 BIO excessive memory allocation
5956
5957   When ASN.1 data is read from a BIO using functions such as d2i_CMS_bio()
5958   a short invalid encoding can cause allocation of large amounts of memory
5959   potentially consuming excessive resources or exhausting memory.
5960
5961   Any application parsing untrusted data through d2i BIO functions is
5962   affected. The memory based functions such as d2i_X509() are *not* affected.
5963   Since the memory based functions are used by the TLS library, TLS
5964   applications are not affected.
5965
5966   This issue was reported by Brian Carpenter.
5967   ([CVE-2016-2109])
5968
5969   *Stephen Henson*
5970
5971 * EBCDIC overread
5972
5973   ASN1 Strings that are over 1024 bytes can cause an overread in applications
5974   using the X509_NAME_oneline() function on EBCDIC systems. This could result
5975   in arbitrary stack data being returned in the buffer.
5976
5977   This issue was reported by Guido Vranken.
5978   ([CVE-2016-2176])
5979
5980   *Matt Caswell*
5981
5982 * Modify behavior of ALPN to invoke callback after SNI/servername
5983   callback, such that updates to the SSL_CTX affect ALPN.
5984
5985   *Todd Short*
5986
5987 * Remove LOW from the DEFAULT cipher list.  This removes singles DES from the
5988   default.
5989
5990   *Kurt Roeckx*
5991
5992 * Only remove the SSLv2 methods with the no-ssl2-method option. When the
5993   methods are enabled and ssl2 is disabled the methods return NULL.
5994
5995   *Kurt Roeckx*
5996
5997### Changes between 1.0.2f and 1.0.2g [1 Mar 2016]
5998
5999* Disable weak ciphers in SSLv3 and up in default builds of OpenSSL.
6000  Builds that are not configured with "enable-weak-ssl-ciphers" will not
6001  provide any "EXPORT" or "LOW" strength ciphers.
6002
6003  *Viktor Dukhovni*
6004
6005* Disable SSLv2 default build, default negotiation and weak ciphers.  SSLv2
6006  is by default disabled at build-time.  Builds that are not configured with
6007  "enable-ssl2" will not support SSLv2.  Even if "enable-ssl2" is used,
6008  users who want to negotiate SSLv2 via the version-flexible SSLv23_method()
6009  will need to explicitly call either of:
6010
6011      SSL_CTX_clear_options(ctx, SSL_OP_NO_SSLv2);
6012  or
6013      SSL_clear_options(ssl, SSL_OP_NO_SSLv2);
6014
6015  as appropriate.  Even if either of those is used, or the application
6016  explicitly uses the version-specific SSLv2_method() or its client and
6017  server variants, SSLv2 ciphers vulnerable to exhaustive search key
6018  recovery have been removed.  Specifically, the SSLv2 40-bit EXPORT
6019  ciphers, and SSLv2 56-bit DES are no longer available.
6020  ([CVE-2016-0800])
6021
6022   *Viktor Dukhovni*
6023
6024 * Fix a double-free in DSA code
6025
6026   A double free bug was discovered when OpenSSL parses malformed DSA private
6027   keys and could lead to a DoS attack or memory corruption for applications
6028   that receive DSA private keys from untrusted sources.  This scenario is
6029   considered rare.
6030
6031   This issue was reported to OpenSSL by Adam Langley(Google/BoringSSL) using
6032   libFuzzer.
6033   ([CVE-2016-0705])
6034
6035   *Stephen Henson*
6036
6037 * Disable SRP fake user seed to address a server memory leak.
6038
6039   Add a new method SRP_VBASE_get1_by_user that handles the seed properly.
6040
6041   SRP_VBASE_get_by_user had inconsistent memory management behaviour.
6042   In order to fix an unavoidable memory leak, SRP_VBASE_get_by_user
6043   was changed to ignore the "fake user" SRP seed, even if the seed
6044   is configured.
6045
6046   Users should use SRP_VBASE_get1_by_user instead. Note that in
6047   SRP_VBASE_get1_by_user, caller must free the returned value. Note
6048   also that even though configuring the SRP seed attempts to hide
6049   invalid usernames by continuing the handshake with fake
6050   credentials, this behaviour is not constant time and no strong
6051   guarantees are made that the handshake is indistinguishable from
6052   that of a valid user.
6053   ([CVE-2016-0798])
6054
6055   *Emilia Käsper*
6056
6057 * Fix BN_hex2bn/BN_dec2bn NULL pointer deref/heap corruption
6058
6059   In the BN_hex2bn function the number of hex digits is calculated using an
6060   int value `i`. Later `bn_expand` is called with a value of `i * 4`. For
6061   large values of `i` this can result in `bn_expand` not allocating any
6062   memory because `i * 4` is negative. This can leave the internal BIGNUM data
6063   field as NULL leading to a subsequent NULL ptr deref. For very large values
6064   of `i`, the calculation `i * 4` could be a positive value smaller than `i`.
6065   In this case memory is allocated to the internal BIGNUM data field, but it
6066   is insufficiently sized leading to heap corruption. A similar issue exists
6067   in BN_dec2bn. This could have security consequences if BN_hex2bn/BN_dec2bn
6068   is ever called by user applications with very large untrusted hex/dec data.
6069   This is anticipated to be a rare occurrence.
6070
6071   All OpenSSL internal usage of these functions use data that is not expected
6072   to be untrusted, e.g. config file data or application command line
6073   arguments. If user developed applications generate config file data based
6074   on untrusted data then it is possible that this could also lead to security
6075   consequences. This is also anticipated to be rare.
6076
6077   This issue was reported to OpenSSL by Guido Vranken.
6078   ([CVE-2016-0797])
6079
6080   *Matt Caswell*
6081
6082 * Fix memory issues in `BIO_*printf` functions
6083
6084   The internal `fmtstr` function used in processing a "%s" format string in
6085   the `BIO_*printf` functions could overflow while calculating the length of a
6086   string and cause an OOB read when printing very long strings.
6087
6088   Additionally the internal `doapr_outch` function can attempt to write to an
6089   OOB memory location (at an offset from the NULL pointer) in the event of a
6090   memory allocation failure. In 1.0.2 and below this could be caused where
6091   the size of a buffer to be allocated is greater than INT_MAX. E.g. this
6092   could be in processing a very long "%s" format string. Memory leaks can
6093   also occur.
6094
6095   The first issue may mask the second issue dependent on compiler behaviour.
6096   These problems could enable attacks where large amounts of untrusted data
6097   is passed to the `BIO_*printf` functions. If applications use these functions
6098   in this way then they could be vulnerable. OpenSSL itself uses these
6099   functions when printing out human-readable dumps of ASN.1 data. Therefore
6100   applications that print this data could be vulnerable if the data is from
6101   untrusted sources. OpenSSL command line applications could also be
6102   vulnerable where they print out ASN.1 data, or if untrusted data is passed
6103   as command line arguments.
6104
6105   Libssl is not considered directly vulnerable. Additionally certificates etc
6106   received via remote connections via libssl are also unlikely to be able to
6107   trigger these issues because of message size limits enforced within libssl.
6108
6109   This issue was reported to OpenSSL Guido Vranken.
6110   ([CVE-2016-0799])
6111
6112   *Matt Caswell*
6113
6114 * Side channel attack on modular exponentiation
6115
6116   A side-channel attack was found which makes use of cache-bank conflicts on
6117   the Intel Sandy-Bridge microarchitecture which could lead to the recovery
6118   of RSA keys.  The ability to exploit this issue is limited as it relies on
6119   an attacker who has control of code in a thread running on the same
6120   hyper-threaded core as the victim thread which is performing decryptions.
6121
6122   This issue was reported to OpenSSL by Yuval Yarom, The University of
6123   Adelaide and NICTA, Daniel Genkin, Technion and Tel Aviv University, and
6124   Nadia Heninger, University of Pennsylvania with more information at
6125   <http://cachebleed.info>.
6126   ([CVE-2016-0702])
6127
6128   *Andy Polyakov*
6129
6130 * Change the `req` command to generate a 2048-bit RSA/DSA key by default,
6131   if no keysize is specified with default_bits. This fixes an
6132   omission in an earlier change that changed all RSA/DSA key generation
6133   commands to use 2048 bits by default.
6134
6135   *Emilia Käsper*
6136
6137### Changes between 1.0.2e and 1.0.2f [28 Jan 2016]
6138
6139 * DH small subgroups
6140
6141   Historically OpenSSL only ever generated DH parameters based on "safe"
6142   primes. More recently (in version 1.0.2) support was provided for
6143   generating X9.42 style parameter files such as those required for RFC 5114
6144   support. The primes used in such files may not be "safe". Where an
6145   application is using DH configured with parameters based on primes that are
6146   not "safe" then an attacker could use this fact to find a peer's private
6147   DH exponent. This attack requires that the attacker complete multiple
6148   handshakes in which the peer uses the same private DH exponent. For example
6149   this could be used to discover a TLS server's private DH exponent if it's
6150   reusing the private DH exponent or it's using a static DH ciphersuite.
6151
6152   OpenSSL provides the option SSL_OP_SINGLE_DH_USE for ephemeral DH (DHE) in
6153   TLS. It is not on by default. If the option is not set then the server
6154   reuses the same private DH exponent for the life of the server process and
6155   would be vulnerable to this attack. It is believed that many popular
6156   applications do set this option and would therefore not be at risk.
6157
6158   The fix for this issue adds an additional check where a "q" parameter is
6159   available (as is the case in X9.42 based parameters). This detects the
6160   only known attack, and is the only possible defense for static DH
6161   ciphersuites. This could have some performance impact.
6162
6163   Additionally the SSL_OP_SINGLE_DH_USE option has been switched on by
6164   default and cannot be disabled. This could have some performance impact.
6165
6166   This issue was reported to OpenSSL by Antonio Sanso (Adobe).
6167   ([CVE-2016-0701])
6168
6169   *Matt Caswell*
6170
6171 * SSLv2 doesn't block disabled ciphers
6172
6173   A malicious client can negotiate SSLv2 ciphers that have been disabled on
6174   the server and complete SSLv2 handshakes even if all SSLv2 ciphers have
6175   been disabled, provided that the SSLv2 protocol was not also disabled via
6176   SSL_OP_NO_SSLv2.
6177
6178   This issue was reported to OpenSSL on 26th December 2015 by Nimrod Aviram
6179   and Sebastian Schinzel.
6180   ([CVE-2015-3197])
6181
6182   *Viktor Dukhovni*
6183
6184### Changes between 1.0.2d and 1.0.2e [3 Dec 2015]
6185
6186 * BN_mod_exp may produce incorrect results on x86_64
6187
6188   There is a carry propagating bug in the x86_64 Montgomery squaring
6189   procedure. No EC algorithms are affected. Analysis suggests that attacks
6190   against RSA and DSA as a result of this defect would be very difficult to
6191   perform and are not believed likely. Attacks against DH are considered just
6192   feasible (although very difficult) because most of the work necessary to
6193   deduce information about a private key may be performed offline. The amount
6194   of resources required for such an attack would be very significant and
6195   likely only accessible to a limited number of attackers. An attacker would
6196   additionally need online access to an unpatched system using the target
6197   private key in a scenario with persistent DH parameters and a private
6198   key that is shared between multiple clients. For example this can occur by
6199   default in OpenSSL DHE based SSL/TLS ciphersuites.
6200
6201   This issue was reported to OpenSSL by Hanno Böck.
6202   ([CVE-2015-3193])
6203
6204   *Andy Polyakov*
6205
6206 * Certificate verify crash with missing PSS parameter
6207
6208   The signature verification routines will crash with a NULL pointer
6209   dereference if presented with an ASN.1 signature using the RSA PSS
6210   algorithm and absent mask generation function parameter. Since these
6211   routines are used to verify certificate signature algorithms this can be
6212   used to crash any certificate verification operation and exploited in a
6213   DoS attack. Any application which performs certificate verification is
6214   vulnerable including OpenSSL clients and servers which enable client
6215   authentication.
6216
6217   This issue was reported to OpenSSL by Loïc Jonas Etienne (Qnective AG).
6218   ([CVE-2015-3194])
6219
6220   *Stephen Henson*
6221
6222 * X509_ATTRIBUTE memory leak
6223
6224   When presented with a malformed X509_ATTRIBUTE structure OpenSSL will leak
6225   memory. This structure is used by the PKCS#7 and CMS routines so any
6226   application which reads PKCS#7 or CMS data from untrusted sources is
6227   affected. SSL/TLS is not affected.
6228
6229   This issue was reported to OpenSSL by Adam Langley (Google/BoringSSL) using
6230   libFuzzer.
6231   ([CVE-2015-3195])
6232
6233   *Stephen Henson*
6234
6235 * Rewrite EVP_DecodeUpdate (base64 decoding) to fix several bugs.
6236   This changes the decoding behaviour for some invalid messages,
6237   though the change is mostly in the more lenient direction, and
6238   legacy behaviour is preserved as much as possible.
6239
6240   *Emilia Käsper*
6241
6242 * In DSA_generate_parameters_ex, if the provided seed is too short,
6243   return an error
6244
6245   *Rich Salz and Ismo Puustinen <ismo.puustinen@intel.com>*
6246
6247### Changes between 1.0.2c and 1.0.2d [9 Jul 2015]
6248
6249 * Alternate chains certificate forgery
6250
6251   During certificate verification, OpenSSL will attempt to find an
6252   alternative certificate chain if the first attempt to build such a chain
6253   fails. An error in the implementation of this logic can mean that an
6254   attacker could cause certain checks on untrusted certificates to be
6255   bypassed, such as the CA flag, enabling them to use a valid leaf
6256   certificate to act as a CA and "issue" an invalid certificate.
6257
6258   This issue was reported to OpenSSL by Adam Langley/David Benjamin
6259   (Google/BoringSSL).
6260
6261   *Matt Caswell*
6262
6263### Changes between 1.0.2b and 1.0.2c [12 Jun 2015]
6264
6265 * Fix HMAC ABI incompatibility. The previous version introduced an ABI
6266   incompatibility in the handling of HMAC. The previous ABI has now been
6267   restored.
6268
6269   *Matt Caswell*
6270
6271### Changes between 1.0.2a and 1.0.2b [11 Jun 2015]
6272
6273 * Malformed ECParameters causes infinite loop
6274
6275   When processing an ECParameters structure OpenSSL enters an infinite loop
6276   if the curve specified is over a specially malformed binary polynomial
6277   field.
6278
6279   This can be used to perform denial of service against any
6280   system which processes public keys, certificate requests or
6281   certificates.  This includes TLS clients and TLS servers with
6282   client authentication enabled.
6283
6284   This issue was reported to OpenSSL by Joseph Barr-Pixton.
6285   ([CVE-2015-1788])
6286
6287   *Andy Polyakov*
6288
6289 * Exploitable out-of-bounds read in X509_cmp_time
6290
6291   X509_cmp_time does not properly check the length of the ASN1_TIME
6292   string and can read a few bytes out of bounds. In addition,
6293   X509_cmp_time accepts an arbitrary number of fractional seconds in the
6294   time string.
6295
6296   An attacker can use this to craft malformed certificates and CRLs of
6297   various sizes and potentially cause a segmentation fault, resulting in
6298   a DoS on applications that verify certificates or CRLs. TLS clients
6299   that verify CRLs are affected. TLS clients and servers with client
6300   authentication enabled may be affected if they use custom verification
6301   callbacks.
6302
6303   This issue was reported to OpenSSL by Robert Swiecki (Google), and
6304   independently by Hanno Böck.
6305   ([CVE-2015-1789])
6306
6307   *Emilia Käsper*
6308
6309 * PKCS7 crash with missing EnvelopedContent
6310
6311   The PKCS#7 parsing code does not handle missing inner EncryptedContent
6312   correctly. An attacker can craft malformed ASN.1-encoded PKCS#7 blobs
6313   with missing content and trigger a NULL pointer dereference on parsing.
6314
6315   Applications that decrypt PKCS#7 data or otherwise parse PKCS#7
6316   structures from untrusted sources are affected. OpenSSL clients and
6317   servers are not affected.
6318
6319   This issue was reported to OpenSSL by Michal Zalewski (Google).
6320   ([CVE-2015-1790])
6321
6322   *Emilia Käsper*
6323
6324 * CMS verify infinite loop with unknown hash function
6325
6326   When verifying a signedData message the CMS code can enter an infinite loop
6327   if presented with an unknown hash function OID. This can be used to perform
6328   denial of service against any system which verifies signedData messages using
6329   the CMS code.
6330   This issue was reported to OpenSSL by Johannes Bauer.
6331   ([CVE-2015-1792])
6332
6333   *Stephen Henson*
6334
6335 * Race condition handling NewSessionTicket
6336
6337   If a NewSessionTicket is received by a multi-threaded client when attempting to
6338   reuse a previous ticket then a race condition can occur potentially leading to
6339   a double free of the ticket data.
6340   ([CVE-2015-1791])
6341
6342   *Matt Caswell*
6343
6344 * Only support 256-bit or stronger elliptic curves with the
6345   'ecdh_auto' setting (server) or by default (client). Of supported
6346   curves, prefer P-256 (both).
6347
6348   *Emilia Kasper*
6349
6350### Changes between 1.0.2 and 1.0.2a [19 Mar 2015]
6351
6352 * ClientHello sigalgs DoS fix
6353
6354   If a client connects to an OpenSSL 1.0.2 server and renegotiates with an
6355   invalid signature algorithms extension a NULL pointer dereference will
6356   occur. This can be exploited in a DoS attack against the server.
6357
6358   This issue was was reported to OpenSSL by David Ramos of Stanford
6359   University.
6360   ([CVE-2015-0291])
6361
6362   *Stephen Henson and Matt Caswell*
6363
6364 * Multiblock corrupted pointer fix
6365
6366   OpenSSL 1.0.2 introduced the "multiblock" performance improvement. This
6367   feature only applies on 64 bit x86 architecture platforms that support AES
6368   NI instructions. A defect in the implementation of "multiblock" can cause
6369   OpenSSL's internal write buffer to become incorrectly set to NULL when
6370   using non-blocking IO. Typically, when the user application is using a
6371   socket BIO for writing, this will only result in a failed connection.
6372   However if some other BIO is used then it is likely that a segmentation
6373   fault will be triggered, thus enabling a potential DoS attack.
6374
6375   This issue was reported to OpenSSL by Daniel Danner and Rainer Mueller.
6376   ([CVE-2015-0290])
6377
6378   *Matt Caswell*
6379
6380 * Segmentation fault in DTLSv1_listen fix
6381
6382   The DTLSv1_listen function is intended to be stateless and processes the
6383   initial ClientHello from many peers. It is common for user code to loop
6384   over the call to DTLSv1_listen until a valid ClientHello is received with
6385   an associated cookie. A defect in the implementation of DTLSv1_listen means
6386   that state is preserved in the SSL object from one invocation to the next
6387   that can lead to a segmentation fault. Errors processing the initial
6388   ClientHello can trigger this scenario. An example of such an error could be
6389   that a DTLS1.0 only client is attempting to connect to a DTLS1.2 only
6390   server.
6391
6392   This issue was reported to OpenSSL by Per Allansson.
6393   ([CVE-2015-0207])
6394
6395   *Matt Caswell*
6396
6397 * Segmentation fault in ASN1_TYPE_cmp fix
6398
6399   The function ASN1_TYPE_cmp will crash with an invalid read if an attempt is
6400   made to compare ASN.1 boolean types. Since ASN1_TYPE_cmp is used to check
6401   certificate signature algorithm consistency this can be used to crash any
6402   certificate verification operation and exploited in a DoS attack. Any
6403   application which performs certificate verification is vulnerable including
6404   OpenSSL clients and servers which enable client authentication.
6405   ([CVE-2015-0286])
6406
6407   *Stephen Henson*
6408
6409 * Segmentation fault for invalid PSS parameters fix
6410
6411   The signature verification routines will crash with a NULL pointer
6412   dereference if presented with an ASN.1 signature using the RSA PSS
6413   algorithm and invalid parameters. Since these routines are used to verify
6414   certificate signature algorithms this can be used to crash any
6415   certificate verification operation and exploited in a DoS attack. Any
6416   application which performs certificate verification is vulnerable including
6417   OpenSSL clients and servers which enable client authentication.
6418
6419   This issue was was reported to OpenSSL by Brian Carpenter.
6420   ([CVE-2015-0208])
6421
6422   *Stephen Henson*
6423
6424 * ASN.1 structure reuse memory corruption fix
6425
6426   Reusing a structure in ASN.1 parsing may allow an attacker to cause
6427   memory corruption via an invalid write. Such reuse is and has been
6428   strongly discouraged and is believed to be rare.
6429
6430   Applications that parse structures containing CHOICE or ANY DEFINED BY
6431   components may be affected. Certificate parsing (d2i_X509 and related
6432   functions) are however not affected. OpenSSL clients and servers are
6433   not affected.
6434   ([CVE-2015-0287])
6435
6436   *Stephen Henson*
6437
6438 * PKCS7 NULL pointer dereferences fix
6439
6440   The PKCS#7 parsing code does not handle missing outer ContentInfo
6441   correctly. An attacker can craft malformed ASN.1-encoded PKCS#7 blobs with
6442   missing content and trigger a NULL pointer dereference on parsing.
6443
6444   Applications that verify PKCS#7 signatures, decrypt PKCS#7 data or
6445   otherwise parse PKCS#7 structures from untrusted sources are
6446   affected. OpenSSL clients and servers are not affected.
6447
6448   This issue was reported to OpenSSL by Michal Zalewski (Google).
6449   ([CVE-2015-0289])
6450
6451   *Emilia Käsper*
6452
6453 * DoS via reachable assert in SSLv2 servers fix
6454
6455   A malicious client can trigger an OPENSSL_assert (i.e., an abort) in
6456   servers that both support SSLv2 and enable export cipher suites by sending
6457   a specially crafted SSLv2 CLIENT-MASTER-KEY message.
6458
6459   This issue was discovered by Sean Burford (Google) and Emilia Käsper
6460   (OpenSSL development team).
6461   ([CVE-2015-0293])
6462
6463   *Emilia Käsper*
6464
6465 * Empty CKE with client auth and DHE fix
6466
6467   If client auth is used then a server can seg fault in the event of a DHE
6468   ciphersuite being selected and a zero length ClientKeyExchange message
6469   being sent by the client. This could be exploited in a DoS attack.
6470   ([CVE-2015-1787])
6471
6472   *Matt Caswell*
6473
6474 * Handshake with unseeded PRNG fix
6475
6476   Under certain conditions an OpenSSL 1.0.2 client can complete a handshake
6477   with an unseeded PRNG. The conditions are:
6478   - The client is on a platform where the PRNG has not been seeded
6479   automatically, and the user has not seeded manually
6480   - A protocol specific client method version has been used (i.e. not
6481   SSL_client_methodv23)
6482   - A ciphersuite is used that does not require additional random data from
6483   the PRNG beyond the initial ClientHello client random (e.g. PSK-RC4-SHA).
6484
6485   If the handshake succeeds then the client random that has been used will
6486   have been generated from a PRNG with insufficient entropy and therefore the
6487   output may be predictable.
6488
6489   For example using the following command with an unseeded openssl will
6490   succeed on an unpatched platform:
6491
6492   openssl s_client -psk 1a2b3c4d -tls1_2 -cipher PSK-RC4-SHA
6493   ([CVE-2015-0285])
6494
6495   *Matt Caswell*
6496
6497 * Use After Free following d2i_ECPrivatekey error fix
6498
6499   A malformed EC private key file consumed via the d2i_ECPrivateKey function
6500   could cause a use after free condition. This, in turn, could cause a double
6501   free in several private key parsing functions (such as d2i_PrivateKey
6502   or EVP_PKCS82PKEY) and could lead to a DoS attack or memory corruption
6503   for applications that receive EC private keys from untrusted
6504   sources. This scenario is considered rare.
6505
6506   This issue was discovered by the BoringSSL project and fixed in their
6507   commit 517073cd4b.
6508   ([CVE-2015-0209])
6509
6510   *Matt Caswell*
6511
6512 * X509_to_X509_REQ NULL pointer deref fix
6513
6514   The function X509_to_X509_REQ will crash with a NULL pointer dereference if
6515   the certificate key is invalid. This function is rarely used in practice.
6516
6517   This issue was discovered by Brian Carpenter.
6518   ([CVE-2015-0288])
6519
6520   *Stephen Henson*
6521
6522 * Removed the export ciphers from the DEFAULT ciphers
6523
6524   *Kurt Roeckx*
6525
6526### Changes between 1.0.1l and 1.0.2 [22 Jan 2015]
6527
6528 * Facilitate "universal" ARM builds targeting range of ARM ISAs, e.g.
6529   ARMv5 through ARMv8, as opposite to "locking" it to single one.
6530   So far those who have to target multiple platforms would compromise
6531   and argue that binary targeting say ARMv5 would still execute on
6532   ARMv8. "Universal" build resolves this compromise by providing
6533   near-optimal performance even on newer platforms.
6534
6535   *Andy Polyakov*
6536
6537 * Accelerated NIST P-256 elliptic curve implementation for x86_64
6538   (other platforms pending).
6539
6540   *Shay Gueron & Vlad Krasnov (Intel Corp), Andy Polyakov*
6541
6542 * Add support for the SignedCertificateTimestampList certificate and
6543   OCSP response extensions from RFC6962.
6544
6545   *Rob Stradling*
6546
6547 * Fix ec_GFp_simple_points_make_affine (thus, EC_POINTs_mul etc.)
6548   for corner cases. (Certain input points at infinity could lead to
6549   bogus results, with non-infinity inputs mapped to infinity too.)
6550
6551   *Bodo Moeller*
6552
6553 * Initial support for PowerISA 2.0.7, first implemented in POWER8.
6554   This covers AES, SHA256/512 and GHASH. "Initial" means that most
6555   common cases are optimized and there still is room for further
6556   improvements. Vector Permutation AES for Altivec is also added.
6557
6558   *Andy Polyakov*
6559
6560 * Add support for little-endian ppc64 Linux target.
6561
6562   *Marcelo Cerri (IBM)*
6563
6564 * Initial support for AMRv8 ISA crypto extensions. This covers AES,
6565   SHA1, SHA256 and GHASH. "Initial" means that most common cases
6566   are optimized and there still is room for further improvements.
6567   Both 32- and 64-bit modes are supported.
6568
6569   *Andy Polyakov, Ard Biesheuvel (Linaro)*
6570
6571 * Improved ARMv7 NEON support.
6572
6573   *Andy Polyakov*
6574
6575 * Support for SPARC Architecture 2011 crypto extensions, first
6576   implemented in SPARC T4. This covers AES, DES, Camellia, SHA1,
6577   SHA256/512, MD5, GHASH and modular exponentiation.
6578
6579   *Andy Polyakov, David Miller*
6580
6581 * Accelerated modular exponentiation for Intel processors, a.k.a.
6582   RSAZ.
6583
6584   *Shay Gueron & Vlad Krasnov (Intel Corp)*
6585
6586 * Support for new and upcoming Intel processors, including AVX2,
6587   BMI and SHA ISA extensions. This includes additional "stitched"
6588   implementations, AESNI-SHA256 and GCM, and multi-buffer support
6589   for TLS encrypt.
6590
6591   This work was sponsored by Intel Corp.
6592
6593   *Andy Polyakov*
6594
6595 * Support for DTLS 1.2. This adds two sets of DTLS methods: DTLS_*_method()
6596   supports both DTLS 1.2 and 1.0 and should use whatever version the peer
6597   supports and DTLSv1_2_*_method() which supports DTLS 1.2 only.
6598
6599   *Steve Henson*
6600
6601 * Use algorithm specific chains in SSL_CTX_use_certificate_chain_file():
6602   this fixes a limitation in previous versions of OpenSSL.
6603
6604   *Steve Henson*
6605
6606 * Extended RSA OAEP support via EVP_PKEY API. Options to specify digest,
6607   MGF1 digest and OAEP label.
6608
6609   *Steve Henson*
6610
6611 * Add EVP support for key wrapping algorithms, to avoid problems with
6612   existing code the flag EVP_CIPHER_CTX_WRAP_ALLOW has to be set in
6613   the EVP_CIPHER_CTX or an error is returned. Add AES and DES3 wrap
6614   algorithms and include tests cases.
6615
6616   *Steve Henson*
6617
6618 * Add functions to allocate and set the fields of an ECDSA_METHOD
6619   structure.
6620
6621   *Douglas E. Engert, Steve Henson*
6622
6623 * New functions OPENSSL_gmtime_diff and ASN1_TIME_diff to find the
6624   difference in days and seconds between two tm or ASN1_TIME structures.
6625
6626   *Steve Henson*
6627
6628 * Add -rev test option to s_server to just reverse order of characters
6629   received by client and send back to server. Also prints an abbreviated
6630   summary of the connection parameters.
6631
6632   *Steve Henson*
6633
6634 * New option -brief for s_client and s_server to print out a brief summary
6635   of connection parameters.
6636
6637   *Steve Henson*
6638
6639 * Add callbacks for arbitrary TLS extensions.
6640
6641   *Trevor Perrin <trevp@trevp.net> and Ben Laurie*
6642
6643 * New option -crl_download in several openssl utilities to download CRLs
6644   from CRLDP extension in certificates.
6645
6646   *Steve Henson*
6647
6648 * New options -CRL and -CRLform for s_client and s_server for CRLs.
6649
6650   *Steve Henson*
6651
6652 * New function X509_CRL_diff to generate a delta CRL from the difference
6653   of two full CRLs. Add support to "crl" utility.
6654
6655   *Steve Henson*
6656
6657 * New functions to set lookup_crls function and to retrieve
6658   X509_STORE from X509_STORE_CTX.
6659
6660   *Steve Henson*
6661
6662 * Print out deprecated issuer and subject unique ID fields in
6663   certificates.
6664
6665   *Steve Henson*
6666
6667 * Extend OCSP I/O functions so they can be used for simple general purpose
6668   HTTP as well as OCSP. New wrapper function which can be used to download
6669   CRLs using the OCSP API.
6670
6671   *Steve Henson*
6672
6673 * Delegate command line handling in s_client/s_server to SSL_CONF APIs.
6674
6675   *Steve Henson*
6676
6677 * `SSL_CONF*` functions. These provide a common framework for application
6678   configuration using configuration files or command lines.
6679
6680   *Steve Henson*
6681
6682 * SSL/TLS tracing code. This parses out SSL/TLS records using the
6683   message callback and prints the results. Needs compile time option
6684   "enable-ssl-trace". New options to s_client and s_server to enable
6685   tracing.
6686
6687   *Steve Henson*
6688
6689 * New ctrl and macro to retrieve supported points extensions.
6690   Print out extension in s_server and s_client.
6691
6692   *Steve Henson*
6693
6694 * New functions to retrieve certificate signature and signature
6695   OID NID.
6696
6697   *Steve Henson*
6698
6699 * Add functions to retrieve and manipulate the raw cipherlist sent by a
6700   client to OpenSSL.
6701
6702   *Steve Henson*
6703
6704 * New Suite B modes for TLS code. These use and enforce the requirements
6705   of RFC6460: restrict ciphersuites, only permit Suite B algorithms and
6706   only use Suite B curves. The Suite B modes can be set by using the
6707   strings "SUITEB128", "SUITEB192" or "SUITEB128ONLY" for the cipherstring.
6708
6709   *Steve Henson*
6710
6711 * New chain verification flags for Suite B levels of security. Check
6712   algorithms are acceptable when flags are set in X509_verify_cert.
6713
6714   *Steve Henson*
6715
6716 * Make tls1_check_chain return a set of flags indicating checks passed
6717   by a certificate chain. Add additional tests to handle client
6718   certificates: checks for matching certificate type and issuer name
6719   comparison.
6720
6721   *Steve Henson*
6722
6723 * If an attempt is made to use a signature algorithm not in the peer
6724   preference list abort the handshake. If client has no suitable
6725   signature algorithms in response to a certificate request do not
6726   use the certificate.
6727
6728   *Steve Henson*
6729
6730 * If server EC tmp key is not in client preference list abort handshake.
6731
6732   *Steve Henson*
6733
6734 * Add support for certificate stores in CERT structure. This makes it
6735   possible to have different stores per SSL structure or one store in
6736   the parent SSL_CTX. Include distinct stores for certificate chain
6737   verification and chain building. New ctrl SSL_CTRL_BUILD_CERT_CHAIN
6738   to build and store a certificate chain in CERT structure: returning
6739   an error if the chain cannot be built: this will allow applications
6740   to test if a chain is correctly configured.
6741
6742   Note: if the CERT based stores are not set then the parent SSL_CTX
6743   store is used to retain compatibility with existing behaviour.
6744
6745   *Steve Henson*
6746
6747 * New function ssl_set_client_disabled to set a ciphersuite disabled
6748   mask based on the current session, check mask when sending client
6749   hello and checking the requested ciphersuite.
6750
6751   *Steve Henson*
6752
6753 * New ctrls to retrieve and set certificate types in a certificate
6754   request message. Print out received values in s_client. If certificate
6755   types is not set with custom values set sensible values based on
6756   supported signature algorithms.
6757
6758   *Steve Henson*
6759
6760 * Support for distinct client and server supported signature algorithms.
6761
6762   *Steve Henson*
6763
6764 * Add certificate callback. If set this is called whenever a certificate
6765   is required by client or server. An application can decide which
6766   certificate chain to present based on arbitrary criteria: for example
6767   supported signature algorithms. Add very simple example to s_server.
6768   This fixes many of the problems and restrictions of the existing client
6769   certificate callback: for example you can now clear an existing
6770   certificate and specify the whole chain.
6771
6772   *Steve Henson*
6773
6774 * Add new "valid_flags" field to CERT_PKEY structure which determines what
6775   the certificate can be used for (if anything). Set valid_flags field
6776   in new tls1_check_chain function. Simplify ssl_set_cert_masks which used
6777   to have similar checks in it.
6778
6779   Add new "cert_flags" field to CERT structure and include a "strict mode".
6780   This enforces some TLS certificate requirements (such as only permitting
6781   certificate signature algorithms contained in the supported algorithms
6782   extension) which some implementations ignore: this option should be used
6783   with caution as it could cause interoperability issues.
6784
6785   *Steve Henson*
6786
6787 * Update and tidy signature algorithm extension processing. Work out
6788   shared signature algorithms based on preferences and peer algorithms
6789   and print them out in s_client and s_server. Abort handshake if no
6790   shared signature algorithms.
6791
6792   *Steve Henson*
6793
6794 * Add new functions to allow customised supported signature algorithms
6795   for SSL and SSL_CTX structures. Add options to s_client and s_server
6796   to support them.
6797
6798   *Steve Henson*
6799
6800 * New function SSL_certs_clear() to delete all references to certificates
6801   from an SSL structure. Before this once a certificate had been added
6802   it couldn't be removed.
6803
6804   *Steve Henson*
6805
6806 * Integrate hostname, email address and IP address checking with certificate
6807   verification. New verify options supporting checking in openssl utility.
6808
6809   *Steve Henson*
6810
6811 * Fixes and wildcard matching support to hostname and email checking
6812   functions. Add manual page.
6813
6814   *Florian Weimer (Red Hat Product Security Team)*
6815
6816 * New functions to check a hostname email or IP address against a
6817   certificate. Add options x509 utility to print results of checks against
6818   a certificate.
6819
6820   *Steve Henson*
6821
6822 * Fix OCSP checking.
6823
6824   *Rob Stradling <rob.stradling@comodo.com> and Ben Laurie*
6825
6826 * Initial experimental support for explicitly trusted non-root CAs.
6827   OpenSSL still tries to build a complete chain to a root but if an
6828   intermediate CA has a trust setting included that is used. The first
6829   setting is used: whether to trust (e.g., -addtrust option to the x509
6830   utility) or reject.
6831
6832   *Steve Henson*
6833
6834 * Add -trusted_first option which attempts to find certificates in the
6835   trusted store even if an untrusted chain is also supplied.
6836
6837   *Steve Henson*
6838
6839 * MIPS assembly pack updates: support for MIPS32r2 and SmartMIPS ASE,
6840   platform support for Linux and Android.
6841
6842   *Andy Polyakov*
6843
6844 * Support for linux-x32, ILP32 environment in x86_64 framework.
6845
6846   *Andy Polyakov*
6847
6848 * Experimental multi-implementation support for FIPS capable OpenSSL.
6849   When in FIPS mode the approved implementations are used as normal,
6850   when not in FIPS mode the internal unapproved versions are used instead.
6851   This means that the FIPS capable OpenSSL isn't forced to use the
6852   (often lower performance) FIPS implementations outside FIPS mode.
6853
6854   *Steve Henson*
6855
6856 * Transparently support X9.42 DH parameters when calling
6857   PEM_read_bio_DHparameters. This means existing applications can handle
6858   the new parameter format automatically.
6859
6860   *Steve Henson*
6861
6862 * Initial experimental support for X9.42 DH parameter format: mainly
6863   to support use of 'q' parameter for RFC5114 parameters.
6864
6865   *Steve Henson*
6866
6867 * Add DH parameters from RFC5114 including test data to dhtest.
6868
6869   *Steve Henson*
6870
6871 * Support for automatic EC temporary key parameter selection. If enabled
6872   the most preferred EC parameters are automatically used instead of
6873   hardcoded fixed parameters. Now a server just has to call:
6874   SSL_CTX_set_ecdh_auto(ctx, 1) and the server will automatically
6875   support ECDH and use the most appropriate parameters.
6876
6877   *Steve Henson*
6878
6879 * Enhance and tidy EC curve and point format TLS extension code. Use
6880   static structures instead of allocation if default values are used.
6881   New ctrls to set curves we wish to support and to retrieve shared curves.
6882   Print out shared curves in s_server. New options to s_server and s_client
6883   to set list of supported curves.
6884
6885   *Steve Henson*
6886
6887 * New ctrls to retrieve supported signature algorithms and
6888   supported curve values as an array of NIDs. Extend openssl utility
6889   to print out received values.
6890
6891   *Steve Henson*
6892
6893 * Add new APIs EC_curve_nist2nid and EC_curve_nid2nist which convert
6894   between NIDs and the more common NIST names such as "P-256". Enhance
6895   ecparam utility and ECC method to recognise the NIST names for curves.
6896
6897   *Steve Henson*
6898
6899 * Enhance SSL/TLS certificate chain handling to support different
6900   chains for each certificate instead of one chain in the parent SSL_CTX.
6901
6902   *Steve Henson*
6903
6904 * Support for fixed DH ciphersuite client authentication: where both
6905   server and client use DH certificates with common parameters.
6906
6907   *Steve Henson*
6908
6909 * Support for fixed DH ciphersuites: those requiring DH server
6910   certificates.
6911
6912   *Steve Henson*
6913
6914 * New function i2d_re_X509_tbs for re-encoding the TBS portion of
6915   the certificate.
6916   Note: Related 1.0.2-beta specific macros X509_get_cert_info,
6917   X509_CINF_set_modified, X509_CINF_get_issuer, X509_CINF_get_extensions and
6918   X509_CINF_get_signature were reverted post internal team review.
6919
6920OpenSSL 1.0.1
6921-------------
6922
6923### Changes between 1.0.1t and 1.0.1u [22 Sep 2016]
6924
6925 * OCSP Status Request extension unbounded memory growth
6926
6927   A malicious client can send an excessively large OCSP Status Request
6928   extension. If that client continually requests renegotiation, sending a
6929   large OCSP Status Request extension each time, then there will be unbounded
6930   memory growth on the server. This will eventually lead to a Denial Of
6931   Service attack through memory exhaustion. Servers with a default
6932   configuration are vulnerable even if they do not support OCSP. Builds using
6933   the "no-ocsp" build time option are not affected.
6934
6935   This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.)
6936   ([CVE-2016-6304])
6937
6938   *Matt Caswell*
6939
6940 * In order to mitigate the SWEET32 attack, the DES ciphers were moved from
6941   HIGH to MEDIUM.
6942
6943   This issue was reported to OpenSSL Karthikeyan Bhargavan and Gaetan
6944   Leurent (INRIA)
6945   ([CVE-2016-2183])
6946
6947   *Rich Salz*
6948
6949 * OOB write in MDC2_Update()
6950
6951   An overflow can occur in MDC2_Update() either if called directly or
6952   through the EVP_DigestUpdate() function using MDC2. If an attacker
6953   is able to supply very large amounts of input data after a previous
6954   call to EVP_EncryptUpdate() with a partial block then a length check
6955   can overflow resulting in a heap corruption.
6956
6957   The amount of data needed is comparable to SIZE_MAX which is impractical
6958   on most platforms.
6959
6960   This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.)
6961   ([CVE-2016-6303])
6962
6963   *Stephen Henson*
6964
6965 * Malformed SHA512 ticket DoS
6966
6967   If a server uses SHA512 for TLS session ticket HMAC it is vulnerable to a
6968   DoS attack where a malformed ticket will result in an OOB read which will
6969   ultimately crash.
6970
6971   The use of SHA512 in TLS session tickets is comparatively rare as it requires
6972   a custom server callback and ticket lookup mechanism.
6973
6974   This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.)
6975   ([CVE-2016-6302])
6976
6977   *Stephen Henson*
6978
6979 * OOB write in BN_bn2dec()
6980
6981   The function BN_bn2dec() does not check the return value of BN_div_word().
6982   This can cause an OOB write if an application uses this function with an
6983   overly large BIGNUM. This could be a problem if an overly large certificate
6984   or CRL is printed out from an untrusted source. TLS is not affected because
6985   record limits will reject an oversized certificate before it is parsed.
6986
6987   This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.)
6988   ([CVE-2016-2182])
6989
6990   *Stephen Henson*
6991
6992 * OOB read in TS_OBJ_print_bio()
6993
6994   The function TS_OBJ_print_bio() misuses OBJ_obj2txt(): the return value is
6995   the total length the OID text representation would use and not the amount
6996   of data written. This will result in OOB reads when large OIDs are
6997   presented.
6998
6999   This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.)
7000   ([CVE-2016-2180])
7001
7002   *Stephen Henson*
7003
7004 * Pointer arithmetic undefined behaviour
7005
7006   Avoid some undefined pointer arithmetic
7007
7008   A common idiom in the codebase is to check limits in the following manner:
7009   "p + len > limit"
7010
7011   Where "p" points to some malloc'd data of SIZE bytes and
7012   limit == p + SIZE
7013
7014   "len" here could be from some externally supplied data (e.g. from a TLS
7015   message).
7016
7017   The rules of C pointer arithmetic are such that "p + len" is only well
7018   defined where len <= SIZE. Therefore the above idiom is actually
7019   undefined behaviour.
7020
7021   For example this could cause problems if some malloc implementation
7022   provides an address for "p" such that "p + len" actually overflows for
7023   values of len that are too big and therefore p + len < limit.
7024
7025   This issue was reported to OpenSSL by Guido Vranken
7026   ([CVE-2016-2177])
7027
7028   *Matt Caswell*
7029
7030 * Constant time flag not preserved in DSA signing
7031
7032   Operations in the DSA signing algorithm should run in constant time in
7033   order to avoid side channel attacks. A flaw in the OpenSSL DSA
7034   implementation means that a non-constant time codepath is followed for
7035   certain operations. This has been demonstrated through a cache-timing
7036   attack to be sufficient for an attacker to recover the private DSA key.
7037
7038   This issue was reported by César Pereida (Aalto University), Billy Brumley
7039   (Tampere University of Technology), and Yuval Yarom (The University of
7040   Adelaide and NICTA).
7041   ([CVE-2016-2178])
7042
7043   *César Pereida*
7044
7045 * DTLS buffered message DoS
7046
7047   In a DTLS connection where handshake messages are delivered out-of-order
7048   those messages that OpenSSL is not yet ready to process will be buffered
7049   for later use. Under certain circumstances, a flaw in the logic means that
7050   those messages do not get removed from the buffer even though the handshake
7051   has been completed. An attacker could force up to approx. 15 messages to
7052   remain in the buffer when they are no longer required. These messages will
7053   be cleared when the DTLS connection is closed. The default maximum size for
7054   a message is 100k. Therefore the attacker could force an additional 1500k
7055   to be consumed per connection. By opening many simulataneous connections an
7056   attacker could cause a DoS attack through memory exhaustion.
7057
7058   This issue was reported to OpenSSL by Quan Luo.
7059   ([CVE-2016-2179])
7060
7061   *Matt Caswell*
7062
7063 * DTLS replay protection DoS
7064
7065   A flaw in the DTLS replay attack protection mechanism means that records
7066   that arrive for future epochs update the replay protection "window" before
7067   the MAC for the record has been validated. This could be exploited by an
7068   attacker by sending a record for the next epoch (which does not have to
7069   decrypt or have a valid MAC), with a very large sequence number. This means
7070   that all subsequent legitimate packets are dropped causing a denial of
7071   service for a specific DTLS connection.
7072
7073   This issue was reported to OpenSSL by the OCAP audit team.
7074   ([CVE-2016-2181])
7075
7076   *Matt Caswell*
7077
7078 * Certificate message OOB reads
7079
7080   In OpenSSL 1.0.2 and earlier some missing message length checks can result
7081   in OOB reads of up to 2 bytes beyond an allocated buffer. There is a
7082   theoretical DoS risk but this has not been observed in practice on common
7083   platforms.
7084
7085   The messages affected are client certificate, client certificate request
7086   and server certificate. As a result the attack can only be performed
7087   against a client or a server which enables client authentication.
7088
7089   This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.)
7090   ([CVE-2016-6306])
7091
7092   *Stephen Henson*
7093
7094### Changes between 1.0.1s and 1.0.1t [3 May 2016]
7095
7096 * Prevent padding oracle in AES-NI CBC MAC check
7097
7098   A MITM attacker can use a padding oracle attack to decrypt traffic
7099   when the connection uses an AES CBC cipher and the server support
7100   AES-NI.
7101
7102   This issue was introduced as part of the fix for Lucky 13 padding
7103   attack ([CVE-2013-0169]). The padding check was rewritten to be in
7104   constant time by making sure that always the same bytes are read and
7105   compared against either the MAC or padding bytes. But it no longer
7106   checked that there was enough data to have both the MAC and padding
7107   bytes.
7108
7109   This issue was reported by Juraj Somorovsky using TLS-Attacker.
7110   ([CVE-2016-2107])
7111
7112   *Kurt Roeckx*
7113
7114 * Fix EVP_EncodeUpdate overflow
7115
7116   An overflow can occur in the EVP_EncodeUpdate() function which is used for
7117   Base64 encoding of binary data. If an attacker is able to supply very large
7118   amounts of input data then a length check can overflow resulting in a heap
7119   corruption.
7120
7121   Internally to OpenSSL the EVP_EncodeUpdate() function is primarly used by
7122   the `PEM_write_bio*` family of functions. These are mainly used within the
7123   OpenSSL command line applications, so any application which processes data
7124   from an untrusted source and outputs it as a PEM file should be considered
7125   vulnerable to this issue. User applications that call these APIs directly
7126   with large amounts of untrusted data may also be vulnerable.
7127
7128   This issue was reported by Guido Vranken.
7129   ([CVE-2016-2105])
7130
7131   *Matt Caswell*
7132
7133 * Fix EVP_EncryptUpdate overflow
7134
7135   An overflow can occur in the EVP_EncryptUpdate() function. If an attacker
7136   is able to supply very large amounts of input data after a previous call to
7137   EVP_EncryptUpdate() with a partial block then a length check can overflow
7138   resulting in a heap corruption. Following an analysis of all OpenSSL
7139   internal usage of the EVP_EncryptUpdate() function all usage is one of two
7140   forms. The first form is where the EVP_EncryptUpdate() call is known to be
7141   the first called function after an EVP_EncryptInit(), and therefore that
7142   specific call must be safe. The second form is where the length passed to
7143   EVP_EncryptUpdate() can be seen from the code to be some small value and
7144   therefore there is no possibility of an overflow. Since all instances are
7145   one of these two forms, it is believed that there can be no overflows in
7146   internal code due to this problem. It should be noted that
7147   EVP_DecryptUpdate() can call EVP_EncryptUpdate() in certain code paths.
7148   Also EVP_CipherUpdate() is a synonym for EVP_EncryptUpdate(). All instances
7149   of these calls have also been analysed too and it is believed there are no
7150   instances in internal usage where an overflow could occur.
7151
7152   This issue was reported by Guido Vranken.
7153   ([CVE-2016-2106])
7154
7155   *Matt Caswell*
7156
7157 * Prevent ASN.1 BIO excessive memory allocation
7158
7159   When ASN.1 data is read from a BIO using functions such as d2i_CMS_bio()
7160   a short invalid encoding can casuse allocation of large amounts of memory
7161   potentially consuming excessive resources or exhausting memory.
7162
7163   Any application parsing untrusted data through d2i BIO functions is
7164   affected. The memory based functions such as d2i_X509() are *not* affected.
7165   Since the memory based functions are used by the TLS library, TLS
7166   applications are not affected.
7167
7168   This issue was reported by Brian Carpenter.
7169   ([CVE-2016-2109])
7170
7171   *Stephen Henson*
7172
7173 * EBCDIC overread
7174
7175   ASN1 Strings that are over 1024 bytes can cause an overread in applications
7176   using the X509_NAME_oneline() function on EBCDIC systems. This could result
7177   in arbitrary stack data being returned in the buffer.
7178
7179   This issue was reported by Guido Vranken.
7180   ([CVE-2016-2176])
7181
7182   *Matt Caswell*
7183
7184 * Modify behavior of ALPN to invoke callback after SNI/servername
7185   callback, such that updates to the SSL_CTX affect ALPN.
7186
7187   *Todd Short*
7188
7189 * Remove LOW from the DEFAULT cipher list.  This removes singles DES from the
7190   default.
7191
7192   *Kurt Roeckx*
7193
7194 * Only remove the SSLv2 methods with the no-ssl2-method option. When the
7195   methods are enabled and ssl2 is disabled the methods return NULL.
7196
7197   *Kurt Roeckx*
7198
7199### Changes between 1.0.1r and 1.0.1s [1 Mar 2016]
7200
7201* Disable weak ciphers in SSLv3 and up in default builds of OpenSSL.
7202  Builds that are not configured with "enable-weak-ssl-ciphers" will not
7203  provide any "EXPORT" or "LOW" strength ciphers.
7204
7205  *Viktor Dukhovni*
7206
7207* Disable SSLv2 default build, default negotiation and weak ciphers.  SSLv2
7208  is by default disabled at build-time.  Builds that are not configured with
7209  "enable-ssl2" will not support SSLv2.  Even if "enable-ssl2" is used,
7210  users who want to negotiate SSLv2 via the version-flexible SSLv23_method()
7211  will need to explicitly call either of:
7212
7213      SSL_CTX_clear_options(ctx, SSL_OP_NO_SSLv2);
7214  or
7215      SSL_clear_options(ssl, SSL_OP_NO_SSLv2);
7216
7217  as appropriate.  Even if either of those is used, or the application
7218  explicitly uses the version-specific SSLv2_method() or its client and
7219  server variants, SSLv2 ciphers vulnerable to exhaustive search key
7220  recovery have been removed.  Specifically, the SSLv2 40-bit EXPORT
7221  ciphers, and SSLv2 56-bit DES are no longer available.
7222  ([CVE-2016-0800])
7223
7224  *Viktor Dukhovni*
7225
7226 * Fix a double-free in DSA code
7227
7228   A double free bug was discovered when OpenSSL parses malformed DSA private
7229   keys and could lead to a DoS attack or memory corruption for applications
7230   that receive DSA private keys from untrusted sources.  This scenario is
7231   considered rare.
7232
7233   This issue was reported to OpenSSL by Adam Langley(Google/BoringSSL) using
7234   libFuzzer.
7235   ([CVE-2016-0705])
7236
7237   *Stephen Henson*
7238
7239 * Disable SRP fake user seed to address a server memory leak.
7240
7241   Add a new method SRP_VBASE_get1_by_user that handles the seed properly.
7242
7243   SRP_VBASE_get_by_user had inconsistent memory management behaviour.
7244   In order to fix an unavoidable memory leak, SRP_VBASE_get_by_user
7245   was changed to ignore the "fake user" SRP seed, even if the seed
7246   is configured.
7247
7248   Users should use SRP_VBASE_get1_by_user instead. Note that in
7249   SRP_VBASE_get1_by_user, caller must free the returned value. Note
7250   also that even though configuring the SRP seed attempts to hide
7251   invalid usernames by continuing the handshake with fake
7252   credentials, this behaviour is not constant time and no strong
7253   guarantees are made that the handshake is indistinguishable from
7254   that of a valid user.
7255   ([CVE-2016-0798])
7256
7257   *Emilia Käsper*
7258
7259 * Fix BN_hex2bn/BN_dec2bn NULL pointer deref/heap corruption
7260
7261   In the BN_hex2bn function the number of hex digits is calculated using an
7262   int value `i`. Later `bn_expand` is called with a value of `i * 4`. For
7263   large values of `i` this can result in `bn_expand` not allocating any
7264   memory because `i * 4` is negative. This can leave the internal BIGNUM data
7265   field as NULL leading to a subsequent NULL ptr deref. For very large values
7266   of `i`, the calculation `i * 4` could be a positive value smaller than `i`.
7267   In this case memory is allocated to the internal BIGNUM data field, but it
7268   is insufficiently sized leading to heap corruption. A similar issue exists
7269   in BN_dec2bn. This could have security consequences if BN_hex2bn/BN_dec2bn
7270   is ever called by user applications with very large untrusted hex/dec data.
7271   This is anticipated to be a rare occurrence.
7272
7273   All OpenSSL internal usage of these functions use data that is not expected
7274   to be untrusted, e.g. config file data or application command line
7275   arguments. If user developed applications generate config file data based
7276   on untrusted data then it is possible that this could also lead to security
7277   consequences. This is also anticipated to be rare.
7278
7279   This issue was reported to OpenSSL by Guido Vranken.
7280   ([CVE-2016-0797])
7281
7282   *Matt Caswell*
7283
7284 * Fix memory issues in `BIO_*printf` functions
7285
7286   The internal `fmtstr` function used in processing a "%s" format string in
7287   the `BIO_*printf` functions could overflow while calculating the length of a
7288   string and cause an OOB read when printing very long strings.
7289
7290   Additionally the internal `doapr_outch` function can attempt to write to an
7291   OOB memory location (at an offset from the NULL pointer) in the event of a
7292   memory allocation failure. In 1.0.2 and below this could be caused where
7293   the size of a buffer to be allocated is greater than INT_MAX. E.g. this
7294   could be in processing a very long "%s" format string. Memory leaks can
7295   also occur.
7296
7297   The first issue may mask the second issue dependent on compiler behaviour.
7298   These problems could enable attacks where large amounts of untrusted data
7299   is passed to the `BIO_*printf` functions. If applications use these functions
7300   in this way then they could be vulnerable. OpenSSL itself uses these
7301   functions when printing out human-readable dumps of ASN.1 data. Therefore
7302   applications that print this data could be vulnerable if the data is from
7303   untrusted sources. OpenSSL command line applications could also be
7304   vulnerable where they print out ASN.1 data, or if untrusted data is passed
7305   as command line arguments.
7306
7307   Libssl is not considered directly vulnerable. Additionally certificates etc
7308   received via remote connections via libssl are also unlikely to be able to
7309   trigger these issues because of message size limits enforced within libssl.
7310
7311   This issue was reported to OpenSSL Guido Vranken.
7312   ([CVE-2016-0799])
7313
7314   *Matt Caswell*
7315
7316 * Side channel attack on modular exponentiation
7317
7318   A side-channel attack was found which makes use of cache-bank conflicts on
7319   the Intel Sandy-Bridge microarchitecture which could lead to the recovery
7320   of RSA keys.  The ability to exploit this issue is limited as it relies on
7321   an attacker who has control of code in a thread running on the same
7322   hyper-threaded core as the victim thread which is performing decryptions.
7323
7324   This issue was reported to OpenSSL by Yuval Yarom, The University of
7325   Adelaide and NICTA, Daniel Genkin, Technion and Tel Aviv University, and
7326   Nadia Heninger, University of Pennsylvania with more information at
7327   <http://cachebleed.info>.
7328   ([CVE-2016-0702])
7329
7330   *Andy Polyakov*
7331
7332 * Change the req command to generate a 2048-bit RSA/DSA key by default,
7333   if no keysize is specified with default_bits. This fixes an
7334   omission in an earlier change that changed all RSA/DSA key generation
7335   commands to use 2048 bits by default.
7336
7337   *Emilia Käsper*
7338
7339### Changes between 1.0.1q and 1.0.1r [28 Jan 2016]
7340
7341 * Protection for DH small subgroup attacks
7342
7343   As a precautionary measure the SSL_OP_SINGLE_DH_USE option has been
7344   switched on by default and cannot be disabled. This could have some
7345   performance impact.
7346
7347   *Matt Caswell*
7348
7349 * SSLv2 doesn't block disabled ciphers
7350
7351   A malicious client can negotiate SSLv2 ciphers that have been disabled on
7352   the server and complete SSLv2 handshakes even if all SSLv2 ciphers have
7353   been disabled, provided that the SSLv2 protocol was not also disabled via
7354   SSL_OP_NO_SSLv2.
7355
7356   This issue was reported to OpenSSL on 26th December 2015 by Nimrod Aviram
7357   and Sebastian Schinzel.
7358   ([CVE-2015-3197])
7359
7360   *Viktor Dukhovni*
7361
7362 * Reject DH handshakes with parameters shorter than 1024 bits.
7363
7364   *Kurt Roeckx*
7365
7366### Changes between 1.0.1p and 1.0.1q [3 Dec 2015]
7367
7368 * Certificate verify crash with missing PSS parameter
7369
7370   The signature verification routines will crash with a NULL pointer
7371   dereference if presented with an ASN.1 signature using the RSA PSS
7372   algorithm and absent mask generation function parameter. Since these
7373   routines are used to verify certificate signature algorithms this can be
7374   used to crash any certificate verification operation and exploited in a
7375   DoS attack. Any application which performs certificate verification is
7376   vulnerable including OpenSSL clients and servers which enable client
7377   authentication.
7378
7379   This issue was reported to OpenSSL by Loïc Jonas Etienne (Qnective AG).
7380   ([CVE-2015-3194])
7381
7382   *Stephen Henson*
7383
7384 * X509_ATTRIBUTE memory leak
7385
7386   When presented with a malformed X509_ATTRIBUTE structure OpenSSL will leak
7387   memory. This structure is used by the PKCS#7 and CMS routines so any
7388   application which reads PKCS#7 or CMS data from untrusted sources is
7389   affected. SSL/TLS is not affected.
7390
7391   This issue was reported to OpenSSL by Adam Langley (Google/BoringSSL) using
7392   libFuzzer.
7393   ([CVE-2015-3195])
7394
7395   *Stephen Henson*
7396
7397 * Rewrite EVP_DecodeUpdate (base64 decoding) to fix several bugs.
7398   This changes the decoding behaviour for some invalid messages,
7399   though the change is mostly in the more lenient direction, and
7400   legacy behaviour is preserved as much as possible.
7401
7402   *Emilia Käsper*
7403
7404 * In DSA_generate_parameters_ex, if the provided seed is too short,
7405   use a random seed, as already documented.
7406
7407   *Rich Salz and Ismo Puustinen <ismo.puustinen@intel.com>*
7408
7409### Changes between 1.0.1o and 1.0.1p [9 Jul 2015]
7410
7411 * Alternate chains certificate forgery
7412
7413   During certificate verfification, OpenSSL will attempt to find an
7414   alternative certificate chain if the first attempt to build such a chain
7415   fails. An error in the implementation of this logic can mean that an
7416   attacker could cause certain checks on untrusted certificates to be
7417   bypassed, such as the CA flag, enabling them to use a valid leaf
7418   certificate to act as a CA and "issue" an invalid certificate.
7419
7420   This issue was reported to OpenSSL by Adam Langley/David Benjamin
7421   (Google/BoringSSL).
7422   ([CVE-2015-1793])
7423
7424   *Matt Caswell*
7425
7426 * Race condition handling PSK identify hint
7427
7428   If PSK identity hints are received by a multi-threaded client then
7429   the values are wrongly updated in the parent SSL_CTX structure. This can
7430   result in a race condition potentially leading to a double free of the
7431   identify hint data.
7432   ([CVE-2015-3196])
7433
7434   *Stephen Henson*
7435
7436### Changes between 1.0.1n and 1.0.1o [12 Jun 2015]
7437
7438 * Fix HMAC ABI incompatibility. The previous version introduced an ABI
7439   incompatibility in the handling of HMAC. The previous ABI has now been
7440   restored.
7441
7442### Changes between 1.0.1m and 1.0.1n [11 Jun 2015]
7443
7444 * Malformed ECParameters causes infinite loop
7445
7446   When processing an ECParameters structure OpenSSL enters an infinite loop
7447   if the curve specified is over a specially malformed binary polynomial
7448   field.
7449
7450   This can be used to perform denial of service against any
7451   system which processes public keys, certificate requests or
7452   certificates.  This includes TLS clients and TLS servers with
7453   client authentication enabled.
7454
7455   This issue was reported to OpenSSL by Joseph Barr-Pixton.
7456   ([CVE-2015-1788])
7457
7458   *Andy Polyakov*
7459
7460 * Exploitable out-of-bounds read in X509_cmp_time
7461
7462   X509_cmp_time does not properly check the length of the ASN1_TIME
7463   string and can read a few bytes out of bounds. In addition,
7464   X509_cmp_time accepts an arbitrary number of fractional seconds in the
7465   time string.
7466
7467   An attacker can use this to craft malformed certificates and CRLs of
7468   various sizes and potentially cause a segmentation fault, resulting in
7469   a DoS on applications that verify certificates or CRLs. TLS clients
7470   that verify CRLs are affected. TLS clients and servers with client
7471   authentication enabled may be affected if they use custom verification
7472   callbacks.
7473
7474   This issue was reported to OpenSSL by Robert Swiecki (Google), and
7475   independently by Hanno Böck.
7476   ([CVE-2015-1789])
7477
7478   *Emilia Käsper*
7479
7480 * PKCS7 crash with missing EnvelopedContent
7481
7482   The PKCS#7 parsing code does not handle missing inner EncryptedContent
7483   correctly. An attacker can craft malformed ASN.1-encoded PKCS#7 blobs
7484   with missing content and trigger a NULL pointer dereference on parsing.
7485
7486   Applications that decrypt PKCS#7 data or otherwise parse PKCS#7
7487   structures from untrusted sources are affected. OpenSSL clients and
7488   servers are not affected.
7489
7490   This issue was reported to OpenSSL by Michal Zalewski (Google).
7491   ([CVE-2015-1790])
7492
7493   *Emilia Käsper*
7494
7495 * CMS verify infinite loop with unknown hash function
7496
7497   When verifying a signedData message the CMS code can enter an infinite loop
7498   if presented with an unknown hash function OID. This can be used to perform
7499   denial of service against any system which verifies signedData messages using
7500   the CMS code.
7501   This issue was reported to OpenSSL by Johannes Bauer.
7502   ([CVE-2015-1792])
7503
7504   *Stephen Henson*
7505
7506 * Race condition handling NewSessionTicket
7507
7508   If a NewSessionTicket is received by a multi-threaded client when attempting to
7509   reuse a previous ticket then a race condition can occur potentially leading to
7510   a double free of the ticket data.
7511   ([CVE-2015-1791])
7512
7513   *Matt Caswell*
7514
7515 * Reject DH handshakes with parameters shorter than 768 bits.
7516
7517   *Kurt Roeckx and Emilia Kasper*
7518
7519 * dhparam: generate 2048-bit parameters by default.
7520
7521   *Kurt Roeckx and Emilia Kasper*
7522
7523### Changes between 1.0.1l and 1.0.1m [19 Mar 2015]
7524
7525 * Segmentation fault in ASN1_TYPE_cmp fix
7526
7527   The function ASN1_TYPE_cmp will crash with an invalid read if an attempt is
7528   made to compare ASN.1 boolean types. Since ASN1_TYPE_cmp is used to check
7529   certificate signature algorithm consistency this can be used to crash any
7530   certificate verification operation and exploited in a DoS attack. Any
7531   application which performs certificate verification is vulnerable including
7532   OpenSSL clients and servers which enable client authentication.
7533   ([CVE-2015-0286])
7534
7535   *Stephen Henson*
7536
7537 * ASN.1 structure reuse memory corruption fix
7538
7539   Reusing a structure in ASN.1 parsing may allow an attacker to cause
7540   memory corruption via an invalid write. Such reuse is and has been
7541   strongly discouraged and is believed to be rare.
7542
7543   Applications that parse structures containing CHOICE or ANY DEFINED BY
7544   components may be affected. Certificate parsing (d2i_X509 and related
7545   functions) are however not affected. OpenSSL clients and servers are
7546   not affected.
7547   ([CVE-2015-0287])
7548
7549   *Stephen Henson*
7550
7551 * PKCS7 NULL pointer dereferences fix
7552
7553   The PKCS#7 parsing code does not handle missing outer ContentInfo
7554   correctly. An attacker can craft malformed ASN.1-encoded PKCS#7 blobs with
7555   missing content and trigger a NULL pointer dereference on parsing.
7556
7557   Applications that verify PKCS#7 signatures, decrypt PKCS#7 data or
7558   otherwise parse PKCS#7 structures from untrusted sources are
7559   affected. OpenSSL clients and servers are not affected.
7560
7561   This issue was reported to OpenSSL by Michal Zalewski (Google).
7562   ([CVE-2015-0289])
7563
7564   *Emilia Käsper*
7565
7566 * DoS via reachable assert in SSLv2 servers fix
7567
7568   A malicious client can trigger an OPENSSL_assert (i.e., an abort) in
7569   servers that both support SSLv2 and enable export cipher suites by sending
7570   a specially crafted SSLv2 CLIENT-MASTER-KEY message.
7571
7572   This issue was discovered by Sean Burford (Google) and Emilia Käsper
7573   (OpenSSL development team).
7574   ([CVE-2015-0293])
7575
7576   *Emilia Käsper*
7577
7578 * Use After Free following d2i_ECPrivatekey error fix
7579
7580   A malformed EC private key file consumed via the d2i_ECPrivateKey function
7581   could cause a use after free condition. This, in turn, could cause a double
7582   free in several private key parsing functions (such as d2i_PrivateKey
7583   or EVP_PKCS82PKEY) and could lead to a DoS attack or memory corruption
7584   for applications that receive EC private keys from untrusted
7585   sources. This scenario is considered rare.
7586
7587   This issue was discovered by the BoringSSL project and fixed in their
7588   commit 517073cd4b.
7589   ([CVE-2015-0209])
7590
7591   *Matt Caswell*
7592
7593 * X509_to_X509_REQ NULL pointer deref fix
7594
7595   The function X509_to_X509_REQ will crash with a NULL pointer dereference if
7596   the certificate key is invalid. This function is rarely used in practice.
7597
7598   This issue was discovered by Brian Carpenter.
7599   ([CVE-2015-0288])
7600
7601   *Stephen Henson*
7602
7603 * Removed the export ciphers from the DEFAULT ciphers
7604
7605   *Kurt Roeckx*
7606
7607### Changes between 1.0.1k and 1.0.1l [15 Jan 2015]
7608
7609 * Build fixes for the Windows and OpenVMS platforms
7610
7611   *Matt Caswell and Richard Levitte*
7612
7613### Changes between 1.0.1j and 1.0.1k [8 Jan 2015]
7614
7615 * Fix DTLS segmentation fault in dtls1_get_record. A carefully crafted DTLS
7616   message can cause a segmentation fault in OpenSSL due to a NULL pointer
7617   dereference. This could lead to a Denial Of Service attack. Thanks to
7618   Markus Stenberg of Cisco Systems, Inc. for reporting this issue.
7619   ([CVE-2014-3571])
7620
7621   *Steve Henson*
7622
7623 * Fix DTLS memory leak in dtls1_buffer_record. A memory leak can occur in the
7624   dtls1_buffer_record function under certain conditions. In particular this
7625   could occur if an attacker sent repeated DTLS records with the same
7626   sequence number but for the next epoch. The memory leak could be exploited
7627   by an attacker in a Denial of Service attack through memory exhaustion.
7628   Thanks to Chris Mueller for reporting this issue.
7629   ([CVE-2015-0206])
7630
7631   *Matt Caswell*
7632
7633 * Fix issue where no-ssl3 configuration sets method to NULL. When openssl is
7634   built with the no-ssl3 option and a SSL v3 ClientHello is received the ssl
7635   method would be set to NULL which could later result in a NULL pointer
7636   dereference. Thanks to Frank Schmirler for reporting this issue.
7637   ([CVE-2014-3569])
7638
7639   *Kurt Roeckx*
7640
7641 * Abort handshake if server key exchange message is omitted for ephemeral
7642   ECDH ciphersuites.
7643
7644   Thanks to Karthikeyan Bhargavan of the PROSECCO team at INRIA for
7645   reporting this issue.
7646   ([CVE-2014-3572])
7647
7648   *Steve Henson*
7649
7650 * Remove non-export ephemeral RSA code on client and server. This code
7651   violated the TLS standard by allowing the use of temporary RSA keys in
7652   non-export ciphersuites and could be used by a server to effectively
7653   downgrade the RSA key length used to a value smaller than the server
7654   certificate. Thanks for Karthikeyan Bhargavan of the PROSECCO team at
7655   INRIA or reporting this issue.
7656   ([CVE-2015-0204])
7657
7658   *Steve Henson*
7659
7660 * Fixed issue where DH client certificates are accepted without verification.
7661   An OpenSSL server will accept a DH certificate for client authentication
7662   without the certificate verify message. This effectively allows a client to
7663   authenticate without the use of a private key. This only affects servers
7664   which trust a client certificate authority which issues certificates
7665   containing DH keys: these are extremely rare and hardly ever encountered.
7666   Thanks for Karthikeyan Bhargavan of the PROSECCO team at INRIA or reporting
7667   this issue.
7668   ([CVE-2015-0205])
7669
7670   *Steve Henson*
7671
7672 * Ensure that the session ID context of an SSL is updated when its
7673   SSL_CTX is updated via SSL_set_SSL_CTX.
7674
7675   The session ID context is typically set from the parent SSL_CTX,
7676   and can vary with the CTX.
7677
7678   *Adam Langley*
7679
7680 * Fix various certificate fingerprint issues.
7681
7682   By using non-DER or invalid encodings outside the signed portion of a
7683   certificate the fingerprint can be changed without breaking the signature.
7684   Although no details of the signed portion of the certificate can be changed
7685   this can cause problems with some applications: e.g. those using the
7686   certificate fingerprint for blacklists.
7687
7688   1. Reject signatures with non zero unused bits.
7689
7690   If the BIT STRING containing the signature has non zero unused bits reject
7691   the signature. All current signature algorithms require zero unused bits.
7692
7693   2. Check certificate algorithm consistency.
7694
7695   Check the AlgorithmIdentifier inside TBS matches the one in the
7696   certificate signature. NB: this will result in signature failure
7697   errors for some broken certificates.
7698
7699   Thanks to Konrad Kraszewski from Google for reporting this issue.
7700
7701   3. Check DSA/ECDSA signatures use DER.
7702
7703   Re-encode DSA/ECDSA signatures and compare with the original received
7704   signature. Return an error if there is a mismatch.
7705
7706   This will reject various cases including garbage after signature
7707   (thanks to Antti Karjalainen and Tuomo Untinen from the Codenomicon CROSS
7708   program for discovering this case) and use of BER or invalid ASN.1 INTEGERs
7709   (negative or with leading zeroes).
7710
7711   Further analysis was conducted and fixes were developed by Stephen Henson
7712   of the OpenSSL core team.
7713
7714   ([CVE-2014-8275])
7715
7716   *Steve Henson*
7717
7718 * Correct Bignum squaring. Bignum squaring (BN_sqr) may produce incorrect
7719   results on some platforms, including x86_64. This bug occurs at random
7720   with a very low probability, and is not known to be exploitable in any
7721   way, though its exact impact is difficult to determine. Thanks to Pieter
7722   Wuille (Blockstream) who reported this issue and also suggested an initial
7723   fix. Further analysis was conducted by the OpenSSL development team and
7724   Adam Langley of Google. The final fix was developed by Andy Polyakov of
7725   the OpenSSL core team.
7726   ([CVE-2014-3570])
7727
7728   *Andy Polyakov*
7729
7730 * Do not resume sessions on the server if the negotiated protocol
7731   version does not match the session's version. Resuming with a different
7732   version, while not strictly forbidden by the RFC, is of questionable
7733   sanity and breaks all known clients.
7734
7735   *David Benjamin, Emilia Käsper*
7736
7737 * Tighten handling of the ChangeCipherSpec (CCS) message: reject
7738   early CCS messages during renegotiation. (Note that because
7739   renegotiation is encrypted, this early CCS was not exploitable.)
7740
7741   *Emilia Käsper*
7742
7743 * Tighten client-side session ticket handling during renegotiation:
7744   ensure that the client only accepts a session ticket if the server sends
7745   the extension anew in the ServerHello. Previously, a TLS client would
7746   reuse the old extension state and thus accept a session ticket if one was
7747   announced in the initial ServerHello.
7748
7749   Similarly, ensure that the client requires a session ticket if one
7750   was advertised in the ServerHello. Previously, a TLS client would
7751   ignore a missing NewSessionTicket message.
7752
7753   *Emilia Käsper*
7754
7755### Changes between 1.0.1i and 1.0.1j [15 Oct 2014]
7756
7757 * SRTP Memory Leak.
7758
7759   A flaw in the DTLS SRTP extension parsing code allows an attacker, who
7760   sends a carefully crafted handshake message, to cause OpenSSL to fail
7761   to free up to 64k of memory causing a memory leak. This could be
7762   exploited in a Denial Of Service attack. This issue affects OpenSSL
7763   1.0.1 server implementations for both SSL/TLS and DTLS regardless of
7764   whether SRTP is used or configured. Implementations of OpenSSL that
7765   have been compiled with OPENSSL_NO_SRTP defined are not affected.
7766
7767   The fix was developed by the OpenSSL team.
7768   ([CVE-2014-3513])
7769
7770   *OpenSSL team*
7771
7772 * Session Ticket Memory Leak.
7773
7774   When an OpenSSL SSL/TLS/DTLS server receives a session ticket the
7775   integrity of that ticket is first verified. In the event of a session
7776   ticket integrity check failing, OpenSSL will fail to free memory
7777   causing a memory leak. By sending a large number of invalid session
7778   tickets an attacker could exploit this issue in a Denial Of Service
7779   attack.
7780   ([CVE-2014-3567])
7781
7782   *Steve Henson*
7783
7784 * Build option no-ssl3 is incomplete.
7785
7786   When OpenSSL is configured with "no-ssl3" as a build option, servers
7787   could accept and complete a SSL 3.0 handshake, and clients could be
7788   configured to send them.
7789   ([CVE-2014-3568])
7790
7791   *Akamai and the OpenSSL team*
7792
7793 * Add support for TLS_FALLBACK_SCSV.
7794   Client applications doing fallback retries should call
7795   SSL_set_mode(s, SSL_MODE_SEND_FALLBACK_SCSV).
7796   ([CVE-2014-3566])
7797
7798   *Adam Langley, Bodo Moeller*
7799
7800 * Add additional DigestInfo checks.
7801
7802   Re-encode DigestInto in DER and check against the original when
7803   verifying RSA signature: this will reject any improperly encoded
7804   DigestInfo structures.
7805
7806   Note: this is a precautionary measure and no attacks are currently known.
7807
7808   *Steve Henson*
7809
7810### Changes between 1.0.1h and 1.0.1i [6 Aug 2014]
7811
7812 * Fix SRP buffer overrun vulnerability. Invalid parameters passed to the
7813   SRP code can be overrun an internal buffer. Add sanity check that
7814   g, A, B < N to SRP code.
7815
7816   Thanks to Sean Devlin and Watson Ladd of Cryptography Services, NCC
7817   Group for discovering this issue.
7818   ([CVE-2014-3512])
7819
7820   *Steve Henson*
7821
7822 * A flaw in the OpenSSL SSL/TLS server code causes the server to negotiate
7823   TLS 1.0 instead of higher protocol versions when the ClientHello message
7824   is badly fragmented. This allows a man-in-the-middle attacker to force a
7825   downgrade to TLS 1.0 even if both the server and the client support a
7826   higher protocol version, by modifying the client's TLS records.
7827
7828   Thanks to David Benjamin and Adam Langley (Google) for discovering and
7829   researching this issue.
7830   ([CVE-2014-3511])
7831
7832   *David Benjamin*
7833
7834 * OpenSSL DTLS clients enabling anonymous (EC)DH ciphersuites are subject
7835   to a denial of service attack. A malicious server can crash the client
7836   with a null pointer dereference (read) by specifying an anonymous (EC)DH
7837   ciphersuite and sending carefully crafted handshake messages.
7838
7839   Thanks to Felix Gröbert (Google) for discovering and researching this
7840   issue.
7841   ([CVE-2014-3510])
7842
7843   *Emilia Käsper*
7844
7845 * By sending carefully crafted DTLS packets an attacker could cause openssl
7846   to leak memory. This can be exploited through a Denial of Service attack.
7847   Thanks to Adam Langley for discovering and researching this issue.
7848   ([CVE-2014-3507])
7849
7850   *Adam Langley*
7851
7852 * An attacker can force openssl to consume large amounts of memory whilst
7853   processing DTLS handshake messages. This can be exploited through a
7854   Denial of Service attack.
7855   Thanks to Adam Langley for discovering and researching this issue.
7856   ([CVE-2014-3506])
7857
7858   *Adam Langley*
7859
7860 * An attacker can force an error condition which causes openssl to crash
7861   whilst processing DTLS packets due to memory being freed twice. This
7862   can be exploited through a Denial of Service attack.
7863   Thanks to Adam Langley and Wan-Teh Chang for discovering and researching
7864   this issue.
7865   ([CVE-2014-3505])
7866
7867   *Adam Langley*
7868
7869 * If a multithreaded client connects to a malicious server using a resumed
7870   session and the server sends an ec point format extension it could write
7871   up to 255 bytes to freed memory.
7872
7873   Thanks to Gabor Tyukasz (LogMeIn Inc) for discovering and researching this
7874   issue.
7875   ([CVE-2014-3509])
7876
7877   *Gabor Tyukasz*
7878
7879 * A malicious server can crash an OpenSSL client with a null pointer
7880   dereference (read) by specifying an SRP ciphersuite even though it was not
7881   properly negotiated with the client. This can be exploited through a
7882   Denial of Service attack.
7883
7884   Thanks to Joonas Kuorilehto and Riku Hietamäki (Codenomicon) for
7885   discovering and researching this issue.
7886   ([CVE-2014-5139])
7887
7888   *Steve Henson*
7889
7890 * A flaw in OBJ_obj2txt may cause pretty printing functions such as
7891   X509_name_oneline, X509_name_print_ex et al. to leak some information
7892   from the stack. Applications may be affected if they echo pretty printing
7893   output to the attacker.
7894
7895   Thanks to Ivan Fratric (Google) for discovering this issue.
7896   ([CVE-2014-3508])
7897
7898   *Emilia Käsper, and Steve Henson*
7899
7900 * Fix ec_GFp_simple_points_make_affine (thus, EC_POINTs_mul etc.)
7901   for corner cases. (Certain input points at infinity could lead to
7902   bogus results, with non-infinity inputs mapped to infinity too.)
7903
7904   *Bodo Moeller*
7905
7906### Changes between 1.0.1g and 1.0.1h [5 Jun 2014]
7907
7908 * Fix for SSL/TLS MITM flaw. An attacker using a carefully crafted
7909   handshake can force the use of weak keying material in OpenSSL
7910   SSL/TLS clients and servers.
7911
7912   Thanks to KIKUCHI Masashi (Lepidum Co. Ltd.) for discovering and
7913   researching this issue. ([CVE-2014-0224])
7914
7915   *KIKUCHI Masashi, Steve Henson*
7916
7917 * Fix DTLS recursion flaw. By sending an invalid DTLS handshake to an
7918   OpenSSL DTLS client the code can be made to recurse eventually crashing
7919   in a DoS attack.
7920
7921   Thanks to Imre Rad (Search-Lab Ltd.) for discovering this issue.
7922   ([CVE-2014-0221])
7923
7924   *Imre Rad, Steve Henson*
7925
7926 * Fix DTLS invalid fragment vulnerability. A buffer overrun attack can
7927   be triggered by sending invalid DTLS fragments to an OpenSSL DTLS
7928   client or server. This is potentially exploitable to run arbitrary
7929   code on a vulnerable client or server.
7930
7931   Thanks to Jüri Aedla for reporting this issue. ([CVE-2014-0195])
7932
7933   *Jüri Aedla, Steve Henson*
7934
7935 * Fix bug in TLS code where clients enable anonymous ECDH ciphersuites
7936   are subject to a denial of service attack.
7937
7938   Thanks to Felix Gröbert and Ivan Fratric at Google for discovering
7939   this issue. ([CVE-2014-3470])
7940
7941   *Felix Gröbert, Ivan Fratric, Steve Henson*
7942
7943 * Harmonize version and its documentation. -f flag is used to display
7944   compilation flags.
7945
7946   *mancha <mancha1@zoho.com>*
7947
7948 * Fix eckey_priv_encode so it immediately returns an error upon a failure
7949   in i2d_ECPrivateKey.
7950
7951   *mancha <mancha1@zoho.com>*
7952
7953 * Fix some double frees. These are not thought to be exploitable.
7954
7955   *mancha <mancha1@zoho.com>*
7956
7957### Changes between 1.0.1f and 1.0.1g [7 Apr 2014]
7958
7959 * A missing bounds check in the handling of the TLS heartbeat extension
7960   can be used to reveal up to 64k of memory to a connected client or
7961   server.
7962
7963   Thanks for Neel Mehta of Google Security for discovering this bug and to
7964   Adam Langley <agl@chromium.org> and Bodo Moeller <bmoeller@acm.org> for
7965   preparing the fix ([CVE-2014-0160])
7966
7967   *Adam Langley, Bodo Moeller*
7968
7969 * Fix for the attack described in the paper "Recovering OpenSSL
7970   ECDSA Nonces Using the FLUSH+RELOAD Cache Side-channel Attack"
7971   by Yuval Yarom and Naomi Benger. Details can be obtained from:
7972   <http://eprint.iacr.org/2014/140>
7973
7974   Thanks to Yuval Yarom and Naomi Benger for discovering this
7975   flaw and to Yuval Yarom for supplying a fix ([CVE-2014-0076])
7976
7977   *Yuval Yarom and Naomi Benger*
7978
7979 * TLS pad extension: draft-agl-tls-padding-03
7980
7981   Workaround for the "TLS hang bug" (see FAQ and PR#2771): if the
7982   TLS client Hello record length value would otherwise be > 255 and
7983   less that 512 pad with a dummy extension containing zeroes so it
7984   is at least 512 bytes long.
7985
7986   *Adam Langley, Steve Henson*
7987
7988### Changes between 1.0.1e and 1.0.1f [6 Jan 2014]
7989
7990 * Fix for TLS record tampering bug. A carefully crafted invalid
7991   handshake could crash OpenSSL with a NULL pointer exception.
7992   Thanks to Anton Johansson for reporting this issues.
7993   ([CVE-2013-4353])
7994
7995 * Keep original DTLS digest and encryption contexts in retransmission
7996   structures so we can use the previous session parameters if they need
7997   to be resent. ([CVE-2013-6450])
7998
7999   *Steve Henson*
8000
8001 * Add option SSL_OP_SAFARI_ECDHE_ECDSA_BUG (part of SSL_OP_ALL) which
8002   avoids preferring ECDHE-ECDSA ciphers when the client appears to be
8003   Safari on OS X.  Safari on OS X 10.8..10.8.3 advertises support for
8004   several ECDHE-ECDSA ciphers, but fails to negotiate them.  The bug
8005   is fixed in OS X 10.8.4, but Apple have ruled out both hot fixing
8006   10.8..10.8.3 and forcing users to upgrade to 10.8.4 or newer.
8007
8008   *Rob Stradling, Adam Langley*
8009
8010### Changes between 1.0.1d and 1.0.1e [11 Feb 2013]
8011
8012 * Correct fix for CVE-2013-0169. The original didn't work on AES-NI
8013   supporting platforms or when small records were transferred.
8014
8015   *Andy Polyakov, Steve Henson*
8016
8017### Changes between 1.0.1c and 1.0.1d [5 Feb 2013]
8018
8019 * Make the decoding of SSLv3, TLS and DTLS CBC records constant time.
8020
8021   This addresses the flaw in CBC record processing discovered by
8022   Nadhem Alfardan and Kenny Paterson. Details of this attack can be found
8023   at: <http://www.isg.rhul.ac.uk/tls/>
8024
8025   Thanks go to Nadhem Alfardan and Kenny Paterson of the Information
8026   Security Group at Royal Holloway, University of London
8027   (www.isg.rhul.ac.uk) for discovering this flaw and Adam Langley and
8028   Emilia Käsper for the initial patch.
8029   ([CVE-2013-0169])
8030
8031   *Emilia Käsper, Adam Langley, Ben Laurie, Andy Polyakov, Steve Henson*
8032
8033 * Fix flaw in AESNI handling of TLS 1.2 and 1.1 records for CBC mode
8034   ciphersuites which can be exploited in a denial of service attack.
8035   Thanks go to and to Adam Langley <agl@chromium.org> for discovering
8036   and detecting this bug and to Wolfgang Ettlinger
8037   <wolfgang.ettlinger@gmail.com> for independently discovering this issue.
8038   ([CVE-2012-2686])
8039
8040   *Adam Langley*
8041
8042 * Return an error when checking OCSP signatures when key is NULL.
8043   This fixes a DoS attack. ([CVE-2013-0166])
8044
8045   *Steve Henson*
8046
8047 * Make openssl verify return errors.
8048
8049   *Chris Palmer <palmer@google.com> and Ben Laurie*
8050
8051 * Call OCSP Stapling callback after ciphersuite has been chosen, so
8052   the right response is stapled. Also change SSL_get_certificate()
8053   so it returns the certificate actually sent.
8054   See <http://rt.openssl.org/Ticket/Display.html?id=2836>.
8055
8056   *Rob Stradling <rob.stradling@comodo.com>*
8057
8058 * Fix possible deadlock when decoding public keys.
8059
8060   *Steve Henson*
8061
8062 * Don't use TLS 1.0 record version number in initial client hello
8063   if renegotiating.
8064
8065   *Steve Henson*
8066
8067### Changes between 1.0.1b and 1.0.1c [10 May 2012]
8068
8069 * Sanity check record length before skipping explicit IV in TLS
8070   1.2, 1.1 and DTLS to fix DoS attack.
8071
8072   Thanks to Codenomicon for discovering this issue using Fuzz-o-Matic
8073   fuzzing as a service testing platform.
8074   ([CVE-2012-2333])
8075
8076   *Steve Henson*
8077
8078 * Initialise tkeylen properly when encrypting CMS messages.
8079   Thanks to Solar Designer of Openwall for reporting this issue.
8080
8081   *Steve Henson*
8082
8083 * In FIPS mode don't try to use composite ciphers as they are not
8084   approved.
8085
8086   *Steve Henson*
8087
8088### Changes between 1.0.1a and 1.0.1b [26 Apr 2012]
8089
8090 * OpenSSL 1.0.0 sets SSL_OP_ALL to 0x80000FFFL and OpenSSL 1.0.1 and
8091   1.0.1a set SSL_OP_NO_TLSv1_1 to 0x00000400L which would unfortunately
8092   mean any application compiled against OpenSSL 1.0.0 headers setting
8093   SSL_OP_ALL would also set SSL_OP_NO_TLSv1_1, unintentionally disabling
8094   TLS 1.1 also. Fix this by changing the value of SSL_OP_NO_TLSv1_1 to
8095   0x10000000L Any application which was previously compiled against
8096   OpenSSL 1.0.1 or 1.0.1a headers and which cares about SSL_OP_NO_TLSv1_1
8097   will need to be recompiled as a result. Letting be results in
8098   inability to disable specifically TLS 1.1 and in client context,
8099   in unlike event, limit maximum offered version to TLS 1.0 [see below].
8100
8101   *Steve Henson*
8102
8103 * In order to ensure interoperability SSL_OP_NO_protocolX does not
8104   disable just protocol X, but all protocols above X *if* there are
8105   protocols *below* X still enabled. In more practical terms it means
8106   that if application wants to disable TLS1.0 in favor of TLS1.1 and
8107   above, it's not sufficient to pass `SSL_OP_NO_TLSv1`, one has to pass
8108   `SSL_OP_NO_TLSv1|SSL_OP_NO_SSLv3|SSL_OP_NO_SSLv2`. This applies to
8109   client side.
8110
8111   *Andy Polyakov*
8112
8113### Changes between 1.0.1 and 1.0.1a [19 Apr 2012]
8114
8115 * Check for potentially exploitable overflows in asn1_d2i_read_bio
8116   BUF_mem_grow and BUF_mem_grow_clean. Refuse attempts to shrink buffer
8117   in CRYPTO_realloc_clean.
8118
8119   Thanks to Tavis Ormandy, Google Security Team, for discovering this
8120   issue and to Adam Langley <agl@chromium.org> for fixing it.
8121   ([CVE-2012-2110])
8122
8123   *Adam Langley (Google), Tavis Ormandy, Google Security Team*
8124
8125 * Don't allow TLS 1.2 SHA-256 ciphersuites in TLS 1.0, 1.1 connections.
8126
8127   *Adam Langley*
8128
8129 * Workarounds for some broken servers that "hang" if a client hello
8130   record length exceeds 255 bytes.
8131
8132   1. Do not use record version number > TLS 1.0 in initial client
8133      hello: some (but not all) hanging servers will now work.
8134   2. If we set OPENSSL_MAX_TLS1_2_CIPHER_LENGTH this will truncate
8135      the number of ciphers sent in the client hello. This should be
8136      set to an even number, such as 50, for example by passing:
8137      -DOPENSSL_MAX_TLS1_2_CIPHER_LENGTH=50 to config or Configure.
8138      Most broken servers should now work.
8139   3. If all else fails setting OPENSSL_NO_TLS1_2_CLIENT will disable
8140      TLS 1.2 client support entirely.
8141
8142   *Steve Henson*
8143
8144 * Fix SEGV in Vector Permutation AES module observed in OpenSSH.
8145
8146   *Andy Polyakov*
8147
8148### Changes between 1.0.0h and 1.0.1  [14 Mar 2012]
8149
8150 * Add compatibility with old MDC2 signatures which use an ASN1 OCTET
8151   STRING form instead of a DigestInfo.
8152
8153   *Steve Henson*
8154
8155 * The format used for MDC2 RSA signatures is inconsistent between EVP
8156   and the RSA_sign/RSA_verify functions. This was made more apparent when
8157   OpenSSL used RSA_sign/RSA_verify for some RSA signatures in particular
8158   those which went through EVP_PKEY_METHOD in 1.0.0 and later. Detect
8159   the correct format in RSA_verify so both forms transparently work.
8160
8161   *Steve Henson*
8162
8163 * Some servers which support TLS 1.0 can choke if we initially indicate
8164   support for TLS 1.2 and later renegotiate using TLS 1.0 in the RSA
8165   encrypted premaster secret. As a workaround use the maximum permitted
8166   client version in client hello, this should keep such servers happy
8167   and still work with previous versions of OpenSSL.
8168
8169   *Steve Henson*
8170
8171 * Add support for TLS/DTLS heartbeats.
8172
8173   *Robin Seggelmann <seggelmann@fh-muenster.de>*
8174
8175 * Add support for SCTP.
8176
8177   *Robin Seggelmann <seggelmann@fh-muenster.de>*
8178
8179 * Improved PRNG seeding for VOS.
8180
8181   *Paul Green <Paul.Green@stratus.com>*
8182
8183 * Extensive assembler packs updates, most notably:
8184
8185   - x86[_64]:     AES-NI, PCLMULQDQ, RDRAND support;
8186   - x86[_64]:     SSSE3 support (SHA1, vector-permutation AES);
8187   - x86_64:       bit-sliced AES implementation;
8188   - ARM:          NEON support, contemporary platforms optimizations;
8189   - s390x:        z196 support;
8190   - `*`:            GHASH and GF(2^m) multiplication implementations;
8191
8192   *Andy Polyakov*
8193
8194 * Make TLS-SRP code conformant with RFC 5054 API cleanup
8195   (removal of unnecessary code)
8196
8197   *Peter Sylvester <peter.sylvester@edelweb.fr>*
8198
8199 * Add TLS key material exporter from RFC 5705.
8200
8201   *Eric Rescorla*
8202
8203 * Add DTLS-SRTP negotiation from RFC 5764.
8204
8205   *Eric Rescorla*
8206
8207 * Add Next Protocol Negotiation,
8208   <http://tools.ietf.org/html/draft-agl-tls-nextprotoneg-00>. Can be
8209   disabled with a no-npn flag to config or Configure. Code donated
8210   by Google.
8211
8212   *Adam Langley <agl@google.com> and Ben Laurie*
8213
8214 * Add optional 64-bit optimized implementations of elliptic curves NIST-P224,
8215   NIST-P256, NIST-P521, with constant-time single point multiplication on
8216   typical inputs. Compiler support for the nonstandard type `__uint128_t` is
8217   required to use this (present in gcc 4.4 and later, for 64-bit builds).
8218   Code made available under Apache License version 2.0.
8219
8220   Specify "enable-ec_nistp_64_gcc_128" on the Configure (or config) command
8221   line to include this in your build of OpenSSL, and run "make depend" (or
8222   "make update"). This enables the following EC_METHODs:
8223
8224           EC_GFp_nistp224_method()
8225           EC_GFp_nistp256_method()
8226           EC_GFp_nistp521_method()
8227
8228   EC_GROUP_new_by_curve_name() will automatically use these (while
8229   EC_GROUP_new_curve_GFp() currently prefers the more flexible
8230   implementations).
8231
8232   *Emilia Käsper, Adam Langley, Bodo Moeller (Google)*
8233
8234 * Use type ossl_ssize_t instead of ssize_t which isn't available on
8235   all platforms. Move ssize_t definition from e_os.h to the public
8236   header file e_os2.h as it now appears in public header file cms.h
8237
8238   *Steve Henson*
8239
8240 * New -sigopt option to the ca, req and x509 utilities. Additional
8241   signature parameters can be passed using this option and in
8242   particular PSS.
8243
8244   *Steve Henson*
8245
8246 * Add RSA PSS signing function. This will generate and set the
8247   appropriate AlgorithmIdentifiers for PSS based on those in the
8248   corresponding EVP_MD_CTX structure. No application support yet.
8249
8250   *Steve Henson*
8251
8252 * Support for companion algorithm specific ASN1 signing routines.
8253   New function ASN1_item_sign_ctx() signs a pre-initialised
8254   EVP_MD_CTX structure and sets AlgorithmIdentifiers based on
8255   the appropriate parameters.
8256
8257   *Steve Henson*
8258
8259 * Add new algorithm specific ASN1 verification initialisation function
8260   to EVP_PKEY_ASN1_METHOD: this is not in EVP_PKEY_METHOD since the ASN1
8261   handling will be the same no matter what EVP_PKEY_METHOD is used.
8262   Add a PSS handler to support verification of PSS signatures: checked
8263   against a number of sample certificates.
8264
8265   *Steve Henson*
8266
8267 * Add signature printing for PSS. Add PSS OIDs.
8268
8269   *Steve Henson, Martin Kaiser <lists@kaiser.cx>*
8270
8271 * Add algorithm specific signature printing. An individual ASN1 method
8272   can now print out signatures instead of the standard hex dump.
8273
8274   More complex signatures (e.g. PSS) can print out more meaningful
8275   information. Include DSA version that prints out the signature
8276   parameters r, s.
8277
8278   *Steve Henson*
8279
8280 * Password based recipient info support for CMS library: implementing
8281   RFC3211.
8282
8283   *Steve Henson*
8284
8285 * Split password based encryption into PBES2 and PBKDF2 functions. This
8286   neatly separates the code into cipher and PBE sections and is required
8287   for some algorithms that split PBES2 into separate pieces (such as
8288   password based CMS).
8289
8290   *Steve Henson*
8291
8292 * Session-handling fixes:
8293   - Fix handling of connections that are resuming with a session ID,
8294     but also support Session Tickets.
8295   - Fix a bug that suppressed issuing of a new ticket if the client
8296     presented a ticket with an expired session.
8297   - Try to set the ticket lifetime hint to something reasonable.
8298   - Make tickets shorter by excluding irrelevant information.
8299   - On the client side, don't ignore renewed tickets.
8300
8301   *Adam Langley, Bodo Moeller (Google)*
8302
8303 * Fix PSK session representation.
8304
8305   *Bodo Moeller*
8306
8307 * Add RC4-MD5 and AESNI-SHA1 "stitched" implementations.
8308
8309   This work was sponsored by Intel.
8310
8311   *Andy Polyakov*
8312
8313 * Add GCM support to TLS library. Some custom code is needed to split
8314   the IV between the fixed (from PRF) and explicit (from TLS record)
8315   portions. This adds all GCM ciphersuites supported by RFC5288 and
8316   RFC5289. Generalise some `AES*` cipherstrings to include GCM and
8317   add a special AESGCM string for GCM only.
8318
8319   *Steve Henson*
8320
8321 * Expand range of ctrls for AES GCM. Permit setting invocation
8322   field on decrypt and retrieval of invocation field only on encrypt.
8323
8324   *Steve Henson*
8325
8326 * Add HMAC ECC ciphersuites from RFC5289. Include SHA384 PRF support.
8327   As required by RFC5289 these ciphersuites cannot be used if for
8328   versions of TLS earlier than 1.2.
8329
8330   *Steve Henson*
8331
8332 * For FIPS capable OpenSSL interpret a NULL default public key method
8333   as unset and return the appropriate default but do *not* set the default.
8334   This means we can return the appropriate method in applications that
8335   switch between FIPS and non-FIPS modes.
8336
8337   *Steve Henson*
8338
8339 * Redirect HMAC and CMAC operations to FIPS module in FIPS mode. If an
8340   ENGINE is used then we cannot handle that in the FIPS module so we
8341   keep original code iff non-FIPS operations are allowed.
8342
8343   *Steve Henson*
8344
8345 * Add -attime option to openssl utilities.
8346
8347   *Peter Eckersley <pde@eff.org>, Ben Laurie and Steve Henson*
8348
8349 * Redirect DSA and DH operations to FIPS module in FIPS mode.
8350
8351   *Steve Henson*
8352
8353 * Redirect ECDSA and ECDH operations to FIPS module in FIPS mode. Also use
8354   FIPS EC methods unconditionally for now.
8355
8356   *Steve Henson*
8357
8358 * New build option no-ec2m to disable characteristic 2 code.
8359
8360   *Steve Henson*
8361
8362 * Backport libcrypto audit of return value checking from 1.1.0-dev; not
8363   all cases can be covered as some introduce binary incompatibilities.
8364
8365   *Steve Henson*
8366
8367 * Redirect RSA operations to FIPS module including keygen,
8368   encrypt, decrypt, sign and verify. Block use of non FIPS RSA methods.
8369
8370   *Steve Henson*
8371
8372 * Add similar low-level API blocking to ciphers.
8373
8374   *Steve Henson*
8375
8376 * low-level digest APIs are not approved in FIPS mode: any attempt
8377   to use these will cause a fatal error. Applications that *really* want
8378   to use them can use the `private_*` version instead.
8379
8380   *Steve Henson*
8381
8382 * Redirect cipher operations to FIPS module for FIPS builds.
8383
8384   *Steve Henson*
8385
8386 * Redirect digest operations to FIPS module for FIPS builds.
8387
8388   *Steve Henson*
8389
8390 * Update build system to add "fips" flag which will link in fipscanister.o
8391   for static and shared library builds embedding a signature if needed.
8392
8393   *Steve Henson*
8394
8395 * Output TLS supported curves in preference order instead of numerical
8396   order. This is currently hardcoded for the highest order curves first.
8397   This should be configurable so applications can judge speed vs strength.
8398
8399   *Steve Henson*
8400
8401 * Add TLS v1.2 server support for client authentication.
8402
8403   *Steve Henson*
8404
8405 * Add support for FIPS mode in ssl library: disable SSLv3, non-FIPS ciphers
8406   and enable MD5.
8407
8408   *Steve Henson*
8409
8410 * Functions FIPS_mode_set() and FIPS_mode() which call the underlying
8411   FIPS modules versions.
8412
8413   *Steve Henson*
8414
8415 * Add TLS v1.2 client side support for client authentication. Keep cache
8416   of handshake records longer as we don't know the hash algorithm to use
8417   until after the certificate request message is received.
8418
8419   *Steve Henson*
8420
8421 * Initial TLS v1.2 client support. Add a default signature algorithms
8422   extension including all the algorithms we support. Parse new signature
8423   format in client key exchange. Relax some ECC signing restrictions for
8424   TLS v1.2 as indicated in RFC5246.
8425
8426   *Steve Henson*
8427
8428 * Add server support for TLS v1.2 signature algorithms extension. Switch
8429   to new signature format when needed using client digest preference.
8430   All server ciphersuites should now work correctly in TLS v1.2. No client
8431   support yet and no support for client certificates.
8432
8433   *Steve Henson*
8434
8435 * Initial TLS v1.2 support. Add new SHA256 digest to ssl code, switch
8436   to SHA256 for PRF when using TLS v1.2 and later. Add new SHA256 based
8437   ciphersuites. At present only RSA key exchange ciphersuites work with
8438   TLS v1.2. Add new option for TLS v1.2 replacing the old and obsolete
8439   SSL_OP_PKCS1_CHECK flags with SSL_OP_NO_TLSv1_2. New TLSv1.2 methods
8440   and version checking.
8441
8442   *Steve Henson*
8443
8444 * New option OPENSSL_NO_SSL_INTERN. If an application can be compiled
8445   with this defined it will not be affected by any changes to ssl internal
8446   structures. Add several utility functions to allow openssl application
8447   to work with OPENSSL_NO_SSL_INTERN defined.
8448
8449   *Steve Henson*
8450
8451 * A long standing patch to add support for SRP from EdelWeb (Peter
8452   Sylvester and Christophe Renou) was integrated.
8453   *Christophe Renou <christophe.renou@edelweb.fr>, Peter Sylvester
8454   <peter.sylvester@edelweb.fr>, Tom Wu <tjw@cs.stanford.edu>, and
8455   Ben Laurie*
8456
8457 * Add functions to copy EVP_PKEY_METHOD and retrieve flags and id.
8458
8459   *Steve Henson*
8460
8461 * Permit abbreviated handshakes when renegotiating using the function
8462   SSL_renegotiate_abbreviated().
8463
8464   *Robin Seggelmann <seggelmann@fh-muenster.de>*
8465
8466 * Add call to ENGINE_register_all_complete() to
8467   ENGINE_load_builtin_engines(), so some implementations get used
8468   automatically instead of needing explicit application support.
8469
8470   *Steve Henson*
8471
8472 * Add support for TLS key exporter as described in RFC5705.
8473
8474   *Robin Seggelmann <seggelmann@fh-muenster.de>, Steve Henson*
8475
8476 * Initial TLSv1.1 support. Since TLSv1.1 is very similar to TLS v1.0 only
8477   a few changes are required:
8478
8479     Add SSL_OP_NO_TLSv1_1 flag.
8480     Add TLSv1_1 methods.
8481     Update version checking logic to handle version 1.1.
8482     Add explicit IV handling (ported from DTLS code).
8483     Add command line options to s_client/s_server.
8484
8485   *Steve Henson*
8486
8487OpenSSL 1.0.0
8488-------------
8489
8490### Changes between 1.0.0s and 1.0.0t [3 Dec 2015]
8491
8492 * X509_ATTRIBUTE memory leak
8493
8494   When presented with a malformed X509_ATTRIBUTE structure OpenSSL will leak
8495   memory. This structure is used by the PKCS#7 and CMS routines so any
8496   application which reads PKCS#7 or CMS data from untrusted sources is
8497   affected. SSL/TLS is not affected.
8498
8499   This issue was reported to OpenSSL by Adam Langley (Google/BoringSSL) using
8500   libFuzzer.
8501   ([CVE-2015-3195])
8502
8503   *Stephen Henson*
8504
8505 * Race condition handling PSK identify hint
8506
8507   If PSK identity hints are received by a multi-threaded client then
8508   the values are wrongly updated in the parent SSL_CTX structure. This can
8509   result in a race condition potentially leading to a double free of the
8510   identify hint data.
8511   ([CVE-2015-3196])
8512
8513   *Stephen Henson*
8514
8515### Changes between 1.0.0r and 1.0.0s [11 Jun 2015]
8516
8517 * Malformed ECParameters causes infinite loop
8518
8519   When processing an ECParameters structure OpenSSL enters an infinite loop
8520   if the curve specified is over a specially malformed binary polynomial
8521   field.
8522
8523   This can be used to perform denial of service against any
8524   system which processes public keys, certificate requests or
8525   certificates.  This includes TLS clients and TLS servers with
8526   client authentication enabled.
8527
8528   This issue was reported to OpenSSL by Joseph Barr-Pixton.
8529   ([CVE-2015-1788])
8530
8531   *Andy Polyakov*
8532
8533 * Exploitable out-of-bounds read in X509_cmp_time
8534
8535   X509_cmp_time does not properly check the length of the ASN1_TIME
8536   string and can read a few bytes out of bounds. In addition,
8537   X509_cmp_time accepts an arbitrary number of fractional seconds in the
8538   time string.
8539
8540   An attacker can use this to craft malformed certificates and CRLs of
8541   various sizes and potentially cause a segmentation fault, resulting in
8542   a DoS on applications that verify certificates or CRLs. TLS clients
8543   that verify CRLs are affected. TLS clients and servers with client
8544   authentication enabled may be affected if they use custom verification
8545   callbacks.
8546
8547   This issue was reported to OpenSSL by Robert Swiecki (Google), and
8548   independently by Hanno Böck.
8549   ([CVE-2015-1789])
8550
8551   *Emilia Käsper*
8552
8553 * PKCS7 crash with missing EnvelopedContent
8554
8555   The PKCS#7 parsing code does not handle missing inner EncryptedContent
8556   correctly. An attacker can craft malformed ASN.1-encoded PKCS#7 blobs
8557   with missing content and trigger a NULL pointer dereference on parsing.
8558
8559   Applications that decrypt PKCS#7 data or otherwise parse PKCS#7
8560   structures from untrusted sources are affected. OpenSSL clients and
8561   servers are not affected.
8562
8563   This issue was reported to OpenSSL by Michal Zalewski (Google).
8564   ([CVE-2015-1790])
8565
8566   *Emilia Käsper*
8567
8568 * CMS verify infinite loop with unknown hash function
8569
8570   When verifying a signedData message the CMS code can enter an infinite loop
8571   if presented with an unknown hash function OID. This can be used to perform
8572   denial of service against any system which verifies signedData messages using
8573   the CMS code.
8574   This issue was reported to OpenSSL by Johannes Bauer.
8575   ([CVE-2015-1792])
8576
8577   *Stephen Henson*
8578
8579 * Race condition handling NewSessionTicket
8580
8581   If a NewSessionTicket is received by a multi-threaded client when attempting to
8582   reuse a previous ticket then a race condition can occur potentially leading to
8583   a double free of the ticket data.
8584   ([CVE-2015-1791])
8585
8586   *Matt Caswell*
8587
8588### Changes between 1.0.0q and 1.0.0r [19 Mar 2015]
8589
8590 * Segmentation fault in ASN1_TYPE_cmp fix
8591
8592   The function ASN1_TYPE_cmp will crash with an invalid read if an attempt is
8593   made to compare ASN.1 boolean types. Since ASN1_TYPE_cmp is used to check
8594   certificate signature algorithm consistency this can be used to crash any
8595   certificate verification operation and exploited in a DoS attack. Any
8596   application which performs certificate verification is vulnerable including
8597   OpenSSL clients and servers which enable client authentication.
8598   ([CVE-2015-0286])
8599
8600   *Stephen Henson*
8601
8602 * ASN.1 structure reuse memory corruption fix
8603
8604   Reusing a structure in ASN.1 parsing may allow an attacker to cause
8605   memory corruption via an invalid write. Such reuse is and has been
8606   strongly discouraged and is believed to be rare.
8607
8608   Applications that parse structures containing CHOICE or ANY DEFINED BY
8609   components may be affected. Certificate parsing (d2i_X509 and related
8610   functions) are however not affected. OpenSSL clients and servers are
8611   not affected.
8612   ([CVE-2015-0287])
8613
8614   *Stephen Henson*
8615
8616 * PKCS7 NULL pointer dereferences fix
8617
8618   The PKCS#7 parsing code does not handle missing outer ContentInfo
8619   correctly. An attacker can craft malformed ASN.1-encoded PKCS#7 blobs with
8620   missing content and trigger a NULL pointer dereference on parsing.
8621
8622   Applications that verify PKCS#7 signatures, decrypt PKCS#7 data or
8623   otherwise parse PKCS#7 structures from untrusted sources are
8624   affected. OpenSSL clients and servers are not affected.
8625
8626   This issue was reported to OpenSSL by Michal Zalewski (Google).
8627   ([CVE-2015-0289])
8628
8629   *Emilia Käsper*
8630
8631 * DoS via reachable assert in SSLv2 servers fix
8632
8633   A malicious client can trigger an OPENSSL_assert (i.e., an abort) in
8634   servers that both support SSLv2 and enable export cipher suites by sending
8635   a specially crafted SSLv2 CLIENT-MASTER-KEY message.
8636
8637   This issue was discovered by Sean Burford (Google) and Emilia Käsper
8638   (OpenSSL development team).
8639   ([CVE-2015-0293])
8640
8641   *Emilia Käsper*
8642
8643 * Use After Free following d2i_ECPrivatekey error fix
8644
8645   A malformed EC private key file consumed via the d2i_ECPrivateKey function
8646   could cause a use after free condition. This, in turn, could cause a double
8647   free in several private key parsing functions (such as d2i_PrivateKey
8648   or EVP_PKCS82PKEY) and could lead to a DoS attack or memory corruption
8649   for applications that receive EC private keys from untrusted
8650   sources. This scenario is considered rare.
8651
8652   This issue was discovered by the BoringSSL project and fixed in their
8653   commit 517073cd4b.
8654   ([CVE-2015-0209])
8655
8656   *Matt Caswell*
8657
8658 * X509_to_X509_REQ NULL pointer deref fix
8659
8660   The function X509_to_X509_REQ will crash with a NULL pointer dereference if
8661   the certificate key is invalid. This function is rarely used in practice.
8662
8663   This issue was discovered by Brian Carpenter.
8664   ([CVE-2015-0288])
8665
8666   *Stephen Henson*
8667
8668 * Removed the export ciphers from the DEFAULT ciphers
8669
8670   *Kurt Roeckx*
8671
8672### Changes between 1.0.0p and 1.0.0q [15 Jan 2015]
8673
8674 * Build fixes for the Windows and OpenVMS platforms
8675
8676   *Matt Caswell and Richard Levitte*
8677
8678### Changes between 1.0.0o and 1.0.0p [8 Jan 2015]
8679
8680 * Fix DTLS segmentation fault in dtls1_get_record. A carefully crafted DTLS
8681   message can cause a segmentation fault in OpenSSL due to a NULL pointer
8682   dereference. This could lead to a Denial Of Service attack. Thanks to
8683   Markus Stenberg of Cisco Systems, Inc. for reporting this issue.
8684   ([CVE-2014-3571])
8685
8686   *Steve Henson*
8687
8688 * Fix DTLS memory leak in dtls1_buffer_record. A memory leak can occur in the
8689   dtls1_buffer_record function under certain conditions. In particular this
8690   could occur if an attacker sent repeated DTLS records with the same
8691   sequence number but for the next epoch. The memory leak could be exploited
8692   by an attacker in a Denial of Service attack through memory exhaustion.
8693   Thanks to Chris Mueller for reporting this issue.
8694   ([CVE-2015-0206])
8695
8696   *Matt Caswell*
8697
8698 * Fix issue where no-ssl3 configuration sets method to NULL. When openssl is
8699   built with the no-ssl3 option and a SSL v3 ClientHello is received the ssl
8700   method would be set to NULL which could later result in a NULL pointer
8701   dereference. Thanks to Frank Schmirler for reporting this issue.
8702   ([CVE-2014-3569])
8703
8704   *Kurt Roeckx*
8705
8706 * Abort handshake if server key exchange message is omitted for ephemeral
8707   ECDH ciphersuites.
8708
8709   Thanks to Karthikeyan Bhargavan of the PROSECCO team at INRIA for
8710   reporting this issue.
8711   ([CVE-2014-3572])
8712
8713   *Steve Henson*
8714
8715 * Remove non-export ephemeral RSA code on client and server. This code
8716   violated the TLS standard by allowing the use of temporary RSA keys in
8717   non-export ciphersuites and could be used by a server to effectively
8718   downgrade the RSA key length used to a value smaller than the server
8719   certificate. Thanks for Karthikeyan Bhargavan of the PROSECCO team at
8720   INRIA or reporting this issue.
8721   ([CVE-2015-0204])
8722
8723   *Steve Henson*
8724
8725 * Fixed issue where DH client certificates are accepted without verification.
8726   An OpenSSL server will accept a DH certificate for client authentication
8727   without the certificate verify message. This effectively allows a client to
8728   authenticate without the use of a private key. This only affects servers
8729   which trust a client certificate authority which issues certificates
8730   containing DH keys: these are extremely rare and hardly ever encountered.
8731   Thanks for Karthikeyan Bhargavan of the PROSECCO team at INRIA or reporting
8732   this issue.
8733   ([CVE-2015-0205])
8734
8735   *Steve Henson*
8736
8737 * Correct Bignum squaring. Bignum squaring (BN_sqr) may produce incorrect
8738   results on some platforms, including x86_64. This bug occurs at random
8739   with a very low probability, and is not known to be exploitable in any
8740   way, though its exact impact is difficult to determine. Thanks to Pieter
8741   Wuille (Blockstream) who reported this issue and also suggested an initial
8742   fix. Further analysis was conducted by the OpenSSL development team and
8743   Adam Langley of Google. The final fix was developed by Andy Polyakov of
8744   the OpenSSL core team.
8745   ([CVE-2014-3570])
8746
8747   *Andy Polyakov*
8748
8749 * Fix various certificate fingerprint issues.
8750
8751   By using non-DER or invalid encodings outside the signed portion of a
8752   certificate the fingerprint can be changed without breaking the signature.
8753   Although no details of the signed portion of the certificate can be changed
8754   this can cause problems with some applications: e.g. those using the
8755   certificate fingerprint for blacklists.
8756
8757   1. Reject signatures with non zero unused bits.
8758
8759   If the BIT STRING containing the signature has non zero unused bits reject
8760   the signature. All current signature algorithms require zero unused bits.
8761
8762   2. Check certificate algorithm consistency.
8763
8764   Check the AlgorithmIdentifier inside TBS matches the one in the
8765   certificate signature. NB: this will result in signature failure
8766   errors for some broken certificates.
8767
8768   Thanks to Konrad Kraszewski from Google for reporting this issue.
8769
8770   3. Check DSA/ECDSA signatures use DER.
8771
8772   Reencode DSA/ECDSA signatures and compare with the original received
8773   signature. Return an error if there is a mismatch.
8774
8775   This will reject various cases including garbage after signature
8776   (thanks to Antti Karjalainen and Tuomo Untinen from the Codenomicon CROSS
8777   program for discovering this case) and use of BER or invalid ASN.1 INTEGERs
8778   (negative or with leading zeroes).
8779
8780   Further analysis was conducted and fixes were developed by Stephen Henson
8781   of the OpenSSL core team.
8782
8783   ([CVE-2014-8275])
8784
8785   *Steve Henson*
8786
8787### Changes between 1.0.0n and 1.0.0o [15 Oct 2014]
8788
8789 * Session Ticket Memory Leak.
8790
8791   When an OpenSSL SSL/TLS/DTLS server receives a session ticket the
8792   integrity of that ticket is first verified. In the event of a session
8793   ticket integrity check failing, OpenSSL will fail to free memory
8794   causing a memory leak. By sending a large number of invalid session
8795   tickets an attacker could exploit this issue in a Denial Of Service
8796   attack.
8797   ([CVE-2014-3567])
8798
8799   *Steve Henson*
8800
8801 * Build option no-ssl3 is incomplete.
8802
8803   When OpenSSL is configured with "no-ssl3" as a build option, servers
8804   could accept and complete a SSL 3.0 handshake, and clients could be
8805   configured to send them.
8806   ([CVE-2014-3568])
8807
8808   *Akamai and the OpenSSL team*
8809
8810 * Add support for TLS_FALLBACK_SCSV.
8811   Client applications doing fallback retries should call
8812   SSL_set_mode(s, SSL_MODE_SEND_FALLBACK_SCSV).
8813   ([CVE-2014-3566])
8814
8815   *Adam Langley, Bodo Moeller*
8816
8817 * Add additional DigestInfo checks.
8818
8819   Reencode DigestInto in DER and check against the original when
8820   verifying RSA signature: this will reject any improperly encoded
8821   DigestInfo structures.
8822
8823   Note: this is a precautionary measure and no attacks are currently known.
8824
8825   *Steve Henson*
8826
8827### Changes between 1.0.0m and 1.0.0n [6 Aug 2014]
8828
8829 * OpenSSL DTLS clients enabling anonymous (EC)DH ciphersuites are subject
8830   to a denial of service attack. A malicious server can crash the client
8831   with a null pointer dereference (read) by specifying an anonymous (EC)DH
8832   ciphersuite and sending carefully crafted handshake messages.
8833
8834   Thanks to Felix Gröbert (Google) for discovering and researching this
8835   issue.
8836   ([CVE-2014-3510])
8837
8838   *Emilia Käsper*
8839
8840 * By sending carefully crafted DTLS packets an attacker could cause openssl
8841   to leak memory. This can be exploited through a Denial of Service attack.
8842   Thanks to Adam Langley for discovering and researching this issue.
8843   ([CVE-2014-3507])
8844
8845   *Adam Langley*
8846
8847 * An attacker can force openssl to consume large amounts of memory whilst
8848   processing DTLS handshake messages. This can be exploited through a
8849   Denial of Service attack.
8850   Thanks to Adam Langley for discovering and researching this issue.
8851   ([CVE-2014-3506])
8852
8853   *Adam Langley*
8854
8855 * An attacker can force an error condition which causes openssl to crash
8856   whilst processing DTLS packets due to memory being freed twice. This
8857   can be exploited through a Denial of Service attack.
8858   Thanks to Adam Langley and Wan-Teh Chang for discovering and researching
8859   this issue.
8860   ([CVE-2014-3505])
8861
8862   *Adam Langley*
8863
8864 * If a multithreaded client connects to a malicious server using a resumed
8865   session and the server sends an ec point format extension it could write
8866   up to 255 bytes to freed memory.
8867
8868   Thanks to Gabor Tyukasz (LogMeIn Inc) for discovering and researching this
8869   issue.
8870   ([CVE-2014-3509])
8871
8872   *Gabor Tyukasz*
8873
8874 * A flaw in OBJ_obj2txt may cause pretty printing functions such as
8875   X509_name_oneline, X509_name_print_ex et al. to leak some information
8876   from the stack. Applications may be affected if they echo pretty printing
8877   output to the attacker.
8878
8879   Thanks to Ivan Fratric (Google) for discovering this issue.
8880   ([CVE-2014-3508])
8881
8882   *Emilia Käsper, and Steve Henson*
8883
8884 * Fix ec_GFp_simple_points_make_affine (thus, EC_POINTs_mul etc.)
8885   for corner cases. (Certain input points at infinity could lead to
8886   bogus results, with non-infinity inputs mapped to infinity too.)
8887
8888   *Bodo Moeller*
8889
8890### Changes between 1.0.0l and 1.0.0m [5 Jun 2014]
8891
8892 * Fix for SSL/TLS MITM flaw. An attacker using a carefully crafted
8893   handshake can force the use of weak keying material in OpenSSL
8894   SSL/TLS clients and servers.
8895
8896   Thanks to KIKUCHI Masashi (Lepidum Co. Ltd.) for discovering and
8897   researching this issue. ([CVE-2014-0224])
8898
8899   *KIKUCHI Masashi, Steve Henson*
8900
8901 * Fix DTLS recursion flaw. By sending an invalid DTLS handshake to an
8902   OpenSSL DTLS client the code can be made to recurse eventually crashing
8903   in a DoS attack.
8904
8905   Thanks to Imre Rad (Search-Lab Ltd.) for discovering this issue.
8906   ([CVE-2014-0221])
8907
8908   *Imre Rad, Steve Henson*
8909
8910 * Fix DTLS invalid fragment vulnerability. A buffer overrun attack can
8911   be triggered by sending invalid DTLS fragments to an OpenSSL DTLS
8912   client or server. This is potentially exploitable to run arbitrary
8913   code on a vulnerable client or server.
8914
8915   Thanks to Jüri Aedla for reporting this issue. ([CVE-2014-0195])
8916
8917   *Jüri Aedla, Steve Henson*
8918
8919 * Fix bug in TLS code where clients enable anonymous ECDH ciphersuites
8920   are subject to a denial of service attack.
8921
8922   Thanks to Felix Gröbert and Ivan Fratric at Google for discovering
8923   this issue. ([CVE-2014-3470])
8924
8925   *Felix Gröbert, Ivan Fratric, Steve Henson*
8926
8927 * Harmonize version and its documentation. -f flag is used to display
8928   compilation flags.
8929
8930   *mancha <mancha1@zoho.com>*
8931
8932 * Fix eckey_priv_encode so it immediately returns an error upon a failure
8933   in i2d_ECPrivateKey.
8934
8935   *mancha <mancha1@zoho.com>*
8936
8937 * Fix some double frees. These are not thought to be exploitable.
8938
8939   *mancha <mancha1@zoho.com>*
8940
8941 * Fix for the attack described in the paper "Recovering OpenSSL
8942   ECDSA Nonces Using the FLUSH+RELOAD Cache Side-channel Attack"
8943   by Yuval Yarom and Naomi Benger. Details can be obtained from:
8944   <http://eprint.iacr.org/2014/140>
8945
8946   Thanks to Yuval Yarom and Naomi Benger for discovering this
8947   flaw and to Yuval Yarom for supplying a fix ([CVE-2014-0076])
8948
8949   *Yuval Yarom and Naomi Benger*
8950
8951### Changes between 1.0.0k and 1.0.0l [6 Jan 2014]
8952
8953 * Keep original DTLS digest and encryption contexts in retransmission
8954   structures so we can use the previous session parameters if they need
8955   to be resent. ([CVE-2013-6450])
8956
8957   *Steve Henson*
8958
8959 * Add option SSL_OP_SAFARI_ECDHE_ECDSA_BUG (part of SSL_OP_ALL) which
8960   avoids preferring ECDHE-ECDSA ciphers when the client appears to be
8961   Safari on OS X.  Safari on OS X 10.8..10.8.3 advertises support for
8962   several ECDHE-ECDSA ciphers, but fails to negotiate them.  The bug
8963   is fixed in OS X 10.8.4, but Apple have ruled out both hot fixing
8964   10.8..10.8.3 and forcing users to upgrade to 10.8.4 or newer.
8965
8966   *Rob Stradling, Adam Langley*
8967
8968### Changes between 1.0.0j and 1.0.0k [5 Feb 2013]
8969
8970 * Make the decoding of SSLv3, TLS and DTLS CBC records constant time.
8971
8972   This addresses the flaw in CBC record processing discovered by
8973   Nadhem Alfardan and Kenny Paterson. Details of this attack can be found
8974   at: <http://www.isg.rhul.ac.uk/tls/>
8975
8976   Thanks go to Nadhem Alfardan and Kenny Paterson of the Information
8977   Security Group at Royal Holloway, University of London
8978   (www.isg.rhul.ac.uk) for discovering this flaw and Adam Langley and
8979   Emilia Käsper for the initial patch.
8980   ([CVE-2013-0169])
8981
8982   *Emilia Käsper, Adam Langley, Ben Laurie, Andy Polyakov, Steve Henson*
8983
8984 * Return an error when checking OCSP signatures when key is NULL.
8985   This fixes a DoS attack. ([CVE-2013-0166])
8986
8987   *Steve Henson*
8988
8989 * Call OCSP Stapling callback after ciphersuite has been chosen, so
8990   the right response is stapled. Also change SSL_get_certificate()
8991   so it returns the certificate actually sent.
8992   See <http://rt.openssl.org/Ticket/Display.html?id=2836>.
8993   (This is a backport)
8994
8995   *Rob Stradling <rob.stradling@comodo.com>*
8996
8997 * Fix possible deadlock when decoding public keys.
8998
8999   *Steve Henson*
9000
9001### Changes between 1.0.0i and 1.0.0j [10 May 2012]
9002
9003[NB: OpenSSL 1.0.0i and later 1.0.0 patch levels were released after
9004OpenSSL 1.0.1.]
9005
9006 * Sanity check record length before skipping explicit IV in DTLS
9007   to fix DoS attack.
9008
9009   Thanks to Codenomicon for discovering this issue using Fuzz-o-Matic
9010   fuzzing as a service testing platform.
9011   ([CVE-2012-2333])
9012
9013   *Steve Henson*
9014
9015 * Initialise tkeylen properly when encrypting CMS messages.
9016   Thanks to Solar Designer of Openwall for reporting this issue.
9017
9018   *Steve Henson*
9019
9020### Changes between 1.0.0h and 1.0.0i [19 Apr 2012]
9021
9022 * Check for potentially exploitable overflows in asn1_d2i_read_bio
9023   BUF_mem_grow and BUF_mem_grow_clean. Refuse attempts to shrink buffer
9024   in CRYPTO_realloc_clean.
9025
9026   Thanks to Tavis Ormandy, Google Security Team, for discovering this
9027   issue and to Adam Langley <agl@chromium.org> for fixing it.
9028   ([CVE-2012-2110])
9029
9030   *Adam Langley (Google), Tavis Ormandy, Google Security Team*
9031
9032### Changes between 1.0.0g and 1.0.0h [12 Mar 2012]
9033
9034 * Fix MMA (Bleichenbacher's attack on PKCS #1 v1.5 RSA padding) weakness
9035   in CMS and PKCS7 code. When RSA decryption fails use a random key for
9036   content decryption and always return the same error. Note: this attack
9037   needs on average 2^20 messages so it only affects automated senders. The
9038   old behaviour can be re-enabled in the CMS code by setting the
9039   CMS_DEBUG_DECRYPT flag: this is useful for debugging and testing where
9040   an MMA defence is not necessary.
9041   Thanks to Ivan Nestlerode <inestlerode@us.ibm.com> for discovering
9042   this issue. ([CVE-2012-0884])
9043
9044   *Steve Henson*
9045
9046 * Fix CVE-2011-4619: make sure we really are receiving a
9047   client hello before rejecting multiple SGC restarts. Thanks to
9048   Ivan Nestlerode <inestlerode@us.ibm.com> for discovering this bug.
9049
9050   *Steve Henson*
9051
9052### Changes between 1.0.0f and 1.0.0g [18 Jan 2012]
9053
9054 * Fix for DTLS DoS issue introduced by fix for CVE-2011-4109.
9055   Thanks to Antonio Martin, Enterprise Secure Access Research and
9056   Development, Cisco Systems, Inc. for discovering this bug and
9057   preparing a fix. ([CVE-2012-0050])
9058
9059   *Antonio Martin*
9060
9061### Changes between 1.0.0e and 1.0.0f [4 Jan 2012]
9062
9063 * Nadhem Alfardan and Kenny Paterson have discovered an extension
9064   of the Vaudenay padding oracle attack on CBC mode encryption
9065   which enables an efficient plaintext recovery attack against
9066   the OpenSSL implementation of DTLS. Their attack exploits timing
9067   differences arising during decryption processing. A research
9068   paper describing this attack can be found at:
9069   <http://www.isg.rhul.ac.uk/~kp/dtls.pdf>
9070   Thanks go to Nadhem Alfardan and Kenny Paterson of the Information
9071   Security Group at Royal Holloway, University of London
9072   (www.isg.rhul.ac.uk) for discovering this flaw and to Robin Seggelmann
9073   <seggelmann@fh-muenster.de> and Michael Tuexen <tuexen@fh-muenster.de>
9074   for preparing the fix. ([CVE-2011-4108])
9075
9076   *Robin Seggelmann, Michael Tuexen*
9077
9078 * Clear bytes used for block padding of SSL 3.0 records.
9079   ([CVE-2011-4576])
9080
9081   *Adam Langley (Google)*
9082
9083 * Only allow one SGC handshake restart for SSL/TLS. Thanks to George
9084   Kadianakis <desnacked@gmail.com> for discovering this issue and
9085   Adam Langley for preparing the fix. ([CVE-2011-4619])
9086
9087   *Adam Langley (Google)*
9088
9089 * Check parameters are not NULL in GOST ENGINE. ([CVE-2012-0027])
9090
9091   *Andrey Kulikov <amdeich@gmail.com>*
9092
9093 * Prevent malformed RFC3779 data triggering an assertion failure.
9094   Thanks to Andrew Chi, BBN Technologies, for discovering the flaw
9095   and Rob Austein <sra@hactrn.net> for fixing it. ([CVE-2011-4577])
9096
9097   *Rob Austein <sra@hactrn.net>*
9098
9099 * Improved PRNG seeding for VOS.
9100
9101   *Paul Green <Paul.Green@stratus.com>*
9102
9103 * Fix ssl_ciph.c set-up race.
9104
9105   *Adam Langley (Google)*
9106
9107 * Fix spurious failures in ecdsatest.c.
9108
9109   *Emilia Käsper (Google)*
9110
9111 * Fix the BIO_f_buffer() implementation (which was mixing different
9112   interpretations of the `..._len` fields).
9113
9114   *Adam Langley (Google)*
9115
9116 * Fix handling of BN_BLINDING: now BN_BLINDING_invert_ex (rather than
9117   BN_BLINDING_invert_ex) calls BN_BLINDING_update, ensuring that concurrent
9118   threads won't reuse the same blinding coefficients.
9119
9120   This also avoids the need to obtain the CRYPTO_LOCK_RSA_BLINDING
9121   lock to call BN_BLINDING_invert_ex, and avoids one use of
9122   BN_BLINDING_update for each BN_BLINDING structure (previously,
9123   the last update always remained unused).
9124
9125   *Emilia Käsper (Google)*
9126
9127 * In ssl3_clear, preserve s3->init_extra along with s3->rbuf.
9128
9129   *Bob Buckholz (Google)*
9130
9131### Changes between 1.0.0d and 1.0.0e [6 Sep 2011]
9132
9133 * Fix bug where CRLs with nextUpdate in the past are sometimes accepted
9134   by initialising X509_STORE_CTX properly. ([CVE-2011-3207])
9135
9136   *Kaspar Brand <ossl@velox.ch>*
9137
9138 * Fix SSL memory handling for (EC)DH ciphersuites, in particular
9139   for multi-threaded use of ECDH. ([CVE-2011-3210])
9140
9141   *Adam Langley (Google)*
9142
9143 * Fix x509_name_ex_d2i memory leak on bad inputs.
9144
9145   *Bodo Moeller*
9146
9147 * Remove hard coded ecdsaWithSHA1 signature tests in ssl code and check
9148   signature public key algorithm by using OID xref utilities instead.
9149   Before this you could only use some ECC ciphersuites with SHA1 only.
9150
9151   *Steve Henson*
9152
9153 * Add protection against ECDSA timing attacks as mentioned in the paper
9154   by Billy Bob Brumley and Nicola Tuveri, see:
9155   <http://eprint.iacr.org/2011/232.pdf>
9156
9157   *Billy Bob Brumley and Nicola Tuveri*
9158
9159### Changes between 1.0.0c and 1.0.0d [8 Feb 2011]
9160
9161 * Fix parsing of OCSP stapling ClientHello extension. CVE-2011-0014
9162
9163   *Neel Mehta, Adam Langley, Bodo Moeller (Google)*
9164
9165 * Fix bug in string printing code: if *any* escaping is enabled we must
9166   escape the escape character (backslash) or the resulting string is
9167   ambiguous.
9168
9169   *Steve Henson*
9170
9171### Changes between 1.0.0b and 1.0.0c  [2 Dec 2010]
9172
9173 * Disable code workaround for ancient and obsolete Netscape browsers
9174   and servers: an attacker can use it in a ciphersuite downgrade attack.
9175   Thanks to Martin Rex for discovering this bug. CVE-2010-4180
9176
9177   *Steve Henson*
9178
9179 * Fixed J-PAKE implementation error, originally discovered by
9180   Sebastien Martini, further info and confirmation from Stefan
9181   Arentz and Feng Hao. Note that this fix is a security fix. CVE-2010-4252
9182
9183   *Ben Laurie*
9184
9185### Changes between 1.0.0a and 1.0.0b  [16 Nov 2010]
9186
9187 * Fix extension code to avoid race conditions which can result in a buffer
9188   overrun vulnerability: resumed sessions must not be modified as they can
9189   be shared by multiple threads. CVE-2010-3864
9190
9191   *Steve Henson*
9192
9193 * Fix WIN32 build system to correctly link an ENGINE directory into
9194   a DLL.
9195
9196   *Steve Henson*
9197
9198### Changes between 1.0.0 and 1.0.0a  [01 Jun 2010]
9199
9200 * Check return value of int_rsa_verify in pkey_rsa_verifyrecover
9201   ([CVE-2010-1633])
9202
9203   *Steve Henson, Peter-Michael Hager <hager@dortmund.net>*
9204
9205### Changes between 0.9.8n and 1.0.0  [29 Mar 2010]
9206
9207 * Add "missing" function EVP_CIPHER_CTX_copy(). This copies a cipher
9208   context. The operation can be customised via the ctrl mechanism in
9209   case ENGINEs want to include additional functionality.
9210
9211   *Steve Henson*
9212
9213 * Tolerate yet another broken PKCS#8 key format: private key value negative.
9214
9215   *Steve Henson*
9216
9217 * Add new -subject_hash_old and -issuer_hash_old options to x509 utility to
9218   output hashes compatible with older versions of OpenSSL.
9219
9220   *Willy Weisz <weisz@vcpc.univie.ac.at>*
9221
9222 * Fix compression algorithm handling: if resuming a session use the
9223   compression algorithm of the resumed session instead of determining
9224   it from client hello again. Don't allow server to change algorithm.
9225
9226   *Steve Henson*
9227
9228 * Add load_crls() function to commands tidying load_certs() too. Add option
9229   to verify utility to allow additional CRLs to be included.
9230
9231   *Steve Henson*
9232
9233 * Update OCSP request code to permit adding custom headers to the request:
9234   some responders need this.
9235
9236   *Steve Henson*
9237
9238 * The function EVP_PKEY_sign() returns <=0 on error: check return code
9239   correctly.
9240
9241   *Julia Lawall <julia@diku.dk>*
9242
9243 * Update verify callback code in `apps/s_cb.c` and `apps/verify.c`, it
9244   needlessly dereferenced structures, used obsolete functions and
9245   didn't handle all updated verify codes correctly.
9246
9247   *Steve Henson*
9248
9249 * Disable MD2 in the default configuration.
9250
9251   *Steve Henson*
9252
9253 * In BIO_pop() and BIO_push() use the ctrl argument (which was NULL) to
9254   indicate the initial BIO being pushed or popped. This makes it possible
9255   to determine whether the BIO is the one explicitly called or as a result
9256   of the ctrl being passed down the chain. Fix BIO_pop() and SSL BIOs so
9257   it handles reference counts correctly and doesn't zero out the I/O bio
9258   when it is not being explicitly popped. WARNING: applications which
9259   included workarounds for the old buggy behaviour will need to be modified
9260   or they could free up already freed BIOs.
9261
9262   *Steve Henson*
9263
9264 * Extend the uni2asc/asc2uni => OPENSSL_uni2asc/OPENSSL_asc2uni
9265   renaming to all platforms (within the 0.9.8 branch, this was
9266   done conditionally on Netware platforms to avoid a name clash).
9267
9268   *Guenter <lists@gknw.net>*
9269
9270 * Add ECDHE and PSK support to DTLS.
9271
9272   *Michael Tuexen <tuexen@fh-muenster.de>*
9273
9274 * Add CHECKED_STACK_OF macro to safestack.h, otherwise safestack can't
9275   be used on C++.
9276
9277   *Steve Henson*
9278
9279 * Add "missing" function EVP_MD_flags() (without this the only way to
9280   retrieve a digest flags is by accessing the structure directly. Update
9281   `EVP_MD_do_all*()` and `EVP_CIPHER_do_all*()` to include the name a digest
9282   or cipher is registered as in the "from" argument. Print out all
9283   registered digests in the dgst usage message instead of manually
9284   attempting to work them out.
9285
9286   *Steve Henson*
9287
9288 * If no SSLv2 ciphers are used don't use an SSLv2 compatible client hello:
9289   this allows the use of compression and extensions. Change default cipher
9290   string to remove SSLv2 ciphersuites. This effectively avoids ancient SSLv2
9291   by default unless an application cipher string requests it.
9292
9293   *Steve Henson*
9294
9295 * Alter match criteria in PKCS12_parse(). It used to try to use local
9296   key ids to find matching certificates and keys but some PKCS#12 files
9297   don't follow the (somewhat unwritten) rules and this strategy fails.
9298   Now just gather all certificates together and the first private key
9299   then look for the first certificate that matches the key.
9300
9301   *Steve Henson*
9302
9303 * Support use of registered digest and cipher names for dgst and cipher
9304   commands instead of having to add each one as a special case. So now
9305   you can do:
9306
9307           openssl sha256 foo
9308
9309   as well as:
9310
9311           openssl dgst -sha256 foo
9312
9313   and this works for ENGINE based algorithms too.
9314
9315   *Steve Henson*
9316
9317 * Update Gost ENGINE to support parameter files.
9318
9319   *Victor B. Wagner <vitus@cryptocom.ru>*
9320
9321 * Support GeneralizedTime in ca utility.
9322
9323   *Oliver Martin <oliver@volatilevoid.net>, Steve Henson*
9324
9325 * Enhance the hash format used for certificate directory links. The new
9326   form uses the canonical encoding (meaning equivalent names will work
9327   even if they aren't identical) and uses SHA1 instead of MD5. This form
9328   is incompatible with the older format and as a result c_rehash should
9329   be used to rebuild symbolic links.
9330
9331   *Steve Henson*
9332
9333 * Make PKCS#8 the default write format for private keys, replacing the
9334   traditional format. This form is standardised, more secure and doesn't
9335   include an implicit MD5 dependency.
9336
9337   *Steve Henson*
9338
9339 * Add a $gcc_devteam_warn option to Configure. The idea is that any code
9340   committed to OpenSSL should pass this lot as a minimum.
9341
9342   *Steve Henson*
9343
9344 * Add session ticket override functionality for use by EAP-FAST.
9345
9346   *Jouni Malinen <j@w1.fi>*
9347
9348 * Modify HMAC functions to return a value. Since these can be implemented
9349   in an ENGINE errors can occur.
9350
9351   *Steve Henson*
9352
9353 * Type-checked OBJ_bsearch_ex.
9354
9355   *Ben Laurie*
9356
9357 * Type-checked OBJ_bsearch. Also some constification necessitated
9358   by type-checking.  Still to come: TXT_DB, bsearch(?),
9359   OBJ_bsearch_ex, qsort, CRYPTO_EX_DATA, ASN1_VALUE, ASN1_STRING,
9360   CONF_VALUE.
9361
9362   *Ben Laurie*
9363
9364 * New function OPENSSL_gmtime_adj() to add a specific number of days and
9365   seconds to a tm structure directly, instead of going through OS
9366   specific date routines. This avoids any issues with OS routines such
9367   as the year 2038 bug. New `*_adj()` functions for ASN1 time structures
9368   and X509_time_adj_ex() to cover the extended range. The existing
9369   X509_time_adj() is still usable and will no longer have any date issues.
9370
9371   *Steve Henson*
9372
9373 * Delta CRL support. New use deltas option which will attempt to locate
9374   and search any appropriate delta CRLs available.
9375
9376   This work was sponsored by Google.
9377
9378   *Steve Henson*
9379
9380 * Support for CRLs partitioned by reason code. Reorganise CRL processing
9381   code and add additional score elements. Validate alternate CRL paths
9382   as part of the CRL checking and indicate a new error "CRL path validation
9383   error" in this case. Applications wanting additional details can use
9384   the verify callback and check the new "parent" field. If this is not
9385   NULL CRL path validation is taking place. Existing applications won't
9386   see this because it requires extended CRL support which is off by
9387   default.
9388
9389   This work was sponsored by Google.
9390
9391   *Steve Henson*
9392
9393 * Support for freshest CRL extension.
9394
9395   This work was sponsored by Google.
9396
9397   *Steve Henson*
9398
9399 * Initial indirect CRL support. Currently only supported in the CRLs
9400   passed directly and not via lookup. Process certificate issuer
9401   CRL entry extension and lookup CRL entries by bother issuer name
9402   and serial number. Check and process CRL issuer entry in IDP extension.
9403
9404   This work was sponsored by Google.
9405
9406   *Steve Henson*
9407
9408 * Add support for distinct certificate and CRL paths. The CRL issuer
9409   certificate is validated separately in this case. Only enabled if
9410   an extended CRL support flag is set: this flag will enable additional
9411   CRL functionality in future.
9412
9413   This work was sponsored by Google.
9414
9415   *Steve Henson*
9416
9417 * Add support for policy mappings extension.
9418
9419   This work was sponsored by Google.
9420
9421   *Steve Henson*
9422
9423 * Fixes to pathlength constraint, self issued certificate handling,
9424   policy processing to align with RFC3280 and PKITS tests.
9425
9426   This work was sponsored by Google.
9427
9428   *Steve Henson*
9429
9430 * Support for name constraints certificate extension. DN, email, DNS
9431   and URI types are currently supported.
9432
9433   This work was sponsored by Google.
9434
9435   *Steve Henson*
9436
9437 * To cater for systems that provide a pointer-based thread ID rather
9438   than numeric, deprecate the current numeric thread ID mechanism and
9439   replace it with a structure and associated callback type. This
9440   mechanism allows a numeric "hash" to be extracted from a thread ID in
9441   either case, and on platforms where pointers are larger than 'long',
9442   mixing is done to help ensure the numeric 'hash' is usable even if it
9443   can't be guaranteed unique. The default mechanism is to use "&errno"
9444   as a pointer-based thread ID to distinguish between threads.
9445
9446   Applications that want to provide their own thread IDs should now use
9447   CRYPTO_THREADID_set_callback() to register a callback that will call
9448   either CRYPTO_THREADID_set_numeric() or CRYPTO_THREADID_set_pointer().
9449
9450   Note that ERR_remove_state() is now deprecated, because it is tied
9451   to the assumption that thread IDs are numeric.  ERR_remove_state(0)
9452   to free the current thread's error state should be replaced by
9453   ERR_remove_thread_state(NULL).
9454
9455   (This new approach replaces the functions CRYPTO_set_idptr_callback(),
9456   CRYPTO_get_idptr_callback(), and CRYPTO_thread_idptr() that existed in
9457   OpenSSL 0.9.9-dev between June 2006 and August 2008. Also, if an
9458   application was previously providing a numeric thread callback that
9459   was inappropriate for distinguishing threads, then uniqueness might
9460   have been obtained with &errno that happened immediately in the
9461   intermediate development versions of OpenSSL; this is no longer the
9462   case, the numeric thread callback will now override the automatic use
9463   of &errno.)
9464
9465   *Geoff Thorpe, with help from Bodo Moeller*
9466
9467 * Initial support for different CRL issuing certificates. This covers a
9468   simple case where the self issued certificates in the chain exist and
9469   the real CRL issuer is higher in the existing chain.
9470
9471   This work was sponsored by Google.
9472
9473   *Steve Henson*
9474
9475 * Removed effectively defunct crypto/store from the build.
9476
9477   *Ben Laurie*
9478
9479 * Revamp of STACK to provide stronger type-checking. Still to come:
9480   TXT_DB, bsearch(?), OBJ_bsearch, qsort, CRYPTO_EX_DATA, ASN1_VALUE,
9481   ASN1_STRING, CONF_VALUE.
9482
9483   *Ben Laurie*
9484
9485 * Add a new SSL_MODE_RELEASE_BUFFERS mode flag to release unused buffer
9486   RAM on SSL connections.  This option can save about 34k per idle SSL.
9487
9488   *Nick Mathewson*
9489
9490 * Revamp of LHASH to provide stronger type-checking. Still to come:
9491   STACK, TXT_DB, bsearch, qsort.
9492
9493   *Ben Laurie*
9494
9495 * Initial support for Cryptographic Message Syntax (aka CMS) based
9496   on RFC3850, RFC3851 and RFC3852. New cms directory and cms utility,
9497   support for data, signedData, compressedData, digestedData and
9498   encryptedData, envelopedData types included. Scripts to check against
9499   RFC4134 examples draft and interop and consistency checks of many
9500   content types and variants.
9501
9502   *Steve Henson*
9503
9504 * Add options to enc utility to support use of zlib compression BIO.
9505
9506   *Steve Henson*
9507
9508 * Extend mk1mf to support importing of options and assembly language
9509   files from Configure script, currently only included in VC-WIN32.
9510   The assembly language rules can now optionally generate the source
9511   files from the associated perl scripts.
9512
9513   *Steve Henson*
9514
9515 * Implement remaining functionality needed to support GOST ciphersuites.
9516   Interop testing has been performed using CryptoPro implementations.
9517
9518   *Victor B. Wagner <vitus@cryptocom.ru>*
9519
9520 * s390x assembler pack.
9521
9522   *Andy Polyakov*
9523
9524 * ARMv4 assembler pack. ARMv4 refers to v4 and later ISA, not CPU
9525   "family."
9526
9527   *Andy Polyakov*
9528
9529 * Implement Opaque PRF Input TLS extension as specified in
9530   draft-rescorla-tls-opaque-prf-input-00.txt.  Since this is not an
9531   official specification yet and no extension type assignment by
9532   IANA exists, this extension (for now) will have to be explicitly
9533   enabled when building OpenSSL by providing the extension number
9534   to use.  For example, specify an option
9535
9536           -DTLSEXT_TYPE_opaque_prf_input=0x9527
9537
9538   to the "config" or "Configure" script to enable the extension,
9539   assuming extension number 0x9527 (which is a completely arbitrary
9540   and unofficial assignment based on the MD5 hash of the Internet
9541   Draft).  Note that by doing so, you potentially lose
9542   interoperability with other TLS implementations since these might
9543   be using the same extension number for other purposes.
9544
9545   SSL_set_tlsext_opaque_prf_input(ssl, src, len) is used to set the
9546   opaque PRF input value to use in the handshake.  This will create
9547   an internal copy of the length-'len' string at 'src', and will
9548   return non-zero for success.
9549
9550   To get more control and flexibility, provide a callback function
9551   by using
9552
9553           SSL_CTX_set_tlsext_opaque_prf_input_callback(ctx, cb)
9554           SSL_CTX_set_tlsext_opaque_prf_input_callback_arg(ctx, arg)
9555
9556   where
9557
9558           int (*cb)(SSL *, void *peerinput, size_t len, void *arg);
9559           void *arg;
9560
9561   Callback function 'cb' will be called in handshakes, and is
9562   expected to use SSL_set_tlsext_opaque_prf_input() as appropriate.
9563   Argument 'arg' is for application purposes (the value as given to
9564   SSL_CTX_set_tlsext_opaque_prf_input_callback_arg() will directly
9565   be provided to the callback function).  The callback function
9566   has to return non-zero to report success: usually 1 to use opaque
9567   PRF input just if possible, or 2 to enforce use of the opaque PRF
9568   input.  In the latter case, the library will abort the handshake
9569   if opaque PRF input is not successfully negotiated.
9570
9571   Arguments 'peerinput' and 'len' given to the callback function
9572   will always be NULL and 0 in the case of a client.  A server will
9573   see the client's opaque PRF input through these variables if
9574   available (NULL and 0 otherwise).  Note that if the server
9575   provides an opaque PRF input, the length must be the same as the
9576   length of the client's opaque PRF input.
9577
9578   Note that the callback function will only be called when creating
9579   a new session (session resumption can resume whatever was
9580   previously negotiated), and will not be called in SSL 2.0
9581   handshakes; thus, SSL_CTX_set_options(ctx, SSL_OP_NO_SSLv2) or
9582   SSL_set_options(ssl, SSL_OP_NO_SSLv2) is especially recommended
9583   for applications that need to enforce opaque PRF input.
9584
9585   *Bodo Moeller*
9586
9587 * Update ssl code to support digests other than SHA1+MD5 for handshake
9588   MAC.
9589
9590   *Victor B. Wagner <vitus@cryptocom.ru>*
9591
9592 * Add RFC4507 support to OpenSSL. This includes the corrections in
9593   RFC4507bis. The encrypted ticket format is an encrypted encoded
9594   SSL_SESSION structure, that way new session features are automatically
9595   supported.
9596
9597   If a client application caches session in an SSL_SESSION structure
9598   support is transparent because tickets are now stored in the encoded
9599   SSL_SESSION.
9600
9601   The SSL_CTX structure automatically generates keys for ticket
9602   protection in servers so again support should be possible
9603   with no application modification.
9604
9605   If a client or server wishes to disable RFC4507 support then the option
9606   SSL_OP_NO_TICKET can be set.
9607
9608   Add a TLS extension debugging callback to allow the contents of any client
9609   or server extensions to be examined.
9610
9611   This work was sponsored by Google.
9612
9613   *Steve Henson*
9614
9615 * Final changes to avoid use of pointer pointer casts in OpenSSL.
9616   OpenSSL should now compile cleanly on gcc 4.2
9617
9618   *Peter Hartley <pdh@utter.chaos.org.uk>, Steve Henson*
9619
9620 * Update SSL library to use new EVP_PKEY MAC API. Include generic MAC
9621   support including streaming MAC support: this is required for GOST
9622   ciphersuite support.
9623
9624   *Victor B. Wagner <vitus@cryptocom.ru>, Steve Henson*
9625
9626 * Add option -stream to use PKCS#7 streaming in smime utility. New
9627   function i2d_PKCS7_bio_stream() and PEM_write_PKCS7_bio_stream()
9628   to output in BER and PEM format.
9629
9630   *Steve Henson*
9631
9632 * Experimental support for use of HMAC via EVP_PKEY interface. This
9633   allows HMAC to be handled via the `EVP_DigestSign*()` interface. The
9634   EVP_PKEY "key" in this case is the HMAC key, potentially allowing
9635   ENGINE support for HMAC keys which are unextractable. New -mac and
9636   -macopt options to dgst utility.
9637
9638   *Steve Henson*
9639
9640 * New option -sigopt to dgst utility. Update dgst to use
9641   `EVP_Digest{Sign,Verify}*`. These two changes make it possible to use
9642   alternative signing parameters such as X9.31 or PSS in the dgst
9643   utility.
9644
9645   *Steve Henson*
9646
9647 * Change ssl_cipher_apply_rule(), the internal function that does
9648   the work each time a ciphersuite string requests enabling
9649   ("foo+bar"), moving ("+foo+bar"), disabling ("-foo+bar", or
9650   removing ("!foo+bar") a class of ciphersuites: Now it maintains
9651   the order of disabled ciphersuites such that those ciphersuites
9652   that most recently went from enabled to disabled not only stay
9653   in order with respect to each other, but also have higher priority
9654   than other disabled ciphersuites the next time ciphersuites are
9655   enabled again.
9656
9657   This means that you can now say, e.g., "PSK:-PSK:HIGH" to enable
9658   the same ciphersuites as with "HIGH" alone, but in a specific
9659   order where the PSK ciphersuites come first (since they are the
9660   most recently disabled ciphersuites when "HIGH" is parsed).
9661
9662   Also, change ssl_create_cipher_list() (using this new
9663   functionality) such that between otherwise identical
9664   ciphersuites, ephemeral ECDH is preferred over ephemeral DH in
9665   the default order.
9666
9667   *Bodo Moeller*
9668
9669 * Change ssl_create_cipher_list() so that it automatically
9670   arranges the ciphersuites in reasonable order before starting
9671   to process the rule string.  Thus, the definition for "DEFAULT"
9672   (SSL_DEFAULT_CIPHER_LIST) now is just "ALL:!aNULL:!eNULL", but
9673   remains equivalent to `"AES:ALL:!aNULL:!eNULL:+aECDH:+kRSA:+RC4:@STRENGTH"`.
9674   This makes it much easier to arrive at a reasonable default order
9675   in applications for which anonymous ciphers are OK (meaning
9676   that you can't actually use DEFAULT).
9677
9678   *Bodo Moeller; suggested by Victor Duchovni*
9679
9680 * Split the SSL/TLS algorithm mask (as used for ciphersuite string
9681   processing) into multiple integers instead of setting
9682   "SSL_MKEY_MASK" bits, "SSL_AUTH_MASK" bits, "SSL_ENC_MASK",
9683   "SSL_MAC_MASK", and "SSL_SSL_MASK" bits all in a single integer.
9684   (These masks as well as the individual bit definitions are hidden
9685   away into the non-exported interface ssl/ssl_locl.h, so this
9686   change to the definition of the SSL_CIPHER structure shouldn't
9687   affect applications.)  This give us more bits for each of these
9688   categories, so there is no longer a need to coagulate AES128 and
9689   AES256 into a single algorithm bit, and to coagulate Camellia128
9690   and Camellia256 into a single algorithm bit, which has led to all
9691   kinds of kludges.
9692
9693   Thus, among other things, the kludge introduced in 0.9.7m and
9694   0.9.8e for masking out AES256 independently of AES128 or masking
9695   out Camellia256 independently of AES256 is not needed here in 0.9.9.
9696
9697   With the change, we also introduce new ciphersuite aliases that
9698   so far were missing: "AES128", "AES256", "CAMELLIA128", and
9699   "CAMELLIA256".
9700
9701   *Bodo Moeller*
9702
9703 * Add support for dsa-with-SHA224 and dsa-with-SHA256.
9704   Use the leftmost N bytes of the signature input if the input is
9705   larger than the prime q (with N being the size in bytes of q).
9706
9707   *Nils Larsch*
9708
9709 * Very *very* experimental PKCS#7 streaming encoder support. Nothing uses
9710   it yet and it is largely untested.
9711
9712   *Steve Henson*
9713
9714 * Add support for the ecdsa-with-SHA224/256/384/512 signature types.
9715
9716   *Nils Larsch*
9717
9718 * Initial incomplete changes to avoid need for function casts in OpenSSL
9719   some compilers (gcc 4.2 and later) reject their use. Safestack is
9720   reimplemented.  Update ASN1 to avoid use of legacy functions.
9721
9722   *Steve Henson*
9723
9724 * Win32/64 targets are linked with Winsock2.
9725
9726   *Andy Polyakov*
9727
9728 * Add an X509_CRL_METHOD structure to allow CRL processing to be redirected
9729   to external functions. This can be used to increase CRL handling
9730   efficiency especially when CRLs are very large by (for example) storing
9731   the CRL revoked certificates in a database.
9732
9733   *Steve Henson*
9734
9735 * Overhaul of by_dir code. Add support for dynamic loading of CRLs so
9736   new CRLs added to a directory can be used. New command line option
9737   -verify_return_error to s_client and s_server. This causes real errors
9738   to be returned by the verify callback instead of carrying on no matter
9739   what. This reflects the way a "real world" verify callback would behave.
9740
9741   *Steve Henson*
9742
9743 * GOST engine, supporting several GOST algorithms and public key formats.
9744   Kindly donated by Cryptocom.
9745
9746   *Cryptocom*
9747
9748 * Partial support for Issuing Distribution Point CRL extension. CRLs
9749   partitioned by DP are handled but no indirect CRL or reason partitioning
9750   (yet). Complete overhaul of CRL handling: now the most suitable CRL is
9751   selected via a scoring technique which handles IDP and AKID in CRLs.
9752
9753   *Steve Henson*
9754
9755 * New X509_STORE_CTX callbacks lookup_crls() and lookup_certs() which
9756   will ultimately be used for all verify operations: this will remove the
9757   X509_STORE dependency on certificate verification and allow alternative
9758   lookup methods.  X509_STORE based implementations of these two callbacks.
9759
9760   *Steve Henson*
9761
9762 * Allow multiple CRLs to exist in an X509_STORE with matching issuer names.
9763   Modify get_crl() to find a valid (unexpired) CRL if possible.
9764
9765   *Steve Henson*
9766
9767 * New function X509_CRL_match() to check if two CRLs are identical. Normally
9768   this would be called X509_CRL_cmp() but that name is already used by
9769   a function that just compares CRL issuer names. Cache several CRL
9770   extensions in X509_CRL structure and cache CRLDP in X509.
9771
9772   *Steve Henson*
9773
9774 * Store a "canonical" representation of X509_NAME structure (ASN1 Name)
9775   this maps equivalent X509_NAME structures into a consistent structure.
9776   Name comparison can then be performed rapidly using memcmp().
9777
9778   *Steve Henson*
9779
9780 * Non-blocking OCSP request processing. Add -timeout option to ocsp
9781   utility.
9782
9783   *Steve Henson*
9784
9785 * Allow digests to supply their own micalg string for S/MIME type using
9786   the ctrl EVP_MD_CTRL_MICALG.
9787
9788   *Steve Henson*
9789
9790 * During PKCS7 signing pass the PKCS7 SignerInfo structure to the
9791   EVP_PKEY_METHOD before and after signing via the EVP_PKEY_CTRL_PKCS7_SIGN
9792   ctrl. It can then customise the structure before and/or after signing
9793   if necessary.
9794
9795   *Steve Henson*
9796
9797 * New function OBJ_add_sigid() to allow application defined signature OIDs
9798   to be added to OpenSSLs internal tables. New function OBJ_sigid_free()
9799   to free up any added signature OIDs.
9800
9801   *Steve Henson*
9802
9803 * New functions EVP_CIPHER_do_all(), EVP_CIPHER_do_all_sorted(),
9804   EVP_MD_do_all() and EVP_MD_do_all_sorted() to enumerate internal
9805   digest and cipher tables. New options added to openssl utility:
9806   list-message-digest-algorithms and list-cipher-algorithms.
9807
9808   *Steve Henson*
9809
9810 * Change the array representation of binary polynomials: the list
9811   of degrees of non-zero coefficients is now terminated with -1.
9812   Previously it was terminated with 0, which was also part of the
9813   value; thus, the array representation was not applicable to
9814   polynomials where t^0 has coefficient zero.  This change makes
9815   the array representation useful in a more general context.
9816
9817   *Douglas Stebila*
9818
9819 * Various modifications and fixes to SSL/TLS cipher string
9820   handling.  For ECC, the code now distinguishes between fixed ECDH
9821   with RSA certificates on the one hand and with ECDSA certificates
9822   on the other hand, since these are separate ciphersuites.  The
9823   unused code for Fortezza ciphersuites has been removed.
9824
9825   For consistency with EDH, ephemeral ECDH is now called "EECDH"
9826   (not "ECDHE").  For consistency with the code for DH
9827   certificates, use of ECDH certificates is now considered ECDH
9828   authentication, not RSA or ECDSA authentication (the latter is
9829   merely the CA's signing algorithm and not actively used in the
9830   protocol).
9831
9832   The temporary ciphersuite alias "ECCdraft" is no longer
9833   available, and ECC ciphersuites are no longer excluded from "ALL"
9834   and "DEFAULT".  The following aliases now exist for RFC 4492
9835   ciphersuites, most of these by analogy with the DH case:
9836
9837           kECDHr   - ECDH cert, signed with RSA
9838           kECDHe   - ECDH cert, signed with ECDSA
9839           kECDH    - ECDH cert (signed with either RSA or ECDSA)
9840           kEECDH   - ephemeral ECDH
9841           ECDH     - ECDH cert or ephemeral ECDH
9842
9843           aECDH    - ECDH cert
9844           aECDSA   - ECDSA cert
9845           ECDSA    - ECDSA cert
9846
9847           AECDH    - anonymous ECDH
9848           EECDH    - non-anonymous ephemeral ECDH (equivalent to "kEECDH:-AECDH")
9849
9850   *Bodo Moeller*
9851
9852 * Add additional S/MIME capabilities for AES and GOST ciphers if supported.
9853   Use correct micalg parameters depending on digest(s) in signed message.
9854
9855   *Steve Henson*
9856
9857 * Add engine support for EVP_PKEY_ASN1_METHOD. Add functions to process
9858   an ENGINE asn1 method. Support ENGINE lookups in the ASN1 code.
9859
9860   *Steve Henson*
9861
9862 * Initial engine support for EVP_PKEY_METHOD. New functions to permit
9863   an engine to register a method. Add ENGINE lookups for methods and
9864   functional reference processing.
9865
9866   *Steve Henson*
9867
9868 * New functions `EVP_Digest{Sign,Verify)*`. These are enhanced versions of
9869   `EVP_{Sign,Verify}*` which allow an application to customise the signature
9870   process.
9871
9872   *Steve Henson*
9873
9874 * New -resign option to smime utility. This adds one or more signers
9875   to an existing PKCS#7 signedData structure. Also -md option to use an
9876   alternative message digest algorithm for signing.
9877
9878   *Steve Henson*
9879
9880 * Tidy up PKCS#7 routines and add new functions to make it easier to
9881   create PKCS7 structures containing multiple signers. Update smime
9882   application to support multiple signers.
9883
9884   *Steve Henson*
9885
9886 * New -macalg option to pkcs12 utility to allow setting of an alternative
9887   digest MAC.
9888
9889   *Steve Henson*
9890
9891 * Initial support for PKCS#5 v2.0 PRFs other than default SHA1 HMAC.
9892   Reorganize PBE internals to lookup from a static table using NIDs,
9893   add support for HMAC PBE OID translation. Add a EVP_CIPHER ctrl:
9894   EVP_CTRL_PBE_PRF_NID this allows a cipher to specify an alternative
9895   PRF which will be automatically used with PBES2.
9896
9897   *Steve Henson*
9898
9899 * Replace the algorithm specific calls to generate keys in "req" with the
9900   new API.
9901
9902   *Steve Henson*
9903
9904 * Update PKCS#7 enveloped data routines to use new API. This is now
9905   supported by any public key method supporting the encrypt operation. A
9906   ctrl is added to allow the public key algorithm to examine or modify
9907   the PKCS#7 RecipientInfo structure if it needs to: for RSA this is
9908   a no op.
9909
9910   *Steve Henson*
9911
9912 * Add a ctrl to asn1 method to allow a public key algorithm to express
9913   a default digest type to use. In most cases this will be SHA1 but some
9914   algorithms (such as GOST) need to specify an alternative digest. The
9915   return value indicates how strong the preference is 1 means optional and
9916   2 is mandatory (that is it is the only supported type). Modify
9917   ASN1_item_sign() to accept a NULL digest argument to indicate it should
9918   use the default md. Update openssl utilities to use the default digest
9919   type for signing if it is not explicitly indicated.
9920
9921   *Steve Henson*
9922
9923 * Use OID cross reference table in ASN1_sign() and ASN1_verify(). New
9924   EVP_MD flag EVP_MD_FLAG_PKEY_METHOD_SIGNATURE. This uses the relevant
9925   signing method from the key type. This effectively removes the link
9926   between digests and public key types.
9927
9928   *Steve Henson*
9929
9930 * Add an OID cross reference table and utility functions. Its purpose is to
9931   translate between signature OIDs such as SHA1WithrsaEncryption and SHA1,
9932   rsaEncryption. This will allow some of the algorithm specific hackery
9933   needed to use the correct OID to be removed.
9934
9935   *Steve Henson*
9936
9937 * Remove algorithm specific dependencies when setting PKCS7_SIGNER_INFO
9938   structures for PKCS7_sign(). They are now set up by the relevant public
9939   key ASN1 method.
9940
9941   *Steve Henson*
9942
9943 * Add provisional EC pkey method with support for ECDSA and ECDH.
9944
9945   *Steve Henson*
9946
9947 * Add support for key derivation (agreement) in the API, DH method and
9948   pkeyutl.
9949
9950   *Steve Henson*
9951
9952 * Add DSA pkey method and DH pkey methods, extend DH ASN1 method to support
9953   public and private key formats. As a side effect these add additional
9954   command line functionality not previously available: DSA signatures can be
9955   generated and verified using pkeyutl and DH key support and generation in
9956   pkey, genpkey.
9957
9958   *Steve Henson*
9959
9960 * BeOS support.
9961
9962   *Oliver Tappe <zooey@hirschkaefer.de>*
9963
9964 * New make target "install_html_docs" installs HTML renditions of the
9965   manual pages.
9966
9967   *Oliver Tappe <zooey@hirschkaefer.de>*
9968
9969 * New utility "genpkey" this is analogous to "genrsa" etc except it can
9970   generate keys for any algorithm. Extend and update EVP_PKEY_METHOD to
9971   support key and parameter generation and add initial key generation
9972   functionality for RSA.
9973
9974   *Steve Henson*
9975
9976 * Add functions for main EVP_PKEY_method operations. The undocumented
9977   functions `EVP_PKEY_{encrypt,decrypt}` have been renamed to
9978   `EVP_PKEY_{encrypt,decrypt}_old`.
9979
9980   *Steve Henson*
9981
9982 * Initial definitions for EVP_PKEY_METHOD. This will be a high level public
9983   key API, doesn't do much yet.
9984
9985   *Steve Henson*
9986
9987 * New function EVP_PKEY_asn1_get0_info() to retrieve information about
9988   public key algorithms. New option to openssl utility:
9989   "list-public-key-algorithms" to print out info.
9990
9991   *Steve Henson*
9992
9993 * Implement the Supported Elliptic Curves Extension for
9994   ECC ciphersuites from draft-ietf-tls-ecc-12.txt.
9995
9996   *Douglas Stebila*
9997
9998 * Don't free up OIDs in OBJ_cleanup() if they are in use by EVP_MD or
9999   EVP_CIPHER structures to avoid later problems in EVP_cleanup().
10000
10001   *Steve Henson*
10002
10003 * New utilities pkey and pkeyparam. These are similar to algorithm specific
10004   utilities such as rsa, dsa, dsaparam etc except they process any key
10005   type.
10006
10007   *Steve Henson*
10008
10009 * Transfer public key printing routines to EVP_PKEY_ASN1_METHOD. New
10010   functions EVP_PKEY_print_public(), EVP_PKEY_print_private(),
10011   EVP_PKEY_print_param() to print public key data from an EVP_PKEY
10012   structure.
10013
10014   *Steve Henson*
10015
10016 * Initial support for pluggable public key ASN1.
10017   De-spaghettify the public key ASN1 handling. Move public and private
10018   key ASN1 handling to a new EVP_PKEY_ASN1_METHOD structure. Relocate
10019   algorithm specific handling to a single module within the relevant
10020   algorithm directory. Add functions to allow (near) opaque processing
10021   of public and private key structures.
10022
10023   *Steve Henson*
10024
10025 * Implement the Supported Point Formats Extension for
10026   ECC ciphersuites from draft-ietf-tls-ecc-12.txt.
10027
10028   *Douglas Stebila*
10029
10030 * Add initial support for RFC 4279 PSK TLS ciphersuites. Add members
10031   for the psk identity [hint] and the psk callback functions to the
10032   SSL_SESSION, SSL and SSL_CTX structure.
10033
10034   New ciphersuites:
10035           PSK-RC4-SHA, PSK-3DES-EDE-CBC-SHA, PSK-AES128-CBC-SHA,
10036           PSK-AES256-CBC-SHA
10037
10038   New functions:
10039           SSL_CTX_use_psk_identity_hint
10040           SSL_get_psk_identity_hint
10041           SSL_get_psk_identity
10042           SSL_use_psk_identity_hint
10043
10044   *Mika Kousa and Pasi Eronen of Nokia Corporation*
10045
10046 * Add RFC 3161 compliant time stamp request creation, response generation
10047   and response verification functionality.
10048
10049   *Zoltán Glózik <zglozik@opentsa.org>, The OpenTSA Project*
10050
10051 * Add initial support for TLS extensions, specifically for the server_name
10052   extension so far.  The SSL_SESSION, SSL_CTX, and SSL data structures now
10053   have new members for a host name.  The SSL data structure has an
10054   additional member `SSL_CTX *initial_ctx` so that new sessions can be
10055   stored in that context to allow for session resumption, even after the
10056   SSL has been switched to a new SSL_CTX in reaction to a client's
10057   server_name extension.
10058
10059   New functions (subject to change):
10060
10061           SSL_get_servername()
10062           SSL_get_servername_type()
10063           SSL_set_SSL_CTX()
10064
10065   New CTRL codes and macros (subject to change):
10066
10067           SSL_CTRL_SET_TLSEXT_SERVERNAME_CB
10068                               - SSL_CTX_set_tlsext_servername_callback()
10069           SSL_CTRL_SET_TLSEXT_SERVERNAME_ARG
10070                                    - SSL_CTX_set_tlsext_servername_arg()
10071           SSL_CTRL_SET_TLSEXT_HOSTNAME           - SSL_set_tlsext_host_name()
10072
10073   openssl s_client has a new '-servername ...' option.
10074
10075   openssl s_server has new options '-servername_host ...', '-cert2 ...',
10076   '-key2 ...', '-servername_fatal' (subject to change).  This allows
10077   testing the HostName extension for a specific single host name ('-cert'
10078   and '-key' remain fallbacks for handshakes without HostName
10079   negotiation).  If the unrecognized_name alert has to be sent, this by
10080   default is a warning; it becomes fatal with the '-servername_fatal'
10081   option.
10082
10083   *Peter Sylvester,  Remy Allais, Christophe Renou*
10084
10085 * Whirlpool hash implementation is added.
10086
10087   *Andy Polyakov*
10088
10089 * BIGNUM code on 64-bit SPARCv9 targets is switched from bn(64,64) to
10090   bn(64,32). Because of instruction set limitations it doesn't have
10091   any negative impact on performance. This was done mostly in order
10092   to make it possible to share assembler modules, such as bn_mul_mont
10093   implementations, between 32- and 64-bit builds without hassle.
10094
10095   *Andy Polyakov*
10096
10097 * Move code previously exiled into file crypto/ec/ec2_smpt.c
10098   to ec2_smpl.c, and no longer require the OPENSSL_EC_BIN_PT_COMP
10099   macro.
10100
10101   *Bodo Moeller*
10102
10103 * New candidate for BIGNUM assembler implementation, bn_mul_mont,
10104   dedicated Montgomery multiplication procedure, is introduced.
10105   BN_MONT_CTX is modified to allow bn_mul_mont to reach for higher
10106   "64-bit" performance on certain 32-bit targets.
10107
10108   *Andy Polyakov*
10109
10110 * New option SSL_OP_NO_COMP to disable use of compression selectively
10111   in SSL structures. New SSL ctrl to set maximum send fragment size.
10112   Save memory by setting the I/O buffer sizes dynamically instead of
10113   using the maximum available value.
10114
10115   *Steve Henson*
10116
10117 * New option -V for 'openssl ciphers'. This prints the ciphersuite code
10118   in addition to the text details.
10119
10120   *Bodo Moeller*
10121
10122 * Very, very preliminary EXPERIMENTAL support for printing of general
10123   ASN1 structures. This currently produces rather ugly output and doesn't
10124   handle several customised structures at all.
10125
10126   *Steve Henson*
10127
10128 * Integrated support for PVK file format and some related formats such
10129   as MS PUBLICKEYBLOB and PRIVATEKEYBLOB. Command line switches to support
10130   these in the 'rsa' and 'dsa' utilities.
10131
10132   *Steve Henson*
10133
10134 * Support for PKCS#1 RSAPublicKey format on rsa utility command line.
10135
10136   *Steve Henson*
10137
10138 * Remove the ancient ASN1_METHOD code. This was only ever used in one
10139   place for the (very old) "NETSCAPE" format certificates which are now
10140   handled using new ASN1 code equivalents.
10141
10142   *Steve Henson*
10143
10144 * Let the TLSv1_method() etc. functions return a 'const' SSL_METHOD
10145   pointer and make the SSL_METHOD parameter in SSL_CTX_new,
10146   SSL_CTX_set_ssl_version and SSL_set_ssl_method 'const'.
10147
10148   *Nils Larsch*
10149
10150 * Modify CRL distribution points extension code to print out previously
10151   unsupported fields. Enhance extension setting code to allow setting of
10152   all fields.
10153
10154   *Steve Henson*
10155
10156 * Add print and set support for Issuing Distribution Point CRL extension.
10157
10158   *Steve Henson*
10159
10160 * Change 'Configure' script to enable Camellia by default.
10161
10162   *NTT*
10163
10164OpenSSL 0.9.x
10165-------------
10166
10167### Changes between 0.9.8m and 0.9.8n [24 Mar 2010]
10168
10169 * When rejecting SSL/TLS records due to an incorrect version number, never
10170   update s->server with a new major version number.  As of
10171   - OpenSSL 0.9.8m if 'short' is a 16-bit type,
10172   - OpenSSL 0.9.8f if 'short' is longer than 16 bits,
10173   the previous behavior could result in a read attempt at NULL when
10174   receiving specific incorrect SSL/TLS records once record payload
10175   protection is active.  ([CVE-2010-0740])
10176
10177   *Bodo Moeller, Adam Langley <agl@chromium.org>*
10178
10179 * Fix for CVE-2010-0433 where some kerberos enabled versions of OpenSSL
10180   could be crashed if the relevant tables were not present (e.g. chrooted).
10181
10182   *Tomas Hoger <thoger@redhat.com>*
10183
10184### Changes between 0.9.8l and 0.9.8m [25 Feb 2010]
10185
10186 * Always check bn_wexpand() return values for failure.  ([CVE-2009-3245])
10187
10188   *Martin Olsson, Neel Mehta*
10189
10190 * Fix X509_STORE locking: Every 'objs' access requires a lock (to
10191   accommodate for stack sorting, always a write lock!).
10192
10193   *Bodo Moeller*
10194
10195 * On some versions of WIN32 Heap32Next is very slow. This can cause
10196   excessive delays in the RAND_poll(): over a minute. As a workaround
10197   include a time check in the inner Heap32Next loop too.
10198
10199   *Steve Henson*
10200
10201 * The code that handled flushing of data in SSL/TLS originally used the
10202   BIO_CTRL_INFO ctrl to see if any data was pending first. This caused
10203   the problem outlined in PR#1949. The fix suggested there however can
10204   trigger problems with buggy BIO_CTRL_WPENDING (e.g. some versions
10205   of Apache). So instead simplify the code to flush unconditionally.
10206   This should be fine since flushing with no data to flush is a no op.
10207
10208   *Steve Henson*
10209
10210 * Handle TLS versions 2.0 and later properly and correctly use the
10211   highest version of TLS/SSL supported. Although TLS >= 2.0 is some way
10212   off ancient servers have a habit of sticking around for a while...
10213
10214   *Steve Henson*
10215
10216 * Modify compression code so it frees up structures without using the
10217   ex_data callbacks. This works around a problem where some applications
10218   call CRYPTO_cleanup_all_ex_data() before application exit (e.g. when
10219   restarting) then use compression (e.g. SSL with compression) later.
10220   This results in significant per-connection memory leaks and
10221   has caused some security issues including CVE-2008-1678 and
10222   CVE-2009-4355.
10223
10224   *Steve Henson*
10225
10226 * Constify crypto/cast (i.e., <openssl/cast.h>): a CAST_KEY doesn't
10227   change when encrypting or decrypting.
10228
10229   *Bodo Moeller*
10230
10231 * Add option SSL_OP_LEGACY_SERVER_CONNECT which will allow clients to
10232   connect and renegotiate with servers which do not support RI.
10233   Until RI is more widely deployed this option is enabled by default.
10234
10235   *Steve Henson*
10236
10237 * Add "missing" ssl ctrls to clear options and mode.
10238
10239   *Steve Henson*
10240
10241 * If client attempts to renegotiate and doesn't support RI respond with
10242   a no_renegotiation alert as required by RFC5746.  Some renegotiating
10243   TLS clients will continue a connection gracefully when they receive
10244   the alert. Unfortunately OpenSSL mishandled this alert and would hang
10245   waiting for a server hello which it will never receive. Now we treat a
10246   received no_renegotiation alert as a fatal error. This is because
10247   applications requesting a renegotiation might well expect it to succeed
10248   and would have no code in place to handle the server denying it so the
10249   only safe thing to do is to terminate the connection.
10250
10251   *Steve Henson*
10252
10253 * Add ctrl macro SSL_get_secure_renegotiation_support() which returns 1 if
10254   peer supports secure renegotiation and 0 otherwise. Print out peer
10255   renegotiation support in s_client/s_server.
10256
10257   *Steve Henson*
10258
10259 * Replace the highly broken and deprecated SPKAC certification method with
10260   the updated NID creation version. This should correctly handle UTF8.
10261
10262   *Steve Henson*
10263
10264 * Implement RFC5746. Re-enable renegotiation but require the extension
10265   as needed. Unfortunately, SSL3_FLAGS_ALLOW_UNSAFE_LEGACY_RENEGOTIATION
10266   turns out to be a bad idea. It has been replaced by
10267   SSL_OP_ALLOW_UNSAFE_LEGACY_RENEGOTIATION which can be set with
10268   SSL_CTX_set_options(). This is really not recommended unless you
10269   know what you are doing.
10270
10271   *Eric Rescorla <ekr@networkresonance.com>, Ben Laurie, Steve Henson*
10272
10273 * Fixes to stateless session resumption handling. Use initial_ctx when
10274   issuing and attempting to decrypt tickets in case it has changed during
10275   servername handling. Use a non-zero length session ID when attempting
10276   stateless session resumption: this makes it possible to determine if
10277   a resumption has occurred immediately after receiving server hello
10278   (several places in OpenSSL subtly assume this) instead of later in
10279   the handshake.
10280
10281   *Steve Henson*
10282
10283 * The functions ENGINE_ctrl(), OPENSSL_isservice(),
10284   CMS_get1_RecipientRequest() and RAND_bytes() can return <=0 on error
10285   fixes for a few places where the return code is not checked
10286   correctly.
10287
10288   *Julia Lawall <julia@diku.dk>*
10289
10290 * Add --strict-warnings option to Configure script to include devteam
10291   warnings in other configurations.
10292
10293   *Steve Henson*
10294
10295 * Add support for --libdir option and LIBDIR variable in makefiles. This
10296   makes it possible to install openssl libraries in locations which
10297   have names other than "lib", for example "/usr/lib64" which some
10298   systems need.
10299
10300   *Steve Henson, based on patch from Jeremy Utley*
10301
10302 * Don't allow the use of leading 0x80 in OIDs. This is a violation of
10303   X690 8.9.12 and can produce some misleading textual output of OIDs.
10304
10305   *Steve Henson, reported by Dan Kaminsky*
10306
10307 * Delete MD2 from algorithm tables. This follows the recommendation in
10308   several standards that it is not used in new applications due to
10309   several cryptographic weaknesses. For binary compatibility reasons
10310   the MD2 API is still compiled in by default.
10311
10312   *Steve Henson*
10313
10314 * Add compression id to {d2i,i2d}_SSL_SESSION so it is correctly saved
10315   and restored.
10316
10317   *Steve Henson*
10318
10319 * Rename uni2asc and asc2uni functions to OPENSSL_uni2asc and
10320   OPENSSL_asc2uni conditionally on Netware platforms to avoid a name
10321   clash.
10322
10323   *Guenter <lists@gknw.net>*
10324
10325 * Fix the server certificate chain building code to use X509_verify_cert(),
10326   it used to have an ad-hoc builder which was unable to cope with anything
10327   other than a simple chain.
10328
10329   *David Woodhouse <dwmw2@infradead.org>, Steve Henson*
10330
10331 * Don't check self signed certificate signatures in X509_verify_cert()
10332   by default (a flag can override this): it just wastes time without
10333   adding any security. As a useful side effect self signed root CAs
10334   with non-FIPS digests are now usable in FIPS mode.
10335
10336   *Steve Henson*
10337
10338 * In dtls1_process_out_of_seq_message() the check if the current message
10339   is already buffered was missing. For every new message was memory
10340   allocated, allowing an attacker to perform an denial of service attack
10341   with sending out of seq handshake messages until there is no memory
10342   left. Additionally every future message was buffered, even if the
10343   sequence number made no sense and would be part of another handshake.
10344   So only messages with sequence numbers less than 10 in advance will be
10345   buffered.  ([CVE-2009-1378])
10346
10347   *Robin Seggelmann, discovered by Daniel Mentz*
10348
10349 * Records are buffered if they arrive with a future epoch to be
10350   processed after finishing the corresponding handshake. There is
10351   currently no limitation to this buffer allowing an attacker to perform
10352   a DOS attack with sending records with future epochs until there is no
10353   memory left. This patch adds the pqueue_size() function to determine
10354   the size of a buffer and limits the record buffer to 100 entries.
10355   ([CVE-2009-1377])
10356
10357   *Robin Seggelmann, discovered by Daniel Mentz*
10358
10359 * Keep a copy of frag->msg_header.frag_len so it can be used after the
10360   parent structure is freed.  ([CVE-2009-1379])
10361
10362   *Daniel Mentz*
10363
10364 * Handle non-blocking I/O properly in SSL_shutdown() call.
10365
10366   *Darryl Miles <darryl-mailinglists@netbauds.net>*
10367
10368 * Add `2.5.4.*` OIDs
10369
10370   *Ilya O. <vrghost@gmail.com>*
10371
10372### Changes between 0.9.8k and 0.9.8l  [5 Nov 2009]
10373
10374 * Disable renegotiation completely - this fixes a severe security
10375   problem ([CVE-2009-3555]) at the cost of breaking all
10376   renegotiation. Renegotiation can be re-enabled by setting
10377   SSL3_FLAGS_ALLOW_UNSAFE_LEGACY_RENEGOTIATION in s3->flags at
10378   run-time. This is really not recommended unless you know what
10379   you're doing.
10380
10381   *Ben Laurie*
10382
10383### Changes between 0.9.8j and 0.9.8k  [25 Mar 2009]
10384
10385 * Don't set val to NULL when freeing up structures, it is freed up by
10386   underlying code. If `sizeof(void *) > sizeof(long)` this can result in
10387   zeroing past the valid field. ([CVE-2009-0789])
10388
10389   *Paolo Ganci <Paolo.Ganci@AdNovum.CH>*
10390
10391 * Fix bug where return value of CMS_SignerInfo_verify_content() was not
10392   checked correctly. This would allow some invalid signed attributes to
10393   appear to verify correctly. ([CVE-2009-0591])
10394
10395   *Ivan Nestlerode <inestlerode@us.ibm.com>*
10396
10397 * Reject UniversalString and BMPString types with invalid lengths. This
10398   prevents a crash in ASN1_STRING_print_ex() which assumes the strings have
10399   a legal length. ([CVE-2009-0590])
10400
10401   *Steve Henson*
10402
10403 * Set S/MIME signing as the default purpose rather than setting it
10404   unconditionally. This allows applications to override it at the store
10405   level.
10406
10407   *Steve Henson*
10408
10409 * Permit restricted recursion of ASN1 strings. This is needed in practice
10410   to handle some structures.
10411
10412   *Steve Henson*
10413
10414 * Improve efficiency of mem_gets: don't search whole buffer each time
10415   for a '\n'
10416
10417   *Jeremy Shapiro <jnshapir@us.ibm.com>*
10418
10419 * New -hex option for openssl rand.
10420
10421   *Matthieu Herrb*
10422
10423 * Print out UTF8String and NumericString when parsing ASN1.
10424
10425   *Steve Henson*
10426
10427 * Support NumericString type for name components.
10428
10429   *Steve Henson*
10430
10431 * Allow CC in the environment to override the automatically chosen
10432   compiler. Note that nothing is done to ensure flags work with the
10433   chosen compiler.
10434
10435   *Ben Laurie*
10436
10437### Changes between 0.9.8i and 0.9.8j  [07 Jan 2009]
10438
10439 * Properly check EVP_VerifyFinal() and similar return values
10440   ([CVE-2008-5077]).
10441
10442   *Ben Laurie, Bodo Moeller, Google Security Team*
10443
10444 * Enable TLS extensions by default.
10445
10446   *Ben Laurie*
10447
10448 * Allow the CHIL engine to be loaded, whether the application is
10449   multithreaded or not. (This does not release the developer from the
10450   obligation to set up the dynamic locking callbacks.)
10451
10452   *Sander Temme <sander@temme.net>*
10453
10454 * Use correct exit code if there is an error in dgst command.
10455
10456   *Steve Henson; problem pointed out by Roland Dirlewanger*
10457
10458 * Tweak Configure so that you need to say "experimental-jpake" to enable
10459   JPAKE, and need to use -DOPENSSL_EXPERIMENTAL_JPAKE in applications.
10460
10461   *Bodo Moeller*
10462
10463 * Add experimental JPAKE support, including demo authentication in
10464   s_client and s_server.
10465
10466   *Ben Laurie*
10467
10468 * Set the comparison function in v3_addr_canonize().
10469
10470   *Rob Austein <sra@hactrn.net>*
10471
10472 * Add support for XMPP STARTTLS in s_client.
10473
10474   *Philip Paeps <philip@freebsd.org>*
10475
10476 * Change the server-side SSL_OP_NETSCAPE_REUSE_CIPHER_CHANGE_BUG behavior
10477   to ensure that even with this option, only ciphersuites in the
10478   server's preference list will be accepted.  (Note that the option
10479   applies only when resuming a session, so the earlier behavior was
10480   just about the algorithm choice for symmetric cryptography.)
10481
10482   *Bodo Moeller*
10483
10484### Changes between 0.9.8h and 0.9.8i  [15 Sep 2008]
10485
10486 * Fix NULL pointer dereference if a DTLS server received
10487   ChangeCipherSpec as first record ([CVE-2009-1386]).
10488
10489   *PR #1679*
10490
10491 * Fix a state transition in s3_srvr.c and d1_srvr.c
10492   (was using SSL3_ST_CW_CLNT_HELLO_B, should be `..._ST_SW_SRVR_...`).
10493
10494   *Nagendra Modadugu*
10495
10496 * The fix in 0.9.8c that supposedly got rid of unsafe
10497   double-checked locking was incomplete for RSA blinding,
10498   addressing just one layer of what turns out to have been
10499   doubly unsafe triple-checked locking.
10500
10501   So now fix this for real by retiring the MONT_HELPER macro
10502   in crypto/rsa/rsa_eay.c.
10503
10504   *Bodo Moeller; problem pointed out by Marius Schilder*
10505
10506 * Various precautionary measures:
10507
10508   - Avoid size_t integer overflow in HASH_UPDATE (md32_common.h).
10509
10510   - Avoid a buffer overflow in d2i_SSL_SESSION() (ssl_asn1.c).
10511     (NB: This would require knowledge of the secret session ticket key
10512     to exploit, in which case you'd be SOL either way.)
10513
10514   - Change bn_nist.c so that it will properly handle input BIGNUMs
10515     outside the expected range.
10516
10517   - Enforce the 'num' check in BN_div() (bn_div.c) for non-BN_DEBUG
10518     builds.
10519
10520   *Neel Mehta, Bodo Moeller*
10521
10522 * Allow engines to be "soft loaded" - i.e. optionally don't die if
10523   the load fails. Useful for distros.
10524
10525   *Ben Laurie and the FreeBSD team*
10526
10527 * Add support for Local Machine Keyset attribute in PKCS#12 files.
10528
10529   *Steve Henson*
10530
10531 * Fix BN_GF2m_mod_arr() top-bit cleanup code.
10532
10533   *Huang Ying*
10534
10535 * Expand ENGINE to support engine supplied SSL client certificate functions.
10536
10537   This work was sponsored by Logica.
10538
10539   *Steve Henson*
10540
10541 * Add CryptoAPI ENGINE to support use of RSA and DSA keys held in Windows
10542   keystores. Support for SSL/TLS client authentication too.
10543   Not compiled unless enable-capieng specified to Configure.
10544
10545   This work was sponsored by Logica.
10546
10547   *Steve Henson*
10548
10549 * Fix bug in X509_ATTRIBUTE creation: don't set attribute using
10550   ASN1_TYPE_set1 if MBSTRING flag set. This bug would crash certain
10551   attribute creation routines such as certificate requests and PKCS#12
10552   files.
10553
10554   *Steve Henson*
10555
10556### Changes between 0.9.8g and 0.9.8h  [28 May 2008]
10557
10558 * Fix flaw if 'Server Key exchange message' is omitted from a TLS
10559   handshake which could lead to a client crash as found using the
10560   Codenomicon TLS test suite ([CVE-2008-1672])
10561
10562   *Steve Henson, Mark Cox*
10563
10564 * Fix double free in TLS server name extensions which could lead to
10565   a remote crash found by Codenomicon TLS test suite ([CVE-2008-0891])
10566
10567   *Joe Orton*
10568
10569 * Clear error queue in SSL_CTX_use_certificate_chain_file()
10570
10571   Clear the error queue to ensure that error entries left from
10572   older function calls do not interfere with the correct operation.
10573
10574   *Lutz Jaenicke, Erik de Castro Lopo*
10575
10576 * Remove root CA certificates of commercial CAs:
10577
10578   The OpenSSL project does not recommend any specific CA and does not
10579   have any policy with respect to including or excluding any CA.
10580   Therefore it does not make any sense to ship an arbitrary selection
10581   of root CA certificates with the OpenSSL software.
10582
10583   *Lutz Jaenicke*
10584
10585 * RSA OAEP patches to fix two separate invalid memory reads.
10586   The first one involves inputs when 'lzero' is greater than
10587   'SHA_DIGEST_LENGTH' (it would read about SHA_DIGEST_LENGTH bytes
10588   before the beginning of from). The second one involves inputs where
10589   the 'db' section contains nothing but zeroes (there is a one-byte
10590   invalid read after the end of 'db').
10591
10592   *Ivan Nestlerode <inestlerode@us.ibm.com>*
10593
10594 * Partial backport from 0.9.9-dev:
10595
10596   Introduce bn_mul_mont (dedicated Montgomery multiplication
10597   procedure) as a candidate for BIGNUM assembler implementation.
10598   While 0.9.9-dev uses assembler for various architectures, only
10599   x86_64 is available by default here in the 0.9.8 branch, and
10600   32-bit x86 is available through a compile-time setting.
10601
10602   To try the 32-bit x86 assembler implementation, use Configure
10603   option "enable-montasm" (which exists only for this backport).
10604
10605   As "enable-montasm" for 32-bit x86 disclaims code stability
10606   anyway, in this constellation we activate additional code
10607   backported from 0.9.9-dev for further performance improvements,
10608   namely BN_from_montgomery_word.  (To enable this otherwise,
10609   e.g. x86_64, try `-DMONT_FROM_WORD___NON_DEFAULT_0_9_8_BUILD`.)
10610
10611   *Andy Polyakov (backport partially by Bodo Moeller)*
10612
10613 * Add TLS session ticket callback. This allows an application to set
10614   TLS ticket cipher and HMAC keys rather than relying on hardcoded fixed
10615   values. This is useful for key rollover for example where several key
10616   sets may exist with different names.
10617
10618   *Steve Henson*
10619
10620 * Reverse ENGINE-internal logic for caching default ENGINE handles.
10621   This was broken until now in 0.9.8 releases, such that the only way
10622   a registered ENGINE could be used (assuming it initialises
10623   successfully on the host) was to explicitly set it as the default
10624   for the relevant algorithms. This is in contradiction with 0.9.7
10625   behaviour and the documentation. With this fix, when an ENGINE is
10626   registered into a given algorithm's table of implementations, the
10627   'uptodate' flag is reset so that auto-discovery will be used next
10628   time a new context for that algorithm attempts to select an
10629   implementation.
10630
10631   *Ian Lister (tweaked by Geoff Thorpe)*
10632
10633 * Backport of CMS code to OpenSSL 0.9.8. This differs from the 0.9.9
10634   implementation in the following ways:
10635
10636   Lack of EVP_PKEY_ASN1_METHOD means algorithm parameters have to be
10637   hard coded.
10638
10639   Lack of BER streaming support means one pass streaming processing is
10640   only supported if data is detached: setting the streaming flag is
10641   ignored for embedded content.
10642
10643   CMS support is disabled by default and must be explicitly enabled
10644   with the enable-cms configuration option.
10645
10646   *Steve Henson*
10647
10648 * Update the GMP engine glue to do direct copies between BIGNUM and
10649   mpz_t when openssl and GMP use the same limb size. Otherwise the
10650   existing "conversion via a text string export" trick is still used.
10651
10652   *Paul Sheer <paulsheer@gmail.com>*
10653
10654 * Zlib compression BIO. This is a filter BIO which compressed and
10655   uncompresses any data passed through it.
10656
10657   *Steve Henson*
10658
10659 * Add AES_wrap_key() and AES_unwrap_key() functions to implement
10660   RFC3394 compatible AES key wrapping.
10661
10662   *Steve Henson*
10663
10664 * Add utility functions to handle ASN1 structures. ASN1_STRING_set0():
10665   sets string data without copying. X509_ALGOR_set0() and
10666   X509_ALGOR_get0(): set and retrieve X509_ALGOR (AlgorithmIdentifier)
10667   data. Attribute function X509at_get0_data_by_OBJ(): retrieves data
10668   from an X509_ATTRIBUTE structure optionally checking it occurs only
10669   once. ASN1_TYPE_set1(): set and ASN1_TYPE structure copying supplied
10670   data.
10671
10672   *Steve Henson*
10673
10674 * Fix BN flag handling in RSA_eay_mod_exp() and BN_MONT_CTX_set()
10675   to get the expected BN_FLG_CONSTTIME behavior.
10676
10677   *Bodo Moeller (Google)*
10678
10679 * Netware support:
10680
10681   - fixed wrong usage of ioctlsocket() when build for LIBC BSD sockets
10682   - fixed do_tests.pl to run the test suite with CLIB builds too (CLIB_OPT)
10683   - added some more tests to do_tests.pl
10684   - fixed RunningProcess usage so that it works with newer LIBC NDKs too
10685   - removed usage of BN_LLONG for CLIB builds to avoid runtime dependency
10686   - added new Configure targets netware-clib-bsdsock, netware-clib-gcc,
10687     netware-clib-bsdsock-gcc, netware-libc-bsdsock-gcc
10688   - various changes to netware.pl to enable gcc-cross builds on Win32
10689     platform
10690   - changed crypto/bio/b_sock.c to work with macro functions (CLIB BSD)
10691   - various changes to fix missing prototype warnings
10692   - fixed x86nasm.pl to create correct asm files for NASM COFF output
10693   - added AES, WHIRLPOOL and CPUID assembler code to build files
10694   - added missing AES assembler make rules to mk1mf.pl
10695   - fixed order of includes in `apps/ocsp.c` so that `e_os.h` settings apply
10696
10697   *Guenter Knauf <eflash@gmx.net>*
10698
10699 * Implement certificate status request TLS extension defined in RFC3546.
10700   A client can set the appropriate parameters and receive the encoded
10701   OCSP response via a callback. A server can query the supplied parameters
10702   and set the encoded OCSP response in the callback. Add simplified examples
10703   to s_client and s_server.
10704
10705   *Steve Henson*
10706
10707### Changes between 0.9.8f and 0.9.8g  [19 Oct 2007]
10708
10709 * Fix various bugs:
10710   + Binary incompatibility of ssl_ctx_st structure
10711   + DTLS interoperation with non-compliant servers
10712   + Don't call get_session_cb() without proposed session
10713   + Fix ia64 assembler code
10714
10715   *Andy Polyakov, Steve Henson*
10716
10717### Changes between 0.9.8e and 0.9.8f  [11 Oct 2007]
10718
10719 * DTLS Handshake overhaul. There were longstanding issues with
10720   OpenSSL DTLS implementation, which were making it impossible for
10721   RFC 4347 compliant client to communicate with OpenSSL server.
10722   Unfortunately just fixing these incompatibilities would "cut off"
10723   pre-0.9.8f clients. To allow for hassle free upgrade post-0.9.8e
10724   server keeps tolerating non RFC compliant syntax. The opposite is
10725   not true, 0.9.8f client can not communicate with earlier server.
10726   This update even addresses CVE-2007-4995.
10727
10728   *Andy Polyakov*
10729
10730 * Changes to avoid need for function casts in OpenSSL: some compilers
10731   (gcc 4.2 and later) reject their use.
10732   *Kurt Roeckx <kurt@roeckx.be>, Peter Hartley <pdh@utter.chaos.org.uk>,
10733    Steve Henson*
10734
10735 * Add RFC4507 support to OpenSSL. This includes the corrections in
10736   RFC4507bis. The encrypted ticket format is an encrypted encoded
10737   SSL_SESSION structure, that way new session features are automatically
10738   supported.
10739
10740   If a client application caches session in an SSL_SESSION structure
10741   support is transparent because tickets are now stored in the encoded
10742   SSL_SESSION.
10743
10744   The SSL_CTX structure automatically generates keys for ticket
10745   protection in servers so again support should be possible
10746   with no application modification.
10747
10748   If a client or server wishes to disable RFC4507 support then the option
10749   SSL_OP_NO_TICKET can be set.
10750
10751   Add a TLS extension debugging callback to allow the contents of any client
10752   or server extensions to be examined.
10753
10754   This work was sponsored by Google.
10755
10756   *Steve Henson*
10757
10758 * Add initial support for TLS extensions, specifically for the server_name
10759   extension so far.  The SSL_SESSION, SSL_CTX, and SSL data structures now
10760   have new members for a host name.  The SSL data structure has an
10761   additional member `SSL_CTX *initial_ctx` so that new sessions can be
10762   stored in that context to allow for session resumption, even after the
10763   SSL has been switched to a new SSL_CTX in reaction to a client's
10764   server_name extension.
10765
10766   New functions (subject to change):
10767
10768           SSL_get_servername()
10769           SSL_get_servername_type()
10770           SSL_set_SSL_CTX()
10771
10772   New CTRL codes and macros (subject to change):
10773
10774           SSL_CTRL_SET_TLSEXT_SERVERNAME_CB
10775                               - SSL_CTX_set_tlsext_servername_callback()
10776           SSL_CTRL_SET_TLSEXT_SERVERNAME_ARG
10777                                    - SSL_CTX_set_tlsext_servername_arg()
10778           SSL_CTRL_SET_TLSEXT_HOSTNAME           - SSL_set_tlsext_host_name()
10779
10780   openssl s_client has a new '-servername ...' option.
10781
10782   openssl s_server has new options '-servername_host ...', '-cert2 ...',
10783   '-key2 ...', '-servername_fatal' (subject to change).  This allows
10784   testing the HostName extension for a specific single host name ('-cert'
10785   and '-key' remain fallbacks for handshakes without HostName
10786   negotiation).  If the unrecognized_name alert has to be sent, this by
10787   default is a warning; it becomes fatal with the '-servername_fatal'
10788   option.
10789
10790   *Peter Sylvester,  Remy Allais, Christophe Renou, Steve Henson*
10791
10792 * Add AES and SSE2 assembly language support to VC++ build.
10793
10794   *Steve Henson*
10795
10796 * Mitigate attack on final subtraction in Montgomery reduction.
10797
10798   *Andy Polyakov*
10799
10800 * Fix crypto/ec/ec_mult.c to work properly with scalars of value 0
10801   (which previously caused an internal error).
10802
10803   *Bodo Moeller*
10804
10805 * Squeeze another 10% out of IGE mode when in != out.
10806
10807   *Ben Laurie*
10808
10809 * AES IGE mode speedup.
10810
10811   *Dean Gaudet (Google)*
10812
10813 * Add the Korean symmetric 128-bit cipher SEED (see
10814   <http://www.kisa.or.kr/kisa/seed/jsp/seed_eng.jsp>) and
10815   add SEED ciphersuites from RFC 4162:
10816
10817           TLS_RSA_WITH_SEED_CBC_SHA      =  "SEED-SHA"
10818           TLS_DHE_DSS_WITH_SEED_CBC_SHA  =  "DHE-DSS-SEED-SHA"
10819           TLS_DHE_RSA_WITH_SEED_CBC_SHA  =  "DHE-RSA-SEED-SHA"
10820           TLS_DH_anon_WITH_SEED_CBC_SHA  =  "ADH-SEED-SHA"
10821
10822   To minimize changes between patchlevels in the OpenSSL 0.9.8
10823   series, SEED remains excluded from compilation unless OpenSSL
10824   is configured with 'enable-seed'.
10825
10826   *KISA, Bodo Moeller*
10827
10828 * Mitigate branch prediction attacks, which can be practical if a
10829   single processor is shared, allowing a spy process to extract
10830   information.  For detailed background information, see
10831   <http://eprint.iacr.org/2007/039> (O. Aciicmez, S. Gueron,
10832   J.-P. Seifert, "New Branch Prediction Vulnerabilities in OpenSSL
10833   and Necessary Software Countermeasures").  The core of the change
10834   are new versions BN_div_no_branch() and
10835   BN_mod_inverse_no_branch() of BN_div() and BN_mod_inverse(),
10836   respectively, which are slower, but avoid the security-relevant
10837   conditional branches.  These are automatically called by BN_div()
10838   and BN_mod_inverse() if the flag BN_FLG_CONSTTIME is set for one
10839   of the input BIGNUMs.  Also, BN_is_bit_set() has been changed to
10840   remove a conditional branch.
10841
10842   BN_FLG_CONSTTIME is the new name for the previous
10843   BN_FLG_EXP_CONSTTIME flag, since it now affects more than just
10844   modular exponentiation.  (Since OpenSSL 0.9.7h, setting this flag
10845   in the exponent causes BN_mod_exp_mont() to use the alternative
10846   implementation in BN_mod_exp_mont_consttime().)  The old name
10847   remains as a deprecated alias.
10848
10849   Similarly, RSA_FLAG_NO_EXP_CONSTTIME is replaced by a more general
10850   RSA_FLAG_NO_CONSTTIME flag since the RSA implementation now uses
10851   constant-time implementations for more than just exponentiation.
10852   Here too the old name is kept as a deprecated alias.
10853
10854   BN_BLINDING_new() will now use BN_dup() for the modulus so that
10855   the BN_BLINDING structure gets an independent copy of the
10856   modulus.  This means that the previous `BIGNUM *m` argument to
10857   BN_BLINDING_new() and to BN_BLINDING_create_param() now
10858   essentially becomes `const BIGNUM *m`, although we can't actually
10859   change this in the header file before 0.9.9.  It allows
10860   RSA_setup_blinding() to use BN_with_flags() on the modulus to
10861   enable BN_FLG_CONSTTIME.
10862
10863   *Matthew D Wood (Intel Corp)*
10864
10865 * In the SSL/TLS server implementation, be strict about session ID
10866   context matching (which matters if an application uses a single
10867   external cache for different purposes).  Previously,
10868   out-of-context reuse was forbidden only if SSL_VERIFY_PEER was
10869   set.  This did ensure strict client verification, but meant that,
10870   with applications using a single external cache for quite
10871   different requirements, clients could circumvent ciphersuite
10872   restrictions for a given session ID context by starting a session
10873   in a different context.
10874
10875   *Bodo Moeller*
10876
10877 * Include "!eNULL" in SSL_DEFAULT_CIPHER_LIST to make sure that
10878   a ciphersuite string such as "DEFAULT:RSA" cannot enable
10879   authentication-only ciphersuites.
10880
10881   *Bodo Moeller*
10882
10883 * Update the SSL_get_shared_ciphers() fix CVE-2006-3738 which was
10884   not complete and could lead to a possible single byte overflow
10885   ([CVE-2007-5135]) [Ben Laurie]
10886
10887### Changes between 0.9.8d and 0.9.8e  [23 Feb 2007]
10888
10889 * Since AES128 and AES256 (and similarly Camellia128 and
10890   Camellia256) share a single mask bit in the logic of
10891   ssl/ssl_ciph.c, the code for masking out disabled ciphers needs a
10892   kludge to work properly if AES128 is available and AES256 isn't
10893   (or if Camellia128 is available and Camellia256 isn't).
10894
10895   *Victor Duchovni*
10896
10897 * Fix the BIT STRING encoding generated by crypto/ec/ec_asn1.c
10898   (within i2d_ECPrivateKey, i2d_ECPKParameters, i2d_ECParameters):
10899   When a point or a seed is encoded in a BIT STRING, we need to
10900   prevent the removal of trailing zero bits to get the proper DER
10901   encoding.  (By default, crypto/asn1/a_bitstr.c assumes the case
10902   of a NamedBitList, for which trailing 0 bits need to be removed.)
10903
10904   *Bodo Moeller*
10905
10906 * Have SSL/TLS server implementation tolerate "mismatched" record
10907   protocol version while receiving ClientHello even if the
10908   ClientHello is fragmented.  (The server can't insist on the
10909   particular protocol version it has chosen before the ServerHello
10910   message has informed the client about his choice.)
10911
10912   *Bodo Moeller*
10913
10914 * Add RFC 3779 support.
10915
10916   *Rob Austein for ARIN, Ben Laurie*
10917
10918 * Load error codes if they are not already present instead of using a
10919   static variable. This allows them to be cleanly unloaded and reloaded.
10920   Improve header file function name parsing.
10921
10922   *Steve Henson*
10923
10924 * extend SMTP and IMAP protocol emulation in s_client to use EHLO
10925   or CAPABILITY handshake as required by RFCs.
10926
10927   *Goetz Babin-Ebell*
10928
10929### Changes between 0.9.8c and 0.9.8d  [28 Sep 2006]
10930
10931 * Introduce limits to prevent malicious keys being able to
10932   cause a denial of service.  ([CVE-2006-2940])
10933
10934   *Steve Henson, Bodo Moeller*
10935
10936 * Fix ASN.1 parsing of certain invalid structures that can result
10937   in a denial of service.  ([CVE-2006-2937])  [Steve Henson]
10938
10939 * Fix buffer overflow in SSL_get_shared_ciphers() function.
10940   ([CVE-2006-3738]) [Tavis Ormandy and Will Drewry, Google Security Team]
10941
10942 * Fix SSL client code which could crash if connecting to a
10943   malicious SSLv2 server.  ([CVE-2006-4343])
10944
10945   *Tavis Ormandy and Will Drewry, Google Security Team*
10946
10947 * Since 0.9.8b, ciphersuite strings naming explicit ciphersuites
10948   match only those.  Before that, "AES256-SHA" would be interpreted
10949   as a pattern and match "AES128-SHA" too (since AES128-SHA got
10950   the same strength classification in 0.9.7h) as we currently only
10951   have a single AES bit in the ciphersuite description bitmap.
10952   That change, however, also applied to ciphersuite strings such as
10953   "RC4-MD5" that intentionally matched multiple ciphersuites --
10954   namely, SSL 2.0 ciphersuites in addition to the more common ones
10955   from SSL 3.0/TLS 1.0.
10956
10957   So we change the selection algorithm again: Naming an explicit
10958   ciphersuite selects this one ciphersuite, and any other similar
10959   ciphersuite (same bitmap) from *other* protocol versions.
10960   Thus, "RC4-MD5" again will properly select both the SSL 2.0
10961   ciphersuite and the SSL 3.0/TLS 1.0 ciphersuite.
10962
10963   Since SSL 2.0 does not have any ciphersuites for which the
10964   128/256 bit distinction would be relevant, this works for now.
10965   The proper fix will be to use different bits for AES128 and
10966   AES256, which would have avoided the problems from the beginning;
10967   however, bits are scarce, so we can only do this in a new release
10968   (not just a patchlevel) when we can change the SSL_CIPHER
10969   definition to split the single 'unsigned long mask' bitmap into
10970   multiple values to extend the available space.
10971
10972   *Bodo Moeller*
10973
10974### Changes between 0.9.8b and 0.9.8c  [05 Sep 2006]
10975
10976 * Avoid PKCS #1 v1.5 signature attack discovered by Daniel Bleichenbacher
10977   ([CVE-2006-4339])  [Ben Laurie and Google Security Team]
10978
10979 * Add AES IGE and biIGE modes.
10980
10981   *Ben Laurie*
10982
10983 * Change the Unix randomness entropy gathering to use poll() when
10984   possible instead of select(), since the latter has some
10985   undesirable limitations.
10986
10987   *Darryl Miles via Richard Levitte and Bodo Moeller*
10988
10989 * Disable "ECCdraft" ciphersuites more thoroughly.  Now special
10990   treatment in ssl/ssl_ciph.s makes sure that these ciphersuites
10991   cannot be implicitly activated as part of, e.g., the "AES" alias.
10992   However, please upgrade to OpenSSL 0.9.9[-dev] for
10993   non-experimental use of the ECC ciphersuites to get TLS extension
10994   support, which is required for curve and point format negotiation
10995   to avoid potential handshake problems.
10996
10997   *Bodo Moeller*
10998
10999 * Disable rogue ciphersuites:
11000
11001   - SSLv2 0x08 0x00 0x80 ("RC4-64-MD5")
11002   - SSLv3/TLSv1 0x00 0x61 ("EXP1024-RC2-CBC-MD5")
11003   - SSLv3/TLSv1 0x00 0x60 ("EXP1024-RC4-MD5")
11004
11005   The latter two were purportedly from
11006   draft-ietf-tls-56-bit-ciphersuites-0[01].txt, but do not really
11007   appear there.
11008
11009   Also deactivate the remaining ciphersuites from
11010   draft-ietf-tls-56-bit-ciphersuites-01.txt.  These are just as
11011   unofficial, and the ID has long expired.
11012
11013   *Bodo Moeller*
11014
11015 * Fix RSA blinding Heisenbug (problems sometimes occurred on
11016   dual-core machines) and other potential thread-safety issues.
11017
11018   *Bodo Moeller*
11019
11020 * Add the symmetric cipher Camellia (128-bit, 192-bit, 256-bit key
11021   versions), which is now available for royalty-free use
11022   (see <http://info.isl.ntt.co.jp/crypt/eng/info/chiteki.html>).
11023   Also, add Camellia TLS ciphersuites from RFC 4132.
11024
11025   To minimize changes between patchlevels in the OpenSSL 0.9.8
11026   series, Camellia remains excluded from compilation unless OpenSSL
11027   is configured with 'enable-camellia'.
11028
11029   *NTT*
11030
11031 * Disable the padding bug check when compression is in use. The padding
11032   bug check assumes the first packet is of even length, this is not
11033   necessarily true if compression is enabled and can result in false
11034   positives causing handshake failure. The actual bug test is ancient
11035   code so it is hoped that implementations will either have fixed it by
11036   now or any which still have the bug do not support compression.
11037
11038   *Steve Henson*
11039
11040### Changes between 0.9.8a and 0.9.8b  [04 May 2006]
11041
11042 * When applying a cipher rule check to see if string match is an explicit
11043   cipher suite and only match that one cipher suite if it is.
11044
11045   *Steve Henson*
11046
11047 * Link in manifests for VC++ if needed.
11048
11049   *Austin Ziegler <halostatue@gmail.com>*
11050
11051 * Update support for ECC-based TLS ciphersuites according to
11052   draft-ietf-tls-ecc-12.txt with proposed changes (but without
11053   TLS extensions, which are supported starting with the 0.9.9
11054   branch, not in the OpenSSL 0.9.8 branch).
11055
11056   *Douglas Stebila*
11057
11058 * New functions EVP_CIPHER_CTX_new() and EVP_CIPHER_CTX_free() to support
11059   opaque EVP_CIPHER_CTX handling.
11060
11061   *Steve Henson*
11062
11063 * Fixes and enhancements to zlib compression code. We now only use
11064   "zlib1.dll" and use the default `__cdecl` calling convention on Win32
11065   to conform with the standards mentioned here:
11066   <http://www.zlib.net/DLL_FAQ.txt>
11067   Static zlib linking now works on Windows and the new --with-zlib-include
11068   --with-zlib-lib options to Configure can be used to supply the location
11069   of the headers and library. Gracefully handle case where zlib library
11070   can't be loaded.
11071
11072   *Steve Henson*
11073
11074 * Several fixes and enhancements to the OID generation code. The old code
11075   sometimes allowed invalid OIDs (1.X for X >= 40 for example), couldn't
11076   handle numbers larger than ULONG_MAX, truncated printing and had a
11077   non standard OBJ_obj2txt() behaviour.
11078
11079   *Steve Henson*
11080
11081 * Add support for building of engines under engine/ as shared libraries
11082   under VC++ build system.
11083
11084   *Steve Henson*
11085
11086 * Corrected the numerous bugs in the Win32 path splitter in DSO.
11087   Hopefully, we will not see any false combination of paths any more.
11088
11089   *Richard Levitte*
11090
11091### Changes between 0.9.8 and 0.9.8a  [11 Oct 2005]
11092
11093 * Remove the functionality of SSL_OP_MSIE_SSLV2_RSA_PADDING
11094   (part of SSL_OP_ALL).  This option used to disable the
11095   countermeasure against man-in-the-middle protocol-version
11096   rollback in the SSL 2.0 server implementation, which is a bad
11097   idea.  ([CVE-2005-2969])
11098
11099   *Bodo Moeller; problem pointed out by Yutaka Oiwa (Research Center
11100   for Information Security, National Institute of Advanced Industrial
11101   Science and Technology [AIST], Japan)*
11102
11103 * Add two function to clear and return the verify parameter flags.
11104
11105   *Steve Henson*
11106
11107 * Keep cipherlists sorted in the source instead of sorting them at
11108   runtime, thus removing the need for a lock.
11109
11110   *Nils Larsch*
11111
11112 * Avoid some small subgroup attacks in Diffie-Hellman.
11113
11114   *Nick Mathewson and Ben Laurie*
11115
11116 * Add functions for well-known primes.
11117
11118   *Nick Mathewson*
11119
11120 * Extended Windows CE support.
11121
11122   *Satoshi Nakamura and Andy Polyakov*
11123
11124 * Initialize SSL_METHOD structures at compile time instead of during
11125   runtime, thus removing the need for a lock.
11126
11127   *Steve Henson*
11128
11129 * Make PKCS7_decrypt() work even if no certificate is supplied by
11130   attempting to decrypt each encrypted key in turn. Add support to
11131   smime utility.
11132
11133   *Steve Henson*
11134
11135### Changes between 0.9.7h and 0.9.8  [05 Jul 2005]
11136
11137[NB: OpenSSL 0.9.7i and later 0.9.7 patch levels were released after
11138OpenSSL 0.9.8.]
11139
11140 * Add libcrypto.pc and libssl.pc for those who feel they need them.
11141
11142   *Richard Levitte*
11143
11144 * Change CA.sh and CA.pl so they don't bundle the CSR and the private
11145   key into the same file any more.
11146
11147   *Richard Levitte*
11148
11149 * Add initial support for Win64, both IA64 and AMD64/x64 flavors.
11150
11151   *Andy Polyakov*
11152
11153 * Add -utf8 command line and config file option to 'ca'.
11154
11155   *Stefan <stf@udoma.org*
11156
11157 * Removed the macro des_crypt(), as it seems to conflict with some
11158   libraries.  Use DES_crypt().
11159
11160   *Richard Levitte*
11161
11162 * Correct naming of the 'chil' and '4758cca' ENGINEs. This
11163   involves renaming the source and generated shared-libs for
11164   both. The engines will accept the corrected or legacy ids
11165   ('ncipher' and '4758_cca' respectively) when binding. NB,
11166   this only applies when building 'shared'.
11167
11168   *Corinna Vinschen <vinschen@redhat.com> and Geoff Thorpe*
11169
11170 * Add attribute functions to EVP_PKEY structure. Modify
11171   PKCS12_create() to recognize a CSP name attribute and
11172   use it. Make -CSP option work again in pkcs12 utility.
11173
11174   *Steve Henson*
11175
11176 * Add new functionality to the bn blinding code:
11177   - automatic re-creation of the BN_BLINDING parameters after
11178     a fixed number of uses (currently 32)
11179   - add new function for parameter creation
11180   - introduce flags to control the update behaviour of the
11181     BN_BLINDING parameters
11182   - hide BN_BLINDING structure
11183   Add a second BN_BLINDING slot to the RSA structure to improve
11184   performance when a single RSA object is shared among several
11185   threads.
11186
11187   *Nils Larsch*
11188
11189 * Add support for DTLS.
11190
11191   *Nagendra Modadugu <nagendra@cs.stanford.edu> and Ben Laurie*
11192
11193 * Add support for DER encoded private keys (SSL_FILETYPE_ASN1)
11194   to SSL_CTX_use_PrivateKey_file() and SSL_use_PrivateKey_file()
11195
11196   *Walter Goulet*
11197
11198 * Remove buggy and incomplete DH cert support from
11199   ssl/ssl_rsa.c and ssl/s3_both.c
11200
11201   *Nils Larsch*
11202
11203 * Use SHA-1 instead of MD5 as the default digest algorithm for
11204   the `apps/openssl` commands.
11205
11206   *Nils Larsch*
11207
11208 * Compile clean with "-Wall -Wmissing-prototypes
11209   -Wstrict-prototypes -Wmissing-declarations -Werror". Currently
11210   DEBUG_SAFESTACK must also be set.
11211
11212   *Ben Laurie*
11213
11214 * Change ./Configure so that certain algorithms can be disabled by default.
11215   The new counterpiece to "no-xxx" is "enable-xxx".
11216
11217   The patented RC5 and MDC2 algorithms will now be disabled unless
11218   "enable-rc5" and "enable-mdc2", respectively, are specified.
11219
11220   (IDEA remains enabled despite being patented.  This is because IDEA
11221   is frequently required for interoperability, and there is no license
11222   fee for non-commercial use.  As before, "no-idea" can be used to
11223   avoid this algorithm.)
11224
11225   *Bodo Moeller*
11226
11227 * Add processing of proxy certificates (see RFC 3820).  This work was
11228   sponsored by KTH (The Royal Institute of Technology in Stockholm) and
11229   EGEE (Enabling Grids for E-science in Europe).
11230
11231   *Richard Levitte*
11232
11233 * RC4 performance overhaul on modern architectures/implementations, such
11234   as Intel P4, IA-64 and AMD64.
11235
11236   *Andy Polyakov*
11237
11238 * New utility extract-section.pl. This can be used specify an alternative
11239   section number in a pod file instead of having to treat each file as
11240   a separate case in Makefile. This can be done by adding two lines to the
11241   pod file:
11242
11243   =for comment openssl_section:XXX
11244
11245   The blank line is mandatory.
11246
11247   *Steve Henson*
11248
11249 * New arguments -certform, -keyform and -pass for s_client and s_server
11250   to allow alternative format key and certificate files and passphrase
11251   sources.
11252
11253   *Steve Henson*
11254
11255 * New structure X509_VERIFY_PARAM which combines current verify parameters,
11256   update associated structures and add various utility functions.
11257
11258   Add new policy related verify parameters, include policy checking in
11259   standard verify code. Enhance 'smime' application with extra parameters
11260   to support policy checking and print out.
11261
11262   *Steve Henson*
11263
11264 * Add a new engine to support VIA PadLock ACE extensions in the VIA C3
11265   Nehemiah processors. These extensions support AES encryption in hardware
11266   as well as RNG (though RNG support is currently disabled).
11267
11268   *Michal Ludvig <michal@logix.cz>, with help from Andy Polyakov*
11269
11270 * Deprecate `BN_[get|set]_params()` functions (they were ignored internally).
11271
11272   *Geoff Thorpe*
11273
11274 * New FIPS 180-2 algorithms, SHA-224/-256/-384/-512 are implemented.
11275
11276   *Andy Polyakov and a number of other people*
11277
11278 * Improved PowerPC platform support. Most notably BIGNUM assembler
11279   implementation contributed by IBM.
11280
11281   *Suresh Chari, Peter Waltenberg, Andy Polyakov*
11282
11283 * The new 'RSA_generate_key_ex' function now takes a BIGNUM for the public
11284   exponent rather than 'unsigned long'. There is a corresponding change to
11285   the new 'rsa_keygen' element of the RSA_METHOD structure.
11286
11287   *Jelte Jansen, Geoff Thorpe*
11288
11289 * Functionality for creating the initial serial number file is now
11290   moved from CA.pl to the 'ca' utility with a new option -create_serial.
11291
11292   (Before OpenSSL 0.9.7e, CA.pl used to initialize the serial
11293   number file to 1, which is bound to cause problems.  To avoid
11294   the problems while respecting compatibility between different 0.9.7
11295   patchlevels, 0.9.7e  employed 'openssl x509 -next_serial' in
11296   CA.pl for serial number initialization.  With the new release 0.9.8,
11297   we can fix the problem directly in the 'ca' utility.)
11298
11299   *Steve Henson*
11300
11301 * Reduced header interdependencies by declaring more opaque objects in
11302   ossl_typ.h. As a consequence, including some headers (eg. engine.h) will
11303   give fewer recursive includes, which could break lazy source code - so
11304   this change is covered by the OPENSSL_NO_DEPRECATED symbol. As always,
11305   developers should define this symbol when building and using openssl to
11306   ensure they track the recommended behaviour, interfaces, [etc], but
11307   backwards-compatible behaviour prevails when this isn't defined.
11308
11309   *Geoff Thorpe*
11310
11311 * New function X509_POLICY_NODE_print() which prints out policy nodes.
11312
11313   *Steve Henson*
11314
11315 * Add new EVP function EVP_CIPHER_CTX_rand_key and associated functionality.
11316   This will generate a random key of the appropriate length based on the
11317   cipher context. The EVP_CIPHER can provide its own random key generation
11318   routine to support keys of a specific form. This is used in the des and
11319   3des routines to generate a key of the correct parity. Update S/MIME
11320   code to use new functions and hence generate correct parity DES keys.
11321   Add EVP_CHECK_DES_KEY #define to return an error if the key is not
11322   valid (weak or incorrect parity).
11323
11324   *Steve Henson*
11325
11326 * Add a local set of CRLs that can be used by X509_verify_cert() as well
11327   as looking them up. This is useful when the verified structure may contain
11328   CRLs, for example PKCS#7 signedData. Modify PKCS7_verify() to use any CRLs
11329   present unless the new PKCS7_NO_CRL flag is asserted.
11330
11331   *Steve Henson*
11332
11333 * Extend ASN1 oid configuration module. It now additionally accepts the
11334   syntax:
11335
11336   shortName = some long name, 1.2.3.4
11337
11338   *Steve Henson*
11339
11340 * Reimplemented the BN_CTX implementation. There is now no more static
11341   limitation on the number of variables it can handle nor the depth of the
11342   "stack" handling for BN_CTX_start()/BN_CTX_end() pairs. The stack
11343   information can now expand as required, and rather than having a single
11344   static array of bignums, BN_CTX now uses a linked-list of such arrays
11345   allowing it to expand on demand whilst maintaining the usefulness of
11346   BN_CTX's "bundling".
11347
11348   *Geoff Thorpe*
11349
11350 * Add a missing BN_CTX parameter to the 'rsa_mod_exp' callback in RSA_METHOD
11351   to allow all RSA operations to function using a single BN_CTX.
11352
11353   *Geoff Thorpe*
11354
11355 * Preliminary support for certificate policy evaluation and checking. This
11356   is initially intended to pass the tests outlined in "Conformance Testing
11357   of Relying Party Client Certificate Path Processing Logic" v1.07.
11358
11359   *Steve Henson*
11360
11361 * bn_dup_expand() has been deprecated, it was introduced in 0.9.7 and
11362   remained unused and not that useful. A variety of other little bignum
11363   tweaks and fixes have also been made continuing on from the audit (see
11364   below).
11365
11366   *Geoff Thorpe*
11367
11368 * Constify all or almost all d2i, c2i, s2i and r2i functions, along with
11369   associated ASN1, EVP and SSL functions and old ASN1 macros.
11370
11371   *Richard Levitte*
11372
11373 * BN_zero() only needs to set 'top' and 'neg' to zero for correct results,
11374   and this should never fail. So the return value from the use of
11375   BN_set_word() (which can fail due to needless expansion) is now deprecated;
11376   if OPENSSL_NO_DEPRECATED is defined, BN_zero() is a void macro.
11377
11378   *Geoff Thorpe*
11379
11380 * BN_CTX_get() should return zero-valued bignums, providing the same
11381   initialised value as BN_new().
11382
11383   *Geoff Thorpe, suggested by Ulf Möller*
11384
11385 * Support for inhibitAnyPolicy certificate extension.
11386
11387   *Steve Henson*
11388
11389 * An audit of the BIGNUM code is underway, for which debugging code is
11390   enabled when BN_DEBUG is defined. This makes stricter enforcements on what
11391   is considered valid when processing BIGNUMs, and causes execution to
11392   assert() when a problem is discovered. If BN_DEBUG_RAND is defined,
11393   further steps are taken to deliberately pollute unused data in BIGNUM
11394   structures to try and expose faulty code further on. For now, openssl will
11395   (in its default mode of operation) continue to tolerate the inconsistent
11396   forms that it has tolerated in the past, but authors and packagers should
11397   consider trying openssl and their own applications when compiled with
11398   these debugging symbols defined. It will help highlight potential bugs in
11399   their own code, and will improve the test coverage for OpenSSL itself. At
11400   some point, these tighter rules will become openssl's default to improve
11401   maintainability, though the assert()s and other overheads will remain only
11402   in debugging configurations. See bn.h for more details.
11403
11404   *Geoff Thorpe, Nils Larsch, Ulf Möller*
11405
11406 * BN_CTX_init() has been deprecated, as BN_CTX is an opaque structure
11407   that can only be obtained through BN_CTX_new() (which implicitly
11408   initialises it). The presence of this function only made it possible
11409   to overwrite an existing structure (and cause memory leaks).
11410
11411   *Geoff Thorpe*
11412
11413 * Because of the callback-based approach for implementing LHASH as a
11414   template type, lh_insert() adds opaque objects to hash-tables and
11415   lh_doall() or lh_doall_arg() are typically used with a destructor callback
11416   to clean up those corresponding objects before destroying the hash table
11417   (and losing the object pointers). So some over-zealous constifications in
11418   LHASH have been relaxed so that lh_insert() does not take (nor store) the
11419   objects as "const" and the `lh_doall[_arg]` callback wrappers are not
11420   prototyped to have "const" restrictions on the object pointers they are
11421   given (and so aren't required to cast them away any more).
11422
11423   *Geoff Thorpe*
11424
11425 * The tmdiff.h API was so ugly and minimal that our own timing utility
11426   (speed) prefers to use its own implementation. The two implementations
11427   haven't been consolidated as yet (volunteers?) but the tmdiff API has had
11428   its object type properly exposed (MS_TM) instead of casting to/from
11429   `char *`. This may still change yet if someone realises MS_TM and
11430   `ms_time_***`
11431   aren't necessarily the greatest nomenclatures - but this is what was used
11432   internally to the implementation so I've used that for now.
11433
11434   *Geoff Thorpe*
11435
11436 * Ensure that deprecated functions do not get compiled when
11437   OPENSSL_NO_DEPRECATED is defined. Some "openssl" subcommands and a few of
11438   the self-tests were still using deprecated key-generation functions so
11439   these have been updated also.
11440
11441   *Geoff Thorpe*
11442
11443 * Reorganise PKCS#7 code to separate the digest location functionality
11444   into PKCS7_find_digest(), digest addition into PKCS7_bio_add_digest().
11445   New function PKCS7_set_digest() to set the digest type for PKCS#7
11446   digestedData type. Add additional code to correctly generate the
11447   digestedData type and add support for this type in PKCS7 initialization
11448   functions.
11449
11450   *Steve Henson*
11451
11452 * New function PKCS7_set0_type_other() this initializes a PKCS7
11453   structure of type "other".
11454
11455   *Steve Henson*
11456
11457 * Fix prime generation loop in crypto/bn/bn_prime.pl by making
11458   sure the loop does correctly stop and breaking ("division by zero")
11459   modulus operations are not performed. The (pre-generated) prime
11460   table crypto/bn/bn_prime.h was already correct, but it could not be
11461   re-generated on some platforms because of the "division by zero"
11462   situation in the script.
11463
11464   *Ralf S. Engelschall*
11465
11466 * Update support for ECC-based TLS ciphersuites according to
11467   draft-ietf-tls-ecc-03.txt: the KDF1 key derivation function with
11468   SHA-1 now is only used for "small" curves (where the
11469   representation of a field element takes up to 24 bytes); for
11470   larger curves, the field element resulting from ECDH is directly
11471   used as premaster secret.
11472
11473   *Douglas Stebila (Sun Microsystems Laboratories)*
11474
11475 * Add code for kP+lQ timings to crypto/ec/ectest.c, and add SEC2
11476   curve secp160r1 to the tests.
11477
11478   *Douglas Stebila (Sun Microsystems Laboratories)*
11479
11480 * Add the possibility to load symbols globally with DSO.
11481
11482   *Götz Babin-Ebell <babin-ebell@trustcenter.de> via Richard Levitte*
11483
11484 * Add the functions ERR_set_mark() and ERR_pop_to_mark() for better
11485   control of the error stack.
11486
11487   *Richard Levitte*
11488
11489 * Add support for STORE in ENGINE.
11490
11491   *Richard Levitte*
11492
11493 * Add the STORE type.  The intention is to provide a common interface
11494   to certificate and key stores, be they simple file-based stores, or
11495   HSM-type store, or LDAP stores, or...
11496   NOTE: The code is currently UNTESTED and isn't really used anywhere.
11497
11498   *Richard Levitte*
11499
11500 * Add a generic structure called OPENSSL_ITEM.  This can be used to
11501   pass a list of arguments to any function as well as provide a way
11502   for a function to pass data back to the caller.
11503
11504   *Richard Levitte*
11505
11506 * Add the functions BUF_strndup() and BUF_memdup().  BUF_strndup()
11507   works like BUF_strdup() but can be used to duplicate a portion of
11508   a string.  The copy gets NUL-terminated.  BUF_memdup() duplicates
11509   a memory area.
11510
11511   *Richard Levitte*
11512
11513 * Add the function sk_find_ex() which works like sk_find(), but will
11514   return an index to an element even if an exact match couldn't be
11515   found.  The index is guaranteed to point at the element where the
11516   searched-for key would be inserted to preserve sorting order.
11517
11518   *Richard Levitte*
11519
11520 * Add the function OBJ_bsearch_ex() which works like OBJ_bsearch() but
11521   takes an extra flags argument for optional functionality.  Currently,
11522   the following flags are defined:
11523
11524      OBJ_BSEARCH_VALUE_ON_NOMATCH
11525      This one gets OBJ_bsearch_ex() to return a pointer to the first
11526      element where the comparing function returns a negative or zero
11527      number.
11528
11529      OBJ_BSEARCH_FIRST_VALUE_ON_MATCH
11530      This one gets OBJ_bsearch_ex() to return a pointer to the first
11531      element where the comparing function returns zero.  This is useful
11532      if there are more than one element where the comparing function
11533      returns zero.
11534
11535   *Richard Levitte*
11536
11537 * Make it possible to create self-signed certificates with 'openssl ca'
11538   in such a way that the self-signed certificate becomes part of the
11539   CA database and uses the same mechanisms for serial number generation
11540   as all other certificate signing.  The new flag '-selfsign' enables
11541   this functionality.  Adapt CA.sh and CA.pl.in.
11542
11543   *Richard Levitte*
11544
11545 * Add functionality to check the public key of a certificate request
11546   against a given private.  This is useful to check that a certificate
11547   request can be signed by that key (self-signing).
11548
11549   *Richard Levitte*
11550
11551 * Make it possible to have multiple active certificates with the same
11552   subject in the CA index file.  This is done only if the keyword
11553   'unique_subject' is set to 'no' in the main CA section (default
11554   if 'CA_default') of the configuration file.  The value is saved
11555   with the database itself in a separate index attribute file,
11556   named like the index file with '.attr' appended to the name.
11557
11558   *Richard Levitte*
11559
11560 * Generate multi-valued AVAs using '+' notation in config files for
11561   req and dirName.
11562
11563   *Steve Henson*
11564
11565 * Support for nameConstraints certificate extension.
11566
11567   *Steve Henson*
11568
11569 * Support for policyConstraints certificate extension.
11570
11571   *Steve Henson*
11572
11573 * Support for policyMappings certificate extension.
11574
11575   *Steve Henson*
11576
11577 * Make sure the default DSA_METHOD implementation only uses its
11578   dsa_mod_exp() and/or bn_mod_exp() handlers if they are non-NULL,
11579   and change its own handlers to be NULL so as to remove unnecessary
11580   indirection. This lets alternative implementations fallback to the
11581   default implementation more easily.
11582
11583   *Geoff Thorpe*
11584
11585 * Support for directoryName in GeneralName related extensions
11586   in config files.
11587
11588   *Steve Henson*
11589
11590 * Make it possible to link applications using Makefile.shared.
11591   Make that possible even when linking against static libraries!
11592
11593   *Richard Levitte*
11594
11595 * Support for single pass processing for S/MIME signing. This now
11596   means that S/MIME signing can be done from a pipe, in addition
11597   cleartext signing (multipart/signed type) is effectively streaming
11598   and the signed data does not need to be all held in memory.
11599
11600   This is done with a new flag PKCS7_STREAM. When this flag is set
11601   PKCS7_sign() only initializes the PKCS7 structure and the actual signing
11602   is done after the data is output (and digests calculated) in
11603   SMIME_write_PKCS7().
11604
11605   *Steve Henson*
11606
11607 * Add full support for -rpath/-R, both in shared libraries and
11608   applications, at least on the platforms where it's known how
11609   to do it.
11610
11611   *Richard Levitte*
11612
11613 * In crypto/ec/ec_mult.c, implement fast point multiplication with
11614   precomputation, based on wNAF splitting: EC_GROUP_precompute_mult()
11615   will now compute a table of multiples of the generator that
11616   makes subsequent invocations of EC_POINTs_mul() or EC_POINT_mul()
11617   faster (notably in the case of a single point multiplication,
11618   scalar * generator).
11619
11620   *Nils Larsch, Bodo Moeller*
11621
11622 * IPv6 support for certificate extensions. The various extensions
11623   which use the IP:a.b.c.d can now take IPv6 addresses using the
11624   formats of RFC1884 2.2 . IPv6 addresses are now also displayed
11625   correctly.
11626
11627   *Steve Henson*
11628
11629 * Added an ENGINE that implements RSA by performing private key
11630   exponentiations with the GMP library. The conversions to and from
11631   GMP's mpz_t format aren't optimised nor are any montgomery forms
11632   cached, and on x86 it appears OpenSSL's own performance has caught up.
11633   However there are likely to be other architectures where GMP could
11634   provide a boost. This ENGINE is not built in by default, but it can be
11635   specified at Configure time and should be accompanied by the necessary
11636   linker additions, eg;
11637           ./config -DOPENSSL_USE_GMP -lgmp
11638
11639   *Geoff Thorpe*
11640
11641 * "openssl engine" will not display ENGINE/DSO load failure errors when
11642   testing availability of engines with "-t" - the old behaviour is
11643   produced by increasing the feature's verbosity with "-tt".
11644
11645   *Geoff Thorpe*
11646
11647 * ECDSA routines: under certain error conditions uninitialized BN objects
11648   could be freed. Solution: make sure initialization is performed early
11649   enough. (Reported and fix supplied by Nils Larsch <nla@trustcenter.de>
11650   via PR#459)
11651
11652   *Lutz Jaenicke*
11653
11654 * Key-generation can now be implemented in RSA_METHOD, DSA_METHOD
11655   and DH_METHOD (eg. by ENGINE implementations) to override the normal
11656   software implementations. For DSA and DH, parameter generation can
11657   also be overridden by providing the appropriate method callbacks.
11658
11659   *Geoff Thorpe*
11660
11661 * Change the "progress" mechanism used in key-generation and
11662   primality testing to functions that take a new BN_GENCB pointer in
11663   place of callback/argument pairs. The new API functions have `_ex`
11664   postfixes and the older functions are reimplemented as wrappers for
11665   the new ones. The OPENSSL_NO_DEPRECATED symbol can be used to hide
11666   declarations of the old functions to help (graceful) attempts to
11667   migrate to the new functions. Also, the new key-generation API
11668   functions operate on a caller-supplied key-structure and return
11669   success/failure rather than returning a key or NULL - this is to
11670   help make "keygen" another member function of RSA_METHOD etc.
11671
11672   Example for using the new callback interface:
11673
11674           int (*my_callback)(int a, int b, BN_GENCB *cb) = ...;
11675           void *my_arg = ...;
11676           BN_GENCB my_cb;
11677
11678           BN_GENCB_set(&my_cb, my_callback, my_arg);
11679
11680           return BN_is_prime_ex(some_bignum, BN_prime_checks, NULL, &cb);
11681           /* For the meaning of a, b in calls to my_callback(), see the
11682            * documentation of the function that calls the callback.
11683            * cb will point to my_cb; my_arg can be retrieved as cb->arg.
11684            * my_callback should return 1 if it wants BN_is_prime_ex()
11685            * to continue, or 0 to stop.
11686            */
11687
11688   *Geoff Thorpe*
11689
11690 * Change the ZLIB compression method to be stateful, and make it
11691   available to TLS with the number defined in
11692   draft-ietf-tls-compression-04.txt.
11693
11694   *Richard Levitte*
11695
11696 * Add the ASN.1 structures and functions for CertificatePair, which
11697   is defined as follows (according to X.509_4thEditionDraftV6.pdf):
11698
11699           CertificatePair ::= SEQUENCE {
11700              forward         [0]     Certificate OPTIONAL,
11701              reverse         [1]     Certificate OPTIONAL,
11702              -- at least one of the pair shall be present -- }
11703
11704   Also implement the PEM functions to read and write certificate
11705   pairs, and defined the PEM tag as "CERTIFICATE PAIR".
11706
11707   This needed to be defined, mostly for the sake of the LDAP
11708   attribute crossCertificatePair, but may prove useful elsewhere as
11709   well.
11710
11711   *Richard Levitte*
11712
11713 * Make it possible to inhibit symlinking of shared libraries in
11714   Makefile.shared, for Cygwin's sake.
11715
11716   *Richard Levitte*
11717
11718 * Extend the BIGNUM API by creating a function
11719           void BN_set_negative(BIGNUM *a, int neg);
11720   and a macro that behave like
11721           int  BN_is_negative(const BIGNUM *a);
11722
11723   to avoid the need to access 'a->neg' directly in applications.
11724
11725   *Nils Larsch*
11726
11727 * Implement fast modular reduction for pseudo-Mersenne primes
11728   used in NIST curves (crypto/bn/bn_nist.c, crypto/ec/ecp_nist.c).
11729   EC_GROUP_new_curve_GFp() will now automatically use this
11730   if applicable.
11731
11732   *Nils Larsch <nla@trustcenter.de>*
11733
11734 * Add new lock type (CRYPTO_LOCK_BN).
11735
11736   *Bodo Moeller*
11737
11738 * Change the ENGINE framework to automatically load engines
11739   dynamically from specific directories unless they could be
11740   found to already be built in or loaded.  Move all the
11741   current engines except for the cryptodev one to a new
11742   directory engines/.
11743   The engines in engines/ are built as shared libraries if
11744   the "shared" options was given to ./Configure or ./config.
11745   Otherwise, they are inserted in libcrypto.a.
11746   /usr/local/ssl/engines is the default directory for dynamic
11747   engines, but that can be overridden at configure time through
11748   the usual use of --prefix and/or --openssldir, and at run
11749   time with the environment variable OPENSSL_ENGINES.
11750
11751   *Geoff Thorpe and Richard Levitte*
11752
11753 * Add Makefile.shared, a helper makefile to build shared
11754   libraries.  Adapt Makefile.org.
11755
11756   *Richard Levitte*
11757
11758 * Add version info to Win32 DLLs.
11759
11760   *Peter 'Luna' Runestig" <peter@runestig.com>*
11761
11762 * Add new 'medium level' PKCS#12 API. Certificates and keys
11763   can be added using this API to created arbitrary PKCS#12
11764   files while avoiding the low-level API.
11765
11766   New options to PKCS12_create(), key or cert can be NULL and
11767   will then be omitted from the output file. The encryption
11768   algorithm NIDs can be set to -1 for no encryption, the mac
11769   iteration count can be set to 0 to omit the mac.
11770
11771   Enhance pkcs12 utility by making the -nokeys and -nocerts
11772   options work when creating a PKCS#12 file. New option -nomac
11773   to omit the mac, NONE can be set for an encryption algorithm.
11774   New code is modified to use the enhanced PKCS12_create()
11775   instead of the low-level API.
11776
11777   *Steve Henson*
11778
11779 * Extend ASN1 encoder to support indefinite length constructed
11780   encoding. This can output sequences tags and octet strings in
11781   this form. Modify pk7_asn1.c to support indefinite length
11782   encoding. This is experimental and needs additional code to
11783   be useful, such as an ASN1 bio and some enhanced streaming
11784   PKCS#7 code.
11785
11786   Extend template encode functionality so that tagging is passed
11787   down to the template encoder.
11788
11789   *Steve Henson*
11790
11791 * Let 'openssl req' fail if an argument to '-newkey' is not
11792   recognized instead of using RSA as a default.
11793
11794   *Bodo Moeller*
11795
11796 * Add support for ECC-based ciphersuites from draft-ietf-tls-ecc-01.txt.
11797   As these are not official, they are not included in "ALL";
11798   the "ECCdraft" ciphersuite group alias can be used to select them.
11799
11800   *Vipul Gupta and Sumit Gupta (Sun Microsystems Laboratories)*
11801
11802 * Add ECDH engine support.
11803
11804   *Nils Gura and Douglas Stebila (Sun Microsystems Laboratories)*
11805
11806 * Add ECDH in new directory crypto/ecdh/.
11807
11808   *Douglas Stebila (Sun Microsystems Laboratories)*
11809
11810 * Let BN_rand_range() abort with an error after 100 iterations
11811   without success (which indicates a broken PRNG).
11812
11813   *Bodo Moeller*
11814
11815 * Change BN_mod_sqrt() so that it verifies that the input value
11816   is really the square of the return value.  (Previously,
11817   BN_mod_sqrt would show GIGO behaviour.)
11818
11819   *Bodo Moeller*
11820
11821 * Add named elliptic curves over binary fields from X9.62, SECG,
11822   and WAP/WTLS; add OIDs that were still missing.
11823
11824   *Sheueling Chang Shantz and Douglas Stebila (Sun Microsystems Laboratories)*
11825
11826 * Extend the EC library for elliptic curves over binary fields
11827   (new files ec2_smpl.c, ec2_smpt.c, ec2_mult.c in crypto/ec/).
11828   New EC_METHOD:
11829
11830           EC_GF2m_simple_method
11831
11832   New API functions:
11833
11834           EC_GROUP_new_curve_GF2m
11835           EC_GROUP_set_curve_GF2m
11836           EC_GROUP_get_curve_GF2m
11837           EC_POINT_set_affine_coordinates_GF2m
11838           EC_POINT_get_affine_coordinates_GF2m
11839           EC_POINT_set_compressed_coordinates_GF2m
11840
11841   Point compression for binary fields is disabled by default for
11842   patent reasons (compile with OPENSSL_EC_BIN_PT_COMP defined to
11843   enable it).
11844
11845   As binary polynomials are represented as BIGNUMs, various members
11846   of the EC_GROUP and EC_POINT data structures can be shared
11847   between the implementations for prime fields and binary fields;
11848   the above `..._GF2m functions` (except for EX_GROUP_new_curve_GF2m)
11849   are essentially identical to their `..._GFp` counterparts.
11850   (For simplicity, the `..._GFp` prefix has been dropped from
11851   various internal method names.)
11852
11853   An internal 'field_div' method (similar to 'field_mul' and
11854   'field_sqr') has been added; this is used only for binary fields.
11855
11856   *Sheueling Chang Shantz and Douglas Stebila (Sun Microsystems Laboratories)*
11857
11858 * Optionally dispatch EC_POINT_mul(), EC_POINT_precompute_mult()
11859   through methods ('mul', 'precompute_mult').
11860
11861   The generic implementations (now internally called 'ec_wNAF_mul'
11862   and 'ec_wNAF_precomputed_mult') remain the default if these
11863   methods are undefined.
11864
11865   *Sheueling Chang Shantz and Douglas Stebila (Sun Microsystems Laboratories)*
11866
11867 * New function EC_GROUP_get_degree, which is defined through
11868   EC_METHOD.  For curves over prime fields, this returns the bit
11869   length of the modulus.
11870
11871   *Sheueling Chang Shantz and Douglas Stebila (Sun Microsystems Laboratories)*
11872
11873 * New functions EC_GROUP_dup, EC_POINT_dup.
11874   (These simply call ..._new  and ..._copy).
11875
11876   *Sheueling Chang Shantz and Douglas Stebila (Sun Microsystems Laboratories)*
11877
11878 * Add binary polynomial arithmetic software in crypto/bn/bn_gf2m.c.
11879   Polynomials are represented as BIGNUMs (where the sign bit is not
11880   used) in the following functions [macros]:
11881
11882           BN_GF2m_add
11883           BN_GF2m_sub             [= BN_GF2m_add]
11884           BN_GF2m_mod             [wrapper for BN_GF2m_mod_arr]
11885           BN_GF2m_mod_mul         [wrapper for BN_GF2m_mod_mul_arr]
11886           BN_GF2m_mod_sqr         [wrapper for BN_GF2m_mod_sqr_arr]
11887           BN_GF2m_mod_inv
11888           BN_GF2m_mod_exp         [wrapper for BN_GF2m_mod_exp_arr]
11889           BN_GF2m_mod_sqrt        [wrapper for BN_GF2m_mod_sqrt_arr]
11890           BN_GF2m_mod_solve_quad  [wrapper for BN_GF2m_mod_solve_quad_arr]
11891           BN_GF2m_cmp             [= BN_ucmp]
11892
11893   (Note that only the 'mod' functions are actually for fields GF(2^m).
11894   BN_GF2m_add() is misnomer, but this is for the sake of consistency.)
11895
11896   For some functions, an the irreducible polynomial defining a
11897   field can be given as an 'unsigned int[]' with strictly
11898   decreasing elements giving the indices of those bits that are set;
11899   i.e., p[] represents the polynomial
11900           f(t) = t^p[0] + t^p[1] + ... + t^p[k]
11901   where
11902           p[0] > p[1] > ... > p[k] = 0.
11903   This applies to the following functions:
11904
11905           BN_GF2m_mod_arr
11906           BN_GF2m_mod_mul_arr
11907           BN_GF2m_mod_sqr_arr
11908           BN_GF2m_mod_inv_arr        [wrapper for BN_GF2m_mod_inv]
11909           BN_GF2m_mod_div_arr        [wrapper for BN_GF2m_mod_div]
11910           BN_GF2m_mod_exp_arr
11911           BN_GF2m_mod_sqrt_arr
11912           BN_GF2m_mod_solve_quad_arr
11913           BN_GF2m_poly2arr
11914           BN_GF2m_arr2poly
11915
11916   Conversion can be performed by the following functions:
11917
11918           BN_GF2m_poly2arr
11919           BN_GF2m_arr2poly
11920
11921   bntest.c has additional tests for binary polynomial arithmetic.
11922
11923   Two implementations for BN_GF2m_mod_div() are available.
11924   The default algorithm simply uses BN_GF2m_mod_inv() and
11925   BN_GF2m_mod_mul().  The alternative algorithm is compiled in only
11926   if OPENSSL_SUN_GF2M_DIV is defined (patent pending; read the
11927   copyright notice in crypto/bn/bn_gf2m.c before enabling it).
11928
11929   *Sheueling Chang Shantz and Douglas Stebila (Sun Microsystems Laboratories)*
11930
11931 * Add new error code 'ERR_R_DISABLED' that can be used when some
11932   functionality is disabled at compile-time.
11933
11934   *Douglas Stebila <douglas.stebila@sun.com>*
11935
11936 * Change default behaviour of 'openssl asn1parse' so that more
11937   information is visible when viewing, e.g., a certificate:
11938
11939   Modify asn1_parse2 (crypto/asn1/asn1_par.c) so that in non-'dump'
11940   mode the content of non-printable OCTET STRINGs is output in a
11941   style similar to INTEGERs, but with '[HEX DUMP]' prepended to
11942   avoid the appearance of a printable string.
11943
11944   *Nils Larsch <nla@trustcenter.de>*
11945
11946 * Add 'asn1_flag' and 'asn1_form' member to EC_GROUP with access
11947   functions
11948           EC_GROUP_set_asn1_flag()
11949           EC_GROUP_get_asn1_flag()
11950           EC_GROUP_set_point_conversion_form()
11951           EC_GROUP_get_point_conversion_form()
11952   These control ASN1 encoding details:
11953   - Curves (i.e., groups) are encoded explicitly unless asn1_flag
11954     has been set to OPENSSL_EC_NAMED_CURVE.
11955   - Points are encoded in uncompressed form by default; options for
11956     asn1_for are as for point2oct, namely
11957           POINT_CONVERSION_COMPRESSED
11958           POINT_CONVERSION_UNCOMPRESSED
11959           POINT_CONVERSION_HYBRID
11960
11961   Also add 'seed' and 'seed_len' members to EC_GROUP with access
11962   functions
11963           EC_GROUP_set_seed()
11964           EC_GROUP_get0_seed()
11965           EC_GROUP_get_seed_len()
11966   This is used only for ASN1 purposes (so far).
11967
11968   *Nils Larsch <nla@trustcenter.de>*
11969
11970 * Add 'field_type' member to EC_METHOD, which holds the NID
11971   of the appropriate field type OID.  The new function
11972   EC_METHOD_get_field_type() returns this value.
11973
11974   *Nils Larsch <nla@trustcenter.de>*
11975
11976 * Add functions
11977           EC_POINT_point2bn()
11978           EC_POINT_bn2point()
11979           EC_POINT_point2hex()
11980           EC_POINT_hex2point()
11981   providing useful interfaces to EC_POINT_point2oct() and
11982   EC_POINT_oct2point().
11983
11984   *Nils Larsch <nla@trustcenter.de>*
11985
11986 * Change internals of the EC library so that the functions
11987           EC_GROUP_set_generator()
11988           EC_GROUP_get_generator()
11989           EC_GROUP_get_order()
11990           EC_GROUP_get_cofactor()
11991   are implemented directly in crypto/ec/ec_lib.c and not dispatched
11992   to methods, which would lead to unnecessary code duplication when
11993   adding different types of curves.
11994
11995   *Nils Larsch <nla@trustcenter.de> with input by Bodo Moeller*
11996
11997 * Implement compute_wNAF (crypto/ec/ec_mult.c) without BIGNUM
11998   arithmetic, and such that modified wNAFs are generated
11999   (which avoid length expansion in many cases).
12000
12001   *Bodo Moeller*
12002
12003 * Add a function EC_GROUP_check_discriminant() (defined via
12004   EC_METHOD) that verifies that the curve discriminant is non-zero.
12005
12006   Add a function EC_GROUP_check() that makes some sanity tests
12007   on a EC_GROUP, its generator and order.  This includes
12008   EC_GROUP_check_discriminant().
12009
12010   *Nils Larsch <nla@trustcenter.de>*
12011
12012 * Add ECDSA in new directory crypto/ecdsa/.
12013
12014   Add applications 'openssl ecparam' and 'openssl ecdsa'
12015   (these are based on 'openssl dsaparam' and 'openssl dsa').
12016
12017   ECDSA support is also included in various other files across the
12018   library.  Most notably,
12019   - 'openssl req' now has a '-newkey ecdsa:file' option;
12020   - EVP_PKCS82PKEY (crypto/evp/evp_pkey.c) now can handle ECDSA;
12021   - X509_PUBKEY_get (crypto/asn1/x_pubkey.c) and
12022     d2i_PublicKey (crypto/asn1/d2i_pu.c) have been modified to make
12023     them suitable for ECDSA where domain parameters must be
12024     extracted before the specific public key;
12025   - ECDSA engine support has been added.
12026
12027   *Nils Larsch <nla@trustcenter.de>*
12028
12029 * Include some named elliptic curves, and add OIDs from X9.62,
12030   SECG, and WAP/WTLS.  Each curve can be obtained from the new
12031   function
12032           EC_GROUP_new_by_curve_name(),
12033   and the list of available named curves can be obtained with
12034           EC_get_builtin_curves().
12035   Also add a 'curve_name' member to EC_GROUP objects, which can be
12036   accessed via
12037           EC_GROUP_set_curve_name()
12038           EC_GROUP_get_curve_name()
12039
12040   *Nils Larsch <larsch@trustcenter.de, Bodo Moeller*
12041
12042 * Remove a few calls to bn_wexpand() in BN_sqr() (the one in there
12043   was actually never needed) and in BN_mul().  The removal in BN_mul()
12044   required a small change in bn_mul_part_recursive() and the addition
12045   of the functions bn_cmp_part_words(), bn_sub_part_words() and
12046   bn_add_part_words(), which do the same thing as bn_cmp_words(),
12047   bn_sub_words() and bn_add_words() except they take arrays with
12048   differing sizes.
12049
12050   *Richard Levitte*
12051
12052### Changes between 0.9.7l and 0.9.7m  [23 Feb 2007]
12053
12054 * Cleanse PEM buffers before freeing them since they may contain
12055   sensitive data.
12056
12057   *Benjamin Bennett <ben@psc.edu>*
12058
12059 * Include "!eNULL" in SSL_DEFAULT_CIPHER_LIST to make sure that
12060   a ciphersuite string such as "DEFAULT:RSA" cannot enable
12061   authentication-only ciphersuites.
12062
12063   *Bodo Moeller*
12064
12065 * Since AES128 and AES256 share a single mask bit in the logic of
12066   ssl/ssl_ciph.c, the code for masking out disabled ciphers needs a
12067   kludge to work properly if AES128 is available and AES256 isn't.
12068
12069   *Victor Duchovni*
12070
12071 * Expand security boundary to match 1.1.1 module.
12072
12073   *Steve Henson*
12074
12075 * Remove redundant features: hash file source, editing of test vectors
12076   modify fipsld to use external fips_premain.c signature.
12077
12078   *Steve Henson*
12079
12080 * New perl script mkfipsscr.pl to create shell scripts or batch files to
12081   run algorithm test programs.
12082
12083   *Steve Henson*
12084
12085 * Make algorithm test programs more tolerant of whitespace.
12086
12087   *Steve Henson*
12088
12089 * Have SSL/TLS server implementation tolerate "mismatched" record
12090   protocol version while receiving ClientHello even if the
12091   ClientHello is fragmented.  (The server can't insist on the
12092   particular protocol version it has chosen before the ServerHello
12093   message has informed the client about his choice.)
12094
12095   *Bodo Moeller*
12096
12097 * Load error codes if they are not already present instead of using a
12098   static variable. This allows them to be cleanly unloaded and reloaded.
12099
12100   *Steve Henson*
12101
12102### Changes between 0.9.7k and 0.9.7l  [28 Sep 2006]
12103
12104 * Introduce limits to prevent malicious keys being able to
12105   cause a denial of service.  ([CVE-2006-2940])
12106
12107   *Steve Henson, Bodo Moeller*
12108
12109 * Fix ASN.1 parsing of certain invalid structures that can result
12110   in a denial of service.  ([CVE-2006-2937])  [Steve Henson]
12111
12112 * Fix buffer overflow in SSL_get_shared_ciphers() function.
12113   ([CVE-2006-3738]) [Tavis Ormandy and Will Drewry, Google Security Team]
12114
12115 * Fix SSL client code which could crash if connecting to a
12116   malicious SSLv2 server.  ([CVE-2006-4343])
12117
12118   *Tavis Ormandy and Will Drewry, Google Security Team*
12119
12120 * Change ciphersuite string processing so that an explicit
12121   ciphersuite selects this one ciphersuite (so that "AES256-SHA"
12122   will no longer include "AES128-SHA"), and any other similar
12123   ciphersuite (same bitmap) from *other* protocol versions (so that
12124   "RC4-MD5" will still include both the SSL 2.0 ciphersuite and the
12125   SSL 3.0/TLS 1.0 ciphersuite).  This is a backport combining
12126   changes from 0.9.8b and 0.9.8d.
12127
12128   *Bodo Moeller*
12129
12130### Changes between 0.9.7j and 0.9.7k  [05 Sep 2006]
12131
12132 * Avoid PKCS #1 v1.5 signature attack discovered by Daniel Bleichenbacher
12133   ([CVE-2006-4339])  [Ben Laurie and Google Security Team]
12134
12135 * Change the Unix randomness entropy gathering to use poll() when
12136   possible instead of select(), since the latter has some
12137   undesirable limitations.
12138
12139   *Darryl Miles via Richard Levitte and Bodo Moeller*
12140
12141 * Disable rogue ciphersuites:
12142
12143   - SSLv2 0x08 0x00 0x80 ("RC4-64-MD5")
12144   - SSLv3/TLSv1 0x00 0x61 ("EXP1024-RC2-CBC-MD5")
12145   - SSLv3/TLSv1 0x00 0x60 ("EXP1024-RC4-MD5")
12146
12147   The latter two were purportedly from
12148   draft-ietf-tls-56-bit-ciphersuites-0[01].txt, but do not really
12149   appear there.
12150
12151   Also deactivate the remaining ciphersuites from
12152   draft-ietf-tls-56-bit-ciphersuites-01.txt.  These are just as
12153   unofficial, and the ID has long expired.
12154
12155   *Bodo Moeller*
12156
12157 * Fix RSA blinding Heisenbug (problems sometimes occurred on
12158   dual-core machines) and other potential thread-safety issues.
12159
12160   *Bodo Moeller*
12161
12162### Changes between 0.9.7i and 0.9.7j  [04 May 2006]
12163
12164 * Adapt fipsld and the build system to link against the validated FIPS
12165   module in FIPS mode.
12166
12167   *Steve Henson*
12168
12169 * Fixes for VC++ 2005 build under Windows.
12170
12171   *Steve Henson*
12172
12173 * Add new Windows build target VC-32-GMAKE for VC++. This uses GNU make
12174   from a Windows bash shell such as MSYS. It is autodetected from the
12175   "config" script when run from a VC++ environment. Modify standard VC++
12176   build to use fipscanister.o from the GNU make build.
12177
12178   *Steve Henson*
12179
12180### Changes between 0.9.7h and 0.9.7i  [14 Oct 2005]
12181
12182 * Wrapped the definition of EVP_MAX_MD_SIZE in a #ifdef OPENSSL_FIPS.
12183   The value now differs depending on if you build for FIPS or not.
12184   BEWARE!  A program linked with a shared FIPSed libcrypto can't be
12185   safely run with a non-FIPSed libcrypto, as it may crash because of
12186   the difference induced by this change.
12187
12188   *Andy Polyakov*
12189
12190### Changes between 0.9.7g and 0.9.7h  [11 Oct 2005]
12191
12192 * Remove the functionality of SSL_OP_MSIE_SSLV2_RSA_PADDING
12193   (part of SSL_OP_ALL).  This option used to disable the
12194   countermeasure against man-in-the-middle protocol-version
12195   rollback in the SSL 2.0 server implementation, which is a bad
12196   idea.  ([CVE-2005-2969])
12197
12198   *Bodo Moeller; problem pointed out by Yutaka Oiwa (Research Center
12199   for Information Security, National Institute of Advanced Industrial
12200   Science and Technology [AIST, Japan)]*
12201
12202 * Minimal support for X9.31 signatures and PSS padding modes. This is
12203   mainly for FIPS compliance and not fully integrated at this stage.
12204
12205   *Steve Henson*
12206
12207 * For DSA signing, unless DSA_FLAG_NO_EXP_CONSTTIME is set, perform
12208   the exponentiation using a fixed-length exponent.  (Otherwise,
12209   the information leaked through timing could expose the secret key
12210   after many signatures; cf. Bleichenbacher's attack on DSA with
12211   biased k.)
12212
12213   *Bodo Moeller*
12214
12215 * Make a new fixed-window mod_exp implementation the default for
12216   RSA, DSA, and DH private-key operations so that the sequence of
12217   squares and multiplies and the memory access pattern are
12218   independent of the particular secret key.  This will mitigate
12219   cache-timing and potential related attacks.
12220
12221   BN_mod_exp_mont_consttime() is the new exponentiation implementation,
12222   and this is automatically used by BN_mod_exp_mont() if the new flag
12223   BN_FLG_EXP_CONSTTIME is set for the exponent.  RSA, DSA, and DH
12224   will use this BN flag for private exponents unless the flag
12225   RSA_FLAG_NO_EXP_CONSTTIME, DSA_FLAG_NO_EXP_CONSTTIME, or
12226   DH_FLAG_NO_EXP_CONSTTIME, respectively, is set.
12227
12228   *Matthew D Wood (Intel Corp), with some changes by Bodo Moeller*
12229
12230 * Change the client implementation for SSLv23_method() and
12231   SSLv23_client_method() so that is uses the SSL 3.0/TLS 1.0
12232   Client Hello message format if the SSL_OP_NO_SSLv2 option is set.
12233   (Previously, the SSL 2.0 backwards compatible Client Hello
12234   message format would be used even with SSL_OP_NO_SSLv2.)
12235
12236   *Bodo Moeller*
12237
12238 * Add support for smime-type MIME parameter in S/MIME messages which some
12239   clients need.
12240
12241   *Steve Henson*
12242
12243 * New function BN_MONT_CTX_set_locked() to set montgomery parameters in
12244   a threadsafe manner. Modify rsa code to use new function and add calls
12245   to dsa and dh code (which had race conditions before).
12246
12247   *Steve Henson*
12248
12249 * Include the fixed error library code in the C error file definitions
12250   instead of fixing them up at runtime. This keeps the error code
12251   structures constant.
12252
12253   *Steve Henson*
12254
12255### Changes between 0.9.7f and 0.9.7g  [11 Apr 2005]
12256
12257[NB: OpenSSL 0.9.7h and later 0.9.7 patch levels were released after
12258OpenSSL 0.9.8.]
12259
12260 * Fixes for newer kerberos headers. NB: the casts are needed because
12261   the 'length' field is signed on one version and unsigned on another
12262   with no (?) obvious way to tell the difference, without these VC++
12263   complains. Also the "definition" of FAR (blank) is no longer included
12264   nor is the error ENOMEM. KRB5_PRIVATE has to be set to 1 to pick up
12265   some needed definitions.
12266
12267   *Steve Henson*
12268
12269 * Undo Cygwin change.
12270
12271   *Ulf Möller*
12272
12273 * Added support for proxy certificates according to RFC 3820.
12274   Because they may be a security thread to unaware applications,
12275   they must be explicitly allowed in run-time.  See
12276   docs/HOWTO/proxy_certificates.txt for further information.
12277
12278   *Richard Levitte*
12279
12280### Changes between 0.9.7e and 0.9.7f  [22 Mar 2005]
12281
12282 * Use (SSL_RANDOM_VALUE - 4) bytes of pseudo random data when generating
12283   server and client random values. Previously
12284   (SSL_RANDOM_VALUE - sizeof(time_t)) would be used which would result in
12285   less random data when sizeof(time_t) > 4 (some 64 bit platforms).
12286
12287   This change has negligible security impact because:
12288
12289   1. Server and client random values still have 24 bytes of pseudo random
12290      data.
12291
12292   2. Server and client random values are sent in the clear in the initial
12293      handshake.
12294
12295   3. The master secret is derived using the premaster secret (48 bytes in
12296      size for static RSA ciphersuites) as well as client server and random
12297      values.
12298
12299   The OpenSSL team would like to thank the UK NISCC for bringing this issue
12300   to our attention.
12301
12302   *Stephen Henson, reported by UK NISCC*
12303
12304 * Use Windows randomness collection on Cygwin.
12305
12306   *Ulf Möller*
12307
12308 * Fix hang in EGD/PRNGD query when communication socket is closed
12309   prematurely by EGD/PRNGD.
12310
12311   *Darren Tucker <dtucker@zip.com.au> via Lutz Jänicke, resolves #1014*
12312
12313 * Prompt for pass phrases when appropriate for PKCS12 input format.
12314
12315   *Steve Henson*
12316
12317 * Back-port of selected performance improvements from development
12318   branch, as well as improved support for PowerPC platforms.
12319
12320   *Andy Polyakov*
12321
12322 * Add lots of checks for memory allocation failure, error codes to indicate
12323   failure and freeing up memory if a failure occurs.
12324
12325   *Nauticus Networks SSL Team <openssl@nauticusnet.com>, Steve Henson*
12326
12327 * Add new -passin argument to dgst.
12328
12329   *Steve Henson*
12330
12331 * Perform some character comparisons of different types in X509_NAME_cmp:
12332   this is needed for some certificates that re-encode DNs into UTF8Strings
12333   (in violation of RFC3280) and can't or won't issue name rollover
12334   certificates.
12335
12336   *Steve Henson*
12337
12338 * Make an explicit check during certificate validation to see that
12339   the CA setting in each certificate on the chain is correct.  As a
12340   side effect always do the following basic checks on extensions,
12341   not just when there's an associated purpose to the check:
12342
12343   - if there is an unhandled critical extension (unless the user
12344     has chosen to ignore this fault)
12345   - if the path length has been exceeded (if one is set at all)
12346   - that certain extensions fit the associated purpose (if one has
12347     been given)
12348
12349   *Richard Levitte*
12350
12351### Changes between 0.9.7d and 0.9.7e  [25 Oct 2004]
12352
12353 * Avoid a race condition when CRLs are checked in a multi threaded
12354   environment. This would happen due to the reordering of the revoked
12355   entries during signature checking and serial number lookup. Now the
12356   encoding is cached and the serial number sort performed under a lock.
12357   Add new STACK function sk_is_sorted().
12358
12359   *Steve Henson*
12360
12361 * Add Delta CRL to the extension code.
12362
12363   *Steve Henson*
12364
12365 * Various fixes to s3_pkt.c so alerts are sent properly.
12366
12367   *David Holmes <d.holmes@f5.com>*
12368
12369 * Reduce the chances of duplicate issuer name and serial numbers (in
12370   violation of RFC3280) using the OpenSSL certificate creation utilities.
12371   This is done by creating a random 64 bit value for the initial serial
12372   number when a serial number file is created or when a self signed
12373   certificate is created using 'openssl req -x509'. The initial serial
12374   number file is created using 'openssl x509 -next_serial' in CA.pl
12375   rather than being initialized to 1.
12376
12377   *Steve Henson*
12378
12379### Changes between 0.9.7c and 0.9.7d  [17 Mar 2004]
12380
12381 * Fix null-pointer assignment in do_change_cipher_spec() revealed
12382   by using the Codenomicon TLS Test Tool ([CVE-2004-0079])
12383
12384   *Joe Orton, Steve Henson*
12385
12386 * Fix flaw in SSL/TLS handshaking when using Kerberos ciphersuites
12387   ([CVE-2004-0112])
12388
12389   *Joe Orton, Steve Henson*
12390
12391 * Make it possible to have multiple active certificates with the same
12392   subject in the CA index file.  This is done only if the keyword
12393   'unique_subject' is set to 'no' in the main CA section (default
12394   if 'CA_default') of the configuration file.  The value is saved
12395   with the database itself in a separate index attribute file,
12396   named like the index file with '.attr' appended to the name.
12397
12398   *Richard Levitte*
12399
12400 * X509 verify fixes. Disable broken certificate workarounds when
12401   X509_V_FLAGS_X509_STRICT is set. Check CRL issuer has cRLSign set if
12402   keyUsage extension present. Don't accept CRLs with unhandled critical
12403   extensions: since verify currently doesn't process CRL extensions this
12404   rejects a CRL with *any* critical extensions. Add new verify error codes
12405   for these cases.
12406
12407   *Steve Henson*
12408
12409 * When creating an OCSP nonce use an OCTET STRING inside the extnValue.
12410   A clarification of RFC2560 will require the use of OCTET STRINGs and
12411   some implementations cannot handle the current raw format. Since OpenSSL
12412   copies and compares OCSP nonces as opaque blobs without any attempt at
12413   parsing them this should not create any compatibility issues.
12414
12415   *Steve Henson*
12416
12417 * New md flag EVP_MD_CTX_FLAG_REUSE this allows md_data to be reused when
12418   calling EVP_MD_CTX_copy_ex() to avoid calling OPENSSL_malloc(). Without
12419   this HMAC (and other) operations are several times slower than OpenSSL
12420   < 0.9.7.
12421
12422   *Steve Henson*
12423
12424 * Print out GeneralizedTime and UTCTime in ASN1_STRING_print_ex().
12425
12426   *Peter Sylvester <Peter.Sylvester@EdelWeb.fr>*
12427
12428 * Use the correct content when signing type "other".
12429
12430   *Steve Henson*
12431
12432### Changes between 0.9.7b and 0.9.7c  [30 Sep 2003]
12433
12434 * Fix various bugs revealed by running the NISCC test suite:
12435
12436   Stop out of bounds reads in the ASN1 code when presented with
12437   invalid tags (CVE-2003-0543 and CVE-2003-0544).
12438
12439   Free up ASN1_TYPE correctly if ANY type is invalid ([CVE-2003-0545]).
12440
12441   If verify callback ignores invalid public key errors don't try to check
12442   certificate signature with the NULL public key.
12443
12444   *Steve Henson*
12445
12446 * New -ignore_err option in ocsp application to stop the server
12447   exiting on the first error in a request.
12448
12449   *Steve Henson*
12450
12451 * In ssl3_accept() (ssl/s3_srvr.c) only accept a client certificate
12452   if the server requested one: as stated in TLS 1.0 and SSL 3.0
12453   specifications.
12454
12455   *Steve Henson*
12456
12457 * In ssl3_get_client_hello() (ssl/s3_srvr.c), tolerate additional
12458   extra data after the compression methods not only for TLS 1.0
12459   but also for SSL 3.0 (as required by the specification).
12460
12461   *Bodo Moeller; problem pointed out by Matthias Loepfe*
12462
12463 * Change X509_certificate_type() to mark the key as exported/exportable
12464   when it's 512 *bits* long, not 512 bytes.
12465
12466   *Richard Levitte*
12467
12468 * Change AES_cbc_encrypt() so it outputs exact multiple of
12469   blocks during encryption.
12470
12471   *Richard Levitte*
12472
12473 * Various fixes to base64 BIO and non blocking I/O. On write
12474   flushes were not handled properly if the BIO retried. On read
12475   data was not being buffered properly and had various logic bugs.
12476   This also affects blocking I/O when the data being decoded is a
12477   certain size.
12478
12479   *Steve Henson*
12480
12481 * Various S/MIME bugfixes and compatibility changes:
12482   output correct application/pkcs7 MIME type if
12483   PKCS7_NOOLDMIMETYPE is set. Tolerate some broken signatures.
12484   Output CR+LF for EOL if PKCS7_CRLFEOL is set (this makes opening
12485   of files as .eml work). Correctly handle very long lines in MIME
12486   parser.
12487
12488   *Steve Henson*
12489
12490### Changes between 0.9.7a and 0.9.7b  [10 Apr 2003]
12491
12492 * Countermeasure against the Klima-Pokorny-Rosa extension of
12493   Bleichbacher's attack on PKCS #1 v1.5 padding: treat
12494   a protocol version number mismatch like a decryption error
12495   in ssl3_get_client_key_exchange (ssl/s3_srvr.c).
12496
12497   *Bodo Moeller*
12498
12499 * Turn on RSA blinding by default in the default implementation
12500   to avoid a timing attack. Applications that don't want it can call
12501   RSA_blinding_off() or use the new flag RSA_FLAG_NO_BLINDING.
12502   They would be ill-advised to do so in most cases.
12503
12504   *Ben Laurie, Steve Henson, Geoff Thorpe, Bodo Moeller*
12505
12506 * Change RSA blinding code so that it works when the PRNG is not
12507   seeded (in this case, the secret RSA exponent is abused as
12508   an unpredictable seed -- if it is not unpredictable, there
12509   is no point in blinding anyway).  Make RSA blinding thread-safe
12510   by remembering the creator's thread ID in rsa->blinding and
12511   having all other threads use local one-time blinding factors
12512   (this requires more computation than sharing rsa->blinding, but
12513   avoids excessive locking; and if an RSA object is not shared
12514   between threads, blinding will still be very fast).
12515
12516   *Bodo Moeller*
12517
12518 * Fixed a typo bug that would cause ENGINE_set_default() to set an
12519   ENGINE as defaults for all supported algorithms irrespective of
12520   the 'flags' parameter. 'flags' is now honoured, so applications
12521   should make sure they are passing it correctly.
12522
12523   *Geoff Thorpe*
12524
12525 * Target "mingw" now allows native Windows code to be generated in
12526   the Cygwin environment as well as with the MinGW compiler.
12527
12528   *Ulf Moeller*
12529
12530### Changes between 0.9.7 and 0.9.7a  [19 Feb 2003]
12531
12532 * In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked
12533   via timing by performing a MAC computation even if incorrect
12534   block cipher padding has been found.  This is a countermeasure
12535   against active attacks where the attacker has to distinguish
12536   between bad padding and a MAC verification error. ([CVE-2003-0078])
12537
12538   *Bodo Moeller; problem pointed out by Brice Canvel (EPFL),
12539   Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and
12540   Martin Vuagnoux (EPFL, Ilion)*
12541
12542 * Make the no-err option work as intended.  The intention with no-err
12543   is not to have the whole error stack handling routines removed from
12544   libcrypto, it's only intended to remove all the function name and
12545   reason texts, thereby removing some of the footprint that may not
12546   be interesting if those errors aren't displayed anyway.
12547
12548   NOTE: it's still possible for any application or module to have its
12549   own set of error texts inserted.  The routines are there, just not
12550   used by default when no-err is given.
12551
12552   *Richard Levitte*
12553
12554 * Add support for FreeBSD on IA64.
12555
12556   *dirk.meyer@dinoex.sub.org via Richard Levitte, resolves #454*
12557
12558 * Adjust DES_cbc_cksum() so it returns the same value as the MIT
12559   Kerberos function mit_des_cbc_cksum().  Before this change,
12560   the value returned by DES_cbc_cksum() was like the one from
12561   mit_des_cbc_cksum(), except the bytes were swapped.
12562
12563   *Kevin Greaney <Kevin.Greaney@hp.com> and Richard Levitte*
12564
12565 * Allow an application to disable the automatic SSL chain building.
12566   Before this a rather primitive chain build was always performed in
12567   ssl3_output_cert_chain(): an application had no way to send the
12568   correct chain if the automatic operation produced an incorrect result.
12569
12570   Now the chain builder is disabled if either:
12571
12572   1. Extra certificates are added via SSL_CTX_add_extra_chain_cert().
12573
12574   2. The mode flag SSL_MODE_NO_AUTO_CHAIN is set.
12575
12576   The reasoning behind this is that an application would not want the
12577   auto chain building to take place if extra chain certificates are
12578   present and it might also want a means of sending no additional
12579   certificates (for example the chain has two certificates and the
12580   root is omitted).
12581
12582   *Steve Henson*
12583
12584 * Add the possibility to build without the ENGINE framework.
12585
12586   *Steven Reddie <smr@essemer.com.au> via Richard Levitte*
12587
12588 * Under Win32 gmtime() can return NULL: check return value in
12589   OPENSSL_gmtime(). Add error code for case where gmtime() fails.
12590
12591   *Steve Henson*
12592
12593 * DSA routines: under certain error conditions uninitialized BN objects
12594   could be freed. Solution: make sure initialization is performed early
12595   enough. (Reported and fix supplied by Ivan D Nestlerode <nestler@MIT.EDU>,
12596   Nils Larsch <nla@trustcenter.de> via PR#459)
12597
12598   *Lutz Jaenicke*
12599
12600 * Another fix for SSLv2 session ID handling: the session ID was incorrectly
12601   checked on reconnect on the client side, therefore session resumption
12602   could still fail with a "ssl session id is different" error. This
12603   behaviour is masked when SSL_OP_ALL is used due to
12604   SSL_OP_MICROSOFT_SESS_ID_BUG being set.
12605   Behaviour observed by Crispin Flowerday <crispin@flowerday.cx> as
12606   followup to PR #377.
12607
12608   *Lutz Jaenicke*
12609
12610 * IA-32 assembler support enhancements: unified ELF targets, support
12611   for SCO/Caldera platforms, fix for Cygwin shared build.
12612
12613   *Andy Polyakov*
12614
12615 * Add support for FreeBSD on sparc64.  As a consequence, support for
12616   FreeBSD on non-x86 processors is separate from x86 processors on
12617   the config script, much like the NetBSD support.
12618
12619   *Richard Levitte & Kris Kennaway <kris@obsecurity.org>*
12620
12621### Changes between 0.9.6h and 0.9.7  [31 Dec 2002]
12622
12623[NB: OpenSSL 0.9.6i and later 0.9.6 patch levels were released after
12624OpenSSL 0.9.7.]
12625
12626 * Fix session ID handling in SSLv2 client code: the SERVER FINISHED
12627   code (06) was taken as the first octet of the session ID and the last
12628   octet was ignored consequently. As a result SSLv2 client side session
12629   caching could not have worked due to the session ID mismatch between
12630   client and server.
12631   Behaviour observed by Crispin Flowerday <crispin@flowerday.cx> as
12632   PR #377.
12633
12634   *Lutz Jaenicke*
12635
12636 * Change the declaration of needed Kerberos libraries to use EX_LIBS
12637   instead of the special (and badly supported) LIBKRB5.  LIBKRB5 is
12638   removed entirely.
12639
12640   *Richard Levitte*
12641
12642 * The hw_ncipher.c engine requires dynamic locks.  Unfortunately, it
12643   seems that in spite of existing for more than a year, many application
12644   author have done nothing to provide the necessary callbacks, which
12645   means that this particular engine will not work properly anywhere.
12646   This is a very unfortunate situation which forces us, in the name
12647   of usability, to give the hw_ncipher.c a static lock, which is part
12648   of libcrypto.
12649   NOTE: This is for the 0.9.7 series ONLY.  This hack will never
12650   appear in 0.9.8 or later.  We EXPECT application authors to have
12651   dealt properly with this when 0.9.8 is released (unless we actually
12652   make such changes in the libcrypto locking code that changes will
12653   have to be made anyway).
12654
12655   *Richard Levitte*
12656
12657 * In asn1_d2i_read_bio() repeatedly call BIO_read() until all content
12658   octets have been read, EOF or an error occurs. Without this change
12659   some truncated ASN1 structures will not produce an error.
12660
12661   *Steve Henson*
12662
12663 * Disable Heimdal support, since it hasn't been fully implemented.
12664   Still give the possibility to force the use of Heimdal, but with
12665   warnings and a request that patches get sent to openssl-dev.
12666
12667   *Richard Levitte*
12668
12669 * Add the VC-CE target, introduce the WINCE sysname, and add
12670   INSTALL.WCE and appropriate conditionals to make it build.
12671
12672   *Steven Reddie <smr@essemer.com.au> via Richard Levitte*
12673
12674 * Change the DLL names for Cygwin to cygcrypto-x.y.z.dll and
12675   cygssl-x.y.z.dll, where x, y and z are the major, minor and
12676   edit numbers of the version.
12677
12678   *Corinna Vinschen <vinschen@redhat.com> and Richard Levitte*
12679
12680 * Introduce safe string copy and catenation functions
12681   (BUF_strlcpy() and BUF_strlcat()).
12682
12683   *Ben Laurie (CHATS) and Richard Levitte*
12684
12685 * Avoid using fixed-size buffers for one-line DNs.
12686
12687   *Ben Laurie (CHATS)*
12688
12689 * Add BUF_MEM_grow_clean() to avoid information leakage when
12690   resizing buffers containing secrets, and use where appropriate.
12691
12692   *Ben Laurie (CHATS)*
12693
12694 * Avoid using fixed size buffers for configuration file location.
12695
12696   *Ben Laurie (CHATS)*
12697
12698 * Avoid filename truncation for various CA files.
12699
12700   *Ben Laurie (CHATS)*
12701
12702 * Use sizeof in preference to magic numbers.
12703
12704   *Ben Laurie (CHATS)*
12705
12706 * Avoid filename truncation in cert requests.
12707
12708   *Ben Laurie (CHATS)*
12709
12710 * Add assertions to check for (supposedly impossible) buffer
12711   overflows.
12712
12713   *Ben Laurie (CHATS)*
12714
12715 * Don't cache truncated DNS entries in the local cache (this could
12716   potentially lead to a spoofing attack).
12717
12718   *Ben Laurie (CHATS)*
12719
12720 * Fix various buffers to be large enough for hex/decimal
12721   representations in a platform independent manner.
12722
12723   *Ben Laurie (CHATS)*
12724
12725 * Add CRYPTO_realloc_clean() to avoid information leakage when
12726   resizing buffers containing secrets, and use where appropriate.
12727
12728   *Ben Laurie (CHATS)*
12729
12730 * Add BIO_indent() to avoid much slightly worrying code to do
12731   indents.
12732
12733   *Ben Laurie (CHATS)*
12734
12735 * Convert sprintf()/BIO_puts() to BIO_printf().
12736
12737   *Ben Laurie (CHATS)*
12738
12739 * buffer_gets() could terminate with the buffer only half
12740   full. Fixed.
12741
12742   *Ben Laurie (CHATS)*
12743
12744 * Add assertions to prevent user-supplied crypto functions from
12745   overflowing internal buffers by having large block sizes, etc.
12746
12747   *Ben Laurie (CHATS)*
12748
12749 * New OPENSSL_assert() macro (similar to assert(), but enabled
12750   unconditionally).
12751
12752   *Ben Laurie (CHATS)*
12753
12754 * Eliminate unused copy of key in RC4.
12755
12756   *Ben Laurie (CHATS)*
12757
12758 * Eliminate unused and incorrectly sized buffers for IV in pem.h.
12759
12760   *Ben Laurie (CHATS)*
12761
12762 * Fix off-by-one error in EGD path.
12763
12764   *Ben Laurie (CHATS)*
12765
12766 * If RANDFILE path is too long, ignore instead of truncating.
12767
12768   *Ben Laurie (CHATS)*
12769
12770 * Eliminate unused and incorrectly sized X.509 structure
12771   CBCParameter.
12772
12773   *Ben Laurie (CHATS)*
12774
12775 * Eliminate unused and dangerous function knumber().
12776
12777   *Ben Laurie (CHATS)*
12778
12779 * Eliminate unused and dangerous structure, KSSL_ERR.
12780
12781   *Ben Laurie (CHATS)*
12782
12783 * Protect against overlong session ID context length in an encoded
12784   session object. Since these are local, this does not appear to be
12785   exploitable.
12786
12787   *Ben Laurie (CHATS)*
12788
12789 * Change from security patch (see 0.9.6e below) that did not affect
12790   the 0.9.6 release series:
12791
12792   Remote buffer overflow in SSL3 protocol - an attacker could
12793   supply an oversized master key in Kerberos-enabled versions.
12794   ([CVE-2002-0657])
12795
12796   *Ben Laurie (CHATS)*
12797
12798 * Change the SSL kerb5 codes to match RFC 2712.
12799
12800   *Richard Levitte*
12801
12802 * Make -nameopt work fully for req and add -reqopt switch.
12803
12804   *Michael Bell <michael.bell@rz.hu-berlin.de>, Steve Henson*
12805
12806 * The "block size" for block ciphers in CFB and OFB mode should be 1.
12807
12808   *Steve Henson, reported by Yngve Nysaeter Pettersen <yngve@opera.com>*
12809
12810 * Make sure tests can be performed even if the corresponding algorithms
12811   have been removed entirely.  This was also the last step to make
12812   OpenSSL compilable with DJGPP under all reasonable conditions.
12813
12814   *Richard Levitte, Doug Kaufman <dkaufman@rahul.net>*
12815
12816 * Add cipher selection rules COMPLEMENTOFALL and COMPLEMENTOFDEFAULT
12817   to allow version independent disabling of normally unselected ciphers,
12818   which may be activated as a side-effect of selecting a single cipher.
12819
12820   (E.g., cipher list string "RSA" enables ciphersuites that are left
12821   out of "ALL" because they do not provide symmetric encryption.
12822   "RSA:!COMPLEMEMENTOFALL" avoids these unsafe ciphersuites.)
12823
12824   *Lutz Jaenicke, Bodo Moeller*
12825
12826 * Add appropriate support for separate platform-dependent build
12827   directories.  The recommended way to make a platform-dependent
12828   build directory is the following (tested on Linux), maybe with
12829   some local tweaks:
12830
12831           # Place yourself outside of the OpenSSL source tree.  In
12832           # this example, the environment variable OPENSSL_SOURCE
12833           # is assumed to contain the absolute OpenSSL source directory.
12834           mkdir -p objtree/"`uname -s`-`uname -r`-`uname -m`"
12835           cd objtree/"`uname -s`-`uname -r`-`uname -m`"
12836           (cd $OPENSSL_SOURCE; find . -type f) | while read F; do
12837                   mkdir -p `dirname $F`
12838                   ln -s $OPENSSL_SOURCE/$F $F
12839           done
12840
12841   To be absolutely sure not to disturb the source tree, a "make clean"
12842   is a good thing.  If it isn't successful, don't worry about it,
12843   it probably means the source directory is very clean.
12844
12845   *Richard Levitte*
12846
12847 * Make sure any ENGINE control commands make local copies of string
12848   pointers passed to them whenever necessary. Otherwise it is possible
12849   the caller may have overwritten (or deallocated) the original string
12850   data when a later ENGINE operation tries to use the stored values.
12851
12852   *Götz Babin-Ebell <babinebell@trustcenter.de>*
12853
12854 * Improve diagnostics in file reading and command-line digests.
12855
12856   *Ben Laurie aided and abetted by Solar Designer <solar@openwall.com>*
12857
12858 * Add AES modes CFB and OFB to the object database.  Correct an
12859   error in AES-CFB decryption.
12860
12861   *Richard Levitte*
12862
12863 * Remove most calls to EVP_CIPHER_CTX_cleanup() in evp_enc.c, this
12864   allows existing EVP_CIPHER_CTX structures to be reused after
12865   calling `EVP_*Final()`. This behaviour is used by encryption
12866   BIOs and some applications. This has the side effect that
12867   applications must explicitly clean up cipher contexts with
12868   EVP_CIPHER_CTX_cleanup() or they will leak memory.
12869
12870   *Steve Henson*
12871
12872 * Check the values of dna and dnb in bn_mul_recursive before calling
12873   bn_mul_comba (a non zero value means the a or b arrays do not contain
12874   n2 elements) and fallback to bn_mul_normal if either is not zero.
12875
12876   *Steve Henson*
12877
12878 * Fix escaping of non-ASCII characters when using the -subj option
12879   of the "openssl req" command line tool. (Robert Joop <joop@fokus.gmd.de>)
12880
12881   *Lutz Jaenicke*
12882
12883 * Make object definitions compliant to LDAP (RFC2256): SN is the short
12884   form for "surname", serialNumber has no short form.
12885   Use "mail" as the short name for "rfc822Mailbox" according to RFC2798;
12886   therefore remove "mail" short name for "internet 7".
12887   The OID for unique identifiers in X509 certificates is
12888   x500UniqueIdentifier, not uniqueIdentifier.
12889   Some more OID additions. (Michael Bell <michael.bell@rz.hu-berlin.de>)
12890
12891   *Lutz Jaenicke*
12892
12893 * Add an "init" command to the ENGINE config module and auto initialize
12894   ENGINEs. Without any "init" command the ENGINE will be initialized
12895   after all ctrl commands have been executed on it. If init=1 the
12896   ENGINE is initialized at that point (ctrls before that point are run
12897   on the uninitialized ENGINE and after on the initialized one). If
12898   init=0 then the ENGINE will not be initialized at all.
12899
12900   *Steve Henson*
12901
12902 * Fix the 'app_verify_callback' interface so that the user-defined
12903   argument is actually passed to the callback: In the
12904   SSL_CTX_set_cert_verify_callback() prototype, the callback
12905   declaration has been changed from
12906           int (*cb)()
12907   into
12908           int (*cb)(X509_STORE_CTX *,void *);
12909   in ssl_verify_cert_chain (ssl/ssl_cert.c), the call
12910           i=s->ctx->app_verify_callback(&ctx)
12911   has been changed into
12912           i=s->ctx->app_verify_callback(&ctx, s->ctx->app_verify_arg).
12913
12914   To update applications using SSL_CTX_set_cert_verify_callback(),
12915   a dummy argument can be added to their callback functions.
12916
12917   *D. K. Smetters <smetters@parc.xerox.com>*
12918
12919 * Added the '4758cca' ENGINE to support IBM 4758 cards.
12920
12921   *Maurice Gittens <maurice@gittens.nl>, touchups by Geoff Thorpe*
12922
12923 * Add and OPENSSL_LOAD_CONF define which will cause
12924   OpenSSL_add_all_algorithms() to load the openssl.cnf config file.
12925   This allows older applications to transparently support certain
12926   OpenSSL features: such as crypto acceleration and dynamic ENGINE loading.
12927   Two new functions OPENSSL_add_all_algorithms_noconf() which will never
12928   load the config file and OPENSSL_add_all_algorithms_conf() which will
12929   always load it have also been added.
12930
12931   *Steve Henson*
12932
12933 * Add the OFB, CFB and CTR (all with 128 bit feedback) to AES.
12934   Adjust NIDs and EVP layer.
12935
12936   *Stephen Sprunk <stephen@sprunk.org> and Richard Levitte*
12937
12938 * Config modules support in openssl utility.
12939
12940   Most commands now load modules from the config file,
12941   though in a few (such as version) this isn't done
12942   because it couldn't be used for anything.
12943
12944   In the case of ca and req the config file used is
12945   the same as the utility itself: that is the -config
12946   command line option can be used to specify an
12947   alternative file.
12948
12949   *Steve Henson*
12950
12951 * Move default behaviour from OPENSSL_config(). If appname is NULL
12952   use "openssl_conf" if filename is NULL use default openssl config file.
12953
12954   *Steve Henson*
12955
12956 * Add an argument to OPENSSL_config() to allow the use of an alternative
12957   config section name. Add a new flag to tolerate a missing config file
12958   and move code to CONF_modules_load_file().
12959
12960   *Steve Henson*
12961
12962 * Support for crypto accelerator cards from Accelerated Encryption
12963   Processing, www.aep.ie.  (Use engine 'aep')
12964   The support was copied from 0.9.6c [engine] and adapted/corrected
12965   to work with the new engine framework.
12966
12967   *AEP Inc. and Richard Levitte*
12968
12969 * Support for SureWare crypto accelerator cards from Baltimore
12970   Technologies.  (Use engine 'sureware')
12971   The support was copied from 0.9.6c [engine] and adapted
12972   to work with the new engine framework.
12973
12974   *Richard Levitte*
12975
12976 * Have the CHIL engine fork-safe (as defined by nCipher) and actually
12977   make the newer ENGINE framework commands for the CHIL engine work.
12978
12979   *Toomas Kiisk <vix@cyber.ee> and Richard Levitte*
12980
12981 * Make it possible to produce shared libraries on ReliantUNIX.
12982
12983   *Robert Dahlem <Robert.Dahlem@ffm2.siemens.de> via Richard Levitte*
12984
12985 * Add the configuration target debug-linux-ppro.
12986   Make 'openssl rsa' use the general key loading routines
12987   implemented in `apps.c`, and make those routines able to
12988   handle the key format FORMAT_NETSCAPE and the variant
12989   FORMAT_IISSGC.
12990
12991   *Toomas Kiisk <vix@cyber.ee> via Richard Levitte*
12992
12993 * Fix a crashbug and a logic bug in hwcrhk_load_pubkey().
12994
12995   *Toomas Kiisk <vix@cyber.ee> via Richard Levitte*
12996
12997 * Add -keyform to rsautl, and document -engine.
12998
12999   *Richard Levitte, inspired by Toomas Kiisk <vix@cyber.ee>*
13000
13001 * Change BIO_new_file (crypto/bio/bss_file.c) to use new
13002   BIO_R_NO_SUCH_FILE error code rather than the generic
13003   ERR_R_SYS_LIB error code if fopen() fails with ENOENT.
13004
13005   *Ben Laurie*
13006
13007 * Add new functions
13008           ERR_peek_last_error
13009           ERR_peek_last_error_line
13010           ERR_peek_last_error_line_data.
13011   These are similar to
13012           ERR_peek_error
13013           ERR_peek_error_line
13014           ERR_peek_error_line_data,
13015   but report on the latest error recorded rather than the first one
13016   still in the error queue.
13017
13018   *Ben Laurie, Bodo Moeller*
13019
13020 * default_algorithms option in ENGINE config module. This allows things
13021   like:
13022   default_algorithms = ALL
13023   default_algorithms = RSA, DSA, RAND, CIPHERS, DIGESTS
13024
13025   *Steve Henson*
13026
13027 * Preliminary ENGINE config module.
13028
13029   *Steve Henson*
13030
13031 * New experimental application configuration code.
13032
13033   *Steve Henson*
13034
13035 * Change the AES code to follow the same name structure as all other
13036   symmetric ciphers, and behave the same way.  Move everything to
13037   the directory crypto/aes, thereby obsoleting crypto/rijndael.
13038
13039   *Stephen Sprunk <stephen@sprunk.org> and Richard Levitte*
13040
13041 * SECURITY: remove unsafe setjmp/signal interaction from ui_openssl.c.
13042
13043   *Ben Laurie and Theo de Raadt*
13044
13045 * Add option to output public keys in req command.
13046
13047   *Massimiliano Pala madwolf@openca.org*
13048
13049 * Use wNAFs in EC_POINTs_mul() for improved efficiency
13050   (up to about 10% better than before for P-192 and P-224).
13051
13052   *Bodo Moeller*
13053
13054 * New functions/macros
13055
13056           SSL_CTX_set_msg_callback(ctx, cb)
13057           SSL_CTX_set_msg_callback_arg(ctx, arg)
13058           SSL_set_msg_callback(ssl, cb)
13059           SSL_set_msg_callback_arg(ssl, arg)
13060
13061   to request calling a callback function
13062
13063           void cb(int write_p, int version, int content_type,
13064                   const void *buf, size_t len, SSL *ssl, void *arg)
13065
13066   whenever a protocol message has been completely received
13067   (write_p == 0) or sent (write_p == 1).  Here 'version' is the
13068   protocol version  according to which the SSL library interprets
13069   the current protocol message (SSL2_VERSION, SSL3_VERSION, or
13070   TLS1_VERSION).  'content_type' is 0 in the case of SSL 2.0, or
13071   the content type as defined in the SSL 3.0/TLS 1.0 protocol
13072   specification (change_cipher_spec(20), alert(21), handshake(22)).
13073   'buf' and 'len' point to the actual message, 'ssl' to the
13074   SSL object, and 'arg' is the application-defined value set by
13075   SSL[_CTX]_set_msg_callback_arg().
13076
13077   'openssl s_client' and 'openssl s_server' have new '-msg' options
13078   to enable a callback that displays all protocol messages.
13079
13080   *Bodo Moeller*
13081
13082 * Change the shared library support so shared libraries are built as
13083   soon as the corresponding static library is finished, and thereby get
13084   openssl and the test programs linked against the shared library.
13085   This still only happens when the keyword "shard" has been given to
13086   the configuration scripts.
13087
13088   NOTE: shared library support is still an experimental thing, and
13089   backward binary compatibility is still not guaranteed.
13090
13091   *"Maciej W. Rozycki" <macro@ds2.pg.gda.pl> and Richard Levitte*
13092
13093 * Add support for Subject Information Access extension.
13094
13095   *Peter Sylvester <Peter.Sylvester@EdelWeb.fr>*
13096
13097 * Make BUF_MEM_grow() behaviour more consistent: Initialise to zero
13098   additional bytes when new memory had to be allocated, not just
13099   when reusing an existing buffer.
13100
13101   *Bodo Moeller*
13102
13103 * New command line and configuration option 'utf8' for the req command.
13104   This allows field values to be specified as UTF8 strings.
13105
13106   *Steve Henson*
13107
13108 * Add -multi and -mr options to "openssl speed" - giving multiple parallel
13109   runs for the former and machine-readable output for the latter.
13110
13111   *Ben Laurie*
13112
13113 * Add '-noemailDN' option to 'openssl ca'.  This prevents inclusion
13114   of the e-mail address in the DN (i.e., it will go into a certificate
13115   extension only).  The new configuration file option 'email_in_dn = no'
13116   has the same effect.
13117
13118   *Massimiliano Pala madwolf@openca.org*
13119
13120 * Change all functions with names starting with `des_` to be starting
13121   with `DES_` instead.  Add wrappers that are compatible with libdes,
13122   but are named `_ossl_old_des_*`.  Finally, add macros that map the
13123   `des_*` symbols to the corresponding `_ossl_old_des_*` if libdes
13124   compatibility is desired.  If OpenSSL 0.9.6c compatibility is
13125   desired, the `des_*` symbols will be mapped to `DES_*`, with one
13126   exception.
13127
13128   Since we provide two compatibility mappings, the user needs to
13129   define the macro OPENSSL_DES_LIBDES_COMPATIBILITY if libdes
13130   compatibility is desired.  The default (i.e., when that macro
13131   isn't defined) is OpenSSL 0.9.6c compatibility.
13132
13133   There are also macros that enable and disable the support of old
13134   des functions altogether.  Those are OPENSSL_ENABLE_OLD_DES_SUPPORT
13135   and OPENSSL_DISABLE_OLD_DES_SUPPORT.  If none or both of those
13136   are defined, the default will apply: to support the old des routines.
13137
13138   In either case, one must include openssl/des.h to get the correct
13139   definitions.  Do not try to just include openssl/des_old.h, that
13140   won't work.
13141
13142   NOTE: This is a major break of an old API into a new one.  Software
13143   authors are encouraged to switch to the `DES_` style functions.  Some
13144   time in the future, des_old.h and the libdes compatibility functions
13145   will be disable (i.e. OPENSSL_DISABLE_OLD_DES_SUPPORT will be the
13146   default), and then completely removed.
13147
13148   *Richard Levitte*
13149
13150 * Test for certificates which contain unsupported critical extensions.
13151   If such a certificate is found during a verify operation it is
13152   rejected by default: this behaviour can be overridden by either
13153   handling the new error X509_V_ERR_UNHANDLED_CRITICAL_EXTENSION or
13154   by setting the verify flag X509_V_FLAG_IGNORE_CRITICAL. A new function
13155   X509_supported_extension() has also been added which returns 1 if a
13156   particular extension is supported.
13157
13158   *Steve Henson*
13159
13160 * Modify the behaviour of EVP cipher functions in similar way to digests
13161   to retain compatibility with existing code.
13162
13163   *Steve Henson*
13164
13165 * Modify the behaviour of EVP_DigestInit() and EVP_DigestFinal() to retain
13166   compatibility with existing code. In particular the 'ctx' parameter does
13167   not have to be to be initialized before the call to EVP_DigestInit() and
13168   it is tidied up after a call to EVP_DigestFinal(). New function
13169   EVP_DigestFinal_ex() which does not tidy up the ctx. Similarly function
13170   EVP_MD_CTX_copy() changed to not require the destination to be
13171   initialized valid and new function EVP_MD_CTX_copy_ex() added which
13172   requires the destination to be valid.
13173
13174   Modify all the OpenSSL digest calls to use EVP_DigestInit_ex(),
13175   EVP_DigestFinal_ex() and EVP_MD_CTX_copy_ex().
13176
13177   *Steve Henson*
13178
13179 * Change ssl3_get_message (ssl/s3_both.c) and the functions using it
13180   so that complete 'Handshake' protocol structures are kept in memory
13181   instead of overwriting 'msg_type' and 'length' with 'body' data.
13182
13183   *Bodo Moeller*
13184
13185 * Add an implementation of SSL_add_dir_cert_subjects_to_stack for Win32.
13186
13187   *Massimo Santin via Richard Levitte*
13188
13189 * Major restructuring to the underlying ENGINE code. This includes
13190   reduction of linker bloat, separation of pure "ENGINE" manipulation
13191   (initialisation, etc) from functionality dealing with implementations
13192   of specific crypto interfaces. This change also introduces integrated
13193   support for symmetric ciphers and digest implementations - so ENGINEs
13194   can now accelerate these by providing EVP_CIPHER and EVP_MD
13195   implementations of their own. This is detailed in
13196   [crypto/engine/README.md](crypto/engine/README.md)
13197   as it couldn't be adequately described here. However, there are a few
13198   API changes worth noting - some RSA, DSA, DH, and RAND functions that
13199   were changed in the original introduction of ENGINE code have now
13200   reverted back - the hooking from this code to ENGINE is now a good
13201   deal more passive and at run-time, operations deal directly with
13202   RSA_METHODs, DSA_METHODs (etc) as they did before, rather than
13203   dereferencing through an ENGINE pointer any more. Also, the ENGINE
13204   functions dealing with `BN_MOD_EXP[_CRT]` handlers have been removed -
13205   they were not being used by the framework as there is no concept of a
13206   BIGNUM_METHOD and they could not be generalised to the new
13207   'ENGINE_TABLE' mechanism that underlies the new code. Similarly,
13208   ENGINE_cpy() has been removed as it cannot be consistently defined in
13209   the new code.
13210
13211   *Geoff Thorpe*
13212
13213 * Change ASN1_GENERALIZEDTIME_check() to allow fractional seconds.
13214
13215   *Steve Henson*
13216
13217 * Change mkdef.pl to sort symbols that get the same entry number,
13218   and make sure the automatically generated functions `ERR_load_*`
13219   become part of libeay.num as well.
13220
13221   *Richard Levitte*
13222
13223 * New function SSL_renegotiate_pending().  This returns true once
13224   renegotiation has been requested (either SSL_renegotiate() call
13225   or HelloRequest/ClientHello received from the peer) and becomes
13226   false once a handshake has been completed.
13227   (For servers, SSL_renegotiate() followed by SSL_do_handshake()
13228   sends a HelloRequest, but does not ensure that a handshake takes
13229   place.  SSL_renegotiate_pending() is useful for checking if the
13230   client has followed the request.)
13231
13232   *Bodo Moeller*
13233
13234 * New SSL option SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION.
13235   By default, clients may request session resumption even during
13236   renegotiation (if session ID contexts permit); with this option,
13237   session resumption is possible only in the first handshake.
13238
13239   SSL_OP_ALL is now 0x00000FFFL instead of 0x000FFFFFL.  This makes
13240   more bits available for options that should not be part of
13241   SSL_OP_ALL (such as SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION).
13242
13243   *Bodo Moeller*
13244
13245 * Add some demos for certificate and certificate request creation.
13246
13247   *Steve Henson*
13248
13249 * Make maximum certificate chain size accepted from the peer application
13250   settable (`SSL*_get/set_max_cert_list()`), as proposed by
13251   "Douglas E. Engert" <deengert@anl.gov>.
13252
13253   *Lutz Jaenicke*
13254
13255 * Add support for shared libraries for Unixware-7
13256   (Boyd Lynn Gerber <gerberb@zenez.com>).
13257
13258   *Lutz Jaenicke*
13259
13260 * Add a "destroy" handler to ENGINEs that allows structural cleanup to
13261   be done prior to destruction. Use this to unload error strings from
13262   ENGINEs that load their own error strings. NB: This adds two new API
13263   functions to "get" and "set" this destroy handler in an ENGINE.
13264
13265   *Geoff Thorpe*
13266
13267 * Alter all existing ENGINE implementations (except "openssl" and
13268   "openbsd") to dynamically instantiate their own error strings. This
13269   makes them more flexible to be built both as statically-linked ENGINEs
13270   and self-contained shared-libraries loadable via the "dynamic" ENGINE.
13271   Also, add stub code to each that makes building them as self-contained
13272   shared-libraries easier (see [README-Engine.md](README-Engine.md)).
13273
13274   *Geoff Thorpe*
13275
13276 * Add a "dynamic" ENGINE that provides a mechanism for binding ENGINE
13277   implementations into applications that are completely implemented in
13278   self-contained shared-libraries. The "dynamic" ENGINE exposes control
13279   commands that can be used to configure what shared-library to load and
13280   to control aspects of the way it is handled. Also, made an update to
13281   the [README-Engine.md](README-Engine.md) file
13282   that brings its information up-to-date and
13283   provides some information and instructions on the "dynamic" ENGINE
13284   (ie. how to use it, how to build "dynamic"-loadable ENGINEs, etc).
13285
13286   *Geoff Thorpe*
13287
13288 * Make it possible to unload ranges of ERR strings with a new
13289   "ERR_unload_strings" function.
13290
13291   *Geoff Thorpe*
13292
13293 * Add a copy() function to EVP_MD.
13294
13295   *Ben Laurie*
13296
13297 * Make EVP_MD routines take a context pointer instead of just the
13298   md_data void pointer.
13299
13300   *Ben Laurie*
13301
13302 * Add flags to EVP_MD and EVP_MD_CTX. EVP_MD_FLAG_ONESHOT indicates
13303   that the digest can only process a single chunk of data
13304   (typically because it is provided by a piece of
13305   hardware). EVP_MD_CTX_FLAG_ONESHOT indicates that the application
13306   is only going to provide a single chunk of data, and hence the
13307   framework needn't accumulate the data for oneshot drivers.
13308
13309   *Ben Laurie*
13310
13311 * As with "ERR", make it possible to replace the underlying "ex_data"
13312   functions. This change also alters the storage and management of global
13313   ex_data state - it's now all inside ex_data.c and all "class" code (eg.
13314   RSA, BIO, SSL_CTX, etc) no longer stores its own STACKS and per-class
13315   index counters. The API functions that use this state have been changed
13316   to take a "class_index" rather than pointers to the class's local STACK
13317   and counter, and there is now an API function to dynamically create new
13318   classes. This centralisation allows us to (a) plug a lot of the
13319   thread-safety problems that existed, and (b) makes it possible to clean
13320   up all allocated state using "CRYPTO_cleanup_all_ex_data()". W.r.t. (b)
13321   such data would previously have always leaked in application code and
13322   workarounds were in place to make the memory debugging turn a blind eye
13323   to it. Application code that doesn't use this new function will still
13324   leak as before, but their memory debugging output will announce it now
13325   rather than letting it slide.
13326
13327   Besides the addition of CRYPTO_cleanup_all_ex_data(), another API change
13328   induced by the "ex_data" overhaul is that X509_STORE_CTX_init() now
13329   has a return value to indicate success or failure.
13330
13331   *Geoff Thorpe*
13332
13333 * Make it possible to replace the underlying "ERR" functions such that the
13334   global state (2 LHASH tables and 2 locks) is only used by the "default"
13335   implementation. This change also adds two functions to "get" and "set"
13336   the implementation prior to it being automatically set the first time
13337   any other ERR function takes place. Ie. an application can call "get",
13338   pass the return value to a module it has just loaded, and that module
13339   can call its own "set" function using that value. This means the
13340   module's "ERR" operations will use (and modify) the error state in the
13341   application and not in its own statically linked copy of OpenSSL code.
13342
13343   *Geoff Thorpe*
13344
13345 * Give DH, DSA, and RSA types their own `*_up_ref()` function to increment
13346   reference counts. This performs normal REF_PRINT/REF_CHECK macros on
13347   the operation, and provides a more encapsulated way for external code
13348   (crypto/evp/ and ssl/) to do this. Also changed the evp and ssl code
13349   to use these functions rather than manually incrementing the counts.
13350
13351   Also rename "DSO_up()" function to more descriptive "DSO_up_ref()".
13352
13353   *Geoff Thorpe*
13354
13355 * Add EVP test program.
13356
13357   *Ben Laurie*
13358
13359 * Add symmetric cipher support to ENGINE. Expect the API to change!
13360
13361   *Ben Laurie*
13362
13363 * New CRL functions: X509_CRL_set_version(), X509_CRL_set_issuer_name()
13364   X509_CRL_set_lastUpdate(), X509_CRL_set_nextUpdate(), X509_CRL_sort(),
13365   X509_REVOKED_set_serialNumber(), and X509_REVOKED_set_revocationDate().
13366   These allow a CRL to be built without having to access X509_CRL fields
13367   directly. Modify 'ca' application to use new functions.
13368
13369   *Steve Henson*
13370
13371 * Move SSL_OP_TLS_ROLLBACK_BUG out of the SSL_OP_ALL list of recommended
13372   bug workarounds. Rollback attack detection is a security feature.
13373   The problem will only arise on OpenSSL servers when TLSv1 is not
13374   available (sslv3_server_method() or SSL_OP_NO_TLSv1).
13375   Software authors not wanting to support TLSv1 will have special reasons
13376   for their choice and can explicitly enable this option.
13377
13378   *Bodo Moeller, Lutz Jaenicke*
13379
13380 * Rationalise EVP so it can be extended: don't include a union of
13381   cipher/digest structures, add init/cleanup functions for EVP_MD_CTX
13382   (similar to those existing for EVP_CIPHER_CTX).
13383   Usage example:
13384
13385           EVP_MD_CTX md;
13386
13387           EVP_MD_CTX_init(&md);             /* new function call */
13388           EVP_DigestInit(&md, EVP_sha1());
13389           EVP_DigestUpdate(&md, in, len);
13390           EVP_DigestFinal(&md, out, NULL);
13391           EVP_MD_CTX_cleanup(&md);          /* new function call */
13392
13393   *Ben Laurie*
13394
13395 * Make DES key schedule conform to the usual scheme, as well as
13396   correcting its structure. This means that calls to DES functions
13397   now have to pass a pointer to a des_key_schedule instead of a
13398   plain des_key_schedule (which was actually always a pointer
13399   anyway): E.g.,
13400
13401           des_key_schedule ks;
13402
13403           des_set_key_checked(..., &ks);
13404           des_ncbc_encrypt(..., &ks, ...);
13405
13406   (Note that a later change renames 'des_...' into 'DES_...'.)
13407
13408   *Ben Laurie*
13409
13410 * Initial reduction of linker bloat: the use of some functions, such as
13411   PEM causes large amounts of unused functions to be linked in due to
13412   poor organisation. For example pem_all.c contains every PEM function
13413   which has a knock on effect of linking in large amounts of (unused)
13414   ASN1 code. Grouping together similar functions and splitting unrelated
13415   functions prevents this.
13416
13417   *Steve Henson*
13418
13419 * Cleanup of EVP macros.
13420
13421   *Ben Laurie*
13422
13423 * Change historical references to `{NID,SN,LN}_des_ede` and ede3 to add the
13424   correct `_ecb suffix`.
13425
13426   *Ben Laurie*
13427
13428 * Add initial OCSP responder support to ocsp application. The
13429   revocation information is handled using the text based index
13430   use by the ca application. The responder can either handle
13431   requests generated internally, supplied in files (for example
13432   via a CGI script) or using an internal minimal server.
13433
13434   *Steve Henson*
13435
13436 * Add configuration choices to get zlib compression for TLS.
13437
13438   *Richard Levitte*
13439
13440 * Changes to Kerberos SSL for RFC 2712 compliance:
13441   1. Implemented real KerberosWrapper, instead of just using
13442      KRB5 AP_REQ message.  [Thanks to Simon Wilkinson <sxw@sxw.org.uk>]
13443   2. Implemented optional authenticator field of KerberosWrapper.
13444
13445   Added openssl-style ASN.1 macros for Kerberos ticket, ap_req,
13446   and authenticator structs; see crypto/krb5/.
13447
13448   Generalized Kerberos calls to support multiple Kerberos libraries.
13449   *Vern Staats <staatsvr@asc.hpc.mil>, Jeffrey Altman <jaltman@columbia.edu>
13450   via Richard Levitte*
13451
13452 * Cause 'openssl speed' to use fully hard-coded DSA keys as it
13453   already does with RSA. testdsa.h now has 'priv_key/pub_key'
13454   values for each of the key sizes rather than having just
13455   parameters (and 'speed' generating keys each time).
13456
13457   *Geoff Thorpe*
13458
13459 * Speed up EVP routines.
13460   Before:
13461crypt
13462pe              8 bytes     64 bytes    256 bytes   1024 bytes   8192 bytes
13463s-cbc           4408.85k     5560.51k     5778.46k     5862.20k     5825.16k
13464s-cbc           4389.55k     5571.17k     5792.23k     5846.91k     5832.11k
13465s-cbc           4394.32k     5575.92k     5807.44k     5848.37k     5841.30k
13466crypt
13467s-cbc           3482.66k     5069.49k     5496.39k     5614.16k     5639.28k
13468s-cbc           3480.74k     5068.76k     5510.34k     5609.87k     5635.52k
13469s-cbc           3483.72k     5067.62k     5504.60k     5708.01k     5724.80k
13470   After:
13471crypt
13472s-cbc           4660.16k     5650.19k     5807.19k     5827.13k     5783.32k
13473crypt
13474s-cbc           3624.96k     5258.21k     5530.91k     5624.30k     5628.26k
13475
13476   *Ben Laurie*
13477
13478 * Added the OS2-EMX target.
13479
13480   *"Brian Havard" <brianh@kheldar.apana.org.au> and Richard Levitte*
13481
13482 * Rewrite commands to use `NCONF` routines instead of the old `CONF`.
13483   New functions to support `NCONF` routines in extension code.
13484   New function `CONF_set_nconf()`
13485   to allow functions which take an `NCONF` to also handle the old `LHASH`
13486   structure: this means that the old `CONF` compatible routines can be
13487   retained (in particular w.rt. extensions) without having to duplicate the
13488   code. New function `X509V3_add_ext_nconf_sk()` to add extensions to a stack.
13489
13490   *Steve Henson*
13491
13492 * Enhance the general user interface with mechanisms for inner control
13493   and with possibilities to have yes/no kind of prompts.
13494
13495   *Richard Levitte*
13496
13497 * Change all calls to low-level digest routines in the library and
13498   applications to use EVP. Add missing calls to HMAC_cleanup() and
13499   don't assume HMAC_CTX can be copied using memcpy().
13500
13501   *Verdon Walker <VWalker@novell.com>, Steve Henson*
13502
13503 * Add the possibility to control engines through control names but with
13504   arbitrary arguments instead of just a string.
13505   Change the key loaders to take a UI_METHOD instead of a callback
13506   function pointer.  NOTE: this breaks binary compatibility with earlier
13507   versions of OpenSSL [engine].
13508   Adapt the nCipher code for these new conditions and add a card insertion
13509   callback.
13510
13511   *Richard Levitte*
13512
13513 * Enhance the general user interface with mechanisms to better support
13514   dialog box interfaces, application-defined prompts, the possibility
13515   to use defaults (for example default passwords from somewhere else)
13516   and interrupts/cancellations.
13517
13518   *Richard Levitte*
13519
13520 * Tidy up PKCS#12 attribute handling. Add support for the CSP name
13521   attribute in PKCS#12 files, add new -CSP option to pkcs12 utility.
13522
13523   *Steve Henson*
13524
13525 * Fix a memory leak in 'sk_dup()' in the case reallocation fails. (Also
13526   tidy up some unnecessarily weird code in 'sk_new()').
13527
13528   *Geoff, reported by Diego Tartara <dtartara@novamens.com>*
13529
13530 * Change the key loading routines for ENGINEs to use the same kind
13531   callback (pem_password_cb) as all other routines that need this
13532   kind of callback.
13533
13534   *Richard Levitte*
13535
13536 * Increase ENTROPY_NEEDED to 32 bytes, as Rijndael can operate with
13537   256 bit (=32 byte) keys. Of course seeding with more entropy bytes
13538   than this minimum value is recommended.
13539
13540   *Lutz Jaenicke*
13541
13542 * New random seeder for OpenVMS, using the system process statistics
13543   that are easily reachable.
13544
13545   *Richard Levitte*
13546
13547 * Windows apparently can't transparently handle global
13548   variables defined in DLLs. Initialisations such as:
13549
13550           const ASN1_ITEM *it = &ASN1_INTEGER_it;
13551
13552   won't compile. This is used by the any applications that need to
13553   declare their own ASN1 modules. This was fixed by adding the option
13554   EXPORT_VAR_AS_FN to all Win32 platforms, although this isn't strictly
13555   needed for static libraries under Win32.
13556
13557   *Steve Henson*
13558
13559 * New functions X509_PURPOSE_set() and X509_TRUST_set() to handle
13560   setting of purpose and trust fields. New X509_STORE trust and
13561   purpose functions and tidy up setting in other SSL functions.
13562
13563   *Steve Henson*
13564
13565 * Add copies of X509_STORE_CTX fields and callbacks to X509_STORE
13566   structure. These are inherited by X509_STORE_CTX when it is
13567   initialised. This allows various defaults to be set in the
13568   X509_STORE structure (such as flags for CRL checking and custom
13569   purpose or trust settings) for functions which only use X509_STORE_CTX
13570   internally such as S/MIME.
13571
13572   Modify X509_STORE_CTX_purpose_inherit() so it only sets purposes and
13573   trust settings if they are not set in X509_STORE. This allows X509_STORE
13574   purposes and trust (in S/MIME for example) to override any set by default.
13575
13576   Add command line options for CRL checking to smime, s_client and s_server
13577   applications.
13578
13579   *Steve Henson*
13580
13581 * Initial CRL based revocation checking. If the CRL checking flag(s)
13582   are set then the CRL is looked up in the X509_STORE structure and
13583   its validity and signature checked, then if the certificate is found
13584   in the CRL the verify fails with a revoked error.
13585
13586   Various new CRL related callbacks added to X509_STORE_CTX structure.
13587
13588   Command line options added to 'verify' application to support this.
13589
13590   This needs some additional work, such as being able to handle multiple
13591   CRLs with different times, extension based lookup (rather than just
13592   by subject name) and ultimately more complete V2 CRL extension
13593   handling.
13594
13595   *Steve Henson*
13596
13597 * Add a general user interface API (crypto/ui/).  This is designed
13598   to replace things like des_read_password and friends (backward
13599   compatibility functions using this new API are provided).
13600   The purpose is to remove prompting functions from the DES code
13601   section as well as provide for prompting through dialog boxes in
13602   a window system and the like.
13603
13604   *Richard Levitte*
13605
13606 * Add "ex_data" support to ENGINE so implementations can add state at a
13607   per-structure level rather than having to store it globally.
13608
13609   *Geoff*
13610
13611 * Make it possible for ENGINE structures to be copied when retrieved by
13612   ENGINE_by_id() if the ENGINE specifies a new flag: ENGINE_FLAGS_BY_ID_COPY.
13613   This causes the "original" ENGINE structure to act like a template,
13614   analogous to the RSA vs. RSA_METHOD type of separation. Because of this
13615   operational state can be localised to each ENGINE structure, despite the
13616   fact they all share the same "methods". New ENGINE structures returned in
13617   this case have no functional references and the return value is the single
13618   structural reference. This matches the single structural reference returned
13619   by ENGINE_by_id() normally, when it is incremented on the pre-existing
13620   ENGINE structure.
13621
13622   *Geoff*
13623
13624 * Fix ASN1 decoder when decoding type ANY and V_ASN1_OTHER: since this
13625   needs to match any other type at all we need to manually clear the
13626   tag cache.
13627
13628   *Steve Henson*
13629
13630 * Changes to the "openssl engine" utility to include;
13631   - verbosity levels ('-v', '-vv', and '-vvv') that provide information
13632     about an ENGINE's available control commands.
13633   - executing control commands from command line arguments using the
13634     '-pre' and '-post' switches. '-post' is only used if '-t' is
13635     specified and the ENGINE is successfully initialised. The syntax for
13636     the individual commands are colon-separated, for example;
13637           openssl engine chil -pre FORK_CHECK:0 -pre SO_PATH:/lib/test.so
13638
13639   *Geoff*
13640
13641 * New dynamic control command support for ENGINEs. ENGINEs can now
13642   declare their own commands (numbers), names (strings), descriptions,
13643   and input types for run-time discovery by calling applications. A
13644   subset of these commands are implicitly classed as "executable"
13645   depending on their input type, and only these can be invoked through
13646   the new string-based API function ENGINE_ctrl_cmd_string(). (Eg. this
13647   can be based on user input, config files, etc). The distinction is
13648   that "executable" commands cannot return anything other than a boolean
13649   result and can only support numeric or string input, whereas some
13650   discoverable commands may only be for direct use through
13651   ENGINE_ctrl(), eg. supporting the exchange of binary data, function
13652   pointers, or other custom uses. The "executable" commands are to
13653   support parameterisations of ENGINE behaviour that can be
13654   unambiguously defined by ENGINEs and used consistently across any
13655   OpenSSL-based application. Commands have been added to all the
13656   existing hardware-supporting ENGINEs, noticeably "SO_PATH" to allow
13657   control over shared-library paths without source code alterations.
13658
13659   *Geoff*
13660
13661 * Changed all ENGINE implementations to dynamically allocate their
13662   ENGINEs rather than declaring them statically. Apart from this being
13663   necessary with the removal of the ENGINE_FLAGS_MALLOCED distinction,
13664   this also allows the implementations to compile without using the
13665   internal engine_int.h header.
13666
13667   *Geoff*
13668
13669 * Minor adjustment to "rand" code. RAND_get_rand_method() now returns a
13670   'const' value. Any code that should be able to modify a RAND_METHOD
13671   should already have non-const pointers to it (ie. they should only
13672   modify their own ones).
13673
13674   *Geoff*
13675
13676 * Made a variety of little tweaks to the ENGINE code.
13677   - "atalla" and "ubsec" string definitions were moved from header files
13678     to C code. "nuron" string definitions were placed in variables
13679     rather than hard-coded - allowing parameterisation of these values
13680     later on via ctrl() commands.
13681   - Removed unused "#if 0"'d code.
13682   - Fixed engine list iteration code so it uses ENGINE_free() to release
13683     structural references.
13684   - Constified the RAND_METHOD element of ENGINE structures.
13685   - Constified various get/set functions as appropriate and added
13686     missing functions (including a catch-all ENGINE_cpy that duplicates
13687     all ENGINE values onto a new ENGINE except reference counts/state).
13688   - Removed NULL parameter checks in get/set functions. Setting a method
13689     or function to NULL is a way of cancelling out a previously set
13690     value.  Passing a NULL ENGINE parameter is just plain stupid anyway
13691     and doesn't justify the extra error symbols and code.
13692   - Deprecate the ENGINE_FLAGS_MALLOCED define and move the area for
13693     flags from engine_int.h to engine.h.
13694   - Changed prototypes for ENGINE handler functions (init(), finish(),
13695     ctrl(), key-load functions, etc) to take an (ENGINE*) parameter.
13696
13697   *Geoff*
13698
13699 * Implement binary inversion algorithm for BN_mod_inverse in addition
13700   to the algorithm using long division.  The binary algorithm can be
13701   used only if the modulus is odd.  On 32-bit systems, it is faster
13702   only for relatively small moduli (roughly 20-30% for 128-bit moduli,
13703   roughly 5-15% for 256-bit moduli), so we use it only for moduli
13704   up to 450 bits.  In 64-bit environments, the binary algorithm
13705   appears to be advantageous for much longer moduli; here we use it
13706   for moduli up to 2048 bits.
13707
13708   *Bodo Moeller*
13709
13710 * Rewrite CHOICE field setting in ASN1_item_ex_d2i(). The old code
13711   could not support the combine flag in choice fields.
13712
13713   *Steve Henson*
13714
13715 * Add a 'copy_extensions' option to the 'ca' utility. This copies
13716   extensions from a certificate request to the certificate.
13717
13718   *Steve Henson*
13719
13720 * Allow multiple 'certopt' and 'nameopt' options to be separated
13721   by commas. Add 'namopt' and 'certopt' options to the 'ca' config
13722   file: this allows the display of the certificate about to be
13723   signed to be customised, to allow certain fields to be included
13724   or excluded and extension details. The old system didn't display
13725   multicharacter strings properly, omitted fields not in the policy
13726   and couldn't display additional details such as extensions.
13727
13728   *Steve Henson*
13729
13730 * Function EC_POINTs_mul for multiple scalar multiplication
13731   of an arbitrary number of elliptic curve points
13732           \sum scalars[i]*points[i],
13733   optionally including the generator defined for the EC_GROUP:
13734           scalar*generator +  \sum scalars[i]*points[i].
13735
13736   EC_POINT_mul is a simple wrapper function for the typical case
13737   that the point list has just one item (besides the optional
13738   generator).
13739
13740   *Bodo Moeller*
13741
13742 * First EC_METHODs for curves over GF(p):
13743
13744   EC_GFp_simple_method() uses the basic BN_mod_mul and BN_mod_sqr
13745   operations and provides various method functions that can also
13746   operate with faster implementations of modular arithmetic.
13747
13748   EC_GFp_mont_method() reuses most functions that are part of
13749   EC_GFp_simple_method, but uses Montgomery arithmetic.
13750
13751   *Bodo Moeller; point addition and point doubling
13752   implementation directly derived from source code provided by
13753   Lenka Fibikova <fibikova@exp-math.uni-essen.de>*
13754
13755 * Framework for elliptic curves (crypto/ec/ec.h, crypto/ec/ec_lcl.h,
13756   crypto/ec/ec_lib.c):
13757
13758   Curves are EC_GROUP objects (with an optional group generator)
13759   based on EC_METHODs that are built into the library.
13760
13761   Points are EC_POINT objects based on EC_GROUP objects.
13762
13763   Most of the framework would be able to handle curves over arbitrary
13764   finite fields, but as there are no obvious types for fields other
13765   than GF(p), some functions are limited to that for now.
13766
13767   *Bodo Moeller*
13768
13769 * Add the -HTTP option to s_server.  It is similar to -WWW, but requires
13770   that the file contains a complete HTTP response.
13771
13772   *Richard Levitte*
13773
13774 * Add the ec directory to mkdef.pl and mkfiles.pl. In mkdef.pl
13775   change the def and num file printf format specifier from "%-40sXXX"
13776   to "%-39s XXX". The latter will always guarantee a space after the
13777   field while the former will cause them to run together if the field
13778   is 40 of more characters long.
13779
13780   *Steve Henson*
13781
13782 * Constify the cipher and digest 'method' functions and structures
13783   and modify related functions to take constant EVP_MD and EVP_CIPHER
13784   pointers.
13785
13786   *Steve Henson*
13787
13788 * Hide BN_CTX structure details in bn_lcl.h instead of publishing them
13789   in <openssl/bn.h>.  Also further increase BN_CTX_NUM to 32.
13790
13791   *Bodo Moeller*
13792
13793 * Modify `EVP_Digest*()` routines so they now return values. Although the
13794   internal software routines can never fail additional hardware versions
13795   might.
13796
13797   *Steve Henson*
13798
13799 * Clean up crypto/err/err.h and change some error codes to avoid conflicts:
13800
13801   Previously ERR_R_FATAL was too small and coincided with ERR_LIB_PKCS7
13802   (= ERR_R_PKCS7_LIB); it is now 64 instead of 32.
13803
13804   ASN1 error codes
13805           ERR_R_NESTED_ASN1_ERROR
13806           ...
13807           ERR_R_MISSING_ASN1_EOS
13808   were 4 .. 9, conflicting with
13809           ERR_LIB_RSA (= ERR_R_RSA_LIB)
13810           ...
13811           ERR_LIB_PEM (= ERR_R_PEM_LIB).
13812   They are now 58 .. 63 (i.e., just below ERR_R_FATAL).
13813
13814   Add new error code 'ERR_R_INTERNAL_ERROR'.
13815
13816   *Bodo Moeller*
13817
13818 * Don't overuse locks in crypto/err/err.c: For data retrieval, CRYPTO_r_lock
13819   suffices.
13820
13821   *Bodo Moeller*
13822
13823 * New option '-subj arg' for 'openssl req' and 'openssl ca'.  This
13824   sets the subject name for a new request or supersedes the
13825   subject name in a given request. Formats that can be parsed are
13826           'CN=Some Name, OU=myOU, C=IT'
13827   and
13828           'CN=Some Name/OU=myOU/C=IT'.
13829
13830   Add options '-batch' and '-verbose' to 'openssl req'.
13831
13832   *Massimiliano Pala <madwolf@hackmasters.net>*
13833
13834 * Introduce the possibility to access global variables through
13835   functions on platform were that's the best way to handle exporting
13836   global variables in shared libraries.  To enable this functionality,
13837   one must configure with "EXPORT_VAR_AS_FN" or defined the C macro
13838   "OPENSSL_EXPORT_VAR_AS_FUNCTION" in crypto/opensslconf.h (the latter
13839   is normally done by Configure or something similar).
13840
13841   To implement a global variable, use the macro OPENSSL_IMPLEMENT_GLOBAL
13842   in the source file (foo.c) like this:
13843
13844           OPENSSL_IMPLEMENT_GLOBAL(int,foo)=1;
13845           OPENSSL_IMPLEMENT_GLOBAL(double,bar);
13846
13847   To declare a global variable, use the macros OPENSSL_DECLARE_GLOBAL
13848   and OPENSSL_GLOBAL_REF in the header file (foo.h) like this:
13849
13850           OPENSSL_DECLARE_GLOBAL(int,foo);
13851           #define foo OPENSSL_GLOBAL_REF(foo)
13852           OPENSSL_DECLARE_GLOBAL(double,bar);
13853           #define bar OPENSSL_GLOBAL_REF(bar)
13854
13855   The #defines are very important, and therefore so is including the
13856   header file everywhere where the defined globals are used.
13857
13858   The macro OPENSSL_EXPORT_VAR_AS_FUNCTION also affects the definition
13859   of ASN.1 items, but that structure is a bit different.
13860
13861   The largest change is in util/mkdef.pl which has been enhanced with
13862   better and easier to understand logic to choose which symbols should
13863   go into the Windows .def files as well as a number of fixes and code
13864   cleanup (among others, algorithm keywords are now sorted
13865   lexicographically to avoid constant rewrites).
13866
13867   *Richard Levitte*
13868
13869 * In BN_div() keep a copy of the sign of 'num' before writing the
13870   result to 'rm' because if rm==num the value will be overwritten
13871   and produce the wrong result if 'num' is negative: this caused
13872   problems with BN_mod() and BN_nnmod().
13873
13874   *Steve Henson*
13875
13876 * Function OCSP_request_verify(). This checks the signature on an
13877   OCSP request and verifies the signer certificate. The signer
13878   certificate is just checked for a generic purpose and OCSP request
13879   trust settings.
13880
13881   *Steve Henson*
13882
13883 * Add OCSP_check_validity() function to check the validity of OCSP
13884   responses. OCSP responses are prepared in real time and may only
13885   be a few seconds old. Simply checking that the current time lies
13886   between thisUpdate and nextUpdate max reject otherwise valid responses
13887   caused by either OCSP responder or client clock inaccuracy. Instead
13888   we allow thisUpdate and nextUpdate to fall within a certain period of
13889   the current time. The age of the response can also optionally be
13890   checked. Two new options -validity_period and -status_age added to
13891   ocsp utility.
13892
13893   *Steve Henson*
13894
13895 * If signature or public key algorithm is unrecognized print out its
13896   OID rather that just UNKNOWN.
13897
13898   *Steve Henson*
13899
13900 * Change OCSP_cert_to_id() to tolerate a NULL subject certificate and
13901   OCSP_cert_id_new() a NULL serialNumber. This allows a partial certificate
13902   ID to be generated from the issuer certificate alone which can then be
13903   passed to OCSP_id_issuer_cmp().
13904
13905   *Steve Henson*
13906
13907 * New compilation option ASN1_ITEM_FUNCTIONS. This causes the new
13908   ASN1 modules to export functions returning ASN1_ITEM pointers
13909   instead of the ASN1_ITEM structures themselves. This adds several
13910   new macros which allow the underlying ASN1 function/structure to
13911   be accessed transparently. As a result code should not use ASN1_ITEM
13912   references directly (such as &X509_it) but instead use the relevant
13913   macros (such as ASN1_ITEM_rptr(X509)). This option is to allow
13914   use of the new ASN1 code on platforms where exporting structures
13915   is problematical (for example in shared libraries) but exporting
13916   functions returning pointers to structures is not.
13917
13918   *Steve Henson*
13919
13920 * Add support for overriding the generation of SSL/TLS session IDs.
13921   These callbacks can be registered either in an SSL_CTX or per SSL.
13922   The purpose of this is to allow applications to control, if they wish,
13923   the arbitrary values chosen for use as session IDs, particularly as it
13924   can be useful for session caching in multiple-server environments. A
13925   command-line switch for testing this (and any client code that wishes
13926   to use such a feature) has been added to "s_server".
13927
13928   *Geoff Thorpe, Lutz Jaenicke*
13929
13930 * Modify mkdef.pl to recognise and parse preprocessor conditionals
13931   of the form `#if defined(...) || defined(...) || ...` and
13932   `#if !defined(...) && !defined(...) && ...`.  This also avoids
13933   the growing number of special cases it was previously handling.
13934
13935   *Richard Levitte*
13936
13937 * Make all configuration macros available for application by making
13938   sure they are available in opensslconf.h, by giving them names starting
13939   with `OPENSSL_` to avoid conflicts with other packages and by making
13940   sure e_os2.h will cover all platform-specific cases together with
13941   opensslconf.h.
13942   Additionally, it is now possible to define configuration/platform-
13943   specific names (called "system identities").  In the C code, these
13944   are prefixed with `OPENSSL_SYSNAME_`.  e_os2.h will create another
13945   macro with the name beginning with `OPENSSL_SYS_`, which is determined
13946   from `OPENSSL_SYSNAME_*` or compiler-specific macros depending on
13947   what is available.
13948
13949   *Richard Levitte*
13950
13951 * New option -set_serial to 'req' and 'x509' this allows the serial
13952   number to use to be specified on the command line. Previously self
13953   signed certificates were hard coded with serial number 0 and the
13954   CA options of 'x509' had to use a serial number in a file which was
13955   auto incremented.
13956
13957   *Steve Henson*
13958
13959 * New options to 'ca' utility to support V2 CRL entry extensions.
13960   Currently CRL reason, invalidity date and hold instruction are
13961   supported. Add new CRL extensions to V3 code and some new objects.
13962
13963   *Steve Henson*
13964
13965 * New function EVP_CIPHER_CTX_set_padding() this is used to
13966   disable standard block padding (aka PKCS#5 padding) in the EVP
13967   API, which was previously mandatory. This means that the data is
13968   not padded in any way and so the total length much be a multiple
13969   of the block size, otherwise an error occurs.
13970
13971   *Steve Henson*
13972
13973 * Initial (incomplete) OCSP SSL support.
13974
13975   *Steve Henson*
13976
13977 * New function OCSP_parse_url(). This splits up a URL into its host,
13978   port and path components: primarily to parse OCSP URLs. New -url
13979   option to ocsp utility.
13980
13981   *Steve Henson*
13982
13983 * New nonce behavior. The return value of OCSP_check_nonce() now
13984   reflects the various checks performed. Applications can decide
13985   whether to tolerate certain situations such as an absent nonce
13986   in a response when one was present in a request: the ocsp application
13987   just prints out a warning. New function OCSP_add1_basic_nonce()
13988   this is to allow responders to include a nonce in a response even if
13989   the request is nonce-less.
13990
13991   *Steve Henson*
13992
13993 * Disable stdin buffering in `load_cert()` (`apps/apps.c`) so that no certs are
13994   skipped when using openssl x509 multiple times on a single input file,
13995   e.g. `(openssl x509 -out cert1; openssl x509 -out cert2) <certs`.
13996
13997   *Bodo Moeller*
13998
13999 * Make ASN1_UTCTIME_set_string() and ASN1_GENERALIZEDTIME_set_string()
14000   set string type: to handle setting ASN1_TIME structures. Fix ca
14001   utility to correctly initialize revocation date of CRLs.
14002
14003   *Steve Henson*
14004
14005 * New option SSL_OP_CIPHER_SERVER_PREFERENCE allows the server to override
14006   the clients preferred ciphersuites and rather use its own preferences.
14007   Should help to work around M$ SGC (Server Gated Cryptography) bug in
14008   Internet Explorer by ensuring unchanged hash method during stepup.
14009   (Also replaces the broken/deactivated SSL_OP_NON_EXPORT_FIRST option.)
14010
14011   *Lutz Jaenicke*
14012
14013 * Make mkdef.pl recognise all DECLARE_ASN1 macros, change rijndael
14014   to aes and add a new 'exist' option to print out symbols that don't
14015   appear to exist.
14016
14017   *Steve Henson*
14018
14019 * Additional options to ocsp utility to allow flags to be set and
14020   additional certificates supplied.
14021
14022   *Steve Henson*
14023
14024 * Add the option -VAfile to 'openssl ocsp', so the user can give the
14025   OCSP client a number of certificate to only verify the response
14026   signature against.
14027
14028   *Richard Levitte*
14029
14030 * Update Rijndael code to version 3.0 and change EVP AES ciphers to
14031   handle the new API. Currently only ECB, CBC modes supported. Add new
14032   AES OIDs.
14033
14034   Add TLS AES ciphersuites as described in RFC3268, "Advanced
14035   Encryption Standard (AES) Ciphersuites for Transport Layer
14036   Security (TLS)".  (In beta versions of OpenSSL 0.9.7, these were
14037   not enabled by default and were not part of the "ALL" ciphersuite
14038   alias because they were not yet official; they could be
14039   explicitly requested by specifying the "AESdraft" ciphersuite
14040   group alias.  In the final release of OpenSSL 0.9.7, the group
14041   alias is called "AES" and is part of "ALL".)
14042
14043   *Ben Laurie, Steve  Henson, Bodo Moeller*
14044
14045 * New function OCSP_copy_nonce() to copy nonce value (if present) from
14046   request to response.
14047
14048   *Steve Henson*
14049
14050 * Functions for OCSP responders. OCSP_request_onereq_count(),
14051   OCSP_request_onereq_get0(), OCSP_onereq_get0_id() and OCSP_id_get0_info()
14052   extract information from a certificate request. OCSP_response_create()
14053   creates a response and optionally adds a basic response structure.
14054   OCSP_basic_add1_status() adds a complete single response to a basic
14055   response and returns the OCSP_SINGLERESP structure just added (to allow
14056   extensions to be included for example). OCSP_basic_add1_cert() adds a
14057   certificate to a basic response and OCSP_basic_sign() signs a basic
14058   response with various flags. New helper functions ASN1_TIME_check()
14059   (checks validity of ASN1_TIME structure) and ASN1_TIME_to_generalizedtime()
14060   (converts ASN1_TIME to GeneralizedTime).
14061
14062   *Steve Henson*
14063
14064 * Various new functions. EVP_Digest() combines EVP_Digest{Init,Update,Final}()
14065   in a single operation. X509_get0_pubkey_bitstr() extracts the public_key
14066   structure from a certificate. X509_pubkey_digest() digests the public_key
14067   contents: this is used in various key identifiers.
14068
14069   *Steve Henson*
14070
14071 * Make sk_sort() tolerate a NULL argument.
14072
14073   *Steve Henson reported by Massimiliano Pala <madwolf@comune.modena.it>*
14074
14075 * New OCSP verify flag OCSP_TRUSTOTHER. When set the "other" certificates
14076   passed by the function are trusted implicitly. If any of them signed the
14077   response then it is assumed to be valid and is not verified.
14078
14079   *Steve Henson*
14080
14081 * In PKCS7_set_type() initialise content_type in PKCS7_ENC_CONTENT
14082   to data. This was previously part of the PKCS7 ASN1 code. This
14083   was causing problems with OpenSSL created PKCS#12 and PKCS#7 structures.
14084   *Steve Henson, reported by Kenneth R. Robinette
14085                              <support@securenetterm.com>*
14086
14087 * Add CRYPTO_push_info() and CRYPTO_pop_info() calls to new ASN1
14088   routines: without these tracing memory leaks is very painful.
14089   Fix leaks in PKCS12 and PKCS7 routines.
14090
14091   *Steve Henson*
14092
14093 * Make X509_time_adj() cope with the new behaviour of ASN1_TIME_new().
14094   Previously it initialised the 'type' argument to V_ASN1_UTCTIME which
14095   effectively meant GeneralizedTime would never be used. Now it
14096   is initialised to -1 but X509_time_adj() now has to check the value
14097   and use ASN1_TIME_set() if the value is not V_ASN1_UTCTIME or
14098   V_ASN1_GENERALIZEDTIME, without this it always uses GeneralizedTime.
14099   *Steve Henson, reported by Kenneth R. Robinette
14100                              <support@securenetterm.com>*
14101
14102 * Fixes to BN_to_ASN1_INTEGER when bn is zero. This would previously
14103   result in a zero length in the ASN1_INTEGER structure which was
14104   not consistent with the structure when d2i_ASN1_INTEGER() was used
14105   and would cause ASN1_INTEGER_cmp() to fail. Enhance s2i_ASN1_INTEGER()
14106   to cope with hex and negative integers. Fix bug in i2a_ASN1_INTEGER()
14107   where it did not print out a minus for negative ASN1_INTEGER.
14108
14109   *Steve Henson*
14110
14111 * Add summary printout to ocsp utility. The various functions which
14112   convert status values to strings have been renamed to:
14113   OCSP_response_status_str(), OCSP_cert_status_str() and
14114   OCSP_crl_reason_str() and are no longer static. New options
14115   to verify nonce values and to disable verification. OCSP response
14116   printout format cleaned up.
14117
14118   *Steve Henson*
14119
14120 * Add additional OCSP certificate checks. These are those specified
14121   in RFC2560. This consists of two separate checks: the CA of the
14122   certificate being checked must either be the OCSP signer certificate
14123   or the issuer of the OCSP signer certificate. In the latter case the
14124   OCSP signer certificate must contain the OCSP signing extended key
14125   usage. This check is performed by attempting to match the OCSP
14126   signer or the OCSP signer CA to the issuerNameHash and issuerKeyHash
14127   in the OCSP_CERTID structures of the response.
14128
14129   *Steve Henson*
14130
14131 * Initial OCSP certificate verification added to OCSP_basic_verify()
14132   and related routines. This uses the standard OpenSSL certificate
14133   verify routines to perform initial checks (just CA validity) and
14134   to obtain the certificate chain. Then additional checks will be
14135   performed on the chain. Currently the root CA is checked to see
14136   if it is explicitly trusted for OCSP signing. This is used to set
14137   a root CA as a global signing root: that is any certificate that
14138   chains to that CA is an acceptable OCSP signing certificate.
14139
14140   *Steve Henson*
14141
14142 * New '-extfile ...' option to 'openssl ca' for reading X.509v3
14143   extensions from a separate configuration file.
14144   As when reading extensions from the main configuration file,
14145   the '-extensions ...' option may be used for specifying the
14146   section to use.
14147
14148   *Massimiliano Pala <madwolf@comune.modena.it>*
14149
14150 * New OCSP utility. Allows OCSP requests to be generated or
14151   read. The request can be sent to a responder and the output
14152   parsed, outputted or printed in text form. Not complete yet:
14153   still needs to check the OCSP response validity.
14154
14155   *Steve Henson*
14156
14157 * New subcommands for 'openssl ca':
14158   `openssl ca -status <serial>` prints the status of the cert with
14159   the given serial number (according to the index file).
14160   `openssl ca -updatedb` updates the expiry status of certificates
14161   in the index file.
14162
14163   *Massimiliano Pala <madwolf@comune.modena.it>*
14164
14165 * New '-newreq-nodes' command option to CA.pl.  This is like
14166   '-newreq', but calls 'openssl req' with the '-nodes' option
14167   so that the resulting key is not encrypted.
14168
14169   *Damien Miller <djm@mindrot.org>*
14170
14171 * New configuration for the GNU Hurd.
14172
14173   *Jonathan Bartlett <johnnyb@wolfram.com> via Richard Levitte*
14174
14175 * Initial code to implement OCSP basic response verify. This
14176   is currently incomplete. Currently just finds the signer's
14177   certificate and verifies the signature on the response.
14178
14179   *Steve Henson*
14180
14181 * New SSLeay_version code SSLEAY_DIR to determine the compiled-in
14182   value of OPENSSLDIR.  This is available via the new '-d' option
14183   to 'openssl version', and is also included in 'openssl version -a'.
14184
14185   *Bodo Moeller*
14186
14187 * Allowing defining memory allocation callbacks that will be given
14188   file name and line number information in additional arguments
14189   (a `const char*` and an int).  The basic functionality remains, as
14190   well as the original possibility to just replace malloc(),
14191   realloc() and free() by functions that do not know about these
14192   additional arguments.  To register and find out the current
14193   settings for extended allocation functions, the following
14194   functions are provided:
14195
14196           CRYPTO_set_mem_ex_functions
14197           CRYPTO_set_locked_mem_ex_functions
14198           CRYPTO_get_mem_ex_functions
14199           CRYPTO_get_locked_mem_ex_functions
14200
14201   These work the same way as CRYPTO_set_mem_functions and friends.
14202   `CRYPTO_get_[locked_]mem_functions` now writes 0 where such an
14203   extended allocation function is enabled.
14204   Similarly, `CRYPTO_get_[locked_]mem_ex_functions` writes 0 where
14205   a conventional allocation function is enabled.
14206
14207   *Richard Levitte, Bodo Moeller*
14208
14209 * Finish off removing the remaining LHASH function pointer casts.
14210   There should no longer be any prototype-casting required when using
14211   the LHASH abstraction, and any casts that remain are "bugs". See
14212   the callback types and macros at the head of lhash.h for details
14213   (and "OBJ_cleanup" in crypto/objects/obj_dat.c as an example).
14214
14215   *Geoff Thorpe*
14216
14217 * Add automatic query of EGD sockets in RAND_poll() for the unix variant.
14218   If /dev/[u]random devices are not available or do not return enough
14219   entropy, EGD style sockets (served by EGD or PRNGD) will automatically
14220   be queried.
14221   The locations /var/run/egd-pool, /dev/egd-pool, /etc/egd-pool, and
14222   /etc/entropy will be queried once each in this sequence, querying stops
14223   when enough entropy was collected without querying more sockets.
14224
14225   *Lutz Jaenicke*
14226
14227 * Change the Unix RAND_poll() variant to be able to poll several
14228   random devices, as specified by DEVRANDOM, until a sufficient amount
14229   of data has been collected.   We spend at most 10 ms on each file
14230   (select timeout) and read in non-blocking mode.  DEVRANDOM now
14231   defaults to the list "/dev/urandom", "/dev/random", "/dev/srandom"
14232   (previously it was just the string "/dev/urandom"), so on typical
14233   platforms the 10 ms delay will never occur.
14234   Also separate out the Unix variant to its own file, rand_unix.c.
14235   For VMS, there's a currently-empty rand_vms.c.
14236
14237   *Richard Levitte*
14238
14239 * Move OCSP client related routines to ocsp_cl.c. These
14240   provide utility functions which an application needing
14241   to issue a request to an OCSP responder and analyse the
14242   response will typically need: as opposed to those which an
14243   OCSP responder itself would need which will be added later.
14244
14245   OCSP_request_sign() signs an OCSP request with an API similar
14246   to PKCS7_sign(). OCSP_response_status() returns status of OCSP
14247   response. OCSP_response_get1_basic() extracts basic response
14248   from response. OCSP_resp_find_status(): finds and extracts status
14249   information from an OCSP_CERTID structure (which will be created
14250   when the request structure is built). These are built from lower
14251   level functions which work on OCSP_SINGLERESP structures but
14252   won't normally be used unless the application wishes to examine
14253   extensions in the OCSP response for example.
14254
14255   Replace nonce routines with a pair of functions.
14256   OCSP_request_add1_nonce() adds a nonce value and optionally
14257   generates a random value. OCSP_check_nonce() checks the
14258   validity of the nonce in an OCSP response.
14259
14260   *Steve Henson*
14261
14262 * Change function OCSP_request_add() to OCSP_request_add0_id().
14263   This doesn't copy the supplied OCSP_CERTID and avoids the
14264   need to free up the newly created id. Change return type
14265   to OCSP_ONEREQ to return the internal OCSP_ONEREQ structure.
14266   This can then be used to add extensions to the request.
14267   Deleted OCSP_request_new(), since most of its functionality
14268   is now in OCSP_REQUEST_new() (and the case insensitive name
14269   clash) apart from the ability to set the request name which
14270   will be added elsewhere.
14271
14272   *Steve Henson*
14273
14274 * Update OCSP API. Remove obsolete extensions argument from
14275   various functions. Extensions are now handled using the new
14276   OCSP extension code. New simple OCSP HTTP function which
14277   can be used to send requests and parse the response.
14278
14279   *Steve Henson*
14280
14281 * Fix the PKCS#7 (S/MIME) code to work with new ASN1. Two new
14282   ASN1_ITEM structures help with sign and verify. PKCS7_ATTR_SIGN
14283   uses the special reorder version of SET OF to sort the attributes
14284   and reorder them to match the encoded order. This resolves a long
14285   standing problem: a verify on a PKCS7 structure just after signing
14286   it used to fail because the attribute order did not match the
14287   encoded order. PKCS7_ATTR_VERIFY does not reorder the attributes:
14288   it uses the received order. This is necessary to tolerate some broken
14289   software that does not order SET OF. This is handled by encoding
14290   as a SEQUENCE OF but using implicit tagging (with UNIVERSAL class)
14291   to produce the required SET OF.
14292
14293   *Steve Henson*
14294
14295 * Have mk1mf.pl generate the macros OPENSSL_BUILD_SHLIBCRYPTO and
14296   OPENSSL_BUILD_SHLIBSSL and use them appropriately in the header
14297   files to get correct declarations of the ASN.1 item variables.
14298
14299   *Richard Levitte*
14300
14301 * Rewrite of PKCS#12 code to use new ASN1 functionality. Replace many
14302   PKCS#12 macros with real functions. Fix two unrelated ASN1 bugs:
14303   asn1_check_tlen() would sometimes attempt to use 'ctx' when it was
14304   NULL and ASN1_TYPE was not dereferenced properly in asn1_ex_c2i().
14305   New ASN1 macro: DECLARE_ASN1_ITEM() which just declares the relevant
14306   ASN1_ITEM and no wrapper functions.
14307
14308   *Steve Henson*
14309
14310 * New functions or ASN1_item_d2i_fp() and ASN1_item_d2i_bio(). These
14311   replace the old function pointer based I/O routines. Change most of
14312   the `*_d2i_bio()` and `*_d2i_fp()` functions to use these.
14313
14314   *Steve Henson*
14315
14316 * Enhance mkdef.pl to be more accepting about spacing in C preprocessor
14317   lines, recognize more "algorithms" that can be deselected, and make
14318   it complain about algorithm deselection that isn't recognised.
14319
14320   *Richard Levitte*
14321
14322 * New ASN1 functions to handle dup, sign, verify, digest, pack and
14323   unpack operations in terms of ASN1_ITEM. Modify existing wrappers
14324   to use new functions. Add NO_ASN1_OLD which can be set to remove
14325   some old style ASN1 functions: this can be used to determine if old
14326   code will still work when these eventually go away.
14327
14328   *Steve Henson*
14329
14330 * New extension functions for OCSP structures, these follow the
14331   same conventions as certificates and CRLs.
14332
14333   *Steve Henson*
14334
14335 * New function X509V3_add1_i2d(). This automatically encodes and
14336   adds an extension. Its behaviour can be customised with various
14337   flags to append, replace or delete. Various wrappers added for
14338   certificates and CRLs.
14339
14340   *Steve Henson*
14341
14342 * Fix to avoid calling the underlying ASN1 print routine when
14343   an extension cannot be parsed. Correct a typo in the
14344   OCSP_SERVICELOC extension. Tidy up print OCSP format.
14345
14346   *Steve Henson*
14347
14348 * Make mkdef.pl parse some of the ASN1 macros and add appropriate
14349   entries for variables.
14350
14351   *Steve Henson*
14352
14353 * Add functionality to `apps/openssl.c` for detecting locking
14354   problems: As the program is single-threaded, all we have
14355   to do is register a locking callback using an array for
14356   storing which locks are currently held by the program.
14357
14358   *Bodo Moeller*
14359
14360 * Use a lock around the call to CRYPTO_get_ex_new_index() in
14361   SSL_get_ex_data_X509_STORE_idx(), which is used in
14362   ssl_verify_cert_chain() and thus can be called at any time
14363   during TLS/SSL handshakes so that thread-safety is essential.
14364   Unfortunately, the ex_data design is not at all suited
14365   for multi-threaded use, so it probably should be abolished.
14366
14367   *Bodo Moeller*
14368
14369 * Added Broadcom "ubsec" ENGINE to OpenSSL.
14370
14371   *Broadcom, tweaked and integrated by Geoff Thorpe*
14372
14373 * Move common extension printing code to new function
14374   X509V3_print_extensions(). Reorganise OCSP print routines and
14375   implement some needed OCSP ASN1 functions. Add OCSP extensions.
14376
14377   *Steve Henson*
14378
14379 * New function X509_signature_print() to remove duplication in some
14380   print routines.
14381
14382   *Steve Henson*
14383
14384 * Add a special meaning when SET OF and SEQUENCE OF flags are both
14385   set (this was treated exactly the same as SET OF previously). This
14386   is used to reorder the STACK representing the structure to match the
14387   encoding. This will be used to get round a problem where a PKCS7
14388   structure which was signed could not be verified because the STACK
14389   order did not reflect the encoded order.
14390
14391   *Steve Henson*
14392
14393 * Reimplement the OCSP ASN1 module using the new code.
14394
14395   *Steve Henson*
14396
14397 * Update the X509V3 code to permit the use of an ASN1_ITEM structure
14398   for its ASN1 operations. The old style function pointers still exist
14399   for now but they will eventually go away.
14400
14401   *Steve Henson*
14402
14403 * Merge in replacement ASN1 code from the ASN1 branch. This almost
14404   completely replaces the old ASN1 functionality with a table driven
14405   encoder and decoder which interprets an ASN1_ITEM structure describing
14406   the ASN1 module. Compatibility with the existing ASN1 API (i2d,d2i) is
14407   largely maintained. Almost all of the old asn1_mac.h macro based ASN1
14408   has also been converted to the new form.
14409
14410   *Steve Henson*
14411
14412 * Change BN_mod_exp_recp so that negative moduli are tolerated
14413   (the sign is ignored).  Similarly, ignore the sign in BN_MONT_CTX_set
14414   so that BN_mod_exp_mont and BN_mod_exp_mont_word work
14415   for negative moduli.
14416
14417   *Bodo Moeller*
14418
14419 * Fix BN_uadd and BN_usub: Always return non-negative results instead
14420   of not touching the result's sign bit.
14421
14422   *Bodo Moeller*
14423
14424 * BN_div bugfix: If the result is 0, the sign (res->neg) must not be
14425   set.
14426
14427   *Bodo Moeller*
14428
14429 * Changed the LHASH code to use prototypes for callbacks, and created
14430   macros to declare and implement thin (optionally static) functions
14431   that provide type-safety and avoid function pointer casting for the
14432   type-specific callbacks.
14433
14434   *Geoff Thorpe*
14435
14436 * Added Kerberos Cipher Suites to be used with TLS, as written in
14437   RFC 2712.
14438   *Veers Staats <staatsvr@asc.hpc.mil>,
14439   Jeffrey Altman <jaltman@columbia.edu>, via Richard Levitte*
14440
14441 * Reformat the FAQ so the different questions and answers can be divided
14442   in sections depending on the subject.
14443
14444   *Richard Levitte*
14445
14446 * Have the zlib compression code load ZLIB.DLL dynamically under
14447   Windows.
14448
14449   *Richard Levitte*
14450
14451 * New function BN_mod_sqrt for computing square roots modulo a prime
14452   (using the probabilistic Tonelli-Shanks algorithm unless
14453   p == 3 (mod 4)  or  p == 5 (mod 8),  which are cases that can
14454   be handled deterministically).
14455
14456   *Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller*
14457
14458 * Make BN_mod_inverse faster by explicitly handling small quotients
14459   in the Euclid loop. (Speed gain about 20% for small moduli [256 or
14460   512 bits], about 30% for larger ones [1024 or 2048 bits].)
14461
14462   *Bodo Moeller*
14463
14464 * New function BN_kronecker.
14465
14466   *Bodo Moeller*
14467
14468 * Fix BN_gcd so that it works on negative inputs; the result is
14469   positive unless both parameters are zero.
14470   Previously something reasonably close to an infinite loop was
14471   possible because numbers could be growing instead of shrinking
14472   in the implementation of Euclid's algorithm.
14473
14474   *Bodo Moeller*
14475
14476 * Fix BN_is_word() and BN_is_one() macros to take into account the
14477   sign of the number in question.
14478
14479   Fix BN_is_word(a,w) to work correctly for w == 0.
14480
14481   The old BN_is_word(a,w) macro is now called BN_abs_is_word(a,w)
14482   because its test if the absolute value of 'a' equals 'w'.
14483   Note that BN_abs_is_word does *not* handle w == 0 reliably;
14484   it exists mostly for use in the implementations of BN_is_zero(),
14485   BN_is_one(), and BN_is_word().
14486
14487   *Bodo Moeller*
14488
14489 * New function BN_swap.
14490
14491   *Bodo Moeller*
14492
14493 * Use BN_nnmod instead of BN_mod in crypto/bn/bn_exp.c so that
14494   the exponentiation functions are more likely to produce reasonable
14495   results on negative inputs.
14496
14497   *Bodo Moeller*
14498
14499 * Change BN_mod_mul so that the result is always non-negative.
14500   Previously, it could be negative if one of the factors was negative;
14501   I don't think anyone really wanted that behaviour.
14502
14503   *Bodo Moeller*
14504
14505 * Move `BN_mod_...` functions into new file `crypto/bn/bn_mod.c`
14506   (except for exponentiation, which stays in `crypto/bn/bn_exp.c`,
14507   and `BN_mod_mul_reciprocal`, which stays in `crypto/bn/bn_recp.c`)
14508   and add new functions:
14509
14510           BN_nnmod
14511           BN_mod_sqr
14512           BN_mod_add
14513           BN_mod_add_quick
14514           BN_mod_sub
14515           BN_mod_sub_quick
14516           BN_mod_lshift1
14517           BN_mod_lshift1_quick
14518           BN_mod_lshift
14519           BN_mod_lshift_quick
14520
14521   These functions always generate non-negative results.
14522
14523   `BN_nnmod` otherwise is `like BN_mod` (if `BN_mod` computes a remainder `r`
14524   such that `|m| < r < 0`, `BN_nnmod` will output `rem + |m|` instead).
14525
14526   `BN_mod_XXX_quick(r, a, [b,] m)` generates the same result as
14527   `BN_mod_XXX(r, a, [b,] m, ctx)`, but requires that `a` [and  `b`]
14528   be reduced modulo `m`.
14529
14530   *Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller*
14531
14532<!--
14533   The following entry accidentally appeared in the CHANGES file
14534   distributed with OpenSSL 0.9.7.  The modifications described in
14535   it do *not* apply to OpenSSL 0.9.7.
14536
14537 * Remove a few calls to bn_wexpand() in BN_sqr() (the one in there
14538   was actually never needed) and in BN_mul().  The removal in BN_mul()
14539   required a small change in bn_mul_part_recursive() and the addition
14540   of the functions bn_cmp_part_words(), bn_sub_part_words() and
14541   bn_add_part_words(), which do the same thing as bn_cmp_words(),
14542   bn_sub_words() and bn_add_words() except they take arrays with
14543   differing sizes.
14544
14545   *Richard Levitte*
14546-->
14547
14548 * In 'openssl passwd', verify passwords read from the terminal
14549   unless the '-salt' option is used (which usually means that
14550   verification would just waste user's time since the resulting
14551   hash is going to be compared with some given password hash)
14552   or the new '-noverify' option is used.
14553
14554   This is an incompatible change, but it does not affect
14555   non-interactive use of 'openssl passwd' (passwords on the command
14556   line, '-stdin' option, '-in ...' option) and thus should not
14557   cause any problems.
14558
14559   *Bodo Moeller*
14560
14561 * Remove all references to RSAref, since there's no more need for it.
14562
14563   *Richard Levitte*
14564
14565 * Make DSO load along a path given through an environment variable
14566   (SHLIB_PATH) with shl_load().
14567
14568   *Richard Levitte*
14569
14570 * Constify the ENGINE code as a result of BIGNUM constification.
14571   Also constify the RSA code and most things related to it.  In a
14572   few places, most notable in the depth of the ASN.1 code, ugly
14573   casts back to non-const were required (to be solved at a later
14574   time)
14575
14576   *Richard Levitte*
14577
14578 * Make it so the openssl application has all engines loaded by default.
14579
14580   *Richard Levitte*
14581
14582 * Constify the BIGNUM routines a little more.
14583
14584   *Richard Levitte*
14585
14586 * Add the following functions:
14587
14588           ENGINE_load_cswift()
14589           ENGINE_load_chil()
14590           ENGINE_load_atalla()
14591           ENGINE_load_nuron()
14592           ENGINE_load_builtin_engines()
14593
14594   That way, an application can itself choose if external engines that
14595   are built-in in OpenSSL shall ever be used or not.  The benefit is
14596   that applications won't have to be linked with libdl or other dso
14597   libraries unless it's really needed.
14598
14599   Changed 'openssl engine' to load all engines on demand.
14600   Changed the engine header files to avoid the duplication of some
14601   declarations (they differed!).
14602
14603   *Richard Levitte*
14604
14605 * 'openssl engine' can now list capabilities.
14606
14607   *Richard Levitte*
14608
14609 * Better error reporting in 'openssl engine'.
14610
14611   *Richard Levitte*
14612
14613 * Never call load_dh_param(NULL) in s_server.
14614
14615   *Bodo Moeller*
14616
14617 * Add engine application.  It can currently list engines by name and
14618   identity, and test if they are actually available.
14619
14620   *Richard Levitte*
14621
14622 * Improve RPM specification file by forcing symbolic linking and making
14623   sure the installed documentation is also owned by root.root.
14624
14625   *Damien Miller <djm@mindrot.org>*
14626
14627 * Give the OpenSSL applications more possibilities to make use of
14628   keys (public as well as private) handled by engines.
14629
14630   *Richard Levitte*
14631
14632 * Add OCSP code that comes from CertCo.
14633
14634   *Richard Levitte*
14635
14636 * Add VMS support for the Rijndael code.
14637
14638   *Richard Levitte*
14639
14640 * Added untested support for Nuron crypto accelerator.
14641
14642   *Ben Laurie*
14643
14644 * Add support for external cryptographic devices.  This code was
14645   previously distributed separately as the "engine" branch.
14646
14647   *Geoff Thorpe, Richard Levitte*
14648
14649 * Rework the filename-translation in the DSO code. It is now possible to
14650   have far greater control over how a "name" is turned into a filename
14651   depending on the operating environment and any oddities about the
14652   different shared library filenames on each system.
14653
14654   *Geoff Thorpe*
14655
14656 * Support threads on FreeBSD-elf in Configure.
14657
14658   *Richard Levitte*
14659
14660 * Fix for SHA1 assembly problem with MASM: it produces
14661   warnings about corrupt line number information when assembling
14662   with debugging information. This is caused by the overlapping
14663   of two sections.
14664
14665   *Bernd Matthes <mainbug@celocom.de>, Steve Henson*
14666
14667 * NCONF changes.
14668   NCONF_get_number() has no error checking at all.  As a replacement,
14669   NCONF_get_number_e() is defined (`_e` for "error checking") and is
14670   promoted strongly.  The old NCONF_get_number is kept around for
14671   binary backward compatibility.
14672   Make it possible for methods to load from something other than a BIO,
14673   by providing a function pointer that is given a name instead of a BIO.
14674   For example, this could be used to load configuration data from an
14675   LDAP server.
14676
14677   *Richard Levitte*
14678
14679 * Fix for non blocking accept BIOs. Added new I/O special reason
14680   BIO_RR_ACCEPT to cover this case. Previously use of accept BIOs
14681   with non blocking I/O was not possible because no retry code was
14682   implemented. Also added new SSL code SSL_WANT_ACCEPT to cover
14683   this case.
14684
14685   *Steve Henson*
14686
14687 * Added the beginnings of Rijndael support.
14688
14689   *Ben Laurie*
14690
14691 * Fix for bug in DirectoryString mask setting. Add support for
14692   X509_NAME_print_ex() in 'req' and X509_print_ex() function
14693   to allow certificate printing to more controllable, additional
14694   'certopt' option to 'x509' to allow new printing options to be
14695   set.
14696
14697   *Steve Henson*
14698
14699 * Clean old EAY MD5 hack from e_os.h.
14700
14701   *Richard Levitte*
14702
14703### Changes between 0.9.6l and 0.9.6m  [17 Mar 2004]
14704
14705 * Fix null-pointer assignment in do_change_cipher_spec() revealed
14706   by using the Codenomicon TLS Test Tool ([CVE-2004-0079])
14707
14708   *Joe Orton, Steve Henson*
14709
14710### Changes between 0.9.6k and 0.9.6l  [04 Nov 2003]
14711
14712 * Fix additional bug revealed by the NISCC test suite:
14713
14714   Stop bug triggering large recursion when presented with
14715   certain ASN.1 tags ([CVE-2003-0851])
14716
14717   *Steve Henson*
14718
14719### Changes between 0.9.6j and 0.9.6k  [30 Sep 2003]
14720
14721 * Fix various bugs revealed by running the NISCC test suite:
14722
14723   Stop out of bounds reads in the ASN1 code when presented with
14724   invalid tags (CVE-2003-0543 and CVE-2003-0544).
14725
14726   If verify callback ignores invalid public key errors don't try to check
14727   certificate signature with the NULL public key.
14728
14729   *Steve Henson*
14730
14731 * In ssl3_accept() (ssl/s3_srvr.c) only accept a client certificate
14732   if the server requested one: as stated in TLS 1.0 and SSL 3.0
14733   specifications.
14734
14735   *Steve Henson*
14736
14737 * In ssl3_get_client_hello() (ssl/s3_srvr.c), tolerate additional
14738   extra data after the compression methods not only for TLS 1.0
14739   but also for SSL 3.0 (as required by the specification).
14740
14741   *Bodo Moeller; problem pointed out by Matthias Loepfe*
14742
14743 * Change X509_certificate_type() to mark the key as exported/exportable
14744   when it's 512 *bits* long, not 512 bytes.
14745
14746   *Richard Levitte*
14747
14748### Changes between 0.9.6i and 0.9.6j  [10 Apr 2003]
14749
14750 * Countermeasure against the Klima-Pokorny-Rosa extension of
14751   Bleichbacher's attack on PKCS #1 v1.5 padding: treat
14752   a protocol version number mismatch like a decryption error
14753   in ssl3_get_client_key_exchange (ssl/s3_srvr.c).
14754
14755   *Bodo Moeller*
14756
14757 * Turn on RSA blinding by default in the default implementation
14758   to avoid a timing attack. Applications that don't want it can call
14759   RSA_blinding_off() or use the new flag RSA_FLAG_NO_BLINDING.
14760   They would be ill-advised to do so in most cases.
14761
14762   *Ben Laurie, Steve Henson, Geoff Thorpe, Bodo Moeller*
14763
14764 * Change RSA blinding code so that it works when the PRNG is not
14765   seeded (in this case, the secret RSA exponent is abused as
14766   an unpredictable seed -- if it is not unpredictable, there
14767   is no point in blinding anyway).  Make RSA blinding thread-safe
14768   by remembering the creator's thread ID in rsa->blinding and
14769   having all other threads use local one-time blinding factors
14770   (this requires more computation than sharing rsa->blinding, but
14771   avoids excessive locking; and if an RSA object is not shared
14772   between threads, blinding will still be very fast).
14773
14774   *Bodo Moeller*
14775
14776### Changes between 0.9.6h and 0.9.6i  [19 Feb 2003]
14777
14778 * In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked
14779   via timing by performing a MAC computation even if incorrect
14780   block cipher padding has been found.  This is a countermeasure
14781   against active attacks where the attacker has to distinguish
14782   between bad padding and a MAC verification error. ([CVE-2003-0078])
14783
14784   *Bodo Moeller; problem pointed out by Brice Canvel (EPFL),
14785   Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and
14786   Martin Vuagnoux (EPFL, Ilion)*
14787
14788### Changes between 0.9.6g and 0.9.6h  [5 Dec 2002]
14789
14790 * New function OPENSSL_cleanse(), which is used to cleanse a section of
14791   memory from its contents.  This is done with a counter that will
14792   place alternating values in each byte.  This can be used to solve
14793   two issues: 1) the removal of calls to memset() by highly optimizing
14794   compilers, and 2) cleansing with other values than 0, since those can
14795   be read through on certain media, for example a swap space on disk.
14796
14797   *Geoff Thorpe*
14798
14799 * Bugfix: client side session caching did not work with external caching,
14800   because the session->cipher setting was not restored when reloading
14801   from the external cache. This problem was masked, when
14802   SSL_OP_NETSCAPE_REUSE_CIPHER_CHANGE_BUG (part of SSL_OP_ALL) was set.
14803   (Found by Steve Haslam <steve@araqnid.ddts.net>.)
14804
14805   *Lutz Jaenicke*
14806
14807 * Fix client_certificate (ssl/s2_clnt.c): The permissible total
14808   length of the REQUEST-CERTIFICATE message is 18 .. 34, not 17 .. 33.
14809
14810   *Zeev Lieber <zeev-l@yahoo.com>*
14811
14812 * Undo an undocumented change introduced in 0.9.6e which caused
14813   repeated calls to OpenSSL_add_all_ciphers() and
14814   OpenSSL_add_all_digests() to be ignored, even after calling
14815   EVP_cleanup().
14816
14817   *Richard Levitte*
14818
14819 * Change the default configuration reader to deal with last line not
14820   being properly terminated.
14821
14822   *Richard Levitte*
14823
14824 * Change X509_NAME_cmp() so it applies the special rules on handling
14825   DN values that are of type PrintableString, as well as RDNs of type
14826   emailAddress where the value has the type ia5String.
14827
14828   *stefank@valicert.com via Richard Levitte*
14829
14830 * Add a SSL_SESS_CACHE_NO_INTERNAL_STORE flag to take over half
14831   the job SSL_SESS_CACHE_NO_INTERNAL_LOOKUP was inconsistently
14832   doing, define a new flag (SSL_SESS_CACHE_NO_INTERNAL) to be
14833   the bitwise-OR of the two for use by the majority of applications
14834   wanting this behaviour, and update the docs. The documented
14835   behaviour and actual behaviour were inconsistent and had been
14836   changing anyway, so this is more a bug-fix than a behavioural
14837   change.
14838
14839   *Geoff Thorpe, diagnosed by Nadav Har'El*
14840
14841 * Don't impose a 16-byte length minimum on session IDs in ssl/s3_clnt.c
14842   (the SSL 3.0 and TLS 1.0 specifications allow any length up to 32 bytes).
14843
14844   *Bodo Moeller*
14845
14846 * Fix initialization code race conditions in
14847           SSLv23_method(),  SSLv23_client_method(),   SSLv23_server_method(),
14848           SSLv2_method(),   SSLv2_client_method(),    SSLv2_server_method(),
14849           SSLv3_method(),   SSLv3_client_method(),    SSLv3_server_method(),
14850           TLSv1_method(),   TLSv1_client_method(),    TLSv1_server_method(),
14851           ssl2_get_cipher_by_char(),
14852           ssl3_get_cipher_by_char().
14853
14854   *Patrick McCormick <patrick@tellme.com>, Bodo Moeller*
14855
14856 * Reorder cleanup sequence in SSL_CTX_free(): only remove the ex_data after
14857   the cached sessions are flushed, as the remove_cb() might use ex_data
14858   contents. Bug found by Sam Varshavchik <mrsam@courier-mta.com>
14859   (see [openssl.org #212]).
14860
14861   *Geoff Thorpe, Lutz Jaenicke*
14862
14863 * Fix typo in OBJ_txt2obj which incorrectly passed the content
14864   length, instead of the encoding length to d2i_ASN1_OBJECT.
14865
14866   *Steve Henson*
14867
14868### Changes between 0.9.6f and 0.9.6g  [9 Aug 2002]
14869
14870 * [In 0.9.6g-engine release:]
14871   Fix crypto/engine/vendor_defns/cswift.h for WIN32 (use `_stdcall`).
14872
14873   *Lynn Gazis <lgazis@rainbow.com>*
14874
14875### Changes between 0.9.6e and 0.9.6f  [8 Aug 2002]
14876
14877 * Fix ASN1 checks. Check for overflow by comparing with LONG_MAX
14878   and get fix the header length calculation.
14879   *Florian Weimer <Weimer@CERT.Uni-Stuttgart.DE>,
14880   Alon Kantor <alonk@checkpoint.com> (and others), Steve Henson*
14881
14882 * Use proper error handling instead of 'assertions' in buffer
14883   overflow checks added in 0.9.6e.  This prevents DoS (the
14884   assertions could call abort()).
14885
14886   *Arne Ansper <arne@ats.cyber.ee>, Bodo Moeller*
14887
14888### Changes between 0.9.6d and 0.9.6e  [30 Jul 2002]
14889
14890 * Add various sanity checks to asn1_get_length() to reject
14891   the ASN1 length bytes if they exceed sizeof(long), will appear
14892   negative or the content length exceeds the length of the
14893   supplied buffer.
14894
14895   *Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>*
14896
14897 * Fix cipher selection routines: ciphers without encryption had no flags
14898   for the cipher strength set and where therefore not handled correctly
14899   by the selection routines (PR #130).
14900
14901   *Lutz Jaenicke*
14902
14903 * Fix EVP_dsa_sha macro.
14904
14905   *Nils Larsch*
14906
14907 * New option
14908        SSL_OP_DONT_INSERT_EMPTY_FRAGMENTS
14909   for disabling the SSL 3.0/TLS 1.0 CBC vulnerability countermeasure
14910   that was added in OpenSSL 0.9.6d.
14911
14912   As the countermeasure turned out to be incompatible with some
14913   broken SSL implementations, the new option is part of SSL_OP_ALL.
14914   SSL_OP_ALL is usually employed when compatibility with weird SSL
14915   implementations is desired (e.g. '-bugs' option to 's_client' and
14916   's_server'), so the new option is automatically set in many
14917   applications.
14918
14919   *Bodo Moeller*
14920
14921 * Changes in security patch:
14922
14923   Changes marked "(CHATS)" were sponsored by the Defense Advanced
14924   Research Projects Agency (DARPA) and Air Force Research Laboratory,
14925   Air Force Materiel Command, USAF, under agreement number
14926   F30602-01-2-0537.
14927
14928 * Add various sanity checks to asn1_get_length() to reject
14929   the ASN1 length bytes if they exceed sizeof(long), will appear
14930   negative or the content length exceeds the length of the
14931   supplied buffer. ([CVE-2002-0659])
14932
14933   *Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>*
14934
14935 * Assertions for various potential buffer overflows, not known to
14936   happen in practice.
14937
14938   *Ben Laurie (CHATS)*
14939
14940 * Various temporary buffers to hold ASCII versions of integers were
14941   too small for 64 bit platforms. ([CVE-2002-0655])
14942   *Matthew Byng-Maddick <mbm@aldigital.co.uk> and Ben Laurie (CHATS)>*
14943
14944 * Remote buffer overflow in SSL3 protocol - an attacker could
14945   supply an oversized session ID to a client. ([CVE-2002-0656])
14946
14947   *Ben Laurie (CHATS)*
14948
14949 * Remote buffer overflow in SSL2 protocol - an attacker could
14950   supply an oversized client master key. ([CVE-2002-0656])
14951
14952   *Ben Laurie (CHATS)*
14953
14954### Changes between 0.9.6c and 0.9.6d  [9 May 2002]
14955
14956 * Fix crypto/asn1/a_sign.c so that 'parameters' is omitted (not
14957   encoded as NULL) with id-dsa-with-sha1.
14958
14959   *Nils Larsch <nla@trustcenter.de>; problem pointed out by Bodo Moeller*
14960
14961 * Check various `X509_...()` return values in `apps/req.c`.
14962
14963   *Nils Larsch <nla@trustcenter.de>*
14964
14965 * Fix BASE64 decode (EVP_DecodeUpdate) for data with CR/LF ended lines:
14966   an end-of-file condition would erroneously be flagged, when the CRLF
14967   was just at the end of a processed block. The bug was discovered when
14968   processing data through a buffering memory BIO handing the data to a
14969   BASE64-decoding BIO. Bug fund and patch submitted by Pavel Tsekov
14970   <ptsekov@syntrex.com> and Nedelcho Stanev.
14971
14972   *Lutz Jaenicke*
14973
14974 * Implement a countermeasure against a vulnerability recently found
14975   in CBC ciphersuites in SSL 3.0/TLS 1.0: Send an empty fragment
14976   before application data chunks to avoid the use of known IVs
14977   with data potentially chosen by the attacker.
14978
14979   *Bodo Moeller*
14980
14981 * Fix length checks in ssl3_get_client_hello().
14982
14983   *Bodo Moeller*
14984
14985 * TLS/SSL library bugfix: use s->s3->in_read_app_data differently
14986   to prevent ssl3_read_internal() from incorrectly assuming that
14987   ssl3_read_bytes() found application data while handshake
14988   processing was enabled when in fact s->s3->in_read_app_data was
14989   merely automatically cleared during the initial handshake.
14990
14991   *Bodo Moeller; problem pointed out by Arne Ansper <arne@ats.cyber.ee>*
14992
14993 * Fix object definitions for Private and Enterprise: they were not
14994   recognized in their shortname (=lowercase) representation. Extend
14995   obj_dat.pl to issue an error when using undefined keywords instead
14996   of silently ignoring the problem (Svenning Sorensen
14997   <sss@sss.dnsalias.net>).
14998
14999   *Lutz Jaenicke*
15000
15001 * Fix DH_generate_parameters() so that it works for 'non-standard'
15002   generators, i.e. generators other than 2 and 5.  (Previously, the
15003   code did not properly initialise the 'add' and 'rem' values to
15004   BN_generate_prime().)
15005
15006   In the new general case, we do not insist that 'generator' is
15007   actually a primitive root: This requirement is rather pointless;
15008   a generator of the order-q subgroup is just as good, if not
15009   better.
15010
15011   *Bodo Moeller*
15012
15013 * Map new X509 verification errors to alerts. Discovered and submitted by
15014   Tom Wu <tom@arcot.com>.
15015
15016   *Lutz Jaenicke*
15017
15018 * Fix ssl3_pending() (ssl/s3_lib.c) to prevent SSL_pending() from
15019   returning non-zero before the data has been completely received
15020   when using non-blocking I/O.
15021
15022   *Bodo Moeller; problem pointed out by John Hughes*
15023
15024 * Some of the ciphers missed the strength entry (SSL_LOW etc).
15025
15026   *Ben Laurie, Lutz Jaenicke*
15027
15028 * Fix bug in SSL_clear(): bad sessions were not removed (found by
15029   Yoram Zahavi <YoramZ@gilian.com>).
15030
15031   *Lutz Jaenicke*
15032
15033 * Add information about CygWin 1.3 and on, and preserve proper
15034   configuration for the versions before that.
15035
15036   *Corinna Vinschen <vinschen@redhat.com> and Richard Levitte*
15037
15038 * Make removal from session cache (SSL_CTX_remove_session()) more robust:
15039   check whether we deal with a copy of a session and do not delete from
15040   the cache in this case. Problem reported by "Izhar Shoshani Levi"
15041   <izhar@checkpoint.com>.
15042
15043   *Lutz Jaenicke*
15044
15045 * Do not store session data into the internal session cache, if it
15046   is never intended to be looked up (SSL_SESS_CACHE_NO_INTERNAL_LOOKUP
15047   flag is set). Proposed by Aslam <aslam@funk.com>.
15048
15049   *Lutz Jaenicke*
15050
15051 * Have ASN1_BIT_STRING_set_bit() really clear a bit when the requested
15052   value is 0.
15053
15054   *Richard Levitte*
15055
15056 * [In 0.9.6d-engine release:]
15057   Fix a crashbug and a logic bug in hwcrhk_load_pubkey().
15058
15059   *Toomas Kiisk <vix@cyber.ee> via Richard Levitte*
15060
15061 * Add the configuration target linux-s390x.
15062
15063   *Neale Ferguson <Neale.Ferguson@SoftwareAG-USA.com> via Richard Levitte*
15064
15065 * The earlier bugfix for the SSL3_ST_SW_HELLO_REQ_C case of
15066   ssl3_accept (ssl/s3_srvr.c) incorrectly used a local flag
15067   variable as an indication that a ClientHello message has been
15068   received.  As the flag value will be lost between multiple
15069   invocations of ssl3_accept when using non-blocking I/O, the
15070   function may not be aware that a handshake has actually taken
15071   place, thus preventing a new session from being added to the
15072   session cache.
15073
15074   To avoid this problem, we now set s->new_session to 2 instead of
15075   using a local variable.
15076
15077   *Lutz Jaenicke, Bodo Moeller*
15078
15079 * Bugfix: Return -1 from ssl3_get_server_done (ssl3/s3_clnt.c)
15080   if the SSL_R_LENGTH_MISMATCH error is detected.
15081
15082   *Geoff Thorpe, Bodo Moeller*
15083
15084 * New 'shared_ldflag' column in Configure platform table.
15085
15086   *Richard Levitte*
15087
15088 * Fix EVP_CIPHER_mode macro.
15089
15090   *"Dan S. Camper" <dan@bti.net>*
15091
15092 * Fix ssl3_read_bytes (ssl/s3_pkt.c): To ignore messages of unknown
15093   type, we must throw them away by setting rr->length to 0.
15094
15095   *D P Chang <dpc@qualys.com>*
15096
15097### Changes between 0.9.6b and 0.9.6c  [21 dec 2001]
15098
15099 * Fix BN_rand_range bug pointed out by Dominikus Scherkl
15100   <Dominikus.Scherkl@biodata.com>.  (The previous implementation
15101   worked incorrectly for those cases where range = `10..._2`  and
15102   `3*range`  is two bits longer than  range.)
15103
15104   *Bodo Moeller*
15105
15106 * Only add signing time to PKCS7 structures if it is not already
15107   present.
15108
15109   *Steve Henson*
15110
15111 * Fix crypto/objects/objects.h: "ld-ce" should be "id-ce",
15112   OBJ_ld_ce should be OBJ_id_ce.
15113   Also some ip-pda OIDs in crypto/objects/objects.txt were
15114   incorrect (cf. RFC 3039).
15115
15116   *Matt Cooper, Frederic Giudicelli, Bodo Moeller*
15117
15118 * Release CRYPTO_LOCK_DYNLOCK when CRYPTO_destroy_dynlockid()
15119   returns early because it has nothing to do.
15120
15121   *Andy Schneider <andy.schneider@bjss.co.uk>*
15122
15123 * [In 0.9.6c-engine release:]
15124   Fix mutex callback return values in crypto/engine/hw_ncipher.c.
15125
15126   *Andy Schneider <andy.schneider@bjss.co.uk>*
15127
15128 * [In 0.9.6c-engine release:]
15129   Add support for Cryptographic Appliance's keyserver technology.
15130   (Use engine 'keyclient')
15131
15132   *Cryptographic Appliances and Geoff Thorpe*
15133
15134 * Add a configuration entry for OS/390 Unix.  The C compiler 'c89'
15135   is called via tools/c89.sh because arguments have to be
15136   rearranged (all '-L' options must appear before the first object
15137   modules).
15138
15139   *Richard Shapiro <rshapiro@abinitio.com>*
15140
15141 * [In 0.9.6c-engine release:]
15142   Add support for Broadcom crypto accelerator cards, backported
15143   from 0.9.7.
15144
15145   *Broadcom, Nalin Dahyabhai <nalin@redhat.com>, Mark Cox*
15146
15147 * [In 0.9.6c-engine release:]
15148   Add support for SureWare crypto accelerator cards from
15149   Baltimore Technologies.  (Use engine 'sureware')
15150
15151   *Baltimore Technologies and Mark Cox*
15152
15153 * [In 0.9.6c-engine release:]
15154   Add support for crypto accelerator cards from Accelerated
15155   Encryption Processing, www.aep.ie.  (Use engine 'aep')
15156
15157   *AEP Inc. and Mark Cox*
15158
15159 * Add a configuration entry for gcc on UnixWare.
15160
15161   *Gary Benson <gbenson@redhat.com>*
15162
15163 * Change ssl/s2_clnt.c and ssl/s2_srvr.c so that received handshake
15164   messages are stored in a single piece (fixed-length part and
15165   variable-length part combined) and fix various bugs found on the way.
15166
15167   *Bodo Moeller*
15168
15169 * Disable caching in BIO_gethostbyname(), directly use gethostbyname()
15170   instead.  BIO_gethostbyname() does not know what timeouts are
15171   appropriate, so entries would stay in cache even when they have
15172   become invalid.
15173   *Bodo Moeller; problem pointed out by Rich Salz <rsalz@zolera.com>*
15174
15175 * Change ssl23_get_client_hello (ssl/s23_srvr.c) behaviour when
15176   faced with a pathologically small ClientHello fragment that does
15177   not contain client_version: Instead of aborting with an error,
15178   simply choose the highest available protocol version (i.e.,
15179   TLS 1.0 unless it is disabled).  In practice, ClientHello
15180   messages are never sent like this, but this change gives us
15181   strictly correct behaviour at least for TLS.
15182
15183   *Bodo Moeller*
15184
15185 * Fix SSL handshake functions and SSL_clear() such that SSL_clear()
15186   never resets s->method to s->ctx->method when called from within
15187   one of the SSL handshake functions.
15188
15189   *Bodo Moeller; problem pointed out by Niko Baric*
15190
15191 * In ssl3_get_client_hello (ssl/s3_srvr.c), generate a fatal alert
15192   (sent using the client's version number) if client_version is
15193   smaller than the protocol version in use.  Also change
15194   ssl23_get_client_hello (ssl/s23_srvr.c) to select TLS 1.0 if
15195   the client demanded SSL 3.0 but only TLS 1.0 is enabled; then
15196   the client will at least see that alert.
15197
15198   *Bodo Moeller*
15199
15200 * Fix ssl3_get_message (ssl/s3_both.c) to handle message fragmentation
15201   correctly.
15202
15203   *Bodo Moeller*
15204
15205 * Avoid infinite loop in ssl3_get_message (ssl/s3_both.c) if a
15206   client receives HelloRequest while in a handshake.
15207
15208   *Bodo Moeller; bug noticed by Andy Schneider <andy.schneider@bjss.co.uk>*
15209
15210 * Bugfix in ssl3_accept (ssl/s3_srvr.c): Case SSL3_ST_SW_HELLO_REQ_C
15211   should end in 'break', not 'goto end' which circumvents various
15212   cleanups done in state SSL_ST_OK.   But session related stuff
15213   must be disabled for SSL_ST_OK in the case that we just sent a
15214   HelloRequest.
15215
15216   Also avoid some overhead by not calling ssl_init_wbio_buffer()
15217   before just sending a HelloRequest.
15218
15219   *Bodo Moeller, Eric Rescorla <ekr@rtfm.com>*
15220
15221 * Fix ssl/s3_enc.c, ssl/t1_enc.c and ssl/s3_pkt.c so that we don't
15222   reveal whether illegal block cipher padding was found or a MAC
15223   verification error occurred.  (Neither SSLerr() codes nor alerts
15224   are directly visible to potential attackers, but the information
15225   may leak via logfiles.)
15226
15227   Similar changes are not required for the SSL 2.0 implementation
15228   because the number of padding bytes is sent in clear for SSL 2.0,
15229   and the extra bytes are just ignored.  However ssl/s2_pkt.c
15230   failed to verify that the purported number of padding bytes is in
15231   the legal range.
15232
15233   *Bodo Moeller*
15234
15235 * Add OpenUNIX-8 support including shared libraries
15236   (Boyd Lynn Gerber <gerberb@zenez.com>).
15237
15238   *Lutz Jaenicke*
15239
15240 * Improve RSA_padding_check_PKCS1_OAEP() check again to avoid
15241   'wristwatch attack' using huge encoding parameters (cf.
15242   James H. Manger's CRYPTO 2001 paper).  Note that the
15243   RSA_PKCS1_OAEP_PADDING case of RSA_private_decrypt() does not use
15244   encoding parameters and hence was not vulnerable.
15245
15246   *Bodo Moeller*
15247
15248 * BN_sqr() bug fix.
15249
15250   *Ulf Möller, reported by Jim Ellis <jim.ellis@cavium.com>*
15251
15252 * Rabin-Miller test analyses assume uniformly distributed witnesses,
15253   so use BN_pseudo_rand_range() instead of using BN_pseudo_rand()
15254   followed by modular reduction.
15255
15256   *Bodo Moeller; pointed out by Adam Young <AYoung1@NCSUS.JNJ.COM>*
15257
15258 * Add BN_pseudo_rand_range() with obvious functionality: BN_rand_range()
15259   equivalent based on BN_pseudo_rand() instead of BN_rand().
15260
15261   *Bodo Moeller*
15262
15263 * s3_srvr.c: allow sending of large client certificate lists (> 16 kB).
15264   This function was broken, as the check for a new client hello message
15265   to handle SGC did not allow these large messages.
15266   (Tracked down by "Douglas E. Engert" <deengert@anl.gov>.)
15267
15268   *Lutz Jaenicke*
15269
15270 * Add alert descriptions for TLSv1 to `SSL_alert_desc_string[_long]()`.
15271
15272   *Lutz Jaenicke*
15273
15274 * Fix buggy behaviour of BIO_get_num_renegotiates() and BIO_ctrl()
15275   for BIO_C_GET_WRITE_BUF_SIZE ("Stephen Hinton" <shinton@netopia.com>).
15276
15277   *Lutz Jaenicke*
15278
15279 * Rework the configuration and shared library support for Tru64 Unix.
15280   The configuration part makes use of modern compiler features and
15281   still retains old compiler behavior for those that run older versions
15282   of the OS.  The shared library support part includes a variant that
15283   uses the RPATH feature, and is available through the special
15284   configuration target "alpha-cc-rpath", which will never be selected
15285   automatically.
15286
15287   *Tim Mooney <mooney@dogbert.cc.ndsu.NoDak.edu> via Richard Levitte*
15288
15289 * In ssl3_get_key_exchange (ssl/s3_clnt.c), call ssl3_get_message()
15290   with the same message size as in ssl3_get_certificate_request().
15291   Otherwise, if no ServerKeyExchange message occurs, CertificateRequest
15292   messages might inadvertently be reject as too long.
15293
15294   *Petr Lampa <lampa@fee.vutbr.cz>*
15295
15296 * Enhanced support for IA-64 Unix platforms (well, Linux and HP-UX).
15297
15298   *Andy Polyakov*
15299
15300 * Modified SSL library such that the verify_callback that has been set
15301   specifically for an SSL object with SSL_set_verify() is actually being
15302   used. Before the change, a verify_callback set with this function was
15303   ignored and the verify_callback() set in the SSL_CTX at the time of
15304   the call was used. New function X509_STORE_CTX_set_verify_cb() introduced
15305   to allow the necessary settings.
15306
15307   *Lutz Jaenicke*
15308
15309 * Initialize static variable in crypto/dsa/dsa_lib.c and crypto/dh/dh_lib.c
15310   explicitly to NULL, as at least on Solaris 8 this seems not always to be
15311   done automatically (in contradiction to the requirements of the C
15312   standard). This made problems when used from OpenSSH.
15313
15314   *Lutz Jaenicke*
15315
15316 * In OpenSSL 0.9.6a and 0.9.6b, crypto/dh/dh_key.c ignored
15317   dh->length and always used
15318
15319           BN_rand_range(priv_key, dh->p).
15320
15321   BN_rand_range() is not necessary for Diffie-Hellman, and this
15322   specific range makes Diffie-Hellman unnecessarily inefficient if
15323   dh->length (recommended exponent length) is much smaller than the
15324   length of dh->p.  We could use BN_rand_range() if the order of
15325   the subgroup was stored in the DH structure, but we only have
15326   dh->length.
15327
15328   So switch back to
15329
15330           BN_rand(priv_key, l, ...)
15331
15332   where 'l' is dh->length if this is defined, or BN_num_bits(dh->p)-1
15333   otherwise.
15334
15335   *Bodo Moeller*
15336
15337 * In
15338
15339           RSA_eay_public_encrypt
15340           RSA_eay_private_decrypt
15341           RSA_eay_private_encrypt (signing)
15342           RSA_eay_public_decrypt (signature verification)
15343
15344   (default implementations for RSA_public_encrypt,
15345   RSA_private_decrypt, RSA_private_encrypt, RSA_public_decrypt),
15346   always reject numbers >= n.
15347
15348   *Bodo Moeller*
15349
15350 * In crypto/rand/md_rand.c, use a new short-time lock CRYPTO_LOCK_RAND2
15351   to synchronize access to 'locking_thread'.  This is necessary on
15352   systems where access to 'locking_thread' (an 'unsigned long'
15353   variable) is not atomic.
15354
15355   *Bodo Moeller*
15356
15357 * In crypto/rand/md_rand.c, set 'locking_thread' to current thread's ID
15358   *before* setting the 'crypto_lock_rand' flag.  The previous code had
15359   a race condition if 0 is a valid thread ID.
15360
15361   *Travis Vitek <vitek@roguewave.com>*
15362
15363 * Add support for shared libraries under Irix.
15364
15365   *Albert Chin-A-Young <china@thewrittenword.com>*
15366
15367 * Add configuration option to build on Linux on both big-endian and
15368   little-endian MIPS.
15369
15370   *Ralf Baechle <ralf@uni-koblenz.de>*
15371
15372 * Add the possibility to create shared libraries on HP-UX.
15373
15374   *Richard Levitte*
15375
15376### Changes between 0.9.6a and 0.9.6b  [9 Jul 2001]
15377
15378 * Change ssleay_rand_bytes (crypto/rand/md_rand.c)
15379   to avoid a SSLeay/OpenSSL PRNG weakness pointed out by
15380   Markku-Juhani O. Saarinen <markku-juhani.saarinen@nokia.com>:
15381   PRNG state recovery was possible based on the output of
15382   one PRNG request appropriately sized to gain knowledge on
15383   'md' followed by enough consecutive 1-byte PRNG requests
15384   to traverse all of 'state'.
15385
15386   1. When updating 'md_local' (the current thread's copy of 'md')
15387      during PRNG output generation, hash all of the previous
15388      'md_local' value, not just the half used for PRNG output.
15389
15390   2. Make the number of bytes from 'state' included into the hash
15391      independent from the number of PRNG bytes requested.
15392
15393   The first measure alone would be sufficient to avoid
15394   Markku-Juhani's attack.  (Actually it had never occurred
15395   to me that the half of 'md_local' used for chaining was the
15396   half from which PRNG output bytes were taken -- I had always
15397   assumed that the secret half would be used.)  The second
15398   measure makes sure that additional data from 'state' is never
15399   mixed into 'md_local' in small portions; this heuristically
15400   further strengthens the PRNG.
15401
15402   *Bodo Moeller*
15403
15404 * Fix crypto/bn/asm/mips3.s.
15405
15406   *Andy Polyakov*
15407
15408 * When only the key is given to "enc", the IV is undefined. Print out
15409   an error message in this case.
15410
15411   *Lutz Jaenicke*
15412
15413 * Handle special case when X509_NAME is empty in X509 printing routines.
15414
15415   *Steve Henson*
15416
15417 * In dsa_do_verify (crypto/dsa/dsa_ossl.c), verify that r and s are
15418   positive and less than q.
15419
15420   *Bodo Moeller*
15421
15422 * Don't change `*pointer` in CRYPTO_add_lock() is add_lock_callback is
15423   used: it isn't thread safe and the add_lock_callback should handle
15424   that itself.
15425
15426   *Paul Rose <Paul.Rose@bridge.com>*
15427
15428 * Verify that incoming data obeys the block size in
15429   ssl3_enc (ssl/s3_enc.c) and tls1_enc (ssl/t1_enc.c).
15430
15431   *Bodo Moeller*
15432
15433 * Fix OAEP check.
15434
15435   *Ulf Möller, Bodo Möller*
15436
15437 * The countermeasure against Bleichbacher's attack on PKCS #1 v1.5
15438   RSA encryption was accidentally removed in s3_srvr.c in OpenSSL 0.9.5
15439   when fixing the server behaviour for backwards-compatible 'client
15440   hello' messages.  (Note that the attack is impractical against
15441   SSL 3.0 and TLS 1.0 anyway because length and version checking
15442   means that the probability of guessing a valid ciphertext is
15443   around 2^-40; see section 5 in Bleichenbacher's CRYPTO '98
15444   paper.)
15445
15446   Before 0.9.5, the countermeasure (hide the error by generating a
15447   random 'decryption result') did not work properly because
15448   ERR_clear_error() was missing, meaning that SSL_get_error() would
15449   detect the supposedly ignored error.
15450
15451   Both problems are now fixed.
15452
15453   *Bodo Moeller*
15454
15455 * In crypto/bio/bf_buff.c, increase DEFAULT_BUFFER_SIZE to 4096
15456   (previously it was 1024).
15457
15458   *Bodo Moeller*
15459
15460 * Fix for compatibility mode trust settings: ignore trust settings
15461   unless some valid trust or reject settings are present.
15462
15463   *Steve Henson*
15464
15465 * Fix for blowfish EVP: its a variable length cipher.
15466
15467   *Steve Henson*
15468
15469 * Fix various bugs related to DSA S/MIME verification. Handle missing
15470   parameters in DSA public key structures and return an error in the
15471   DSA routines if parameters are absent.
15472
15473   *Steve Henson*
15474
15475 * In versions up to 0.9.6, RAND_file_name() resorted to file ".rnd"
15476   in the current directory if neither $RANDFILE nor $HOME was set.
15477   RAND_file_name() in 0.9.6a returned NULL in this case.  This has
15478   caused some confusion to Windows users who haven't defined $HOME.
15479   Thus RAND_file_name() is changed again: e_os.h can define a
15480   DEFAULT_HOME, which will be used if $HOME is not set.
15481   For Windows, we use "C:"; on other platforms, we still require
15482   environment variables.
15483
15484 * Move 'if (!initialized) RAND_poll()' into regions protected by
15485   CRYPTO_LOCK_RAND.  This is not strictly necessary, but avoids
15486   having multiple threads call RAND_poll() concurrently.
15487
15488   *Bodo Moeller*
15489
15490 * In crypto/rand/md_rand.c, replace 'add_do_not_lock' flag by a
15491   combination of a flag and a thread ID variable.
15492   Otherwise while one thread is in ssleay_rand_bytes (which sets the
15493   flag), *other* threads can enter ssleay_add_bytes without obeying
15494   the CRYPTO_LOCK_RAND lock (and may even illegally release the lock
15495   that they do not hold after the first thread unsets add_do_not_lock).
15496
15497   *Bodo Moeller*
15498
15499 * Change bctest again: '-x' expressions are not available in all
15500   versions of 'test'.
15501
15502   *Bodo Moeller*
15503
15504### Changes between 0.9.6 and 0.9.6a  [5 Apr 2001]
15505
15506 * Fix a couple of memory leaks in PKCS7_dataDecode()
15507
15508   *Steve Henson, reported by Heyun Zheng <hzheng@atdsprint.com>*
15509
15510 * Change Configure and Makefiles to provide EXE_EXT, which will contain
15511   the default extension for executables, if any.  Also, make the perl
15512   scripts that use symlink() to test if it really exists and use "cp"
15513   if it doesn't.  All this made OpenSSL compilable and installable in
15514   CygWin.
15515
15516   *Richard Levitte*
15517
15518 * Fix for asn1_GetSequence() for indefinite length constructed data.
15519   If SEQUENCE is length is indefinite just set c->slen to the total
15520   amount of data available.
15521
15522   *Steve Henson, reported by shige@FreeBSD.org*
15523
15524   *This change does not apply to 0.9.7.*
15525
15526 * Change bctest to avoid here-documents inside command substitution
15527   (workaround for FreeBSD /bin/sh bug).
15528   For compatibility with Ultrix, avoid shell functions (introduced
15529   in the bctest version that searches along $PATH).
15530
15531   *Bodo Moeller*
15532
15533 * Rename 'des_encrypt' to 'des_encrypt1'.  This avoids the clashes
15534   with des_encrypt() defined on some operating systems, like Solaris
15535   and UnixWare.
15536
15537   *Richard Levitte*
15538
15539 * Check the result of RSA-CRT (see D. Boneh, R. DeMillo, R. Lipton:
15540   On the Importance of Eliminating Errors in Cryptographic
15541   Computations, J. Cryptology 14 (2001) 2, 101-119,
15542   <http://theory.stanford.edu/~dabo/papers/faults.ps.gz>).
15543
15544   *Ulf Moeller*
15545
15546 * MIPS assembler BIGNUM division bug fix.
15547
15548   *Andy Polyakov*
15549
15550 * Disabled incorrect Alpha assembler code.
15551
15552   *Richard Levitte*
15553
15554 * Fix PKCS#7 decode routines so they correctly update the length
15555   after reading an EOC for the EXPLICIT tag.
15556
15557   *Steve Henson*
15558
15559   *This change does not apply to 0.9.7.*
15560
15561 * Fix bug in PKCS#12 key generation routines. This was triggered
15562   if a 3DES key was generated with a 0 initial byte. Include
15563   PKCS12_BROKEN_KEYGEN compilation option to retain the old
15564   (but broken) behaviour.
15565
15566   *Steve Henson*
15567
15568 * Enhance bctest to search for a working bc along $PATH and print
15569   it when found.
15570
15571   *Tim Rice <tim@multitalents.net> via Richard Levitte*
15572
15573 * Fix memory leaks in err.c: free err_data string if necessary;
15574   don't write to the wrong index in ERR_set_error_data.
15575
15576   *Bodo Moeller*
15577
15578 * Implement ssl23_peek (analogous to ssl23_read), which previously
15579   did not exist.
15580
15581   *Bodo Moeller*
15582
15583 * Replace rdtsc with `_emit` statements for VC++ version 5.
15584
15585   *Jeremy Cooper <jeremy@baymoo.org>*
15586
15587 * Make it possible to reuse SSLv2 sessions.
15588
15589   *Richard Levitte*
15590
15591 * In copy_email() check for >= 0 as a return value for
15592   X509_NAME_get_index_by_NID() since 0 is a valid index.
15593
15594   *Steve Henson reported by Massimiliano Pala <madwolf@opensca.org>*
15595
15596 * Avoid coredump with unsupported or invalid public keys by checking if
15597   X509_get_pubkey() fails in PKCS7_verify(). Fix memory leak when
15598   PKCS7_verify() fails with non detached data.
15599
15600   *Steve Henson*
15601
15602 * Don't use getenv in library functions when run as setuid/setgid.
15603   New function OPENSSL_issetugid().
15604
15605   *Ulf Moeller*
15606
15607 * Avoid false positives in memory leak detection code (crypto/mem_dbg.c)
15608   due to incorrect handling of multi-threading:
15609
15610   1. Fix timing glitch in the MemCheck_off() portion of CRYPTO_mem_ctrl().
15611
15612   2. Fix logical glitch in is_MemCheck_on() aka CRYPTO_is_mem_check_on().
15613
15614   3. Count how many times MemCheck_off() has been called so that
15615      nested use can be treated correctly.  This also avoids
15616      inband-signalling in the previous code (which relied on the
15617      assumption that thread ID 0 is impossible).
15618
15619   *Bodo Moeller*
15620
15621 * Add "-rand" option also to s_client and s_server.
15622
15623   *Lutz Jaenicke*
15624
15625 * Fix CPU detection on Irix 6.x.
15626   *Kurt Hockenbury <khockenb@stevens-tech.edu> and
15627   "Bruce W. Forsberg" <bruce.forsberg@baesystems.com>*
15628
15629 * Fix X509_NAME bug which produced incorrect encoding if X509_NAME
15630   was empty.
15631
15632   *Steve Henson*
15633
15634   *This change does not apply to 0.9.7.*
15635
15636 * Use the cached encoding of an X509_NAME structure rather than
15637   copying it. This is apparently the reason for the libsafe "errors"
15638   but the code is actually correct.
15639
15640   *Steve Henson*
15641
15642 * Add new function BN_rand_range(), and fix DSA_sign_setup() to prevent
15643   Bleichenbacher's DSA attack.
15644   Extend BN_[pseudo_]rand: As before, top=1 forces the highest two bits
15645   to be set and top=0 forces the highest bit to be set; top=-1 is new
15646   and leaves the highest bit random.
15647
15648   *Ulf Moeller, Bodo Moeller*
15649
15650 * In the `NCONF_...`-based implementations for `CONF_...` queries
15651   (crypto/conf/conf_lib.c), if the input LHASH is NULL, avoid using
15652   a temporary CONF structure with the data component set to NULL
15653   (which gives segmentation faults in lh_retrieve).
15654   Instead, use NULL for the CONF pointer in CONF_get_string and
15655   CONF_get_number (which may use environment variables) and directly
15656   return NULL from CONF_get_section.
15657
15658   *Bodo Moeller*
15659
15660 * Fix potential buffer overrun for EBCDIC.
15661
15662   *Ulf Moeller*
15663
15664 * Tolerate nonRepudiation as being valid for S/MIME signing and certSign
15665   keyUsage if basicConstraints absent for a CA.
15666
15667   *Steve Henson*
15668
15669 * Make SMIME_write_PKCS7() write mail header values with a format that
15670   is more generally accepted (no spaces before the semicolon), since
15671   some programs can't parse those values properly otherwise.  Also make
15672   sure BIO's that break lines after each write do not create invalid
15673   headers.
15674
15675   *Richard Levitte*
15676
15677 * Make the CRL encoding routines work with empty SEQUENCE OF. The
15678   macros previously used would not encode an empty SEQUENCE OF
15679   and break the signature.
15680
15681   *Steve Henson*
15682
15683   *This change does not apply to 0.9.7.*
15684
15685 * Zero the premaster secret after deriving the master secret in
15686   DH ciphersuites.
15687
15688   *Steve Henson*
15689
15690 * Add some EVP_add_digest_alias registrations (as found in
15691   OpenSSL_add_all_digests()) to SSL_library_init()
15692   aka OpenSSL_add_ssl_algorithms().  This provides improved
15693   compatibility with peers using X.509 certificates
15694   with unconventional AlgorithmIdentifier OIDs.
15695
15696   *Bodo Moeller*
15697
15698 * Fix for Irix with NO_ASM.
15699
15700   *"Bruce W. Forsberg" <bruce.forsberg@baesystems.com>*
15701
15702 * ./config script fixes.
15703
15704   *Ulf Moeller, Richard Levitte*
15705
15706 * Fix 'openssl passwd -1'.
15707
15708   *Bodo Moeller*
15709
15710 * Change PKCS12_key_gen_asc() so it can cope with non null
15711   terminated strings whose length is passed in the passlen
15712   parameter, for example from PEM callbacks. This was done
15713   by adding an extra length parameter to asc2uni().
15714
15715   *Steve Henson, reported by <oddissey@samsung.co.kr>*
15716
15717 * Fix C code generated by 'openssl dsaparam -C': If a BN_bin2bn
15718   call failed, free the DSA structure.
15719
15720   *Bodo Moeller*
15721
15722 * Fix to uni2asc() to cope with zero length Unicode strings.
15723   These are present in some PKCS#12 files.
15724
15725   *Steve Henson*
15726
15727 * Increase s2->wbuf allocation by one byte in ssl2_new (ssl/s2_lib.c).
15728   Otherwise do_ssl_write (ssl/s2_pkt.c) will write beyond buffer limits
15729   when writing a 32767 byte record.
15730
15731   *Bodo Moeller; problem reported by Eric Day <eday@concentric.net>*
15732
15733 * In `RSA_eay_public_{en,ed}crypt` and RSA_eay_mod_exp (rsa_eay.c),
15734   obtain lock CRYPTO_LOCK_RSA before setting `rsa->_method_mod_{n,p,q}`.
15735
15736   (RSA objects have a reference count access to which is protected
15737   by CRYPTO_LOCK_RSA [see rsa_lib.c, s3_srvr.c, ssl_cert.c, ssl_rsa.c],
15738   so they are meant to be shared between threads.)
15739   *Bodo Moeller, Geoff Thorpe; original patch submitted by
15740   "Reddie, Steven" <Steven.Reddie@ca.com>*
15741
15742 * Fix a deadlock in CRYPTO_mem_leaks().
15743
15744   *Bodo Moeller*
15745
15746 * Use better test patterns in bntest.
15747
15748   *Ulf Möller*
15749
15750 * rand_win.c fix for Borland C.
15751
15752   *Ulf Möller*
15753
15754 * BN_rshift bugfix for n == 0.
15755
15756   *Bodo Moeller*
15757
15758 * Add a 'bctest' script that checks for some known 'bc' bugs
15759   so that 'make test' does not abort just because 'bc' is broken.
15760
15761   *Bodo Moeller*
15762
15763 * Store verify_result within SSL_SESSION also for client side to
15764   avoid potential security hole. (Re-used sessions on the client side
15765   always resulted in verify_result==X509_V_OK, not using the original
15766   result of the server certificate verification.)
15767
15768   *Lutz Jaenicke*
15769
15770 * Fix ssl3_pending: If the record in s->s3->rrec is not of type
15771   SSL3_RT_APPLICATION_DATA, return 0.
15772   Similarly, change ssl2_pending to return 0 if SSL_in_init(s) is true.
15773
15774   *Bodo Moeller*
15775
15776 * Fix SSL_peek:
15777   Both ssl2_peek and ssl3_peek, which were totally broken in earlier
15778   releases, have been re-implemented by renaming the previous
15779   implementations of ssl2_read and ssl3_read to ssl2_read_internal
15780   and ssl3_read_internal, respectively, and adding 'peek' parameters
15781   to them.  The new ssl[23]_{read,peek} functions are calls to
15782   ssl[23]_read_internal with the 'peek' flag set appropriately.
15783   A 'peek' parameter has also been added to ssl3_read_bytes, which
15784   does the actual work for ssl3_read_internal.
15785
15786   *Bodo Moeller*
15787
15788 * Initialise "ex_data" member of RSA/DSA/DH structures prior to calling
15789   the method-specific "init()" handler. Also clean up ex_data after
15790   calling the method-specific "finish()" handler. Previously, this was
15791   happening the other way round.
15792
15793   *Geoff Thorpe*
15794
15795 * Increase BN_CTX_NUM (the number of BIGNUMs in a BN_CTX) to 16.
15796   The previous value, 12, was not always sufficient for BN_mod_exp().
15797
15798   *Bodo Moeller*
15799
15800 * Make sure that shared libraries get the internal name engine with
15801   the full version number and not just 0.  This should mark the
15802   shared libraries as not backward compatible.  Of course, this should
15803   be changed again when we can guarantee backward binary compatibility.
15804
15805   *Richard Levitte*
15806
15807 * Fix typo in get_cert_by_subject() in by_dir.c
15808
15809   *Jean-Marc Desperrier <jean-marc.desperrier@certplus.com>*
15810
15811 * Rework the system to generate shared libraries:
15812
15813   - Make note of the expected extension for the shared libraries and
15814     if there is a need for symbolic links from for example libcrypto.so.0
15815     to libcrypto.so.0.9.7.  There is extended info in Configure for
15816     that.
15817
15818   - Make as few rebuilds of the shared libraries as possible.
15819
15820   - Still avoid linking the OpenSSL programs with the shared libraries.
15821
15822   - When installing, install the shared libraries separately from the
15823     static ones.
15824
15825   *Richard Levitte*
15826
15827 * Fix SSL_CTX_set_read_ahead macro to actually use its argument.
15828
15829   Copy SSL_CTX's read_ahead flag to SSL object directly in SSL_new
15830   and not in SSL_clear because the latter is also used by the
15831   accept/connect functions; previously, the settings made by
15832   SSL_set_read_ahead would be lost during the handshake.
15833
15834   *Bodo Moeller; problems reported by Anders Gertz <gertz@epact.se>*
15835
15836 * Correct util/mkdef.pl to be selective about disabled algorithms.
15837   Previously, it would create entries for disabled algorithms no
15838   matter what.
15839
15840   *Richard Levitte*
15841
15842 * Added several new manual pages for SSL_* function.
15843
15844   *Lutz Jaenicke*
15845
15846### Changes between 0.9.5a and 0.9.6  [24 Sep 2000]
15847
15848 * In ssl23_get_client_hello, generate an error message when faced
15849   with an initial SSL 3.0/TLS record that is too small to contain the
15850   first two bytes of the ClientHello message, i.e. client_version.
15851   (Note that this is a pathologic case that probably has never happened
15852   in real life.)  The previous approach was to use the version number
15853   from the record header as a substitute; but our protocol choice
15854   should not depend on that one because it is not authenticated
15855   by the Finished messages.
15856
15857   *Bodo Moeller*
15858
15859 * More robust randomness gathering functions for Windows.
15860
15861   *Jeffrey Altman <jaltman@columbia.edu>*
15862
15863 * For compatibility reasons if the flag X509_V_FLAG_ISSUER_CHECK is
15864   not set then we don't setup the error code for issuer check errors
15865   to avoid possibly overwriting other errors which the callback does
15866   handle. If an application does set the flag then we assume it knows
15867   what it is doing and can handle the new informational codes
15868   appropriately.
15869
15870   *Steve Henson*
15871
15872 * Fix for a nasty bug in ASN1_TYPE handling. ASN1_TYPE is used for
15873   a general "ANY" type, as such it should be able to decode anything
15874   including tagged types. However it didn't check the class so it would
15875   wrongly interpret tagged types in the same way as their universal
15876   counterpart and unknown types were just rejected. Changed so that the
15877   tagged and unknown types are handled in the same way as a SEQUENCE:
15878   that is the encoding is stored intact. There is also a new type
15879   "V_ASN1_OTHER" which is used when the class is not universal, in this
15880   case we have no idea what the actual type is so we just lump them all
15881   together.
15882
15883   *Steve Henson*
15884
15885 * On VMS, stdout may very well lead to a file that is written to
15886   in a record-oriented fashion.  That means that every write() will
15887   write a separate record, which will be read separately by the
15888   programs trying to read from it.  This can be very confusing.
15889
15890   The solution is to put a BIO filter in the way that will buffer
15891   text until a linefeed is reached, and then write everything a
15892   line at a time, so every record written will be an actual line,
15893   not chunks of lines and not (usually doesn't happen, but I've
15894   seen it once) several lines in one record.  BIO_f_linebuffer() is
15895   the answer.
15896
15897   Currently, it's a VMS-only method, because that's where it has
15898   been tested well enough.
15899
15900   *Richard Levitte*
15901
15902 * Remove 'optimized' squaring variant in BN_mod_mul_montgomery,
15903   it can return incorrect results.
15904   (Note: The buggy variant was not enabled in OpenSSL 0.9.5a,
15905   but it was in 0.9.6-beta[12].)
15906
15907   *Bodo Moeller*
15908
15909 * Disable the check for content being present when verifying detached
15910   signatures in pk7_smime.c. Some versions of Netscape (wrongly)
15911   include zero length content when signing messages.
15912
15913   *Steve Henson*
15914
15915 * New BIO_shutdown_wr macro, which invokes the BIO_C_SHUTDOWN_WR
15916   BIO_ctrl (for BIO pairs).
15917
15918   *Bodo Möller*
15919
15920 * Add DSO method for VMS.
15921
15922   *Richard Levitte*
15923
15924 * Bug fix: Montgomery multiplication could produce results with the
15925   wrong sign.
15926
15927   *Ulf Möller*
15928
15929 * Add RPM specification openssl.spec and modify it to build three
15930   packages.  The default package contains applications, application
15931   documentation and run-time libraries.  The devel package contains
15932   include files, static libraries and function documentation.  The
15933   doc package contains the contents of the doc directory.  The original
15934   openssl.spec was provided by Damien Miller <djm@mindrot.org>.
15935
15936   *Richard Levitte*
15937
15938 * Add a large number of documentation files for many SSL routines.
15939
15940   *Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>*
15941
15942 * Add a configuration entry for Sony News 4.
15943
15944   *NAKAJI Hiroyuki <nakaji@tutrp.tut.ac.jp>*
15945
15946 * Don't set the two most significant bits to one when generating a
15947   random number < q in the DSA library.
15948
15949   *Ulf Möller*
15950
15951 * New SSL API mode 'SSL_MODE_AUTO_RETRY'.  This disables the default
15952   behaviour that SSL_read may result in SSL_ERROR_WANT_READ (even if
15953   the underlying transport is blocking) if a handshake took place.
15954   (The default behaviour is needed by applications such as s_client
15955   and s_server that use select() to determine when to use SSL_read;
15956   but for applications that know in advance when to expect data, it
15957   just makes things more complicated.)
15958
15959   *Bodo Moeller*
15960
15961 * Add RAND_egd_bytes(), which gives control over the number of bytes read
15962   from EGD.
15963
15964   *Ben Laurie*
15965
15966 * Add a few more EBCDIC conditionals that make `req` and `x509`
15967   work better on such systems.
15968
15969   *Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>*
15970
15971 * Add two demo programs for PKCS12_parse() and PKCS12_create().
15972   Update PKCS12_parse() so it copies the friendlyName and the
15973   keyid to the certificates aux info.
15974
15975   *Steve Henson*
15976
15977 * Fix bug in PKCS7_verify() which caused an infinite loop
15978   if there was more than one signature.
15979
15980   *Sven Uszpelkat <su@celocom.de>*
15981
15982 * Major change in util/mkdef.pl to include extra information
15983   about each symbol, as well as presenting variables as well
15984   as functions.  This change means that there's n more need
15985   to rebuild the .num files when some algorithms are excluded.
15986
15987   *Richard Levitte*
15988
15989 * Allow the verify time to be set by an application,
15990   rather than always using the current time.
15991
15992   *Steve Henson*
15993
15994 * Phase 2 verify code reorganisation. The certificate
15995   verify code now looks up an issuer certificate by a
15996   number of criteria: subject name, authority key id
15997   and key usage. It also verifies self signed certificates
15998   by the same criteria. The main comparison function is
15999   X509_check_issued() which performs these checks.
16000
16001   Lot of changes were necessary in order to support this
16002   without completely rewriting the lookup code.
16003
16004   Authority and subject key identifier are now cached.
16005
16006   The LHASH 'certs' is X509_STORE has now been replaced
16007   by a STACK_OF(X509_OBJECT). This is mainly because an
16008   LHASH can't store or retrieve multiple objects with
16009   the same hash value.
16010
16011   As a result various functions (which were all internal
16012   use only) have changed to handle the new X509_STORE
16013   structure. This will break anything that messed round
16014   with X509_STORE internally.
16015
16016   The functions X509_STORE_add_cert() now checks for an
16017   exact match, rather than just subject name.
16018
16019   The X509_STORE API doesn't directly support the retrieval
16020   of multiple certificates matching a given criteria, however
16021   this can be worked round by performing a lookup first
16022   (which will fill the cache with candidate certificates)
16023   and then examining the cache for matches. This is probably
16024   the best we can do without throwing out X509_LOOKUP
16025   entirely (maybe later...).
16026
16027   The X509_VERIFY_CTX structure has been enhanced considerably.
16028
16029   All certificate lookup operations now go via a get_issuer()
16030   callback. Although this currently uses an X509_STORE it
16031   can be replaced by custom lookups. This is a simple way
16032   to bypass the X509_STORE hackery necessary to make this
16033   work and makes it possible to use more efficient techniques
16034   in future. A very simple version which uses a simple
16035   STACK for its trusted certificate store is also provided
16036   using X509_STORE_CTX_trusted_stack().
16037
16038   The verify_cb() and verify() callbacks now have equivalents
16039   in the X509_STORE_CTX structure.
16040
16041   X509_STORE_CTX also has a 'flags' field which can be used
16042   to customise the verify behaviour.
16043
16044   *Steve Henson*
16045
16046 * Add new PKCS#7 signing option PKCS7_NOSMIMECAP which
16047   excludes S/MIME capabilities.
16048
16049   *Steve Henson*
16050
16051 * When a certificate request is read in keep a copy of the
16052   original encoding of the signed data and use it when outputting
16053   again. Signatures then use the original encoding rather than
16054   a decoded, encoded version which may cause problems if the
16055   request is improperly encoded.
16056
16057   *Steve Henson*
16058
16059 * For consistency with other BIO_puts implementations, call
16060   buffer_write(b, ...) directly in buffer_puts instead of calling
16061   BIO_write(b, ...).
16062
16063   In BIO_puts, increment b->num_write as in BIO_write.
16064
16065   *Peter.Sylvester@EdelWeb.fr*
16066
16067 * Fix BN_mul_word for the case where the word is 0. (We have to use
16068   BN_zero, we may not return a BIGNUM with an array consisting of
16069   words set to zero.)
16070
16071   *Bodo Moeller*
16072
16073 * Avoid calling abort() from within the library when problems are
16074   detected, except if preprocessor symbols have been defined
16075   (such as REF_CHECK, BN_DEBUG etc.).
16076
16077   *Bodo Moeller*
16078
16079 * New openssl application 'rsautl'. This utility can be
16080   used for low-level RSA operations. DER public key
16081   BIO/fp routines also added.
16082
16083   *Steve Henson*
16084
16085 * New Configure entry and patches for compiling on QNX 4.
16086
16087   *Andreas Schneider <andreas@ds3.etech.fh-hamburg.de>*
16088
16089 * A demo state-machine implementation was sponsored by
16090   Nuron (<http://www.nuron.com/>) and is now available in
16091   demos/state_machine.
16092
16093   *Ben Laurie*
16094
16095 * New options added to the 'dgst' utility for signature
16096   generation and verification.
16097
16098   *Steve Henson*
16099
16100 * Unrecognized PKCS#7 content types are now handled via a
16101   catch all ASN1_TYPE structure. This allows unsupported
16102   types to be stored as a "blob" and an application can
16103   encode and decode it manually.
16104
16105   *Steve Henson*
16106
16107 * Fix various signed/unsigned issues to make a_strex.c
16108   compile under VC++.
16109
16110   *Oscar Jacobsson <oscar.jacobsson@celocom.com>*
16111
16112 * ASN1 fixes. i2d_ASN1_OBJECT was not returning the correct
16113   length if passed a buffer. ASN1_INTEGER_to_BN failed
16114   if passed a NULL BN and its argument was negative.
16115
16116   *Steve Henson, pointed out by Sven Heiberg <sven@tartu.cyber.ee>*
16117
16118 * Modification to PKCS#7 encoding routines to output definite
16119   length encoding. Since currently the whole structures are in
16120   memory there's not real point in using indefinite length
16121   constructed encoding. However if OpenSSL is compiled with
16122   the flag PKCS7_INDEFINITE_ENCODING the old form is used.
16123
16124   *Steve Henson*
16125
16126 * Added BIO_vprintf() and BIO_vsnprintf().
16127
16128   *Richard Levitte*
16129
16130 * Added more prefixes to parse for in the strings written
16131   through a logging bio, to cover all the levels that are available
16132   through syslog.  The prefixes are now:
16133
16134           PANIC, EMERG, EMR       =>      LOG_EMERG
16135           ALERT, ALR              =>      LOG_ALERT
16136           CRIT, CRI               =>      LOG_CRIT
16137           ERROR, ERR              =>      LOG_ERR
16138           WARNING, WARN, WAR      =>      LOG_WARNING
16139           NOTICE, NOTE, NOT       =>      LOG_NOTICE
16140           INFO, INF               =>      LOG_INFO
16141           DEBUG, DBG              =>      LOG_DEBUG
16142
16143   and as before, if none of those prefixes are present at the
16144   beginning of the string, LOG_ERR is chosen.
16145
16146   On Win32, the `LOG_*` levels are mapped according to this:
16147
16148           LOG_EMERG, LOG_ALERT, LOG_CRIT, LOG_ERR => EVENTLOG_ERROR_TYPE
16149           LOG_WARNING                             => EVENTLOG_WARNING_TYPE
16150           LOG_NOTICE, LOG_INFO, LOG_DEBUG         => EVENTLOG_INFORMATION_TYPE
16151
16152   *Richard Levitte*
16153
16154 * Made it possible to reconfigure with just the configuration
16155   argument "reconf" or "reconfigure".  The command line arguments
16156   are stored in Makefile.ssl in the variable CONFIGURE_ARGS,
16157   and are retrieved from there when reconfiguring.
16158
16159   *Richard Levitte*
16160
16161 * MD4 implemented.
16162
16163   *Assar Westerlund <assar@sics.se>, Richard Levitte*
16164
16165 * Add the arguments -CAfile and -CApath to the pkcs12 utility.
16166
16167   *Richard Levitte*
16168
16169 * The obj_dat.pl script was messing up the sorting of object
16170   names. The reason was that it compared the quoted version
16171   of strings as a result "OCSP" > "OCSP Signing" because
16172   " > SPACE. Changed script to store unquoted versions of
16173   names and add quotes on output. It was also omitting some
16174   names from the lookup table if they were given a default
16175   value (that is if SN is missing it is given the same
16176   value as LN and vice versa), these are now added on the
16177   grounds that if an object has a name we should be able to
16178   look it up. Finally added warning output when duplicate
16179   short or long names are found.
16180
16181   *Steve Henson*
16182
16183 * Changes needed for Tandem NSK.
16184
16185   *Scott Uroff <scott@xypro.com>*
16186
16187 * Fix SSL 2.0 rollback checking: Due to an off-by-one error in
16188   RSA_padding_check_SSLv23(), special padding was never detected
16189   and thus the SSL 3.0/TLS 1.0 countermeasure against protocol
16190   version rollback attacks was not effective.
16191
16192   In s23_clnt.c, don't use special rollback-attack detection padding
16193   (RSA_SSLV23_PADDING) if SSL 2.0 is the only protocol enabled in the
16194   client; similarly, in s23_srvr.c, don't do the rollback check if
16195   SSL 2.0 is the only protocol enabled in the server.
16196
16197   *Bodo Moeller*
16198
16199 * Make it possible to get hexdumps of unprintable data with 'openssl
16200   asn1parse'.  By implication, the functions ASN1_parse_dump() and
16201   BIO_dump_indent() are added.
16202
16203   *Richard Levitte*
16204
16205 * New functions ASN1_STRING_print_ex() and X509_NAME_print_ex()
16206   these print out strings and name structures based on various
16207   flags including RFC2253 support and proper handling of
16208   multibyte characters. Added options to the 'x509' utility
16209   to allow the various flags to be set.
16210
16211   *Steve Henson*
16212
16213 * Various fixes to use ASN1_TIME instead of ASN1_UTCTIME.
16214   Also change the functions X509_cmp_current_time() and
16215   X509_gmtime_adj() work with an ASN1_TIME structure,
16216   this will enable certificates using GeneralizedTime in validity
16217   dates to be checked.
16218
16219   *Steve Henson*
16220
16221 * Make the NEG_PUBKEY_BUG code (which tolerates invalid
16222   negative public key encodings) on by default,
16223   NO_NEG_PUBKEY_BUG can be set to disable it.
16224
16225   *Steve Henson*
16226
16227 * New function c2i_ASN1_OBJECT() which acts on ASN1_OBJECT
16228   content octets. An i2c_ASN1_OBJECT is unnecessary because
16229   the encoding can be trivially obtained from the structure.
16230
16231   *Steve Henson*
16232
16233 * crypto/err.c locking bugfix: Use write locks (`CRYPTO_w_[un]lock`),
16234   not read locks (`CRYPTO_r_[un]lock`).
16235
16236   *Bodo Moeller*
16237
16238 * A first attempt at creating official support for shared
16239   libraries through configuration.  I've kept it so the
16240   default is static libraries only, and the OpenSSL programs
16241   are always statically linked for now, but there are
16242   preparations for dynamic linking in place.
16243   This has been tested on Linux and Tru64.
16244
16245   *Richard Levitte*
16246
16247 * Randomness polling function for Win9x, as described in:
16248   Peter Gutmann, Software Generation of Practically Strong
16249   Random Numbers.
16250
16251   *Ulf Möller*
16252
16253 * Fix so PRNG is seeded in req if using an already existing
16254   DSA key.
16255
16256   *Steve Henson*
16257
16258 * New options to smime application. -inform and -outform
16259   allow alternative formats for the S/MIME message including
16260   PEM and DER. The -content option allows the content to be
16261   specified separately. This should allow things like Netscape
16262   form signing output easier to verify.
16263
16264   *Steve Henson*
16265
16266 * Fix the ASN1 encoding of tags using the 'long form'.
16267
16268   *Steve Henson*
16269
16270 * New ASN1 functions, `i2c_*` and `c2i_*` for INTEGER and BIT
16271   STRING types. These convert content octets to and from the
16272   underlying type. The actual tag and length octets are
16273   already assumed to have been read in and checked. These
16274   are needed because all other string types have virtually
16275   identical handling apart from the tag. By having versions
16276   of the ASN1 functions that just operate on content octets
16277   IMPLICIT tagging can be handled properly. It also allows
16278   the ASN1_ENUMERATED code to be cut down because ASN1_ENUMERATED
16279   and ASN1_INTEGER are identical apart from the tag.
16280
16281   *Steve Henson*
16282
16283 * Change the handling of OID objects as follows:
16284
16285   - New object identifiers are inserted in objects.txt, following
16286     the syntax given in [crypto/objects/README.md](crypto/objects/README.md).
16287   - objects.pl is used to process obj_mac.num and create a new
16288     obj_mac.h.
16289   - obj_dat.pl is used to create a new obj_dat.h, using the data in
16290     obj_mac.h.
16291
16292   This is currently kind of a hack, and the perl code in objects.pl
16293   isn't very elegant, but it works as I intended.  The simplest way
16294   to check that it worked correctly is to look in obj_dat.h and
16295   check the array nid_objs and make sure the objects haven't moved
16296   around (this is important!).  Additions are OK, as well as
16297   consistent name changes.
16298
16299   *Richard Levitte*
16300
16301 * Add BSD-style MD5-based passwords to 'openssl passwd' (option '-1').
16302
16303   *Bodo Moeller*
16304
16305 * Addition of the command line parameter '-rand file' to 'openssl req'.
16306   The given file adds to whatever has already been seeded into the
16307   random pool through the RANDFILE configuration file option or
16308   environment variable, or the default random state file.
16309
16310   *Richard Levitte*
16311
16312 * mkstack.pl now sorts each macro group into lexical order.
16313   Previously the output order depended on the order the files
16314   appeared in the directory, resulting in needless rewriting
16315   of safestack.h .
16316
16317   *Steve Henson*
16318
16319 * Patches to make OpenSSL compile under Win32 again. Mostly
16320   work arounds for the VC++ problem that it treats func() as
16321   func(void). Also stripped out the parts of mkdef.pl that
16322   added extra typesafe functions: these no longer exist.
16323
16324   *Steve Henson*
16325
16326 * Reorganisation of the stack code. The macros are now all
16327   collected in safestack.h . Each macro is defined in terms of
16328   a "stack macro" of the form `SKM_<name>(type, a, b)`. The
16329   DEBUG_SAFESTACK is now handled in terms of function casts,
16330   this has the advantage of retaining type safety without the
16331   use of additional functions. If DEBUG_SAFESTACK is not defined
16332   then the non typesafe macros are used instead. Also modified the
16333   mkstack.pl script to handle the new form. Needs testing to see
16334   if which (if any) compilers it chokes and maybe make DEBUG_SAFESTACK
16335   the default if no major problems. Similar behaviour for ASN1_SET_OF
16336   and PKCS12_STACK_OF.
16337
16338   *Steve Henson*
16339
16340 * When some versions of IIS use the 'NET' form of private key the
16341   key derivation algorithm is different. Normally MD5(password) is
16342   used as a 128 bit RC4 key. In the modified case
16343   MD5(MD5(password) + "SGCKEYSALT")  is used instead. Added some
16344   new functions i2d_RSA_NET(), d2i_RSA_NET() etc which are the same
16345   as the old Netscape_RSA functions except they have an additional
16346   'sgckey' parameter which uses the modified algorithm. Also added
16347   an -sgckey command line option to the rsa utility. Thanks to
16348   Adrian Peck <bertie@ncipher.com> for posting details of the modified
16349   algorithm to openssl-dev.
16350
16351   *Steve Henson*
16352
16353 * The evp_local.h macros were using 'c.##kname' which resulted in
16354   invalid expansion on some systems (SCO 5.0.5 for example).
16355   Corrected to 'c.kname'.
16356
16357   *Phillip Porch <root@theporch.com>*
16358
16359 * New X509_get1_email() and X509_REQ_get1_email() functions that return
16360   a STACK of email addresses from a certificate or request, these look
16361   in the subject name and the subject alternative name extensions and
16362   omit any duplicate addresses.
16363
16364   *Steve Henson*
16365
16366 * Re-implement BN_mod_exp2_mont using independent (and larger) windows.
16367   This makes DSA verification about 2 % faster.
16368
16369   *Bodo Moeller*
16370
16371 * Increase maximum window size in `BN_mod_exp_...` to 6 bits instead of 5
16372   (meaning that now 2^5 values will be precomputed, which is only 4 KB
16373   plus overhead for 1024 bit moduli).
16374   This makes exponentiations about 0.5 % faster for 1024 bit
16375   exponents (as measured by "openssl speed rsa2048").
16376
16377   *Bodo Moeller*
16378
16379 * Rename memory handling macros to avoid conflicts with other
16380   software:
16381           Malloc         =>  OPENSSL_malloc
16382           Malloc_locked  =>  OPENSSL_malloc_locked
16383           Realloc        =>  OPENSSL_realloc
16384           Free           =>  OPENSSL_free
16385
16386   *Richard Levitte*
16387
16388 * New function BN_mod_exp_mont_word for small bases (roughly 15%
16389   faster than BN_mod_exp_mont, i.e. 7% for a full DH exchange).
16390
16391   *Bodo Moeller*
16392
16393 * CygWin32 support.
16394
16395   *John Jarvie <jjarvie@newsguy.com>*
16396
16397 * The type-safe stack code has been rejigged. It is now only compiled
16398   in when OpenSSL is configured with the DEBUG_SAFESTACK option and
16399   by default all type-specific stack functions are "#define"d back to
16400   standard stack functions. This results in more streamlined output
16401   but retains the type-safety checking possibilities of the original
16402   approach.
16403
16404   *Geoff Thorpe*
16405
16406 * The STACK code has been cleaned up, and certain type declarations
16407   that didn't make a lot of sense have been brought in line. This has
16408   also involved a cleanup of sorts in safestack.h to more correctly
16409   map type-safe stack functions onto their plain stack counterparts.
16410   This work has also resulted in a variety of "const"ifications of
16411   lots of the code, especially `_cmp` operations which should normally
16412   be prototyped with "const" parameters anyway.
16413
16414   *Geoff Thorpe*
16415
16416 * When generating bytes for the first time in md_rand.c, 'stir the pool'
16417   by seeding with STATE_SIZE dummy bytes (with zero entropy count).
16418   (The PRNG state consists of two parts, the large pool 'state' and 'md',
16419   where all of 'md' is used each time the PRNG is used, but 'state'
16420   is used only indexed by a cyclic counter. As entropy may not be
16421   well distributed from the beginning, 'md' is important as a
16422   chaining variable. However, the output function chains only half
16423   of 'md', i.e. 80 bits.  ssleay_rand_add, on the other hand, chains
16424   all of 'md', and seeding with STATE_SIZE dummy bytes will result
16425   in all of 'state' being rewritten, with the new values depending
16426   on virtually all of 'md'.  This overcomes the 80 bit limitation.)
16427
16428   *Bodo Moeller*
16429
16430 * In ssl/s2_clnt.c and ssl/s3_clnt.c, call ERR_clear_error() when
16431   the handshake is continued after ssl_verify_cert_chain();
16432   otherwise, if SSL_VERIFY_NONE is set, remaining error codes
16433   can lead to 'unexplainable' connection aborts later.
16434
16435   *Bodo Moeller; problem tracked down by Lutz Jaenicke*
16436
16437 * Major EVP API cipher revision.
16438   Add hooks for extra EVP features. This allows various cipher
16439   parameters to be set in the EVP interface. Support added for variable
16440   key length ciphers via the EVP_CIPHER_CTX_set_key_length() function and
16441   setting of RC2 and RC5 parameters.
16442
16443   Modify EVP_OpenInit() and EVP_SealInit() to cope with variable key length
16444   ciphers.
16445
16446   Remove lots of duplicated code from the EVP library. For example *every*
16447   cipher init() function handles the 'iv' in the same way according to the
16448   cipher mode. They also all do nothing if the 'key' parameter is NULL and
16449   for CFB and OFB modes they zero ctx->num.
16450
16451   New functionality allows removal of S/MIME code RC2 hack.
16452
16453   Most of the routines have the same form and so can be declared in terms
16454   of macros.
16455
16456   By shifting this to the top level EVP_CipherInit() it can be removed from
16457   all individual ciphers. If the cipher wants to handle IVs or keys
16458   differently it can set the EVP_CIPH_CUSTOM_IV or EVP_CIPH_ALWAYS_CALL_INIT
16459   flags.
16460
16461   Change lots of functions like EVP_EncryptUpdate() to now return a
16462   value: although software versions of the algorithms cannot fail
16463   any installed hardware versions can.
16464
16465   *Steve Henson*
16466
16467 * Implement SSL_OP_TLS_ROLLBACK_BUG: In ssl3_get_client_key_exchange, if
16468   this option is set, tolerate broken clients that send the negotiated
16469   protocol version number instead of the requested protocol version
16470   number.
16471
16472   *Bodo Moeller*
16473
16474 * Call dh_tmp_cb (set by `..._TMP_DH_CB`) with correct 'is_export' flag;
16475   i.e. non-zero for export ciphersuites, zero otherwise.
16476   Previous versions had this flag inverted, inconsistent with
16477   rsa_tmp_cb (..._TMP_RSA_CB).
16478
16479   *Bodo Moeller; problem reported by Amit Chopra*
16480
16481 * Add missing DSA library text string. Work around for some IIS
16482   key files with invalid SEQUENCE encoding.
16483
16484   *Steve Henson*
16485
16486 * Add a document (doc/standards.txt) that list all kinds of standards
16487   and so on that are implemented in OpenSSL.
16488
16489   *Richard Levitte*
16490
16491 * Enhance c_rehash script. Old version would mishandle certificates
16492   with the same subject name hash and wouldn't handle CRLs at all.
16493   Added -fingerprint option to crl utility, to support new c_rehash
16494   features.
16495
16496   *Steve Henson*
16497
16498 * Eliminate non-ANSI declarations in crypto.h and stack.h.
16499
16500   *Ulf Möller*
16501
16502 * Fix for SSL server purpose checking. Server checking was
16503   rejecting certificates which had extended key usage present
16504   but no ssl client purpose.
16505
16506   *Steve Henson, reported by Rene Grosser <grosser@hisolutions.com>*
16507
16508 * Make PKCS#12 code work with no password. The PKCS#12 spec
16509   is a little unclear about how a blank password is handled.
16510   Since the password in encoded as a BMPString with terminating
16511   double NULL a zero length password would end up as just the
16512   double NULL. However no password at all is different and is
16513   handled differently in the PKCS#12 key generation code. NS
16514   treats a blank password as zero length. MSIE treats it as no
16515   password on export: but it will try both on import. We now do
16516   the same: PKCS12_parse() tries zero length and no password if
16517   the password is set to "" or NULL (NULL is now a valid password:
16518   it wasn't before) as does the pkcs12 application.
16519
16520   *Steve Henson*
16521
16522 * Bugfixes in `apps/x509.c`: Avoid a memory leak; and don't use
16523   perror when PEM_read_bio_X509_REQ fails, the error message must
16524   be obtained from the error queue.
16525
16526   *Bodo Moeller*
16527
16528 * Avoid 'thread_hash' memory leak in crypto/err/err.c by freeing
16529   it in ERR_remove_state if appropriate, and change ERR_get_state
16530   accordingly to avoid race conditions (this is necessary because
16531   thread_hash is no longer constant once set).
16532
16533   *Bodo Moeller*
16534
16535 * Bugfix for linux-elf makefile.one.
16536
16537   *Ulf Möller*
16538
16539 * RSA_get_default_method() will now cause a default
16540   RSA_METHOD to be chosen if one doesn't exist already.
16541   Previously this was only set during a call to RSA_new()
16542   or RSA_new_method(NULL) meaning it was possible for
16543   RSA_get_default_method() to return NULL.
16544
16545   *Geoff Thorpe*
16546
16547 * Added native name translation to the existing DSO code
16548   that will convert (if the flag to do so is set) filenames
16549   that are sufficiently small and have no path information
16550   into a canonical native form. Eg. "blah" converted to
16551   "libblah.so" or "blah.dll" etc.
16552
16553   *Geoff Thorpe*
16554
16555 * New function ERR_error_string_n(e, buf, len) which is like
16556   ERR_error_string(e, buf), but writes at most 'len' bytes
16557   including the 0 terminator.  For ERR_error_string_n, 'buf'
16558   may not be NULL.
16559
16560   *Damien Miller <djm@mindrot.org>, Bodo Moeller*
16561
16562 * CONF library reworked to become more general.  A new CONF
16563   configuration file reader "class" is implemented as well as a
16564   new functions (`NCONF_*`, for "New CONF") to handle it.  The now
16565   old `CONF_*` functions are still there, but are reimplemented to
16566   work in terms of the new functions.  Also, a set of functions
16567   to handle the internal storage of the configuration data is
16568   provided to make it easier to write new configuration file
16569   reader "classes" (I can definitely see something reading a
16570   configuration file in XML format, for example), called `_CONF_*`,
16571   or "the configuration storage API"...
16572
16573   The new configuration file reading functions are:
16574
16575           NCONF_new, NCONF_free, NCONF_load, NCONF_load_fp, NCONF_load_bio,
16576           NCONF_get_section, NCONF_get_string, NCONF_get_numbre
16577
16578           NCONF_default, NCONF_WIN32
16579
16580           NCONF_dump_fp, NCONF_dump_bio
16581
16582   NCONF_default and NCONF_WIN32 are method (or "class") choosers,
16583   NCONF_new creates a new CONF object.  This works in the same way
16584   as other interfaces in OpenSSL, like the BIO interface.
16585   `NCONF_dump_*` dump the internal storage of the configuration file,
16586   which is useful for debugging.  All other functions take the same
16587   arguments as the old `CONF_*` functions with the exception of the
16588   first that must be a `CONF *` instead of a `LHASH *`.
16589
16590   To make it easier to use the new classes with the old `CONF_*` functions,
16591   the function CONF_set_default_method is provided.
16592
16593   *Richard Levitte*
16594
16595 * Add '-tls1' option to 'openssl ciphers', which was already
16596   mentioned in the documentation but had not been implemented.
16597   (This option is not yet really useful because even the additional
16598   experimental TLS 1.0 ciphers are currently treated as SSL 3.0 ciphers.)
16599
16600   *Bodo Moeller*
16601
16602 * Initial DSO code added into libcrypto for letting OpenSSL (and
16603   OpenSSL-based applications) load shared libraries and bind to
16604   them in a portable way.
16605
16606   *Geoff Thorpe, with contributions from Richard Levitte*
16607
16608### Changes between 0.9.5 and 0.9.5a  [1 Apr 2000]
16609
16610 * Make sure _lrotl and _lrotr are only used with MSVC.
16611
16612 * Use lock CRYPTO_LOCK_RAND correctly in ssleay_rand_status
16613   (the default implementation of RAND_status).
16614
16615 * Rename openssl x509 option '-crlext', which was added in 0.9.5,
16616   to '-clrext' (= clear extensions), as intended and documented.
16617   *Bodo Moeller; inconsistency pointed out by Michael Attili
16618   <attili@amaxo.com>*
16619
16620 * Fix for HMAC. It wasn't zeroing the rest of the block if the key length
16621   was larger than the MD block size.
16622
16623   *Steve Henson, pointed out by Yost William <YostW@tce.com>*
16624
16625 * Modernise PKCS12_parse() so it uses STACK_OF(X509) for its ca argument
16626   fix a leak when the ca argument was passed as NULL. Stop X509_PUBKEY_set()
16627   using the passed key: if the passed key was a private key the result
16628   of X509_print(), for example, would be to print out all the private key
16629   components.
16630
16631   *Steve Henson*
16632
16633 * des_quad_cksum() byte order bug fix.
16634   *Ulf Möller, using the problem description in krb4-0.9.7, where
16635   the solution is attributed to Derrick J Brashear <shadow@DEMENTIA.ORG>*
16636
16637 * Fix so V_ASN1_APP_CHOOSE works again: however its use is strongly
16638   discouraged.
16639
16640   *Steve Henson, pointed out by Brian Korver <briank@cs.stanford.edu>*
16641
16642 * For easily testing in shell scripts whether some command
16643   'openssl XXX' exists, the new pseudo-command 'openssl no-XXX'
16644   returns with exit code 0 iff no command of the given name is available.
16645   'no-XXX' is printed in this case, 'XXX' otherwise.  In both cases,
16646   the output goes to stdout and nothing is printed to stderr.
16647   Additional arguments are always ignored.
16648
16649   Since for each cipher there is a command of the same name,
16650   the 'no-cipher' compilation switches can be tested this way.
16651
16652   ('openssl no-XXX' is not able to detect pseudo-commands such
16653   as 'quit', 'list-XXX-commands', or 'no-XXX' itself.)
16654
16655   *Bodo Moeller*
16656
16657 * Update test suite so that 'make test' succeeds in 'no-rsa' configuration.
16658
16659   *Bodo Moeller*
16660
16661 * For SSL_[CTX_]set_tmp_dh, don't create a DH key if SSL_OP_SINGLE_DH_USE
16662   is set; it will be thrown away anyway because each handshake creates
16663   its own key.
16664   ssl_cert_dup, which is used by SSL_new, now copies DH keys in addition
16665   to parameters -- in previous versions (since OpenSSL 0.9.3) the
16666   'default key' from SSL_CTX_set_tmp_dh would always be lost, meaning
16667   you effectively got SSL_OP_SINGLE_DH_USE when using this macro.
16668
16669   *Bodo Moeller*
16670
16671 * New s_client option -ign_eof: EOF at stdin is ignored, and
16672   'Q' and 'R' lose their special meanings (quit/renegotiate).
16673   This is part of what -quiet does; unlike -quiet, -ign_eof
16674   does not suppress any output.
16675
16676   *Richard Levitte*
16677
16678 * Add compatibility options to the purpose and trust code. The
16679   purpose X509_PURPOSE_ANY is "any purpose" which automatically
16680   accepts a certificate or CA, this was the previous behaviour,
16681   with all the associated security issues.
16682
16683   X509_TRUST_COMPAT is the old trust behaviour: only and
16684   automatically trust self signed roots in certificate store. A
16685   new trust setting X509_TRUST_DEFAULT is used to specify that
16686   a purpose has no associated trust setting and it should instead
16687   use the value in the default purpose.
16688
16689   *Steve Henson*
16690
16691 * Fix the PKCS#8 DSA private key code so it decodes keys again
16692   and fix a memory leak.
16693
16694   *Steve Henson*
16695
16696 * In util/mkerr.pl (which implements 'make errors'), preserve
16697   reason strings from the previous version of the .c file, as
16698   the default to have only downcase letters (and digits) in
16699   automatically generated reasons codes is not always appropriate.
16700
16701   *Bodo Moeller*
16702
16703 * In ERR_load_ERR_strings(), build an ERR_LIB_SYS error reason table
16704   using strerror.  Previously, ERR_reason_error_string() returned
16705   library names as reason strings for SYSerr; but SYSerr is a special
16706   case where small numbers are errno values, not library numbers.
16707
16708   *Bodo Moeller*
16709
16710 * Add '-dsaparam' option to 'openssl dhparam' application.  This
16711   converts DSA parameters into DH parameters. (When creating parameters,
16712   DSA_generate_parameters is used.)
16713
16714   *Bodo Moeller*
16715
16716 * Include 'length' (recommended exponent length) in C code generated
16717   by 'openssl dhparam -C'.
16718
16719   *Bodo Moeller*
16720
16721 * The second argument to set_label in perlasm was already being used
16722   so couldn't be used as a "file scope" flag. Moved to third argument
16723   which was free.
16724
16725   *Steve Henson*
16726
16727 * In PEM_ASN1_write_bio and some other functions, use RAND_pseudo_bytes
16728   instead of RAND_bytes for encryption IVs and salts.
16729
16730   *Bodo Moeller*
16731
16732 * Include RAND_status() into RAND_METHOD instead of implementing
16733   it only for md_rand.c  Otherwise replacing the PRNG by calling
16734   RAND_set_rand_method would be impossible.
16735
16736   *Bodo Moeller*
16737
16738 * Don't let DSA_generate_key() enter an infinite loop if the random
16739   number generation fails.
16740
16741   *Bodo Moeller*
16742
16743 * New 'rand' application for creating pseudo-random output.
16744
16745   *Bodo Moeller*
16746
16747 * Added configuration support for Linux/IA64
16748
16749   *Rolf Haberrecker <rolf@suse.de>*
16750
16751 * Assembler module support for Mingw32.
16752
16753   *Ulf Möller*
16754
16755 * Shared library support for HPUX (in shlib/).
16756
16757   *Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE> and Anonymous*
16758
16759 * Shared library support for Solaris gcc.
16760
16761   *Lutz Behnke <behnke@trustcenter.de>*
16762
16763### Changes between 0.9.4 and 0.9.5  [28 Feb 2000]
16764
16765 * PKCS7_encrypt() was adding text MIME headers twice because they
16766   were added manually and by SMIME_crlf_copy().
16767
16768   *Steve Henson*
16769
16770 * In bntest.c don't call BN_rand with zero bits argument.
16771
16772   *Steve Henson, pointed out by Andrew W. Gray <agray@iconsinc.com>*
16773
16774 * BN_mul bugfix: In bn_mul_part_recursion() only the a>a[n] && b>b[n]
16775   case was implemented. This caused BN_div_recp() to fail occasionally.
16776
16777   *Ulf Möller*
16778
16779 * Add an optional second argument to the set_label() in the perl
16780   assembly language builder. If this argument exists and is set
16781   to 1 it signals that the assembler should use a symbol whose
16782   scope is the entire file, not just the current function. This
16783   is needed with MASM which uses the format label:: for this scope.
16784
16785   *Steve Henson, pointed out by Peter Runestig <peter@runestig.com>*
16786
16787 * Change the ASN1 types so they are typedefs by default. Before
16788   almost all types were #define'd to ASN1_STRING which was causing
16789   STACK_OF() problems: you couldn't declare STACK_OF(ASN1_UTF8STRING)
16790   for example.
16791
16792   *Steve Henson*
16793
16794 * Change names of new functions to the new get1/get0 naming
16795   convention: After 'get1', the caller owns a reference count
16796   and has to call `..._free`; 'get0' returns a pointer to some
16797   data structure without incrementing reference counters.
16798   (Some of the existing 'get' functions increment a reference
16799   counter, some don't.)
16800   Similarly, 'set1' and 'add1' functions increase reference
16801   counters or duplicate objects.
16802
16803   *Steve Henson*
16804
16805 * Allow for the possibility of temp RSA key generation failure:
16806   the code used to assume it always worked and crashed on failure.
16807
16808   *Steve Henson*
16809
16810 * Fix potential buffer overrun problem in BIO_printf().
16811   *Ulf Möller, using public domain code by Patrick Powell; problem
16812   pointed out by David Sacerdote <das33@cornell.edu>*
16813
16814 * Support EGD <http://www.lothar.com/tech/crypto/>.  New functions
16815   RAND_egd() and RAND_status().  In the command line application,
16816   the EGD socket can be specified like a seed file using RANDFILE
16817   or -rand.
16818
16819   *Ulf Möller*
16820
16821 * Allow the string CERTIFICATE to be tolerated in PKCS#7 structures.
16822   Some CAs (e.g. Verisign) distribute certificates in this form.
16823
16824   *Steve Henson*
16825
16826 * Remove the SSL_ALLOW_ADH compile option and set the default cipher
16827   list to exclude them. This means that no special compilation option
16828   is needed to use anonymous DH: it just needs to be included in the
16829   cipher list.
16830
16831   *Steve Henson*
16832
16833 * Change the EVP_MD_CTX_type macro so its meaning consistent with
16834   EVP_MD_type. The old functionality is available in a new macro called
16835   EVP_MD_md(). Change code that uses it and update docs.
16836
16837   *Steve Henson*
16838
16839 * `..._ctrl` functions now have corresponding `..._callback_ctrl` functions
16840   where the `void *` argument is replaced by a function pointer argument.
16841   Previously `void *` was abused to point to functions, which works on
16842   many platforms, but is not correct.  As these functions are usually
16843   called by macros defined in OpenSSL header files, most source code
16844   should work without changes.
16845
16846   *Richard Levitte*
16847
16848 * `<openssl/opensslconf.h>` (which is created by Configure) now contains
16849   sections with information on -D... compiler switches used for
16850   compiling the library so that applications can see them.  To enable
16851   one of these sections, a pre-processor symbol `OPENSSL_..._DEFINES`
16852   must be defined.  E.g.,
16853           #define OPENSSL_ALGORITHM_DEFINES
16854           #include <openssl/opensslconf.h>
16855   defines all pertinent `NO_<algo>` symbols, such as NO_IDEA, NO_RSA, etc.
16856
16857   *Richard Levitte, Ulf and Bodo Möller*
16858
16859 * Bugfix: Tolerate fragmentation and interleaving in the SSL 3/TLS
16860   record layer.
16861
16862   *Bodo Moeller*
16863
16864 * Change the 'other' type in certificate aux info to a STACK_OF
16865   X509_ALGOR. Although not an AlgorithmIdentifier as such it has
16866   the required ASN1 format: arbitrary types determined by an OID.
16867
16868   *Steve Henson*
16869
16870 * Add some PEM_write_X509_REQ_NEW() functions and a command line
16871   argument to 'req'. This is not because the function is newer or
16872   better than others it just uses the work 'NEW' in the certificate
16873   request header lines. Some software needs this.
16874
16875   *Steve Henson*
16876
16877 * Reorganise password command line arguments: now passwords can be
16878   obtained from various sources. Delete the PEM_cb function and make
16879   it the default behaviour: i.e. if the callback is NULL and the
16880   usrdata argument is not NULL interpret it as a null terminated pass
16881   phrase. If usrdata and the callback are NULL then the pass phrase
16882   is prompted for as usual.
16883
16884   *Steve Henson*
16885
16886 * Add support for the Compaq Atalla crypto accelerator. If it is installed,
16887   the support is automatically enabled. The resulting binaries will
16888   autodetect the card and use it if present.
16889
16890   *Ben Laurie and Compaq Inc.*
16891
16892 * Work around for Netscape hang bug. This sends certificate request
16893   and server done in one record. Since this is perfectly legal in the
16894   SSL/TLS protocol it isn't a "bug" option and is on by default. See
16895   the bugs/SSLv3 entry for more info.
16896
16897   *Steve Henson*
16898
16899 * HP-UX tune-up: new unified configs, HP C compiler bug workaround.
16900
16901   *Andy Polyakov*
16902
16903 * Add -rand argument to smime and pkcs12 applications and read/write
16904   of seed file.
16905
16906   *Steve Henson*
16907
16908 * New 'passwd' tool for crypt(3) and apr1 password hashes.
16909
16910   *Bodo Moeller*
16911
16912 * Add command line password options to the remaining applications.
16913
16914   *Steve Henson*
16915
16916 * Bug fix for BN_div_recp() for numerators with an even number of
16917   bits.
16918
16919   *Ulf Möller*
16920
16921 * More tests in bntest.c, and changed test_bn output.
16922
16923   *Ulf Möller*
16924
16925 * ./config recognizes MacOS X now.
16926
16927   *Andy Polyakov*
16928
16929 * Bug fix for BN_div() when the first words of num and divisor are
16930   equal (it gave wrong results if `(rem=(n1-q*d0)&BN_MASK2) < d0)`.
16931
16932   *Ulf Möller*
16933
16934 * Add support for various broken PKCS#8 formats, and command line
16935   options to produce them.
16936
16937   *Steve Henson*
16938
16939 * New functions BN_CTX_start(), BN_CTX_get() and BT_CTX_end() to
16940   get temporary BIGNUMs from a BN_CTX.
16941
16942   *Ulf Möller*
16943
16944 * Correct return values in BN_mod_exp_mont() and BN_mod_exp2_mont()
16945   for p == 0.
16946
16947   *Ulf Möller*
16948
16949 * Change the `SSLeay_add_all_*()` functions to `OpenSSL_add_all_*()` and
16950   include a #define from the old name to the new. The original intent
16951   was that statically linked binaries could for example just call
16952   SSLeay_add_all_ciphers() to just add ciphers to the table and not
16953   link with digests. This never worked because SSLeay_add_all_digests()
16954   and SSLeay_add_all_ciphers() were in the same source file so calling
16955   one would link with the other. They are now in separate source files.
16956
16957   *Steve Henson*
16958
16959 * Add a new -notext option to 'ca' and a -pubkey option to 'spkac'.
16960
16961   *Steve Henson*
16962
16963 * Use a less unusual form of the Miller-Rabin primality test (it used
16964   a binary algorithm for exponentiation integrated into the Miller-Rabin
16965   loop, our standard modexp algorithms are faster).
16966
16967   *Bodo Moeller*
16968
16969 * Support for the EBCDIC character set completed.
16970
16971   *Martin Kraemer <Martin.Kraemer@Mch.SNI.De>*
16972
16973 * Source code cleanups: use const where appropriate, eliminate casts,
16974   use `void *` instead of `char *` in lhash.
16975
16976   *Ulf Möller*
16977
16978 * Bugfix: ssl3_send_server_key_exchange was not restartable
16979   (the state was not changed to SSL3_ST_SW_KEY_EXCH_B, and because of
16980   this the server could overwrite ephemeral keys that the client
16981   has already seen).
16982
16983   *Bodo Moeller*
16984
16985 * Turn DSA_is_prime into a macro that calls BN_is_prime,
16986   using 50 iterations of the Rabin-Miller test.
16987
16988   DSA_generate_parameters now uses BN_is_prime_fasttest (with 50
16989   iterations of the Rabin-Miller test as required by the appendix
16990   to FIPS PUB 186[-1]) instead of DSA_is_prime.
16991   As BN_is_prime_fasttest includes trial division, DSA parameter
16992   generation becomes much faster.
16993
16994   This implies a change for the callback functions in DSA_is_prime
16995   and DSA_generate_parameters: The callback function is called once
16996   for each positive witness in the Rabin-Miller test, not just
16997   occasionally in the inner loop; and the parameters to the
16998   callback function now provide an iteration count for the outer
16999   loop rather than for the current invocation of the inner loop.
17000   DSA_generate_parameters additionally can call the callback
17001   function with an 'iteration count' of -1, meaning that a
17002   candidate has passed the trial division test (when q is generated
17003   from an application-provided seed, trial division is skipped).
17004
17005   *Bodo Moeller*
17006
17007 * New function BN_is_prime_fasttest that optionally does trial
17008   division before starting the Rabin-Miller test and has
17009   an additional BN_CTX * argument (whereas BN_is_prime always
17010   has to allocate at least one BN_CTX).
17011   'callback(1, -1, cb_arg)' is called when a number has passed the
17012   trial division stage.
17013
17014   *Bodo Moeller*
17015
17016 * Fix for bug in CRL encoding. The validity dates weren't being handled
17017   as ASN1_TIME.
17018
17019   *Steve Henson*
17020
17021 * New -pkcs12 option to CA.pl script to write out a PKCS#12 file.
17022
17023   *Steve Henson*
17024
17025 * New function BN_pseudo_rand().
17026
17027   *Ulf Möller*
17028
17029 * Clean up BN_mod_mul_montgomery(): replace the broken (and unreadable)
17030   bignum version of BN_from_montgomery() with the working code from
17031   SSLeay 0.9.0 (the word based version is faster anyway), and clean up
17032   the comments.
17033
17034   *Ulf Möller*
17035
17036 * Avoid a race condition in s2_clnt.c (function get_server_hello) that
17037   made it impossible to use the same SSL_SESSION data structure in
17038   SSL2 clients in multiple threads.
17039
17040   *Bodo Moeller*
17041
17042 * The return value of RAND_load_file() no longer counts bytes obtained
17043   by stat().  RAND_load_file(..., -1) is new and uses the complete file
17044   to seed the PRNG (previously an explicit byte count was required).
17045
17046   *Ulf Möller, Bodo Möller*
17047
17048 * Clean up CRYPTO_EX_DATA functions, some of these didn't have prototypes
17049   used `char *` instead of `void *` and had casts all over the place.
17050
17051   *Steve Henson*
17052
17053 * Make BN_generate_prime() return NULL on error if ret!=NULL.
17054
17055   *Ulf Möller*
17056
17057 * Retain source code compatibility for BN_prime_checks macro:
17058   BN_is_prime(..., BN_prime_checks, ...) now uses
17059   BN_prime_checks_for_size to determine the appropriate number of
17060   Rabin-Miller iterations.
17061
17062   *Ulf Möller*
17063
17064 * Diffie-Hellman uses "safe" primes: DH_check() return code renamed to
17065   DH_CHECK_P_NOT_SAFE_PRIME.
17066   (Check if this is true? OpenPGP calls them "strong".)
17067
17068   *Ulf Möller*
17069
17070 * Merge the functionality of "dh" and "gendh" programs into a new program
17071   "dhparam". The old programs are retained for now but will handle DH keys
17072   (instead of parameters) in future.
17073
17074   *Steve Henson*
17075
17076 * Make the ciphers, s_server and s_client programs check the return values
17077   when a new cipher list is set.
17078
17079   *Steve Henson*
17080
17081 * Enhance the SSL/TLS cipher mechanism to correctly handle the TLS 56bit
17082   ciphers. Before when the 56bit ciphers were enabled the sorting was
17083   wrong.
17084
17085   The syntax for the cipher sorting has been extended to support sorting by
17086   cipher-strength (using the strength_bits hard coded in the tables).
17087   The new command is `@STRENGTH` (see also `doc/apps/ciphers.pod`).
17088
17089   Fix a bug in the cipher-command parser: when supplying a cipher command
17090   string with an "undefined" symbol (neither command nor alphanumeric
17091   *A-Za-z0-9*, ssl_set_cipher_list used to hang in an endless loop. Now
17092   an error is flagged.
17093
17094   Due to the strength-sorting extension, the code of the
17095   ssl_create_cipher_list() function was completely rearranged. I hope that
17096   the readability was also increased :-)
17097
17098   *Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>*
17099
17100 * Minor change to 'x509' utility. The -CAcreateserial option now uses 1
17101   for the first serial number and places 2 in the serial number file. This
17102   avoids problems when the root CA is created with serial number zero and
17103   the first user certificate has the same issuer name and serial number
17104   as the root CA.
17105
17106   *Steve Henson*
17107
17108 * Fixes to X509_ATTRIBUTE utilities, change the 'req' program so it uses
17109   the new code. Add documentation for this stuff.
17110
17111   *Steve Henson*
17112
17113 * Changes to X509_ATTRIBUTE utilities. These have been renamed from
17114   `X509_*()` to `X509at_*()` on the grounds that they don't handle X509
17115   structures and behave in an analogous way to the X509v3 functions:
17116   they shouldn't be called directly but wrapper functions should be used
17117   instead.
17118
17119   So we also now have some wrapper functions that call the X509at functions
17120   when passed certificate requests. (TO DO: similar things can be done with
17121   PKCS#7 signed and unsigned attributes, PKCS#12 attributes and a few other
17122   things. Some of these need some d2i or i2d and print functionality
17123   because they handle more complex structures.)
17124
17125   *Steve Henson*
17126
17127 * Add missing #ifndefs that caused missing symbols when building libssl
17128   as a shared library without RSA.  Use #ifndef NO_SSL2 instead of
17129   NO_RSA in `ssl/s2*.c`.
17130
17131   *Kris Kennaway <kris@hub.freebsd.org>, modified by Ulf Möller*
17132
17133 * Precautions against using the PRNG uninitialized: RAND_bytes() now
17134   has a return value which indicates the quality of the random data
17135   (1 = ok, 0 = not seeded).  Also an error is recorded on the thread's
17136   error queue. New function RAND_pseudo_bytes() generates output that is
17137   guaranteed to be unique but not unpredictable. RAND_add is like
17138   RAND_seed, but takes an extra argument for an entropy estimate
17139   (RAND_seed always assumes full entropy).
17140
17141   *Ulf Möller*
17142
17143 * Do more iterations of Rabin-Miller probable prime test (specifically,
17144   3 for 1024-bit primes, 6 for 512-bit primes, 12 for 256-bit primes
17145   instead of only 2 for all lengths; see BN_prime_checks_for_size definition
17146   in crypto/bn/bn_prime.c for the complete table).  This guarantees a
17147   false-positive rate of at most 2^-80 for random input.
17148
17149   *Bodo Moeller*
17150
17151 * Rewrite ssl3_read_n (ssl/s3_pkt.c) avoiding a couple of bugs.
17152
17153   *Bodo Moeller*
17154
17155 * New function X509_CTX_rget_chain() (renamed to X509_CTX_get1_chain
17156   in the 0.9.5 release), this returns the chain
17157   from an X509_CTX structure with a dup of the stack and all
17158   the X509 reference counts upped: so the stack will exist
17159   after X509_CTX_cleanup() has been called. Modify pkcs12.c
17160   to use this.
17161
17162   Also make SSL_SESSION_print() print out the verify return
17163   code.
17164
17165   *Steve Henson*
17166
17167 * Add manpage for the pkcs12 command. Also change the default
17168   behaviour so MAC iteration counts are used unless the new
17169   -nomaciter option is used. This improves file security and
17170   only older versions of MSIE (4.0 for example) need it.
17171
17172   *Steve Henson*
17173
17174 * Honor the no-xxx Configure options when creating .DEF files.
17175
17176   *Ulf Möller*
17177
17178 * Add PKCS#10 attributes to field table: challengePassword,
17179   unstructuredName and unstructuredAddress. These are taken from
17180   draft PKCS#9 v2.0 but are compatible with v1.2 provided no
17181   international characters are used.
17182
17183   More changes to X509_ATTRIBUTE code: allow the setting of types
17184   based on strings. Remove the 'loc' parameter when adding
17185   attributes because these will be a SET OF encoding which is sorted
17186   in ASN1 order.
17187
17188   *Steve Henson*
17189
17190 * Initial changes to the 'req' utility to allow request generation
17191   automation. This will allow an application to just generate a template
17192   file containing all the field values and have req construct the
17193   request.
17194
17195   Initial support for X509_ATTRIBUTE handling. Stacks of these are
17196   used all over the place including certificate requests and PKCS#7
17197   structures. They are currently handled manually where necessary with
17198   some primitive wrappers for PKCS#7. The new functions behave in a
17199   manner analogous to the X509 extension functions: they allow
17200   attributes to be looked up by NID and added.
17201
17202   Later something similar to the X509V3 code would be desirable to
17203   automatically handle the encoding, decoding and printing of the
17204   more complex types. The string types like challengePassword can
17205   be handled by the string table functions.
17206
17207   Also modified the multi byte string table handling. Now there is
17208   a 'global mask' which masks out certain types. The table itself
17209   can use the flag STABLE_NO_MASK to ignore the mask setting: this
17210   is useful when for example there is only one permissible type
17211   (as in countryName) and using the mask might result in no valid
17212   types at all.
17213
17214   *Steve Henson*
17215
17216 * Clean up 'Finished' handling, and add functions SSL_get_finished and
17217   SSL_get_peer_finished to allow applications to obtain the latest
17218   Finished messages sent to the peer or expected from the peer,
17219   respectively.  (SSL_get_peer_finished is usually the Finished message
17220   actually received from the peer, otherwise the protocol will be aborted.)
17221
17222   As the Finished message are message digests of the complete handshake
17223   (with a total of 192 bits for TLS 1.0 and more for SSL 3.0), they can
17224   be used for external authentication procedures when the authentication
17225   provided by SSL/TLS is not desired or is not enough.
17226
17227   *Bodo Moeller*
17228
17229 * Enhanced support for Alpha Linux is added. Now ./config checks if
17230   the host supports BWX extension and if Compaq C is present on the
17231   $PATH. Just exploiting of the BWX extension results in 20-30%
17232   performance kick for some algorithms, e.g. DES and RC4 to mention
17233   a couple. Compaq C in turn generates ~20% faster code for MD5 and
17234   SHA1.
17235
17236   *Andy Polyakov*
17237
17238 * Add support for MS "fast SGC". This is arguably a violation of the
17239   SSL3/TLS protocol. Netscape SGC does two handshakes: the first with
17240   weak crypto and after checking the certificate is SGC a second one
17241   with strong crypto. MS SGC stops the first handshake after receiving
17242   the server certificate message and sends a second client hello. Since
17243   a server will typically do all the time consuming operations before
17244   expecting any further messages from the client (server key exchange
17245   is the most expensive) there is little difference between the two.
17246
17247   To get OpenSSL to support MS SGC we have to permit a second client
17248   hello message after we have sent server done. In addition we have to
17249   reset the MAC if we do get this second client hello.
17250
17251   *Steve Henson*
17252
17253 * Add a function 'd2i_AutoPrivateKey()' this will automatically decide
17254   if a DER encoded private key is RSA or DSA traditional format. Changed
17255   d2i_PrivateKey_bio() to use it. This is only needed for the "traditional"
17256   format DER encoded private key. Newer code should use PKCS#8 format which
17257   has the key type encoded in the ASN1 structure. Added DER private key
17258   support to pkcs8 application.
17259
17260   *Steve Henson*
17261
17262 * SSL 3/TLS 1 servers now don't request certificates when an anonymous
17263   ciphersuites has been selected (as required by the SSL 3/TLS 1
17264   specifications).  Exception: When SSL_VERIFY_FAIL_IF_NO_PEER_CERT
17265   is set, we interpret this as a request to violate the specification
17266   (the worst that can happen is a handshake failure, and 'correct'
17267   behaviour would result in a handshake failure anyway).
17268
17269   *Bodo Moeller*
17270
17271 * In SSL_CTX_add_session, take into account that there might be multiple
17272   SSL_SESSION structures with the same session ID (e.g. when two threads
17273   concurrently obtain them from an external cache).
17274   The internal cache can handle only one SSL_SESSION with a given ID,
17275   so if there's a conflict, we now throw out the old one to achieve
17276   consistency.
17277
17278   *Bodo Moeller*
17279
17280 * Add OIDs for idea and blowfish in CBC mode. This will allow both
17281   to be used in PKCS#5 v2.0 and S/MIME.  Also add checking to
17282   some routines that use cipher OIDs: some ciphers do not have OIDs
17283   defined and so they cannot be used for S/MIME and PKCS#5 v2.0 for
17284   example.
17285
17286   *Steve Henson*
17287
17288 * Simplify the trust setting structure and code. Now we just have
17289   two sequences of OIDs for trusted and rejected settings. These will
17290   typically have values the same as the extended key usage extension
17291   and any application specific purposes.
17292
17293   The trust checking code now has a default behaviour: it will just
17294   check for an object with the same NID as the passed id. Functions can
17295   be provided to override either the default behaviour or the behaviour
17296   for a given id. SSL client, server and email already have functions
17297   in place for compatibility: they check the NID and also return "trusted"
17298   if the certificate is self signed.
17299
17300   *Steve Henson*
17301
17302 * Add d2i,i2d bio/fp functions for PrivateKey: these convert the
17303   traditional format into an EVP_PKEY structure.
17304
17305   *Steve Henson*
17306
17307 * Add a password callback function PEM_cb() which either prompts for
17308   a password if usr_data is NULL or otherwise assumes it is a null
17309   terminated password. Allow passwords to be passed on command line
17310   environment or config files in a few more utilities.
17311
17312   *Steve Henson*
17313
17314 * Add a bunch of DER and PEM functions to handle PKCS#8 format private
17315   keys. Add some short names for PKCS#8 PBE algorithms and allow them
17316   to be specified on the command line for the pkcs8 and pkcs12 utilities.
17317   Update documentation.
17318
17319   *Steve Henson*
17320
17321 * Support for ASN1 "NULL" type. This could be handled before by using
17322   ASN1_TYPE but there wasn't any function that would try to read a NULL
17323   and produce an error if it couldn't. For compatibility we also have
17324   ASN1_NULL_new() and ASN1_NULL_free() functions but these are faked and
17325   don't allocate anything because they don't need to.
17326
17327   *Steve Henson*
17328
17329 * Initial support for MacOS is now provided. Examine INSTALL.MacOS
17330   for details.
17331
17332   *Andy Polyakov, Roy Woods <roy@centicsystems.ca>*
17333
17334 * Rebuild of the memory allocation routines used by OpenSSL code and
17335   possibly others as well.  The purpose is to make an interface that
17336   provide hooks so anyone can build a separate set of allocation and
17337   deallocation routines to be used by OpenSSL, for example memory
17338   pool implementations, or something else, which was previously hard
17339   since Malloc(), Realloc() and Free() were defined as macros having
17340   the values malloc, realloc and free, respectively (except for Win32
17341   compilations).  The same is provided for memory debugging code.
17342   OpenSSL already comes with functionality to find memory leaks, but
17343   this gives people a chance to debug other memory problems.
17344
17345   With these changes, a new set of functions and macros have appeared:
17346
17347     CRYPTO_set_mem_debug_functions()         [F]
17348     CRYPTO_get_mem_debug_functions()         [F]
17349     CRYPTO_dbg_set_options()                 [F]
17350     CRYPTO_dbg_get_options()                 [F]
17351     CRYPTO_malloc_debug_init()               [M]
17352
17353   The memory debug functions are NULL by default, unless the library
17354   is compiled with CRYPTO_MDEBUG or friends is defined.  If someone
17355   wants to debug memory anyway, CRYPTO_malloc_debug_init() (which
17356   gives the standard debugging functions that come with OpenSSL) or
17357   CRYPTO_set_mem_debug_functions() (tells OpenSSL to use functions
17358   provided by the library user) must be used.  When the standard
17359   debugging functions are used, CRYPTO_dbg_set_options can be used to
17360   request additional information:
17361   CRYPTO_dbg_set_options(V_CYRPTO_MDEBUG_xxx) corresponds to setting
17362   the CRYPTO_MDEBUG_xxx macro when compiling the library.
17363
17364   Also, things like CRYPTO_set_mem_functions will always give the
17365   expected result (the new set of functions is used for allocation
17366   and deallocation) at all times, regardless of platform and compiler
17367   options.
17368
17369   To finish it up, some functions that were never use in any other
17370   way than through macros have a new API and new semantic:
17371
17372     CRYPTO_dbg_malloc()
17373     CRYPTO_dbg_realloc()
17374     CRYPTO_dbg_free()
17375
17376   All macros of value have retained their old syntax.
17377
17378   *Richard Levitte and Bodo Moeller*
17379
17380 * Some S/MIME fixes. The OID for SMIMECapabilities was wrong, the
17381   ordering of SMIMECapabilities wasn't in "strength order" and there
17382   was a missing NULL in the AlgorithmIdentifier for the SHA1 signature
17383   algorithm.
17384
17385   *Steve Henson*
17386
17387 * Some ASN1 types with illegal zero length encoding (INTEGER,
17388   ENUMERATED and OBJECT IDENTIFIER) choked the ASN1 routines.
17389
17390   *Frans Heymans <fheymans@isaserver.be>, modified by Steve Henson*
17391
17392 * Merge in my S/MIME library for OpenSSL. This provides a simple
17393   S/MIME API on top of the PKCS#7 code, a MIME parser (with enough
17394   functionality to handle multipart/signed properly) and a utility
17395   called 'smime' to call all this stuff. This is based on code I
17396   originally wrote for Celo who have kindly allowed it to be
17397   included in OpenSSL.
17398
17399   *Steve Henson*
17400
17401 * Add variants des_set_key_checked and des_set_key_unchecked of
17402   des_set_key (aka des_key_sched).  Global variable des_check_key
17403   decides which of these is called by des_set_key; this way
17404   des_check_key behaves as it always did, but applications and
17405   the library itself, which was buggy for des_check_key == 1,
17406   have a cleaner way to pick the version they need.
17407
17408   *Bodo Moeller*
17409
17410 * New function PKCS12_newpass() which changes the password of a
17411   PKCS12 structure.
17412
17413   *Steve Henson*
17414
17415 * Modify X509_TRUST and X509_PURPOSE so it also uses a static and
17416   dynamic mix. In both cases the ids can be used as an index into the
17417   table. Also modified the X509_TRUST_add() and X509_PURPOSE_add()
17418   functions so they accept a list of the field values and the
17419   application doesn't need to directly manipulate the X509_TRUST
17420   structure.
17421
17422   *Steve Henson*
17423
17424 * Modify the ASN1_STRING_TABLE stuff so it also uses bsearch and doesn't
17425   need initialising.
17426
17427   *Steve Henson*
17428
17429 * Modify the way the V3 extension code looks up extensions. This now
17430   works in a similar way to the object code: we have some "standard"
17431   extensions in a static table which is searched with OBJ_bsearch()
17432   and the application can add dynamic ones if needed. The file
17433   crypto/x509v3/ext_dat.h now has the info: this file needs to be
17434   updated whenever a new extension is added to the core code and kept
17435   in ext_nid order. There is a simple program 'tabtest.c' which checks
17436   this. New extensions are not added too often so this file can readily
17437   be maintained manually.
17438
17439   There are two big advantages in doing things this way. The extensions
17440   can be looked up immediately and no longer need to be "added" using
17441   X509V3_add_standard_extensions(): this function now does nothing.
17442   Side note: I get *lots* of email saying the extension code doesn't
17443   work because people forget to call this function.
17444   Also no dynamic allocation is done unless new extensions are added:
17445   so if we don't add custom extensions there is no need to call
17446   X509V3_EXT_cleanup().
17447
17448   *Steve Henson*
17449
17450 * Modify enc utility's salting as follows: make salting the default. Add a
17451   magic header, so unsalted files fail gracefully instead of just decrypting
17452   to garbage. This is because not salting is a big security hole, so people
17453   should be discouraged from doing it.
17454
17455   *Ben Laurie*
17456
17457 * Fixes and enhancements to the 'x509' utility. It allowed a message
17458   digest to be passed on the command line but it only used this
17459   parameter when signing a certificate. Modified so all relevant
17460   operations are affected by the digest parameter including the
17461   -fingerprint and -x509toreq options. Also -x509toreq choked if a
17462   DSA key was used because it didn't fix the digest.
17463
17464   *Steve Henson*
17465
17466 * Initial certificate chain verify code. Currently tests the untrusted
17467   certificates for consistency with the verify purpose (which is set
17468   when the X509_STORE_CTX structure is set up) and checks the pathlength.
17469
17470   There is a NO_CHAIN_VERIFY compilation option to keep the old behaviour:
17471   this is because it will reject chains with invalid extensions whereas
17472   every previous version of OpenSSL and SSLeay made no checks at all.
17473
17474   Trust code: checks the root CA for the relevant trust settings. Trust
17475   settings have an initial value consistent with the verify purpose: e.g.
17476   if the verify purpose is for SSL client use it expects the CA to be
17477   trusted for SSL client use. However the default value can be changed to
17478   permit custom trust settings: one example of this would be to only trust
17479   certificates from a specific "secure" set of CAs.
17480
17481   Also added X509_STORE_CTX_new() and X509_STORE_CTX_free() functions
17482   which should be used for version portability: especially since the
17483   verify structure is likely to change more often now.
17484
17485   SSL integration. Add purpose and trust to SSL_CTX and SSL and functions
17486   to set them. If not set then assume SSL clients will verify SSL servers
17487   and vice versa.
17488
17489   Two new options to the verify program: -untrusted allows a set of
17490   untrusted certificates to be passed in and -purpose which sets the
17491   intended purpose of the certificate. If a purpose is set then the
17492   new chain verify code is used to check extension consistency.
17493
17494   *Steve Henson*
17495
17496 * Support for the authority information access extension.
17497
17498   *Steve Henson*
17499
17500 * Modify RSA and DSA PEM read routines to transparently handle
17501   PKCS#8 format private keys. New *_PUBKEY_* functions that handle
17502   public keys in a format compatible with certificate
17503   SubjectPublicKeyInfo structures. Unfortunately there were already
17504   functions called *_PublicKey_* which used various odd formats so
17505   these are retained for compatibility: however the DSA variants were
17506   never in a public release so they have been deleted. Changed dsa/rsa
17507   utilities to handle the new format: note no releases ever handled public
17508   keys so we should be OK.
17509
17510   The primary motivation for this change is to avoid the same fiasco
17511   that dogs private keys: there are several incompatible private key
17512   formats some of which are standard and some OpenSSL specific and
17513   require various evil hacks to allow partial transparent handling and
17514   even then it doesn't work with DER formats. Given the option anything
17515   other than PKCS#8 should be dumped: but the other formats have to
17516   stay in the name of compatibility.
17517
17518   With public keys and the benefit of hindsight one standard format
17519   is used which works with EVP_PKEY, RSA or DSA structures: though
17520   it clearly returns an error if you try to read the wrong kind of key.
17521
17522   Added a -pubkey option to the 'x509' utility to output the public key.
17523   Also rename the `EVP_PKEY_get_*()` to `EVP_PKEY_rget_*()`
17524   (renamed to `EVP_PKEY_get1_*()` in the OpenSSL 0.9.5 release) and add
17525   `EVP_PKEY_rset_*()` functions (renamed to `EVP_PKEY_set1_*()`)
17526   that do the same as the `EVP_PKEY_assign_*()` except they up the
17527   reference count of the added key (they don't "swallow" the
17528   supplied key).
17529
17530   *Steve Henson*
17531
17532 * Fixes to crypto/x509/by_file.c the code to read in certificates and
17533   CRLs would fail if the file contained no certificates or no CRLs:
17534   added a new function to read in both types and return the number
17535   read: this means that if none are read it will be an error. The
17536   DER versions of the certificate and CRL reader would always fail
17537   because it isn't possible to mix certificates and CRLs in DER format
17538   without choking one or the other routine. Changed this to just read
17539   a certificate: this is the best we can do. Also modified the code
17540   in `apps/verify.c` to take notice of return codes: it was previously
17541   attempting to read in certificates from NULL pointers and ignoring
17542   any errors: this is one reason why the cert and CRL reader seemed
17543   to work. It doesn't check return codes from the default certificate
17544   routines: these may well fail if the certificates aren't installed.
17545
17546   *Steve Henson*
17547
17548 * Code to support otherName option in GeneralName.
17549
17550   *Steve Henson*
17551
17552 * First update to verify code. Change the verify utility
17553   so it warns if it is passed a self signed certificate:
17554   for consistency with the normal behaviour. X509_verify
17555   has been modified to it will now verify a self signed
17556   certificate if *exactly* the same certificate appears
17557   in the store: it was previously impossible to trust a
17558   single self signed certificate. This means that:
17559   openssl verify ss.pem
17560   now gives a warning about a self signed certificate but
17561   openssl verify -CAfile ss.pem ss.pem
17562   is OK.
17563
17564   *Steve Henson*
17565
17566 * For servers, store verify_result in SSL_SESSION data structure
17567   (and add it to external session representation).
17568   This is needed when client certificate verifications fails,
17569   but an application-provided verification callback (set by
17570   SSL_CTX_set_cert_verify_callback) allows accepting the session
17571   anyway (i.e. leaves x509_store_ctx->error != X509_V_OK
17572   but returns 1): When the session is reused, we have to set
17573   ssl->verify_result to the appropriate error code to avoid
17574   security holes.
17575
17576   *Bodo Moeller, problem pointed out by Lutz Jaenicke*
17577
17578 * Fix a bug in the new PKCS#7 code: it didn't consider the
17579   case in PKCS7_dataInit() where the signed PKCS7 structure
17580   didn't contain any existing data because it was being created.
17581
17582   *Po-Cheng Chen <pocheng@nst.com.tw>, slightly modified by Steve Henson*
17583
17584 * Add a salt to the key derivation routines in enc.c. This
17585   forms the first 8 bytes of the encrypted file. Also add a
17586   -S option to allow a salt to be input on the command line.
17587
17588   *Steve Henson*
17589
17590 * New function X509_cmp(). Oddly enough there wasn't a function
17591   to compare two certificates. We do this by working out the SHA1
17592   hash and comparing that. X509_cmp() will be needed by the trust
17593   code.
17594
17595   *Steve Henson*
17596
17597 * SSL_get1_session() is like SSL_get_session(), but increments
17598   the reference count in the SSL_SESSION returned.
17599
17600   *Geoff Thorpe <geoff@eu.c2.net>*
17601
17602 * Fix for 'req': it was adding a null to request attributes.
17603   Also change the X509_LOOKUP and X509_INFO code to handle
17604   certificate auxiliary information.
17605
17606   *Steve Henson*
17607
17608 * Add support for 40 and 64 bit RC2 and RC4 algorithms: document
17609   the 'enc' command.
17610
17611   *Steve Henson*
17612
17613 * Add the possibility to add extra information to the memory leak
17614   detecting output, to form tracebacks, showing from where each
17615   allocation was originated: CRYPTO_push_info("constant string") adds
17616   the string plus current file name and line number to a per-thread
17617   stack, CRYPTO_pop_info() does the obvious, CRYPTO_remove_all_info()
17618   is like calling CYRPTO_pop_info() until the stack is empty.
17619   Also updated memory leak detection code to be multi-thread-safe.
17620
17621   *Richard Levitte*
17622
17623 * Add options -text and -noout to pkcs7 utility and delete the
17624   encryption options which never did anything. Update docs.
17625
17626   *Steve Henson*
17627
17628 * Add options to some of the utilities to allow the pass phrase
17629   to be included on either the command line (not recommended on
17630   OSes like Unix) or read from the environment. Update the
17631   manpages and fix a few bugs.
17632
17633   *Steve Henson*
17634
17635 * Add a few manpages for some of the openssl commands.
17636
17637   *Steve Henson*
17638
17639 * Fix the -revoke option in ca. It was freeing up memory twice,
17640   leaking and not finding already revoked certificates.
17641
17642   *Steve Henson*
17643
17644 * Extensive changes to support certificate auxiliary information.
17645   This involves the use of X509_CERT_AUX structure and X509_AUX
17646   functions. An X509_AUX function such as PEM_read_X509_AUX()
17647   can still read in a certificate file in the usual way but it
17648   will also read in any additional "auxiliary information". By
17649   doing things this way a fair degree of compatibility can be
17650   retained: existing certificates can have this information added
17651   using the new 'x509' options.
17652
17653   Current auxiliary information includes an "alias" and some trust
17654   settings. The trust settings will ultimately be used in enhanced
17655   certificate chain verification routines: currently a certificate
17656   can only be trusted if it is self signed and then it is trusted
17657   for all purposes.
17658
17659   *Steve Henson*
17660
17661 * Fix assembler for Alpha (tested only on DEC OSF not Linux or `*BSD`).
17662   The problem was that one of the replacement routines had not been working
17663   since SSLeay releases.  For now the offending routine has been replaced
17664   with non-optimised assembler.  Even so, this now gives around 95%
17665   performance improvement for 1024 bit RSA signs.
17666
17667   *Mark Cox*
17668
17669 * Hack to fix PKCS#7 decryption when used with some unorthodox RC2
17670   handling. Most clients have the effective key size in bits equal to
17671   the key length in bits: so a 40 bit RC2 key uses a 40 bit (5 byte) key.
17672   A few however don't do this and instead use the size of the decrypted key
17673   to determine the RC2 key length and the AlgorithmIdentifier to determine
17674   the effective key length. In this case the effective key length can still
17675   be 40 bits but the key length can be 168 bits for example. This is fixed
17676   by manually forcing an RC2 key into the EVP_PKEY structure because the
17677   EVP code can't currently handle unusual RC2 key sizes: it always assumes
17678   the key length and effective key length are equal.
17679
17680   *Steve Henson*
17681
17682 * Add a bunch of functions that should simplify the creation of
17683   X509_NAME structures. Now you should be able to do:
17684   X509_NAME_add_entry_by_txt(nm, "CN", MBSTRING_ASC, "Steve", -1, -1, 0);
17685   and have it automatically work out the correct field type and fill in
17686   the structures. The more adventurous can try:
17687   X509_NAME_add_entry_by_txt(nm, field, MBSTRING_UTF8, str, -1, -1, 0);
17688   and it will (hopefully) work out the correct multibyte encoding.
17689
17690   *Steve Henson*
17691
17692 * Change the 'req' utility to use the new field handling and multibyte
17693   copy routines. Before the DN field creation was handled in an ad hoc
17694   way in req, ca, and x509 which was rather broken and didn't support
17695   BMPStrings or UTF8Strings. Since some software doesn't implement
17696   BMPStrings or UTF8Strings yet, they can be enabled using the config file
17697   using the dirstring_type option. See the new comment in the default
17698   openssl.cnf for more info.
17699
17700   *Steve Henson*
17701
17702 * Make crypto/rand/md_rand.c more robust:
17703   - Assure unique random numbers after fork().
17704   - Make sure that concurrent threads access the global counter and
17705     md serializably so that we never lose entropy in them
17706     or use exactly the same state in multiple threads.
17707     Access to the large state is not always serializable because
17708     the additional locking could be a performance killer, and
17709     md should be large enough anyway.
17710
17711   *Bodo Moeller*
17712
17713 * New file `apps/app_rand.c` with commonly needed functionality
17714   for handling the random seed file.
17715
17716   Use the random seed file in some applications that previously did not:
17717           ca,
17718           dsaparam -genkey (which also ignored its '-rand' option),
17719           s_client,
17720           s_server,
17721           x509 (when signing).
17722   Except on systems with /dev/urandom, it is crucial to have a random
17723   seed file at least for key creation, DSA signing, and for DH exchanges;
17724   for RSA signatures we could do without one.
17725
17726   gendh and gendsa (unlike genrsa) used to read only the first byte
17727   of each file listed in the '-rand' option.  The function as previously
17728   found in genrsa is now in app_rand.c and is used by all programs
17729   that support '-rand'.
17730
17731   *Bodo Moeller*
17732
17733 * In RAND_write_file, use mode 0600 for creating files;
17734   don't just chmod when it may be too late.
17735
17736   *Bodo Moeller*
17737
17738 * Report an error from X509_STORE_load_locations
17739   when X509_LOOKUP_load_file or X509_LOOKUP_add_dir failed.
17740
17741   *Bill Perry*
17742
17743 * New function ASN1_mbstring_copy() this copies a string in either
17744   ASCII, Unicode, Universal (4 bytes per character) or UTF8 format
17745   into an ASN1_STRING type. A mask of permissible types is passed
17746   and it chooses the "minimal" type to use or an error if not type
17747   is suitable.
17748
17749   *Steve Henson*
17750
17751 * Add function equivalents to the various macros in asn1.h. The old
17752   macros are retained with an `M_` prefix. Code inside the library can
17753   use the `M_` macros. External code (including the openssl utility)
17754   should *NOT* in order to be "shared library friendly".
17755
17756   *Steve Henson*
17757
17758 * Add various functions that can check a certificate's extensions
17759   to see if it usable for various purposes such as SSL client,
17760   server or S/MIME and CAs of these types. This is currently
17761   VERY EXPERIMENTAL but will ultimately be used for certificate chain
17762   verification. Also added a -purpose flag to x509 utility to
17763   print out all the purposes.
17764
17765   *Steve Henson*
17766
17767 * Add a CRYPTO_EX_DATA to X509 certificate structure and associated
17768   functions.
17769
17770   *Steve Henson*
17771
17772 * New `X509V3_{X509,CRL,REVOKED}_get_d2i()` functions. These will search
17773   for, obtain and decode and extension and obtain its critical flag.
17774   This allows all the necessary extension code to be handled in a
17775   single function call.
17776
17777   *Steve Henson*
17778
17779 * RC4 tune-up featuring 30-40% performance improvement on most RISC
17780   platforms. See crypto/rc4/rc4_enc.c for further details.
17781
17782   *Andy Polyakov*
17783
17784 * New -noout option to asn1parse. This causes no output to be produced
17785   its main use is when combined with -strparse and -out to extract data
17786   from a file (which may not be in ASN.1 format).
17787
17788   *Steve Henson*
17789
17790 * Fix for pkcs12 program. It was hashing an invalid certificate pointer
17791   when producing the local key id.
17792
17793   *Richard Levitte <levitte@stacken.kth.se>*
17794
17795 * New option -dhparam in s_server. This allows a DH parameter file to be
17796   stated explicitly. If it is not stated then it tries the first server
17797   certificate file. The previous behaviour hard coded the filename
17798   "server.pem".
17799
17800   *Steve Henson*
17801
17802 * Add -pubin and -pubout options to the rsa and dsa commands. These allow
17803   a public key to be input or output. For example:
17804   openssl rsa -in key.pem -pubout -out pubkey.pem
17805   Also added necessary DSA public key functions to handle this.
17806
17807   *Steve Henson*
17808
17809 * Fix so PKCS7_dataVerify() doesn't crash if no certificates are contained
17810   in the message. This was handled by allowing
17811   X509_find_by_issuer_and_serial() to tolerate a NULL passed to it.
17812
17813   *Steve Henson, reported by Sampo Kellomaki <sampo@mail.neuronio.pt>*
17814
17815 * Fix for bug in d2i_ASN1_bytes(): other ASN1 functions add an extra null
17816   to the end of the strings whereas this didn't. This would cause problems
17817   if strings read with d2i_ASN1_bytes() were later modified.
17818
17819   *Steve Henson, reported by Arne Ansper <arne@ats.cyber.ee>*
17820
17821 * Fix for base64 decode bug. When a base64 bio reads only one line of
17822   data and it contains EOF it will end up returning an error. This is
17823   caused by input 46 bytes long. The cause is due to the way base64
17824   BIOs find the start of base64 encoded data. They do this by trying a
17825   trial decode on each line until they find one that works. When they
17826   do a flag is set and it starts again knowing it can pass all the
17827   data directly through the decoder. Unfortunately it doesn't reset
17828   the context it uses. This means that if EOF is reached an attempt
17829   is made to pass two EOFs through the context and this causes the
17830   resulting error. This can also cause other problems as well. As is
17831   usual with these problems it takes *ages* to find and the fix is
17832   trivial: move one line.
17833
17834   *Steve Henson, reported by ian@uns.ns.ac.yu (Ivan Nejgebauer)*
17835
17836 * Ugly workaround to get s_client and s_server working under Windows. The
17837   old code wouldn't work because it needed to select() on sockets and the
17838   tty (for keypresses and to see if data could be written). Win32 only
17839   supports select() on sockets so we select() with a 1s timeout on the
17840   sockets and then see if any characters are waiting to be read, if none
17841   are present then we retry, we also assume we can always write data to
17842   the tty. This isn't nice because the code then blocks until we've
17843   received a complete line of data and it is effectively polling the
17844   keyboard at 1s intervals: however it's quite a bit better than not
17845   working at all :-) A dedicated Windows application might handle this
17846   with an event loop for example.
17847
17848   *Steve Henson*
17849
17850 * Enhance RSA_METHOD structure. Now there are two extra methods, rsa_sign
17851   and rsa_verify. When the RSA_FLAGS_SIGN_VER option is set these functions
17852   will be called when RSA_sign() and RSA_verify() are used. This is useful
17853   if rsa_pub_dec() and rsa_priv_enc() equivalents are not available.
17854   For this to work properly RSA_public_decrypt() and RSA_private_encrypt()
17855   should *not* be used: RSA_sign() and RSA_verify() must be used instead.
17856   This necessitated the support of an extra signature type NID_md5_sha1
17857   for SSL signatures and modifications to the SSL library to use it instead
17858   of calling RSA_public_decrypt() and RSA_private_encrypt().
17859
17860   *Steve Henson*
17861
17862 * Add new -verify -CAfile and -CApath options to the crl program, these
17863   will lookup a CRL issuers certificate and verify the signature in a
17864   similar way to the verify program. Tidy up the crl program so it
17865   no longer accesses structures directly. Make the ASN1 CRL parsing a bit
17866   less strict. It will now permit CRL extensions even if it is not
17867   a V2 CRL: this will allow it to tolerate some broken CRLs.
17868
17869   *Steve Henson*
17870
17871 * Initialize all non-automatic variables each time one of the openssl
17872   sub-programs is started (this is necessary as they may be started
17873   multiple times from the "OpenSSL>" prompt).
17874
17875   *Lennart Bang, Bodo Moeller*
17876
17877 * Preliminary compilation option RSA_NULL which disables RSA crypto without
17878   removing all other RSA functionality (this is what NO_RSA does). This
17879   is so (for example) those in the US can disable those operations covered
17880   by the RSA patent while allowing storage and parsing of RSA keys and RSA
17881   key generation.
17882
17883   *Steve Henson*
17884
17885 * Non-copying interface to BIO pairs.
17886   (still largely untested)
17887
17888   *Bodo Moeller*
17889
17890 * New function ASN1_tag2str() to convert an ASN1 tag to a descriptive
17891   ASCII string. This was handled independently in various places before.
17892
17893   *Steve Henson*
17894
17895 * New functions UTF8_getc() and UTF8_putc() that parse and generate
17896   UTF8 strings a character at a time.
17897
17898   *Steve Henson*
17899
17900 * Use client_version from client hello to select the protocol
17901   (s23_srvr.c) and for RSA client key exchange verification
17902   (s3_srvr.c), as required by the SSL 3.0/TLS 1.0 specifications.
17903
17904   *Bodo Moeller*
17905
17906 * Add various utility functions to handle SPKACs, these were previously
17907   handled by poking round in the structure internals. Added new function
17908   NETSCAPE_SPKI_print() to print out SPKAC and a new utility 'spkac' to
17909   print, verify and generate SPKACs. Based on an original idea from
17910   Massimiliano Pala <madwolf@comune.modena.it> but extensively modified.
17911
17912   *Steve Henson*
17913
17914 * RIPEMD160 is operational on all platforms and is back in 'make test'.
17915
17916   *Andy Polyakov*
17917
17918 * Allow the config file extension section to be overwritten on the
17919   command line. Based on an original idea from Massimiliano Pala
17920   <madwolf@comune.modena.it>. The new option is called -extensions
17921   and can be applied to ca, req and x509. Also -reqexts to override
17922   the request extensions in req and -crlexts to override the crl extensions
17923   in ca.
17924
17925   *Steve Henson*
17926
17927 * Add new feature to the SPKAC handling in ca.  Now you can include
17928   the same field multiple times by preceding it by "XXXX." for example:
17929   1.OU="Unit name 1"
17930   2.OU="Unit name 2"
17931   this is the same syntax as used in the req config file.
17932
17933   *Steve Henson*
17934
17935 * Allow certificate extensions to be added to certificate requests. These
17936   are specified in a 'req_extensions' option of the req section of the
17937   config file. They can be printed out with the -text option to req but
17938   are otherwise ignored at present.
17939
17940   *Steve Henson*
17941
17942 * Fix a horrible bug in enc_read() in crypto/evp/bio_enc.c: if the first
17943   data read consists of only the final block it would not decrypted because
17944   EVP_CipherUpdate() would correctly report zero bytes had been decrypted.
17945   A misplaced 'break' also meant the decrypted final block might not be
17946   copied until the next read.
17947
17948   *Steve Henson*
17949
17950 * Initial support for DH_METHOD. Again based on RSA_METHOD. Also added
17951   a few extra parameters to the DH structure: these will be useful if
17952   for example we want the value of 'q' or implement X9.42 DH.
17953
17954   *Steve Henson*
17955
17956 * Initial support for DSA_METHOD. This is based on the RSA_METHOD and
17957   provides hooks that allow the default DSA functions or functions on a
17958   "per key" basis to be replaced. This allows hardware acceleration and
17959   hardware key storage to be handled without major modification to the
17960   library. Also added low-level modexp hooks and CRYPTO_EX structure and
17961   associated functions.
17962
17963   *Steve Henson*
17964
17965 * Add a new flag to memory BIOs, BIO_FLAG_MEM_RDONLY. This marks the BIO
17966   as "read only": it can't be written to and the buffer it points to will
17967   not be freed. Reading from a read only BIO is much more efficient than
17968   a normal memory BIO. This was added because there are several times when
17969   an area of memory needs to be read from a BIO. The previous method was
17970   to create a memory BIO and write the data to it, this results in two
17971   copies of the data and an O(n^2) reading algorithm. There is a new
17972   function BIO_new_mem_buf() which creates a read only memory BIO from
17973   an area of memory. Also modified the PKCS#7 routines to use read only
17974   memory BIOs.
17975
17976   *Steve Henson*
17977
17978 * Bugfix: ssl23_get_client_hello did not work properly when called in
17979   state SSL23_ST_SR_CLNT_HELLO_B, i.e. when the first 7 bytes of
17980   a SSLv2-compatible client hello for SSLv3 or TLSv1 could be read,
17981   but a retry condition occurred while trying to read the rest.
17982
17983   *Bodo Moeller*
17984
17985 * The PKCS7_ENC_CONTENT_new() function was setting the content type as
17986   NID_pkcs7_encrypted by default: this was wrong since this should almost
17987   always be NID_pkcs7_data. Also modified the PKCS7_set_type() to handle
17988   the encrypted data type: this is a more sensible place to put it and it
17989   allows the PKCS#12 code to be tidied up that duplicated this
17990   functionality.
17991
17992   *Steve Henson*
17993
17994 * Changed obj_dat.pl script so it takes its input and output files on
17995   the command line. This should avoid shell escape redirection problems
17996   under Win32.
17997
17998   *Steve Henson*
17999
18000 * Initial support for certificate extension requests, these are included
18001   in things like Xenroll certificate requests. Included functions to allow
18002   extensions to be obtained and added.
18003
18004   *Steve Henson*
18005
18006 * -crlf option to s_client and s_server for sending newlines as
18007   CRLF (as required by many protocols).
18008
18009   *Bodo Moeller*
18010
18011### Changes between 0.9.3a and 0.9.4  [09 Aug 1999]
18012
18013 * Install libRSAglue.a when OpenSSL is built with RSAref.
18014
18015   *Ralf S. Engelschall*
18016
18017 * A few more `#ifndef NO_FP_API / #endif` pairs for consistency.
18018
18019   *Andrija Antonijevic <TheAntony2@bigfoot.com>*
18020
18021 * Fix -startdate and -enddate (which was missing) arguments to 'ca'
18022   program.
18023
18024   *Steve Henson*
18025
18026 * New function DSA_dup_DH, which duplicates DSA parameters/keys as
18027   DH parameters/keys (q is lost during that conversion, but the resulting
18028   DH parameters contain its length).
18029
18030   For 1024-bit p, DSA_generate_parameters followed by DSA_dup_DH is
18031   much faster than DH_generate_parameters (which creates parameters
18032   where `p = 2*q + 1`), and also the smaller q makes DH computations
18033   much more efficient (160-bit exponentiation instead of 1024-bit
18034   exponentiation); so this provides a convenient way to support DHE
18035   ciphersuites in SSL/TLS servers (see ssl/ssltest.c).  It is of
18036   utter importance to use
18037           SSL_CTX_set_options(s_ctx, SSL_OP_SINGLE_DH_USE);
18038   or
18039           SSL_set_options(s_ctx, SSL_OP_SINGLE_DH_USE);
18040   when such DH parameters are used, because otherwise small subgroup
18041   attacks may become possible!
18042
18043   *Bodo Moeller*
18044
18045 * Avoid memory leak in i2d_DHparams.
18046
18047   *Bodo Moeller*
18048
18049 * Allow the -k option to be used more than once in the enc program:
18050   this allows the same encrypted message to be read by multiple recipients.
18051
18052   *Steve Henson*
18053
18054 * New function OBJ_obj2txt(buf, buf_len, a, no_name), this converts
18055   an ASN1_OBJECT to a text string. If the "no_name" parameter is set then
18056   it will always use the numerical form of the OID, even if it has a short
18057   or long name.
18058
18059   *Steve Henson*
18060
18061 * Added an extra RSA flag: RSA_FLAG_EXT_PKEY. Previously the rsa_mod_exp
18062   method only got called if p,q,dmp1,dmq1,iqmp components were present,
18063   otherwise bn_mod_exp was called. In the case of hardware keys for example
18064   no private key components need be present and it might store extra data
18065   in the RSA structure, which cannot be accessed from bn_mod_exp.
18066   By setting RSA_FLAG_EXT_PKEY rsa_mod_exp will always be called for
18067   private key operations.
18068
18069   *Steve Henson*
18070
18071 * Added support for SPARC Linux.
18072
18073   *Andy Polyakov*
18074
18075 * pem_password_cb function type incompatibly changed from
18076           typedef int pem_password_cb(char *buf, int size, int rwflag);
18077   to
18078           ....(char *buf, int size, int rwflag, void *userdata);
18079   so that applications can pass data to their callbacks:
18080   The `PEM[_ASN1]_{read,write}...` functions and macros now take an
18081   additional void * argument, which is just handed through whenever
18082   the password callback is called.
18083
18084   *Damien Miller <dmiller@ilogic.com.au>; tiny changes by Bodo Moeller*
18085
18086   New function SSL_CTX_set_default_passwd_cb_userdata.
18087
18088   Compatibility note: As many C implementations push function arguments
18089   onto the stack in reverse order, the new library version is likely to
18090   interoperate with programs that have been compiled with the old
18091   pem_password_cb definition (PEM_whatever takes some data that
18092   happens to be on the stack as its last argument, and the callback
18093   just ignores this garbage); but there is no guarantee whatsoever that
18094   this will work.
18095
18096 * The -DPLATFORM="\"$(PLATFORM)\"" definition and the similar -DCFLAGS=...
18097   (both in crypto/Makefile.ssl for use by crypto/cversion.c) caused
18098   problems not only on Windows, but also on some Unix platforms.
18099   To avoid problematic command lines, these definitions are now in an
18100   auto-generated file crypto/buildinf.h (created by crypto/Makefile.ssl
18101   for standard "make" builds, by util/mk1mf.pl for "mk1mf" builds).
18102
18103   *Bodo Moeller*
18104
18105 * MIPS III/IV assembler module is reimplemented.
18106
18107   *Andy Polyakov*
18108
18109 * More DES library cleanups: remove references to srand/rand and
18110   delete an unused file.
18111
18112   *Ulf Möller*
18113
18114 * Add support for the free Netwide assembler (NASM) under Win32,
18115   since not many people have MASM (ml) and it can be hard to obtain.
18116   This is currently experimental but it seems to work OK and pass all
18117   the tests. Check out INSTALL.W32 for info.
18118
18119   *Steve Henson*
18120
18121 * Fix memory leaks in s3_clnt.c: All non-anonymous SSL3/TLS1 connections
18122   without temporary keys kept an extra copy of the server key,
18123   and connections with temporary keys did not free everything in case
18124   of an error.
18125
18126   *Bodo Moeller*
18127
18128 * New function RSA_check_key and new openssl rsa option -check
18129   for verifying the consistency of RSA keys.
18130
18131   *Ulf Moeller, Bodo Moeller*
18132
18133 * Various changes to make Win32 compile work:
18134   1. Casts to avoid "loss of data" warnings in p5_crpt2.c
18135   2. Change unsigned int to int in b_dump.c to avoid "signed/unsigned
18136      comparison" warnings.
18137   3. Add `sk_<TYPE>_sort` to DEF file generator and do make update.
18138
18139   *Steve Henson*
18140
18141 * Add a debugging option to PKCS#5 v2 key generation function: when
18142   you #define DEBUG_PKCS5V2 passwords, salts, iteration counts and
18143   derived keys are printed to stderr.
18144
18145   *Steve Henson*
18146
18147 * Copy the flags in ASN1_STRING_dup().
18148
18149   *Roman E. Pavlov <pre@mo.msk.ru>*
18150
18151 * The x509 application mishandled signing requests containing DSA
18152   keys when the signing key was also DSA and the parameters didn't match.
18153
18154   It was supposed to omit the parameters when they matched the signing key:
18155   the verifying software was then supposed to automatically use the CA's
18156   parameters if they were absent from the end user certificate.
18157
18158   Omitting parameters is no longer recommended. The test was also
18159   the wrong way round! This was probably due to unusual behaviour in
18160   EVP_cmp_parameters() which returns 1 if the parameters match.
18161   This meant that parameters were omitted when they *didn't* match and
18162   the certificate was useless. Certificates signed with 'ca' didn't have
18163   this bug.
18164
18165   *Steve Henson, reported by Doug Erickson <Doug.Erickson@Part.NET>*
18166
18167 * Memory leak checking (-DCRYPTO_MDEBUG) had some problems.
18168   The interface is as follows:
18169   Applications can use
18170           CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_ON) aka MemCheck_start(),
18171           CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_OFF) aka MemCheck_stop();
18172   "off" is now the default.
18173   The library internally uses
18174           CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_DISABLE) aka MemCheck_off(),
18175           CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_ENABLE) aka MemCheck_on()
18176   to disable memory-checking temporarily.
18177
18178   Some inconsistent states that previously were possible (and were
18179   even the default) are now avoided.
18180
18181   -DCRYPTO_MDEBUG_TIME is new and additionally stores the current time
18182   with each memory chunk allocated; this is occasionally more helpful
18183   than just having a counter.
18184
18185   -DCRYPTO_MDEBUG_THREAD is also new and adds the thread ID.
18186
18187   -DCRYPTO_MDEBUG_ALL enables all of the above, plus any future
18188   extensions.
18189
18190   *Bodo Moeller*
18191
18192 * Introduce "mode" for SSL structures (with defaults in SSL_CTX),
18193   which largely parallels "options", but is for changing API behaviour,
18194   whereas "options" are about protocol behaviour.
18195   Initial "mode" flags are:
18196
18197   SSL_MODE_ENABLE_PARTIAL_WRITE   Allow SSL_write to report success when
18198                                   a single record has been written.
18199   SSL_MODE_ACCEPT_MOVING_WRITE_BUFFER  Don't insist that SSL_write
18200                                   retries use the same buffer location.
18201                                   (But all of the contents must be
18202                                   copied!)
18203
18204   *Bodo Moeller*
18205
18206 * Bugfix: SSL_set_options ignored its parameter, only SSL_CTX_set_options
18207   worked.
18208
18209 * Fix problems with no-hmac etc.
18210
18211   *Ulf Möller, pointed out by Brian Wellington <bwelling@tislabs.com>*
18212
18213 * New functions RSA_get_default_method(), RSA_set_method() and
18214   RSA_get_method(). These allows replacement of RSA_METHODs without having
18215   to mess around with the internals of an RSA structure.
18216
18217   *Steve Henson*
18218
18219 * Fix memory leaks in DSA_do_sign and DSA_is_prime.
18220   Also really enable memory leak checks in openssl.c and in some
18221   test programs.
18222
18223   *Chad C. Mulligan, Bodo Moeller*
18224
18225 * Fix a bug in d2i_ASN1_INTEGER() and i2d_ASN1_INTEGER() which can mess
18226   up the length of negative integers. This has now been simplified to just
18227   store the length when it is first determined and use it later, rather
18228   than trying to keep track of where data is copied and updating it to
18229   point to the end.
18230   *Steve Henson, reported by Brien Wheeler <bwheeler@authentica-security.com>*
18231
18232 * Add a new function PKCS7_signatureVerify. This allows the verification
18233   of a PKCS#7 signature but with the signing certificate passed to the
18234   function itself. This contrasts with PKCS7_dataVerify which assumes the
18235   certificate is present in the PKCS#7 structure. This isn't always the
18236   case: certificates can be omitted from a PKCS#7 structure and be
18237   distributed by "out of band" means (such as a certificate database).
18238
18239   *Steve Henson*
18240
18241 * Complete the `PEM_*` macros with DECLARE_PEM versions to replace the
18242   function prototypes in pem.h, also change util/mkdef.pl to add the
18243   necessary function names.
18244
18245   *Steve Henson*
18246
18247 * mk1mf.pl (used by Windows builds) did not properly read the
18248   options set by Configure in the top level Makefile, and Configure
18249   was not even able to write more than one option correctly.
18250   Fixed, now "no-idea no-rc5 -DCRYPTO_MDEBUG" etc. works as intended.
18251
18252   *Bodo Moeller*
18253
18254 * New functions CONF_load_bio() and CONF_load_fp() to allow a config
18255   file to be loaded from a BIO or FILE pointer. The BIO version will
18256   for example allow memory BIOs to contain config info.
18257
18258   *Steve Henson*
18259
18260 * New function "CRYPTO_num_locks" that returns CRYPTO_NUM_LOCKS.
18261   Whoever hopes to achieve shared-library compatibility across versions
18262   must use this, not the compile-time macro.
18263   (Exercise 0.9.4: Which is the minimum library version required by
18264   such programs?)
18265   Note: All this applies only to multi-threaded programs, others don't
18266   need locks.
18267
18268   *Bodo Moeller*
18269
18270 * Add missing case to s3_clnt.c state machine -- one of the new SSL tests
18271   through a BIO pair triggered the default case, i.e.
18272   SSLerr(...,SSL_R_UNKNOWN_STATE).
18273
18274   *Bodo Moeller*
18275
18276 * New "BIO pair" concept (crypto/bio/bss_bio.c) so that applications
18277   can use the SSL library even if none of the specific BIOs is
18278   appropriate.
18279
18280   *Bodo Moeller*
18281
18282 * Fix a bug in i2d_DSAPublicKey() which meant it returned the wrong value
18283   for the encoded length.
18284
18285   *Jeon KyoungHo <khjeon@sds.samsung.co.kr>*
18286
18287 * Add initial documentation of the X509V3 functions.
18288
18289   *Steve Henson*
18290
18291 * Add a new pair of functions PEM_write_PKCS8PrivateKey() and
18292   PEM_write_bio_PKCS8PrivateKey() that are equivalent to
18293   PEM_write_PrivateKey() and PEM_write_bio_PrivateKey() but use the more
18294   secure PKCS#8 private key format with a high iteration count.
18295
18296   *Steve Henson*
18297
18298 * Fix determination of Perl interpreter: A perl or perl5
18299   *directory* in $PATH was also accepted as the interpreter.
18300
18301   *Ralf S. Engelschall*
18302
18303 * Fix demos/sign/sign.c: well there wasn't anything strictly speaking
18304   wrong with it but it was very old and did things like calling
18305   PEM_ASN1_read() directly and used MD5 for the hash not to mention some
18306   unusual formatting.
18307
18308   *Steve Henson*
18309
18310 * Fix demos/selfsign.c: it used obsolete and deleted functions, changed
18311   to use the new extension code.
18312
18313   *Steve Henson*
18314
18315 * Implement the PEM_read/PEM_write functions in crypto/pem/pem_all.c
18316   with macros. This should make it easier to change their form, add extra
18317   arguments etc. Fix a few PEM prototypes which didn't have cipher as a
18318   constant.
18319
18320   *Steve Henson*
18321
18322 * Add to configuration table a new entry that can specify an alternative
18323   name for unistd.h (for pre-POSIX systems); we need this for NeXTstep,
18324   according to Mark Crispin <MRC@Panda.COM>.
18325
18326   *Bodo Moeller*
18327
18328 * DES CBC did not update the IV. Weird.
18329
18330   *Ben Laurie*
18331lse
18332   des_cbc_encrypt does not update the IV, but des_ncbc_encrypt does.
18333   Changing the behaviour of the former might break existing programs --
18334   where IV updating is needed, des_ncbc_encrypt can be used.
18335ndif
18336
18337 * When bntest is run from "make test" it drives bc to check its
18338   calculations, as well as internally checking them. If an internal check
18339   fails, it needs to cause bc to give a non-zero result or make test carries
18340   on without noticing the failure. Fixed.
18341
18342   *Ben Laurie*
18343
18344 * DES library cleanups.
18345
18346   *Ulf Möller*
18347
18348 * Add support for PKCS#5 v2.0 PBE algorithms. This will permit PKCS#8 to be
18349   used with any cipher unlike PKCS#5 v1.5 which can at most handle 64 bit
18350   ciphers. NOTE: although the key derivation function has been verified
18351   against some published test vectors it has not been extensively tested
18352   yet. Added a -v2 "cipher" option to pkcs8 application to allow the use
18353   of v2.0.
18354
18355   *Steve Henson*
18356
18357 * Instead of "mkdir -p", which is not fully portable, use new
18358   Perl script "util/mkdir-p.pl".
18359
18360   *Bodo Moeller*
18361
18362 * Rewrite the way password based encryption (PBE) is handled. It used to
18363   assume that the ASN1 AlgorithmIdentifier parameter was a PBEParameter
18364   structure. This was true for the PKCS#5 v1.5 and PKCS#12 PBE algorithms
18365   but doesn't apply to PKCS#5 v2.0 where it can be something else. Now
18366   the 'parameter' field of the AlgorithmIdentifier is passed to the
18367   underlying key generation function so it must do its own ASN1 parsing.
18368   This has also changed the EVP_PBE_CipherInit() function which now has a
18369   'parameter' argument instead of literal salt and iteration count values
18370   and the function EVP_PBE_ALGOR_CipherInit() has been deleted.
18371
18372   *Steve Henson*
18373
18374 * Support for PKCS#5 v1.5 compatible password based encryption algorithms
18375   and PKCS#8 functionality. New 'pkcs8' application linked to openssl.
18376   Needed to change the PEM_STRING_EVP_PKEY value which was just "PRIVATE
18377   KEY" because this clashed with PKCS#8 unencrypted string. Since this
18378   value was just used as a "magic string" and not used directly its
18379   value doesn't matter.
18380
18381   *Steve Henson*
18382
18383 * Introduce some semblance of const correctness to BN. Shame C doesn't
18384   support mutable.
18385
18386   *Ben Laurie*
18387
18388 * "linux-sparc64" configuration (ultrapenguin).
18389
18390   *Ray Miller <ray.miller@oucs.ox.ac.uk>*
18391   "linux-sparc" configuration.
18392
18393   *Christian Forster <fo@hawo.stw.uni-erlangen.de>*
18394
18395 * config now generates no-xxx options for missing ciphers.
18396
18397   *Ulf Möller*
18398
18399 * Support the EBCDIC character set (work in progress).
18400   File ebcdic.c not yet included because it has a different license.
18401
18402   *Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>*
18403
18404 * Support BS2000/OSD-POSIX.
18405
18406   *Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>*
18407
18408 * Make callbacks for key generation use `void *` instead of `char *`.
18409
18410   *Ben Laurie*
18411
18412 * Make S/MIME samples compile (not yet tested).
18413
18414   *Ben Laurie*
18415
18416 * Additional typesafe stacks.
18417
18418   *Ben Laurie*
18419
18420 * New configuration variants "bsdi-elf-gcc" (BSD/OS 4.x).
18421
18422   *Bodo Moeller*
18423
18424### Changes between 0.9.3 and 0.9.3a  [29 May 1999]
18425
18426 * New configuration variant "sco5-gcc".
18427
18428 * Updated some demos.
18429
18430   *Sean O Riordain, Wade Scholine*
18431
18432 * Add missing BIO_free at exit of pkcs12 application.
18433
18434   *Wu Zhigang*
18435
18436 * Fix memory leak in conf.c.
18437
18438   *Steve Henson*
18439
18440 * Updates for Win32 to assembler version of MD5.
18441
18442   *Steve Henson*
18443
18444 * Set #! path to perl in `apps/der_chop` to where we found it
18445   instead of using a fixed path.
18446
18447   *Bodo Moeller*
18448
18449 * SHA library changes for irix64-mips4-cc.
18450
18451   *Andy Polyakov*
18452
18453 * Improvements for VMS support.
18454
18455   *Richard Levitte*
18456
18457### Changes between 0.9.2b and 0.9.3  [24 May 1999]
18458
18459 * Bignum library bug fix. IRIX 6 passes "make test" now!
18460   This also avoids the problems with SC4.2 and unpatched SC5.
18461
18462   *Andy Polyakov <appro@fy.chalmers.se>*
18463
18464 * New functions sk_num, sk_value and sk_set to replace the previous macros.
18465   These are required because of the typesafe stack would otherwise break
18466   existing code. If old code used a structure member which used to be STACK
18467   and is now STACK_OF (for example cert in a PKCS7_SIGNED structure) with
18468   sk_num or sk_value it would produce an error because the num, data members
18469   are not present in STACK_OF. Now it just produces a warning. sk_set
18470   replaces the old method of assigning a value to sk_value
18471   (e.g. sk_value(x, i) = y) which the library used in a few cases. Any code
18472   that does this will no longer work (and should use sk_set instead) but
18473   this could be regarded as a "questionable" behaviour anyway.
18474
18475   *Steve Henson*
18476
18477 * Fix most of the other PKCS#7 bugs. The "experimental" code can now
18478   correctly handle encrypted S/MIME data.
18479
18480   *Steve Henson*
18481
18482 * Change type of various DES function arguments from des_cblock
18483   (which means, in function argument declarations, pointer to char)
18484   to des_cblock * (meaning pointer to array with 8 char elements),
18485   which allows the compiler to do more typechecking; it was like
18486   that back in SSLeay, but with lots of ugly casts.
18487
18488   Introduce new type const_des_cblock.
18489
18490   *Bodo Moeller*
18491
18492 * Reorganise the PKCS#7 library and get rid of some of the more obvious
18493   problems: find RecipientInfo structure that matches recipient certificate
18494   and initialise the ASN1 structures properly based on passed cipher.
18495
18496   *Steve Henson*
18497
18498 * Belatedly make the BN tests actually check the results.
18499
18500   *Ben Laurie*
18501
18502 * Fix the encoding and decoding of negative ASN1 INTEGERS and conversion
18503   to and from BNs: it was completely broken. New compilation option
18504   NEG_PUBKEY_BUG to allow for some broken certificates that encode public
18505   key elements as negative integers.
18506
18507   *Steve Henson*
18508
18509 * Reorganize and speed up MD5.
18510
18511   *Andy Polyakov <appro@fy.chalmers.se>*
18512
18513 * VMS support.
18514
18515   *Richard Levitte <richard@levitte.org>*
18516
18517 * New option -out to asn1parse to allow the parsed structure to be
18518   output to a file. This is most useful when combined with the -strparse
18519   option to examine the output of things like OCTET STRINGS.
18520
18521   *Steve Henson*
18522
18523 * Make SSL library a little more fool-proof by not requiring any longer
18524   that `SSL_set_{accept,connect}_state` be called before
18525   `SSL_{accept,connect}` may be used (`SSL_set_..._state` is omitted
18526   in many applications because usually everything *appeared* to work as
18527   intended anyway -- now it really works as intended).
18528
18529   *Bodo Moeller*
18530
18531 * Move openssl.cnf out of lib/.
18532
18533   *Ulf Möller*
18534
18535 * Fix various things to let OpenSSL even pass "egcc -pipe -O2 -Wall
18536   -Wshadow -Wpointer-arith -Wcast-align -Wmissing-prototypes
18537   -Wmissing-declarations -Wnested-externs -Winline" with EGCS 1.1.2+
18538
18539   *Ralf S. Engelschall*
18540
18541 * Various fixes to the EVP and PKCS#7 code. It may now be able to
18542   handle PKCS#7 enveloped data properly.
18543
18544   *Sebastian Akerman <sak@parallelconsulting.com>, modified by Steve*
18545
18546 * Create a duplicate of the SSL_CTX's CERT in SSL_new instead of
18547   copying pointers.  The cert_st handling is changed by this in
18548   various ways (and thus what used to be known as ctx->default_cert
18549   is now called ctx->cert, since we don't resort to `s->ctx->[default_]cert`
18550   any longer when s->cert does not give us what we need).
18551   ssl_cert_instantiate becomes obsolete by this change.
18552   As soon as we've got the new code right (possibly it already is?),
18553   we have solved a couple of bugs of the earlier code where s->cert
18554   was used as if it could not have been shared with other SSL structures.
18555
18556   Note that using the SSL API in certain dirty ways now will result
18557   in different behaviour than observed with earlier library versions:
18558   Changing settings for an `SSL_CTX *ctx` after having done s = SSL_new(ctx)
18559   does not influence s as it used to.
18560
18561   In order to clean up things more thoroughly, inside SSL_SESSION
18562   we don't use CERT any longer, but a new structure SESS_CERT
18563   that holds per-session data (if available); currently, this is
18564   the peer's certificate chain and, for clients, the server's certificate
18565   and temporary key.  CERT holds only those values that can have
18566   meaningful defaults in an SSL_CTX.
18567
18568   *Bodo Moeller*
18569
18570 * New function X509V3_EXT_i2d() to create an X509_EXTENSION structure
18571   from the internal representation. Various PKCS#7 fixes: remove some
18572   evil casts and set the enc_dig_alg field properly based on the signing
18573   key type.
18574
18575   *Steve Henson*
18576
18577 * Allow PKCS#12 password to be set from the command line or the
18578   environment. Let 'ca' get its config file name from the environment
18579   variables "OPENSSL_CONF" or "SSLEAY_CONF" (for consistency with 'req'
18580   and 'x509').
18581
18582   *Steve Henson*
18583
18584 * Allow certificate policies extension to use an IA5STRING for the
18585   organization field. This is contrary to the PKIX definition but
18586   VeriSign uses it and IE5 only recognises this form. Document 'x509'
18587   extension option.
18588
18589   *Steve Henson*
18590
18591 * Add PEDANTIC compiler flag to allow compilation with gcc -pedantic,
18592   without disallowing inline assembler and the like for non-pedantic builds.
18593
18594   *Ben Laurie*
18595
18596 * Support Borland C++ builder.
18597
18598   *Janez Jere <jj@void.si>, modified by Ulf Möller*
18599
18600 * Support Mingw32.
18601
18602   *Ulf Möller*
18603
18604 * SHA-1 cleanups and performance enhancements.
18605
18606   *Andy Polyakov <appro@fy.chalmers.se>*
18607
18608 * Sparc v8plus assembler for the bignum library.
18609
18610   *Andy Polyakov <appro@fy.chalmers.se>*
18611
18612 * Accept any -xxx and +xxx compiler options in Configure.
18613
18614   *Ulf Möller*
18615
18616 * Update HPUX configuration.
18617
18618   *Anonymous*
18619
18620 * Add missing `sk_<type>_unshift()` function to safestack.h
18621
18622   *Ralf S. Engelschall*
18623
18624 * New function SSL_CTX_use_certificate_chain_file that sets the
18625   "extra_cert"s in addition to the certificate.  (This makes sense
18626   only for "PEM" format files, as chains as a whole are not
18627   DER-encoded.)
18628
18629   *Bodo Moeller*
18630
18631 * Support verify_depth from the SSL API.
18632   x509_vfy.c had what can be considered an off-by-one-error:
18633   Its depth (which was not part of the external interface)
18634   was actually counting the number of certificates in a chain;
18635   now it really counts the depth.
18636
18637   *Bodo Moeller*
18638
18639 * Bugfix in crypto/x509/x509_cmp.c: The SSLerr macro was used
18640   instead of X509err, which often resulted in confusing error
18641   messages since the error codes are not globally unique
18642   (e.g. an alleged error in ssl3_accept when a certificate
18643   didn't match the private key).
18644
18645 * New function SSL_CTX_set_session_id_context that allows to set a default
18646   value (so that you don't need SSL_set_session_id_context for each
18647   connection using the SSL_CTX).
18648
18649   *Bodo Moeller*
18650
18651 * OAEP decoding bug fix.
18652
18653   *Ulf Möller*
18654
18655 * Support INSTALL_PREFIX for package builders, as proposed by
18656   David Harris.
18657
18658   *Bodo Moeller*
18659
18660 * New Configure options "threads" and "no-threads".  For systems
18661   where the proper compiler options are known (currently Solaris
18662   and Linux), "threads" is the default.
18663
18664   *Bodo Moeller*
18665
18666 * New script util/mklink.pl as a faster substitute for util/mklink.sh.
18667
18668   *Bodo Moeller*
18669
18670 * Install various scripts to $(OPENSSLDIR)/misc, not to
18671   $(INSTALLTOP)/bin -- they shouldn't clutter directories
18672   such as /usr/local/bin.
18673
18674   *Bodo Moeller*
18675
18676 * "make linux-shared" to build shared libraries.
18677
18678   *Niels Poppe <niels@netbox.org>*
18679
18680 * New Configure option `no-<cipher>` (rsa, idea, rc5, ...).
18681
18682   *Ulf Möller*
18683
18684 * Add the PKCS#12 API documentation to openssl.txt. Preliminary support for
18685   extension adding in x509 utility.
18686
18687   *Steve Henson*
18688
18689 * Remove NOPROTO sections and error code comments.
18690
18691   *Ulf Möller*
18692
18693 * Partial rewrite of the DEF file generator to now parse the ANSI
18694   prototypes.
18695
18696   *Steve Henson*
18697
18698 * New Configure options --prefix=DIR and --openssldir=DIR.
18699
18700   *Ulf Möller*
18701
18702 * Complete rewrite of the error code script(s). It is all now handled
18703   by one script at the top level which handles error code gathering,
18704   header rewriting and C source file generation. It should be much better
18705   than the old method: it now uses a modified version of Ulf's parser to
18706   read the ANSI prototypes in all header files (thus the old K&R definitions
18707   aren't needed for error creation any more) and do a better job of
18708   translating function codes into names. The old 'ASN1 error code embedded
18709   in a comment' is no longer necessary and it doesn't use .err files which
18710   have now been deleted. Also the error code call doesn't have to appear all
18711   on one line (which resulted in some large lines...).
18712
18713   *Steve Henson*
18714
18715 * Change #include filenames from `<foo.h>` to `<openssl/foo.h>`.
18716
18717   *Bodo Moeller*
18718
18719 * Change behaviour of ssl2_read when facing length-0 packets: Don't return
18720   0 (which usually indicates a closed connection), but continue reading.
18721
18722   *Bodo Moeller*
18723
18724 * Fix some race conditions.
18725
18726   *Bodo Moeller*
18727
18728 * Add support for CRL distribution points extension. Add Certificate
18729   Policies and CRL distribution points documentation.
18730
18731   *Steve Henson*
18732
18733 * Move the autogenerated header file parts to crypto/opensslconf.h.
18734
18735   *Ulf Möller*
18736
18737 * Fix new 56-bit DES export ciphersuites: they were using 7 bytes instead of
18738   8 of keying material. Merlin has also confirmed interop with this fix
18739   between OpenSSL and Baltimore C/SSL 2.0 and J/SSL 2.0.
18740
18741   *Merlin Hughes <merlin@baltimore.ie>*
18742
18743 * Fix lots of warnings.
18744
18745   *Richard Levitte <levitte@stacken.kth.se>*
18746
18747 * In add_cert_dir() in crypto/x509/by_dir.c, break out of the loop if
18748   the directory spec didn't end with a LIST_SEPARATOR_CHAR.
18749
18750   *Richard Levitte <levitte@stacken.kth.se>*
18751
18752 * Fix problems with sizeof(long) == 8.
18753
18754   *Andy Polyakov <appro@fy.chalmers.se>*
18755
18756 * Change functions to ANSI C.
18757
18758   *Ulf Möller*
18759
18760 * Fix typos in error codes.
18761
18762   *Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>, Ulf Möller*
18763
18764 * Remove defunct assembler files from Configure.
18765
18766   *Ulf Möller*
18767
18768 * SPARC v8 assembler BIGNUM implementation.
18769
18770   *Andy Polyakov <appro@fy.chalmers.se>*
18771
18772 * Support for Certificate Policies extension: both print and set.
18773   Various additions to support the r2i method this uses.
18774
18775   *Steve Henson*
18776
18777 * A lot of constification, and fix a bug in X509_NAME_oneline() that could
18778   return a const string when you are expecting an allocated buffer.
18779
18780   *Ben Laurie*
18781
18782 * Add support for ASN1 types UTF8String and VISIBLESTRING, also the CHOICE
18783   types DirectoryString and DisplayText.
18784
18785   *Steve Henson*
18786
18787 * Add code to allow r2i extensions to access the configuration database,
18788   add an LHASH database driver and add several ctx helper functions.
18789
18790   *Steve Henson*
18791
18792 * Fix an evil bug in bn_expand2() which caused various BN functions to
18793   fail when they extended the size of a BIGNUM.
18794
18795   *Steve Henson*
18796
18797 * Various utility functions to handle SXNet extension. Modify mkdef.pl to
18798   support typesafe stack.
18799
18800   *Steve Henson*
18801
18802 * Fix typo in SSL_[gs]et_options().
18803
18804   *Nils Frostberg <nils@medcom.se>*
18805
18806 * Delete various functions and files that belonged to the (now obsolete)
18807   old X509V3 handling code.
18808
18809   *Steve Henson*
18810
18811 * New Configure option "rsaref".
18812
18813   *Ulf Möller*
18814
18815 * Don't auto-generate pem.h.
18816
18817   *Bodo Moeller*
18818
18819 * Introduce type-safe ASN.1 SETs.
18820
18821   *Ben Laurie*
18822
18823 * Convert various additional casted stacks to type-safe STACK_OF() variants.
18824
18825   *Ben Laurie, Ralf S. Engelschall, Steve Henson*
18826
18827 * Introduce type-safe STACKs. This will almost certainly break lots of code
18828   that links with OpenSSL (well at least cause lots of warnings), but fear
18829   not: the conversion is trivial, and it eliminates loads of evil casts. A
18830   few STACKed things have been converted already. Feel free to convert more.
18831   In the fullness of time, I'll do away with the STACK type altogether.
18832
18833   *Ben Laurie*
18834
18835 * Add `openssl ca -revoke <certfile>` facility which revokes a certificate
18836   specified in `<certfile>` by updating the entry in the index.txt file.
18837   This way one no longer has to edit the index.txt file manually for
18838   revoking a certificate. The -revoke option does the gory details now.
18839
18840   *Massimiliano Pala <madwolf@openca.org>, Ralf S. Engelschall*
18841
18842 * Fix `openssl crl -noout -text` combination where `-noout` killed the
18843   `-text` option at all and this way the `-noout -text` combination was
18844   inconsistent in `openssl crl` with the friends in `openssl x509|rsa|dsa`.
18845
18846   *Ralf S. Engelschall*
18847
18848 * Make sure a corresponding plain text error message exists for the
18849   X509_V_ERR_CERT_REVOKED/23 error number which can occur when a
18850   verify callback function determined that a certificate was revoked.
18851
18852   *Ralf S. Engelschall*
18853
18854 * Bugfix: In test/testenc, don't test `openssl <cipher>` for
18855   ciphers that were excluded, e.g. by -DNO_IDEA.  Also, test
18856   all available ciphers including rc5, which was forgotten until now.
18857   In order to let the testing shell script know which algorithms
18858   are available, a new (up to now undocumented) command
18859   `openssl list-cipher-commands` is used.
18860
18861   *Bodo Moeller*
18862
18863 * Bugfix: s_client occasionally would sleep in select() when
18864   it should have checked SSL_pending() first.
18865
18866   *Bodo Moeller*
18867
18868 * New functions DSA_do_sign and DSA_do_verify to provide access to
18869   the raw DSA values prior to ASN.1 encoding.
18870
18871   *Ulf Möller*
18872
18873 * Tweaks to Configure
18874
18875   *Niels Poppe <niels@netbox.org>*
18876
18877 * Add support for PKCS#5 v2.0 ASN1 PBES2 structures. No other support,
18878   yet...
18879
18880   *Steve Henson*
18881
18882 * New variables $(RANLIB) and $(PERL) in the Makefiles.
18883
18884   *Ulf Möller*
18885
18886 * New config option to avoid instructions that are illegal on the 80386.
18887   The default code is faster, but requires at least a 486.
18888
18889   *Ulf Möller*
18890
18891 * Got rid of old SSL2_CLIENT_VERSION (inconsistently used) and
18892   SSL2_SERVER_VERSION (not used at all) macros, which are now the
18893   same as SSL2_VERSION anyway.
18894
18895   *Bodo Moeller*
18896
18897 * New "-showcerts" option for s_client.
18898
18899   *Bodo Moeller*
18900
18901 * Still more PKCS#12 integration. Add pkcs12 application to openssl
18902   application. Various cleanups and fixes.
18903
18904   *Steve Henson*
18905
18906 * More PKCS#12 integration. Add new pkcs12 directory with Makefile.ssl and
18907   modify error routines to work internally. Add error codes and PBE init
18908   to library startup routines.
18909
18910   *Steve Henson*
18911
18912 * Further PKCS#12 integration. Added password based encryption, PKCS#8 and
18913   packing functions to asn1 and evp. Changed function names and error
18914   codes along the way.
18915
18916   *Steve Henson*
18917
18918 * PKCS12 integration: and so it begins... First of several patches to
18919   slowly integrate PKCS#12 functionality into OpenSSL. Add PKCS#12
18920   objects to objects.h
18921
18922   *Steve Henson*
18923
18924 * Add a new 'indent' option to some X509V3 extension code. Initial ASN1
18925   and display support for Thawte strong extranet extension.
18926
18927   *Steve Henson*
18928
18929 * Add LinuxPPC support.
18930
18931   *Jeff Dubrule <igor@pobox.org>*
18932
18933 * Get rid of redundant BN file bn_mulw.c, and rename bn_div64 to
18934   bn_div_words in alpha.s.
18935
18936   *Hannes Reinecke <H.Reinecke@hw.ac.uk> and Ben Laurie*
18937
18938 * Make sure the RSA OAEP test is skipped under -DRSAref because
18939   OAEP isn't supported when OpenSSL is built with RSAref.
18940
18941   *Ulf Moeller <ulf@fitug.de>*
18942
18943 * Move definitions of IS_SET/IS_SEQUENCE inside crypto/asn1/asn1.h
18944   so they no longer are missing under -DNOPROTO.
18945
18946   *Soren S. Jorvang <soren@t.dk>*
18947
18948### Changes between 0.9.1c and 0.9.2b  [22 Mar 1999]
18949
18950 * Make SSL_get_peer_cert_chain() work in servers. Unfortunately, it still
18951   doesn't work when the session is reused. Coming soon!
18952
18953   *Ben Laurie*
18954
18955 * Fix a security hole, that allows sessions to be reused in the wrong
18956   context thus bypassing client cert protection! All software that uses
18957   client certs and session caches in multiple contexts NEEDS PATCHING to
18958   allow session reuse! A fuller solution is in the works.
18959
18960   *Ben Laurie, problem pointed out by Holger Reif, Bodo Moeller (and ???)*
18961
18962 * Some more source tree cleanups (removed obsolete files
18963   crypto/bf/asm/bf586.pl, test/test.txt and crypto/sha/asm/f.s; changed
18964   permission on "config" script to be executable) and a fix for the INSTALL
18965   document.
18966
18967   *Ulf Moeller <ulf@fitug.de>*
18968
18969 * Remove some legacy and erroneous uses of malloc, free instead of
18970   Malloc, Free.
18971
18972   *Lennart Bang <lob@netstream.se>, with minor changes by Steve*
18973
18974 * Make rsa_oaep_test return non-zero on error.
18975
18976   *Ulf Moeller <ulf@fitug.de>*
18977
18978 * Add support for native Solaris shared libraries. Configure
18979   solaris-sparc-sc4-pic, make, then run shlib/solaris-sc4.sh. It'd be nice
18980   if someone would make that last step automatic.
18981
18982   *Matthias Loepfe <Matthias.Loepfe@AdNovum.CH>*
18983
18984 * ctx_size was not built with the right compiler during "make links". Fixed.
18985
18986   *Ben Laurie*
18987
18988 * Change the meaning of 'ALL' in the cipher list. It now means "everything
18989   except NULL ciphers". This means the default cipher list will no longer
18990   enable NULL ciphers. They need to be specifically enabled e.g. with
18991   the string "DEFAULT:eNULL".
18992
18993   *Steve Henson*
18994
18995 * Fix to RSA private encryption routines: if p < q then it would
18996   occasionally produce an invalid result. This will only happen with
18997   externally generated keys because OpenSSL (and SSLeay) ensure p > q.
18998
18999   *Steve Henson*
19000
19001 * Be less restrictive and allow also `perl util/perlpath.pl
19002   /path/to/bin/perl` in addition to `perl util/perlpath.pl /path/to/bin`,
19003   because this way one can also use an interpreter named `perl5` (which is
19004   usually the name of Perl 5.xxx on platforms where an Perl 4.x is still
19005   installed as `perl`).
19006
19007   *Matthias Loepfe <Matthias.Loepfe@adnovum.ch>*
19008
19009 * Let util/clean-depend.pl work also with older Perl 5.00x versions.
19010
19011   *Matthias Loepfe <Matthias.Loepfe@adnovum.ch>*
19012
19013 * Fix Makefile.org so CC,CFLAG etc are passed to 'make links' add
19014   advapi32.lib to Win32 build and change the pem test comparison
19015   to fc.exe (thanks to Ulrich Kroener <kroneru@yahoo.com> for the
19016   suggestion). Fix misplaced ASNI prototypes and declarations in evp.h
19017   and crypto/des/ede_cbcm_enc.c.
19018
19019   *Steve Henson*
19020
19021 * DES quad checksum was broken on big-endian architectures. Fixed.
19022
19023   *Ben Laurie*
19024
19025 * Comment out two functions in bio.h that aren't implemented. Fix up the
19026   Win32 test batch file so it (might) work again. The Win32 test batch file
19027   is horrible: I feel ill....
19028
19029   *Steve Henson*
19030
19031 * Move various #ifdefs around so NO_SYSLOG, NO_DIRENT etc are now selected
19032   in e_os.h. Audit of header files to check ANSI and non ANSI
19033   sections: 10 functions were absent from non ANSI section and not exported
19034   from Windows DLLs. Fixed up libeay.num for new functions.
19035
19036   *Steve Henson*
19037
19038 * Make `openssl version` output lines consistent.
19039
19040   *Ralf S. Engelschall*
19041
19042 * Fix Win32 symbol export lists for BIO functions: Added
19043   BIO_get_ex_new_index, BIO_get_ex_num, BIO_get_ex_data and BIO_set_ex_data
19044   to ms/libeay{16,32}.def.
19045
19046   *Ralf S. Engelschall*
19047
19048 * Second round of fixing the OpenSSL perl/ stuff. It now at least compiled
19049   fine under Unix and passes some trivial tests I've now added. But the
19050   whole stuff is horribly incomplete, so a README.1ST with a disclaimer was
19051   added to make sure no one expects that this stuff really works in the
19052   OpenSSL 0.9.2 release.  Additionally I've started to clean the XS sources
19053   up and fixed a few little bugs and inconsistencies in OpenSSL.{pm,xs} and
19054   openssl_bio.xs.
19055
19056   *Ralf S. Engelschall*
19057
19058 * Fix the generation of two part addresses in perl.
19059
19060   *Kenji Miyake <kenji@miyake.org>, integrated by Ben Laurie*
19061
19062 * Add config entry for Linux on MIPS.
19063
19064   *John Tobey <jtobey@channel1.com>*
19065
19066 * Make links whenever Configure is run, unless we are on Windoze.
19067
19068   *Ben Laurie*
19069
19070 * Permit extensions to be added to CRLs using crl_section in openssl.cnf.
19071   Currently only issuerAltName and AuthorityKeyIdentifier make any sense
19072   in CRLs.
19073
19074   *Steve Henson*
19075
19076 * Add a useful kludge to allow package maintainers to specify compiler and
19077   other platforms details on the command line without having to patch the
19078   Configure script every time: One now can use
19079   `perl Configure <id>:<details>`,
19080   i.e. platform ids are allowed to have details appended
19081   to them (separated by colons). This is treated as there would be a static
19082   pre-configured entry in Configure's %table under key `<id>` with value
19083   `<details>` and `perl Configure <id>` is called.  So, when you want to
19084   perform a quick test-compile under FreeBSD 3.1 with pgcc and without
19085   assembler stuff you can use `perl Configure "FreeBSD-elf:pgcc:-O6:::"`
19086   now, which overrides the FreeBSD-elf entry on-the-fly.
19087
19088   *Ralf S. Engelschall*
19089
19090 * Disable new TLS1 ciphersuites by default: they aren't official yet.
19091
19092   *Ben Laurie*
19093
19094 * Allow DSO flags like -fpic, -fPIC, -KPIC etc. to be specified
19095   on the `perl Configure ...` command line. This way one can compile
19096   OpenSSL libraries with Position Independent Code (PIC) which is needed
19097   for linking it into DSOs.
19098
19099   *Ralf S. Engelschall*
19100
19101 * Remarkably, export ciphers were totally broken and no-one had noticed!
19102   Fixed.
19103
19104   *Ben Laurie*
19105
19106 * Cleaned up the LICENSE document: The official contact for any license
19107   questions now is the OpenSSL core team under openssl-core@openssl.org.
19108   And add a paragraph about the dual-license situation to make sure people
19109   recognize that _BOTH_ the OpenSSL license _AND_ the SSLeay license apply
19110   to the OpenSSL toolkit.
19111
19112   *Ralf S. Engelschall*
19113
19114 * General source tree makefile cleanups: Made `making xxx in yyy...`
19115   display consistent in the source tree and replaced `/bin/rm` by `rm`.
19116   Additionally cleaned up the `make links` target: Remove unnecessary
19117   semicolons, subsequent redundant removes, inline point.sh into mklink.sh
19118   to speed processing and no longer clutter the display with confusing
19119   stuff. Instead only the actually done links are displayed.
19120
19121   *Ralf S. Engelschall*
19122
19123 * Permit null encryption ciphersuites, used for authentication only. It used
19124   to be necessary to set the preprocessor define SSL_ALLOW_ENULL to do this.
19125   It is now necessary to set SSL_FORBID_ENULL to prevent the use of null
19126   encryption.
19127
19128   *Ben Laurie*
19129
19130 * Add a bunch of fixes to the PKCS#7 stuff. It used to sometimes reorder
19131   signed attributes when verifying signatures (this would break them),
19132   the detached data encoding was wrong and public keys obtained using
19133   X509_get_pubkey() weren't freed.
19134
19135   *Steve Henson*
19136
19137 * Add text documentation for the BUFFER functions. Also added a work around
19138   to a Win95 console bug. This was triggered by the password read stuff: the
19139   last character typed gets carried over to the next fread(). If you were
19140   generating a new cert request using 'req' for example then the last
19141   character of the passphrase would be CR which would then enter the first
19142   field as blank.
19143
19144   *Steve Henson*
19145
19146 * Added the new 'Includes OpenSSL Cryptography Software' button as
19147   doc/openssl_button.{gif,html} which is similar in style to the old SSLeay
19148   button and can be used by applications based on OpenSSL to show the
19149   relationship to the OpenSSL project.
19150
19151   *Ralf S. Engelschall*
19152
19153 * Remove confusing variables in function signatures in files
19154   ssl/ssl_lib.c and ssl/ssl.h.
19155
19156   *Lennart Bong <lob@kulthea.stacken.kth.se>*
19157
19158 * Don't install bss_file.c under PREFIX/include/
19159
19160   *Lennart Bong <lob@kulthea.stacken.kth.se>*
19161
19162 * Get the Win32 compile working again. Modify mkdef.pl so it can handle
19163   functions that return function pointers and has support for NT specific
19164   stuff. Fix mk1mf.pl and VC-32.pl to support NT differences also. Various
19165   #ifdef WIN32 and WINNTs sprinkled about the place and some changes from
19166   unsigned to signed types: this was killing the Win32 compile.
19167
19168   *Steve Henson*
19169
19170 * Add new certificate file to stack functions,
19171   SSL_add_dir_cert_subjects_to_stack() and
19172   SSL_add_file_cert_subjects_to_stack().  These largely supplant
19173   SSL_load_client_CA_file(), and can be used to add multiple certs easily
19174   to a stack (usually this is then handed to SSL_CTX_set_client_CA_list()).
19175   This means that Apache-SSL and similar packages don't have to mess around
19176   to add as many CAs as they want to the preferred list.
19177
19178   *Ben Laurie*
19179
19180 * Experiment with doxygen documentation. Currently only partially applied to
19181   ssl/ssl_lib.c.
19182   See <http://www.stack.nl/~dimitri/doxygen/index.html>, and run doxygen with
19183   openssl.doxy as the configuration file.
19184
19185   *Ben Laurie*
19186
19187 * Get rid of remaining C++-style comments which strict C compilers hate.
19188
19189   *Ralf S. Engelschall, pointed out by Carlos Amengual*
19190
19191 * Changed BN_RECURSION in bn_mont.c to BN_RECURSION_MONT so it is not
19192   compiled in by default: it has problems with large keys.
19193
19194   *Steve Henson*
19195
19196 * Add a bunch of SSL_xxx() functions for configuring the temporary RSA and
19197   DH private keys and/or callback functions which directly correspond to
19198   their SSL_CTX_xxx() counterparts but work on a per-connection basis. This
19199   is needed for applications which have to configure certificates on a
19200   per-connection basis (e.g. Apache+mod_ssl) instead of a per-context basis
19201   (e.g. s_server).
19202      For the RSA certificate situation is makes no difference, but
19203   for the DSA certificate situation this fixes the "no shared cipher"
19204   problem where the OpenSSL cipher selection procedure failed because the
19205   temporary keys were not overtaken from the context and the API provided
19206   no way to reconfigure them.
19207      The new functions now let applications reconfigure the stuff and they
19208   are in detail: SSL_need_tmp_RSA, SSL_set_tmp_rsa, SSL_set_tmp_dh,
19209   SSL_set_tmp_rsa_callback and SSL_set_tmp_dh_callback.  Additionally a new
19210   non-public-API function ssl_cert_instantiate() is used as a helper
19211   function and also to reduce code redundancy inside ssl_rsa.c.
19212
19213   *Ralf S. Engelschall*
19214
19215 * Move s_server -dcert and -dkey options out of the undocumented feature
19216   area because they are useful for the DSA situation and should be
19217   recognized by the users.
19218
19219   *Ralf S. Engelschall*
19220
19221 * Fix the cipher decision scheme for export ciphers: the export bits are
19222   *not* within SSL_MKEY_MASK or SSL_AUTH_MASK, they are within
19223   SSL_EXP_MASK.  So, the original variable has to be used instead of the
19224   already masked variable.
19225
19226   *Richard Levitte <levitte@stacken.kth.se>*
19227
19228 * Fix `port` variable from `int` to `unsigned int` in crypto/bio/b_sock.c
19229
19230   *Richard Levitte <levitte@stacken.kth.se>*
19231
19232 * Change type of another md_len variable in pk7_doit.c:PKCS7_dataFinal()
19233   from `int` to `unsigned int` because it is a length and initialized by
19234   EVP_DigestFinal() which expects an `unsigned int *`.
19235
19236   *Richard Levitte <levitte@stacken.kth.se>*
19237
19238 * Don't hard-code path to Perl interpreter on shebang line of Configure
19239   script. Instead use the usual Shell->Perl transition trick.
19240
19241   *Ralf S. Engelschall*
19242
19243 * Make `openssl x509 -noout -modulus`' functional also for DSA certificates
19244   (in addition to RSA certificates) to match the behaviour of `openssl dsa
19245   -noout -modulus` as it's already the case for `openssl rsa -noout
19246   -modulus`.  For RSA the -modulus is the real "modulus" while for DSA
19247   currently the public key is printed (a decision which was already done by
19248   `openssl dsa -modulus` in the past) which serves a similar purpose.
19249   Additionally the NO_RSA no longer completely removes the whole -modulus
19250   option; it now only avoids using the RSA stuff. Same applies to NO_DSA
19251   now, too.
19252
19253   *Ralf S.  Engelschall*
19254
19255 * Add Arne Ansper's reliable BIO - this is an encrypted, block-digested
19256   BIO. See the source (crypto/evp/bio_ok.c) for more info.
19257
19258   *Arne Ansper <arne@ats.cyber.ee>*
19259
19260 * Dump the old yucky req code that tried (and failed) to allow raw OIDs
19261   to be added. Now both 'req' and 'ca' can use new objects defined in the
19262   config file.
19263
19264   *Steve Henson*
19265
19266 * Add cool BIO that does syslog (or event log on NT).
19267
19268   *Arne Ansper <arne@ats.cyber.ee>, integrated by Ben Laurie*
19269
19270 * Add support for new TLS ciphersuites, TLS_RSA_EXPORT56_WITH_RC4_56_MD5,
19271   TLS_RSA_EXPORT56_WITH_RC2_CBC_56_MD5 and
19272   TLS_RSA_EXPORT56_WITH_DES_CBC_SHA, as specified in "56-bit Export Cipher
19273   Suites For TLS", draft-ietf-tls-56-bit-ciphersuites-00.txt.
19274
19275   *Ben Laurie*
19276
19277 * Add preliminary config info for new extension code.
19278
19279   *Steve Henson*
19280
19281 * Make RSA_NO_PADDING really use no padding.
19282
19283   *Ulf Moeller <ulf@fitug.de>*
19284
19285 * Generate errors when private/public key check is done.
19286
19287   *Ben Laurie*
19288
19289 * Overhaul for 'crl' utility. New function X509_CRL_print. Partial support
19290   for some CRL extensions and new objects added.
19291
19292   *Steve Henson*
19293
19294 * Really fix the ASN1 IMPLICIT bug this time... Partial support for private
19295   key usage extension and fuller support for authority key id.
19296
19297   *Steve Henson*
19298
19299 * Add OAEP encryption for the OpenSSL crypto library. OAEP is the improved
19300   padding method for RSA, which is recommended for new applications in PKCS
19301   #1 v2.0 (RFC 2437, October 1998).
19302   OAEP (Optimal Asymmetric Encryption Padding) has better theoretical
19303   foundations than the ad-hoc padding used in PKCS #1 v1.5. It is secure
19304   against Bleichbacher's attack on RSA.
19305   *Ulf Moeller <ulf@fitug.de>, reformatted, corrected and integrated by
19306   Ben Laurie*
19307
19308 * Updates to the new SSL compression code
19309
19310   *Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)*
19311
19312 * Fix so that the version number in the master secret, when passed
19313   via RSA, checks that if TLS was proposed, but we roll back to SSLv3
19314   (because the server will not accept higher), that the version number
19315   is 0x03,0x01, not 0x03,0x00
19316
19317   *Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)*
19318
19319 * Run extensive memory leak checks on SSL commands. Fixed *lots* of memory
19320   leaks in `ssl/` relating to new `X509_get_pubkey()` behaviour. Also fixes
19321   in `apps/` and an unrelated leak in `crypto/dsa/dsa_vrf.c`.
19322
19323   *Steve Henson*
19324
19325 * Support for RAW extensions where an arbitrary extension can be
19326   created by including its DER encoding. See `apps/openssl.cnf` for
19327   an example.
19328
19329   *Steve Henson*
19330
19331 * Make sure latest Perl versions don't interpret some generated C array
19332   code as Perl array code in the crypto/err/err_genc.pl script.
19333
19334   *Lars Weber <3weber@informatik.uni-hamburg.de>*
19335
19336 * Modify ms/do_ms.bat to not generate assembly language makefiles since
19337   not many people have the assembler. Various Win32 compilation fixes and
19338   update to the INSTALL.W32 file with (hopefully) more accurate Win32
19339   build instructions.
19340
19341   *Steve Henson*
19342
19343 * Modify configure script 'Configure' to automatically create crypto/date.h
19344   file under Win32 and also build pem.h from pem.org. New script
19345   util/mkfiles.pl to create the MINFO file on environments that can't do a
19346   'make files': perl util/mkfiles.pl >MINFO should work.
19347
19348   *Steve Henson*
19349
19350 * Major rework of DES function declarations, in the pursuit of correctness
19351   and purity. As a result, many evil casts evaporated, and some weirdness,
19352   too. You may find this causes warnings in your code. Zapping your evil
19353   casts will probably fix them. Mostly.
19354
19355   *Ben Laurie*
19356
19357 * Fix for a typo in asn1.h. Bug fix to object creation script
19358   obj_dat.pl. It considered a zero in an object definition to mean
19359   "end of object": none of the objects in objects.h have any zeros
19360   so it wasn't spotted.
19361
19362   *Steve Henson, reported by Erwann ABALEA <eabalea@certplus.com>*
19363
19364 * Add support for Triple DES Cipher Block Chaining with Output Feedback
19365   Masking (CBCM). In the absence of test vectors, the best I have been able
19366   to do is check that the decrypt undoes the encrypt, so far. Send me test
19367   vectors if you have them.
19368
19369   *Ben Laurie*
19370
19371 * Correct calculation of key length for export ciphers (too much space was
19372   allocated for null ciphers). This has not been tested!
19373
19374   *Ben Laurie*
19375
19376 * Modifications to the mkdef.pl for Win32 DEF file creation. The usage
19377   message is now correct (it understands "crypto" and "ssl" on its
19378   command line). There is also now an "update" option. This will update
19379   the util/ssleay.num and util/libeay.num files with any new functions.
19380   If you do a:
19381   perl util/mkdef.pl crypto ssl update
19382   it will update them.
19383
19384   *Steve Henson*
19385
19386 * Overhauled the Perl interface:
19387   - ported BN stuff to OpenSSL's different BN library
19388   - made the perl/ source tree CVS-aware
19389   - renamed the package from SSLeay to OpenSSL (the files still contain
19390     their history because I've copied them in the repository)
19391   - removed obsolete files (the test scripts will be replaced
19392     by better Test::Harness variants in the future)
19393
19394   *Ralf S. Engelschall*
19395
19396 * First cut for a very conservative source tree cleanup:
19397   1. merge various obsolete readme texts into doc/ssleay.txt
19398   where we collect the old documents and readme texts.
19399   2. remove the first part of files where I'm already sure that we no
19400   longer need them because of three reasons: either they are just temporary
19401   files which were left by Eric or they are preserved original files where
19402   I've verified that the diff is also available in the CVS via "cvs diff
19403   -rSSLeay_0_8_1b" or they were renamed (as it was definitely the case for
19404   the crypto/md/ stuff).
19405
19406   *Ralf S. Engelschall*
19407
19408 * More extension code. Incomplete support for subject and issuer alt
19409   name, issuer and authority key id. Change the i2v function parameters
19410   and add an extra 'crl' parameter in the X509V3_CTX structure: guess
19411   what that's for :-) Fix to ASN1 macro which messed up
19412   IMPLICIT tag and add f_enum.c which adds a2i, i2a for ENUMERATED.
19413
19414   *Steve Henson*
19415
19416 * Preliminary support for ENUMERATED type. This is largely copied from the
19417   INTEGER code.
19418
19419   *Steve Henson*
19420
19421 * Add new function, EVP_MD_CTX_copy() to replace frequent use of memcpy.
19422
19423   *Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)*
19424
19425 * Make sure `make rehash` target really finds the `openssl` program.
19426
19427   *Ralf S. Engelschall, Matthias Loepfe <Matthias.Loepfe@adnovum.ch>*
19428
19429 * Squeeze another 7% of speed out of MD5 assembler, at least on a P2. I'd
19430   like to hear about it if this slows down other processors.
19431
19432   *Ben Laurie*
19433
19434 * Add CygWin32 platform information to Configure script.
19435
19436   *Alan Batie <batie@aahz.jf.intel.com>*
19437
19438 * Fixed ms/32all.bat script: `no_asm` -> `no-asm`
19439
19440   *Rainer W. Gerling <gerling@mpg-gv.mpg.de>*
19441
19442 * New program nseq to manipulate netscape certificate sequences
19443
19444   *Steve Henson*
19445
19446 * Modify crl2pkcs7 so it supports multiple -certfile arguments. Fix a
19447   few typos.
19448
19449   *Steve Henson*
19450
19451 * Fixes to BN code.  Previously the default was to define BN_RECURSION
19452   but the BN code had some problems that would cause failures when
19453   doing certificate verification and some other functions.
19454
19455   *Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)*
19456
19457 * Add ASN1 and PEM code to support netscape certificate sequences.
19458
19459   *Steve Henson*
19460
19461 * Add ASN1 and PEM code to support netscape certificate sequences.
19462
19463   *Steve Henson*
19464
19465 * Add several PKIX and private extended key usage OIDs.
19466
19467   *Steve Henson*
19468
19469 * Modify the 'ca' program to handle the new extension code. Modify
19470   openssl.cnf for new extension format, add comments.
19471
19472   *Steve Henson*
19473
19474 * More X509 V3 changes. Fix typo in v3_bitstr.c. Add support to 'req'
19475   and add a sample to openssl.cnf so req -x509 now adds appropriate
19476   CA extensions.
19477
19478   *Steve Henson*
19479
19480 * Continued X509 V3 changes. Add to other makefiles, integrate with the
19481   error code, add initial support to X509_print() and x509 application.
19482
19483   *Steve Henson*
19484
19485 * Takes a deep breath and start adding X509 V3 extension support code. Add
19486   files in crypto/x509v3. Move original stuff to crypto/x509v3/old. All this
19487   stuff is currently isolated and isn't even compiled yet.
19488
19489   *Steve Henson*
19490
19491 * Continuing patches for GeneralizedTime. Fix up certificate and CRL
19492   ASN1 to use ASN1_TIME and modify print routines to use ASN1_TIME_print.
19493   Removed the versions check from X509 routines when loading extensions:
19494   this allows certain broken certificates that don't set the version
19495   properly to be processed.
19496
19497   *Steve Henson*
19498
19499 * Deal with irritating shit to do with dependencies, in YAAHW (Yet Another
19500   Ad Hoc Way) - Makefile.ssls now all contain local dependencies, which
19501   can still be regenerated with "make depend".
19502
19503   *Ben Laurie*
19504
19505 * Spelling mistake in C version of CAST-128.
19506
19507   *Ben Laurie, reported by Jeremy Hylton <jeremy@cnri.reston.va.us>*
19508
19509 * Changes to the error generation code. The perl script err-code.pl
19510   now reads in the old error codes and retains the old numbers, only
19511   adding new ones if necessary. It also only changes the .err files if new
19512   codes are added. The makefiles have been modified to only insert errors
19513   when needed (to avoid needlessly modifying header files). This is done
19514   by only inserting errors if the .err file is newer than the auto generated
19515   C file. To rebuild all the error codes from scratch (the old behaviour)
19516   either modify crypto/Makefile.ssl to pass the -regen flag to err_code.pl
19517   or delete all the .err files.
19518
19519   *Steve Henson*
19520
19521 * CAST-128 was incorrectly implemented for short keys. The C version has
19522   been fixed, but is untested. The assembler versions are also fixed, but
19523   new assembler HAS NOT BEEN GENERATED FOR WIN32 - the Makefile needs fixing
19524   to regenerate it if needed.
19525   *Ben Laurie, reported (with fix for C version) by Jun-ichiro itojun
19526    Hagino <itojun@kame.net>*
19527
19528 * File was opened incorrectly in randfile.c.
19529
19530   *Ulf Möller <ulf@fitug.de>*
19531
19532 * Beginning of support for GeneralizedTime. d2i, i2d, check and print
19533   functions. Also ASN1_TIME suite which is a CHOICE of UTCTime or
19534   GeneralizedTime. ASN1_TIME is the proper type used in certificates et
19535   al: it's just almost always a UTCTime. Note this patch adds new error
19536   codes so do a "make errors" if there are problems.
19537
19538   *Steve Henson*
19539
19540 * Correct Linux 1 recognition in config.
19541
19542   *Ulf Möller <ulf@fitug.de>*
19543
19544 * Remove pointless MD5 hash when using DSA keys in ca.
19545
19546   *Anonymous <nobody@replay.com>*
19547
19548 * Generate an error if given an empty string as a cert directory. Also
19549   generate an error if handed NULL (previously returned 0 to indicate an
19550   error, but didn't set one).
19551
19552   *Ben Laurie, reported by Anonymous <nobody@replay.com>*
19553
19554 * Add prototypes to SSL methods. Make SSL_write's buffer const, at last.
19555
19556   *Ben Laurie*
19557
19558 * Fix the dummy function BN_ref_mod_exp() in rsaref.c to have the correct
19559   parameters. This was causing a warning which killed off the Win32 compile.
19560
19561   *Steve Henson*
19562
19563 * Remove C++ style comments from crypto/bn/bn_local.h.
19564
19565   *Neil Costigan <neil.costigan@celocom.com>*
19566
19567 * The function OBJ_txt2nid was broken. It was supposed to return a nid
19568   based on a text string, looking up short and long names and finally
19569   "dot" format. The "dot" format stuff didn't work. Added new function
19570   OBJ_txt2obj to do the same but return an ASN1_OBJECT and rewrote
19571   OBJ_txt2nid to use it. OBJ_txt2obj can also return objects even if the
19572   OID is not part of the table.
19573
19574   *Steve Henson*
19575
19576 * Add prototypes to X509 lookup/verify methods, fixing a bug in
19577   X509_LOOKUP_by_alias().
19578
19579   *Ben Laurie*
19580
19581 * Sort openssl functions by name.
19582
19583   *Ben Laurie*
19584
19585 * Get the `gendsa` command working and add it to the `list` command. Remove
19586   encryption from sample DSA keys (in case anyone is interested the password
19587   was "1234").
19588
19589   *Steve Henson*
19590
19591 * Make *all* `*_free` functions accept a NULL pointer.
19592
19593   *Frans Heymans <fheymans@isaserver.be>*
19594
19595 * If a DH key is generated in s3_srvr.c, don't blow it by trying to use
19596   NULL pointers.
19597
19598   *Anonymous <nobody@replay.com>*
19599
19600 * s_server should send the CAfile as acceptable CAs, not its own cert.
19601
19602   *Bodo Moeller <3moeller@informatik.uni-hamburg.de>*
19603
19604 * Don't blow it for numeric `-newkey` arguments to `apps/req`.
19605
19606   *Bodo Moeller <3moeller@informatik.uni-hamburg.de>*
19607
19608 * Temp key "for export" tests were wrong in s3_srvr.c.
19609
19610   *Anonymous <nobody@replay.com>*
19611
19612 * Add prototype for temp key callback functions
19613   SSL_CTX_set_tmp_{rsa,dh}_callback().
19614
19615   *Ben Laurie*
19616
19617 * Make DH_free() tolerate being passed a NULL pointer (like RSA_free() and
19618   DSA_free()). Make X509_PUBKEY_set() check for errors in d2i_PublicKey().
19619
19620   *Steve Henson*
19621
19622 * X509_name_add_entry() freed the wrong thing after an error.
19623
19624   *Arne Ansper <arne@ats.cyber.ee>*
19625
19626 * rsa_eay.c would attempt to free a NULL context.
19627
19628   *Arne Ansper <arne@ats.cyber.ee>*
19629
19630 * BIO_s_socket() had a broken should_retry() on Windoze.
19631
19632   *Arne Ansper <arne@ats.cyber.ee>*
19633
19634 * BIO_f_buffer() didn't pass on BIO_CTRL_FLUSH.
19635
19636   *Arne Ansper <arne@ats.cyber.ee>*
19637
19638 * Make sure the already existing X509_STORE->depth variable is initialized
19639   in X509_STORE_new(), but document the fact that this variable is still
19640   unused in the certificate verification process.
19641
19642   *Ralf S. Engelschall*
19643
19644 * Fix the various library and `apps/` files to free up pkeys obtained from
19645   X509_PUBKEY_get() et al. Also allow x509.c to handle netscape extensions.
19646
19647   *Steve Henson*
19648
19649 * Fix reference counting in X509_PUBKEY_get(). This makes
19650   demos/maurice/example2.c work, amongst others, probably.
19651
19652   *Steve Henson and Ben Laurie*
19653
19654 * First cut of a cleanup for `apps/`. First the `ssleay` program is now named
19655   `openssl` and second, the shortcut symlinks for the `openssl <command>`
19656   are no longer created. This way we have a single and consistent command
19657   line interface `openssl <command>`, similar to `cvs <command>`.
19658
19659   *Ralf S. Engelschall, Paul Sutton and Ben Laurie*
19660
19661 * ca.c: move test for DSA keys inside #ifndef NO_DSA. Make pubkey
19662   BIT STRING wrapper always have zero unused bits.
19663
19664   *Steve Henson*
19665
19666 * Add CA.pl, perl version of CA.sh, add extended key usage OID.
19667
19668   *Steve Henson*
19669
19670 * Make the top-level INSTALL documentation easier to understand.
19671
19672   *Paul Sutton*
19673
19674 * Makefiles updated to exit if an error occurs in a sub-directory
19675   make (including if user presses ^C) [Paul Sutton]
19676
19677 * Make Montgomery context stuff explicit in RSA data structure.
19678
19679   *Ben Laurie*
19680
19681 * Fix build order of pem and err to allow for generated pem.h.
19682
19683   *Ben Laurie*
19684
19685 * Fix renumbering bug in X509_NAME_delete_entry().
19686
19687   *Ben Laurie*
19688
19689 * Enhanced the err-ins.pl script so it makes the error library number
19690   global and can add a library name. This is needed for external ASN1 and
19691   other error libraries.
19692
19693   *Steve Henson*
19694
19695 * Fixed sk_insert which never worked properly.
19696
19697   *Steve Henson*
19698
19699 * Fix ASN1 macros so they can handle indefinite length constructed
19700   EXPLICIT tags. Some non standard certificates use these: they can now
19701   be read in.
19702
19703   *Steve Henson*
19704
19705 * Merged the various old/obsolete SSLeay documentation files (doc/xxx.doc)
19706   into a single doc/ssleay.txt bundle. This way the information is still
19707   preserved but no longer messes up this directory. Now it's new room for
19708   the new set of documentation files.
19709
19710   *Ralf S. Engelschall*
19711
19712 * SETs were incorrectly DER encoded. This was a major pain, because they
19713   shared code with SEQUENCEs, which aren't coded the same. This means that
19714   almost everything to do with SETs or SEQUENCEs has either changed name or
19715   number of arguments.
19716
19717   *Ben Laurie, based on a partial fix by GP Jayan <gp@nsj.co.jp>*
19718
19719 * Fix test data to work with the above.
19720
19721   *Ben Laurie*
19722
19723 * Fix the RSA header declarations that hid a bug I fixed in 0.9.0b but
19724   was already fixed by Eric for 0.9.1 it seems.
19725
19726   *Ben Laurie - pointed out by Ulf Möller <ulf@fitug.de>*
19727
19728 * Autodetect FreeBSD3.
19729
19730   *Ben Laurie*
19731
19732 * Fix various bugs in Configure. This affects the following platforms:
19733   nextstep
19734   ncr-scde
19735   unixware-2.0
19736   unixware-2.0-pentium
19737   sco5-cc.
19738
19739   *Ben Laurie*
19740
19741 * Eliminate generated files from CVS. Reorder tests to regenerate files
19742   before they are needed.
19743
19744   *Ben Laurie*
19745
19746 * Generate Makefile.ssl from Makefile.org (to keep CVS happy).
19747
19748   *Ben Laurie*
19749
19750### Changes between 0.9.1b and 0.9.1c  [23-Dec-1998]
19751
19752 * Added OPENSSL_VERSION_NUMBER to crypto/crypto.h and
19753   changed SSLeay to OpenSSL in version strings.
19754
19755   *Ralf S. Engelschall*
19756
19757 * Some fixups to the top-level documents.
19758
19759   *Paul Sutton*
19760
19761 * Fixed the nasty bug where rsaref.h was not found under compile-time
19762   because the symlink to include/ was missing.
19763
19764   *Ralf S. Engelschall*
19765
19766 * Incorporated the popular no-RSA/DSA-only patches
19767   which allow to compile an RSA-free SSLeay.
19768
19769   *Andrew Cooke / Interrader Ldt., Ralf S. Engelschall*
19770
19771 * Fixed nasty rehash problem under `make -f Makefile.ssl links`
19772   when "ssleay" is still not found.
19773
19774   *Ralf S. Engelschall*
19775
19776 * Added more platforms to Configure: Cray T3E, HPUX 11,
19777
19778   *Ralf S. Engelschall, Beckmann <beckman@acl.lanl.gov>*
19779
19780 * Updated the README file.
19781
19782   *Ralf S. Engelschall*
19783
19784 * Added various .cvsignore files in the CVS repository subdirs
19785   to make a "cvs update" really silent.
19786
19787   *Ralf S. Engelschall*
19788
19789 * Recompiled the error-definition header files and added
19790   missing symbols to the Win32 linker tables.
19791
19792   *Ralf S. Engelschall*
19793
19794 * Cleaned up the top-level documents;
19795   o new files: CHANGES and LICENSE
19796   o merged VERSION, HISTORY* and README* files a CHANGES.SSLeay
19797   o merged COPYRIGHT into LICENSE
19798   o removed obsolete TODO file
19799   o renamed MICROSOFT to INSTALL.W32
19800
19801   *Ralf S. Engelschall*
19802
19803 * Removed dummy files from the 0.9.1b source tree:
19804   crypto/asn1/x crypto/bio/cd crypto/bio/fg crypto/bio/grep crypto/bio/vi
19805   crypto/bn/asm/......add.c crypto/bn/asm/a.out crypto/dsa/f crypto/md5/f
19806   crypto/pem/gmon.out crypto/perlasm/f crypto/pkcs7/build crypto/rsa/f
19807   crypto/sha/asm/f crypto/threads/f ms/zzz ssl/f ssl/f.mak test/f
19808   util/f.mak util/pl/f util/pl/f.mak crypto/bf/bf_locl.old apps/f
19809
19810   *Ralf S. Engelschall*
19811
19812 * Added various platform portability fixes.
19813
19814   *Mark J. Cox*
19815
19816 * The Genesis of the OpenSSL rpject:
19817   We start with the latest (unreleased) SSLeay version 0.9.1b which Eric A.
19818   Young and Tim J. Hudson created while they were working for C2Net until
19819   summer 1998.
19820
19821   *The OpenSSL Project*
19822
19823### Changes between 0.9.0b and 0.9.1b  [not released]
19824
19825 * Updated a few CA certificates under certs/
19826
19827   *Eric A. Young*
19828
19829 * Changed some BIGNUM api stuff.
19830
19831   *Eric A. Young*
19832
19833 * Various platform ports: OpenBSD, Ultrix, IRIX 64bit, NetBSD,
19834   DGUX x86, Linux Alpha, etc.
19835
19836   *Eric A. Young*
19837
19838 * New COMP library [crypto/comp/] for SSL Record Layer Compression:
19839   RLE (dummy implemented) and ZLIB (really implemented when ZLIB is
19840   available).
19841
19842   *Eric A. Young*
19843
19844 * Add -strparse option to asn1pars program which parses nested
19845   binary structures
19846
19847   *Dr Stephen Henson <shenson@bigfoot.com>*
19848
19849 * Added "oid_file" to ssleay.cnf for "ca" and "req" programs.
19850
19851   *Eric A. Young*
19852
19853 * DSA fix for "ca" program.
19854
19855   *Eric A. Young*
19856
19857 * Added "-genkey" option to "dsaparam" program.
19858
19859   *Eric A. Young*
19860
19861 * Added RIPE MD160 (rmd160) message digest.
19862
19863   *Eric A. Young*
19864
19865 * Added -a (all) option to "ssleay version" command.
19866
19867   *Eric A. Young*
19868
19869 * Added PLATFORM define which is the id given to Configure.
19870
19871   *Eric A. Young*
19872
19873 * Added MemCheck_XXXX functions to crypto/mem.c for memory checking.
19874
19875   *Eric A. Young*
19876
19877 * Extended the ASN.1 parser routines.
19878
19879   *Eric A. Young*
19880
19881 * Extended BIO routines to support REUSEADDR, seek, tell, etc.
19882
19883   *Eric A. Young*
19884
19885 * Added a BN_CTX to the BN library.
19886
19887   *Eric A. Young*
19888
19889 * Fixed the weak key values in DES library
19890
19891   *Eric A. Young*
19892
19893 * Changed API in EVP library for cipher aliases.
19894
19895   *Eric A. Young*
19896
19897 * Added support for RC2/64bit cipher.
19898
19899   *Eric A. Young*
19900
19901 * Converted the lhash library to the crypto/mem.c functions.
19902
19903   *Eric A. Young*
19904
19905 * Added more recognized ASN.1 object ids.
19906
19907   *Eric A. Young*
19908
19909 * Added more RSA padding checks for SSL/TLS.
19910
19911   *Eric A. Young*
19912
19913 * Added BIO proxy/filter functionality.
19914
19915   *Eric A. Young*
19916
19917 * Added extra_certs to SSL_CTX which can be used
19918   send extra CA certificates to the client in the CA cert chain sending
19919   process. It can be configured with SSL_CTX_add_extra_chain_cert().
19920
19921   *Eric A. Young*
19922
19923 * Now Fortezza is denied in the authentication phase because
19924   this is key exchange mechanism is not supported by SSLeay at all.
19925
19926   *Eric A. Young*
19927
19928 * Additional PKCS1 checks.
19929
19930   *Eric A. Young*
19931
19932 * Support the string "TLSv1" for all TLS v1 ciphers.
19933
19934   *Eric A. Young*
19935
19936 * Added function SSL_get_ex_data_X509_STORE_CTX_idx() which gives the
19937   ex_data index of the SSL context in the X509_STORE_CTX ex_data.
19938
19939   *Eric A. Young*
19940
19941 * Fixed a few memory leaks.
19942
19943   *Eric A. Young*
19944
19945 * Fixed various code and comment typos.
19946
19947   *Eric A. Young*
19948
19949 * A minor bug in ssl/s3_clnt.c where there would always be 4 0
19950   bytes sent in the client random.
19951
19952   *Edward Bishop <ebishop@spyglass.com>*
19953
19954<!-- Links -->
19955
19956[CVE-2024-13176]: https://www.openssl.org/news/vulnerabilities.html#CVE-2024-13176
19957[CVE-2024-9143]: https://www.openssl.org/news/vulnerabilities.html#CVE-2024-9143
19958[CVE-2024-6119]: https://www.openssl.org/news/vulnerabilities.html#CVE-2024-6119
19959[CVE-2024-5535]: https://www.openssl.org/news/vulnerabilities.html#CVE-2024-5535
19960[CVE-2024-4741]: https://www.openssl.org/news/vulnerabilities.html#CVE-2024-4741
19961[CVE-2024-4603]: https://www.openssl.org/news/vulnerabilities.html#CVE-2024-4603
19962[CVE-2024-2511]: https://www.openssl.org/news/vulnerabilities.html#CVE-2024-2511
19963[CVE-2024-0727]: https://www.openssl.org/news/vulnerabilities.html#CVE-2024-0727
19964[CVE-2023-6237]: https://www.openssl.org/news/vulnerabilities.html#CVE-2023-6237
19965[CVE-2023-6129]: https://www.openssl.org/news/vulnerabilities.html#CVE-2023-6129
19966[CVE-2023-5678]: https://www.openssl.org/news/vulnerabilities.html#CVE-2023-5678
19967[CVE-2023-5363]: https://www.openssl.org/news/vulnerabilities.html#CVE-2023-5363
19968[CVE-2023-4807]: https://www.openssl.org/news/vulnerabilities.html#CVE-2023-4807
19969[CVE-2023-3817]: https://www.openssl.org/news/vulnerabilities.html#CVE-2023-3817
19970[CVE-2023-3446]: https://www.openssl.org/news/vulnerabilities.html#CVE-2023-3446
19971[CVE-2023-2975]: https://www.openssl.org/news/vulnerabilities.html#CVE-2023-2975
19972[RFC 2578 (STD 58), section 3.5]: https://datatracker.ietf.org/doc/html/rfc2578#section-3.5
19973[CVE-2023-2650]: https://www.openssl.org/news/vulnerabilities.html#CVE-2023-2650
19974[CVE-2023-1255]: https://www.openssl.org/news/vulnerabilities.html#CVE-2023-1255
19975[CVE-2023-0466]: https://www.openssl.org/news/vulnerabilities.html#CVE-2023-0466
19976[CVE-2023-0465]: https://www.openssl.org/news/vulnerabilities.html#CVE-2023-0465
19977[CVE-2023-0464]: https://www.openssl.org/news/vulnerabilities.html#CVE-2023-0464
19978[CVE-2023-0401]: https://www.openssl.org/news/vulnerabilities.html#CVE-2023-0401
19979[CVE-2023-0286]: https://www.openssl.org/news/vulnerabilities.html#CVE-2023-0286
19980[CVE-2023-0217]: https://www.openssl.org/news/vulnerabilities.html#CVE-2023-0217
19981[CVE-2023-0216]: https://www.openssl.org/news/vulnerabilities.html#CVE-2023-0216
19982[CVE-2023-0215]: https://www.openssl.org/news/vulnerabilities.html#CVE-2023-0215
19983[CVE-2022-4450]: https://www.openssl.org/news/vulnerabilities.html#CVE-2022-4450
19984[CVE-2022-4304]: https://www.openssl.org/news/vulnerabilities.html#CVE-2022-4304
19985[CVE-2022-4203]: https://www.openssl.org/news/vulnerabilities.html#CVE-2022-4203
19986[CVE-2022-3996]: https://www.openssl.org/news/vulnerabilities.html#CVE-2022-3996
19987[CVE-2022-2274]: https://www.openssl.org/news/vulnerabilities.html#CVE-2022-2274
19988[CVE-2022-2097]: https://www.openssl.org/news/vulnerabilities.html#CVE-2022-2097
19989[CVE-2020-1971]: https://www.openssl.org/news/vulnerabilities.html#CVE-2020-1971
19990[CVE-2020-1967]: https://www.openssl.org/news/vulnerabilities.html#CVE-2020-1967
19991[CVE-2019-1563]: https://www.openssl.org/news/vulnerabilities.html#CVE-2019-1563
19992[CVE-2019-1559]: https://www.openssl.org/news/vulnerabilities.html#CVE-2019-1559
19993[CVE-2019-1552]: https://www.openssl.org/news/vulnerabilities.html#CVE-2019-1552
19994[CVE-2019-1551]: https://www.openssl.org/news/vulnerabilities.html#CVE-2019-1551
19995[CVE-2019-1549]: https://www.openssl.org/news/vulnerabilities.html#CVE-2019-1549
19996[CVE-2019-1547]: https://www.openssl.org/news/vulnerabilities.html#CVE-2019-1547
19997[CVE-2019-1543]: https://www.openssl.org/news/vulnerabilities.html#CVE-2019-1543
19998[CVE-2018-5407]: https://www.openssl.org/news/vulnerabilities.html#CVE-2018-5407
19999[CVE-2018-0739]: https://www.openssl.org/news/vulnerabilities.html#CVE-2018-0739
20000[CVE-2018-0737]: https://www.openssl.org/news/vulnerabilities.html#CVE-2018-0737
20001[CVE-2018-0735]: https://www.openssl.org/news/vulnerabilities.html#CVE-2018-0735
20002[CVE-2018-0734]: https://www.openssl.org/news/vulnerabilities.html#CVE-2018-0734
20003[CVE-2018-0733]: https://www.openssl.org/news/vulnerabilities.html#CVE-2018-0733
20004[CVE-2018-0732]: https://www.openssl.org/news/vulnerabilities.html#CVE-2018-0732
20005[CVE-2017-3738]: https://www.openssl.org/news/vulnerabilities.html#CVE-2017-3738
20006[CVE-2017-3737]: https://www.openssl.org/news/vulnerabilities.html#CVE-2017-3737
20007[CVE-2017-3736]: https://www.openssl.org/news/vulnerabilities.html#CVE-2017-3736
20008[CVE-2017-3735]: https://www.openssl.org/news/vulnerabilities.html#CVE-2017-3735
20009[CVE-2017-3733]: https://www.openssl.org/news/vulnerabilities.html#CVE-2017-3733
20010[CVE-2017-3732]: https://www.openssl.org/news/vulnerabilities.html#CVE-2017-3732
20011[CVE-2017-3731]: https://www.openssl.org/news/vulnerabilities.html#CVE-2017-3731
20012[CVE-2017-3730]: https://www.openssl.org/news/vulnerabilities.html#CVE-2017-3730
20013[CVE-2016-7055]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-7055
20014[CVE-2016-7054]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-7054
20015[CVE-2016-7053]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-7053
20016[CVE-2016-7052]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-7052
20017[CVE-2016-6309]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-6309
20018[CVE-2016-6308]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-6308
20019[CVE-2016-6307]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-6307
20020[CVE-2016-6306]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-6306
20021[CVE-2016-6305]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-6305
20022[CVE-2016-6304]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-6304
20023[CVE-2016-6303]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-6303
20024[CVE-2016-6302]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-6302
20025[CVE-2016-2183]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-2183
20026[CVE-2016-2182]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-2182
20027[CVE-2016-2181]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-2181
20028[CVE-2016-2180]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-2180
20029[CVE-2016-2179]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-2179
20030[CVE-2016-2178]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-2178
20031[CVE-2016-2177]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-2177
20032[CVE-2016-2176]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-2176
20033[CVE-2016-2109]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-2109
20034[CVE-2016-2107]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-2107
20035[CVE-2016-2106]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-2106
20036[CVE-2016-2105]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-2105
20037[CVE-2016-0800]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-0800
20038[CVE-2016-0799]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-0799
20039[CVE-2016-0798]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-0798
20040[CVE-2016-0797]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-0797
20041[CVE-2016-0705]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-0705
20042[CVE-2016-0702]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-0702
20043[CVE-2016-0701]: https://www.openssl.org/news/vulnerabilities.html#CVE-2016-0701
20044[CVE-2015-3197]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-3197
20045[CVE-2015-3196]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-3196
20046[CVE-2015-3195]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-3195
20047[CVE-2015-3194]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-3194
20048[CVE-2015-3193]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-3193
20049[CVE-2015-1793]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-1793
20050[CVE-2015-1792]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-1792
20051[CVE-2015-1791]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-1791
20052[CVE-2015-1790]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-1790
20053[CVE-2015-1789]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-1789
20054[CVE-2015-1788]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-1788
20055[CVE-2015-1787]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-1787
20056[CVE-2015-0293]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0293
20057[CVE-2015-0291]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0291
20058[CVE-2015-0290]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0290
20059[CVE-2015-0289]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0289
20060[CVE-2015-0288]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0288
20061[CVE-2015-0287]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0287
20062[CVE-2015-0286]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0286
20063[CVE-2015-0285]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0285
20064[CVE-2015-0209]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0209
20065[CVE-2015-0208]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0208
20066[CVE-2015-0207]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0207
20067[CVE-2015-0206]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0206
20068[CVE-2015-0205]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0205
20069[CVE-2015-0204]: https://www.openssl.org/news/vulnerabilities.html#CVE-2015-0204
20070[CVE-2014-8275]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-8275
20071[CVE-2014-5139]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-5139
20072[CVE-2014-3572]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3572
20073[CVE-2014-3571]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3571
20074[CVE-2014-3570]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3570
20075[CVE-2014-3569]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3569
20076[CVE-2014-3568]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3568
20077[CVE-2014-3567]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3567
20078[CVE-2014-3566]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3566
20079[CVE-2014-3513]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3513
20080[CVE-2014-3512]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3512
20081[CVE-2014-3511]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3511
20082[CVE-2014-3510]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3510
20083[CVE-2014-3509]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3509
20084[CVE-2014-3508]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3508
20085[CVE-2014-3507]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3507
20086[CVE-2014-3506]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3506
20087[CVE-2014-3505]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3505
20088[CVE-2014-3470]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-3470
20089[CVE-2014-0224]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-0224
20090[CVE-2014-0221]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-0221
20091[CVE-2014-0195]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-0195
20092[CVE-2014-0160]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-0160
20093[CVE-2014-0076]: https://www.openssl.org/news/vulnerabilities.html#CVE-2014-0076
20094[CVE-2013-6450]: https://www.openssl.org/news/vulnerabilities.html#CVE-2013-6450
20095[CVE-2013-4353]: https://www.openssl.org/news/vulnerabilities.html#CVE-2013-4353
20096[CVE-2013-0169]: https://www.openssl.org/news/vulnerabilities.html#CVE-2013-0169
20097[CVE-2013-0166]: https://www.openssl.org/news/vulnerabilities.html#CVE-2013-0166
20098[CVE-2012-2686]: https://www.openssl.org/news/vulnerabilities.html#CVE-2012-2686
20099[CVE-2012-2333]: https://www.openssl.org/news/vulnerabilities.html#CVE-2012-2333
20100[CVE-2012-2110]: https://www.openssl.org/news/vulnerabilities.html#CVE-2012-2110
20101[CVE-2012-0884]: https://www.openssl.org/news/vulnerabilities.html#CVE-2012-0884
20102[CVE-2012-0050]: https://www.openssl.org/news/vulnerabilities.html#CVE-2012-0050
20103[CVE-2012-0027]: https://www.openssl.org/news/vulnerabilities.html#CVE-2012-0027
20104[CVE-2011-4619]: https://www.openssl.org/news/vulnerabilities.html#CVE-2011-4619
20105[CVE-2011-4577]: https://www.openssl.org/news/vulnerabilities.html#CVE-2011-4577
20106[CVE-2011-4576]: https://www.openssl.org/news/vulnerabilities.html#CVE-2011-4576
20107[CVE-2011-4109]: https://www.openssl.org/news/vulnerabilities.html#CVE-2011-4109
20108[CVE-2011-4108]: https://www.openssl.org/news/vulnerabilities.html#CVE-2011-4108
20109[CVE-2011-3210]: https://www.openssl.org/news/vulnerabilities.html#CVE-2011-3210
20110[CVE-2011-3207]: https://www.openssl.org/news/vulnerabilities.html#CVE-2011-3207
20111[CVE-2011-0014]: https://www.openssl.org/news/vulnerabilities.html#CVE-2011-0014
20112[CVE-2010-4252]: https://www.openssl.org/news/vulnerabilities.html#CVE-2010-4252
20113[CVE-2010-4180]: https://www.openssl.org/news/vulnerabilities.html#CVE-2010-4180
20114[CVE-2010-3864]: https://www.openssl.org/news/vulnerabilities.html#CVE-2010-3864
20115[CVE-2010-1633]: https://www.openssl.org/news/vulnerabilities.html#CVE-2010-1633
20116[CVE-2010-0740]: https://www.openssl.org/news/vulnerabilities.html#CVE-2010-0740
20117[CVE-2010-0433]: https://www.openssl.org/news/vulnerabilities.html#CVE-2010-0433
20118[CVE-2009-4355]: https://www.openssl.org/news/vulnerabilities.html#CVE-2009-4355
20119[CVE-2009-3555]: https://www.openssl.org/news/vulnerabilities.html#CVE-2009-3555
20120[CVE-2009-3245]: https://www.openssl.org/news/vulnerabilities.html#CVE-2009-3245
20121[CVE-2009-1386]: https://www.openssl.org/news/vulnerabilities.html#CVE-2009-1386
20122[CVE-2009-1379]: https://www.openssl.org/news/vulnerabilities.html#CVE-2009-1379
20123[CVE-2009-1378]: https://www.openssl.org/news/vulnerabilities.html#CVE-2009-1378
20124[CVE-2009-1377]: https://www.openssl.org/news/vulnerabilities.html#CVE-2009-1377
20125[CVE-2009-0789]: https://www.openssl.org/news/vulnerabilities.html#CVE-2009-0789
20126[CVE-2009-0591]: https://www.openssl.org/news/vulnerabilities.html#CVE-2009-0591
20127[CVE-2009-0590]: https://www.openssl.org/news/vulnerabilities.html#CVE-2009-0590
20128[CVE-2008-5077]: https://www.openssl.org/news/vulnerabilities.html#CVE-2008-5077
20129[CVE-2008-1678]: https://www.openssl.org/news/vulnerabilities.html#CVE-2008-1678
20130[CVE-2008-1672]: https://www.openssl.org/news/vulnerabilities.html#CVE-2008-1672
20131[CVE-2008-0891]: https://www.openssl.org/news/vulnerabilities.html#CVE-2008-0891
20132[CVE-2007-5135]: https://www.openssl.org/news/vulnerabilities.html#CVE-2007-5135
20133[CVE-2007-4995]: https://www.openssl.org/news/vulnerabilities.html#CVE-2007-4995
20134[CVE-2006-4343]: https://www.openssl.org/news/vulnerabilities.html#CVE-2006-4343
20135[CVE-2006-4339]: https://www.openssl.org/news/vulnerabilities.html#CVE-2006-4339
20136[CVE-2006-3738]: https://www.openssl.org/news/vulnerabilities.html#CVE-2006-3738
20137[CVE-2006-2940]: https://www.openssl.org/news/vulnerabilities.html#CVE-2006-2940
20138[CVE-2006-2937]: https://www.openssl.org/news/vulnerabilities.html#CVE-2006-2937
20139[CVE-2005-2969]: https://www.openssl.org/news/vulnerabilities.html#CVE-2005-2969
20140[CVE-2004-0112]: https://www.openssl.org/news/vulnerabilities.html#CVE-2004-0112
20141[CVE-2004-0079]: https://www.openssl.org/news/vulnerabilities.html#CVE-2004-0079
20142[CVE-2003-0851]: https://www.openssl.org/news/vulnerabilities.html#CVE-2003-0851
20143[CVE-2003-0545]: https://www.openssl.org/news/vulnerabilities.html#CVE-2003-0545
20144[CVE-2003-0544]: https://www.openssl.org/news/vulnerabilities.html#CVE-2003-0544
20145[CVE-2003-0543]: https://www.openssl.org/news/vulnerabilities.html#CVE-2003-0543
20146[CVE-2003-0078]: https://www.openssl.org/news/vulnerabilities.html#CVE-2003-0078
20147[CVE-2002-0659]: https://www.openssl.org/news/vulnerabilities.html#CVE-2002-0659
20148[CVE-2002-0657]: https://www.openssl.org/news/vulnerabilities.html#CVE-2002-0657
20149[CVE-2002-0656]: https://www.openssl.org/news/vulnerabilities.html#CVE-2002-0656
20150[CVE-2002-0655]: https://www.openssl.org/news/vulnerabilities.html#CVE-2002-0655
20151